Comparison of Fish Catch Between an Electric Seine and Backpack Electrofishers in Shoal Habitat of the Chipola River, Florida

Gear selection for fish sampling is a crucial step in any study design, but choosing the most effective sampling gear is especially challenging in lotic environments with variable depth and flow. Certain gear types may be more suitable than others depending on study objectives, the habitat being sampled, and the selectivity of the gear for certain species or sizes of fish (Buckmeier and Schlechte 2009). Gear comparison studies offer a decisive way to select the gear that most effectively samples fish in a target habitat. Backpack electrofishers are widely used to collect freshwater fish in shallow, lotic habitats because of their compact size, relative safety, small crew size requirement, and ease in standardizing sampling effort (Onorato et al. 1998). Additionally, they are commonly used to sample hard-to-reach stream sites and wadeable areas that are inaccessible by boat.

A less commonly used electrofishing device is the electric seine. The electric seine was introduced by Haskell (1940) and has taken on various forms since, likely because it is not commercially available and has been adapted to local sampling needs (Haskell 1950). An electric seine consists of several dropper electrodes suspended from horizontal cables that are connected to a generator and an electrofishing box (typically a boat electrofishing box). Electric seines have been successfully used as a fish sampling method in streams (Larimore 1961; Bayley et al. 1989; Angermeier et al. 1991). Dowling et al. (1990) provided the first technical reference for constructing an electric seine that broadcasts an electric field over a large area (Bestgen et al. 2007) and reduces fish escapement (Bayley et al. 1989).

Previous gear comparison studies in lotic environments have compared a wide variety of gears including boat electrofishing, hoop netting, backpack electrofishing, shore-based electrofishing, mesh seining, prepositioned area shocking, beach seining, electric seining, and parallel wires (Dauble and Gray 1980; Wiley and Tsai 1983; Onorato et al. 1998; Pugh and Schramm 1998; Ensign et al. 2002; Walsh et al. 2002; Burns 2007). The few studies that have compared electric seines with backpack electrofishers suggested that electric seines are more efficient at describing species richness and total numbers present and provide a desirable mix of efficiency and operability (Bayley et al. 1989; Bestgen et al. 2007). However, to our knowledge, none of these studies has been conducted in large river shoals (i.e., rivers too wide to allow block nets to be reasonably stretched from bank to bank) characterized by the high species diversity typical of the southeastern United States, nor have they compared the effectiveness of using two backpack electrofishers vs. one electric seine.

The Florida Fish and Wildlife Conservation Commission routinely performs long-term monitoring of fish communities in rivers such as the Chipola River. Shoal habitat in the Chipola River is difficult to sample effectively because of large areas of shallow water and rock outcrops (Wheeler and Allen 2003). Therefore, gears such as barge electrofishers, boat electrofishers, and mesh seines are ineffective or impractical to use because they run aground and easily snag on the jagged river bottom. The inability to effectively sample shoal habitat may lead to a misrepresentation of the river's ichthyofauna, as the fish assemblage of shoal habitat likely differs from that of deeper portions of the river typically sampled using a boat electrofisher.

Before starting a long-term monitoring program on the Chipola River, we conducted a gear comparison study to determine which of two gear types would be more effective at sampling fish in shoal habitat. Our objective was to compare the relative effectiveness of two backpack electrofishers with that of an electric seine in a large river shoal habitat. Personnel, construction, and financial comparisons are also discussed. The results of this study may help facilitate study design considerations for other fish community sampling studies in large river shoal habitats that occur in the southeastern United States and may be used in meta-analyses between these two gear types.

The Chipola River is formed by the confluence of Marshall Creek and Cowarts Creek in Jackson County, Florida and is a tributary of the Apalachicola River. Sixty-three active springs provide relatively stable water temperatures and low turbidity (Parsons and Crittenden 1959; Bass and Cox 1985) for approximately 40 species of fish (Florida Fish and Wildlife Conservation Commission, unpublished data). The karst geology of the area distinguishes the river's substrate from that of other rivers in northwest Florida (Bass and Cox 1985) and creates unique habitats such as boulders, rock shelves, smooth bedrock, and shoals with jagged rock outcroppings. Shoals are shallow areas (<1m) with increased surface agitation and rocky substrate, with some substrate breaking the surface. Shoal habitat can be found in numerous rivers across the southeastern United States primarily at or below the Fall Line (Wynn 2012). The presence of eelgrass Vallisneria americana and a widening of the stream were often (but not always) associated with shoals in the Chipola River. Four shoals were sampled near the town of Altha, Florida, between Peacock Landing and Johnny Boy Landing (Figure 1).

