Repositorio Institucional de Documentos

2014 76 Yacine Kouba Distribución espacial, dinámica espacio-temporal, regeneración y diversidad en las comunidades de Quercus faginea del Pirineo Central Aragonés Departamento Director/es Geografía y Ordenación del Territorio López Alados, Concepción Director/es Tesis Doctoral Autor Repositorio de la Universidad de Zaragoza – Zaguan http://zaguan.unizar.es UNIVERSIDAD DE ZARAGOZA Departamento Director/es Yacine Kouba DISTRIBUCIÓN ESPACIAL, DINÁMICA ESPACIOTEMPORAL, REGENERACIÓN Y DIVERSIDAD EN LAS COMUNIDADES DE QUERCUS FAGINEA DEL PIRINEO CENTRAL ARAGONÉS Director/es Geografía y Ordenación del Territorio López Alados, Concepción Tesis Doctoral Autor 2014 Repositorio de la Universidad de Zaragoza – Zaguan http://zaguan.unizar.es UNIVERSIDAD DE ZARAGOZA Departamento Director/es Director/es Tesis Doctoral Autor Repositorio de la Universidad de Zaragoza – Zaguan http://zaguan.unizar.es UNIVERSIDAD DE ZARAGOZA Distribución espacial, dinámica espacio-temporal, regeneración y diversidad en las comunidades de Quercus faginea del Pirineo Central Aragonés Spatial distribution, spatio-temporal dynamic, regeneration, and diversity in the Quercus faginea communities of the Aragón’s Central Pyrenees ©Yacine Kouba Tesis doctoral 2014 TESIS DOCTORAL Distribución espacial, dinámica espacio-temporal, regeneración y diversidad en las comunidades de Quercus faginea del Pirineo Central Aragonés Memoria presentada por Yacine Kouba para optar al grado de Doctor en Geografía por la Universidad de Zaragoza 2014 DIRECTOR: Concepción L. Alados Profesor de investigación Instituto Pirenaico de Ecología (CSIC) Departamento de Geografía y Ordenación del territorio Facultad de Filosofía y Letras Universidad de Zaragoza A mes parents pour leur amour et surtout pour leur compréhension durant toutes ces années que j'ai passé loin d'eux A ma fiancée Rania pour son immense amour et son soutien moral et surtout pour avoir accepte de partage le rojet de sa vie avec moi “If you cannot do great things, do small things in a great way . ” Napoleon Hill (1883-1970) American writer III References 143 Capitulo 6: The importance of historical land use in shaping contemporary patterns of plant community in Mediterranean forests 152 Abstract 152 Introduction 153 Methods 154 Results 158 Discussion 163 Conclusions 167 References 167 Discusión general y conclusiones 172 Discusión general 172 Conclusiones 176 Referencias 178 IV Resumen/Summary 1 Resumen/Summary En muchos bosques Mediterráneos, Quercus faginea se considera como un componente estructural importante de las comunidades nativas porque ofrece hábitat para una amplia diversidad de comunidades de plantas y animales, y por lo tanto, es de gran interés para la conservación de los ecosistemas. A pesar de su importancia, esta especies es poco estudiada en comparación con otras especies tal como Q. ilex y Fagus syvatica. Esta tesis doctoral (i) analiza el efecto de los factores abióticos y el uso antrópico en la distribución de Q. faginea en el Pirineo Central, (ii) examina la dinámica espaciotemporal de los bosques de Q. faginea del Prepirineo Central durante el periodo entre 1957 y 2006, (iii) evalúa el papel del cambio del uso del suelo y el cambio climático en el establecimiento y la dinámica re-generacional de los bosques de Q. faginea, (iv) analiza la relación entre la expansión de Q. faginea observada en algunos campos abandonados y los cambios socioeconómicos en siete municipios del Prepirineo Central durante la segunda mitad del siglo XX, (v) estudia la organización de la diversidad florística a múltiples escalas espaciales e identifica los factores ambientales que han influenciado en la distribución espacial de la diversidad florística en los bosque de Q. faginea, e (vi) investiga el efecto del uso histórico del suelo en las comunidades vegetales (diversidad y composición florística) de los bosques de Q. faginea del Prepirineo Central. En primer lugar hemos examinado el efecto de los factores abióticos y el uso antrópico en la distribución de Q. faginea en el Pirineo Central. Los mapas de presencia-ausencia de Q. faginea, de las variables abióticas y del uso antrópico se han derivado de los mapas disponibles, como por ejemplo el mapa del tercer Inventario Nacional Forestal; INF3, y fotografías aéreas, usando el sistema de información geográfica. El efecto de las variables abióticas y del uso del suelo en la distribución de Q. faginea se ha cuantificado usando el modelo bayesiano “Bayesian Model Averaging (BMA)”. A gran escala, las variables abióticas (clima y litología), fueron los factores que tuvieron el mayor efecto en la distribución espacial de Q. faginea, sin embargo, las plantaciones introducidas recientemente y la presión del ganado de las primeras décadas del siglo XX han afectado negativamente la distribución de Q. faginea en esta zona. Resumen/Summary 2 Para evaluar la dinámica espaciotemporal de los bosques de Q. faginea durante el periodo entre 1957 y 2006, hemos realizado los mapas de distribución de Q. faginea en el Prepirineo Central durante los años 2006 y 1957 usando el mapa del tercer Inventario Nacional Forestal (INF3) y las fotografías aéreas tomadas en 2006 y 1957. Las ganancias y pérdidas en los bosques de Q. faginea se cuantificaron usando la matriz de cambios. La fragmentación, el aislamiento, y la irregularidad de las manchas se evaluaron usando una seria de índices del paisaje. Los factores más probables que influyeron en estos cambios se identificaron por medio de modelos estadísticos. Los resultados mostraron cambios substanciales en los patrones espaciales de los bosques de Q. faginea en el Prepirineo Central durante los últimos 50 años. Estos cambios se han reflejado claramente en un incremento considerable de la fragmentación, del grado de aislamiento, y en la irregularidad de las manchas. Los cambios en los patrones espaciales de los bosques de Q. faginea se asocian particularmente al aumento en las plantaciones de pinos y la antigua deforestación asociada con el aumento de las tierras cultivadas y los pastos. Además las carreteras actuaron como “atractores” para los cambios de uso del suelo y la deforestación, lo que influyó fuertemente en la variabilidad espacial de los bosques de Q. faginea. Para entender cómo los cambios del uso del suelo y el cambio climático han influenciado en la dinámica de las masas de Q. faginea y cómo han afectado al establecimiento de los individuos de esta especies durante el siglo XX, hemos seleccionado diez masas forestales dominadas por Q. faginea en el Prepirineo Central. En cada masa de Q. faginea se ha establecido un transecto de 500-m en un sitio elegido al azar. Cada transecto tiene 20 puntos de muestreo elegidos a un intervalo de 25-m. Usando el método “Point-quarter” en cada punto se ha identificado el árbol adulto de Q. faginea más cercano a este punto en los cuatro direcciones cardinales. Las siguientes variables se han recogido para cada uno de los cuatro árboles: diámetro a la altura del pecho (DBH) (cm), la altura del árbol (m), hábitat (monte bajo o terraza abandonada). Además se ha estimado la edad de 290 árboles en las diez masas muestreadas. Los datos de distribución de edad se analizaron usando la función “Power function” para estimar el reclutamiento de Q. faginea a lo largo del siglo XX. Para evaluar los efectos de las variables climáticas y del uso del suelo en el establecimiento de Q. faginea, hemos calculado los Resumen/Summary 3 coeficientes de correlación de Spearman (r s ) entre las variables climáticas y las que reflejan el uso de suelo, y el número de árboles establecidos observado y los residuales obtenidos de la función ajustada “Power function”. Desde finales de 1930, el reclutamiento de los nuevos individuos en los bosques de Q. faginea, ha mostrado una alta variabilidad interanual y el reclutamiento más alto se produjo entre 1965 y 1975. El establecimiento de los árboles se correlacionó negativamente con la temperatura media de las máximas del verano, tamaño de la población de los pueblos cercanos, y la cantidad de ganado, y se correlacionó positivamente con las precipitaciones anuales, precipitaciones del invierno, y las precipitaciones del invierno-primavera. Para estudiar la relación entre la expansión de Q. faginea y los cambios socioeconómicos en siete municipios del Prepirineo Central durante la segunda mitad del siglo XX, hemos cuantificado en primer lugar las ganancias (ha) en los bosques de Q. faginea por municipio entre 1957 y 2006, luego hemos correlacionado estas ganancias con los cambios -entre 1957 y 2006en el tamaño de la población del municipio, y el número de agricultores y cabezas de ganado por cada municipio. La superficie de las nuevas áreas adquiridas por los bosques de Q. faginea ha aumentado significativamente con el decremento del tamaño de población, y el número de agricultores y cabezas de ganado. En general, la emigración rural que ha ocurrido en esta área desde los años sesenta ha generalizado un abandono masivo de las tierras agrícolas y una disminución en la presión ganadera, lo que ha permitido a Q. faginea colonizar algunas de las tierras abandonadas sobre todo en los municipios con una disminución notable del número de agricultores. Para entender la organización de la diversidad florística a diferentes escalas espaciales (transecto, masa forestal, región) e identificar los factores ambientales que influyen en los patrones de la diversidad florística en los bosque de Q. faginea, se ha cuantificado la riqueza y la abundancia de las plantas vasculares en 30 transectos de 500-m establecidos en las 10 masas de Q. faginea muestreadas (3 transectos por masa) usando el método “Point-intercept”. Además, se ha estimado la edad media por masa, el coeficiente de variación de edad de la masa forestal, la abundancia de Q. faginea (en cada uno de los 30 transectos), el tipo de bosque Resumen/Summary 4 (monte bajo vs. bosque secundario), el área de la masa forestal, y la complejidad de la forma de la masa forestal (perímetro/área). Hemos usado la partición jerárquica-multiplicativa de la diversidad y el índice de Hill para evaluar los patrones de la diversidad florística a distintas escalas espaciales, y el análisis de redundancia (RDA) para examinar el efecto de las variables ambientales que describen las características de las masas de Q. faginea en la variación de la composición florística. Los resultados han revelado que la mayor parte de la diversidad florística (diversidad gama) en las masas de Q. faginea del Prepirineo Central es el resultado de la diferenciación florística entre las masas, dicho de otra manera, es resultado de la diversidad beta entre las masas de Q. faginea. Esta misma diversidad -diversidad betaestá estructurada en gran parte por unos factores que son muy influenciados por el uso histórico de la masa forestal como por ejemplo la edad media de la masa forestal, el coeficiente de variación de la edad, y la abundancia de Q. faginea. Para investigar los efectos de la gestión histórica de los bosques y el uso histórico de las tierras agrícolas en las comunidades vegetales actuales de los bosques de Q. faginea, hemos comparado la diversidad y la composición de la comunidad vegetal entre masas de Q. faginea que han establecido en las terrazas abandonadas y las masas del monte bajo. Además las masas de Q. faginea se han evaluado basándose en su edad (jóvenes vs. viejas) y la intensidad de la gestión histórica. Los resultados han revelado diferencias significativas en cuanto a la composición florística entre las masas de Q. faginea. Esta variación florística es debida principalmente a diferencias en riqueza y equitabilidad “evenness” de especies que desarrollan en diferentes estadios sucesionales: especies de la fase inicial de la sucesión vegetal, especies de la fase intermedia de la sucesión vegetal, y especies de la fase final de la sucesión vegetal. Los resultados han revelado también que las masas viejas de Q. faginea albergan pocas especies que se consideran como especialistas de los bosques. Ello respalda la idea de que la recolonización de los bosques alterados por especies especialistas es muy difícil incluso cuando las masas forestales se dejan sin ningún tipo de gestión por mucho tiempo. Resumen/Summary 5 In many mesic forests in Mediterranean environments, Q. faginea is an important structural component of native plant communities because it provides habitat for a wide diversity of plant and animal communities and, therefore, is of great interest for ecosystem conservation. In spite of the importance of this species, exist few studies compared to other species such as Q. ilex and Fagus syvaltica. This PhD thesis (i) analyzes the effects of abiotic and anthropogenic factors on the distribution of Q. faginea forests in the Central Pyrenees, (ii) examine the spatiotemporal dynamics of Q. faginea forests over period from 1957 to 2006, (iii) assess the roles of land-use and climate change on the establishment and regeneration dynamics of Q. faginea forests in the Central Pre-Pyrenees, (iv) analyzes the relationship between the observed expansion of Q. faginea in some abandoned lands and socioeconomic changes (i.e. population number, number of farmers and livestock) in seven municipalities of the Central Pre-Pyrenees during the second half of the 20th century, (v) study the organization of plant diversity at multiple spatial scales and identify the environmental factors that might have patterned plant species diversity in humanaltered oak forests, and (vi) investigate the effects of the previous land management on contemporary plant communities (plant diversity and composition) in the oak forests of the Central Pre-Pyrenees. Firstly we examined the effects of abiotic and anthropic factors on the distribution of Q. faginea in the Central Pyrenees. Information on the presence-absence of Q. faginea, and abiotic and anthropic variables, were derived using GIS based on the available maps (e.g. the third Spanish National Forest Inventory map; IFN3) and aerial photographs. The effect of abiotic and land use variables on Q. faginea distribution was quantified using Bayesian Model Averaging (BMA). On a broad scale, abiotic variables; i.e. climate and lithology, were the factors that had the greatest effect on the spatial distribution of Q. faginea; although, recently introduced pine plantations and previous livestock pressure have had a negative effect on the distribution of Q. faginea in the region. To assess the spatiotemporal dynamics of Q. faginea forests over period from 1957 to 2006 we created maps of Q. faginea distribution over the Central Pre-Pyrenees in 2006 and 1957 using the third Spanish National Forest Inventory map (IFN3) and aerial photographs from 2006 and 1957. Gains and losses in Q. faginea forests were Resumen/Summary 6 quantified by means of construction of matrix of changes. Patch fragmentation, isolation, and irregularity were assessed using a set of standard landscape metrics. We also identified, the probable factors influencing these changes using statistical models (e.g. Bayesian Model Averaging; BMA). The results revealed substantial changes in the spatial patterns of Q. faginea forests over the last 50 years. These changes have been clearly reflected in noteworthy increase of fragmentation, isolation degrees, and patch irregularity. Changes in the spatial patterns of Q. faginea forests were particularly driven by the vast introduction of pine plantations and the former deforestation, associated with increasing the amount of croplands and pastures. In addition, roads acted as attractors for changes in land use and deforestation, which influenced strongly the spatial variability of Q. faginea forests. To understand how changes in land use and climate influence the dynamics of Q. faginea forest stands and how they affected tree establishment in the 20th century we selected ten stands that were dominated by Q. faginea in the Central PrePyrenees. Within each stand, a 500-m linear transect was established at a randomly chosen location. Each transect had sampling points (n=20) at 25-m intervals. Using the “Point-quarter” method at each point, we identified the closest adult Q. faginea tree in each of the four cardinal directions. The following variables were recorded for each of the four trees: diameter at breast height (DBH) (cm), tree height (m), habitat (coppice stand or abandoned terrace). The age of 290 trees were also estimated in the teen sampled stands. The age distribution data was analyzed by using “Power function” to estimate the establishment of Q. faginea trees along the 20th century. To assess the effects of climate and land use variables on Q. faginea establishment, we calculated Spearman correlation coefficients (rs) between the climate and land use variables and both the observed number of trees established and the residuals obtained from the fitted “Power function”. Since the late 1930s, Q. faginea became established episodically, and the highest peak occurred between 1965 and 1975. Tree establishment was negatively correlated with mean summer maximum temperature, population size of nearby villages, and the amount of livestock, but was positively correlated with annual, winter, and winter-spring precipitation. To study the relation between the expansion of Q. faginea and socioeconomic changes in seven municipalities of the Central Pre-Pyrenees over the second half of the 20th Resumen/Summary 7 century, we first quantified the gains (ha) in Q. faginea forests per municipality between 1957 and 2006, then we correlated these gains with changes -between 1957 and 2006in population size and the number of farmers, and livestock in each municipality. The amount of gains in Q. faginea increased significantly with a decrease in population size and the number of farmers and livestock. Overall, the rural emigration that occurred in this area since 1960s has generated a massive abandonment of agricultural lands and a decrease in livestock pressure, which allowed Q. faginea to colonize some of the abandoned lands, essentially in the municipalities that known a high reduction in the number of farmers. To understand how plant species diversity changes across different spatial scales (i.e. transect, stand, and entire region) and to identify the environmental factors that might have patterned plant species diversity in Q. faginea forests, the richness and abundance of all vascular plant species were quantified for 30 500-m transects established in the ten sampled oak stands (3 transects per stand) by using “Pointintercept” method. Furthermore, for each stand we estimated the mean age, coefficient of variation of tree age, Q. faginea abundance (in each of the 30 transects), stand size, and stand-form complexity (Perimeter/area of stand). We used multiplicative diversity partitioning and Hill Index to assess plant diversity patterns at the three spatial scales, and redundancy analysis (RDA) to test the effects of environmental variables that describe Q. faginea stands’ characteristics on the compositional variation of plant species. The results revealed that a great part of plant diversity (gama diversity) in the Q. faginea stands of the Central Pre-Pyrenees is a result of floristic differentiation among stands (i.e. among-stands beta diversity). This beta diversity is mainly structured by the factors that are strongly influenced by the historical use of oak stands, such as mean stand age, the coefficient of variation of tree age, and Q. faginea abundance. To investigate the effects of previous forest management and agricultural land use on contemporary plant communities in oak forests, we compared the plant diversity and composition of abandoned coppices and secondary growth stands. In addition, the stands were assessed based on their ages (young vs. old stands), and historical management intensity. The finding revealed a significant compositional differentiation between stands. This compositional variation is due to differences in Resumen/Summary 8 the richness and evenness of plant species of different habitat preferences (i.e. early-, mid-, late-successional species). The results showed also that the old oak stands harbored a considerably small share of forest specialists, which support the suggestion that the re-colonization by forest specialists can be difficult, even if the stand is left unmanaged for a long time. Introducción general 15 aclarado por perturbaciones (Figura 3). Los estratos arbustivo y herbáceo tienen una densidad variable dependiendo del grado de apertura del dosel arbóreo. Desde un punto de vista fitosociológico, se encuentra mezclado con pinares (Pinus ssp) naturales o repoblados o encinares (Quercus ilex). Hay además, algunas masas de Q. faginea puro con presencia de algunas especie arbustivas como Buxus sepervirens, Quercus coccifera y Juniperus communis. En esta comunidad también podemos encontrar Genista scorpius, Arctostaphylos uva-ursi. Protegidos por el Buxus sepervirens aparecen a finales de invierno narcisos (Narcissus ssp) y Viola alba. En sus claros hallaremos plantas mediterráneas como Thymus vulgares, Lavandula latifolia, Linum suffruticosum, Linum narbonense. De un punto de vista de la ecología del paisaje, los bosques de Q. faginea forman en algunos sitios masas grandes continuas y en otros sitios forma teselas pequeñas fragmentadas y aisladas. En su límite altitudinal superior, Q. faginea contacta en función de la humedad con formaciones de pino silvestre (Pinus sylvestris) o con hayedos (Fagus sylvatica). Es importante señalar también que en Pirineo Central hay dos tipos de bosques de quejigo: 1. Los quejigares que se consideran como montes bajos: desde la antigüedad se han explotado bastante para leñas, carbón y pastos, por lo que forman masas fragmentadas y muy aclaradas (Sancho et al., 1998). En los alrededores de los pueblos se han conservado algunos árboles centenarios por su producción de bellota. 