Kite Escape Device: a New Approach to Reduce Bycatch in Shrimp Trawls

Abstract

Shrimp trawls are one of the main bycatch generators in artisanal fisheries. Our work evaluated the adoption of low cost devices for artisanal bottom trawling to reduce bycatch assimilation and improve shrimp catches. To evaluate the proposed changes, we evaluated the performance of four different trawl gears from May to July 2015 in Pinheira Beach, Brazil. The hauls were divided into four treatments as follows: (1) hauls with unmodified control gear, (2) hauls with gear that included the kite escape device (KED), (3) hauls with gear that included long sweeps, and (4) hauls with gear that included both the KED and long sweeps. The second trawling group (2) showed the highest reduction in bycatch for all trawl gears. Additionally, all hauls that included the KED and/or long sweeps had improved shrimp catches during the study compared with trawl nets without the KED. During the hauls with KEDs we got the best shrimp-to-bycatch proportion (1:8.9). The modifications proposed can improve trawl gear selectivity while maintaining the shrimp catches.

Received December 12, 2016; accepted June 20, 2017

Usually, commercial fishing methods contribute to large catches of a target species. However, during fishing operations it is also possible to observe catches of other species, which are typically called bycatch and not targeted by fishing activities (Alverson et al. 1994; Kelleher 2005). Shrimp trawling becomes a major bycatch generator once the weight of shrimp is exceeded by that of the assimilated bycatch, which can be composed of hundreds of fish and aquatic invertebrates species and be responsible for nearly 25% of the bycatch generated worldwide (Gray et al. 1990; Kenelly et al. 1998; Stobutzki et al. 2001; Stelle et al. 2002; Kelleher 2005). This problem has an impact on the food chain, creating changes to the abundance and spatial distribution of many species (Jones 1992; Dayton et al. 1995).

Some authors have recommended the implementation of technologies to reduce bycatch (bycatch reduction devices [BRDs]) thereby minimizing the impacts of shrimp trawls in the aquatic environment (Broadhurst 2000; Hall et al. 2000). These technological changes have been adopted in some countries since the 1960s, and several strategies have already been tested.

In Brazil, experiments with technological changes in shrimp trawls are still incipient. In fact, the first example of their implementation is attributed to square meshes in the cod end of the trawl adopted by Conolly (1992). Recent experiments using Nordmore grids and square meshes were performed by Cattani et al. (2011) and Silva et al. (2011, 2012a, 2012b).

According to Branco and Verani (2006), the Brazilian artisanal shrimp trawls can catch nontarget species in a shrimp: bycatch proportion by weight as high as 1:19. Considering the relevance of this subject and the lack of efforts to minimize the impact of this catch method in a regional context, this study evaluated the performance of a new, low-cost BRD called a kite escape device (KED), developed to be implemented in typical Brazilian trawling gear. Additionally, this study evaluated the use of long sweeps between the doors and the edge of the wings on the bottom trawl. Notably, the sweeps have the potential to increase the swept area and make the gear more efficient for catching demersal fishes at the front of the trawl mouth (Kim and Wardle 1998; Weinberg and Somerton 2006). In our study, we evaluated the potential of these devices to increase shrimp catches.

METHODS

Kite escape device

The focus of this study was to evaluate the impacts of the KED and sweeps on the catch composition in a typical shrimp trawl gear design adopted by Brazilian artisanal fisheries. Thus, our trawl gear included characteristics of the main local trawl gear (Table 1; Figure 1). We assumed that the environmental conditions were constant during the field activities because all trawls were conducted under the same tide and moon conditions.

TABLE 1. Typical measurements of artisanal shrimp trawls in the southern region of Brazil.

