Kirsch, J.E., J.T. Peterson, A. Duarte, D. Barnard, A. Goodman, S. Hugentobler, J. Julienne, M. Meek, R.W. Perry, C. Phillis, L. Smith, and J. Stewart. 2024. Fish misidentification and incomplete detection affects inferences from data informing water operation decisions. North American Journal of Fish Management https://doi.org/10.1002/nafm.10974
Abstract
The alteration of river flow regimes and water diversions caused by water management infrastructure is a major stressor to freshwater fishes. The gate-controlled Delta Cross Channel (DCC) at Walnut Grove, California was constructed to divert fresh water from the Sacramento River to supply freshwater to two export facilities that pump water into canals for agricultural and domestic uses. The alteration of river flows due to the operation of these gates can negatively influence the migration pathways and survival of endangered juvenile Sacramento River winter-run Chinook salmon Oncorhynchus tshawytscha. Count data of these salmon at fixed monitoring sites in the Sacramento River has informed DCC operations during October through December since 1996. Although clear guidelines were developed for DCC operations using the salmon count data, there is uncertainty about how environmental conditions influence salmon occupancy and if those relationships could be affected by fish sampling and identification error. We assessed the effect of environmental conditions along with incomplete detection and misidentification error on juvenile winter-run Chinook salmon occupancy by analyzing data collected at nine monitoring sites from 1996 to 2019 using hierarchical multi-state occupancy models. To incorporate misidentification into the modeling framework, we used accuracy data derived from fish identification examinations along with genetic sampling. We found that occupancy varied considerably and was related to flow characteristics, water clarity, weather, time of year, and whether occupancy was detected during the previous sampling day. However, we found that these relationships and their significance changed considerably when we accounted for incomplete detection and the nominal probability (0.056) of misidentifying individual salmon. Detection was relatively low and influenced by factors including a site effect, flow, water clarity and volume sampled. Our study demonstrates the need to account for identification and detection error while using monitoring data to assess factors influencing fish occupancy and inform future management decisions.