Abandoned, Lost or Otherwise Discarded Fishing Gear (ALDFG) in Tuna Pole-and-Line Fisheries

The shing industry is recognized as one of the primary sources of at-sea marine litter, largely through its contributions via abandoned, lost or otherwise discarded shing gear (ALDFG). Individual eet’s contributions to ALDFG vary signicantly across this global industry. While much information is available for some sheries, the rate of ALDFG remains poorly known for many techniques. In this study, we used data collected by sheries observers onboard pole-and-line shing vessels in the Azores (Atlantic Ocean) and the Maldives (Indian Ocean) to provide an accurate and representative estimate of ALDFG for this gear. Our analysis of 993 shing events demonstrated ALDFG contributions much lower than have been recorded for any other commercial tuna shing gear. Overall, we found that an angler loses some monolament line in 1.4% (±0.2) of shing events. This informs that for every thousand tonnes of tuna harvested using this shing technique, 0.3 kg of nylon is entering the marine environment. Globally, we estimate that all pole-and-line sheries together contribute to 96 kg ± 42.6 kg of ALDFG per year. These results further evidence the low environmental impact of this traditional shing practice, as well as the need for other methods to convert to less damaging gears.


Introduction
Fisheries are considered to be a principal source of sea-based litter globally, with 640,000 mt of shing gear estimated to be lost or abandoned in the oceans annually 1 . Abandoned, lost or otherwise discarded shing gear (ALDFG 1 ) are therefore an important component of the debris oating at the ocean surface 2 , stranded on coastlines 3 and accumulating on the sea oor 4 . Over recent decades, shing effort intensi cation and ongoing increases in the use of synthetic materials to construct shing gears have been the main factors responsible for increasing amounts of ALDFG in our oceans worldwide 1 .
Some of the greatest ecosystem impacts caused by ALDFG result from their continuous capture and entanglement of target and non-target species, sometimes for several years after loss, through a phenomenon known as "ghost shing" 5,6 . This impact is unsurprising when passive shing gears (e.g., nets, traps, longlines) have been purposely designed to withstand long periods in harsh marine environments (minimizing repair or replacement costs) while continuing to capture sh without requiring real time sher oversight.
Charismatic megafauna such as marine mammals, seabirds, sharks and sea turtles, are regular victims of ghost shing, which is recognized as a major source of mortality in some populations 7,8,9 . These animals typically become entangled by ALDFG when pursuing prey species that associate with lost shing gear such as drifting sh aggregating devices (FADs) 7,9,10,11 , when interacting with ALDFG out of curiosity, when seeking shelter amongst ALDFG, or when using ALDFG items as nesting material 12 .
Beyond direct entanglements of megafauna, ALDFG can also cause signi cant direct damage to marine habitats, including sensitive habitats such coral reefs, sea-grass meadows and mangroves that serve as nurseries for many marine species. As an example from the western central Paci c, tuna purse seine eets deploy between 44,700 and 64,900 drifting FADs annually with about 86% of these being lost or abandoned. Most of those devices (92%) are found on reefs, impacting multiple sensitive habitats during stranding and causing long term or permanent damages 13 . With the majority of these abandoned devices eventually sinking to the sea oor, their damage to deeper benthic marine ecosystems remains yet to be assessed and effectively quanti ed. In addition, ALDFG can represent navigational hazards or other threats to safety at sea, with the capacity to cause signi cant socio-economic costs when they damage boats and/or other still in use shing equipment 14,1,15 . As a result, there have been growing calls for large and increasingly common ALDFG components, such as drifting FADs, to have radar re ectors and navigational lights attached 16 .