2. Los quejigares recientemente formados en las terrazas abandonadas: Estos bosques se han establecido en las tierras abandonadas, antiguamente usadas como tierras agrícolas. Se han instalado especialmente durante la segunda parte del siglo Figura 3: Una foto de un bosque de quejigo en el Prepirineo (elaboración propia) Introducción general 16 veinte como resultado de la sucesión vegetal natural (invasión del quejigo en las terrazas abandonadas). La gestión histórica de Quercus faginea en el Pirineo Central Aragonés Los usos de esta especie han variado poco a lo largo de la historia, pero sí ha existido variación en su intensidad de acuerdo a los niveles de desarrollo demográfico y tecnológico de las sociedades humanas (Sancho et al., 1998). Los principales productos obtenidos tradicionalmente de esta especie han sido las leñas, seguidos con menor importancia por la madera y la bellota. La utilización de los quejigares como fuente de combustible, ya sea en forma de leña o de carbono se debe a la buena calidad de este producto. Desde la antigüedad existen testimonios de esta dedicación, así Teofrasto en su “Historia de las Plantas” y Plinio el Viejo en su “Historia Natural” hacen referencias frecuentes al uso generalizado y buena calidad de la leña y el carbón de los quejigos en particular y los robles en general (Anbré, 1962; Díaz-Regaños, 1988; Sancho et al., 1998). La leña y el carbón del quejigo se han empleado como fuentes de energía en las actividades domésticas de las sociedades rurales pero también en actividades industriales (Sancho et al., 1998). Este aprovechamiento ha sido la principal causa de la deforestación y alteración de los quejigares. A partir de la edad moderna la presión sobre los bosques en general se acelera debido al aumento demográfico con el consiguiente aumento de la demanda de combustible, madera y terreno para el cultivo (Sancho et al., 1998; Barbero et al., 1990). También se ha constatado el uso de madera de esta especie para la obtención de vigas y pilares en la construcción, pero se trata de un uso puntual debido a la dificultad de encontrar ejemplares con porte adecuado. A parir de la primera mitad del siglo veinte, el descuaje de masas de quejigo para el cultivo agrícola y la extensión de los pastos han causado fuerte impacto en los bosques del quejigo del Pirineo Aragonés, determinando la configuración actual de la estructura de estos bosques. En el caso más extremo ha conducido a la disminución del área ocupada por los bosques de quejigo y la sustitución de la especie. La introducción de las plantaciones de pinos (especialmente Pinus niga y Pinus syvestris) a partir de segunda mitad del siglo veinte ha reducido el área total ocupada por los bosques de quejigo. Estas especies, que se caracterizan por un Introducción general 17 crecimiento rápido, están ocupando áreas que estaban ocupadas anteriormente por el quejigo. El abandono de las tierras no fértiles, que ha sucedido en el Pirineo Aragonés, particularmente desde la segunda mitad del siglo veinte, ha permitido al quejigo recolonizar algunas zonas de estas tierras abandonadas como consecuencia de un proceso de sucesión vegetal natural. Zonas de estudio Situación geográfica y descripción de la zona de estudio En primer lugar se ha seleccionado una zona muy amplia del Pirineo Central Aragonés con un área de 4394 Km 2 (Figura 4). Esta zona abarca la mayor parte de los bosques del quejigo en el Pirineo Central Aragonés. Geográficamente el área de estudio está situada en la provincia de Huesca y está delimitada por el río Cinca en el este, el río Aragón en el oeste, la cuidad de Huesca en el sur y por las fronteras con Francia en el norte. La zona tiene un carácter rural, incluye 324 núcleos de población, ninguno de ellos, a excepción de los municipios de Jaca, Fraga y Sabiñánigo con población mayor de 4000 habitantes. Esta zona se ha elegido para estudiar a gran escala el efecto de los factores abióticos y el uso de suelo sobre la Figura 4: Modelo Degital de Elevaciones (MDE; CINTA, 2013) y delimitación de las áreas de estudio Introducción general 18 distribución espacial de Q. faginea. En segundo lugar se ha elegido una zona reducida en el Prepirineo Aragonés con un área de 1363 Km 2 (Figura 4). La zona incluye una gran parte de la sierra de Guara, la sierra de Javierre, sierra de Bones, sierra de Belarre, sierra Alta, sierra Caballera, sierra de la Gabardiella, sierra de Aineto, sierra de San Pedro, sierra de Portiello, sierra de Bescos, sierra de Villacampa, y sierra de Picardiello. Esta zona se ha seleccionado para estudiar a escala media la evolucion espaciotemporal de los bosques de quejigo, asõ como identificar los factores del uso antropico que han conducido la dinamica espaciotemporal de esta especies. Por ultimo se han seleccionado diez masas forestales dominadas por el quejigo (elegidas segun el tipo de habitat, es decir si es un monte bajo o una terraza abandonada) en el Prepirineo. Las diez masas estan situadas cerca de los siguientes pueblos: Abena (AB), Ara (AR), Arguõs (AG), Belsue (BL), Ipies (IP), Ibort (IB), Lucera (LU), Nocito (NO), Rapun (RP), y Rasal (RA) (Figura 4). En estas masas se ha estudiado la dinamica de las masas de quejigo durante el siglo XX, asõ como se ha estudiado el efecto del uso historico de estas masas forestales en la diversidad y composicion florõstica. Litología, geomorfología y relieve Tradicionalmente se han distinguido dos unidades geológicas y geomorfológicas en el Pirineo: El Pirineo Axial formada por materiales del ciclo hercínico (principalmente sedimentos paleozoicos y plutones graníticos) y el Prepirineo constituido por sedimentos deformados durante el ciclo alpino con predominio de materiales calcáreos (Soler & Puigdefábregas, 1970, 1972). Este último consta a su vez de cuatro partes (Figura 5): las Sierras Interiores (dominadas por calizas “Cretáceo-Paleoceno” y areniscas del “flysch luteciense”), la Depresión Prepirenaica (dominada por margas, areniscas, y los conglomerados del Oligoceno), las Sierras Exteriores (dominadas por areniscas ludiense, margas, calizas, y conglomerados), y la Depresión del Ebro (viene dominada por relieves planos con cerros tabulares o sasos entre vales u hondonadas y materiales sedimentarios miocénicos) (Villar et al., 1997). Hay que señalar que la Depresión del Ebro no forma parte de nuestra zona de estudio. Introducción general 19 Clima De oeste a este se establece una transición climática, de forma que la parte más occidental tiene características oceánicas mientras que en la parte más oriental se percibe mejor la influencia mediterránea (Lasanta et al., 2002). Las principales características climatológicas del área de estudio son las siguientes: Las precipitaciones: Las precipitaciones anuales superan los 500 mm en toda el área de estudio, aproximándose a los 2000 mm en las crestas de los picos más altos (Lasanta et al., 2002). La cantidad de lluvia desciende rápidamente hacia el sur. Así, Sabiñánigo recibe entre 800 y 900 mm anuales de promedio, y Aínsa unos 1000 mm (Del Valle, 2000). La precipitación vuelve a aumentar algo en el Prepirineo, sobre todo en el Occidental (Loarre) y en el central (Guara), donde se superan los 1000 mm (Del Valle, 2000). Las estaciones más lluviosas son la primavera y el otoño (Del Valle, 2000). El verano es la estación más seca, pero en ocasiones pueden caer precipitaciones intensas de tipo tormentoso (Del Valle, 2000). La temperatura: La temperatura media anual oscila entre 9º C y 11º C para las localidades más bajas, con importantes contrastes estacionales y diarios (Lasanta Figura 5: Bloque-diagrama geológico y estructural del Pirineo Aragonés (Villar et al., 1997) Introducción general 20 et al., 2002). Por encima de 1600 m la temperatura media anual no supera 6º C (Creus, 1987). La isoterma de 0º C se sitúa a 1600 m para el periodo de diciembre a marzo (Lasanta et al., 2002). La temperatura media desciende rápidamente con la altura, a razón de 0,6º C por cada 100 m de ascenso aproximadamente. Por ello, la temperatura depende mucho del factor topográfico. La temperatura media anual en Candanchú es de 5,2º C y los meses con este valor por debajo de 0º C son diciembre, enero y febrero. En Jaca la temperatura anual asciende hasta 11º C y ningún mes tiene valores medios por debajo de 0º C (Del Valle, 2000). El mes más frío suele ser enero, y el más cálido julio. En ocasiones se registran valores mínimos muy bajos, próximos a -20º C, pero muy esporádicamente, sólo cuando se producen invasiones de aire ártico o siberiano en invierno (Del Valle, 2000). Diferencias espaciales en precipitación y temperatura: Las características medias descritas hasta ahora varían mucho entre unas zonas y otras. Estas variaciones espaciales principalmente son: -A medida que avanzamos hacia el sur, la temperatura tiende a aumentar y la precipitación a disminuir. -A medida que nos desplazamos desde el oeste hacia el este la precipitación también tiende a disminuir y la temperatura a aumentar (suponiendo que nos mantenemos a la misma altura), aunque estas tendencias son compensadas en parte por la mayor elevación de la Cordillera. Paisaje y pisos bioclimáticos En el Pirineo Aragonés los paisajes se ordenan en función de la altitud, pues ésta determina las condiciones climáticas, que es un factor decisivo en la instalación de la vegetación y las actividades humanas (Benlloch, 2002). Por otro lado las variaciones ambientales y orográficas de esta zona condicionan la existencia de los diferentes pisos bioclimáticos (Figura 6). Las cumbres de las montañas, por encima de los 2800 m de altitud, constituyen el piso nival, dónde sólo unas pocas plantas localizadas en ambientes favorables consiguen sobrevivir. Por debajo del nival se extiende el piso alpino que se encuentra por encima de 2200-2300 m de altitud y se caracteriza por unas condiciones climáticas extremadamente duras. En este piso únicamente pueden desarrollarse plantas herbáceas y subarbustivas que en gran Introducción general 21 parte permanecen muchos meses (más de seis) bajo la nieve, lo que les protege (Benlloch, 2002). En las zonas donde se ha llegado a formar algo de suelo, la vegetación natural correspondiente a pastizales alpinos densos con dominio de especies vivaces caméfitas hemicriptófitas (Benlloch, 2002; Bueno, 2011) entre las que se encuentran muchos endemismos (Villar et al., 1997). El piso subalpino (entre 1800 y 2200 m aproximadamente) está ocupado principalmente por pinares de pino negro (Pinus uncinata) y en enclaves más húmedos aparecen abetales (Abies alba) (Benlloch, 2002). Hay que señalar que la potencialidad de los pastos de alta montaña llevó a los pirenaicos a ampliar el área de pastoreo a costa de estos pinares de pino negro. Así, muchas de las zonas más accesibles y de mejores suelos han sido deforestadas para su aprovechamiento como pastos Figura 6: Pisos bioclimáticos y sus comunidades vegetales en el Pirineo Aragonés (Villar et al., 1997) Introducción general 22 (García-Ruíz, 1976; Monserrat, 1968; Benlloch 2002). El piso montano húmedo (superior) está ocupado por hayedos (Fagus silvatica) y abetales (Abies alba), que se sitúan entre los 1000 y 1700 metros. El pino silvestre (Pinus silvestris) es el árbol típico del piso montano seco (inferior), donde convive con el abetal (Abies alba), llegando en ocasiones a subir hasta los dominios del pino negro (Pinus uncinata) (Benlloch, 2002). La explotación generalizada de todos los bosques del piso montano ha favorecido la dispersión del pino silvestre (Pinus silvestris), que espontáneamente ocupa los claros del quejigo (Q. faginea), los del hayedo (Fagus silvatica) y los del abetal (Abies alba). En los claros de los pinares de zonas secas aparece el erizón (Echinospartum horridum). En las zonas húmedas de los valles se asientan los bosques mixtos de frondosas, con gran variedad de especies (Benlloch, 2002). El piso basal del pirineo, fue destruido en gran parte para destinarlo al aprovechamiento agrícola. El quejigo ocupa una banda que se extiende por todo el Prepirineo, preferentemente en alturas comprendidas entre los 500 y 1000 metros. A menor altura que el quejigo crece la carrasca (Quercus ilex) que ocupa las solanas calizas del Prepirineo, con ella conviven Buxus sepervirens, Juniperus communis, Juniperus sabina y Arctostaphylos uva-ursi. Uso de suelo En función de las características climáticas y del relieve se escalonan los usos del suelo en la zona de estudio. El nivel superior del Pirineo Aragonés corresponde a los pastos supraforestales, que aparecen frecuentemente por encima de los 1600 m. Inmediatamente por debajo de los pastos supraforetales se sitúa el nivel forestal, mejor conservado en las umbrías (Lasanta, 2002). Como consecuencia de la deforestación - 5.000 10.000 15.000 20.000 25.000 30.000 35.000 40.000 45.000 50.000 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 Figura 7: Evolución de la población en la zona de estudio entre 1900 y 2000 (elaboración propia a partir de los datos del Instituto Aragonés de estadística). Introducción general 23 tradicional para el cultivo, el espacio agrícola se sitúa en el nivel inferior del Pirineo ocupando los fondos de valle y la parte baja de las laderas, si bien durante la segunda mitad del siglo XIX y las primeras décadas del siglo XX, la extensión del espacio cultivado fue muy superior, ocupando las actividades agrícolas sectores de los niveles superiores del Pirineo Aragonés (Lasanta, 2002). A partir de los años Figura 8: Mapa de usos de suelo en la zona de estudio (elaboración propia a partir del mapa de CORINE Land Cover 2000) Bosque Corrientes de agua Matorral de alta montaña Matorral Mediterráneo Megaforbios Pastos Mediterráneos Pastos supraforestales Rocas, gleras y heleros Zonas antropizadas Zonas húmedas Introducción general 24 sesenta se produjo una emigración masiva de la población local (Figura 7) hacia las ciudades en búsqueda de mejores condiciones de vida, lo que ha generalizado una despoblación y envejecimiento de los pueblos y un abandono masivo de las tierras y campos agrícolas. Como consecuencia de este abandono la superficie labrada se reduce a un tercio (Garcia Ruiz & Lasanta, 1990). Una parte de las áreas abandonadas las repuebla el Patrimonio Forestal de Estado, mientras el resto se recupera de forma natural. La Figura 8 demuestra los diferentes usos del suelo en la zona de estudio. Justificación de la tesis Pocos trabajos han tratado de estudiar los bosques de quejigo del Pirineo Aragonés aunque es una especie que juega un papel muy importante en la conservación de la diversidad biológica en esta zona. En el Pirineo Aragonés además de la antigua deforestación severa de los bosques de quejigo para el incremento de las tierras agrícolas y los pastos, esta especie está condicionada por los factores abióticos como el clima y la litología, por lo tanto definir los factores abióticos y los relacionados con el cambio del uso de suelo que afectan a la distribución y la evolución espaciotemporal de esta especie tiene una gran importancia a la hora de realizar los planes de gestión de estos bosques. Por otra parte se desconoce el efecto del cambio climático y el cambio del uso de suelo en la dinámica regeneracional de las masas del quejigo. Estudios realizados en otros ecosistemas han mostrado que estos factores pueden afectar negativamente al establecimiento de los árboles y el reclutamiento de nuevos individuos (Améztegui et al., 2010; Chauchard et al., 2010; Gimmi et al., 2010). Se desconoce también el efecto del uso histórico de los bosques de quejigo en la riqueza, diversidad, y composición de las especies que en ellos habitan, aunque muchos estudios han destacado la importancia del uso histórico en la diversidad y la composición de las especies vegetales que albergan (Estevan et al., 2007; Fahey & Puettmann, 2008; Lomba et al., 2011). En esta tesis se realiza un estudio multidisciplinario sobre los bosques del quejigo que aborda distintos enfoques: la distribución espacial, la evolución espaciotemporal, la dinámica de las masas forestales, y la diversidad florística. Introducción general 31 miden variables y compara el valor de la variable en una localidad con el valor en las localidades vecinas. Dado pares de localidades separadas por ciertas distancias, es la propiedad de variables aleatorias de tomar valores que son más similares (autocorrelación positiva) o menos similares (autocorrelación negativa) que lo esperado por pares de localidades aleatoreamente asociados (Legendre, 1993). El Índice de Moran se ha usado en los capítulos 1 y2 para definir la distancia mínima (entre las observaciones) a partir de la cual desaparece la autocorrelación espacial (Índice de Moran igual a cero) en las variables respuesta. #4. Análisis de la Varianza Multivalente semi-paramétrico (PERMANOVA): Técnica de análisis multivariante que permite cubrir los casos dónde hay más de una variable dependiente que no pueden ser combinadas de manera simple (por ejemplo, usando ANOVA) (Anderson, 2001). PERMANOVA utiliza las distancias entre cada par de observaciones para obtener una matriz de distancia (como el análisis de componentes principales normalmente conocido por sus siglas en inglés, PCA) sobre la que luego se calcula la significación de las variables explicativas con simulaciones de Monte Carlo. Aplica análisis de permutaciones sobre las matrices de distancia. Este análisis nos permite saber si las variables explicativas tienen un efecto sobre el conjunto de las variables respuesta (ejemplo: la matriz de sitios × especies). Calcula también la variabilidad explicada por cada una de las variables explicativas. Este método se ha usado en el capítulo 6 para identificar los factores ambientales que controlan la variación de la composición florística en los bosques de Q. faginea. #5. Escalamiento multidimensional no métrico (NMDS): El NMDS es una técnica multivariante de interdependencia que trata de representar en un espacio geométrico de pocas dimensiones las proximidades existentes entre un conjunto de objetos. El NMDS es un método de ordenación adecuado para datos que no son normales o que están en una escala discontinua o arbitraria (Minchin, 1987). NMDS es una técnica ampliamente utilizada en ecología para detectar gradientes en comunidades biológicas, tiene la ventaja de permite reducir la dimensionalidad de los datos originales y visualizar los resultados en un gráfico de ordenación. Los ejes resultantes de la ordenación NMDS se pueden también relacionar con distintas Introducción general 32 variables ambientales para determinar de manera indirecta el efecto de éstas sobre la matriz de sitios × especies. Hemos usado este método (capítulo 6) para examinar las diferencias en la composición florística entre masas forestales con distintas características (estructurales, espaciales, y topográficas). #6. Análisis de Redundancia (RDA): Es una técnica multivariante que permite representar en un espacio geométrico de pocas dimensiones las proximidades existentes entre un conjunto de objetos condicionado por una serie de variables predictoras. El RDA es una técnica de ordenación restringida (constrained ordination; Legendre & Legendre, 2012), lo que significa que la ordenación de los objetos representa solamente la estructura de los datos que maximiza la relación con una segunda matriz de variables predictoras. RDA relaciona dos matrices: la matriz de variables dependientes (por ejemplo una matriz de sitios × especies) y la matriz de variables independientes (por ejemplo una matriz de variables ambientales). La relación entre ambas matrices se hace por medio de una combinación de técnicas de regresión multivariante y análisis de componentes principales (Bocard et al., 2011). En esta tesis (capítulo 5), RDA se ha usado para estudiar la relación entre las características estructurales, espaciales, y topográficas de las masas forestales y la composición florística. #7. Partición Multiplicativa de la diversidad (MP): Este método está basado en la idea de Wittaker (1972): diversidad gama () es igual a la diversidad alfa () multiplicada por la diversidad beta (). Jost (2006) ha recomendado el uso de los números equivalentes (Hill numbers) en lugar de los índices de diversidad conocidos (p.e. riqueza, Shannon), y ha demostrado que el uso de los números equivalentes permite satisfacer los requisitos de la MP. La MP se ha usado en esta tesis (capítulo 5) para evaluar los patrones de biodiversidad a múltiples escalas espaciales (transecto, masa forestal, y área de estudio). #8. Métricas de paisaje: Una métrica describe la estructura espacial de un paisaje en un tiempo determinado. Se utilizan como herramientas para caracterizar la geometría y las propiedades espaciales de un parche (una entidad espacialmente homogénea) o un mosaico de parches (Leitao & Ahern, 2002). En esta tesis Introducción general 33 (capítulo 2) se han usado las siguientes métricas para cuantificar y comparar la configuración espacial de las manchas de Q. faginea entre 1957 y 2006: (1) el número total de manchas de Q. faginea, (2) el tamaño medio de mancha (ha), (3) la distancia media entre las manchas (m), (4) el area total ocupada por Q. faginea (ha), (5) la longitud total de los bordes de las manchas (km), (6) la media de la ratio perímetro-área de mancha. Referencias Améztegui, A., Brotons, L., Coll, L., 2010. Land-use changes as major drivers of mountain pine (Pinus uncinata Ram.) expansion in the Pyrenees. Global Ecology and Biogeography 19, 632-641. André, J., 1962. Plinie L’ancie. Histoire Naturelle, Livre XVI. Les belle lettres. París, 198 pp. Anderson, M.J., 2001. A new method for non-parametric multivariate analysis of variance. Austral Ecology 26, 32-46. Arroyo, S., 2002. Presencia, vitalidad y regeneración del Quercus faginea en Torres. Sumuntán, 16: 89-100. Aubert M., Bureau F., Alard D., Bardat J., 2004. Effect of tree mixture on the humic epipedon and vegetation diversity in managed beech forests (Normandy, France). Canadian Journal of Forest Research 34, 233-248. Augusto, L., Ranger, J., Binkley, D., Rothe A., 2002. Impact of several common tree species of European temperate forests on soil fertility. Annals of Forest Science 59, 233-253. Barbero, M., Bonin, G., Loisel, R., Quézel, P., 1990. Changes and disturbances of forest ecosystems caused by human activities in the western part of the Mediterranean basin. Plant Ecology 87,151-173. Battles, J.J., Shlisky, A.J., Barrett, R.H., Heald, R.C., Allen-Diaz, B.H., 2001. The effects of forest management on plant species diversity in a Sierran conifer forest. Forest Ecology and Management 146, 211-222. Introducción general 34 Benlloch, P.I., 2002. El medio natural de los Pirineos: límites y condicionantes para el desarrollo de actividades económicas. Ager 2, 9-42. Borcard, D., Gillet, F., Legendre, P., 2011. Numerical Ecology with R. Spriger, New York. Bueno, C.G., 2011. Las perturbaciones de jabalí en los pastos alpinos del Pirineo Central: Una aproximación multiescalar. Tesis doctoral, 190pp. Chauchard, S., Beilhe, F., Denis, N., Carcaillet, C., 2010. An increase in the upper treelimit of silver fir (Abies alba Mill.) in the Alps since the mid-20th century: a landuse change phenomenon. Forest Ecology and Management 259: 1406-1415. Chevan, A., Sutherland, M., 1991. Hierarchical partitioning. American Statistic. 