Comparison of Two Escapement Designs for Western Painted Turtles Captured in Modified Fyke Nets

Many reptile species are imperiled. Böhm et al. (2013) noted nearly 1 in 5 species, of the 1,500 studied, are threatened with extinction; and freshwater environments, tropical regions, and oceanic islands had the highest proportion of threatened reptile species. The primary threats to reptiles are habitat degradation, pollution, overexploitation, introduction of invasive species, and disease (Wilcove et al. 1998; Gibbons et al. 2000). An often overlooked threat to aquatic reptiles is bycatch mortality. Bycatch has been extensively studied in marine systems but is poorly understood in freshwater environments. Raby et al. (2011) noted only 37 of 1,152 papers on bycatch were devoted to freshwater systems. In freshwater environments, high rates of bycatch mortality have been observed in a variety of passive fishing gears (Barko et al. 2004), in commercial fisheries (Larocque et al. 2012a) and in fisheries management sampling (Moos and Blackwell 2017). However, studies on freshwater turtles have indicated that population declines can occur with only a slight increase in adult mortality (Brooks et al. 1991; Congdon et al. 1993; Midwood et al. 2015).

Research directed at reducing freshwater fisheries bycatch, although sparse, has increased in recent years. Much of that research has focused on reducing bycatch in commercial fisheries (Lowry et al. 2005; Larocque et al. 2012b). The research on bycatch reduction that is applicable to fish population sampling studied catch dynamics for a limited number of fish species (Fratto et al. 2008a) or has indicated noteworthy impacts on catch dynamics of target species (Fratto et al. 2008b; Larocque et al. 2012b). More research is necessary to refine bycatch reduction methodology during biological sampling.

To date, research to reduce bycatch through gear alteration has focused on exclusion and escapement. Exclusion is altering gear to prevent the capture of nontarget species. Tested exclusion devices include exclusion bars, panels, and rings (Lowry et al. 2005; Fratto et al. 2008a, 2008b; Larocque et al. 2012b; Cairns et al. 2013). Escapement is altering fish capture gear to allow captured nontarget species to escape while retaining the target species. Escapement designs have included chimneys (Fratto et al. 2008a; Larocque et al. 2012b), loose-weave mesh panels (Fratto et al. 2008b), and escapement chutes (Lowry et al. 2005). Designing successful exclusion and escapement methods involve understanding the physical and behavioral differences of target and nontarget species and altering sampling gear to take advantage of the observed differences.

Exclusion designs work best when the target is one species, with a specific body shape or size, as is common in commercial fisheries. Exclusion modifications have proven effective for commercial fisheries targeting Long-finned Eel Anguilla reinhardtii (Lowry et al. 2005), Common Carp Cyprinus carpio (Lowry et al. 2005), and blue crab Callinectes sapidus (Guillory and Prejean 1998). However, the goal of many fish population assessments is to catch a representative sample of species from a fishery, including all sizes of targeted species. Fyke nets have been recommended for targeting crappie Pomoxis species and sunfish Lepomis species (Miranda and Boxrucker 2009).

Inherent biases affect all passive sampling gears; however, a bycatch reduction device (BRD) may introduce additional capture biases that need to be understood and accounted for before conclusions can be made. Unfortunately, finding bycatch reduction devices that introduce minimal biases has been challenging. For instance, in their evaluation of a constriction rectangle exclusion device, Cairns et al. (2013) noted, relative to control nets, a significant reduction in the catch per unit effort for total target species, total fish bycatch, and Pumpkinseed Lepomis gibbosus, while concurrently documenting a significant increase in mean total length for Bluegill Lepomis macrochirus. Larocque et al. (2012b) observed a significantly smaller mean total length of Bluegill captured in fyke nets having escapement bars compared with control nets. A significant difference in length distribution of White Crappie Pomoxis annularis and Bluegill was observed in Wisconsin-type fyke nets modified with exclusion bars versus control nets (Fratto et al. 2008b). Escapement may be more applicable than exclusion for reduction of bycatch during fish population assessments because there may be differences in the way fish and turtles react to and utilize escapement devices. However, escapement designs have resulted in loss of targeted fish. Lower fish catches were observed in modified nets with loose-weave mesh and escapement chimneys compared with control nets (Fratto et al. 2008a). Cairns et al. (2016) noted an 18% decrease in target fish captures in their escapement design.

