Wintering ducks, swans and geese in Japan did not move north with 1970–2026 winter warming; their January centroid moved 12–22 km per decade south in 1996–2026

Question: did wintering ducks, swans and geese (Anatidae) in Japan move north as winters became warmer? Data: the national mid-January waterfowl census of the Ministry of the Environment of Japan, 57 censuses from 1970 to 2026: prefecture x species totals for all years, and site counts for 1996–2026 with 5,045 sites counted in all 31 years. Temperature: December–January GHCN-CAMS grid, checked against 73 homogeneous JMA stations. Method: for each species and year we calculated the count-weighted mean latitude (centroid). We compared trends with the isotherm shift. Winter warming was 0.22–0.33 °C per decade and the spatial gradient was 1.10 °C per degree latitude, so isotherms moved north by 22–34 km per decade. We also measured the centroid response to warm winters (detrended), tested a temperature-only model fitted to 1970–1997 on 1998–2026, and split fixed sites by their artificial-feeding history. Results: the species-mean centroid did not move north. It moved −12 km per decade in 1970–2026 (22 species, 95% species-bootstrap interval −25 to −1), −22 km per decade in 1996–2026 from prefecture totals (23 species, −32 to −12; 15 species significantly south, 3 north), and −12 km per decade at the fixed sites (−21 to −3). In 19 robustness variants the upper limit was never above +5 km per decade. In warm winters the birds were further north, but only by 6–9 km per °C, less than 10% of the isotherm shift per °C. The 1970–1997 temperature model placed the 1998–2026 centroids 36 km too far north (7 to 71). Regular feeding stopped at many northern sites after the 2008–2011 avian influenza outbreaks. The share of Whooper Swans and Northern Pintails at fed sites fell from 62–71% to 10–30%. This gives most of the Pintail shift and about one quarter of the mean shift. At never-fed sites the mean trend was −8 km per decade (−17 to +3). The occupancy centroid did not change, so the shift is a fall of numbers at large northern sites, not a change of the occupied sites. Limits: the census sees only birds in Japan; before 1996 we used fixed prefecture points; counts are not corrected for detection; we did not test land-use causes.

waterfowl · Anatidae · wintering range · range shift · climate tracking · isotherm · Japan · January waterfowl census · artificial feeding · avian influenza · centroid · ducks · swans · geese · out-of-sample test