Juvenile and Trash Fish Excluder Device (JTED) for Taiwanese Bottom Trawl

Abstract

Juvenile and trash fish excluder devices (JTEDs) are used to mitigate the ecological impact of trawling. Trawling has a large ecological footprint in Taiwan. Thus, this study designed three JTEDs, which differed only in grid interval, for single boat bottom otter trawls that are commonly used in Taiwan. The designed JTEDs were tested in the waters off of Kaohsiung on a Kaohsiung-based bottom trawler, and the results were as follows. (1) At least 300 species were caught, and 189 of these species had little commercial value; their weight accounted for 52.5% of the total catch. (2) The three JTEDs had good exclusion performance. (3) The ratios of the weight of bycatch excluded to the weight of total catch were 7.9%, 29.1%, and 36.2%, for grid intervals of 10, 15, and 20 mm, respectively. (4) JTEDs with grid intervals of 10, 15, and 20 mm were associated with a 0.4%, 11.5%, and 19.0% loss in revenue, respectively. (5) Mature (and thus commercially valuable) Loliginidae tended to be excluded when the grid interval was 20 or (especially) 15 mm. In general, the designed JTEDs should be used at a grid size of 15 mm because this strikes the best balance between profitability and sustainability.

1. Introduction

Bycatch and discarded catch greatly contribute to the depletion of marine fishery resources worldwide and must be managed to ensure the sustainability of marine ecosystems [1,2,3]. In 2009, Davies et al. estimated that bycatch accounted for 40.4 percent of global marine catches [4]. Pérez Roda et al. estimated that the annual discards in global marine fisheries from 2010 to 2014 were about 9.1 million tons, and bottom trawlers had the most serious discards, accounting for 45% of the total discard [3,5]. Marine wildlife that is discarded after trawling is unlikely to survive [6,7,8].

Bycatch reduction devices (BRDs) have been widely installed on trawls to reduce bycatch without compromising fishing yield. For example, a study reported that the turtle excluder devices and square-mesh codend used by shrimp trawl fisheries in Queensland, Australia, reduced the bycatch weight ratio by 29% without affecting the fishing yield of Melicertus plebejus [9]. Another study reported that Nordmøre grids and radial escape sections reduced the bycatch weight ratio by 74% and 36%, respectively, at a loss of 4% and 11% in the fishing yield, respectively, when used by shrimp trawl fisheries [10]. In another study, rope grids reduced the weight ratio of four main bycatch types by 36–50% without affecting the fishing yield of Pandalus borealis when used by shrimp trawl fisheries in the Gulf of Maine in the United States [11]. Size-sorting grids were also discovered to reduce the bycatch weight ratio by 33% at a slight loss to the fishing yield [12]. A Japanese study designed a BRD, called SURF-BRD, that was better able to exclude the small size of crabs, because the catch was more likely to come into contact with the front panel of the BRD [13,14]. A novel double-grid system in which 80% of a catch comes into contact with at least one grid has also been used by Norwegian cod trawlers. However, this grid system did not outperform the conventional Sort-V grid in excluding Sebastes spp. [15]. Nonetheless, the double-grid system reduced the bycatch weight ratio of small deep-sea shrimp, juvenile Sebastes spp., and juvenile Reinhardtius hippoglossoides by 45%, 16%, and 32%, respectively, when used by Norwegian deep-sea shrimp trawlers [16].

In Taiwan, trawling is commonly used for coastal fishing and has a large ecological footprint. At the end of 2020, Taiwan had more than 1300 trawlers; 830 were trawlers weighing less than 50 t and operating in sea areas no less than 3 nautical miles from the shore, and 491 were trawlers weighing more than 50 t and operating in sea areas no less than 12 nautical miles from the shore. Trawl fisheries once accounted for over 50% of the production of Taiwanese coastal fisheries and have maintained a stable annual catch of 32,000 t over the past decade. Fish from trawling is affordable on the Taiwanese market and is a pillar of food self-sufficiency in Taiwan. However, coastal fishery resources in Taiwan have long been strained by the large amounts of bycatch (of juvenile and trash fish) from trawlers.

Data on bycatch tend to be scarce because bycatch from Taiwanese trawlers is typically conveyed back to port and traded outside the commercial market. Therefore, fishery authorities in Taiwan launched a port inspection system 10 years ago to gather data on the catch of coastal fisheries. According to such data for <50 t trawlers based in Kaohsiung and Pingtung over the last 10 years, bycatch has accounted for 41.5–65.3% of the total catch by weight.

Since 1998, the Southeast Asian Fisheries Development Center has developed four types of juvenile and trash fish excluder devices (JTEDs) with the support of the Food and Agriculture Organization of the United Nations. One such JTED has a semicurved rigid sorting grid, and it was found to be effective in tests involving shrimp trawls in Southeast Asia [17]. However, Taiwanese trawl fisheries differ from their Southeast Asian counterparts. For example, Southeast Asian trawl fisheries primarily catch shrimp, whereas Taiwanese trawl fisheries catch various types of fish. Consequently, foreign JTEDs introduced directly to Taiwan might not be appropriate for Taiwanese trawls. Thus, this study referenced Southeast Asian designs [18,19,20,21,22,23,24,25,26,27,28] to design a JTED that is appropriate to the trawl types, fishery characteristics, and marine ecosystems found in Taiwanese waters.

Taiwanese trawlers are more interested in commercial profitability than ecological sustainability. Thus, this study focuses on the use of underwater photography systems to observe the behavior of catches (e.g., fish, crustaceans, and cephalopods) on grids, and the loss of income from the use of JTED.

