Choosing best practices for managing impacts of trawl fishing on seabed habitats and biota

Dyfyniad o'r fersiwn a gyhoeddwyd / Citation for published version (APA): McConnaughey, R. A., Hiddink, J. G., Jennings, S., Pitcher, R., Kaiser, M., Suuronen, P., Sciberras, M., Rijnsdorp, A. D., Collie, J., Mazor, T., Amoroso, R. O., Parma, A., & Hilborn, R. (2020). Choosing best practices for managing impacts of trawl fishing on seabed habitats and biota. Fish and Fisheries, 21(2), 319-337. https://doi.org/10.1111/faf.12431


| INTRODUC TI ON
Fish and shellfish caught with bottom otter trawls, beam trawls and shellfish dredges (hereafter 'bottom trawls') account for around one quarter of the global capture-fisheries landings (Amoroso et al., 2018).
However, bottom trawling is often one of the most significant forms of physical disturbance on the seabed (Eastwood, Mills, Aldridge, Houghton, & Rogers, 2007;Foden, Rogers, & Jones, 2011). The extent of this disturbance is highly variable, and the proportion of seabed area exposed to bottom trawling ranges from <1% to >80% in different regions of the world (Amoroso et al., 2018). Trawling may modify sediment texture (grain size), the presence and nature of bedforms and chemical exchange processes (Oberle, Storlazzi, & Hanebuth, 2016;Simpson & Watling, 2006). Trawling can also have direct and indirect impacts on populations and communities of benthic invertebrates, with significant reductions in abundance, biomass, species diversity, body size and productivity reported in many studies (Collie, Hall, Kaiser, & Poiner, 2000;Hiddink et al., 2017;Kaiser et al., 2006;McConnaughey, Syrjala, & Dew, 2005;Sciberras et al., 2018). Exposure is widespread because trawls can be adapted for use in diverse habitats and are readily scaled to a wide range of vessels, target species, fishing conditions and geographical settings (Løkkeborg, 2005;Suuronen et al., 2012;Valdemarsen, Jørgensen, & Engås, 2007).
The impacts of bottom trawling at a particular location are determined by the design of the gear and its operation, the frequency and intensity of trawling, the susceptibilities of biota (which influence depletion) and the life histories of the biota (which influence recovery). Environments exposed to different physical regimes have different sensitivities to bottom trawling, reflecting characteristics of the benthic fauna (e.g., Snelgrove & Butman 1994;Hiddink et al., 2019;Kaiser, Hormbrey, Booth, Hinz, & Hiddink, 2018) and the background level of natural disturbance (e.g., McConnaughey & Syrjala, 2014). The footprint of trawling (the geographical area that is directly contacted by trawls at least once in a specified time period) is determined by multiple factors including the distribution and catchability of fish or shellfish, technical capacity of the fleet, production costs and market prices, ruggedness of the seabed, environmental conditions (e.g., prevailing weather patterns), the state of fishery development and changes in management measures. Each of these factors varies in space and time such that the footprint may move, contract or grow from year to year (e.g., Jennings, Lee, & Hiddink, 2012;Kaiser, 2005), although at broad scales, the distributions of bottom trawling tend to be consistent from year to year (Amoroso et al., 2018).
The impacts of trawling on the environment and biodiversity are the focus of societal debates about the benefits and costs of seafood production, and an increasing focus of fisheries and environmental management regulation and certification processes. This is especially the case when trawling occurs on or close to vulnerable marine ecosystems (VMEs) or ecologically and biologically significant areas (EBSAs), but also for other types of habitat (Garcia, Rice, & Charles, 2014). A range of management measures and voluntary industry actions have been adopted to reduce or prevent trawling impacts on seabed habitats. However, the knowledge base to evaluate the effectiveness of these measures or combinations of measures, and the extent to which they represent best practices, is fragmented.
Here, we review and evaluate the effectiveness of management measures and industry actions that are intended to minimize the im- 2018), we address multiple knowledge gaps identified in a prioritization exercise concerned with reducing the environmental impacts of trawling (Kaiser et al., 2016). The resulting guidance on best practices is intended to help managers and the industry minimize environmental impacts of trawling per unit weight or value of landed fish, while achieving a sustainable level of fish production.

| MANAG EMENT OBJEC TIVE S AND TR ADE-OFFS
Managers of trawl fisheries are frequently faced with the need to reconcile multiple and often conflicting societal, environmental and economic objectives. Foremost among the management objectives is usually the need for sustainable exploitation of the targeted stocks resulting in employment, income and food security. In most countries and regions, there are also stated objectives to accomplish this exploitation with minimal habitat impacts or losses of ecosystem services and to ensure the unintended bycatch is minimized. Habitat protection measures may therefore limit exploitation benefits because of trade-offs. Resolving the fundamental conundrum between biological and socio-economic objectives remains one of the major challenges of fishery management.