Due to limitations of previously tested BRDs, additional designs need to be tested to identify effective methods for reducing bycatch during fish sampling. Thus, our goals were to design and test two BRDs to determine 1) the effectiveness of each design for escapement of western painted turtles Chrysemys picta bellii, and 2) potential impacts of each design on catch dynamics during simulated fish population assessments. An effective BRD would provide a tool for reducing the impact of bycatch on freshwater turtle populations while ensuring minimal biases are introduced in fish sampling.

We constructed BRD devices on modified fyke nets of 1.9-cm mesh (bar measure), with a 15.2-m lead, two 0.9 × 1.5 m rectangular frames, and three 0.9-m-diameter hoops. Design A had a 20.3 × 12.7 cm hole, with a maximum stretch size of 25.4 × 22.9 cm, cut in the top of the cab area between the first and second hoop at the front end of the cab (Figure 1). We sewed a 35.6 × 22.9 cm mesh flap inside of the cab around the exterior opening. The flap sloped toward the interior throat, and we tied it to the throat to keep the mesh stretched tight. A single opening at the front end of the flap was the only entrance to the exterior opening. We hypothesized the flap would make it difficult for fish to escape through the exterior opening. However, turtles seeking air would swim to the top of the net and find the flap and crawl through to the exterior opening. Our design A was similar to a device used by Lowry et al. (2005) with an opening cut in the top of the carp trap and netting partially blocking the opening. In their design, large mesh netting prevented Common Carp from utilizing the opening while allowing turtles to pass through.

Design B consisted of a 27.9 × 12.7 cm hole, with a maximum stretch of 33.0 × 20.3 cm, cut in the top of the cab area between the first and second hoop at the front end of the cab (Figure 2). We sewed a mesh throat 91.4 × 40.6 cm on the exterior of the cab over the opening sloping back to the rear of the cab and tied it to the last hoop. The throat is not held open but is compressed through the rear half of the throat. Anything trying to escape must force its way through to the opening. We hypothesized fish, especially laterally compressed fish, would be reluctant to swim through the compressed opening, but turtles would be able to squeeze through the opening. We did not find any tested BRD designs similar to design B, therefore; we consider it to be a novel design. We used <$10 (U.S. currency) in material and supplies (net mesh, string and floats) for construction of either BRD design per net. The time required for one person to build and install each BRD was approximately 2 h per net.

To determine the number of turtles and fish that escaped each net design, we installed a holding pen above each BRD (Figures 1 and 2). We attached floats to the holding pen to keep it from resting on the net; and the shape and size of the pen prevented it from contacting the BRD, thereby preventing it from interfering with escapement. We placed a mesh throat at the top of each holding pen to facilitate removal of fish and turtles and tied it shut when the net was set.

We conducted trials from 2012 to 2014 on two small impoundments—Pigors Lake (Brown County, South Dakota) and Pierpont Lake (Day County, South Dakota)—with two trials completed in 2012, three in 2013, and one in 2014. We conducted one trial on Pigors Lake and five on Pierpont Lake. The day before trials were to be conducted we set modified fyke nets in shallow areas of each lake to catch turtles to be introduced into nets for testing escapement. We deployed modified fyke nets overnight and removed turtles in the morning. We moved turtles from capture nets to the trial nets with no rest period. With no rest period, stress from capture could have affected escapement rates. Condition of turtles was important; therefore, we set each capture net in shallow water to allow turtles access to air, thereby reducing net-induced stress on captured turtles. We conducted trials only when surface water temperature was <20°C. Our primary focus was reduced stress on turtles, and at temperatures ≥20°C stress and mortality rates increase for western painted turtles captured in modified fyke nets (Moos and Blackwell 2017). We did not record condition and mortality of turtles in this study. Conducting trials in water <20°C could affect escapement of ectotherms, including fish and turtles, because of sluggish behavior due to slower metabolic rate. Conducting trials at different temperatures would likely yield varied escapement results for each species of fish and turtle. We conducted five trials during the month of May and one trial in September.

