EVALUATION OF A TOPLESS BOTTOM-TRAWL DESIGN FOR FISH CAPTURE IN THE SUMMER FLOUNDER FISHERY

One of the major threats to sea turtles is the incidental capture as bycatch in marine commercial trawl fisheries. A gear-based approach has been suggested to reduce sea turtle bycatch levels in the summer flounder fishery (Paralicthys dentatus). Previous conservation experiments using a turtle excluder device (TED) in the summer flounder fishery resulted in a significant loss of the target species summer flounder, about 35% on average. A topless-trawl design was proposed as an alternative gear design to mitigate sea turtle bycatch. Previous testing showed that a topless-trawl with a headrope length of 48.7 m (160 ft) was effective at reducing sea turtle catch, but had a significant loss of target species, ranging from 51-74% on average, compared to a traditional trawl net with a 19.8-m (65-ft) headrope. In an effort to improve performance of the experimental trawl, a model of the 48.7-m (160-ft) headrope trawl was evaluated at the flume tank at the Memorial University in St. John’s, Newfoundland. This experimental net was optimally reconfigured with thirty 20-cm (8-in) plastic floats on the headrope and two restrictor lines. The 48.7-m (160ft) topless-trawl with 30 floats and two restrictor lines was tested in the summer flounder fishery in the summer of 2013 to assess its ability to catch summer flounder with two different float configurations. With the optimal float arrangement, the 48.7-m (160-ft) headrope topless-trawl with two restrictor lines had a significant loss of target species (p=0.008), with 22.7% loss compared to a traditional trawl. With this same float arrangement, the topless-trawl had a 12% loss of skate species (the majority of the catch) with no significance from zero (p=0.057). The experimental topless-trawl reduced the capture of all species overall, including the target species, summer

Cookie Sweep: Protective ground gear equipped with rubber rollers ("cookies") that roll over and keep the net off of obstructions along the ocean bottom and help minimize trawl damage.
Fishing Circle: The mouth of the trawl net.
Footrope: The bottom rope of the fishing circle.
Headrope: The top rope of the fishing circle, usually equipped with plastic floats.
Trawl Doors: Doors (made of wood or steel) that flow through the water at an angle, causing them to spread apart and aid in opening the net horizontally. Doors also kick up bottom sediment as they are towed along the bottom, initializing the herding of fish.

Regulatory Action
In 1977, the National Marine Fisheries Service (NMFS) and the United States

Sea Turtles and Fishing Gear
A gear-based approach would work to reduce bycatch levels because specific gear is more prone to incidentally capture non-target species (Mitchell et al. 1995). 2 Sea turtle bycatch regulations are addressed fishery-by-fishery and the focus of this study is the interactions within the summer flounder trawl fishery. The loggerhead species will be the primary sea turtle species of focus because of its relative abundance in the northeast and mid-Atlantic waters and interactions with fishing gear in these areas.
Sub-adult and adult loggerhead sea turtles' prey consists of benthic species such as crustaceans, mollusks, and invertebrates in the bottom sediment (Ruckdeschel and Shoop 2006). From May to October, when sea surface temperatures seasonally increase, loggerheads are found as far north as southern New England to forage before migrating south for the winter (Ruckdeschel and Shoop 2006). In areas south of Virginia, summer is the breeding season for the loggerheads and other sea turtle species, so they are consistently found near shore between nesting cycles (Spotila 2011).
According to the U.S. National Bycatch Report, in 2005 an estimated total of 11,772 individual sea turtles were caught in all U.S. fisheries and an estimated total of 1,062 individual sea turtles, all loggerhead species, were caught as bycatch in the Northeast fisheries (Karp et al. 2011). In general, some of the fisheries with the highest bycatch ratios were bottom trawl fisheries. In addition to the ecological effects of high bycatch levels, there are also economic consequences because the cost of sorting through bycatch is high and time consuming (Kumar and Deepthi 2006).
Trawl gear specifically is a large threat to sea turtles. Historical stranding and observer data have shown that trawl nets have a strong ability to consistently catch sea turtles and have been addressed by NOAA as a priority in reducing sea turtle bycatch 3 ). A trawl net is a mobile, non-selective gear type with high bycatch levels of species that typically have little to no commercial value (Kumar and Deepthi 2006). Trawl nets catch prey through entrainment/herding by actively towing the nets through the water column and catching non-target species (Sasso and Epperly 2006).
According to the U.S. National Bycatch Report (Karp et al. 2011) the primary discards in the northeast trawl fisheries are non-marketable species and are discarded as waste.
The low selectivity of trawl gear results in the catch of endangered species such as sea turtles, marine mammals, and elasmobranchs. When air-breathing animals, such as sea turtles, are caught in nets they are forcibly submerged and typically drown. Non-target species, such as sea turtles, that are caught as bycatch have key roles in the marine food webs and ecosystem. The natural biodiversity of these systems can be strongly impacted when these species are removed (Kumar and Deepthi 2006). Limited capture of these federally protected species is permitted by the Magnuson Stevenson Act (Magnuson-Stevens Fishery Conservation and Management Act sec. 206).