Fisheries are operating across different environmental settings, from shallow waters to the open and deep ocean, using a wide variety of techniques to catch their target species. Fishermen lose, abandon or discard their shing gears for various intentional and unintentional reasons 17 , and the rates of such losses are highly variable between gear types and regions 18 . Preventing the loss of shing gear has become a global concern over the past decades 14 .
In order to appropriately inform sheries management on the need to reduce ALDFG, and additional stock mortalities caused by ghost shing, data on the contribution of gear loss by different shing techniques is increasingly needed. A recent meta-analysis of ALDFG highlighted the signi cant knowledge gaps on the rate of gear loss for many shing gear types 18 . Also a recent legal analysis of the 'loss' and 'abandonment' of FADs shows how complex the issues are and suggests some instances where the use of these devices constitutes illegal, unreported and unregulated (IUU) shing among international legislations 19 . The use of drifting FADs is also coming under increasing scrutiny for potentially contravening international marine pollution legislations including MARPOL V and the London Convention 20 .
Tunas are among the most valuable shes on the planet, being exploited across the globe with a variety of shing techniques. Today, purse seine net sheries using drifting FADs are landing the majority of tuna harvests worldwide, followed by pelagic longlines, which together with gillnets represent important contributors to ALDFG 21 . In the 1950s, pole-and-line shing was responsible for the highest proportion of tuna landings globally, but nowadays it represents less than 10% of global tuna catches 22 . Pole-and-line is a shing technique used to harvest various species of tunas, which has recently been popularised as a one-by-one shing method together with handline, troll and rod-and-line shing gears. One-by-one shing methods are recognized as being more environmentally friendly than other shing gears used to commercially harvest tunas, predominantly because they are active shing techniques that catch one sh at a time, yielding low bycatch and discard rates 23 while also reducing the likelihood of over shing, habitat damage or gear loss. Especially when compared to "passive gears" that are deployed and left at sea for later retrieval. Pole-and-line is essentially an artisanal shing technique mainly performed in Maldives, Japan and Indonesia with landings from each of these nations varying between 76,000 and 100,000 tons per year 22  Although pole-and-line sheries elicit many positive social and environmental attributes when compared to purse seining, gillnetting and longlining, the proportion of the tuna catch taken by pole-and-line has been decreasing for many years 22,24 . As a result, there has been substantial effort to demonstrate the assets of pole-and-line tuna shing by different NGOs (e.g. International Pole & Line Foundation (IPNLF), WWF, Greenpeace) but also by government sheries agencies, and private companies. While there have been different quantitative assessments on the environmental ecological footprint of pole-and-line (including bycatch rates 23 , discards 23  Considering the absence of publicly available data and publications assessing gear loss among poleand-line sheries, the objective of this study was therefore to use observer data across 2019 to provide a detailed and robust assessment of the quantities of gear loss associated with this shing technique. This study utilises data collected by trained sheries observers in two case study pole-and-line eets, the Azores and Maldives, to quantify the frequency and amount of lost shing gear (ALDFG) resulting from both of these sheries, operating under different circumstances, and to ultimately provide an accurate estimate of gear loss by pole-and-line sheries at the global level.