To test turtle escapement from each BRD design, we introduced 10 turtles into each net. We measured the curved carapace length with a flexible, metal measuring tape and determined the sex of each turtle by observation of secondary sexual characteristics including foreclaw length and precloacal tail length (Ernst 1971; Gibbons and Lovich 1990; Rowe 1997; Table S1, Supplemental Material). Sex of western painted turtles <10 cm curved carapace length is difficult to differentiate using secondary sexual characteristics; therefore, we classified them as unknown (Table S1, Supplemental Material). We did not control the sex ratio of turtles placed in the nets and it varied substantially. We used three nets of each BRD design in each trial, with all six nets set in close proximity to each other to minimize differences in depth and shoreline slope. All nets were completely submerged to force turtles to find the BRD, and we tied shut the throat in the cab of each net so the only avenue of escape was through the BRD.

We used two time trials: 4-h and 24-h. We used a 4-h trial because the amount of time western painted turtles can survive submerged decreases with increasing water temperature (Herbert and Jackson 1985). We assumed short-term trials would indicate the proportion of turtles that potentially would escape before mortality occurred in warm water temperatures. We set the long-term trials overnight for approximately 24 h to simulate the typical duration of fish population assessments.

We completed three each of the 4-h and 24-h trials over the course of this study. We excluded one net set of design A during the 24-h trials because of net damage. We set nine design A and nine design B nets during the 4-h trials and set eight design A and nine design B nets during the 24-h trial. We calculated escapement rates [(no. escaping/total no.) ×100] for each net set during the trials and reported the range, number escaped, and mean escapement rates. We reported male and female turtle escapement rates by net design for each time trial. We did not conduct statistical analysis of how size and sex influenced escapement because of relatively small sample size and constraints of project design.

We used two natural lakes—Enemy Swim Lake (Day County, South Dakota) and North Buffalo Lake (Marshall County, South Dakota)—and two small impoundments—Richmond Lake (Brown County, South Dakota) and White Lake (Marshall County, South Dakota)—to test fish escapement and turtle escapement in the presence of fish from 2012 to 2014. We selected these lakes because they contain diverse fish communities, including species typically targeted with modified fyke nets. Similar to the turtle escapement trials, we attempted to sample when water temperatures were <20°C; however, we conducted the White Lake trial at 21°C. We conducted one trial during 21–22 May 2013 on Richmond Lake. We conducted two trials in June, with White Lake sampled during 7–8 June 2012, and Enemy Swim Lake sampled during 3–5 June 2014. We conducted two trials in September, with North Buffalo sampled during 25–26 September 2012, and Enemy Swim Lake sampled during 17–19 September 2013.

At three locations on each lake, we set one net of BRD design A and B approximately 75 m apart. We randomly assigned net orders to the three test areas. We set nets overnight to simulate the conditions of fish population assessments. We checked nets the following day and recorded data including the total number (Table S2, Supplemental Material) and size of all fish and turtles captured in the holding pen and cab areas of each net (Tables S3–S7, Supplemental Material). We did not record condition and mortality of fish and turtles during this study. We removed all fish and turtles and then reset the nets in the same locations and order as on the first day. We followed the same procedure the second day, with the exception that we removed nets from the lake after data collection. To account for poor catches, we ran nets for 3 d during each of the two trials on Enemy Swim Lake to gather additional data. During the five trials, we completed 34 overnight net sets with design A and 35 overnight net sets with design B. We removed two net sets of design A and one of design B because of net failure. We determined escapement rates for 13 fish species and western painted turtles; however, sample sizes were small for several fish species. We combined catch and escapement of each species for all trials and reported a cumulative escapement rate [(no. escaping/total no.) × 100] (Table 1).

