IMPACTOS DEL CAMBIO GLOBAL DOCTORAL EN ECOSISTEMAS...

20

15

LA

UR

A H

ERN

ÁN

DEZ

MA

TEO

TE

SIS

DO

CTO

RA

L IMPACTOS DEL CAMBIO GLOBAL EN ECOSISTEMAS FORESTALES IBÉRICOS A

PARTIR DEL INVENTARIO FORESTAL NACIONAL

IMPACTS OF GLOBAL CHANGE ON IBERIAN FOREST ECOSYSTEMS INFERRED FROM THE SPANISH NATIONAL FOREST

INVENTORY

LAURA HERNÁNDEZ MATEO

TESIS DOCTORAL ESCUELA TÉCNICA SUPERIOR DE INGENIEROS DE MONTES

UNIVERSIDAD POLITÉCNICA DE MADRID

2015

UNIVERSIDAD POLITÉCNICA DE MADRID ESCUELA TÉCNICA SUPERIOR DE INGENIEROS DE MONTES

IMPACTOS DEL CAMBIO GLOBAL EN ECOSISTEMAS

FORESTALES IBÉRICOS A PARTIR DEL INVENTARIO

FORESTAL NACIONAL

IMPACTS OF GLOBAL CHANGE ON IBERIAN FOREST ECOSYSTEMS

INFERRED FROM THE SPANISH NATIONAL FOREST INVENTORY

Tesis doctoral

LAURA HERNÁNDEZ MATEO Ingeniero de Montes

2015

PROGRAMA DE DOCTORADO EN INVESTIGACIÓN FORESTAL AVANZADA

ESCUELA TÉCNICA SUPERIOR DE INGENIEROS DE MONTES

IMPACTOS DEL CAMBIO GLOBAL EN ECOSISTEMAS FORESTALES

IBÉRICOS A PARTIR DEL INVENTARIO FORESTAL NACIONAL

IMPACTS OF GLOBAL CHANGE ON IBERIAN FOREST ECOSYSTEMS

INFERRED FROM THE SPANISH NATIONAL FOREST INVENTORY

LAURA HERNÁNDEZ MATEO Ingeniero de Montes

DIRECTORES

ISABEL CAÑELLAS FERNANDO MONTES Dr. Ingeniero de Montes Dr. Ingeniero de Montes

MADRID, 2015

Tribunal nombrado por el Mgfco. y Excmo. Sr. Rector de la Universidad Politécnica de Madrid, el día…….. de …………………………… de 2015 Presidente D ……………………………………………………… Vocal D.……………………………………………………… Vocal D.……………………………………………………… Vocal D.……………………………………………………… Secretario D...……………………………………………………..

Realizado el acto de defensa y lectura de la Tesis el día …. de…………. de 2016 en Madrid.

Calificación……………………………………………………..

EL PRESIDENTE LOS VOCALES

EL SECRETARIO

A mi hija Claudia.

“When one tugs at a single thing in nature, he finds it attached to the rest of the world.”

― John Muir

AGRADECIMIENTOS

“Somos, desde que nacemos, exploradores de un mundo complejo y fascinante”.

― Gerald Durrel Esta frase del famoso naturalista británico introduce muy bien algunos de los factores necesarios para desarrollar una tesis doctoral: potencial para plantearte cuestiones sobre lo que te rodea, capacidad para aportar respuestas y aprender de los errores, y, perseverancia. Aunque sin lugar a dudas, el factor determinante para llevarla a buen puerto es la colaboración y ayuda de los compañeros con los que compartes el viaje. Así, en primer lugar quiero expresar mi más profundo agradecimiento a los directores de la tesis, Isabel Cañellas y Fernando Montes. Hace ya cinco años, Isabel me brindó la oportunidad de realizar una Tesis Doctoral combinando el trabajo técnico relacionado con el IFN que realizo en el INIA y mi pasión por la botánica y la biogeografría. Trabajadora incansable, siempre ha encontrado tiempo para resolver dudas, leer y comentar mis numerosos borradores, y facilitar el compaginar mi labor técnica y científica. Gracias por tu entereza y apoyo durante estos años, y sobre todo por confiar en mi capacidad y darme la suficiente libertad para disfrutar de mi vocación científica. A Fernando tengo que agradecerle muchas horas de trabajo. Gracias por todas esas tardes de tormenta de ideas, horas de programación y mucha prueba y error. Sin tu ayuda, la tesis nunca hubiera llegado a buen puerto. Además, con tu paciencia, has sabido reconducir mí a veces excesiva motivación y ponerme los pies en el suelo (aunque al final, me saliera aquella vez con la mía, y termináramos dejando la frase del urogallo en el artículo…). Mi agradecimiento también a todos los compañeros del CIFOR por su apoyo y empuje, sobre todo en la recta final de la elaboración de la tesis. Especialmente tengo mucho que agradecer a Iciar. La recta final de la tesis ha sido dura, pero mucho más llevadera en su compañía. La alegría y optimismo que desprende, se contagia, y me siento afortunada de compartir el día a día con una persona con su calidad humana. Por su implicación directa o indirecta en la tesis, agradezco también a Ángel esos viajes “exprés” a la Rioja (incluso con metro y medio de nieve, y un frio del carajo) para ir en busca y captura de las fotos aéreas del IFN1; también a Aurora y Ernesto por su ayuda georreferenciándolas; y a Roberto y Vicente, del MAGRAMA, por todo su apoyo y colaboración durante estos años. Nunca me hubiera embarcado en este viaje sin el legado que dejó en mí el paso por la UD de Botánica de la ETSI de Montes durante mi etapa formativa como Ingeniero de Montes. Allí fue donde di mis primeros pasos en este dinámico mundo que es la ciencia y donde adquirí mis primeras herramientas con las que canalizar mi vocación científica. Por todo ello, gracias.

También quiero expresar mi gratitud a los investigadores que me han acogido en sus departamentos en las breves estancias de investigación que he hecho en el camino en el CITA, IPE e INRA. Eustaquio, Chechu, Nacho gracias por vuestro tiempo e inspiración. Además, tengo mucho que agradecer a Rut, no sólo por brindarme la oportunidad de hacer una estancia en la UCM hace unos años, sino por su amistad. Ha sido y es un verdadero placer trabajar en su compañía, y espero que estas primeras colaboraciones sean sólo un ejemplo de las muchas que tenemos por delante.

Mucha de la inspiración y la energía necesaria para realizar la tesis la he encontrado rodeada de naturaleza y montañas. Tengo la suerte de compartir aficiones con grandes personas, científicos y naturalistas de las que he recibido mucho apoyo e inspiración. Iván, Paula, Cesar, Gaby, Irene, Justi, Vir, Pablos, Manu, Javi, Sergio, Antonio, Luis… Gracias. Mi más afectuoso agradecimiento va dirigido a toda mi gente y familia por su apoyo incondicional durante estos años. Sobre todo, agradezco a mis padres, Carmen y Luis, todas las oportunidades que me han ofrecido a lo largo de mi vida, pero sobre todo el haberme trasmitido valores, respeto, responsabilidad, y, por alimentar mi curiosidad y capacidad de asombro. Tanto ellos, como mi hermana Paula, han sido mi punto de referencia y mi más firme sostén siempre. Por último, por su apoyo y comprensión en este viaje con altibajos, tengo mucho que agradecer a Iván, la persona con la que comparto mi vida y el proyecto de vida que en estos momentos siento en mi vientre. A ese nuevo proyecto va dedicada la tesis. A ti, Claudia. Esta tesis doctoral ha sido financiada por las encomiendas de gestión entre el MAGRAMA y el INIA-CIFOR, AEG-09-007 y EG-13-072, y el proyecto de Plan Nacional de I+D+i, AGL2010.21153.

ÍNDICE RESUMEN 1 SUMMARY 3 1. INTRODUCCIÓN GENERAL

1.1. El impacto del cambio global en los bosques 7 1.1.1. Cambios en distribución y abundancia de especies 10 1.1.2. Un nuevo reto ambiental: la invasión de los ecosistemas forestales por flora exótica invasora 15 1.2. Los bosques ibéricos y el cambio global 19 1.3. El Inventario Forestal Nacional como instrumento para responder a los nuevos retos en ecología y gestión forestal 23 1.4. Referencias bibliográficas 25

2. OBJETIVOS Y ESTRUCTURA DE LA TESIS 2.1. Objetivos 39 2.2. Estructura de la tesis 40

3. ASSESSING CHANGES IN SPECIES DISTRIBUTION FROM SEQUENTIAL LARGE SCALE FOREST INVENTORIES Resumen /Abstract 43 3.1. Introduction 45 3.2. Material and Methods

3.2.1 Study area 46 3.2.2. NFI dataset 47 3.2.3. Universal Kriging 48 3.2.4. Crossvalidation of the UK model 49 3.2.5. Testing the significance of the changes in species distribution between inventories 50 3.2.6. Mapping changes in species spatial distribution 50 3.2.7. Software used 50

3.3. Results 3.3.1. Variogram models 51 3.3.2. Mapping forest range change 51 3.3.3. Assessing species distribution change between inventories 53 3.3.3.1. Latitudinal shifts 54 3.3.3.2. Shifts in elevation 54 3.3.3.3. Changes in exposure distribution 54 3.3.4. Trade-off between altitude/exposure and latitude 54

3.4. Discussion 3.4.1. A new method for assessing changes in species distribution from long term forest inventories 55 3.4.2. Changes in the distribution of forests 57 3.4.3. Conclusion 58

3.5. References 59

4. EXPLORING RANGE SHIFTS OF CONTRASTING TREE SPECIES ALONG BIOCLIMATIC AND ELEVATIONAL GRADIENTS Resumen / Abstract 65 4.1. Introduction 67 4.2. Material and Methods

4.2.1. Study area 69 4.2.2. Characterization of the bioclimatic gradient 71 4.2.3. NFI data 71 4.2.4. Statistical analysis

4.2.4.1. Universal kriging models 72 4.2.4.2. Species range shifts along the bioclimatic and elevational gradients 72 4.2.4.3. Changes in composition 73

4.2.4.4. Past land use changes 74 4.2.5. Software employed 74

4.3. Results 4.3.1. Climatic trends in the study area 74 4.3.2. Universal Kriging models 75 4.3.3. Species range shifts along the bioclimatic and elevational gradients 76 4.3.4. Changes in species co-occurrence 79 4.3.5. Past land use changes 81

4.4. Discussion 81 4.5. Conclusions 84 4.6. References 84

5. TRACKING FAGUS SYLVATICA LEADING EDGE IN NORTH-WESTERN IBERIA: HOLOCENE MIGRATION INERTIA, FOREST RECOVERY AND RECENT GLOBAL CHANGE

Resumen / Abstract 91 5.1. Introduction 93 5.2. Material and Methods

5.2.1. Analyzed data 95 5.2.2. Data analysized 96 5.2.3. Data analysis 98

5.3. Results 5.3.1. Mortality rates and dbh structure of living and dead standing trees 100

5.3.2. Basal area increment 101 5.3.3. Recruitment patterns 101 5.3.4. Biotic and abiotic factors affecting basal area increment, dominance and recruitment categories 104 5.4. Discussion 5.4.1. Fagus sylvatica expansion in North-Western Iberia 107 5.4.2. Fagus sylvatica and Quercus petraea biotic interactions 108 5.4.3. Global change and forest conservation 109 5.5. Conclusions 110 5.6. References 110

6. ASSESSING SPATIO-TEMPORAL RATES, PATTERNS AND DETERMINANTS OF BIOLOGICAL INVASIONS IN FOREST ECOSYSTEMS. THE CASE OF ACACIA SPECIES IN NW SPAIN.

Resumen /Abstract 119 6.1. Introduction 121 6.2. Material and Methods

6.2.1. Study area 122 6.2.2. Data used 124 6.2.3. Data analysis 127

6.3. Results 6.3.1. Changes in the distribution of Acacia species 128 6.3.2. Acacia species dynamics and dbh structure 130 6.3.3. Factors involved in the spread of Acacia species 131

6.4. Discussion 6.4.1. Spatio-temporal dynamics 132 6.4.2. Factors associated with Acacia species spread and invisibility 133

6.5. Conclusions 135 6.6. References 136

7. DISCUSIÓN GENERAL 7.1. Cambios en la distribución de los bosques 143 7.2. Un nuevo factor en ecología y gestión forestal: las invasiones biológicas 148 7.3. Referencias biobliográficas 150

8. CONCLUSIONES 159 9. ANEXOS (ARTÍCULOS PUBLICADOS EN REVISTAS SCI) Anexo1 164 Anexo2 179

1

RESUMEN

Los efectos del cambio global sobre los bosques son una de las grandes preocupaciones de la sociedad del siglo XXI. Algunas de sus posibles consecuencias como son los efectos en la producción, la sostenibilidad, la pérdida de biodiversidad o cambios en la distribución y ensamblaje de especies forestales pueden tener grandes repercusiones sociales, ecológicas y económicas. La detección y seguimiento de estos efectos constituyen uno de los retos a los que se enfrentan en la actualidad científicos y gestores forestales. En base a la comparación de series históricas del Inventario Forestal Nacional Español (IFN), esta tesis trata de arrojar luz sobre algunos de los impactos que los cambios socioeconómicos y ambientales de las últimas décadas han generado sobre nuestros bosques. En primer lugar, esta tesis presenta una innovadora metodología con base geoestadística que permite la comparación de diferentes ciclos de inventario sin importar los diferentes métodos de muestreo empleados en cada uno de ellos (Capítulo 3). Esta metodología permite analizar cambios en la dinámica y distribución espacial de especies forestales en diferentes gradientes geográficos. Mediante su aplicación, se constatarán y cuantificarán algunas de las primeras evidencias de cambio en la distribución altitudinal y latitudinal de diferentes especies forestales ibéricas, que junto al estudio de su dinámica poblacional y tasas demográficas, ayudarán a testar algunas hipótesis biogeográficas en un escenario de cambio global en zonas de especial vulnerabilidad (Capítulos 3, 4 y 5). Por último, mediante la comparación de ciclos de parcelas permanentes del IFN se ahondará en el conocimiento de la evolución en las últimas décadas de especies invasoras en los ecosistemas forestales del cuadrante noroccidental ibérico, uno de los más afectados por la invasión de esta flora (Capítulo 6).

Palabras clave: Cambio global, ecosistemas forestales ibéricos, geoestadística, ecología forestal, biogeografía, migraciones latitudinales y altitudinales, dinámica y sucesión forestal, invasiones biológicas.

3

SUMMARY

The effects of global change on forests are one of the major concerns of the XXI century. Some of the potential impacts of global change on forest growth, productivity, biodiversity or changes in species assembly and spatial distribution may have great ecological and economic consequences. The detection and monitoring of these effects are some of the major challenges that scientists and forest managers face nowadays. Based on the comparison of historical series of the Spanish National Forest Inventory (NFI), this thesis tries to shed some light on some of the impacts driven by recent socio-economic and environmental changes on our forest ecosystems. Firstly, this thesis presents an innovative methodology based on geostatistical techniques that allows the comparison of different NFI cycles regardless of the different sampling methods used in each of them (Chapter 3). This methodology, in conjunction with other statistical techniques, allows to analyze changes in the spatial distribution and population dynamics of forest species along different geographic gradients. By its application, this thesis presents some of the first evidences of changes in species distribution along different geographical gradients in the Iberian Peninsula. The analysis of these findings, of species population dynamics and demographic rates will help to test some biogeographical hypothesis on forests under climate change scenarios in areas of particular vulnerability (Chapters 3, 4 and 5). Finally, by comparing NFI cycles with permanent plots, this thesis increases our knowledge about the patterns and processes associated with the recent evolution of invasive species in the forest ecosystems of North-western Iberia, one of the areas most affected by the invasion of allien species at national scale (Chapter 6).

Keywords: Global Change, Iberian forest ecosystems, geostatistics, forest ecology, biogeography, latitudinal and altitudinal migration, forest succession, forest dynamics, biological invasions.

1. Introducción general

CAPITULO1: INTRODUCCIÓN GENERAL

7

1. Introducción

1.1. Efectos del Cambio Global sobre los bosques

Los bosques cubren aproximadamente un 31% de la superficie del planeta (Fig. 1.1.; FAO 2015) y su contribución al bienestar humano es extraordinariamente amplio. Los ecosistemas forestales son uno de los repositorios terrestres de biodiversidad más importantes, albergando cerca de dos tercios de la biodiversidad del planeta (MEA, 2005), juegan un papel clave en la regulación de los ciclos biogeoquímicos globales como el del agua y el carbono (Kappebelle et al. 1999), y proporcionan recursos y servicios esenciales a millones de personas. Los bosques suministran cerca de 5000 productos diferentes (Chiras 1998) y el sector forestal aporta aproximadamente el 2% del PIB mundial (FAO 2001).

Fig. 1.1. Tendencias actuales de la superficie forestal a nivel mundial y según regiones presentadas en el último informe sobre el estado de los bosques (Fuente: modificado de FAO (2015)).

El papel central de los bosques para la conservación de la Biodiversidad a nivel

global y su función como sumidero o fuente de carbono a considerar en las estrategias de mitigación del cambio climático ha sido reconocido a nivel mundial en recientes procesos internacionales, entre ellos la Convención de las Naciones Unidas sobre el Cambio Climático, el Convenio sobre la Diversidad Biológica, la Convención de las Naciones Unidas de Lucha contra la Desertificación y el Foro de las Naciones Unidas sobre los Bosques, entre otros (Keenan et al. 2015). Por todo ello, el impacto del acelerado cambio global contemporáneo en los bosques ha irrumpido como una de las grandes preocupaciones de la sociedad del siglo XXI (FAO, 2001).

El actual cambio ambiental global es el resultado del efecto sinérgico de las

actividades humanas sobre los sistemas terrestres (biosfera, hidrosfera, geosfera y atmósfera) y los procesos que determinan su funcionamiento. Los fenómenos que engloban el cambio global son muchos y presentan dinámicas interrelacionadas que se retroalimentan continuamente. El aumento de la población humana, cambios en el uso del suelo o del territorio, alteraciones en los ciclos biogeoquímicos, sobre todo los del agua y gases de efecto invernadero, alteraciones en la biodiversidad y en la interacción entre especies (principalmente especies invasoras), procesos de


8

desertización y el cambio climático son los factores de cambio más importantes, y todos ellos pueden tener un impacto significativo sobre los bosques. Sin embargo, la influencia relativa del impacto de estos factores sobre los ecosistemas terrestres varía entre ecoregiones y biomas (ver Fig. 1.2) (Chapin III et al. 2001; MEA 2005).

Fig. 1.2. Impacto durante el último siglo de diversos impulsores del cambio global sobre la biodiversidad de diferentes bosques y ecorregiones y las tendencias esperadas a corto plazo. (Fuente: modificado de MEA (2005)). El cuadro negro enmarca los factores, ecosistemas y regiones analizados en la presente tesis doctoral.

A lo largo de su historia, el planeta Tierra ha experimentado cambios ambientales globales derivados de la propia dinámica planetaria (tectónica de placas, vulcanismo, cambios en la órbita terrestre, proliferación de la vida, etc.) o catástrofes naturales (impactos de meteoritos). Salvo excepciones, estos cambios siempre se han producido a una escala temporal geológica, lo que ha posibilitado los procesos de adaptación y persistencia de muchos ecosistemas y de sus correspondientes componentes bióticos y abióticos. La singularidad del actual cambio global radica en que tiene a la especie humana como principal impulsor de alteración y en que se está desarrollando a un ritmo sin precedentes (Duarte et al. 2006). Esta singularidad es especialmente patente en el acelerado ritmo de cambio observado en la dinámica atmosférica como consecuencia directa del aumento en la emisión de gases de efecto invernadero durante el siglo XX (Houghton et al. 2001). Como recogen los últimos informes del Panel Intergubernamental de expertos sobre el Cambio Climático (IPCC) (IPCC 2013) el calentamiento en el sistema climático es inequívoco, y desde la década de 1950, muchos de los cambios observados no han tenido precedentes en los últimos decenios a milenios (Fig.1.3). En el hemisferio norte, 1983-2012 ha sido el período de 30 años más cálido de los últimos 1400 años y parece que la tendencia se mantiene.


9

Las últimas observaciones climáticas confirman que cada una de las tres últimas décadas han sido sucesivamente más cálidas que cualquier decenio anterior desde 1850 (IPCC 2013).

Fig.1.3. Reconstrucción de la variación media de la temperatura del Hemisferio Norte en los últimos 1000 años basado en datos instrumentales (últimos 100 años) y aproximaciones a partir de otros registros (fósiles, isótopos, etc.) (Fuente: modificado de BOM (2006)).

Las especies arbóreas pueden considerarse especies “clave”, ya que su

presencia estructura y caracteriza sustancialmente al conjunto de elementos bióticos y abióticos que definen un ecosistema. Por lo tanto, cambios en su composición, distribución y abundancia pueden suponer importantes consecuencias para toda la biodiversidad y servicios asociados a sus bosques (Walther et al. 2002). Sin embargo, la longevidad y la limitada capacidad de dispersión característica de las especies forestales (Jump y Peñuelas 2005; Hampe y Petit 2005), agravada por factores como la fragmentación del territorio y la interacción con otras especies, pueden dificultar la adaptación de los bosques a los rápidos cambios ambientales contemporáneos (Aitken et al. 2008; Lindner et al. 2010).

La detección y seguimiento de los efectos del cambio global sobre los

ecosistemas forestales constituyen uno de los grandes retos a los que se enfrentan en la actualidad científicos y gestores forestales. Con el fin de gestionar los bienes y servicios asociados a los bosques de forma sostenible es de vital importancia tener una comprensión profunda del alcance de los efectos del cambio global sobre los mismos, sus tendencias y los factores asociados (Hanewinkel et al. 2014; FAO 2015).


10

1.1.1. Cambios en distribución y abundancia de especies como consecuencia del cambio global La fitogeografía o geografía de las plantas es una rama de la biogeografía que

estudia la distribución espaciotemporal de la vegetación y su relación con factores del medio. Esta disciplina que nació hace aproximadamente 200 años cuando Alexander von Humboldt publicó el perfil de la vegetación asociada a la geología y a las bandas térmicas del volcán Chimborazo (Fig. 1.4.; Von Humboldt and Bonpland 1807), ha adquirido una especial relevancia en los últimos años en el estudio de la potencialidad climática de las especies vegetales para realizar predicciones bajo los supuesto de calentamiento global (Thomas et al. 2004; Thuiller et al. 2003), así como para analizar los actuales cambios en la distribución de las especies y analizar los factores de cambio asociados (Lenoir et al. 2009; Callaghan et al. 2013).

Fig. 1.4. Actualización, 200 años después, del primer perfil de vegetación altitudinal realizado por Humbolt, uno de los padres de la geografía de plantas (Fuente: Morueta-Holme et al. (2015)). Se muestra un resumen de los principales cambios en los límites de la distribución de la vegetación y los glaciares en el periodo 1802-2012. Las barras a la derecha representan los principales impulsores de la alteración: clima (el incremento de temperatura) y el cambio en el uso del territorio.


11

Caja.1.1. Relación de términos científicos en castellano e inglés relacionados con el subcapítulo 1.1.1. de la presente tesis doctoral y su signifcado.

En un contexto de cambio global, se espera que la vegetación responda a los rápidos cambios en las condiciones ambientales mediante desplazamientos hacia zonas con mejores condiciones ambientales para su persistencia, adaptándose a las nuevas condiciones del medio, o si esto no ocurre, mostrando señales de decaimiento e incluso extinciones locales (Aitken et al. 2008).

Esta respuesta no es nueva. La paleoecología muestra evidencias de desplazamientos altitudinales o latitudinales en el área de distribución de las especies vegetales, extinciones y reemplazo de unas especies por otras como respuesta a los cambios climáticos de los últimos milenios (Davis y Shaw, 2001; Rubiales et al. 2008). Asimismo, la mayoría de los modelos bioclimáticos pronostican un desplazamiento significativo del área de distribución de la vegetación hacia los polos como respuesta al esperado incremento de temperatura en el futuro (Guisan et al. 1998; Malcolm et al. 2002; Thomas et al. 2004). Desde la segunda mitad del siglo XX se realizó un importante esfuerzo en el desarrollo de modelos de distribución de especies que proyectaran los posibles efectos de cambio climático en la vegetación a diferentes escalas (Parmesan y Yohe 2003; Araújo et al. 2005) y teniendo en cuenta diferentes escenarios climáticos futuros basados en modelos de desarrollo (IPCC special report on emissions scenarios (SRES)). Estas proyecciones debían ser consideradas para la toma de decisiones en la gestión de la naturaleza y su conservación (Benito-Garzón et al. 2008). Sin embargo, debido a la importante incertidumbre que acompaña a estas predicciones, su uso en la elaboración de estrategias prácticas de gestión creó bastante controversia (Thuiller et al. 2008). En la última década y considerando el acelerado cambio global contemporáneo, la necesidad de contar con herramientas y conocimientos sólidos donde basar las estrategias de gestión ambiental (MEA 2005) han hecho derivar muchos de los esfuerzos científicos hacia la búsqueda de cambios recientes en el área de distribución de las especies, así como al análisis de sus alcances y factores asociados (ver la reciente revisión de Lenoir y Svenning 2015).

http://scholar.google.es/scholar?cluster=3063344234117434759&hl=en&as_sdt=2005&sciodt=0,5

http://scholar.google.es/scholar?cluster=3063344234117434759&hl=en&as_sdt=2005&sciodt=0,5


12

El cambio o desplazamiento del área de distribución de una especie (species range shift) se define como el proceso mediante el cual las especies cambian su distribución en el tiempo en busca de las condiciones ambientales que requieren para su desarrollo (Lenoir y Svenning 2013). Teóricamente, el área potencial de distribución de una especie, también conocida como el nicho fundamental o ecológico, está definida por todos los factores abióticos que delimitan su distribución (Hutchinson 1957). Sin embargo, debido a interacciones bióticas y factores históricos como cambios en el uso del suelo o migraciones históricas, la distribución geográfica pocas veces representa plenamente el nicho fundamental de una especie (Randin 2013; Lenoir y Svenning 2013). La proporción de nicho fundamental que la especie ocupa en la realidad se le denomina nicho real (realized niche), y es a lo que se referirá a lo largo de esta Tesis Doctoral como área de distribución de la especie.

Los cambios en el rango de distribución de una especie son el resultado neto de procesos de crecimiento, decaimiento y extinción de sus poblaciones a diferentes escalas temporales (Breshears et al. 2008). Sin embargo, la existencia de estos procesos varía notablemente de una localización a otra a lo largo de su área de distribución (Caughley et al. 1988; Thomas et al. 2001; Hampe y Petit 2005; Anderson et al. 2009), habiendo fuertes contrastes entre los procesos que ocurren entre el centro o núcleo de su distribución, con una mayor abundancia de la especie y una distribución continua de la misma, y su periferia, donde la especie es más escasa y suele encontrase fragmentada (sensu `centre-periphery hipothesis´ de Brown y Lomolino, 1998).

En la zona central de la distribución, también llamado óptimo de distribución, la

especie encuentra generalmente condiciones favorables para su desarrollo, lo que deriva en mayores tasas de regeneración y emigración que de mortalidad e inmigración (Pulliam 2000). De esta forma, y en ausencia de limitaciones bióticas o abióticas para su dispersión, se espera que en el centro de su distribución la especie pueda colonizar y establecer nuevas poblaciones en zonas disponibles (Lenoir y Svenning 2013). En cambio, en los márgenes o zona periférica de la distribución de una especie las poblaciones locales pueden encontrar condiciones límites y desfavorables para su desarrollo, lo que supondría mayores tasas de mortalidad e inmigración que de regeneración y emigración y un posible decaimiento o extinción local (Lawton 1993; Vucetich y Waite 2003). Sin embargo en un escenario de cambio climático como el actual, estos procesos en los márgenes de distribución descritos para un estado ambiental estable se pueden revertir (Lenoir y Svenning 2013).


13

Fig.1.5. Representación esquemática de las características poblacionales y procesos dominantes en los márgenes de la distribución de una especie bajo un escenario de cambio en las condiciones ambientales (Fuente: modificado de Hampe y Petit (2005)).

De esta forma, se espera que el incremento de la temperatura actual tenga

efectos directos y rápidos en la dinámica poblacional en los márgenes de distribución de las especies (Brown y Lomolino 1998; Hampe y Petit 2005). Las poblaciones de los márgenes de distribución antes limitadas por las bajas temperaturas, pueden tener en la actualidad un papel de frente de avance de distribución, con un crecimiento exponencial y mayores tasas de colonización (Fig. 1.5), mientras que los eventos de decaimiento y las tasas de mortalidad podrían verse incrementados en las poblaciones de los márgenes de distribución limitados por altas temperaturas y estrés hídrico (Hampe y Petit 2005). Así, análisis detallados de la dinámica demográfica (Hampe y Petit 2005), genética (Vitasse et al. 2010; Aitken et al. 2008; Sancho-Knapik et al. 2014) y de la estructura poblacional (Hampe y Petit 2005) de las especies en sus márgenes de distribución pueden ser clave para ayudar a identificar las tendencias a largo plazo en sus patrones de distribución.

En el marco del actual calentamiento global y dada la relación directa entre la

distribución de especies y gradientes ambientales como el clima (Hutchinson 1957), el estudio de cambios en la distribución de especies a lo largo de gradientes geográficos ha ganado una creciente atención. Con el objeto de estudiar las tendencias de cambio actual de forma estandarizada, Lenoir y Svenning (2015) han propuesto recientemente una nueva clasificación conceptual de los diferentes tipos de desplazamientos en el área de distribución de las especies a lo largo de un gradiente en un escenario de cambio climático. Basado en clasificaciones anteriores (Breshears et al. 2008; Maggini et al. 2011) y asumiendo conservación de nicho a lo largo del tiempo (Peterson et al. 1999), esta clasificación comprende cinco grandes categorías (Fig. 1.6): expansión (expand), donde dominan los eventos de colonización dentro y fuera del rango de distribución de la especie; avance (march), donde tanto procesos de extinción y


14

colonización ocurren dentro y fuera del rango de distribución de la especie; retracción (retract), donde procesos de extinción aparecen dentro del rango de distribución existente; constricción (lean), donde se producen eventos de colonización dentro del área de distribución de la especie; y colisión (crash), donde eventos de decaimiento aparecen a lo largo de toda las distribución de la especie.

Fig. 1.6. Representación conceptual de las cinco grandes categorías de cambio en la distribución de una especie a lo largo de un gradiente (latitud/longitud/altitud) en un escenario de cambio climático. La curva gris representa la distribución de la especie antes del cambio ambiental, mientras que la verde corresponde con la distribución después del cambio. Las líneas rojas, naranjas y violetas representan las tendencias (persistencia/desplazamientos) teóricas esperadas si el calentamiento continuara en el futuro (Fuente: modificado de Lenoir y Svenning (2015)).

En la actualidad existe un elevado número de evidencias de cambios en la distribución de las especies relacionado con el calentamiento global contemporáneo (véase la reciente revisión de Lenoir y Svenning 2015). Acorde con las proyecciones sobre alteraciones en la distribución de especies bajo los supuestos de cambio climático, están particularmente bien representados los cambios direccionales en la distribución de especies a lo largo de gradientes latitudinales y altitudinales hacia los polos o hacia mayores elevaciones como respuesta al aumento de temperatura (Kullman 2000; Colwell et al. 2008; Dolanc et al. 2013).

Sin embargo, en la última década la aparición de un elevado número de

evidencias contra-intuitivas que muestran descensos en la distribución altitudinal de algunas especies o inesperados `estancamientos´ de otras (Lenoir et al. 2010; Bertrand et al. 2011; Crimmins et al. 2011; Callaghan et al. 2015; Bodin et al. 2013; Hernández et al. 2014a) cuestionan la perspectiva unidireccional y univariable de los cambios esperados en la distribución de especies. Estas inesperadas respuestas de la vegetación se atribuyen principalmente a diversos factores bióticos y abióticos como la modificación humana del territorio (Lenoir et al. 2010; Callaghan et al. 2015),


15

interacciones bióticas, restricciones geográficas en la dispersión y a complejas condiciones ambientales regionales y locales (Lenoir y Svenning 2015). Sin embargo, a pesar de la importancia que tiene conocer los impulsores y los mecanismos subyacentes a los cambios recientes en distribución de especies (Thuiller et al. 2008), este es un fenómeno poco estudiado a día de hoy (Crimmins et al. 2011; Lenoir y Svenning 2015).

Por otro lado, existen grandes discrepancias en los esfuerzos en investigación

derivados al estudio de los cambios en distribución de especies leñosas entre ecoregiones y biomas lo que supone un importante sesgo en el conocimiento de este impacto a escala global (Lenoir y Svenning 2015). Debido a la esperada sensibilidad al cambio climático de las poblaciones vegetales en los márgenes de distribución, la mayor parte de trabajos que muestran evidencias de cambios en la distribución de las especies se enmarcan en los frentes de distribución `leading edge´ de la zona boreal del planeta (Kullman 2000; Kullman y Kjällgren 2006) o en el límite de distribución altitudinal en zonas montañosas del hemisferio norte (Gehrig-Fasel et al. 2007; Gottfried et al. 2012). Sin embargo, a pesar de la importancia para la conservación de la diversidad genética y filogenética de las poblaciones situadas en la retaguardia de su distribución (Hampe y Petit 2005), se ha centrado escasa atención en estas poblaciones así como en las situadas en los cinturones de vegetación de menor altitud (Jump et al. 2009; Lenoir y Svenning 2015).

1.1.2. Un nuevo reto ambiental: la invasión de los ecosistemas forestales por flora exótica invasora

Caja.1.2. Relación de términos científicos en castellano e inglés relacionados con el subcapítulo 1.1.2. de la presente tesis doctoral y su signifcado.

Las invasiones biológicas se definen como el proceso de introducción, establecimiento y propagación de especies exóticas (Williamson y Fitter 1996; Richardson et al. 2000; Balaguer 2004; Vilà et al. 2008). El último informe sobre el estado mundial de los bosques estima en 79 millones de hectáreas la superficie forestal invadida por especies exóticas, con una tendencia a aumentar (FAO 2015). Hoy en día las invasiones biológicas se consideran un problema a nivel mundial, no solo por sus repercusiones económicas (Pimentel et al. 2005), sino porque se


16

considera una de las principales amenazas para la conservación de la biodiversidad después de la pérdida de hábitat y la fragmentación (Vitousek et al. 1997; Mack et al. 2000). El Libro Rojo de la UICN de 2004 hace responsable a las especies exóticas invasoras de poner en peligro a un 5.4% de las especies con algún grado de amenaza a nivel mundial (Vilá et al. 2008).

Las especies invasoras pueden causar impactos a muchas escalas, desde el individuo hasta la estructura y funcionalidad del ecosistema receptor (Parker et al. 2009). La severidad de estos impactos dependerá de la interacción entre las características del hábitat receptor o su susceptibilidad a ser invadido (invasibility) y las características de la especie invasora o invasividad (invasiveness) (Alpert et al. 2000). Alguno de los ejemplos de estos impactos son la alteración de los ciclos de nutrientes y del agua (Castro-Díez et al. 2012; Ehrenfeld 2010; Glenn y Nagler 2005), de las características del suelo (May y Attiwill 2003; Marchante et al. 2003), de la composición de la comunidad (Levine et al. 2003; Lorenzo et al. 2010; González Muñoz et al. 2012) o de la interacción entre especies (Holmes y Cowling 1997; Morales y Traveset 2009). Por estos motivos, la expansión de especies exóticas invasoras es uno de los temas más urgentes a tratar en la agenda actual de los organismos para la conservación de la naturaleza a nivel mundial (CDB 2010, PNUMA 2010), siendo, por ejemplo, la identificación de vías de propagación, para evitar o controlar su introducción y establecimiento, uno de los principales objetivos de la Estrategia para la Diversidad Biológica de la Unión Europea 2020 (EC 2011).

No todas las especies exóticas son invasoras, ni todos los ambientes son

igualmente propensos a ser invadidos (Balaguer 2004). A pesar de que hasta hace poco se aceptaba que el proceso de invasión seguía la “regla de los 10” según la cual un promedio del 10% de la especies introducidas pasaban a ser subespontáneas, de éstas un 10% a naturalizadas, y de éstas un 10% a invasoras (Williamson y Fitter 1996), recientemente se ha demostrado que dicha regla no siempre se cumple y que el proceso de invasión de una especie (Fig. 1.7) depende de los filtros bióticos y abióticos que la especie exótica debe superar antes de convertirse en invasora (Richardson et al. 2000; Richardson y Pysek, 2006).

En este marco, en el proceso de invasión, pueden distinguirse cuatro etapas

fundamentales, cuyos filtros, asociados a la invasividad de la especie exótica y a la invasibilidad del hábitat receptor, determinarán el éxito o fracaso de la invasión: el transporte, la colonización, el establecimiento y la expansión (Fig.1.7; Sakai 2001; Teoharides y Dukes 2007).


17

Fig.1.7. Ilustración de las cuatro etapas del proceso de invasión y los factores o filtros que afectan el éxito en cada uno de ellos. Abajo se presentan las características de la especie invasora (invasividad) y del hábitat receptor (invasibilidad) jugando un papel importante en el éxito de la invasión (Fuente: modificado de Teoharides y Dukes (2007)).

Como ilustra la Fig. 1.7 el proceso de invasión comienza con el transporte, normalmente asociado a actividades humanas, de una especie desde su lugar de origen a un nuevo territorio (Vilá et al. 2008). Los filtros que determinan la supervivencia al transporte están relacionados con algunas de las características inherentes a la especie exótica como altas tasas de producción de semilla, resistencia a enfermedades, tolerancia a la sequía, durabilidad de semillas, etc. La colonización empieza cuando la especie llega al nuevo territorio (Theoharides y Dukes 2007) y el éxito de esta etapa depende tanto de la invasividad de la especie como de la susceptibilidad del hábitat a su invasión. La presión de propágulos de la especie es un elemento clave en esta etapa, ya que aumenta la diversidad genética y la probabilidad de dominar el banco de semillas y de encontrar un ambiente adecuado para su posterior establecimiento (Williamson y Fitter 1996; Lonsdale 1993). Características abióticas del hábitat como el clima, el suelo y la disponibilidad de recursos juegan pues un papel importante en la colonización. El establecimiento requiere que la especie invasora pueda crear poblaciones estables por sí mismas y aunque los factores abióticos siguen siendo importantes para su éxito, los filtros bióticos y las relaciones de competencia o facilitación que se establezcan con el resto de organismos del sistema son claves. Finalmente, la capacidad de dispersión de la especie y las características a escala de paisaje, como la fragmentación, la existencia de corredores y el régimen de perturbación, son los factores más relevantes para que la especie invasora se expanda en el nuevo territorio.


18

Como se observa en lo anteriormente expuesto, el proceso de invasión está muy relacionado con la invasividad de la especie invasora y la invasibilidad del hábitat receptor y en los últimos años la investigación de las invasiones biológicas ha hecho un gran esfuerzo en analizar las propiedades que hacen a algunas especies más propensas a ser invasoras, y las características que hacen a algunos hábitats más susceptibles a la invasión. Así, algunas propiedades relacionadas con la invasividad son alta producción de semillas y capacidad de reproducción vegetativa (Richardson y Pysek 2006; van Kleunen et al. 2010), una alta capacidad competitiva (Vilà y Weiner 2004), periodos juveniles cortos, y alta plasticidad fenotípica (Rejmánek y Richardson 1996; Davison et al. 2011). Además, el éxito de algunas especies invasoras se ha relacionado con su habilidad para producir alelopatías y aumentar su dominancia y competitividad en el nuevo nicho (sensu ´the novel weapon hypothesis` (Callaway y Aschehoug 2000)). Respecto a la invasibilidad, hay numerosas teorías que tratan de explicar qué hace que un hábitat sea más o menos susceptible a la invasión y están muy relacionadas tanto con el régimen de perturbación y la disponibilidad de recursos, como con la fragmentación/conectividad o heterogeneidad del territorio y la facilitación a la dispersión de propágulos (Vilà y Pujadas 2001; Hansen y Clevenger 2005). Así, entre otras, la teoría de nicho vacío (empty niche) (Levine y D´Antonio, 1999) sugiere que la ausencia de una forma de vida en un territorio puede favorecer el establecimiento de especies exóticas pertenecientes a ese tipo de forma de vida que será capaz de explotar un nicho vacante o vacío. La hipótesis de la perturbación (the disturbance hypothesis) (Mack et al. 2000) o la de la disponibilidad de recursos fluctuantes (Davis et al. 2000) proponen que las perturbaciones, a veces puntuales, pueden aumentar la cantidad de recursos disponibles y promover la sucesión biológica o re-ensamblaje de especies, dando las mismas oportunidades a especies exóticas y nativas. Otras hipótesis están relacionadas con la ventaja que tienen algunas especies exóticas sobre las nativas al no tener enemigos naturales en los nuevos territorios (Keane y Crawley 2002; Blossey y Notzold 1995).

Aunque el significativo número de trabajos publicados en la última década ha

mejorado nuestro conocimiento teórico sobre las invasiones biológicas en torno a tres grandes cuestiones, impactos, invasividad e invasibilidad, el conocimiento práctico sobre qué especies son potencialmente invasoras, y qué hábitat y condiciones ambientales son más susceptibles a la invasión es aún limitado (Pino et al. 2005). Además, en los últimos años se han identificado nuevos retos como la necesidad de generar nuevo conocimiento sobre la dinámica temporal de las invasiones biológicas (Richardson et al. 2010). Esta línea de investigación se ve obstaculizada por la escasa disponibilidad de información espaciotemporal detallada sobre especies invasoras a grandes escalas (Pyšek et al. 2002). Aunque el acceso a listados de flora invasora y a mapas ha mejorado en los últimos tiempos (ver DAISIE Project; Foxcroft et al. 2010), la escasez de inventarios sistemáticos y periódicos, que permitan la posibilidad de explorar tasas demográficas, patrones espacio-temporales y sus factores asociados, limita la existencia de conocimientos clave para llevar a cabo medidas de prevención y control (ver sin embargo Hernández et al. 2014b).


19

1.2. Los bosques ibéricos y el cambio global

La Península Ibérica, como otras regiones del Mediterráneo, constituyó un importante refugio de flora durante las glaciaciones del Cuaternario (Hewitt 1999). Este hecho, sumado a factores geográficos como una amplia heterogeneidad topográfica y climática (en la Península Ibérica interactúan las regiones biogeográficas Atlántica, Alpina y Mediterránea), hacen de la península un punto caliente de biodiversidad (Cowling et al.1996; Medail y Quezel 1997), con un alto número de endemismos (Galán 1998; Sainz-Ollero y Moreno 2002) y donde muchas especies encuentran su límite de distribución a nivel continental, normalmente asociadas a los diferentes sistemas montañosos ibéricos (Costa et al. 1997).

Estos factores biogeográficos, junto con la alta presión histórica del hombre en el paisaje (García-Ruiz et al. 1996; Valbuena-Carabaña et al. 2010) y la sensibilidad de los ecosistemas forestales a los contrastes en las condiciones ambientales, hacen que la Península Ibérica, dentro de la Cuenca Mediterránea, haya sido identificada como una de las áreas que más intensamente pueden verse afectadas por el cambio global (MEA 2005; Bakkenes et al. 2006) y ha sido propuesta como una de las zonas prioritarias para el estudio de los efectos del cambio climático sobre sus sistemas naturales (Lavorel et al. 1998; Engler et al. 2011).

Los registros climáticos del último siglo muestran un claro aumento de olas de calor y recurrencia de sequías, así como un aumento de las temperaturas medias anuales en Europa (Fig.1.8, EEA 2015; IPCC 2013) y más concretamente en la cuenca Mediterránea (Bücher y Dessens 1991; Alpert et al. 2008). Así, la temperatura media en Europa se ha incrementado de media en 1,3ºC en comparación con la época pre-industrial (1850-1899) (Brohan et al. 2006) y el ritmo de cambio se ha acelerado en la tres últimas décadas alcanzando incrementos de 0,18ºC a 0,25ºC por década (IPCC 2013). Por el contrario, los registros de precipitación media anual no muestran tendencias claras de cambio, aunque empiezan a verse patrones relacionados con un descenso de la precipitación estacional, sobre todo de invierno, en el sur de Europa (Berg et al. 2013).

Diferentes modelos climáticos para Europa confirman la tendencia actual de calentamiento climático para las próximas décadas para todos los escenarios de emisiones (IPCC 2013). Estos modelos proyectan un incremento en las temperaturas, sobre todo en verano en el sur de Europa, y en el invierno, en el norte, con incrementos medios de entre 2.5ºC y 4ºC de media para finales del siglo XXI, pero con tendencias menos claras para la precipitación (Kjellström et al. 2011).


20

Fig.1.8. Tendencia al alza en las anomalías de la temperatura media anual (ºC) entre 1850 y 2014 y teniendo como referencia la era pre-industrial en Europa (Fuente: modificado de EEA (2015)).

Basándose en estos futuros escenarios de cambio climático, diversos modelos bioclimáticos pronostican una contracción general en el área potencial de distribución de muchas especies forestales Ibéricas para un futuro próximo (ver ejemplo en Fig. 1.9; Benito-Garzón et al. 2008; Felicísimo et al. 2011; Ruiz-Labourdette et al. 2012). Estos estudios prevén una rápida contracción en la distribución de especies que encuentran en la península su límite meridional de distribución, siendo más crítica ésta en las especies de coníferas de montaña, ya que las especies planocaducifolias parecen tener una mayor probabilidad de desplazamiento a zonas de mayor altitud. Estas proyecciones también indican una disminución notable en la superficie de especies con área de distribución restringida, como son las especies Submediterráneas (Sánchez de Dios et al. 2009). En contraste, se prevé un menor impacto en el área de distribución de especies típicamente Mediterráneas que incluso podrían aumentar su dominancia en los paisajes ibéricos (Benito-Garzón et al. 2008; Ruiz-Labourdette et al. 2013).

Estos potenciales efectos a largo plazo en la distribución de las especies forestales Ibéricas se ven en ocasiones corroborados por resultados que muestran el efecto negativo del aumento de la temperatura y la sequía sobre el crecimiento (Jump et al. 2006; Linares y Camarero 2012; Gea et al. 2013, 2014), las tasas demográficas (Carnicer et al. 2011; Vilá-Cabrera et al. 2011, 2012; Benavides et al. 2015), la interacción entre especies (del Río et al. 2014; Ameztegui et al. 2015) o la productividad (Vayreda et al. 2012; Gea y Cañellas 2014) de especies forestales ibéricas generalmente limitadas por la disponibilidad hídrica.


21

Fig. 1.9. Predicción de área potencial de Pinus sylvestris, como ejemplo de conífera de montaña, en la actualidad (arriba) y para 2020 (abajo) utilizando los escenarios A1 (izquierda) y A2 (derecha) del modelo CSIRO (azul) y el escenario A2 del modelo HadCM3 (verde). (Fuente: modificado de Benito-Garzón (2008)).

Sin embargo, en los últimos años numerosos trabajos muestran la importancia que tienen, además del clima, otros factores, como variaciones en el uso del territorio (Chauchard et al. 2007; Ameztegui et al. 2010), la gestión histórica de muchas especies forestales (Camarero et al. 2011) o la plasticidad fenotípica de las especies (Benito Garzón et al. 2011), en la respuesta de los bosques ibéricos al cambio global. Estas nuevas evidencias hacen reconsiderar la certidumbre en las proyecciones de cambio en distribución de especies basados sólo en cambio climático. Pero a pesar de la importancia de conocer el alcance, las tendencias y factores del impacto del cambio global sobre la distribución de los bosques ibéricos, hasta la fecha han sido pocos los esfuerzos derivados a estudiarlos (ver sin embargo San-Elorza et al. 2003; Peñuelas y Boada 2003; Urli et al. 2014; Hernández et al. 2014a; Hernández et al. en rev.).

En el contexto de las invasiones biológicas, el grado de invasión en la Península Ibérica es muy heterogéneo. Se estima la existencia de unos 123 taxones de plantas con carácter invasor, lo que supone aproximadamente un 12% de la flora autóctona a nivel peninsular (Sanz Elorza et al. 2004). Este porcentaje varía entre regiones, suponiendo desde un 6-8% de la flora del sureste peninsular al 18% en el noroeste (Romero-Buján et al. 2007). Además como ilustra la Fig.1.10 la mayor riqueza de flora exótica, relacionada con áreas metropolitanas y la líneas de costa, se distribuye por el Mediterráneo oriental (Vilá et al. 2008) debido al efecto facilitador de una temperatura suave y la transformación antrópica del territorio. Sin embargo, todavía es escaso el


22

conocimiento de qué hábitats son en mayor o menor medida invadidos (Pino et al. 2005) y en qué grado.

Fig.1.10. Riqueza de plantas invasoras en cuadrículas de 10 x 10 km en la España Peninsular (Fuente: Sanz Elorza et al. 2004).

Por otro lado, como ilustra la Fig 1.11. la mayor parte de especies de flora invasora presentes hoy en día en la Península no ha alcanzado ni la mitad de su área potencial de distribución en España (Gassó et al. 2012) y es esperable que muchos táxones, favorecidos por la dispersión humana, aumente su grado de invasión en un futuro próximo.

Fig.1.11. Distribución de la frecuencia de la ocupación relativa (proporción de la actual área potencial ocupada) de especies invasoras en España (Fuente: modificado de Gassó et al. (2012)).


23

1.3. El Inventario Forestal Nacional como instrumento para responder a los nuevos retos en ecología y gestión forestal Hoy en día es indiscutible el valor que constituyen las fuentes de datos de los Inventarios Forestales Nacionales (IFN) (Tomppo et al. 2010) por su sólido diseño estadístico de muestreo y la representatividad objetiva que presentan sobre los bosques a grandes escalas espaciales y temporales. Tanto es así, que en la última década se ha incrementado exponencialmente la creación de nuevos IFN a nivel mundial, y la FAO (2015) estima que desde 1970 a 2014 un 82% de la superficie forestal (aproximadamente 3,3 billones de hectáreas) queda representada por los diferentes IFNs. Estas fuentes de datos forestales a gran escala fueron desarrolladas desde mediados del siglo XX en un gran número de países (Tomppo et al. 2010). Al principio su primordial objetivo era la estimación de los recursos madereros de los bosques, pero con el paso del tiempo, y a medida que la visión sobre el papel de los bosques para la humanidad cambió, se hicieron necesarias nuevas tomas de datos adicionales relacionados con el papel de los bosques como reservorios de carbono y biodiversidad (Chirici et al. 2011). En los años sesenta del siglo pasado se proyecta en España el primer ciclo de inventario en toda la superficie forestal arbolada y siendo la unidad de muestreo la provincia. El IFN1 se llevó a cabo entre 1965 y 1975, y aunque el ciclo de inventario se había fijado en 10 años, diferentes acontecimientos históricos hicieron que no se pudiera realizar el segundo ciclo (IFN2) hasta 20 años después, en el intervalo de tiempo entre 1986 y 1996 (Vallejo y Villanueva 2002). El tercer ciclo de inventario se llevó a cabo entre 1997 y 2008, y actualmente se está realizando el cuarto, del que contamos ya en 2015 con datos de diez comunidades autónomas. Los avances en la ciencia estadística y forestal, así como en los sistemas de información geográfica supusieron incorporar actualizaciones y mejoras en cada nuevo ciclo respecto al anterior. El IFN1 sirvió para conocer el estado de partida de nuestros montes (Vallejo y Villanueva 2002), pero es a partir del IFN2 donde se diseña un inventario continuo que permitiera analizar la evolución de nuestras masas forestales mediante la comparación con los posteriores ciclos (Hernández et al. 2014c). Así, a partir del IFN2 las parcelas de muestreo de campo se sitúan en las zonas arboladas sobre los vértices de la malla UTM de 1 km, remidiendo, en la medida de lo posible, las parcelas levantadas en los anteriores inventarios. En cada punto de la malla de muestreo, los árboles son medidos en parcelas circulares concéntricas con radio variable según el diámetro (Bravo et al. 2002). Además, desde el IFN3 la toma de datos clásica del inventario se complementa con una adicional cuyo principal objetivo es la estimación de nuevas variables relacionadas con el estado de la biodiversidad de nuestros bosques, que se centra en la composición y la estructura principalmente (Alberdi et al. 2014). La representación sistemática y detallada de la superficie forestal española recogida en las bases de datos del IFN, hace de éste una herramienta muy útil para conocer el estado general de los bosques, aportando la información base para diferentes informes estadísticos nacionales (IEPNB 2013) e internacionales (FAO 2001; 2015) y para la toma de decisiones en política y gestión forestal (Fig.1.12). Además, propiedades inherentes a los IFN le hacen muy útiles para avanzar en la


24

ciencia forestal (Fig.1.12). Por ejemplo, la habitual georeferenciación de la parcelas del IFN hace posible la asociación de los datos recogidos en las parcelas con información geográfica. Esta disponibilidad de datos a escalas espaciales amplias asociados a gradientes abióticos y bióticos es fundamental a la hora de parametrizar modelos estadísticos que analicen variaciones en procesos ecológicos (Sagarin y Pauchard 2009). Además, la posibilidad de tener variables ecológicas remedidas en diferentes ciclos de IFN continuos a lo largo de dilatados espacios de tiempo constituye una herramienta muy útil para analizar la dinámica de los ecosistemas forestales (Lund et al. 1998), máxime para especies longevas con lentas tasas de cambio como son los árboles.