Study areas and pole-and-line sheries
We focused our assessment on two distinct pole-and-line eets for which all operational aspects are well described 23,27 . Both eets have well-established observer programs 23,27 which are using comparable data collection methods and simultaneously offering two distinct ecological and social settings.

The Azores
The Azores is an oceanic archipelago located in the NE Atlantic ( Figure 1). These nine Portuguese islands are inhabited by a relatively small population of 242,846 people in 2019. The economy is mainly supported by agriculture (dairy farming), tourism and sheries. Tuna shing, using exclusively pole-and-line ( Figure 2) and hand-line techniques with livebait, is the second most important shery in the region.
In some years it represents more than half of the total seafood landed in the EEZ (~1 million km 2 ) 28 . The Azorean pole-and-line eet is classi ed as an artisanal shery using small to medium sized vessels (up to 30 m long) and catching mostly skipjack (Katsuwonus pelamis) and bigeye (Thunnus obesus) tunas; with albacore (Thunnus alalunga) and yellow n (Thunnus albacares) tunas being caught in much smaller quantities. The occurrence of tuna in the Azores is highly seasonal, with bigeye tuna being more abundant from May to July, and skipjack tuna from July onwards. As a result, tuna sheries operate mainly from May to November.

The Maldives
As an archipelagic nation located in the central Indian Ocean and an exclusive economic zone (EEZ) covering an area of over 900 000 km 2 (3 000 times its land mass), the Maldives has a long and ongoing history of being heavily dependent on its marine resources 29 . The pole-and-line tuna shery is both the oldest and still the largest and primary shery in the Maldives, and has been a mainstay in the country for centuries 30,31,32 . As a result, the tuna sector is one of the most important sectors of the national economy, and it provides about 85 percent of the total protein consumed by Maldivians 33  Observers at both locations followed a similar and comparable protocol to record the loss of shing gear components, and to record other relevant operational and catch data for both bait sh and tuna shing events. For each bait sh shing event, the observers recorded the number of net deployments for livebait and any losses of gear components, while also recording total weight of the bait sh catch and its species composition. Out of the 172 observed events targeting bait sh, only a single scoop net was lost.
Therefore, ALDFG production in bait sh catch was considered as unsigni cant and not included in the results.
For each pole-and-line tuna shing event (de ned as periods of active shing separated by at least 10 minutes), observers recorded the geographic position of the event, its duration, number of active shers, type of poles (carbon vs. bamboo), total tuna catch (number of individuals, species and average size/weight). School association was also recorded in the following six categories: anchored FAD (aFAD), drifting FAD (dFAD), oating debris, underneath the shing vessel, seamount, or free school. Schools were de ned as associated to a FAD when the start of the shing event was located within 1 nautical mile of the FAD (or other oating object); while for seamounts, the schools were de ned as being associated when shing initiated within 5 nautical miles from the seamount 23 .
At the end of each shing event, the observers recorded the number of all gear components that were lost during that particular shing event. The gear components included poles (bamboo or carbon), gaffs, lures and lines. We did not include the loss of hooks due to their small sizes and limited environmental implications. When mono lament nylon leaders were lost, the observers estimated the length (meters) of each fragment that was lost. Total length of lost leaders were converted into mass using nylon runnage metrics for the 1. To investigate whether the amount of gear loss was in uenced by the species composition of the catch, each pole-and-line event was classi ed depending upon its catch composition; being "single species" when the catch was composed by a unique species and "mixed species" when more than one species was caught during the event. The rationale behind such grouping is that it is expected that when shers are gearing up for a particular species, the probability of gear loss is higher if another larger species joins the feeding school during the same shing event.
We used general additive models (GAM) to determine the in uence of four different predictor variables on the probability of gear loss per event: (1) average length of the tuna caught; (2) the type of school association; (3) the species composition; (4) the eet (Azores vs. Maldives) and (5) the total tuna catch (kg). Both the length of tuna and total catch were included in the model as a smooth term. School association was included as a factor and grouped into; (1) free schools, (2) associated (i.e. anchored FAD (aFAD), seamounts; drifting FAD (dFAD), or when tuna have aggregated around the boat). The inclusion of association variables followed the assumption that schools that are associated with oating objects are more frequently composed of different species compared to free schools that are generally composed of a single species 23 . Similarly, the model also included a factor variable which distinguished the shing event from being composed by single species or mixed. Finally, a case study approach was included as a factor to evaluate potential differences between the assessed eets (Maldives and Azores). Finally, the number of anglers was included as an offset in the model in order to account for differences in shing effort. The model was tted using a negative binomial distribution due to the large amount of zeros. Model assumptions were evaluated through the inspection of diagnostic plots. All analyses were performed using R statistical software 36 and GAM models were developed using the mgcv library 37 .
Based on the observed average gear loss to tuna catch ratio, we estimated total line loss for both eets using total landed catch for

Results
Across both national eets, sheries observers monitored a total of 993 pole-and-line shing events, corresponding to a total catch of 1,021 tonnes of tunas, mostly skipjack tuna (60.2% of total), across both regions. In the Azores, skipjack tuna represented 56.6% of the total observed catch by weight, while the remaining catch was composed of albacore (24.3%) and bigeye tuna (18.9%). In the Maldives skipjack tuna composed 81.5% of the observed catch, while the rest was yellow n tuna (18.5%).
Across both eets, gear loss was registered in 83 pole-and-line shing events, indicating that gear loss occurred in 8.4% of the total number of pole-and-line events monitored. A total of 111 gear components were lost during those 83 events, the majority being leaders, representing 90% of the total number of items lost ( Table 1). Considering that an average of 8 anglers (± 4) was operating per shing event, the average rate of gear loss was estimated at 1.4% (± 0.2) per angler. The average loss of gear components was 0.1 items (± 0.02) per shing event. The average length of lost line per event was estimated to be 0.1 meters (± 0.02), with a corresponding mass of 0.13 g (± 0.02).
Results from the GAM suggested a signi cantly higher probability of gear loss for Maldives eets compared to the Azores ( Figure 3). Furthermore, the model revealed that the probability of gear loss was higher for shing events involving multiple species compared to shing events composed of single species (Figure 3). Increasing tuna size and total catch also resulted in an increase in gear loss ( Figure 3). The GAM also highlighted that the gear loss for shing events associated with drifting or xed features was signi cantly higher than that of free schools ( Figure 3).

