Use of light-emitting diode (LED) lamps in combination with metal halide(MH) lamps reduce fuel consumption in the Vietnamese purse seine fishery

Khnh Q. Nguyen, Phu D. Trn, Luong T. Nguyen, Phuong V. To, Corey J. Morris

aNha Trang University, 2 Nguyen Dinh Chieu, Nha Trang, Viet Nam

bFisheries and Oceans Canada, St. John’s, NL A1C 5X1, Canada

Keywords:

ABSTRACT The use of high-power lights during night-time purse seining is common in Vietnam. Typically, metal halide(MH) lamps are used in the commercial fishery to attract fish, however these lights require more energy, have a shorter lifespan, and lower chromatic performance than light emitting diode (LED) lamps. This study examined catch efficiency and fuel consumption when using LED lamps in combination with reduced numbers of MH lamps(10.24 kW), compared to conventional lighting (28.6 kW), used during purse seining off the coast of Ninh Thuan province, Vietnam. The economic performance associated with using LED lamps in this fishery was also assessed.We found no significant differences in catch rates between the different light treatments, however fuel consumption was significantly reduced. Fuel consumption per nightly trip using LED with MH lamps was 70.8 l(11.1 l/h) compared to114 l (17.45 l/hr) using MH lamps alone, an estimated 37.9% reduction in fuel consumption. An investment in LED lamps by a fishing enterprise will require additional initial costs, however our analysis revealed the financial break-event point can be reached after approximately 101 nightly trips when the fuel price is at the 2015 level of USD $0.74 per l. Fishing enterprises can increase their profitability, and reduce CO2 emissions, by using LED lamps in the Vietnamese purse seine fishery.

1.Introduction

For thousands of years, starting with bonfires on a beach, people have used artificial lights to attract fish in order to improve catch-rates(Ben-Yami, 1976, p. 121), and this technique continues today as an important part of purse seining, stick held lift nets, squid jigging, scoop netting, drop netting, and hook-and-line (Nguyen & Winger, 2019).Fishing with lights has been one of the most advanced and successful methods for catching squids, hairtail, anchovy, and other pelagic species for centuries (An, He, Arimoto, & Jang, 2017; Matsushita, Azuno, &Yamashita, 2012; Nguyen & Tran, 2015; Sofijanto, Arfiati, Lelono, &Muntaha, 2019; Suuronen et al., 2012; Yamashita, Matsushita, & Azuno,2012). Light emitting diode (LED) lamp technology provides lower energy consumption rates, longer lifespan, higher efficiency, better chromatic performance, and reduced environmental impacts compared to traditional lighting technology (Nguyen & Winger, 2019).

Purse seining is a commercially important fishing technique for pelagic species, often used during darkness and relying on artificial light to gather positive phototactic fish (Ben-Yami, 1976, p. 121). The typical operation of purse-seining with artificial lights involves a main vessel equipped with powerful lamps and a purse seine net, and a smaller boat equipped with only lamps to attract fish within the fishing area (Nguyen& Nguyen, 2011; Sofijanto et al., 2019). Several factors influence catching efficiency of purse seines, such as purse seine length, encircling speed, net sinking speed, and pursing speed (Ben-Yami, 1976, p. 121),however, characteristics of the light source is an important component of this fishery (Sofijanto et al., 2019).

Of the total number of 2853 fishing vessels in Ninh Thuan province,approximately 1304 are purse seine vessels (DCFRP, 2014). The vietnamease purse seine fishery operates year-round except during full moon periods when the conditions are less favorable for light fishing(Nguyen, 2006; Nguyen & Nguyen, 2011). Fishermen determine the type, number and power of fishing lights based on their personal experience (Nguyen, 2006). Four typical light types used in the purse seine fishery in Ninh Thuan province include compact, incandescent, fluorescent tubes, and metal-halide lamps, that typically have a short lifespan and high electric power requirements (Nguyen, 2006; Nguyen &Nguyen, 2011; Nguyen & Tran, 2015). Fishermen generally believe that brighter lamps can attract more fish and as a result the amount of lighting and energy requirements used during fishing has gradually increased from a few kW in earlly 2000s to 100 kW or more in recent years (Nguyen, 2006; Nguyen & Tran, 2015). Larger electric generators are needed to power these energetically demanding light systems, and their fuel cost can represent as much as 60% of the total cost per fishing trip (Cao, Eide, Armstrong, & Le, 2020; Pham, Flaaten, & Nguyen,2013). The purse seine fishery has an annual profit margin of 11% (Cao et al., 2020), the reported cost of fishing increased from USD $12,000 USD in 2005 to USD $70,000 in 2016 (excluding labour costs; USD $1 =21,458 VND in 2015) (Cao et al., 2020; Nguyen, 2006; Pham et al.,2013), with average annual revenue per vessel of $21,000 to $162,690 over the same period (Cao et al., 2020).