Turtle Excluder Devices (TED's)
One of the more effective ways to reduce bycatch in trawl fisheries is to alter trawl designs and use gear modifications to increase the selectivity of trawl gear (Kumar andDeepthi 2006, Karp et al. 2011). An example of a trawl fishery that has historically had a high level of sea turtle bycatch is the U.S. shrimp fishery that typically uses a bottom otter trawl to entrain prey (Lutz et al. 2002 Before the required use of a bycatch reduction device (BRD), incidental capture within 4 shrimp trawls was considered the highest human source of mortality on sea turtle populations. Shrimp trawls killed approximately 44,000 sea turtles annually and the main species caught were juvenile loggerheads and Kemp's Ridleys (NRC 1990, Spotila 2011. Due to high levels of bycatch, especially in the Gulf of Mexico, shrimp fisheries have had to implement regulations and use gear modifications to reduce these levels of bycatch. One gear alteration that has been made to reduce bycatch is the Turtle Excluder Device (TED), which was first introduced to shrimp fisheries in the late 1980's (Mitchell et al. 1995). A TED is an array of angled, spaced bars, positioned in the direction of trawl flow and placed in the trawl net immediately before the cod end to allow entrapped sea turtles to escape (Saunders 1988, Lutz et al. 2002. If the opening of the TED is large enough, all sea turtles should effectively escape. NMFS requires that TED's be 97% effective in sea turtle escapement (Crowder et al. 1994). TED regulations and requirements continuously changed after their initial requirement and as of 1994 TED's were required year round for all inshore and offshore shrimp trawls (Crowder et al. 1994). Within these fisheries, the TED has proven to be a reliable BRD by effectively reducing sea turtle catch and catching shrimp with less than 5% loss of shrimp (Department of Commerce 1987).

