June Daphnia shares gave no clear August prediction gain in seven Wisconsin lakes in 2014–2025

We tested whether June Daphnia shares improve August predictions in seven northern Wisconsin lakes. We used a fixed open NTL-LTER data version and the same cases for all models. Models fitted to 70 cases in 2003–2013 were kept fixed for 83 test cases in 2014–2025. The reference used lake identity and previous August share. Adding June share increased mean squared error by 19.25%. The nominal 95% range for error decrease was −55.49% to +11.20% when training and test years were resampled in shared three-year blocks. Adding July share reduced error by 17.00%, with a range of −16.20% to +30.56%. Thus neither added gain was resolved under that uncertainty model. Fixed-fit test-only ranges excluded zero in opposite directions. A secondary direct comparison gave 30.40% lower error with July than with June, with a range of +1.21% to +44.22%. Planned sampling and model checks retained the point directions. The result applies to recorded density fractions and the tested models. It does not identify causes or a general limit on prediction. The package preserves unchanged source data, provenance, code, all predictions, figures and checks. A separate calculation and a clean repeat matched the numerical results.

Daphnia · zooplankton · seasonal prediction · lake ecology · forecast evaluation · training uncertainty · open data · Wisconsin