Measuring Indian giant squirrel (Ratufa indica) abundance in southern India using distance sampling

Abstract

A large body of work on the ecology of sciurids is based on comparing patterns of abundance across either space or time. However, in most cases investigators choose to use surrogate measures of abundance, such as indices based on species or sign encounter rates, or trapping rates. This requires the assumption that detection probabilities are equal at all sites (or time periods) sampled, an assumption that is difficult to meet under field conditions. We demonstrate the application of line transect-based distance sampling, a technique that explicitly models and accounts for detection probability, to estimate ecological densities of Indian giant squirrels in forested habitats. Line transect surveys were carried out at several sites and the number of detections included: 86 (Bandipur), 152 (Nalkeri), 110 (Sunkadakatte), 304 (Muthodi) and 236 (Lakkavalli). The encounter rates ranged from 0.179/km in Bandipur through 0.296/km (Nalkeri), 0.368/km (Sunkadakatte), and 0.625/km (Lakkavalli), to 0.779/km in Muthodi, while the estimated probabilities of detection were 0.517 (Bandipur), 0.532 (Nalkeri), 0.531 (Sunkadakatte) 0.548 (Lakkavalli) and 0.604 (Muthodi). The estimated mean squirrel densities (± standard error of the density) ranged from 2.37 (0.33) squirrels/km2 in Bandipur through 4.55 (0.44) squirrels/km² in Nalkeri, 4.86 (0.62) squirrels/km² in Sunkadakatte, to 10.20 (0.82) squirrels/km² and 12.26 (1.10) squirrels/km² in Muthodi and Lakkavalli respectively. We discuss design, field survey and data analytic considerations for rigorously estimating squirrel density and abundance.