45, 90-96. CINTA, 2013. Centro de Información Territorial de Aragón, Gobierno de Aragón, Zaragoza, Spain. http://sitar.aragon.es Cottam, G., Curtis, J.T., 1956. The use of distance measures in phytosociological sampling. Ecology 37: 451-460. Creus, J., 1987. Algunas características climáticas de la alta montaña en los Pirineos centrales. Actas del X congreso Nacional de Geografia. 137-146, Zaragoza. Del valle, J., 2000. El clima de montaña. Aplicación al Pirineo aragonés. Boreas Natural. http://www.aragonesasi.com/boreas/articulos/clima1.htm DMA, 2007. Atlas Climático de Aragón. Departamento de Medio Ambiente. Gobierno de Aragón, Zaragoza. Díaz-Regañon, J.M., 1988. Teofrasto. Historia de las plantas. Editorial Gredos. Madrid, 531 pp. Estevan H., Lloret F., Vayreda J., Terradas J., 2007. Determinants of woody species richness in Scot pine and beech forests: climate, forest patch size and forest structure. Acta Oecologica 31, 325-331. Introducción general 35 Ewers, R.M., Kliskey, A.D., Walker, S., Rutledge, D., Harding, J.S., Didham, R.K., 2006. Past and future trajectories of forest loss in New Zealand. Biological Conservation 133,312-325. Fahey, R.T., Puettmann, K.J., 2008. Patterns in spatial extent of gap influence on understory plant communities. Forest Ecology and Management 255, 28012810. Flinn, K.M., Vellend, M., 2005. Recovery of forest plant communities in postagricultural landscapes. Frontiers in Ecology and the Environment 3, 243250. Freitas, S.R., Hawbaker, T.J., Metzger, J.P., 2010. Effects of roads, topography, and land use on forest cover dynamics in the Brazilian Atlantic Forest. Forest Ecology and Management 259, 410-417. García-Ruiz, J.M., 1976. “Modos de vida y niveles de renta en el Prepirineo del Alto Aragón Occidental”, Jaca, Monografías del Instituto de Estudios Pirenaicos nº 106. Garcia-Ruiz, J.M., Lasanta, T., 1990. Land-use change in the Spanish Pyrenees. Mountain Research and Development 10, 267-279. Gimmi, U., Wohlgemuth, T., Rigling, A., Hoffmann, C.W., Bürgi, M., 2010. Land-use and climate change effects in forest compositional trajectories in a dry CentralAlpine valley. Annals of Forest Science 67, 701. Härdtle, W., Von Oheimb, G., Westphal, C., 2003. The effects of light and soil conditions on the species richness of the ground vegetation of deciduous forests in northern Germany (Schleswig-Holstein). Forest Ecology and Management 182, 327-338. Hermy, M., Verheyen, K., 2007. Legacies of the past in the present-day forest biodiversity: a review of past land-use effects on forest plant species composition and diversity. Ecological Research 22, 361-371. Hart, S.A., Chen, H.Y.H., 2006. Understory vegetation dynamics of North American boreal forests. Critical Reviews in Plant Sciences 25, 381-397. Introducción general 36 Hoeting, J.A., Madigan, D., Raftery, A.E., Volinsky, C.T., 1999. Bayesian model averaging: a tutorial. Statistical Science 14, 382-401. IPCC, 2007. Climate Change 2007. The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge. Jost, L., 2006. Entropy and diversity. Oikos 113:363-375. Kobayashi, Y., Koike, F., 2010. Separating the effects of land-use history and topography on the distribution of woody plant populations in a traditional rural landscape in Japan. Landscape and Urban Planning 95, 34-45. Larcher, W., 2000. Temperature stress and survival ability of Mediterranean sclerophyllous plants. Plant Biosystems 134, 279-295. 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 2,173-195. Legendre, P., 1993. Spatial autocorrelation: trouble or new paradigm? Ecology 74, 1659-1673. Legendre, P., Legendre, L., 2012. Numerical ecology, 3rd ed. Elsevier, New York Leitao, A.B., Ahern, J., 2002. Applying landscape ecological concepts and metrics in sustainable landscape planning. Landscape Urban Planning 59, 65-93. Linares, J.C., Camarero, J.J., Carreira, J.A., 2010. Competition modulates the adaptation capacity of forests to climatic stress: insights from recent growth decline and death in relict stands of the Mediterranean fir Abies pinsapo. Journal of Ecology 98, 592-603. Lomba, A., Vicente, J., Moreira, F., Honrado, J., 2011. Effects of multiple factors on plant diversity of forest fragments in intensive farmland of Northern Portugal. Forest Ecology and Management 262, 2219-2228. Introducción general 37 Minchin, P.R., 1987. An evaluation of relative robustness of techniques for ecological ordinations. Vegetatio 69, 89-107. Monserrat, P., 1968. “Los pastos pirenaicos y su importancia económica”. Pirineos 87-90, 133-152. Moran, P.A.P., 1950. Notes on continuous stochastic phenomena. Biometrika 37, 17-23. Ogaya, R., Peñuelas, J., Martínez-Vilalta, J., Mangirón, M., 2003. Effect of drought on diameter increment of Quercus ilex, Phillyrea latifolia, and Arbutus unedo in a holm oak forest of NE Spain. Forest Ecology and Management 180, 175-184. PNOA, 2006. Plan Nacional de Ortofotografía Aérea http://www.ign.es/wmsinspire/pnoa-ma. Pulido, F.J., Díaz, M., 2005. Regeneration of a Mediterranean oak: a whole-cycle approach. Ecoscience 12, 92-102. Rademacher, C., Neuert, C., Grundmann, V., Wissel, C., Grimm, V., 2004. Reconstructing spatiotemporal dynamics of Central European natural beech forests: the rule-based forest model BEFORE. Forest Ecology and Management 194,349-368. Raftery, A.E., Madigan, D., Hoeting, J.A., 1997. Bayesian model averaging for linear regression models. Journal of American Statistical Association 92,179-191. Ruíz de la Torre, J., 1988. Árboles y arbustos de la península ibérica. I.C.O.N.A. Sancho, M.P.J., Fernández, P.M.D., Albertos, S.M., Sánchez, L.G., 1998. Regiones de procedencia de Quercus pyrenaica Willd. Quercus faginea Lam. Quercus canariensis Willd. OAPN, Madrid. Sawchik, J., Dufrêne, M., Lebrun, P., Schtickzelle, N., Baguette, M., 2002. Metapopulation dynamics of the bog fritillary butterfly: modelling the effect of habitat fragmentation. Acta Oecologica 23, 287-296. Schwarz, O. 1964. Quercus. In Tutin et al. Flora europea. Vol I. Cambridge. Introducción general 38 Soler, M., and Puigdefábregas, J., 1970. Lineas generales de la Geología del Alto Aragón Occidental. Pirineos 96, 5-20. Soler, M., and Puigdefábregas, J., 1972. Esquema litológico del Alto Aragón Occidental. Pirineos 106, 5-15. Thomson, J.R., Nally, R.M., Fleishman, E., Horrocks, G., 2007. Predicting bird species distributions in reconstructed landscapes. Conservation Biology, 21:752-766 Vicente-Serrano, S.M., Lasanta, T., Gracia, C., 2010. Aridification determines changes in forest growth in Pinus halepensis forests under semiarid Mediterranean climate conditions. Agricultural and Forest Meteorology 150, 614-628. Vicioso, C., 1950. Revision del género Quercus en España. Boletín del Instituto Forestal de Investigaciones y Experiencias, 51. I.F.I.E. Ministerio de Agricultura. Madris, 194 pp. Villar, L., Sesé, J.A., Ferrández, J.V., 1997. Atlas de la Flora del Pirineo Aragonés. Vol I, Huesca y Zaragoza, Instituto de Estudios Altoaragoneses -Consejo de Protección de la Naturaleza de Aragón. Whittaker, R.H., 1972. Evolution and Measurement of Species Diversity. Taxon 21, 213-251. Capitulo 1 39 Plant Ecology, Volume 212, Issue 6, Page 999-107, December 2010 CAPITULO 1 Effects of abiotic and anthropogenic factors on the spatial distribution of Quercus faginea in the Spanish Central Pyrenees Y. Kouba* 1 , C. L. Alados 1 and C.G. Bueno 2 1 Pyrenean Institute of Ecology (CSIC), Avda. Montañana 1005. P. O. Box 202, E50080 Zaragoza, Spain; 2 Pyrenean Institute of Ecology (CSIC), Avda Rgmto Galicia s/n P. O. Box 64, Jaca E-22700, Huesca, Spain;*Corresponding author: Tel: +34 976 716034; fax: +34 976 716019; Email: [email protected] Abstract Abiotic factors often are the most important factors influencing a species’ distribution. Nevertheless, when investigating the underlying causes of a species’ distribution, it is important to assess both the abiotic and the anthropogenic factors (land-use variables) that might have influenced the species’ distribution. That is especially true in the Mediterranean Basin, where natural ecosystems have undergone significant changes in response to anthropogenic pressures in the region. In this study, we examined the effects of abiotic and anthropogenic factors on the distribution of Quercus faginea in the Spanish Central Pyrenees. Information on the presence-absence of Q. faginea, and abiotic and anthropogenic variables, were derived using GIS based on digital maps and aerial photographs. To identify and quantify the factors that have affected significantly the spatial distribution of Q. faginea, we used Bayesian Model Averaging and hierarchical partitioning. In the Spanish Central Pyrenees, on a broad scale, abiotic variables; i.e. climate and lithology, were the factors that had the greatest effect on the spatial distribution of Q. faginea; however, recently introduced pine plantations and previous livestock pressure has had a negative effect on the distribution of Q. faginea in the region. Capitulo 1 40 Keywords: Species distribution models (SDM), explanatory models, Bayesian Model Averaging, hierarchical partitioning, land-use variables. Introduction An understanding of the factors that influence the distribution of a species is important because it enables us to estimate the drivers of a species’ distribution within a region. To model the relationship between a species and its environment, ecologists often use empirically based statistical models (e.g., Coudun et al. 2006; Segurado and Araújo 2004; Thuiller et al. 2004). Statistical models of species distributions (i.e., presence-absence of species) quantify the relationships between the dependent variable and a set of explanatory variables such as temperature, slope, and elevation. These models can be used as (1) explanatory: To make inferences about which variables are important in controlling the distribution of the focal species , i.e., examine the underlying causes of a species’ distribution by examining the statistical significance of an explanatory variable influence on dependent variable or (2) predictive models: To estimate the spatial distribution of environments that are suitable for species distribution by creating a predictive map using the relationship between dependent variable and the predictors. But, in many cases, these models are used only to predict the potential distribution area of a species (i.e. as predictive models), while the underlying causes tend to be a secondary consideration. Consequently, few modeling studies have addressed those causes (e.g., Graham et al. 2004; Nally 2000). Usually, the spatial distributions of terrestrial species are studied in the context of abiotic variables only, i.e., climate (e.g. Araújo et al. 2005), climate and topography (e.g. Raxworthy et al. 2007), climate and soil conditions (e.g. Coudun et al. 2006), and climate and lithology (e.g. Gastón et al. 2009). These abiotic variables are postulated to be the most important factors influencing the distribution of a species. However, human land use (e.g., agricultural and livestock activities and reforestations) may affect the distribution of plant species (Randin et al. 2009; Dirböck et al. 2003) or communities (Fischer 1990). In the Mediterranean Basin, one of the world’s biodiversity hotspots (Myers et al. 2000), human activities have Capitulo 1 47 2007). Explanatory variables that had values of Pr (vs. ¹ 0) > 0.75 were identified as ‘‘key factors’’ (Nally et al. 2008; Viallefont et al. 2001). More than one key factor was identified in our analysis; therefore, to determine relative importance of each key factor, we used Hierarchical Partitioning (HP) in the ‘‘hier.part’’ package in R (Walsh and Nally 2008). HP estimated the “independent” contribution of each “key factor” to the total variance explained by the model (Chevan and Sutherland 1991; Nally et al. 2008). Log-Likelihood goodness of fit measure was used. A logistic model was most appropriate because the dependent variable was binary (presence or absence). Note that hierarchical partitioning, as currently implemented in the “hier.part” package, assumes a monotonic relationship between the dependent and the explanatory variables (Luoto et al. 2006). To assess the nature of the relationship between the dependent variable and each one of the key factors (i.e., linear or nonlinear) we used univariate Generalized Additive Models (univariate GAMs); the smoothed function was plotted for each univariate GAM. The linear and quadratic terms were used in the HP analysis for those key factors that showed a nonlinear relationship with the dependent variable. Results “Key factors” influencing the distribution of Q. faginea BMA identified the factors that had the most effect on the distribution of Q. faginea in the Spanish Central Pyrenees. Among the 11 explanatory variables included in the model, the following six were identified as key factors: three abiotic variables (lithology, slope, and water balance) and three anthropogenic variables (cost distance to pastures, cost distance to livestock roads, and distance to nearest plantation). BMA suggested that insolation, terrain curvature, number of frost days per year, distance to the nearest village, and distance to the nearest mine did not have a statistically significant effect on the distribution of Q. faginea in the region [Pr (vs. ¹ 0) < 0.25]. All of the key factors had a high probability of being included in the final model [Pr (vs. ¹ 0) > 0.94], which reflected their strong relationship with the occurrence of Q. faginea. Water balance and cost distance to the livestock roads were negatively correlated with the occurrence of Q. faginea because the posterior means (PM+SD) of the coefficients associated with each variable were negative Capitulo 1 48 (Table 1); i.e., the probability of occurrence of Q. faginea increased with a decrease in the water balance, and in accessible areas from livestock roads. Slope, lithology, distance to the nearest plantation, and cost distance to pastures had a positive effect on the likelihood that Q. faginea was present (Table 1), which indicated that the probability that Q. faginea was present increased with an increase in the slope (the cross tabulation between slope and the Q. faginea distribution maps indicated that almost all (> 90%) of the areas occupied by Q. faginea had a slope between 5º and 30º. This result indicated also that the probability that Q. faginea was present was higher in lithological zones that were formed by flysch-limestone rocks than it was in other zones. The probability of occurrence increased with an increase in the distance to the nearest plantation, and cost distance to pastures, which indicated that Q. faginea was more likely to be found away from pine plantations and in unreachable areas from pastures, than in other sites. Table 1 Bayesian Averaging Model (BMA) and hierarchical partitioning (HP) used for identifying the most important factors that affecting the distribution of Q. faginea over the Spanish Central Pyrenees. Pr (bvs. ¹ 0) is the posterior proaility that a variale had a non-zero coefficient, and (PM+SD) is the posterior means and standard deviation of the coefficients associated with each variale Variales BMA HP (%) Pr (bvs. ¹ 0) PM+ SD Lithology 100% 0.136±0.087 17.431 Slope 100% 0.4104±0.009 16.003 Curvature 03.45% 0.007±0.004 0 Water alance 100% -0.511±0.011 33.661 Insolation 21.20% -0.023±0.021 0 Numer of frosts days 06.05% 0.041±0.025 0 Cost distance to pastures 100% 0.072±0.004 10.103 Cost distance to livestock roads 99.76% -0.011±0.008 08.093 Distance to nearest mine 14.10% -0.012±0.027 0 Distance to nearest village 07.50% -0.026±0.231 0 Distance to the nearest plantation 94.70% 0.106±0.120 14.709 Independent explained variance HP analysis (Tale 1) suggested that the aiotic factors explained > 67% of the total independent variance, which reflects the importance of those factors on the Capitulo 1 49 distribution of Q. faginea. Water balance (33.66%), lithology (17.43 %), and slope (16.00 %) made the greatest independent contributions. Anthropogenic factors explained about 33% of the total independent variance, with distance to the nearest plantation (14.70%) and cost distance to pastures (10.10 %) making the greatest contribution. Those two variables had a negative effect on the probability of Q. faginea presence. Cost distance to livestock had a lowest contribution (08.09%) to the total independent explained variance. Discussion Effects of abiotic factors In large part, abiotic factors, particularly climate, lithology and slope, explained the spatial distribution of Q. faginea in the Spanish Central Pyrenees; precipitation and soil water recharge can have a significant effect on the establishment of Q. faginea plants (Corcuera et al. 2004), and a high soil water recharge favors the growth of Q. faginea (Corcuera et al. 2004). Our study indicated that water balance and the occurrence of Q. faginea were negatively correlated because of a particularity of the Central Pyrenees, where water balance increases with an increase in elevation and where temperatures are very low values in winter. Water balance was strongly correlated with elevation and mean monthly minimum temperatures (see statistical analysis). The cross tabulation between elevation and Q. faginea distribution maps indicated that, in the Central Pyrenees, all of the areas occupied by this species were between 450 m and 1500 m a.s.l.. The species does not occur above 1500 m because freezing temperatures hinder the establishment and growth of seedlings. As in our study, Sánchez de Dios et al. (2006) found that Q. faginea forests on the Iberian Peninsula were associated with continental areas that have low precipitation. The phenology of Q. faginea is similar to that of species that produce roots that can access deep water reserves (Castro and Montserrat 1998). In the Central Pyrenees, the probability of Q. faginea occurrence was high in areas that were characterized lithologically by flysch-limestone rocks and were rich in calcium carbonate, which generates calcareous soils that are suited to the establishment and growth of this species (Ceballos and Torre 1979; Sancho et al. 1998). In addition, the results showed that the probability of Q. faginea occurrence increases with a raise in slope values; the cross tabulation between slope and the Capitulo 1 50 Q. faginea distribution maps revealed that Q. faginea occupied areas that had a slope between 5º and 30º. In the Central Pyrenees those areas were mainly occupied by croplands, grasslands, and abandoned fields. Particularly, in grasslands and abandoned fields the high radiation and low water availability during summer, and herbaceous plants competition (i.e. herbaceous plants are strong competitor for resources, particularly water) limit the establishment of Q. faginea seedling (Rey Benayas et al. 2005), and therefore, prevent this specie from spreading in those areas. Effects