Summer rain gave no clear repeated link to next-year butterfly record loss in England and Wales in 1984-2023

We tested whether summer rain predicted loss of the same Ringlet or Meadow Brown record in the next year. We used 54,904 current site-species rows at 3,255 sites in England and Wales, with separate tests for 1984-2003 and 2004-2023. Models used 20 annual observations each and controlled summer temperature and linear time. Rain slopes ranged from -1.86 to +4.14 percentage points per 100 mm. All four adjusted intervals included zero. Loss of all site records accounted for most missing study rows in each sample. This is a partition of archive loss, not proof of survey closure or extinction. Country weather, changing site membership, missing visit histories and short annual series limit inference. The result does not show random missingness or no biological drought response. The package includes the full risk set, source hashes, code, uncertainty, checks and offline reproduction inputs.

butterflies · observation bias · rainfall · archive loss · missing records · regional replication · null result