How Quickly Do Fish Communities Recover From Boat Electrofishing in Large Lakes?

Boat electrofishing is commonly used by fisheries biologists to sample freshwater fishes in the littoral zone of standing waters (Bonar et al. 2009). In large lakes, researchers often collect samples at randomly selected sites over fixed lengths of shoreline or periods of electrofishing time and express results as a measure of relative abundance (i.e., catch per unit effort [CPUE], or number of fish per mile or hour; Miranda and Boxrucker 2009; Reynolds and Kolz 2012). Researchers have widely applied this for general fisheries assessments, but it may not adequately monitor trends at specific locations that might serve as habitat for rare or invasive species. Consequently, some monitoring programs have expanded their efforts to include both probabilistic (random) and nonprobabilistic (targeted) site sampling (e.g., Gutreuter 1993; Hoffman et al. 2010).

Sampling designs that incorporate both random and targeted sites can be complex, especially when random sites are selected that partially overlap frequently targeted sites. In addition, there may be circumstances when researchers desire to estimate fish abundance at targeted sites using mark–recapture methods (Hall 1986; Mitro and Zale 2002). In both cases, caught and released fish should have equal catchability to those not captured between successive electrofishing passes (Van Den Avyle and Hayward 1999; Temple and Pearsons 2006). While fisheries biologists have studied the amount of time necessary to allow fish to recover and redistribute between electrofishing sampling events (termed “recovery period”) in cold-water streams (Temple and Pearsons 2006), information on shoreline habitats in lakes is limited. Most authors have applied a 24–48-h recovery period between electrofishing passes (Hall 1986; Mesa and Shreck 1989; Peterson et al. 2004) and some have questioned the validity of mark–recapture estimates when the recovery period is less than 6 h for some species (Shreck et al. 1976; Temple and Pearsons 2006). However, researchers do not know if shorter recovery periods would be equally effective when sampling near-shore habitats.

Monitoring for new invasive species may require resampling of targeted habitats within relatively short time durations to ensure adequate coverage of unique habitats. New nonnative species in the introduction phase occur at low abundance, making them difficult to detect (Jerde et al. 2011), particularly in large waters which are not as easily sampled, as opposed to smaller waters such as streams where researchers can block off and deplete targeted reaches (Lonzarich et al. 1998). In the Laurentian Great Lakes, where new aquatic invasive species (AISs) are introduced on a regular basis (Mills et al. 1993), it is paramount to find new invaders quickly and reliably. Within the U.S. Fish and Wildlife Service (USFWS), most AIS early detection and monitoring programs on the Great Lakes focus much of their efforts on large port cities with high connectivity to other waters. Most of these locations have highly modified shorelines (e.g., seawalls, rip-rap) that limit potential areas in which researchers can use boat electrofishing, which is often the most effective gear for characterizing fish communities in these nearshore areas (B. Smith, personal observation). At most high-priority sampling areas in Lake Michigan, researchers often sample boat electrofishing sites more than once to ensure complete coverage of limited available habitat and collection of all species. However, we currently do not know the influence of initial electrofishing runs on subsequent electrofishing runs.

Our objective was to determine if CPUE, size structure, and community metrics of near-shore fishes in targeted habitats were affected by recovery periods of differing duration. Our goal was to establish the shortest recovery period necessary to estimate fish abundance, size structure, and community metrics and to evaluate if resampling of electrofishing sites could be performed within the same day while yielding reliable fish community data and potentially collecting additional species not collected during earlier runs. If initial electrofishing runs influence catch, size structure, and community metrics of subsequent electrofishing runs, we would expect larger differences for short recovery periods than for longer recovery periods.