Trawl gear characteristics	Size
Layout	2 panels
Headrope	10 m
Cod end	3.56 m
Total length (without cod end)	8.94 m
Total length (with cod end)	12.5 m
Stretched mesh (wings and tunnel)	0.024 m
Stretched mesh (cod end)	0.02 m
Total weight (with foot rope)	7.70 kg

FIGURE 1. Typical shrimp trawl gear configuration used by artisanal fishers in the southern region of Brazil. The upper and lower portions of the gear are shown on the left and right, respectively. Twine thicknesses are given in the Den scale (g/9,000 m; e.g., 20/9 = 9 yarns of 20 g/9,000 m each); the letters A–E refer to the different net panels; PP = polypropylene, PE = polyethylene, and ϕ = the diameter of the foot and head ropes; the numbers right beside the net panels are the numbers of meshes; the numbers 24 and 20 are the sizes (mm) of the stretched meshes along the fishing gear; finally, the number–letter combinations (e.g., 4B1N) represent the taper cuts. According to the FAO (1990), three different taper cuts are possible: N = a cut through both strands at the side of a knot, advancing the taper cut in the netting one mesh in the direction perpendicular to the general course of the twine; T = a cut through both strands at the top or bottom of a knot, advancing the taper cut in the netting one mesh in the twinewise direction; and B = a cut through one strand at a knot, advancing the taper cut in the netting half a mesh across the strip in the direction perpendicular to the general course of the twine plus half a mesh either way along the strip in the twinewise direction. More details about these taper cuts are available in FAO (1990).

The KED is a low-cost device prototype derived from the union of two BRDs previously described in the literature: the fisheye and the escape window (Broadhurst 2000; Eayrs 2007). The fisheye is a device installed on the cod end of the trawl that provides the largest escape of flatfish through an elliptical opening. The escape is effective once the fish swim against the current and return to the cod end during the fishing operation. The escape window consists of a window with larger square meshes usually located in the cod end (upper panel) and is effective in providing an escape for smaller fish. For this device to work, the escape can occur during the fish’s passage to the cod end; that is, it is not necessary for the fish to swim against the current.

In order to combine the advantages observed between both BRDs mentioned above, we developed a KED that adapted the two exclusion mechanisms into a single device to increase the efficiency of bycatch reduction (Figure 2c). The KED was made with galvanized wire 51.5 mm in diameter and was 0.44 kg in weight (see plots b and c in Figure 2). Associated with this wire frame were square meshes comprising 65-mm stretched mesh. This device was installed on the upper cod end panel to provide bycatch reduction. Additionally, we implemented four polystyrene floats in the device to provide better buoyancy during fishing, which avoided the possibility that the upper cod end panel would touch the lower cod end paneland thus obstruct fish passage to the end of the fishing gear.

FIGURE 2. Designs of trawl gear components. (a) Escape window with square meshes. (b) Fisheye device. (c) KED prototype (escape window plus fisheye). (d) Layout of long sweeps implementation.

The BRD placement on the cod end is very important because if it is installed too close to the end of the cod end, the shrimp can escape during hauls. In contrast, if the BRD is installed too close to the end portion of the tunnel, the fish may be prevented from escaping the trawl gear (Eayrs 2007). Thus, the KED was placed at a distance of 105 meshes, counted from the end of the cod end, resulting in a space of 2.10 m between the insertion point of the BRD and the end of the cod end.

Sweeps evaluation and experimental design

The sweeps were designed to increase the swept area using cables allocated between the doors and the wing edges. Several authors (Andrew et al. 1991; Kim and Wardle 1998; Somerton and Munro 2001; Weinberg and Somerton 2006) have reported the effectiveness of this device in increasing fish catches; however, the effectiveness of sweeps on shrimp catch was not described. So the sweeps were positioned in the same substrata line to maximize friction with the bottom and possibly increase shrimp catches (Santos et al. 2006).

The sweeps consisted of polyethylene cables 6 mm in diameter, fitted with rings 65 mm in diameter to provide the best layout. Based on the structural knowledge that sweeps can increase the swept area, we expected to see increases in shrimp catches. It is important to highlight that sweep performance, that is, the horizontal and vertical opening of the doors, was calculated using the equations suggested by FAO (1990). Thus, it was possible to evaluate the sweeps relative efficiency in increasing the swept area (see plot c in Figure 2).

The experiment was performed at Pinheira Beach located in the southern Brazilian region (29.87°S, 48.58°W; sand and mud substrata; approximately 5–7 m depth) in 2015 during autumn and winter using a artisanal fishing boat 7 m in length and powered by an 18-hp motor. During the study period we conducted 24 hauls with shrimp trawls organized into four distinct gear types with six trials per gear type: (1) control fishing gear, (2) trawl gear with KED (CD), (3) trawl gear with long sweeps (SDCT), and (4) trawl gear with KED plus long sweeps (CDCT).