The performance of LED lamps among commercial fisheries is known to vary across gear types and species. For example, the fixed lift net using LED lamps caught anchovy (

Stolephorus

sp.) in comparable amounts to traditional lamps (i.e. pressured kerosene lamp; Susanto, Irnawati,Mustahal, & Syabana, 2017), and the catch performance of LED lamps in the hairtail (

Trichiurus lepturus

) angling fishery during the winter was better than the metal halide lamps (An et al., 2017). However, catch rates for squid (

Todarodes pacificus

) jigging and hairtail angling (summer season) using LED lamps alone were significantly less than conventional operations with metal halide lamps, in spite of reduced fuel consumption (An, Bae, Lee, Park, & Bae, 2012; Matsushita et al., 2012; Matsushita & Yamashita, 2012; Park et al., 2017; Yamashita et al., 2012).Catch variation could also exist due to differences in light characterises,such as wave length or intensity, or their application that could affect underwater irradiance or the spatial distribution, influencing fish behaviour (An et al., 2012; Matsushita et al., 2012; Matsushita &Yamashita, 2012; Nguyen & Winger, 2019; Park et al., 2017; Sofijanto et al., 2019; Susanto et al., 2017; Yamashita et al., 2012). The effect of fishing lamps on fish behaviour is not fully understood, however, studies have shown that a combination of LED lamps with a sufficient number of conventional MH lamps have maintained catch rates of squid, hairtail,and anchovy, with a significant reduction in fuel consumptions (An et al., 2012; Matsushita et al., 2012; Matsushita & Yamashita, 2012;Park et al., 2017; Yamashita et al., 2012).

Several studies have demonstrated the economic and environmental benefits of using artificial light in other fisheries (An et al., 2017;Nguyen & Tran, 2015; Park et al., 2017). These studies have shown that a key challenge in adopting LED lamp is ensuring a return on investment and increased profit thereafter. This study compares fishing performance and fuel consumption between the traditional MH lamps and LED lamps in combination with substantial reduced number of MH lamps for the purse sein fishery in Ninh Thuan province, Vietnam. The economic performance of a typical fishing enterprise using LED lamps is considered to demonstrate increased profitability.

2.Methods

2.1.Sea trials and data collection

This study was conducted onboard a commercial purse seiner(

NT90578TS)

, measuring 19.5 m LOA with a 370 HP engine. A 45 HP Yanmar-3SMGGE engine and STC-40KW generator was used to power the fishing lamps, which had a separate fuel tank and gauges that measured fuel consumption. The purse seine used in this experiment was 568.7 m long and 91.2 m deep, with stretched mesh sizes in the wing and bunt measuring 80 mm and 20 mm, respectively. Comparative fishing experiments were conducted during the annual stewardship fishery, in the nearshore waters of Ninh Thuan province, approximately 25 nautical miles directly east from the port of My Tan (Fig. 1). Fishing experiments were conducted from March 10 to May 27, 2015 with 10 days break each month (from 11th to 19th of lunar month) to avoid low catch rates during the full moon phases, typical for the purse seine fishery (Nguyen & Nguyen, 2011).

Fig. 1.The map of study area off the east coast of Ninh Thuan. The blue rectangle on the top right panel of the figure denotes location of Ninh Thuan. The red rectangle on the bottom panel of the figure indicates the sampling area. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Daylight white rectangular-shaped LED lamps (

LED

-

Toplight

Deluxe

) and LED tube lamps (

INABA LED KYOSERA

) as shown in Fig. 2 were used. The vessel was also equipped with MH lamps (manufactured by

Philips

) and fluorescent tube lamps (manufactured by

Rang Dong

). A total of 20 rectangular-shaped LED lamps (100 W each x 20 lamps =2000 W) and 24 LED tube lamps (10 W each x 24 lamps =240 W) were used, with equal numbers on the starboard and port sides of the vessel(each side consisted of 10 rectangular-shaped and 12 LED tube lamps)placed in parallel with the vessel’s existing light system, which consisted of 27 MH lamps (1000 W each x 27 lamp units =27,000 W) and 40 fluorescent tube lamps (40 W each x 40 lamp units =1600 W) (see Fig. 2). The LED lamps were installed at an angle of approximately 10toward the water’s surface. We measured the distribution of spectral wavelengths released from the LED lamps and MH lamps using a spectrofluorometer outlined by Nguyen, Winger, Morris, and Grant (2017).The steady-state intensity was acquired using a Photon Technologies International QuantaMaster 8000 spectrofluorometer. The emission of light (i.e. luminescence) was detected by a Hamamatsu R-928 five-stage photomultiplier tube in photon-counting mode contained within a PTI Model 814 PMT housing, which in turn was enclosed in a Products for Research S600 phosphorescence detection cooling device to minimize contributions from dark current spectral artifacts. The optical characteristics of white rectangular-shaped LED lamps and MH lamps are shown in Fig. 3.

