Sensitivity of migratory connectivity estimates to spatial sampling design

Abstract Background The use of statistical methods to quantify the strength migratory connectivity is commonplace. However, little attention has been given their sensitivity spatial sampling designs and scales inference. Methods We examine sources bias imprecision in most widely used methodology, Mantel correlations, under a range plausible regimes using simulated populations. Results As correlations depend fundamentally on scale configuration sampling, unbiased inferences about population-scale can only be made certain regimes. Within contiguous population, samples drawn from smaller subsets generate lower metrics than as whole, even when underlying ecology population constant across population. Random individuals species ranges therefore underestimate connectivity. Where multiple discrete sites are used, by contrast, overestimation arise due being biased towards larger between-individual pairwise distances seasonal where occurs (typically breeding). Severity all biases was greater for populations with levels true When were applied realistic populations, accuracy measures maximised increasing number ensuring an spread full range. Conclusions These results suggest strong potential making quantitative statistics. Researchers wishing apply these should limit inference extent maximise sites, avoid drawing conclusions based small sample sizes.