Data analysis

The catches were separated on board into two categories: shrimp and bycatch. The fish caught were identified to the family level and type according to specific taxonomic keys. The total biomass per haul was used to calculate the CPUE (kg/h) for shrimp and fish caught during all experiments.

The performance of the control gear, CD, SDCT, and CDCT hauls were evaluated using shrimp assimilation and bycatch exclusion. The bycatch exclusion capability was calculated from equation (1), as suggested by Rico-Meija and Rueda (2007). Additionally, equation (2) measures the shrimp catches capacity. Thus, by considering the relative CPUE, it was possible to evaluate the effectiveness of CD, SDCT, and CDCT, in relation to the control gear.

(1)

(2)

We implemented the Kruskal–Wallis test (Siegel and Castellan-Junior 1988) on the assimilated bycatch biomasses in the control gear, CD, SDCT, and CDCT to check for statistical differences between bycatch by trawl gear type. Additionally, we adapted the Mann–Whitney U-test (Lehmann 2006) to test for statistical differences between experiments. Thus, it was possible to evaluate the effectiveness of each shrimp trawl gear to reduce bycatch, highlighting which of the tested hauls had the best (and worst) selectivity performance. We emphasize that both tests were applied considering a one-tailed distribution and a 0.05 significance level.

RESULTS

During the trials we caught four distinct shrimp species: red spotted shrimp Farfantepenaeus brasiliensis, southern white shrimp Litopenaeus schmitti, Argentine red shrimp Pleoticus muelleri, and Atlantic seabob shrimp Xiphopenaeus kroyeri. The main species caught in all hauls was Atlantic seabob shrimp, corresponding to 97.4% of the total catch. The other species represented just 2.6% of the catch.

The trawls caught 7,278 bycatch fish from 21 families and 42 species (Table 2). The Sciaenidae and Carangidae families represented 90.38% of the total catch. The main species caught were: Rake Stardrum Stellifer rastrifer, Atlantic Bumper Chloroscombrus chrysurus, American Coastal Pellona Pellona harroweri, Whitemouth Croaker Micropogonias furnieri, and Banded Croaker Paralonchurus brasiliensis,; these species represented more than 90% of the total catch (Table 2). Rake Stardrum and Atlantic Bumper alone represented 81%. The total catch obtained across all trawls was 8,737 individuals. The fish bycatch contained 7,278 individuals while the shrimp catch contained 1,459 individuals.

TABLE 2. Shrimp species and the five teleost fishes most often caught by treatment. SD = control trawl gear, CD = trawl gear with KED, SDCT = trawl gear with long sweeps, CDCT = trawl gear with KED plus long sweeps.

Family	Scientific name	Catch numbers by treatment				Total number
		SD	CD	SDCT	CDCT
Shrimp species
Penaeidae	Farfantepenaeus brasiliensis	1			1	2
	Litopenaeus schmitti	9	6	10	9	34
	Xiphopenaeus kroyeri	238	303	401	479	1,421
Solenoceridae	Pleoticus muelleri			1	1	2
Teleost fish species
Carangidae	Chloroscombrus chrysurus	2,278	46	397	63	2,784
Clupeidae	Pellona harroweri	357	81	12	6	456
Sciaenidae	Micropogonias furnieri	12	10	63	77	162
	Paralonchurus brasiliensis	18	10	62	63	153
	Stellifer rastrifer	524	652	1,020	908	3,104

The gear efficiency that accurred when sweeps were present (CT) compared with the control gear suggests that CT trawls had a greater horizontal opening. In one case, CT trawls showed a 21% greater horizontal opening (7.1 m) compared with those without sweeps (ST) (5.0 m) (Table 3). Thus, it is possible to confirm that the CT trawls covered a larger swept area and therefore had an increased trawl sampling area than did those of the ST.

TABLE 3. Results for Kruskal–Wallis test (variance analysis) and Mann–Whitney U-test (U-value). SD = control trawl gear, CD = trawl gear with KED, SDCT = trawl gear with long sweeps, CDCT = trawl gear with KED plus long sweeps.