Summer Flounder Fishery
The summer flounder fishery is a commercially important trawl fishery in the mid-Atlantic and northeast, ranging from Cape Hatteras, NC to Cape Cod, MA (Terceiro 2006). The summer flounder fishery typically uses a bottom otter trawl net, 5 similar to the shrimp fisheries (Terceiro 2006). Summer flounder (Paralichthys dentatus) is a demersal flatfish ranging from the Gulf of Maine to South Carolina (Terceiro 2006). The interactions between sea turtles and fisheries occur because of the overlap in habitat use. The summer flounder is predominantly found in the same sandy substrates as the loggerhead's prey (Perrine et al. 2003).
The bycatch of loggerhead sea turtles caught within the summer flounder fishery was estimated at 192 turtles annually for 2000-2004(Murray 2008. With an increase in bycatch in Northeast regions, it is anticipated that BRD's will soon be required in the Northeast fisheries (Karp et al. 2011). Since 1992, all bottom trawls for summer flounder south of Cape Charles, VA have been required to use a NMFSapproved Flounder TED. The current TED in the summer flounder fishery is a hardgrid design with ≥89 cm width (≥ 35 in) and ≥32 cm (≥12 in) height (Gallagher 2010). The northern regions of the flounder fishery including southern New England and the mid-Atlantic currently do not require the use of a TED, but that is likely to change in the future (Gallagher 2010). One of the proposed ways to reduce bycatch in trawl fisheries is to alter trawl designs and focus on changing fishing methods (Kumar andDeepthi 2006, Karp et al. 2011).
The catch efficiency of the NMFS-certified TED required in the Mid-Atlantic summer flounder fishery has been tested in existing fisheries inshore and offshore in the mid-Atlantic regions (DeAlteris and Parkins 2009). Trawl nets with the NMFScertified flounder TED had a significant reduction of target catch species ranging from 28% to 35% loss (Lawson et al. 2007, DeAlteris and Parkins 2009, DeAlteris and Parkins 2012. A standard TED and experimental TED were tested and both were 6 found to have substantial loss of target species (Lawson et al. 2007). It was suggested that a gear-based approach and alternative net design be made to exclude sea turtles because of the significant loss of target species.

Topless-trawls
The proposed alternative gear design for the summer flounder fishery was a topless-trawl (DeAlteris and Parkins 2012). A topless-trawl is designed by increasing the setback of the headrope, which eliminates the overhung panel (square) to allow pelagic species to exit the net (He et al. 2007, Pol et al. 2003. This type of gear modification considers the behavior of the target species and modifies the gear to match this behavior (Ryer 2007). This trawl type theoretically would not significantly reduce flatfish retention because of the herding behavior of flatfish (Ryer 2007).
Flatfish species generally use anti-predator behavior when encountering a trawl net and bury in the sediment or herd low to the seafloor close to the lower panel (Ryer 2007, Thomsen 1993. This increase of headrope length however would allow pelagic species to escape the trawl once they encounter the footrope by swimming upwards. Topless-trawls have been proposed in other flatfish fisheries as a method of reducing the bycatch of roundfish (Pol et al. 2003, Ryer 2007, Thomsen 1993. In contrast to flatfish, when roundfish encounter a trawl they are able to use bursts of sustained swimming to swim off of the seafloor (Ryer 2007). In a traditional bottom trawl the headrope is forward of the footrope, preventing fish from escaping through swimming upwards once they encounter the footrope (Revill et al. 2006). A toplesstrawl was first designed to reduce cod and other roundfish bycatch in flatfish trawls in 7 the Faroe Islands (Thomsen 1993). Underwater video footage showed that flatfish stayed at the lower panel of the trawl and roundfish swam upwards to effectively escape with a setback headrope and this study resulted in a significant reduction of roundfish bycatch with no flatfish catch reduction (Thomsen 1993). A topless-trawl design in the yellowtail flounder fishery was tested as a successful method of reducing juvenile target species and other bycatch (Pol et al. 2003). A similar design, referred to as a "cutaway trawl" was effective at reducing whiting bycatch without any loss of Nephrops (Norway lobster) in a European Nephrops fishery (Revill et al. 2006). A similar design with the Gulf of Maine pink shrimp fishery was tested as a way to reduce pelagic finfish species and was successful in reducing bycatch (He et al. 2007).
The effectiveness of topless-trawls in other groundfish fisheries suggested that this design could be an effective way to reduce sea turtle bycatch in the summer flounder fishery. An increased headrope for given footrope length increases the escape time for sea turtles and the demersal target species would herd and swim towards the cod end of the net and remain captured (DeAlteris and Parkins 2012). By excluding sea turtles before they are captured, the negative effects of forced submergence can be avoided. Reducing bycatch of large sea turtles (>100 cm length) is challenging in a fishery where the target species is also large (25 cm-70 cm length). It is challenging to develop gear modifications to exclude large bycatch species but retain large target species so behavior of these species and trawl nets must be identified to effectively do both. In regard to sea turtles, a topless-trawl would ideally work by allowing sea turtles to escape in a similar manner to roundfish in a trawl fishery. Sea turtles would 8 ideally swim upward once they feel the footrope and escape before entering the trawl due to the setback of the headrope and escape the net similarly to roundfish. stabilizing the TED in water and prevent the trawl from rolling over during deployment and retrieval (Mitchell et al. 1995).
The purpose of this study was to increase the catch performance of the toplesstrawl and test a modified topless-trawl in the summer flounder fishery for the retention of target species, summer flounder. A model net was designed and tested in the flume tank to create a set of gear modifications and optimally configure the trawl before it 9 was tested in the field. The null hypothesis was that there would be no significant difference of catch between the control net and the experimental topless-trawl at α=0.05. The alternative hypothesis was that the control net would catch on average more summer flounder than the experimental topless-trawl. Because this net was designed for use in a commercial fishery, it was also important to consider the size distribution of the summer flounder captured. The null hypothesis was that there would be no significant difference between the length frequency of the flounder with α=0.05. The alternative hypothesis was that there would be a difference between the two net types in the length frequency for summer flounder. and sixteen 20-cm (8-in) floats on the headrope (used in previous field studies). This configuration was tested as a comparison because in the field it effectively caught summer flounder but ineffectively reduced sea turtle catch.
All riggings (16) of the net were tested at a towing speed of 5.5 km-per-hour (3 knots), and at target bridle angles of 11 and 15 degrees. Rigs No. 6-10 were also tested at a target angle of 9 degrees to gather additional measurements. Measurements of the upper wing spread, left wing spread, mean wing end spread, wing height, headrope height, port tension, starboard tension, total tension, mouth area, mouth drag, and bridle angle were recorded.