of anthropogenic factors In the Central Pyrenees, anthropogenic factors appear to have played a secondary role in influencing the spatial distribution of Q. faginea, acting to restrain the expansion of this species. In particular, livestock grazing and introduced plantations have affected negatively the distribution of Q. faginea in the region. Elsewhere, the probability that Q. faginea was present was lowest in areas near pine plantations, which indicates that native forests can be severely affected by introduced species, particularly those that are fast growing and have strong dispersal abilities (Teixido et al. 2010). In the Central Pyrenees Q. faginea forests were extensively deforested to increase the amount of croplands and pastures (Lasanta 1989), but latter reforested with pine plantations (Amo et al 2007). Currently these plantations occupy areas that were previously covered by Q. faginea forests. In addition, the introduced species are characterized as fast growing and have dispersal ability, which allows them to be great competitors to Q. faginea, i.e., creating a competition that prevents Q. faginea from spreading in those plantations. This study showed that Q. faginea was less likely to occur in areas close to pastures that were accessible to livestock. In areas that experience livestock overgrazing, the regeneration of tree populations is practically impossible (Barbero et al. 1990). Livestock eliminates seedlings, which diminishes recruitment and, consequently, hinders species regeneration (Cierjacks and Hensen 2004). In some areas of the Central Pyrenees, Q. faginea forests were used in a “dehesas” system (Barbero et al. 1990; Montserrat 1990); i.e., a silvo-pastoral system that had sparse Q. faginea and perennial grass layers. In these particular ecosystems, the high grazing pressure increases the acorns consumption and Capitulo 1 51 creates abiotic conditions unfavorable for seedling establishment (Pulido and Díaz 2005, Plieninger 2006) which hamper the regeneration of Q. faginea and therefore lead to its disappearance from those sites. Between 1200-1500 m a.s.l., Q. faginea forests were harvested and summer livestock pastures were created (Lasanta et al. 2005). Conversely, Q. faginea seemed to be favored in accessible areas from livestock roads, which is probably a consequence of silvicultural practices (thinning, pruning) used by the forestry service alongside the livestock roads, which helped to maintain the health of Q. faginea stands near livestock roads. BMA suggested that distance to the nearest village did not affect the likelihood of Q. faginea occurrence, which means that the spatial distribution of this species was not directly affected by the human activities, concentrated around villages, probably because of the important decrease in the anthropogenic pressure on the territory (e.g. the abandonment of agricultural fields) as consequence of human exodus that occurred in the region during the second half of the 20th C. Conclusions This study has demonstrated that abiotic variables operating on a broad scale; e.g., lithology and climate were the main factors influencing the distribution of Q. faginea in the Spanish Central Pyrenees, and anthropogenic factors, particularly the recent addition of plantations and previous livestock pressure, affected negatively the distribution of Q. faginea in the study area. Much attention has been focused on the role of abiotic variables as main factors influencing the distributions of terrestrial tree species; however, our study showed that anthropomorphic changes in land use can affect the distribution of tree species, especially in the Mediterranean region, where natural ecosystems underwent substantial modifications caused by changes in the anthropomorphic use of land. Acknowledgments The Spanish CICYT CGL2008-00655/BOS Project supported this research financially. The first author was supported through MAEC-AECID grant from the Spanish Agency for International Cooperation and Development. We thank Bruce MacWhirter for improving the English. We are also grateful to the reviewers of the manuscript for their valuable comments. Capitulo 1 52 References Amo L, López P, Martín J (2007) Natural oak forest vs. ancient pine plantations: lizard microhabitat use may explain the effects of ancient reforestations on distribution and conservation of Iberian lizards. Biodivers Conserv 16: 34093422 Araújo MB, Pearson RG (2005) Equilibrium of species’ distributions with climate. Ecogr 28: 693-695 Austin M (2007) Species distribution models and ecological theory: A critical assessment and some possible new approaches. Ecol Model 200: 1-19 Barbero M, Bonin G, Loisel R, Quézel P (1990) Changes and disturbances of forest ecosystems caused by human activities in the western part of the Mediterranean basin. Plant Ecol 87: 151-173 Castro DP, Montserrat MG (1998) Phenological pattern of fifteen Mediterranean phanaerophytes from shape Quercus ilex communities of NE-Spain. Plant Ecol 139: 103-112 Ceballos L, Torre JRD (1979) Árboles y arbustos de la España peninsular. ETSIM, Madrid CHE (2009) Mapa de núcleos de población 1:50 000. Confederación Hidrográfica del Ebro, Zaragoza, Spain. http://www.chebro.es/. Accessed December 2009 Chevan A, Sutherland M (1991) Hierarchical partitioning. Am Stat 45: 90-96 Cierjacks A, Hensen I (2004) Variation of stand structure and regeneration of Mediterranean holm oak along a grazing intensity gradient. Plant Ecol 173: 215-223 CITA (2007) Mapa geológico de la comunidad autónoma de Aragón 1:50 000. Centro de Información Territorial de Aragón, Gobierno de Aragón, Zaragoza, Spain. http://sitar.aragon.es/. Accessed December 2009 CITA (2009) Modelo Digital del Terrino de la comunidad autónoma de Aragón. Centro de Información Territorial de Aragón, Gobierno de Aragón, Zaragoza, Spain. http://sitar.aragon.es/. Accessed December 2009 Cliff A, Ord J (1973) Spatial autocorrelation. Pion Press, London Congalton RG (1991) A review of assessing the accuracy of classifications of remotely sensed data. Remote Sens Environ 37: 35-46 Capitulo 1 53 Corcuera L, Camarero J, Gil-Pelegrin E (2004) Effects of a severe drought on growth and wood anatomical properties of Quercus faginea. IAWA J 25: 185204 Coudun C, Gégout JC, Piedallu C, Rameau JC (2006) Soil nutritional factors improve models of plant species distribution: an illustration with Acer campestre (L.) in France. J Biogeogr 33: 1750-1763 Dirnböck T, Dullinger S, Grabherr G (2003) A regional impact assessment of climate and land-use change on alpine vegetation. J Biogeogr 30: 401-417 DMA (2007) Atlas Climático de Aragón. Departamento de Medio Ambiente, Gobierno de Aragón, Zaragoza, Spain DMA (2010) Mapa de vías pecuarias de la comunidad autónoma de Aragón 1:50 000. Departamento de Medioambiente, Gobierno de Aragón, Zaragoza, Spain ESRI (2006) ArcGIS Help (9.2). Environmental Systems Research Institute, Inc., New York, USA Fischer HS (1990) Simulating the distribution of plant communities in an alpine landscape. Coenoses 5: 37-43 Gastón A, Soriano C, Miguel GV (2009) Lithologic data improve plant species distribution models based on coarse-grained ocurrence data. Sist Recur For 18: 42-49 Graham CH, Ron SR, Santos JC, Schneider CJ, Moritz C (2004) Integrating phylogenetics and environmental niche models to explore speciation mechanisms in dendrobatid frogs. Evol 58: 1781-1793 Guisan A, Zimmermann NE (2000) Predictive habitat distribution models in ecology. Ecol Model 135: 147-186 Hargreaves GH (1975) Moisture availability and crop production. Trans. Am. Soc. Agric. Eng 18: 980-984 Himrane H, Camarero JJ, Gil-Pelegrín E (2004) Morphological and ecophysiological variation of the hybrid oak Quercus subpyrenaica ( Q. faginea × Q. pubescens). Trees Struct Funct 18: 566-575 Hoeting JA, Madigan D, Raftery AE, Volinsky CT (1999) Bayesian model averaging: A tutorial. Stat Sci 14: 382-401 Capitulo 1 54 IGN (2002) Corine 2000: descripción de la nomenclatura del Corine Land Cover al nivel 5º. Actualización 2000. Instituto Geográfico Nacional, Centro Nacional de Información Geográfica, Ministerio de Fomento, Madrid, Spain IGN (2006) Plan Nacional de Ortofotografía Aérea 1: 30 000. Instituto Geográfico Nacional, Centro Nacional de Información Geográfica, Ministerio de Fomento, Madrid, Spain Lansac AR, Zaballos JP, Martin A (1994) Seasonal water potential changes and proline accumulation in Mediterranean shrubland species. Vegetatio 113: 141-154 Lasanta T (1989) Evolución reciente de la agricultura de montaña. Geoforma Edición, Logroño 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: 173-195 Lasanta T, Arnáez J, Errea MP, Ortigosa L, Ruiz-Flaño P (2009) Mountain pastures, environmental degradation, and landscape remediation: the example of a Mediterranean policy initiative. Appl Geogr 29: 308-319 Lasanta T, Vicente-Serrano S, Guadrat J (2005) Spatial-temporal variability of the plant landscape in the Mediterranean highlands due to the abandonment of traditional land uses: a study of the Spanish Central Pyrenees. Appl. Geogr 25: 47-65 Legendre P (1993) Spatial Autocorrelation: Trouble or New Paradigm? Ecology 74: 1659-1673 Loidi J, Herrera M (1990) The Quercus pubescens and Quercus faginea forests in the Basque Country (Spain): distribution and typology in relation to climatic factors. Plant Ecol 90: 81-92 Luoto M, Heikkinen RK, Pöyry J, Saarinen K (2006) Determinants of the biogeographical distribution of butterflies in boreal regions. J Biogeogr 33: 1764-1778 Madigan D, Raftery AE (1994) Model selection and accounting for model uncertainty in graphical models using Occam's window. J Am Stat Assoc 89: 1535-1546 Capitulo 1 55 Maltez-Mouro S, García L, Freitas H (2009) Influence of forest structure and environmental variables on recruit survival and performance of two Mediterranean tree species (Quercus faginea L. and Q. suber Lam.). Eur J For Res 128: 27-36 Mediavilla S, Escudero A (2004) Stomatal responses to drought of mature trees and seedlings of two co-occurring Mediterranean oaks. For Ecol Manag 187: 281-294 Millington J, Perry G, Romero-Calcerrada R (2007) Regression techniques for examining land use/cover change: a case study of a Mediterranean landscape. Ecosyst 10: 562-578 MMA (2007) Tercer inventario forestal nacional (IFN3) 1:50 000. Ministerio de Medio Ambiente y Medio Rural y Marino, Madrid, Spain Montserrat G (1990) Estudio de la colonización vegetal de los campos abandonados del valle de Aísa (Jaca, Huesca), Informe del proyecto LUCDEME: Erosión y colonización vegetal en campos abandonados, 77p, Jaca Myers N, Mittermeier RA, Mittermeier CG, da Fonseca GAB, Kent J (2000) Biodiversity hotspots for conservation priorities. Nat 403: 853-858 Nally RM (2000) Regression and model-building in conservation biology, biogeography and ecology: The distinction between – and reconciliation of – ‘predictive’ and ‘explanatory’ models. Biodivers Conserv 9: 655-671 Nally RM, Vries LD, Thomson JR (2008) Are replanted floodplain forests in southeastern Australia providing bird biodiversity benefits? Restor Ecol 18: 85-94 Perry DA (1998) Landscape pattern and forest pest. Northwest Environ. J. 4: 213228 Plieninger T (2006) Habitat loss, fragmentation, and alteration: quantifying the impact of land-use changes on a Spanish dehesa landscape by use of aerial photography and GIS. Landsc Ecol 21:91-105 Pulido FJ, Díaz M 2005. Regeneration of a Mediterranean oak: a whole-cycle approach. Ecosci 12: 92-102 Purves DW, Zavala MA, Ogle K, Prieto F, Benayas JMR (2007) Environmental heterogeneity, bird-mediated directed dispersal, and oak woodland dynamics in Mediterranean Spain. Ecol Monogr 77: 77-97 Capitulo 1 56 Raftery AE, Madigan D, Hoeting JA (1997) Bayesian model averaging for linear regression models. J Ame Stat Assoc 92: 179-191 Raxworthy CJ, Ingram CM, Rabibisoa N, Pearson RG (2007) Applications of ecological niche modeling for species delimitation: A review and empirical evaluation using day geckos (Phelsuma) from Madagascar. Syst Biol 56:907923 Raftery E, Hoeting J, Volinsky C, Painter I, Yeung K (20099) BMA: Bayesian model averaging. R package version 3.12. http://CRAN.R-project.org/package=BMA R Development Core Team (2009) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3900051-07-0, URL http://www.R-project.org Randin CF, Jaccard H, Vittoz P, Yoccoz NG, Guisan A (2009) Land use improves spatial predictions of mountain plant abundance but not presence-absence. J Veg Sci 20: 996–1008 Rey Benayas JM, Navarro J, Espigares T, Nicolau JM, Zavala MA (2005) Effects of artificial shading and weed mowing in reforestation of Mediterranean abandoned cropland with contrasting Quercus species. For Ecol Manag 212: 302-314 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 Segurado P, Araújo MB (2004) An evaluation of methods for modelling species distributions. J Biogeogr 31: 1555-1568 Teixido AL, Quintanilla LG, Carreño F, Gutiérrez D (2010) Impacts of changes in land use and fragmentation patterns on Atlantic coastal forests in northern Spain. J Environ Manag 91: 879-886 Thomson JR, Nally RM, Fleishman E, Horrocks G (2007) Predicting bird species distributions in reconstructed landscapes. Conserv Biol 21: 752-766 Thuiller W, Araújo MB, Lavorel S (2004) Do we need land-cover data to model species distributions in Europe? J Biogeogr 31: 353-361 Viallefont V, Raftery AE, Richardson S (2001) Variable selection and Bayesian model averaging in case-control studies. Stat Med 20: 3215-3230 Capitulo 2 63 as either the first, second, or third most dominant tree species based on a field visual estimate of the proportional coverage of each species in the patch. All patches, in which Q. faginea was among the three most dominant species, were used to create the distribution map of this species within the study area. Overall, the contribution of Q. faginea to patches varied from 20% (in the patches where Q. faginea was the third most dominant species) to 90% (in the patches where it was the most dominant species). The distribution map was imported into a GIS for further processing, including correcting the edges of patches and the removal for each patch of the areas where Q. faginea was absent. This process was done by visual inspection of ortho-rectified, 0.5-m-resolution aerial photographs (1:30,000) from the Spanish National Plan of Aerial Orthophotographs (IGN 2006). The accuracy of the final map (96%) was quantified by sampling 200 random points and calculating confusion matrices (Congalton 1991). The 1957 map was produced by visual interpretation of the aerial photographs of the United States Army from 1956 to 1957 provided by Spain’s Centro Cartográfico y Fotográfico del Ejército del Aire. The 1957 aerial photographs were taken on panchromatic film with a scale of 1:32,000 and were obtained as 24 x 24 contact prints. A total of 170 contact prints covering the study area were scanned into digital format at a resolution of 1,000 dpi, and then geo-referenced with the software Topol 9.5, using 12 ground control points per photograph, for a final resolution of approximately 0.5 m. However, it is worthwhile indicate that the polygons defined in the 2006 map were used, as a guide, to locate the Q. faginea forest patches in the 1957 aerial photographs by means of the overlapping tool in GIS environment. It should be noted that a minimum mapping unit of 0.1 ha per patch was defined in both maps (Fig. 1). Change analysis Based on published scientific literature (Molinillo et al. 1997; Montserrat 1990) and our interpretation of the aerial photographs, we assumed that only patches that were occupied by shrubland in 1957 could have converted to Q. faginea through natural vegetation succession by 2006; therefore, the patches of shrubland were included in the 1957 map. By 2006, however, the patches that Capitulo 2 64 were occupied by Q. faginea in 1957 might still be Q. faginea or had been converted to other land uses. Visual inspection of the 1957 and 2006 photographs revealed that some patches of Q. faginea had been converted to croplands, pine plantations, or shrublands, and this information was included in the 2006 map. The changes in Q. faginea forests were mapped by overlaying maps from 1957 and 2006. Changes map was summarized by calculating change rates (probabilities of change) and the area covered by each change class (Table 1). The probabilities of change were calculated using the following formula: P ij = A j2006 /A i1957 where P ij is the probability that a section changes from class i to class j, A j 2006 is the area of the class j in 2006, and A i 1957 is the area of the class i in 1957. Three types of change were defined (1) gain (i.e., i = {shrubland} and j = {Q. faginea forest}), (2) loss (i.e., i = {Q. faginea forest} and j = {pine plantation, shrubland, cropland}), (3) remnants (no change) (i.e., i = {Q. faginea forest} and j = {Q. faginea forest}). To avoid errors arose from misregistration between two dates, only changes classes covering an area more than 0.1 ha were considered. Fragmentation analysis Quantification and temporal comparison of the spatial configuration of Q. faginea forest patches between 1957 and 2006 was conducted based on a following set of standard landscape metrics: (1) Number of Q. faginea patches; (2) Mean patch size (ha); (3) Total area (ha) occupied by Q. faginea patches; (4) Mean patch distance (m) (the average of the nearest distances between the edges of Q. faginea patches); (5) Total patches edge length (km) of all Q. faginea patches; (6) Mean perimeter-to-area ratio (the mean of the ratio of perimeter (m) to area (m 2 ) of all Q. faginea patches). Most of these metrics have been applied in diverse forest fragmentation studies (Echeverria et al. 2006; Sitzia et al. 2010; Teixido et al. 2010), and have enabled the assessment of spatial attributes in fragmented landscapes. Capitulo 2 65 Number of patches was used as a proxy of patch subdivision, i.e., an increase in the number of Q. faginea patches means that some of them were broken up into separate patches (fragments). Mean patch size and total area, were used for the assessment of patch fragmentation and loss, a decline in these metrics involves an increase of patch fragmentation and loss. In the same way, mean patch distance was used as a measure of patch isolation, i.e., used to compare the degree of isolation among Q. faginea patches between 1957 and 2006. Finally, total patches edge length and mean perimeter-to-area ratio were used to assess the dynamic of irregularity of Q. faginea patches (i.e., patch-shape complexity). The following equation quantified the relative change (R) of each index: R = (A 2006 - A 1957 ) × 100/A 1957 Where A 2006 and A 1957 are the number of patches, mean patch size, total area, mean patch distance, total edge length, or mean perimeter-to-area ratio in 2006 and 1957, respectively. Statistical analysis Dependent and independent variables The response variables were derived by reclassifying and dividing the changes map in two binary maps. One map displayed the gains in Q. faginea forests (i.e., gain/no gain) and the other one displayed the losses in Q. faginea forests (i.e., loss/no loss). The predictor variables included topo-climatic and land use variables that were suspected of causing changes in the Q. faginea forests between 1957 and 2006. Elevation (m a.s.l.), slope (º), and insolation (WH/m 2 ) were derived from the Digital Elevation Model of Aragón (CITA 2009) at 20m of resolution using ArcGIS 9.2 (ESRI 2006). A map of number of frost days per year (Frost_days) was obtained from the Digital Climatic Atlas of Aragón (DMA 2007); the data in this map was averaged for the period 1971-2000. Topography can have a strong effect on the dynamics of vegetation (Carmel and Kadmon 1999), and elevation strongly Capitulo 2 66 influences temperature and rainfall in mountains (Barry 1992). Thus, elevation is often a proxy for climatic gradients (Gallego Fernández et al. 2004; Pueyo and Beguería 2007). Slope gradient influences hydrological and erosion processes in the soil (Florinsky et al. 2002) and insolation influences soil-vegetation, evapotranspiration and, therefore, soil water content, and it might have a significant effect on the spatial dynamic of Q. faginea. Number of frost days per year is postulated to have a direct influence on the establishment and distribution of plant species (Coudun et al. 2006). The extensive pine plantations that were created in the area within the last 50 years might have influenced the distribution of Q. faginea; therefore, distance to the nearest pine plantation (Distance_plant) was included in the analyses and calculated in a GIS using the straight line distance function. Distance to the nearest village (Distance_village) and distance to the nearest road (Distance_road) were measures of the intensity of human activity; activity is more likely to occur close to these structures. The road map in the study area was digitized on 2006 aerial photographs. The road network map of Aragón (CITA 2010) was used to identify the roads locations (all road types of this map were considered, i.e., primary, secondary, and unpaved roads). The distance to the nearest road was quantified using the straight line distance function in the GIS. Similarly, the distance to the nearest village was derived from a map of settlements in Aragon (CHE 2009). To assess the extent to which livestock and