Predictive Distribution Modelling of Critical Habitat Availability and
Prioritization for the Vultures in the Greater Panna Landscape, India
Abstract
Vultures are a specialized species group, utilizing wide habitat and
forage niches and their long-term survival depends on the protection of
their critical habitats. Taking a landscape approach, we modelled the
distribution of nest sites (n = 30) and roost sites (n = 31) of
cliff-nesting vultures (four species) in the Greater Panna Landscape
(GPL), central India. We performed Random Forest (RF), Generalized
Linear Model (GLM) and Boosted Regression Tree (BRT) algorithms. The AUC
values for the predictive distribution of nests were 0.97, 0.90, 0.97
for RF, GLM and BRT, respectively, while for roost distribution it was
found to be 0.76, 0.63, 0.74 for RF, GLM and BRT, respectively. We
ensembled the predictions of all three methods for better accuracy and
combined the model outputs. We then performed zonation analysis on the
final map and used Human footprint as a proxy for conservation cost to
define spatial prioritization for conservation inputs. The results
reveal that the GPL has a total of 9,402 sq. km. area within the top 20
ranks in terms of conservation prioritization for nesting and roosting.
Given the cost value, the top 20 ranked units will account for
approximately 60% of the critical habitats and these may be the focus
of long-term conservation inputs to sustain the vulture populations in
the landscape. The spatially explicit outputs based on the robust
methodology involving intensive fieldwork and ensembled modelling offer
a basis for local scale and landscape scale actions, which can be
replicated in other parts of the vulture distribution ranges.