Tree populations are facing new and rapidly changing selective pressures such as more frequent extreme droughts which threat the conservation of forests and related ecosystem services (Lindner et al., 2010). Drought-induced forest decline is a major emerging threat for forests in drought-prone areas (Allen et al., 2010) but also in mesic temperate forests (van Mantgem & Stephenson, 2007). At broad biogeographical scales, drought-related forest decline has been usually reported near the species southernmost distribution limit (rear edge sensu Hampe & Petit, 2005) where drought stress is often presumed to be more limiting for the species persistence than in the core area of distribution (Jump et al., 2009).

One of the major research gaps identified by Allen et al. (2010) is a better knowledge on within-species genetic variability as related to forest decline which may allow understanding the differential responses of rear-edge tree populations to drought stress. First, rear-edge populations may be genetically impoverished because of their historical isolation or due to further fragmentation. Second, rear-edge tree populations living near the species xeric distribution limit may have adapted to drought stress and represent genetic reservoirs of drought-resistant genotypes. Genetic impoverishment may lead to high vulnerability against drought-induced decline. On the contrary, selection of drought-resistant genotypes may lead to populations of a high adaptive value in a warmer and dryer climatic scenario.

Silver fir (Abies alba) constitutes a suitable species to test these contrasting ideas since (i) its genetic variation and paleogeography across many European populations has been quantitatively evaluated (e.g., Liepelt et al., 2009; Alba-Sánchez et al., 2010), and (ii) it has experienced drought-induced decline in its western distribution limit in the Spanish Pyrenees since the late 1980s (Camarero et al., 2011). The current geographical distribution of silver fir, which dominates wide forested areas in central Europe, is the result of postglacial recolonization, which started about 11000 years ago from glacial refugia located in southern Europe such as the Pyrenees (Liepelt et al., 2009). However, while its distribution is quite continuous in the center of the range, this species shows a fragmented distribution patterns at the periphery (Gömöry et al., 2012). This is the case of the Pyrenean species’ rear edge (Camarero et al., 2004).

The reduced population sizes and the habitat fragmentation of Pyrenean silver fir populations may result in reduced genetic variation, increased levels of inbreeding within populations, and also elevation genetic differentiation among populations (Young et al., 1996; Leimu et al., 2010). In fact, previous studies have shown a lower genetic variability for silver fir in the Pyrenees compared to other European fir populations using either isozymes (Konnert & Bergmann, 1995) or chloroplast microsatellites (Vendramin et al., 1999). The performance of these populations may also be compromised by high levels of genetic drift (Bacles & Jump, 2010), leading to increased homozygosity, which may negatively affect individual fitness (Leimu et al., 2006). As a long-term consequence, these processes could induce a reduction in the adaptive potential of these fragmented populations reducing their resilience ability to cope with changing climatic conditions (Pertoldi et al., 2007).

Rear-edge populations respond more to drought stress than less marginal populations in terms of radial growth (Carrer et al., 2010), being the former more prone to extirpation due to stochastic events than the latter (Lesica & Allendorf, 1995). Thus, rear-edge populations may be very susceptible to rising temperatures and associated aridification trends, which may trigger drought-induced forest decline (Macias et al., 2006). Moreover, selective logging has been proposed as an additional contributing factor of decline through the negative selection of slow-growing trees (Camarero et al., 2011), and the opening of abundant gaps thus reducing air humidity which is a main ecophysiological constrain of silver fir (Peguero-Pina et al., 2007).

Thereby, the genetic status in such peripheral rear-edge silver fir populations is a decisive aspect for assessing their adaptive ability and resilience potential against increasing aridification trends. Thus, the genetic structure of Pyrenean silver fir populations may be hypothesized to be a risk factor for predisposing them to drought-induced decline. On that account we characterize the genetic structure of ten silver fir populations located in the Spanish Pyrenees as compared with two Central European populations using nuclear microsatellite markers. We analyze climatic variables and long-term climatic trends and we characterize forest structure in declining and non-declining Pyrenean sites. Finally, a comprehensive assessment of climatic trends, silver-fir decline patterns and genetic data is made by relating them, concluding with a consideration of potential risk of these interactions on local extinction processes of rear-edge tree populations.

Rear edge tree populations inhabiting the southernmots (or lowermost) margins of the species distribution range may act as an important reserve of genetic diversity in changing climates because they have often been long-isolated and lack gene flow, display strong differentiation, and may exhibit local adaptation in response to strong selective pressures such as drought stress (Hampe & Petit, 2005; Parisod & Joost, 2010; Gugger et al., 2011). The studied silver fir populations in the Spanish Pyrenees can be considered as the western “stable” rear edge for the species. Hampe & Petit (2005) defined “stable” rear edges as situations where at least some populations have persisted in situ at suitable growing sites across Quaternary climatic oscillations, while the species expanded its range into other regions. This is the case of silver fir in the Spanish Pyrenees (Alba-Sánchez et al., 2010). Moreover, the studied populations are isolated and small sized, which also matches the definition of a “stable” edge situation. This, in turn, can be a reason for the comparatively small genetic diversity found within the studied A. alba populations in comparison to German ones (Table 3). Bulgarian silver fir populations also show higher diversity than Spanish and German ones with diversity values of up to N _e = 8.7 and A = 18.2 (data in preparation). A low genetic variation for rear edge populations as a consequence of isolation and the small size has been previously reported in other plant species (Chang et al., 2004).