Discussion
The results of this study suggest very low gear loss frequency and volumes resulting from pole-and-line shing for tunas. On average, 1.4% (± 0.2) of the lines used are lost during pole-and-lines shing events, relating to 0.1 meters (± 0.02) of mono lament line. The ecological and economic impacts of ALDFG stemming from this gear type are limited, especially when compared to other types of shing techniques (such as gillnets or pot gears) for which loss events generally represent losing the entire gear. Overall, we estimate that pole-and-line sheries across the globe are losing a total of 96 kg of mono lament lines per year, consisting of small lengths of lost line that pose a negligible risk of causing entanglement mortalities and other ghost shing. While the extrapolation of our data to the world's pole-and-line eet has quite an associated level of uncertainty, we acknowledge that regional variation exists, it nonetheless provides a research informed preliminary measure of the magnitude of gear loss resulting from this gear type globally, which can be re ned in the future. Regardless of uncertainty related to this metric, the potential ALDFG contribution of pole & line shing is clearly not comparable to losses by other tuna shing gears that contribute many tons of plastic ALDFG to the oceans each year. As such, it is fair to assume that the contribution of gear loss by pole-and-line shing is at the lowest end of the spectrum when compared to other sheries, including longlining which have a higher gear loss rate (19-22%) and are also known to lose longer fragments of mono lament when partial losses occur 18 . While lost mono lament lines can be responsible for entanglement of marine organisms 38 , the average length of lost gear in pole-and-line shing gear was 0.1 meters, which represent a very limited potential for causing entanglements and ghost shing.
Our results contrast signi cantly with the meta-analysis by Richardson et al. (2019), which predicted that proportions of gear losses for pole-and-line was between 62 and 69%, the highest among "hooks and lines" sheries. While their calculations were based on a general study from the early 1990s 26 , our assessment is centred on a dataset covering 993 pole-and-line events monitored by trained sheries observers across two pole-and-line eets within distinct environmental settings. Therefore, we are con dent that our results are trustworthy, reliable and representative of this shing technique. Detailed collection of ALDFG data by sheries observers is the most reliable method to quantify gear loss, a method that should be applied and more transparently reported across all future shing activities.
The importance of gear loss is challenging to compare between shing techniques. The most common metric used to report gear loss is expressed as a proportion of deployed gear 18 . While this can be a valuable metric, it often does not provide a useful comparison of the contributions of ALDFG between different gear types, because it fails to appropriately account for the actual size of the lost equipment, and therefore the volumetric contribution that each loss event actually represents. We suggest that reporting the ALDFG quantity (in weight and number) as totals, and as a proportion of achieved total sh catch. This approach will offer more suitable metrics of gear loss that are also more directly comparable between different shing gears and eets. We believe the ultimate aim of such comparisons should be to help inform the minimisation of total and relative (per tonne of catch) ALDFG contributions by sheries both between and within gear type categories.
Our data also provide relevant information on the causes of gear loss in pole-and-line sheries. The most important factors in uencing gear loss rate were the type of tuna school being shed (associated with oating objects or not), and its species composition. We found that free schools were typically monospeci c composed of single species, and resulted in a lower gear loss compared to mixed schools, where smaller species (e.g. skipjack tuna) are mixed with larger species (bigeye tunas in the Azores or yellow n tunas in the Maldives). Accordingly, the probability of gear loss also unsurprisingly increased with increasing sh size.
This study further demonstrates the sustainability bene ts of pole-and-line sheries. Along with being a selective shing technique with minimal bycatch and limited impacts to the broader marine ecosystem, the overall contribution of pole-and-line sheries towards ALDFG is also a minor fraction of that seen from its counterpart tuna eets using more damaging gears. Similar studies should be rigorously performed across all different shing techniques and regions, using trained observers and comparable techniques, to suitably quantify the environmental footprint of various sheries and their gear types. Figure 1 Location of both case study areas where observers onboard pole-and-line tuna vessels collected data on gear loss.

Declarations
Page 16/17  Generalized additive model (GAM) derived effects of the investigated parameters (average length of tunas, total catch, school type, catch composition and location) on the probability of gear loss during