| MANAG EMENT ME A SURE S AND INDUS TRY AC TI ON S
In the following section, we present nine management measures and voluntary industry actions (hereafter 'measures') that can be used to reduce and manage trawling effects on seabed habitats and biota (Table 1). These measures can be grouped into four classes: (1) technical measures that refer to changes in gear design and operations, (2) spatial controls that include gear-specific prohibitions, freezing the trawling footprint, nearshore restrictions and coastal zoning, prohibitions by habitat type including real time (i.e. 'move-on rules') and multipurpose habitat management (e.g., marine-protected areas, MPAs), (3) impact quotas that are output controls that include invertebrate bycatch or habitat-impact quotas and (4) effort controls that affect the overall amount and distribution of trawling. Several of these measures can be used simultaneously, where their relative effects depend on characteristics of the fishery, environment and management system in which they are applied.
Measures can be evaluated using both qualitative and quantitative performance metrics, recognizing that the preferred metrics will depend on the local, national or regional context. Our proposed metrics for evaluation include the positive and negative effects of trawling on (a) benthic biota, (b) sustainable fish populations and food production, (c) ecosystems and ecosystem services and (d) economic performance of the fishery. In the following sections, we evaluate the efficacy of selected measures using one or a combination of these four performance metrics and predictions from a simple impact-yield model.

| Gear design and operations
The design and operation of trawls may be modified to reduce impacts on the benthos, while maintaining an acceptable level of performance ( Figure 1; Jennings & Revill, 2007;Valdemarsen et al., 2007;Valdemarsen & Suuronen, 2003 contact also reduce trawling impacts, as the mortality of benthic invertebrates caused by trawl gears is correlated with the penetration of the gear into the seabed . A number of gear modifications will reduce the direct impacts of bottom trawling on benthos by reducing physical contact and penetration depth of gear within the seabed. For example, large-diameter rubber bobbins separated by rows of small-diameter discs create openings under the footrope that reduce unobserved mortality of commercially valuable crab species (Hammond, Conquest, & Rose, 2013;Rose, Gauvin, & Hammond, 2010). This design requirement reduced habitat disturbance by 24% since it was implemented in the Bering Sea (2011) and central Gulf of Alaska (2014) flatfish fisheries (50 CFR § 679.24; Smeltz, Harris, Olson, & Sethi, 2019). Fly-wires attached to the warps (fork-rigged trawl), shortening of the warp-length-to-depth ratio and lighter/high-aspect-ratio/maneuverable semi-pelagic trawl doors also reduce the contact area of otter trawls (Brewer, Eayrs, Mounsey, & Wang, 1996;Broadhurst, Sterling, & Millar, 2015;Ramm, Mounsey, Xiao, & Poole, 1993;Valdemarsen et al., 2007). A wing that skims just above the bottom on a sumwing beam trawl reduces both penetration depth and fuel consumption by 10% (van Marlen et al., 2009).
Pulse trawls in the North Sea have been shown to increase target species (Solea solea, Soleidae) selectivity and to catch 40% less benthos and undersized fishes compared to conventional beam trawls rigged with multiple tickler chains ( van Marlen, Wiegerinck, Os-Koomen, & Barneveld, 2014). In addition, pulse trawls are towed at a lower speed, around five knots as compared to six to seven knots with beam trawls, and the electrodes penetrate less deeply into the seabed (Depestele et al., 2016(Depestele et al., , 2018. Concerns still remain about limited knowledge of the effects of electricity on marine organisms and the benthic ecosystem (Soetaert, Decostere, Polet, Verschueren, & Chiers, 2015).
Gear modifications that limit the weight and durability of gear may also influence the trawling footprint by discouraging use in rough areas of the seabed that commonly support sensitive benthic species and habitats. For example, it has been proposed that 'rockhopper' gear, which uses large tires on the footrope and a separate tension line to lift the net off the seabed and prevent gear damage after contact with a large boulder, should be banned (Norse, 2005). Pelagic trawls, on the other hand, are frequently fished in smooth-bottom areas where they make occasional contact with the seafloor, particularly when targeted species are in close proximity to the seabed. Although it is common practice, bottom contact of pelagic trawls is discouraged in the Gulf of Alaska by prohibiting devices that protect the footrope (50 CFR § 679.2). Industry-sponsored studies have shown that alternative designs and materials can reduce the penetration depth and overall weight of scallop dredges, thereby reducing gear wear, fuel consumption, bycatch and seabed impacts, while increasing catch efficiency (Abram, 2009;Hinz, Murray, Malcolm, & Kaiser, 2012;Humphrey, 2009).
Operational changes by fishers combined with innovative technology can further reduce the impacts of trawling, due to efficiency gains that reduce the level of effort required to catch the quota. For example, 'smart capture systems' that improve control of the gear can eliminate the need for excessive weight used to stabilize gear (CRISP, 2014). Technologies such as the use of acoustic and video imaging for pre-catch identification and catch monitoring could potentially increase catch rates of target species and correspondingly reduce the trawling footprint (e.g., Barents Sea cod). Regulation in the Gulf of Alaska limits the proportion of time pelagic trawls may be on bottom to 10% (50 CFR § 679.24), which is a 75% reduction on previous estimates of unregulated bottom-contact time (NMFS, 2005).
Gear modifications that reduce bottom contact will usually reduce impacts on benthic species and habitats per unit of effort, relative to more localized reductions achieved with spatial controls alone. However, there may be offsetting effects that are difficult to quantify. For example, elevated footropes that reduce the number of contact points on the seafloor may concentrate pressure forces over a smaller area of the seabed, which could potentially increase unobserved mortality and injury (Hammond et al., 2013;Mensink et al., 2000). The use of the more efficient electrical pulse trawl and expansion to new trawling grounds (ICES, 2018) may cause conflicts with other fisheries that experience reduced catch rates on shared fishing grounds (Sys, Poos, Van Meensel, Polet, & Buysse, 2016). Although potentially advantageous, modifying existing gears (or substituting other gears) is often problematic because their effectiveness often relies on close contact with the seabed due to the behaviour of many target species (Creutzberg, Duineveld, & Noort, 1987;. Reduced catch rates, however, may be acceptable when offset by lower operating costs and less wear of the gear, once the capital costs for new lower-impact and/or energy-efficient gear are recovered. To assess such costs, the UK Sea Fish Industry Authority has produced a tool to evaluate the economic performance of gear designs, and thus, their commercial viability, before fishers embark on costly investments in innovation (Witteveen, Curtis, Johnson, & Noble, 2017). Furthermore, the effects of changing gear design on benthic communities can be estimated through the strong relationship between penetration depths of fishing gears and depletion of benthic biota Szostek et al., 2017).

