Comparison of four types of sampling gears for estimating age-0 yellow perch density

Abstract

To aid biologists in obtaining reliable and efficient estimates of age-0 yellow perch (Perca flavescens) abundance, we compared operational effort and catch characteristics (i.e., density, length frequencies, and precision) of four gear types (beach seines, benthic sleds, drop nets, and push trawls) in littoral habitats in two South Dakota glacial lakes. Gear types were selected on the basis that the volume of water sampled could be determined and, thus, density (number/m³) of age-0 yellow perch could be computed for each. Age-0 yellow perch were collected on three occasions with each gear in both lakes in August 2011. Differences in gear operational effort (i.e., the time required for set-up, deployment, and fish sorting) and age-0 yellow perch density and length frequencies were compared among sampling gears. Mean operational effort ranged from 9 to 65 min, mean density from 0.07 to 4.1 age-0 yellow perch/m³, coefficients of variation of mean age-0 perch density from 33 to 134, and number of samples required to estimate a 25% change in mean age-0 perch abundance from 22 to 305. Beach seines and drop nets may selectively sample larger age-0 yellow perch than benthic sleds and push trawls. We recommend use of beach seines or benthic sleds for sampling age-0 yellow perch in littoral habitats of glacial lakes. However, all sampling gears tested in this study have associated costs and benefits and one may be more suitable than another under different circumstances.

Keywords:

• yellow perch
• population density
• beach seine
• benthic sled
• push trawl
• drop net

Introduction

Yellow perch (Perca flavescens) are recreationally and ecologically important in northern United States Great Plains glacial lakes (Fisher et al. 1999). In South Dakota, USA, previous research on yellow perch has addressed the larval stage, relating larval perch abundance to suites of climatological variables (e.g., Ward et al. 2004) and evaluating relationships between larval perch abundance and recruitment and year-class strength (Anderson et al. 1998; Isermann and Willis 2008). However, little work has examined age-0 yellow perch dynamics, as development of appropriate sampling protocols has proven difficult (Fisher et al. 1999). Difficulties in sampling age-0 yellow perch and other fishes are often encountered due to patchy distribution and heterogeneous habitat use (Paradis et al. 2008; Dembkowski et al. 2011), dense submerged macrophyte coverage (Serafy et al. 1988; Dewey 1992), net avoidance owing to increased swimming capability with increasing fish size (Morse 1989; Gartz et al. 1999; Tischler et al. 2000), and other factors resulting from ontogenetic changes occurring in age-0 fishes (e.g., Whiteside et al. 1985; Post and McQueen 1988; Ruzycki and Wurtsbaugh 1999; Kolar et al. 2003).

Additional difficulties in sampling age-0 fish may also arise from inherent gear limitations. For example, gears designed to sample age-0 fish in open water habitats may not be suitable for sampling age-0 fish in densely vegetated littoral areas (Paradis et al. 2008). Another gear-related limitation is that different gears are often selective for different sizes of individuals within a cohort (Hubert 1996; Neumann and Allen 2007). In a comparison of age-0 shad (Dorosoma spp.) catch characteristics among different trawling gears, Michaletz et al. (1995) found that a Tucker trawl captured a higher proportion of larger shad than did a frame trawl with the same mesh size. Although the use of multiple gears is recommended to overcome individual gear limitations (Tischler et al. 2000), use of multiple gears may be accompanied by logistical complications because the use of multiple gears may require more specialized equipment or more personnel. Furthermore, gear standardization and comparison may prove difficult (Kjelson et al. 1975).

Standardized sampling protocols for postlarval (>25 mm total length (TL)) age-0 yellow perch (hereafter referred to as age-0 yellow perch) in South Dakota glacial lakes include shoreline seining for nearshore age-0 perch and bottom trawling for offshore age-0 perch (e.g., Fisher et al. 1999). Previous studies indicate that age-0 yellow perch in northeastern South Dakota glacial lakes typically are situated in nearshore areas (Lucchesi 1994; Dembkowski et al. 2011). However, no known evaluation of shoreline seining efforts for age-0 yellow perch in northern Great Plains glacial lakes exists nor have catch characteristics of seine hauls been compared with those of other potentially more effective and reliable gears, which may provide more accurate and precise estimates of age-0 yellow perch abundance and require less effort than seining operations.