Fig. 1.12. La utilidad de la información de los IFNs es muy amplia, siendo importante su contribución al conocimiento del estado de los bosques a nivel mundial (FAO 2001), a las estadísticas nacionales, o, al avance en la ciencia forestal aplicada y teórica. Así, en base a las fuentes de datos del IFN es posible llevar a cabo trabajos de investigación a diferentes escalas espaciales y temporales y bajo diferentes escenarios de cambio, desde socioeconómicos a climáticos. En la última década las fuentes de datos del IFN español han sido profusamente utilizadas para avanzar en nuestro conocimiento aplicado (Torras y Saura 2008; Herrero y Bravo 2008; Fernández-Alonso et al. 2013) y teórico (Benito-Garzón et al. 2011; García-Valdés et al. 2013; Ruiz-Benito et al. 2013) en ecología forestal. A partir de la información de un ciclo de IFN se pueden hacer análisis diferenciales de diferentes variables ecológicas (estructura poblacional, densidad, dominancia, productividad, estado del regenerado, captura de carbono acumulado, biodiversidad) entre poblaciones, comunidades y ecosistemas (ver ejemplos en Gil Tena et al. 2009; Martín-Queller et al. 2011; Hernández et al. 2014b, 2014c). Mientras que mediante la comparación de ciclos de inventario se pueden hacer inferencias sobre dinámica poblacional (cambios en composición y estructura), procesos demográficos (tasas/ratios de crecimiento,


25

mortalidad) (Gómez-Aparicio et al. 2011; Carnicer et al. 2011; Hernández et al. 2014b; Hernández et al. en rev.; Benavides et al. 2015), así como estimar cambios en la distribución espacial potencial (Thuiller et al. 2003; Benito-Garzón et al. 2008) o reciente (Urli et al. 2014; Hernández et al. 2014a; Hernández et al. en rev.) de especies forestales o de variables de masas a lo largo de gradientes ambientales.

1.4. Referencias bibliográficas

Aitken SN, Yeaman S, Holliday Ja, Wang T, Curtis-Mclane S (2008) Adaptation, migration or extirpation: climate change outcomes for tree populations. Evol. Appl. 1: 95–111.

Alberdi I, Cañellas I, Condés S (2014) A long-scale biodiversity monitoring methodology for Spanish National Forest Inventory. Application to Álava region. Forest Syst. 23(1): 93-110

Alpert P, Bone E, Holzapfel C (2000) Invasiveness, invasibility and the role of environmental stress in the spread of non-native plants. Perspective Plant Ecol. Evol. Systematics. 3: 52–66.

Alpert P, Krichak S, Shafir H, Haim D, Osentinsky I (2008) Climatic trends to extremes employing regional modelling and statistical interpretation over the E. Mediterranean. Glob. Planet Change. 63: 163-170.

Ameztegui, A, Brotons L, Coll L (2010) Land-use changes as major drivers of mountain pine (Pinus uncinata Ram.) expansion in the Pyrenees. Glob. Ecol. Biogeog.19: 632-641.

Ameztegui A, Coll L, Messier C (2015) Modeling the effect of climate-induced changes in recruitment and juvenile growth on mixed-forest dynamics: The case of montane-subalpine Pyrenean ecotones. Ecol. Model. 313: 84-93. Anderson BJ, Akçakaya HR, Araújo MB et al. (2009) Dynamics of range margins for metapopulations under climate change. P. Roy. Soc. Lond. B. Bio., 276(1661): 1415-1420.

Araújo MB, Pearson RG, Thuiller W, Erhard M (2005) Validation of species–climate impact models under climate change. Glob. Change Biol. 11(9): 1504-1513.

Bakkenes B, Eickhout B, Alkemade JRM (2006) Impacts of different climate stabilisation scenarios on plant species in Europe. Global. Environmental. Politics. 45:123-145.

Balaguer L (2004) Las plantas invasoras, ¿el reflejo de una sociedad crispada o una amenaza científicamente contrastada? Historia Natural. 5:32-41.

Benavides R, Escudero A, Coll L et al. (2015) Survival vs. growth trade-off in early recruitment challenges global warming impacts on Mediterranean mountain trees. Perspect. Plant. Ecol. En prensa.


26

Benito-Garzón M, Sánchez de Dios R, Sainz -Ollero H (2008) Effects of climate change on the distribution of Iberian forests. Appl. Veg. Sci.11: 169–178.

Benito-Garzón M, Alía R, Robson TM, Zavala MA (2011) Intra-specific variability and plasticity influence potential tree species distributions under climate change. Global Ecol. Biogeogr. 20:766–778.

Berg P, Moseley C, Haerter JO (2013) Strong increase in convective precipitation in response to higher temperatures. Nat. Geosci. 6(3): 181-185. Bertrand R, Lenoir J, Piedallu C et al. (2011). Changes in plant community composition lag behind climate warming in lowland forests. Nature, 479(7374), 517-520.

Blossey B, Notzold R (1995) Evolution of increased competitive ability in invasive non indigenous plants: a hypothesis. J. Ecol. 83: 887-889.

Bodin J, Badeau V, Bruno E, Cluzeau C, Moisselin JM, Walther GR, Dupouey JL (2013) Shifts of forest species along an elevational gradient in Southeast France: climate change or stand maturation? J. Veg. Sci. 24: 269–283.

BOM (2006) Bureau of Meteorology, Commonwealth of Australia. http://www.bom.gov.au/info/climate/change/gallery/

Bravo F, Río M, Pando V, San Martin R, Montero G, Ordoñez C, Cañellas I (2002) El diseño de las parcelas del Inventario Forestal Nacional y la estimación de variables dasométricas. En: El Inventario Forestal Nacional como elemento clave para la gestión forestal sostenible (Bravo et al. eds.). Fundación Gral. de la Universidad de Valladolid. pp: 19-35.

Breshears DD, Huxman TE, Adams HD, Zou CB, Davison JE (2008) Vegetation synchronously leans upslope as climate warms. P. Natl. Acad. Sci. Biol.105: 11591–11592.

Brohan P, Kennedy JJ, Harris I, Tett S, Jones PD (2006) Uncertainty estimates in regional and global observed temperature changes: a new data set from 1850. J. Geophy. Res.111: 12106. Brown JH and Lomolino MV (1998) Biogeography. Sunderland: Sinauer Associates. Bücher A, Dessens J (1991) Secular trend of surface temperature at an elevated observatory in the Pyrenees. J. Climate. 4: 859-868.

Callaghan TV, Jonasson C, Thierfelder T, Yang Z, Hedenås H, Johansson M, Sloan V L (2013) Ecosystem change and stability over multiple decades in the Swedish subarctic: complex processes and multiple drivers. Philos. T. Roy. Soc. B. 368(1624): 20120488.

Callaway RM, Aschehoug ET (2000) Invasive plants versus their new and old neighbours: a mechanism for exotic invasion. Science 290: 521-523.


27

Camarero JJ; Bigler C; Linares JC; Gil-Pelegrín E (2011) Synergistic effects of past historical logging and drought on the decline of Pyrenean silver fir forests. Forest Ecol. Manag. , 262:759-769. Carnicer J, Coll M, Ninyerola M, Pons X, Sánchez G, Peñuelas J (2011) Widespread crown condition decline, food web disruption, and amplified tree mortality with increased climate change-type drought. P. Natl. A. Sci. India. B.108: 1474-1478.

Castro-Díez P, Fierro-Brunnenmeister N, González-Muñoz N, Gallardo A (2012) Effects of exotic and native tree leaf litter on soil properties of two contrasting sites in the Iberian Peninsula. Plant. Soil. 350: 179-191. Caughley G, Grice D, Barker R, Brown B (1988) The edge of the range. J. Anim. Ecol. 57, 771–785.

Chapin III FS, Sala OE, Huber-Sannwald E (2001) Potential biodiversity change: global patterns and biome comparisons. In: Global Biodiversity in a Changing Environment: Scenarios for the 21st Century (Eds: Chapin III FS, Sala OE, Huber-Sannwald E), pp351–367. Springer-Verlag. New York, NY, USA.

Chauchard S, Carcaillet C, Guibal F (2007). Patterns of Land-use Abandonment Control Tree-recruitment and Forest Dynamics in Mediterranean Mountains. Ecosystems. 10 (6): 936-948.

Chiras D (1998) Environmental Science: A Systems Approach to Sustainable Development. Waldsworth Publishing Company, Belmont, California, USA.

Chirici G, Winter S, McRoberts RE (Eds.) (2011) National Forest Inventories: Contributions to Forest Biodiversity Assessments: Contributions to Forest Biodiversity Assessments (Vol. 20). Springer Science & Business Media.

Colwell RK et al. (2008) Global warming, elevational range shifts, and lowland biotic attrition in the wet tropics. Science. 322: 258 – 261.

Costa M, Morla C, Sainz H (Eds) (1997) Los bosques ibéricos. Una interpretación geobotánica. Planeta, Barcelona. pp 572.

Cowling PW, Rundel BB, Lamont M, Arroyo MK (1996) Plant diversity in Mediterranean-climate regions. Trends Ecol. Evol., 11:362–366. Crimmins SM, Dobrowski SZ, Greenberg JA, Abatzoglou JT, Mynsberge AR (2011) Changes in climatic water balance drive downhill shifts in plant species’ optimum elevations. Science 331: 324 – 327.

DAISIE Project. Delivering Alien Invasive Species Inventories for Europe. http://www.europe-aliens.org/

Davidson AM, Jennions M, Nicotra AB (2011) Do invasive species show higher phenotypic plasticity than native species and, if so, is it adaptative? A meta-analysis. Ecol. Lett. 14: 419-431.


28

Davis MA, Grime JP, Thompson K (2000) Fluctuating resources in plant communities: a general theory of invasibility. J. Ecol. 88: 528-534. Davis MB, Shaw RG (2001) Range shifts and adaptive responses to Quaternary climate change. Science 292: 673-679.

del Río M, Condés S, Pretzsch H (2014). Analyzing size-symmetric vs. size-asymmetric and intra-vs. inter-specific competition in beech (Fagus sylvatica L.) mixed stands. Forest Ecol. Manag. 325: 90-98.

Dolanc CR et al. (2013) Widespread shifts in the demographic structure of subalpine forests in the Sierra Nevada, California, 1934 to 2007. Global Ecol. Biogeogr. 22: 264 – 276.

Duarte CM et al. (2006). Cambio global. Impacto de la actividad humana sobre el sistema Tierra. CSIC.

EC (2011) Communication from the Commission to the European Parliament, the Council, the Economic and Social Committee and the Committee of the Regions – Our Life Insurance, Our Natural Capital: An EU Biodiversity Strategy to 2020. COM (2011) 244 pp.

EEA (2015) Impacts of Europe's changing climate: an indicator-based assessment, EEA Report. Available: http://www.eea.europa.eu/data-and-maps/indicators/global-and-european-temperature-1/assessment Ehrenfeld JG (2010) Ecosystem consequences of biological invasions. Ann. Rev. Ecol. Evol. S. 41: 59-80.

Engler R, Randin CF, Thuiller J et al. (2011) 21st century climate change threatens mountain flora unequally across Europe. Glob Change Biol 17: 2330-2341.

FAO (2001) Global forest resources assessment. State of the World’s Forests 2001. Food and Agriculture Organization of the United Nations. Forestry Paper. Rome.200 pp. FAO (2015) Food and Agriculture Organization of the United Nations. Global forest resources assessment 2015: Main report. How are the world’s forests changing? www.fao.org/forestry/fra Felicísimo ÁM, Muñoz J, Villalba CJ, Mateo RG (2011) Impactos, vulnerabilidad y adaptación al cambio climático de la biodiversidad española. Flora y vegetación. Ministerio de Medio Ambiente y Medio Rural y Marino, Madrid.

Fernández-Alonso JM, Alberdi I, Álvarez-González JG, Vega JA., Cañellas I, Ruiz-González AD (2013). Canopy fuel characteristics in relation to crown fire potential in pine stands: analysis, modelling and classification. Eur. J. For. Res. , 132(2), 363-377.

http://www.eea.europa.eu/data-and-maps/indicators/global-and-european-temperature-1/assessment

http://www.eea.europa.eu/data-and-maps/indicators/global-and-european-temperature-1/assessment

http://www.fao.org/forestry/fra


29

Foxcroft LC, Richardson DM, Rejmánek M, Pyšek P (2010) Alien plant invasions in tropical and sub-tropical savannas: patterns, processes and prospects. Biol Invasions. 12(12): 3913-3933.

Galán P, Gamarra R, García JI (1998) Árboles y arbustos de la península Ibérica e Islas Baleares. Jaguar, Madrid.

Garcia-Ruiz JM, Lasanta T, Ruiz-Flano P, Ortigosa L, White S, González C, Martí C (1996) Land-use changes and sustainable development in mountain areas: a case study in the Spanish Pyrenees. Landscape Ecol. 11(5): 267-277.

García-Valdés R, Zavala MA, Araújo MB, Purves DW (2013) Chasing a moving target: projecting climate change-induced changes in non-equilibrial tree species distributions. J. Ecol. 329: 206-213.

Gassó N, Thuiller W, Pino J, Vilà M (2012) Potential distribution range of invasive plant species in Spain. NeoBiota, 12: 25-40.

Gea-Izquierdo G, Fernández de Uña L, Cañellas I (2013) Growth projections reveal local vulnerability of Mediterranean oaks with rising temperatures. Forest Ecol. Manag. 305: 282-293. Gea-Izquierdo G, Cañellas I (2014) Local climate forces instability in long-term productivity of a Mediterranean oak along climatic gradients. Ecosystems 17: 228-241. Gea-Izquierdo G, Viguera B, Cabrera M, Cañellas I (2014) Drought induced decline could portend widespread pine mortality at the xeric ecotone in managed Mediterranean pine-oak woodlands. Forest Ecol. Manag. 320, 70-82. Gehrig-Fasel J, Guisan A, Zimmermann NE (2007) Tree line shifts in the Swiss Alps: Climate change or land abandonment? J.Veg.Sci. 18(4), 571-582.

Gil-Tena A, Brotons L, Saura S (2009) Mediterranean forest dynamics and forest bird distribution changes in the late 20th century. Glob Change Biol. 15: 474-485.

Glenn EW, Nagler PL (2005) Comparative ecophysiology of Tamarix ramosissima and native trees in western US riparian zones. J. Arid. Enviro. 61: 419- 446. Gómez-Aparicio L, García-Valdés R, Ruiz-Benito P, Zavala MA (2011) Disentangling the relative importance of climate, size and competition on tree growth in Iberian forests: implications for management under global change. Glob Change Biol., 17: 2400-2414.

González-Muñoz N, Costa-Tenorio M, Espigares T (2012) Invasion of alien Acacia dealbata on Spanish Quercus robur forests: Impact on soils and vegetation. Forest Ecol Manage. 269: 214-221.

Gottfried M, Pauli H, Futschik A et al. (2012) Continent-wide response of mountain vegetation to climate change. Nature. 2(2): 111-115.


30

Guisan A, Theurillat JP, Kienast F (1998) Predicting the potential distribution of plant species in an alpine environment. J.Veg.Sci. 9:65–74.

Hampe A, Petit RJ (2005) Conserving biodiversity under climate change: the rear edge matters. Ecol. Lett., 8(5): 461-467.

Hanewinkel M, Peyron JL (2014). Tackling climate change—the contribution of scientific knowledge in forestry. Ann. For. Sci. 71(2): 113-115.

Hansen M, Clevenger AP (2005) The influence of disturbances and habitat on the frequency of nonnative plant species along transportation corridors. Biol. Cons. 125: 249−259.

Hernández L, Cañellas I, Alberdi I, Torres I, Montes F (2014a). Assessing changes in species distribution from sequential large-scale forest inventories. Ann. For. Sci. 71:161-171.

Hernández L, Martínez-Fernández J, Cañellas I, de la Cueva AV (2014b) Assessing spatio-temporal rates, patterns and determinants of biological invasions in forest ecosystems. The case of Acacia species in NW Spain. Forest Ecol. Manag. 329: 206-213.

Hernández L, Sandoval VJ, Vallejo R. et al. (2014c) The forest biodiversity of Andalusia: estimation based on the Spanish NFI data analysis. Ed: INIA. MINECO. ISSN: 1575-6106.

Hernández L, Sánchez de Dios R, Montes F, Sainz-Ollero H, Cañellas I. Exploring range shifts of contrasting tree species along bioclimatic and elevational gradients. Forest Ecol. Manag. En rev.

Herrero C, Bravo F (2008) Can we get an operational indicator of forest carbon sequestration?: A case study from two forest regions in Spain. Ecol. Indic. 17: 120-126.

Hewitt MC (1999) Post-glacial re-colonization of European biota. Biol. J. Linn. Soc. 68:87–112. Holmes PM, Cowling RM (1997) The effects of invasion by Acacia saligna on the guild structure and regeneration capabilities of South African fynbos shrublands. J. App. Ecol. 34: 317-332.

Houghton J, Ding Y, Griggs DJ et al. (2001). IPCC 2001: Climate Change 2001. The Climate change Contribution of Working Group I to the Third Assessment Report of the Intergovemmental Panel on Climate Change, 159.

Hutchinson GE (1957) Population studies – animal ecology and demography – concluding remarks. Cold Spring Harbor Symposia on Quantitative Biology, 22:415–427.

IEPNB (2013). Inventario Español del Patrimonio Natural y de la Biodiversidad. MAGRAMA.


31

IPCC (2013) “Resumen para responsables de políticas. En: Cambio Climático 2013: Bases físicas. Contribución del Grupo de trabajo I al Quinto Informe de Evaluación del Grupo Intergubernamental de Expertos sobre el Cambio Climático” [Stocker, T. F., D. Qin, G.-K. Plattner, M. Tignor, S. K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex y P.M. Midgley (eds.)]. Cambridge University Press, Cambridge, Reino Unido y Nueva York, NY, Estados Unidos de América.

Jump AS, Peñuelas J (2005) Running to stand still: adaptation and the response of plants to rapid climate change. Ecol. Lett., 8: 1010-1020.

Jump AS, Hunt JM, Peñuelas J (2006) Rapid climate change-related growth decline at the southern range edge of Fagus sylvatica. Glob Change Biol. 12: 2163-2174.

Jump AS, Mátyás CA, Peñuelas J (2009) The altitude-for-latitude disparity in the range retractions of woody species. Trends Ecol.Evol. 24:694-701.

Kappelle M, Van Vuuren MMI, Baas P (1999) Effects of climate change on biodiver-sity: a review and identification of key research issues. Biodivers. Conserv. 8: 1383–1397.

Keane RM, Crawley MJ (2002) Exotic plant invasions and the enemy release hypothesis. Trends Ecol. Evol S.17: 164-170. Keenan RJ, Reams GA, Achard F, de Freitas JV, Grainger A, Lindquist E (2015). Dynamics of global forest area: results from the FAO Global Forest Resources Assessment 2015. Forest Ecol. Manag. , 352, 9-20. Kjellström E, Nikulin G, Hansson U, Strandberg H, Ullerstig A (2011) 21st century changes in the European climate: uncertainties derived from an ensemble of regional climate model simulations. Tellus, 63A(1): 24-40. Kullman L (2000) Tree-limit rise and recent warming: a geoecological case study from the Swedish Scandes. Norsk Geografisk Tidsskrift, 54(2): 49-59.

Kullman L, Kjällgren L (2006) Holocene pine tree‐line evolution in the Swedish Scandes: Recent tree‐line rise and climate change in a long‐term perspective. Boreas, 35(1): 159-168.

Lavorel S, Canadell J, Rambal S, Terradas J (1998) Mediterranean terrestrial ecosystems: research priorities on global change effects. Global Ecol. Biogeogr. 7(3): 157-166. Lawton JH (1993) Range, population abundance and conservation. Trends Ecol. Evol. S. 8: 409–413. Lenoir J, Gégout JC, Pierrat JC, et al. (2009) Differences between tree species seedling and adult altitudinal distribution in mountain forests during the recent warm period (1986–2006). Ecography. 32: 765–777.


32

Lenoir J, Gégout JC, Guisan A, et al. (2010) Going against the flow: potential mechanisms for unexpected downslope range shifts in a warming climate. Ecography. 33(2): 295-303. Lenoir J, Svenning JC (2013) Latitudinal and elevational range shifts under contemporary climate change. In: Encyclopedia of Biodiversity (Ed. Levin SA), pp. 599–611.

Lenoir J, Svenning JC (2015) Climate‐related range shifts–a global multidimensional synthesis and new research directions. Ecography. 38: 15–28.

Levine JM, D´Antonio CM (1999) Elton revisited: a review of evidence linking diversity and invasibility. Oikos 87: 15-26.

Levine JM, Vilà M, D'Antonio CM, Dukes JS, Grigulis K, Lavorel S (2003) Mechanisms underlying the impacts of exotic plant invasions. Proc R Soc Lond B Biol Sci. 270: 775-781. Linares JC, Camarero JJ (2012) Growth patterns and sensitivity to climate predict silver fir decline in the Spanish Pyrenees. Eur. J. For. Res. 131: 1001-1012.

Lindner M, Maroschek M, Netherer S et al. (2010) Climate change impacts, adaptive capacity, and vulnerability of European forest ecosystems. Forest Ecol. Manag. 259: 698-709.

Lonsdale WM (1993) Rates of spread of an invading species – Mimosa nigra in northern Australia. J Ecol. 81: 513–521.

Lorenzo P, González L, Reigosa MJ (2010) The genus Acacia as invader: the characteristic case of Acacia dealbata Link in Europe. Ann For Sci. 67(1): 101.

Lund HG (1998) A comparison of multipurpose resource inventories (MRIs) throughout the world. Working PaperNo. 14. Joensuu, Finland: European Forest Institute. 46 pp.

Mack RN, Simberloff D, Mark Lonsdale W, Evans H, Clout M, Bazzaz FA (2000) Biotic invasions: causes, epidemiology, global consequences, and control. Ecol Appl. 10(3): 689-710.

Maggini R, Lehmann A, Kery M, et al. (2011) Are Swiss birds tracking climate change? Detecting elevational shifts using response curve shapes. Ecological Modelling 222: 21–32.

Malcolm JR, Markham A, Neilson RP, Garaci M (2002) Estimated migration rates under scenarios of global climate change. J.Biogeogr . 29: 835–849.

Marchante H, Marchante E, Freitas H (2003) Invasion of the Portuguese dune ecosystems by the exotic species Acacia longifolia (Andrews) Willd: effects at the community level. In Child, L.E., Brock, J.H., Brundu, G., Prach, K., Pyšek, P., Wade, P.M. & Williamson, M. (Eds.), Plant Invasions: Ecological Threats and Management Solutions. Backhuys Publishers. The Netherlands. pp. 75-85.


33

Martín-Queller E, Gil-Tena A, Saura S (2011) Species richness of woody plants in the landscapes of Central Spain: the role of management disturbances, environment and non-stationarity. J.Veg.Sci. . 22: 238-250

May BM, Attiwill PM (2003) Nitrogen-fixation by Acacia dealbata and changes in soil properties 5 years after mechanical disturbance or slash -burning following timber harvest. Forest Ecol. Manag. 181:339-355.

MEA (2005) Millennium Ecosystem Assessment. Ecosystem and human well-being: biodiversity synthesis. World Resources Institute, Washington D.C.

Medail P, Quezel P (1997) Hot-Spots analysis for conservation of plant biodiversity in the Mediterranean Basin. Ann. Mo. Bot. Gard. 84:112–127. Morales CL, Traveset A (2009) A meta-analysis of impacts of alien vs. native plants on pollinator visitation and reproductive success of co-flowering native plants. Ecol. Lett. 12: 1-13. Morueta-Holme N, Engemann K, Sandoval-Acuña P, Jonas JD, Segnitz RM, Svenning JC (2015) Strong upslope shifts in Chimborazo's vegetation over two centuries since Humboldt. Proc Natl Acad Sci. 09938. Parmesan C, Yohe G (2003) A globally coherent fingerprint of climate change impacts across natural systems. Nature, 421:37–42.

Peñuelas J, Boada M (2003) A global change-induced biome shift in the Montseny mountains (NE Spain). Glob Change Biol 9: 131-140.

Peterson AT, Soberón J, Sánchez-Cordero V (1999) Conservatism of ecological niches in evolutionary time. Science, 285: 1265–1267.

Pimentel D, Zúniga R, Morrison D (2005) Update on the environmental and the economic costs associated with alien-invasive species in the United States. Ecol. Econom. 52: 273-288.

Pino J, Font X, Carbó J, Jové M, Pallarès L (2005) Large-scale correlates of alien plant invasion in Catalonia (NE of Spain). Biol. Cons. 122: 339-350.

Pulliam HR (2000) On the relationship between niche and distribution. Ecol. Lett. 3: 349–361.

Pyšek P, Jarošik V, Kuèera T (2002) Patterns of invasion in a temperate nature reserves. Biol. Cons. 104: 13-24.

Randin CF, Paulsen J, Vitasse Y, Kollas C, Wohlgemuth T, Zimmermann NE, Körner C (2013) Do the elevational limits of deciduous tree species match their thermal latitudinal limits? Global Ecol. Biogeogr. , 22(8): 913-923.

Rejmánek M, Richardson DM (1996) What attributes make some plant species more invasive? Ecology. 77: 1655–1661.


34

Richardson DM, Allsopp N, D’Antonio CM. Milton SJ, Rejmánek M (2000) Plant invasions: the role of mutualisms. Biol. Rev. 7: 65-93.

Richardson D, Pysek P (2006) Plant invasions: merging the concepts of species invasiveness and community invasivility. Prog. Phys. Geog. 30(3): 409-431.

Richardson DM, Daehler CC, Leishman MR, Pauchard A, Pyšek P (2010) Plant invasions: theoretical and practical challenges. Biol Invasions 12: 3907-3911.

Romero Buján MI R (2007). Flora exótica de Galicia (noroeste ibérico). Botanica Complutensis. 31: 113-125.

Rubiales JM, García-Amorena I, García Álvarez S, Gómez Manzaneque F (2008) The Late Holocene extinction of Pinus sylvestris in the western Cantabrian Range (Spain). J.Biogeogr. 35: 1840-1850.

Ruiz‐Benito P, Gómez‐Aparicio L, Paquette A, Messier C, Kattge J, Zavala MA (2013) Diversity increases carbon storage and tree productivity in Spanish forests. Global Ecol. Biogeogr.44:112-123.

Ruiz-Labourdette D, Nogués Bravo D, Ollero HS, Schmitz MF, Pineda FD (2012) Forest composition in Mediterranean mountains is projected to shift along the entire elevational gradient under climate change. J. Biogeog. 39: 162-176.

Ruiz-Labourdette D, Schmitz MF, Pineda FD (2013) Changes in tree species composition in Mediterranean Mountains under climate change: indicators for conservation planning. Ecol.Indi.24: 310-323

Sagarin R, Pauchard A (2009) Observational approaches in ecology open new around in a changing world. Frontiers in Ecology and the Environment. 8: 379-386.

Sakai AK (2001) The population biology of invasive species. Annu. Rev. Ecol. Syst. 32: 305–332.

Sánchez de Dios R, Benito-Garzón M, Sainz-Ollero H (2009) Present and future extension of the Iberian Submediterranean territories as determined from the distribution of marcescent oaks. Plant Ecology. 204: 189-205.

Sancho Knapik D, Peguero Pina JJ, Cremer E et al. (2014) Genetic and environmental characterization of Abies alba Mill. populations at its western rear edge. Pirineos. 169: e007.

Sainz Ollero H, Moreno Saiz J (2002) Flora vascular endémica española, cap.14. En: La diversidad biológica de España (Pineda F et al. Eds). Pearson Educación, Madrid, pp. 175-195.

Sanz-Elorza M, Dana ED, González A, Sobrino E (2003) Changes in the high-mountain vegetation of the Central Iberian Peninsula as a probable sign of global warming. Ann. Bot. London. 92: 273–280.


35

Sanz-Elorza M, Dana ED, Sobrino E (2004) Atlas de las plantas alóctonas invasoras de España. Madrid, D.G. de la Biodiversidad, Ministerio de Medio Ambiente, 384 pp.

Teoharides KA, Dukes JS (2007) Plant invasion across space and time: factors affecting non-indigenous species success during four stages of invasion. New Phytol.176: 256-273.

Thomas CD, Cameron A, Green RE, et al. (2004) Extinction risk from climate change. Nature. 427: 145–8.

Thuiller W, Vayreda J, Pino J, Sabate S, Lavorel S, Gracia C (2003) Large-scale environmental correlates of forest tree distribution in Catalonia (NE Spain). Global Ecol. Biogeogr 12: 313–325.

Thuiller, W., Albert, C., Araújo, et al. (2008). Predicting global change impacts on plant species’ distributions: future challenges. Perspect. Plant. Ecol. 9: 137–152.

Tomppo E, Gschwantner T, Lawrence M et al. (2010) National forest inventories. Pathways for Common Reporting. European Science Foundation, 541-553. Springer Science & Business Media.

Torras O, Saura S (2008) Effects of silvicultural treatments on forest biodiversity indicators in the Mediterranean. Forest Ecol. Manag. 255: 3322-3330

UNEP (2010) Report of the Seventh Meeting of the Conference of the Parties to the Convention on Biological Diversity (UNEP/CBD/COP/35/4). Decision XI/28, UNEP.

Urli M, Delzon S, Eyermann A, Couallier V, García-Valdés R, Zavala MA, Porté AJ (2014) Inferring shifts in tree species distribution using asymmetric distribution curves: a case study in the Iberian mountains. J.Veg.Sci. 25(1): 147-159.

Valbuena-Carabaña M, de Heredia UL, Fuentes-Utrilla P, González-Doncel I, Gil, L (2010) Historical and recent changes in the Spanish forests: a socio-economic process. Rev. Palaeobo. Palyno. 162(3): 492-506.

Vallejo R, Villanueva JA (2002) El banco de datos de la naturaleza. En: El Inventario Forestal Nacional como elemento clave para la gestión forestal sostenible (Bravo et al. eds.). Fundación Gral. de la Universidad de Valladolid, pp: 19-35.

van Kleunen M, Weber E, Fischer M (2010) A metaanalysis of trait differences between invasive and noninvasive plant species. Ecol. Lett. 13: 235-245.

Vayreda J, Martinez-Vilalta J, Gracia M, Retana J (2012) Recent climate changes interact with stand structure and management to determine changes in tree carbon stocks in Spanish forests. Glob. Change Biol. 18: 1028-1041. Vilà M, Pujadas J (2001) Land-use and socio-economic correlates of plant invasions in European and North African countries. Biol. Cons. 100: 397- 401. Vilà M, Weiner J (2004) Are invasive plant species better competitors tan native plant species? Evidence from pair-wise experiments. Oikos 105: 229-238.


36

Vilá M, Valladares F, Traveset A, Santamaría L, Castro-Díez P (Eds.) (2008) Invasiones Biológicas. CSIC, Madrid. Vilà-Cabrera A, Martínez-Vilalta J, Galiano L, Retana J (2012) Patterns of forest decline and regeneration across scots pine populations. Ecosystems 13: 323-335.

Vilá-Cabrera A, Martínez-Vilalta J, Vayreda J, Retana J (2011) Structural and climatic determinants of demographic rates of Scots pine forests across the Iberian Peninsula. Ecol. Appl 21: 1162–1172.

Vitasse Y, Bresson CC, Kremer A, Michalet R and Delzon S (2010) Quantifying phenological plasticity to temperature in two temperate tree species. Func. Ecol. 24:1211- 1218. Vitousek PM, Mooney HA, Lubchenco J, Melillo J (1997) Human domination of Earth´s ecosystems. Science. 277:494-499.

Von Humboldt A, Bonpland A (1807) Ideenzu einer Geographie der Pflanzen nebst einem Naturgemailde der Tropenlander. Cotta. Tübingen.

Vucetich JA, Waite TA (2003) Spatial patterns of demography and genetic processes across the species range: null hypotheses for landscape conservation genetics. Conserv. Gen. 4: 639–645.

Walther GR, Post E, Convey P, et al. (2002) Ecological responses to recent climate change. Nature. 416: 389-95.

Williamson M, Fitter A (1996) The varying success of invaders. Ecology. 77(6): 1661-1666.

2. Objetivos y estructura de la Tesis

CAPÍTULO2: OBJETIVOS Y ESTRUCTURA DE LA TESIS

39

2. Objetivos y Estructura de la Tesis

2.1. Objetivos

Tras contextualizar el potencial impacto del cambio global en los bosques, esta Tesis Doctoral trata de avanzar en el conocimiento de dos grandes líneas de investigación en esta materia: el estudio del cambio en la distribución de especies a lo largo de gradientes ambientales; y la evolución de las invasiones biológicas en ecosistemas forestales. Mediante la comparación de la información recogida en las bases del Inventario Forestal Nacional Español (IFN), que abarca un periodo de hasta cuatro décadas, y usando técnicas apropiadas para su análisis, por un lado esta tesis trata de arrojar algunas de las primeras evidencias de cambio en la distribución de especies en zonas de especial sensibilidad, intentando testar así hipótesis biogeográficas previas bajo escenarios de cambio climático. Por otro lado, esta tesis propone por primera vez el uso de la información detallada y a gran escala de los IFN para analizar en detalle la evolución de las invasiones biológicas en ecosistemas forestales ibéricos.

De forma específica, en el capítulo 3 esta tesis propone la comparación de la serie completa de ciclos del IFN español con el objetivo de analizar cambios en la distribución espacial de especies a lo largo de gradientes ambientales mediante la aplicación de técnicas geoestadísticas. Para ello se presenta una nueva aproximación metodológica usando un tipo de regresión espacial denominado Universal Kriging (UK) que considera la autocorrelación espacial y su relación con diferentes variables ambientales para predecir la probabilidad de presencia de especies. Mediante la aplicación práctica de esta metodología en algunos casos o mediante el análisis de tasas demográficas e interacciones biológicas, los capítulos 3, 4 y 5 analizan los cambios observados en el área de distribución y dinámica de algunas de las especies de mayor sensibilidad al cambio global: especies que encuentran su límite meridional de distribución en la Península Ibérica, o especies que interactúan en zonas de transición bioclimática. Además se analizan la coincidencia o no de los cambios observados con las proyecciones de modelos bioclimáticos y se proponen nuevos marcos conceptuales para explicar las posibles discrepancias encontradas. Por último, en el capítulo 6 y en el marco de estudio de las invasiones biológicas, este último trabajo propone el uso de los diferentes ciclos del IFN para avanzar en el conocimiento de la dinámica de invasión de especies del género Acacia en el noroeste Peninsular. Además este capítulo ahondará en el conocimiento práctico y teórico de los factores que determinan su expansión y de la susceptibilidad de diferentes ecosistemas forestales a ser invadidos.

CAPÍTULO2: OBJETIVOS Y ESTRUCTURA DE LA TESIS

40

2.2. Estructura de la Tesis

La Tesis Doctoral está estructurada en ocho capítulos y una última sección de Anexos donde se han adjuntado al documento dos de los trabajos ya publicados en revistas incluidas en la base datos Science Citation Index (SCI) resultado de la Tesis. Tras un primer capítulo introductorio en español donde se contextualiza la tesis, en el segundo capítulo se exponen los objetivos generales y específicos y la estructura del documento. A continuación se presentan cuatro capítulos (capítulos 3-6) en inglés con sus secciones de resumen, introducción, material y métodos, resultados, discusión y referencias bibliográficas. Estos capítulos reproducen el contenido de artículos publicados (capítulos 3, 6) y en revisión (capítulos 4 y 5). Por último, en el capítulo 7 se presenta una discusión general de los resultados de la tesis, y en el capítulo 8 las conclusiones generales. La Tabla 2.1. presenta un resumen del contexto científico de cada capítulo, así como de sus objetivos, área de estudio y resultados científicos.

Tabla 2.1. Resumen de capítulos, resultados en forma de publicación, objetivos, área de estudio, periodo y especies focales de los capítulos 3-6 de la presente tesis doctoral.

Capítulo

y Resultados ______________

Objetivos generales

______________________

Área de estudio y periodo

Especies focales

_______________

Cap

ítulo

3

Her

nánd

ez e

t al.

(201

4).

Anna

ls o

f For

est

Scie

nces

.

Nueva aproximación metodológica para analizar cambios en distribución de especies a partir de series históricas de IFN. Aplicación del método a especies que encuentran su límite de distribución meridional en la Península Ibérica.

Pirineo occidental 1970-2010

Pinus sylvestris L. Fagus sylvatica L.

Cap

ítulo

4

Her

nánd

ez e

t al.

(En

2ª re

visi

ón e

n Fo

rest

Eco

logy

and

M

anag

emen

t). Análisis de cambio en distribución

espacial de especies a lo largo de gradientes bioclimático y altitudinal. Estudio de los factores asociados a los cambios observados.

Navarra Zona de transición Submediterránea

Gradiente altitudinal 1970-2010

Quercus subpyrenaica E.H del Villar Quercus ilex subsp. ballota (Desf.) Samp. Pinus sylvestris L. Fagus sylvatica L.

Cap

ítulo

5

Her

nánd

ez,e

t al.

(En

revi

sión

en

Pers

pect

ives

in

Plan

t Eco

logy

, Ev

olut

ion

and

Syst

emat

ics)

Análisis de dinámica, tasas demográficas, sucesión vegetal y factores bióticos y abióticos subyacentes.

Región biogeográfica

Atlántica Ibérica 1986-2012

Fagus sylvatica L. Quercus petraea Matt. Liebl.

Cap

ítulo

6

Her

nánd

ez e

t al.

(201

4). F

ores

t Ec

olog

y an

d M

anag

emen

t. Análisis de tasas demográficas, procesos espacio-temporales y determinantes de la expansión de dos especies exóticas de carácter invasor.

Galicia 1998-2008

Acacia dealbata Link Acacia melanoxylon R. Br. in W.T. Aiton

3. Assessing changes in species distribution from sequential large scale

forest inventories

Este capítulo reproduce íntegramente el texto del siguiente manuscrito:

Hernández L, Cañellas I, Alberdi I, Torres I, Montes F (2014) Assesing changes in species distribution from sequential large scale forest inventories. Annals of Forest

Science. DOI:10.1007/s13595-013-0308-6.

CAPÍTULO3: ASSESSING CHANGES ON FOREST FROM NFI

43

Resumen

Se asume que el cambio global está afectando en la actualidad la composición, estructura y distribución de los ecosistemas forestales, sin embargo, aún son escasas las evidencias precisas de cambios latitudinales y altitudinales en su distribución. El objetivo de este trabajo es presentar un método basado en las bases de datos de los Inventarios Forestales Nacionales (IFN) para estimar cambios espacio-temporales en la distribución de especies. Para ello hemos desarrollado una nueva aproximación basada en Universal Kriging que permite comparar modelos de distribución de especies a partir de los secuenciales ciclos de inventario. La aproximación permite construir test de significación que tienen en cuenta las diferencias en la intensidad de muestreo entre ciclos. Esta metodología ha sido aplicada en un periodo de 40 años en el Pirineo Occidental donde algunas poblaciones de especies como Pinus sylvestris L. y Fagus sylvatica L. alcanzan su límite meridional de distribución, y donde son más vulnerables a sufrir los efectos del cambio global. Los resultados arrojan evidencias de un incremento en la presencia de las dos especies en la región en el periodo de estudio. La distribución del P. sylvestris se ha desplazado 1,5 km hacia el norte, mientras que F. sylvatica ha expandido su distribución 2 km hacia el sur. Además el óptimo de la distribución altitudinal de las dos especies ha ascendido en aproximadamente 200 m. Como resultados de estos cambios la zona donde las dos especies coexisten se ha extendido. Este trabajo proporciona una herramienta útil para comparar ciclos de IFN independientemente de los métodos de muestreo utilizados, así como para cuantificar y testar cambios en distribución espacial de las especies a lo largo de gradientes geográficos.

Palabras clave: Inventario Forestal Nacional, Universal Kriging, cambio en distribución de especies, Pinus sylvestris, Fagus sylvatica, Pirineos.


44

Abstract

It is assumed that global change is already affecting the composition, structure and distribution of forest ecosystems, however, detailed evidences of altitudinal and latitudinal shifts are still scarce. This study aims to propose a method based on NFI to assess spatio-temporal changes in species distributions. We develop an approach based on Universal Kriging to compare species distribution models from the different NFI cycles and regardless of the differences in the sampling schemes used. Furthermore, a confidence interval approach is used to assess significant changes in species distribution. The approach is applied to some of the southernmost populations of Pinus sylvestris and Fagus sylvatica in the Western Pyrenees over the last 40 years. An increase of the presence of the two species in the region was observed. P. sylvestris distribution has shifted about 1.5 km northwards over recent decades, whereas F. sylvatica has extended its distribution southwards by about 2 km. Furthermore, the optimum altitude for both species has risen by about 200 m. As a result, the zone in which the two species coexist has been enlarged. This approach provides a useful tool to compare NFI data from different sampling schemes, quantifying and testing significant shifts in tree species distribution over recent decades across geographical gradients.

Keywords: National Forest Inventory, Universal Kriging, shifts, Pinus sylvestris, Fagus sylvatica, Pyrenees.


45

3.1. Introduction

The impact of global change on forests is emerging as a major concern for 21st century society (FAO 2000). Forestry and conservation policy-makers need to understand how the distributions of species are affected by global change in order to tackle its effects on forests. Although it is assumed that global change is already affecting the composition, structure and distribution of forest ecosystems at different spatial and temporal scales, reported evidences of altitudinal and latitudinal shifts are still scarce (Peñuelas and Boada 2003).

Large-scale forest surveys such as National Forest Inventories (NFI) can provide a valuable tool for monitoring changes in forest biodiversity and can be of great relevance in the conservation and management of natural resources. In the second half of the 20th century most developed countries started undertaking periodical NFIs covering the entire forest area. Although NFIs were primarily designed to estimate forest resources, they are increasingly being employed to assess the impact of global change on forest ecosystems (Thuiller et al. 2003). When permanent plots are inventoried sequentially, forest evolution can be analysed through direct comparison between inventories (Poljanec et al. 2010; Vilá-Cabrera et al. 2011), otherwise species-environment relationships should be analysed from species distribution models (SDMs) (Guisan and Zimmermann 2000). One problem that arises when comparing species distributions is whether or not the changes are related to the different sampling schemes employed across successive NFIs. This shortcoming can be dealt with in a similar way to change-of-support problems by using block kriging techniques (Yoo and Trgovac 2011).

Environmental variables such as precipitation, temperature and elevation exhibit spatial dependence, which is partly responsible for the spatial pattern observed in vegetation distribution (Miller et al. 2007). In addition, spatial autocorrelation can also result from ecological processes involved in forest dynamics occurring at different scales, leading to spatial continuity in species distribution (Bellehumeur and Legendre 1998). As a consequence, spatial dependence in biogeographical data has recently been identified as an important area of SDMs research, especially for species with a broad distribution, which are generally better modelled by including spatial autocorrelation in the model (Segurado and Araújo 2004). Universal kriging may be considered a type of spatial regression, incorporating spatial autocorrelation as well as relationships between environmental variables in the spatial prediction of species distribution (Montes and Ledo 2010). It may be important to consider this latter quality of universal kriging in SDMs as species distribution varies over environmental gradients. The altitudinal range where species with a broad latitudinal distribution appear varies as a result of a decline in temperature with increasing elevation and more northerly latitude. This variation in modelled relationships over space, known as non-stationarity, shows interdependence with spatial scale, so local and landscape approaches may reveal species-environment relationships that global models average out (Osborne et al. 2007; Miller and Hanham 2011).


46

The Iberian Peninsula, like other Mediterranean regions, constituted an important refuge for flora during the Quaternary glaciations. This fact, together with the current interaction of the Eurosiberian, Mediterranean and Alpine biogeographical zones, made the Iberian Peninsula a biodiversity hotspot where many species find the trailing edge of their distribution, many of them highly fragmented in different mountain ranges. This is the case of Scots pine (Pinus sylvestris L.) and European beech (Fagus sylvatica L.), two species widely distributed across Europe and which have some of their southernmost populations in the mountain ranges of the Iberian Peninsula. The most recent climatic records for the Mediterranean region reveal an increase in the number of heat waves and droughts during the last century (Alpert et al. 2008). This pattern, together with the important impact of human activity and the sensitivity of the ecosystems to contrasts in climatic conditions, make the mountain ecosystems of the Iberian Peninsula particularly vulnerable to the effects of global change (Engler et al. 2011). It is expected that woody species will respond to changes in environmental conditions by migration towards suitable areas or through adaptation to the new conditions. Otherwise, they may decline or even become locally extinct (Aitken et al. 2008). Taking into account not only these premises but also the results of previous studies in European mountain forests (Lenoir et al. 2008, Poljanec et al. 2010) we hypothesize an upward shift in the distribution of Scots pine and European beech at regional scale over recent decades in the Western Pyrenees.

The aim of this study is to develop a new method based on universal kriging models to assess spatio-temporal changes in species distribution from large-scale sequential forest inventories regardless of the differences in the sampling schemes used. For testing the usefulness of this approach we also aim to answer the following questions: i) Is it possible to detect significant shifts in the spatial distribution of Pinus sylvestris and Fagus sylvativa in the Western Pyrenees over recent decades? ii) Can we quantify the extent of these shifts across geographical gradients?

3.2. Material and methods

3.2.1. Study area

This study focuses on the Western Pyrenees in the Navarra region of Spain between E.D. 50 UTM (Universal Transverse Mercator) 4.734 S-4.764 N and 642 E-685 W. The study area, which covers 741.60 km2, is characterized by a steep altitudinal gradient, which ranges from approximately 600 m a.s.l. at the southern limit to 2,400 m a.s.l at the northern limit. Due to its geographical location, where several biogeographical regions overlap, the region has a variable climate. At more northerly latitudes, where the highest altitudes are found and the oceanic influence of the Cantabrian Sea is less notable, the mean annual precipitation is 1,063 mm, and the mean annual temperature is about 8 ºC, which reflects the typical continental characteristics of the Pyrenees. By contrast, in the mid latitudes, mean annual precipitation levels are higher (1,300 mm), as are the temperatures (10.5ºC). The southernmost area, however, is influenced by a sub-Mediterranean climate with moderate summer drought. Hence, the mean annual rainfall in this part of the study area is lower (1,044 mm) although the mean annual temperature is higher (11 ºC) (Ninyerola et al. 2005).


47

Scots pine and beech form the naturally occurring uneven-aged forests that dominate the study area. In the Iberian Peninsula, the majority of Scots pine stands can be found in the more continental, south-central Pyrenees; becoming more scarce in the northwest (Costa et al. 1997), i.e., in the areas influenced by the Cantabrian Sea, where beech stands are more common. Beech and Scots pine occur in the montane altitudinal belt between 600 and 1,600 m a.s.l., with beech occupying higher and shadier slopes. In the lower montane altitudinal belt, Pinus sylvestris can be found alongside marcescent Quercus species like Q. faginea Lam., Q. pyrenaica Willd. and Q. pubescens Willd., whereas above these forests, the sub-alpine belt (1,600-2,300 m a.s.l.) is dominated by Pinus uncinata Ram. and the alpine belt (over 2,300 m a.s.l.) is the domain of herbaceous, sometimes shrubby vegetation. Both species can co-occur in some zones, where European beech is found in more mesic conditions and Scots pine where more xeric conditions prevail.

3.2.2. NFI dataset

The Spanish NFI is a forest monitoring system covering the entire forested area of Spain. Inventories are undertaken approximately every 10 years with an intensity of one sampling point every 1 km2. At each sampling point, trees are measured in concentric circular plots with increasing radii from 5 to 25 m. In the 5-m-radius plot, trees with dbh (diameter at breast height) > 75 mm and saplings with 25 ≤ dbh < 75 mm are measured, while the recruitment density is monitored (individuals of less than 1.30 m height or dbh < 25 mm). In the 10-m-radius plot, trees with dbh > 125 mm are measured; in the 15-m-radius plot, trees with dbh > 225 mm are measured; and in the 25-m-radius plot, trees with dbh > 425 mm are measured. In the Navarra region, four cycles of the Spanish NFI have been completed. Although the NFI2 (1986-1995) and NFI3 (1997-2006) cycles have been used in numerous studies (Vilá-Cabrera et al. 2011), research involving the comparison of data from the four NFIs conducted since 1960 has never been undertaken. Since different sampling designs were used in the various NFI cycles, the number of plots and locations varied from one to another (Table 3.1).

Table 3.1. Spanish National Forest Inventory (NFI) cycles sampling periods, source of information used for forest area assessment, sampling period and number of plots measured in the study area in each NFI cycle.

Cycle NFI

sampling period

Forest area information source Study area sampling period

Study area total number

of plots

NFI1 1965-1974 Aerial photographs from 1969 1971 203

NFI2 1986-1996 Map of land use 1:50.000 1989-1990 241

NFI3 1997-2007 Spanish forestry map 1:50.000 1999-2000 437

NFI4 2008-2017 Spanish forestry map 1:25.000 2009-2010 497


48

In order to analyze changes in the spatial distribution of species using the NFI databases, presence/absence indicator variables, with a value of 1 if a species was present in a plot and 0 if it was absent were defined for both species. The presence of a species in a given plot was registered according to size: recruitment, saplings and trees with dbh ≥ 75 mm. Therefore, the presence/absence indicator variable is a regionalized variable that (for the xy location of the plot centre) will be 1 if there are individuals within the respective radius and 0 if not. The variable-radius sampling design of the NFI plots smoothes the sampling estimator surface and minimizes the variance of the estimator (Williams 2001).