We constructed an electric seine similar to that of Dowling et al. (1990), with some structural modifications. The electric seine was approximately 4 m long and consisted of six 6.35-mm-wide steel drop electrodes (droppers) that were 91 cm long and spaced 76 cm apart. Three anode droppers (+) and three cathode droppers (−) were alternated along the length of the electric seine. Six gill-net floats (15.2 × 9 cm) were threaded over the cable to float the seine (Figure 2A). A Briggs and Stratton 3,250-W, 240/120-V alternating current generator and a Smith-Root type VI boat electrofisher supplied and regulated power to the seine. A junction box combined the electrical connections from a standard foot-pedal switch and two Smith-Root anode poles. The generator, electrofisher, junction box, and foot pedal were all located in a 16-foot aluminum boat during sampling. The poles served as the proximal and distal probes and acted as two additional safety switches. As a safety mechanism, power was cut off if either probe switch or the foot-pedal switch was not depressed. Electrical output was set to 60 Hz (frequency), 100 V, and 2.5 ms (15% duty cycle). The average output amperage was approximately 0.6 amps, which was sufficient to immobilize fish of all sizes in preliminary gear testing.

Two Smith-Root model 12b backpack electrofishers were used for sampling fish for comparison with the electric seine. We chose to use two backpack electrofishers rather than one because two units can cover an area comparable with that covered by the electric seine. Electrical output on the backpacks was set to 60 Hz, 300–400 V, and 6 ms, and average amperage was approximately 0.8 amps.

Although electrical settings differed between the two gear types, our goal was to immobilize the majority of observed fish to optimize catchability across multiple families of fish. Limitations in output setting available for each gear type prevented us from using settings that were identical to each other. Field trials conducted before the study indicated that the settings we used limited fish escapement and optimized catchability. We observed little to no escapement of electroshocked fish during the study for both gear types.

Before sampling began, we recorded coordinates delineating the boundary of each of four shoals with a Garmin 76CSx global positioning system (GPS). The GPS coordinates were connected to create boundary lines for each shoal using ArcGIS 10 (Esri 2010). A grid of rectangles 4 m wide × 5 m long was then layered over each shoal using the Create Fishnet tool in XTools Pro v. 11 to visualize all possible sites. This grid size was subjectively chosen on the basis of our block net size and gear operability within that length and width. Site coordinates were determined for each rectangle within the grid using the Add XYZ Coordinates feature in XTools Pro. v. 11. A random-number generator was used to randomly select 25 sites for backpack electrofisher sampling and 25 sites for electric seine sampling. The number of sites per gear was chosen on the basis of a balance between performing enough samples to allow statistical testing and the number that could reasonably be sampled. All sites were sampled from August 28 to October 3, 2014, and the number of sites per shoal per gear type was kept as equal as possible given time, personnel, and available sampling area constraints.

Each electric seine site was sampled by a crew of five—two people handled the probes, two kickers carried dip nets and walked behind the seine (Figure 2B), and one person operated the foot-pedal switch in the boat positioned near the shoal. Kickers were used because preliminary gear testing revealed that after being shocked, darters and other benthic fish species remained on the bottom in either dense eelgrass or interstitial spaces in the rocky substrate. The kicking helped bring immobilized fish up into the water column, where stream flow would carry them into the block net. If possible, the kickers also captured fish with dip nets. Site sampling began by locating a randomly drawn coordinate using the GPS. The apex (i.e., middle) of a block net (14 × 1.5 m with 3-mm mesh) was then positioned over the location of the coordinate with both ends pointing upstream to create a U shape (Figure 2B). Bricks or large rocks were then placed along the base of the block net to maintain bottom contact and prevent fish escapement. The net was held in position during sampling by tethering the poles on each end of the net to an anchor that was hooked onto bedrock. Great care was taken not to disturb the block-netted area before sampling by staying on the outside of the net. After the block net had been set, the electric seine was stretched across the width of the site area at the upstream end of the block net. Electric current was applied for approximately 2 min for each site. Biologists handling the probes on either end of the electric seine slowly walked downstream outside the block net while the kickers within the block net followed the seine while disturbing the water column, substrate, and vegetation. This continued until the seine reached the end of the block net, where power was cut off and the end of the block net quickly lifted out of the water like a hammock. After removing all fish the end of the block net was laid back into the water at its original location.

