Understanding imperfect detection in a San Francisco Estuary long‐term larval and juvenile fish monitoring programme

Imperfect detection can present a significant challenge when monitoring for a rare and imperilled species. Here, a long‐term larval and early‐juvenile fish monitoring programme in the upper San Francisco Estuary was examined to evaluate its overall reliability in detecting various fish species, including the imperilled delta smelt, Hypomesus transpacificus McAllister, for which the programme was designed. Using occupancy modelling, detection probability of species with pelagic larval or juvenile life stages was found to be generally high (≥.95) based on the current sampling effort of three larval net tows per site. However, detection probability can vary considerably from year to year depending on the species’ level of larval production. Water temperature and turbidity were identified as important predictors of occurrence for young‐of‐year delta smelt, longfin smelt Spirinchus thaleichthys (Ayres) and striped bass Morone saxatilis (Walbaum), and there was evidence for fish size selectivity by the sampling gear in all three species. These results highlight the need to consider adaptively managing detection probability by increasing sampling effort in years when young‐of‐year delta smelt abundance is expected to be low, especially when information on the species’ occurrence at a particular region is critical.