Data from: Does movement behaviour predict population densities? a test with 25 butterfly species

Diffusion, which approximates a correlated random walk, has been used by
ecologists to describe movement, and forms the basis for many theoretical
models. However, it is often criticized as too simple a model to describe
animal movement in real populations. We test a key prediction of diffusion
models, namely, that animals should be more abundant in land cover classes
through which they move more slowly. This relationship between density and
diffusion has rarely been tested across multiple species within a given
landscape. We estimated diffusion rates and corresponding densities of 25
Israeli butterfly species from flight path data and visual surveys. The
data were collected across 19 sites in heterogeneous landscapes with four
land cover classes: semi-natural habitat, olive groves, wheat fields and
field margins. As expected from theory, species tended to have higher
densities in land cover classes through which they moved more slowly and
lower densities in land cover classes through which they moved more
quickly. Two components of movement (move length and turning angle) were
not associated with density, nor was expected net squared displacement.
Move time, however, was associated with density, and animals spent more
time per move step in areas with higher density. The broad association we
document between movement behaviour and density suggests that diffusion is
a good first approximation of movement in butterflies. Moreover, our
analyses demonstrate that dispersal is not a species-invariant trait, but
rather one that depends on landscape context. Thus, land cover classes
with high diffusion rates are likely to have low densities and be
effective conduits for movement.