Each backpack electrofisher site was sampled by a crew of four—two operating electrofishers and two kickers walking behind carrying dip nets. Methods for deploying the block net and sampling were identical to those used for electric seine sampling. All fishes were identified to species and counted. Most fishes < 200 mm total length were placed in a jar of 10% formalin and processed at the laboratory.

Depth (m) and surface current velocity (m/s; Marsh McBirney Flow-Mate 2000) were recorded at the upstream, middle, and downstream ends of each site after fish collection. One reading each of water surface temperature and ambient conductivity were also taken at each site after fish collection. The primary and secondary substrate types were visually quantified within the site area by percent coverage using four substrate types similar to those of Bitz et al. (2015): boulder, rocky fine, sand/pea gravel, and smooth bedrock. Percent area coverage of eelgrass and filamentous algae (the only two types of aquatic vegetation present) was also visually quantified within each site.

Habitat variables for sites were compared between gears to determine whether any significant differences existed between habitats sampled by the two gear types. Mean depth and mean surface current velocity were analyzed using linear mixed models in SAS v9.4 (Cary, NC) using the MIXED procedure. Shoal was included as a random variable to account for both gear types being tested at each shoal, and sampling gear was included as a fixed factor. Because percent coverage of eelgrass and filamentous algae was not normally distributed we used a generalized linear mixed model assuming a log-normal distribution to analyze each of these habitat variables. Again, shoal was included as a random variable and gear was included as a fixed factor. Because we observed so many sites with zero filamentous algae cover we added a very small amount (0.1% coverage) to all observations to enable analysis.

Total catch of fish per site and species richness per site were also analyzed using generalized linear mixed models; we assumed a negative binomial distribution to account for the discrete nature of the data (counts of fish or species). Shoal was included as a random variable and gear was included as a fixed factor. All generalized linear mixed models were analyzed in SAS v9.4 using the GLIMMIX procedure.

Fish assemblage indices such as Simpson's dominance, Shannon's diversity, and Pielou's evenness were compared between gear types using the Diversity Permutation Test module in Paleontological Statistics software (Hammer et al. 2001). This module takes species-by-site matrices (i.e., raw total catch data by site), pools the matrices of the two gear types, bootstraps the pooled data to create 1,000 new data matrices with the same number of species and sites as the original data set, computes diversity indices on each new matrix, and calculates P values (α = 0.05) to evaluate differences in fish assemblage indices between gear types.

Sample-based species accumulation curves were generated using the Sample Rarefaction module in Paleontological Statistics software for each gear type to compare the average number of species collected relative to the total number of sites sampled. Curves were linearized using an arcsine transformation fit to a third-order polynomial. Differences between rates of species accumulation between the gear types were compared using analysis of covariance in SAS version 9.4.

Nonmetric multidimensional scaling was performed on fourth-root-transformed catch data with the Bray–Curtis distance measure to visually depict site similarity/dissimilarity between gear types. One backpack electrofisher site outlier was removed before the analysis because zero fish had been collected there. A permutation-based multivariate analysis of variance was then used to test the null hypothesis that the two gear types did not differ significantly in fish assemblages collected. The permutation-based multivariate analysis of variance and nonmetric multidimensional scaling were performed using Paleontological Statistics software. All statistical tests were conducted with a significance level (α) of 0.05.

Mean water temperature during the study was 24.2°C (range: 20.9–26.5°C), and mean ambient conductivity was 249.7 μS (range: 218.9–269.5 μS). Shoal habitats sampled differed between gear types (Table 1). Depth (F_1,45 = 4.06, P = 0.0499) and velocity (F_1,39 = 13.11, P < 0.001) differed significantly between sites covered by backpack electrofishers and electric seines. The ranges of mean depth and mean velocity for backpack electrofisher and electric seine sites was 0.22–0.66 m and 0.19–0.81 m, and 0.10–1.63 m/s and 0.27–2.81 m/s, respectively. Eelgrass coverage did not differ between sites sampled (F_1,33 = 0.15, P = 0.7047), whereas filamentous algae coverage differed significantly between sites sampled (F_1,44 = 11.01, P = 0.0018). Substrate composition in sites between the two gear types, though not statistically evaluated, was considered to be similar, as both were primarily sand dominated; however, almost twice as many sites were dominated by rocky fine substrate in electric seine sites compared with backpack sites (Table 1).