We recorded the total length of fish and curved carapace length of turtles in 1.0-cm size classes (e.g., 10.0–10.9 cm). We compared length frequency of specimens in the holding pen with those remaining in the cab within each net design for Black Bullhead Ameiurus melas, Black Crappie Pomoxis nigromaculatus, Bluegill, Rock Bass Ambloplites rupestris, and western painted turtles using the Kolmogorov–Smirnov test. We used raw data (Table S3–S7, Supplemental Material) and conducted statistical analysis in SYSTAT 13 (Systat Software Inc. 2009) with alpha set at 0.05. However, to account for variable sample sizes, we presented data using percentage of catch by 1.0-cm size class in the holding pen and net cab for each species (Figures 3–5) to improve clarity of results.

To determine whether reentry of fish and turtles into the cab from the holding pens affected the observed escapement rates, we ran one trial in 2014 with the three nets of each escapement design. We tied the throat in the cab shut, preventing escapement from the cab; and we placed 15 Black Bullheads, 20 Bluegills, and 6 western painted turtles into the holding pen of each net. We checked nets after 1 h and recorded the number of fish and turtles that reentered the cab. We used a 1-h trial for reentry to get a snap-shot of the potential for fish and turtle movement back into the cab of the fyke net.

Escapement rates for western painted turtles during the 4-h trials for design A, when considering individual net sets, ranged from 60 to 100%, with 64 of 90 turtles (71.1%) escaping; whereas, escapement ranged from 0 to 100% among individual net sets for design B, with 50 of 90 turtles (55.6%) escaping. Escapement rates during the 24-h trials for design A ranged from 50 to 80% among individual net sets, with 51 of 80 turtles (63.8%) escaping; whereas, escapement from design B ranged from 20 to 90% among individual net sets, with 45 of 90 turtles (50.0%) escaping. The mean curved carapace length of male and female turtles used in the trials were 16.5 cm (SE = 1.76) and 18.0 cm (SE = 2.84), respectively. Escapement rates of male turtles during the 4-h trials were 73.2% from design A and 66.7% from design B. Female escapement rates were 66.7% and 61.5% from designs A and B, respectively. During the 24-h trials escapement rates for males were 66.7% from design A and 52.9% from design B. Female escapement rates were 61.5% and 47.1% from designs A and B, respectively.

Design A had escapement rates ranging from 0 to 33% among fish species captured (Table 1). We observed escapement rates of 0–27% for centrarchids, and western painted turtles escaped at the highest rate (49%; Table 1). We observed reentry of Bluegills (11%), Black Bullheads (70%), and western painted turtles (44%) from the holding pen into the fyke net in design A. Comparing the length frequency distribution of individuals in the holding pen and cab of design A, we observed no difference for Black Bullhead (P = 0.09; Figure 3), Black Crappie (P = 0.08; Figure 3), and Rock Bass (P = 0.56; Figure 4). We observed differences for Bluegill (P < 0.01; Figure 4) and western painted turtles (P = 0.05; Figure 5). The size structure was skewed toward smaller Bluegill and larger turtles in the holding pen compared with the cab.

Design B escapement rates ranged from 0 to 65% among fish species captured, with low escapement rates for most species, including 0–12% observed escapement of centrarchids (Table 1). Black Bullhead (20%), White Sucker (43%), and Channel Catfish (65%) escaped at moderate to high rates (Table 1). Only 11% of captured western painted turtles escaped (Table 1). Reentry was minimal, with 3% of Bluegill and 0% of Black Bullhead or western painted turtles reentering the fyke net in design B. Comparing the length frequency distribution of individuals in the holding pen and cab of design B, we observed no difference for western painted turtles (P = 0.85; Figure 5). We observed differences for Black Bullhead (P < 0.01; Figure 3), Black Crappie (P = 0.04; Figure 3), Bluegill (P = 0.03; Figure 4), and Rock Bass (P = 0.01; Figure 4). Within the holding pen, the size structure was skewed toward larger sizes of all four species compared with the cab.