2. Materials and Methods

2.1. Design of Proposed JTED

Single boat bottom otter trawlers are commonly used in Taiwan. They are mostly distributed along the west coast of Taiwan (including Taiwan’s offshore islands) and primarily used to specifically catch various types of fish. This study used Kaohsiung-based 20-to-50-t trawlers employing single boat bottom otter trawls as the vessels of interest. The structure of these trawlers and their trawls were analyzed to develop a JTED for the codend of bottom trawls. The JTED designed by this study comprises a funnel net, sorter, guide buoy, escape opening, interception net, guide board, catch entrance, and codend (Figure 1).

The funnel net coaxes the catch that has entered the codend toward the sorting area of the sorter. If the size of the catch is smaller than the grid size of the sorting area, the catch can escape through the grid; this design provides improved exclusion of juvenile and trash fish relative to existing ones. The funnel net is made of polyethylene and is 150-cm long. The mesh size is 23.3 mm. The front of the funnel net has a large circular opening with a diameter equal to that of the codend; the fringe of the circular opening is stitched to the codend. The front opening is 330 cm from the bottom of the sorter. The rear of the funnel net has a small circular opening with a diameter of approximately half that of the front opening. The rear opening is 160 cm from the top of the sorter. The top of the rear opening is stitched to the top of the barrel of the codend, and the bottom of the rear opening is left to dangle. This design allows the rear opening and the sorting area of the sorter to remain at the same level, and the catch is more likely to come into contact with the sorting area as a result.

The guide buoy is a plastic spherical buoy with a diameter of 25.4 cm. It is attached to the top of the sorter frame and ensures that the JTED-equipped section of the trawl remains buoyant. It also stabilizes the direction of the JTED and prevents the JTED from rotating substantially during trawling.

The escape opening is located behind (relative to a vector pointing toward the bottom of the codend) the top of the sorter. The escape opening has a length and width of 100 and 90 cm, respectively. The escape opening allows bycatch fish that have passed through the sorter to escape the trawl.

The interception net has small meshes that prevent bycatch fish that have passed through the sorter from swimming back to the codend [29]. The bottom of the interception net is connected to the bottom of the guide board of the sorter. The interception net extends toward the end of the escape opening (i.e., toward the bottom of the codend) and its top is pulled taut and stitched to the end of the escape opening. The left and right sides of the interception net are stitched along the barrel of the codend to guide the juvenile and trash fish excluded by the sorter to the escape opening, from which they can escape into the sea.

In essence, the sorter is the key component that determines the JTED’s exclusion performance; its meshes must be small enough to allow the larvae of large marketable fish or small fish with little commercial value to escape but must be large enough to trap larger catch. This study thus designed three sorters that differed in their grid intervals; the sorters were designed as such for the following reasons.

• This study referenced 20-to-50-t Taiwanese trawlers with single boat bottom otter trawls, and these trawls had sorters that were made of stainless steel and had a circular shape for durability and rust resistance. The upper half of the sorter was the sorting area; catch that came into contact with this screen could escape from the trawl. The lower half was divided into two sections. The upper section was the guide board, which was bent 45° from the bottom of the sorting area toward the end of the codend. It guided the catch that could not pass through the sorter toward the bottom of the trawl. The lower section was an empty region that drove the catch that could not pass through the sorter to swim toward the end of codend, where it was trapped.
• Considering the codend specifications and the width of the net hauling channel at the quarter of the reference trawlers, this study set the frame diameters in two sorter designs to 131 cm (Figure 2 and Figure 3). The reference trawlers usually carried two or more trawls. One trawl was rolled on the net hauler for use, whereas spare trawls were stored in the net hold at the quarter of the fishing boat. However, because the aforementioned frame diameters in the two sorters were larger than the opening of the net hold (the maximum diagonal length was 115 cm), the diameter of the third designed sorter was set at 110 cm (Figure 4).
• In the present day, Taiwan’s bottom otter trawlers use codends with 20-mm meshes, and the diamond-shaped meshes open horizontally during trawling. If the trawl is completely open (90°) during trawling, the maximum horizontal opening of the mesh is 14.3 mm. Therefore, a grid interval of 15 mm was chosen for the first sorter design (JTED_(15mm)). Because the diamond-shaped meshes of codends used by Taiwan’s bottom otter trawlers open 60° during trawling and because the opening is approximately 10 mm, a grid interval of 10 mm was chosen for the second sorter design (JTED_(10mm)). For the third sorter design, a larger grid interval (20 mm) and mesh angle were selected (JTED_(20mm)).

2.2. Experimental Vessel and JTED Installation Method

The aforementioned JTED design was evaluated on a Kaohsiung-based trawler, named Shin Fu Jin, that had a bottom otter trawl. The trawler had a gross tonnage of 32.8 t and length of 20.9 m. It was equipped with a six-cylinder, 640-hp diesel engine with a maximum speed of 1650 rpm. The fish finder was a Suzuki Navigation Sounder ES-1029. The vessel had an 880-cm-long codend with 400 meshes along its width; the mesh was 20 mm.

Three JTEDs (each with one of the three sorter designs) were installed on the original codend of the experimental vessel. The bottom of the sorters was installed 460 cm from the end of the codend. According to a Japanese study, bycatch exclusion is improved if more catch comes into contact with the sorter [13]. Referencing previous Taiwanese studies, the present study tilted the top of the sorters 10° toward the headrope of the trawl. A 440-cm-long cover net with 300 meshes along its width was stitched behind the escape opening (Figure 5); the mesh was 20 mm. This net was used to determine the ability of the JTEDs to both exclude bycatch and retain target catch. An underwater photography system comprising a GoPro HERO8 camera, Recsea WHG-HERO8 deep diving waterproof case, and SUPE Scubalamp V6K Pro v2 photoflood lamp (the luminous flux was set at 6000 lumen; Figure 6) was also installed on the sorter frame; this system was used to observe the activity of the catch in the codend during each haul.