A total of 2,191 fish representing nine families and 25 species was collected during the study. The electric seine captured 1,036 fish and 22 species and backpack electrofishers captured 1,155 fish and 21 species. Species unique to electric seine sites included Shadow Bass Ambloplites ariommus, Redear Sunfish Lepomis microlophus, Florida Sand Darter Ammocrypta bifascia, and Eastern Mosquitofish Gambusia holbrooki (Table 2). Species unique to backpack electrofisher sites included American Eel Anguilla rostrata, Spotted Sucker Minytrema melanops, and Pirate Perch Aphredoderus sayanus (Table 2). Species unique to each gear constituted < 1% of the total catch.

Mean number of fish collected per site was similar between the electric seine (41.40 fish/site) and backpack electrofishers (46.20 fish/site [F_1,45 = 0.68, P = 0.4136]). Mean species richness for each site was also similar between the electric seine (6.12 species/site) and backpack electrofishers (6.20 species/site [F_1,45 = 0.04, P = 0.8499]). Simpson's dominance, Shannon's diversity, and Pielou's evenness were similar between gear types (Table 3). The rate of species accumulation relative to the number of sites was also similar between gear types (F_1,42 = 2.60, P = 0.11) (Figure 3).

The two-dimensional nonmetric multidimensional scaling ordination (stress = 0.17) indicated that fish assemblages collected with each gear type were similar in composition, as shown by the mixing of individual sites in ordination space (Figure 4). The permutation-based multivariate analysis of variance confirmed the nonmetric multidimensional scaling ordination result that fish assemblages collected with each gear type were similar (F = 0.57, P = 0.73).

Sampling shoal habitats in the Chipola River with two backpack electrofishers and an electric seine revealed no differences in fish catch overall or at the site level, or in the rate of species accumulation between the two gear types. Greater differences in total catch for some species like Weed Shiner Notropis texanus and Redbreast Sunfish Lepomis auritus likely reflected natural variation in catch between sites rather than the greater effectiveness of one gear type over the other. For example, 102 Weed Shiners (32% of species total) were collected in one backpack electrofisher site and 13 Redbreast Sunfish (57% of species total) were collected in one electric seine site, which made it appear that there were large differences in catch between the two gear types for those species. American Eels were easier to collect with backpack electrofishers because the anode could be actively placed next to the fish to maintain electric shock, whereas the anodes and cathodes of the electric seine could not be moved toward an individual fish. We sometimes observed American Eels being shocked with the electric seine yet avoiding capture; no eels were collected by the electric seine. It is unclear why we collected more centrarchids and ictalurids in electric seine sites and more cyprinids in backpack electrofisher sites.

We attempted to complete sampling with both gear types in a short time period and in similar habitats to minimize bias and strengthen comparisons in fish catch. We had planned to alternate sampling days with backpack electrofishers and the electric seine, but electrical connection problems with the electric seine early in the study forced us to sample all backpack electrofisher sites in the first 2 wk of the study period and all but two electric seine sites in the final 4 wk of the study period. Sampling occurred over a 5-wk period encompassing September, when water temperatures are relatively stable in north Florida. Therefore, we expect that fish movement to other habitats and reduced effectiveness of the electrofishing gears as a result of changing water temperatures did not occur. Rain events that started in the second week of the study led to differences in mean surface current velocity and depth in sites sampled between gear types. However, differences in mean depth among gears were minimal (1–12 cm), and we do not believe that overall differences in surface current velocity were substantial enough to bias results (i.e., we did not observe any differences in fish behavior or escapement due to the increase in current velocity). Limitations of our study include results based on a single year of data from a single stream using one range of electrical settings. Further, our number of sites per gear may have inflated sampled habitat differences. However, despite the limited scope of our study, we believe that our results provide a valid comparison between gear types because of the number of replicate samples.

The sites we sampled were also similar in regard to substrate and habitat. Eelgrass and interstitial space among rocks are two abundant fish habitats in Chipola River shoals. We detected no significant difference in mean percent area coverage of eelgrass between the gear types and we also conclude that substrate type values were comparable between the gear types. Although shoals contain extensive rocky substrate, sand/pea gravel was often the dominant substrate type because it is prevalent in between shoal rocks. Additionally, we believe that the statistical difference in percent coverage of filamentous algae was not biologically significant given that we observed few fish using filamentous algae as cover and differences in percent coverage were relatively small overall. On the basis of field observations we believe that habitats sampled by the two gear types were comparable.