During the sea trails, two experimental treatments were investigated:(1) all of 27 MH lamps and 40 fluorescent tube lamps with a total wattage of 28.6 kW were used as the control treatment (Ctr); (2) 20 daylight rectangular-shaped LED lamps and 24 fluorescent LED tube lamps, combined with 8 MH lamps (four for each side) with a total wattage of 10.24 kW were used as the experimental treatment (Exp).The number of MH lamps and LED lamps selected for the Exp were based on our pre-sampling of similar sea surface illuminance for both light treatments (Ctr vs. Exp) when using 8 MH lamps. Fishing trips for the Ctr was conducted on even days of the lunar month (i.e. March 20, 22, 24),while experimental trips were undertaken on odd days of the lunar month (i.e. March 21, 23, 25). We conducted sea trials using the purse seiner onboard one boat to minimize effects of different fishing equipment (i.e. purse winch), gear structure (i.e. mesh size and hanging ratio),fishing technology, and operational practice (i.e. skipper experience) on catch rate. Both experimental and control trips were conducted on the same fishing ground (Fig. 1). The only difference between Exp and Ctr fishing was the lighting methods using LED lamps for the Exp and MH lamps for Ctr. Over the course of the study period (March 10 to May 27,2015), we conduced 60 nights of fishing and successfully deployed a total of 120 sets (60 Exp sets, and 60 Ctr sets) at depths ranging from 50 to 85 m.

The fishing vessel typically left the My Tan fishing port in the afternoon and arrived at the fishing grounds 2–4 h later, and around sunset. The anchor was placed, and the fishing lamps were turned on for experimental fishing operations. Lighting duration was typically 2–4 h in duration, depending on the estimated abundance of fish detected by the fishing sonar (Furuno CH-250), which was used to determine fishing sets. When the density of fish was high, or after a maximum of 4 h lighting, one-by-one pairs of lamps (one lamp in each side) was turned off every minute to gradually reduce the illumination. Before the final pair of lamps was turned off, a light raft equipped with five fluorescent tube lamps powered by a battery, was quickly deployed to maintain attraction of the fish, while the purse seine was set to encircle fish at the maximum speed of the vessel. The pursing wire and net wall was hauled using the purse winch and the net hauler. The process from turning off the first lamp to fish retention occurred over the course of approximately 1 h. After finishing the first set, the vessel moved a short distance (i.e. 1 h, approximate 5–8 knot) to another location and continued this process throughout the night-time period, conducting an average of 2 sets per night, until daylight the following day.

Fig. 2.LED lamps installed parallelly with MH lamps.

Fig. 3.Normalized fluorescence of rectangular-shaped LED lamps and MH lamps obtained following the procedure outlined in Nguyen et al. (2017).

All experimental data was recorded onboard the vessel on survey sheets, including departure and arrival times, lighting start and end times, and fishing positions and depths. Upon the retrieval of purse seines, all catch was sorted to the species level, and the catch was recorded as the number of baskets, each weighing 20 kg. To re-evaluate the catch volume, we obtained data of consignment sales for the catch from both Exp trips and Ctr trips through the record of marine products sales. For each nightly trip, measurement of the luminous intensity was taken at the sea surface at the freeboards (0 m) and at 5 m distance intervals on both sides (port and starboard side). Luminous intensity was measured for both Ctr and Exp to the nearest lux, using the Testo-545 auto digital lux meter, Victor merk with a capacity up to 100,000 lux following the procedure outlined by Nguyen (2006); Nguyen and Tran(2015). We assumed that luminous intensity at the sea surface and underwater were proportional based on earlier work by Susanto et al.(2017). Luminous intensity measurements in air are more accurate and required less effort than underwater measurements.

2.2.Statistical analysis

We conducted all analysis, data preparation, and produced all figures using R Statistical Software (R Core Team, 2019). Generalized Linear Mixed Models (GLMMs) were used to compare the catch rates of Ctr and Exp, accounting for non-normality and random effects (Bolker et al.,2009). Analyses were conducted separately for each species via the

glmmadmb

function based on the

glmmADMB

package (Bolker, Skaug,Magnusson, & Nielsen, 2012).

