The distribution and direct impacts of marine debris on the commercial shrimping industry

Commercial shrimpers frequently encounter marine debris in their nets, resulting in the loss of time and catch, and added repair costs. Prior to this study, no information existed on the spatial and temporal distribution of marine debris that shrimpers encounter and the subsequent economic impact on commercial shrimping. To characterize the quantity and impacts of marine debris, twenty commercial shrimpers participated in a comprehensive data collection program within the north central Gulf of Mexico, USA. Results showed that derelict crab traps were an overwhelming issue for shrimpers, and the type of shing gear used (skimmer vs. otter trawls) inuenced both the type of marine debris encountered and the subsequent economic impacts. Surveyed shrimpers encountered marine debris on 19% of tows and lost an average of 18.21 minutes, 7.88 kg of catch, and $6.37 (USD) in gear damage per tow with encounters, resulting in losses of $7,683 (USD) per year, per shrimper. This study is the rst to our knowledge to quantify the distribution and types of MD encountered by commercial shrimpers and its economic impact. Posadas et al. (2021) analyzed the results from the preliminary survey used to select shermen for this study. Many of the shermen who participated in the 2018 survey participated in the data collection program for this study. This survey assessed the perceived frequency and impacts caused by MD in 2018. Ninety-eight (98%) of shrimpers reported that they encountered MD during their shing trips (i.e., round trip of departure and return to harbor) with 85% encountering it frequently, and most shrimpers indicated reduced catch (80%), lost shing time (82%), and/or vessel repairs (75%) due to MD 12 . Similarly, the shrimpers who participated in this study reported making an average of 7 tows per day, and accurately estimated that 19% of all tows encountered MD. Shrimpers who participated in the qualitative survey, reported the crab traps and other abandoned shing gear were the most common and most destructive types of MD encountered 12 , which agreed with the results of this study. Logbook results showed that derelict crab traps accounted for 79% of MD encounters followed by other types of derelict shing gear (DFG) in the Mississippi Sound and north-central Gulf of Mexico.


Introduction
Commercial seafood industries have shaped cities and economies along coastlines worldwide. Along the US Gulf Coast, commercial seafood industries have not only established economies, but also have immense cultural signi cance 1,2 . The shrimping industry is the most economically valuable of shing industries along the US Gulf Coast and South Atlantic 3 . In Mississippi alone, the entire seafood industry contributed a total of $465.4 (USD) million to the state's economy in 2015 with the shrimping industry accounting for over 46% of that total 4 . Mississippi's shrimping industry contributed a total of $215.4 million (USD), and the total personal income for commercial shrimpers was $88.5 million in 2015 4 . However, this industry is fragile and is exposed to a variety of natural and anthropogenic stressors including lack of stewardship practices, climate change, severe weather, varying and evolving regulations, and ocean pollution 5,6,7 , which has led to steady declines in the number of Mississippi shrimpers and associated landings over the last 16 years 8 .
One stressor that shrimpers must adapt to is marine debris (MD). MD is de ned as any persistent solid material that is manufactured or processed and directly or indirectly, intentionally, or unintentionally, disposed of or abandoned into the marine environment or the Great Lakes (33 U.S.C. 1951-1958 (2006)).
MD and its impacts can be found on virtually any ocean, sea oor, and beach worldwide. Estimates show that about 6.4 anthropogenic tons of waste is produced each year, most of which primarily plastic and other hard, durable materials 9,10 . With the continued increasing production of plastics, the generation of MD will likely increase as well 11 .
Derelict shing gear (DFG) is one common type of MD shrimpers encounter 12,13 ; DFG is any recreational or commercial shing equipment that has been lost, abandoned, or otherwise discarded 3 . Derelict crab traps are a common type of DFG which are found globally and responsible for signi cant ecological and economic impacts 14 . These crab traps are often lost when the lines attaching the traps to a oat are broken by wave action or being run over by propellers, which makes the now unmarked traps di cult to recover. Arthur et al. (2020) 15 estimate that the Mississippi Sound has an estimated 22,000 actively shed crab traps in the state's shery, and nearly 5,500 derelict crab traps while Louisiana has an estimated 188,000 derelict crab traps and Alabama has over 8,000 derelict crab traps.
Qualitative studies of Mississippi shrimpers have shown they frequently encounter MD, mostly DFG, and it has signi cant impacts on their operations, indicating that MD has a large impact on the industry 12 .
However, no paired quantitative estimate of the distribution and direct economic impact of marine debris on the commercial shrimping industry has occurred. As the amount of MD increases around the world, its impacts are becoming increasingly apparent. The economic state of the Gulf Coast is heavily dependent on the health of sheries in the Gulf of Mexico. This fragile ecosystem and economy have suffered tremendously in recent decades from natural and anthropogenic disasters. Studying the distribution and effects of marine debris on the commercial shrimping industry is important to understand and potentially manage yet another stressor facing this industry.