agricultural activities affected the spatiotemporal dynamics of Q. faginea between 1957 and 2006, cost distance to the nearest pasture (Cosdistance_pasture) and cost distance to the nearest cropland (Cosdistance_crop) were derived from a CORINE Land Cover map and included in the analyses. Maps of the cost distance to the nearest pasture and cropland were measures of transportation costs, i.e., calculate the least accumulative cost for moving from the source pixel (in this case pasture or cropland) to each of other pixels using slope as cost layer. The cost increase with an increasing of slope values up to 35º, beyond this steepness, areas were considered inaccessible to man and livestock (Kouba et al. 2010). All maps were subset to identical extents at a spatial resolution of 20-m. Moran’s I correlogram (Legendre and Legendre 1998) was used to assess the Capitulo 2 67 spatial autocorrelations (SAC) of the dependent variables. If present in the data, SAC violate the assumption about the independence of residuals and call into question the validity of hypothesis testing (Dormann et al. 2007). Moran’s I correlogram show a decrease from some level of SAC to a value of 0 (or below), indicating no SAC at some distance between pixels. In our study, the SAC declined monotonically above a lag of 10 pixels (~ 200 m) in the map of gain and 12 pixels (~ 240 m) in the map of loss; therefore, a length of 300 m was used as a minimum threshold in extracting pixels (Millington et al. 2007). The selected sample (i.e., extracted pixels) of the response variables values was intersected with the corresponding values of the 10 predictor variables layers and the resulting dataset was imported into R (R Development Core Team 2009) for statistical analyses. Statistical models To examine the effect of each predictor on each of the dependent variables, we first used generalized additive models (GAM; Hastie and Tibshirani 1990). We created a univariate GAM model for each potential predictor variable and each of the two binary dependent variables, i.e., gain/no gain and loss/no loss, and selected the best predictors from these models based on their statistical significance and explained deviance (D 2 ) (Rutherford et al. 2008). To determine whether the dependent variable exhibited a linear or a non-linear response to the predictor variable, the smoothed function was plotted for each univariate GAM model (Guisan and Zimmermann 2000). If the response of the dependent variable to the predictor is non-linear, the quadratic terms should be included in subsequent analyses. When there is curvature in the trend, the inclusion of the quadratic term increases the precision of the linear term estimation (Hair et al. 1998). In order to avoid the strong correlations between the linear and quadratic terms, the input variables were ‘‘centered’’ by subtracting the sample mean from all values before being squared (Schielzeth 2010). Collinearity was detected in the predictor variables using the Pearson correlation coefficients, with a threshold of 0.8 (Menard 2002; Rutherford et al. 2008). If Pearson correlation coefficient between two independent variables Capitulo 2 68 exceeded 0.8, one of the variables was excluded from the analyses. The final models were generated using Bayesian model averaging (BMA; Madigan and Raftery 1994) and adaptative regression by mixing with model screening (ARMS; Yuan and Ghosh 2008), which deal with uncertainty in the selection of models and add inference about the most important predictor variables. BMA uses Bayesian information criterion (BIC) to find good candidate models for inclusion in the final model (Hoeting et al. 1999). ARMS involves the following main steps (Morfin and Makowski 2009): (1) the sample is split into a training set and a test set; (2) each model is fitted by least square or maximum likelihood; (3) a set of models is selected based on Akaike’s information criterion (AIC) and Bayesian information criterion (BIC); and (4) the response values are predicted in the test set using the fitted models obtained from the training set. The models are weighted using likelihood ‘‘likeli’’ or Akaike’s information criterion ‘‘AIC’’. BMA and ARMS models were fitted using the predictor variables that had significant predictive power in the univariate GAMs and were not correlated with other predictor variables. The overall fit of the BMA and the ARMS was evaluated using the received operating characteristic (ROC) curve (Hanley and McNeil 1982). The area under the curve (AUC) was calculated using fivefold cross-validation (Millington et al. 2010). BMA, ARMS, and AUC were implemented using the MMIX package (Morfin and Makowski 2009) of R software (R Development Core Team 2009), functions ‘‘bmaBIC’’, ‘‘arms,’’ and ‘‘aucCV,’’ respectively. Results Gains, losses, and fragmentation of Q. faginea forests In the Central Pre-Pyrenees of Spain, the total area occupied by Q. faginea forests decreased by ~ 9% between 1957 and 2006. In 1957, Q. faginea forests covered 7% (9,149 ha) of the study area, but by 2006, they were reduced to 6% (8,336 ha) of the area. The changes matrix (Table 1) revealed that Q. faginea forests gained 626 ha in some areas through natural transitions from shrubland to Q. faginea forests and lost 1,438 ha in others. The transition to pine plantations and shrubland was the most important source of losses in Q. faginea forests: ~ 924 ha were converted to pine plantations and ~ 390 ha to shrubland. Moreover, the Capitulo 2 69 changes matrix revealed that ~ 125 ha of Q. faginea forests were converted to cropland. Table 1 Changes matrix calculated for Q. faginea forests in the Spanish Central PrePyrenees between 1957 and 2006 Change classes P a Area (ha) b Type of change 1957 2006 Q. faginea Q. faginea 0.774 7110 No change Q. faginea Pine plantations 0.100 924 Loss Q. faginea Shrubland 0.042 390 Loss Q. faginea Cropland 0.013 125 Loss Shrubland Q. faginea 0.068 626 Gain a : The probability of change. b : The area covered by each change class. In the study area, the number of patches of Q. faginea forests increased from 104 in 1957 to 118 in 2006 (~ 13.5%) (Table 2). In 1957, 30 patches (29%) > 100 ha contributed > 81% of the total Q. faginea forests. By 2006, the number of large patches (> 100 ha) was 24, while the total area occupied by these patches was ~ 70% (Fig. 2a, b). The number of small patches (< 10 ha) decreased from 34 in 1957 to 30 in 2006, but the number of medium-sized patches (10-100 ha) increased from 40 in 1975 to 64 in 2006 (Fig. 2a). Table 2 Landscape metrics used in an analysis of the fragmentation of Q. faginea forests in the Spanish Central Pre-Pyrenees between 1957 and 2006 Landscape metric 1957 2006 R (%) 1957-2006 Number of patches 104 118 13.5 Mean patch size (ha) 87.7 70.6 -19.5 Total patch area (ha) 9149 8336 -8.65 Mean patch distance (m) 1087 1179.4 8.50 Total patches edge length (km) 637 660 3.60 perimeter-to-area ratio 95.5 109.5 14.65 R: is the relative change in each metric (Positive values indicate an increase, and negative values indicate a decrease). Capitulo 2 70 Fig. 2 Proportion (%) of (a) Q. faginea patches and (b) area occupied by each patch category as a function of the size of Q. faginea patches, between 1957 and 2006, in the Spanish Central PrePyrenees. The number above each bar represents (a) the number of patches or (b) the area (ha) occupied by each patch category In addition to the general decline in total patch area and the increase of number of patches, our results (Table 2) showed substantial changes in the spatial patterns of Q. faginea forests. These changes implied the reduction of mean patch size (from 87.7 ha in 1957 to 70.6 ha in 2006), the increase of mean patch distance (augmented by approximately 8.5% between 1957 and 2006), as well as the increase of total edge length and mean perimeter-to-area ratio by 3.6 and 14.65%, respectively. Factors correlated with changes in Q. faginea forests 7417 1573 159 5819 2345 172 0 10 20 30 40 50 60 70 80 90 0-10 10-100 >100 Patch size (ha) % of occupied area 1957 2006 34 40 30 64 30 24 0 10 20 30 40 50 60 0-10 10-100 >100 Patch size (ha) % of patches 1957 2006 a b Capitulo 2 71 The univariate GAMs revealed that gains in Q. faginea forests were significantly (P < 0.05) correlated with elevation, number of frost days ‘‘Frost_days’’, insolation, distance to the nearest road ‘‘Distance_road’’, cost distance to the nearest cropland ‘‘Cosdistance_crop’’, distance to the nearest village ‘‘Distance_village’’, elevation and distance to the nearest road explained the most deviance (Table 3). Losses of Q. faginea forests were significantly (P £ 0.05) correlated with all land use variables and slope (Table 3). All variables that were significant (P £ 0.05) in the univariate GAMs were included in the BMA and ARMS, except insolation, which was excluded from the gains model because it was highly correlated with elevation (data not shown), but the latter explained more of the deviance in the univariate GAM model (Table 3). Table 3Univariate GAM models for each predictor variable against the two dependent variables (Gains and Losses). Variables statistically significant at P<0.05 are shown in bold Variables Gains Losses P D 2 P D 2 Elevation <0.05 0.26 0.25 0.06 Slope 0.65 0.07 <0.05 0.10 Insolation <0.05 0.15 0.11 0.02 Frost_days <0.05 0.13 0.57 0.11 Distance_road <0.05 0.22 <0.05 0.08 Cosdistance_crop <0.05 0.09 <0.05 0.09 Distance_village <0.05 0.11 <0.05 0.08 Cosdistance_pastur 0.25 0.01 <0.05 0.11 Distance_plant 0.58 0.03 <0.05 0.14 BMA and ARMS indicated that, in the gains model, elevation, ‘‘Frost_days’’, and ‘‘Distance_road’’ had high probabilities [Pr (B vs ¹ 0) ³ 0.90], and ‘‘Cosdistance_crop’’ and ‘‘Cosdistance_village,’’ low proailities [Pr (B vs ¹ 0) £ 0.41] of eing in the est-candidate model (Tale 4). In the losses model, ‘‘Cosdistance_crop’’, ‘‘Cosdistance_pastur’’, ‘‘Distance_plant’’, ‘‘Distance_road’’ and slope had high proailities [Pr (B vs ¹ 0) ³ 0.90] and ‘‘Distance_village’’ low proailities [Pr (B vs ¹ 0) £ 0.37] of eing in the est-candidate model (Tale 4). Capitulo 2 72 Millington et al. (2010) argued that only the variables that have a great probability of being in the best-candidate model are useful; therefore, here we considered only those predictor variables that had a [Pr (B vs ¹ 0) ³ 0.90] in both BMA and ARMS as factors that have had a significant influence on the changes in the Q. faginea forests. Table 4 Bayesian Averaging Model (BMA) and Adaptative Regression by Mixing with Model Screening (ARMS) used for identifying the most important factors that affecting the gains “gainsmodel” and losses “losses-model” in Q. faginea forests over the Spanish Central Pre-Pyrenees between 1957 and 2006 Variables BMA ARMS Mean  (±SD) Pr ( vs ¹ 0) Mean  Pr ( vs ¹ 0) Gains-model Elevation -0.046± 0.040 1.00 -0.046 1.00 Frost_days -0.549± 0.005 1.00 -0.559 1.00 Distance_road 0.360± 0.009 1.00 0.370 0.92 Cosdistance_crop 0.002± 0.006 0.38 0.001 0.40 Distance_village 0.400± 0.070 0.41 0.410 0.39 AUC.CV 0.889 0.891 Losses-model Slope -0.002± 0.283 0.94 -0.001 0.98 Distance_road -0.003± 0.003 0.93 -0.004 0.97 Cosdistance_crop -0.190± 0.001 0.99 -0.302 0.97 Distance_village 0.009± 0.002 0.33 0.009 0.37 Cosdistance_pastur -0.005± 0.050 1.00 -0.005 1.00 Distance_plant -0.150± 0.080 1.00 -0.130 1.00 AUC.CV 0.872 0.878 Variables have a Pr ( vs ¹ 0) ³ 0.9 (i.e. drivers of change) are shown in bold Greater expansion of Q. faginea forests in areas distant to roads, given the significant positive relationship between ‘‘Distance_road’’ and Q. faginea forests gains in both BAM and ARMS (Table 4). Both BAM and ARMS indicated that the expansion of Q. faginea forests (i.e., the probability of gains) increased as elevation and number of days with frost decreased. Patches of Q. faginea forests close to pine Capitulo 2 79 Aspinall R (2004) Modelling land use change with generalized linear models--a multi-model analysis of change between 1860 and 2000 in Gallatin Valley, Montana. J Environ Manag 72: 91-103. doi: 10.1016/j.jenvman.2004.02.009 Barbero M, Bonin G, Loisel R, Quézel P, (1990) Changes and disturbances of forest ecosystems caused by human activities in the western part of the mediterranean basin. Plant Ecol 87: 151-173 Blanco E, Casado MA, Costa M, Escribano R, García M, Genova M, Gomez A, Gomez F, Moreno JC, Morla C, Regato P, Sáinz H (1997) Los busques ibericos: una interpretación geobotánica. Ed. Planeta, Madrid Barrio G, Alvera B, Puigdefabregas J, Diez C (1997) Response of high mountain landscape to topographic variables: Central pyrenees. Landsc Ecol 12: 95115. doi: 10.1007/BF02698210. Barry RG (1992) MountainWeather and Climate, second ed, Meuthe, London Callaway RM, Davis FW (1993) Vegetation Dynamics, Fire, and the Physical Environment in Coastal Central California. Ecol 74: 1567-1578 Carmel Y, Kadmon R (1999) Effects of grazing and topography on long-term vegetation changes in a Mediterranean ecosystem in Israel. Plant Ecol 145: 243-254 CHE (2009) Mapa de núcleos de población 1:50 000. Confederación Hidrográfica del Ebro, Zaragoza, Spain. http://www.chebro.es/. Accessed December 2009 CITA (2009) Modelo Digital del Terrino de la comunidad autónoma de Aragón. Centro de Información Territorial de Aragón, Gobierno de Aragón, Zaragoza, Spain. http://sitar.aragon.es/. Accessed December 2009 CITA (2010) Mapa de redes de carreteras de Aragón 1:300 000. Centro de Información Territorial de Aragón, Gobierno de Aragón, Zaragoza, Spain. http://sitar.aragon.es/. Accessed January 2010 Congalton RG (1991) A review of assessing the accuracy of classifications of remotely sensed data. Remote Sens Environ 37: 35-46 Coudun C, Gégout JC, Piedallu C, Rameau JC (2006) Soil nutritional factors improve models of plant species distribution: an illustration with Acer campestre (L.) in France. J Biogeogr 33: 1750-1763 Capitulo 2 80 Díaz M, González E, Muñoz-Pulido R, Naveso MA (1996) Habitat selection patterns of common cranes Grus grus wintering in holm oak Quercus ilex dehesas of central Spain: effects of human management. Biol Conserv 75: 119-123 DMA (2007) Atlas Climático de Aragón. Departamento de Medio Ambiente, Gobierno de Aragón, Zaragoza, Spain Dormann, CF, McPherson JM et al. (2007) "Methods for account for spatial autocorrelation in the analysis of species distributional data: a review." Ecogr 30: 609-628 Echeverria C, Coomes D, Salas J, Rey-Benayas JM, Lara A, Newton A (2006) Rapid deforestation and fragmentation of Chilean Temperate Forests. Biol Conserv 130: 481-494. doi: 10.1016/j.biocon.2006.01.017 ESRI (2006) ArcGIS Help (9.2). Environmental Systems Research Institute, Inc., New York, USA Ewers RM, Kliskey AD, Walker S, Rutledge D, Harding JS, Didham RK (2006) Past and future trajectories of forest loss in New Zealand. Biol Conserv 133: 312325. doi: 10.1016/j.biocon.2006.06.018 Florinsky IV, Eilers RG, Manning GR, Fuller LG (2002) Prediction of soil properties by digital terrain modelling. Environ Model Softw 17: 295-311. doi: 10.1016/S1364-8152(01)00067-6 Freitas SR, Hawbaker TJ, Metzger JP (2010) Effects of roads, topography, and land use on forest cover dynamics in the Brazilian Atlantic Forest. For Ecol Manage 259: 410-417. doi: 10.1016/j.foreco.2009.10.036 Gallego Fernández J.B., Rosario García Mora M. and GarcÃa Novo F. 2004. Vegetation dynamics of Mediterranean shrublands in former cultural landscape at Grazalema Mountains, South Spain. Plant Ecol 172: 83-94 García D, Quevedo M, Ramón-Obeso J, Abajo A (2005) Fragmentation patterns and protection of montane forest in the Cantabrian range (NW Spain). For Ecol Manage 208: 29-43 Gracia M, Retana J, Roig P (2002) Mid-term successional patterns after fire of mixed pine-oak forests in NE Spain. Acta Oecol 23: 405-411. doi: 10.1016/S1146-609X(02)01169-4 Capitulo 2 81 Guirado M, Pino J, Rodá F, Basnou C (2008) Quercus and Pinus cover are determined by landscape structure and dynamics in peri-urban Mediterranean forest patches. Plant Ecol 194: 109-119 Guisan A, Zimmermann NE (2000) Predictive habitat distribution models in ecology. Ecol Model 135: 147-186. doi: 10.1016/S0304-3800(00)00354-9 Hair JF, Anderson RE, Tatham RL, Black WC (1998) Multivariate Data Analysis. Prentice-Hall, Upper Saddle River, NJ Hanley JA, McNeil BJ (1982) The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiol 143: 29-36 Hastie TJ, Tibshirani RJ (1990) Generalized Additive Models. London. Chapman, Hall. Hoeting JA, Madigan D, Raftery AE, Volinsky CT (1999) Bayesian Model Averaging: A Tutorial. Stat Sci 14: 382-401 IGN (2006) Plan Nacional de Ortofotografía Aérea 1: 30 000. Instituto Geográfico Nacional, Centro Nacional de Información Geográfica, Ministerio de Fomento, Madrid, Spain Jump AS, Peñuelas J (2006) Genetic effects of chronic habitat fragmentation in a wind-pollinated tree. Proceedings of the National Academy of Sciences of the United States of America 103: 8096-100. doi: 10.1073/pnas.0510127103 Kobayashi Y, Koike F (2010) Separating the effects of land-use history and topography on the distribution of woody plant populations in a traditional rural landscape in Japan. landsc Urban Plan 95: 34-45. doi: 10.1016/j.landurbplan.2009.11.003 Kouba Y, Alados C L, Bueno G C (2010) Effects of abiotic and anthropogenic factors on the spatial distribution of Quercus faginea in the Spain Central Pyrenees. Plant Ecol. doi: 10.1007/s11258-010-9880-0 Lasanta-Martínez T (1989) Evolución reciente de la agricultura de montaña. Geoforma Edición, Logroño Lasanta-Martínez T, Vicente-Serrano SM, Cuadrat-Prats JM (2005) Mountain Mediterranean landscape evolution caused by the abandonment of traditional primary activities: a study of the Spanish Central Pyrenees. Appl Geogr 25: 47-65. doi: 10.1016/j.apgeog.2004.11.001 Legendre P, Legendre L (1998) Numerical ecology. Elsevier Capitulo 2 82 Madigan D, Raftery AE (1994) Model Selection and Accounting for Model Uncertainty in Graphical Models Using Occam's Window. J Am Stat Assoc 89: 1535–1546 Maltez-Mouro S, García L, Freitas H (2009) Influence of forest structure and environmental variables on recruit survival and performance of two Mediterranean tree species (Quercus faginea L. and Q. suber Lam.). Eur J For Res 128: 27-36 Menard S (2002) Applied Logistic Regression Analysis. Sag Publications, California Millington J, Perry G, Romero-Calcerrada Rl (2007) Regression Techniques for Examining Land Use/Cover Change: A Case Study of a Mediterranean Landscape. Ecosyst 10: 562-578 Millington JDA, Walters MB, Matonis MS, Liu J (2010) Effects of local and regional landscape characteristics on wildlife distribution across managed forests. For Ecol Manage 259: 1102-1110. doi: 10.1016/j.foreco.2009.12.020 MMA (2007) Tercer inventario forestal nacional (IFN3) 1:50 000. Ministerio de Medio Ambiente y Medio Rural y Marino, Madrid, Spain Molinillo M, Lasanta T, García-Ruiz JM (1997) RESEARCH: Managing Mountainous Degraded Landscapes After Farmland Abandonment in the Central Spanish Pyrenees. Environ Manage 21: 587-598 Montserrat G (1990) Estudio de la colonización vegetal de los campos abandonados del valle de Aísa, Jaca, Huesca Morfin M, Makowski D (2009) MMIX: Model selection uncertainty and model mixing. R package version 1.1.http://CRAN.R-project.org/package=MMIX Palik BJ, Murphy PG (1990) Disturbance versus edge effects in sugar-maple/beech forest fragments. For Ecol Manage 32: 187-202. doi: 10.1016/03781127(90)90170-G Plieninger T (2006) Habitat loss, fragmentation, and alteration: quantifying the impact of land-use changes on a Spanish dehesa landscape by use of aerial photography and GIS. Landsc Ecol 21:91-105 Plieninger T, Schaar M (2008) Modification of Land Cover in a Traditional Agroforestry System in Spain: Processes of Tree Expansion and Regression. Ecol Soc 13: 7 Capitulo 2 83 Poyatos R. Latron J, Llorens P (2003) Land Use and Land Cover Change After Agricultural Abandonment. Mountain Research and Development 23: 362368. doi: 10.1659/0276-4741(2003)023. Pueyo Y, Alados CL (2007) Effects of fragmentation, abiotic factors and land use on vegetation recovery in a semi-arid Mediterranean area. Basic Appl Ecol 8: 158-170. doi: 10.1016/j.baae.2006.03.009 Pueyo Y, Beguería S (2007) Modelling the rate of secondary succession after farmland abandonment in a Mediterranean mountain area. Landsc Urban Plan 83: 245-254. doi: 10.1016/j.landurbplan.2007.04.008 Pulido FJ, Díaz M (2005) Regeneration of a Mediterranean oak: a whole-cycle approach. Ecosci 12: 92-102 Puyravaud JP (2003) Standardizing the calculation of the annual rate of deforestation. For Ecol Manage 177: 593-596 Rademacher C, Neuert C, Grundmann V, Wissel C, Grimm V (2004) Reconstructing spatiotemporal dynamics of Central European natural beech forests: the rulebased forest model BEFORE. For Ecol Manage 194: 349-368. doi: 10.1016/j.foreco.2004.02.022 Rey Benayas JM, Navarro J, Espigares T, Nicolau JM, Zavala MA (2005) Effects of artificial shading and weed mowing in reforestation of Mediterranean abandoned cropland with contrasting Quercus species. For Ecol Manage 212: 302-314. doi: 10.1016/j.foreco.2005.03.032 Rhemtulla JM, Mladenoff DJ, Clayton MK (2009) Legacies of historical land use on regional forest composition and structure in Wisconsin, USA (mid-1800s1930s-2000s). Ecol Appl 19: 1061-1078 Rivas-Martínez S (1987) Memoria del mapa de Series de Vegetación de España. Escala 1:400.000. I.C.O.N.A. Ministerio de Agricultura. Pesca