The goal of this study was to identify the most suitable gear for use in future studies relating to age-0 yellow perch recruitment in glacial lakes. We evaluated four different gears (beach seine, benthic sled, drop net, and push trawl) for their efficiency and reliability in obtaining quantitative estimates of age-0 yellow perch abundance and size structure in two northeastern South Dakota glacial lakes. Although other studies have compared catch characteristics and precision of different gears for sampling age-0 yellow perch (e.g., Paradis et al. 2008), this study is the first to evaluate a benthic sled and push trawl for sampling age-0 yellow perch in littoral areas of glacial lakes.

Methods

Study lakes

This study was conducted in two northeastern South Dakota, USA, glacial lakes during August 2011. Pickerel Lake (Day County) is mesotrophic (trophic state index (TSI) = 48.8; Carlson 1977), has a surface area of 397 ha, mean depth of 4.8 m, and shoreline development index of 2.2 (Stueven and Stewart 1996). Clear Lake (Marshall County) is eutrophic (TSI = 52.6), has a surface area of 474 ha, mean depth of 3.8 m, and shoreline development index of 1.5. Pickerel Lake has a relatively steep basin morphometry compared to Clear Lake. Human development (i.e., cabins and homes) has resulted in the elimination of most natural riparian vegetation at both lakes. Littoral habitat consists largely of rock and sand substrate interrupted by sparse submerged macrophytes. Submerged macrophytes in Pickerel and Clear Lakes are predominantly sago pondweed (Stuckenia pectinata) and coontail (Ceratophyllum demersum) with limited emergent stands of bulrushes (Scirpus spp.) and cattails (Typha spp.) in shallow, protected, and undeveloped areas of the lakes. Yellow perch are listed as a primary management species in both lakes (various unpublished South Dakota Department of Game, Fish and Parks collection reports).

Data collection

Age-0 yellow perch were sampled from Pickerel and Clear Lakes using four different sampling gear types (beach seine, benthic sled, push trawl, and drop net). Preliminary sampling indicated that age-0 yellow perch in Clear and Pickerel Lakes were distributed around patches of submerged vegetation in nearshore areas and maintain a mostly demersal existence (Dembkowski et al. 2011). Thus, this study focused on gear performance only in littoral areas of Pickerel and Clear Lakes. Gears were selected on the basis that they could effectively sample demersal habitats and that water volume and, thus, density (number/m³) of age-0 yellow perch could be computed for each. Each lake was sampled on 1 August 2011, 8 August 2011, and 15 August 2011. At each lake on each date, one transect was sampled with the beach seine, benthic sled, and push trawl; and two locations were sampled with the drop net. More samples were collected with the drop net to equalize the volume of water sampled with each gear. No samples were collected with the drop net on 15 August 2011 because of temporal sampling constraints. Thus, a total of three beach seine, benthic sled, and push trawl samples, and four drop-net samples were collected at each lake.

A 27.4 × 1.8 m bag seine (3-mm bar mesh) was deployed in a circle by wading, using an onshore point as the starting and ending point. To enclose the sample, the lead-line was pulled toward shore from both ends until the collected fish were confined in the seine bag. The volume of water sampled by the beach seine was computed as the area of the circle enclosed by the net during each deployment (range = 28.3–63.5 m²) multiplied by mean depth within the circle during the deployment.

The benthic sled (sensu Niles and Hartman 2007) was 3-m long and was constructed of 3-mm bar mesh net attached to a rigid, galvanized steel frame (1.2 × 0.9 m). The frame and net assembly were fastened to two 1.2-m galvanized steel skis separated by a 102 mm diameter polyvinyl chloride (PVC) roll bar, which enabled the sled to roll over obstructions in the water. The skis of the benthic sled were filled with sand and capped to provide ballast. The benthic sled was outfitted with towing bridles that were attached to lines running down booms extending outward from the bow of a 5.5-m boat equipped with a 115-hp outboard motor and was pushed along each transect for a target of 100 m at 1–2 m/s. A main winch located under the bow deck was attached to the boom arm assembly to lower and raise the benthic sled into and out of the water. Depth of the benthic sled was controlled by two winches bolted to either side of the boom assembly to allow for better navigation along benthic contours. The volume of water sampled by the benthic sled was computed as the distance trawled multiplied by the surface area of the mouth of the net (1.1 m²).