Methods for bycatch reduction likely would be widely used if the designs were easily incorporated into existing gears and did not substantially alter the fish-catching ability of the gears. We observed higher turtle escapement rates from BRD design A than design B in all trials. Turtles were able to exit the cab through the escapement opening and the flap did not appear to inhibit escapement. The compressed throat in BRD design B likely deterred some turtles from escaping. However, turtles escaped at moderate rates during the 4-h and 24-h trials, indicating they will push through the throat. Unfortunately the two BRD designs used in this study had high fish escapement rates and exhibited decreased turtle escapement during the fish-population assessment simulation; therefore, we cannot recommend their inclusion as part of fish-sampling gear protocol without modification and further testing.

Turtle escapement rates were lower during fish-population assessment simulation trials, especially from design B. The reason for the decreased escapement rates is unknown. The presence of fish in the net may have altered turtle behavior because fish are a common component of the painted turtle's diet (Ernst and Lovich 2009). Captured turtles may have focused their attention on eating rather than escaping; however, we did not record data pertaining to predation of fish by turtles in the net sets. Another possibility is the fish prevented turtle escapement by blocking the opening. However, we observed no fish blocking the opening when we retrieved the nets. Another possible explanation for the difference in escapement of turtles between the turtle escapement and fish assessment simulation trials could be the design of the trials. Turtles captured during the assessment simulation could swim into the nets at any time during the overnight set; and, during the turtle escapement trials, we placed 10 turtles into the nets at the same time. Lowry et al. (2005) observed turtle escapement in pulses, suggesting a “follow the leader” behavior. A group of turtles placed in the net at the same time is more likely to exhibit the “follow the leader” escapement behavior than the independent capture of turtles during the assessment simulation.

We expected higher escapement rates during the 24-h trials than 4-h trials because of the increased duration of time available for turtles to find the escapement openings. However, we observed lower escapement rates for both designs during the 24-h trials. Decreased escapement in design A could be due to reentry of turtles into the cab from the holding pen. The cause of decreased escapement rates in design B during the 24-h trials is not known because we did not observe reentry. Lowry et al. (2005) noted 71% of turtles escaped within 90 min and 77% escaped during the entire 4-h trial indicating most escapements occurred within 90 min. The findings of Lowry et al. (2005) support our observation of substantial escapement during the 4-h trial and no additional escapement during the 24-h trials. Cairns et al. (2016) noted decreased activity with increased duration of submergence, which likely explains why no additional escapement occurred during the 24-h trials.

Although design A had high rate of turtle escapement, it also likely had high rates of fish escapement. We were unable to gain a true estimate of fish escapement because of reentry of fish into the cab of the net. The numbers reported should be considered minimum estimated escapement, with actual escapement likely exceeding our observations. The unknown effect of movement between the holding pen and cab precludes conclusions about size biases in this BRD design for Bluegill and western painted turtles.

Physical and behavioral differences among fish species appeared to affect escapement rates in design B. Typically benthic species (including Black Bullhead, White Sucker, and Channel Catfish) escaped at higher rates than other species, indicating they are more willing to push through the compressed throat. Laterally compressed, open-water fish species (such as Bluegill, Black Crappie, and Rock Bass) escaped at lower rates, indicating a reluctance to push through the throat. Smith et al. (2016) observed Black Bullhead escapement from modified fyke nets with a restricted throat or unrestricted throat, while Bluegill and Black Crappie readily escaped from the nets with unrestricted throats but few escaped from the restricted throats. The findings of Smith et al. (2016) support our observations of species-specific utilization of potential avenues of escape. Size of specimens also affected escapement from design B, with all four fish species analyzed exhibiting significantly larger size structure in the holding pens. It is not known whether smaller individuals were reluctant to attempt to use the escapement throat or if they were physically incapable of pushing through the opening. Size biases of target species are common with BRD designs installed in fyke nets (Fratto et al. 2008b; Larocque et al. 2012b; Cairns et al. 2013). Estimated escapement from design B is likely close to actual escapement because of minimal reentry of fish and no observed reentry of turtles into the cab. Although not suitable for reduction of turtle bycatch, design B could be useful for reduction of unwanted fish bycatch when targeting laterally compressed fishes.