We conducted boat electrofishing during late June, July, and early September in the southernmost 3 km of Green Bay, Lake Michigan, hereafter referred to as Lower Green Bay (Figure 1). We randomly selected 30 sites along the shoreline having depths of 0.5–2 m. We applied a 200-m buffer between sites during the selection process to reduce potential interference from treatments. Some sites partially overlapped but we sampled them weeks or months apart, negating any influence of the original sampling event. We randomly assigned each site one of six recovery durations (0.25, 1, 3, 6, 24, or 48 h), for a total of five replicates per treatment. The 48-h recovery period was the control. We conducted electrofishing during daylight hours (0800–1800 hours) with either a 6.1-m or 5.5-m–long Kann two-boom electrofishing boat equipped with an Infinity control box (Midwest Lake Electrofishing Systems, Polo, MO). We standardized settings for pulsed DC electrofishing with duty cycle set at 30%, pulse rate at 60 pulses per second, 118–325 V, 20–30 A, and power of approximately 5,000–5,500 W. Sampling began at randomly selected points and proceeded along the shoreline for 10 min of electrofishing time. We recorded the boat's path using a boat-mounted global positioning system. We identified fish to species and gave all individuals ≥ 150 mm a unique partial fin clip/hole punch combination that was site and run specific. We measured the first 50 individuals of each species during each run for total length to the nearest millimeter; we counted all other fish. The time for recovery began once fish were released in the middle of the sampling tract. The second electrofishing run replicated the first, by following the same tract for the same pedal time. We again collected fish, identified them to species, measured total length to the nearest millimeter, and recorded any fin clips. We also gave newly captured fish on the second run the same site- and run-specific fin clip to indicate they had been captured. Marking and recapture runs were performed with the same electrofishing boat and crew.

For each electrofishing run we calculated the number of recaptured fish, CPUE, mean total length, species richness, Shannon's diversity index (H′), and Shannon's evenness (J′).We computed CPUE as the number of fish captured per minute. We conducted analyses on differences in measures between the first and second run for each variable (i.e., residuals), except number of recaptured fish, because recaptured fish were only available during the second electrofishing run. We chose to use residuals because we wanted to identify site-specific differences between marking and recapture runs. The response variable for recapture rate was the proportion of marked fish that were recaptured in the site where they were initially marked and released. We tested all variables for normality using the Shapiro–Wilk test and tested homogeneity of variances using Levene's test. We log-transformed nonnormal data to meet the assumption of normality.

We conducted multivariate analysis of variance (MANOVA) to investigate if differences existed among the recovery durations for each of the residual variables. We used Wilk's lambda to calculate the multivariate F statistic. If overall MANOVAs were significant, we used ANOVAs to test for differences among treatment types using type III sums of square with a priori α = 0.10 for all comparisons. If we performed ANOVAs, we employed the false discovery rate to control type I error rates for multiple tests (Benjamini and Hochberg 1995, Verhoeven et al. 2005). We constructed bar plots to help visualize differences in response variables (mean ± SE) between marking and recapture runs for varying levels of recovery duration. We performed all analyses using the R statistical software program (R Core Team 2016).

Recovery duration had no effect on measured population metrics between marking and recapture runs. We found similar differences between marking and recapture runs in log-transformed CPUE, log-transformed mean total length, species richness, Shannon's diversity, and Shannon's evenness, despite variable recovery duration (Λ_Wilks = 0.459, F_5,25 = 0.741, P = 0.799; Data S1, Supplemental Material). Because the overall MANOVAs did not detect significant differences, we did not perform individual ANOVAs for each response variable. Due to poor weather we were unable to revisit one site in time to resample and ended up sampling later, resulting in one additional 24-h replicate (n = 6). Bar plots comparing measured population metrics (mean ± SE) between marking and recapture runs further demonstrate similarity between marking and recapture runs; no detectible trend is apparent for any response variable (Figure 2).