The push trawl consisted of a 3-m long, 3-mm bar mesh bottom trawl with a zipper-style cod-end, and a 3.75-m head rope. Because the push trawl did not have a rigid frame, a rope was threaded around the mouth of the net in a rectangular fashion, allowing the mouth to open and maintain a surface area of 1.1 m², after which a stop-knot engaged and prevented the mouth from opening any further. We assumed that the mouth of the push trawl opened to its fullest extent during each transect throughout the study. On 1 August and 15 August at Pickerel Lake, submerged vegetation at the selected transect location was too thick, and the mouth of the push trawl collapsed. In these instances, the sample was not collected and a different, less-vegetated location was selected nearby. The push trawl was also attached to the boom arm assembly and pushed along each transect for a target of 100 m at 1–2 m/s. The volume of water sampled by the push trawl was computed as the distance trawled multiplied by the surface area of the net mouth (1.1 m²).

The drop net (Kahl 1963) consisted of a cast net (6.2-m diameter; 5-mm bar mesh) suspended from a 4.8 m² floating 20-mm-diameter PVC frame. Each corner of the PVC frame was connected to a 1.2-m vertical piece of aluminum pipe with a buoy attached to the surface end. The edges of the cast net were pinched in four areas to form corners and attached to a 100-mm long piece of steel pipe (hereafter referred to as collars), which were slightly larger in diameter than the corner posts. The collars were fitted over the corner posts and rested on top of a lynchpin attached to a trip-cord, enabling the drop net to be deployed remotely (Dewey 1992). On removal of the lynchpins by pulling on the trip-cord, the collars slid down the corner posts, and the net fell to the lake bottom. The sample was enclosed within the cast net by tightening the cast net hand line from above. The drop net was set and left undisturbed for 120 min before removing the lynchpins. During net deployment, the boat was positioned approximately 50 m away from the drop net to avoid frightening fish toward or away from the net. One person would approach the net from the shoreline or water (depending on density of riparian vegetation), pull the trip-cord, and ensure that the net had released from the corner posts properly. The cast net hand line was then passed to another person on the boat who would tighten the hand line, enclose the sample, and lift the entire net into a trough on the boat. The volume of water sampled by the drop net was computed as the surface area of the drop net (4.4 m²) multiplied by mean depth of the water column directly below the net.

Densities of age-0 yellow perch were reported as the number of age-0 perch/m³. A random subsample of assumed age-0 yellow perch captured on each sampling occasion with each gear was measured to TL (mm), preserved in 75% ethanol, and transported to the laboratory where sagittal otoliths were removed and ages were estimated by two independent readers to verify that collected fish were age-0. Densities and length frequencies of age-0 yellow perch, sampling precision, and operational effort in the field (i.e., the time required for set-up, deployment, and fish sorting for each gear) were compared among gears at Pickerel and Clear Lakes.

Analysis

Potential differences in age-0 yellow perch densities among gears at each lake were evaluated using a one-way repeated measures analysis of variance (ANOVA). Post-hoc multiple comparisons were performed using a Student-Newman-Keuls (SNK) test. Although we had a relatively small sample size, parametric analyses of variance and post-hoc multiple comparisons are relatively robust to small sample size and deviations from normality (Brenden et al. 2003). The coefficients of variation (CV) and standard error (SE) of age-0 yellow perch densities estimated with each gear were used as measures of sampling precision. Length frequencies of fish were compared among gears at each lake using Kolmogorov-Smirnov (K-S) tests (Neumann and Allen 2007).

The sample size required to detect a 25% difference in mean age-0 yellow perch density was estimated for each gear in each lake. Estimation of sampling effort needed to detect a predetermined change in density is important because it can aid in determination of the feasibility of a selected sampling protocol. Power analysis calculations to estimate sample size were performed using the formula suggested by Snedecor and Cochran (1989),

where n is the estimated sampling effort, z _α is the z-distribution deviate for the probability of a type I error at a given level of significance, z _β is the z-distribution deviate for the probability of a type II error at a given level of statistical power, s is the standard deviation of the mean age-0 yellow perch catch per effort for each gear at each lake, and d is the specified effect size. Sample size estimation was performed at an alpha of 0.10, beta of 0.20, and power (1-β) of 0.80.

