Many studies in ecology and conservation investigate gene flow among populations and aim to determine if genetic structure is mainly shaped by isolation by distance, or by current or historic landscape patterns (e.g. Keller & Largiader 2003; Epps et al. 2006; Arens et al. 2007). Considering environmental/ landscape variations on population or individual allele variations is a powerful tool to provide new insights into selection processes in natural populations. Landscape genetics helps us to integrate the effect of landscape connectivity into gene flow analysis and further allows us to better understand local adaptation processes by helping to come up with new hypothesis on potential selection pressures (Manel et al. 2003). State of the art and recent advances in statistic and molecular tools that allow us to address such questions were discussed at the workshop on ‘Landscape Genetics’ (University of Grenoble 20–23 October 2008). The main objective of this workshop, sponsored by the European Scientific Foundation (ConGen programme), was to explore how spatial and environmental data can be combined and implemented in relevant questions related to conservation. We met this objective by bringing together both leading scientists in the field of spatial and landscape genetics and young researchers, who presented various case studies and new methodological approaches. This workshop not only summarized the most recent developments but also tried to bridge the gap between theoretical and case approaches in conservation scenarios. Linking theoretical and practical approaches has the potential to gain even more importance in the future mainly because we expect the availability of genetic data sets to increase exponentially over the next years. The workshop focused on two main issues. First, we addressed the question to what extent landscape genetics could be used to assess the impact of landscape fragmentation on both genetic diversity and connectivity. In addition, we discussed how genome scans and landscape genetic could be used to assess the genetic basis of local adaptation.
The impact of landscape fragmentation on genetic connectivity was highlighted by Marie-Josée Fortin, who started the workshop by outlining global perspectives on quantifying spatial heterogeneity in ecological and genetic data (Jacquez et al. 2008). This ranges from quantifying either the degree of habitat fragmentation (e.g. delineating spatial homogeneous patches or groups) or the degree of connectivity (e.g. identifying potential paths or corridors between patches or individuals). Connectivity will probably play a major role in the future of landscape genetics (Cushman 2006; Cushman et al. 2006). This was highlighted through a case study presented by Sam Cushman, who showed how connectivity could be related to the gradient concept of population structure. The study illustrated how gene flow, selection and other evolutionary processes acted differentially across space as functions of local gradients of landscape resistance. Gradient-based approaches can thus provide an alternative to classic models of population genetics which assume that panmictic populations exist in a homogeneous environment. The gradient concept should help to disentangle whether observed spatial genetic patterns are (i) due to isolation by distance, or (ii) shaped by additional structure due to current or past landscape variables, or (iii) the result of demographic history. Olivier Hardy introduced a new approach based on interlocus correlations, which aims distinguishing between these different scenarios by assessing whether a given model (e.g. isolation by distance, demographic history of populations) is sufficient to account for the observed genetic structure. More realistic concepts and powerful statistical approaches for inferring spatial genetic structure and for identifying genetic entities are still vibrant and ongoing (e.g. Hardy et al. 2006; Manel et al. 2007; Corander et al. 2008; Philibert et al. 2008). We expect that several new and updated software packages that implement such novel statistical approaches will become available in the close future. For example, Olivier François introduced an approach for combining individual genetic data, spatial information and covariates (e.g. environmental variables) which will be implemented in a shortly available version of TESS (Francois et al. 2006). Thibaut Jombart further presented a spatial principal component analysis implemented in an R library (Adegenet) (Jombart et al. 2008), which allows to deal with large data sets and thus provides a promising analysis tool, especially for future applications, when enormous data sets will be available. Lisette Waits presented a case study (North American amphibians) and pointed out that the vast amount of available landscape genetic methods often makes it difficult for users to choose an appropriate approach. This demonstrates the urgent need for a comprehensive evaluation of methodological approaches based on realistic spatial simulations. One example of simulation evaluation focusing on boundary detection methods was presented by Toni Safner. Simulations were also presented by Bryan Epperson to illustrate the importance of stochastic processes in determining geographical patterns of genetic variation (Epperson 2003).
Landscape genetics and genome scans also provide a promising tool to detect candidate genes that are potentially under selection. Stéphane Joost described how correlations between allele frequencies and environmental/landscape variables could help us to come up with assumptions on selective pressures (Joost et al. 2007). However, Rolf Holderegger pointed out that at the moment, we are still far from understanding adaptive genetic diversity in real landscapes (Holderegger et al. 2008). Even when markers were demonstrated to be adaptive, only few studies have investigated the adaptive relevance of the identified marker sets. There is thus an urgent need for further experiments and simulation studies. Progress in the field of adaptive landscape genetics is closely related to the knowledge of the studied genome.
Pierre Taberlet gave an overview of the current development in the field of sequencing technologies and stressed that the availability of whole genome information in the near future is likely to change the research field dramatically. This next generation sequencing technologies (Coombs 2008) will strongly influence the questions that ecologists will be able to address. So far in studies of local adaptation, for example, massive parallel pyrosequencing (Margulies et al. 2006) has the potential to produce large genome scans traditionally carried out by AFLPs. This highly innovative approach will boost studies on landscape genetics, by allowing to quickly spot the chromosomal regions (and even the single nucleotide polymorphisms) that are associated with environmental parameters. Even though such techniques will readily provide us with huge amounts of data, there will still be a bottleneck in adequately processing and interpreting these data. A proper sampling design based on relevant biological questions will become even more important for future studies that incorporate these techniques. Despite all technical advances, it will be essential to integrate knowledge from different disciplines like spatial statistics, population genetics and landscape ecology to unravel the underlying processes of how landscape and environmental patterns affect genetic variation.