Regional winter temperature gave no clear forecast gain for 12 common California larval fish taxa in 2000-2022

We tested whether temperature predicts fish-larva occurrence more reliably than density when larvae are caught. We used 1,743 CalCOFI spring tows and 12 common taxa in two sectors. Models fitted in 1978-1999 were fixed for tests in 2000-2011 and 2012-2022. Regional January-March sea surface temperature was added to survey controls and a year trend. All 12 corrected primary skill intervals included zero. Occurrence point skill ranged from -0.024 to +0.006. Positive-density skill ranged from -0.044 to +0.018. Paired differences changed sign across sectors and periods. An exploratory check with annual mean errors gave no clear positive gain. Removing the year trend gave a small clear total-density gain in early South only. These results do not support a general claim that occurrence is easier to forecast than positive density in this sample. Sparse positive catches, shared regional climate, short tests and fixed model forms limit inference. This is a realized-weather hindcast of catches, not a forecast of adult stocks or population growth.