3.2.3. Universal kriging

Universal kriging (UK) is a spatial regression procedure that incorporates the spatial autocorrelation in the estimation of a regionalized variable. In the universal kriging model, the value of the variable Z(so) at location so is expressed as a

polynomial of the auxiliary variables (i.e., mean function), which account

for a spatial trend, and a spatially autocorrelated residual process (s0) (Matheron 1969):

[1]

Universal kriging was used to model the relationships between species presence/absence indicator variables from the NFI dataset and the latitude (y, E.D. 50 UTM, in km), elevation (h, m a.s.l.) and exposure (cos(φ), the cosine of the aspect azimuth) derived from a 25 m resolution DEM of Spain, for each inventory. In the case of the Scots pine presence/absence indicator variable, exploratory analyses indicated differences in the altitude at which presence prediction was maximum across the latitudinal gradient; therefore, a polynomial of order 2 including h2, h and the h×y interaction was used to model the latitudinal variation in altitudinal maximum in the mean function. The model performance was notably improved by incorporating the cosine of the aspect azimuth and the cos(φ)×y interaction to model the latitudinal variation of this variable. Therefore, the mean function for the Scots pine presence/absence indicator variable was the following:

[2]

In the case of the beech presence/absence indicator variable, preliminary analyses did not show any interaction between altitude and latitude. However, the altitudinal distribution clearly showed a maximum presence probability. In addition, a latitudinal trend was noted for cos(φ); thus, the mean function finally chosen was the following:

[3]

p

k

kk sf0

0 )(

)()()( 00

00 ssfsZp

k

kk

yhyhhsfp

k

kk )cos()cos()( 5432

2100

0

yyhhsfp

k

kk )cos()cos()( 5432

2100

0


49

The universal kriging prediction p(Z,s0) of the regionalized variable Z(s0) can be interpreted as the probability of the presence of each species analyzed in s0 (Goovaerts 1994) and is calculated as:

[4]

under the unbiasedness conditions:

[5]

where Z(si) is the value of the variable at n sampling points. The variance of the universal kriging prediction (σUK(s0)) was calculated as in Cressie (1993).

The variogram, that describes the degree of spatial dependence of the regionalized variable, was modelled using the spherical variogram defined by the following parameters: the nugget (the semivariance value at the origin), the autocorrelation range (the distance at which the semivariance stabilizes) and the sill (the semivariance for distances greater than the autocorrelation range). One of the crucial steps when using universal kriging is the variogram parameters and β coefficients fitting. The VLS method used for the variogram and mean function estimation seems more suitable than Maximum Likelihood methods for the presence/absence indicator variable because it does not require multi-gaussian distributional assumption (Montes and Ledo 2010).

The kriging prediction may give values outside the range [0,1], although these deviations are usually of small magnitude (Goovaerts 1994). There are different techniques used for constraining the indicator kriging prediction to [0,1] (Tolosana-Delgado et al. 2008). For the comparison between inventories pursued in this study such constrains are not necessary, however, to map the occurrence of the species (Fig. 3.1), the kriging indicator predictions outside [0,1] were set to the nearest bound (upward-downward correction –Deustch and Journel 1992).

3.2.4. Cross-validation of the universal kriging model

A leaving-one-out cross-validation was used to determine the prediction bias and accuracy of the prediction variance estimates of the fitted model (Cressie, 1993). Differences between the predicted and observed values (SEE, sum of estimation errors) were used to assess the bias.

To test the accuracy of the prediction variance estimation, the ratio of the variance of the residuals to the prediction variance was estimated through the fitted variogram (σUK(s0)).VSEE (variance of the standardized estimation errors) was calculated. Values close to 1 indicated a similar distribution for the observation and prediction errors.

in

i

i sZsZ 1

0,p

itf

pktftf

i

n

i

kiki

1,

...0),(),(

0

10

i

0i

s

ss


50

3.2.5. Testing the significance of changes in species distribution between inventories

To test the statistical significance of the changes in species distribution along the gradients of the analysed variables between inventories, a novel significance test based on block kriging variances (Isaaks and Srivastava 1989) was developed. The block kriging estimates the mean of the variable for a determined region (or block) through averaging the universal kriging weights λi for the points resulting from the discretizacion of the block. The block mean and the block prediction variance were estimated for the strata determined by the classification of the forest area i) in 10 altitude classes, ii) in 7 latitude classes and iii) in 9 exposure classes, which constitute the blocks in this case. The block mean of the presence/absence indicator variable can be interpreted as the mean probability of species presence in each stratum.

As kriging provides unbiased estimators, the block mean does not depend on the number of plots in each inventory, although the kriging variance depends on the spatial autocorrelation and the distance between samples, which varies with the plot density. To assess differences between inventories at 95% confidence level, confidence intervals for the presence/absence block prediction at each altitude, latitude and exposure class were calculated from the kriging variance for each inventory date (Isaaks and Srivastava 1989):

0000 96.1,,96.1, ssZps UKUKsZp

To determine the significance of the differences between the indicator variable kriging predictions of the different NFI cycles, these were compared with the size of the confidence intervals. The use of confidence intervals allows the differences to be presented as a range of magnitudes, rather than just assess the statistical significance (Katz 1992). The absence of overlapping between the confidence intervals at two different inventory cycles gives a conservative test at 95% confidence for the difference between the predicted values. Note that the presence/absence prediction mean is estimated for each class, so bias due to differences in forest area between inventories is avoided.

3.2.6. Mapping changes in species distribution

To graphically assess the species distribution at each inventory date, a threshold probability indicating the species presence was set at 0.5 (Montes et al. 2005), checking that the proportion of predicted points above the indicator threshold was unbiased. The indicator variables of both species were inferred in a 500 m x 500 m grid covering the forest area at each inventory cycle to build the prediction maps.

3.2.7. Software used

ArcGis 9.2 was used to interpolate 25-m-resolution digital elevation models (DEM) from the topographical map of Spain and to derive latitude, altitude and cosine of the aspect azimuth at sample and prediction points. Geostatistical analyses were performed using a Microsoft® VisualBasic® application developed by the authors.


51

3.3. Results

3.3.1. Variogram models

The variogram models fitted for the Scots pine and beech differed considerably. The nugget effect was smaller for the Scots pine (Table 3.2), whereas the spatial autocorrelation range was larger for beech (Table 3.3). The autocorrelation range in the Scots pine model progressively diminished from approximately 3 km in 1971 to 0.86 km in 2010, which indicates a decrease in the spatial continuity of the Scots pine indicator variable in the study area. In contrast, the progressively larger autocorrelation ranges of beech (from 9 km in 1971 to 22 km in 2010) indicated increasing continuity in the spatial distribution. The analyses of cross-validation residuals showed that the universal kriging models fitted performed satisfactorily in terms of bias and kriging variance estimation in both cases (Tables 3.2 and 3.3).

3.3.2. Mapping forest range changes

The most notable changes in the universal kriging prediction maps (Fig. 3.1) were the expansion of the Scots pine from 71% to 79% of the plots from 1971 to 2010 and an increase from 53% to 72% in the case of beech. In addition, there were changes in the distribution areas of Scots pine and beech. For instance, in the NFI1, the two species predominantly occupied distinct areas (coexisting in 31% of the study area), whereas in the NFI3 and NFI4 the area in which the two species coexisted had increased to approximately 42% of the study area.

Fig. 3.1 Universal kriging predictions for the distribution areas of Pinus sylvestris (black), Fagus sylvatica (dark grey) and areas where both species coexist (light grey) from NFI1 (1971), NFI2 (1989-1990), NFI3 (1999-2000) and NFI4 (2009-2010).


52

Table 3.2 Universal kriging models, VLS estimations of variogram parameters and β coefficients for Pinus sylvestris and explanatory variables; elevation, square of the elevation, exposure (Expo) and the interaction between latitude, elevation and exposure. Cross-validation sum of residuals (SEE) and the residual variance/prediction variance ratio (VSEE) are also shown.

Table 3.3 Universal kriging model, VLS estimations of variogram parameters and β coefficients for Fagus sylvatica and explanatory variables; elevation, the square of the elevation, latitude, exposure (Expo) and the latitudinal interaction with the exposure. The cross-validation sum of residuals (SEE) and the residual variance/prediction variance ratio (VSEE) are also shown.

Pinus sylvestris Variogram parameters β coefficients of the auxiliary variables Cross-validation Inventory Nugget Sill Range (km) β 0 Elevation Elevation 2 Elevation*Latitude Expo Expo*Latitude SEE VSEE NFI1 (1971) 0 0.177 3.042 0.565 0.003 - 5.9 E-07 -5.5 E-07 31.319 -0.007 -0.0090 1.10 NFI2 (1989-1990) 0 0.161 1.134 0.063 0.134 - 5.9 E-07 -2.792 E-05 45.307 -0.010 -0.00021 1.01 NFI3 (1999-2000) 0.042 0.102 0.895 -0.499 0.138 -1.08 E-06 -2.852 E-05 39.783 -0.008 -0.00047 1.01 NFI4 (2009-2010) 0.039 0.121 0.868 -0.240 0.125 -8.3 E-07 -2.602 E-05 43.433 -0.009 -0.00056 1.01

Fagus sylvatica Variogram parameters β coefficients of the auxiliary variables Cross-validation Inventory Nugget Sill Range (km) β 0 Elevation Elevation 2 Latitude Expo Expo*Latitude SEE VSEE NFI1 (1971) 0.081 0.098 14.569 -132.070 0.005 - 2.1 E-06 0.027 47.276 -0.010 0.0041 1.21 NFI2 (1989-1990) 0.043 0.146 9.205 -154.671 0.002 - 1 E-06 0.032 30.136 -0.006 0.0036 1.70 NFI3 (1999-2000) 0.076 0.118 15.825 -110.065 0.004 -1.5 E-06 0.023 44.392 -0.009 0.00002 1.36 NFI4 (2009-2010) 0.108 0.093 22.952 214.585 0.007 -2.9 E-06 -0.046 36.204 -0.008 0.0001 0.99


53

3.3.3. Assessing species distribution changes between inventories

Differences for a range of magnitude between species distribution across the NFI are assessed through the confidence intervals of the block kriging mean predictions for each of the latitude, altitude and exposure classes considered. The NFI1 and NFI2 show wider confidence intervals due to the lower sampling density. The most notable differences for both species arose at mid-latitudes (4747-4752 km), where the block mean kriging prediction for NFI3 and NFI4 falls outside the confidence intervals for the NFI1 and NFI2 (Fig. 3.2). These differences were significant at 95% level (no overlapping between prediction confidence intervals) for beech between the NFI1 and the NFI4.

Fig. 3.2 Universal kriging block mean prediction for the presence/absence indicator variables by latitudinal (above), altitudinal (mid) and exposure (below) classes for Pinus sylvestris (a, c, e) and Fagus sylvatica (b, d, f) from the NFI1, NFI2, NFI3 and NFI4 (from light to dark grey) data. Error bar lines represent the 95% confidence intervals of the block mean for each inventory and species. The black arrow indicates the direction of the shifting trend of each tree species in the study area.


54

3.3.3.1. Latitudinal shifts

The results revealed changes in the latitudinal distribution range of both species at the spatial scale and time frame studied. The block kriging prediction for the 10 latitudinal classes suggests a major increase in the presence of Scots pine in the mid-northern latitudes of the study area from 1971 to 2000 and in the northern latitudes from 2000 to 2010 (Fig. 3.2a). This increase was concomitant with a decrease in Scots pine presence prediction in southern latitudes from 2000 to 2010, which indicates a possible shift in the distribution range of this species towards northern latitudes within the study area. In contrast, the presence of beech generally increased in the region both in northern latitudes (mainly from 1971 to 1990) and mid-southern latitudes (from 1990 to 2010) (Fig. 3.2b), where this species was scarce in the first inventories.

3.3.3.2. Shifts in elevation

Figure 3.2c and 3.2d reveal a general shift in the presence/distribution of both species towards higher elevations. In the case of Scots pine, the optimum altitude (corresponding to the maximum presence prediction values) was found at about 700 m in 1971 and 800-900 m in 2010. The maximum altitude at which the presence prediction was above the 0.50 threshold was 1,250 m in 1971 and 1,450 m in 2010 (Fig. 3.2c). In the case of beech, the optimum altitudinal prediction varied over the period 1971-2010, from 1,300-1,400 m to 1,450-1,600 m. The maximum altitude at which the species was found shifted from 1,550 to 1,750 m (Fig. 3.2d).

3.3.3.3. Changes in exposure distribution

The block averages of the presence by exposure class show that this environmental factor is not determinant for Scots pine distribution (Fig. 3.2e). In contrast, the presence prediction for beech has increased for Northern exposures, where this species is more frequent (Fig. 3.2f).

3.3.3.4. Trade-off between altitude/exposure and latitude

The inclusion of altitude for latitude and exposure for latitude interactions in the universal kriging model provided a deeper insight into the species distribution at regional scale. In the study area, there was a South-North elevation gradient that influenced the altitudinal species distribution. However, figures 3.3a and 3.4 show that for both species, the largest elevation shifts took place at mid and southern latitudes, where the sub-Mediterranean influence is greater and the elevations are lower. The presence increased for northern exposures and southern latitudes for both species (Fig. 3.3b), following the elevation gradient.


55

Fig. 3.3 Universal kriging block predictions of the mean distribution likelihood of each species and NFI, considering the interactions between altitude and latitude (above) as well as exposure (as the cosine of the aspect azimuth (cos(φ)) and latitude (below).

3.4. Discussion

3.4.1. A new method for assessing changes in species distribution from long term forest inventories.

In this study we have presented a new method based on universal kriging for the early detection of changes in the distribution of forest species using long term information derived from the NFI. Kriging techniques provide optimal and unbiased estimates of a regionalized variable in the presence of spatial autocorrelation, allowing the uncertainty of the estimates to be quantified (Mandallaz 2000). In our case study, spatial autocorrelation accounted for a large part of the semivariance of the presence/absence indicator variable for both Pinus sylvestris and Fagus sylvatica. This fact has also been reported for other species with widespread distribution area (Segurado and Araújo 2004). The spatial autocorrelation reflects the spatial continuity of the main factors involved in species dynamics (dispersal, mortality, environmental/physical barriers, historical biogeography and socioeconomic factors). Furthermore, the changes observed in the variogram of each species across the


56

successive inventories provide relevant information for understanding the spatial dynamics and patterns of the species in the region.

The distribution of the analysed species is non-stationary at the scale of the study: whereas Fagus sylvatica dominates in the Atlantic influenced north-western forests, Pinus sylvestris is distributed in the southernmost areas of the study region. Universal kriging, as opposed to other kriging techniques, is capable of explaining the non-stationarity of the species distribution in the study area through the mean function depending on the auxiliary variables included in the model (Cressie 1993). Our results reveal how local environmental variations can also constrain the species distribution, highlighting the benefit of using local and landscape approaches to identify species-environment relationships that are averaged out by global models (Osborne et al. 2007). Furthermore, the species-environment relationships can also display non-stationarity for broadly distributed species (Miller and Hanham 2011), such as the upward shift in the altitudinal distribution of the Scots pine across the latitudinal gradient in the study area or the preference of both species for less exposed locations as the Mediterranean influence increases at southern latitudes (Jump et al. 2009). These latitudinal changes in the species-environment relationships are incorporated through the altitude × latitude and exposure × latitude interactions in the universal kriging model. The species-environment interactions can be a key aspect of analysis when studying changes in the species distribution ranges in transitional bioclimatic regions.

Direct comparison of NFI plot data may lead to overestimation of changes in species distributions due to the fluctuating population dynamics, changes in forest inventory area and variations in sampling density across the successive NFIs (Loiselle 2003). The block kriging prediction resolves this problem by providing an unbiased estimator of the mean for the blocks from the information contained in the entire data set. Block kriging incorporates the change of support from point to block, smoothing variations caused by the different sampling schemes used (Yoo and Trgovac 2011) and allowing direct comparison. Figure 3.2 shows the wider confidence intervals derived from the block kriging variance for the NFI1 and NFI2, which had lower sampling densities than the NFI3 and NFI4. The confidence interval approach was found to be a useful technique for identifying trends in the differences between inventories along environmental gradients. This approach provides information about the magnitude of the differences between inventories relative to the accuracy of the kriging prediction, which may be an alternative to hypothesis testing (Katz 1992) to assess the significance of the changes observed in species distribution.

This new method may provide the basis for more studies at a broader scale. Although in this study we have analysed presence/absence data to identify evidence of changes in the spatial distribution of species, other variables of biological interest derived from plot level information of NFIs could be analysed using these techniques. Based on the changes in species distribution observed using this methodological approach, future research could focus on disentangling the effect of land use changes and climate change on forest dynamics.


57

3.4.2. Changes in the distribution of forests

The methodological approach based on the universal kriging model allowed us to obtain evidence of some of the impacts of global change on species distribution over recent decades at regional scale. Firstly, the universal kriging model indicated a progressive increase in the occurrence of the two species in the study area, which agrees with previous studies reporting a global increase in forests in mountainous areas of other developed countries (Gellrich and Zimmermann 2007). Secondly, with regard to the altitudinal distribution of both Scots pine and beech, the approach based on the universal kriging model revealed a vertical shift of approximately 200 m towards higher elevations between 1971 and 2010; a finding supported by other research conducted at European scale (Lenoir et al. 2008).

As expected, the distribution of Scots pine in the study area was found to have shifted over recent decades towards northern latitudes and higher altitudes. This result is consistent with previous predictions for pine species in the Iberian Peninsula (Benito Garzón et al. 2008). In addition, the spatial autocorrelation range of Scots pine decreased between 1971 and 2010, implying a probable fragmentation and loss of continuity of its distribution area. This finding, together with the decreasing presence of the species over the last decade in the southernmost areas, where the sub-Mediterranean influence is higher, may reflect an increase in the presence of other tree species in the most recent NFI. In these zones, it may be that Scots pine is being replaced by broadleaved taxa like Quercus pubescens, which is more adaptable to drier conditions, as reported in studies of other trailing edge populations of this species (Gimmi et al. 2010). The shift towards the northernmost latitudes and higher altitudes along with both the decreasing presence in the southernmost latitudes and the fragmentation of its distribution area, may suggest a northward retraction of the Scots pine in the region.

Beech populations have shown the greatest expansion in the study area over the four decades covered by this study, increasing its presence at higher altitudes. This seems to confirm the ‘deciduous tree invasion’ in the sub-alpine belt predicted by Kräuchi and Kienast (1993). The progressive increase in the spatial autocorrelation range of the beech indicator variable reflects an increasingly continuous distribution of the species in the study area, as reported in other areas of Europe with similar scenarios (Poljanec et al. 2010). The presence of beech increased significantly between NFI1 and NFI2 in the northernmost mountains, where this species was already widely distributed. However, in the most recent NFIs, the presence of this species unexpectedly increased at mid-latitudes, where it might be expected that the temperate, humid ecological requirements of this species would not be satisfied due to the weaker Atlantic influence. However, it must be considered that this increase mainly occurred on north facing slopes, reflecting the ability of beech to spread easily, although only to favourable biotopes.


58

The altitudinal and latitudinal shifts in the distribution range of both species have led to an extension of the area where the Scots pine and beech coexist. These findings provide new evidence of changes in the forest composition. Scots pine and beech can be considered ‘engineer species’ because their presence substantially characterizes the environment of a site and can define habitats. Therefore, variations in the forest composition resulting from the expansion or retraction of these species populations might have profound implications on the diversity and distribution of other species associated with their forests. For example, endangered species of folivores such as the capercaillie (Tetrao urogallus), very sensitive to habitat fragmentation, find their westernmost Pyrenean habitat in these forests (Rodríguez and Obeso 2000). In the future, the biotic interactions established between the two species analysed in this study will play a crucial role in the persistence and stability of these forest areas. Moreover, these sensitive transitional forest areas will be central to monitoring the effects of global change and to the development of successful conservation strategies for the region.

Fig. 3.4. Schematic reconstruction of Pinus sylvestris and Fagus sylvatica forest dynamics from 1971 to 2010 in the Western Pyrenees. The spatial expansion induced by latitudinal and altitudinal shifts of both species is shown based on the combination of universal kriging results for the two species in the NFI1 and NFI4. Light grey trees represent the sporadic occurrence of each species with a presence prediction value below 0.5.

3.4.3. Conclusion

This work provides a new tool for the early detection of changes in forest species and constitutes a first step towards tackling the effects of global change on forests. The methodological approach proposed allows the spatial and temporal shifts in species distribution over recent decades to be analysed using long term NFI and


59

regardless of differences in the sampling schemes used in each. Furthermore, the confidence interval approach proposed allows us to assess whether the differences detected in the distribution of species are significant.

3.5. References

Aitken SN, Yeaman S, Holliday JA, Wang T, Curtis-McLane S (2008) Adaptation, migration or extirpation: climate change outcomes for tree populations. Evol. Appl. 1: 95–111.

Alpert P, Krichak S, Shafir H, Haim D, Osentinsky I (2008) Climatic trends to extremes employing regional modelling and statistical interpretation over the E. Mediterranean. Glob. Planet Change 63: 163-170.

Bellehumeur C, Legendre P (1998) Multiscale sources of variation in ecological variables: modelling spatial dispersion, elaborating sampling designs. Landscape Ecol 13: 15-25.

Benito Garzón M, Sánchez de Dios R, Sainz Ollero H (2008) Effects of climate change on the distribution of Iberian tree species. Appl. Veg. Sci 11: 169-178.

Costa M, Morla C, Sainz H (Eds) (1997) Los bosques ibéricos. Una interpretación geobotánica. Planeta, Barcelona. pp 572.

Cressie NAC (1993) Statistics for Spatial Data. Wiley, New York.

Deutsch CV, Journel AG (1992) GSLIB Geostatistical Software Library and User's Guide. Oxford University Press, 340 pp.

Engler R, Randin CF, Thuiller W et al. (2011) 21st century climate change threatens mountain flora unequally across Europe. Glob Change Biol. 17: 2330-2341.

FAO (Food and Agriculture Organization of the United Nations) (2000) Global forest resources assessment. FAO Forestry Paper 140. Rome.

Gellrich M, Zimmermann NE (2007) Investigating the regional-scale pattern of agricultural land abandonment in the Swiss mountains: a spatial statistical modelling approach. Landscape Urban Plan 79: 65–76.

Gimmi U, Wohlgemuth T, Rigling A, Hoffmann CW, Bürgi M (2010) Land-use and climate change effects in forest compositional trajectories in a dry Central-Alpine valley. Ann. For. Sci. 67 (7) 701

Goovaerts P (1994) Comparison of CoIK, IK and mIK performances for modeling conditional probabilities of categorical variables. In R. Dimitrakopoulos (ed) Geostatistics for the Next Century, Kluwer, Dordrecht, pp18-29.

Guisan A, Zimmermann NE (2000) Predictive habitat distribution models in ecology. Ecol. Model 135: 147–186.


60

Isaaks EH, Srivastava RM (1989) Applied geostatistics. Oxford University Press, New York.

Jump AS, Mátyás CA, Peñuelas J (2009) The altitude-for-latitude disparity in the range retractions of woody species. Tree. 24: 694-701.

Katz RW (1992) Role of statistics in the validation of general circulation models. Clim. Res 2: 35-45.

Kräuchi N, Kienast F (1993) Modelling subalpine forest dynamics as influenced by a changing environment. Water Air Soil Pollut 68: 185-197.

Lenoir J, Gégout JC, Marquet PA, de Ruffray P, Brisse H (2008) A significant upward shift in plant species optimum elevation during the 20th century. Science 320: 1768-1771.

Loiselle BA, Howell CA, Graham CH, Goerck JM, Brooks T, Smith KG, Williams PH (2003) Avoiding Pitfalls of Using Species Distribution Models in Conservation Planning. Cons. Biol. 17: 1591-1600.

Mandallaz D (2000) Estimation of the spatial covariance in universal kriging: application to forest inventory. Environ. Ecol. Stat. 7(3): 263-284.

Matheron G (1969) Le krigeage Universel. Cahiers du Centre de Morphologie Mathematique, 1. Fontainebleau, France.

Miller J, Franklin J, Aspinall R (2007) Incorporating spatial dependence in predictive vegetation models. Ecol. Model 202: 225–42.

Miller J, Hanham RQ (2011) Spatial nonstationarity and the scale of species-environment relationships in the Mojave Desert, California, USA. Int. J. Geogr. Inf. Sci. 25: 423-438.

Montes F, Hernández MJ, Cañellas I (2005) A geostatistical aproach to cork production sampling estimation in Quercus suber L. forests. Can. J. For. Res 35: 2787-2796.

Montes F, Ledo A (2010) Incorporating environmental and geographical information in forest data analysis: a new fitting approach for universal kriging. Can. J. For. Res 40, 1852-1861.

Ninyerola M, Pons X, Roure JM (2005) Atlas Climático Digital de la Península Ibérica.

Osborne PE, Foody GM, Suárez-Seoane S (2007) Non-stationarity and local approaches to modeling the distributions of wildlife. Divers. Distrib 13: 313-323.


Poljanec A, Ficko A, Boncina A (2010) Spatio-temporal dynamic of European beech (Fagus sylvatica L.) in Slovenia, 1970–2005. For. Ecol. Manage 259: 2183-2190.


61

Rodríguez AE, Obeso JR (2000) Diet of the Cantabrian capercaillie: geographic variation and energetic content. Ardeola, 47:77-83.

Segurado P, Araújo MB (2004) An evaluation of methods for modelling species distributions. J. Biogeog 31: 1555–1568.

Thuiller W, Vayreda J, Pino J, Sabate S, Lavorel S, Gracia C (2003) Large-scale environmental correlates of forest tree distribution in Catalonia (NE Spain). Global Ecol. Biogeogr 12: 313–325.

Tolosana-Delgado R, Pawlowsky-Glahn V, Egozcue JJ (2008) Indicator Kriging without Order Relation Violations. Math. Geosci 40: 327-347.

Vilá-Cabrera A, Martínez-Vilalta J, Vayreda J, Retana J (2011) Structural and climatic determinants of demographic rates of Scots pine forests across the Iberian Peninsula. Ecol. Appl 21: 1162–1172.

Williams MS (2001) New approach to areal sampling in ecological surveys. For. Ecol. Manage 154: 11–22.

Yoo EH, Trgovac AB (2011) Scale effects in uncertainty modeling of presettlement vegetation distribution. Int. J. Geogr. Inf. Sci 25: 405-421.

4. Exploring range shifts of contrasting tree species along bioclimatic and

elevational gradients.


Hernández L, Sánchez de Dios R, Montes F, Sainz-Ollero H, Cañellas I. Exploring range shifts of contrasting tree species along bioclimatic and elevational gradients. (En

2a revisión). Forest Ecology and Management.

CAPÍTULO4: SPECIES SHIFTS ALONG BIOCLIMATIC AND ELEVATIONAL GRADIENTS

65

Resumen

Los gradientes bioclimáticos juegan un papel importante en la determinación de flujos de biodiversidad y son especialmente útiles para captar tendencias continentales en aproximaciones a mucha menor escala. En este trabajo tratamos de detectar cambios en la distribución espacial de diferentes especies forestales ibéricas que interactúan en el gradiente bioclimático Templado-Mediterráneo del norte de España, donde es posible analizar cambios en su distribución a lo largo de gradientes climáticos y geográficos y donde se pueden testar la concurrencia de los cambios observados con las proyecciones basadas en escenarios de cambio climático. A través de la comparación de largas series de datos del Inventario Forestal Nacional (1971-2010) y enfocado en la zona de transición Submediterránea, se evalúan los cambios en la distribución espacial de una especie Submediterránea, una conífera montana, una especie templada caducifolia y una especie esclerófila, mediante modelos de Universal Kriging. Después se analizan los mecanismos que inducen dichos cambios analizando los cambios en la distribución altitudinal de las diferentes especies diferenciando clases de edad y cuantificando la importancia de los cambios de uso de suelo en el área de distribución actual. La especie Submediterránea, la templada y la conífera montana muestran un avance en su área de distribución desde las zonas Submediterráneas hacia zonas más frías y húmedas. Simultáneamente las áreas de distribución de las especies templada y Submediterránea también están avanzando hacia mayores elevaciones. La especie de conífera montana parece retraerse en altitud y hacia menores elevaciones. La especie mediterránea en cambio parece tener una expansión general de su área de distribución. Como consecuencia de estos cambios, se ha producido un aumento en el área donde estas especies coexisten. Los diferentes cambios identificados para cada especie están determinados por la combinación del actual cambio climático, cambios de uso de suelo, disponibilidad de nicho y el manejo forestal al que se han visto sometidas. Los hallazgos de este trabajo contribuyen a mejorar nuestra comprensión de los complejos mecanismos y efectos del cambio global contemporáneo en la distribución de los bosques.

Palabras clave: gradientes bioclimático y altitudinal, cambio en distribución de especies, cambio de uso de suelo, disponibilidad de nicho, cambio climático, pasado manejo forestal, Quercus subpyrenaica, Quercus ilex subesp. ballota, Pinus sylvestris, Fagus sylvatica.


66

Abstract

Bioclimatic gradients play a key role in inducing biodiversity flows and are especially useful as a small-scale proxy for continental-scale patterns. In this study we attempt to detect shifts in the ranges of contrasting Iberian tree species interacting in the Temperate-Mediterranean bioclimatic gradient in northern Spain, where the overall climatic and altitudinal range of the species can be explored and where the forthcoming shifts in species ranges, based on climatic scenarios, can be tested. Data from long-term forest monitoring (1971 and 2010) is compared through universal kriging models to assess the shifts in the ranges of a Submediterranean, a Mediterranean, a temperate deciduous and a mountain conifer species focusing on the Submediterranean transitional zone. The mechanism underlying these shifts in ranges of the species is then explored by, on the one hand, comparing the altitudinal range at different life-stages, while on the other, quantifying the importance of past land use changes on the current range of the studied species. The Submediterranean, temperate and mountain conifer species show march range shifts from the Submediterranean zone towards more humid, colder territories along the bioclimatic gradient. These trends are concomitant with elevational march range shifts towards higher altitudes in the case of the Submediterranean and temperate species and a downwards lean shift for the conifer species range. The Mediterranean species, in contrast, is undergoing a general range expansion. As a consequence, there has been an increase in the size of the areas in which the studied species coexist. The different shifts identified for the studied species along the analyzed gradients are determined by a combination of ongoing climate warming, past land use changes, niche availability and forest management legacy partially contradicting previous projections based on climate change scenarios. The findings contribute to furthering our understanding of the complex mechanisms and effects of contemporary global change in the distribution of forests.

Key words: bioclimatic and elevational gradients, shifts, land use change, niche availability, climate warming, forest management legacy, Quercus subpyrenaica, Quercus ilex subesp. ballota, Pinus sylvestris, Fagus sylvatica.


67

4.1. Introduction

Forest ecosystems are arguably one of the most important terrestrial repositories of biodiversity and carbon; playing a key role in regulating global biogeochemical cycles and providing essential resources for millions of people (Kappelle et al. 1999). Since global change is expected to alter the geographic distribution of plant species at different scales (Thuiller et al. 2008), it is vital to both understand and to report on cases of species range shifts and their potential consequences for forest biodiversity and productivity (Hughes 2000; Walther et al. 2005).

A species range shift is defined as the process whereby species change their distribution over time, seeking the environmental conditions which they require (Lenoir and Svenning 2013). Theoretically, a species range, also known as the fundamental or ecological niche, is defined by all the abiotic constraints that shape the species distribution (Hutchinson 1957). However, due to species interaction and historical factors such as land use and migration history, the geographic distribution of a species is unlikely to fully represent its fundamental niche. The fraction of the fundamental niche that the species is actually occupying is called ‘realized niche’, which is referred to as the species range throughout this article.

Given the direct relationship between species distribution and environmental gradients such as climate (Hutchinson 1957) and the need to determine the likely responses of species to global change, the study of species range shifts along geographic gradients has gained increased attention over recent decades. Consequently, based on previous classifications (Breshears et al. 2008; Maggini et al. 2011) and assuming niche conservatism over time (Peterson et al. 1999), a new conceptual classification of species range shifts along an environmental gradient has recently been proposed that comprises five categories (Lenoir and Svenning 2015): expand, where colonization events inside and outside the existing range dominate; march, where both extinction and colonization events inside and outside the existing range occur; retract, where extinction events occur within the existing range; lean, where colonization events occur within the species range; and crash, defined as general species decline across the existing range.

Ample evidences exits of species range shifts associated with contemporary climate warming (see the recent review of Lenoir and Svenning 2015), latitudinal and altitudinal directional range shifts being particularly well documented. However, over the last decade, this unidirectional and univariate perspective on changes in species´ ranges has been challenged by an increasing amount of counterintuitive evidence, such as stasis or even downward shifts (Lenoir et al. 2010; Crimmins et al. 2011; Callaghan et al. 2015). These unpredicted responses are mainly attributed to human land use modification (Lenoir et al. 2010; Callagham et al. 2015) together with biotic interactions, dispersal constraints, lags in the effects of climate warming or complex regional and local environmental conditions (Lenoir and Svenning 2015). However, despite the importance of developing appropriate conservation strategies (Thuiller et al. 2008), the drivers and associated mechanism underlying species range shifts is an understudied phenomenon (Crimmins et al. 2011; Lenoir and Svenning 2015).


68

Theory (Thomas et al. 2001; Hampe and Petit 2005) and empirical data (Caughley et al. 1988; Anderson et al. 2009) suggest that species reach their greatest abundance at the centre of their range and decline in abundance toward the edges of their range. Thus, it is expected that the first and most critical signs of a broader species response to environmental change will be evident in the shifts and dynamics of populations along the edge of the range. Consequently, most contemporary range shifts in woody species have been reported at the leading/front edge and on the limits of the altitudinal distribution of the species in mountain areas (Jump et al. 2009). However, scarce research has focused on the trailing/rear edge of the distribution or on the lowland range of woody plant species (Jump et al. 2009; Lenoir and Svenning 2015). Similarly, the overall environmental gradient of a species range is an aspect which has received little attention (see however Lenoir et al. 2009; Suzuki et al. 2015). Despite of the incomplete observation driven by focusing on one point of a species range, the changes along the overall geographic gradients or multispecies distribution ranges remain poorly explored. In this regard, regional studies of longlived species range changes along bioclimatic gradients, where the overall geographical and climatic range of many species are found, are suitable for testing hypotheses with regard to wider, much less manageable scales (Gosz 1992). On a continental scale, the transitional zone between bioclimatic units supports high levels of biological diversity (Di Castri and Hansen 1992) and is deemed to play a crucial role in inducing biodiversity patterns and ecological flows between adjacent biomes (Risser 1995).

In the Iberian Peninsula, transitional territories exhibiting intermediate characteristics between Mediterranean and Temperate bioclimatic regions are generically referred to as Submediterranean (Ozenda 1994). These transitional territories account for a large area (19%) of the forested surface in Spain (Maldonado et al. 2001). They harbor temperate, mountain conifer and Mediterranean species as well as genuine Submediterranean plant communities not found in either of the adjacent regions (Sánchez de Dios et al. 2006). As a result, this transitional region provides a perfect scenario to study vegetation range shifts. Additionally, on a continental scale, the trailing edge populations of many species are found in these bioclimatic transitional zones.

Based on future climate change scenarios, a general contraction in the potential range of many Iberian tree species is predicted, this contraction being particularly notable for some of the species interacting in the Submediterranean zone (Benito-Garzón et al. 2008; Ruiz-Labourdette et al. 2012). These studies forecast a dramatic and rapid range contraction for mountain conifer species as well as range contraction and upward shifts for temperate species, both finding here some of their southernmost populations. These projections also suggest a notable decline in Submediterranean species (Sánchez de Dios et al. 2009). In contrast, typical Mediterranean species are predicted to be less affected by climate change (Benito-Garzón et al. 2008; Ruiz-Labourdette et al. 2013).

Within this conceptual framework and with the aim of testing previous projections and general patterns in species range response to global change, we analyze the range shifts of four ecologically contrasted species (Quercus ilex subsp. ballota (Desf.) Samp., Fagus sylvatica L., Pinus sylvestris L. and Q. subpyrenaica E.H del Villar) along the Temperate-Mediterranean bioclimatic gradient in northern Spain


69

from 1971 to 2010 based on national forest inventory (NFI) data, as well as the changes in their altitudinal range during the same period differentiating saplings and adult life-stages. Then, we quantified the importance of past land use changes in the current distribution of the studied species in order to correctly identify the mechanism underlying the species range shifts observed. We hypothesize that tree species have shifted their ranges between 1971 and 2010 and that the patterns and drivers of the observed shifts are species-dependent.

4.2. Material and methods

4.2.1. Study area

The study focuses on the Navarra region in Central Northern Spain (Fig. 4.1). The region is characterized by a steep climatic gradient from the Temperate northern territories to the Mediterranean territories in the South (hereinafter Temperate-Mediterranean gradient). In addition, the alpine biogeographical region is also present in the northeast where the Pyrenees Mountains reach altitudes above 2,000 m. This climatic gradient is even steeper if a NW-SE axis is followed due to two opposite influences. On the one hand, the hyper humid conditions associated with the Bay of Biscay (Atlantic Ocean), bringing 2,500 mm of annual precipitation, and on the other, the continental semiarid Mediterranean conditions associated with the Ebro Basin, with just 337 mm of annual precipitation. These climatic gradients have led to the existence of a broad band of Temperate-Mediterranean transitional territories (hereinafter Submediterranean territories) crossing the center of the region from West to East (Fig.4.1).

The vegetation reflects this marked climatic variation (Fig.4.1). The forests of the NW comprise broadleaved species such as European beech (Fagus sylvatica) and oak forests (Quercus robur L., Q. petraea (Matt.) Liebl.). Travelling east, the humid influence from the west decreases and the Pyrenean altitude increases, leading to the gradual appearance of coniferous species (Pinus sylvestris and Pinus uncinata Ramond ex DC.). Within the southern, Mediterranean areas, sclerophyllous forests of holm oak (Quercus ilex subsp. ballota) dominate the vegetal landscapes; while Aleppo pine forests (Pinus halepensis Mill.) appear scattered across the most xeric territories. Lastly, the Submediterranean territories in Navarra act as a ‘hybridization’ zone for two transitional oaks: Quercus pubescens Willd. and Q. faginea Lam., leading to forests of the hybrid Q. subpyrenaica (Sánchez de Dios et al. 2006).

To define the climatic trends in the region, we selected 11 climatological stations across the study area (Fig.4.1), which represented the most complete climate series available (data was gathered from the Navarra Climatic Network (Meteo-Navarra)). Mean annual temperature (ºC) and total annual precipitation (mm) were obtained from daily climatic data series covering the period from 1938 to 2005. Missing daily data were filled in using data from nearby stations.


70

Fig.4.1. Study area: A) Location of Navarra region in southern Europe between the Alpine, Mediterranean and Temperate bioclimatic regions. B) Illustration of the Temperate-Mediterranean gradient (ranking from SUBM5 to SUBM1) from Sánchez de Dios et al. (2009). C) Distribution of the focal species in the study area in the last inventory (2010) (species co-occurrence is not represented). D) Types of land use change in the period 1956-2008 (1. Agricultural systems abandonment, 2.Tree plantations). E) From left to right, pictures of Quercus ilex subsp. ballota, Quercus subpyrenaica, Pinus sylvestris and Fagus sylvatica. (The theoretical limits of Mediterranean, Alpine and Temperate bioclimatic regions for Europe (EEA, 2011) are shown in A) and B) panels).


71

4.2.2. Characterization of the Submediterranean area.

A transitional zone is a scale-independent concept occurring in areas with a steep environmental gradient which affects key ecological processes and where existing thresholds of gradual environmental gradients cause large changes in ecosystem dynamics (Risser 1995). Since it shares certain characteristics with adjoining areas, it is very difficult to define the limits of a transition zone (Sánchez de Dios et al. 2009). However, the existence of unique plant communities, not found in either of the adjacent regions, makes it possible to define the climatic conditions of these singular areas. In this study, the delimitation of the gradual boundaries of the Iberian Temperate-Mediterranean gradient (SUBM) is based on the distribution of unique Submediterranean tree species, as proposed by Sánchez de Dios et al. (2009). This gradient is demarcated by the ranges of five key climatic variables that define the distribution of Iberian Submediterranean Quercus: total precipitation, summer precipitation and mean annual temperatures (mean annual temperature, mean maximum temperature for the hottest month and mean minimum temperature for the coldest month). A score of 0–5 was assigned, depending on whether all (5) or none (0) of the climatic variables fell within the ranges established. In accordance, we considered Submediterranean territories to be those with a score of SUBM5. Scores from SUBM4-SUBM0 show a gradient towards less Submediterranean climatic conditions (either Mediterranean or Temperate). As can be observed in Fig. 4.1, there were no 0 scores in the study region.

4.2.3. NFI data

To assess changes in the spatial distribution of the selected species, data from two cycles of the Spanish NFI that comprise the largest historical record available (covering forty years of forest dynamics, from 1971 to 2010) were used. The Spanish NFI is a spatial monitoring system covering the entire forest area of Spain. The sampling unit is the region or province and it is performed approximately every 10 years with an intensity of one sampling plot every 1 km2. Although a continuous design with permanent plots has been used in the more recent Spanish NFI cycles, between the NFI 1971 and the NFI 2010 the number of plots and location of these plots varied.

In a first step, a presence/absence indicator variable for each species, comprising different life-stages in each georeferenced inventory plot, was used to define the overall spatial distribution range of the species in each NFI cycle to be considered in the models. In both cycles, individuals with a DBH (diameter at breast height) of between 2.5 cm and 7.5 cm were recorded as saplings in NFI plots. Individuals with a DBH equal to or more than 7.5 cm were recorded as adult life-stage trees. The species codes for F. sylvatica, P. sylvestris and Q. ilex subsp. ballota were harmonized between the two inventory cycles (NFI1971 and NFI2010). Trees identified as Q. faginea and Q. pubescens in the study area in the 2010 inventory were considered as Q. subpyrenaica, whereas trees identified as Q. lusitanica in 1971 were considered as Q. subpyrenaica for this study.


72

4.2.4. Statistical analysis

4.2.4.1. Universal kriging models

Since different sampling designs were used in the two NFI cycles analyzed in this work, forest evolution could not be analysed through direct comparison between plots. Consequently, changes in the spatial distribution of species were identified by using the method based on the Universal Kriging model proposed by Hernández et al. (2014). Using Universal Kriging models (UK), spatial autocorrelation as well as the relationships between the species distribution and species-specific explanatory variables were considered in order to calculate comparable species distribution maps from the two sequential NFI cycles. In the case of Q. subpyrenaica, Q. ilex and F.sylvatica, an exploratory analysis indicated better model performance with the inclusion of altitude, square of the altitude, latitude and expoure as auxiliary variables. In the case of P. sylvestris the auxiliary variables chosen were altitude, square of the altitude, exposure and longitude. The variogram that describes the degree of spatial dependence of the spatial distribution of each species was modelled using the spherical model. The variogram parameters and β regression parameters of the explanatory variables were estimated using the least squares method based on the decomposition of the variance (VLS) proposed by Montes and Ledo (2010). A leaving-one-out cross-validation procedure was then used to determine the bias and the accuracy of the prediction variance estimates of the fitted model (Cressie 1993). This validation set was also used to evaluate the UK prediction performance and accuracy, employing the area under the receiver operating curve (AUC) test (Pearce and Ferrier 2000). The values of sensitivity, specificity and true statistic skill (TSS =sensitivity + specificity -1) were then used to assess the accuracy of the UK interpolation, whereas the proportion of correctly identified presence and absence records was used to check for the absence of bias in the assessment of the distribution area from the UK probability maps for each species and date (Liu et al. 2005). These accuracy values were also used to select the best probability threshold to assess the species presence, checking that the proportion of predicted points above the indicator threshold was unbiased.

4.2.4.2. Species range shifts along the bioclimatic and elevational gradients

The differences in the species range maps for the two dates [(2010 and 1971)] allow the colonization (analysed through the dichotomous dependent variable which takes value 1 where the species was not present in 1971 but was present in 2010 and 0 otherwise) or extinction events (analysed through the dichotomous variable which takes value 1 where the species was present in 1971 but disappeared in 2010) to be assessed. Then, a dependent categorical variable DYN with three levels (colonization, extinction and persistence events), was also derived for each species to analyze the relationship between the spatial dynamics of the species and the Temperate-Mediterranean gradient (SUBM).


73

We used the kappa statistic as a descriptive coefficient to examine differences between the absence/presence maps for each species in the two years considered (Monserud and Leeman 1992). If the rates of presence/absence between the two dates are in complete agreement, the value of the kappa coefficient would be 1, meaning that there have been no important changes in the distribution of the species over the intervening period. If there is no agreement between the two years other than what would be expected by chance, the kappa coefficient would be 0.

To determine the independence of the biological processes observed (DYN: species extinction, colonization, and persistence events) for each studied species with the Temperate-Mediterranean gradient (ranking from SUBM5 to SUBM1), Pearson´s Chi-squared tests were performed. The Chi-square statistic is significant at the 0.05 level with a two sided test. Tests are adjusted for all pairwise comparisons using the Bonferroni correction in a two-way contingency table. In order to correctly interpret the results it must be considered that for the more temperate species considered (F. sylvatica and P. sylvestris) located at higher latitudes, the higher the score of the gradient, the greater the Mediterranean characteristics (warmer mean and maximum annual temperatures and lower total precipitation). In contrast, in the case of the Mediterranean species considered (Q. ilex) located at lower latitudes, the higher the score, the more Temperate the conditions (colder and more humid climate) (Fig.4.1).

Bearing in mind the direct relationship between elevation and temperature gradients, we attempted to discern recent and past climate-related species responses by comparing different life stages when analyzing the species’ altitudinal range width (ecological amplitude in the elevational gradient) and optimum (altitude at which the species reaches maximum abundance within its altitudinal range) predicted at different moments in time. To test the statistical significance of the altitudinal changes in the distribution of each species at different life stages and dates, we used the method based on the block kriging variance (Isaaks and Srivastava 1989). The block mean and the block prediction variance were estimated for each of the strata determined by seven altitude classes in the study area. The block mean of the species indicator is interpreted as the mean probability of the species occurrence in each stratum. At each altitude, confidence intervals for each indicator block prediction were calculated at a confidence level of 95% to assess the differences between each date (Hernández et al. 2014).

4.2.4.3. Changes in composition

To analyse the variation in forest composition in the study zone over the period, the variable species co-occurrence variable was estimated as the number of the studied species with a presence probability greater than the selected threshold in each prediction plot. Then we calculated the differences in the prediction maps between the two dates within the Temperate-Mediterranean gradient (SUBM), considering the co-occurrence of the studied species. The Welch test (Welch 1947) was used to examine in greater depth the differences found due to the non-homogeneity of the variance, and Tamhane´s T2 (Tamahane 1979) test was performed for post hoc multiple comparisons of mean values.


74

4.2.4.4. Past land use changes

Since major land use changes occurred between 1960 and 1970 in rural areas of the Iberian Peninsula as a result of depopulation and the abandonment of traditional economic activities (García-Ruiz et al. 1996), we quantified the importance of past land use changes in the current distribution of the studied species. For this we used a map of land use change (LUC) (1:10,000) between 1956 and 2008 (Tracasa 2008). The land use changes were reclassified into two general types: abandonment of agricultural system and tree plantations (Fig.4.1). A geographical Information System (GIS) was then used to extract the value of the reclassified land use change grid cell from the LUC map to each 2010 NFI plot. After that, the NFI plots were classified as agricultural systems abandonment, plantations or plots without land use changes. Finally, we quantified the importance of past land use changes in the current distribution of each studied species as the percentage of the total plots where each species occurs in each class of land use change.

4.2.4. Software employed

ArcGis 9.3 (ESRI Inc., Redlands, CA, USA) was used to derive the topographical variables from the DEM and to interpolate topographical and climatic variables into the prediction grid. Geostatistical analyses were performed using a MatLab® software package developed by the authors. R software was used to perform the AUC/ROC method and SPSS17.0 for all the other statistical analysis.

4.3. Results

4.3.1. Climatic trends in the study area

The climatic data reveals that the mean annual temperature has risen and precipitation has decreased during recent decades (1971-2005) in comparison to the previous 40 years (1938-1970) (Fig. 4.2). However, while no consistent trend in total annual precipitation was detected over any of the studied periods, mean annual temperature shows a significant increase (of approximately 1ºC) in the most recent past.


75

Fig.4.2. Trends in total annual precipitation (above) and mean annual temperatures (below) through both studied time periods (1938-1971) and (1971-2005). In both cases each dot corresponds to the average precipitation (P) in mm and temperature (T) in ºC. The line indicates the tendency.

4.3.2. Universal Kriging models

Table 4.1.presents the parameters that define the fitted variogram model for each species, reflecting their different ecology and spatial distribution. The nugget effect that defines part of the variance explained by the spatial correlation of the species has higher values in P. sylvestris and F. sylvatica than in the other two species. In these cases, the nugget effect is about 50 % of the sill, probably due to a more continuous and widespread distribution in the study zone compared with the Quercus species, which present a more fragmented spatial distribution. The analyses of cross-validation residuals showed that the fitted Universal Kriging models performed satisfactorily in terms of bias and kriging variance estimation in all cases. The AUC-ROC and accuracy values (sensitivity, specificity and TSS) are high, which indicates a satisfactory performance of the UK models and high agreement between observations and predictions for all the studied species (Table 4.1). Based on the values of the accuracy coefficients for different probability thresholds, the species presence was set at 0.5 (Table 4.2).


76

Table 4.1. UK models results: VLS estimates of the variogram parameters and β coefficients for the UK models for Q. subpyrenaica (Qsub), Q. ilex subsp. ballota (Qilex), P. sylvestris (Psyl) and F. sylvatica (Fsyl) and date, considering as explanatory variables elevation (Elev), square of the elevation (Elev2), exposure (Expo), longitude and latitude. Cross-validation sum of residuals (SEE) and the residual variance ratio (VSEE) are also shown.

Table 4.2. Comparison of the UK species distribution predictions accuracy using different occurrence thresholds. Codes: Q. subpyrenaica (Qsub), Q. ilex subsp. ballota (Qilex), P. sylvestris (Psyl) and F. sylvatica (Fsyl). Prop. Correct. (Proportion of the presence and absence records correctly identified). TSS (true statistic skill).

Species Threshold AUC Sensitivity Specificity Prop. Correct TSS

Qsub

0.33 0.9 0.86 0.92 0.89 0.78 0.50 0.83 0.80 0.94 0.91 0.78 0.66 0.79 0.72 0.90 0.9 0.62

Qilex

0.33 0.92 0.92 0.91 0.91 0.83 0.50 0.90 0.86 0.94 0.92 0.80 0.66 0.84 0.72 0.96 0.91 0.68

Psyl

0.33 0.92 0.93 0.92 0.92 0.85 0.50 0.89 0.86 0.93 0.92 0.79 0.66 0.83 0.72 0.96 0.90 0.68

Fsyl

0.33 0.90 0.90 0.91 0.90 0.81 0.50 0.87 0.80 0.94 0.91 0.74 0.66 0.83 0.83 0.96 0.91 0.79

4.3.3. Species range shifts within the bioclimatic and elevational gradients

The comparison of the four species distribution maps predicted by the UK model between the two NFI cycles (1971-2010) reveals a notable expansion of the forested areas (Fig. 4.3). Q. subpyrenaica and Q. ilex show the highest expansion rates; 7.1% and 5.3% respectively from 1971 to 2010, while for P. sylvestris and F. sylvatica the rates are 1.3% and 1.6%. Although contraction was less frequent, the rate was lowest for P. sylvestris, 0.8 while the values for Q. subpyrenaica, Q. ilex and F. sylvatica were higher at 1.3%, 2.8% and 1% respectively. These results are corroborated by the Kappa statistic. This coefficient indicates only small variations in the distribution of P. sylvestris (0.9) over the period 1971-2010.

UK model Variogram parameters β coefficients of the auxiliary variables Cross-validation

Species IFN Nugget Sill Range (km) β 0 Elev Elev 2 Expo Long Lat SEE VSEE

Qsub 1971 0.072 0.061 44.738 1.037 3.9.x10 4 3.5 x10 7 -0.001 - 4.5 x10 5 0 1.460

2010 0.121 0.046 34.260 -3.229 4.5 x10 4 3.5x10 7 -0.053 - 9.5 x10 5 0 0.510

Qilex 1971 0.056 0.060 58.340 -3.019 3.3 x10 4 2.4x10 7 -0.009 - 6.4 x10 5 0 0.613

2010 0.016 0.064 69.680 -0.920 1.3x10 4 1.3x10 7 -0.008 - 2.2 x10 5 0 1.931

Psyl 1971 0.052 0.101 29.255 -3.148 6.5x10 4 4.3x10 7 - 0.005 - 0 0.791

2010 0.065 0.128 73.790 -1.490 4.0x10 4 2.5x10 7 - 0.002 - 0 0.958

Fsyl 1971 0.000 0.200 9.537 -20.112 1.9x10 4 8.8x10 7 0.062 - 4.1 x10 5 0 1.601

2010 0.000 0.204 8.519 -18.442 1.2x10 4 3.7x10 7 0.084 - 3.8 x10 5 0 1.991


77

However, the lower kappa statistic values for Q. subpyrenaica, Q. ilex and F. sylvatica (0.5, 0.5 and 0.6 respectively) suggest that their distributions have suffered more changes. Q.subpyrenaica was the species which displayed the greatest differences between 1971 and 2010.

