Comparative assessment of avian diversity in Gilgit city using acoustic monitoring (BirdNET) and conventional point count surveys

Abstract

The process of urbanization tends to streamline bird communities, favouring generalist species. The present acoustic survey is the first of its kind in Gilgit City (Pakistan), aiming to describe the urban bird community using BirdNET (a deep-learning sound identification tool) alongside the parallel point count method. We deployed a portable recorder at 32 selected points across urban, suburban, peri-urban, and riverine environments over ten months (winter, spring, and summer). We processed the recordings using BirdNET, while point-count surveys were subjectively verified. BirdNET identified 47 species, while point counts recorded 45. Both approaches revealed a rich but biased assemblage: a few ubiquitous generalists dominated (e.g., House Sparrow, Streaked Laughing Thrush, Rose-ringed Parakeet), whereas many species were detected only once. Species diversity was moderate overall but declined from peri-urban areas (highest richness: ~31 species) to the urban core (highest richness: ~19 species). Richness peaked seasonally, with winter migrants (e.g., White Wagtail) increasing diversity, and post-breeding flocks causing surges in late summer. Our results corroborate existing evidence that BirdNET-based acoustic monitoring can effectively complement traditional field surveys, while demonstrating its applicability in the high-altitude urban and riverine landscapes of Gilgit City, particularly for detecting nocturnal or cryptic species (e.g., owls, night jars, etc.). These findings align with global evidence of urban avifaunal homogenization and highlight the role of peri-urban green spaces as key biodiversity reservoirs. The current effort is the preliminary comparative quantitative analysis of urban avifauna of Gilgit city and sets crucial data to be used in urban conservation planning.