Results a) Distribution of marine debris
The participating shermen submitted a total of 1,067 tow records. However, 897 tow records were used for data analysis; 170 were excluded because they were either submitted for dates outside of the range of the study, missing information, or not representative of a single tow (> 4 hours). Out of the 897 tows, 218 (24%) reported encountering MD; however, 50 (5.8%) of those encounters were attributed to organic materials, such as vegetation, which was not a focus of this study, and were excluded from the rest of the analyses.
MD encounters varied by location (Kruskal-Wallis Rank Sum Test -p = 0.010) with the highest chance of MD encounters occurring in Area 8 ( Figure 1). Dunn's post hoc tests indicated that the probability of encountering MD in Area 2 was very low; with Areas 2 and 6 (p = 0.040) and Areas 2 and 8 (p = 0.030) being signi cantly different from each other. Nearest neighbor tests showed that the observed clustered pattern in marine debris was not random (nearest neighbor ratio = 0.317; p < 0.001), which could indicate overarching anthropogenic and environmental drivers that in uence these patterns. MD encounters also varied by month (Kruskal-Wallis Rank Sum Test -p < 0.001) with a higher percentage of encounters occurring in November (37%) and lowest occurring in September (12%) than other months ( Figure 2).
Dunn's post hoc comparison showed that 3 comparisons were signi cant (e.g., July and September p = 0.004, August and November p = 0.023, and September and November p < 0.001; Figure 2).
Overall, shrimpers reported encountering MD on 19% of tows. The dominant type of debris encountered by shrimpers was reported to be derelict crab traps (79% of the tows with MD encounters) followed by other types of shing gear (5%), single use plastics (5%), and unknown trash items (4%; Figure 2). Generally, the assemblage of MD encountered was similar in each area (Kruskal-Wallis Rank Sum Testp = 0.430) but varied by month (Kruskal-Wallis Rank Sum Test -p = 0.007). However, that monthly result was driven by the difference in MD assemblage between October and November (Dunn's post hoc test-p = 0.019) with all other monthly pairwise comparison showing no statistical difference (p > 0.050; Figure  2).
The pattern of MD encounters was heavily in uenced by type of shing gear (Kruskal-Wallis Rank Sum Test -p = 0.002). The two distinct gear types used were skimmer (29% of all reported tows) and otter trawls (52% of all reported tows). The remaining 19% of tows reported using either both types of gear or other, unidenti ed gear. The difference in likelihood of MD encounters between these two dominant gear types was signi cant with over 31% of Skimmer and only 13% of Otter tows reporting encountering MD (Kruskal-Wallis Rank Sum Test -p < 0.001). The probability of MD encounters between the two gear types also had a spatial variation (Kruskal Wallis Rank Sum Test-p-value < 0.001; Figures 3 and 4). The use of skimmers resulted in more MD encounters in Area 8 while the use of otter trawls resulted in more MD encounters inshore and mostly in Area 4. Additionally, the types of debris caught by these two gear types were different (Kruskal Wallis Rank Sum Test -p < 0.001; Table 1). Skimmers caught both more and a wider variety of MD than otter trawls. Table 1 Marine debris caught and the probability of occurrence. Amount of each type of marine debris documented by shrimpers and the probability of encountering each type based on the type of gear used.