2.3. Test Conditions

Eight exploratory fishing trips with five hauls each trip were undertaken; these trips were undertaken between 30 April and 26 August 2021, in a sea area off of Kaohsiung approximately 4.3–9.7 nautical miles from the nearest shore (Figure 7). Each trawling test lasted approximately 1 h; 10 min was spent releasing the trawl, 35 min was spent trawling, and 15 min was spent hauling the trawl back in. JTED_(10mm), JTED_(15mm), and JTED_(20mm) were evaluated in two, four, and two exploratory fishing trips, respectively. The trawling speed was 3.2–4.5 knots, and the trawling depth was 26–127 m.

JTED performance was indicated by the weight ratio of desirable (i.e., large and marketable) catch to bycatch. To determine this ratio, the catches in the cover net (coded A) and codend (coded B) were weighed after each haul. For catches >2 kg, a random sample comprising all target catches and a subset of bycatches weighing 2 ± 0.5 kg were taken, and the weights of these two types of catches were used to estimate the actual weight ratio.

3. Results

3.1. Notable Observations from Underwater Photographs

3.1.1. Fish

Fish continued to swim toward the headrope of the codend. Large fish were strong and often swam at the same speed as the trawl. Therefore, they maintained an adequate distance from the sorter in the codend (Figure 8a). However, over time, some fatigued fish rested (i.e., stopped) on the surface of the sorting area (Figure 8b) before continuing to swim. Smaller fish did not swim as fast, and the current forced them to come into contact with the surface of the sorter after they passed through the funnel net. They successfully passed through the sorting area and escaped (Figure 8c). The juvenile and trash fish that escaped the JTED during the test did not actually escape, and some of them rested on the surface of the interception net behind the JTED sorter (Figure 8d,e); these fish waited until the trawler had slowed down to change direction or until the net was hauled before passing back through the sorter and into the codend (Figure 8f).

3.1.2. Cephalopods

Cephalopods do not have streamlined bodies; therefore, unlike small fish, they rarely passed directly through the JTED sorter when they came into contact with the sorting area. They typically used their tentacles to attach themselves to the surface of the sorting area and remained there (Figure 9a). Because cephalopods, particularly Loliginidae, have soft bodies, they could squeeze through the sorter and escape by using their ability to climb (Figure 9b). Some cephalopods could not support themselves and fell off the surface of the sorter (Figure 9c). The underwater photography footage indicated that, unlike small fish, the cephalopods that escaped through the sorter did not return to the codend.

3.1.3. Crabs

Crabs are relatively slow swimmers because of their large weight. Therefore, most crabs dropped to the catch entrance below the sorter after they passed through the funnel net (Figure 9d), directly entering the end of the codend. Only a few crabs were able to come into contact with the sorting area in the upper half of the sorter and cling to it using their claws (Figure 9e,f). However, they eventually fell off the sorting area and were transported to the end of the codend by being swept along by the current or by being knocked down by other catch.

3.2. Composition of Catches

The total weight of the randomly collected catches was 548 kg, and that of the catches after extrapolation was 814 kg (

Table 1. Weight of catches in tests of proposed JTEDs.

Haul	Date	Grid Interval (mm)	(a) Weight of Samples (kg)			(b) Extrapolated Total Weight (kg)			Weight Ratio of Excluded Catch (%) (1/3)
			Cover Net	Codend	Subtotal	Cover Net (1)	Codend (2)	Subtotal (3)
1	20210430	15	0.97	11.51	12.5	0.97	11.51	12.5	7.8
2	20210430	15	1.16	21.24	22.4	1.16	21.24	22.4	5.2
3	20210430	15	2.30	10.33	12.6	9.78	26.77	36.5	26.8
4	20210430	15	2.04	10.33	12.4	5.49	18.37	23.9	23.0
5	20210430	15	1.87	7.82	9.7	4.30	14.56	18.9	22.8
6	20210513	15	4.88	11.55	16.4	29.30	11.55	40.8	71.7
7	20210513	15	4.04	10.87	14.9	12.12	11.41	23.5	51.5
8	20210513	15	2.63	5.49	8.1	7.88	10.96	18.8	41.8
9	20210513	15	1.24	8.25	9.5	1.24	8.25	9.5	13.0
10	20210513	15	1.25	18.60	19.9	1.25	18.60	19.9	6.3
11	20210603	15	2.72	13.81	16.5	5.44	22.18	27.6	19.7
12	20210603	15	2.15	10.44	12.6	5.90	12.21	18.1	32.6
13	20210603	15	1.72	9.92	11.6	1.72	9.92	11.6	14.8
14	20210603	15	2.91	7.67	10.6	2.91	7.67	10.6	27.5
15	20210603	15	3.00	10.72	13.7	3.00	10.72	13.7	21.8
16	20210610	15	2.84	14.49	17.3	8.53	29.11	37.6	22.7
17	20210610	15	2.11	13.85	16.0	4.79	13.85	18.6	25.7
18	20210610	15	1.71	9.10	10.8	1.71	9.10	10.8	15.8
19	20210610	15	3.29	7.14	10.4	8.47	11.23	19.7	43.0
20	20210610	15	2.48	9.68	12.2	2.48	9.68	12.2	20.4
21	20210709	20	4.88	12.47	17.3	11.51	12.47	24.0	48.0
22	20210709	20	4.51	7.88	12.4	4.51	7.88	12.4	36.4
23	20210709	20	1.86	12.99	14.9	1.86	12.99	14.9	12.5
24	20210709	20	4.75	9.46	14.2	47.52	79.99	127.5	37.3
25	20210709	20	2.28	5.91	8.2	2.28	5.91	8.2	27.8
26	20210819	20	4.67	9.82	14.5	7.93	9.82	17.7	44.7
27	20210819	20	3.84	6.65	10.5	3.84	6.65	10.5	36.6
28	20210819	20	1.93	8.12	10.0	1.93	8.12	10.0	19.2
29	20210819	20	5.99	12.69	18.7	5.99	12.69	18.7	32.1
30	20210819	20	4.48	5.07	9.5	4.48	5.07	9.5	46.9
31	20210823	10	2.54	9.65	12.2	2.54	9.65	12.2	20.9
32	20210823	10	2.26	18.03	20.3	2.26	18.03	20.3	11.1
33	20210823	10	2.14	21.09	23.2	2.14	27.14	29.3	7.3
34	20210823	10	1.04	14.87	15.9	1.04	14.87	15.9	6.5
35	20210823	10	0.43	36.15	36.6	0.43	36.15	36.6	1.2
36	20210826	10	1.57	9.10	10.7	1.57	9.10	10.7	14.7
37	20210826	10	0.11	4.77	4.9	0.11	4.77	4.9	2.3
38	20210826	10	0.32	6.50	6.8	0.32	6.50	6.8	4.7
39	20210826	10	0.79	4.17	5.0	0.79	4.17	5.0	15.9
40	20210826	10	0.97	10.68	11.7	0.97	10.68	11.7	8.4
Subtotal		10	12.17	135.01	147.17	12.17	141.06	153.22	7.9
		15	47.28	222.78	270.06	118.41	288.87	407.28	29.1
		20	39.19	91.03	130.22	91.85	161.57	253.42	36.2
Total			98.6	448.8	547.5	222.4	591.5	813.9	27.3