We collected a total of 5,108 fish from 36 species and marked 727 fish of 23 species during marking and recapture runs (Table 1). The most commonly collected species were juvenile Gizzard Shad Dorosoma cepedianum (n = 1,778) and Yellow Perch Perca flavescens (n = 1,855). Common Carp Micropterus dolomieu (n = 194), Freshwater Drum Aplodinotus grunniens (n = 173), and Smallmouth Bass Micropterus dolomieu (n = 90) were the most common species marked and released. We marked 349 fish during marking runs (≥ 150 mm) but only one was recaptured in the same area where it was released (Smallmouth Bass; 371 mm); accordingly, we did not have sufficient data to perform analysis between varying levels of recovery duration. We recaptured five other fish including two Smallmouth Bass (192 and 402 mm), two Common Carp Cyprinus carpio (544 and 734 mm), and one Rock Bass Ambloplites rupestris (166 mm) in areas outside the tract where they were initially captured and released. Fish recapture was unrelated to recovery period duration. We recaptured one fish 1 h after release, two after 3 h, two after 6 h, and one at 24 h. We marked and released six Burbot Lota lota (169–249 mm) and three Muskellunge Esox masquinongy (343–345 mm) but they did not recover from handling stress in warm water (22.3–27.1 C) so we excluded them from the pool of marked fish available for recapture.

Overall, resampling at each site provided reliable fish community data and led to the capture of additional species. We only captured Bigmouth Buffalo Ictiobus cyprinellus, Black Crappie Pomoxis nigromaculatus, Fathead Minnow Pimephales promelas, and Trout-Perch Percopsis omiscomaycus during recapture runs and only captured seven other species during marking runs (Table 1). We captured four known AISs during marking and recapture runs, including Alewife Alosa pseudoharengus, Common Carp, Round Goby Neogobius melanostomus, and White Perch Morone americana. We captured one Shortnose Gar Lepisosteus platostomus, a native transplant from the Mississippi River basin, during a marking run.

Failure to detect differences in CPUE, size structure, population metrics, and recapture rates for any length of recovery duration between marking and recapture runs indicates that our sampling areas were not significantly impacted by electrofishing. If electrofishing had a strong effect on study reaches, we would expect that the shortest recovery durations (i.e., 0.25 h) would yield lower abundance, smaller size structure, and fewer species during recapture runs, but this is not what we observed. We identified at least two reasons (i.e., geographical scale, water quality) related to electrofishing efficiency to explain why we did not detect differences between marking and recapture runs with varying recovery durations.

Lower Green Bay is a large, complex, open system and a transitional zone between the Fox River, a major tributary, and the greater portion of Green Bay and by extent Lake Michigan. Most electrofishing recolonization and recovery studies have focused on small, relatively confined study areas such as streams (Temple and Pearsons 2006) and reservoirs (Maceina et al. 1995). By comparison, our study area is orders of magnitude larger and directly connected to other large adjacent water bodies. Following electrofishing, we observed similar fish abundance and diversity in sampled areas, and with virtually no recaptures, indicating that fish captured during recapture runs were missed during the marking run or were immigrants to the study area. Once we marked fish and returned them to the water, they either dispersed out of the study area or they were less vulnerable to recapture. We suspect that most released fish dispersed out of the study area—especially at longer recovery durations. Similar problems have been documented in rivers where larger fish captured by electrofishing quickly leave the sample area upon release (Nordwall 1999) while immigration into the study area remains constant (Young and Schmetterling 2004). Even small warm-water streams experimentally defaunated can recover their original fish abundance and diversity within a few days (Peterson and Bailey 1993; Lonzarich et al. 1998) but in a large, productive, open system like Lower Green Bay this process may only take minutes to hours.

The effective area sampled by boat electrofishing in this experiment was small compared to the overall size of the study area. During stream electrofishing surveys the entire width of the stream may be effectively sampled (Thompson and Rahel 1996; Bertrand et al. 2006) and some stream fishes such as Rainbow Trout Oncorhynchus mykiss may have similar catchability within 3 h after capture by electrofishing (Temple and Pearsons 2006). In cove reservoir sampling (2.3–3.7 ha), researchers can effectively estimate Largemouth Bass Micropterus salmoides abundance by depletion with several concurrent electrofishing events, and with similar catchability of fish within hours after initial capture and marking (Maceina et al. 1995). By comparison, our study area was open, allowing fish to easily immigrate in and emigrate out of the electrofishing tract, making any meaningful depletion or attempt at making population estimates futile.

