The forecast gain from ice-off dates was uncertain for sampled Daphnia peaks at two Wisconsin lakes in 2007–2024

We tested whether lake ice-off dates improve forecasts of the sampled spring Daphnia density peak at Lakes Mendota and Monona, Wisconsin, USA. We used fixed open data versions and required survey coverage from April to July. Each forecast used at least ten eligible earlier years from the same lake. The main test had 24 lake-year targets in 12 shared years between 2007 and 2024. The fitted ice-off model had a mean absolute error of 10.80 days. A forecast based on the mean earlier peak date had an error of 11.14 days. The difference was −0.33 days, with a 95% three-year block resample interval of −1.88 to +0.82 days. One- and five-year block intervals also included zero. If 2024 was removed from scoring, the difference was +0.21 days. Some other date definitions and model forms gave larger gains. The result applies to the largest recorded total density from April 1 to July 15. It does not test a water temperature threshold model or identify a causal climate effect. The two lakes share weather and water, survey years are missing, and sample spacing limits peak timing. The package has unchanged source data and metadata, code, all forecasts, survey exclusions, figures, source hashes and independent arithmetic checks.

Daphnia · lake ice · phenology · ecological forecasting · chronological validation · open data · Lake Mendota · Lake Monona