Data Analysis
Locations, expressed in Latitude and Longitude, were recorded by GPS for all starting points on the tow and were recorded for the duration of the field work (Tables   2, 3, and 4, Figure 4). Door spread was calculated in feet and sensor data were collected for a sample of tows from field trials 1 and 2, and analyzed as an effort to evaluate the opening of the net throughout the duration of the tows (Tables 7 and   Figures 6, 7, and 8). Data were compiled using Microsoft Excel and R for analysis.
For sensor analysis, the measurements are the differences between two sensors vertically stacked on the headrope and footrope of the starboard-wing and are measured at hundreds of time points throughout each tow. The "mean opening" is the 15 overall average for the duration of the tow, "start opening" is the average at the beginning of the tow, and "end opening" is the average at the end of the tow (Tables 7   and 8).
Catch weights were compared using one-tailed paired T-tests to compare the catch of summer flounder and the bycatch between the experimental net and control net with a significance value of p<0.05. Mean catch weights for the topless and control trawls for the two topless-trawl float configurations are reported in kilograms per tow.
A catch ratio (experimental catch/control catch) was calculated for all paired tows to determine the percent loss of catch overall for summer flounder and skates. Length frequency graphs were created using "R" to evaluate the difference in size distributions between the control and experimental net for float configurations#1 and #2. A Kolmogorov-Smirnov test was performed to determine if there was a significant difference between length frequency distributions between the control and experimental nets for float configuration #1 and float configuration #2.
In the bycatch analysis, the skate complex included little, winter, and clearnose skate species. All other bycatch was sorted by species. The only species considered in this analysis of the net performance are summer flounder and skate species. Skate catch was considered because of the consistent high volumes of catch with both nets.
Because the sampling period took place from June 20-August 11 the species composition varied greatly throughout the summer sampling due to temporal changes in occurrence. The average catch in pounds and percent of total catch was calculated for the bycatch species consistently captured during all field trials (Table 14).

Flume Tank
A summary of the towing speed, bridle angle, upper wing spread, lower wing spread, mean wing end spread, wing height, and headrope height is shown in Table 1 and Figure 3. When floats were added to the headrope for Rigs No. 1-9 the headrope height and wing height gradually increased, but there was a strong difference,   Figures 3 and 4). The optimal configuration was Rig No. 11, with thirty 20-cm (8-in) floats on the headrope and the two restrictor lines (Figure 4).