b) Direct economic impacts
Of the 897 tows analyzed, shrimpers reported a direct economic impact of MD (i.e., lost shing time, lost catch, and/or gear damage) for 10% of them. Lost shing time and catch were more impactful than gear damage ( Figure 5; Table 2). Fishermen reported 56% and 54% of all MD encounters resulted in lost time and catch, respectively, whereas gear damage was only reported for about 7%. Overall, shrimpers reported losing between 0 and 240 min, 0 and 68 kg of shrimp catch, and $0 and $200 (USD) in gear damage per tow due to MD (Table 2). Table 2 Impacts marine debris had on shrimpers. The impact marine debris had on shrimpers per tow including kilograms lost, fishing time lost, total sales lost, and damage costs.
The length of tows for each gear type was not affected by MD encounter ( Figure 6; Kruskal-Wallis Rank Sum Test -p = 0.221). However, the impact of MD on lost shing time (Kruskal-Wallis Rank Sum Testp < 0.001) and lost catch (Kruskal-Wallis Rank Sum Test -p < 0.001) was heavily in uenced by gear type used ( Figure 7). Because gear damage costs did not occur frequently, it did not show an in uence of gear type (Kruskal-Wallis Rank Sum Test -p = 0.466).
Overall, there were 79 and 56 MD encountering tows reported for skimmer and otter trawls respectively (Table 1). While skimmer trawls were over 2 times more likely to encounter MD than otter trawls (31% vs. 13% encounter rate per tow), otter trawls accounted for the costliest MD encounters. Of those MD encounters with skimmers, only 19% reported lost catch and 24% reported lost shing time due to MD.
Conversely, over 88% and 86% of otter tows with MD encounters reported lost catch and shing time respectively. Mean catch lost when MD was encountered was nearly 15x greater for otter trawls with MD encounters (14.23 kg per tow) than skimmer trawls (1.61 kg per tow; Figure 7). Time lost showed a nearly identical pattern with reported means of nearly 28 min and 8 min per tow dealing with MD for otter and skimmer trawls respectively (Figure 7).
When analyzing all tows (i.e., with and without marine debris encounters), shrimpers lost an average of 4.61 (± 15.33) minutes removing and disposing of MD and lost an average of 2 (± 6.42) kg of shrimp per tow. These losses lead to an average of $16.67 (± $51.23; USD) lost in direct sales per tow. The cost of damage is considered a labor income loss, and while not as frequent, an average of $   15 . Therefore, the higher number of crabbers losing traps in this area could explain a higher probability for shrimpers to encounter MD in the waters offshore of Jackson County.
While July through September is generally the peak of the shrimping season 12 , the logbook submissions did not peak until October. The onset of "shelter in place" orders due to the COVID-19 pandemic stalled shing efforts and logbook participation at the beginning of the season which started in June 2020. A general increase in shing efforts and additional participating shermen led to a peak of logbook submissions in October. Following the "shelter in place" orders, shermen dealt with an incredibly busy hurricane season with 8 tropical cyclones making landfall along the Gulf Coast 3,16 . Hurricane Laura made landfall in Louisiana in late August, Hurricane Sally made landfall near Gulf Shores, Alabama in mid-September 2020, and both Hurricane Delta and Zeta made landfall during the month of October in Louisiana with Zeta passing along coastal Mississippi 3 . These hurricanes brought storm surges, winds, and rain that may have moved and/or created MD throughout the study areas. Additionally, the higher tides allowed for shrimpers to sh in areas that were previously too shallow to reach. Peaks in MD encounters during the month of October could be a result of areas shed were that were likely littered with MD that accumulated over long periods of time.
There was little diversity in the type of MD that was encountered throughout the entire study area and a clustered pattern was observed. Crabbers use metal traps that sit on the sea oor with a buoy attached to mark their location. To increase productivity, crabbers typically drop their traps near each other 13,15 ; resulting in the clustered pattern observed with the nearest neighbor analysis. These traps are not extremely mobile except for being dragged by a boat's propellors or strong storm surges affecting the currents and dragging them across the sea oor 13 . Once crab traps are abandoned or lost at sea, they are considered MD 15 . Other types of debris encountered included tires, housing or construction materials, single use plastics, shing gear, clothing, and rubber material.
