British butterflies flew about 2 days later per 10 kg N/ha/yr of deposition in 1986-2024, but tree-host species also did and warmth slopes did not change, so sward cooling was not shown

We tested the nitrogen (N) microclimate-cooling hypothesis on butterfly flight timing. This hypothesis says that N deposition makes swards denser and cooler, and thus delays species whose immature stages develop in the grass or herb layer. We used UKBMS mean flight dates of 20 univoltine species on British transects (60,296 site-species-years at 1,401 sites in 1986-2012, and data to 2024) and the CBED 1986-2012 mean grid-average N deposition. The models had 50 km block by year fixed effects, elevation, rainfall, local 60-day temperature, season-coverage and count filters, and site-clustered errors. We wrote the predictions before we fitted the models. Sward species flew +1.7 days later per 10 kg N/ha/yr (95% interval -0.0 to +3.5; k = 18). The point estimate was positive in three regions and three periods. The pre-set clear rule was not met. With land-cover controls (exploratory), the result was +2.4 (+0.8 to +4.0). Over the 10th-90th percentile N range (10-17 kg), this is about 1.2-1.6 days. The placebo test failed: two tree-host species had a larger positive association (+9.1; +1.2 to +17.0). The temperature slope (-4.25 days per C) did not clearly weaken with N (+0.15; -0.17 to +0.48 days per C per 10 kg). Grid-average N correlated with elevation (r = 0.48) and with 5 km tree cover within blocks (r = 0.52). Thus the small delay at high-N sites is possibly from site confounding, not from sward cooling. Limitations: observational cross-sectional contrast, a 5 km land-cover-weighted exposure that stops in 2012, no microclimate data, and only 2-3 placebo species. The bundle has code, aggregate results and figures, but no third-party records.

nitrogen deposition · butterfly phenology · flight date · microclimate cooling · UKBMS · CBED · placebo test · spatial confounding · observation bias · climate warming · air pollution