| Prohibitions by gear type
An absolute prohibition of bottom trawling in a given region provides the most comprehensive protection of seabed habitats from the effects of trawling and may improve harvests by competing gears. The primary objective of gear-specific prohibitions is to shift fishing to other gears that have lower benthic impacts, such as stationary nets, pots and longlines (Pham et al., 2014;Suuronen et al., 2012). For example, the prohibition of trawling in Venezuela inshore in 2001 and territorial waters in 2009 led to increased catches by small-scale fishers who supplied 70% of annual fisheries production (compared to only 6%, or 70,000 t, by trawlers in 2007). In Qatar, the number of artisanal fishers (+52%), the artisanal catch (+159%) and the size of artisanal-class vessels increased after closure of its bottom-trawl fishery in 1993 (Al-Abdulrazzak, 2013;El Sayed, 1996;FAO, 2003;Walton et al., 2018 Note: Evaluation of each measure/action is based on four evaluation metrics and predicted impacts from the yield-impact model (see text sections for references and details). Impact is expressed as effects on fractional depletion of benthic biomass per trawl pass (d) or catchability of target species (q), recovery rate of the benthos (r b ) and trawling intensity (F) on relative benthic status at regional scales (RBS; Equation 1) and on target-stock biomass (B f ; Equation 2).

TA B L E 1 (Continued)
in 2005 to protect coral reefs, where it has been estimated that one deep-sea bottom trawl will have an impact on cold-water corals similar to 296-1,719 longlines (Pham et al., 2014). implementation of the ban has been delayed in many coastal areas but a substantial part of the industrial trawling has been phased out (Endroyono, 2017).
What is clear from the Indonesian experience is that an immediate ban on trawling will cause considerable societal hardship in the short term with potential positive outcomes only realized over a longer time period (Chong, Dwiponggo, Ilyas, & Martosubroto, 1987;Endroyono, 2017). For example, close to 25,000 trawl fishers were immediately unemployed and shrimp exports dropped by 22% during the first year of the ban (1983), representing ~19% of the total value of fisheries exports. In the medium term (3-5 years), the ban eliminated the supply of trawl-caught so-called trash fish to 13 fishmeal factories, production dropped from 4,000 t in 1980 to 6 t in 1983, and Indonesia began to import fishmeal as a result. There is anecdotal and some fishery information that indicates improvements in catch of shrimp per unit of effort by small-scale fisheries closer to port by the few remaining trawl vessels prior to expiration of their licenses (Chong et al., 1987;Endroyono, 2017). The government of Indonesia is currently exploring alternative fish-capture technologies to exploit its shrimp resources (Endroyono, 2017).
Similarly, the ban in Venezuela directly affected 263 Venezuelan trawlers, as well as Italian and Spanish vessels operating in the area.
Approximately 6,500 workers in the industry were displaced, and as many as 26,000 jobs were affected indirectly; supplies of the cheapest fish in the domestic market were also reduced as a result of the ban (Marquez, 2009