). Fish, crustaceans, cephalopods, conches, and other catches accounted for 71.3%, 13.2%, 11.3%, 2.4%, and 1.8% of the total catch weight, respectively. The catches comprised 111 and 189 species that were marketable and that had little economic value, respectively; they accounted for 47.5% and 52.5% of the total catch weight, respectively.

3.3. Overall JTED Performance

The ratio of excluded bycatch weight to total catch weight for JTED_(10mm), JTED_(15mm), and JTED_(20mm) was 7.9%, 29.1%, and 36.2%, respectively (Table 1). In total, 146 species were excluded by the JTEDs (i.e., the species caught in the cover net in the 40 hauls); 23 of these species each and collectively accounted for >0.5% and 94% of the catch weight in the cover net, respectively. However, seven of these 23 species were marketable catch (21%;

Table 2. Main species of catch excluded by JTEDs.

No.	Species	Weight Ratio of the Catch in the Cover Net (%)				Weight Ratio of the Catch Excluded by JTEDs (Cover Net/Codend)
		Total	10 mm	15 mm	20 mm	10 mm	15 mm	20 mm
Marketable catch		21.0	3.2	21.5	41.2
1	Loliginidae	10.9	1.7	16.7	10.4	9.8	58.1	61.7
2	Trichiurus japonicus	3.8	0.9	1.3	12.3	1.0	25.6	70.1
3	Aurigequula fasciatus	1.7	-	-	7.3	-	-	100.0
4	Psenopsis anomala	1.4	-	-	6.0	-	-	34.0
5	Sepia madokai	1.3	0.3	1.9	1.5	0.8	8.9	23.9
6	Upeneus japonicus	1.0	0.3	1.4	0.9	0.3	17.8	16.3
7	Carangoides equula	0.7	-	0.2	2.8	-	19.9	29.1
Catch with little or no economic value		73.3	94.3	69.2	53.8
1	Equulites rivulatus	21.3	40.1	14.5	12.2	66.7	69.0	68.0
2	Synagrops philippinensis	12.7	28.1	5.0	9.7	38.8	60.7	64.9
3	Acropoma japonicum	10.8	1.7	21.3	1.0	19.7	46.9	38.7
4	Metapenaeopsis provocatoria longiroitris	6.1	4.0	7.5	5.9	36.8	56.7	63.0
5	Benthosema pterotum	4.9	5.4	5.6	2.6	75.9	98.9	93.1
6	Photopectoralis aureus	4.7	3.3	5.0	5.7	24.4	82.7	73.0
7	Champsodon snyderi	3.9	5.3	2.5	5.1	43.8	68.3	74.3
8	Gymnothorax minor	2.1	0.5	3.1	1.9	3.7	57.4	31.1
9	Saurida umeyoshii	1.6	-	2.7	1.4	-	53.4	50.1
10	Psettina tosana	1.0	1.6	0.7	1.1	17.2	59.0	56.3
11	Lancer rose shrimp	0.9	0.7	0.3	2.5	39.5	52.1	77.7
12	Malakichthys wakiyae	0.9	0.7	-	2.8	71.3	N/A	82.2
13	Repomucenus virgis	0.7	1.5	0.1	1.0	34.9	N/A	79.1
14	Synodus fuscus	0.6	0.0	0.0	-	10.9	27.3	N/A
15	Abralia multihamata	0.6	0.1	1.0	0.3	80.0	79.9	73.8
16	Chlorophthalmus nigromarginatus	0.6	1.4	-	0.7	69.6	N/A	93.7
Total		94.3	97.5	90.6	95.0

). The performance of each JTED sorter was as follows.

• JTED_(20mm), JTED_(15mm), and JTED_(10mm) excluded the most to least marketable catches, respectively; the ratios of excluded marketable catch weight to total catch weight were 41.2%, 21.5%, and 3.2%, respectively. JTED_(20mm), JTED_(15mm), and JTED¬_(10mm) excluded the least to most bycatch, respectively; the ratios of excluded bycatch weight to total catch weight were 53.8%, 69.2%, and 94.3%, respectively.
• More marketable catch was excluded when the grid interval was higher. For example, the weight of Trichiurus japonicas fishes excluded comprised 1%, 25.6%, and 70.1% of the total weight of T. japonicas fishes caught, respectively. For Loliginidae, the corresponding figures were 9.8%, 58.1%, and 61.7% for JTED_(20mm), JTED_(15mm), and JTED¬_(10mm), respectively. With regard to bycatch exclusion, JTED_(10mm) performed the worst and JTED_(15mm) and JTED_(20mm) had similar performance except for Equulites rivulatus, Acropoma japonicum, Benthosema pterotum, and Photopectoralis aureus, in which JTED_(15mm) outperformed JTED_(20mm).

