Evaluating detection of temporal trends in long-term freshwater fisheries data to inform future monitoring efforts

ABSTRACT Objective Florida’s Freshwater Fisheries Long-Term Monitoring Program was implemented in 2006 to track changes in freshwater fish populations and communities. As part of an evaluation of the program, this study used a simulation framework to assess trend detection for fish abundance and biomass indices and how sampling intensity (number of samples per year) and frequency (number of years) can influence detection of these trends. Methods Using count and weight data from fall electrofishing samples collected between 2006 and 2021 from 21 lakes, trends were simulated for annual mean count and weight over a 10-year period that ranged from −70% to +200%. In all, simulations were performed for seven game fish species and three management-relevant groups (large nongame, nonnative, and prey species). For sampling intensity, data were simulated with a range of sample sizes, from 10 to 40 electrofishing transects or the maximum number available for a given lake. For sampling frequency, data were simulated for different sampling schedules that included sampling 1 year followed by 1- or 2-year breaks (4–5 years of sampling in a 10-year period), sampling two consecutive years followed by 1- or 2-year breaks (6–7 years of sampling in a 10-year period), and sampling the first 5 years only. Results Simulations based on weight and count data yielded similar results, but the effect of sampling frequency and sampling schedule varied by species, management group, and lake. Trend detection was lower and more variable when mean counts and weights of fish in electrofishing samples were low. Overall, at least a 60% increase or 40% decrease over a 10-year period was typically needed for trends in mean weight and count to be detected at least 80% of the time in at least half of the lakes. Increasing sampling intensity did not substantially improve trend detection for lower-magnitude changes, but reducing sample intensity to a minimum of 10 electrofishing transects per year would have a large negative effect on trend detection in almost all lakes. Detection of trends improved as the number of years sampled increased, but ideally, sampling should be spaced throughout the entire 10-year period to capture the full magnitude of change. Sampling every year generally resulted in better trend detection and for many species and groups was the only sampling schedule that resulted in all study lakes achieving the 80% target detection level. Of the alternative schedules considered, those involving 2 years of consecutive sampling outperformed those with only 1 year of sampling followed by a 1- or 2-year break. Conclusions Relatively large changes in mean count and weight were required to detect trends over a 10-year period, but there was no clear advantage of using count or weight data for monitoring purposes. Further, study results support the current sampling intensity, but trend detection is optimized at higher mean catch and weight values. Although sampling every year is ideal, an alternative schedule involving sampling two consecutive years with 1- or 2-year breaks could be considered in certain situations. These results will be important for informing future decisions regarding Florida’s Freshwater Fisheries Long-Term Monitoring Program and other monitoring initiatives.