Vicia pisiformis L. is a perennial leguminous plant with a main distribution in broadleaved forest-steppes of eastern Europe. The species is classified as endangered (EN) according to the IUCN red-lists in both Norway and Sweden, due to severe fragmentation, small population sizes and continuing population decline. The populations on the Scandinavian Peninsula constitute the northern limit of the species distribution and are mostly restricted to warm stony slopes with predominantly southern aspects. In this study we used the AFLP method, which is a high-resolution genetic fingerprint method. Samples were collected from 22 Scandinavian populations. The overall genetic structure was analysed in an AMOVA, in a Mantel test and through constrained correspondence analysis (CCA). The ordination scores representing non-geographic genetic divergence were extracted from the CCA and analysed in a linear model using habitat variables and population size as explanatory variables. We found (i) a strong geographic structure, (ii) significant genetic divergence between populations, (iii) that this genetic divergence remained significant even after removing the effect of geography in a partial CCA and (iv) that the remaining non-geographic part of genetic divergence (distance from the ordination centre) was associated with aspect, populations with a northern aspect were more genetically divergent. Aspect explains more variation than population size and is the only variable retained in the minimal adequate model. We suggest that local adaptation has caused this divergence from an expected geographical pattern of genetic variation. This explanation is further supported by the association between aspect and specific AFLP fragments. Many plant populations are relics of a different climate (Aguirre-Planter et al. 2000; Despres et al. 2002; Pico and Riba 2002). In response to long-term climate change, populations can either migrate towards a more favourable climate or adapt to the new conditions (delaVega 1996; Jump et al. 2006). Species with limited dispersal ability are at risk of reaching isolated dead-ends of decreasingly suitable habitat, without any suitable habitat within dispersal distance (Colas et al. 1997). Isolated populations have to use their inherent evolutionary potential and adapt to changes in environmental conditions, or they will go extinct. As population fragments go extinct, those that remain will become increasingly isolated from each other both spatially and also genetically as the level of gene flow declines with increasing distance. Such correlation between genetic dissimilarities and geographic distances, known as isolation by distance (Slatkin 1993; Wright 1943), when found, suggests a history of geographically limited gene flow (Kimura and Weiss 1964). On top of an isolation by distance pattern there might be other genetic structures to be found. Occasional long-distance dispersal events for example may disturb geographic patterns with puzzling allele distributions as a result (Nichols and Hewitt 1994). Genetic drift is a process that will affect any pattern of genetic variation in a random fashion. Local adaptation through natural selection is a process that, if sufficiently strong in comparison with gene flow and genetic drift, will create patterns where genetic differentiation is associated with certain environmental conditions (Wright 1951). Several studies have shown the importance of local adaptation of populations (reviewed by Kawecki and Ebert 2004) (see also Bonin et al. 2006; Knight and Miller 2004; Kolseth and Lönn 2005; Lönn et al. 1998). Local adaptation can be strong also at small spatial scales (Snaydon and Davies 1976; Lönn 1993) even though it is sometimes very limited in terms of the number of genes involved (Kärkkainen et al. 2004) Environmental variability provides a base for biological variation by imposing differentiated selection pressures resulting in local adaptation. Topography provides large environmental variation within a relatively small area and thereby provides a basis for small-scale local adaptations. Depending on the local topographic possibilities populations can either migrate up and down slopes or along the same altitude to a different aspect to find a suitable microclimate. The dispersal distance will be much shorter per degree of temperature change during altitudinal migration (Hewitt 1996), than during simple latitudinal migration across a flat landscape. Slope and aspect are two important topographic parameters that determine the influx level of solar radiation, especially towards the poles where the total global radiation decreases (Larcher 2003). Vicia pisiformis is an endangered poorly-dispersed long-lived forest herb with its main distribution across the semi-open broadleaved forest steppes of eastern Europe. The Scandinavian populations are believed to be climate relict populations from warmer times. Earlier genetic studies of V. pisiformis using allozymes, RAPD:s and morhology, have found low to very low levels of genetic variation (Gustafsson and Gustafsson 1994; Black-Samuelsson et al. 1997; Black-Samuelsson and Lascoux 1999). Therefore we used AFLP (amplified fragment length polymorphism) markers, which detect even very small genetic differences between individuals. AFLP mainly analyse neutral variation, as the major fraction of most genomes is assumed to be neutral. However, since the AFLP-fragments are distributed randomly throughout the whole genome some fragments may be situated so close to regions under selection that they become more or less linked to them. This linkage disequilibrium between molecular markers and regions under selection, often referred to as quantitative trait loci (QTL), forms the basis for both QTL-mapping and marker assisted selection (MAS), reviewed by Dekkers and Hospital (2002). Gardner and Latta (2006) for example, found QTL under selection in both natural environments and in the greenhouse. Markers have been found to be connected to biomass production (Cavagnaro et al. 2006) and environmental variation (Bonin et al. 2006; Jump et al. 2006; Porcher et al. 2006). In this study we examine 22 Swedish and Norwegian populations of Vicia pisiformis and ask (i) if there is genetic differentiation between these populations, (ii) if there is can it be explained in its entirety by geographic location or (iii) can it partly be explained by habitat characteristics, suggesting local adaptation, or population size, suggesting genetic drift. We show that populations are differentiated geographically and that genetic variation in addition to the geographical pattern is associated with habitat.
2008. Vol. 145, no 1, 1-8 p.