y Alimentación, Madrid, p. 268 Rodà F, Retana J, Gracia CA, Bellot J (1999) Ecology of Mediterranean evergreen oak forests. Springer-Verlag, Berlin Rutherford GN, Bebi P, Edwards PJ, Zimmermann NE (2008) Assessing land-use statistics to model land cover change in a mountainous landscape in the European Alps. Ecol Model 212: 460-471. doi: 10.1016/j.ecolmodel.2007.10.050 Capitulo 2 84 R Development Core Team (2009) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3900051-07-0, URL http://www.R-project.org Sancho MPJ, Fernández PMD, Albertos SM, Sánchez LG (1998) Regiones de procedencia de Quercus pyrenaica Willd. Quercus faginea Lam. Quercus canariensis Willd. OAPN, Madrid Saunders SC, Mislivets MR, Chen J, Cleland DT (2002) Effects of roads on landscape structure within nested ecological units of the Northern Great Lakes Region, USA. Biol Conserv 103: 209-225. doi: 10.1016/S0006-3207(01)00130-6 Sawchik J, Dufrêne M, Lebrun P, Schtickzelle N, Baguette M (2002) Metapopulation dynamics of the bog fritillary butterfly: modelling the effect of habitat fragmentation. Acta Oecol 23: 287-296. doi: 10.1016/S1146-609X(02)011578 Schielzeth H (2010) Simple means to improve the interpretability of regression coefficients. Methods Ecol Evol. doi: 10.1111/j.2041-210X.2010.00012.x. Sitzia T, Semenzato P, Trentanovi G (2010) Natural reforestation is changing spatial patterns of rural mountain and hill landscapes: A global overview. For Ecol Manage 259:1354-1362. doi: 10.1016/j.foreco.2010.01.048 Staus N, Strittholt J, Dellasala D, Robinson R (2002) Rate and patterns of forest disturbance in the Klamath-Siskiyou ecoregion, USA between 1972 and 1992. Landsc Ecol 17: 455-470 Suc JP (1984) Origin and evolution of the Mediterranean vegetation and climate in Europe. Nature 307: 429-432 Teixido AL, Quintanilla LG, Carreño F, Gutiérrez D (2010) Impacts of changes in land use and fragmentation patterns on Atlantic coastal forests in northern Spain. J Environ Manage 91: 879-886. doi: 10.1016/j.jenvman.2009.11.004 Vicente-Serrano SM, Lasanta T, Gracia C (2010) Aridification determines changes in forest growth in Pinus halepensis forests under semiarid Mediterranean climate conditions. Agri For Meteorol 150: 614-628. Villar-Salvador P, Castro-Díez P, Pérez-Rontomé C, Montserrat-Martí G (1997) Stem xylem features in three Quercus (Fagaceae) species along a climatic gradient in NE Spain. Trees Struct Funct 12: 90-96 Capitulo 2 85 Wahren C, Papst W, Williams R (1994) Long-term Vegetation Change in Relation to Cattle Grazing in Subalpine Grassland and Heathland on the Bogong High Plains: An Analysis of Vegetation Records from 1945 to 1994. Aust J Bot 42: 607-639 Young A., Boyle T, Brown T (1996) The population genetic consequences of habitat fragmentation for plants. Trends Ecol Evol 11: 413-418 Yuan Z, Ghosh D (2008) Combining Multiple Biomarker Models in Logistic Regression. Biom 64, 431-439 Capitulo 3 86 Forest Ecology and Management, Volume 274, Page 143-150, June 2012 CAPITULO 3 Roles of land-use and climate change on the establishment and regeneration dynamics of Mediterranean semi-deciduous oak forests Yacine Kouba a, *, J. Julio Camarero b and Concepción L. Alados a a Pyrenean Institute of Ecology (CSIC). Avda. Montañana 1005, PO Box 202, ES-50080, Zaragoza, Spain; b ARAID, Pyrenean Institute of Ecology (CSIC). Avda. Montañana 1005, PO Box 202, ES-50080, Zaragoza, Spain; *Corresponding author at: Pyrenean Institute of Ecology (CSIC). Avda. Montañana 1005, P. O. Box 202, 50080 Zaragoza, Spain; Email address: [email protected]; Phone: +34 976 716034; fax: +34 976 716019. Abstract Long-term changes in climate and land use have significant effects on the forest dynamics in formerly managed landscapes. To quantify the relative importance of climatic and land use factors on tree establishment at regional scales, retrospective analyses are required. In this paper, we provide an historical reconstruction of the establishment of Mediterranean oak (Quercus faginea) forests in the 20th century within the context of substantial changes in climate and changes in land use in the Spanish Pre-Pyrenees. Since the late 1930s, Q. faginea became established episodically, and the highest peak occurred between 1965 and 1975. Tree establishment was negatively correlated with mean summer maximum temperature, population size of nearby villages, and the amount of livestock, but was positively correlated with annual, winter, and winter-spring precipitation. This study revealed that assessments of the effects of land use and climate changes on historical forest recruitment are vital in understanding the structure of contemporary forests. Capitulo 3 87 Key words: climate change, land use changes, forest dynamics, Pre-Pyrenees, Quercus faginea, tree recruitment. Introduction Changes in land use and climate can have significant influences on tree establishment and on forest structure and dynamics in formerly managed landscapes (Barbero et al., 1990; Améztegui et al., 2010; Chauchard et al. 2010; Gimmi et al., 2010). In the Mediterranean region, the availability of water is the main factor that limits tree growth (Ogaya et al., 2003; Linares et al., 2010) and forest regeneration (Pulido and Díaz, 2005). In addition, inter-annual variability in precipitation significantly affects annual tree recruitment (Urbieta et al., 2008). However, differences in annual rainfall might not be the only climatic factor that influences forest regeneration in inland areas that have a continental Mediterranean climate because in these areas temperature and the amount of precipitation act together to dictate water availability, and low temperatures in winter cause cold stress (Larcher, 2000; Vicente-Serrano et al., 2010). Furthermore, the frequency and intensity of extreme weather events such as severe drought are expected to increase in those inland areas because of global climatic change (IPCC, 2007). In the Mediterranean region, particularly in the southern Pre-Pyrenees, where dry conditions prevail, semi-deciduous oak forests (Quercus faginea) occur in areas that are incur variable drought stress and, therefore, they might be especially sensitive to climate warming and increasing aridity (Alla et al., 2011). Understanding the nature of changes in land use is important for understanding the structure and stand dynamics of contemporary forests (Améztegui et al., 2010; Gimmi et al., 2010). In general, in the mountains of Europe, anthropogenic factors have had a greater influence on the current composition and structure of many forests than have changes in climate (Olano et al., 2008; Tappeiner et al., 2008, Gimmi et al., 2010). Furthermore, the ongoing changes in the policies of the EU for agricultural and rural developmental might lead to even more pronounced changes in the mountain forests (Tappeiner et al., 2008). Capitulo 3 88 In the Central Pre-Pyrenees, changes in land use (i.e., farmland abandonment and grazing cessation) have led to the expansion of forests into formerly cultivated or grazed areas (Lasanta et al. 2006; Améztegui et al., 2010). In particular, Q. faginea has colonized some of the abandoned lands in the Central Pre-Pyrenees through natural transitions from abandoned lands to forests (Kouba and Alados, 2012). Acorns dispersed locally by gravity or through shortor long-distance dispersal mediated by rodents (Pulido and Díaz, 2005) and birds (Gómez, 2003), respectively, are the main means by which Q. faginea seeds reach abandoned fields (Maltez-Mouro et al., 2008). Encroachment by Q. faginea into abandoned lands has led to the formation of two types of forests: (i) Q. faginea stands that were harvested intensively for timber and firewood for centuries and that were used as pastures (Sancho et al., 1998), and (ii) new Q. faginea stands that became established in the abandoned terraces, mainly during the second half of the twentieth century (Kouba and Alados, 2012). Those forests are valued highly because they provide invaluable habitat for maintaining the biodiversity of Mediterranean plant and animal species (Rey Benayas et al., 2005, Kouba et al., 2010, Kouba and Alados, 2012). To understand how changes in land use and climate influence the dynamics of forest regeneration and how they affect tree establishment , both of these factors should be assessed simultaneously (Abrams and Copenheaver, 1999; Camarero and Gutiérrez, 2007; Chauchard et al., 2007, 2010; Copenheaver and Abrams, 2003); however, identifying the importance of changes in land use and recent climate trends on the regeneration dynamics in forests can be attained only by comparing sites that have contrasting histories and climates. The main purpose of this study was to assess our understanding of the potential effects of changes in land use and climate on the regeneration and growth dynamics of Q. faginea forests. Specifically, we aimed to (i) determine whether tree recruitment in Q. faginea forests in the last century was affected by climatic factors (e.g., drought) or changes in land use (e.g., land abandonment), (ii) assess the importance of specific types of forest habitats (i.e., coppice stands and abandoned terraces) on tree growth and performance in Q. faginea, and provide a basis for the development of land-management strategies that can mitigate the effects of global Capitulo 3 95 municipalities. Several municipalities merged in a large municipality, reducing the total number of municipalities in the study area. In the analysis of the data from the second period, we used the number of livestock in the two municipalities that encompassed the four villages near the ten sampling sites. Those data were obtained from the Provincial Service of Agriculture of Huesca Province (2011). Inflections and trends in the landuse variables (number of inhabitants and number of livestock) were assessed in the same way as were climate variables. The effects of changes in human and livestock densities on Q. faginea recruitment were evaluated by comparing the number of inhabitants, livestock numbers, and both the observed number of trees established and the residuals of the fitted model for each of the 5yr age classes by calculating Spearman correlation coefficients. Results Tree-size and age sricures exploraions The distribution of the diameters of all of the Q. faginea trees fit best a negative exponential distribution, and the most abundant class of trees had DBH between 5 and 10 cm (Fig. 2A). The most abundant class of trees was 46 m tall (Fig. 2B). The age distribution of the Q. faginea trees (Fig. 3A) indicated episodic recruitment, with highest recruitment in the late 1960s and early 1970s (Fig. 3B). Between 1935 and the early 1970s, Q. faginea recruitment was greater than the recruitment predicted by a power function, and maximum differences (number of positive residuals) occurred in the late 1960s and early 1970s (Fig. 3A). Furthermore, there were three periods of either reduced recruitment or high mortality (periods in which the predicted tree frequency was much higher than the observed frequency of trees and the residuals were negative): the late 1970s, the late 1980s, and the early 1990s (Fig. 3A). In abandoned terraces, Q. faginea recruitment did not occur until the 1940s. More than 65% of the individuals sampled on abandoned terraces were established between 1965 and Capitulo 3 96 the early 1990s, and most of them recruited in the late 1960s. In coppice stands, the first recruitment peak occurred in the late 1930s, and others occurred in the late 1940s and early 1950s, in the early 1970s and early1980s (Fig. 3B). Fig. 2 Diameter at breast height (DBH) (A) and height (B) of Q. faginea trees at ten sampling sites in the Spanish PrePyrenees. The Q. faginea trees in coppice stands (mean ± SE = 54 ± 4 yr) were significantly (p < 0.05) older than the trees in abandoned terraces (mean ± SE = 43 ± 5 yr) (Table 2); however, the trees on abandoned terraces had mean annual radial and heightgrowth rates that were significantly (p < 0.05) higher than those of the Q. faginea trees in coppice stands (Table 2). Although mean annual radial and heightgrowth rates are agedependent, in this study, the difference between the two habitats in the mean age of the trees was not large (overall mean ± SE = 48 ± 5 yr). Thus, a comparison of the rates was a valid means of detecting differences in the vigor and performance of the trees in the two habitats. In addition, the mean DBH of Q. faginea trees was significantly (p < 0.05) greater on abandoned terraces (13.0 cm) than in coppice stands (10.6 cm) (Table 2). Climae rends In the Central PrePyrenees, mean annual temperatures and mean summer maximum temperatures exhibited moderate interannual variability between 1910 and 1990 (coefficients of variation of 14.3% and 7.2%, respectively). Between 1915 and 1990, mean annual, winter, summer, and winterspring (B) Height (m) 0 2 4 6 8 10 12 (A) DBH (cm) 0 5 10 15 20 25 30 35 40 Number of trees 0 50 100 150 200 250 Capitulo 3 97 precipitation exhibited high variability in comparison to temperature variables (coefficients of variation of 40.0%, 54.5%, 32.1%, and 40.0%, respectively). Fig. 3 Observed and estimated numbers of Q. faginea trees at ten sampling sites in the Spanish PrePyrenees, as a function of their year of establishment (A), and the residuals that correspond to the difference between the observed and predicted number of trees. Comparative histograms between abandoned terraces and coppice stands displaying the number of trees against year of establishment (B). (A) Number of trees -40 -20 0 20 40 60 Observed number of trees estimated number of trees Residuals (B) Year of establishment 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 Number of trees 0 5 10 15 20 25 Coppice stands Abandoned terraces Capitulo 3 98 Table 2 Characteristics and related statistics of linear mixedeffects models that compare the Q. faginea tree variables of coppice stands (C) and abandoned terraces (T) with site as a random factor. Significant (p <0.05) differences between forest and terrace are indicated in bold. Values are mean ± standard error. Variable Terrace (T) Coppice stands (C) Comparaison (T C) F p-value DBH (cm) 13.0 ± 1.0 10.6 ± 0.5 T > C 4.2 0.042 Height (m) 5.4 ± 0.4 4.7 ± 0.2 T - C 3.3 0.071 Age (years) 43 ± 5 54 ± 4 T < C 11 0.001 Radialgrowth rate (mm yr 1 ) 1.5 ± 0.2 1.1 ± 0.1 T > C 11 0.001 Heightgrowth rate (cm yr 1 ) 13.7 ± 1.6 11.9 ± 0.9 T > C 5.4 0.020 In the last century, there have been five significant inflections in mean summer maximum temperature trends (Fig. 4A), with low values in 1925 (turn point test, p < 0.05), 1939 (p < 0.05), and 1972 (p < 0.05), and high values in 1943 (p < 0.01) and 1975 (p < 0.05). Mean summer maximum temperature anomalies decreased significantly (MannKendall test, t = 0.37, p < 0.05) between 1910 and 1925 (Fig. 4A), increased significantly between 1939 and 1943 (t = 0.70, p < 0.05), and, thereafter, decreased until 1972. Since 1975, mean summer maximum temperatures have increased significantly (t = 0.52, p < 0.05). Between 1915 and 1990, seven inflections were detected in the distributions of either winterspring or annual precipitation (Fig. 4B). Mean annual precipitation anomalies increased significantly (p < 0.05) between 1938 and 1943 (t = 0.61) and between 1943 and 1960 (t = 0.52). Between 1960 and 1972, mean annual rainfall was high. From 1973 until 1990, annual rainfall was markedly lower than it was at any other time in the 20th century. Changes in land use The human population was highest in the early 20th century (Fig. 5), declined sharply between 192030, and continued to decline until 1990 (t =  0.96, p < 0.01). The livestock numbers (Fig. 5) increased significantly (t = 0.98, p Capitulo 3 99 < 0.05) between 1890 and 1910, remained high until 1930, and decreased significantly between 1930 and 1970 (t = 0.66, p < 0.01). Between 1970 and 1990, livestock densities have increased substantially and have oscillated in the last 20 yr. Fig. 4 Mean summer maximum temperature fluctuations and anomalies (with respect to the average) from 1910 to 1990 (A), and winter, winterspring, and annual precipitation fluctuations and anomalies from 1915 to 1990 (B) within the study area. The marked points indicate significant (p £ 0.05) inflections in temperature or precipitation, which were identified using the “turnpoints” function. -6 -3 0 3 6 1910 1915 1920 1925 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 Temperature (ºC) -1000 -500 0 500 1000 1500 1910 1915 1920 1925 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 Precipitation (mm) Winter Winter-Spring Annual (A) (B) Capitulo 3 100 Fig. 5 Changes in the size of the human population and the amount of livestock (cattle, sheep, and goats) in the study area, between 1890 and 1990. Table 3 Spearman correlation coefficients (r s ) and related probability levels (p-value) calculated between weather and landuse variables vs. established trees. Calculations were performed using the observed number of established trees and the residuals of fitted power functions. Variable Observed p-value Residuals p-value Weaher variables Mean summer maximum Temperature - - 0.54 0.025 Annual precipitation - - 0.61 0.004 Winter precipitation 0.5 0.048 0.63 0.009 Winterspring precipitation - - 0.62 0.010 Land use variables Local population size 0.72 0.009 - - Number of livestock 0.82 0.000 0.75 0.001 Facors influencing Q. faginea recruimen All of the climate and land use variables that were significantly correlated with either the recruitment residuals of the fitted power function or the numbers of trees established were considered to have affected the recruitment history of Q. faginea forests. Correlation analyses indicated that the residuals of the power function fitted to the observed number of trees established were significantly negatively correlated with mean summer maximum temperatures and the number of livestock, and significantly positively correlated with annual, 0 1000 2000 3000 4000 5000 6000 7000 0 500 1000 1500 2000 2500 1890 1895 1900 1905 1910 1915 1920 1925 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 Number of inhabitants Years Number of inhabitants Number of livestock Number of livestock Capitulo 3 101 winter, and winterspring precipitation (Table 3). The number of trees established and winter precipitation were significantly positively correlated. In addition, the number of trees established was significantly negatively correlated with the numbers of inhabitants and the amount of livestock within the study area (Table 3). Discussion Dynamics of Q. faginea foress The negative exponential distribution of the diameters of the Q. faginea trees and the episodic recruitment revealed by this study reflect unevenaged forests (Smith et al., 1997). The comparison of the ageclass distributions of trees in coppice stands and those in young stands established on abandoned terraces can provide insights into the dynamics of forest development on abandoned lands. The colonization of the abandoned terraces by Q. faginea began in the 1940s; however, more than 65% of the Q. faginea trees present on those abandoned terraces in 2010 became established after 1965. Before Q. faginea began colonizing the abandoned terraces, shrubs (e.g., B. sempervirens, G. scorpius) were occupied them, which indicates that they were abandoned long before 1965 (Montserrat, 1990; Capitanio and Carcaillet, 2008; Kouba and Alados, 2012). The mean annual radial and heightgrowth rates of the Q. faginea trees on the abandoned terraces were higher and the stems were thicker than were those of the trees in coppice stands; probably, because the abandoned terraces and farmlands in the study area are on relatively flat lands and some are in valley bottoms where the soils have the highest amounts of nutrients and water, which might have enhanced tree growth (Lasanta et al., 2000). After abandonment, herbaceous plants and shrubs colonized the terraces before becoming forested by Q. faginea. The process of secondary succession involves significant changes in habitat and microclimate at local scales, including a reduction in runoff and an improvement in soil infiltration, which enhances soil conservation (Molinillo, 1997, Lasata et al., 2000), an increase in litter accumulation, which generates a large amount of organicmatter because of leaf Capitulo 3 102 decomposition (MaltezMouro et al., 2005), and the accumulation of minerals in the flatlands, which increases soil nutrient contents (MaltezMouro et al., 2005). Those changes helped to improve soil fertility (Lasanta et al., 2000). In other studies, Q. faginea trees had the highest growth rates and stands had the highest densities on the shallowest slopes, which had the highest soil fertility (Maltez Mouro et al., 2005). Effecs of climae on Q. faginea recruimen The recent reductions in annual precipitation and increases in summer temperatures in the last 20 yr have had a significant effect on Q. faginea recruitment. The positive correlations between Q. faginea recruitment and annual, winter, and winterspring precipitation reflect how interannual variability in precipitation has affected Q. faginea recruitment. Mean maximum summer temperature and Q. faginea recruitment were negatively correlated, which suggests that most of the recruitment occurred in years that had cool summers, and oak seedlings experienced high mortality