Water clarity may have influenced our results through reduced capture efficiency. This was likely the primary reason for low recaptures at short resting durations (i.e., 0.25 h) because there was less time for fish emigrate out of the study area. Lower Green Bay is shallow (< 5 m), eutrophic to hypereutrophic, and highly turbid (Trebitz et al. 2007; Althouse et al. 2014) with water clarity often < 0.5 m. High turbidity is known to reduce visibility of electroshocked fish to netters, particularly small fishes and those lacking swim bladders (Reynolds 1996; Brousseau et al. 2005; Price and Peterson 2010). Electrofishing capture efficiency in many systems, and for most species, is low and becomes poorer at extreme levels of turbidity (Reynolds 1996; Lyon et al. 2014). Assuming low capture efficiency during marking runs, it is unlikely that we achieved depletion of abundance and diversity in the study area. Assuming immigration into our study area was constant, the fish we removed could be replaced, especially over longer resting durations.

For monitoring AIS, resampling of a site may be required and acceptable. The USFWS currently performs targeted resampling of rivers to detect Bighead Hypophthalmichthys nobilis and Silver Carp Hypophthalmichthys molitrix using environmental DNA; researchers typically sample sites three times per sampling season (USFWS 2017). By reelectrofishing our study areas we were able to find all known AISs and additional species not detected during initial marking runs with no detrimental effect on measured population metrics. By allowing overlapping sampling of a site within a short time period, we increased flexibility in our sampling design while maximizing sampling efficiency.

For specific study questions, like detection of AIS in ideal habitats, targeted electrofishing can be performed concurrently in the same location and provides similar fish community information between electrofishing runs. Resampling a site may also yield species not captured during the initial sampling period. These findings would not be applicable in situations where precise estimates of system-wide target species abundance are desired because overlapping electrofishing runs would not be independent of each other. We conclude that shallow-water fish communities in Lower Green Bay quickly recover after electrofishing, allowing for resampling to be performed in the same day with no measurable impact on calculated fish community metrics. Our results would likely apply to other large lake and reservoir systems, particularly for turbid waters.

Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.

Data S1. Response variable data used to calculate residuals (recapture run − marking run) including log-transformed mean total length, log-transformed catch per unit effort, species richness, Shannon's diversity (H′), and Shannon's evenness (J′) for six recovery-period durations (0.25, 1, 3, 6, 24, and 48 h). We performed sampling during June–September 2015 on Green Bay, Lake Michigan.

Found at DOI: http://dx.doi.org/10.3996/032017-JFWM-021.S1 (17 KB XLSX).

Reference S1. [USFWS] U.S. Fish and Wildlife Service. 2017. Quality Assurance Project Plan eDNA Monitoring of Bighead and Silver Carps. U.S. Fish and Wildlife Service, USFWS Midwest Regions, Bloomington, Minnesota. This document details field, mobile laboratory, and genetics laboratory procedures and is updated annually.

Found at DOI: http://dx.doi.org/10.3996/032017-JFWM-021.S2; also available at https://www.fws.gov/midwest/fisheries/eDNA/ documents/ QAPP.pdf (2125 KB PDF).

This project could not have been performed without the cooperation and support of the Aquatic Invasive Species program staff at the Green Bay Fish and Wildlife Conservation Office, including Ken King, Matt Petasek, Anthony Reith, Dalton Hendricks, Brandon Falish, Lisa LaBudde, Lindsey McKinney, Rachel Richter, Carolyn Malecha, Jessica Finger, and Hailey Farah. I would also like to thank the journal reviewers and the associate editor for their careful consideration of this manuscript.

Any use of trade, product, website, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government.