Fig.4.3. Map showing extinction, colonization and persistence events of A) Quercus subpyrenaica B) Quercus ilex subsp. ballota C) Pinus sylvestris D) Fagus sylvatica ranges in the study area over the last 40 years predicted by the species specific Universal Kriging models.


78

The Chi-squared tests showed significant relationships between the nominal variable of Temperate-Mediterranean gradient (SUBM) and the nominal variable DYN (Fig.4.4). Q.subpyrenaica shows colonization events in SUBM 1, SUBM2, SUBM3, which correspond to areas of less pronounced Submediterranean conditions. However in SUBM5, although the area where it persists predominates, the extinction events are significantly greater than the colonization events. Q. ilex is colonizing SUBM 3, and SUBM4, while it persists at SUBM 5. The range of P.sylvestris shows the least variation between 1971 and 2010. This species shows colonization events towards SUBM 1 and SUBM2, mainly located in the Alpine and Temperate regions of Navarra, while its presence within SUBM5 is decreasing. Finally, F. sylvatica has maintained its distribution area in its most favourable territories SUBM1 and SUBM2, although extinction events are, however, the most important process occurring in SUBM5.

Fig.4.4. Association between the variable DYN (extinction, colonization and persistence events) and the Temperate-Mediterranean gradient (SUBM from 5 to 1) for each species considered. Significant differences are indicated by different letters (Pearson´s Chi-squared tests, p<0.05).

The block kriging results revealed no significant changes in the elevational range of any of the species over the studied period, although the variation in species occurrence across the elevational gradient between 1971 and 2010 (discriminating between life-stages) reveals interesting trends (Fig.4.5). Altitudinal changes between range width and optimum differed between species. As regards range width, the mountain conifer species, P. sylvestris, is the only species which maintains the same altitudinal amplitude in the two dates considered, although if saplings are taken into account, the altitudinal range is currently moving downwards (Fig.4.5). Between 1971 and 2010 Q. ilex has expanded towards lower altitudes (approximately 40 m), whereas Q. subpyrenaica has also shifted towards higher altitudes (approximately 60 m in minimum and maximum altitudes). However, the sapling range of Q. subpyrenaica suggests a shift towards higher altitudes in recent years (Fig.4.5). Finally, F. sylvatica shows a clear trend of contraction at lower altitudes and expansion towards higher


79

elevations (a shift of between 80 m for the minimum altitude and 200 for the maximum). The optimum altitudinal range, where the species are regenerating abundantly, remains unaltered for the majority of the studied species (Fig. 4.5). Only the Q. subpyrenaica and F. sylvatica species have shifted their mean altitudinal optimum by about 60 m and 40 m respectively.

Fig.4.5. Observed altitudinal range width and optimum for mature (solid line) and saplings (dotted line, Reg) in each date (1971 in grey and 2010 in black) inferred from block kriging predictions for A) Quercus subpyrenaica (Qsub), B) Quercus ilex subsp. ballota (Qilex), C) Pinus sylvestris (Psyl) and D) Fagus sylvatica (Fsyl). 4.3.4. Changes in species co-occurrence

Due to changes in the spatial distribution of the species, the composition of the forests in the study region has varied over the period 1971-2010 (Fig.4.6). If we only take into account the occurrence of the four species, there has been an increase of 4.1% in the forest area where some of the species coexist (accounting for 22.9% in 1971 and 26.9% in 2010 of the total forested area). In most cases, these mixed forest areas are composed of two of the four species considered. Forests in which three of the four species are present are rare, accounting for just 0.02% in 1971 and increasing to 0.2 % in 2010. Q. ilex and F. sylvatica usually form monospecific stands; these two species being the least prevalent in mixed forests, which accounted for 16.5% of their distribution area in 1971, rising to 18.4% and 19.3% respectively by 2010. Q. subpyrenaica and P. sylvestris, however, exhibit greater contact with other species, their presence in mixed forests increasing from 29.9% and 35.1% respectively to 36.0 % and 37.9% over the period. By comparing the predicted maps from the two periods, the Welch and post hoc multiple comparison tests (Table 4.3) indicate a significant increase in species co-occurrence in one main area, the Submediterranean zone (SUBM5) compared with all the other analyzed zones. This trend is also occurring in


80

parts of the SUBM 1 zone, which for most of the studied species are mainly located in the alpine and more temperate regions of the study area.

Fig 4.6. Mapping species co-occurrence over the period (1971-2010).

Table 4.3. Changes in species co-occurrence along the Temperate-Mediterranean gradient in the period 1971-2010. In the table, mean differences (Mean diff.) in species co-occurrence per plot within the Temperate-Mediterranean gradient (SUBM ranking from 5 to 1) between dates and multiple comparisons between them are shown. Standard deviation (St. Dev) and number of plots (n) are shown. Significant difference at level p<0.05.

Gradient code

Mean diff. (St. Dev)

SUBM 1 (n=7,312)

SUBM 2 (n=8,898)

SUBM 3 (n=7,074)

SUBM 4 (n=7,497)

SUBM 5 (n=10,376)

SUBM1 -0.137 (0.32) 0.01 0.13* 0.01 -0.03*

SUBM 2 -0.025 (0.34) 0.12* -0.002 -0.04*

SUBM 3 -0.147 (0.53) -0.12* -0.16*

SUBM 4 -0.023 (0.66) -0.04

SUBM 5 0.015 (0.62)


81

4.3.5. Past land use changes

The GIS integration of the LUC map and the 2010 species occurrence indicates that land use changes have played an important role in the configuration of the current distribution range of P. sylvestris, Q. ilex and Q. subpyrenaica in the study area, affecting 30.3%, 20.0% and17.5% respectively of their current distribution, while F. sylvatica is the least influenced (8.1%) (Fig.4.7). The abandonment of past agricultural systems is shown to be the most important type of land use change affecting their distribution.

Fig.4.7. Percentages of Quercus subpyrenaica (Qsub), Quercus ilex subsp. ballota (Qilex), Pinus sylvestris (Psyl) and Fagus sylvatica (Fsyl) current distribution range that experienced different land use changes in the period 1956-2008 in the study region.

(Codes: A, Agricultural system abandonment; B, Tree plantations; no LUC, area with no land use changes).

4.4. Discussion

Our results confirm the prior hypothesis of tree species shifts along geographical gradients over recent decades revealing differing patterns from one species to another. The results point to march shifts (according to the classification of Lenoir and Svenning 2015) in the ranges of the Submediterranean, temperate and mountain conifer species ranges along the Temperate-Mediterranean gradient towards new, suitable territories. These species are retracting within their existing ranges and expanding beyond them. In the case of the Mediterranean species, the results suggest a range expansion type shift as this species has been undergoing colonization events both within and beyond the existing range. In the case of the elevational gradient, the analysis reveals current march shifts toward higher altitudes in the case of the Submediterranean and temperate species, an ongoing lean shift towards lower altitudes for the montane conifer species and a downwards expansion shift for the Mediterranean species.

The expansion shift found in the distribution of the sclerophyllous Mediterranean species towards Submediterranean territories, along with the associated march shift of the temperate, mountain conifer and Submediterranean species would partly support previous predictions of bioclimatic envelope models suggesting an expansion of suitable climatic territories for Mediterranean species such as Q. ilex with a concomitant retraction of more temperate suitable areas (Ohlemüller et al. 2006). The Submediterranean species has expanded towards the margins of bioclimatic gradient territory whereas the Mediterranean species is tending to colonize the core, supporting the hypothesis of a prospective “mediterraneanization” of this singular area (Benito-


82

Garzón et al. 2008; Sánchez de Dios et al. 2009; Ruiz-Labourdette et al. 2013). This trend would also appear to be corroborated by the broad-spectrum of “thermophilation” identified by Gottfried et al. (2012) in mountain plant communities throughout Europe over the last decade. Since the Submediterranean and Mediterranean species display the highest rates of change, our results support recent hypotheses and predictions suggesting that the greatest changes in species distributions are occurring across the entire range and elevations (Jump et al. 2009; Ruiz-Labourdette et al. 2013) and not only at higher elevations and marginal ranges. In addition, the greater range change rates found for the Submediterranean species also corroborate previous assumptions that species with restricted ranges are more sensitive to climate change (Thuiller et al. 2005).

Some of the shifts observed in the distributions of the studied species, towards colder and more humid territories along the bioclimatic and elevational gradients, appear to confirm the importance of increasing temperatures, drought severity and hydric deficit on tree species dynamics over recent decades in the region. As previously reported for other territories (IPPC 2013; WMO 2013), this climatic tendency has been especially evident in the study area over the last four decades (since 1970), which is the period covered by the field sampling analyzed in this study. However, although the findings for the Submediterranean, temperate and mountain pine species reveal a reduction in their presence in the Submediterranean territories, these species also appear capable of maintaining part of their populations and even increasing their distribution area into adjacent territories and elevational belts performing both upward and downward shifts. If the species distribution maps from the last inventory (2010) are compared with the results from models for the Iberian Peninsula and Europe based on climatic scenarios for the years 2020 and 2050, it would appear that the latter overestimate the short term reduction in the geographical range of Submediterranean as well as mountain pine and temperate species communities (Ohlemüller et al. 2006; Benito-Garzón et al. 2008; Sánchez de Dios et al. 2009).

Theoretical models suggest that climate warming and habitat modification, either in combination or separately, can explain unexpected species range changes (Lenoir et al. 2010). In accordance with the aforementioned theoretical model, the temperate species, whose range is the least modified by human activity in the region in recent decades, is that which displays the most evident trend to shift its range towards higher altitudes and colder biomes. In contrast, the Submediterranean, the mountain conifer and the Mediterranean species, whose ranges have been subject to a higher degree of human land use change, do not exhibit such intuitive shifts. However, based on our results together with these two important drivers, we propose that niche availability (mainly driven by biogeographical complexity and/or constraints) and forest management legacy should also be considered when explaining current species range shifts along environmental gradients. The extent of their influence depends on the studied species.


83

To discern the response of species to past and recent environmental changes, we consider that the observed changes in the altitudinal ranges of the species (at mature stage) between 1971 and 2010 could have been driven by a combination of climate warming and past habitat modification, while the changes detected between the current altitudinal range of the species saplings and their past and current mature stage altitudinal range, are more likely to be related to recent climate change. In accordance with these assumptions, in the case of the Submediterranean and temperate species, our analysis points to a prior expansion of their altitudinal range; upwards in the case of the temperate species and both upwards and downwards for the Submediterranean species, probably due to the release of competition in areas previously used by man as well as to climate warming. However, the comparison of the altitudinal sapling range of this species in 2010 with that observed at previous moments in time indicates an ongoing march shift towards higher elevations in both cases. Since these temperate and Submediterranean species currently inhabit the foothills, and mid-slope areas of mountains in their respective ranges, they are capable of elevational displacement toward available and suitable, cooler vegetation belts in response to the recent climate warming detected in the region. Our results support those of previous studies concerning these two species in the Spanish Pyrenees and pre-Pyrenees (Peñuelas and Boada 2003; Kouba et al. 2012; Hernández et al. 2014; Urli et al. 2014). In the case of the Mediterranean species, our findings suggests that habitat modification together with climate warming can have similar effects on the altitudinal range of species since this species shows a downwards shift in its distribution range over the period. This species, which is mainly distributed in lowland areas where there has been a high degree of human induced disturbances in the past (Garcia Ruiz et al. 1996), occupied only part of its potential distribution area. Hence, the consistent downward shift of this species suggests that its pioneer behavior and adaptation to drier and warmer conditions allows it to recover and colonize climatically viable territory in Mediterranean and Submediterranean areas. This pattern coincides with similar shifts found in mountain areas of Northeastern Spain (Peñuelas and Boada 2003) and with what was partly anticipated based on climatic change scenarios (Benito-Garzón et al. 2008; Ruiz-Labourdette et al. 2012). Finally, the conifer species shows the smallest range variation of the species studied along bioclimatic and elevational gradients and it provides a good example of the potential effects of regular forest management and geographical constraints on long term vegetation dynamics. P. sylvestris has long been and continues to be a highly managed species and it was widely employed in the large afforestation plans undertaken in Spain in mid-20th century (Valbuena-Carabaña et al. 2010). As it was previously proposed by Jump et al. (2009), regular forest management, such as artificially supported natural regeneration, can maintain populations of a species and halt range shifts, thereby further contributing to the existence of a time lag in range shifts. Furthermore, since this species is mainly found occupying the timber line in this region (Costa et al. 1997), as climate warming progresses, any upward shift in its distribution and the detection of such shifts may be constrained by the lack of areas of sufficient altitude where the species would be able to find a suitable habitat. However, in the last inventory a mismatch between the mature and early stages can be observed, indicating a downwards lean shift. This pattern could indicate an ongoing contraction of the range that may agree with previous projections for the rear edge populations of the species (Benito-Garzón et al. 2008). In addition, other factors affecting the species distribution response to changes in


84

environmental conditions such as the resilience, inertia and plasticity (Benito-Garzón et al. 2011), partly conferred by the longevity, vegetative propagation and hybrid behavior that characterizes some of the species studied, might also explain some of the lags and unforeseen species range shifts observed.

To conclude, as a consequence of the species shifts detected, there has been an increase in the width of the ecotones in which the four studied species coexist. The results point to two major hotspots where the species co-occurrence has increased: the Submediterranean territories and to a lesser extent, the temperate and alpine zones. This pattern coincides with the widely accepted view of an upward movement in species richness as a result of elevational and latitudinal shifts and vegetation recovery (Walther et al. 2002). Thus, the changes in species ranges and resulting community reorganizations may have a considerable impact on the way species interact that imply consequences for the functioning and services of these forests (Walther 2002). Some of the data we provide are quite recent, which suggests that the process that we describe is currently taking place and will possibly consolidate in the near future. From now on, since most of the potential spaces for colonization have already been occupied, autogenic processes together with biotic interactions and climatic and physical constraints will be the main factors involved in the forest dynamics of these sensitive areas.

4.5. Conclusions

This work provides evidence of contemporary tree species shifts along bioclimatic and elavational gradients that do not entirely coincide with the projections based on climate change. The species range shifts observed, the resulting increase in tree species co-occurrence and the process of ‘mediterraneization’ found in the Submediterranean transitional zone during recent decades, highlight the suitability of bioclimatic gradients for monitoring the effects of global change, as well as their key role as bridges for the potential migration of tree species. The findings highlight the necessity to consider the effect of land use change, climate change, niche availability and forest management legacy when trying to ascertain species range responses to global change.

4.6. References

Anderson B J, Akçakaya H R, Araújo M B, et al. (2009) Dynamics of range margins for metapopulations under climate change. P. Roy. Soc. Lond. B. Bio., 276(1661): 1415-1420.

Benito-Garzón M, Sánchez de Dios R, Sainz -Ollero H (2008) Effects of climate change on the distribution of Iberian forests. Appl. Veg. Sci.11: 169–178.


Breshears DD, Huxman TE, Adams HD, Zou CB, Davison JE (2008) Vegetation synchronously leans upslope as climate warms. P. Natl. Acad. Sci. Biol.105: 11591–11592.


85


Caughley G, Grice D, Barker R, Brown B (1988) The edge of the range. J. Anim. Ecol. 57, 771–785. doi:10.2307/5092.

Costa M., Morla C, Sainz -Ollero H (1997) Los bosques ibéricos. Una interpretación geobotánica. Planeta.

Cressie NAC (1993) Statistics for Spatial Data. Wiley, New York.

Crimmins SM, Dobrowski SZ, Greenberg JA, Abatzoglou JT, Mynsberge AR (2011) Changes in climatic water balance drive downhill shifts in plant species’ optimum elevations. Science 331: 324 – 327.

Di Castri F, Hansen AJ (1992) The environment and development crisis as determinants of landscape dynamics. In: Di Castri F, Hansen AJ (eds) Landscape boundaries. Consequences for biotic diversity and ecological flows. Ecological studies 92. Springer, New York.

EEA (2011) European biogeographic map. http://www.eea.europa.eu/data-andmaps/data/biogeographical-regions-europe


Gosz JR (1992) Ecological functions in a biome transition zone: translating local responses to broad scale dynamics. In: Hansen A.J., De Castri F. (eds) Landscape boundaries. Springer, New York

Gottfried M, Pauli H, Futschik A, Akhalkatsi M, Barančok P, Alonso JLB, Grabherr G (2012) Continent-wide response of mountain vegetation to climate change. Nature Clim. Change, 2(2): 111-115.

Hampe A, Petit RJ (2005) Conserving biodiversity under climate change: the rear edge matters. Ecology letters, 8(5): 461-467.

Hernández L, Cañellas I, Alberdi I, Torres I, Montes F (2014) Assessing changes in species distribution from sequential large scale forest inventories. Ann. For. Sci. 71:161-171.

Hughes L (2000) Biological consequences of global warming: is the signal already apparent? Trends Ecol. Evol.15: 56–61.

Hutchinson GE (1957) Population studies – animal ecology and demography – concluding remarks. Cold Spring Harbor Symposia on Quantitative Biology, 22:415–427.


86

IPCC (2013) Climate Change 2013: The Physical Science Basis. Contribution of Working Group 1 to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. http://www.ipcc.ch/

Isaaks EH, Srivastava RM (1989) Applied geostatistics. OxfordUniversity Press, New York.


Kappelle M, Van Vuuren MMI, Baas P (1999) Effects of climate change on biodiver-sity: a review and identification of key research issues. Biodivers. Conserv. 8: 1383–1397.

Kouba Y, Camarero JJ, Alados CL (2012) Roles of land-use and climate change on the establishment and regeneration dynamics of Mediterranean semi-deciduous oak forests. For. Ecol. Manag. 274: 143-150

Lenoir J, Gégout JC, Pierrat JC, et al. (2009) Differences between tree species seedling and adult altitudinal distribution in mountain forests during the recent warm period (1986–2006). Ecography. 32: 765–777.

Lenoir J, Gégout JC, Guisan A, et al. (2010) Going against the flow: potential mechanisms for unexpected downslope range shifts in a warming climate. Ecography. 33(2): 295-303.

Lenoir J, Svenning JC (2013) Latitudinal and elevational range shifts under contemporary climate change. In: Encyclopedia of Biodiversity (Ed. Levin SA), pp. 599–611.


Liu C, Berry PM, Dawson TP, Pearson RG (2005) Selecting thresholds of occurrence in the prediction of species distributions. Ecography. 28: 385-393.

Maggini R, Lehmann A, Kery M, et al. (2011) Are Swiss birds tracking climate change? Detecting elevational shifts using response curve shapes. Ecological Modelling 222: 21–32.

Maldonado FJ, Sainz Ollero H, Sánchez de Dios R, Xandri P (2001) Distribución y estado de conservación de los bosques españoles: un análisis de las carencias en la red de territorios protegidos. In: Camprodon i Subirachs J, Plana Bach E (eds) Conservación de la biodiversidad y gestión forestal. Universitat de Barcelona, Barcelona.

Monserud RA, Leemans R (1992) Comparing global vegetation maps with the kappa statistics. Ecol. Model. 62:275–293

Montes F, Ledo A (2010) Incorporating environmental and geographical information in forest data analysis: a new fitting approach for universal kriging. Can. J.Forest. Res. 40: 1852-1861.


87

Ohlemüller R, Gritti ES, Sykes MT, Thomas CD (2006) Quantifying components of risk for European woody species under climate change. Glob. Change Biol. 12(9): 1788-1799.

Ozenda P (1994) Végétation du continent européen. Delachaux et Niestlé, Paris.

Pearce J, Ferrier S (2000) Evaluating the predictive performance of habitat models developed using logistic regression. Ecol. Mod. 133: 225-245.

Peñuelas J, Boada M (2003) A global change‐induced biome shift in the Montseny mountains (NE Spain). Glob.Change.Biol. 9(2): 131-140.

Peterson AT, Soberón J, Sánchez-Cordero V (1999) Conservatism of ecological niches in evolutionary time. Science, 285: 1265–1267.

Risser PG (1995) The status of the science examining ecotones. BioScience 45: 318-325.



Sánchez de Dios R, Benito-Garzón M, Sainz-Ollero H (2006) Hybrid zones between two European oaks: a plant community approach. Plant Ecol.187:109–125.

Sánchez de Dios R, Benito-Garzón M, Sainz-Ollero H (2009) Present and future extension of the Iberian Submediterranean territories as determined from the distribution of marcescent oaks. Plant Ecol. 204: 189-205.

Suzuki SN, Ishihara MI, Hidaka A (2015) Regional-scale directional changes in abundance of tree species along a temperature gradient in Japan. Global Change Biology. doi: 10.1111/gcb.12911.

Tamhane AC (1979) A comparison of procedures for multiple comparisons of means with unequal variances. J. Am. Stat. Assoc.74, 471–479.

Thomas CD, Bodsworth EJ, Wilson RJ, et al. (2001) Ecological and evolutionary processes at expanding range margins. Nature, 411: 577–581.

Thuiller W, Lavorel S., Auraújo M.B (2005) Niche properties and geographical extent as predictors of species sensitivity to climate change. Global Ecol. Biogeogr. 14, 347–357.

Thuiller W, Albert C, Araújo MB.et al. (2008) Predicting global change impacts on plant species’ distributions: future challenges. Perspect. Plant Ecol. 9(3): 137-152.


88

Tracasa (2008) Mapa de cambios (MCA 56-08). Ocupación del suelo en Navarra: desarrollos realizados y cambios ocurridos en los últimos 50 años. Sección de Evaluación de Recursos Agrarios del Departamento de Desarrollo Rural y Medio Ambiente del Gobierno de Navarra. Pamplona.




Walther GR, Berger S, Sykes MT (2005) An ecological ‘footprint’ of climate change. P. Roy. Soc. Lond. B. Bio. 272: 1427–1432.

Welch BL (1947) The generalization of ‘‘student’s’’ problem when several different population variances are involved. Biometrika. 34: 28–35.

WMO (2013) The Global Climate 2001–2010: A Decade of Climate Extremes – Summary Report. World Meteorological Organisation, Geneva.

5. Tracking Fagus sylvatica leading edge in North-Western Iberia: Holocene

migration inertia, forest recovery and recent global change


Hernández L, Sánchez de Dios R, Montes F, Sainz-Ollero H, Cañellas I. Tracking Fagus sylvatica leading edge in North-Western Iberia: Holocene migration inertia,

forest recovery and recent global change. (En revision) Perspectives in Plant Ecology, Evolution and Systematics.

CAPÍTULO5: TRACKING FAGUS SYLVATICA LEADING EDGE IN NORTH-WESTERN IBERIA

91

Resumen

La Península Ibérica constituye el límite meridional de distribución europeo de Fagus sylvatica L. y aunque el límite noroccidental peninsular parece ser adecuado climáticamente para la especie, F. sylvatica ahí está ausente. Entre otros motivos se ha sugerido que este hecho se debe a que la especie no está aún en equilibrio con el clima debido a la existencia de un desfase en su expansión post-glacial. El objetivo de este trabajo es dilucidar cuales son los factores bióticos y abióticos determinantes de la dinámica de población y distribución de la especies durante las últimas décadas en la región Atlántica. Además, basado en previas hipótesis que sugieren que la expansión de F. sylvatica en el Noroeste peninsular está relacionada con la retracción de Quercus petraea (Matt.) Leibl., también se estudia la dinámica en la sucesión forestal de la especie en base a procesos demográficos y la interacción de las dos especies. A partir de los datos de los últimos ciclos del Inventario Forestal Nacional (1986-2012) en el área de estudio, se analiza en detalle el cambio en la dominancia, estructura poblacional, mortalidad, regeneración y el incremento en área basimétrica de las dos especies considerando diferentes tipos de bosque. Además, mediante modelos lineales generalizados y técnicas de modelado bayesiano se examinan los factores directos e indirectos que determinan la regeneración y la sucesión forestal. Contrariamente a lo proyectado bajo las actuales condiciones de cambio climático, F. sylvatica está expandiendo su distribución en el noroeste peninsular, donde está incrementando su dominancia y está colonizando zonas de menor altitud. En consecuencia, la región biogeográfica Atlántica Ibérica se puede considerar como un frente de avance o “leading edge” de la distribución de F. sylvatica. Los resultados identifican una relación interespecífica entre Q. petraea y F. sylvatica que afecta negativamente a la primera, arrojando evidencias del reemplazo de las masas de Q. petraea por las de F. sylvatica. Estos hallazgos no sólo confirman previas hipótesis biogeográficas, sino que proporciona nuevas claves a considerar en las estrategias de gestión y conservación de la zona de estudio bajo presentes y futuras condiciones climáticas.

Palabras clave: Fagus sylvatica, región biogeográfica Atlántica, limite de distribución, sucesión forestal, regeneración, demografía, Quercus petraea, Península Ibérica.


92

Abstrat

The Iberian Peninsula constitutes the south-western limit of the European distribution of Fagus sylvatica L. However, although the Iberian North-West extreme would appear to be climatically suitable for the species, F. sylvatica is not present there. This fact has led to the suggestion that the species is not yet in equilibrium with the climate and that there is a migration lag in the post-glacial expansion of the species. The objective of this work is to understand the main biotic and abiotic factors driving F. sylvatica distribution and population dynamics in the Iberian Atlantic biogeographical region over recent decades. Furthermore, in the light of other studies which suggest that the expansion of F. sylvatica in North-Western Iberia might be related to the retraction of Quercus petraea (Matt.) Leibl. forests, we also study the demographic trends of both F. sylvatica and Q. petraea along with oak-beech interaction processes to infer forest succession dynamics. Using data from the last two cycles of the Spanish National Forest Inventory (1986-2012) for the study area; dominance, population structure and recruitment as well as basal area increment were analysed by different forest types for the two target species. General linear models and Bayesian structural equation modelling techniques were also applied to study the direct and indirect drivers of recruitment and forest succession. Contrary to what might be expected under the current conditions of climatic change, the population of F. sylvatica in North-Western Iberia is expanding, the basal area increment of F. sylvatica increasing westwards and new recruitment occurring in the lowlands. Accordingly, the Iberian Atlantic biogeographical region may be considered one of the leading edges of F. sylvatica. Our results also identify the existence of an inter-specific relationship between Q. petraea and F. sylvatica which negatively affects Q. petraea. The findings evidence the fact that Q. petraea forests are being replaced by F. sylvatica forests in North-Western Iberia. These results not only confirm previous biogeographical hypothesis but also provide new leads for forest management and conservation strategies under current and future climatic conditions.

Key words: Fagus sylvatica, Atlantic biogeographical region, leading edge, forest succession, recruitment, demography, Quercus petraea, Iberian Peninsula.


93

5.1. Introduction

European beech (Fagus sylvatica L.) is a widespread late-successional deciduous tree species in Europe. Its latitudinal distribution extends from Sweden and Norway to Sicily. From west to east it is present from the Cantabrian Mountains in North-Western Spain to the Carpathians and Balkan Mountains (Ellenberg 1966). However its modern distribution in Europe dates from very recent times. Both palaeabotanical and phylogenetic studies reveal that the expansion of F. sylvatica from its glacial refugia started during the Late-Glacial (6,000-7,000 cal. yr bp) and that the increase in the area occupied by beech populations was exponential from the Late-Glacial until about 3,500 cal. yr bp, then slowed down towards an equilibrium value (Magri 2008). Due to its late post-glacial expansion in Europe, at the time F. sylvatica started to spread, most of Europe was already covered by forests within which it was difficult for new species to establish (Green 1987; Giesecke et al. 2007; Brewer et al. 2002). Thus, after the end of the last glaciation, European beech occupied no more than approximately 50 % of the area potentially occupied today (Pretzsch et al. 2013). In addition to favorable climatic conditions, other intrinsic biotic and anthropic factors such as competitive interactions, dispersion capacities, genetic adaptations and habitat disturbance induced by human activities have been identified as factors underlying the rapid expansion of F. sylvatica in the late Holocene (Tinner and Lotter 2006; Giesecke et al. 2007; Saltre et al. 2013).

As regards the Iberian Peninsula, although F. sylvatica was present in the Iberian Atlantic mixed forests before the Holocene glaciations (Ramil-Rego et al. 2000; López-Merino et al. 2008), the dominance of the European beech over other Atlantic species as well as over Pinus sylvestris L. and the establishment of monospecific beech forests is very recent (Rubiales et al. 2008; Muñoz-Sobrino et al. 2009). Both palaeopalynologic and genetic data point to a recent East-West Iberian colonization of beech forests from East and Central Europe via the Pyrenees during the late Holocene (3,000 BP) (Magri 2008). The newly arrived, highly competitive F. sylvatica, with a high rate of dispersion, mixed with the European beeches already present in Iberia, expanding from the glacial refugia to conform the modern diversity of F. sylvatica haplotypes found in the Iberian Peninsula (Magri 2008).

Today, F. sylvatica is the most widespread broadleaved tree species in northern

Spanish forests (388,000 ha as dominant species and 80,000 ha in mixed and other forests) (MFE50 2001)). The Iberian Peninsula constitutes the south-western edge of the distribution of the species, where it is mostly present in the montane bioclimatic belt in north-central Iberia (Costa et al. 1997) (Fig. 5.1). In the Spanish Atlantic biogeographical region, the species is found in the Cantabrian range. In these mountains, the abundance of F. sylvatica decreases from East to West and it is absent in the westernmost part, where sessile oak (Quercus petraea (Matt.) Liebl.) forests are more abundant.


94

However, North-Western Iberia would seem to be climatically suitable for F. sylvatica when modelling the species distribution (Benito-Garzón et al. 2008; Meier at al. 2011; Saltre et al. 2013). These findings support recent hypotheses suggesting that the postglacial expansion of beech has been and still is strongly dispersal-limited with multimillenial migration lags and therefore it is not yet in equilibrium with the climate (Svenning and Skov 2004; Fang and Lechowicz 2006; Svenning et al. 2008, Svenning and Sandel 2013). In this regard, Saltre et al. (2013) found that F. sylvatica was only in equilibrium with climate in some parts of its range. The fact that North-Western Iberia was not one of those regions has led some authors to suggest that the expansion of European beech forests in the northern Iberian Peninsula may be still in progress (Rodríguez-Guitán 2004). Others have associated the expansion of F. sylvatica in North-Western Iberia with the retreat of Q. petraea forests (Costa et al. 1997). However, to date there have been no published studies of the recent spatio-temporal dynamics of F. sylvatica populations on a broad scale in the region to support these assumptions.

In the present study we analyse the dynamics over recent decades of F.

sylvatica in the Spanish Atlantic region, which makes up 85% of the distribution area of the species in Spain (MFE50 2001). Our hypothesis is that F. sylvatica is still expanding westward, following the Holocene migration inertia at its leading edge (according to the “centre-periphery hypothesis”, Hampe and Petit (2005)). The leading edge model of colonization states that range expansions involve long-standing increases in species abundance and/or distribution area as a result of recruitment success concomitant with a high level of demographic stochasticity, low mortality rates and an increase in dominance and growth. Moreover, it is thought that we are still witnessing the substitution of Atlantic sessile oak forests by F. sylvatica forests.

The main objectives of this study are: (1) to analyse the distribution and population dynamics of F. sylvatica to assess the existence of colonization shifts; (2) to study both F. sylvatica and Q. petraea demographic trends and oak-beech interaction processes to infer forest succession dynamics; (3) to examine the structure of the causal relationships (direct and indirect) among the explanatory factors that modulate recruitment in both species and (4) to identify the main driving factors (climate, topography, forest succession) that determine the observed patterns of recruitment and basal area increment in F. sylvatica. We used data from the Spanish National Forest Inventory (NFI), but did not include plots with forest management to avoid recent artificial inferences on the successional process.

Identifying the current trends of F. sylvatica in North-Western Spain in relation

to its Holocene dispersion will help us to understand the lack of fit of present and future distribution models under a scenario of global change. In addition, the detailed study of F. sylvatica-Q. petraea interactions in North-Western Iberia provides useful information for forest management and conservation strategies under current and future climatic conditions.


95

5.2. Material and Methods

5.2.1. Study area

The study area covers the distribution area of F. sylvatica within the Atlantic biogeographical region in the Iberian Peninsula (European Environment Agency boundaries 2011) (Fig. 5.1).This region extends westward from the Pyrenees comprising the area between the Atlantic Ocean and the north face of the Cantabrian range. The study area encompasses a broad altitudinal range from 500 m to 1,700 m a.s.l., which covers the entire altitudinal range of F. sylvatica in the region. It is characterized by temperate humid Atlantic conditions without a dry period, with mean annual temperatures ranging from 7ºC to 15 ºC and mean annual rainfall from 900 mm to 2,500 mm.

Fig.5.1. Study area: a) Study area located at the Southwestern limit of Fagus sylvatica European distribution (EURFORGEN 2009). b) Location of the selected NFI plots in the study area where the Atlantic Biogeographical region (European Environment Agency boundaries 2011) is highlighted.

In the study region, F. sylvatica can be found in pure and mixed stands or coexisting in forests dominated by other species. Pure beech forests are mainly associated with foggy conditions and north-facing slopes, under the influence of wet winds from the Cantabrian Sea in lowland hilly areas with higher summer rainfall (Rozas et al. 2015). This species is also present in mixed forest in lowlands, coexisting with other broadleaved species such as Quercus robur L., Fraxinus excelsior L. Castanea sativa Mill. or Acer spp. or at higher altitudes with Quercus petraea or Q.


96

pyrenaica Willd. It is also present in pure stands of the abovementioned broadleaved species and plantations of Pinus radiata D. Don and P. sylvestris.

5.2.2. Data analyzed

The historical cartographical sources available for the studied region, which comprised Forest Maps from 1966 (Ceballos 1966) and 1995 (Ruiz de la Torre 1986-2002), were compared to assess changes in the distribution area of F. sylvatica over the last four decades. However, since the scale of the maps (1:400,000 the 1966 map and 1:200,000 the other) does not allow a thorough analysis, the results obtained can only be considered as indicative.

Hence, the spatially detailed information from the Spanish NFI was used in this study. Two consecutive cycles (T0: 1986-2008 and T1:1996-2012) undertaken in different provinces within the Atlantic biogeographical region of North-Western Spain were selected to cover a time interval of approximately 10 years (see plot distribution in Fig. 5.1). In these NFI cycles, permanent plots were established systematically in the forested area at the intersections of a 1 km × 1 km grid. Field plots consist of four concentric circular areas with radii of 5, 10, 15 and 25 m depending on the dbh (diameter at breast height) of the trees. In the 5 m radii plot, different recruitment categories (trees with dbh<2.5 cm) and saplings (trees with 2.5 cm < dbh <7.5 cm) are counted and their mean height estimated. Furthermore, other forest attributes such as silvicultural treatments are recorded.

We began our analyses with a dataset of 27,543 NFI plots from the two consecutive cycles for the provinces selected in the study region (Fig. 5.1). Then, we selected all plots in which F.sylvatica and/or Q.petraea were present (trees or recruitment) and discarded all plots in which management operations such as thinning or harvesting had been recorded during the period between the studied cycles. The final subsample analyzed consisted of 3,914 NFI plots.

Based on the species composition and dominance from NFI information, the selected plots were classified into seven forest types (Table 5.1). Plots where the basal area of one tree species was greater than 70% of the total basal area were classified as pure stands; plots where the joint basal area of two or more species was greater than 70% were classified as mixed stands (MFE50 2001). Thus, the seven types defined were: pure plots of F. sylvatica (PFsylv), Q. petraea (PQpetr) and other species (POther; e.g. Q. pyrenaica, C. sativa, P.sylvestris, P. radiata) along with mixed plots of the target species (MFsylv, MQpetr, MFsylvQpetr) and others (MOther). In the ongoing analysis, only the forest types for which there was a significant number of plots with the target variable were considered (n>5).

To analyze stand basal area increment and forest succession we summarized

basal area (m2/ha), dead trees per hectare and recruitment/sapling presence for the focal species per plot.

The comparison of the information derived from the plots in which F. sylvatica was present in two consecutive NFI cycles allowed the plot basal area increment (BAI) to be assessed (1,607 plots). This dataset was also used to calculate the mortality


97

rates of the two focal species. Furthermore, in order to analyze the diameter distribution (of living and dead trees) of each species across the study area, the proportion (number of trees per hectare) by dbh classes was also calculated. Table 5.1. Characterization of the different forests types found in the study area.

NT1, number of plots in the second cycle; TMMIN (ºC), mean minimum temperature of the coldest month; TMED (ºC), mean annual temperature; TMMAX (ºC), mean maximum temperature of the hottest month; PTOT (mm), total annual rainfall; PVER (mm), summertime precipitation; N (mean number of trees/ha); G, mean basal area in m2/ha; Expo, exposure as the cosine of the aspect azimut; Alt, altitude in m a.s.l.

We then classified each plot using an indicator of stand succession based on the concurrence (or not) of adult and recruitment individuals of the target species according to Carnicer et al. (2014). Thus, plots with recruitment success are defined as plots in which both saplings and adult trees of the focal species are observed; plots with recruitment failure are those plots with presence of adult trees and absence of saplings; and plots with new recruitment are those plots without adult trees in which sapling and recruitment of the analyzed species are observed. For these analyses, a sample of 3,442 plots for the first cycle (T0) and 2,581 plots for the second cycle (T1) were used.

Finally, a total of 21 predictors were considered as independent variables in order to analyze the relative importance of abiotic and biotic factors determining recent BAI and recruitment of F. sylvatica over recent decades. Some of these variables were also used to characterize the different forest types (Table 5.1). As regards the abiotic factors; topographical variables (altitude, exposure, longitude and latitude) were taken from the digital elevation model of Spain with a spatial resolution of 25 m (U.T.M, ED 50). Climatic data were obtained from MAPA (1974–1990). Five variables were selected: total annual rainfall, summertime rainfall, mean annual temperature, mean maximum temperature of the hottest month and mean minimum temperature of the coldest month. As biotic factors, we considered forest stand attributes derived from NFI datasets such as total basal area (BA) and density at plot level (N), recruitment occurrence of the target species, or species dominance index (Sdi). Sdi was assessed as the numerical strength or conspecific dominance for a selected species in relation to the total number of individuals of other species in the plot for a target variable (basal area (SdiBA), density (SdiN), dbh (SdiDbh), maximum dbh (SdiMaxDbh)). As an

Forest type Forest type code NT1 TMMIN TMED TMMAX PTOT PVER N G Expo Alt

Mixed forests with Q.petraea MQpetr 75 0.2 10.5 23.6 1337.9 167.0 680.2 18.5 0.04 763.4

Mixed forests with F.sylvatica MFsylv 476 1.3 11.3 24.3 1417.6 200.6 620.1 21.2 0.21 711.3

Mixed forests of F.sylvatica and Q.petraea

MFsylvQpetr 155 -0.6 9.9 23.9 1255.7 154.3 590.7 25.9 0.11 968.5

Mixed forest of other species MOther 35 2.0 11.8 24.5 1487.5 211.9 758.8 21.7 0.21 521.6

Pure forests of Q.petraea PQpetr 175 -1.1 9.7 24.2 1249.3 147.7 522.7 20.6 -0.09 1024.2

Pure forests of F. sylvatica PFsylv 1369 0.03 10.4 24.1 1367.9 183.4 492.5 26.1 0.22 993.3

Pure forests of other species POther 296 0.9 11.1 24.3 1334.2 182.2 647.5 25.7 -0.03 710.9


98

example, Sdi BA was determined by the value of the total basal area of a species with respect to the sum of the basal area of the rest of the species in the plot. 5.2.3. Data analysis

Differences in the dbh structure of living and dead standing trees of F. sylvatica and Q. petraea, as well as changes in the basal area of F. sylvatica between cycles were assessed by forest type. Since the mean basal area increment of F. sylvatica did not meet the assumptions of normality and homogeneity of variances, differences in the change in basal area between cycles and forest types were tested through a non-parametric Kruskal-Wallis test. Post- hoc Mann–Whitney U tests followed by Bonferroni correction were then used to determine which pairs of forest types differed (Day and Quinn 1989).

Patterns for each recruitment category (success, failure and new) for both F. sylvatica and Q. petraea in pure or mixed forests were then studied using Pearson Chi square tests. Tests were adjusted for all pairwise comparisons using the Bonferroni correction in a two-way contingency table.

To explore the structure of the direct and indirect effects between the biotic and abiotic variables affecting the recruitment of F. sylvatica and Q. petraea, we used a Bayesian structural equations model (SEM). Specifically, we applied SEM models to determine how the presence and abundance of F. sylvatica affects recruitment of Q. petraea (after accounting for other contributing physical factors such as climate, aspect, longitude and latitude). Figure 5.2 illustrates the set of direct and indirect relationships examined in the SEM models that were developed using Amos software (Arbuckle 2010). Indirect effects are those exerted through the mediation of a third explanatory variable.

Bayesian SEM analysis allowed us to use non-normal data (recruitment variables: recruitment success, recruitment failure and new recruitment are binomial variables). Bayesian approaches treat the unknown model parameters as random variables and assign probabilities to the subsets of the parameter space. First, Bayesian methods start by assuming prior distributions for model parameters. Prior distributions describe the probability assigned to a parameter in advance of any empirical evidence. Then Bayesian methods use Bayes’ theorem to derive the posterior distributions of parameters. The posterior distribution is an updated distribution for the parameters, which reflects a combination of prior belief and empirical evidence (Bolstad 2004). Finally, numerical methods based on Markov chain Monte Carlo (MCMC) simulations are used to derive posterior distributions. Thus the reported parameters are means of marginal posterior distributions generated from MCMC algorithms. To judge model adequacy, posterior predictive P and convergence statistic (CS) values were examined. Posterior predictive P is recommended to be close to 0.5; CS should be close to 1, and values below 1.1 are considered acceptable (Gelman et al. 2004).

The conservative default convergence criterion threshold provided by AMOS software is 1.002. We used a 0.9 acceptance threshold to judge the adequacy of the model. In addition, we examined the significance of each path using 95% Bayesian


99

credibility intervals, since a coefficient can be considered significant when its 95% credibility interval does not overlap zero (Arbuckle 2010).

Fig.5.2. Scheme of the relationships analysed in the structural equation models (SEM).We tested the effects on F. sylvatica and Q. petraea recruitment categories (success, failure and new) of abiotic (climatic variables, exposition, longitude and latitude) and biotic (stand attribute factors such as SdiBA_Fagus: species dominance index (relative basal area of F. sylvatica with respect to the rest of the species of that plot (%)); SdiBA_Petraea: species dominance index (relative basal area of Q. petraea with respect to the rest of the species of that plot (%))).All correlated variables were removed.

Finally, to provide information on the relationships and strengths among Fagus sylvatica basal area increment between NFI cycles and biotic and abiotic predictors, a general linear model (GLM) with gaussian error and identity link was used. We chose a GLM because it accepts a combination of continuous and categorical variables as well as non-normally distributed ones (Hosmer and Lemeshow 1989). Multicollinearity was verified using Pearson correlation coefficients. Some of the explanatory variables that were highly correlated (| r | > 0.8) were excluded prior to building the model. We followed a step-by-step model-building procedure and the fit of the model was tested after the elimination of each variable. Deviance reduction, estimated as: D2 = (null deviance -residual deviance)/null, was used as the measure of discrepancy to assess the goodness-of-fit of the model (Crawley 1993).

ArcGis 9.5 (ESRI 2006) was used as an image analysis tool as well as to extract topographical and climatic variables to the 1km x 1km grid. R software and SPSS17.0 were used for all the other statistical analyses.

Summer

rainfall

Temperature

Annual rainfall

Exposition

Longitude

Latitude

Success_petraea

New_petraea

Failure_fagus

New_fagus

SdiBA_petraea

SdiBA_fagus

Failure_petraea

Success_fagus


100

5.3. Results

After comparing historical forest cartography for the study area (Fig. 5.3) we found that 23.23% of the surface area identified as Q. petraea forests in 1966, was shown as F. sylvatica forests in 1995. In addition, the new area occupied by F.

sylvatica in 1995 was far larger than that of Q. petraea (196,481 ha against 37,510 ha).

Fig.5.3: Evolution of F. sylvatica forest distribution in the studied area between 1966 and 1995. Codes: Q. petraea to F. sylvatica: Q. petraea forests present on the 1966 map which have been replaced by F. sylvatica forests on the 1995 map. New F. sylvatica: forest which were not present on the forest map from 1966 (Ceballos 1966) but that were present on the 1995 map (Ruiz de la Torre 1986-2002). Maintained F. sylvatica: F. sylvatica forests present on both forest maps. Disappeared F. sylvatica: F. sylvatica forests present on the forest map from 1966 but not present on the 1995 map.

5.3.1. Mortality rates and dbh structure of living and dead standing trees.

Mean mortality rates of F. sylvatica and Q. petraea showed no significant differences by forest types, with mean annual values for the two species of around 0.02 in mixed stands, 0.03 in their respective pure stands and 0.004 in pure stands of other species.

The pattern in the proportion of living and dead trees of different dbh classes for the two target species by forest type differs (Fig.5.4). F. sylvatica populations show a similar diametric distribution in all forest types for living and dead trees, presenting high demography stochasticity. The proportion of smallest diameter trees was greater in both cases. Q. petraea shows similar patterns for living and dead trees in the pure stands where it dominates. However, this species shows completely different patterns for the dbh structure of living and dead trees in F. sylvatica pure and mixed stands with a high proportion of medium and large diameter trees, indicating an ageing stage in these forest types.


101

5.3.2. Basal area increment

Differences in mean basal area increment of F. sylvatica are significantly associated with forest types (Kruskal-Wallis, df = 4, χ2 = 154.80, p < 10−16). The post-hoc analysis (Mann-Whitney test at α = 0.05) found significant differences in F. sylvatica mean basal area increment between some forest types. Pure F. sylvatica (PFsylv) forest has significantly higher basal area increments than the other forest types, while the increments in mixed forest of F. sylvatica (MFsylv) and F.sylvatica and Q. petraea (MFsylvQpetr) (not significantly different between them) are greater than in pure forests of other species (POther).

Fig. 5.4. Proportion of living (above) and dead (below) trees by diameter at breast height (dbh) classes by forest type for a) F. sylvatica and b) Q.petraea.

5.3.3. Recruitment patterns

The presence of F. sylvatica in the Iberian Atlantic region is greater than that of Q. petraea. The former was present (either as adults or saplings) in 90.55% of the selected plots whilst Q. petraea was present only in 23.56%. Finally, 14.06% of the studied plots presented recruitment of both species. Figure 5.5. summarizes the observed trends for F. sylvatica and Q. petraea recruitment in both NFI cycles. 62.5% of the plots at T0 and 68% at T1 were detected as having F. sylvatica recruitment success and 3.2% and 5.8% respectively as new recruitment. Whereas F. sylvatica is characterized by higher recruitment success than failure in both time periods, Q. petraea shows more recruitment failure than success at T1. In both time periods, Q. petraea displays more new recruitment than F. sylvatica. Differences were statistically significant at both T0 and T1 (Pearson Chi-square at 0.05 level).


102

Fig. 5.5. Recruitment category trends for the target species from the first NFI cycle (T0) and the second NFI cycle (T1). Percentage of plots with recruitment success, recruitment failure and new recruitment for a) F. sylvatica and b) for Q. petraea.

Fig. 5.6. F. sylvatica/Q. petraea interaction in the second NFI cycle (T1; Nº plots = 2,581): a) Percentage of plots with Q. petraea recruitment arranged by recruitment categories. Plots with presence and plots with no presence of F. sylvatica are indicated. Number of plots with Q. petraea recruitment = 608 plots. b) Percentage of plots with F.sylvatica recruitment arranged by recruitment categories. Plots with presence and plots with no presence of Q. petraea are indicated. Number of plots with F. sylvatica recruitment = 2,336 plots. c) Percentage of plots with Q. petraea recruitment arranged by recruitment categories that also show F. sylvatica categories. Due to the co-occurrence in a single plot of both F. sylvatica and Q. petraea the addition of the number of plots with Q. petraea recruitment plus the number of plots with F. sylvatica recruitment (608+2,336) is greater than the total plot number (2,581).

01020304050607080

Success Failure New

Pe

rce

nte

ge o

f p

lots

Q. petraea recrutiment categories

01020304050607080

Success Failure New

Pe

rce

nte

ge o

f p

lots

F. sylvatica recruitment categories

T0 T1

a) b)

0

1

2

3

4

5

6

Fagus success Fagus failure Fagus new

Perc

enta

ge o

f plo

ts

F. sylvatica recruitment categories

Petraea success

Petraea failure

Petraea new

No F. sylvatica presenceF. sylvatica presence

No Q. petraea presenceQ. petraea presencea) b)

c)


103

In addition, Figure 5.6a shows differences in the Q. petraea recruitment pattern depending on whether F. sylvatica is present in the plot in the second NFI cycle (T1). Thus, the presence of F. sylvatica negatively affects Q. petraea recruitment (Pearson Chi-square statistic at 0.05 level). Comparison of column proportions (z test adjusted using Bonferroni corrections at 0.05 significance level) shows that when F. sylvatica is present in the plot, the probability of Q. petraea recruitment failure is greater than the probability of either Q. petraea recruitment success or new recruitment. By contrast, when F. sylvatica is not present, the probability of Q. petraea recruitment success is greater than the probability of failure. However when studying the effect of the presence of Q. petraea on F. sylvatica recruitment this pattern is not clear since the comparison of column proportions is not statistically significant (Pearson Chi-square at 0.05 level) (Fig. 5.6b). Finally, Figure 5.6c shows that in those plots with Q. petraea recruitment, there is a greater probability of recruitment failure than success in this species if successful recruitment of F. sylvatica is present (Pearson Chi-square statistic at the 0.05 level and column proportions comparison). It can also be seen that when F. sylvatica recruitment failure occurs, the probability of Q. petraea recruitment failure is greater than the probability of successful recruitment or new recruitment of this species.

As regards the way this pattern changes depending of the type of forest (Fig. 5.7), in pure stands of either of the two species, the probability of recruitment failure of the other species is greater than the probability of either new or successful recruitment of that species (Pearson Chi-square at 0.05 level and column proportions comparison). In all forest types, F. sylvatica shows a similar pattern, with more recruitment success than failure except in forests of other species where the probability of new recruitment is greater than the probability of recruitment failure and recruitment success (Fig.5.7a). However Q. petraea only displays greater recruitment success than failure in its pure stands (Fig. 5.7b).

Fig.5.7. Recruitment patterns by forest types in the second NFI cycle (T1; Nº plots = 2,581) for: a) F. sylvatica recruitment categories in each forest type. Number of plots with F. sylvatica recruitment = 2,336 plots. b) Q. petraea recruitment categories in each forest type. Number of plots with Q. petraea recruitment = 608 plots. Due to the co-occurrence in a single plot of both F. sylvatica and Q. petraea the total number of plots (2,336+608) is greater than the total plot number (2,581).

0

1

2

3

4

5

6

Pe

rce

nta

ge o

f p

lots

Forest type

Petraea success

Petraea failure

Petraea new

0

5

10

15

20

25

30

35

40

45

50

Pe

rce

nta

geo

f p

lots

Forest type

Fagus success

Fagus failure

Fagus new

a) b)


104

If we focus on the Q.petraea-F.sylvatica mixed plots, we find that there are no statistically significant differences in their recruitment success or failure (Pearson Chi-square at 0.05 level). This occurs because both have similar recruitment behavior. When one of them fails to recruit the other also fails (43 cases of recruitment failure for both species and 41 cases of recruitment success). However, where this situation does not occur, the number of plots in which F. sylvatica displays recruitment success is more than double the number of plots in which Q.petraea recruits successfully (52 compared to 19).

