Distance-based eigenvector maps (DBEM) to analyse metapopulation structure with irregular sampling

Characterizing spatial patterns due to ecological processes is a major issue for analysing and predicting species distributions. Grasping the non-linear nature of population dynamics over networks of discrete suitable sites is here central, as very specific signatures are expected. In the line of promising results from Fourier analysis of metapopulation maps, we found distance-based eigenvector maps (DBEM) to help disentangle the respective signatures of habitat and metapopulation structuring, with the great advantage of being applicable to irregular sampling schemes, a common feature of ecological surveys. A smoothing procedure was required to obtain the distinguishable signatures, and this may be a critical issue for investigating non-contingent and reliable patterns in spatial ecology.