| Freeze trawling footprint
The impacts of trawling can be limited by confining activities to previously trawled areas. The spatial pattern of trawling is related to the distribution of fish, as well as various constraints of the fleet such as distance to port and operating costs (e.g., Hutton, Mardle, Pascoe, & Clark, 2004). High-resolution studies of vessel- An advantage of freezing the footprint over other forms of spatial management is that it avoids the potentially large negative effects on seabed habitats and biota that are associated with displacement of fishing effort to previously untrawled areas (Abbott & Haynie, 2012;Dinmore, Duplisea, Rackham, Maxwell, & Jennings, 2003;Hiddink, Hutton, Jennings, & Kaiser, 2006). The resulting concentration of effort in the defined ground may be consistent with the choices of fishers who increasingly focus on core areas as competition for space and resources is otherwise reduced (Gillis, Peterman, & Tyler, 1993;Kaiser, 2005). This usually occurs without reductions in yield, at least in the short term, and the potential benefits are especially significant for biogenic and deeper-water habitats with long recovery times (e.g., Clark et al., 2016). However, it is important that this measure is coupled with limited or regulated fishing effort and/or quota controls so as  Coastal zoning is often intended to minimize gear conflicts between artisanal and industrial fishing fleets and to reduce the incidence of gear-related impacts on sensitive nearshore (nursery) habitats such as eelgrass beds that support biodiversity and functional processes.

| Nearshore restrictions and zoning
With the imposition of a nearshore trawling restriction, fishery production from the nearshore is likely to decline until a possible compensatory increase in catches by substitute (artisanal) fishing methods and recovery of habitats and associated fish populations occurs. A regional or national system of spatial zoning by vessel class would be a more formalized approach; it could be used to preferentially benefit local economies dependent on nearshore and recreational activities including small-scale and recreational fishing and eco-tourism. While the unsuitability of many inshore vessels for offshore fishing precludes nearshore trawlers from fishing in offshore zones, offshore effort (and impacts) could increase over time if capital investments are made to upgrade or replace vessels for trawling on deeper, more distant and potentially sensitive fishing grounds. Overall, nearshore restrictions to limit trawling impacts could be particularly effective when technology or resources (e.g., VMS or onboard observers) are not available to monitor and enforce the fine-scale distribution of trawling activity.
In such cases, distinct wheelhouse colours assigned to different harbours combined with self-enforcement could be used as a simple control mechanism among the fishers, as practiced in SE Asia (e.g., 'sasi', Endroyono, 2017).

| Prohibitions by habitat type
Bottom trawling is commonly prohibited over habitat types that are both easily disturbed and slow to recover, such as seagrasses, sponges, corals and other endemic or rare types of seabed communities (Clark et al., 2016;Freese, 2001;Kaiser et al., 2018;Neckles, Short, Barker, & Kopp, 2005 Permanent prohibitions by habitat type provide effective protection when locations of sensitive habitats can be identified and prohibitions can be introduced prior to significant physical disturbance (Howell, Davies, & Narayanaswamy, 2010). The designated areas are usually small, so the benefits to overall ecosystem function and food production are limited, but they may confer economic benefits to local economies that rely on artisanal fisheries or eco-tourism (Gell & Roberts, 2002). Fleets targeting species that are strongly associated with sensitive habitats may suffer reduced yields or increased competition as effort concentrates in the remaining areas (Poos & Rijnsdorp, 2007). Overall, rare and sensitive habitats that are vulnerable to towed bottom-fishing gears can be effectively protected with long-term site-specific prohibitions, assuming adequate enforcement capabilities exist or voluntary initiatives and compliance are effective.
Real-time closures are another type of prohibition, whereby encounter-and-move-on rules substitute for strict avoidance of encounters. Real-time closures typically require a particular vessel to move a minimum distance away from the position of its last tow when the catch from that tow meets or exceeds a threshold weight or volume for a particular taxon. In practice, they do not necessarily minimize or eliminate further adverse effects on VMEs (Auster et al., 2011;Dunn et al., 2014;cf. Wallace et al., 2015). Moreover, fishing effort is likely to be displaced into similar (but less preferred) fishing grounds, which expands the trawling footprint and may increase total effort due to lower catch rates of target species (Kenchington, 2011), thereby increasing overall impacts to seabed habitats and biota. Temporary, move-on rules may thus produce unpredictable changes in effort and impacts overall and may be better considered as secondary to other measures for reducing trawling impacts on sensitive biota.