Similar to the results of this study, Posadas et al. (2021) indicated that MD has higher impacts on reduced shing time and catch than gear damage. This study indicated that 56% and 54% of MD encounters reported lost time and catch whereas only 7% of MD encounters reported direct gear damage. However, the impact of MD on lost shing time and lost catch was heavily in uenced by gear type used. The two most common types of trawls used in Mississippi are otter and skimmer trawls. Otter trawls are towed directly behind the boat and must be taken completely out of the water when MD is caught 17 . Skimmers are mounted on a frame and pushed along either side of the boat, shing the entire water column. Individually, the skimmer nets are about half the width of the otter trawls; however, because two skimmer nets are used at the same time, the two types of gear cover about the same area while shing 17 . The difference in likelihood of MDE's between these two gear types was signi cant with over 31% of skimmer and only 13% of otter tows reporting encountering marine debris.
Due to gear types, shrimpers encounter benthic MD, which sits on the sea oor, more often than encountering oating MD 18 . On the Gulf Coast, benthic MD mostly consists of DFG, including crab traps, which our results show is the likely cause of the most impactful MDE's for shrimpers. There was an average loss of 2 (± 6.42) kg of catch, 4.61 (± 15.33) minutes, and $1.30 (± $14.35; USD) costs in gear damage per tow; these collectively corresponded to about $16.66 (± $51.21; USD) per tow in direct losses per tow. When only considering tows with MDE's, shrimpers lost an average of 18.21 (± 29.13) minutes, 7.88 (± 11.3) kg per tow, and $6.37 (± $31.76; USD) in gear damage per tow.
Although the most frequently shed area reported for both types of gear was near the mouth of Biloxi Bay, the probability of MDE's between the two gear types also had a spatial variation. The use of skimmers resulted in more MDE's offshore of the mouth of Biloxi Bay while the use of otter trawls resulted in more MDE's inshore and mostly just off the coast of Harrison County. A possible explanation for these differences is, skimmers sh the entire water column while otter trawls only sh along the sea oor. Coale's (1994) study comparing rates of bycatch between the two types of gear, showed that skimmers were unable to sh in waters with greater depths than 3.7 meters. The more frequent and wider variety of MD encounters with skimmers may be a result of these nets being susceptible to both benthic and oating debris.
Overall, shrimpers submitted an average of 7 tows per day, at the current rate of MDE's (19% of tows) this would equate to $116.69 lost per day. These shrimpers reported shing an average of 11 days per month 12 during shrimping season (June-December), so these MD impacts can be extrapolated to total $7,701.54 (2020 $USD) lost each season. There were 120 registered shrimpers in the State of Mississippi during 2020, so assuming each shrimper encounters the same level of marine debris as encountered in this study, the total annual negative direct economic impact to the Mississippi shrimping industry is approaching $1,000,000 per year. This impact is signi cant when considering the total dockside sales of Mississippi commercial shrimpers in 2019 was only $15,000,000 27 .
The MD recorded in the data collection is likely an underestimation of the MD in the Mississippi Sound. EPA's Gulf of Mexico Program funded a 2-year collaborative cleanup effort speci cally incentivizing commercial shrimpers to remove derelict crab traps from the sea oor. Nearly 2,300 traps were removed from the waters between 2019 and 2020 19 , so much of the MD that could have impacted shing efforts had already been removed. However, as MD inevitably increases in the future and these negative economic impacts could expand signi cantly.
The expenses caused by MD could become critical for an industry that is already subjected to a variety of stressors such as an aging work force and increasing frequencies of natural and anthropogenic disasters. As development and litter increase worldwide, the MD crisis is expected to escalate as well 10 . For shrimpers, damaging encounters with MD will likely escalate as well. Along with potentially increasing lost sales from lost shing time and catch and costs of damage caused by MD, shrimpers must also balance rising costs of marine diesel 20 and the falling prices of both dockside and wholesale prices for shrimp 21 . The Gulf States have provided over 86% of commercially caught wild shrimp for the nation 21 . Compared to the other Gulf States, Mississippi's shrimping industry is relatively small, and Mississippi has had the least amount of derelict crab traps. The impacts of MD paired with natural and anthropogenic disasters, rising fuel costs, and falling prices of wild shrimp could be crippling for the industry throughout the region.