We modelled the catch of each species per set as the response variable, with fixed covariates (explanatory variables) light treatments (Ctr vs. Exp) and lunar phases (crescent and new moon). Lunar phases were considered in our analysis because the catching performance is impacted by the lunar rhythm and natural light (Matsushita & Yamashita, 2012;Nguyen, 2006; Yamashita et al., 2012). New moon phases were determined from 27th of the month before to the 3rd of the month after, and remaining days determined as crescent periods. These data were obtained from the U.S. Naval Observatory website as used by Ortega--Garcia, Ponce-Diaz, O’Hara, and Meril¨a (2008), and Matsushita and Yamashita (2012). We deployed two sets per night and catch rate was not uniform across sets; therefore, we included set number as a random effect. This allowed us to incorporate the dependency structure among observations within the same study night, and to account for temporal variations in the environment (i.e. water temperature, weather). We had initially tested and found the data were overdispersed – noted by the dispersion parameter for Quasipoisson family greater than 1 for all test of species, thus a negative binomial distribution with a log link function was used. The log link function ensured positive fitted values, and the negative binomial distribution was appropriate for our count data. All assumptions of the GLMM were met with regard to homogeneity of variance, normal distribution of errors, and independence of errors. We fit the following model:

where

Catch

is the

jth

observation in day

i

.

Differences in steaming and lighting time and fuel consumption during those periods were compared using paired t-tests. Comparing the fuel consumption between the experimental trips and control trips was also performed using paired t-tests. A confidence level of p <0.05 was used for all analyses.

We estimated the economic efficiency of using LED lamps by simulating data from the

NT90578TS

in which we assumed two scenarios: (i)the vessel harvests pelagic fishes using LED lamps combining MH lamps(10.24 kW), and (ii) the purse seine fishing vessel equips traditional fishing lamps (MH and fluorescent lamps), 28.6 kW total. We assume that the catching efficiency is the same between Ctr and Exp, as demonstrated in this study. Therefore, total revenue (TR) or gross revenue was defined as the income per trip. Fishing costs of each trip (FC)include fixed cost (the sum of cost associated with maintenance and repair, insurance and fees, depreciation, interest payment on loans), and variable costs (VC) (the sum of cost of lubricant, ice, food, crew labour,and fuel) (Cao et al., 2020; Pham et al., 2013). All FCs and VCs are similar between the two scenarios except fuel cost, which is expected to be reduced by using the LED lamp fishing method as demonstrated in this study.

Table 1 shows the initial equipment costs for LED lamps and MH lamps include lamp costs, lamp frames, wire, and sockets for both scenarios. Increase in LED lamp equipment costs must be balanced by a decrease in variable costs in order for the fishing enterprise to reach a break-even point on the investment by reducing fuel consumption during each fishing trip. The time required to reach a Return on Investment(ROI) is determined by the break-even point, considering the number of fishing trips per year and fuel price, which is represented by the following equation outlined in An et al. (2017):

where Dc is the differences in equipment costs on annual basis (USD), Fs is the fuel-saving per trip when deploying Exp instead of Ctr (l trip), N is the number of trips per year for the purse seine fishery, and Pis price of fuel (USD l). The break-even point occurs when different costs (D)equal zero and a fishing enterprise has earned-back the investment made in purchasing the LED lamps, instead of MH lamps and can begin generating revenue on that investment. As each subsequent fishing trip passes, Drepresents the financial gain realized using the LED lamps,compared to the traditional lighting method.

Table 1 Description of light installation costs for two scenarios: (i) the vessel equips with LED lights combining MH lights (10.24 kW), and (ii) the vessel equips traditional fishing lights (MH and fluorescent lights), 28.6 kW total. Currency exchange rate in 2015: USD $1 =$21,458 VND.

3.Results

The experimental and control treatments had the same light intensity distribution. Measured maximum luminous intensity for Exp was 2545 lux in the port side and 2682.9 lux in the starboard side, compared to 2547.4 lux in the port side and 2549.4 lux in starboard side for the Ctr(Fig. 4). The horizontal lighted area, both experimental and control treatments included 60 m from each side of the vessel (Fig. 4), and the light intensity rapidly decreased over that distance for both light treatments. However, the Exp treatment had a faster decay than the Ctr,particularly at maximum distances.