Test	Value	Status
Kruskal–Wallis test
SD, CD, SDCT and CDCT	269.57	Reject the null hypothesis
Mann–Whitney test
SD versus CD	54.91	Reject the null hypothesis
SD versus SDCT	80.63	Reject the null hypothesis
SD versus CDCT	233.21	Reject the null hypothesis
CD versus SDCT	16.22	Reject the null hypothesis
CD versus CDCT	30.02	Reject the null hypothesis
SDCT versus CDCT	64.39	Reject the null hypothesis

The best trawl performances in terms of shrimp catch were obtained during hauls with sweeps (SDCT and CDCT). In CDCT we caught 490 individuals (1.55 kg), followed by SDCT, which caught 412 individuals (1.23 kg), and CD with 309 individuals (0.92 kg). The control gear hauls had the lowest shrimp catches (248 individuals, 0.64 kg), and there was no loss of shrimp when the modified trawls (KED and sweeps) were employed . In fact, there was an increase in shrimp catches compared with the control gear trawls (Figure 3).

FIGURE 3. Numbers of bycatch (FA) and shrimp catch (CA) caught by differnet trawl gears. SD = control trawl gear, CD = trawl gear with KED, SDCT = trawl gear with long sweeps, CDCT = trawl gear with KED plus long sweeps.

Bycatch was predominant in all fishing gears. The highest incidence of bycatch was observed in the control gear (3,398 individuals) and SDCT (1,764 individuals). By contrast, hauls containing the KED (CD and CDCT) showed significant bycatch reductions. The smallest bycatch was observed in CD (873 individuals) followed by CDCT (1,243 individuals) (Figure 3).

Thus, considering the bycatch by trawl modality, the highest bycatch was observed in the control gear. By comparing the selectivity percentage between trawl treatments (CD, SDCT, and CDCT) relative to the control gear, we observed that SDCT had 48% selectivity and CDCT had 63.5%. The CD had the best selectivity, excluding 75% of bycatch relative to the control gear. So, the CD treatments emerged as the modality with the best selectivity.

We calculated the fraction of bycatch and shrimp catch in each haul. For the control gear the percentage was 92.7% bycatch and 7.3% shrimp catch, indicating that this modality was the least effective. In the modified trawls (CD, SDCT, and CDCT) we observed a significant reduction in the differences between these percentages. In CDCT we achieved 60.6% bycatch and 39.4% shrimp catch, followed by CD (64.4% bycatch and 35.4% shrimp catch) and SDCT (76.7% bycatch and 23.3% shrimp catch). According to our evaluation, CD and CDCT presented the best selectivities for reducing bycatch (Figure 4).

FIGURE 4. Percentage of shrimp catch (CA) and bycatch (FA) caught by different trawl gears. SD = control trawl gear, CD = trawl gear with KED, SDCT = trawl gear with long sweeps, CDCT = trawl gear with KED plus long sweeps.

The percent assimilation of shrimp biomass suggested that the CD, SDCT, and CDCT trawls had higher catches than the control. In one instance the SDCT and CDCT had a chrimp catch of 81% and 136%, respectively, suggesting that sweeps influence the shrimp catch while CD showed catches of 36% compared with the control gear.

The Kruskal–Wallis test revealed that the bycatch biomass caught were significantly different between all treatments. Similar results were observed in the post hoc test (Mann–Whitney U-test), which also revealed that in paired evaluations the control gear, CD, SDCT, and CDCT had different bycatch biomasses (Table 3).

The worst proportions of shrimp catch to bycatch were found in the control gear (1:23.8) and SDCT (1:18.8). However, the trawl nets with KED showed better shrimp catch to bycatch ratios; in CD and CDCT these were 1:8.9 and 1:13.4, respectively.

Rake Stardrum was the most frequently caught fish species across all hauls (3,104 individuals). The largest numbers were caught in SDCT trawls (1,020 individuals). The CDCT trawls caught 908 individuals, CD trawls caught 652 individuals, and the control gear trawls caught 524 individuals. Atlantic Bumper was the second most abundant species in our study (2,784 individuals) and was present in all hauls. This species was most predominantly found in the control gear (2,278 individuals) and SDCT (397 individuals) trawls. However, hauls using the KED reduced their capture. In CD and CDCT trawls 46 and 63 individuals were caught, respectively.

