https://doi.org/10.15447/sfews.2018v16iss4art2

Resource managers rely on long-term monitoring surveys conducted in the San Francisco Estuary to evaluate the status and trends of resident fish populations in this important region. These surveys are potentially confounded because of the incomplete detection of individuals and species, the magnitude of which is often related to the same factors that affect fish populations. We used multistate occupancy estimators to evaluate the distribution, abundance, and detection probability of four fish species collected during 1995–2015 with three long-term surveys. Detection probabilities varied positively with fish abundance and negatively with Secchi depth. Detection varied among species and was greatest for the 20-mm Survey and least for the midwater trawl used for the midwater trawl used in the San Francisco Bay Study. Incomplete detection resulted in underestimates of occupancy and abundance across species and surveys and were greatest for the Bay Study. However, trends in occupancy and abundance of the study period appeared to be unbiased. Fish occupancy and abundance were generally related to salinity or specific conductance, day-of-the year, and water temperature, but the nature of the relations varied among surveys and species. There also was strong spatial and temporal dependence in species-specific occupancy and abundance that changed through time and were unrelated to the covariates considered. Our results suggest that managers consider incorporating methods for estimating detection and adjusting data to ensure data quality. Additionally, the strong spatio-temporal patterns in the monitoring data suggest that existing protocols may need to be modified to ensure that data and inferences reflect system-wide changes rather than changes at a specific set of non-randomly selected locations.