Habitat selection
We characterized second- and third-order habitat
selection to identify habitat characteristics preferred by
the two-and the three-toed sloth (Johnson1980). Second order
habitat selection reflects the habitat features that
individuals use to select the location of their home range,
whereas third-order selection reflects the habitat features
that individuals preferentially use within their home
range. For this analysis, we only included individuals
that had a minimum of 16 relocations and that were
not dispersing from their home range. We estimated
home ranges using 90% fixed kernel methods for 34
adult resident two-toed sloths (median = 64 relocations;
range: 16–179) and 33 (n = 30 adults; n = 3 subadults)
three-toed sloths (median = 50 relocations; range = 20–
212). Home ranges were delineated using ESRI’s ArcGIS
10.1 and the Geospatial Modelling Environment with
least-squares cross-validation (LSCV) as the smoothing
parameter (Millspaugh & Marzluff 2001). To evaluate
second-order habitat selection, we considered used
habitat as the proportion of each habitat type in each sloth’s home range. We treated available habitat as the
proportion of each habitat type within the Minimum
Convex Polygon (MCP) of all sloths relocations of a given
species (i.e. pooled across individuals). Within the MCP,
we calculated available habitat in two different ways:
(1) as the proportion of all five habitats (cocoa, tropical
forest, pastures, monocultures and human development)
and (2) excluding the least-used habitats (monocultures
and human development). Monocultures and human
development habitats were excluded from the second
analysis as sloths are predominantly observed in cocoa,
tropical forest and cattle pastures, and rarely utilize
monocultures or human development (Vaughan et al.
2007). To test for third-order habitat selection, we
considered used habitat to be the proportion of relocations
that occurred in each habitat. We quantified available
habitat by generating random points within each sloth’s
home range (where the number of points was equal to
the number of observed relocations) and calculated the
proportion of locations occurring in each habitat type.
Weused Manly’sAlpha preference index to test for habitat
selection at both levels across all individuals (Manly et al.
2002). We inferred selection for a particular habitat type
if the observed index exceeded the expected value and did
not overlap in the 95% CI. Conversely, we inferred that
sloths avoided a habitat type if the observed index was less
than the expected values and did not overlap in the 95%
CI.