Validity of covariance models for the analysis of geographical variation

Niche modelling and landscape genetics of the yellow-legged hornet (Vespa velutina): An integrative approach for evaluating central-marginal population dynamics in Europe

Genetic diversity is an important biological trait for a successful invasion. During the expansion across a new territory, an invasive species may face unprecedented ecological conditions that will determine its demography and genetic diversity. The first record of the yellow-legged hornet (Vespa velutina) in Europe dates back to 2004 in France, from where it has successfully spread through a large territory in the continent, including Italy, Spain and Portugal. Integrative approaches offer a powerful strategy to detect and understand patterns of genetic variation in central and marginal populations. Here, we have analysed the relationship between genetic diversity parameters inferred from 15 V. velutina nuclear DNA microsatellite loci, and geographical and environmental drivers, such as the distance to the introduction focus, environmental suitability and distance to native and invasive niche centroids. Our results revealed a central-marginal dynamic, where allelic richness decreased towards the edge of the expansion range. The low environmental suitability of the territories invaded by marginal populations could prevent a diverse population from establishing and reducing the genetic diversity in populations at the expansion edge. Moreover, Markov chain Monte Carlo analysis showed both geographical and environmental distances were influencing population genetic differentiation. This study highlights the importance of combining genetic analysis with geographical and environmental drivers to understand genetic trends of invasive species to new environment.

Evaluating insect-host interactions as a driver of species divergence in palm flower weevils

Plants and their specialized flower visitors provide valuable insights into the evolutionary consequences of species interactions. In particular, antagonistic interactions between insects and plants have often been invoked as a major driver of diversification. Here we use a tropical community of palms and their specialized insect flower visitors to test whether antagonisms lead to higher population divergence. Interactions between palms and the insects visiting their flowers range from brood pollination to florivory and commensalism, with the latter being species that feed on decaying-and presumably undefended-plant tissues. We test the role of insect-host interactions in the early stages of diversification of nine species of beetles sharing host plants and geographical ranges by first delimiting cryptic species and then using models of genetic isolation by environment. The degree to which insect populations are structured by the genetic divergence of plant populations varies. A hierarchical model reveals that this variation is largely uncorrelated with the kind of interaction, showing that antagonistic interactions are not associated with higher genetic differentiation. Other aspects of host use that affect plant-associated insects regardless of the outcomes of their interactions, such as sensory biases, are likely more general drivers of insect population divergence.

Model selection with multiple regression on distance matrices leads to incorrect inferences

In landscape genetics, model selection procedures based on Information Theoretic and Bayesian principles have been used with multiple regression on distance matrices (MRM) to test the relationship between multiple vectors of pairwise genetic, geographic, and environmental distance. Using Monte Carlo simulations, we examined the ability of model selection criteria based on Akaike's information criterion (AIC), its small-sample correction (AICc), and the Bayesian information criterion (BIC) to reliably rank candidate models when applied with MRM while varying the sample size. The results showed a serious problem: all three criteria exhibit a systematic bias toward selecting unnecessarily complex models containing spurious random variables and erroneously suggest a high level of support for the incorrectly ranked best model. These problems effectively increased with increasing sample size. The failure of AIC, AICc, and BIC was likely driven by the inflated sample size and different sum-of-squares partitioned by MRM, and the resulting effect on delta values. Based on these findings, we strongly discourage the continued application of AIC, AICc, and BIC for model selection with MRM.