Field Work
Float Configuration #1

Gear Measurements
The  (Table 5). Based on results from flume-tank testing with a scale model of the net, the estimated average observed bridle angle in the field testing was 15.4 degrees for both the topless-trawl and control models (Table 5).
Data from the depth sensors on the wing-opening were gathered for a sample of paired tows with float configuration #1 (Table 7). The topless-trawl starboard-wing height averaged from 0.03 m (0.1 ft) to 1.62 m (5.3 ft) and the control-trawl starboardwing height averaged from 0.8 m (2.8 ft) to 0.95 m (3.1 ft) ( Table 7). The majority of tows for the topless-trawl showed a decrease in wing height from the beginning to the end of the tow with a slope significantly different from zero (Table 7, Figure 6). The three tows for the control trawl had a slope closer to zero for wing-opening over time (Table 7). A linear regression model was fit for the gear measurements for Haul 3 on summer flounder compared to the control trawl and this was a significant difference from zero, (p=0.0016) (Table 10). This is based on a mean catch per tow of 78.9 kg (174.1 lbs) with a standard deviation of 31.9 kg for the topless-trawl as compared to a mean catch per tow of 110.3 kg (243.2 lbs) with a standard deviation of 42.2 kg for the control trawl (Table 9). The catch varied greatly between tows in the first field trial from June 20-June 27, 2013 for both net types, possibly due to a low volume of fish.
For float configuration #1, the summer flounder was on average 7.5% of the total catch for the topless-trawl and 7.4% of the total catch for the control trawl (Table 10).
A Kolmogorov-Smirnov two-sample test was used to compare the length frequency distributions of summer flounder between the control trawl and experimental topless trawl for float configuration #1. The Kolmogorov-Smirnov test indicated a statistically significant difference between the two distributions. However this test is very sensitive to large sample sizes and although the test indicates a significant difference, the distributions still appear similar between the control and experimental nets ( Figure 10).

Bycatch
For float configuration #1, the skate complex was on average 84.8% of the total catch for the topless-trawl and was on average 86.0% of the total catch for the control trawl (Table 12). For float configuration #1, the topless-trawl caught on average 30.1% less skates than the control trawl with significant difference from zero (p=0.0070) (Table12). This is based on a mean catch per tow of 894.5 kg (1972.1 lbs) 19 and standard deviation of 306.2 kg for the topless-trawl compared to 1281.6 kg (2825.5 lbs) and standard deviation of 462.7 kg for the control trawl.

Gear Measurements
For float configuration #2 the average door spread for topless-trawl was 107.3 m (351.9 ft) compared to an average door spread of 112.0 m (367.7 ft) for the control trawl ( Table 6). The average bridle angle for the topless-trawl with an assumed a wingspread of 14.0 m (46 ft) was 14.8 degrees and was compared to the average bridle angle for the control net with an assumed wingspread of 15.2 m (50 ft) was 15.1 degrees (Table 6). Data from the depth sensors were gathered on a subset of paired tows with float configuration #2 for the experimental trawl (Table 8). The toplesstrawl center headrope opening averaged from 0.21 m (0.7 ft) to 1.46 m (4.8 ft) (Table   20   8). All tows for the topless-trawl showed a decrease in headrope opening from the beginning to the end of the tow with a slope significantly different from zero (Table   8). A linear regression model was fit for the gear measurements for Haul 6 on July 14, 2014 for the experimental net as an example of the headrope height changes over time (R 2 =0.573 and p-value<0.0001 ( Figure 10).