On the other hand, high levels of genetic differentiation could be found among the ten analyzed fir populations in the Pyrenees (Table 3). The Spanish silver fir populations show a high genetic differentiation in relation to the maximum geographical distance (ca. 100 km). This might be caused by long-term isolation, fragmentation and reduced gene flow among the silver fir populations. Also as local selection processes as discussed in Comps et al., (2001) and Petit et al., (2003) regarding stable rear edges and glacial refugia might have been played a role. Similar to the fir populations in the Spanish Pyrenees, Gugger et al., (2011) detected a high differentiation among Douglas-fir rear-edge populations in Mexico. Specifically, in our study area, the southernmost Guara population can be considered as the most geographically isolated (Fig. 1). Although the genetic diversity parameters for the silver firs sampled in Guara are comparable to the values of the most eastern populations here studied, inbreeding may be assumed, since there the fixation index F was high (0.127) and deviated from Hardy-Weinberg-equilibrium. An even higher F-value (0.25) was detected in the threatened relict fir species A. ziyuanensis in China (Tang et al., 2008). In contrast, a lower level of genetic differentiation was found in the German silver fir populations, i.e. in the core of the distribution range, which also presented similar geographical distances separating them (Cremer, 2009).

In this study we have evidenced that the genetic differentiation among the studied populations in the Spanish Pyrenees is mainly established by an East-West gradient (Fig. 4b). The most eastern populations (Collubert and Benasque) show a higher genetic diversity along this longitudinal gradient, whereas the western populations (Paco Ezpela, San Juan de la Peña and Oroel) are characterized by the lower genetic diversity values among the studied Spanish populations (Fig. 5).

Peripheral rear-edge populations can occupy habitats more stressful from those faced by populations located near the core of the distribution range (Lesica & Allendorf, 1995). In fact, Peguero-Pina et al., (2007) showed that silver fir stands in the southern Pyrenees occupies more xeric habitats than those reported for this species in Central Europe. Our results agree with these assumptions since summer precipitation and sMAI are comparatively low in the Spanish Pyrenees than in Germany. Specifically, the three populations living under the more xeric conditions are located in the western study area and show high defoliation levels (Paco Ezpela, San Juan de la Peña and Paco Mayor) (Fig. 5). In this western group the Gamueta populations does not show defoliation, probably due to the more mesic climatic conditions existing in this site when compared to the declining populations. For this reason, this population may be more prone to experience decline phenomena in the future under the current context of rising temperatures and summer precipitation decrease. The apparent importance of summer aridity, expressed by low sMAI values, on the performance of these western populations implies that a shift in summer temperatures could threaten the survival of these stands, even in the absence of any modification in the precipitation regime. On the other hand, although the climatic conditions experiencing by the central and eastern populations in this study are more similar to those existing in Germany, the Guara site shows climatic variables regarding aridity (sMAI and Gaussen index) more similar to those found in the western sites. The absence of defoliation in this southernmost population and a similar genetic diversity with the most eastern populations indicates the possibility that silver fir in Guara could be well adapted to the sub-Mediterranean conditions of this site. Additional studies should be done to assess this possibility.

An evident warming trend has been detected since the 1940s in western European mountains, showing that the latest decades were much warmer than any other period of the instrumental records (Díaz & Bradley, 1997). In the Central Pyrenees, mean temperature increased by +0.8 ºC between 1882 and 1970 at a high-elevation site (Bücher & Dessens, 1991). A progressive increase in the length and intensity of the summer aridity period has been detected during the last decades in the studied sites, especially in the westernmost Pyrenean populations (Camarero et al., 2011). This warming trend during the late 20th century involves increasing aridity and should affect the long-term performance of these rear-edge silver fir stands (Macias et al., 2006). If these predictions are confirmed, the present rear edge for A. alba in the Spanish Pyrenees will have to be redrawn in a near future. For this reason, conservation of declining Pyrenean silver fir populations may be an urgent need for maintaining their ecosystem services including their regional genetic diversity. Western Pyrenean sites subjected to dryer conditions in summer, high defoliations levels and low genetic diversity are prone to drought-induced decline and mortality. Since these rear-edge populations are small, isolated and genetically impoverished, regional population dynamics cannot easily compensate local extinction events. Based on these considerations, a conservation of the regional genetic diversity, i.e. of the whole gene pool, of Pyrenean silver fir populations must be associated with the preservation of these single populations.