The nightly catches (all species combine) for the Exp and Ctr ranged from 225 to 6537 kg (mean of 2032 ±SE 289 kg), and from 434 to 5462 kg (mean of 2099 ±SE 260 kg), respectively. Both treatments captured the same number of species including 10 fish species and one species of squid (Table 2). The dominant species were skipjack tuna (

Katsuwonus pelamis

), yellowtail scad (

Decapterus maruadsi

), hairtail (

Trichiurus lepturus

), Indian mackerel (

Rastrelliger kanagurta

), and squid (

Loligo edulis

).Together these five species comprised 98.4% and 98.5% of the total catch of all species captured by the Exp and Ctr, respectively (see Table 2), and only these five species were included in our analysis.During the study, we fished 42 and 78 sets in the new moon phases(21 Ctr sets vs. 21 Exp sets) and crescent period (39 Ctr sets vs. 39 Expsets), respectively. GLMM showed no interaction between light treatments and lunar phases (

p

>0.05 for all four fish species and squid(Table 3). However, new moon phases had significantly higher catch rates than crescent periods for each fish species, and for all species combined (Table 3). Modelled catch rates for skipjack tuna, yellowtail scad, hairtail, and Indian mackerel were 1160, 1346, 391, and 342 kg per single set for new moon phases, respectively, compared to 699, 497,175, and 152 kg per single set for those species caught during the crescent period, respectively. There was no significant difference in catch rates of squid between new moon phases and crescent period.Modelled catch rates of squid were 167 and 69 for new moon phases and crescent periods, respectively.

Table 2 Summary of all species captured during the fishing experiment for experimental treatment (Exp) and control treatment (Ctr).

For each species comparison, the GLMM failed to detect any significant difference in catch rates between Exp and Ctr (Table 3). Modelled catch rates of skipjack tuna, yellowtail scad, hairtail, and Indian mackerel were 459, 370, 157, and 173 for Exp, respectively (Table 3).Modelled catch rates of those species were 436, 363, 121, and 105 for Ctr, respectively (Table 3). Unexpectedly, the LED lamp fishing method captured 17.3% less squid than the Ctr, which was statistically significant. Modelled catch rates of squid were 40 and 49 kg per single set for Exp and Ctr, respectively (Table 2). For all top five species combined,our statistical model revealed that the Exp captured 1202 Kg per single set, slightly more than the Ctr which caught 1108 kg per single set,however the difference was not statistically significant (Fig. 5).

Fig. 6 illustrates catch rates of the top five species combined for each pair of sets (

n

=60 pair sets). Catch comparison for pair sets fluctuated around the 1:1 line, suggesting it was unlikely that catch rates differed between Exp and Ctr throughout the fishing experiments.

Fig. 4.Horizontal light intensity distribution of experimental treatment (Exp) and control treatment (Ctr) in port side (left panel) and starboard side (right panel).Vertical bars are standard errors.

Table 3 Parameter estimates, fit statistics, and variation from the random effect of a GLMM model for the dominant species using pair sets as a random factor (n =60). SE is standard error of the estimate. CI is confidence intervals.

Fig. 5.Mean catch for different experimental treatment (Exp) and control treatment (Ctr). Black points represent mean catch. Bars are standard errors.

Fig. 7 illustrates the steaming time (to, from, and between the fishing grounds) and lighting time (the actual process of fishing when lights were turned on) for different light treatments. The steaming-time ranging from 6.1 to 9.2 h per nightly trip (mean of 8 ±0.09 SE hours per nightly trip) and was significantly greater than the lighting-time that ranged from 4.5 to 7.5 h, with a mean of 6.5 ±0.09 SE hours per nightly trip (t =−13.48, df =59, p <0.001). On average, there were no significant differences in steaming for Exp trips and Ctr trips (t =−1.69, df=29, p =0.102), which were 7.9 and 8.1 h per nightly trip, respectively(Fig. 7). The lighting for Exp trips and Ctr trips also were not significantly different (t =−1.21, df =29, p =0.235), which were 6.4 and 6.5 h per daily trip, respectively (Fig. 7).

Fig. 8 illustrates the fuel consumption of the Exp and Ctr during fishing operations, which included the steaming and lighting. The total fuel consumption for steaming consisted of 55%, and 45% for lighting.Average fuel consumption for lighting, at 92.4 l per nightly trip, was less than that used during steaming, which was 114.2 l per daily trip. There was no difference in steaming fuel consumption between Exp and Ctr operations (t =−2.14, df =29, p =0.116), with average of 110.5 l and 117.8 l per daily trip, respectively. Exp trips consumed a mean of 37.9%less fuel for lighting than the Ctr trips, corresponding with an average of 70.8 l and 114 l per nightly trip, respectively, which was statistically significant (t =−33.83, df =29, p <0.001). On average, the fuel consumption per hour was 17.45 l for Ctr compared to 11.08 l per hour for Exp. Fishing with LED lamp captured 61.09 kg of fish per kW of light power per lighting hour. This is significantly higher than Ctr fishing trips, having a mean of 22.12 kg (p <0.001) of fish.

Fig. 6.Catch comparison of top five species between the Exp (experimental treatment) plotted on the y axis and Ctr (control treatment) plotted on the x axis. Each blank dot represents one pair of sets. The dashed line denotes the Exp and Ctr had the same catch rate (i.e., 1:1 line). Dots above the 1:1 line denote that the Exp captured more than Ctr in each pair set, and vice versa.