Mean operational effort was compared among gears at each lake using an ANOVA. Because the drop net was left undisturbed for 120 min after setting before we returned to collect the sample, estimates of true operational effort would be artificially over-inflated by the time required for the net to be left undisturbed. Thus, 120 min was subtracted from the recorded times in the field, and the resulting numbers were used in the comparison. If an overall statistically significant difference was detected among gears, post-hoc comparisons to evaluate individual gear differences were made using an SNK multiple comparison test. All comparisons were assessed for statistical significance at α = 0.10, and significance levels were adjusted using Bonferroni corrections to control overall experimentwise error rates. Adjusted significance was designated at α = 0.03. All statistical analyses were performed using the Statistical Analysis System software package (SAS Institute 2010).

Results

Density comparisons

Densities of age-0 yellow perch were variable among gears at each lake (Table 1) and were significantly different among gears at Pickerel Lake (F _5,7 = 8.23; p = 0.008) and Clear Lake (F _5,7 = 6.90; p = 0.01). At Pickerel Lake, the beach seine (mean = 3.04 yellow perch/m³) provided the highest density estimate, followed by the drop net (mean = 1.04 yellow perch/m³), push trawl (mean = 0.09 yellow perch/m³), and benthic sled (mean = 0.07 yellow perch/m³; Table 1). Differences in age-0 yellow perch densities between the beach seine and the benthic sled and between the beach seine and push trawl were significant. Although the beach seine provided a higher age-0 yellow perch density estimate than the drop net, this difference was not significant. CV and SEs of mean age-0 yellow perch densities ranged from 53 (beach seine) to 120 (drop net), and 0.03 (benthic sled) to 0.93 (beach seine), respectively.

Table 1. Mean density (standard error) of age-0 yellow perch, operational effort, coefficient of variation of mean densities (CV), and the number of samples (n) required to detect a 25% change in mean age-0 yellow perch abundance for four different gears in Pickerel and Clear lakes, South Dakota, during August 2011.

Lake	Gear	Fish density (number/m^3)	CV	n	Operational effort (min)
Pickerel	Beach seine	3.04 (0.93)^a	53	40	23 (1.8)^b
	Benthic sled	0.07 (0.03)^b	79	95	12 (1.0)^b
	Push trawl	0.09 (0.06)^b	106	166	13 (2.1)^b
	Drop net	1.04 (0.63)^b	120	205	45 (9.3)^a
Clear	Beach seine	4.10 (0.71)^a	33	22	23 (2.1)^a,b
	Benthic sled	0.65 (0.39)^b	79	90	13 (1.5)^b
	Push trawl	1.64 (0.82)^b	86	106	17 (2.1)^a,b
	Drop net	0.07 (0.05)^b	134	305	28 (4.3)^a
Note: For each lake, means with the same letter(s) are not significantly different at α = 0.10.

At Clear Lake, the beach seine (mean = 4.10 yellow perch/m³) provided the highest estimate of age-0 yellow perch density, followed by the push trawl (mean = 1.64 yellow perch/m³), the benthic sled (mean = 0.65 yellow perch/m³), and the drop net (mean = 0.07 yellow perch/m³; Table 1). Differences in age-0 yellow perch density estimates between the beach seine and the other three gears were all significant, with the beach seine providing greater estimates in all cases. CV and SEs of mean age-0 yellow perch catch rates ranged from 33 (beach seine) to 134 (drop net), and 0.05 (drop net) to 0.82 (push trawl), respectively.

Sample size estimation

In both lakes, the beach seine required the fewest samples to detect a 25% change in mean age-0 yellow perch density, followed by the benthic sled, push trawl, and drop net (Table 1). Effort required to detect a 25% change in mean density ranged from 40 (beach seine) to 205 (drop net) at Pickerel Lake and from 22 (beach seine) to 305 (drop net) at Clear Lake. At Pickerel Lake, the benthic sled required about twice, the push trawl about four times, and the drop net about six times as many samples as the beach seine. At Clear Lake, the benthic sled required about four times, the push trawl about five times, and the drop net about 15 times as many samples as the beach seine. The number of samples required to detect a 25% change in mean density was lower at Clear Lake for all gears except the drop net.

Length frequencies

Age-0 yellow perch length frequencies were variable among gears at Clear Lake, but not at Pickerel Lake (Figure 1). At Clear Lake, mean TL of age-0 yellow perch collected with the beach seine was similar to that collected with the drop net, and mean TL collected with the benthic sled was similar to that collected with the push trawl. In addition, mean TL of age-0 yellow perch collected with the beach seine and drop net was generally greater than those collected with the benthic sled and push trawl (Figure 1). At Pickerel Lake, mean TL of age-0 yellow perch was similar among all gears (Figure 1).