in years that had hot, dry summers (see also Rey Benayas et al., 2005). The establishment of Q. faginea was very low in the late 1970s, late 1980s, and early 1990s and high in the late 1960s and early 1970s. Periods of low recruitment coincided with hot summers and low precipitation in winter and spring, which caused the most severe droughts in the region during the 20th century (VicenteSerrano, 2006) and might have caused high mortality among Q. faginea recruits. In the Mediterranean Basin, the weather in summer is one of the main factors that influence recruitment (i.e., seed germination, seedling emergence and survival) in oak species (Pulido and Díaz, 2005; Urbieta et al., 2008). Harsh conditions such as hot and dry summers are major causes of seedling mortality in Q. faginea and other Mediterranean oak species (Valladares et al., 2000; EstesoMartínez et al., 2006; MaltezMouro et al., 2008). Sufficient precipitation in winter and spring, and cool summer temperatures in the 1960s and early 1970s, especially between 1970 and 1972, might have produced the recruitment pulses that occurred between 1965 and 1975. The Capitulo 3 103 amount of moisture available in the soil has a strong influence on the survival of Q. faginea seedlings, which usually germinate in early spring (EstesoMartínez et al., 2006; MaltezMouro et al., 2007). In our study, high precipitation in winter and spring increased soil moisture, which can increase seedling survival if the subsequent summer is not exceptionally hot and dry. Other studies revealed also that low water availability reduces the growth of Q. faginea (Rey Benayas et al., 2005). Effecs of human and livesock populaions on Q. faginea recruimen The first expansion of Q. faginea into the study area occurred in the late 1930s, when local human populations declined, which might have reduced anthropogenic pressures on the territory in the area. Furthermore, the high Q. faginea recruitment in the late 1960s and early 1970s coincided with an increase in the recruitment rate of Q. faginea on the abandoned terraces and the decline in the local human population that had begun about 40 yr earlier. The negative correlation between Q. faginea recruitment and the number of livestock suggests that grazing pressure had a significant negative effect on the establishment of Q. faginea, particularly between 1890 and 1930. In addition, the first peak in Q. faginea recruitment occurred when grazing pressure began to decline, and the highest recruitment peaks occurred in late 1960s and early 1970s, which coincided with the lowest numbers of livestock. Livestock overgrazing constrains the regeneration of tree species (Barbero et al., 1990; Carmel and Kadmon 1999). Livestock eliminate seedlings, which diminishes recruitment and, consequently, hinders forest regeneration (Cierjacks and Hensen, 2004; Callaway and Davis 1993; Wahren et al. 1994). The increase in the number of livestock that began in the 1970s was not accompanied by an increase in grazing pressure because of significant changes in livestock husbandry in the Central PrePyrenees (GarcíaRuiz et al., 1996). Since the 1970s, the number of livestock grazing freely in the mountain grasslands and rangelands of the study area in summer has decreased sharply (Molinillo et al., 1997; Lasanta et al., 2006). Capitulo 3 104 Conclusions In the 20th Century, changes in land use and climate have strongly influenced the dynamics of Q. faginea forests in the Central PrePyrenees. The history of recruitment in those forests involved the following stages: (i) before 1935, the establishment of Q. faginea was restricted mainly to coppice stands because of extensive farmland cultivation on the mountain terraces and livestock overgrazing; (ii) between 1935 and the early 1960s, reductions in human land use and livestock pressure favored Q. faginea recruitment and expansion; (iii) in the late 1960s and early 1970s, the encroachment of abandoned terraces by Q. faginea was enhanced by favorable climatic conditions; and (iv) since 1975, Q. faginea recruitment has been stressed by drought (insufficient amount of rainfall in winter and spring, and high temperature in summer). high rates of tree growth and recruitment in this species should be maintained by using improved management of the forests based on drought alerts and mitigation adaptive systems (e.g., preventive thinning in very dense forests) and by the enhancement and regulation of Q. faginea colonization in formerly cultivated and grazed lands (e.g., selection of vigorous or reproductive trees in encroached abandoned terraces). Acknowledgements The Spanish Ministry of Education and Science (projects CICYT CGL2008 00655/BOS and CGL200804847C0201/BOS) funded this research. A MAEC AECID grant from the Spanish Agency for International Cooperation and Development supported the first author. J.J. Camarero is grateful for the support of ARAID and the Globimed Network (www.globimed.net). We thank A.Q. Alla, G. SangüesaBarreda, and P.N. Galve for assisting in the collecting of field data, and the Agencia Estatal de Meteorología for providing weather data. Capitulo 4 111 Pirineos (accepted) CAPITULO 4 La expansión del quejigo (Quercus faginea) en el Prepirineo Aragonés durante la segunda mitad del siglo XX The expansion of Quercus faginea in the Aragón’s Pre-Pyrenees over the second half of the twentieth century Yacine Kouba 1 y Concepción L. Alados 1 1 Insiuo Pirenaico de Ecología, CSIC. Avda. Monañana 1005, PO Box 13034, ES50192 Zaragoza. [email protected] Resumen Se investigó la expansión de Quercus faginea Lam. (quejigo) debida a la disminución de la presión antrópica en siete municipios del Prepirineo Aragonés durante la segunda mitad del siglo XX. La emigración rural que ha ocurrido en esta zona a partir de los años sesenta ha producido un abandono masivo de las tierras agrícolas junto con la disminución de la presión pastoral. Como resultado algunos de los campos abandonados han sido colonizados por el quejigo, sobre todo en los municipios con una disminución notable en el número de agricultores y cabezas de ganado. Se espera que, con la continua disminución de la presión humana en el territorio, los quejigares colonicen nuevas áreas en los próximos años como resultado de la sucesión secundaria. Palabras clave: Montes bajos. Terrazas abandonadas. Bosques segundarios. Población y ganado. Presión humana en el territorio. ABSTRACT.-The expansion of Quercus faginea Lam. -due to a decrease in human pressurewas investigated in seven municipalities of the Aragón’s Pre-Pyrenees over the second half of the twentieth century. The rural emigration that occurred in this area since 1960s has generated a massive abandonment of agricultural lands and a Capitulo 4 112 decrease in livestock pressure. As a result, some abandoned fields have been colonized by Q. faginea; specially, in the municipalities that have known a marked decrease in the numbers of farmers and livestock. With the continued decrease of human pressure in the territory, it is expected that Q. faginea will colonize new areas during the next years as a result of secondary succession. Key words: Coppices. Abandoned terraces. Secondary forests. Population and livestock. Human pressure in the territory. Introducción Desde hace miles de años la cuenca mediterránea está sometida a cambios constantes en los usos de suelo (Quezel & Barbero, 1990; Hoùerou, 1981), habiendo dominado la deforestación para aumentar la superficie de las tierras cultivadas y los pastos ( Barbero et al., 1990). Durante la segunda parte del siglo veinte, sin embargo, esta dinámica ha variado debido a cambios socioeconómicos (Margaris et al., 1996; Lasanta & Vicente-Serrano, 2007). Mientras que en los países de la parte sur de la cuenca mediterránea la deforestación y la presión sobre las tierras continúa, en los países euro-mediterráneos se han abandonado muchos campos de cultivo y se ha reducido la presión pastoral (MacDonald et al., 2000). Estos cambios en el uso del suelo han provocado en estos últimos países un cambio significativo en el paisaje, en particular la regeneración natural de la vegetación en los campos abandonados (Margaris et al., 1996; MacDonald et al., 2000). En el Prepirineo Aragonés las zonas rurales han experimentado cambios socioeconómicos profundos, con un gran despoblamiento, muy especialmente entre las décadas sesenta y ochenta. Ello ha supuesto el abandono de importantes superficies anteriormente cultivadas, la disminución de la presión pastoral y cambios en las fuentes de alimentación del ganado (García-Ruiz & Lasanta, 1990; Lasanta et al., 2010). Estos cambios han afectado a la dinámica de los ecosistemas, desencadenándose un proceso de revegetación con avance del bosque y de los matorrales como consecuencia de la sucesión secundaria (Lasanta et al., 2000, 2010; Vicente-Serrano et al., 2000). Capitulo 4 113 Algunos estudios previos han argumentado que los campos abandonados solo llegan a ser totalmente recubiertos con matorral o bosque en las áreas donde la sucesión no ha sido interrumpida por las actividades humanas, principalmente el pastoreo y la agricultura (Lasanta et al., 2010). Otros estudios recientes han documentado la instalación de nuevas masas de Quercus faginea Lam. (quejigo) en el Prepirineo Aragonés (Kouba & Alados, 2012; Kouba et al., 2012), especialmente en las terrazas abandonadas, durante la segunda parte del siglo XX como resultado de la sucesión (Kouba et al., 2012). Aunque los bosques de quejigo tienen un papel relevante en la conservación de la diversidad biológica (Kouba et al., 2011; MaltezMouro et al., 2009; Rey Benayas et al., 2005), pocos trabajos han tratado de estudiar la dinámica de los quejigares del Prepirineo Aragonés. Este trabajo parte de la hipótesis de que existe una relación entre los cambios socioeconómicos que han ocurrido en el Prepirineo Aragonés durante la segunda mitad del siglo veinte y la expansión del quejigo. Particularmente, queremos saber si la colonización del quejigo de algunas zonas del Prepirineo es realmente el resultado de la reducción de la presión antrópica en el territorio como consecuencia del abandono de las tierras agrícolas y la disminución de la presión pastoral. Materiales y métodos Situación geográfica y descripción de la zona de estudio Se ha seleccionado un área de 1363 km 2 en el Prepirineo Aragonés que incluye siete municipios: Jaca, Sabin anigo, Arguis, Nueno, Caldearinas, Loarre y Las Pinas de Riglos (Figura 1). El area se caracteriza por una gran variacion altitudinal, desde los 500 m en las Sierras Interiores a los 2000 m en las zonas más altas. El clima es sub-Mediterraneo con influencia continental en la parte norte y con influencia Mediterranea en la parte Sur. La precipitacion media anual es de 1317 mm y la temperatura media anual es de 11,5 ºC (Kouba et al., 2012). Las precipitaciones presentan fuerte variacion estacional y el periodo con mayor pluviosidad ocurre entre octubre y junio (Lasanta et al., 2000). El sustrato litologico esta dominado por rocas areniscas, lutitas, margas y calizas. La cobertura vegetal es muy variada e incluye pinares de P. sylvestris, P. nigra (naturales o repoblados), Fagus sylvatica, Q. ilex, y Q. faginea. Los bosques de quejigo se extienden principalmente en las zonas Capitulo 4 114 del flysch y rocas detríticas eoceno-oligocenas de la depresión media pirenaica (800-2000 m). Tambien existen matorrales dominados por Q. coccifera y Buxus sempervirens,pastizales, campos abandonados y terrenos agrõcolas. Figura 1: Mapas de distribución de masas forestales de Q. faginea y de matorral en el área de estudio en los años (A) 1957 y (B) 2006. Figure 1:DistributionmapsofQ.fagineastandsandshrublandsinthestudyareaduring(A)1957 and(B)2006. Identificación y cuantificación de los cambios en la superficie ocupada por los bosquesdequejigoentre1957-2006 En primer lugar se han elaborado los mapas de distribución de Q.faginea en los años 1957 y 2006 (Figura 1). El mapa de distribución en 1957 se ha elaborado mediante la interpretación visual de las fotos aéreas del año 1957 proporcionados por el Centro Cartográfico y Fotográfico del Ejército del Aire. Basándonos en los trabajos publicados (Molinillo etal., 1997; Monserrat, 1990) y en nuestra propia inspección de las fotos aéreas, hemos asumido que sólo las zonas que eran ocupadas por matorral en 1957 podrían haberse transformado en manchas de Q. faginea en 2006, por lo que las manchas de matorral se han incluido en el mapa de 1957. El mapa de distribución en 2006 fue elaborado a partir del tercer Inventario Nacional Forestal (IFN3; MMA, 2007). Para ello se seleccionaron las manchas de ² 0 2010 Km (A) (B) Matorral Quercus faginea Matorral Quercus faginea Capitulo 4 115 bosque en las que el quejigo era una de las tres especies arbóreas más abundantes. Las manchas de distribución de la vegetación fueron corregidas con la ayuda de las fotos aéreas orto-rectificadas a escala 1:30000 del Plan Nacional de Ortofotografía Aérea (PNOA, 2006). Debido a que Q. faginea se hibrida con Q. pubescens, haciendo difícil la identificación de individuos (Himrane et al., 2004; Loidi & Herrera, 1990), hemos incluido los híbridos (principalmente Q. subpyrenaica) con Q. faginea. Hay que señalar que hemos usado una unidad mínima cartografiable (UMC) de 0,1 ha por mancha en ambos mapas. Los cambios en la superficie ocupada por Q. faginea se obtuvieron a partir de la superposición de los mapas de 1957 y de 2006. El siguiente paso fue cuantificar las ganancias en la superficie ocupada por Q. faginea en cada uno de los siete municipios de la zona de estudio, mediante la superposición del mapa de las ganancias con el mapa de los límites administrativos de los municipios. Es importante señalar aunque el área de estudio no incluye la totalidad de los términos municipales implicados, envuelve la mayor parte de los bosques de quejigo en estos municipios. Caracterización de las zonas con presencia de quejigo y definición de los indicadores socioeconómicos Las características de las zonas con presencia de Q. faginea se relacionan con variables climáticas que pueden influir directa o indirectamente en Q. faginea y con variables antrópicas que reflejan las perturbaciones causadas por las actividades humanas (Kouba et al., 2011). La variable climática es el balance hídrico (mm), que refleja las condiciones de sequía en la zona de estudio, esta variable se obtuvo del Atlas Climático de Aragón (DMA, 2007) con una resolución de 100-m, para el periodo 1971-2000. Las variables antrópicas que se incluyen son: la distancia a la repoblación de pinar más próxima y la distancia de coste a pastizales (Kouba et al., 2011). Estas variables se han usado para saber si las extensivas reforestaciones con pinos durante los últimos 50 años y el pastoreo han influenciado a la distribución espacial de Q. faginea. La distancia a las repoblaciones de pinares se realizó calculando la distancia euclidiana entre cada pixel y la reforestación más próxima. La distancia de coste es una combinación de la distancia entre dos puntos y la pendiente que los separa. Capitulo 4 116 Los cambios en el número de habitantes (1960-2006), el número de agricultores (1970-2006) y el número de cabezas de ganado (1970-2006) se han usado como indicadores de los cambios socioeconómicos en la zona de estudio. Dichos cambios se han calculado por cada municipio de la zona de estudio usando los datos de evolución de la población humana y los censos de población obtenidos del Instituto Aragonés de Estadística (IAE) y el Instituto Nacional de Estadística (INE), respectivamente, así como los datos de ganado (ovino, bovino, y caprino) obtenidos del IAE y el Servicio Provincial de Agricultura de Huesca. Hay que señalar que para el cálculo del número total de cabezas de ganado se ha considerado que una vaca es equivalente a seis ovejas (García-González and Marinas, 2008). Análisis de los datos Para definir las características de las zonas con presencia de Q. faginea se ha usado el mapa de distribución de Q. faginea en 2006 para extraer aleatoriamente 1000 puntos, la mitad con presencia y la otra mitad con ausencia de Q. faginea. Para asegurar la representatividad de la muestra, cada mancha de Q. faginea se ha representado por cinco puntos o más. El valor de cada variable en cada punto se obtuvo usando la función “simple” del software ArcGIS 10.1 (ESRI, 2013). La base de datos resultante fue importada al software estadístico R (R Development Core, 2013). Para examinar las diferencias entre las zonas con presencia y las con ausencia de Q. faginea se ha usado la prueba de Wilcoxon. Las relaciones entre el cambio en el número de habitantes, agricultores y cabezas de gado y el cambio en la superficie ocupada por Q. faginea han sido examinadas mediante coeficientes de correlación de Spearman. Resultados La mayor parte de los bosques de Q. faginea del Prepirineo están situados en zonas que se caracterizan por un balance hídrico negativo e inferior a las zonas sin Q. faginea (Figura 2a). Una gran parte de los bosques de Q. faginea del Prepirineo Capitulo 4 117 están relativamente alejados de las repoblaciones de pinos, con un intervalo de distancia de 1 a 6 km y un promedio de 2,5 km (Figura 2b). Las zonas con presencia de Q. faginea parecen ser de muy difícil acceso desde los pastos; en otras palabras, son zonas inaccesibles al ganado (Figura 2c). Figura 2: Comparación de las áreas con presencia y con ausencia de Q. faginea. El valor de la prueba de Wilcoxon (W) y el nivel de significación estadística (p) son mostrados por cada variable. Figure 2: The comparison between the areas with presence and those with absence of Q. faginea. The value of Wilcoxon test (W) and its significance level (p) are shown for each variable. Capitulo 4 118 Figura 3: Cambios en el número de (A) habitantes, (B) agricultores, y (C) cabezas de ganado en los municipios de la zona de estudio (media ± error estándar) entre 1960 y 2006. Figure 3: Changes in the number of (A) inhabitants, (B) farmers, and (C) livestock in the municipalities of the study area (mean ± standard error) between 1960 and 2006. 1960 1970 1980 1990 2000 2010 0 20 40 60 80 100 120 140 Número de habitantes 1960 1970 1980 1990 2000 2010 0 100 200 300 400 500 600 Número de agricultores 1960 1970 1980 1990 2000 2010 0 2000 4000 6000 8000 10000 12000 14000 Número de cabezas de ganado (A) (B) (C) Capitulo 4 119 Se ha producido un decremento de la población humana y del número de agricultores en los municipios de la zona de estudio durante el periodo (19602006) (Figura 3A, B). Sin embargo, el número de cabezas de ganado ha aumentado durante el periodo 1960-2006 (Figura 3C). Existe una relación negativa entre la expansión de Q. faginea y los cambios en el número de habitantes (Figura 4), número de agricultores (Figura 5) y número de cabezas de ganado (Figura 6) en los municipios de la zona de estudio. Discusión El hecho de que Q. faginea se encuentre en zonas con un balance de agua negativo se debe a que, en la zona de estudio, este aumenta con la altitud, donde a su vez disminuyen ampliamente las temperaturas, impidiendo el establecimiento y crecimiento de las plántulas de Q. faginea (Kouba et al., 2011). Así, el área de distribución del Q. faginea en la zona de estudio se encuentra entre las cotas 450 m a 1500 m (Kouba et al., 2011). Variación del número de habitantes (1960-2006) -1000 -500 0 500 1000 1500 2000 2500 Expansión de Q. faginea (ha) 20 40 60 80 100 120 140 160 r = - 0.67; p < 0.05 Figura 4: Relación entre la expansión de Q. faginea y el cambio en el número de habitantes entre 1970 y 2006 en los municipios de la zona de estudio. El valor del coeficiente de correlación de Spearman (r) y el nivel de significación estadística (p) son mostrados. Figure 4: Relationship between the expansion of Q. faginea and changes in the number of inhabitants during the period 1970-2006 in the municipalities of the study area. The value of Spearman correlation coefficient (r) and its significance level (p) are shown. Capitulo 4 120 Variación del número de agricultores (1970-2006) -200 -150 -100 -50 0 Expansión de Q. faginea (ha) 20 40 60 80 100 120 140 160 r = - 0.85; p < 0.05 Figure 5: Relación entre la expansión de Q. faginea y la variación en el número de agricultores entre 1970 y 2006 en los municipios de la zona de estudio. El valor del coeficiente de correlación de Spearman (r) y el nivel de significación estadística (p) son mostrados. Figure 5: Relationship between the expansion of Q. faginea and changes in the number of farmers during the period 1970-2006 in the municipalities of the study area. The value of Spearman correlation coefficient (r) and its significance level (p) are shown. Variación del número de cabezas de ganado (1970-2006) -5000 0 5000 10000 15000 20000 25000 30000 Expansión de Q. faginea (ha) 20 40 60 80 100 120 140 160 r = - 0.86; p < 0.05 Figure 6: Relación entre la expansión de Q. faginea y el cambio en el número de cabezas de ganado entre 1970 y 2006 en los municipios de la zona de estudio. El valor del coeficiente de correlación de Spearman (r) y el nivel de significación estadística (p) son mostrados. Figure 6: Relationship between the expansion of Q. faginea and changes in the number of farmers during the period 1970-2006 in the municipalities of the study area. The value of Spearman correlation coefficient (r) and its significance level (p) are shown. Capitulo 