3.4. Commercial Loss from JTEDs

In general, trawl fishers are more interested in commercial profitability than ecological sustainability. Thus, this study calculated the commercial loss caused by from the exclusion of marketable catch (

Table 3. Main marketable catches excluded by JTEDs and the associated loss in income.

Species		Number of Catch (n)					Ratio of the Number of Catch that Escaped Through JTED (%)					Size of Excluded Catch(cm)				Income of Catch (NT$)					Loss of Income (NT$)					Ratio of Loss of Income (%)
		Total	10 mm	15 mm	20 mm		Total	10 mm	15 mm	20 mm		10 mm	15 mm	20 mm		10 mm	15 mm	20 mm	Subtotal		10 mm	15 mm	20 mm	Subtotal		10 mm	15 mm	20 mm
(Fish)																22,119	21,522	10,106	53,748		45	430	2039	2514		0	2	20
※	Priacanthus macracanthus	157	15	105	37		-	-	-	-		N/A	N/A	N/A		1787	12,121	4008	17,916		-	-	-	-		-	-	-
※	Evynnis cardinalis	256	141	75	40		-	-	-	-		N/A	N/A	N/A		1528	679	365	2573		-	-	-	-		-	-	-
	Sphyrna lewini	4	-	4	-		-	N/A	-	N/A		N/A	N/A	N/A		-	691	-	691		-	-	-	-		N/A	-	N/A
	Drepane punctate	5	-	5	-		-	N/A	-	N/A		N/A	N/A	N/A		-	864	-	864		-	-	-	-		N/A	-	N/A
	Trichiurus nanhaiensis	13	10	3	-		-	-	-	N/A		N/A	N/A	N/A		305	110	-	415		-	-	-	-		-	-	N/A
	Plectorhinchus cinctus	3	-	3	-		-	N/A	-	N/A		N/A	N/A	N/A		-	1178	-	1178		-	-	-	-		N/A	-	N/A
※	Epinephelus awoara	10	3	5	2		-	-	-	-		N/A	N/A	N/A		793	1354	45	2193		-	-	-	-		-	-	-
	Sphyraena forsteri	10	1	8	1		-	-	-	-		N/A	N/A	N/A		29	249	30	308		-	-	-	-		-	-	-
	Johnius macrorhynus	13	-	13	-		-	N/A	-	N/A		N/A	N/A	N/A		-	260	-	260		-	-	-	-		N/A	-	N/A
	Glaucosoma buergeri	1	1	-	-		-	-	N/A	N/A		N/A	N/A	N/A		1170	-	-	1170		-	-	-	-		-	N/A	N/A
	Epinephelus coioides	1	-	1	-		-	N/A	-	N/A		N/A	N/A	N/A		-	764	-	764		-	-	-	-		N/A	-	N/A
	Sphyraena barracuda	1	-	1	-		-	N/A	-	N/A		N/A	N/A	N/A		-	197	-	197		-	-	-	-		N/A	-	N/A
	Zeus faber	1	1	-	-		-	-	N/A	N/A		N/A	N/A	N/A		195	-	-	195		-	-	-	-		-	N/A	N/A
	Ariomma indicum	6	3	-	3		-	-	N/A	-		N/A	N/A	N/A		70	-	62	132		-	-	-	-		-	N/A	-
	Pennahia pawak	10	10	-	-		-	-	N/A	N/A		N/A	N/A	N/A		93	-	-	93		-	-	-	-		-	N/A	N/A
	Branchiostegus auratus	2	1	1	-		-	-	-	N/A		N/A	N/A	N/A		68	52	-	120		-	-	-	-		-	-	N/A
	Pennahia argentata	1	-	1	-		-	N/A	-	N/A		N/A	N/A	N/A		-	7	-	7		-	-	-	-		N/A	-	N/A
	Sphyraena japonica	1	1	-	-		-	-	N/A	N/A		N/A	N/A	N/A		18	-	-	18		-	-	-	-		-	N/A	N/A
※	Trachurus japonicus	1270	1234	-	36		0.6	0.2	N/A	13.9		16~17 (FL)	N/A	13~15 (FL)		3149	1	75	3225		8	-	8	16		0.3	-	10.7
	Saurida elongata	84	23	48	13		2.4	-	2.1	7.7		N/A	14 (FL)	16 (FL)		25	221	36	282		-	1	1	2		-	0.5	3.7
	Nemipterus bathybius	104	47	47	10		2.9	-	4.3	10.0		N/A	9~10 (FL)	17 (FL)		256	274	57	587		-	2	8	11		-	0.9	14.5
	Decapterus maruadsi	76	-	67	9		5.3	N/A	3.0	22.2		N/A	11~20 (FL)	5~21 (FL)		-	288	27	316		-	1	4	4		N/A	0.2	14.3
※	Upeneus japonicus	1063	796	187	80		6.9	0.6	29.4	16.3		11~13 (FL)	5~13 (FL)	10~14 (FL)		2587	451	304	3342		15	126	52	193		0.6	28.0	17.0
※	Branchiostegus albus	6	3	3	-		16.7	-	33.3	N/A		N/A	14 (TL)	N/A		2925	38	-	2963		-	9	-	9		-	22.6	N/A
※	Psenopsis anomala	167	38	23	106		19.2	-	17.4	26.4		N/A	15~16 (FL)	13~20 (FL)		795	256	2389	3440		-	-	593	593		-	-	24.8
	Carangoides equula	126	4	16	106		19.8	-	31.3	18.9		N/A	11~14 (FL)	11~17 (FL)		64	154	1141	1358		-	37	224	261		-	24.0	19.6
	Nemipterus virgatus	37	3	22	12		21.6	-	22.7	25.0		N/A	5~11 (FL)	13~15 (FL)		10	144	43	197		-	5	14	19		-	3.7	33.2
	Dentex hypselosomus	134	45	71	18		27.6	-	38.0	55.6		N/A	5~9 (FL)	9~14 (FL)		694	327	200	1221		-	85	105	191		-	26.0	52.7
※	Trichiurus japonicus	637	323	126	188		36.7	2.8	53.2	84.0		23~50 (TL)	12~57 (TL)	16~70 (TL)		1787	12,121	4008	17,916		21	164	1029	1214		0.4	19.5	77.7
(Curstaceans)																1447	2518	871	4836		-	10	-	10		-	0.4	-
	Penaeus japonicus	3	1	-	2		-	-	N/A	-		N/A	N/A	N/A		68	-	130	198		-	-	-	-		-	-	-
	Penaeus marginatus	12	2	7	3		-	-	-	-		N/A	N/A	N/A		15	45	41	101		-	-	-	-		-	-	-
	Metapenaeus ensis	6	-	5	1		-	N/A	-	-		N/A	N/A	N/A		-	6	1	7		-	-	-	-		-	-	-
※	Charybdis feriatus	15	4	10	1		-	-	-	-		N/A	N/A	N/A		595	1409	293	2296		-	-	-	-		-	-	-
	Portunus sanguinolentus	16	-	16	-		-	N/A	-	N/A		N/A	N/A	N/A		-	846	-	846		-	-	-	-		-	-	-
	Charybdis miles	40	34	2	4		-	-	-	-		N/A	N/A	N/A		770	14	87	870		-	-	-	-		-	-	-
	Charybdis amboinensis	40	-	19	21		2.5	N/A	5.3	-		N/A	2 (CW)	N/A		-	199	320	518		-	10	-	10		-	5.0	-
(Cephalopod)																6303	8616	4945	19,865		69	3300	993	4362		1.1	38.3	20.1
※	Amphioctopus aegina	98	38	44	16		3.1	-	6.8	-		N/A	4 (ML)	N/A		409	800	256	1465		-	9	-	9		-	1.2	-
※	Sepia madokai	726	195	421	110		42.0	20.5	50.6	47.3		1~4 (ML)	1~6 (ML)	1~9 (ML)		4946	2847	930	8723		19	254	108	380		0.4	8.9	11.6
※	Loliginidae	5708	146	4882	680		79.5	12.3	89.9	19.0		2~9 (ML)	1~13 (ML)	2~12 (ML)		948	4970	3759	9677		49	3037	885	3972		5.2	61.1	23.6
Total																29,870	32,656	15,922	78,448		113	3740	3032	6885		0.4	11.5	19.0