American Coastal Pellona was the third most abundant species captured in trawl trials (456 individuals). The largest catches of this teleost were observed in the control gear (357 individuals) and the lowest were observed in CDCT (six individuals). Whitemouth Croaker was the fourth most abundant bycatch species found in our study (162 individuals). The largest catches of this species were observed in CDCT (77 individuals) and SDCT (63 individuals). Much smaller numbers were noted in the control gear (12 individuals) and CD (10 individuals).

Finally, Banded Croaker was the fifth most abundant species (153 individuals). The largest catches occurred in CDCT (63 individuals) and SDCT (62 individuals). The smallest catches were observed in the control gear (18 individuals) and CD (10 individuals).

According to the shrimp catch results, Atlantic seabob shrimp was the most proeminent shrimp species caught across all hauls (97.4%). The largest catches were observed in CDCT (479 individuals), followed by SDCT (401 individuals). The smallest catches were observed in CD (303 individuals) and the control gear (238 individuals).

DISCUSSION

Trawl trials that included sweeps (SDCT and CDCT) resulted in a greater shrimp assimilation compared with those without sweeps. In paired evaluations, SDCT caught 81% more shrimp biomass than did the control gear and CD, and CDCT caught 135% more shrimp biomass than did the control gear and CD. It is possible the larger catches reflect the larger area swept during those trawls; the areas swept by SDCT and CDCT were 21% higher than those swept by the control gear and CD.

Also, the configuration of the sweeps may have contributed to the increase in shrimp catch. According to Kim and Wardle (1998), trawl gears with sweeps usually catch a greater biomass. Notably, the friction generated by the sweeps in contact with the marine substrate contributes to greater sediment disturbance, forcing shrimp and demersal fishes towards the front of the trawl mouth.

Trawls with KED and sweeps together exhibited greater shrimp catches than the control gear. Comparing the control gear with CD, we observed a 36% increase of shrimp catch in CD compared with the control gear. It is important to highlight that SDCT and CDCT performed better. In paired evaluations, CDCT had 55% higher shrimp biomass assimilation compared with SDCT. Thus, the use of these proposed devices can improve shrimp catches. In previous work, Brewer et al. (1998) suggested that using fisheyes in bottom trawls (note: KED is a fisheye plus an escape window) can increase the size of shrimp catches. Other authors found the same pattern when using fisheyes in shrimp trawls (Heales et al. 2008).

Usually, fishers do not adopt the BRDs because of the possibility of reducing the target species catches. Some authors (Broadhurst et al. 1996, 2002) suggested that rigid grids on trawl gear can reduce the bycatch but can also contribute to reductions in shrimp catch. This pattern was not observed in our study; our results suggest the use of BRDs in Brazilian artisanal trawl fisheries will increase shrimp catch and reduce bony fish bycatch.

The KED performance can be linked to its location on the cod end. The distance of this BRD to the cod end can significantly affect bycatch as well as shrimp catch. Alió et al. (2009) implemented the fisheye at 75 and 150 cm from the cod end. In the first case they observed a 40% loss in shrimp, while in the second implementation only 5% of the shrimp escaped from the fishing gear. In this study, the KED distance from the cod end was 210 cm. This is similar to the ideal distance established by Alió et al. (2009). However, these authors adopted the fisheye without an escape window; i.e., the device only provided an elliptical opening from which fish could escape.

The CD hauls presented the highest bycatch reduction compared with the other modalities. The CD excluded 75% of bycatch, followed by CDCT (63.5%) and SDCT (48%). Trawl gear that included sweeps showed a higher bycatch biomass assimilation compared with the control gear. The negative exclusion percentages in SDCT (−53%) and CDCT (−33%) suggest that the increase in bycatch biomass resulting from the use of sweeps was a consequence of catching larger individuals compared with those caught in the control gear (Table 4). Somerton et al. (2007) explained that the effect of sweeps on the catch composition depends on the species and its behavior. Additionally, those authors suggested that smaller fish can escape more easily from this device. Engas and Godo (1989) and Andrew et al. (1991) found an increase in the length of the fish caught when sweeps were present and a positive correlation between the size of fish caught and the length of the sweeps. Shrimp catches can have a negative impact on the marine environment (Broadhurst 2000). Thus, one of the main challenges of this fishing activity is to reduce bycatch, which is partially composed of small individuals without commercial value (Graça-Lopes 1996). So, the implementation of sweeps may assist with this problem, maximizing the exclusion of juveniles.

