Demographics of free-roaming dogs in Guwahati, India: Assessing population characteristics and comparing population...

India has the highest global burden of rabies, accounting for approximately 35 % of all human rabies deaths worldwide with free-roaming dogs (FRDs) playing a critical role in the spread of the disease. While several studies have assessed FRD demographics in India, none have focused on Northeast India, particularly Assam. This study 1) characterizes FRD demographics in two urban settings in Guwahati, Assam, 2) estimates FRD populations using Program MARK's Huggin's heterogeneity models and the Application SuperDuplicates (AS), and 3) evaluates AS as a resource effective alternative to Huggin's heterogeneity models for population estimation. Photographic capture-recapture surveys were conducted in two distinct locations in Guwahati: a market area without defined boundaries and a gated university campus. We detected 183 and 141 unique FRDs at the market and campus sites, respectively. The FRD in the market area were significantly younger, exhibited lower sterilization rates, and tended to be in poorer physical condition compared to those in the campus area; differences in body condition and collar presence were not statistically significant. Population estimates using Huggin's heterogeneity models revealed a FRD population size of 252 individuals (using the Chao estimator in the Huggin's model) at the market site, while campus estimates varied between 161 during morning transects and 153 during evening transects (using the Jackknife estimator) and with morning and evening combined (using the Chao estimator) of 164 FRD. The AS estimated 246 FRD in the market and 178 on campus. Since the AS only requires data from two transect surveys, all possible survey combinations were built to estimate the FRD population size via AS and over 70 % similarity was observed between all these AS estimates compared to Huggin's heterogeneity model results. Within the same urban setting, variations free-roaming dog population characteristics and estimates across study sites were likely based on differences in local environmental and socio-spatial factors. Application SuperDuplicates showed to be a reliable, efficient alternative to Huggin's heterogeneity models for estimating FRD populations, with compatibility above the 70 % threshold whilst requiring information from two surveys only.