A total equipment costs of scenario 1 (LED lamp and MH lamp combination) was USD $5599.45 (USD $1 =21,458 VND in 2015), and scenario 2 (MH lamp only) was USD $2343.08, a difference of USD$3256.37 between the two scenarios (Table 1). Fuel savings for each nightly trip was 43.2 l corresponding to USD $32 (USD $0.74 per l in 2015), comparing scenario 1 vs scenario 2. The ROI relates fuel price and the number of daily trips as shown in Fig. 9. The location of the intersection at the given fuel price and break-even curve indicates that a fishing enterprise needs approximately 101 trips (about 5 months) to reach the break-even point (D=0) on an investment in LED lamps(scenario 1), and could begin to realize a savings in variable costs after five months of fishing. Because the break-even point is associated with fuel price (Fig. 9), higher fuel prices would require less time to reach the break-even point. The zone above and to the right of the breakeven line indicates that LED lamps would be profitable while the zone below and to the left of the break-even line indicates that the continued use of MH lamps is recommended (Fig. 9).

Fig. 7.Operation time for experimental treatment (Exp) and control treatment(Ctr). Red and blue diamonds represent mean time for lighting and steaming,respectively. (For interpretation of the references to colour in this figure legend,the reader is referred to the Web version of this article.)

Fig. 8.Fuel consumption for experimental treatment (Exp) and control treatment (Ctr). Red and blue diamonds represent mean time for lighting and steaming, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

4.Discussion

In this study we found that incorporating LED lamps with MH lamps in the commercial Vietnamese purse seining fishery reduced fuel consumed to produce light by 37.9%, which reduced overall fishery costs by as much as USD $32 per nightly trip, thereby significantly increasing fishing profitability and reducing greenhouse gas emissions from this fishery. An installation of LED lamps required higher initial costs compared to MH lights, and approximately five months (depending on fuel price) to reach the break-even point, after which fishing enterprises begin earning profit due to reduced fuel consumption.

Fig. 9.The intersection of the curves designates the break-even line (black solid curve) that corresponds with the time until Return on Investment (ROI) in relation to the number of fishing trip and fuel price for the vessels with the LED lights (scenario i), instead of using MH lights alone (scenario ii). The zone to the above of the break-even line indicates the financial gain realized when using LED lights, meanwhile the zone in the below of break-even line indicates the initial loss when investing LED lights. Red dotted line indicates current fuel price (0.74 USD per l) that the vessel harvests about 100 trips reaching breakevent point. (For interpretation of the references to colour in this figure legend,the reader is referred to the Web version of this article.)

Although the total electrical power requirements for the Exp was 2.79 times less than that of the Ctr (10.24 kW versus 28.6 kW), both treatments produced similar light intensities. However, the illumination pattern was different at the greatest distances away from the vessel (i.e.≥45 m). For example, light intensity of experimental treatment was slightly less than the intensity produced by control treatment throughout 45–60 m away from the vessel. This difference might be due to the lower beam angle for LED lamps and installation method. The light distribution of the MH lamp illuminated the area along the sides of the lamp, while the LED lamp illuminated the area directly under the bulb (An et al., 2017; Baskoro, Riyanto, & Mawardi, 2019; Sofijanto et al., 2019; Susanto et al., 2017). In our study, light systems were installed at an angle of approximately 10and 0toward the water’s surface for the LED and MH lamps, respectively, indicating more energy from the LED lamps was directed toward the water, with narrower beam angle and smaller area of illumination than the MH lamps. This agrees with Sofijanto et al. (2019) who reported that the LED lamp illumination was focused at smaller angle with potentially more light penetration into the water although we did not measure light intensity under water in our study. It is likely that the MH lamps had a larger beam angle, and spread more light further from the vessel and above the water (Nguyen & Tran,2015; Susanto et al., 2017).

Light is the prerequisite for vision, which is one of the most important means of orientation for most animals (Ben-Yami, 1976, p. 121).The colour (i.e. wavelength) and intensity (i.e. lux) produced by an artificial light strongly affects the response behaviour in marine fish(Nguyen & Winger, 2019). Each species has an optimal wavelength and illumination, usually with peak absorbances between 400 and 494 nm(Nguyen & Winger, 2019). LED lamps used in this study have a peak wavelength of 457 nm (Fig. 3). Each species has different behaviour in response to light intensity. For example, anchovy (

Engraulidae

) and mackerel (

Scomber scombrus

) prefer the underwater illuminance of 0.03–6.00 lux and 2.40–39.50 lux, respectively, while squid (