5.3.4. Biotic and abiotic factors affecting basal area increment, dominance and recruitment categories

Among the 21 biotic and abiotic predictors previously considered to explain basal area increment in F. sylvatica, the best GLM fits were obtained for 8 different predictors with which the basal area increment displayed significant relationships (Table 5.2). The final model accounted for 26.3% of the observed variability. As regards abiotic factors, altitude, longitude and latitude exhibited negative relationships with the basal area increment; altitude showing the strongest association. This indicates an association between F. sylvatica basal area increment and low altitudes, western longitudes and northern latitudes. Among the biotic factors, the dominance of F. sylvatica in the previous cycle presented a strong and positive association with basal area increment together with the current plot density. Conversely, the species dominance index of other tree species in the plot showed a negative relationship.

Table 5.2. GLM results for the basal area increment (BAI) of F. sylvatica between NFI cycles and biotic and abiotic predictors in the study area (significance codes: *** (p> 0.001); ** (p > 0.01), *(p>0.05)).

SdiBA_othersp_T1: species dominance index. Relative basal area of the rest of species of the plot (%) in the second NFI cycle. SdiBA_Fagus_T0: species dominance index. Relative basal area of F. sylvatica with respect to the rest of the species of that plot (%) in the first NFI cycle.N: density or number of trees per hectare.

F. sylvativa BAI Predictors

Effect D2

Intercept -0.8 - Forest types - 1,241.1*** Longitude -4.3E-06 1.7** Latitude -1.5E-01 8.4* Altitude -2.05E-03 376.6*** Summer rainfall 3.6E-03 32.1* SdiBA-othersp_T1 -0.4 174.6*** N 1.7E-03 629.9*** SdiBA_Fagus_T0 0.1 2232.6*** Deviance explained(D2) 26.25%


105

The SEM models identified mean annual temperature and orientation (exposition) as key environmental factor directly related to the dominance (SdiBA) of both F. sylvatica and Q. petraea (Table 5.3). This is in consonance with the GLM results since mean annual temperature and altitude are correlated variables. However, whilst temperature has the same negative effect on the dominance of both species, F. sylvatica dominance increases in north-facing locations whilst Q. petraea prefers south-facing sites.

SEM models for different recruitment classes revealed that F. sylvatica recruitment success is negatively affected by mean annual temperature (more success in colder places and therefore higher altitude) and that both failure and new recruitment are occurring in warmer, lowland plots. New recruitment of Q. petraea is only affected by orientation, with a preference for south-facing slopes.

As regards indirect effects, recruitment failure in F. sylvatica and Q. petraea displays different patterns, the former being positively affected by summer rainfall and negatively affected by total rainfall while the opposite relationship was found for Q.petraea.

The SEM models identified the dominance of F. sylvatica as a key negative factor for Q. petraea recruitment success and new recruitment. In contrast, Q. petraea dominance has no significant direct effect on F. sylvatica recruitment.

As regards recruitment failure, F. sylvatica shows an expected trend of greater failure in those plots where its dominance is lower. However, it is interesting that F. sylvatica recruitment failure is also greater when Q. petraea dominance is lower, indicating a shade preference for recruitment. Q. petraea, however, showed a different pattern of recruitment failure, the likelihood of failure being greater when F. sylvatica dominance is lower, although greater dominance of Q. petraea is also associated with higher recruitment failure in this species.

Finally, both longitude and latitude have a significant and positive indirect effect on F. sylvatica recruitment success, indicating more success eastwards (where the core of the species distribution is located) and northwards. In contrast, Q. petraea recruits successfully westwards and southwards.


106

Table 5.3. Results of the Bayesian SEM model in the second NFI cycle (T1) where the direct and indirect effects of different factors in the recruitment and dominance of the target species was analyzed. Standardized parameter estimates for direct and indirect effects are reported. Significant parameter estimates are in bold. An effect is considered significant when its 95% credibility interval does not overlap 0.

Standarized direct effects Standarized indirect effects

Latit

ude

Long

itude

Expo

sitio

n

Sum

mer

rain

fall

Tota

l rai

nfal

l

Mea

n an

nual

te

mpe

ratu

re

Sd

iAB

_F

ag

us

Sd

iAB

_P

etr

ae

a

Succ

ess_

Fagu

s

New

_Fag

us

Latit

ude

Long

itude

Expo

sitio

n

Sum

mer

rain

fall

Tota

l rai

nfal

l

Mea

n an

nual

te

mpe

ratu

re

Sd

iBA

_Fag

us

Sd

iBA

_Pet

raea

Succ

ess_

Fagu

s

New

_Fag

us

SdiBA_Fagus 0,13 0,35 0,13 0,02 -0,00 -0,35 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00 SdiBA_Petraea -0,07 -0,18 -0,07 -0,30 0,29 -0,05 -0,42 0,00 0,00 0,00 -0,05 -0,15 -0,05 -0,01 0,00 0,15 0,00 0,00 0,00 0,00 Success_Fagus -0,01 0,15 0,00 -0,05 0,02 -0,08 0,53 -0,04 0,00 0,00 0,07 0,20 0,08 0,02 -0,01 -0,19 0,01 0,00 0,00 0,00 New_Fagus -0,03 0,00 -0,05 -0,03 -0,04 0,14 0,00 -0,02 0,00 0,00 0,01 0,01 0,01 0,01 -0,00 -0,00 0,01 0,00 0,00 0,00 Failure_Fagus 0,08 -0,09 0,02 0,14 -0,12 0,06 -0,18 -0,15 0,00 0,00 -0,01 -0,01 -0,01 0,04 -0,04 0,05 0,06 0,00 0,00 0,00 Failure_Petraea 0,06 -0,09 -0,00 -0,04 -0,00 -0,06 -0,07 0,24 0,00 0,00 -0,04 -0,11 -0,04 -0,07 0,07 0,05 -0,10 0,00 0,00 0,00 Success_Petraea 0,01 0,01 -0,02 -0,08 0,06 0,01 -0,05 0,69 0,02 -0,01 -0,10 -0,24 -0,09 -0,21 0,20 0,07 -0,28 -0,01 0,00 0,00 New_Petraea 0,05 -0,03 -0,05 0,05 -0,06 -0,03 -0,10 0,00 -0,01 -0,06 -0,01 -0,04 -0,01 0,00 0,01 0,03 -0,01 0,01 0,00 0,00

SdiBA_Fagus: species dominance index. Relative basal area of F. sylvatica with respect to the rest of the species of that plot (%). SdiBA_Petraea: species dominance index. Relative basal area of Q. petraea with respect to the rest of the species of that plot (%).


107

5.4. Discussion

5.4.1. Fagus sylvatica expansion in North-Western Iberia

Our results show the vigorous status of F. sylvatica populations at the Southwestern limit of the species distribution, where the area of forests containing European beech has increased over recent decades as reported in previous studies (Sánchez de Medina 2005; FAO 2010; Sevilla Martínez 2011). The low mortality rates, high diversity and balanced dbh (age) structure along with the recruitment success patterns found for F. sylvatica in the studied region do not indicate a population decline as might be expected in a rear edge population according to the “centre-periphery hypothesis” (Hampe and Petit 2005). Moreover, the results show that the presence of F. sylvatica is increasing in the core of the studied area (Eastern and northern parts) and that new recruitment is occurring in the lowlands. They also reveal that the basal area increment of F. sylvatica increases westwards. Although changes in species abundance within their ranges should not be strictly classified as species range shifts per se, they can be considered as intermediate states in an ongoing shifting process (Maggini et al. 2011; Lenoir and Svenning 2015). In this sense, our results indicate the existence of F. sylvatica shift within the species distribution range.

Hence, according to Maggini et al. (2011) and Lenoir and Svenning (2015), the

range shift of F. sylvatica could be classified as “expand” (colonization and establishment events inside and outside the existing range in the studied area). This aspect, together with the population attributes and processes found at the studied edge of the species range regarding demographic stochasticity, increased abundance and population growth (Hampe and Petit 2005; Maggini et al. 2011), may reveal the Iberian Atlantic biogeographic region as one of the leading edges of F. sylvatica. In contrast, as further discussed in the next section, our results seem to suggest that this region might also be considered a rear edge of Q. petraea distribution. Moreover, other studies concerning the genetic structure of F. sylvatica and Q. petraea populations in the studied area (Olalde et al. 2002; Petit et al. 2002; Magri et al. 2006) appear to confirm this assertion, although further research is required.

The spatiotemporal pattern of recruitment is an important factor influencing

population dynamics of plant communities. Our results reveal new F. sylvatica recruitment, concomitant with basal area increment, occurring at low altitudes. This is contrary to what would be expected under climate change scenarios (Zhu et al. 2012). These findings are consistent with the results obtained by Rabasa et al. (2013), where F. sylvatica showed a greater abundance of juveniles than adults at lower altitudes throughout Europe.

On the other hand, the widespread abandonment of land, fire suppression and

reduced management has favoured forest recovery in many areas of Spain (Carnicer et al. 2014). Since F. sylvatica is highly dependent on ground disturbances for successful establishment (see for example Lindbladh et al. 2000), human disturbance might be a key factor explaining the expansion of F. sylvatica in suitable climatic areas, such as the lowlands of North-Western Iberia, with higher summer rainfall.


108

Thus, the expansion of F. sylvatica seems to be related to vegetation disequilibrium at the leading edge caused by post-glaciar migrational lags and forest recovery in suitable areas after land use abandonment in combination with recent climate change inducing vegetation shifts.

5.4.2. Biotic interactions of Fagus sylvatica and Quercus petraea

Our findings point to the existence of an inter-specific relationship between the two species which negatively affects Q. petraea. On the one hand, the results reveal that recruitment success in Q. petraea is limited by the presence of F. sylvatica. Furthermore, the two species show different population structures in mixed oak-beech forests plots, which present older Q. petraea individuals and younger F. sylvatica ones. This pattern indicates a recent colonization and replacement of Q. petraea forests by F. sylvatica in North-Western Iberia.

The replacement of oak

forests by beech forests has been reported in other European areas (see for example, Kunstler et al. 2004, 2005; Petrian et al. 2014) although this question has not yet been properly addressed in Spain. Nevertheless, young beech forests with presence of ageing Q. petraea individuals are relatively common in the studied area (Fig. 5.8).

Fig. 5.8: Example of ageing Q. petraea trees in young forests dominated by F.sylvatica located in Tarna

mountain pass (Asturias) (2011 by L. Hernández).

Furthermore, our results indicate that dominance and new recruitment of Q. petraea are associated with southern slopes. The Cantabrian range constitutes the Atlantic-Mediterranean border in Northern Iberia. Thus, the southern slopes of the study area receive greater Mediterranean influence, to which Q. petraea seems to be better adapted than F. sylvatica. Accordingly, although many studies have demonstrated the superior below- and above-ground competitive ability of F. sylvatica over Q. petraea in mature mixed forests (Leuschner et al. 2001; Hein and Dhôte 2006; Cavin et al. 2013) a number of previous studies have also shown that Q. petraea has a greater tolerance to drought than F. sylvatica (Aranda et al. 2005; Leuschner et al. 2001; Zapater et al. 2011; Cavin et al. 2013). Thus, mixed stands of sessile oak and beech can be expected to occur naturally on sites where drought and warmth restrict the competitive potential of beech to such an extent that oak can compete successfully (Pretzsch et al. 2013).


109

In this regard, the behaviour of Q. petraea in the core of its distribution area (central Europe) seems to differ from that observed in the study area. Although this species requires wetter conditions in its central and Northern European provenances, it is able to colonize areas with more Mediterranean conditions in Southern Europe (Costa et al. 1997; Bussotti et al. 2015). Furthermore, the ability of Q. petraea to hybridize with other more Mediterranean Quercus species (particularly Quercus pyrenaica) in the studied area should be taken into account. In oaks, hybridization is a key factor in genetic adaptation, allowing migration and adaptation (Petit et al. 2004, Sánchez de Dios et al. 2006; Alberto et al. 2010). Thus, the Q. petraea-Q. pyrenaica gene flow in the study area might constitute an adaptive advantage for the persistence of Q. petraea under a climate change scenario (Valbuena-Carabaña et al. 2005).

5.4.3. Global change and forest conservation

All predictive distribution models point to a reduction in F. sylvatica surface area in the Iberian Peninsula under climate change scenarios (Benito-Garzón et al. 2008; Meier et al. 2011; Saltre et al. 2015). Moreover, various local studies confirm the current retraction in the range of F. sylvatica forests in areas with more Mediterranean influence (Peñuelas and Boada 2003; Jump et al. 2006; Peñuelas et al. 2007, 2008; Piovesan et al. 2008). However, our findings reveal an increase in the distribution area of the species in the Spanish Atlantic region over the last half of the 20th Century and Urli et al. (2014) found no evidence of shifts in F. sylvatica in North-East and Central Spain over a 10 year period. This tendency has been also reported in central and northern Europe (see for example Bolte et al. 2007; Poljanec et al. 2010). At the same time, in other parts of Southeast Europe, an increase in F. sylvatica growing stock has also been reported in areas where extinction associated with climate change had been predicted (Tegel et al. 2014). Thus, although the range of F. sylvatica is indeed influenced by climate, it is not closely tracking the expected climatic changes. In this regard, several studies suggest that when exposed to climate change, the distributions of tree species would not change dramatically for a period of approximately one century (Svenning and Sandel 2013; García-Valdés et al. 2013, 2015; Lenoir and Svenning 2015). However, sooner or later tree distribution shifts will occur and future habitat loss must be taken into account in forest management and conservation plans (García-Valdés et al. 2015). Furthermore, climate warming may not only affect species distribution by altering abiotic conditions but also by changing the importance and intensity of their interaction with other species (Hughes 2000; Tylianakis et al. 2008; Lenoir et al. 2010, Rabasa et al. 2013). Hence, the effect of the predicted warming temperatures on Q. petraea-F.sylvatica interactions should be considered in future research. Our findings suggest that the competitive balance between sessile oak and European beech will be affected by climate change and will tip in favor of oak as water-stress levels increase (Aranda et al. 2005; Calvin et al. 2013; Pretzsch et al. 2013). Thus, these results suggest that the future expansion/contraction of F. sylvatica forests with ongoing climate change will be a key process which indirectly controls the demographic responses of Q. petraea in North-Western Iberia.


110

Finally, in recent decades, species distribution models (SDMs) have tended to assume equilibrium between vegetation and climate (see for example Araújo and Perarson 2005). Our results agree with those studies suggesting the importance of demographic and evolutionary dispersion mechanisms as important factors constraining species distributions and causing the lack of accuracy of SDMs’ (Beale and Lennon 2012; Fordham et al. 2012, 2013). This is especially important when predicting extinction risks under human induced global change (Saltre et al. 2015).

5.5. Conclusions

Contrary to what might be expected under current conditions of climatic change, the population of F. sylvatica in North-Western Iberia is expanding. Colonization is occurring in the lowlands and basal area increment is increasing westwards. Thus, our results indicate the existence of a shift in F. sylvatica within the specie’s distribution range, which may be classified as “expand”, following the Holocene inertia of the species and forest recovery in the area. As a result, the Iberian Atlantic biogeographic region can be considered one of the leading edges of F. sylvatica. In addition, the analysis of recruitment patterns and population structure of both F. sylvatica and Q. petraea confirmed the existence of a recent colonization and replacement of Q. petraea forests by F. sylvatica forests. However, it is possible that the population of F. sylvatica is not yet responding to climate change and that future habitat loss as well as changes in the importance and intensity of F. sylvatica-Q.petraea interaction should be expected in the near future. Our findings help to further our understanding of F. sylvatica population dynamics at the Southwestern limit of its distribution range, providing useful information for regional forest management and conservation strategies under current and future climatic conditions.

5.6. References

Alberto F, Niort J, Derory J, Lepais O, Vitalis R, Galop D, Kremer A (2010). Population differentiation of sessile oak at the altitudinal front of migration in the French Pyrenees. Mol. Ecol. 19 (13): 2626-2639

Aranda I, Gil L, Pardos JA (2005) Seasonal changes in apparent hydraulic conductance and their implications for water use of European beech (Fagus sylvatica L.) and sessile oak (Quercus petraea (Matt.) Liebl) in South Europe. Plant Ecol. 179: 155–167.

Araújo MB, Pearson RG (2005) Equilibrium of species’ distributions with climate. Ecography, 28: 693–695.

Arbuckle JL (2010) Amos (version 19.0). SPSS, Chicago, IL.

Beale CM, Lennon JJ (2012) Incorporating uncertainty in predictive species distribution modelling. Philos. T. Roy. Soc. B. 367: 247–258.

Benito-Garzón M, Sánchez de Dios R, Sainz Ollero H (2008) Effects of climate change on the distribution of Iberian forests. Appl. Veg. Sci. 11: 169–178.

Bolstad WM (2004) Introduction to Bayesian Statistics. Hoboken, NJ, John Wiley and Sons.


111

Bolte A, Czajkowski T, Kompa T (2007) The north-eastern distribution range of European beech — a review. Forestry 80 (4): 413-429.

Brewer S, Cheddadi R, de Beaulieu JL, Reille M (2002) The spread of deciduous Quercus throughout Europe since the last glacial period. Forest Ecol. Manag. 156: 27–48.

Bussotti F, Pollastrini M, Holland V, Brüggemann W (2015) Functional traits and adaptive capacity of European forests to climate change. Environ. Exp. Bot. 111: 91-113.

Carnicer J, Coll M, Pons X, Ninyerola M, Vayreda J, Peñuelas J (2014) Large-scale recruitment limitation in Mediterranean pines, the role of Quercus ilex and forest successional advance as key regional drivers. Glob. Ecol. Biogeog. 23: 371–384.

Cavin L, Mountford EP, Peterken GF, Jump AS (2013) Extreme drought alters competitive dominance within and between tree species in a mixed forest stand. Funct. Ecol. 27: 1424–1435.

Crawley MJ (1993) GLM for Ecologists. Blackwell Scientific, Oxford.

Ceballos L (1966) Mapa forestal de España 1,400,000. Ministerio de Agricultura. Madrid.

Costa M; Morla C, Sainz H (Eds.) (1997) Los Bosques ibéricos. Planeta. Barcelona. Day RW, Quinn GP (1989) Comparisons of treatments after an analysis of variance in ecology. Ecol. Monogr. 59 (4): 433-463. European Environment Agency boundaries (2011) European biogeographic map. http,//www.eea.europa.eu/data-andmaps/data/biogeographical-regions-europe

EUFORGEN (2009) Distribution map of European beech (Fagus sylvatica). www.euforgen.org.

Ellenberg H (1996) Vegetation Mitteleuropas mit den Alpen, 5th edn. Ulmer, Stuttgart, Germany, pp. 1095

Fang J, Lechowicz MJ (2006) Climatic limits for the present distribution of beech (Fagus L.) species in the world. J. Biogeogr. 33: 1804 – 1819.

Fordham DA, Akcakaya RH, Araújo MB et al. (2012) Plant extinction risk under climate change, are forecast range shifts alone a good indicator of species vulnerability to global warming? Glob. Change Biol. 18: 1357–1371.

Fordham DA, Mellin C, Russell M et al. (2013) Population dynamics can be more important than physiological limits for determining range shifts under climate change. Glob. Change Biol. 19: 3224–3237.

FAO (2010) Evaluación de los recursos forestales mundiales 2010. Informe nacional, España. FRA2010/196. Rome (Italy)

http://www.eea.europa.eu/data-andmaps/data/biogeographical-regions-europe

http://www.euforgen.org/


112

García-Valdés R, Zavala MA, Araújo MB, Purves DW (2013) Chasing a moving target, projecting climate change-induced shifts in nonequilibrial tree species distributions. J. Ecol. 101: 441–453.

García-Valdés R, Svenning JC, Zavala MA, Purves DW, Araújo MB (2015) Evaluating the combined effects of climate and land-use change on tree species distributions. J. Appl. Ecol. 52 (4): 902-912.

Gelman A, Carlin JB, Stern HS, Rubin DB (2004) Bayesian Data Analysis. 2nd ed. Boca Raton, Chapman and Hall/CRC.

Giesecke T, Hickler T, Kunkel T, Sykes MT, Bradshaw RHW (2007) Towards an understanding of the Holocene distribution of Fagus sylvatica L. J. Biogeogr. 34: 118-131.

Green DG (1987) Pollen evidence for the postglacial origins of Nova Scotia’s forests. Can. J. Botany. 65: 1163–1179.

Hampe A, Petit R (2005) Conserving biodiversity under climate change, the rear edge matters. Ecol. Lett. 8 (5): 461-467.

Hein S, Dhôte JF (2006) Effect of species composition, stand density and site index on the basal area increment of oak trees (Quercus sp.) in mixed stands with beech (Fagus sylvatica L.) in northern France. Ann. For. Sci. 63: 457–46.

Hosmer D, Lemeshow S (1989) Applied Logistic Regression. New York, John Wiley and Sons.

Hughes L (2000) Biological consequences of global warming, is the signal already apparent? Trends Ecol. Evol. 15: 56-61.

Jump AS, Hunt JM, Peñuelas J (2006) Rapid climate change-related growth decline at the southern range edge of Fagus sylvatica. Glob. Change Biol.12: 2163-2174.

Kunstler G, Curt T, Lepart J (2004) Spatial pattern of beech (Fagus sylvatica L.) and oak (Quercus pubescens Mill.) seedlings in natural pine (Pinus sylvestris L.) woodlands. Eur. J. For. Res. 123: 331–337

Kunstler G, Curt T, Bouchaud M, Lepart J (2005) Growth, mortality, and morphological response of European beech and downy oak along a light gradient in sub-Mediterranean forest. Can. J. For. Res. 35: 1657-1668

Lenoir J, Gégout JC, Guisan A, Vittoz P, Wohlgemuth T, Zimmermann NE, Dullinger S, Pauli H, Willner W, Svenning JC (2010) Going against the flow, potential mechanisms for unexpected downslope range shifts in a warming climate. Ecography 33: 295-303

Lenoir J, Svenning JC (2015) Climate-related range shifts – a global multidimensional synthesis. Ecography 38 (1): 15–28.

Leuschner C, Hertel D, Coners H, Büttner V (2001) Root competition between beech and oak, a hypothesis. Oecologia, 126: 276–284.


113

Lindbladh M, Bradshaw R, Holmquist B (2000) Pattern and process in south Swedish forests during the last 3000 years at stand and regional scales. J. Ecol. 88: 113–128

López-Merino L, López-Sáez JA, Ruiz Zapata MB, Gil García MJ (2008) Reconstructing the history of beech (Fagus sylvatica L.) in the North-Western Iberian Range (Spain), From Late-Glacial refugia to the Holocene anthropic-induced forests. Rev. Palaeobot. Palyno. 152: 58–65

Maggini R, Lehmann A, Kéry M, Schmid H, Beniston M, Jenni L, Zbinden N (2011) Are Swiss birds tracking climate change? Detecting elevational shifts using response curve shapes. Ecol. Model. 222: 21 – 32.

Magri D, Vendramin GG, Comps B, et al. (2006) A new scenario for the Quaternary history of European beech populations, palaeobotanical evidence and genetic consequences. New Phytol. 171: 199–221.

Magri D (2008) Patterns of post-glacial spread and the extent of glacial refugia of European beech (Fagus sylvatica). J. Biogeogr. 35: 450-463

M.A.P.A(1974–1990) Caracterizaciones agroclimáticas de las provincias españolas. Ministerio de Agricultura, Pesca y Alimentación, Madrid

MFE50 (2001) Mapa forestal de España. B. D. D. de la Naturaleza. Escala, 1:50,000.

Meier ES, Edwards TC, Kienast F, Dobbertin M, Zimmermann NE (2011) Co-occurrence patterns of trees along macro-climatic gradients and their potential influence on the present and future distribution of Fagus sylvatica L. J. Biogeogr. 38: 371–382

Muñoz-Sobrino C, Ramil-Rego P, Gómez-Orellana L, Ferreiro da Costa J, Díaz Varela RA (2009) Climatic and human effects on the post-glacial dynamics of Fagus sylvatica L. in NW Iberia. Plant Ecol. 203: 317–340

Olalde M, Herrán A, Espinel S, Goicoechea S (2002) White oaks phylogeography in the Iberian Peninsula. Forest Ecol. Manag. 156: 89–102

Peñuelas, J., Boada, M., 2003. A global change-induced biome shift in the Montseny Mountains (NE Spain). Glob. Change Biol. 9, 131–140.

Peñuelas J, Ogaya R, Boada M, Jump AS (2007) Migration, invasion and decline, changes in recruitment and forest structure in a warming-linked shift of European beech forest in Catalonia (NE Spain). Ecography 30: 830–838.

Peñuelas J, Hunt JM, Ogaya R, Boada M, Jump AS (2008) Twentieth century changes of tree-ring δ13C at the southern range-edge of Fagus sylvatica, increasing water-use efficiency does not avoid the growth decline induced by warming at low altitudes. Glob. Change Biol. 14: 1076–1088

Petit RJ, Brewer S, Bordács S, et al (2002) Identification of refugia and postglacial colonization routes of European white oaks based on chloroplast DNA and fossil pollen evidence. Forest Ecol. Manag. 156: 49-74.


114

Petit RJ, Bialozyta R, Garnier-Géré P, Hampe A (2004) Ecology and genetics of tree invasions, from recent introductions to Quaternary migrations. Forest Ecol. Manag. 197: 117–137.

Petritan I, Marzano R, Petritan AM, Lingua E (2014) Overstory succession in a mixed Quercus petraea–Fagus sylvatica old growth forest revealed through the spatial pattern of competition and mortality. Forest Ecol. Manag. 326: 9-17.

Piovesan G, Biondi F, Di Filippo A, Alessandrini A, Maugeri M (2008) Drought-driven growth reduction in old beech (Fagus sylvatica L.) forests of the central Apennines. Italy. Glob. Change Biol. 14: 1265–1281.

Poljanec A, Ficko A, Boncina A (2010) Spatiotemporal dynamic of European beech (Fagus sylvatica L.) in Slovenia, 1970–2005. Forest Ecol. Manag. 259 (11): 2183-2190.

Pretzsch H, Bielak K, Block J, et al. (2013) Productivity of mixed versus pure stands of oak (Quercus petraea (Matt.) Liebl. and Quercus robur L.) and European beech (Fagus sylvatica L.) along an ecological gradient. Eur. J. For. Res. 132: 263–280

Rabasa S, Granda E, Benavides E, et al. (2013) Disparity in elevational shifts of European trees in response to recent climate warming. Glob. Change Biol. 19: 2490–2499

Ramil-Rego P, Rodríguez Guitián MA, Muñoz Sobrino C, Gomez-Orellana L (2000) Some considerations about the postglacial history and recent distribution of Fagus sylvatica in the NW Iberian Peninsula. Folia Geobot. 35:241–271.

Rodriguez-Guitán MA (2004) Aplicacion de criterios botánicos para a proposta de modelos de xestion sustentable das masas arborizadas autóctonas do subsector Galaico-Asturiano Septentrional. Ph.D. thesis, Universidade de Santiago de Compostela, Santiago de Compostela, Spain.

Rozas V, Camarero JJ, Sangüesa-Barreda G, Souto M, García-González I (2015) Summer drought and ENSO-related cloudiness distinctly drive Fagus sylvatica growth near the species rear-edge in northern Spain. Agr. Forest Meteorol. 201: 153-164.

Rubiales JM, García Amorena I, García Álvarez S, Gómez Manzaneque F (2008) The Late Holocene extinction of Pinus sylvestris in the west of the Cantabrian Range. J. Biogeogr. 35: 1840-1850.

Ruiz de la Torre (Dir) (1986-2002) Mapa forestal de España 1,200.000. Ministerio de Agricultura Pesca y Alimentación, Madrid.

Saltre F, St-Amant R, Gritti ES, Brewer S, Gaucherel C, Davis B.A.S and Chuine I (2013) Climate or migration, what limited European beech post-glacial colonization? Global Ecol. Biogeogr. 22: 1217–1227.

Saltre F, Dutié A, Gaucherel C, Chuine I (2015) How climate, migration ability and habitat fragmentation affect the projected future distribution of European beech. Glob. Change Biol. 21: 897–910.


115

Sánchez de Dios R, Benito Garzón M, Sainz Ollero H (2006) Hybrid zones between two European oaks, a plant community approach. Plant Ecol. 187:109–125

Sánchez de Medina A (2005) Diagnóstico de los hayedos burgaleses e identificación de tipologías estructurales para la asignación de tratamientos selvícolas. PhD thesis. Escuela técnica superior de ingenieros de montes. UPM, Madrid. Spain

Sevilla-Martínez F (2011) Evolución temporal del régimen de renovaciones en la montaña Cantábrica. In, Ezquerra, F.J. y Rey, E. (Coord.). La evolución del paisaje vegetal y el uso del fuego en la cordillera Cantábrica. Pp. 32-46. Fundación Patrimonio Natural de Castilla y León, Valladolid. Spain

Svenning JC, Skov F (2004) Limited filling of the potential range in European tree species. Ecol. Lett. 7: 565–573

Svenning JC, Normand S, Skov F (2008) Postglacial dispersal limitation of widespread forest plant species in nemoral Europe. Ecography 31: 316326

Svenning JC, Sandel B (2013) Disequilibrium vegetation dynamics under future climate change. Am. J. Bot. 100(7): 1266 - 1286

Tegel W, Seim A, Hakelberg D, Hoffmann S, Panev M, Westphal T, Büntgen U (2014) A recent growth increase of European beech (Fagus sylvatica L.) at its Mediterranean distribution limit contradicts drought stress. Eur. J. For. Res. 133: 61-71

Tinner W, Lotter AF (2006) Holocene expansions of Fagus silvatica and Abies alba in Central Europe, Where are we after eight decades of debate? Quaternary Sci. Rev. 25: 526 – 549

Tylianakis JM; Didham RK; Bascompte J,Wardle DA (2008) Global change and species interactions in terrestrial ecosystems. Ecol. Lett. 11(12), 1351-1363

Urli M, Delzon S, Eyermann A, Couallier V, García-Valdés R, Zavala MA, Port AJ (2014) Inferring shifts in tree species distribution using asymmetric distribution curves, a case study in the Iberian mountains. J. Veg. Sci. 25: 147–159

Valbuena-Carabaña M, Gonzalez-Martinez SC, Sork VL, Collada C, Soto A, Goicoechea PG, Gil L (2005) Gene flow and hybridisation in a mixed oak forest (Quercus pyrenaica Willd. and Quercus petraea (Matts.) Liebl.) in central Spain. Heredity 95: 457–485

Zapater M, Hossann C, Bréda N, Bréchet C, Bonal D, Granier A (2011). Evidence of hydraulic lift in a young beech and oak mixed forest using 18O soil water labelling. Trees 25: 885–894

Zhu K, Woodall CW, Clark JS (2012) Failure to migrate, lack of tree range expansion in response to climate change. Glob. Change Biol. 18: 1042–1052.

6. Assessing spatio-temporal rates, patterns and determinants of biological

invasions in forest ecosystems. The case of Acacia species in NW Spain.


Hernández L, Martínez-Fernández J, Cañellas I, Vázquez de la Cueva A (2014) Assessing spatio-temporal rates, patterns and determinants of biological invasions in

forest ecosystems. The case of Acacia species in NW Spain. Forest Ecology and Management. DOI:10.1016/j.foreco.2014.05.058

CAPÍTULO 6: SPATIO-TEMPORAL RATES, PATTERNS AND DETERMINANTS OF BIOLOGICAL INVASIONS

119

Resumen

Actualmente la invasión de flora exótica invasora se está convirtiendo en uno de los mayores desafíos ambientales por su consecuencia en la flora nativa y en la pérdida de Biodiversidad. El conocimiento de su evolución y los factores asociados a su expansión es el primer paso para el desarrollo de medidas eficaces de control. Este trabajo sugiere la utilidad de la precisa información espacio-temporal proveniente de los Inventarios Forestales Nacionales (IFN) para estimar los ratios demográficos, tendencias espacio-temporales y los determinantes de la expansión de especies invasoras en ecosistemas forestales. Con este objetivo se seleccionaron dos de las especies invasoras más extendidas en Europa: Acacia dealbata y Acacia melanoxylon. Enmarcado en los sistemas forestales del noroeste peninsular y en base a la comparación de los dos últimos ciclos de IFN español, este trabajo analiza la dinámica de las dos especies de Acacia entre 1998 y 2008 a través de cambios en su distribución, dominancia, estructura diamétrica y regeneración. Para determinar que tipos de hábitat forestal son más susceptibles a la invasión, la superficie forestal en estudio fue clasificada como diferentes tipos de bosque. Finalmente, en base a modelos lineales generalizados, se analiza la importancia relativa de factores bióticos y abióticos en la expansión de las dos especies de Acacia en los bosques de la zona de estudio en la última década. Los resultados confirman la rápida expansión de las especies de Acacia en los bosques del noroeste peninsular, con ratios de expansión anual de aproximadamente un 0,1%. Las dos especies están aumentando su presencia en la mayor parte de los bosques del área de estudio, empezando a dominar en el estrato de regeneración de algunos de ellos. Factores ambientales y la conectividad entre poblaciones de Acacia son los determinantes más importantes asociados con la expansión de estas especies en nuevas áreas. Además, la combinación de procesos perturbadores y factores bióticos asociados con la estructura de la masa (área basal total, riqueza de especies y cobertura arbórea) parecen estar asociados con la vulnerabilidad o resistencia de algunos tipos de bosque a su expansión. La etapa temprana de invasión detectada en las dos especies de Acacia en la zona de estudio pone en evidencia la potencialidad de ambas especies para continuar con su expansión. La combinación de este factor y el elevado régimen de perturbaciones (principalmente incendios) en esta región, podría ser crítico en la la configuración futura de los paisajes forestales en el noroeste peninsular. Este estudio demuestra la utilidad de considerar la información detallada proveniente de los IFN para estudiar la evolución de las invasiones biológicas en ecosistemas forestales. Los datos espacialmente explícitos obtenidos de estas bases de datos pueden contribuir no sólo a fomentar nuestros conocimientos en materia de invasiones biológicas, sino también para desarrollar estrategias de conservación y manejo más eficientes.

Palabras clave: Acacia dealbata, Acacia melanoxylon, ratios de expansión, dinámica de especies invasoras, invasibilidad, factores de invasión, Inventario Forestal Nacional.


120

Abstract

Invasive species currently pose a major environmental challenge. Understanding their development and the factors associated with their expansion is the first step towards developing effective control measures. This work proposes the use of detailed spatio-temporal information from forest monitoring systems to assess the demographic rates, spatio-temporal patterns and spread determinants of invasive plants in forest ecosystems. For this purpose, we selected two of the most widespread non-native plants in Europe: Acacia dealbata and Acacia melanoxylon. Focusing on the forested area of northwest Spain and based on the comparison of two cycles of the Spanish National Forest Inventory, this study analyses the dynamics of Acacia species between 1998 and 2008 in regards to changes in their spatial distribution, dominance, abundance, diametrical (dbh) structure and regeneration. In addition, the forested area was classified into forest types to identify the forests which are more susceptible to invasion. Finally, through general linear models, this study analyses the relative importance of abiotic and biotic factors determining the spread of Acacia species over the studied period. The results confirm a rapid expansion in the presence of Acacia species in the forests of NW Spain, with annual spread rates around 0.1%. These two species are increasing their dominance across most forest types in the study area, where they are becoming the dominant species in the regeneration of some of them. Environmental factors and connectivity between Acacia populations are identified as the main factors associated with their spread into new areas. Additionally, the combination of disturbances and biotic factors associated with stand structure (total basal area, richness and tree cover) appear to determine the vulnerability or resistance of some forest to their spread. The early stage of invasion detected highlights the potential of Acacia species to continue spreading. This aspect, in conjunction with the high degree of disturbances (mainly fires) in this region, could be critical in determining the configuration of future forest landscapes in NW Spain. This study demonstrates the value of considering broad-scale periodic forest surveys to monitor biological invasions in forests ecosystems. The spatially-explicit data obtained from these surveys can contribute not only to furthering our knowledge with regard to invasion biology but also to developing more efficient conservation and management strategies.

Key words: Acacia dealbata, Acacia melanoxylon, expansion rates, invasive species dynamics, invasibility, spread factors, National Forest Inventories.


121

6.1. Introduction Invasive alien species pose one of the most important direct threats to the structure and function of ecosystem diversity. The spread of invasive alien species is therefore among the most urgent nature conservation issues to be faced at global scale (CBD 2010, UNEP 2010), the identification of alien spread pathways to prevent their introduction and establishment being one of the main targets of the EU 2020 Strategic Plan for Biodiversity (EC 2011). The considerable research into invasion ecology in recent decades has centred around three main questions: invasiveness (Rejmánek and Richardson 1996), invasibility (Chytrý et al. 2008) and impacts (Hejda et al. 2009). As a result, our theoretical and practical knowledge of plant invasions has improved substantially (Richardson 2011), although the capability to address further challenges in this line of research may be hindered by the lack of availability, detail and heterogeneity of information concerning invaders (Pyšek et al. 2002). One of these new challenges is to provide new insights into invasion dynamics (Richardson et al. 2010). Although access to detailed lists and maps of non-native species has improved at global level in recent times (Foxcroft et al. 2010), there is a lack of broad scale periodic surveys providing the possibility to identify detailed demographic rates, spatio-temporal patterns and determinants of invasive plant spread. Statistically-designed inventories such as National Forest Inventories (NFI) based on periodic re-measurements of permanent sample units constitute a valuable tool for monitoring forest dynamics (Lund et al. 1998). Consequently, the inclusion of non-native species data in these surveys, based on biodiversity monitoring programs, (Corona et al. 2011) provides a valuable opportunity not only to examine the broad-scale evolution of plant invasions in forest ecosystems but also to test ecological hypotheses in invasion biology. For this study we selected two different Acacia species in an attempt to determine general patterns in forest dynamics and invasion ecology over the last decade in the northwest of Spain, one of the most affected regions of the country (Romero Buján 2007). Acacia species are among the most widespread invasive plants in Europe, and two of the most aggressive are A. dealbata Link and A. melanoxylon R. Br. in W.T. Aiton. Today, they are widely naturalized and have become an environmental problem in Southwestern Europe (Carballeira and Reignosa 1999; Hussein et al. 2011) where they pose a threat to native species and have been declared “invaders” (Sanz-Elorza et al. 2004). The invasive success of Acacia is mainly attributed to its rapid growth rate, prolific production of seeds with high longevity, germination stimulated by fire, allelopathic effects and the absence of natural enemies (Marchante et al. 2003). In the Iberian Peninsula, these two species have not yet reached their potential distribution range (Gassó et al. 2012), although it is likely that they will be able to reach this potential range in the near future. Identifying the determinants of invasion in the early stages is crucial to the development of realistic predictive models of invasion risk (Kolar and Logde 2001) and to mitigate the potential ecological impact.


122

This study constitutes a first attempt at using national forests monitoring data to study the evolution of biological invasions. Based on the analysis of the sequential broad-scale databases from two cycles of the Spanish NFI, the primary aim of this study is to examine the spatio-temporal changes in the distribution, abundance and dominance of A. melanoxylon and A. dealbata in the forests of NW Spain over the period 1998-2008. The second objective is to identify the types of forest which are most vulnerable to invasion and the level of invasion reached. Finally, an attempt is made to disentangle the relative importance of the biotic and abiotic factors underlying the spread of these two species in different forest types. For this purpose we test two hypotheses: the first is that Acacia species have expanded and increased their dominance in the forests of NW Spain between 1998 and 2008. The second is that the expansion rates and dominance of these species differ from one forest type to another. 6.2. Material and Methods

6.2.1. Study area

This study is based on NFI information for the provinces which comprise the region of Galicia in the northwest of the Iberian Peninsula (Fig.6.1). Due to a combination of bioclimatic and human factors, the percentage of non-native flora in this area of NW Spain (14%) (Romero Buján 2007) is higher than the for the Iberian Peninsula as a whole (12%) (Sanz Elorza et al. 2004). This region presents a climatic gradient from the coast towards inland areas, but there is a dominant humid Atlantic climate with mild temperatures (mean annual temperature of 13 ºC) and abundant precipitation (mean annual rainfall of 1,400 mm). Soils are acidic and the area exhibits a complex topography, with altitudes ranging from sea level up to 2,124 m.

Fig. 6.1. A) Known distribution area of Acacia melanoxylon and Acacia dealbata in Europe. B) Known distribution area of Acacia species in NW Spain. C) Distribution of the different forest types in NW Spain.


123

Table 6.1. Characteristics of the different forest types in NW Spain and the studied Acacia species.

Total number of plots/Forested area (%) (2008)

Mea annual precipitation

(mm)

Mean annual temperature

(ºC)

Mean elevation

(m)

Mean daily temperature oscillation

(ºC)

Distance to the

coastline (km)

Plots by forest type

MARPIN Maritime pine forests (Pinus pinaster) 1,395/16.2 1,447.4+278.5 12.8+1.1 360.3+229.7 12.0+1.9 38.3+36.1

MIXB Mixed broadleaf forests 1,180/13.7 1,381.3+223.4 11.8+1.2 471.6+376.7 12.4+1.4 47.2+26.5

EUC Non-site native species plantations of Eucalyptus spp. 2,106/24.5 1,519.5+258.3 12.7+0.9 253.6+343.8 10.4+1.1 12.7+11.8

FLOOD Floodplain/Riparian forests 321/3.7 1,361.1+252.7 12.6+1.2 276.2+845.4 12.2+1.9 42.0+35.6 MIXCB Mixed coniferous broadleaf forests 2,177/25.3 1,403.2+284.4 12.4+1.3 419.0+258.1 12.1+1.9 42.7+35.4 SHR Shrubland 245/2.9 1,438.2+229.6 12.0+1.3 470.8+258.8 11.7+1.8 36.0+31.6

RADPIN Non-native species plantations of Pinus radiata 483/5.6 1,356.1+218.5 11.5+0.8 496.7+169.6 11.7+1.4 41.9+23.7

SYLPIN Plantations of Pinus sylvestris 241/2.8 1,415.6+204.5 9.6+1.1 1,060.4+250.0 13.9+1.5 91.3+33.7 PYROAK Oak forests of Quercus pyrenaica 368/4.3 1,185.8+212.3 11.3+1.3 723.4+261.3 14.3+0.9 91.7+26.8 Plots with Acacia spp. presence Acacia melanoxylon Acacia dealbata

262/- 1,565. 8+275.1 13.3+0.9 237.2+143.8 10.9+1.2 10.3+11.6 199/- 1,347. 7+264.2 13.4+0.9 277.3+155.1 12.9+1.4 42.0+28.5


124

Today, almost 50% of the forests in NW Spain comprise plantations of non-site-native species such as Pinus pinaster Ait., Eucalyptus spp., Pinus sylvestris L. and Pinus radiata D. Don (Table 6.1).The native forest types present in the area are floodplain/riparian forests, scattered coastline forests of Pinus pinaster, atlantic mixed broad-leaved forest where Quercus robur L. is abundant and oak forests of Quercus pyrenaica Willd. in the transition zone between the Atlantic and Mediterranean biogeoregions (Fig.6.1C). Acacia species are Australian N2-fixing trees that were introduced into Europe as an ornamental species in the 19th century (Sheppard et al. 2006). In Spain, they are mainly distributed in the most NW territories where A. melanoxylon occurs close to the coastline in temperate locations while A. dealbata, with a broader ecological valence, can be found in more continental areas (Fig.6.1B, Table 6.1). 6.2.2. Data used The study is based on spatially detailed information from two consecutives cycles of the Spanish NFI performed in NW Spain in 1998 (NFI3) and in 2008 (NFI4), a time interval of 10 years (13,159 plots). In these Spanish NFI cycles, permanent plots were established systematically in the forested area at the intersections of a 1km×1km grid. Field plots consist of four concentric circular areas with radii of 5, 10, 15 and 25 m. Depending on the dbh (diameter at breast height) of the tree species, different dendrometric characteristics are measured within each plot and for each radius such as dbh or height of trees with dbh > 7.5 cm and height > 1.30 m. Furthermore, other forest attributes and conditions are measured (tree and shrub species composition, density, covers, recruitment, saplings (trees with 2.5 cm < dbh <7.5 cm), silvicultural treatments (clear-cutting, groundwork and crown treatments), etc.). A total of 20 predictors were considered as independent variables to analyze the relative importance of abiotic and biotic factors determining the spread of Acacia species over the period considered (Table 6.2). Some of these variables were also used to characterize the climatic and physical ranges of the two Acacia species and the different forest types, as well as their disturbance level (Fig. 6.2). As regards the abiotic factors; topographical variables (altitude, aspect, exposure and closest distance to sea) were taken from the digital elevation model of Spain with a spatial resolution of 25 m (U.T.M, ED 50). Climatic variables were extracted from Gonzalo (2010). The impact of human disturbances on the spread of Acacias over the time frame considered were discerned using variables such as silviculture treatments obtained from NFI plot databases, the forest-urban and forest-crop interface calculated from land use maps (Heymann et al. 1994; EEA 2012) and the fire incidence from MODIS burned area products (Boschetti 2009). As for biotic factors, we considered several attributes of forest structure derived from NFI datasets such as species richness, tree cover and basal area at plot level. Furthermore, to analyse the importance of propagule pressure or distance from invasion loci, we consider the connectivity between plots containing Acacia species.


125

Table 6.2. Variables used for studying current distribution range, evolution and spread of the studied Acacia species in NW Iberian Peninsula over the period 1998-2008. Variable Type Description

Presence/Absence (1/0) Dep/Nom Proportion of Acacia species

ΔG, ΔBA, ΔN, ΔS Dep/Cont Differences in the growing stock (m3/ha), basal area (m2/ha), number of trees per ha and area

NSap; NSaprel Dep/Cont Total and relative number of saplings per hectare per plot

COL Dep/Nom Proportion of plots by forest type with new Acacia spp. presence

DIS Dep/Nom Proportion of plots by forest type with Acacia spp. gone

IVI Dep/Cont Particularized Importance Index value

Spr_Amel;Spr_Adeal (1/0) Dep/Nom Acacias species spread

Forest types Ind/Nom Main forest types of NW Spain

Topographical variables Elev Ind/Cont Elevation in m a.s.l

Aspect Ind/Ord Classes of aspect (NE;E;SE;S;SW;W;NW,N)

Slope Ind/Cont Values in degrees from 0 to 90.

Dsea Ind/Cont Closest distance from the field plot to the coast line (km)

Climatic variables Ptot Ind/Cont Mean annual precipitation (mm)

Tm Ind/Cont Mean annual temperature(◦C)


126

Table 6.2. Variables used for studying current distribution range, evolution and spread of the studied Acacia species in NW Iberian Peninsula over the period 1998-2008. Variable Type Description

Osc Ind/Cont Daily average of the thermal oscillation

Other abiotic variables

Clearcut Ind/Dich Clear cutting

SitePrep Ind/Dich Ground preparation treatments

CrownTreat Ind/Dich Crown-release treatments

FUint Ind/Cont Length (m) of the forest-urban interface in a buffer of 450 m around the NFI plot. Source: Corine Land Cover 2000 raster 100 meters (EEA 2012)

FCInt Ind/Cont Length (m) of the forest crop interface in a buffer of 450 m around the NFI plot. Source: Corine Land Cover 2000 raster 100 meters (EEA 2012)

Fire Ind/Cont Burned pixels detected in the NFI plots by MODIS in the period 2000-2007 (Boschetti et al. 2009)

Biotic variables BA Ind/Cont Total basal area in the plot (m2/ha)

Forest cover Ind/Cont Total forest cover in the plot as the sum of shrub and tree covers (%)

Treecover Ind/Cont Tree cover in percentage of the total forest cover (%)

TreeRichness Ind/Cont Total number of trees identified in the NFI plot

ShrubRichness Ind/Cont Total number of shrubs identified in the NFI plot

Trichness Ind/Cont Total number of species as the sum of tree and shrub richness

Connect Ind/Cont Connectivity between plots with Acacia species (m) as the mean value of the distance of a centroid of a plot to the centroid of the nearest five plots with Acacia species.


127

6.2.3. Data analysis Spanish NFI records of the presence of A. melanoxylon and A. dealbata (Fig. 6.1B), along with other cartographical sources such as botanical Atlases (Sanz Elorza 2004, Anthos 2012), provided valuable information for mapping the current spatial distribution of the two Acacia species in the study area (Fig.6.2A). Based on NFI plot information such as species dominance and forest management, the forested area of NW Spain was then classified into nine forest types (see Fig. 6.2 and Table 6.2 for main characteristics and abbreviations) according to the definitions proposed for Europe (Barbati et al. 2007).

Fig. 6.2. Description of the forest types of NW Spain. Left: Richness: mean and standard deviation of tree, shrub and total richness per plot; Right: Disturbance regime: percentage of total plots with silvicultural treatments and fire incidence recorded in the period 2000-2008. The comparison of the information derived from the plots in which Acacia species were present in the two sequential NFI allowed the total spatial expansion, density and growing stock rates of Acacia species to be assessed (114 plots). To analyze the stage of invasion in the plots during the time frame considered, changes in Acacia species abundance by dbh class were also calculated. It must be considered that due to the dynamic nature of this species and the frequency of disturbances in the forests where these two species are found, the number of remeasured plots used in these analyses was limited and was lower than the proportion used when studying the distribution or spread derived from the presence/absence indicator. The vulnerability to the invasion in the different forest types was analyzed through the particularized Important Value Index (IVI) from Curtis (1959) as the sum of the relative density and dominance of Acacia species in the plots by forest types (114 plots). Relative density was defined as the numerical strength of a species in relation to the total number of individuals of all the species. It was calculated as: [(Number of individual of the species/Number of individual of all the species) x 100]. Relative dominance was determined by the value of the total basal area of a species with respect to the sum of basal area of the rest of the species in the plot and it was calculated as: [(Total basal area of the species/Total basal area of all the species) x 100]. The IVI was used to determine the overall importance of Acacia species in the plots, providing a good indicator of invasive species dominance. In these approaches, only the forest types in which there was a significant number of plots with presence of the two Acacia species were considered (n>5). Since there were different sample sizes and non-homogeneity of the variance, the mean differences in Acacia IVI by forest type

02468

101214

Mea

n ric

hnes

s pe

r plo

t

Forest types

Mean tree richness Mean shrubs richness

0%10%20%30%40%50%60%70%

Perc

enta

ge o

f tot

al a

rea

Forest type

Forest treatments


128

were assessed through the Welch Test. Tamhane´s T2 test was then used for post-hoc multiple comparisons of mean values between forest types (Hollander and Wolfe 1999). Two dichotomous dependent variables (Col, colonization, and Dis, disappearance) (757 plots) were then created to analyze changes in spatial distribution and the level of invasion of the two Acacia species in the different forest types in NW Spain during the period (1998-2008). The current invasibility of the different forest types was partly analysed through the total number and proportion of saplings of the two Acacia species in the regeneration of each forest type according to the last NFI (2008) (92 plots). The proportion of the species in the regeneration defines the numerical strength of that species in relation to the total number of individuals of all the species, indicating conspecific abundance or dominance. The results were then analysed through box-and-whisker plots to show the overall patterns and range of the current regeneration of the two Acacia species in the different forest types they have invaded. Finally, a dichotomous dependent variable Spr was created to assess the effect of different biotic and abiotic factors on the spread of the Acacia species in the forests of NW Spain over the period 1998-2008 through general linear model (GLM) analysis with binomial error and logit link. This was analyzed in the plots in which Acacia species were not present in 1998 but were present in 2008 (251 plots). The logistic regression models provide information on the relationships and strengths among dependent and independent variables. Furthermore, logistic regression accepts a combination of continuous and categorical variables as well as not normally distributed ones (Hosmer and Lemeshow 1989). Multicollinearity was verified using Pearson correlation coefficients. Some of the explanatory variables that were highly correlated (| r | > 0.8) were excluded prior to building the model. We followed a step-by-step model-building procedure and the fit of the model was tested after the elimination of each variable. Deviance reduction, estimated as: D2 = (null deviance -residual deviance)/null, was used as the measure of discrepancy to assess the goodness-of-fit of the model (Crawley 1993).