※: Main marketable catch in this study. “-“: The value or quantity is zero. “N/A”: Not applicable (because there is no corresponding value or quantity to calculate.

). The JTEDs excluded 29, 7, and 3 marketable species of fish, crustaceans, and cephalopods, respectively. These species collectively accounted for 90% of the total catch by weight. In general, JTED_(10mm), JTED_(15mm), and JTED_(20mm) resulted in 0.4%, 11.5%, and 19% lower revenue, respectively. These results were determined as follows.

3.4.1. Fish

Of the 29 species of common marketable fish caught, 18 species, including Priacanthus macracanthus, escaped through the JTEDs, resulting in commercial loss. Less than 10% of T. japonicas, Saurida elongate, Nemipterus bathybius, Decapterus maruadsi, and Upeneus japonicas fishes escaped through the JTEDs. Almost no fish could escape through JTED_(10mm), and the fish that escaped through JTED_(15mm) and JTED_(20mm) were relatively small and had little commercial value overall. The fishes from the aforementioned five species that escaped from the JTEDs over the study period had a combined market value of NT$227. Together with these escaped fishes, fishes from the aforementioned five species that were caught had a combined market value of NT$7751. Thus, the JTEDs resulted in an approximately 2.9% loss in revenue.

Six Branchiostegus albus fishes were also caught in this study, and only one small 14-cm B. albus (1/6 = 16.7%) escaped through JTED_(15mm). Nonetheless, small B. albus is not of commercial value. Psenopsis anomala, Carangoides equula, N. virgatus, and Dentex hypselosomus had a 0% probability of escaping JTED_(10mm) and 20–30% probability of escaping JTED_(15mm) and JTED_(20mm). The species that passed through JTED_(15mm) had little commercial importance because they were relatively small. In addition, although T. japonicas passed through all three JTEDs, only JTED_(20mm) excluded mature T. japonicas easily and thus resulted in a commercial loss.

In terms of fish catch, JTED_(15mm), JTED_(15mm), and JTED_(20mm) resulted in (almost) 0%, 2%, and 20% lower revenue, respectively.

3.4.2. Crustaceans

The JTEDs did not exclude any crustaceans. The seven species of common marketable crustaceans caught in this study were Penaeus japonicas, Pe. marginatus, Metapenaeus ensis, Charybdis feriatus, Portunus sanguinolentus, Ch. miles, and Ch. amboinensis.

3.4.3. Cephalopods

Three common marketable cephalopods were caught in this study: Amphioctopus aegina, Sepia madokai, and Loliginidae. Hardly any A. aegina escaped. Although many Se. madokai and Loliginidae escaped through the JTEDs (probabilities of escape: 20.5–47.3% and 12.3–89.9%, respectively), most of the Se. madokai that escaped were relatively small and thus had little commercial value. For example, 50% of the Se. madokai excluded in the three types of JTED had a mantle length of ≤4.5 cm.

In terms of cephalopod catch, JTED_(15mm), JTED_(15mm), and JTED_(20mm) resulted in an (almost) 0%, 38%, and 20% lower revenue, respectively.