Todarodes pacificus

) and yellowtail scad (

Decapterus maruadsi

) have much lower light intensity threshold ranging from 0.0034 to 0.03 lux, and from 0.001 to 0.01 lux, respectively (Nguyen, 2006; Nguyen & Winger, 2019).Images obtained from sonar (Furuno CH-250) showed that fish schools stayed near sea surface, less than 20 m depth, when Ctr lighting was deployed. By contrast, fish were concentrated under the light sources between 25 and 30 m depth from sea surface when LED lamps were turned on. We speculated that light from LED lamps penetrated further into the water than the MH lamps. This is consistent with previous studies which have revealed LED lamps illuminated greater areas underwater comparted to fluorescent and MH lamps (Nguyen & Tran,2015; Susanto et al., 2017; Sofijanto, Arfiati, Lelono, & Muntaha, 2018).

A combination of LED lamps and MH lamps maintained similar catch rates to that of using only MH lamps during purse seine fishing, and less electrical output was required resulting in substantially lower fuel consumption. This is consistent with most other fisheries that have switched to LED light sources. For example, An et al. (2012) reported that daily fuel consumption when using LED lamps in combination with MH lamps for hairtail angling could be reduced up to 33% compared to MH lamps. Fuel consumption rate for the squid jigging fishery decreased on average 0.28 l/kW h corresponding with a savings of 24% using LED lamps with a reduced number of MH lamps (Matsushita et al., 2012;Yamashita et al., 2012). Fuel consumption for lighting during the purse seine fishery in Vietnam is a significant proportion of the total operational cost, and depends upon the number of lamps, generator capacity,and operational time. In this study the engine used to produce light required one-third less fuel compared to conventional operation. The reduced demand could also allow a reduced generator capacity to match the reduced power output when using LED lamps that could reduce equipment costs.

In this study, we fixed the number of LED lamps and MH lamps across experiments consistent with lighting levels used during commercial fishery, however it is unknown whether these lighting levels are optimum for this purse seine fishery, despite substantial increases in lighting levels over recent years (Nguyen & Nguyen, 2011; Nguyen & Tran,2015). The optimal combination of LED and MH lamps, as well as the relationship between catch and amount of electric output needed to power fishing lamps, are important factors affecting fishery profitability(Matsushita & Yamashita, 2012). The light saturation level beyond which there is no added benefit of light on the catch rate, is reportedly variable among fisheries, lamp types, and fishing season. Yamashita et al. (2012) reported that 50 blue LED lamps with 24–36 MH lamps(117 kW total) appeared to saturate the fishing effect for squid jigging.Similarly, Matsushita and Yamashita (2012) reported that the combination of a 9 kW LED panel lamp and 36 MH lamps (108 kW total)optimized the catch rate of Japanese common squid taking fuel consumption into consideration. For the hairtail angling fishery, 18 kW LED with 40 MH lamps (60 kW total) seemed an optimal number, however,the catch efficiency was dependent on the season (An et al., 2012).

Fuel consumption per fishing trip included steaming (traveling from the port to the fishing ground, between fishing grounds, and returning)and the fuel used for lighting which was used only during fishing operations. The fuel cost ratio between lighting: steaming is different among fisheries, for example it was 1:1.6 for squid jigging (Yamashita et al., 2012), and 1:2.9 for hairtail angling (An et al., 2017). These ratios are lower than that for our study, which was nearly 1:1. The difference in fuel consumption ratios among fisheries is likely related to different fishery specific strategies and travel distances. In our study, fuel consumption for lighting operations was a significant portion of overall fuel cost, and reductions could improve the economical profitability.

No difference was detected between the control treatment and experimental treatment in terms of total catch or the fish species caught,with the exception of squid catch rates that were substantial reduced during the experimental light treatment. We speculate that squid could respond differently to LED lamps, compared to MH lamp as suggested by Matsui, Takayama, and Sakurai (2016). In addition, squid are attracted towards the light at night, but avoid highly illuminated areas, often aggregating in the shadow zone below the vessel (An, 2013). Squid were observed moving away from the underwater lamp when it was used,leading to squid jigging catch rates reduced up to 25% (An, 2013).

Although the price of squid was highest (USD $2.6 per kg), compared to other species (USD $1.1, 1, 0.8, 0.6 for hairtail, skipjack tuna, Indian mackerel, and yellowtail scad, respectively), the catch of squid accounted for only 2.3% in total (Table 2). As a result, there were no differences in landed values between control and experimental treatment, which were USD $3832 and USD $3829 per nightly trip, respectively. The similar comparison was found in other fisheries (i.e. hairtail angling and fixed lift net). For example, there were no significant differences between the landed value for angling vessel using LED lamps and MH lamps. The mean landed values were USD $1319 for the angling vessel using LED lamps, compared to USD $1054 for the MH lamp vessel(An et al., 2017).