| Multipurpose habitat management
Trawling is commonly prohibited in designated areas, as part of a multipurpose habitat-conservation programme with much broader objectives than preventing local trawling impacts (Gell & Roberts, 2002

| Invertebrate-bycatch quotas
These measures establish quotas that limit trawl bycatch of vulnerable structure-forming species, such as coral and sponge. At present, they are being implemented for groundfish management in British Columbia, Canada, where fleet-wide and individual limits are intended to reduce and manage impacts on corals and sponges (Wallace et al., 2015). This management approach meets conservation goals for sensitive biota, without reducing landings of target species or displacing much fishing effort. The primary limitation is the substantial resource and cost associated with 100% observer coverage and enforcement. However, in data-poor or resource-limited situations at smaller geographical scales, a self-enforcement strategy among the fishers could substitute for the fishery management authority.
Although the only current known use of the measure is in the British Columbia groundfish fishery, favourable conditions seem to exist in other regions, such as Alaska, Australia, and parts of Europe, where such an approach could be implemented.

| Habitat-impact quotas
This management measure combines detailed mapping of sensitive habitats with vessel-location tracking to monitor the aggregate impacts of trawling by each vessel in relation to an overall impact quota, as measured by fishing activity (e.g., time or swept-area) in pre-defined habitat types (Holland & Schnier, 2006a, 2006b). Vessels, for example, could use their habitat quota by fishing for long periods on less-sensitive habitats or short periods on more-sensitive habitats, with their choice of location governed by the trade-off between catch rates of target species and the rate of use of habitat quota.
The primary advantage of habitat-impact quotas over invertebratebycatch quotas is that they do not rely on onboard observers, but on remote-vessel-tracking systems such as VMS that are a less expensive means to monitor fleet activity. The primary disadvantage is that bycatch controls based on fishing activity rather than actual bycatch may be inherently less precise. Habitat-impact quotas also require stakeholder agreement on the veracity of high-resolution habitat-sensitivity maps. These may not exist in many regions and are also expensive to create.
Habitat impact quotas have not been implemented in real fisheries to date, but they would be powerful management tools if the objective is to limit benthic impacts from trawling. Results from a dynamic, spatially explicit fishery-simulation model indicate that individual habitat quotas were more cost-effective for achieving habitat-management objectives than both fixed and rotating closures, although effectiveness depended on characteristics of the target-species fishery (Holland & Schnier, 2006a). The primary advantage of habitat-impact quotas over permanent closures is that they allow trawlers to evaluate where they fish in relation to catch returns per unit area of habitat disturbed. A negative aspect of this system is that it leaves open the possibility for some disturbance of sensitive habitats. Maintaining overall benthic habitat status would require a habitat-quota system that imposes a tariff that is proportional to the reduction in benthic status.

| Removal of effort
Total trawling effort is related to fleet capacity and the level of fishing activity. Fleet capacity encompasses the equipment and operational characteristics of vessels operating in a fishery and is commonly expressed in terms of total vessel tonnage (or length) and total engine power, or more simply as the number of vessels (Felthoven & Paul, 2004). Lowering trawling effort tends to cause a reduction in footprint and a contraction to core areas that are repeatedly fished (Kaiser, 2005), with a corresponding reduction in the extent of benthic impacts. The total catch of target species may decline at first, but if the target stock is overfished then fishery production should eventually improve in response to increased survivorship of the stock and reduced habitat impacts. Reduced competition should improve the economic performance of the remaining fishers. However, some of the benefits of regulations intended to change the level of effort can be countered by changes in one or more of the other controlling factors that affect catching power and which may not be regulated, such as changes in vessel or engine size when effort is regulated by days at sea (e.g., Eigaard, Marchal, Gislason, & Rijnsdorp, 2014). Total benthic impacts could inadvertently increase despite removal of effort, for example, if fishers invest their buyback grants to increase fishing capacity and move to other fisheries in more vulnerable habitats. In general, limiting effort will indirectly reduce the distribution and intensity of trawling and the associated impacts on benthic biota and may have more positive effects than implementation of MPAs which lead to fleet redistribution (Abbott & Haynie, 2012;Dinmore et al., 2003;Hiddink et al., 2006). However, effort reductions can be problematic to implement, especially in developing countries where one of the goals of management may be to employ a large number of people. It is noteworthy that the economic and societal costs of buyback programmes are immediate (Ye et al., 2013), whereas the potential ecological benefits of reduced effort tend to accrue more slowly and permit a more gradual societal readjustment to the management changes.