Summer Flounder
Float configuration #2 had an average loss of 22.7% summer flounder when compared to the control trawl and this was a significant difference from zero (p=0.008) (Table 11). This is based on a mean catch per tow of 126.1 kg (277.9 lbs) with a standard deviation of 47.2 kg for the topless-trawl as compared to a mean catch per tow of 163.2 kg (359.7 lbs) with a standard deviation of 99.1 kg for the control trawl (Table 9). For float configuration #2, the summer flounder was on average 11.2% of the total catch for the topless-trawl and 12.2% of the total catch for the control trawl (Table11).
A Kolmogorov-Smirnov two-sample test was used to compare the length frequency distributions of summer flounder between the control trawl and experimental topless trawl for float configuration #2. The Kolmogorov-Smirnov test indicated a statistically significant difference between the two distributions. However this test is very sensitive to large sample sizes and although the test indicates a significant difference, the distributions still appear similar between the control and experimental nets (Figure 10).
For float configuration #2, a total of 21 pairs were completed during the day, and for these pairs the topless-trawl caught on average 16.0% less summer flounder 21 than the control net (Table 11). On average, the percent of summer flounder in the total catch for paired tows conducted during the day was 8.7% for the topless-trawl and 9.3% for the control trawl (Table 11). A total of seven night pairs were completed with float configuration #2. For night pairs, the topless-trawl caught on average 30.5% less summer flounder than the control trawl (Table11). On average, for the seven night pairs completed, the percent summer flounder of total catch was 23.6% for the toplesstrawl and 23.6% for the control trawl (Table 11).

Bycatch
For float configuration #2, the topless-trawl caught on average 12.0% less skates than the control net which was not significantly different from zero (p=0.0570) (Table 14). This is based on a mean catch per tow of 948.8 kg (2019.8 lbs) and a standard deviation of 842.7 kg for the topless-trawl as compared to 1080.2 kg (2381.5 lbs) and a standard deviation of 822.3 kg for the control trawl. On average, the percent skate of total catch was 84.6% for the topless-trawl and 81.0% for the control trawl (Table 14). For the 21 paired tows completed during the day, the topless-trawl caught on average 10.0% less skates than the control trawl (Table 14). For the pairs completed during the day, on average skates were 81.2% of the total catch for the topless-trawl and 79.5% of the total catch for the control trawl (Table 14). For night pairs, the topless-trawl caught on average 23.5% less skates than the control trawl (Table 14). For the seven night pairs completed, on average skates were 71.4% of the total catch for the topless-trawl and 64.8% of the total catch for the control trawl (

Flume Tank
The flume tank work allowed us to evaluate why the original 48.7-m (160-ft) topless-trawl had a significant loss of target species. Without restrictor lines, the headrope height was low at 1.4 m (4.5 ft), and it is assumed that flounder were escaping in the wing section because the wings of the net were lying flat (Table 1). It appeared that there was not enough lift of the headrope because the wings of the net were lying flat due to the significant setback of the headrope. With an increased bridle angle the spread increased and the headrope height decreased to 0.9 m (3.1 ft). At the maximum spread, the wings of the net spread wider and the headrope height was low allowing for even less retention of summer flounder (Figure 3). Attempting to shorten extensions did not have an influence on the configuration. The additional float on Rigs No. 1-9 increased the headrope height slightly but the wings continued to lie flat at the increased bridle angle (Figure 3). The addition of floats allowed for a larger wing height opening, but not enough of a difference to address the fishing issues.
With the addition of two restrictor lines the net took proper shape and the wings and headrope increased in height. The addition of the restrictor ropes strongly influenced the upper and lower wing spread, bringing the two wings closer together and increasing the wing height and headrope height (Figure 3). For Rigs No. 11-15, the configuration included the two restrictor ropes in the same position as Rig No. 10 but the number of floats was altered to now reduce the height of the headrope. At the maximum number of floats (thirty-seven 20-cm (8-in)), the headrope height was 2.7 m (7 ft) and there was concern that the net would not effectively decrease sea turtle catch 24 (Table 1). With the decrease of floats, the wing-opening decreased overall and the headrope opening decreased slightly (Figures 4 and 5). Before the cod end of the net, a "pocket" developed when the meshes between the headrope and cod end should have remained streamlined. The optimal configuration, Rig No. 11, was with thirty 20-cm (8-in) floats along the headrope and was the model used to design the gear for the field study ( Figure 4).