Conclusion
This study is the rst to quantify the impacts of MD on the commercial shrimping industry. However, Arthur et al. (2020) analyzed the bene ts of a derelict crab trap cleanup along the Gulf Coast. This study concluded the removal of these traps would be bene cial for both the blue crab and n sh sheries with additional bene ts for the economy, marine mammals and sea turtles, and boating tra c 15 . Shrimpers and crabbers generally sh in the same areas, so when traps become unmarked or derelict, they are likely to sit on the sea oor until a shrimping trawl picks them up 13 .
The higher number of MD occurrences per tow documented in October and November suggest that distribution and patterns of encountering MD could have been in uenced by hurricanes in the Gulf of Mexico during the study period. Additionally, the type of gear used in uenced the type and magnitude of debris caught as well as the economic impact. To alleviate and prevent these MD encounters, shrimpers may want to consider the type of gear that they choose, and a benthic MD focused cleanup should be done before the start of the hurricane/shrimping seasons each year 15 .

Methods a) Shrimper recruitment
Forty-four (44) shrimpers were surveyed in late 2018 to gather location, vessel and gear characteristics, shing effort, and to gauge interest in participating in an incentivized data collection program. This survey was approved by the Institutional Review Board in 2019 (MSU IRB-18-533) 12 . From the participants that expressed interest in participating in an incentivized data collection study, we selected twenty (20) that represented a diverse and representative group of shrimpers. They were chosen based on the length of their boat, the type of shing gear used (e.g., skimmer, otter trawl, and others), which coastal county their boat resides in, and shing effort (i.e., the number of reported trips in 2018). Shrimpers that successfully completed the data collection procedures for this study were provided a stipend of $300 (USD) per month from July to September 2020 and $500 (USD) per month October to December 2020 (i.e., $2,400 (USD) per shrimper over the shrimping season).

b) Data collection
Shrimpers used logbooks to collect data. In each logbook, they reported data associated with every tow completed over the 2020 shrimping season (e.g., July through December). Speci c elds from the logbook are listed below: • Speci c times shing nets were placed in the water and removed • General location shed speci ed by gridded map (Figure 8) • The types of marine debris encountered (e.g., plastic, shing gear, metal, etc.) • The amount of catch lost due to encountering marine debris (pounds) • The amount of shing time lost due to encountering marine debris (minutes) • The damaged caused to shing gear/vessel by marine debris (e.g., torn net, tangled motor, etc.) • The estimated cost of the damage that occurred ($) c) Analyses

Spatial Distribution Analyses
While shrimpers reported location information within individual cells based on the grid map, these cells were grouped into 9 larger shing areas for spatial analysis (Figure 9). While it was important to group the grid cells as evenly as possible, the shing areas were grouped by distance from shore, and numbered northwest to southeast. An additional factor that was considered when creating the shing zones was where the shermen would logically sh during a single day of shing. While area of the shing zone was considered and maintained by grouping 8 grid cells into each zone, the shape of each zone differed from one another. The nearshore study areas (0 to 1.85 KM from shore) include areas 1, 2, 4, and 9 while the offshore (1.86 to 75 KM) study areas include 3, 5, 6, 7, and 8. Some surveys documented multiple areas from the grid map shed. For these, the most northwest quadrant was chosen to use for both spatial and economic impact analyses.
The Shapiro-Wilk normality test was used to assess the assumption of nonparametric data distribution 22 .
The Kruskal-Wallis rank sum test 23 was used to determine the probability of encountering marine debris and trends in abundance and diversity of marine debris for each of these 9 study areas. Using the Euclidean Distance, a nearest neighbor analysis 24 was used to evaluate the spatial distribution of MD reported by shrimpers. The default search area was used to encompass all MDE's. Analyses with signi cant results were then applied to ArcGIS to create a series of choropleth maps showing the probability of MD encounters using base maps derived from ESRI.

Economic Impacts Analyses
The logbooks kept by shrimpers during the 2020 season did not account for weight of shrimp caught per tow; because of this, data from 2019 was used in the formula for sales lost due to shing time lost due to MDE's. Sales lost due to shing time lost during MDE's was calculated by multiplying 2020 shing time lost per tow from the logbooks by the 2019 average catch per minute, 0.364 kg (± 0.395), and the 2019 average dockside price per kilogram ($4.53 (USD)) 25 .
Due to lack of normality, non-parametric tests were used (e.g., Kruskal-Wallis and Dunn's post hoc test) were used to assess the impacts of MD. To assess the effect of shing location (n=9) and month (n=6) on the response variables of pounds lost, time lost, damage costs, and cumulative direct economic impact, multiple Kruskal-Wallis Rank Sum tests were conducted following the procedures of Queen and Keough (2002) 26 . Each Kruskal-Wallis test assessed the effect of shing location and month on an individual response variable (i.e., pounds lost, time lost, and damage costs) across all the records (with and without MD occurrences) and separately for records with MD encounters (i.e., 2 separate ANOVAs for each response variable). If location was signi cant, but month and the interaction between location and month were not, dates were pooled for Dunn's post hoc comparisons among locations. If a signi cant interaction between location and month occurred, or location and month were both signi cant, but the interaction was not, post hoc comparisons were done on each date separately.

Declarations
Data Availability Statement: Figure 5 Comparison of the observed impacts when analyzing only marine debris encounters.

Figure 6
Comparison of the length of tows for each gear type with and without marine debris encounters.

Figure 7
Comparison of impacts caused by marine debris for both gear types (skimmer and otter trawls).