The cost of each gear, required construction time, number of personnel needed, and setup time in the field plays a role in determining which gear researchers use when sampling. Gathering materials and constructing the electric seine took a considerable amount of time given the inexperience of the biologists in electric wiring and task of finding and ordering parts. We ordered parts and constructed the seine over a long time period, making an estimate of construction time difficult, but we estimate that it could be completed in 1 to 4 mo, depending on selected parts, part availability, and skill/knowledge of the personnel building the seine. We gathered price quotes from three companies for an electrofisher box and four companies for backpack electrofishers to provide a range of prices from which to compare the two gear types. The cost of two new pulsed direct-current backpack electrofishers, including anode, cathode, and batteries, excluding tax and shipping was between $12,312.00 and $25,700.00 The cost of building the electric seine, including the generator used in our study, electrofisher box capable of alternating current output (excluding tax and shipping), and electric seine parts was between $7,460.00 and $17,000.00. Therefore, compared with an electric seine, two new backpack electrofishers would be more expensive on average, with the cost difference depending on brand and model, but no construction time would be required.

Setup for sampling with the electric seine took more time and effort than that for backpack electrofishers. The electrical cord extending from the boat to the electric seine had to be periodically freed from rocks and eelgrass mats, and the need for a generator, electrofisher, and junction box added extra sampling gear as noted by Bayley et al. (1989). Sampling with the electric seine also required an additional crew member.

The small (4 × 5 m) block-netted sampling area might have limited our ability to detect differences in fish catch between the two gear types. One benefit of an electric seine is that it creates a large, continuous electric field, which reduces opportunities for escapement (Angermeier et al. 1991). Conversely, backpack electrofishers have a smaller electric field and are better suited for collecting fish associated with cover (Dauble and Gray 1980; Bayley et al. 1989). Using two backpack electrofishers in a relatively small, 4-m-wide sampling area may create a similar electric field as using one electric seine in the same size sampling area. Bayley et al. (1989) compared an electric seine and one backpack electrofisher in stream sections as wide as 10.4 m and determined that the electric seine more efficiently captured cyprinids and catostomids because of its ability to encircle fish, but we collected more cyprinids using backpack electrofishers. Burns (2007) also found that as site width increased, backpack units did not effectively cover as much of the sampling area as a parallel wire unit. Using only one backpack electrofisher in the same site width or two backpack electrofishers in a larger site width could produce results different from ours.

We determined no significant differences in the fish assemblages or rates of species accumulation between collections made with an electric seine and two backpack electrofishers in shoal habitats of the Chipola River. Although each gear type collected several unique species and different numbers of individual species, differences in total count were low, and neither gear type provided statistically better fish sampling results. Therefore, we consider the two gear types to be equally effective for sampling large river shoals when sites are block netted and narrow. The choice of gear in future studies on large river shoal habitat may be determined by project objectives, target species, available funds, equipment, or personnel. In general, backpack electrofishing required a smaller crew, and the equipment was more manageable. In contrast, the electric seine was more difficult to manage because of the size, number of parts, and larger crew required. The cost of a new electrofisher box and construction of an electric seine would be less on average than buying two new backpack electrofishers.

Table S1. Environmental variable data from all sites sampled with two backpack electrofishers or an electric seine in the Chipola River, Florida from August 28 to October 3, 2014. Included are date, gear type, shoal name, site number, depth measurements, flow measurements, primary and secondary vegetation types, dominant substrate, water temperature, dissolved oxygen, conductivity, latitude and longitude.

Found at DOI: http://dx.doi.org/10.3996/032017-JFWM-026.S1 (19 KB XLSX).

Table S2. Fish catch data from all sites sampled with two backpack electrofishers or an electric seine in the Chipola River, Florida from August 28 to October 3, 2014. Included are date, gear type, shoal name, site number, species, total count, centimeter group, total length, and weight.

Found at DOI: http://dx.doi.org/10.3996/032017-JFWM-026.S2 (41 KB XLSX).

This project was funded by Sportfish Restoration Grant FL F-F14AF00915—Florida Freshwater Fisheries Research. We thank Jordan Hults, Jason O'Connor, Chester Copperpot, Neil Branson, Cameron Bodine, and Reuben Smit for fieldwork assistance, and Justin Hill for help with the schematic drawing of the electric seine. Several reviewers greatly improved this paper, including Bland Crowder, Travis Tuten, Kim Bonvechio, and Chris Anderson.

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.