Brood parasitism, the reproductive strategy where an individual relies on others to feed its young, is prevalent among bird species and determines the reproductive output of both the host and the parasite. Despite this importance, broad-scale studies on brood parasitism among birds remain scarce due to the lack of data. We first present a comprehensive dataset of brood parasitism incidences by Brown-headed Cowbird (Molothrus ater), a common brood parasite in North America, from the citizen science platform eBird. By combining large language models (LLMs) with manual screening, we extracted 27,316 incidences of brood parasitism from >9 million eBird observation comments across the breeding range of the Brown-headed Cowbird (Canada, United States, and Mexico). Combined with the date and GPS coordinates associated with each eBird observation, this dataset offers unique opportunities to analyze the broad-scale spatial and temporal patterns of brood parasitism. With this dataset, we are now analyzing the spatial and temporal changes in brood parasitism phenology compared to the phenological shifts of hosts. This dataset and subsequent analysis is uniquely enabled by our data collection pipeline using LLM.
This is an ongoing collaboration with Thabassum Hashmi Hajamaideen (undergraduate research assistant, University of Michigan), Robert Ernst (Georgia Institute of Technology), Benjamin Freeman (Georgia Institute of Technology), Kai Zhu (University of Michigan), and Brian Weeks (University of Michigan). We currently have several projects in progress, including a data paper that describes the dataset. Stay tuned!