Data from: Edge effects and geometric constraints: a landscape-level empirical test

Edge effects are pervasive in landscapes yet their causal mechanisms are
still poorly understood. Traditionally, edge effects have been attributed
to differences in habitat quality along the edge-interior gradient of
habitat patches, under the assumption that no edge effects would occur if
habitat quality was uniform. This assumption was questioned recently after
the recognition that geometric constraints tend to reduce population
abundances near the edges of habitat patches, the so-called “geometric
edge effect” (GEE). Here we present the first empirical, landscape-level
evaluation of the importance of the GEE in shaping abundance patterns in
fragmented landscapes. Using a dataset on the distribution of small
mammals across 18 forest fragments, we assessed whether the incorporation
of the GEE into the analysis changes the interpretation of edge effects
and the degree to which predictions based on the GEE match observed
responses. Quantitative predictions were generated for each fragment using
simulations that took into account home range, density and matrix use for
each species. The incorporation of the GEE into the analysis changed
substantially the interpretation of overall observed edge responses at the
landscape scale. Observed abundances alone would lead to the conclusion
that the small mammals as a group have no consistent preference for forest
edges or interiors, and that the black-eared opossum Didelphis aurita (a
numerically dominant species in the community) has on average a preference
for forest interiors. In contrast, incorporation of the GEE suggested that
the small mammal community as a whole has a preference for forest edges,
whereas D. aurita has no preference for forest edges or interiors.
Unexplained variance in edge responses was reduced by the incorporation of
GEE, but remained large, varying greatly on a fragment-by-fragment basis.
This study demonstrates how to model and incorporate the GEE in analyses
of edge effects, and that this incorporation is necessary to properly
interpret edge effects in landscapes. It also suggests that geometric
constraints alone are unlikely to explain the variability in edge
responses of a same species among different areas, highlighting the need
to incorporate other ecological factors into explanatory models of edge
effects.