ArcGis 9.5 (ESRI 2006) was used as an image analysis tool as well as to extract topographical, climatic, land use and fire variables to the 1km x 1km grid. R software was used to fit the GLM model and SPSS17.0 (SPSS Statistics 2009) for all the other statistical analysis.

6.3. Results

6.3.1. Changes in the distribution of Acacia species

In 1998, A. melanoxylon and A. dealbata were present in 2.2% and 1.5% respectively of the total forested area of the study region. By 2008 they had expanded by 1% and 0.83% respectively (Table 6.3), with an expansion rate (proportion of total hectares invaded per year) of approximately 3,100 ha yr-1 and 2,500 ha yr-1. The intensity of the changes in the spatial distribution of Acacia species differs according to forest type. There has been a notable increase in the presence of A. melanoxylon in riparian forests and Eucalyptus spp. plantations, reaching 7% and 3.5% of the total


129

forest area of these two forest types respectively (Fig.6.3). In these two forest types, the increase was almost four times greater than for other forest types. Conversely, there has been a notable decrease in the presence of this species in mixed broadleaf forests and shrublands, although in 2008, it was present in 1% and 3% of these two forest types respectively. The presence of A. dealbata has increased significantly in the majority of forest types in the study region, even occurring in Quercus pyrenaica oak forest in 2008, where it was absent in 1998. This increase is particularly notable in mixed conifer-broadleaf forests, the most abundant forest type in the study area, where it is present in almost 3.5%. A small decrease in the presence of this species was detected in riparian forest and shrubland, occurring in 3% and 0.4% of each type respectively (Fig.6.3).

Table 6.3. Changes in distribution (ΔS) and stock (ΔG, ΔBA, ΔN) of A. dealbata and A. melanoxylon, 1998-2008.

A. melanoxylon A. dealbata

1998 2008 Increment 1998 2008 Increment

Number of plots with presence 167 262 95 129 199 70

Percentage of total forest area (%) 2.2 3.2 1.00 1.5 2.4 0.8

Area (ha) 57,300 88,250 31,000 40,900 66,300

25,400

Mean number of trees per ha 2.6 3.1 0.5 2.4 3.8 1.4 Relative density: % of Acacia sp. in total tree per ha 0.4 0.4 0.02 0.4 0.5 0.1

Mean basal area (m2/ha) 0.03 0.06 0.03 0.02 0.04 0.02 Relative dominance: %of Acacia sp. in total basal area (%) 0.2 0.2 0.1 0.1 0.2 0.1

Mean growing stock(m3/ha) 0.2 0.4 0.2 0.1 0.2 0.1

Proportion of Acacia sp. in total growing stock (%) 0.2 0.3 0.1 0.1 0.1 0.03

Fig.6.3. Changes in the proportion of the target Acacia species in the different forest types of NW Spain, 1998-2008.

0.00%

1.00%

2.00%

3.00%

4.00%

5.00%

6.00%

7.00%

8.00%

Pres

ence

A.melanoxylon (1998) A.melanoxylon (2008)

Forest

(+ 0,4)

(- 1,8)

(- 0,6)(+ 2,1)

(+ 2,6)

(-1,3)(+ 0,6)

0.00%0.50%1.00%1.50%2.00%2.50%3.00%3.50%4.00%4.50%

Pres

ence

A.dealbata (1998) A.dealbata (2008)

Forest type

(+ 0.03)

(-0.8)

(+1.4)(-0.9)

(+ 0.5)

(+ 0.6)(+ 0.7)

(+ 0.8)


130

6.3.2. Acacia species dynamics and dbh structure

Analysis of the dynamics of NFI plots in which Acacia species were present in 1998 and 2008 show that the number of trees per hectare, basal area and growing stock of these two invasive species almost doubled in all cases during this period (Table 6.3). However, while in the case of A. melanoxylon the increase has been greater in terms of basal area and proportion of growing stock in the stands where it occurs, A.dealbata has undergone a greater increase in the number of trees per hectare (Table 6.3).

The pattern of variation in the proportion of trees of different dbh classes over the studied period differs between the two species. In the case of A. melanoxylon, the proportion of trees with the smallest diameters has suffered a slight drop over the period, while the proportion of trees with medium-large diameter has increased (Fig.6.4). However, in the case of A. dealbata the proportion of small diameter trees has risen, whereas the proportion of medium diameter trees has decreased slightly (Fig. 6.4B).

Fig.6.4. Evolution of the total number of trees per dbh class during the period 1998-2008 for A) A. melanoxylon and B) A. dealbata in the forests of NW Spain.

The analysis of Acacia species saplings did not identify significant differences between forest types due to the high level of stochasticity in the associated variables. However, as can be seen from Table 6.4, the relative and absolute density of the regeneration varied from one forest type to another. A. melanoxylon regenerates more abundantly in mixed conifer-broadleaf forests than in Eucalyptus spp. plantations, the only two forest types in which saplings of this species were found. In addition, as revealed by the relative results, this species tends to dominate in the regeneration of mixed conifer-broadleaf forests. In the case of A.dealbata, despite the high dispersion of the saplings associated variables, the results show a clear tendency to dominate in the regeneration of all the forest types in which this species was found: mixed broadleaf forests, Eucalyptus spp. plantations and mixed conifer-broadleaf forest, regenerating in similar abundance.

0%

25%

50%

75%

100%

2.5-7.5 7.6-12.5 12.6-22.5 22.6-42.5 >42.6

Perc

enta

ge o

f num

ber p

er h

a

Dbh clases (cm)

1998 2008

A)

0%

25%

50%

75%

100%

2.5-7.5 7.6-12.5 12.6-22.5 22.6-42.5

Perc

enta

ge o

f num

ber p

er h

a

Dbh clases (cm)

1998 2008

B)


131

Table 6.4. Mean and standard deviation (Stand- Dev) of the absolute (Nsap) and relative (Nsaprel (%)) number of A. melanoxylon and A.dealbata saplings in the regeneration per hectare. For forest type abbreviations see Table 6.1

As indicated by the Welch and Post hoc tests through mean differences in the IVI over the period 1998-2008, the dominance of the two invasive species increased in all the forests types in which they were present (Table 6.5). However, these changes differed significantly depending on the forest type. The increase in overall importance of A.melanoxylon and A. dealbata was significantly greater in mixed conifer-broadleaf forests than in other forest types. Additionally, in the case of A. melanoxylon, this increase in dominance was also more significant in Eucalyptus spp. plantations than in other forest types. Table 6.5. Mean differences in IVI (Importance Value Index) of 1) A. melanoxylon and 2) A. dealbata, between the invaded forest types of NW Spain. (* Significant difference at level p> 0.05, for forest type abbreviations see Table 6.1).

6.3.3. Factors involved in the spread of Acacia species Among the 20 biotic and abiotic predictors previously considered to explain the spread of the Acacia species from 1998 to 2008 (Table 6.2), the best GLM fits were obtained for 11 different predictors in the case of A. melanoxylon and 10 for A.dealbata with which the spread of the two species over the period displayed significant relationships. The final models accounted for 52% and 48% of the observed variability in the spread of A. melanoxylon and A.dealbata respectively over the period (Table 6.3). As regards the climatic abiotic factors, mean annual temperature has significant and positive effects on both species spread over the studied period, while to a lesser extent, annual rainfall and the mean temperature oscillation has a negative one. The factor ‘closest distance to the sea’ also has a positive relationship with the spread of A.


Forest Type

Mean Nsap (Stand. Dev)

Mean Nsaprel (Stand. Dev)

Mean Nsap (Stand. Dev)

Mean Nsaprel (Stand. Dev)

EUC 875.4 (756.7) 0.5 (0.4) 2,772.8 (1,930.5) 0.7 (0.4)

MIXCB 2018.9 (1,036.4) 0.8 (0.3) 2,113.6 (1,916.2) 0.9 (0.1)

MIXB - - 3,028.5 (2,921.2) 0.9 (0.2)

1) Mean IVI (Stand.Dev) Forest type MARPIN

(n=10) MIXB (n=7)

EUC (n=19)

MIXCB (n=32)

0.14 (0.08) MARPIN -0.227 -0.64* -0,86* 0.36 (0.23) MIXB -0.420 -0.64 0.79 (0.61) EUC -0.22 1.00 (0.89) MIXCB

2) Mean IVI (Stand.Dev.)

Forest type MIXB (n=7)

EUC (n=9)

MIXCB (n=30)

0.62 (0.57) MIXB -0.25 -0,66* 0.87 (0.69) EUC -0.41 1.28 (0.69) MIXCB


132

melanoxylon in the study zone. Regarding human-mediated and other disturbances, clear-cutting, urban-forest interfaces and fire incidence had a positive and significant effect on both species. Among the biotic factors connectivity between stands shows a strong association with the spread. As for other biotic interactions, stand structure characteristics such as tree cover and total basal area exhibited a negative relationship while total richness has a positive association with their spread. Table 6.6. Binomial GLM results for the spread of A. melanoxylon and A.dealbata during the period 1998-2008. (Significance codes: ** (p> 0.01); * (p > 0.05)).

A. melanoxylon A.dealbata

Predictors

Effect D2

Effect

D2 Intercept -0.35 - 5.59 -

Ptot -0.0125 1.99* -0.0037 8.22** Tm 0.31 44.73** 0.22 68.43** Dsea 4.08E-05 62.89** - - Osc -0.28 9.7** -0.42 11.52** Clearcut 0.96 2.09** - - Connect -3.18E-04 85.33** -1.80E-04 66.15** Treecover -0.01 12.13** -0.04 10.48** Fire 0.93 2.06* 0.68 2.87* BA -0.04 4.47** -0.0407 3.85** TRichness 0.13 4.35** 0.22 12.31** FUInt - - 0.08 7.23** Deviance (D2) 51.623% 48.026% 6.4. Discussion The results confirm the prior hypothesis of expansion and upward dominance of Acacia species in the forested area of NW Spain over the last decade, revealing different invasion patterns from one forest type to another.

6.4.1. Spatio-temporal dynamics The quantification of the area occupied by invasive plants in a given zone, the identification of the stage of invasion, their spread rates and the potential to invade new areas are of critical importance in determining the ability of organisms to shift their ranges and to detect their invasive success and persistence (Higgins and Richardson 1996; Higgins et al. 2001). The percentage of forested area in NW Spain occupied by the two Acacia species reached 3.2% and 2.4% in 2008, exhibiting annual rates of invasion (proportion of total area invaded per year) of 0.1% and 0.08% for A. melanoxylon and A. dealbata respectively. To date, most of the empirical data on invasion rates worldwide are based on estimations from past aerial photographs (Lonsdale 1993; Higgins et al. 2001, Müllerová et al. 2005) for broad scale studies or, to a lesser extent, from estimates based on field work for local scale studies (Wangen and Webster 2006). However, to our knowledge this is the first time that these rates have been studied extensively on a detailed spatial grid on such a large scale. Bearing


133

in mind the limitations involved in comparing the spread rates associated with the differing sized areas and populations monitored of previous studies, it would appear that the Acacia species in NW Spain exhibit mid-high invasion rates, since similar or lower estimates have been reported for other invasive tree/shrub species; 0.08% for Pinus radiata (Richardson and Brown 1986), 0.03% for Acacia cyclops or 0.06% for Pinus pinaster (Higgins et al. 2001), all in South Africa. The rates of areal spread of the two Acacia species (31 km2 yr-1 and 25 km2 yr-1) are close to the average for other tree/shrub life forms worldwide, which present a mean spread rate of around 27 km2 yr-

1, ranging from 0.02 km2 yr-1 to 179 km2 yr-1 (from a review carried out by Pyšek and Hulme 2005). The aforementioned spatio-temporal spread was concomitant with a substantial increase in growing stock and dominance over the period 1998-2008, doubling in all the cases the preceding values but displaying different traits. Whereas A. melanoxylon seems to be increasing its dominance in forests where it was present through increments in growing stock and basal area, A. dealbata displays a greater increase in density and regeneration. Although this rise in dominance is general for all forest types analyzed, a significant increase is observed in disturbed forest types (according to our characterization) such as Eucalyptus spp. plantations and mixed conifer-broadleaf forests, but also in habitats which in principle are less altered by human activity such as native mixed-broadleaf forests. This pattern is also observed in the relative proportion of trees in the regeneration where Acacia species tend to dominate and to homogenize this important forest stratum supporting previous results at local scale in the region (Hussein et al. 2011; Gonzalez-Muñoz et al. 2012; Lorenzo et al. 2012). The concurrence of our findings with those of the aforementioned studies may have important implications for the future composition and functional diversity of forests in which Acacia species are becoming naturalized and spreading and where a decrease in structural and compositional complexity is expected. The dbh distribution structure of Acacia species over the studied period suggest an early stage of invasion in the forests of NW Spain being saplings the dominant class. This finding is in accordance with the general increase of Acacia species occurrence observed in the majority of forest types of NW Spain in the last inventory. Bearing in mind the capacity of Acacia species to flower throughout the year and their strong resprouting ability (Lorenzo et al. 2010), these results highlight the potential of Acacia species to continue spreading in the near future, confirming previous predictions suggesting that these species have not yet reached their potential area of distribution within the Iberian Peninsula (Gassó et al. 2005). 6.4.2. Factors associated with Acacia species spread and invasibility

The results of the GLMs point to connectivity and environment as the key factors associated with the expansion of the two invasive species.

It is not surprising that connectivity between Acacia populations has a strong association with the spread since increased availability of propagules between proximate populations raises the chances of establishment, persistence, naturalization and invasion (Alston and Richardson 2006). Furthermore, this result is in accordance


134

with the non-long distance dispersal adaptation more common in Acacia species which are usually dispersed by animals such as birds and ants (Davison and Morton 1984; Lorenzo et al. 2010).

With regard to environmental factors, temperature and distance to the sea are revealed as the most important filters constraining the colonization of new areas. This finding agrees with the habitat compatibility hypothesis (Rejmánek et al. 2005) which states that habitats globally tend to be invaded by species from similar environments at source. The natural distribution range of A. melanoxylon and A. dealbata, mainly in south-eastern Australia, illustrates their preference for oceanic climate locations (Costermans 1985). This partly explains the absence of both Acacia species in certain forest types characterized by higher temperature oscillations, such as Pinus sylvestris plantations located at higher altitudes. Similarly A.melanoxylon is completely absent in oak forests of Quercus pyrenaica in submediterranean climatic transition zones although the most recent inventory indicated a small presence of A.dealbata in these forests. The broader ecological valence of the latter allows this species to invade more continental locations, away from the influence of the coast. However, this significant relationship with certain environmental factors does not necessarily explain the degree of invasion in the different forest types. Moreover, forest types with similar environmental conditions exhibit dissimilar levels of invasion, suggesting that other mechanisms render them more or less susceptible to invasion. Disturbances are considered one of the most important factors behind the invasive spread of Acacia species (Brooks et al. 2004; Lorenzo et al. 2010; Le Maitre et al. 2011). The significant positive association found between the spread of A. melanoxylon and A. dealbata and disturbance events such as fires, clear-cutting and urban-forest interfaces would appear to support this idea. Davis et al. (2000) suggest that habitats might be more susceptible to invasion when there is an increase in the amount of unused resources resulting from disturbance events. Accordingly, some of the habitats with higher levels of disturbance correspond to forest types with a higher degree of invasibility. Eucalyptus spp. plantations are located close to the coast where population is concentrated and where intuitively there would be a high propagule pressure (di Castri 1989) due to their proximity to urban-forest interfaces and communication networks. Furthermore, Eucalyptus spp. plantations and mixed conifer-broadleaf forests, two of the most invaded forest types, had the highest fire incidence over the period (affecting almost 10% and 8% respectively of their extent). Forest fires are one of the main disturbance factors in forest ecosystems across large areas of Spain, including NW Spain (Moreno et al. 1998; de la Cueva et al. 2006). Acacia species are highly resilient to fires and are capable of regenerating both through germination and sprouting from roots and stems after fires (Ough 2001). Some authors (e.g. Lorenzo et al. 2010) have already suggested that the spread of A. dealbata in NW Spain may be assisted by human disturbances such as fires, although this hypothesis has not yet been tested at larger scales (see however de la Cueva (2014) for a local scale study). Hence, one of the key findings of the present study is the positive relationship found between the spread of both Acacia species and fire incidence, which confirms that areas currently occupied by Acacia are often areas which have been affected by fire.


135

Disturbances seem to be an important factor associated with the spread of invasive species, however not always the more anthropogenic disturb forests are the ecosystems with a higher degree of invasion. In such cases, as suggested by Alpert et al. (2000) or Blossey and Nötzold (1995), community structure traits might be influencing the invasibility of some ecosystems. This is the case of Pinus radiata and Pinus pinaster plantations, which present lower levels of invasion than other forests even though they are intensively managed and occupy large areas with as suitable environmental conditions as other more invaded forests. Pine plantations form monospecific forests with higher stand basal area and an evergreen canopy which is more closed than that of Eucalyptus spp. plantations, resulting in a general absence of understory. In these forests, Acacia species would face strong competition for resources, which may be associated with negative relationships found between the spread of these species and both total stand basal area and tree cover. This pattern is also corroborated by the significantly lower increase in dominance found in Pinus pinaster plantations compared with other forest types. Furthermore, these results support the findings of previous studies that identified important relationships between invasion by alien species and percentage of overstorey cover (Alston and Richardson 2006) and canopy closure (Fuentes-Ramírez et al. 2011).

As regards other community structure traits which influence the spread of Acacia species; total richness per plot shows a significant positive relationship supporting previous studies at similar scales (Alpert et al. 2000; Stohlgren et al. 2003; Souza et al. 2011). In accordance with this finding, forest types with higher values of richness such as mixed conifer-broadleaf forests, mixed-broadleaf forests and riparian forests, show some of the highest levels of invasion. Levine and D`Antonio (1999) suggested that species richness may be positively correlated with invasion, because both are promoted by the same factors. Like floodplain forests, the conifer-broadleaf and mixed-broadleaf forests of NW Spain form transitional, dynamic ecosystems that act as corridors for species between adjacent habitats (Richardson et al. 2007) and therefore, support high levels of diversity. The current mixed conifer-broadleaf forests of NW Spain are the result of natural colonization by native species of abandoned monospecific plantations (Saura and Carballal 2004) and the mixed-broadleaf forests are usually found in valleys near populations, with a rich mosaic of land uses and where there would be a strong propagule pressure and more opportunities for Acacia species to be established. In mixed forests, our results also show an alarming tendency for Acacia species to dominate in the regeneration which lead us to suggest that in these cases, once established in the understory, other processes such as competitive ability (Blossey and Nötzold 1995) and allelopathy (Callaway and Aschehoug 2000) may facilitate their persistence and dominance.

6.5. Conclusions Acacia species are spreading rapidly and are becoming the dominant tree species across large areas of forest in NW Spain. As this study suggests, the success of Acacia species in spreading to and invading new areas is not due to a single mechanism but rather to a group of interrelated processes. The distribution ranges of the species in the forests of NW Spain are mainly constrained by environmental filters and the connectivity or propagule pressure between proximate populations.


136

Additionally, the combination of disturbance events and stand structure traits seem to play an important role in determining the level of invasion of the different forests. Our results point to an early stage of invasion, highlighting the potential of Acacia species to continue spreading. This fact, together with the high frequency of disturbances such as fire, may be critical in determining the configuration of future forest landscapes in the region (e.g. de la Cueva et al. 2012).

The empirical results from this research will contribute to the growing reference database on plant invasion rates and may provide practical help in the assessment of level and severity of biological invasions worldwide. Furthermore, the detailed data obtained from this type of study, such as spread rates, spread determinants and forests invisibility is crucial to improving spatially-explicit information on the risk of invasions as well as facilitating the development of efficient policies and management measures for forest conservation. Although limited to forest ecosystems, this work highlights the suitability of using broad-scale periodic forest surveys to monitor invasive plants, as well as their potential to contribute in the future to the necessary practical and theoretical understanding of biological invasions. 6.6. References

Alpert P, Bone E, Holzapfel C (2000) Invasiveness, invasibility and the role of environmental stress in the spread of non-native plants. Perspective Plant Ecol. Evol. Systematics. 3: 52–66.

Alston KP, Richardson DM (2006) The roles of habitat features, disturbance, and distance from putative source populations in structuring alien plant invasions at the urban/wildland interface on the Cape Peninsula, South Africa. Biol Conserv. 132(2): 183-198.

Anthos (2012) Sistema de información de las plantas de España. Real Jardín Botánico, CSIC-Fundación Biodiversidad. Recurso electrónico [http://www.anthos.es. Last Access: March 2014]

Barbati A, Corona P, Marchetti M (2007) A forest typology for monitoring sustainable forest management: The case of European Forest Types. Plant Biosyst. 141(1): 93-103.

Blossey B, Nötzold R (1995) Evolution of increased competitive ability in invasive non-indigenous plants: a hypothesis. J. Ecol. 83: 887– 889.

Boschetti L, Roy DP, Hoffmann AA (2009) MODIS Collection 5 Burned Area Product - MCD45 User’s Guide, Version 2.0, November 2009

Brooks ML, D'Antonio CM, Richardson DM et al (2004) Effects of invasive alien plants on fire regimes. BioScience 54(7): 677- 688.

Callaway RM, Aschehoug ET (2000) Invasive plants versus their new and old neighbors: a mechanism for exotic invasion. Science. 290: 521–523.

Carballeira A, Reigosa MJ (1999) Effects of natural leachates of Acacia dealbata Link in Galicia (NW Spain). Bot. Bull. Acad. Sin. 40: 87-92.


137

Chytrý M, Jarosik V, Pysek P, Hajek O, Knollová I, Tichý L, Danihelka J (2008) Separating habitat invasibility by alien plants from the actual level of invasion. Ecology. 89(6): 1541-1553.

Corona P, Chirici G, McRoberts RE, Winter S, Barbati A (2011) Contribution of large-scale forest inventories to biodiversity assessment and monitoring. Forest Ecol Manage. 262(11): 2061-2069.

Costermans L (1985) Native Trees and Shrubs of South -Eastern Australia. Rigby Publishers, Melbourne.

Crawley MJ (1993) GLIM for ecologists. Blackwell Scientific, Oxford.

Curtis JT (1959) The Vegetation of Wisconsin, An Ordination of Plant Communities. University Wisconsin Press, Madison. Wisconsin.

Davidson DW, Morton SR (1984) Dispersal adaptations of some Acacia species in the Australian arid zone. Ecology. 65: 1038–1051.

Davis MA et al. (2000) Fluctuating resources in plant communities: a general theory of invasibility. J. Ecol. 88: 528–534.

de la Cueva AV, García del Barrio JM, Ortega M, Sánchez-Palomares O (2006) Recent fire regime in peninsular Spain in relation to forest potential productivity and population density. International J. of Wildland Fire. 15: 397-405.

de la Cueva AV, Quintana JR, Cañellas I (2012) Fire activity projections in the SRES A2 and B2 climatic scenarios in peninsular Spain. International J. of Wildland Fire. 21: 653-665.

de la Cueva AV (2014) Case studies of the expansion of Acacia dealbata in the valley of the river Miño (Galicia, Spain). Forest Syst.

Di Castri F (1989) History of biological invasions with special emphasis on the Old World. In: Drake JA, Mooney HA, di Castri F et al. (Eds.), Biological invasions: a global perspective, John Wiley & Sons, New York, USA.

EC (2011) Communication from the Commission to the European Parliament, the Council, the Economic and Social Committee and the Committee of the Regions – Our Life Insurance, Our Natural Capital: An EU Biodiversity Strategy to 2020. COM (2011) pp. 244

EEA (2012) Corine Land Cover 2000 seamless vector data. European Environment Agency.http://www.eea.europa.eu/data-and-maps/data/corine-land-cover-2000-clc2000-seamless-vector-database-4#tab-gis-data. Last access: February 2014.

Foxcroft LC, Richardson DM, Rejmánek M, Pyšek P (2010) Alien plant invasions in tropical and sub-tropical savannas: patterns, processes and prospects. Biol Invasions. 12(12): 3913-3933.

http://www.eea.europa.eu/data-and-maps/data/corine-land-cover-2000-clc2000-seamless-vector-database-4#tab-gis-data

http://www.eea.europa.eu/data-and-maps/data/corine-land-cover-2000-clc2000-seamless-vector-database-4#tab-gis-data


138

Fuentes-Ramírez A, Pauchard A, Cavieres LA, García RA (2011) Survival and growth of Acacia dealbata vs. native trees across an invasion front in south-central Chile. Forest Ecol Manage. 261(6): 1003-1009.



Gonzalo J (2010) Diagnosis Fitoclimática de la España Peninsular. Hacia un modelo de clasificación funcional de la vegetación y de los ecosistemas peninsulares españoles. In: Serie técnica Naturaleza y Parques Nacionales. Organismo Autónomo de Parques Nacionales. Ministerio de Medio Amiente y Medio Rural y Marino de España (ed). Madrid.

Hejda M, Pyšek P, Jarošík V (2009) Impact of invasive plants on the species richness, diversity and composition of invaded communities. J. Ecol. 97(3): 393-403.

Heymann Y, Steenmans C, Croissille G, Bossard M (1994) Corine Land Cover - A technical guide. Office for Official Publications of European Communities. CEC, Luxembourg. http://www.eea.europa.eu/publications/COR0-landcover. Last access: December 2013.

Higgins SI, Richardson DM (1996) A review of models of alien plant spread. Ecol Mod. 87: 249–265.

Higgins SI, Richardson DM, Cowling RM (2001) Validation of a spatial simulation model of a spreading alien plant population. J App Ecol. 38: 571-584.

Hollander M, Wolfe DA (1999) Non-Parametric Statistical Methods, second ed. Wiley, New York.

Hosmer DW, Lemeshow S (2000) Applied Logistic Regression, Second Edition. John Wiley & Sons, Inc., New York.

Hussain MI, Gonzalez L, Reigosa MJ (2011) Allelopathic potential of Acacia melanoxylon on the germination and root growth of native species. Weed Biol Manage. 11(1): 18-28.

Kolar CS, Lodge DM (2001) Progress in invasion biology: predicting invaders. Trends Ecol Evol. 16(4): 199-204.

Le Maitre DC, Gaertner M, Marchante E et al. (2011) Impacts of invasive Australian acacias: implications for management and restoration. Divers Distrib. 17: 1015-1029.

Levine JM, D’Antonio CM (1999) Elton revisited: a review of evidence linking diversity and invasibility. Oikos, 87: 15–26.


139

Levine JM, Vilà M, D'Antonio CM, Dukes JS, Grigulis K, Lavorel S (2003) Mechanisms underlying the impacts of exotic plant invasions. Proc R Soc Lond B Biol Sci. 270: 775-781.



Lorenzo P, Pazos-Malvido E, Rubido-Bará M, Reigosa MJ, González L (2012) Invasion by the leguminous tree Acacia dealbata (Mimosaceae) reduces the native understorey plant species in different communities. Aust J Bot. 60(8): 669-675.

Lund HG (1998) A comparison of multipurpose resource inventories (MRIs) throughout the world. Working PaperNo. 14. Joensuu, Finland: European Forest Institute. 46 pp.

Mack RN, Simberloff D, Mark Lonsdale W, Evans H, Clout M, Bazzaz FA (2000) Biotic invasions: causes, epidemiology, global consequences, and control. Ecol Appl. 10(3): 689-710.

Marchante H, Marchante E, Freitas H (2003) Invasion of the Portuguese dune ecosystems by the exotic species Acacia longifolia (Andrews) Willd: effects at the community level. In Child, L.E., Brock, J.H., Brundu, G., Prach, K., Pyšek, P., Wade, P.M. & Williamson, M. (Eds.), Plant Invasions: Ecological Threats and Management Solutions. Backhuys Publishers. The Netherlands. pp. 75-85.

Moreno JM, Vázquez A, Vélez R (1998) Recent history of forest fires in Spain. In: Moreno JM (ed.) Large Forest Fires, Backhuys Publishers, Leiden, The Netherlands, pp. 159-185.

Pyšek P, Jarošik V, Kuèera T (2002) Patterns of invasion in a temperate nature reserves. Biol. Conserv. 104: 13-24.

Pyšek P, Hulme PE (2005) Spatio-temporal dynamics of plant invasions: linking pattern to process. Ecoscience. 12: 302-315.

Rejmánek M, Richardson DM (1996) What attributes make some plant species more invasive? Ecology. 77: 1655–1661.

Rejmánek M, Richardson DM, Pyšek P (2005) Plant invasions and invasibility of plant communities. In: van der Maarel, E. (ed.) Vegetation ecology, pp. 332-355. Blackwell Science, Oxford, UK.

Richardson DM, Holmes PM, Esler KJ, Galatowitsch SM, Stromberg JC, Kirkman SP, Hobbs RJ (2007) Riparian vegetation: degradation, alien plant invasions, and restoration prospects. Diver Distri. 13(1): 126-139.

Richardson DM, Brown PJ (1986) Invasion of mesic mountain fynbos by Pinus radiata." South African J. Bot. 52: 529-536.


140


Romero Buján MIR (2007). Flora exótica de Galicia (noroeste ibérico). Botanica Complutensis. 31: 113-125.

Sakai AK (2001) The population biology of invasive species. Annu. Rev. Ecol. Syst. 32: 305–332

Sanz-Elorza M, Dana ED, Sobrino E (2004) Atlas de las plantas alóctonas invasoras de España. Madrid, D.G. de la Biodiversidad, Ministerio de Medio Ambiente, 384 pp.

Saura S, Carballal P (2004) Discrimination of native and exotic forest patterns through shape irregularity indices: an analysis in the landscapes of Galicia, Spain. Landscape Ecol. 19(6): 647-662.

Sheppard AW, Shaw RH, Sforza R (2006) Top 20 environmental weeds for classical biological control in Europe: a review of opportunities, regulations and other barriers to adoption. Weed Res. 46: 93–117.

Souza L, Bunn WA, Simberloff D, Lawton RM, Sanders NJ (2011) Biotic and abiotic influences on native and exotic richness relationship across spatial scales: favourable environments for native species are highly invasible. Funct Ecol. 25(5): 1106-1112.

Stohlgren TJ, Barnett DT, Kartesz JT (2003) The rich get richer: patterns of plant invasions in the United States. Front Ecol Enviro. 1(1): 11-14.

UNEP (2010) Report of the Seventh Meeting of the Conference of the Parties to the Convention on Biological Diversity (UNEP/CBD/COP/35/4). Decision XI/28, UNEP.

Wangen SR, Webster CR (2006) Potential for multiple lag phases during biotic invasions: reconstructing an invasion of the exotic tree Acer platanoides. J Appl Ecol. 43(2): 258-268.

7. Discusión general

CAPÍTULO7: DISCUSIÓN GENERAL

143

7. Discusión general

7.1. Cambios en la distribución de los bosques

Los resultados de los capítulos 3, 4 y 5 de esta tesis confirman hipótesis previas que planteaban cambios en la actual distribución de los bosques como consecuencia del acelerado cambio global contemporáneo (FAO 2001; 2015; MEA 2005; Hanewinkel et al. 2014) mostrando patrones y tendencias muy diferentes entre especies, gradientes y regiones estudiadas.

Para ello, y por primera vez, esta tesis propone la comparación de la serie histórica completa de ciclos del Inventario Forestal Nacional (IFN) español, mediante la aplicación de técnicas geoestadísticas, con el objetivo de analizar cambios en la distribución de especies a lo largo de gradientes geográficos (Hernández et al. 2013; Hernández et al. 2014a). La geoestadística engloba un conjunto de herramientas estadísticas que permiten describir la continuidad espacial de un fenómeno natural, incorporando la autocorrelación espacial en la estimación del valor de una variable en sitios no muestreados del espacio (Gallardo y Maestre 2008). La autocorrelación espacial refleja la continuidad espacial de los principales factores que intervienen en la distribución y dinámica de especies forestales como la dispersión, la mortalidad, las barreras geográficas o ciertos factores socioeconómicos (Miller et al. 2007) y se ha identificado como un factor relevante a considerar en los modelos de distribución de especies (Segurado y Araújo 2004). Basado en ésta y otras propiedades de las técnicas geoestadísticas, en el capítulo 3 esta tesis propone una nueva aproximación metodológica (Fig.7.1) para analizar cambios en la distribución espacial de especies forestales en base a la comparación de los modelos de “Universal Kriging” (UK) derivados de la información espacial de cada ciclo de IFN (Hernández et al. 2014a).

Las técnicas geoestadísticas de predicción espacial, conocidas como “Kriging”, constituyen estimadores óptimos e insesgados, en presencia de autocorrelación espacial, y permiten cuantificar la incertidumbre de la estimación (Mandallaz 2000). Así, la aplicación de estas técnicas supone una oportunidad para la comparación de estimaciones homogéneas derivadas de diferentes ciclos de inventario aunque el diseño de la malla de muestreo no sea idéntico, como es el caso del IFN español, donde la ubicación de la parcelas a lo largo de los diferentes ciclos de IFN ha variado (Vallejo y Villanueva 2002; Hernández et al. 2013; Hernández et al. 2014a). En este sentido, la comparación directa de los datos de los ciclos permanentes de IFN puede dar lugar a una sobreestimación de los cambios en la distribución de las especies debido a los cambios en la superficie muestreada o variaciones en el esfuerzo de muestreo a través de los sucesivos inventarios (Loiselle et al. 2003). El uso de la técnica “block Kriging” resuelve este problema proporcionando un estimador insesgado de la media de la presencia de la especie por cada bloque o rango de la variable ambiental a analizar (Yoo y Trgovac 2011), permitiendo así la comparación directa por bloques de la presencia de la especies en cada ciclo de IFN. Además, el uso de los intervalos de confianza de la varianza de la predicción de una especie en base al “block Kriging” propuesto supone un avance metodológico importante que permite determinar si los cambios observados en la distribución de dicha especie a lo largo de


144

los diferentes ciclos de IFN son significativos estadísticamente. Este aspecto es una limitación frecuente en otras aproximaciones estadísticas en el estudio de cambios en los rangos de distribución de especies (Urli et al. 2014). Por último, la inclusión de la interacción entre variables auxiliares en el modelo de UK, permite analizar los cambios de distribución de una especie relacionados con una variable a lo largo del gradiente de otra, un factor importante cuando los estudios se enfocan en áreas de compleja geografía como la Península Ibérica.

La incorporación de variables climáticas en los modelos de UK o de otras variables antrópicas (información sobre masas gestionadas, cambios de usos de suelo, etc.) y el desarrollo de modelos espacio-temporales para determinar tendencias en el tiempo constituyen líneas a avanzar en el futuro. Además, hasta ahora esta metodología se ha aplicado sólo al estudio de la presencia/ausencia de especies a lo largo de gradientes ambientales, sin embargo, otras variables de interés ecológico para el estudio de la evolución de los bosques derivadas del IFN podrían ser también analizadas usando esta aproximación.

Fig. 7.1. Esquema de la metodología propuesta en esta Tesis Doctoral, basada en modelos de “Universal Kriging” y técnicas de “block Kriging”, para estimar cambios en la distribución de especies a partir de comparación de ciclos de IFN (Fuente: modificado de Hernández et al. (2013))

De forma general, los resultados de los capítulos 3, 4 y 5, centrados en especies forestales del Pirineo occidental, especies que interactúan en la zona transicional Submediterránea de Navarra y en la región Atlántica, confirman la expansión generalizada observada en los bosques europeos en los dos últimos decenios (Fig. 1.2., FAO 2015). Este aumento de superficie forestal de las últimas décadas se debe principalmente a la interacción del cambio climático con el abandono de formas de vida tradicionales relacionadas con sistemas agrarios y a la sucesión


145

natural de los bosques (MEA 2005; Chauchard et al. 2007; Gellrich et al. 2007; Bodin et al. 2013).

En España, estos cambios se produjeron a partir de la década de 1960 con la liberalización de la economía y su integración en la economía de mercado, cuando la explotación tradicional de sistemas agrarios y forestales dejó de ser rentable (García-Ruiz 1996; Lasanta 2002; Valbuena-Carabaña et al. 2010). Como consecuencia, se produjo una nueva organización socioeconómica del paisaje, con un despoblamiento generalizado de las zonas rurales y un cese de explotaciones tradicionales ganaderas y forestales con efectos directos en los bosques. El resultado de todos estos cambios ha dado como resultado el aumento progresivo por sucesión natural de las zonas de matorral y arbolado (el llamado “forest recovery”) (Debussche et al. 1999; Vicente-Serrano et al. 2005) y a un aumento de la densificación y la acumulación de biomasa de los bosques, muchos de ellos hoy sin gestionar (Poyatos et al. 2003; Montero y Serrada 2013).

Fig.7.1. Tendencias de ganancias o pérdidas anuales de superficie forestal en hectáreas (ha) en el periodo 1990-2015 mostradas por el último informe sobre el estado de los bosques (FAO 2015). España, dentro de la zona templada del hemisferio norte, muestra una tendencia de ganancia de superficie forestal.

Esta expansión de la superficie encontrada para todas las especies estudiadas, contradice parcialmente las predicciones sobre cambios en distribución de especies forestales basados en diferentes escenarios de cambio climático para Europa (Ohlemüller et al. 2006) y la Península Ibérica (Benito-Garzón et al. 2008; Sánchez de Dios et al. 2009; Ruiz-Labourdette et al. 2013) que proyectaban para fechas próximas (2020, 2050) una generalizada contracción de las áreas de distribución de muchas especies. Sin embargo, las mencionadas proyecciones sólo consideran los cambios en las condiciones climáticas como impulsores de las alteraciones en la distribución de especies, cuando hoy en día numerosos estudios muestran que estos cambios están también determinados por otros factores del cambio global como alteraciones en el uso del territorio, gestión forestal y subsiguiente sucesión natural (Coop y Givnish 2007; Ameztegui et al. 2010; Callaghan et al. 2013; Bodin et al. 2013; Hernández et al. en rev.2), nuevas interacciones bióticas con especies nativas (Hernández et al. en


146

rev.2) o alóctonas (Holmes y Cowling 1997; Levine et al. 2003; Morales y Traveset 2009; Hernández et al. 2014b), factores físicos (Lenoir y Svenning 2015; Hernández et al. en rev.1) y otros factores ecológicos inherentes a la ecología de las especies como la longevidad (Rubiales et al. 2010; Hernández et al. en rev.1) o la plasticidad fenotípica (Benito-Garzón et al. 2011).

En cambio, esta expansión general en el área de distribución de las especies analizada a lo largo de gradientes ambientales (capítulos 3 y 4) confirman, en parte, algunas de las hipótesis biogeográficas bajo escenarios de cambio climático esperadas. Así, en el capítulo 4, el avance observado del área de distribución de las especies Submediterránea (Q. subpyrenaica), templada (F. sylvatica) y conífera montana (P. sylvestris) desde zonas de transición Submediterráneas hacia otras más frías y húmedas a lo largo de gradiente bioclimático Templado-Mediterráneo, en combinación con la expansión hacia la zona de transición de la especie mediterránea, confirma la proyectada “Mediterraneización” de esta zona transicional de la Península Ibérica (Benito Garzón et al. 2008; Sánchez de Dios et al. 2009; Ruiz-Labourdette et al. 2012, 2013) y corrobora las tendencias hacia la “termolización de la composición de especies” encontradas en sistemas montañosos europeos (Gottfried et al. 2012). De la misma forma, los mayores ratios de cambio observados en la especie Submediterránea en el capítulo 4 ratifican asunciones previas que sugerían una mayor sensibilidad a los cambios ambientales de especies con restringida distribución (Thuiller et al. 2005). De hecho, los mayores cambios asociados a esta especie y al taxon mediterráneo apoyan nuevas hipótesis y predicciones que sugieren que los cambios en la distribución de las especies se están produciendo a lo largo de toda la distribución de la especie y no sólo en los márgenes de su distribución (Jump et al. 2009; Ruiz-Labourdette et al. 2013).

En cuanto a los cambios a lo largo del gradiente altitudinal, los capítulos 3 y 4 muestran tendencias claras de desplazamiento hacia mayores elevaciones de la especie templada estudiada (Fagus sylvatica), cuya distribución ha sido la menos afectada por factores antrópicos en las últimas décadas. En el Pirineo occidental, la expansión en altitud de esta especie confirma la esperada invasión del piso sub-alpino por especies caducifolias (Kräuchi y Kienast 1993) como consecuencia del progresivo aumento de las temperaturas. Esta tendencia también se observa en la última década en la distribución de la especie Submediterrénea (capítulo 4) y otras especies a escala europea (Lenoir et al. 2009), lo que corroboraría el efecto del aumento de la temperatura detectado desde 1970 en el área de estudio (Hernández et al. en rev.1) y Europa (IPPC 2013; WMO 2013). Sin embargo, hay que considerar que estas especies ocupan generalmente el piso colino y montano o supramediterráneo en sus respectivas áreas de distribución por lo que tendrían la posibilidad de desplazarse hacia cinturones de vegetación de mayor elevación y más frescos en respuesta al aumento de temperatura. No es el caso de la conífera de montaña estudiada, P. sylvestris, que generalmente ocupa el límite del bosque en altitud en la mayor parte de las montañas Ibéricas donde habita, viendo limitada su capacidad para desplazar su distribución en altitud cuando se analiza todo su rango altitudinal de distribución (Hernández et al. en rev.1). Esto no sucede cuando el estudio se enfoca en una sección más estrecha de un gradiente geográfico, como muestran los resultados de desplazamientos en altitud de P. sylvestris en el Pirineo occidental (capítulo 3). En


147

este caso, se observa un desplazamiento de la especies en las últimas décadas de 200 m hacia mayores elevaciones. Esta disparidad de resultados enfatiza la necesidad de analizar cambios en la distribución de especies en zonas que recojan todo el rango de distribución de las especies a lo largo de gradientes. Si los estudios se centran en secciones más estrechas de un gradiente geográfico, como la altitud en zonas de montaña (Urli et al. 2014; Hernández et al. 2014), probablemente sobreestimaremos los cambios observados, y estaremos detectando una respuesta sesgada de la distribución de la especie a cambios en las condiciones ambientales (Lenoir et al 2009).

Las alteraciones demográficas de las especies tienen repercusiones a largo plazo en sus patrones de distribución (Hampe y Petit 2005; Purves 2009), sin embargo la relación demografía-distribución es un fenómeno muy poco estudiado. En este contexto, el análisis en el capítulo 5 de la dinámica y tasas demográficas en las últimas décadas de las poblaciones de Fagus sylvatica en la región biogeográfica Atlántica confirman teorías previas que proponían la expansión actual de la especie siguiendo la inercia de su migración Holocena hacia el noroeste peninsular (Costa et al. 1997). Esta tendencia parecería contradecir la esperada contracción en el margen meridional de distribución de la especie en un escenario de cambio climático. Sin embargo, confirma resultados previos que encuentran efectos positivos del actual aumento de temperatura y mayores periodos vegetativos en el crecimiento y la productividad de especies europeas en ausencia de limitaciones hídricas (Nabuurs et al. 2003; Martínez-Vilalta et al. 2008; Poljanec et al. 2010; Estiarte y Peñuelas 2015). Este es el caso de las poblaciones actuales de F. sylvatica en la región Atlántica Ibérica donde se ha constatado una tendencia clara de incremento de temperaturas en la últimas décadas, pero no de disminución de precipitación (Berg et al. 2013; IPCC 2013). Además esta expansión parece quedar confirmada por los patrones de regeneración y estructura poblacional que muestran las poblaciones Atlánticas de Q. petraea y F. sylvatica en la última década y que señalan los actuales procesos de colonización y reemplazamiento de Quercus petraea por F. sylvatica en la zona. Sin embargo, en el futuro y con un escenario de cambio climático con progresivo aumento de temperaturas y eventos extremos de sequías (IPCC 2013) la dirección e intensidad de la interacción entre las dos especies podría cambiar y revertir las tendencias de patrones de distribución actuales.

En este sentido y como consecuencia de los cambios en el área de distribución de las especies observados en los capítulos 3, 4 y 5, uno de los resultados más consistentes de esta tesis es la constatación de la expansión del área donde las especies analizadas coexisten. Este patrón coincide con la visión generalizada del aumento de riqueza y nuevos ensamblajes de especies como resultado de desplazamientos latitudinales y altitudinales de las especies (Walther et al. 2002) y parece común en otras zonas de la Península Ibérica (Sanz-Elorza et al. 2003; Peñuelas y Boadas 2003; Ameztegui y Coll 2011). Como consecuencia, las interacciones bióticas (Gómez- Aparicio et al. 2011) que se establezcan entre estas especies en el futuro tendrán un papel crucial en la persistencia, estabilidad y configuración del futuro paisaje forestal de la Península Ibérica así como en los bienes y servicios de los ecosistemas forestales.


148

7.2. Un nuevo factor en ecología y gestión forestal: las invasiones biológicas

En el capítulo 6 se propone por primera vez la comparación de la información detallada de diferentes ciclos de Inventarios Forestales Nacionales (IFNs) a escala regional para determinar la capacidad de colonización, establecimiento y persistencia de especies invasoras en nuevos territorios forestales. En este sentido, la posibilidad que brinda la comparación de IFNs para la cuantificación del área ocupada por la especie invasora, su ratio de expansión y la facilicidad para colonizar nuevos territorios son de importancia crítica para analizar el potencial invasivo de una especie alóctona (Higgins y Richardson 1996; Higgins et al. 2001). Hasta la fecha, la mayor parte de los datos empíricos sobre ratios de invasión de especies exóticas se basaban en estimaciones a partir de comparaciones aproximadas entre fotografías aéreas antiguas y actuales en estudios a escalas amplias (Lonsdale 1993; Higgins et al. 2001), o, en menor medida, a partir de estimaciones basadas en trabajo de campo para estudios a escala local (Wangen y Webster 2006). Sin embargo, este trabajo propone el uso por primera de vez del análisis de ratios de expansión de especies invasoras en una malla espacial georreferenciada a gran escala y con el subsiguiente incremento de precisión y detalle.

Además, la posibilidad de comparar ciclos de IFNs presenta la oportunidad de estudiar la evolución y dinámica de invasión de especies alóctonas, una de las grandes lagunas en el conocimiento de las invasiones biológicas identificadas en años recientes (Pyšek et al. 2002; Richardson et al. 2010). Así, los resultados del estudio en el noroeste peninsular en base a la comparación de los IFNs en el periodo 1998-2008 confirman la expansión y el incremento de dominancia en algunos tipos de bosque de Acacia dealbata y Acacia melanoxylon en la superficie forestal estudiada con unos ratios de expansión medio-altos teniendo en cuenta otros trabajos a escala global. Además se constata que estas especies tienen potencial para seguir invadiendo nuevos territorios por su temprano estado de colonización (Gassó et al. 2012) y por características ecológicas como son su amplio nicho ecológico y su capacidad de propagación rápida, tanto vegetativamente como por semillas (Lorenzo et al. 2010), entre otras. La combinación de la invasividad de las especies de Acacia, con factores como las buenas condiciones ambientales que encuentran estas especies en el clima templado y húmedo del noroeste peninsular y el proyectado aumento de riesgo de incendios en la zona en los próximos años (de la Cueva et al. 2012; Fig. 7.3A) puede tener repercusiones para la futura configuración de los paisajes gallegos. La rápida expansión de estas especies y su capacidad para colonizar y dominar diversos ecosistemas forestales (González-Muñoz et al. 2012) podrían suponer una futura homogeneización de ecosistemas forestales de noroeste peninsular con la resultante pérdida de biodiversidad a largo plazo.


149

Fig.7.3. A) Mapa de predicción para 2071-2100 del cambio en el régimen de incendios para Europa bajo escenarios Seasonal Severity Rating (SSR) con referencia al periodo 1961-1990 (EEA 2015). B) Mapa de grado de invasión para Europa (EEA 2015).

En la actualidad, existen mapas sobre riesgo de invasión a escala Europea (Fig.7.3B) pero la escala y precisión de los mismos los hacen difícilmente aprovechables en gestión medioambiental. En un futuro, los resultados precisos extraídos de este tipo de trabajos utilizando IFNs como son el nicho potencial climático de la especie, la distancia óptima de conectividad entre poblaciones para su dispersión, los ratios de expansión o la probabilidad de invasión de los diferentes hábitats, junto con la distribución actual de la especie en la parcelas georrefereciadas, permitirán hacer mapas precisos y a diferentes escalas del riesgo de invasión de las especies invasoras analizadas en ecosistemas forestales, una herramienta que puede ser de gran utilidad para llevar a cabo estrategias de control y gestión forestal a nivel regional.

Por otro lado, como los resultados del capítulo 6 sugieren, el éxito en la propagación, colonización e invasión de nuevas áreas forestales por las especies de Acacia no se debe a un único mecanismo, sino más bien a un grupo de procesos interrelacionados. El área de distribución potencial de las dos especies de Acacia en la zona de estudio se ve limitada principalmente por filtros ambientales, por la conectividad entre poblaciones próximas que posibiliten la propagación de propágulos, y por el régimen de perturbación del hábitat receptor, relacionado con la disponibilidad de recursos y nuevos nichos vacantes (Alpert et al. 2000). Sin embargo, este estudio también ha identificado factores bióticos asociados con la estructura y variables de masa de los hábitats forestales receptores como determinantes del potencial grado de invasión del hábitat receptor.


150

En el capítulo 6 se ha encontrado por ejemplo una relación positiva entre la riqueza del hábitat y el grado de invasión por especies invasoras. En teoría, las comunidades más diversas deberían ser menos susceptibles a la invasión ya que el uso más completo de los recursos debería de reducir la disponibilidad de nichos que facilitaran la entrada de especies invasoras en el sistema (Tilman 1997). Esta relación contraintuitiva es un resultado recurrente en estudios previos y ha sido ampliamente discutido (Levine y D`Antonio 1999; Stohlgren et al. 2003), sin embargo, a día de hoy no existe una base empírica o teórica sólida que apoye una relación positiva o negativa directa entre la riqueza de especies nativas y invasibilidad. Como está ampliamente documentado (Barbeito et al. 2008), la estructura de los ecosistemas forestales está directamente relacionada con la diversidad, la productividad, el crecimiento neto y la estabilidad de los ecosistemas forestales. Por lo tanto, no sería de extrañar que la composición estructural de los ecosistemas receptores también jugara un papel importante en su vulnerabilidad a ser invadidos. Las interacciones bióticas en hábitats con diferentes grados de mezcla de especies pueden explicar parte del éxito de la especie alóctona introducida. En este contexto, y aunque la comprensión de los factores que hacen susceptible a un hábitat a ser invadido por especies exóticas es clave a la hora de definir estrategias de control y conservación, los mecanismos por los que un ecosistema forestal es invadido con éxito por especies exóticas es aún poco conocido (Pyšek y Chytrý 2014). La información detallada sobre estructura arbórea de los hábitats forestales receptores a la invasión que se pueden derivar del IFN (Alberdi al. 2014) abre un marco de posibilidad de investigación en el futuro (Hernández y Barbeito 2015).

7.3. Referencias bibliográficas

Alberdi I, Cañellas I, Condés S (2014) A long-scale biodiversity monitoring methodology for Spanish National Forest Inventory. Application to Álava region. Forest Syst. 23(1): 93-110

Alpert P, Bone E, Holzapfel C (2000) Invasiveness, invasibility and the role of environmental stress in the spread of non-native plants. Persp.Plant Ecol. Evol. S. 3: 52–66.

Ameztegui A, Coll L (2011) Tree dynamics and co‐existence in the montane–sub‐alpine ecotone: the role of different light‐induced strategies. J. Veg. Sci. 22(6): 1049-1061.

Ameztegui, A, Brotons L, Coll L (2010) Land-use changes as major drivers of mountain pine (Pinus uncinata Ram.) expansion in the Pyrenees. Glob. Ecol. Biogeog.19: 632-641.