| MANAG EMENT C APACIT Y
The success of management measures to reduce trawling impacts on the benthos will depend greatly on the management capacity of the region. Melnychuk, Peterson, Elliott, and Hilborn (2017) have shown that while many of the richer countries have the capacity to identify and enforce fisheries-harvest regulations and to regulate location and gear used in bottom-trawl fisheries, many other countries lack these capacities. For example, the Asia-Pacific region provides a well-studied example that illustrates the challenges of open-access trawl fisheries with full-utilization markets that are managed for the 'triple bottom line', namely economic, environmental and societal goals (e.g., FAO, 2012FAO, , 2014Pho, 2007).
Millions of people are directly and indirectly employed by ~80,000 trawlers operating in mostly coastal areas throughout the region.
Nearshore waters with characteristically sensitive habitats are particularly important; for example, 90% of the marine catch in Vietnam is taken at depths < 30 m (Pho, 2007). Under these circumstances, broadly applicable measures such as spatial controls have been the most widely supported (FAO, 2014). In other cases, much more resource-intensive practices have been successfully implemented with the participation of multiple stakeholders, such as the invertebrate-bycatch quota system in British Columbia.

| INTER AC TI ON S WITH E XIS TING MANAG EMENT SYS TEMS
Based on our review of the effects of different measures, we conclude that there can be positive or negative interactions between these and many existing management systems. Potential interactions would therefore need to be considered systematically when considering the introduction of any new measure and, for this reason, we summarize such interactions and their consequences in Table 2. For example, freezing the trawling footprint to reduce benthic impacts could inadvertently affect existing catch controls (e.g., a TAC) by reducing the probability of achieving quota uptake if stock redistribution occurs but, at the same time, it is unlikely to interact with a measure for closed areas (Table 2). Similarly, the development of pulse trawling in the North Sea highlights the important point that any one measure will have both positive and negative consequences and suites of measures may need to be introduced simultaneously. Furthermore, interactions unrelated to fishery management might also need to be considered, such as protective measures intended for iconic species and de facto trawling prohibitions associated with disputed borders, shipping lanes and hydrocarbon operations.  If modified gear reduces catch efficiency for target species then effort required to take the TAC or quota would increase with consequent risk that effectiveness of measure is reduced or compromised.

| FIS HERY YIELD AND THE REL ATIVE B E NTH I C S TATUS
Closed areas will limit the areas where fisheries using alternate gears may operate; may increase risk of vessel interactions and gear conflicts.
Technical measures may increase the likelihood that any modified gear or operation will also have low environmental impacts per unit catch.
Prohibitions by gear type (Section 3.2) Low risk of unintended consequences.
If prohibited gear is replaced by gears with lower catch efficiency for target species then effort required to take the TAC or quota would increase with consequent risk of increases in total environmental impact.
Closed areas will limit the areas where fisheries using alternate gears may operate; may increase risk of vessel interactions and gear conflicts.
Technical measures may increase the likelihood that any gear substituted for the prohibited gear will also have high environmental impacts per unit catch.
Freeze trawling footprint (Section 3.3) Limits options for the fishery to respond to changes in stock distribution, risk of increasing effort in footprint to maintain catch, leading to lower profitability.
May reduce probability of achieving quota uptake, especially in case when stock distribution is changing.
Low risk of unintended consequences.
Reduce flexibility of industry to respond to consequences of freezing footprint. May prevent changes to gear that would maintain catchability of target species.
Nearshore restrictions and zoning (Section 3.4) Increased vessel interactions and/or gear conflicts in offshore areas.
May reduce probability of achieving quota uptake for species using nearshore areas.
Increase vessel interactions and/or gear conflicts if closed areas are not in nearshore zone.
Reduce flexibility of industry to respond to consequences of nearshore restrictions. May prevent changes to gear that would help maintain catches.
Prohibitions by habitat type (Section 3.5) Increased vessel interactions and/or gear conflicts in areas where trawling is not prohibited.
May reduce probability of achieving quota uptake for species associated with those habitats where trawling is prohibited.
Increase vessel interactions and/or gear conflicts.
Reduce flexibility of industry to respond to consequences of prohibitions by habitat type. May prevent changes to gear that would help to maintain catches.

Low risk of unintended consequences
May reduce probability of achieving quota uptake

Low risk of unintended consequences
Reduce flexibility of industry to respond and to develop and employ gears that reduce habitat impact.
Invertebrate bycatch quota (Section 3.7) Low risk of unintended consequences.
May reduce probability of achieving quota uptake.

Low risk of unintended consequences
Reduce flexibility of industry to respond and to develop and employ gears that reduce invertebrate bycatch.
Habitat impact quotas (Section 3.8) Low risk of unintended consequences.
May reduce probability of achieving quota uptake.