4. Discussion

4.1. Comparison of the Proposed JTED with Existing JTEDs

According to Sistiaga et al. [30], the passageway of a four-panel flexgrid for catch is wider than that of a two-panel flexgrid. Because passageway size affects the probability of catches clogging the flexgrid, the four-panel flexgrid in that study was less likely to be clogged by catch compared with the two-panel flexgrid (p = 0.03). Nevertheless, both the two-panel and four-panel flexgrids increased the flow velocity in the trawl and reduced the probabilities that the grid would be clogged by catch and the trawl would tear [31]. The JTEDs designed in the present study had a funnel net and a sorter with a circular frame. The funnel net guided the catches toward the sorter, which increased the likelihood that the sorter would exclude bycatch. The diameter of the rear opening of the funnel net was approximately half that of the front opening, which increased the water velocity from the rear of the funnel net to the sorter. This guided large catches that could not pass through the sorter to where they could be transported by the current into the end of the codend and prevented catch clogging at the surface of the sorter. The circular frame design used for the sorter in the present study differs from the design of the JTED sorters (rectangular or semicurved rigid sorting grid) used in Southeast Asian countries [19,20,23,25,26,27]. The guide board and passageway for the catch were installed at the lower part of the circular frame. This design is similar to that used for JTEDs in Southeast Asian countries, which have an empty region below the JTED to guide large catches to the end of the codend. However, JTEDs in Southeast Asia do not include a frame to stabilize the passageway for catch, and factors such as the speed and direction of trawling and changes in submarine topography could narrow the passageway and cause catches to clog at the front of the JTED. By contrast, the passageway for catch in the present study’s JTED never deforms.

4.2. Exclusion Performance of JTED Grid Intervals

The three JTEDs had high exclusion performance and caused little loss in commercial revenue. Specifically, the JTEDs excluded 66.7–69.0% of E. rivulatus by weights 38.8–64.9% of Synagrops philippinensis by weight, 36.8–63.0% of Metapenaeopsis provocatoria longiroitris by weight, 75.9–98.9% of B. pterotum by weight, and 43.8–74.3% of Champsodon snyderi by weights (Table 2). These results are similar to those of experiments conducted in Southeast Asia [28].

In the present study, more catches were excluded (by weight) when the grid interval was higher. This finding is similar to those from experiments in Southeast Asian countries, where rectangular or semicurved rigid sorting grids are installed on trawls (

Table 4. Total catch excluded by JTEDs of this study and previous studies by weight.

Country	Literature	Type of JTED	Grid Interval (mm)	Weight Ratio of Excluded (%)
Malaysia	Chokesanguan et al. [19]	A	12	35
		A	20	73
Indonesia	Chokesanguan et al. [20]	B	40	79
Philippines	Chokesanguan et al. [23]	B	10	33
		B	20	77
		B	30	79
Brunei	Chokesanguan et al. [25]	A	10	30
		A	20	78
		A	30	86
Myanmar	Chokesanguan et al. [26]	B	10	30
		B	20	31
		B	30	93
Cambodia	Chokesanguan et al. [27]	B	10	22
		B	20	56
		B	30	76
Taiwan	This study	C	10	8
		C	15	29
		C	20	36

A: rectangular rigid sorting grid. B: semicurved rigid sorting grid. C: grid as designed in present study.

). For example, trawls used in the Philippines have semicurved rigid sorting grids with grid intervals of 10, 20, and 30 mm; in experiments, these grid intervals resulted in the exclusion of 33%, 77%, and 79% of the bycatch by weight, respectively.

At grid intervals of 10 and 20 mm, the JTEDs used in Southeast Asia outperformed those in the present study in terms of the amount of bycatch excluded. This might be attributable to differences in catch composition and the area of the JTED sorting grid. For example, more crustaceans were caught by the trawlers in the present study than by trawlers in studies from Southeast Asia (13% vs. <3%, respectively); the crustaceans caught in this study tended to be heavy crabs that could not easily escape the JTEDs [19,20,26,27].

JTEDs used in Southeast Asia also have a more porous sorting grid. This is primarily because the target catch of Southeast Asian trawl fisheries is shrimp whereas those of Taiwanese trawl fisheries is various types of fish. First, the present study and Southeast Asian studies have both defined the grid interval of JTED sorting grids as the horizontal distance between two vertical bars at the sorting area. However, the sorting area of the JTEDs in the present study contained numerous vertical bars and three horizontal bars spaced 15–18 cm from each other (Figure 2, Figure 3 and Figure 4). By contrast, the sorting area of JTEDs used in Southeast Asia contains no horizontal bars and only vertical bars spaced 50–65 cm apart [17,19]. For JTEDs with a grid interval of 10 mm, the areas of the grid of the JTEDs in this study and in Southeast Asian studies are approximately 15–18 and 50–60 cm², respectively. Consequently, Southeast Asian JTEDs exclude 70–90% of bycatch by weight.

In general, Southeast Asian JTEDs are not straightforwardly superior to the present study’s JTEDs just because they have excluded more bycatch in previous experiments. First, local conditions (such as target catch) differ between Taiwan and Southeast Asia. Second, the commercial loss from the use of Southeast Asia JTEDs has not been tested.

4.3. Discrimination in Excluding Bycatch and Including Target Catch for Each Grid Interval

Compared with JTED_(15mm), JTED_(20mm) was discovered to exclude less bycatch, such as E. rivulatus and P. aureus (

Table 5. Number of catches excluded by proposed JTEDs.