This study has shown that a key challenge in adopting LED lamps is the financial burden of the initial capital investment. The lessons learned suggest that close cooperation among fishermen, scientists, management, agencies, and other stakeholders is a critical component in successfully altering MH lamps using LED lamps purse seine fishery. For example, fishermen can gradually replace the broken MH lamps by LED lamps. We estimate that this evolution is proximately less than two years corresponding with the lifespan of MH lamps. Alternatively, fishing subsidy programs from the government and lamp producers can help fishermen to purchase LED lamps with cheaper prices, which are successfully deployed in south Korea (An et al., 2017; Park et al., 2017).

Assuming an average USD $32 decrease in variable costs of a nightly trip when using a combination of LED lamps and MH lamps, the use of LED lamps is predicted to substantially increase the profitability of the Ninh Thuan purse seine fishery. For example, we estimated gains using LED lamps per vessel, per year, was approximately USD $7000. If this result is representative of the entire purse seine fishery (1304 fishing vessel), the financial gain would be USD $7,238,000 annually. Moreover, the lifespan of MH lamps is about 3000–5000 h, while LED lamps typically have 30,000 h of steady state operating time (An, 2013). In fact, commercial harvesters typically replace the MH lamps every two years (pers. comm. Captain T. Nguyen,

NT90897TS

). Theoretically, LED lamps could last for 10+years, highlighting use of LED lamps to reduce maintenance and replacement costs over a longer-term. For example,the south Korea squid jigging fishery could save USD 54,999,302 annually if its entire fleet changed from MH lamps to LED lamps (Park et al., 2017). An et al. (2017) showed that replacing traditional metal halide lamps with LED lamps on vessels targeting hairtail around the Korean Peninsula, would achieve a break-even point quickly depending on the fuel price and number of fishing trips per year. Similar economic benefits have been documented for squid jigging fisheries (Matsushita et al., 2012), purse seine fisheries (Cao et al., 2020; Kehayias, Bouliopoulos, Chiotis, & Koutra, 2016; Mills, Gengnagel, & Wollburg, 2014;Sofijanto et al., 2019), and lift-net fisheries (Baskoro et al., 2019; Susanto et al., 2017).

Operating the additional generators onboard vessels to produce electricity for lighting contributes to greenhouse gas emissions (Nguyen& Winger, 2019; Park et al., 2017). Fuel consumption by the world’s capture fisheries in 2000 was approximately 50 billion litters, accounting for 1.2% of the global fuel consumption, and directly releasing more than 130 million tons of greenhouse gas into the atmosphere(Tyedmers, Watson, & Pauly, 2005). Incorporating lighting technology with lower energy requirements, such as LED lamps compared to MH lamps, helps to reduce greenhouse gas emissions. A case study of the squid jigging fishery in South Korea revealed that using LED lamps could reduce 172,486 tons of greenhouse gas emissions annually (Park et al.,2017). A total of 1295 l of fuel saved for three months conducting experiments, a conservative assumption of 3.5 tons of COemission was reduced in the environment compared to the conventional lighting method.

In conclusion, this study demonstrated that a combination of LED lamps and MH lamps in the purse seine fishery reduced fuel consumption by 37.9% while maintaining catch rates, compared to conventional lamps. We also showed that fishing enterprises can improve their nearterm financial profitability if they were to install LED lamps, instead of MH lamps. For the experimental vessel, we estimated the initial increase in variable costs would reach a break-even point within approximately five months due to a noticeable reduction in fuel consumption. Changes in the number of fishing trips and fuel prices have direct effects on the economic performance of this fishery.

CRediT authorship contribution statement

Khanh Q. Nguyen: Conceptualization, Funding acquisition, Methodology, Investigation, Formal analysis, Writing - review & editing. Phu D. Tran: Conceptualization, Funding acquisition, Methodology, Investigation, Formal analysis, Writing - original draft. Luong T. Nguyen:Investigation, Data curation, Formal analysis, Writing - original draft.Phuong V. To: Conceptualization, Funding acquisition, Resources.Corey J. Morris: Writing - review & editing.

Acknowledgments

Funding for this study was provided by the Ninh Thuan Department of Science and Technology. However, the funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscripts. We are also grateful to the Institute of Marine Science and Fishing Technology for providing the effort to deploy this study. We thank Nguyễn To`an for his in-kind contribution through the use of his vessel and crew of purse sein fishing vessel

NT90897TS

for their valuable assistance setting-up the experimental equipment and data collection. We would also like to thank the editor and anonymous reviewers who provided valuable critiques and suggestions on how to improve the manuscript.