Low risk of unintended consequences
Reduce flexibility of industry to respond and to develop and employ gears that reduce habitat impact.
Removal of effort (Section 3.9) New measure and existing effort control are compatible.
Removal of effort may reduce probability of achieving quota uptake.
Low risk of unintended consequences.
Low risk of unintended consequences.
The effect of fishing on target stock biomass (B f ) can be described as where q is the catchability of the gear (the fraction of the exploited stock caught in a trawl pass), K f is the carrying capacity for fish and r f is the recovery rate of the fish. Accordingly, fishery yield can be calculated as Predictions from this simple impact-yield model not only serve to reinforce metric-based evaluations, they also provide a useful description of the relationship between target-stock dynamics, maximum sustainable yield (MSY) and the RBS (Figure 2; Table 1). In particular, technological developments of trawling gears such as elevated footropes, which reduce the d to q ratio (Figure 2, curve c) are shown to reduce benthic impact per unit of fisheries yield, while gears with a higher d/q ratio ( Figure 2, curves a, b)  (more sensitive) areas. Better targeting of aggregations of the target species is beneficial, as it will result in a higher catch per unit effort and therefore a higher yield at a lower benthic impact.
Curve (c) in Figure 2 represents fisheries where the ratio of the recovery rates of the benthos and the fish (r b /r f ) is high (i.e. benthic-fauna biomass has relatively higher rates of increase than fish biomass, and MSY is achieved at a lower F), while the curve (a) represents fisheries where the recovery rate for benthos (r b ) is lower than fish (r f ). The relationship shows that it may be possible to achieve a high yield with only a small reduction in RBS for trawl fisheries that exploit fish in resilient benthic habitats (high r b ) using gears that cause a low benthic mortality (d) but catch a large fraction (q) of the exploited stock. This relationship also implies that any form of spatial management that displaces trawling to benthic ecosystems with a higher r b will be beneficial, provided that r f remains the same (i.e., fish redistribute to those areas or have the same amount of food and productivity). Impact quotas in the form of invertebrate-bycatch or habitat-impact effectively increase the RBS by moving trawling effort away from sensitive areas (low r b ) to more resilient (high r b ) areas.
Identifying the point on these curves at which a fishery is currently positioned could assist in the identification of initiatives that may be most effective at reducing benthic impacts while maintaining catches.

| CON CLUS IONS
Our performance-based evaluation showed that best practices and the likelihood of reducing impacts of trawling on seabed habitats and biota will be influenced by the characteristics of the fishery and the ecosystem, as well as the local, regional or national values, priorities and resources. That is, regions where protection of seabed habitats The relationship between the relative benthic status (RBS) and yield of bottom-trawl fisheries for three different scenarios. Relationship (a) is for fisheries with a high d/q ratio-that is where a trawl pass catches a low proportion of the fish present (q) and causes a high mortality of benthos (d), or where the fishery occurs on benthic communities with a low rate of recovery r. Relationships (b) and (c) are for fisheries with a low d/q ratio-that is where a trawl pass catches a high proportion of the fish present while causing low mortality of benthos, or when fishing on benthic fauna with a fast rate of recovery r. On parts of the curve that are not coloured green, a reduction of fishing mortality (as indicated by arrows) increases both yield and RBS. The weight of the lines is proportional to the fishing mortality F, indicating that fishing gears that efficiently catch the target species need a lower F to achieve the optimum yield at a lower benthic impact. The figure illustrates that if the fish stock is fished beyond F MSY , a reduction in F will result in an increase of both yield and the benthic status (arrows in the grey to orange part of the curves). Reducing F from above F MSY to F MSY always reduces impacts on benthic biota and increases fishery yield, especially for gears with a high d and for trawling in sensitive areas with low r. Because this is a heuristic model, parameter values are not specified and no values are given on the x-axis as the conclusions do not depend on these values. No separate figures are shown to separate the effects of, for example, increasing q from decreasing d or increasing in r as they result in equivalent changes and biota is a high priority may choose to accept only a low level of impact or no impact, particularly for sensitive species such as corals and sponges. Other regions may decide that conserving a representative proportion of habitats within a network of MPAs is sufficient or that current trawling footprints are minimal and additional measures are not required. Because of the multiple and potentially interacting policy drivers that influence the management of fisheries and their environmental impact, we anticipate that the best practices for any particular region will enhance or adjust the emphasis of the existing management system, rather than overhaul it. For these reasons, and without regional context, we cannot be prescriptive about the selection of measures to manage these impacts and how to improve trade-offs between food production and environmental protection.
However, we have drawn attention to the broad range of potential practices that exist and that could be considered by managers and industry, as well as the interactions between them and the existing management system.
Based on the issues we have considered, four steps could be followed to help managers, industry, and other stakeholders gather and generate the evidence needed to evaluate potential best practices in their region, and to identify which measures would be most effective at reducing benthic impacts while maintaining fishery yields: first, identify all fisheries, environmental and socio-economic management objectives that may be affected by bottom trawling; second, evaluate the current bottom trawling footprint and concentration of activity within this footprint, preferably using high-resolution effort data but if necessary using data-limited methods; third, evaluate the distribution of sensitive habitats and any other habitats of concern in relation to the footprint of trawling; and fourth, evaluate in a regional context the effects of alternative management measures (Table 1), both individually and in combinations, on the probability of achieving objectives, while taking into account interactions between potential measures, and potential measures and the existing management system ( Table 2). The most suitable measures strongly depend on both the objectives and the data availability, which will differ among jurisdictions, which in turn means that the most suitable measures will also be different among jurisdictions. In Table 3, we identify the data requirements for implementing management measures and for subsequent evaluation of their effectiveness in terms of ecological and socio-economic impacts and compliance.
The main technical considerations when evaluating best practices are the footprint of the trawl fisheries and the gear-specific sensitivities of the benthic habitats and associated fauna (Table 3).
Trawling footprints have already been mapped in many regions (e.g., Amoroso et al., 2018;Eigaard et al., 2017), and when the requisite high-resolution spatial effort data are not available, trawling footprints can be estimated from the relationship between the regional swept-area ratio (area trawled in one year/area of region) and footprint (Amoroso et al., 2018). Regional swept-area ratio can be estimated from the product of mean vessel speed, mean trawl width and hours of trawling summed across fleet segments. Information on the broad-scale distributions of seabed habitats (Jenkins, 1997) and results from experiments describing the gear-and habitat-specific depletion and recovery rates of benthic habitats and biota Sciberras et al., 2018) have also been compiled. These data can be combined with footprint in a quantitative TA B L E 3 Data needed for a preliminary evaluation and implementation (PI), and the subsequent evaluation of effectiveness and fishery monitoring, compliance and surveillance (EC) of management measures and voluntary actions to reduce trawling impacts on seabed habitats and biota (Habitat maps and gear-habitat sensitivities are primarily applicable to PI.) Habitat impact quotas ( §3.8)