Barbeito I, Pardos M, Calama R, Cañellas I (2008) Effect of stand structure on Stone pine (Pinus pinea L.) regeneration dynamics. Forestry. 81(5): 617-629.

Benito-Garzón M, Sánchez de Dios R, Sainz Ollero H (2008) Effects of climate change on the distribution of Iberian forests. Appl. Veg. Sci. 11: 169–178.


151


Berg P, Moseley C,Haerter JO (2013) Strong increase in convective precipitation in response to higher temperatures. Nat. Geosci. 6(3): 181-185.

Bodin J, Badeau V, Bruno E, Cluzeau C, Moisselin JM, Walther GR, Dupouey JL (2013) Shifts of forest species along an elevational gradient in Southeast France: climate change or stand maturation? J. Veg. Sci. 24: 269–283.


Chauchard S, Carcaillet C, Guibal F (2007). Patterns of Land-use Abandonment Control Tree-recruitment and Forest Dynamics in Mediterranean Mountains. Ecosystems. 10 (6): 936-948.

Coop JD, Givnish TJ (2007) Spatial and temporal patterns of recent forest encroachment in montane grasslands of the Valles Caldera, NewMexico, USA. J. Biogeogr. 34: 914–927.

Costa M; Morla C, Sainz H (Eds.) (1997) Los Bosques ibéricos. Planeta. Barcelona.

de la Cueva AV, Quintana JR, Cañellas I (2012) Fire activity projections in the SRES A2 and B2 climatic scenarios in peninsular Spain. International J. of Wildland Fire. 21: 653-665.

Debussche M, Lepart J, Dervieux A (1999) Mediterranean landscape changes: evidence from old postcards. Glob. Ecol. Biogeog. 8: 3–15.

Estiarte M, Peñuelas J (2015). Alteration of the phenology of leaf senescence and fall in winter deciduous species by climate change: effects on nutrient proficiency. Glob. Change Biol. 21(3): 1005-1017.

FAO (2001) Global forest resources assessment. State of the World’s Forests 2001. Food and Agriculture Organization of the United Nations. Forestry Paper. Rome.200 pp.

FAO (2015) Food and Agriculture Organization of the United Nations. Global forest resources assessment 2015: Main report. How are the world’s forests changing? www.fao.org/forestry/fra

Gallardo A, Maestre FT (2008) Métodos geoestadísticos para el análisis de datos ecológicos espacialmente explícitos. En: Maestre FT, Escudero A, Bonet A (Eds.) Introducción al Análisis Espacial de Datos en Ecología y Ciencias Ambientales: Métodos y Aplicaciones. Universidad Rey Juan Carlos, Madrid. pp..215–272.

http://www.fao.org/forestry/fra


152



Gellrich M, Zimmermann NE (2007) Investigating the regional-scale pattern of agricultural land abandonment in the Swiss mountains: a spatial statistical modelling approach. Landscape Urban Plan 79: 65–76.

Gómez-Aparicio L, García-Valdés R, Ruiz-Benito P, Zavala MA (2011) Disentangling the relative importance of climate, size and competition on tree growth in Iberian forests: implications for management under global change. Glob Change Biol., 17: 2400-2414.


Gottfried M, Pauli H, Futschik A et al. (2012) Continent-wide response of mountain vegetation to climate change. Nature. 2(2): 111-115.

Hampe A, Petit R (2005) Conserving biodiversity under climate change, the rear edge matters. Ecol. Lett. 8 (5): 461-467.

Hanewinkel M, Peyron JL (2014). Tackling climate change—the contribution of scientific knowledge in forestry. Ann. For. Sci. 71(2): 113-115.

Hérnandez L, Cañellas I, Alberdi I, Vallejo R, Montes F (2013) ¿Se están produciendo cambios en la distribución de los bosques? En: Montes: Servicios y Desarrollo rural. (Ed.: Sociedad Española de Ciencias Forestales). 6CFE01-424. ISBN978-84-937964-9-5.

Hernández L, Cañellas I, Alberdi I, Torres I, Montes F (2014a). Assessing changes in species distribution from sequential large-scale forest inventories. Ann. For. Sci. 1-11.

Hernández L, Martínez-Fernández J, Cañellas I, de la Cueva AV (2014b) Assessing spatio-temporal rates, patterns and determinants of biological invasions in forest ecosystems. The case of Acacia species in NW Spain. Forest Ecol. Manag. 329: 206-213.

Hernández L, Barbeito I (2015) New mixtures: are richer forests more vulnerable to non-native species invasions? Eumixfor Cost Action Workshop. Brasov. Romania.

Hernández L, Sánchez de Dios R, Montes F, Sainz-Ollero H, Cañellas I. Exploring range shifts of contrasting tree species along bioclimatic and elevational gradients. Forest Ecol. Manag. En rev.1


153

Hernández L, Sánchez de Dios R, Montes F, Sainz-Ollero H, Cañellas I. Tracking Fagus sylvatica leading edge in North-Western Iberia: Holocene migration inertia, forest recovery and recent global change. Persp. Plant Ecol. Evol. S. En rev.2

Higgins SI, Richardson DM (1996) A review of models of alien plant spread. Ecol Mod. 87: 249–265.

Higgins SI, Richardson DM, Cowling RM (2001) Validation of a spatial simulation model of a spreading alien plant population. J. App. Ecol. 38: 571-584.

Holmes PM, Cowling RM (1997) The effects of invasion by Acacia saligna on the guild structure and regeneration capabilities of South African fynbos shrublands. J. App. Ecol. 34: 317-332.

IPCC (2013) Climate Change 2013: The Physical Science Basis. Contribution of Working Group 1 to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. http://www.ipcc.ch/


Kräuchi N, Kienast F (1993) Modelling subalpine forest dynamics as influenced by a changing environment. Water Air Soil Pollut 68: 185-197.

Lasanta T (2002) Los sistemas de gestión en el Pirineo central español durante el siglo XX: del aprovechamiento global de los recursos a la descoordinación espacial en los usos del suelo. Ager. 2:173–196.

Lenoir J, Gégout JC, Pierrat JC, et al. (2009) Differences between tree species seedling and adult altitudinal distribution in mountain forests during the recent warm period (1986–2006). Ecography. 32: 765–777.


Levine JM, D´Antonio CM (1999) Elton revisited: a review of evidence linking diversity and invasibility. Oikos. 87: 15-26.

Levine JM, Vilà M, D'Antonio CM, Dukes JS, Grigulis K, Lavorel S (2003) Mechanisms underlying the impacts of exotic plant invasions. Proc R Soc Lond B Biol Sci. 270: 775-781.

Loiselle BA, Howell CA, Graham CH, Goerck JM, Brooks T, Smith KG, Williams PH (2003) Avoiding Pitfalls of Using Species Distribution Models in Conservation Planning. Cons. Biol. 17: 1591-1600.




154

Mandallaz D (2000) Estimation of the spatial covariance in universal kriging: application to forest inventory. Environ. Ecol. Stat. 7(3): 263-284.

Martinez-Vilalta J, Lopez BC, Adell N, Badiella L, Ninyerola M (2008) Twentieth century increase of Scots pine radial growth in NE Spain shows strong climate interactions. Glob. Change Biol. 14:, 2868-2881.

MEA (2005) Millennium Ecosystem Assessment. Ecosystem and human well-being: biodiversity synthesis. World Resources Institute, Washington D.C.

Miller J, Franklin J, Aspinall R (2007) Incorporating spatial dependence in predictive vegetation models. Ecol. Model 202: 225–42.

Montero G, Serrada R (Eds.) (2013) La situación de los bosques y el sector forestal en España - ISFE 2013. (Ed. Sociedad Española de Ciencias Forestales). Lourizán (Pontevedra). España.

Morales CL, Traveset A (2009) A meta-analysis of impacts of alien vs. native plants on pollinator visitation and reproductive success of co-flowering native plants. Ecol. Lett. 12(7): 716-728

Nabuurs, G. J., Schelhaas, M. J., Mohren, G. M. J. & Field, C. B. (2003) Temporal evolution of the European forest sector carbon sink from 1950 to 1999. Glob. Change Biol. 9: 152-160.

Ohlemüller R, Gritti ES, Sykes MT, Thomas CD (2006) Quantifying components of risk for European woody species under climate change. Glob. Change Biol. 12(9): 1788-1799.


Poljanec A, Ficko A, Boncina A (2010) Spatiotemporal dynamic of European beech (Fagus sylvatica L.) in Slovenia, 1970–2005. Forest Ecol. Manag. 259(11): 2183-2190.

Poyatos R, Latron J, Llorens P (2003) Land use and land cover change after agricultural abandonment - The case of a Mediterranean Mountain Area (Catalan Pre-Pyrenees). Mt. Res. Dev. 23: 362–368.

Purves DW (2009) The demography of range boundaries versus range cores in eastern US tree species. Proceedings of the Royal Society B: Biol. Sci. 276: 1477-1484.

Pyšek P, Jarošik V, Kuèera T (2002) Patterns of invasion in a temperate nature reserves. Biological Conservation 104: 13-24.

Pyšek P, Chytrý M (2014). Habitat invasion research: where vegetation science and invasion ecology meet. J. Veg. Sci. 25(5): 1181-1187.



155

Rubiales JM, García-Amorena I, Hernández L, Génova M, Martínez F, Gómez Manzaneque F, Morla C (2010) Late Quaternary dynamics of pinewoods in the Iberian mountains. Rev. Palaeo. Palyn.162: 476–491.



Sánchez de Dios R, Benito-Garzón M, Sainz-Ollero H (2009) Present and future extension of the Iberian Submediterranean territories as determined from the distribution of marcescent oaks. Plant Ecol. 204: 189-205.

Sanz-Elorza M, Dana ED, González A, Sobrino E (2003) Changes in the high-mountain vegetation of the Central Iberian Peninsula as a probable sign of global warming. Ann. Bot. London. 92: 273–280.

Segurado P, Araújo MB (2004) An evaluation of methods for modelling species distributions. J. Biogeog 31: 1555–1568.

Stohlgren T, David T, Barnett J, Kartesz JT (2003) "The rich get richer: patterns of plant invasions in the United States." Front.Ecol.Env. 1: 11-14.

Thuiller W, Lavorel S., Auraújo M.B (2005) Niche properties and geographical extent as predictors of species sensitivity to climate change. Global Ecol. Biogeogr. 14, 347–357.

Tilman D (1997) Community invasibility, recruitment limitation, and grassland biodiversity. Ecology. 78: 81–92.



Vallejo R, Villanueva JA (2002) El banco de datos de la naturaleza. En: El Inventario Forestal Nacional como elemento clave para la gestión forestal sostenible (Bravo et al. eds.). Fundación Gral. de la Universidad de Valladolid, pp: 19-35.

Vicente-Serrano SM, Lasanta T, Romo A (2005) Analysis of spatial and temporal evolution of vegetation cover in the Spanish central Pyrenees: Role of human management. Enviro. Manage.34: 802–818.



156

Wangen SR, Webster CR (2006) Potential for multiple lag phases during biotic invasions: reconstructing an invasion of the exotic tree Acer platanoides. J Appl Ecol. 43(2): 258-268.

WMO (2013) The Global Climate 2001–2010: A Decade of Climate Extremes – Summary Report. World Meteorological Organisation, Geneva.

Yoo EH, Trgovac AB (2011) Scale effects in uncertainty modeling of presettlement vegetation distribution. Int. J. Geogr. Inf. Sci 25: 405-421.

8. Conclusiones

CAPÍTULO8: CONCLUSIONES

159

1. La presente Tesis Doctoral pone de manifiesto la idoneidad del uso de las bases de datos de los Inventarios Forestales Nacionales para la detección y seguimiento de la dinámica de los ecosistemas forestales en un contexto de cambio global. Los resultados de esta tesis enfatizan la potencialidad de esta herramienta para testar hipótesis biogeográficas bajo escenarios de cambio climático y analizar la evolución reciente de invasiones biológicas en ecosistemas forestales, contribuyendo así a la comprensión teórica y práctica del impacto reciente del cambio global sobre los bosques.

2. La aproximación metodológica propuesta basada en indicator block kriging proporciona una herramienta para la detección temprana de cambios en la distribución de especies forestales a lo largo de gradientes ambientales. Esta técnica permite construir tests de significación que incorporan la incertidumbre ligada a las diferencias en la intensidad de muestreo, permitiendo así la comparación de los sucesivos Inventarios Forestales Nacionales.

3. La superficie forestal asociada a P. sylvestris, F. sylvatica, Quercus ilex y Quercus subpyrenaica se ha incrementado en Navarra desde 1971 a 2010, siendo Q. ilex y Q. subpyrenaica las especies que tienen mayores ratios de expansión. A lo largo del gradiente bioclimático Templado-Mediterráneo P. sylvestris, F. sylvatica y Q. subpyrenaica muestran un avance en su área de distribución desde las zonas de transición Submediterránea hacia zonas más frías y húmedas. Simultáneamente el rango altitudinal de F. sylvatica y Q. subpyrenaica también muestra un desplazamiento hacia mayores elevaciones. Mientras que P. sylvestris parece retraer su distribución en altitud total desplazando su rango altitudinal hacia menores elevaciones. Q. ilex en cambio, parece tener una expansión general de su área de distribución.

4. Como consecuencia de los cambios en el área de distribución observados, se ha producido un aumento del área donde las especies estudiadas coexisten. Variaciones en la composición de los bosques como resultado de la expansión o retracción de estas especies podrían tener profundas implicaciones en la diversidad y distribución de otras especies asociadas a sus ecosistemas, así como en la funcionalidad y servicios ecosistémicos asociados.

5. Los resultados confirman hipótesis previas como la tendencia hacia una futura “Mediterraneización” de la zona de transición Submediterránea. También confirma la sensibilidad al cambio climático de especies de restringida distribución como es la especie Submediterránea (Q. subpyrenaica), que es la que muestra mayores alteraciones en su rango. Por último, los resultados apoyan hipótesis biogeogáficas recientes que proponen que los cambios en la distribución de la especies se están produciendo a lo largo de todas su distribución y no sólo en las poblaciones situadas en los límites climáticos.

CAPÍTULO8: CONCLUSIONES

160

6. Algunos de los cambios en la distribución de especies encontrados no coinciden totalmente con las proyecciones de su distribución para un futuro próximo bajo escenarios de cambio climático. Estos resultados sugieren la necesidad de un nuevo modelo conceptual para explicar los cambios en distribución de especies recientes donde se considere, entre otros, el efecto combinado del actual cambio climático, cambios en el uso de territorio, la disponibilidad de nicho y el manejo forestal histórico al que se han visto sometidas las especies forestales.

7. Contrariamente a lo proyectado bajo las actuales condiciones de cambio climático, F. sylvatica está expandiendo su distribución en el noroeste peninsular, donde está incrementando su dominancia y colonizando zonas de menor altitud gracias a la recuperación de terreno forestal tras cambios en el uso del territorio y siguiendo la inercia de su expansión Holocena. Por lo tanto, la región biogeográfica Atlántica Ibérica se puede considerar como un frente de avance o “leading edge” actual de la distribución de F. sylvatica.

8. El análisis de los patrones de reclutamiento y estructura poblacional identifican una relación inter-específica entre Quercus petraea y F. sylvatica que afecta negativamente a la primera, confirmando el actual reemplazo de las masas de Q. petraea por F. sylvatica.

9. El análisis de la evolución de Acacia dealbata y Acacia melanoxylon confirma la rápida expansión de las dos especies entre 1998 y 2008 en los bosques del noroeste peninsular, con ratios de expansión anual aproximadamente de un 0,1%. Las dos especies están aumentando su presencia en la mayor parte de los bosques del área de estudio, empezando a dominar en el estrato de regeneración de algunos de ellos.

10. Factores ambientales y la conectividad entre poblaciones de Acacia son los determinantes más importantes asociados a la expansión de estas especies en nuevas áreas. Además, la combinación de procesos perturbadores y factores bióticos relacionados con la estructura de la masa (área basal total, riqueza de especies y cobertura arbórea) parecen estar asociados con la vulnerabilidad o resistencia de algunos tipos de bosque a su expansión.

11. La etapa temprana de invasión detectada en las dos especies de Acacia en la zona de estudio pone en evidencia la potencialidad de ambas especies para continuar con su expansión. La combinación de este factor y el elevado régimen de perturbaciones (principalmente incendios) en esta región, podría ser crítico en la futura configuración de los paisajes forestales en el noroeste peninsular.

Anexo1

ORIGINAL PAPER

Assessing changes in species distribution from sequential

large-scale forest inventories

Laura Hernández & Isabel Cañellas & Iciar Alberdi &

Iván Torres & Fernando Montes

Received: 27 August 2012 /Accepted: 13 June 2013 /Published online: 12 July 2013# INRA and Springer-Verlag France 2013

Abstract

& Context It is assumed that global change is already affect-

ing the composition, structure and distribution of forest

ecosystems; however, detailed evidences of altitudinal and

latitudinal shifts are still scarce.

& Aims To develop a method based on National Forest In-

ventory (NFI) to assess spatio-temporal changes in species

distributions.

& Methods We develop an approach based on universal

kriging to compare species distribution models from the

different NFI cycles and regardless of the differences in the

sampling schemes used. Furthermore, a confidence interval

approach is used to assess significant changes in species

distribution. The approach is applied to some of the south-

ernmost populations of Pinus sylvestris and Fagus sylvatica

in the Western Pyrenees over the last 40 years.

& Results An increase of the presence of the two species in the

region was observed. Scots pine distribution has shifted about

1.5 km northwards over recent decades, whereas the European

beech has extended its distribution southwards by about 2 km.

Furthermore, the optimum altitude for both species has risen

by about 200 m. As a result, the zone in which the two species

coexist has been enlarged.

& Conclusions This approach provides a useful tool to com-

pare NFI data from different sampling schemes, quantifying

and testing significant shifts in tree species distribution over

recent decades across geographical gradients.

Keywords National Forest Inventory . Universal kriging .

Shifts . Pinus sylvestris . Fagus sylvatica . Pyrenees

1 Introduction

The impact of global change on forests is emerging as a major

concern for the twenty-first century society (FAO 2000).

Forestry and conservation po1icy-makers need to understand

how the distributions of species are affected by global change

in order to tackle its effects on forests. Although it is assumed

that global change is already affecting the composition, struc-

ture and distribution of forest ecosystems at different spatial

and temporal scales, reported evidences of altitudinal and

latitudinal shifts are still scarce (Peñuelas and Boada 2003).

Large-scale forest surveys such as the National Forest In-

ventory (NFI) can provide a valuable tool for monitoring

changes in forest biodiversity and can be of great relevance in

the conservation and management of natural resources. In the

second half of the twentieth century, most developed countries

started undertaking periodical NFIs covering the entire forest

area. Although NFIs were primarily designed to estimate forest

resources, they are increasingly being employed to assess the

impact of global change on forest ecosystems (Thuiller et al.

2003). When permanent plots are inventoried sequentially,

Handling Editor: Erwin Dreyer

Contribution of the co-authors Isabel Cañellas coordinated the

associate research projects. Iciar Alberdi provided access to all NFI

databases. Fernando Montes and Laura Hernández conceived, designed

and run the data analysis. Fernando Montes also supervised the work.

Laura Hernández conducted manuscript writing. Fernando Montes,

Isabel Cañellas, Iciar Alberdi and Iván Torres conducted manuscript

reviewing.

L. Hernández (*) : I. Cañellas : I. Alberdi : F. Montes

INIA-CIFOR, Ctra. La Coruña, km 7.5, 28040 Madrid, Spain

e-mail: [email protected]

L. Hernández

E.T.S.I. Montes, Polytechnic University of Madrid, 28040 Madrid,

Spain

I. Cañellas

Sustainable Forest Management Research Institute,

University of Valladolid-INIA, 34004 Palencia, Spain

I. Torres

UCLM, University of Castilla-La Mancha, 45071 Toledo, Spain

Annals of Forest Science (2014) 71:161–171

DOI 10.1007/s13595-013-0308-6

forest evolution can be analysed through direct comparison

between inventories (Poljanec et al. 2010; Vilá-Cabrera et al.

2011); otherwise, species–environment relationships should be

analysed from species distribution models (SDMs) (Guisan

and Zimmermann 2000). One problem that arises when com-

paring species distributions is whether or not the changes are

related to the different sampling schemes employed across

successive NFIs. This shortcoming can be dealt with in a

similar way to change-of-support problems by using block

kriging techniques (Yoo and Trgovac 2011).

Environmental variables such as precipitation, temperature

and elevation exhibit spatial dependence, which is partly

responsible for the spatial pattern observed in vegetation

distribution (Miller et al. 2007). In addition, spatial autocorre-

lation can also result from ecological processes involved in

forest dynamics occurring at different scales, leading to spatial

continuity in species distribution (Bellehumeur and Legendre

1998). As a consequence, spatial dependence in biogeograph-

ical data has recently been identified as an important area of

SDM research, especially for species with a broad distribu-

tion, which are generally better modelled by including spatial

autocorrelation in the model (Segurado and Araújo 2004).

Universal kriging may be considered a type of spatial regres-

sion, incorporating spatial autocorrelation as well as relation-

ships between environmental variables in the spatial predic-

tion of species distribution (Montes and Ledo 2010). It may be

important to consider this latter quality of universal kriging in

SDMs as species distribution varies over environmental gra-

dients. The altitudinal range where species with a broad lati-

tudinal distribution appear varies as a result of a decline in

temperature with increasing elevation and more northerly

latitude. This variation in modelled relationships over space,

known as non-stationarity, shows interdependence with spa-

tial scale, so local and landscape approaches may reveal

species–environment relationships that global models average

out (Osborne et al. 2007; Miller and Hanham 2011).

The Iberian Peninsula, like other Mediterranean regions,

constituted an important refuge for flora during the Quaternary

glaciations. This fact, together with the current interaction of the

Eurosiberian, Mediterranean and Alpine biogeographical

zones, made the Iberian Peninsula a biodiversity hotspot where

many species find the limits of their distribution; many of them

highly fragmented in different mountain ranges. This is the case

of Scots pine (Pinus sylvestris L.) and European beech (Fagus

sylvatica L.), two species widely distributed across Europe and

which have some of their southernmost populations in the

mountain ranges of the Iberian Peninsula. The most recent

climatic records for theMediterranean region reveal an increase

in the number of heat waves and droughts during the last

century (Alpert et al. 2008). This pattern, together with the

important impact of human activity and the sensitivity of the

ecosystems to contrasts in climatic conditions, makes the

mountain ecosystems of the Iberian Peninsula particularly

vulnerable to the effects of global change (Engler et al. 2011).

It is expected that woody species will respond to changes in

environmental conditions by migration towards suitable areas

or through adaptation to the new conditions. Otherwise, they

may decline or even become locally extinct (Aitken et al. 2008).

Taking into account not only these premises but also the results

of previous studies in European mountain forests (Lenoir et al.

2008; Poljanec et al. 2010), we hypothesize an upward shift in

the distribution of Scots pine and European beech at regional

scale over recent decades in the Western Pyrenees.

The aim of this study is to develop a new method based on

universal kriging models to assess spatio-temporal changes in

species distribution from large-scale sequential forest inventories

regardless of the differences in the sampling schemes used. For

testing the usefulness of this approach, we also aim to answer the

following questions: (a) Is it possible to detect significant shifts

in the spatial distribution of P. sylvestris and F. sylvatica in the

Western Pyrenees over recent decades? (b) Can we quantify the

extent of these shifts across geographical gradients?

2 Material and methods

2.1 Study area

This study focuses on the Western Pyrenees in the Navarra

region of Spain between E.D. 50 Universal Transverse Merca-

tor (UTM) 4.734 S–4.764 N and 642 E–685W. The study area,

which covers 741,60 km2, is characterized by a steep altitudinal

gradient, which ranges from approximately 600 masl at the

southern limit to 2,400 masl at the northern limit. Due to its

geographical location, where several biogeographical regions

overlap, the region has a variable climate. At more northerly

latitudes, where the highest altitudes are found and the oceanic

influence of the Cantabrian Sea is less notable, the mean annual

precipitation is 1,063 mm, and the mean annual temperature is

about 8 °C, which reflects the typical continental characteristics

of the Pyrenees. By contrast, in the mid latitudes, mean annual

precipitation levels are higher (1,300 mm), as are the tempera-

tures (10.5 °C). The southernmost area, however, is influenced

by a sub-Mediterranean climate with moderate summer

drought. Hence, the mean annual rainfall in this part of the

study area is lower (1,044 mm), although the mean annual

temperature is higher (11 °C) (Ninyerola et al. 2005).

Scots pine and beech form the naturally occurring uneven-

aged forests that dominate the study area. In the Iberian

Peninsula, the majority of Scots pine stands can be found in

the more continental, south-central Pyrenees; becoming scarcer

in the northwest (Costa et al. 1997), i.e. in the areas influenced

by the Cantabrian Sea, where beech stands are more common.

Beech and Scots pine occur in the montane altitudinal belt

between 600 and 1,600 masl, with beech occupying higher

162 L. Hernández et al.

and shadier slopes. In the lower montane altitudinal belt, P.

sylvestris can be found alongside marcescent Quercus species

like Quercus faginea Lam., Quercus pyrenaica Willd. and

Quercus pubescens Willd., whereas above these forests, the

sub-alpine belt (1,600–2,300 masl) is dominated by Pinus

uncinata Ram. and the alpine belt (over 2,300 masl) is the

domain of herbaceous, sometimes shrubby vegetation. Both

species can co-occur in some zones, where European beech is

found in more mesic conditions and Scots pine where more

xeric conditions prevail.

2.2 NFI dataset

The Spanish NFI is a forest monitoring system covering the

entire forested area of Spain. Inventories are undertaken

approximately every 10 years with an intensity of one sam-

pling point every 1 km2. At each sampling point, trees are

measured in concentric circular plots with increasing radii

from 5 to 25 m. In the 5-m-radius plot, trees with a diameter

at breast height (dbh) ≥75 mm and trees with 25≤dbh

<75 mm (saplings) are measured, while the recruitment

density is monitored (individuals of less than 1.30 m height

or dbh<25 mm). In the 10-m-radius plot, trees with

dbh≥125 mm are measured; in the 15-m-radius plot, trees

with dbh≥225 mm are measured; and in the 25-m-radius

plot, trees with dbh≥425 mm are measured. In the Navarra

region, four cycles of the Spanish NFI have been completed.

Although the NFI2 (1986–1995) and NFI3 (1997–2006)

cycles have been used in numerous studies (Vilá-Cabrera

et al. 2011), research involving the comparison of data from

the four NFIs conducted since 1960 has never been under-

taken. Since different sampling designs were used in the

various NFI cycles, the number of plots and locations varied

from one to another (Table 1).

In order to analyse changes in the spatial distribution of

species using the NFI databases, presence/absence indicator

variables with a value of 1 if a species was present in a plot

and 0 if it was absent were defined for both species. The

presence of a species in a given plot was registered according

to size, recruitment, saplings and trees with dbh ≥75 mm.

Therefore, the presence/absence indicator variable is a re-

gionalized variable that (for the xy location of the plot centre)

will be 1 if there are individuals within the respective radius

and 0 if not. The variable-radius sampling design of the NFI

plots smoothes the sampling estimator surface and mini-

mizes the variance of the estimator (Williams 2001).

2.3 Universal kriging

Universal kriging (UK) is a spatial regression procedure that

incorporates the spatial autocorrelation in the estimation of a

regionalized variable. In the universal kriging model, the

value of the variable Z(s0) at location s0 is expressed as a

polynomial of the auxiliary variables ∑k¼0

p

βk f k s0ð Þ (i.e. mean

function), which account for a spatial trend, and a spatially

autocorrelated residual process δ(s0) (Matheron 1969):

Z s0ð Þ ¼X

k¼0

p

βk f k s0ð Þ þ δ s0ð Þ ð1Þ

Universal kriging was used to model the relationships

between species presence/absence indicator variables from

the NFI dataset and the latitude (y, E.D. 50 UTM, in

kilometres), elevation (h, in metres above sea level) and

exposure (cos(8), the cosine of the aspect azimuth) derived

from a 25 m resolution digital elevation model (DEM) of

Spain, for each inventory. In the case of the Scots pine

presence/absence indicator variable, exploratory analyses

indicated differences in the altitude at which presence pre-

diction was maximum across the latitudinal gradient; there-

fore, a polynomial of order 2 including h2, h and the h×y

interaction was used to model the latitudinal variation in

altitudinal maximum in the mean function. The model per-

formance was notably improved by incorporating the cosine

of the aspect azimuth and the cos(8)×y interaction to model

the latitudinal variation of this variable. Therefore, the mean

function for the Scots pine presence/absence indicator vari-

able was the following:

X

k¼0

p

βk fk s0ð Þ ¼ β0 þ β1hþ β2h2 þ β3hyþ β4cos φð Þ

þ β5cos φð Þy ð2Þ

In the case of the beech presence/absence indicator vari-

able, preliminary analyses did not show any interaction

between altitude and latitude. However, the altitudinal dis-

tribution clearly showed a maximum presence probability. In

Table 1 Spanish National For-

est Inventory (NFI) cycles sam-

pling periods, source of infor-

mation used for forest area as-

sessment, sampling period and

number of plots measured in the

study area in each NFI cycle

Cycle NFI

sampling period

Forest area

information source

Study area

sampling period

Study area total

number of plots

NFI1 1965–1974 Aerial photographs from 1969 1971 203

NFI2 1986–1996 Map of land use 1:50,000 1989–1990 241

NFI3 1997–2007 Spanish forestry map 1:50,000 1999–2000 437

NFI4 2008–2017 Spanish forestry map 1:25,000 2009–2010 497

Assessing changes on forests from NFI 163

addition, a latitudinal trend was noted for cos(8); thus, the

mean function finally chosen was the following:

X

k¼0

p

βk f k s0ð Þ ¼ β0 þ β1hþ β2h2 þ β3yþ β4 cos φð Þ

þ β5cos φð Þy ð3Þ

The universal kriging prediction p(Z,s0) of the regional-

ized variable Z(s0) can be interpreted as the probability of the

presence of each species analysed in s0 (Goovaerts 1994) and

is calculated as:

p Z; s0ð Þ ¼X

i¼1

n

λi⋅Z sið Þ ð4Þ

under the unbiasedness conditions:

X

i¼1

n

λi⋅ f k si; tið Þ ¼ f k s0; t0ð Þ k ¼ 0…p f0si; tið Þ≡1 ∀ i

ð5Þ

where Z(si) is the value of the variable at n sampling points.

The variance of the universal kriging prediction (σUK(s0))

was calculated as in Cressie (1993).

The variogram that describes the degree of spatial depen-

dence of the regionalized variable was modelled using the

spherical variogram defined by the following parameters: the

nugget (the semivariance value at the origin), the autocorrela-

tion range (the distance at which the semivariance stabilizes)

and the sill (the semivariance for distances greater than the

autocorrelation range). One of the crucial steps when using the

universal kriging is the variogram parameters and β coefficient

fitting. The variance least square (VLS) method used for the

variogram and mean function estimation seems more suitable

than maximum likelihood methods for the presence/absence

indicator variable because it does not require multi-Gaussian

distributional assumption (Montes and Ledo 2010).

The kriging prediction may give values outside the range

[0,1], although these deviations are usually of small magni-

tude (Goovaerts 1994). There are different techniques used

for constraining the indicator kriging prediction to [0,1]

(Tolosana-Delgado et al. 2008). For the comparison between

inventories pursued in this study, such constrains are not

necessary; however, to map the occurrence of the species

(Fig. 1), the kriging indicator predictions outside [0,1] were

set to the nearest bound (upward–downward correction;

Deutsch and Journel 1992).

2.4 Cross-validation of the universal kriging model

A leaving-one-out cross-validation was used to determine

the prediction bias and accuracy of the prediction variance

estimates of the fitted model (Cressie 1993). Differences

between the predicted and observed values (SEE, sum of

estimation errors) were used to assess the bias.

To test the accuracy of the prediction variance estimation,

the ratio of the variance of the residuals to the prediction

variance was estimated through the fitted variogram

(σUK(s0)). Variance of the standardized estimation errors

(VSEE) was calculated. Values close to 1 indicated a similar

distribution for the observation and prediction errors.

2.5 Testing the significance of changes in species

distribution between inventories

To test the statistical significance of the changes in species

distribution along the gradients of the analysed variables

between inventories, a novel significance test based on block

kriging variances (Isaaks and Srivastava 1989) was devel-

oped. The block kriging estimates the mean of the variable

for a determined region (or block) through averaging the

universal kriging weights λi for the points resulting from

the discretizacion of the block. The block mean and the block

prediction variance were estimated for the strata determined

by the classification of the forest area (a) in ten altitude

classes, (b) in seven latitude classes and (c) in nine exposure

classes, which constitute the blocks in this case. The block

mean of the presence/absence indicator variable can be

interpreted as the mean probability of species presence in

each stratum.

As kriging provides unbiased estimators, the block mean

does not depend on the number of plots in each inventory,

although the kriging variance depends on the spatial auto-

correlation and the distance between samples, which varies

with the plot density. To assess differences between inven-

tories at 95 % confidence level, confidence intervals for the

presence/absence block prediction at each altitude, latitude

and exposure class were calculated from the kriging variance

for each inventory date (Isaaks and Srivastava 1989):

p Z; s0ð Þ−1:96σUK s0ð Þ; p Z; s0ð Þ þ 1:96σUK s0ð Þð Þ

To determine the significance of the differences between

the indicator variable kriging predictions of the different NFI

cycles, these were compared with the size of the confidence

intervals. The use of confidence intervals allows the differ-

ences to be presented as a range of magnitudes, rather than

just assess the statistical significance (Katz 1992). The ab-

sence of overlapping between the confidence intervals at two

different inventory cycles gives a conservative test at 95 %

confidence for the difference between the predicted values.

Note that the presence/absence prediction mean is estimated

for each class, so bias due to differences in forest area

between inventories is avoided.


2.6 Mapping changes in species distribution

To graphically assess the species distribution at each inven-

tory date, a threshold probability indicating the species pres-

ence was set at 0.5 (Montes et al. 2005), checking that the

proportion of predicted points above the indicator threshold

was unbiased. The indicator variables of both species were

inferred in a 500×500-m grid covering the forest area at each

inventory cycle to build the prediction maps.

2.7 Software used

ArcGis 9.2 was used to interpolate 25-m resolution DEM

from the topographical map of Spain and to derive latitude,

altitude and cosine of the aspect azimuth at sample and

prediction points. Geostatistical analyses were performed

using a Microsoft® VisualBasic® application developed by

the authors.

3 Results

3.1 Variogram models

The variogram models fitted for the Scots pine and beech

differed considerably. The nugget effect was smaller for the

Scots pine (Table 2), whereas the spatial autocorrelation

range was larger for beech (Table 3). The autocorrelation

range in the Scots pine model progressively diminished

from approximately 3 km in 1971 to 0.86 km in 2010,

which indicates a decrease in the spatial continuity of

the Scots pine indicator variable in the study area. In

contrast, the progressively larger autocorrelation ranges

of beech (from 9 km in 1971 to 22 km in 2010)

indicated increasing continuity in the spatial distribu-

tion. The analyses of cross-validation residuals showed

that the universal kriging models fitted performed satis-

factorily in terms of bias and kriging variance estima-

tion in both cases (Tables 2 and 3).

Fig. 1 Universal kriging predictions for the distribution areas of P.

sylvestris (black) and F. sylvatica (dark grey) and areas where both

species coexist (light grey) from NFI1 (1971), NFI2 (1989–1990), NFI3

(1999–2000) and NFI4 (2009–2010). Kriging was performed over a

500×500-m grid covering the surveyed areas, and a threshold of 0.5 for

the indicator variable value was used to define the presence/absence of

each species at each prediction point


3.2 Mapping forest range changes

The most notable changes in the universal kriging prediction

maps (Fig. 1) were the expansion of the Scots pine from 71 to

79 % of the plots from 1971 to 2010 and an increase from 53

to 72% in the case of beech. In addition, there were changes in

the distribution areas of Scots pine and beech. For instance, in

the NFI1, the two species predominantly occupied distinct

areas (coexisting in 31 % of the study area), whereas in the

NFI3 and NFI4, the area in which the two species coexisted

had increased to approximately 42 % of the study area.

3.3 Assessing species distribution changes between

inventories

Differences for a range of magnitude between species distri-

bution across the NFI are assessed through the confidence

intervals of the block kriging mean predictions for each of

the latitude, altitude and exposure classes considered. The

NFI1 and NFI2 show wider confidence intervals due to the

lower sampling density. The most notable differences for

both species arose at mid latitudes (4,747–4,752 km),

where the block mean kriging prediction for NFI3 and NFI4

falls outside the confidence intervals for the NFI1 and NFI2

(Fig. 2). These differences were significant at 95 % level (no

overlapping between prediction confidence intervals) for beech

between the NFI1 and the NFI4.

3.3.1 Latitudinal shifts

The results revealed changes in the latitudinal distribution

range of both species at the spatial scale and time frame

studied. The block kriging prediction for the ten latitudinal

classes suggests a major increase in the presence of Scots pine

in the mid-northern latitudes of the study area from 1971 to

2000 and in the northern latitudes from 2000 to 2010 (Fig. 2a).

This increase was concomitant with a decrease in Scots pine

presence prediction in southern latitudes from 2000 to 2010,

which indicates a possible shift in the distribution range of this

species towards northern latitudes within the study area. In

contrast, the presence of beech generally increased in the

region both in northern latitudes (mainly from 1971 to 1990)

and mid-southern latitudes (from 1990 to 2010) (Fig. 2b),

where this species was scarce in the first inventories.

3.3.2 Shifts in elevation

Figure 2c, d reveals a general shift in the presence/distribution

of both species towards higher elevations. In the case of Scots

pine, the optimum altitude (corresponding to the maximum

presence prediction values) was found at about 700 m in 1971

Table 2 Universal kriging models, VLS estimations of variogram pa-

rameters and β coefficients for P. sylvestris and explanatory variables;

elevation, square of the elevation, exposure (Expo) and the interaction

between latitude, elevation and exposure. Cross-validation sum of re-

siduals (SEE) and the residual variance/prediction variance ratio (VSEE)

are also shown

Pinus sylvestris Variogram parameters β coefficients of the auxiliary variables Cross-validation

Inventory Nugget Sill Range

(km)

β0 Elevation Elevation 2 Elevation

× latitude

Expo Expo

× latitude

SEE VSEE

NFI1 (1971) 0 0.177 3.042 0.565 0.003 −5.9 E-07 −5.5 E-07 31.319 −0.007 −0.0090 1.10

NFI2 (1989–1990) 0 0.161 1.134 0.063 0.134 −5.9 E-07 −2.7 E-05 45.307 −0.010 −0.0002 1.01

NFI3 (1999–2000) 0.042 0.102 0.895 −0.499 0.138 −1.1 E-06 −2.8 E-05 39.783 −0.008 −0.0005 1.01

NFI4 (2009–2010) 0.039 0.121 0.868 −0.240 0.125 −8.3 E-07 −2.6 E-05 43.433 −0.009 −0.0006 1.01

Table 3 Universal kriging model, VLS estimations of variogram pa-

rameters and β coefficients for Fagus sylvatica and explanatory vari-

ables; elevation, the square of the elevation, latitude, exposure (Expo)

and the latitudinal interaction with the exposure. The cross-validation

sum of residuals (SEE) and the residual variance/prediction variance

ratio (VSEE) are also shown

Fagus sylvatica Variogram parameters β coefficients of the auxiliary variables Cross-validation

Inventory Nugget Sill Range (km) β0 Elevation Elevation 2 Latitude Expo Expo × latitude SEE VSEE

NFI1 (1971) 0.081 0.098 14.569 −132.070 0.005 −2.1 E-06 0.027 47.276 −0.010 0.0041 1.21

NFI2 (1989–1990) 0.043 0.146 9.205 −154.671 0.002 −1.3 E-06 0.032 30.136 −0.006 0.0036 1.70

NFI3 (1999–2000) 0.076 0.118 15.825 −110.065 0.004 −1.5 E-06 0.023 44.392 −0.009 0.000 1.36

NFI4 (2009–2010) 0.108 0.093 22.952 214.585 0.007 −2.9 E-06 −0.046 36.204 −0.008 0.0001 0.99


and 800–900 m in 2010. The maximum altitude at which the

presence prediction was above the 0.50 threshold was 1,250m

in 1971 and 1,450 m in 2010 (Fig. 2c). In the case of beech,

the optimum altitudinal prediction varied over the period

1971–2010, from 1,300–1,400 to 1,450–1,600 m. The maxi-

mum altitude at which the species was found shifted from

1,550 to 1,750 m (Fig. 2d).

3.3.3 Changes in exposure distribution

The block averages of the presence by exposure class show

that this environmental factor is not a determinant for Scots

pine distribution (Fig. 2e). In contrast, the presence predic-

tion for beech has increased for northern exposures, where

this species is more frequent (Fig. 2f).

Fig. 2 Universal kriging block mean prediction for the presence/ab-

sence indicator variables by latitudinal (above), altitudinal (mid) and

exposure (below) classes for P. sylvestris (a, c, e) and F. sylvatica (b, d,

f) from the NFI1, NFI2, NFI3 and NFI4 (from light to dark grey) data.

Error bar lines represent the 95 % confidence intervals ci-u upper

confident interval, ci-l lower confident interval of the block mean for

each inventory and species. The black arrow indicates the direction of

the shifting trend of each tree species in the study area


3.4 Trade-off between altitude/exposure and latitude

The inclusion of altitude for latitude and exposure for lati-

tude interactions in the universal kriging model provided a

deeper insight into the species distribution at regional scale.

In the study area, there was a south–north elevation gradient

that influenced the altitudinal species distribution. However,

Figs. 3a and 4 show that for both species, the largest eleva-

tion shifts took place at mid- and southern latitudes, where

the sub-Mediterranean influence is greater and the elevations

are lower. The presence increased for northern exposures and

southern latitudes for both species (Fig. 3b), following the

elevation gradient.

4 Discussion

4.1 A new method for assessing changes in species

distribution from long-term forest inventories

In this study, we have presented a new method based on

universal kriging for the early detection of changes in the

distribution of forest species using long-term information

derived from the NFI. Kriging techniques provide optimal

and unbiased estimates of a regionalized variable in the pres-

ence of spatial autocorrelation, allowing the uncertainty of the

estimates to be quantified (Mandallaz 2000). In our case study,

spatial autocorrelation accounted for a large part of the

semivariance of the presence/absence indicator variable for

both P. sylvestris and F. sylvatica. This fact has also been

reported for other species with widespread distribution area

(Segurado and Araújo 2004). The spatial autocorrelation re-

flects the spatial continuity of the main factors involved in

species dynamics (dispersal, mortality, environmental/physical

barriers, historical biogeography and socioeconomic factors).

Furthermore, the changes observed in the variogram of each

species across the successive inventories provide relevant in-

formation for understanding the spatial dynamics and patterns

of the species in the region.

The distribution of the analysed species is non-stationary at

the scale of the study, whereas F. sylvatica dominates in the

Atlantic influenced north-western forests, and P. sylvestris is

distributed in the southernmost areas of the study region.

Universal kriging, as opposed to other kriging techniques, is

capable of explaining the non-stationarity of the species dis-

tribution in the study area through the mean function

Fig. 3 Universal kriging block

predictions of the mean

distribution likelihood of each

species and NFI, considering

the interactions between altitude

and latitude (above) as well as

exposure (as the cosine of the

aspect azimuth (cos(8)) and

latitude (below))


depending on the auxiliary variables included in the model

(Cressie 1993). Our results reveal how local environmental

variations can also constrain the species distribution,

highlighting the benefit of using local and landscape

approaches to identify species–environment relation-

ships that are averaged out by global models (Osborne

et al. 2007). Furthermore, the species–environment re-

lationships can also display non-stationarity for broadly

distributed species (Miller and Hanham 2011), such as

the upward shift in the altitudinal distribution of the

Scots pine across the latitudinal gradient in the study

area or the preference of both species for less exposed

locations as the Mediterranean influence increases at southern

latitudes (Jump et al. 2009). These latitudinal changes in the

species–environment relationships are incorporated through

the altitude × latitude and exposure × latitude interactions in

the universal kriging model. The species–environment inter-

actions can be a key aspect of analysis when studying changes

in the species distribution ranges in transitional bioclimatic

regions.

Direct comparison of NFI plot data may lead to over-

estimation of changes in species distributions due to the

fluctuating population dynamics, changes in forest inventory

area and variations in sampling density across the successive

NFIs (Loiselle et al. 2003). The block kriging prediction

resolves this problem by providing an unbiased estimator

of the mean for the blocks from the information contained in

the entire data set. Block kriging incorporates the change of

support from point to block, smoothing variations caused by

the different sampling schemes used (Yoo and Trgovac 2011)

and allowing direct comparison. Figure 2 shows the wider

confidence intervals derived from the block kriging variance

for the NFI1 and NFI2, which had lower sampling densities

than the NFI3 and NFI4. The confidence interval approach

was found to be a useful technique for identifying trends in

the differences between inventories along environmental

gradients. This approach provides information about the

magnitude of the differences between inventories relative

to the accuracy of the kriging prediction, which may be

an alternative to hypothesis testing (Katz 1992) to as-

sess the significance of the changes observed in species

distribution.

This new method may provide the basis for more studies

at a broader scale. Although in this study we have analysed

presence/absence data to identify evidence of changes in the

spatial distribution of species, other variables of biological

interest derived from plot level information of NFIs could be

analysed using these techniques. Based on the changes in

species distribution observed using this methodological ap-

proach, future research could focus on disentangling the

Fig. 4 Schematic reconstruction of P. sylvestris and F. sylvatica forest

dynamics from 1971 to 2010 in the Western Pyrenees. The spatial

expansion induced by latitudinal and altitudinal shifts of both species is

shown based on the combination of universal kriging results for the two

species in the NFI1 and NFI4. Light grey trees represent the sporadic

occurrence of each species with a presence prediction value below 0.5


effect of land use changes and climate change on forest

dynamics.

4.2 Changes in the distribution of forests

The methodological approach based on the universal kriging

model allowed us to obtain evidence of some of the impacts of

global change on species distribution over recent decades at

regional scale. Firstly, the universal kriging model indicated a

progressive increase in the occurrence of the two species in the

study area, which agreeswith previous studies reporting a global

increase in forests in mountainous areas of other developed

countries (Gellrich and Zimmermann 2007). Secondly, with

regard to the altitudinal distribution of both Scots pine and

beech, the approach based on the universal kriging model

revealed a vertical shift of approximately 200 m towards higher

elevations between 1971 and 2010; a finding supported by other

research conducted at European scale (Lenoir et al. 2008).

As expected, the distribution of Scots pine in the study area

was found to have shifted over recent decades towards northern

latitudes and higher altitudes. This result is consistent with pre-

vious predictions for pine species in the Iberian Peninsula (Benito

Garzón et al. 2008). In addition, the spatial autocorrelation range

of Scots pine decreased between 1971 and 2010, implying a

probable fragmentation and loss of continuity of its distribution

area. This finding, together with the decreasing presence of the

species over the last decade in the southernmost areas, where the

sub-Mediterranean influence is higher, may reflect an increase in

the presence of other tree species in the most recent NFI. In these

zones, it may be that Scots pine is being replaced by broadleaved

taxa like Q. pubescens, which is more adaptable to drier condi-

tions, as reported in studies of other marginal populations of this

species (Gimmi et al. 2010). The shift towards the northernmost

latitudes and higher altitudes along with both the decreasing

presence in the southernmost latitudes and the fragmentation of

its distribution area may suggest a northward retraction of the

Scots pine in the region.

Beech populations have shown the greatest expansion in

the study area over the four decades covered by this study,

increasing its presence at higher altitudes. This seems to

confirm the ‘deciduous tree invasion’ in the sub-alpine belt

predicted by Kräuchi and Kienast (1993). The progressive

increase in the spatial autocorrelation range of the beech

indicator variable reflects an increasingly continuous distri-

bution of the species in the study area, as reported in other

areas of Europe with similar scenarios (Poljanec et al. 2010).

The presence of beech increased significantly between NFI1

and NFI2 in the northernmost mountains, where this species

was already widely distributed. However, in the most recent

NFIs, the presence of this species unexpectedly increased at

mid latitudes, where it might be expected that the temperate,

humid ecological requirements of this species would not be

satisfied due to the weaker Atlantic influence. However, it

must be considered that this increase mainly occurred on

north facing slopes, reflecting the ability of beech to spread

easily, although only to favourable biotopes.

The altitudinal and latitudinal shifts in the distribution range of

both species have led to an extension of the area where the Scots

pine and beech coexist (Figs. 1 and 4). These findings provide

new evidence of changes in the forest composition. Scots pine

and beech can be considered ‘engineer species’ because their

presence substantially characterizes the environment of a site and

can define habitats. Therefore, variations in the forest composi-

tion resulting from the expansion or retraction of these species

populations might have profound implications on the diversity

and distribution of other species associated with their forests. For

example, endangered species of folivores such as the capercaillie

(Tetrao urogallus), very sensitive to habitat fragmentation, find

their westernmost Pyrenean habitat in these forests (Rodríguez

and Obeso 2000). In the future, the biotic interactions established

between the two species analysed in this study will play a crucial

role in the persistence and stability of these forest areas. More-

over, these sensitive transitional forest areas will be central to

monitoring the effects of global change and to the development

of successful conservation strategies for the region.

4.3 Conclusion

This work provides a new tool for the early detection of

changes in forest species and constitutes a first step towards

tackling the effects of global change on forests. The methodo-

logical approach proposed allows the spatial and temporal

shifts in species distribution over recent decades to be analysed

using long-term NFI and regardless of differences in the sam-

pling schemes used in each. Furthermore, the confidence inter-

val approach proposed allows us to assess whether the differ-

ences detected in the distribution of species are significant.

Acknowledgments The authors wish to thank all the staff that makes

possible the development of the NFI but especially Roberto Vallejo, Head

of the Spanish National Forest Inventory, and Dr. Aitor Gastón (E.T.I

Forestales), for kindly providing access to the full Spanish NFI data sets.

The authors thank Adam Collins for the careful English language revision.

Funding This research was supported by the AEG-09-007 agreement

from the Spanish Ministry of Agriculture, Food and Environment

(MAGRAMA) and the AGL2010-21153.00.01 project funded by the

Spanish Ministry of Science and Innovation (MICINN). F. Montes held

a Ramon y Cajal research grant, financed by the MICINN.