Figure 1. Length-frequency histograms and mean total lengths for age-0 yellow perch collected during August 2011 in Pickerel Lake (left panel) and Clear Lake (right panel) with a beach seine, benthic sled, drop net, and push trawl. N = number of individuals; TL = total length (mm); SE = 1 standard error of the mean total length. For each lake, length frequencies of gears with the same letter did not differ significantly (Kolmogorov-Smirnov tests; p > 0.03).

At Clear Lake, length frequencies differed significantly between the beach seine and benthic sled (K-S = 0.32; df = 174; p < 0.001), beach seine and push trawl (K-S = 0.32; df = 210; p < 0.001) benthic sled and drop net (K-S = 0.23; df = 105; p < 0.001), and push trawl and drop net (K-S = 0.20; df = 141; p < 0.001), but not between the benthic sled and push trawl (K-S = 0.06; df = 232; p = 0.42) or beach seine and drop net (K-S = 0.09; df = 83; p = 0.50), suggesting that beach seine and drop-net samples contained larger age-0 yellow perch than benthic sled and push trawl samples (Figure 1). No significant differences in length frequencies were detected among gears at Pickerel Lake (Figure 1).

Operational effort

Mean operational effort in the field differed significantly among gears at Pickerel Lake (F _3,9 = 6.97; p = 0.01) and Clear Lake (F _3,9 = 4.51; p = 0.03; Table 1). At Pickerel Lake, mean operational effort for the drop net was significantly greater than that for the beach seine, benthic sled, and push trawl. At Clear Lake, mean operational effort for the drop net was significantly greater than that for the benthic sled. Other differences in operational effort were noted among the gears, but none were statistically significant (Table 1).

Discussion

Beach seines, benthic sleds, push trawls, and drop nets represent a diverse array of potential gears for obtaining quantitative estimates of age-0 yellow perch abundance. Many tradeoffs were observed with respect to catch characteristics and other measured variables for each gear. For example, at Pickerel Lake, the beach seine required half as many samples, but twice as much operational effort as the benthic sled to detect a 25% change in mean age-0 yellow perch density. Thus, our selection and recommendation of sampling gears is based on the collective results rather than individual catch characteristics such as only catch rates or precision. On the basis of the collective results and on our study objectives and sampling conditions, we recommend use of beach seines or benthic sleds for sampling age-0 yellow perch in littoral areas of glacial lakes.

All gears evaluated by this study have associated advantages and disadvantages as sampling gears for age-0 fish. Beach seines are commonly used because of their relative ease of use in shallow systems with smooth bottoms (Hayes et al. 1996), generally inexpensive cost (a beach seine such as the one used in this study can be purchased for approximately US$600), and minimal equipment requirements (i.e., no specialized boat is needed to operate a beach seine). However, beach seines may be impractical in systems with underwater obstructions (i.e., large boulders or tree stumps) that cause the net to become snagged or entangled or systems with dense aquatic macrophytes that may slow or completely block the seine haul (e.g., Serafy et al. 1988). Similarly, seining may prove ineffective in lakes with steep basin morphologies where nearshore water depth may exceed the height of the seine. Furthermore, seining data often do not accurately represent actual abundance because not all fish present in an area are captured and some species may display more or less avoidance capabilities than others (Lyons 1986). Although sampling difficulties owing to net avoidance may not be exclusive to seining, net avoidance and sampling inefficiencies of benthic sleds, drop nets, and push trawls have been studied to a much lesser extent than those of beach seines (e.g., Serafy et al. 1988; Lorenz et al. 1998). In our study, and probably for most shoreline seining operations, seining required one or two people to leave the boat to operate the seine, which may have biased our samples by pushing fish toward or away from the area to be seined. In addition, results of this study indicate that the beach seine may be selective for larger age-0 yellow perch, but because we did not determine actual cohort size structure at Clear Lake, size-selectivity of the gears could not be estimated (Allen et al. 1999). Our findings of potential selectivity for larger age-0 yellow perch contrast with those of Bayley and Herendeen (2000) and Allen et al. (1992), who reported that seine capture efficiency decreased for larger and faster fish.