5 127 European Journal of Forest Research (under review) CAPITULO 5 Plant -diversity in human-altered forest ecosystems: The importance of the structural, spatial, and topographical characteristics of stands in patterning plant species assemblages Yacine Kouba* 1 , Felipe Martínez-García 2 , Ángel de Frutos 3 and Concepción L. Alados 1 1 Pyrenean Institute of Ecology (CSIC). Avda. Montañana 1005, PO Box 13034, ES50192, Zaragoza, Spain; 2 Dpto. de Silvopascicultura, Escuela de Ingeniería Forestal y del Medio Natural, Universidad Politécnica de Madrid. Paseo de Las Moreras s/n, ES28040, Madrid, Spain; 3 Pyrenean Institute of Ecology (CSIC), Avda Nuestra Señora de la Victoria s/n, ES-22700, Jaca (Huesca), Spain;*Corresponding author: Tel: +34 976 716034; fax: +34 976 716019; Email: [email protected] Abstract An understanding of spatial patterns of plant species diversity and the factors that govern and generate those patterns is critical for the development of appropriate biodiversity management in forest ecosystems. We studied the spatial organization of plants species in human-modified and managed oak forests (primarily, Quercus faginea) in the Central Pre-Pyrenees, Spain. To test whether plant community assemblages varied non-randomly across the spatial scales, we used multiplicative diversity partitioning based on a nested hierarchical design of three increasingly coarser spatial scales (transect, stand, region). To quantify the importance of environmental factors in patterning plant community assemblages and identify the determinants of plant diversity patterns, we used canonical ordination. We observed a high contribution of -diversity to total γ-diversity and found -diversity to be higher and -diversity to be lower than expected by random distributions of individuals at different spatial scales. Environmental Capitulo 5 128 variables that are strongly influenced by historical land use such as mean stand age, the abundance of the dominant tree species (Q. faginea), the age structure of stand, stand size, and the topographical conditions (i.e., slope) were the main factors that explained the compositional variation in plant communities. The results indicate that (1) the structural, spatial, and topographical characteristics of the forest stands have the greatest effect on diversity patterns, (2) forests in landscapes that have different land use histories are environmentally heterogeneous and, therefore, can experience high levels of compositional differentiation, even at local scales (e.g., within the same stand). Maintaining habitat heterogeneity at multiple spatial scales should be considered in the development of management plans for enhancing plant diversity and related functions in human-altered forests. Key words: secondary forests, community assembly, forest structure, compositional dissimilarity, beta diversity, species turnover. Introduction Most studies of forest ecosystems focused on -diversity, i.e., the diversity within a specific site; however, recent studies that have partitioned diversity into hierarchical components have shown that much of the floral diversity is due to differentiation in species composition among sites (-diversity; Arroyo-Rodríguez et al., 2013; Chandy, Gibson, & Robertson, 2006; Gossner et al., 2013). An understanding of how diversity components, particularly, -diversity, vary spatially, and the factors that are responsible for the patterns observed is essential for understanding how species diversity is organized and maintained (Condit et al. 2002; Arroyo-Rodríguez et al. 2013). Particularly, in human-modified and managed forests, the structural, spatial, and topographical characteristics of the forest stands, which are strongly influenced by historical land use-type and intensity, might have a significant role in shaping plant diversity patterns (Flinn and Vellend 2005; Hermy and Verheyen 2007; Berhane et al. 2013). Recent studies have found that forest stands in landscapes that have different land use histories manifest a high environmental heterogeneity, which can lead to high levels of compositional differentiation (i.e., -diversity) even at fine scales (e.g. Arroyo- Capitulo 5 129 Rodríguez et al. 2013). The floristic differentiation can drive successional trajectories and potentially affect the maintenance of biodiversity in such altered forests (Chazdon et al. 2009; Melo et al. 2013; Arroyo-Rodríguez et al. 2013). For centuries, the oak forests (mainly, Quercus faginea) in the western Mediterranean region have been harvested intensively for timber and firewood, and clearcut for agriculture (Sancho et al. 1998), which has reduced them to coppice stands that have different management histories; i.e., different coppicing intensities and time since coppicing ceased (Sancho et al. 1998). In the late 19th and 20th centuries, however, changes in socioeconomic structures and production systems resulted in the abandonment of the poorest arable lands and their subsequent afforestation (Sciama et al. 2009). In particular, in the Central Pyrenees, Spain, the encroachment of some abandoned farmlands by Q. faginea has led to new, secondary growth Q. faginea-dominated stands (Kouba et al. 2012). Probably, the biodiversity that can be observed in these human-modified and managed forests (i.e., either the formerly managed or the new secondary growth forests) does not have value for conservation, per se, because it usually is deprived of any conservation status. Often, however, such islets of habitats are considered biodiversity “refuges”, which allows them to recover many components of the original biodiversity (Chazdon 2008), and provide important ecosystem services such as control of climate and erosion. Therefore, the assessment of plant diversity patterns across multiple spatial scales and the factors that govern and generate those patterns is required to accurately evaluate the impact of historical maninduced disturbances on the spatial dissimilarities in species composition (- diversity) and to gain a better understanding of the mechanisms that contribute to the maintenance of species diversity in such human-modified and managed forests. Although the use of multi-scale analyses to analyze the spatial patterns of faunal diversity has increased, very few studies have used this approach to assess the hierarchical organization of plant species diversity, particularly -diversity, at multiple spatial scales in forest ecosystems (but see Chandy et al. 2006; Chávez and Macdonald 2012; Arroyo-Rodríguez et al. 2013). In this study, we used multiplicative diversity partitioning to understand how plant species diversity changes across three spatial scales (transect, stand, and entire region) as well as to identify the spatial scales at which nonrandom processes have had the greatest Capitulo 5 130 effect. To identify the environmental factors that might have patterned plant species diversity in human-modified and managed oak forests, we used constrained ordination analysis (RDA). We hypothesized that (H1) plant community assemblages vary non-randomly across the spatial scales, (H2) - diversity components contribute more to -diversity than do -diversity components because of high habitat heterogeneity, and (H3) the structural, spatial, and topographical characteristics of the forest stands, which are largely the result of historical land use, are the main factors that structure the compositional variation in plant communities in these human-modified and managed forests. Methods Study area The study was conducted within a 1363-km 2 area at an elevation of 450-1950 m a.s.l. in the Central Pre-Pyrenees, Spain (between 42.32 N to 42.11 N, and 0.31 W to 0.04 W) (Fig. 1). The lithology is mostly conglomerate, limestone, marl, and sandstone developed on Eocene flysch sedimentary formations (Kouba and Alados 2011). The climate is transitional sub-Mediterranean; i.e., influenced by continental effects from the Pyrenees to the north and by milder Mediterranean conditions that prevail from the south (i.e., the Ebro Basin). In the study area, mean annual precipitation is 1317 ± 302 mm (1915-2005) (Kouba et al. 2012) and mean annual air temperature is 11.5 ± 2.8º C (1910-2005) (Kouba et al. 2012). The area has a variety of land-use/cover types including natural forests of Pinus sylvestris, P. nigra, Fagus sylvatica, Q. ilex, and Q. faginea, shrublands of Q. coccifera and Buxus sempervirens, artificial plantations of P. sylvestris and P. nigra, arable farmland, pastures (xeric pastures and subalpine pastures), urban areas, and abandoned farmland. In the second half of the twentieth century, major changes in land use occurred in the area (Lasanta et al. 2005) because of agricultural mechanization and intensification, the introduction of pine plantations, and the abandonment of croplands and pastures, which has led to forest regrowth (Lasanta et al. 2005; Vicente-Serrano et al. 2010). In the area, Q. faginea is one of the most abundant naturally occurring species and the communities in which it occurs constitute a transition zone between Mediterranean forests in which Q. ilex ssp. Capitulo 5 131 ballota or P. halepensis are predominant, and mountain continental or mesic forests of P. sylvestris, P. nigra ssp. salzmannii, and F. sylvatica (Loidi and Herrera 1998; Sancho et al. 1998). The overstorey canopy of those semi-deciduous oak stands is dominated by Q. faginea interspersed with some scattered pines (Pinus sylvestris and P. nigra) and evergreen oak (Q. ilex subsp. ballota). The understory is composed of shrubs (Q. coccifera, B. sempervirens, Genista scorpius, Juniperus communis), forbs (Aphyllanthes monspeliensis, Arenaria montana, Achillea millefolium), and graminoids (Brachypodium pinnatum, Carex halleriana, Festuca rubra, Carex flacca, Bromus erectus). Fig. 1 Location of the study area within Europe (upper right panel), and the locations of the ten Q. faginea forest stands sampled in the Central Pre-Pyrenees, Spain (left panel). The location of the three floristic transects (FT) and the forest structural transect (ST) within each stand (lower right panel). AB = Abena, AG = Arguis, AR = Ara, BE = Belsué, IB = Ibort, IP = Ipies, LU = Lucera, NO = Nocito, RA = Rasal, RP = Rapun Stand selection and surveys Based on the distribution maps of Q. faginea in the study area in 1957 and 2006 (Kouba and Alados 2011) and dendrochronological data that reflect the historical dynamics of Q. faginea stands in the study area (Kouba et al. 2012), ten Q. fagineaETRS_1989_UTM_Zone_30N Table 1 Characteristics of ten oak stands within a 1363-km 2 area in the Central Pre-Pyrenees, Spain. Values are mean ± standard error. ELEVAT = elevation, ORIENT = orientation (S = South, SE = South East, SW = South West, E = East), SLOP = slope, STSIZE = stand size, SHPCOMP = shape complexity, DENSITY = density, QFAB = Q. faginea abundance, DBH = diameter at breast height, TREHEIGHT = tree height, AGE = mean stand tree age, CVAGE = Coefficient of Variation of age of stand, FORTYPE = Forest type (SF = secondary forest, CS = abandoned coppice stand) Stand characteri stics/stand locations Rasal (RA) Belsué (BE) Abena (AB) Ara (AR) Lucera (LU) Ibort (IB) Ipies (IP) Nocito (NO) Arguis (AG) Rapun (RP) Topography ELEVAT (m a.s.l.) 868.3 ± 4.8 1158.5 ± 1.20 970.3 ± 1.50 971.1 ± 2.00 1198.0 ± 7.70 950.8 ± 2.60 852.5 ±2.30 1046.7 ± 2.10 1026.2 ± 1.90 923.3 ± 2.40 ORIENT S S S SE SE S E SW S SW SLOP (º) 9.3 ± 0.50 30.5 ± 0.40 11.7 ± 0.50 19.6 ± 0.50 16.8 ± 1.20 14.8 ± 1.10 7.8 ± 0.60 25.0 ± 0.80 11.0 ± 0.60 17.98 ± 1.40 Spatial attributes STSIZE (ha) a 114 94 73 244 1115 40 146 294 1847 217 SHPCOMP (perimeter/area) a 126.41 119.60 77.49 164.23 244.28 103.17 268.11 267.27 232.38 204.62 Forest structure DENSITY (stems ha-1) 607 ± 0.20 1100 ± 0.10 999 ± 0.10 503 ± 0.30 867 ± 0.10 1088 ± 0.10 812 ± 0.10 983 ± 0.10 818 ± 0.10 540 ± 0.10 QFAB (Tree/Transect) 239±43 362±15 339±18 133±32 173±14 426±7 193±8 389±28 381±13 212±18 DBH (cm) 14.00 ± 1.40 9.0 ± 0.70 13.3 ± 1.30 7.2 ± 0.50 12.0 ± 0.80 13.3 ± 0.80 11.4 ± 0.70 12.3 ± 1.70 13.0 ± 1.40 6.8 ± 0.50 TREHEIGHT (m) 5.10 ± 0.40 4.8 ± 0.30 5.1 ± 0.30 3.4 ± 0.20 5.5 ± 0.30 6.1 ± 0.20 4.3 ± 0.30 5.5 ± 0.41 4.7 ± 0.33 3.9 ± 0.25 AGE (years) 31 ± 3 40 ± 4 50 ± 2 35 ± 1 39 ± 1 63 ± 2 64 ± 2 56 ± 5 50 ± 1 69 ± 2 CVAGE (%)d 31 43 19 17 12 17 15 47 10 9 FORTYPE SF CS SF CS CS CS CS SF CS CS a Calculated based on the distribution map of Q. faginea forests in the study area (for more details, see Kouba et al. 2011) 132 Capitulo 5 133 dominated stands that differed in their structural, spatial, and topographical characteristics were selected within the study area (see Table 1, Fig. 1). Primarily, the stands were surrounded by farmland, pine plantations, abandoned land, and grasslands (see Fig. 1). In 2009 and 2010, during the period of peak growth (May and June), the vascular plant species were surveyed in the ten stands. Within each stand, three 500-m linear transects (30 transects in total) were established (hereafter, floristic transects). To estimate plant abundance and richness within each transect, we used the Point-Intercept Method (Goodall 1952), which involves recording, at 40cm intervals, the identity of all individuals that are in contact with a vertical nail (Alados et al. 2009). We recorded all of the vascular plants that touched the nail and any overstorey species (including Q. faginea) that was above the nail. The abundance of each species in each transect was estimated as the number of individuals (of this species) recorded along the transect. Plant species that could not be identified with certainty in the field were collected, pressed, and brought to the laboratory for identification by botanical experts. Species that have traits that make them difficult to distinguish were only identified to the genera level. Plant nomenclature followed Flora Ibérica (Castroviejo et al. 1986-2012). Plant growth forms represent broad patterns of variation among correlated plant traits that are more related to ecosystem functions, e.g. nutrient use efficiency, protection against abiotic and biotic hazards, and competitive strength (Lavorel et al. 1997; Dorrepaal 2007), and, therefore, are expected to differ in their responses to environmental factors. Accordingly, plant species were grouped based on growth forms: woody (tree and shrubs), graminoids, or forbs. In this study, the relative abundance of Q. faginea (QFAB) in each floristic transect was included in the analyses as surrogate for the amount of canopy cover (%). To quantify the structural properties of each stand (Table 1) a 500-m linear transect (hereafter, forest structure transect) was established within each stand (n =10) and the forest was sampled using the Point-quarter Method (Cottam and Curtis 1956). Each forest structure transect was placed close to the central floristic transect within each stand. Sampling points (n = 20) were at 25-m intervals along each of the transects. At each sampling point, we identified the closest adult Q. faginea tree in each of the four cardinal directions within a maximum distance of 5- Capitulo 5 134 m from the sampling point (Kouba et al. 2012). Adult trees were defined as those > 2 m high or that had a stem diameter at breast height (DBH) ³ 4 cm. The following measurements were recorded: diameter at breast height (DBH) (cm), tree height (m), and age (for details about age estimation, see Kouba et al. 2012). Those data were used to estimate the following variables for each stand: density (DENSITY), mean diameter at breast height (DBH), mean tree height (TREHEIGHT), mean age (AGE), and coefficient of variation of tree age (CVAGE). Furthermore, forest type (FORTYPE; secondary growth stands vs. abandoned coppice stands) was recorded for each stand based on visual observation on the field (see Table 1). To quantify the spatial attributes of each stand (Table 1), we measured stand size (STSIZE) and shape complexity (SHPCOMP) using a digitized Q. faginea distribution map, the ‘Patch Analyst’ extension in ArcGIS 10.1 (ESRI 2013), the Third National Forest Inventory map (IFN3; MAGMARA, 2013), and orthorectified aerial photographs taken in 2006 (CINTA 2013). In addition, the mean elevation (ELEVAT), mean slope (SLOP), and orientation (ORIENT) of each stand were derived from a Digital Elevation Model (CINTA 2013). Partitioning of biodiversity To assess plant diversity patterns across multiple spatial scales, we used multiplicative partitioning because of the advantages of the Hill Number ( q D) and q-metric (see below): q D  = q D  × q D  (Whittaker 1972; Jost 2006, 2007, 2010). Diversity is quantified using the Hill Number ( q D), which has the property to be invariant to changes in absolute numbers; if all species double in abundance, q D remains unchanged. It measures variation in relative, rather than absolute abundance, and it follows the replication principle: Combining two sets of nonoverlapping species that have the same abundance distributions doubles the value of q D (Jost 2006; Scheiner 2012). To quantify diversity patterns based on various weightings for rare and abundant species, we used the q-metric, which reflects the sensitivity of the diversity index to the relative frequencies of species. The analyses included two q-values: (1) q = 0 reflects species richness, which is not sensitive to species abundance and, therefore, assigns disproportionate weight to rare species (Jost 2006), and (2) q = 0.999 (and not q = 1, which would require division by zero) is equivalent to the exponential of Shannon entropy; here, species are weighted in Capitulo 5 135 proportion to their frequency in the sampled community and, therefore, it can be interpreted as the number of ‘typical species’ in the community (Chao et al. 2012). We used a nested hierarchical design of three increasingly coarser spatial scales: individual assemblages at the transect level, pooled assemblages within a stand, and a single, pooled assemblage across the entire region (Fig. 2). Fig. 2 Hierarchical levels in the multiplicative partitioning of plant species diversity in ten oak forest stands in the Central Pre-Pyrenees, Spain The design allowed q D  diversity to be decomposed into within transect ( q D  _ transects ), among transects ( q D  _ transects ), within stand ( q D  _ stands ), and among stands ( q D  _ stands ) components (Fig. 2). To test for significant differences in the spatial partitioning of diversity, the expected values of the measures of diversity were calculated using individual-based randomizations (10 4 permutations; Crist et al. 2003), which evaluated whether the  and  components of diversity differed significantly from a random distribution of individuals among samples (Crist et al. 2003). Those analyses were performed using the ‘vegan’ package (Oksanen et al. 2013) implemented in the R software (R Development Core Team 2013). To test whether differences in species richness might have biased the observed spatial diversity pattern, we additively partitioned -diversity into the two components of spatial turnover and nestedness using the method suggested by Baselga (2010). We performed this analysis using the ‘betapart’ package (function Capitulo 5 136 ‘beta.sample’) (Baselga and Orme 2012) within the R software (R Development Core Team 2013). Partitioning the variation in plant communities in response to environmental factors To identify the environmental variables that explained a significant amount of the variation in species composition, we used Canonical Redundancy Analyses (RDA). The matrices of species abundance were transformed using Hellinger’s Transformation (Legendre and Gallagher 2001). The explanatory variables included in the finale RDA models were selected based on forward stepwise procedure, which provided an estimate of the best set of non-redundant variables for predicting species composition and a ranking of the relative importance of the individual explanatory variables. The spatial autocorrelation of the residuals of the RDA models was tested using a multi-scale ordination (MSO; Borcard et al. 2011; Legendre and Legendre 2012). Initial analyses indicated significant spatial autocorrelation in the residuals of the RDA models and a scale-dependent relationship between the species data and the explanatory variables. To address those problems, the following three steps were followed: (i) the Hellinger-transformed species data matrices and the explanatory variables were detrended along the Y Cartesian geographic coordinates (i.e., the coordinates of transect-central points), which supported the assumption of stationarity in the computation of confidence intervals in the MSO variograms (Legendre and Legendre 2012). (ii) The sampling design was spatially nested; therefore, the function ‘create.MEM.model’ (Borcard et al. 2011; Declerck et al. 2011) was used to construct a staggered spatial matrix of Moran’s eigenvector maps (MEM), and (iii) partial canonical redundancy analyses (partial RDAs) were performed using the detrended data and included the computed MEMs as covariables, which controlled for the effects of spatial structure (i.e., excluded the compositional variation caused by spatial structure; Borcard et al. 2011; Legendre and Legendre 2012). Results In the survey of the 10 oak stands in the Central Pre-Pyrenees, Spain, we identified 238 vascular plant species. On average, the floristic transects contained