Six (arguably) necessary steps forwards in landscape connectivity and genetics

摘要

Landscape heterogeneity and fragmentation affect how organisms are distributed in the landscape, determine the chance of a patch being colonized, reduce inbreeding in small populations and maintain evolutionary potential. Predicting the way in which animals disperse is pivotal for management and conservation purposes. I discuss here the conceptual and methodological weak points of circuit theory and least-cost modelling, the two most commonly-used methods in the scientific literature. I argue that these two methods, although very brilliant and very well supported by freely-available softwares, make use of six axiomatic assumptions: 1) any landscape can be divided into source and sink areas for any considered species; 2) source-sink areas can be a priori defined by the users; 3) any species adopt a global optimization of its dispersal over any landscape; 4) biotic movements are undirected; 5) stability points along dispersal paths are absent; 6) frictional values based on expert opinion are true-to-life. I argue that these axioms are only realistic for a limited number of species with short-range shifts over lowland (or, at least, patchy) landscapes, and for which frictional values can be realistically defined. I also describe an alternative theoretical and methodological approach, called Flow Connectivity, which can fix such weak points.