An important factor in any study using quantitative estimates of fish abundance is the precision and accuracy of estimates of area or volume sampled. Although the beach seine may be relatively simple to operate, inexpensive to purchase and maintain, and provided the highest age-0 yellow perch density estimates, estimation of the volume of water sampled is somewhat subjective and may lead to extreme under- or over-estimation of actual abundance. Although the method of estimating volume sampled in this article has been used successfully in other studies (e.g., Paradis et al. 2008), we urge that caution be taken when using this method.

The benthic sled and push trawl are both designed to actively ride along the substrate and capture benthic and demersal fishes (Herke 1969; Rogers 1985; Niles and Hartman 2007). Both gears provided low but more precise estimates of mean age-0 yellow perch density and similar length-frequency distributions. Operational effort in the field for both gears was minimal compared to the beach seine and drop net, but the benthic sled required less sampling effort than the push trawl to detect a 25% change in mean age-0 yellow perch abundance. However, both gears require specialized boom and winch assemblies and a large motor (>50 hp) to push the nets along the substrate. The benthic sled and push trawl are also limited to nearshore areas in depths <2 m. At greater depths, the strain on the boom and winch assembly becomes excessive and there is a risk of the cod-end of the nets becoming fouled in the boat propeller. The benthic sled and push trawl can be effectively operated by two people, and operation does not require leaving the boat. The benthic sled may be preferable to the push trawl in areas with dense aquatic macrophytes or woody structure as the rigid frame of the benthic sled can withstand forces that otherwise cause the mouth of the push trawl to collapse. Another advantage of the rigid frame is that its surface area is a constant and can simply be multiplied by the distance trawled to provide an accurate and precise measure of water volume sampled. Although there are relatively few data with which to compare our benthic sled and push trawl findings, we are encouraged by their performance as potential sampling gears for age-0 yellow perch.

The drop net used in this study was a modification of that used by Kahl (1963) and was designed to enclose a sample by dropping from the surface of the water to the substrate. Operating the drop net with our modifications proved difficult. Although other studies have used various drop-net designs with success (e.g., Kahl 1963; Kjelson et al. 1975; Hoagman 1977; Evans and Tollmark 1979; Lorenz et al. 1997), none have evaluated efficiency in northern Great Plains glacial lakes. Similar to the beach seine, the drop net may be size-selective for larger age-0 yellow perch, but the tendency of the drop net to capture larger individuals may have been a function of mesh size. The drop net had 5-mm bar mesh, whereas all other gears had 3-mm bar mesh. The drop net was difficult to set from a boat and required three people to set even while wading in calm conditions. In windy conditions (>24 kph), the net would come unsettled from its anchored position, and the lynchpins would pull out, causing the net to drop prematurely before the 120 min settling period had expired. When unimpeded, the drop net fell to the lake bottom, enclosing a discrete sample of fish and an easily measurable volume of water. However, the presence of underwater obstructions (i.e., tree branches or thick aquatic vegetation) hindered the ability of the net to completely enclose the sample and could have allowed some fish to escape. Submerged aquatic macrophytes may also slow the descent of the net through the water column, which could also have affected capture efficiency (Kushlan 1981).

Choice of gears for a given project or sampling protocol is often lake- and study-specific. Results herein provide information regarding catch rates, TL ranges, operational effort, and approximate sample sizes needed to estimate a 25% change in age-0 fish abundance that managers can use to select an appropriate gear tailored to their study objectives and sampling conditions. Given our study objectives and sampling conditions, we recommend that the benthic sled or beach seine be used to quantitatively sample age-0 yellow perch in shallow, littoral areas of glacial lakes. The benthic sled provides a more discrete and quantitative sample and should be used when sampling precision is of concern. The push trawl may be of use in lakes with no or sparse aquatic macrophytes. We do not recommend use of the drop net under its current design as it requires substantial effort in the field and samples a relatively small volume of water compared with most other gears. However, we acknowledge that all sampling gears evaluated have associated costs and benefits, and one gear type may be more suitable than others under certain circumstances.

Acknowledgments

Funding for this project was provided by Federal Aid in Sport Fish Restoration funds (Project F-15-R; Study 1518) administered by South Dakota Department of Game, Fish and Parks, and South Dakota State University. We thank B. Graff and J. Lindgren for field assistance, R. Klumb and D. Shuman for the use of push-trawling equipment, B. Blackwell for the use of seining equipment, M. Kaemingk for provision of the drop-net equipment, and J. Breegemann, D. Deslauriers, C. Hayer, M. Kaemingk, and T. Rapp for helpful reviews of earlier drafts of the manuscript.