Field Trials
Overall, float configuration #1 and float configuration #2 both resulted in a statistically significant reduction of summer flounder. Therefore we can reject the null hypothesis that there is no significant difference between the catch of flounder between the 48.7-m (160-ft) headrope topless-trawl and the control trawl and accept the alternative hypothesis that there is statistical difference between the two nets.
Float configuration #1 resulted in significant loss of target species with 30.4 % loss (p value=0.0016). The percent of fluke of total catch was 7.5 % for the toplesstrawl and 7.4% for the control trawl, indicating that the experimental net was reducing overall catch of all species including summer flounder. The rearrangement of floats after the initial eleven pairs helped to evenly distribute the lift of the headrope, which increased catch efficiency. After this modification, float configuration #2 was more effective at catching summer flounder with the 22.7% loss of summer flounder compared to the control trawl, although there was still significant difference between the two net types (p=0.0080). The tows completed using float configuration #2 had a higher percent of summer flounder of total catch for both net types, indicating that more summer flounder were present at the time. The percent of summer flounder of 25 total catch was 11.2% for the experimental trawl and 12.2% for the control trawl, indicating that the experimental net is reducing overall catch, not just catch of the summer flounder with both float configurations.
For all pairs with float configuration #2, a total of seven paired tows were completed at night. There was a quantitative and observational difference between the night and day pairs. For the seven night pairs completed with float configuration #2, the topless-trawl caught 30.5% less summer flounder and 23.5% less skates than the control trawl. For the twenty-one day pairs completed with float configuration #2, the topless-trawl caught 17.7% less flounder than the control trawl (Table 11). More flounder were caught on average during the night pairs compared to day pairs, but there was less difference between the two nets during the day pairs. So, although the topless-trawl performed better at night overall, there was less of a difference between the two net types during the paired tows completed during the day.
Based on video data obtained from the underwater camera footage collected throughout the study, a pocket that had formed before the cod end of the net that was seen to "clog" with fish throughout the tow. This was similar to the "pocket" seen in the flume tank and the issue addressed with Rig No. 15. It is possible that this pocket caused the headrope center opening and starboard vertical wing-opening to decrease in height throughout the duration of the hauls and could contribute to the loss of flatfish ( Figure 6 and 10). An increase of catch throughout the duration of a trawl could cause physical changes in the net and bring the headrope down and interfere with functionality of the net. This loading of fish before the cod end and decrease in headrope height throughout the tow was not seen in the control trawl. 26 Although there was still significant loss of flounder with the topless-trawl, this design had the lowest loss of target species when compared to past gear modifications.
A loss of 22.7% target species is still a lower loss than seen when using a TED (28%-35% loss), and the original 48.7-m (160-ft) headrope (51%-74% loss) (DeAlteris and Parkins 2012; Lawson et al. 2007). Because the null hypothesis was rejected and there was statistical significance between the two net types, there is not an opportunity for Type II error. A power analysis on a data set is only necessary when the null hypothesis is false, which is a Type II error (Zar 1984). Therefore Type II error did not occur in this study and there is not a need to perform a power analysis on the data examined.
The decrease in loss of target species is due to the additions of the restrictor lines within the net that help to elevate the long wings of the topless-trawl, thus maintaining the proper shape of the trawl opening. So although there is still significant difference between the catch efficiency of the two nets, the 48.7-m (160-ft) headrope design for the topless-trawl with the addition of two restrictor lines could be a positive method to mitigate sea turtle interactions within the mid-Atlantic and southern New England summer flounder fishery. This topless-trawl was later tested for its ability to reduce sea turtle bycatch, which is not included in this thesis. Using the same control trawl and experimental topless-trawl design, a total of 132 paired tows were conducted would be beneficial to conduct further studies in the flume-tank to optimally configure the experimental topless trawl by excluding the "pocket" that was seen before the cod end. Potentially if this "pocket" were removed, the clogging of fish could be reduced and the headline height would remain more stable while fishing rather than decreasing in height.