The taxonomic effect on the dispersal-range size relationship
We found that studies that included species that can only be grouped into a high taxonomic level such as ‘phylum’ or ‘division’ found less often positive dispersal-range size relationships than studies that examined species within lower taxonomic levels (‘family’, ’order’, ’class’). Our separate analyses for plants and animals show that this effect is primarily driven by studies on plants, in which there was a clear distinction between e.g. ‘angiosperm-wide’ studies (126/159 relationships, in contrast to 10/298 relationships in animals that was at ‘phylum’ or higher taxonomic level), versus those performed for a specific ‘family’ or ‘genus’. This suggests that the disparity of dispersal-range size relationships when including plant (or animal) lineages that are phylogenetically distantly related and may therefore differ to a large extent in the traits that capture their dispersal ability (e.g., small seeds for wind-dispersed taxa, large seeds for animal-dispersed taxa, Table 2), will obscure or erase a relationship between dispersal and range size. In addition, including a clade in which many species are missing may also simply bias the initial dataset towards e.g. well-sampled species with relatively large range sizes. This means that a clear hypothesis and expectation on how dispersal affects range size in a studied clade is essential. In addition, including phylogenies to control for clade differences could provide a solution. We found that only 24% and 44% of dispersal-range size relationships that have been performed on animals and on plants, respectively, explicitly considered phylogenetic relationships. Even though phylogenetic correction was not selected in our final statistical model, we should be aware that we may not have had the statistical power to detect its real effect.