Removal of effort ( §3.9)
Note: The summary is based on material cited in the text (e.g., Section 3.1) or is otherwise based on consensus judgement by the authors. Catch and effort refers to aggregate landings/logbook data. Spatial effort by gear refers to the trawling footprint based on VMS or observer information. Light shading indicates the data type would be very useful, while dark shading indicates the data type is required. A particular data type should be considered an important prerequisite for a measure if it is required for PI, EC or both.
risk-assessment framework to estimate the RBS and test options for management Pitcher et al., 2017;. Individual-based simulation models are available to evaluate management options in both biological and economic terms, although the data requirements are considerable (Bastardie, Nielsen, & Miethe, 2014).
The linkage between fishery status and seabed status is another important consideration for management that seeks to reduce bottom-trawling impacts. For regions where bottom-trawl fisheries are implicated in generating high and unsustainable rates of fishing mortality on target stocks, actions taken to meet F MSY reference points are likely to lead to substantial reductions in seabed impact. Amoroso et al. (2018), for example, compared rates of fishing mortality on stocks caught with bottom trawls across a >200-fold gradient in bottom-trawling footprint. In regions with bottom-trawling footprints <10% of seabed area, fishing rates on bottom-dwelling fish stocks as expressed by the ratio of F/F MSY were almost always less than one and were therefore sustainable. But when trawling footprints exceeded 20% of seabed area, F/F MSY consistently exceeded one. Although this relationship is not strictly causal, given many of these stocks are also caught in other fisheries and the varying attributes of the existing management systems (Amoroso et al., 2018), it does imply that achieving sustainable rates of exploitation on target stocks leads to trawl fisheries that leave large areas (typically > 80%) of seabed unimpacted by bottom trawling. Improvements in stock status would also reduce the effort required to take the quota and further reduce benthic impacts per unit catch weight or value ( Figure 2), perhaps obviating the need for additional protective measures outside of particularly sensitive habitats.
Best practices will evolve as knowledge and experience increase or circumstances change. In any management system, it is therefore advisable to include an adaptive process (and funding) to monitor performance and allow for future refinements. Overall, this framework for considering best practices provides a necessary focus for stakeholder engagement in the development and ongoing evaluation of management plans concerned with the impacts of towed bottom-fishing gears on seabed habitats and biota.

ACK N OWLED G EM ENTS
The scientific results and conclusions, as well as any views or opin- Rugolo, R. White (AFSC); and two anonymous reviewers improved the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sharing is not applicable to this article as no new data were created or analysed in this study.