Warm Marches lowered late-March feeder reports of departing winter-visitor birds in eastern North America in 2002-2024, but the pre-registered all-species test was null

Question: do short-distance migrant birds that winter at feeders leave earlier after a warm March? We used Project FeederWatch checklists (seasons 2002-2024), NOAA GHCN_CAMS March temperature, and Breeding Bird Survey data to remove breeding areas. Retention is the reporting rate of a winter-visitor species in 15-31 March divided by its rate in mid-winter at the same sites, minus the same value for resident feeder birds in the same 2-degree cell and season. The analysis plan was written before the models. Over all 356 species-cells, the March slope did not differ clearly from zero in any of four samples (East and West, 2002-2012 and 2013-2024); the pre-registered test failed. Most species-cells (for example Dark-eyed Junco and White-throated Sparrow) do not decrease before 31 March, so the common window could not detect their departure. In an exploratory subset of species-cells that decrease during March, a 1 C warmer March lowered retention in the East by 4.1% (95% interval 1.8 to 7.4) in 2002-2012 and 3.1% (1.1 to 5.0) in 2013-2024, about 3.5 days earlier per 1 C (2.3 to 5.1). Next- and previous-season placebos, a snow control, and removal of insectivores did not remove the East result. West estimates were negative but uncertain. Retention had no clear 2002-2024 trend, and March did not warm clearly at the sites. Limitations: the subset was chosen after the main test; lower reports can mean less feeder use, not departure; species that leave in April were not testable.

spring departure · migration timing · Project FeederWatch · winter birds · phenology · temperature · citizen science · observation bias · pre-registration · null result · Breeding Bird Survey · GHCN_CAMS