No.	Species	Number of Catch (n)				Ratio of Excluded Catch (%)				Chi-Squared Test
		10 mm	15 mm	20 mm		10 mm	15 mm	20 mm		χ2	p
Marketable catch
1	Priacanthus macracanthus	15	105	37		-	-	-		N/A	N/A
2	Evynnis cardinalis	141	75	40		-	-	-		N/A	N/A
3	Upeneus japonicus	796	187	80		6.8	75.3	17.8		208.1	0.00
4	Saurida elongata	23	48	13		-	2.1	7.7		2.2	0.34
5	Trichiurus japonicus	323	126	188		2.8	53.2	84.0		355.9	0.00
6	Amphioctopus aegina	38	44	16		-	6.8	-		3.8	0.15
7	Sepia madokai	195	421	110		20.5	50.6	47.3		51.0	0.00
8	Loliginidae	146	4882	680		12.3	89.9	19.0		2254.3	-
Catch with little or no economic value
1	Equulites rivulatus	5041	29,394	49,515		49.9	86.7	56.9		8055.7	-
2	Acropoma japonicum	804	7738	124		8.0	46.9	53.2		454.6	-
3	Synagrops philippinensis	5371	4352	3714		42.9	46.3	66.3		526.0	-
4	Photopectoralis aureus	392	1114	6364		37.5	48.8	17.6		570.1	-
5	Champsodon snyderi	1493	4180	1666		50.0	42.8	70.0		353.4	-
6	Metapenaeopsis provocatoria longiroitris	1651	12,419	6459		37.5	43.0	44.7		27.6	-
7	Calappa lophos	63	64	33		-	-	-		N/A	N/A

). However, this may have been due to the experimental design rather than to inferior exclusion performance. Specifically, because a cover net was stitched at the escape opening behind the sorters in the experiment, the excluded bycatch did not actually escape. The underwater footage collected in this study indicated that some juvenile and trash fish rested on the surface of the interception net after passing through the sorting area (Figure 8e) and remained there until the trawler slowed to change direction or until the net was hauled in; they then passed back through the sorting area and into the codend (Figure 8f). Therefore, small catches that remained on the cover net were more likely to return to the codend for JTEDs with larger grid intervals. Thus, JTEDs with larger grid intervals had much lower exclusion performance. If the cover net was not installed, the performance of all the JTEDs in excluding juvenile and trash fish might have been enhanced.

Loliginidae is a marketable catch, and the Loliginidae species commonly caught near Taiwan include Uroteuthis chinensis, Ur. edulis, Ur. duvaucelii, and Loliolus beka [32].

The three JTEDs in this study differed considerably in their ability to exclude Loliginidae with discrimination (i.e., excluding immature catch while retaining mature catch; Table 3). In general, JTED_(10mm) could exclude Loliginidae with discrimination but JTED_(15mm) was overly permissive and allowed a large proportion of Loliginidae to escape. JTED_(20mm) also excluded a fair number of mature Loliginidae.

Specifically, for JTED_(10mm), the mantle length of 50% of the catch (L₅₀) was 4.2 cm, indicating that most excluded Loliginidae were immature. By contrast, JTED_(15mm) had an exclusion probability of almost 90% and L₅₀ of 12.7 cm. Overall, 76.8% of Loliginidae captured in the present study (by number) had smaller than 6 cm, and JTED_(10mm), JTED_(15mm), and JTED_(20mm) excluded 40.7%, 91.6%, and 10.2% of these small Loliginidae (by number), respectively. JTED_(15mm) and JTED_(20mm) excluded 85.4% and 20.9% of mature Loliginidae (L₅₀ = 6–10 cm), respectively, which is unacceptably high from a commercial perspective. Nonetheless, some of these Loliginidae can grow to be greater than 10 cm, which makes them more valuable.

All Se. madokai caught in this study were small (L₅₀ < 4.5 cm), and 42% were excluded (305 of all 726 Se. madokai). Thus, the excluded Se. madokai were not commercially valuable. The JTEDs differed in their ability to exclude Loliginidae versus Se. madokai probably because Se. madokai has a hard, boat-shaped calcareous shell within its body but Loliginidae has a transparent, flexible shell covering its body.

4.4. Future Improvements to Proposed JTEDs

The following should be considered in future JTED designs.

The tilt angle of the sorter can be adjusted in future designs to ensure that fewer mature Loliginidae are excluded. Specifically, this study’s underwater footage revealed that Loliginidae rarely passed through the sorter when they came into contact with it. Instead, they used their tentacles to cling to the surface of the sorter and remained there; they then squeezed themselves through the sorter to escape. Some Loliginidae could not support themselves and fell off the surface of the sorter (Figure 9a–c).

To improve the ability of JTED_(15mm) and JTED_(20mm) to exclude Loliginidae, future studies should reference the double and triple BRDs developed in countries such as the United States and Norway [12,15,16,30,31,33].

Future JTED designs can incorporate more grid intervals to better fit a given catch composition. Specifically, the composition of catches of Taiwanese trawl fisheries varies by season, sea area, and operational depth, but the present study’s JTEDs can only accommodate one grid interval at a time.

Future JTEDs should be designed with ease of installation and removal in mind because JTEDs are difficult to install: two or three people are required to do so, and the initial installation takes an entire day. Thus, the replacement of a JTED entails a nonnegligible loss in revenue, time, and labor. Ease of installation also increases the willingness of trawl operators to use a JTED.

Future designs should address the problem of driftwood and weeds becoming caught in the trawl after typhoons. Although only a small amount of driftwood and weeds were stuck at the front of the sorters in this study, large waste (e.g., drift wood, abandoned nets, or marine debris) or large aquatic animals may become stuck at the front of the sorter and thus impede hauling.

5. Conclusions

The JTEDs developed in this study performed well in all tests. The section connecting the trawl to the JTED did not rupture, and the JTEDs were robust against underwater pressure, rupture, impact, and corrosion.

In general, larger grid intervals were associated with better exclusion performance at the cost of greater revenue lost. In particular, JTED_(15mm) rarely excluded common marketable fish and crustaceans. Nonetheless, JTED_(15mm) and, to some extent, JTED_(20mm) excluded a large proportion of mature and commercially valuable Loliginidae. Overall, this study recommends the use of JTED_(15mm) in Taiwanese trawlers because it strikes the best balance between sustainability and profitability.