Population density modelling of mixed polymorphic phenotypes: an application of spatial mark-resight models

Melanism is a form of pigmentation polymorphism where individuals have darker colouration than what is considered the 'wild' phenotype. In the case of leopards, Panthera pardus, melanism occurs at higher frequencies amongst populations in tropical and subtropical moist forests of south and southeast Asia, presenting a unique challenge in estimating and monitoring these populations. Unlike the wild phenotype that is readily recognizable by its rosette patterns, melanism results in individuals being unidentifiable or 'unmarked' through photographic captures obtained using white flash cameras. Spatial mark-resight (SMR) models that require only a subset of the population to be 'marked' offer the opportunity to estimate population density. In this study, we present an application of SMR models to estimate leopard densities using camera trap survey data from three sampling years at Manas National Park (MNP), India. By using an SMR model that allowed us to include captures of unidentified sightings of marked individuals, we were also able to incorporate captures where identity was either not confirmed or only known from a single flank. Following 18 674 trap-days of sampling across three years, we obtained 728 leopard photo-captures, of which 22.6% (165) were melanistic. We estimated leopard densities of 4.33, 2.61 and 3.37 individuals/100 km(2) across the 3 years. To our best knowledge, these represent the first known estimates of leopard densities from populations comprising both melanistic and wild phenotypes. Finally, we highlight that SMR models present an opportunity to revisit past camera trap survey data for leopards and other species such as Jaguars, P. onca, that exhibit phenotypic polymorphism towards generating valuable information on populations.

Automated summary

This study applied SMR models to estimate leopard densities at Manas National Park in India, with a high proportion of melanistic leopards captured. The research highlights the potential of SMR models for revisiting past camera trap survey data and generating valuable information on populations.