TABLE 4. Size of bycatch and its biomass caught by treatment. SD = control trawl gear, CD = trawl gear with KED, SDCT = trawl gear with long sweeps, CDCT = trawl gear with KED plus long sweeps.

Treatments	Number of individuals	Biomass (kg)
SD	3,398	15.34
CD	873	8.31
SDCT	1,764	23.28
CDCT	1,243	20.74
Total	7,278	67.66

There is a large proportion of bycatch relative to shrimp caught in bottom trawls (Branco and Verani 2006). Those authors estimated the proportion of shrimp catch to bycatch at 1:19. In the present study, the the control gear and SDCT exhibited similar proportions to that of Branco and Verani (2006). A greater proportion of bycatch relative to shrimp was observed in the control gear (1:23.8), followed by SDCT (1:18.8). However, in the trawl equipped with a KED we observed a significant reduction in the proportions of shrimp catch to bycatch; in CDCT, the shrimp : bycatch ratio was 1:13.4 and in CD it was 1:8.9.

The results showed that compared with the other treatments the best CPUEs can be achieved by CD and CDCT. The CD was efficient in catching shrimp and excluding bycatch. However, despite the CDCT catching greater bycatch biomass than the control gear and CD, the shrimp catch was also greater than in the other hauls. In principle, this amplified effect suggests that the combined use of sweeps plus KED in the fishing gear will improve selectivity. However, considering the potential negative impacts of sweeps on the benthic environment, CD is the best choice.

When evaluating the size of the bycatch for the most abundant teleost species, the greatest reduction occurred in trawls using KED devices, i.e., CDCT and CD, compared with the control gear and SDCT, for which there was an increase of bycatch. Catches of Atlantic Bumper and American Coastal Pellona were significantly lower in SDCT and CDCT than in the control gear and CD, respectively. According to Fischer et al. (2011), these species are pelagic and inhabit shallow coastal waters, making them susceptible to bottom trawl gears. These results demonstrate that the KED and sweeps are potentially effective at selecting for pelagic species.

Rake Stardrum, Whitemouth Croaker, Shorthead Drum Larimus breviceps, and red spotted shrimp were predominantly found in SDCT and CDCT trawl nets, compared with the control gear and CD. This increased assimilation of these demersal species in SDCT and CDCT corroborates with the literature and can be explained by the increase in the size of the swept areas covered by these hauls (Andrew et al. 1991; Kim and Wardle 1998; Somerton and Munro 2001; Weinberg and Somerton 2006). Therefore, the CD showed better selectivity compared with the other fishing gears.

The use of CDCT performed poorly in reducing the demersal fish assimilation. Most likely the use of KEDs on the trawl’s top panel plus long sweeps may have amplified the potential retention of demersal species. This may explain why the species caught most frequently in these trawls was Rake Seadrum (43% of the total catch), a species with demersal behavior.

CONCLUSIONS

The adoption of sweeps in shrimp trawls can increase bycatch and shrimp catches. Additionally, sweeps can ensure a considerable reduction in juvenile teleost fish catches. Bycatch exclusion can be increased with a KED, a BRD that provides a 15% increase in escape area around the cod end. The results obtained by trawls using a KED suggest that this technology can be adopted as a management alternative for artisanal fishing. This device not only reduces the bycatch assimilation but can also increase the shrimp catch. The CD showed a better selectivity performance, as evident by the lowest shrimp-to-bycatch ratio (1:8.9), confirming that this proposed BRD is an effective alternative solution to reducing bycatch. Thus, we conclude that the optimization of this fishing method will improve shrimp catches and reduce bycatch.

ACKNOWLEDGMENTS

This study is part of the Programa de Apoio à Pesquisa, which was financed by Fundação de Amparo à Pesquisa e Inovação do Estado de Santa Catarina and Universidade do Estado de Santa Catarina.