References

Aitken SN, Yeaman S, Holliday JA, Wang T, Curtis-McLane S (2008)

Adaptation, migration or extirpation: climate change outcomes for

tree populations. Evol Appl 1:95–111


Alpert P, Krichak S, Shafir H, HaimD, Osentinsky I (2008) Climatic trends

to extremes employing regional modelling and statistical interpreta-

tion over the E. Mediterranean Glob Planet Change 63:163–170

Bellehumeur C, Legendre P (1998) Multiscale sources of variation in

ecological variables: modelling spatial dispersion, elaborating

sampling designs. Landscape Ecol 13:15–25

Benito Garzón M, Sánchez de Dios R, Sainz Ollero H (2008) Effects of

climate change on the distribution of Iberian tree species. Appl Veg

Sci 11:169–178

Costa M, Morla C, Sainz H (eds) (1997) Los bosques ibéricos. Una

interpretación geobotánica. Planeta, Barcelona

Cressie NAC (1993) Statistics for spatial data. Wiley, New York

Deutsch CV, Journel AG (1992) GSLIB geostatistical software library

and user’s guide. Oxford University Press, London, p 340

Engler R, Randin CF, Thuiller W, Dullinger S, Zimmermann NE, Araújo

MB, Pearman PB, Le Lay G, Piedallu C, Albert CH, Choler P,

Coldea G, De Lamo X, Dirnböck T, Gégout JC, Gómez-García D,

Grytnes JA, Heegaard E, Høistad F, Nogués-Bravo D, Normand S,

PuşcaşM, Sebastià MT, Stanisci A, Theurillat JP, Trivedi MR, Vittoz

P, Guisan A (2011) 21st century climate change threatens mountain

flora unequally across Europe. Glob Change Biol 17:2330–2341

FAO (Food and Agriculture Organization of the United Nations) (2000)

Global forest resources assessment. FAO Forestry Paper 140. Rome

Gellrich M, Zimmermann NE (2007) Investigating the regional-scale

pattern of agricultural land abandonment in the Swiss mountains: a

spatial statistical modelling approach. Landscape Urban Plan

79:65–76

Gimmi U, Wohlgemuth T, Rigling A, Hoffmann CW, Bürgi M (2010)

Land-use and climate change effects in forest compositional tra-

jectories in a dry Central-Alpine valley. Ann For Sci 67:701

Goovaerts P (1994) Comparison of CoIK, IK and mIK performances

for modeling conditional probabilities of categorical variables. In:

Dimitrakopoulos R (ed) Geostatistics for the next century. Kluwer,

Dordrecht, pp 18–29

Guisan A, Zimmermann NE (2000) Predictive habitat distribution models

in ecology. Ecol Model 135:147–186

Isaaks EH, Srivastava RM (1989) Applied geostatistics. Oxford

University Press, New York

Jump AS, Mátyás CA, Peñuelas J (2009) The altitude-for-latitude dispar-

ity in the range retractions of woody species. Tree 24:694–701

Katz RW (1992) Role of statistics in the validation of general circula-

tion models. Clim Res 2:35–45

Kräuchi N, Kienast F (1993) Modelling subalpine forest dynamics as

influenced by a changing environment. Water Air Soil Pollut

68:185–197

Lenoir J, Gégout JC, Marquet PA, de Ruffray P, Brisse H (2008) A

significant upward shift in plant species optimum elevation during

the 20th century. Science 320:1768–1771

Loiselle BA, Howell CA, Graham CH, Goerck JM, Brooks T, Smith

KG, And Williams PH (2003) Avoiding pitfalls of using species

distribution models in conservation planning. Cons Biol 17:1591–

1600

Mandallaz D (2000) Estimation of the spatial covariance in universal

kriging: application to forest inventory. Environ Ecol Stat 7:263–

284

Matheron G (1969) Le krigeage Universel. Cahiers du Centre de

Morphologie Mathematique, 1. Fontainebleau, France

Miller J, Franklin J, Aspinall R (2007) Incorporating spatial dependence

in predictive vegetation models. Ecol Model 202:225–242

Miller J, Hanham RQ (2011) Spatial nonstationarity and the scale of

species-environment relationships in the Mojave Desert, California,

USA. Int J Geogr Inf Sci 25:423–438

Montes F, HernándezMJ, Cañellas I (2005) A geostatistical approach to

cork production sampling estimation in Quercus suber L. forests.

Can J For Res 35:2787–2796

Montes F, Ledo A (2010) Incorporating environmental and geograph-

ical information in forest data analysis: a new fitting approach for

universal kriging. Can J For Res 40:1852–1861

Ninyerola M, Pons X, Roure JM (2005) Atlas Climático Digital de la

Península Ibérica.

Osborne PE, Foody GM, Suárez-Seoane S (2007) Non-stationarity and

local approaches to modeling the distributions of wildlife. Divers

Distrib 13:313–323

Peñuelas J, BoadaM (2003) A global change-induced biome shift in the

Montseny mountains (NE Spain). Glob Change Biol 9:131–140

Poljanec A, Ficko A, Boncina A (2010) Spatio-temporal dynamic of

European beech (Fagus sylvatica L.) in Slovenia, 1970–2005. For

Ecol Manage 259:2183–2190

Rodríguez AE, Obeso JR (2000) Diet of the Cantabrian capercaillie:

geographic variation and energetic content. Ardeola 47:77–83

Segurado P, Araújo MB (2004) An evaluation of methods for modelling

species distributions. J Biogeog 31:1555–1568

Thuiller W, Vayreda J, Pino J, Sabate S, Lavorel S, Gracia C

(2003) Large-scale environmental correlates of forest tree

distribution in Catalonia (NE Spain). Global Ecol Biogeogr

12:313–325

Tolosana-Delgado R, Pawlowsky-Glahn V, Egozcue JJ (2008) Indicator

kriging without order relation violations. Math Geosci 40:327–347

Vilá-Cabrera A, Martínez-Vilalta J, Vayreda J, Retana J (2011) Structural

and climatic determinants of demographic rates of Scots pine forests

across the Iberian Peninsula. Ecol Appl 21:1162–1172

Williams MS (2001) New approach to areal sampling in ecological

surveys. For Ecol Manage 154:11–22

Yoo EH, Trgovac AB (2011) Scale effects in uncertainty modeling of

presettlement vegetation distribution. Int J Geogr Inf Sci 25:405–

421


Anexo2

Assessing spatio-temporal rates, patterns and determinants of biologicalinvasions in forest ecosystems. The case of Acacia species in NW Spain

Laura Hernández a,b,⇑, Jesús Martínez- Fernández a, Isabel Cañellas a,c, Antonio Vázquez de la Cueva a

a INIA-CIFOR, Ctra. La Coruña, Km 7.5, 28040 Madrid, SpainbUPM, Universidad Politécnica de Madrid, 28040 Madrid, SpaincUniversity of Valladolid and Institute on Sustainable Forest Management, Avda, Madrid, 44, 34004 Palencia, Spain

a r t i c l e i n f o

Article history:

Received 26 March 2014

Received in revised form 28 May 2014

Accepted 31 May 2014

Keywords:

Acacia dealbata

Acacia melanoxylon

Invasive species dynamics

Invasibility

Spread factors

National Forest Inventories

a b s t r a c t

Invasive species currently pose a major environmental challenge. Understanding their development and

the factors associated with their expansion is the first step towards developing effective control

measures. This work proposes the use of detailed spatio-temporal information from forest monitoring

systems to assess the demographic rates, spatio-temporal patterns and spread determinants of invasive

plants in forest ecosystems. For this purpose, we selected two of the most widespread non-native plants

in Europe: Acacia dealbata and Acacia melanoxylon. Focusing on the forested area of northwest Spain and

based on the comparison of two cycles of the Spanish National Forest Inventory, this study analyzes the

dynamics of Acacia species between 1998 and 2008 in regards to changes in their spatial distribution,

dominance, abundance, diametrical (dbh) structure and regeneration. In addition, the forested area

was classified into forest types to identify the forests which are more susceptible to invasion. Finally,

through general linear models, this study analyzes the relative importance of abiotic and biotic factors

determining the spread of Acacia species over the studied period. The results confirm a rapid expansion

in the presence of Acacia species in the forests of NW Spain, with annual spread rates around 0.1%. These

two species are increasing their dominance across most forest types in the study area, where they are

becoming the dominant species in the regeneration of some of them. Environmental factors and connec-

tivity between Acacia populations are identified as the main factors associated with their spread into new

areas. Additionally, the combination of disturbances and biotic factors associated with stand structure

(total basal area, richness and tree cover) appear to determine the vulnerability or resistance of some

forest to their spread. The early stage of invasion detected highlights the potential of Acacia species to

continue spreading. This aspect, in conjunction with the high degree of disturbances (mainly fires) in this

region, could be critical in determining the configuration of future forest landscapes in NW Spain. This

study demonstrates the value of considering broad-scale periodic forest surveys to monitor biological

invasions in forests ecosystems. The spatially-explicit data obtained from these surveys can contribute

not only to furthering our knowledge with regard to invasion biology but also to developing more

efficient conservation and management strategies.

Ó 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND

license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

1. Introduction

Invasive alien species pose one of the most important direct

threats to the structure and function of ecosystem diversity. The

spread of invasive alien species is therefore among the most urgent

nature conservation issues to be faced at global scale (UNEP, 2010),

the identification of alien spread pathways to prevent their

introduction and establishment being one of the main targets of

the EU 2020 Strategic Plan for Biodiversity (EC, 2011).

The considerable research into invasion ecology in recent dec-

ades has centered around three main questions: invasiveness

(Rejmánek and Richardson, 1996), invasibility (Chytry et al.,

2008) and impacts (Hejda et al., 2009). As a result, our theoretical

and practical knowledge of plant invasions has improved

substantially, although the capability to address further challenges

in this line of research may be hindered by the lack of availability,

detail and heterogeneity of information concerning invaders

(Pyšek et al., 2002). One of these new challenges is to provide

new insights into invasion dynamics (Richardson et al., 2010).

http://dx.doi.org/10.1016/j.foreco.2014.05.058

0378-1127/Ó 2014 The Authors. Published by Elsevier B.V.

This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

⇑ Corresponding author at: INIA-CIFOR, Ctra. La Coruña, Km 7.5, 28040 Madrid,

Spain. Tel.: +34 9191 347 3936.

E-mail address: [email protected] (L. Hernández).

Forest Ecology and Management 329 (2014) 206–213

Contents lists available at ScienceDirect

Forest Ecology and Management

journal homepage: www.elsevier .com/ locate/ foreco

Although access to detailed lists and maps of non-native species

has improved at global level in recent times (Foxcroft et al.,

2010), there is a lack of broad scale periodic surveys providing

the possibility to identify detailed demographic rates, spatio-

temporal patterns and determinants of invasive plant spread.

Statistically-designed inventories such as National Forest Inven-

tories (NFI) based on periodic re-measurements of permanent

sample units constitute a valuable tool for monitoring forest

dynamics (Lund, 1998). Consequently, the inclusion of non-native

species data in these surveys, based on biodiversity monitoring

programs, (Corona et al., 2011) provides a valuable opportunity

not only to examine the broad-scale evolution of plant invasions

in forest ecosystems but also to test ecological hypotheses in inva-

sion biology.

For this study we selected two different Acacia species in an

attempt to determine general patterns in forest dynamics and

invasion ecology over the last decade in the northwest of Spain,

one of the most affected regions of the country (Romero Buján,

2007). Acacia species are among the most widespread invasive

plants in Europe, and two of the most aggressive are Acacia dealba-

ta Link and Acacia melanoxylon R. Br. in W.T. Aiton. Today, they are

widely naturalized and have become an environmental problem in

Southwestern Europe (Carballeira and Reigosa, 1999; Hussain

et al., 2011) where they pose a threat to native species and have

been declared ‘‘invaders’’ (Sanz-Elorza et al., 2004). The invasive

success of Acacia is mainly attributed to its rapid growth rate, pro-

lific production of seeds with high longevity, germination stimu-

lated by fire, allelopathic effects and the absence of natural

enemies (Marchante et al., 2003). In the Iberian Peninsula, these

two species have not yet reached their potential distribution range

(Gassó et al., 2012), although it is likely that they will be able to

reach this potential range in the near future. Identifying the deter-

minants of invasion in the early stages is crucial to the develop-

ment of realistic predictive models of invasion risk (Kolar and

Lodge, 2001) and to mitigate the potential ecological impact.

This study constitutes a first attempt at using national forests

monitoring data to study the evolution of biological invasions.

Based on the analysis of the sequential broad-scale databases from

two cycles of the Spanish NFI, the primary aim of this study is to

examine the spatio-temporal changes in the distribution,

abundance and dominance of A. melanoxylon and A. dealbata in

the forests of NW Spain over the period 1998–2008. The second

objective is to identify the types of forest which are most vulnera-

ble to invasion and the level of invasion reached. Finally, an

attempt is made to disentangle the relative importance of the

biotic and abiotic factors underlying the spread of these two

species in different forest types. For this purpose we test two

hypotheses: the first is that Acacia species have expanded and

increased their dominance in the forests of NW Spain between

1998 and 2008. The second is that the expansion rates and

dominance of these species differ from one forest type to another.

2. Material and methods

2.1. Study area

This study is based on NFI information for the provinces which

comprise the region of Galicia in the northwest of the Iberian Pen-

insula (Fig. 1). Due to a combination of bioclimatic and human fac-

tors, the percentage of non-native flora in this area of NW Spain

(14%) (Romero Buján, 2007) is higher than the for the Iberian

Peninsula as a whole (12%) (Sanz-Elorza et al., 2004). This region

presents a climatic gradient from the coast towards inland

areas, but there is a dominant humid Atlantic climate with mild

temperatures (mean annual temperature of 13 °C) and abundant

precipitation (mean annual rainfall of 1400 mm). Soils are acidic

and the area exhibits a complex topography, with altitudes ranging

from sea level up to 2124 m. Today, almost 50% of the forests in NW

Spain comprise plantations of non-site-native species such as Pinus

pinaster Ait., Eucalyptus spp., Pinus sylvestris L. and Pinus radiata D.

Don (Table S2 in Supplementary material). The native forest types

present in the area are floodplain/riparian forests, scattered

coastline forests of P. pinaster, atlantic mixed broad-leaved forest

where Quercus robur L. is abundant and oak forests of Quercus

pyrenaica Willd. in the transition zone between the Atlantic and

Mediterranean biogeoregions (Fig. 1C).

Acacia species are Australian N2-fixing trees that were intro-

duced into Europe as an ornamental species in the 19th century

(Sheppard et al., 2006). In Spain, they are mainly distributed in

the most NW territories where A. melanoxylon occurs close to the

coastline in temperate locations while A. dealbata, with a broader

ecological valence, can be found in more continental areas

(Fig. 1B, Table S2 in Supplementary material).

2.2. Data used

The study is based on spatially detailed information from two

consecutives cycles of the Spanish NFI performed in NW Spain in

1998 (NFI3) and in 2008 (NFI4), a time interval of 10 years

(13,159 plots). In these Spanish NFI cycles, permanent plots were

established systematically in the forested area at the intersections

of a 1 km � 1 km grid. Field plots consist of four concentric circular

areas with radii of 5, 10, 15 and 25 m.

Depending on the dbh (diameter at breast height) of the tree

species, different dendrometric characteristics are measured

within each plot and for each radius such as dbh or height of trees

with dbhP 7.5 cm and heightP 1.30 m. Furthermore, other forest

attributes and conditions are measured (tree and shrub species

composition, density, covers, recruitment, saplings (trees with

2.5 cm 6 dbh < 7.5 cm), silvicultural treatments (clear-cutting,

groundwork and crown treatments), etc.).

A total of 20 predictors were considered as independent

variables to analyze the relative importance of abiotic and biotic

factors determining the spread of Acacia species over the period

considered (Table S1 in Supplementary material). Some of these

variables were also used to characterize the climatic and physical

ranges of the two Acacia species and the different forest types, as

well as their disturbance level (Fig. S1 in Supplementary material).

As regards the abiotic factors; topographical variables (altitude,

aspect, exposure and closest distance to sea) were taken from the

digital elevation model of Spain with a spatial resolution of 25 m

(U.T.M, ED 50). Climatic variables were extracted from Gonzalo

(2010). The impact of human disturbances on the spread of Acacias

over the time frame considered were discerned using variables

such as silviculture treatments obtained from NFI plot databases,

the forest-urban and forest-crop interface calculated from land

use maps (Heymann et al., 1994; EEA, 2012) and the fire incidence

from MODIS burned area products (Boschetti et al., 2009). As for

biotic factors, we considered several attributes of forest structure

derived from NFI datasets such as species richness, tree cover

and basal area at plot level. Furthermore, to analyze the

importance of propagule pressure or distance from invasion loci,

we consider the connectivity between plots containing Acacia

species (Fig. S1 in Supplementary material).

2.3. Data analysis

Spanish NFI records of the presence of A. melanoxylon and

A. dealbata (Fig. 1B), along with other cartographical sources such

as botanical Atlases (Sanz-Elorza et al., 2004; Anthos, 2012),

provided valuable information for mapping the current spatial

L. Hernández et al. / Forest Ecology and Management 329 (2014) 206–213 207

distribution of the two Acacia species in the study area (Fig. 1A).

Based on NFI plot information such as species dominance and for-

est management, the forested area of NW Spain was then classified

into nine forest types (see Fig. 1C and Table S2 in Supplementary

material for main characteristics and abbreviations) according to

the definitions proposed for Europe (Barbati et al., 2007).

The comparison of the information derived from the plots in

which Acacia species were present in the two sequential NFI

allowed the total spatial expansion, density and growing stock

rates of Acacia species to be assessed (114 plots). To analyze the

stage of invasion in the plots during the time frame considered,

changes in Acacia species abundance by dbh class were also calcu-

lated. It must be considered that due to the dynamic nature of this

species and the frequency of disturbances in the forests where

these two species are found, the number of remeasured plots used

in these analyzes was limited and was lower than the proportion

used when studying the distribution or spread derived from the

presence/absence indicator.

The vulnerability to the invasion in the different forest types

was analyzed through the particularized Important Value Index

(IVI) from Curtis (1959) as the sum of the relative density and

dominance of Acacia species in the plots by forest types (114 plots).

Relative density was defined as the numerical strength of a species

in relation to the total number of individuals of all the species. It

was calculated as: [(Number of individual of the species/Number

of individual of all the species) � 100]. Relative dominance was

determined by the value of the total basal area of a species with

respect to the sum of basal area of the rest of the species in the plot

and it was calculated as: [(Total basal area of the species/Total

basal area of all the species) � 100]. The IVI was used to determine

the overall importance of Acacia species in the plots, providing a

good indicator of invasive species dominance. In these approaches,

only the forest types in which there was a significant number of

plots with presence of the two Acacia species were considered

(nP 5). Since there were different sample sizes and non-homoge-

neity of the variance, the mean differences in Acacia IVI by forest

type were assessed through the Welch Test. Tamhane’s T2 test

was then used for post hoc multiple comparisons of mean values

between forest types (Hollander and Wolfe, 1999).

Two dichotomous dependent variables (Col, colonization, and

Dis, disappearance) (757 plots) were then created to analyze

changes in spatial distribution and the level of invasion of the

two Acacia species in the different forest types in NW Spain during

the period (1998–2008).

The current invasibility of the different forest types was partly

analyzed through the total number and proportion of saplings of

the two Acacia species in the regeneration of each forest type

according to the last NFI (2008) (92 plots). The proportion of the

species in the regeneration defines the numerical strength of that

species in relation to the total number of individuals of all the spe-

cies, indicating conspecific abundance or dominance. The results

were then analyzed to show the overall patterns and range of the

current regeneration of the two Acacia species in the different for-

est types they have invaded.

Finally, a dichotomous dependent variable Spr was created to

assess the effect of different biotic and abiotic factors on the spread

of the Acacia species in the forests of NW Spain over the period

1998–2008 through general linear model (GLM) analysis with

binomial error and logit link. This was analyzed in the plots in

which Acacia species were not present in 1998 but were present

in 2008 (251 plots). The logistic regression models provide infor-

mation on the relationships and strengths among dependent and

Fig. 1. (A) Known distribution area of Acacia melanoxylon and Acacia dealbata in Europe. (B) Known distribution area of Acacia species in NW Spain. (C) Distribution of the

different forest types in NW Spain.

208 L. Hernández et al. / Forest Ecology and Management 329 (2014) 206–213

independent variables. Furthermore, logistic regression accepts a

combination of continuous and categorical variables as well as

not normally distributed ones (Hosmer and Lemeshow, 2000).

Multicollinearity was verified using Pearson correlation

coefficients. Some of the explanatory variables that were highly

correlated (|r| > 0.8) were excluded prior to building the model.

We followed a step-by-step model-building procedure and the fit

of the model was tested after the elimination of each variable.

Deviance reduction, estimated as: D2 = (null deviance – residual

deviance)/null, was used as the measure of discrepancy to assess

the goodness-of-fit of the model (Crawley, 1993).

ArcGis 10 (ESRI Inc., Redlands, CA, USA) was used as an image

analysis tool as well as to extract topographical, climatic, land

use and fire variables to the 1 km � 1 km grid. R software was used

to fit the GLMmodel and SPSS17.0 (SPSS Statistics, 2009) for all the

other statistical analysis.

3. Results

3.1. Changes in the distribution of Acacia species

In 1998, A.melanoxylon and A. dealbatawere present in 2.2% and

1.5% respectively of the total forested area of the study region. By

2008 they had expanded by 1% and 0.83% respectively (Table 1),

with an expansion rate (proportion of total hectares invaded per

year) of approximately 3100 ha yrÿ1 and 2500 ha yrÿ1. The inten-

sity of the changes in the spatial distribution of Acacia species dif-

fers according to forest type. There has been a notable increase in

the presence of A. melanoxylon in riparian forests and Eucalyptus

spp. plantations, reaching 7% and 3.5% of the total forest area of

these two forest types respectively (Fig. 2). In these two forest

types, the increase was almost four times greater than for other

forest types. Conversely, there has been a notable decrease in the

presence of this species in mixed broadleaf forests and shrublands,

although in 2008, it was present in 1% and 3% of these two forest

types respectively. The presence of A. dealbata has increased signif-

icantly in the majority of forest types in the study region, even

occurring in Q. pyrenaica oak forest in 2008, where it was absent

in 1998. This increase is particularly notable in mixed conifer-

broadleaf forests, the most abundant forest type in the study area,

where it is present in almost 3.5%. A small decrease in the presence

of this species was detected in riparian forest and shrubland,

occurring in 3% and 0.4% of each type respectively (Fig. 2).

3.2. Acacia species dynamics and dbh structure

Analysis of the dynamics of NFI plots in which Acacia species

were present in 1998 and 2008 show that the number of trees

per hectare, basal area and growing stock of these two invasive

species almost doubled in all cases during this period (Table 1).

However, while in the case of A. melanoxylon the increase has been

greater in terms of basal area and proportion of growing stock in

the stands where it occurs, A. dealbata has undergone a greater

increase in the number of trees per hectare (Table 1).

The pattern of variation in the proportion of trees of different

dbh classes over the studied period differs between the two spe-

cies. In the case of A. melanoxylon, the proportion of trees with

the smallest diameters has suffered a slight drop over the period,

while the proportion of trees with medium-large diameter has

increased (Fig. S2 in Supplementary material). However, in the case

of A. dealbata the proportion of small diameter trees has risen,

whereas the proportion of medium diameter trees has decreased

slightly (Fig. S2).

Table 1

Changes in distribution (DS) and stock (DN,DBA,DG) of A. dealbata and A. melanoxylon, 1998–2008. In bold the decennial increments are highlighted.


1998 2008 Increment 1998 2008 Increment

Number of plots with presence 167 262 95 129 199 70

Percentage of total forest area (%) 2.2 3.2 1.00 1.5 2.4 0.8

Area (ha) 57,300 88,250 31,000 40,900 66,300 25,400

Mean number of trees per ha 2.6 3.1 0.5 2.4 3.8 1.4

Relative density:% of Acacia sp. in total tree per ha 0.4 0.4 0.02 0.4 0.5 0.1

Mean basal area (m2/ha) 0.03 0.06 0.03 0.02 0.04 0.02

Relative dominance: % of Acacia sp. in total basal area (%) 0.2 0.2 0.1 0.1 0.2 0.1

Mean growing stock (m3/ha) 0.2 0.4 0.2 0.1 0.2 0.1

Proportion of Acacia sp. in total growing stock (%) 0.2 0.3 0.1 0.1 0.1 0.03

Fig. 2. Changes in the proportion of the Acacia species in the different forest types

of NW Spain, 1998–2008. See Table S2 in Supplementary material for forest types

abbreviations.


The analysis of Acacia species saplings did not identify signifi-

cant differences between forest types due to the high level of

stochasticity in the associated variables. However, as can be seen

from Table 2, the relative and absolute density of the regeneration

varied from one forest type to another. A. melanoxylon regenerates

more abundantly in mixed conifer-broadleaf forests than in Euca-

lyptus spp. plantations, the only two forest types in which saplings

of this species were found. In addition, as revealed by the relative

results, this species tends to dominate in the regeneration of mixed

conifer-broadleaf forests. In the case of A. dealbata, despite the high

dispersion of the saplings associated variables, the results show a

clear tendency to dominate in the regeneration of all the forest

types in which this species was found: mixed broadleaf forests,

Eucalyptus spp. plantations and mixed conifer-broadleaf forest,

regenerating in similar abundance.

As indicated by the Welch and Post hoc tests through mean dif-

ferences in the IVI over the period 1998–2008, the dominance of

the two invasive species increased in all the forests types in which

they were present (Table 3). However, these changes differed sig-

nificantly depending on the forest type. The increase in overall

importance of A. melanoxylon and A. dealbata was significantly

greater in mixed conifer-broadleaf forests than in other forest

types. Additionally, in the case of A. melanoxylon, this increase in

dominance was also more significant in Eucalyptus spp. plantations

than in other forest types.

3.3. Factors involved in the spread of Acacia species

Among the 20 biotic and abiotic predictors previously consid-

ered to explain the spread of the Acacia species from 1998 to

2008 (Table S1 in Supplementary material), the best GLM fits were

obtained for 11 different predictors in the case of A. melanoxylon

and 10 for A. dealbata with which the spread of the two species

over the period displayed significant relationships. The final mod-

els accounted for 52% and 48% of the observed variability in the

spread of A. melanoxylon and A. dealbata respectively over the

period (Table 4). As regards the climatic abiotic factors, mean

annual temperature has significant and positive effects on both

species spread over the studied period, while to a lesser extent,

annual rainfall and the mean temperature oscillation has a

negative one. The factor ‘closest distance to the sea’ also has a

positive relationship with the spread of A. melanoxylon in the study

zone. Regarding human-mediated and other disturbances, clear-

cutting, urban-forest interfaces and fire incidence had a positive

and significant effect on both species. Among the biotic factors

connectivity between stands shows a strong association with the

spread. As for other biotic interactions, stand structure characteris-

tics such as tree cover and total basal area exhibited a negative

relationship while total richness has a positive association with

their spread.

4. Discussion

The results confirm the prior hypothesis of expansion and

upward dominance of Acacia species in the forested area of NW

Spain over the last decade, revealing different invasion patterns

from one forest type to another.

4.1. Spatio-temporal dynamics

The quantification of the area occupied by invasive plants in a

given zone, the identification of the stage of invasion, their spread

Table 2

Mean and standard deviation (Stand-Dev) of the absolute (Nsap) and relative (Nsaprel (%)) number of A. melanoxylon and A. dealbata saplings in the regeneration per hectare. See

Table S2 of Supplementary material for forest types abbreviations.

Forest type A. melanoxylon A. dealbata

Mean Nsap (Stand. Dev.) Mean Nsaprel (Stand. Dev) Mean Nsap (Stand. Dev) Mean Nsaprel (Stand. Dev)

EUC 875.4 (756.7) 0.5 (0.4) 2772.8 (1930.5) 0.7 (0.4)

MIXCB 2018.9 (1036.4) 0.8 (0.3) 2113.6 (1916.2) 0.9 (0.1)

MIXB – – 3028.5 (2921.2) 0.9 (0.2)

Table 3

Mean differences in IVI (Importance Value Index) of (1) A. melanoxylon and (2) A. dealbata, between the invaded forest types of NW Spain. See Table S2 of Supplementary material

for forest type abbreviations.

(1) Mean IVI (Stand. Dev) Forest type MARPIN (n = 10) MIXB (n = 7) EUC (n = 19) MIXCB (n = 32)

0.14 (0.08) MARPIN ÿ0.227 ÿ0.64* ÿ086*

0.36 (0.23) MIXB ÿ0.420 ÿ0.64

0.79 (0.61) EUC ÿ0.22

1.00 (0.89) MIXCB

(2) Mean IVI (Stand. Dev) Forest type MIXB (n = 7) EUC (n = 9) MIXCB (n = 30)

0.62 (0.57) MIXB ÿ0.25 ÿ066*

0.87 (0.69) EUC ÿ0.41

1.28 (0.69) MIXCB

* Significant difference at level p = 0.05.

Table 4

Binomial GLM results for the spread of A. melanoxylon and A. dealbata during the

period 1998–2008. See Table S1 of Supplementary material for abbreviations.

Predictors A. melanoxylon A. dealbata

Effect D2 Effect D2

Intercept ÿ0.35 – 5.59 –

Ptot ÿ0.0125 1.99* ÿ0.0037 8.22**

Tm 0.31 44.73** 0.22 68.43**

Dsea 4.08Eÿ05 62.89** – –

Osc ÿ0.28 9.7** ÿ0.42 11.52**

Clearcut 0.96 2.09** – –

Connect ÿ3.18Eÿ04 85.33** ÿ1.80Eÿ04 66.15**

Treecover ÿ0.01 12.13** ÿ0.04 10.48**

Fire 0.93 2.06* 0.68 2.87*

BA ÿ0.04 4.47** ÿ0.0407 3.85**

TRichness 0.13 4.35** 0.22 12.31**

FUInt – – 0.08 7.23**

Deviance (D2) 51.623% 48.026%

* Significance codes: p > 0.05** Significance codes: p > 0.01.


rates and the potential to invade new areas are of critical impor-

tance in determining the ability of organisms to shift their ranges

and to detect their invasive success and persistence (Higgins and

Richardson, 1996; Higgins et al., 2001). The percentage of forested

area in NW Spain occupied by the two Acacia species reached 3.2%

and 2.4% in 2008, exhibiting annual rates of invasion (proportion of

total area invaded per year) of 0.1% and 0.08% for A. melanoxylon

and A. dealbata respectively. To date, most of the empirical data

on invasion rates worldwide are based on estimations from past

aerial photographs (Lonsdale, 1993; Higgins et al., 2001) for broad

scale studies or, to a lesser extent, from estimates based on field

work for local scale studies (Wangen and Webster, 2006). How-

ever, to our knowledge this is the first time that these rates have

been studied extensively on a detailed spatial grid on such a large

scale. Bearing in mind the limitations involved in comparing the

spread rates associated with the differing sized areas and popula-

tions monitored of previous studies, it would appear that the Aca-

cia species in NW Spain exhibit mid-high invasion rates, since

similar or lower estimates have been reported for other invasive

tree/shrub species; 0.08% for P. radiata (Richardson and Brown,

1986), 0.03% for Acacia cyclops or 0.06% for P. pinaster (Higgins

et al., 2001), all in South Africa. The rates of areal spread of the

two Acacia species (31 km2 yrÿ1 and 25 km2 yrÿ1) are close to the

average for other tree/shrub life forms worldwide, which present

a mean spread rate of around 27 km2 yrÿ1, ranging from 0.02 km2 -

yrÿ1 to 179 km2 yrÿ1 (from a review carried out by Pyšek and

Hulme, 2005).

The aforementioned spatio-temporal spread was concomitant

with a substantial increase in growing stock and dominance over

the period 1998–2008, doubling in all the cases the preceding val-

ues but displaying different traits. Whereas A. melanoxylon seems

to be increasing its dominance in forests where it was present

through increments in growing stock and basal area, A. dealbata

displays a greater increase in density and regeneration. Although

this rise in dominance is general for all forest types analyzed, a sig-

nificant increase is observed in disturbed forest types (according to

our characterization) such as Eucalyptus spp. plantations and

mixed conifer-broadleaf forests, but also in habitats which in

principle are less altered by human activity such as native

mixed-broadleaf forests. This pattern is also observed in the rela-

tive proportion of trees in the regeneration where Acacia species

tend to dominate and to homogenize this important forest stratum

supporting previous results at local scale in the region (Hussain

et al., 2011; González-Muñoz et al., 2012; Lorenzo et al., 2012).

The concurrence of our findings with those of the aforementioned

studies may have important implications for the future composi-

tion and functional diversity of forests in which Acacia species

are becoming naturalized and spreading and where a decrease in

structural and compositional complexity is expected.

The dbh distribution structure of Acacia species over the studied

period suggest an early stage of invasion in the forests of NW Spain

being saplings the dominant class. This finding is in accordance

with the general increase of Acacia species occurrence observed

in the majority of forest types of NW Spain in the last inventory.

Bearing in mind the capacity of Acacia species to flower throughout

the year and their strong resprouting ability (Lorenzo et al., 2010),

these results highlight the potential of Acacia species to continue

spreading in the near future, confirming previous predictions sug-

gesting that these species have not yet reached their potential area

of distribution within the Iberian Peninsula (Gassó et al., 2012).

4.2. Factors associated with Acacia species spread and invasibility

The results of the GLMs point to connectivity and environment

as the key factors associated with the expansion of the two

invasive species.

It is not surprising that connectivity between Acacia populations

has a strong association with the spread since increased availability

of propagules between proximate populations raises the chances of

establishment, persistence, naturalization and invasion (Alston and

Richardson, 2006). Furthermore, this result is in accordance with

the non-long distance dispersal adaptation more common in Acacia

species which are usually dispersed by animals such as birds and

ants (Davidson and Morton, 1984; Lorenzo et al., 2010).

With regard to environmental factors, temperature and

distance to the sea are revealed as the most important filters

constraining the colonization of new areas. This finding agrees

with the habitat compatibility hypothesis (Rejmánek et al., 2005)

which states that habitats globally tend to be invaded by species

from similar environments at source. The natural distribution

range of A. melanoxylon and A. dealbata, mainly in south-eastern

Australia, illustrates their preference for oceanic climate locations

(Costermans, 1985). This partly explains the absence of both Acacia

species in certain forest types characterized by higher temperature

oscillations, such as P. sylvestris plantations located at higher alti-

tudes. Similarly A. melanoxylon is completely absent in oak forests

of Q. pyrenaica in submediterranean climatic transition zones

although the most recent inventory indicated a small presence of

A. dealbata in these forests. The broader ecological valence of the

latter allows this species to invade more continental locations,

away from the influence of the coast. However, this significant

relationship with certain environmental factors does not necessar-

ily explain the degree of invasion in the different forest types.

Moreover, forest types with similar environmental conditions

exhibit dissimilar levels of invasion, suggesting that other mecha-

nisms render them more or less susceptible to invasion.

Disturbances are considered one of the most important factors

behind the invasive spread of Acacia species (Brooks et al., 2004;

Lorenzo et al., 2010; Le Maitre et al., 2011). The significant positive

association found between the spread of A. melanoxylon and A.

dealbata and disturbance events such as fires, clear-cutting and

urban-forest interfaces would appear to support this idea. Davis

et al. (2000) suggest that habitats might be more susceptible to

invasion when there is an increase in the amount of unused

resources resulting from disturbance events. Accordingly, some

of the habitats with higher levels of disturbance correspond to for-

est types with a higher degree of invasibility. Eucalyptus spp. plan-

tations are located close to the coast where population is

concentrated and where intuitively there would be a high propa-

gule pressure (Di Castri, 1989) due to their proximity to urban-for-

est interfaces and communication networks. Furthermore,

Eucalyptus spp. plantations and mixed conifer-broadleaf forests,

two of the most invaded forest types, had the highest fire incidence

over the period (affecting almost 10% and 8% respectively of their

extent). Forest fires are one of the main disturbance factors in for-

est ecosystems across large areas of Spain, including NW Spain

(Moreno et al., 1998; de la Cueva et al., 2006). Acacia species are

highly resilient to fires and are capable of regenerating both

through germination and sprouting from roots and stems after

fires (Ough, 2001). Some authors (e.g. Lorenzo et al., 2010) have

already suggested that the spread of A. dealbata in NW Spain

may be assisted by human disturbances such as fires, although this

hypothesis has not yet been tested at larger scales (see however

de la Cueva (2014) for a local scale study). Hence, one of the key

findings of the present study is the positive relationship found

between the spread of both Acacia species and fire incidence,

which confirms that areas currently occupied by Acacia are often

areas which have been affected by fire.

Disturbances seem to be an important factor associated with

the spread of invasive species, however not always the more

anthropogenic disturb forests are the ecosystems with a higher

degree of invasion. In such cases, as suggested by Alpert et al.


(2000) or Blossey and Nötzold (1995), community structure traits

might be influencing the invasibility of some ecosystems. This is

the case of P. radiata and P. pinaster plantations, which present

lower levels of invasion than other forests even though they are

intensively managed and occupy large areas with as suitable envi-

ronmental conditions as other more invaded forests. Pine planta-

tions form monospecific forests with higher stand basal area and

an evergreen canopy which is more closed than that of Eucalyptus

spp. plantations, resulting in a general absence of understory. In

these forests, Acacia species would face strong competition for

resources, which may be associated with negative relationships

found between the spread of these species and both total stand

basal area and tree cover. This pattern is also corroborated by the

significantly lower increase in dominance found in P. pinaster plan-

tations compared with other forest types. Furthermore, these

results support the findings of previous studies that identified

important relationships between invasion by alien species and per-

centage of overstorey cover (Alston and Richardson, 2006) and can-

opy closure (Fuentes-Ramírez et al., 2011).

As regards other community structure traits which influence

the spread of Acacia species; total richness per plot shows a signif-

icant positive relationship supporting previous studies at similar

scales (Alpert et al., 2000; Stohlgren et al., 2003; Souza et al.,

2011). In accordance with this finding, forest types with higher val-

ues of richness such as mixed conifer-broadleaf forests, mixed-

broadleaf forests and riparian forests, show some of the highest

levels of invasion. Levine and D’Antonio (1999) suggested that spe-

cies richness may be positively correlated with invasion, because

both are promoted by the same factors. Like floodplain forests,

the conifer-broadleaf and mixed-broadleaf forests of NW Spain

form transitional, dynamic ecosystems that act as corridors for

species between adjacent habitats (Richardson et al., 2007) and

therefore, support high levels of diversity. The current mixed coni-

fer-broadleaf forests of NW Spain are the result of natural coloniza-

tion by native species of abandoned monospecific plantations

(Saura and Carballal, 2004) and the mixed-broadleaf forests are

usually found in valleys near populations, with a rich mosaic of

land uses and where there would be a strong propagule pressure

and more opportunities for Acacia species to be established. In

mixed forests, our results also show an alarming tendency for

Acacia species to dominate in the regeneration which lead us to

suggest that in these cases, once established in the understory,

other processes such as competitive ability (Blossey and Nötzold,

1995) and allelopathy (Callaway and Aschehoug, 2000) may

facilitate their persistence and dominance.

5. Conclusions

Acacia species are spreading rapidly and are becoming the dom-

inant tree species across large areas of forest in NW Spain. As this

study suggests, the success of Acacia species in spreading to and

invading new areas is not due to a single mechanism but rather

to a group of interrelated processes. The distribution ranges of

the species in the forests of NW Spain are mainly constrained by

environmental filters and the connectivity or propagule pressure

between proximate populations. Additionally, the combination of

disturbance events and stand structure traits seem to play an

important role in determining the level of invasion of the different

forests. Our results point to an early stage of invasion, highlighting

the potential of Acacia species to continue spreading. This fact,

together with the high frequency of disturbances such as fire,

may be critical in determining the configuration of future forest

landscapes in the region (e.g. de la Cueva et al., 2012).

The empirical results from this research will contribute to the

growing reference database on plant invasion rates and may

provide practical help in the assessment of level and severity of

biological invasions worldwide. Furthermore, the detailed data

obtained from this type of study, such as spread rates, spread

determinants and forests invisibility is crucial to improving spa-

tially-explicit information on the risk of invasions as well as facil-

itating the development of efficient policies and management

measures for forest conservation. Although limited to forest eco-

systems, this work highlights the suitability of using broad-scale

periodic forest surveys to monitor invasive plants, as well as their

potential to contribute in the future to the necessary practical and

theoretical understanding of biological invasions.

Acknowledgements

The Spanish National Forest Inventory databases for Galicia

provinces were provided by MAGRAMA. This research was sup-

ported by the AEG-09-007 agreement funded by MAGRAMA and

INIA and AGL2010.21153 Project. The authors thank Adam Collins

for the careful editing of the English and to an anonymous reviewer

whose constructive comments improved the manuscript.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, in

the online version, at http://dx.doi.org/10.1016/j.foreco.2014.

05.058.

References

Alpert, P., Bone, E., Holzapfel, C., 2000. Invasiveness, invasibility and the role ofenvironmental stress in the spread of non-native plants. Perspect. Plant Ecol.Evol. Syst. 3, 52–66.

Alston, K.P., Richardson, D.M., 2006. The roles of habitat features, disturbance, anddistance from putative source populations in structuring alien plant invasionsat the urban/wildland interface on the Cape Peninsula, South Africa. Biol.Conserv. 132 (2), 183–198.

Anthos, 2012. Sistema de información de las plantas de España. Real JardínBotánico, CSIC-Fundación Biodiversidad. Recurso electrónico. <http://www.anthos.es> (accessed March 2014).

Barbati, A., Corona, P., Marchetti, M., 2007. A forest typology for monitoringsustainable forest management: The case of European forest types. PlantBiosyst. 141 (1), 93–103.

Blossey, B., Nötzold, R., 1995. Evolution of increased competitive ability in invasivenon-indigenous plants: a hypothesis. J. Ecol. 83, 887–889.

Boschetti, L., Roy D.P., Hoffmann, A.A., 2009. MODIS Collection 5 Burned AreaProduct – MCD45 User’s Guide, Version 2.0, November 2009.

Brooks, M.L., D’Antonio, C.M., Richardson, D.M., Grace, J.B., Keeley, J.E., Di Tomaso,J.M., Hobbs, R.J., Pellant, M., Pyke, D., 2004. Effects of invasive alien plants on fireregimes. BioScience 54 (7), 677–688, <https://earthdata.nasa.gov/sites/default/files/field/document/MODIS_Burned_Area_User_Guide_2.0.pdf>.

Callaway, R.M., Aschehoug, E.T., 2000. Invasive plants versus their new and oldneighbors: a mechanism for exotic invasion. Science 290, 521–523.

Carballeira, A., Reigosa, M.J., 1999. Effects of natural leachates of Acacia dealbataLink in Galicia (NW Spain). Bot. Bull. Acad. Sin. 40, 87–92.

Chytry, M., Jarosik, V., Pysek, P., Hajek, O., Knollová, I., Tichy, L., Danihelka, J., 2008.Separating habitat invasibility by alien plants from the actual level of invasion.Ecology 89 (6), 1541–1553.

Corona, P., Chirici, G., McRoberts, R.E., Winter, S., Barbati, A., 2011. Contribution oflarge-scale forest inventories to biodiversity assessment and monitoring. ForestEcol. Manage. 262 (11), 2061–2069.

Costermans, L., 1985. Native Trees and Shrubs of South-Eastern Australia. RigbyPublishers, Melbourne.

Crawley, M.J., 1993. GLIM for Ecologists. Blackwell Scientific, Oxford.Curtis, J.T., 1959. The Vegetation of Wisconsin: An Ordination of Plant Communities.

University Wisconsin Press, Madison, Wisconsin.Davidson, D.W., Morton, S.R., 1984. Dispersal adaptations of some Acacia species in

the Australian arid zone. Ecology 65, 1038–1051.Davis, M.A. et al., 2000. Fluctuating resources in plant communities: a general

theory of invasibility. J. Ecol. 88, 528–534.de la Cueva, A.V., García del Barrio, J.M., Ortega, M., Sánchez-Palomares, O., 2006.

Recent fire regime in peninsular Spain in relation to forest potentialproductivity and population density. Int J. Wildland Fire 15, 397–405.

de la Cueva, A.V., Quintana, J.R., Cañellas, I., 2012. Fire activity projections in theSRES A2 and B2 climatic scenarios in peninsular Spain. Int. J. Wildland Fire 21,653–665.


de la Cueva, A.V., 2014. Case studies of the expansion of Acacia dealbata in thevalley of the river Miño (Galicia, Spain). Forest Syst. 23 (1), 3–14.

Di Castri, F., 1989. History of biological invasions with special emphasis on the OldWorld. In: Drake, J.A., Mooney, H.A., di Castri, F., Groves, R.H., Kruger, F.J.,Rejmánek, M., Williamson, M. (Eds.), Biological Invasions: A Global Perspective.John Wiley & Sons, New York, USA.

EC, 2011. Communication from the Commission to the European Parliament, theCouncil, the Economic and Social Committee and the Committee of the Regions– Our Life Insurance, Our Natural Capital: An EU Biodiversity Strategy to 2020.COM 2011, 244.

EEA, 2012. Corine Land Cover 2000 seamless vector data. European EnvironmentAgency. <http://www.eea.europa.eu/data-and-maps/data/corine-land-cover-2000-clc2000-seamless-vector-database-4#tab-gis-data> (accessed February2014).

Foxcroft, L.C., Richardson, D.M., Rejmánek, M., Pyšek, P., 2010. Alien plant invasionsin tropical and sub-tropical savannas: patterns, processes and prospects. Biol.Invasions 12 (12), 3913–3933.

Fuentes-Ramírez, A., Pauchard, A., Cavieres, L.A., García, R.A., 2011. Survival andgrowth of Acacia dealbata vs. native trees across an invasion front in south-central Chile. Forest Ecol. Manage. 261 (6), 1003–1009.

Gassó, N., Thuiller, W., Pino, J., Vilà, M., 2012. Potential distribution range of invasiveplant species in Spain. NeoBiota 12, 25–40.

González-Muñoz, N., Costa-Tenorio, M., Espigares, T., 2012. Invasion of alien Acaciadealbata on Spanish Quercus robur forests: Impact on soils and vegetation.Forest Ecol. Manage. 269, 214–221.

Gonzalo J. 2010. Diagnosis Fitoclimática de la España Peninsular. Hacia un modelode clasificación funcional de la vegetación y de los ecosistemas peninsularesespañoles. In: Serie técnica Naturaleza y Parques Nacionales. OrganismoAutónomo de Parques Nacionales. Ministerio de Medio Amiente y MedioRural y Marino de España (ed). Madrid.

Hejda, M., Pyšek, P., Jarošík, V., 2009. Impact of invasive plants on the speciesrichness, diversity and composition of invaded communities. J. Ecol. 97 (3),393–403.

Heymann, Y., Steenmans, C., Croissille, G., Bossard, M., 1994. Corine Land Cover – Atechnical guide. Office for Official Publications of European Communities. CEC,Luxembourg. <http://www.eea.europa.eu/publications/COR0-landcover>(accessed December 2013).

Higgins, S.I., Richardson, D.M., 1996. A review of models of alien plant spread. Ecol.Mod. 87, 249–265.

Higgins, S.I., Richardson, D.M., Cowling, R.M., 2001. Validation of a spatialsimulation model of a spreading alien plant population. J. Appl. Ecol. 38, 571–584.

Hollander, M., Wolfe, D.A., 1999. Non-Parametric Statistical Methods, second ed.Wiley, New York.

Hosmer, D.W., Lemeshow, S., 2000. Applied Logistic Regression, second ed. JohnWiley & Sons Inc., New York.

Hussain, M.I., Gonzalez, L., Reigosa, M.J., 2011. Allelopathic potential of Acaciamelanoxylon on the germination and root growth of native species. Weed Biol.Manage. 11 (1), 18–28.

Kolar, C.S., Lodge, D.M., 2001. Progress in invasion biology: predicting invaders.Trends Ecol. Evol. 16 (4), 199–204.

Le Maitre, D.C., Gaertner, M., Marchante, E., Ens, E.J., Holmes, P.M., Pauchard, A.,O’Farrell, P.J., Rogers, A.M., Blanchard, R., Blignaut, J., Richardson, D.M., 2011.Impacts of invasive Australian acacias: implications for management andrestoration. Divers. Distrib. 17, 1015–1029.

Levine, J.M., D’Antonio, C.M., 1999. Elton revisited: a review of evidence linkingdiversity and invasibility. Oikos 87, 15–26.

Lonsdale, W.M., 1993. Rates of spread of an invading species – Mimosa nigra innorthern Australia. J. Ecol. 81, 513–521.

Lorenzo, P., González, L., Reigosa, M.J., 2010. The genus Acacia as invader: thecharacteristic case of Acacia dealbata Link in Europe. Ann. Forest Sci. 67 (1), 101.

Lorenzo, P., Pazos-Malvido, E., Rubido-Bará, M., Reigosa, M.J., González, L., 2012.Invasion by the leguminous tree Acacia dealbata (Mimosaceae) reduces thenative understorey plant species in different communities. Aust. J. Bot. 60 (8),669–675.

Lund, H.G., 1998. A comparison of multipurpose resource inventories (MRIs)throughout the world. Working Paper No. 14. Joensuu, European ForestInstitute, Finland, pp. 46.

Marchante, H., Marchante, E., Freitas, H., 2003. Invasion of the Portuguese duneecosystems by the exotic species Acacia longifolia (Andrews) Willd: effects atthe community level. In: Child, L.E., Brock, J.H., Brundu, G., Prach, K., Pyšek, P.,Wade, P.M., Williamson, M. (Eds.), Plant Invasions: Ecological Threats andManagement Solutions. Backhuys Publishers, The Netherlands, pp. 75–85.

Moreno, J.M., Vázquez, A., Vélez, R., 1998. Recent history of forest fires in Spain. In:Moreno, J.M. (Ed.), Large Forest Fires. Backhuys Publishers, Leiden, TheNetherlands, pp. 159–185.

Ough, K., 2001. Regeneration of wet forest flora a decade after clear-felling orwildfire- is there a difference? Austral. J. Bot. 46, 645–664.

Pyšek, P., Jarošik, V., Kuèera, T., 2002. Patterns of invasion in a temperate naturereserves. Biol. Conserv. 104, 13–24.

Pyšek, P., Hulme, P.E., 2005. Spatio-temporal dynamics of plant invasions: linkingpattern to process. Ecoscience 12, 302–315.

Rejmánek, M., Richardson, D.M., 1996. What attributes make some plant speciesmore invasive? Ecology 77, 1655–1661.

Rejmánek, M., Richardson, D.M., Pyšek, P., 2005. Plant invasions and invasibility ofplant communities. In: van der Maarel, E. (Ed.), Vegetation Ecology. BlackwellScience, Oxford, UK, pp. 332–355.

Richardson, D.M., Holmes, P.M., Esler, K.J., Galatowitsch, S.M., Stromberg, J.C.,Kirkman, S.P., Hobbs, R.J., 2007. Riparian vegetation: degradation, alien plantinvasions, and restoration prospects. Diver. Distrib. 13 (1), 126–139.

Richardson, D.M., Brown, P.J., 1986. Invasion of mesic mountain fynbos by Pinusradiata. S. Afr. J. Bot. 52, 529–536.

Richardson, D.M., Daehler, C.C., Leishman, M.R., Pauchard, A., Pyšek, P., 2010. Plantinvasions: theoretical and practical challenges. Biol. Invasions 12, 3907–3911.

Romero Buján, M.I.R., 2007. Flora exótica de Galicia (noroeste ibérico). Bot.Complutensis 31, 113–125.

Sanz-Elorza, M., Dana, E.D., Sobrino, E., 2004. Atlas de las plantas alóctonasinvasoras de España. Madrid, D.G. de la Biodiversidad, Ministerio de MedioAmbiente, pp. 384.

Saura, S., Carballal, P., 2004. Discrimination of native and exotic forest patternsthrough shape irregularity indices: an analysis in the landscapes of Galicia,Spain. Landscape Ecol. 19 (6), 647–662.

Sheppard, A.W., Shaw, R.H., Sforza, R., 2006. Top 20 environmental weeds forclassical biological control in Europe: a review of opportunities, regulations andother barriers to adoption. Weed Res. 46, 93–117.

Souza, L., Bunn, W.A., Simberloff, D., Lawton, R.M., Sanders, N.J., 2011. Biotic andabiotic influences on native and exotic richness relationship across spatialscales: favourable environments for native species are highly invasible. Funct.Ecol. 25 (5), 1106–1112.

Stohlgren, T.J., Barnett, D.T., Kartesz, J.T., 2003. The rich get richer: patterns of plantinvasions in the United States. Front Ecol. Environ. 1 (1), 11–14.

UNEP, 2010. Report of the Seventh Meeting of the Conference of the Parties to theConvention on Biological Diversity (UNEP/CBD/COP/35/4). Decision XI/28,UNEP.

Wangen, S.R., Webster, C.R., 2006. Potential for multiple lag phases during bioticinvasions: reconstructing an invasion of the exotic tree Acer platanoides. J. Appl.Ecol. 43 (2), 258–268.


IMPACTOS DEL CAMBIO GLOBAL DOCTORAL EN ECOSISTEMAS...

Documents

Transcript of IMPACTOS DEL CAMBIO GLOBAL DOCTORAL EN ECOSISTEMAS...