Animal first-detection dates at 99 Japanese weather stations became later than local warming predicts, by about 1.1 days per decade more than plant dates, in 1953-2020

Question: did the animal first-detection dates in the Japan Meteorological Agency (JMA) phenology network change in the same way as the plant dates, relative to the local temperature? Does a decrease in detection explain a difference? Approach: we used 2,108 station series (20 animal items, 32 plant items that warm years advance) at 99 stations, 1953-2020, with GHCN-Daily station temperature. For each series, we fit the date against the mean temperature in the pre-event window and a linear year term. The year term (gamma) is the trend that the temperature does not explain. Plant events at fixed reference trees are the control. Results: warm years advanced animals by 3.3 days/C and plants by 4.1 days/C (medians). The windows warmed by about 0.2 C/decade. The plant dates moved earlier (-0.63 days/decade), but the animal dates moved later (+0.66 days/decade). The median gamma was +1.34 days/decade for animals and +0.24 for plants. The difference was +1.11 days/decade (95% station-cluster bootstrap CI 0.83-1.37). It was positive in three regions and in two periods (8 of 9 CIs did not include zero) and with a 365-day window (+1.33). Gamma was largest for frogs, a dragonfly and a lizard (+2.4 to +4.5) and smallest for cicadas and the barn swallow. Detection tests gave partial support. Animal dates next to non-recorded years were later (3.5 days per unit gap share, vs 1.8 for plants; difference p = 0.006). From 1988, nearby records were 1.1 days earlier than on-site records. Sighting items had a larger gamma than song items (1.95 vs 0.93). But the residual scatter of the animal dates did not increase, and the dates before a series stopped early were not later. These two results do not agree with a simple sparse-detection model. Limitation: we cannot divide the gap into a true phenology change and a detection change. The code 0 combines non-detection and no observation. Do not use the JMA animal records as uncorrected climate-response indicators.

phenology · observation bias · first detection date · climate change · Japan Meteorological Agency · temperature sensitivity · amphibians · insects · birds · plants · urbanization · GHCN-Daily