Our results, and those of others, indicate escapement designs may have limited application for reducing bycatch during fish population sampling and, like exclusion designs, may be best suited for commercial fisheries targeting specific species. A promising alternative to reducing bycatch is reducing mortality of bycatch. Recommended methods of reducing bycatch mortality of freshwater turtles include setting nets in shallow water to allow turtles access to air (Bury 2011), sampling at lower water temperature (Moos and Blackwell 2017), and reducing the duration nets are submerged (Bury 2011; Larocque et al. 2012a). Potential gear alterations warranting further study include adding an enclosed chimney onto nets and refining the methods employed by Larocque et al. (2012c) for floating part of the fyke net to the surface using floats. Refinement of bycatch-mortality reduction methodologies would reduce the impact on nontarget species during biological fish population sampling.

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.

Table S1. Water body, date, water temperature (°C), bycatch reduction device (BRD) design, net identification number, time trial, curved carapace length (mm), sex, and escapement of western painted turtles Chrysemys picta bellii during turtle escapement trials utilizing modified fyke nets in northeastern South Dakota from 2012 to 2014.

Found at DOI: http://dx.doi.org/10.3996/112016-JFWM-082.S1 (24 KB XLSX).

Table S2. Water body, date, water temperature (°C), bycatch reduction device (BRD) design, net identification number, species captured, number that were retained in cab, number that escaped into holding pen, and total number captured in each net set during fish population assessment simulation trials utilizing modified fyke nets in northeastern South Dakota from 2012 to 2014.

Found at DOI: http://dx.doi.org/10.3996/112016-JFWM-082.S2 (29 KB XLSX).

Table S3. Net design, 1.0-cm size class (total length), and escapement data for Black Bullhead Ameiurus melas used to evaluate how size affected use of bycatch reduction device (BRD) designs A and B in modified fyke nets during fish population assessment simulation trials in northeastern South Dakota from 2012 to 2014.

Found at DOI: http://dx.doi.org/10.3996/112016-JFWM-082.S3 (100 KB XLSX).

Table S4. Net design, 1.0-cm size class (total length), and escapement data for Black Crappie Pomoxis nigromaculatus used to evaluate how size affected use of bycatch reduction device (BRD) designs A and B in modified fyke nets during fish population assessment simulation trials in northeastern South Dakota from 2012 to 2014.

Found at DOI: http://dx.doi.org/10.3996/112016-JFWM-082.S4 (21 KB XLSX).

Table S5. Net design, 1.0-cm size class (total length), and escapement data for Bluegill Lepomis macrochirus used to evaluate how size affected use of bycatch reduction device (BRD) designs A and B in modified fyke nets during fish population assessment simulation trials in northeastern South Dakota from 2012 to 2014.

Found at DOI: http://dx.doi.org/10.3996/112016-JFWM-082.S5 (18 KB XLSX).

Table S6. Net design, 1.0-cm size class (total length), and escapement data for Rock Bass Ambloplites rupestris used to evaluate how size affected use of bycatch reduction device (BRD) designs A and B in modified fyke nets during fish population assessment simulation trials in northeastern South Dakota from 2012 to 2014.

Found at DOI: http://dx.doi.org/10.3996/112016-JFWM-082.S6 (14 KB XLSX).

Table S7. Net design, 1.0-cm size class (curved carapace length), and escapement data for western painted turtles Chrysemys picta bellii used to evaluate how size affected use of bycatch reduction device (BRD) designs A and B in modified fyke nets during fish population assessment simulation trials in northeastern South Dakota from 2012 to 2014.

Found at DOI: http://dx.doi.org/10.3996/112016-JFWM-082.S7 (16 KB XLSX).

We thank all of the permanent and seasonal South Dakota Game, Fish and Parks employees that assisted with the collection of data, especially R. Braun, S. Kennedy, and T. Kaufman. Substantial improvement to the manuscript was possible thanks to excellent critiques by the Associate Editor and anonymous reviewers.

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.