Résumé
Community genetics aims at understanding how within-species variation, species diversity and environmental factors interact to shape community assembly. An approach that emerged a few years ago has been to quantify the correlation between the neutral genetic diversity of a focal species and species diversity of the surrounding community (species–genetic diversity correlations, or SGDCs). We here review this approach and discuss its interpretative framework in a community ecology context. First, we show that the case for mostly positive SGDCs is probably overstated due to publication bias – only 11% are significantly positive, a fraction comparable to the significantly negative ones. This suggests that variation in area and connectivity among habitat patches, theoretically leading to positive SGDCs, is not the only factor affecting SGDCs. Second, building upon previous contributions, we propose a general framework to identify the multiple factors underpinning SGDCs, and argue that it will help deepen our understanding of community assembly, especially with regard to the ecological factors playing at metacommunity scale. Our framework distinguishes between site and community factors which can affect SGDCs either positively or negatively, depending on whether the focal species and the rest of the community are similar or dissimilar, in terms of realized niches and dispersal abilities. Empirical studies should thus go beyond simply computing SGDCs, and we provide statistical methods (e.g. structural equation modelling) to decompose SGDCs into the multiple contributions of site and community factors. As an example, we use a published dataset (freshwater snail metacommunity), and show how the role of focal population size on SGDCs had hitherto not been detected. We further discuss how considering several focal species and various delimitations of the community may help one to identify clusters of ecologically similar species. We eventually highlight the benefit that SGDC studies would get from integrating beta-diversities.