<b>Staying for food by urban birds: </b><b>insights from neural network analysis into adaptive strategies</b>

The black-headed gull, belonging to the order Lariformes, family Laridae, is a bird species found worldwide. It breeds in the Palearctic region and migrates south to winter in India, Southeast Asia, and the Philippines (MacKinnon 2022). It is widely distributed in all provinces and regions of China and is one of the most common birds in inland and coastal areas (Zheng 2023). The environmental adaptations of black-headed gulls play a crucial role in their widespread geographic distribution, whereas their specific distribution areas are determined by human activities and food availability (Liang et al. 2019). Many black-headed gulls heavily rely on human food sources and are slowly developing alternative foraging techniques distinct from their natural behaviors. Most individuals forage in streets and squares, and actively approach human feeders (Feng and Liang 2020). In November 2023, we collected FID data on black-headed gulls in Kunming, Yunnan, southwest China (24°23′–26°22′ N, 102°10′–103°40′ E). Three gull-watching scenic areas within the city were selected as large regional sample sites: Haigeng Dam, Daguan Park, and Huanhu Road (Figure 1). Haigeng Dam and Daguan Park are located in the city and Huanhu Road is located in the countryside. In these areas, two types of spots were selected: tourist hotspots (for popular photo spot), which are frequently visited by urban residents and international tourists, and where shops selling food specifically for feeding the black-headed gulls are available; and non-tourist hotspots (not for popular photo spot) are located within the same scenic area but away from crowds and with few feeders or sightseers. In this study, field surveys of the flight initiation distance (FID) of black-headed gulls (Chroicocephalus ridibundus) were performed at three scenic spots in Kunming City, Yunnan, southwest China. Our results showed that, within the same area, FID of black-headed gulls was significantly shorter in areas with increased human activity. In areas with prolonged human contact, black-headed gulls showed higher adaptability and shorter FID. The FID data were further analyzed by a multilayer perceptron regression model (MLPRegressor) with a neural network approach to delineate FID thresholds for black-headed gulls in different human disturbance areas. The analysis revealed that black-headed gulls exhibit a high degree of behavioral flexibility in urban areas, with food availability playing a key role in increasing the birds’ tolerance to humans. These findings highlight the significant impact of human behaviors, such as feeding, on wildlife behavior patterns. This further suggests that food availability is a crucial mechanism for the high adaptability of black-headed gulls to urban environments. Understanding this mechanism is essential for understanding the coexistence of humans and wildlife. The establishment of FID models for black-headed gulls will provide new perspectives and tools for animal behavior research.