Exploring trade-offs in mixed fisheries by integrating fleet dynamics into multispecies size-spectrum models

Abstract

1. Ecosystem-based fisheries management aims to ensure ecologically sustainable fishing while maximising socio-economic benefits. Achieving this goal for mixed fisheries requires better understanding of the effects of competing fishing fleets on shared resources and economic performance. Proposed management strategies that promote either specialisation or diversification of catches may result in unintended consequences for ecosystem-based management. Here, we ask the following questions: does increased or decreased competition among fleets lead to better ecological and socio-economic fishery outcomes? How effective are currently proposed management strategies for achieving these outcomes?
2. We integrated fleet dynamics into a multispecies size-spectrum model and parameterised this model to represent Australian Southern and Eastern Scalefish and Shark Mixed Fishery. We compared the fishery status quo to two extreme scenarios: no competition, where each species is fished only by one fleet (specialisation); and maximal competition, where all fleets catch all species (diversification). To answer our second question, we considered three more plausible scenarios resulting from proposed management strategies: decreased competition due to reduced bycatch, and increased competition due to increased catches of under-utilised or valuable species. We used indicators to explore scenarios' outcomes.
3. Our model reproduced observed trends in fishing effort and yield. Extreme scenarios showed that a fishery dependent on single species management structures is more likely to achieve ecosystem-based management objectives if fleets do not compete, while maximal competition can lead to socio-economic loss as management buffers the ecological impact of diversifying. The more plausible scenarios showed little improvement over the status quo, with mixed ecological and negative economic effects.
4. Synthesis and applications. Our model can be applied to assess mixed fisheries ecosystem-based management strategies. Our results show that, under single species management approaches, greatest outcomes can be achieved when fleets are specialised, whereas managing fleets that catch similar species is unlikely to be successful. They question the effectiveness of these management approaches in providing resilience for mixed fisheries facing changes and highlight the need to account for fleet interactions in the evaluation of management strategies to avert unintended risks.

1 INTRODUCTION

Ecosystem-based fisheries management aims to ensure long-term ecologically sustainable resource use while maximising socio-economic benefits. Achieving this goal is complicated by a mismatch between policy and the complexity of socio-ecological systems upon which fisheries rest (Link & Browman, 2017). Management policies state ecological objectives, including maintaining target stocks within safe biological limits and minimising fishing impacts on bycatch species and habitats. But they lack clarity on the socio-economic dimension. The explicit economic objective remains broad, and often translates to maximising net economic yield in fisheries (Dichmont et al., 2010; Link & Browman, 2017). It is implemented at the whole fishery level, and it ignores that multiple fleets competing for a shared resource contribute differently to the economy of the fishery and the welfare of communities relying on fishing (Pascoe et al., 2014). Meeting socio-economic management objectives is only possible if the mechanisms regulating fleets' performance are better understood and if policy mirrors such understanding.

In mixed fisheries, complex ecological and competitive fleet interactions affect species abundance and fleet performance (Gaichas et al., 2017; Ulrich et al., 2012). For example, fishing one species leads to a decrease in its abundance and to changes in the abundance and availability of linked species, such as predators, or prey, or species competing for the same food source (Szuwalski et al., 2017). High catches by one fleet can result in less resource available to other fleets if fleets target the same species or if the target species of one fleet is the bycatch of other fleets. Furthermore, the cost of fishing and the price of fish influence fleet performance, with operational costs affecting competitiveness for fleets targeting the same species. In turn, market demand controls the price of fish which, for example, declines when the supply of fish outstrips the demand. A decline in the price of fish likely leads to a decrease in fleet revenue. The level of system complexity across ecological and economic factors scales up with the number of species, fleets and their interactions and so does the management challenge.

Several management strategies for mixed fisheries have been proposed to meet ecosystem-based management objectives. Among these strategies are two contrasting approaches: either specialisation or diversification of catches (Anderson et al., 2017). Fleets can specialise by minimising bycatch while maintaining sustainable yields of target species or they can diversify by targeting under-utilised species and thus by balancing exploitation across different populations as the abundance of these populations fluctuates due to environmental and depletion patterns (Robinson et al., 2020). These strategies find their counterparts in terrestrial agriculture, where diversification can lead to higher harvests and more stable farmers' revenue (Tamburini et al., 2020). This has laid the groundwork for similar comparative studies in fisheries, which also show increased catches and more stable revenues as fleets diversify (Anderson et al., 2017; Robinson et al., 2020). However, unlike agriculture, fisheries rely on the natural resources and involve multiple fleets that compete for such shared resources. While specialisation results in a reduction of competitive fleet interactions, diversification can lead to higher competition if, for example, all fleets target the same previously under-utilised species. The role of competition on the outcomes of these strategies is critical in understanding management effectiveness.

Our aim is to answer the following questions: (a) does increased or decreased competition among fleets lead to better ecological and economic fishery outcomes? (b) How effective are currently proposed management strategies for achieving these outcomes?

Understanding the effect of fleet competition on highly interconnected marine ecosystems requires analytical tools that can represent the ecological and socio-economic dynamics at play. Whole of system models have previously been used in such circumstances (Fulton et al., 2018) but can take considerable resources to develop, require the calculation and interpretation of many parameters and are challenging to fit to data (Hansen et al., 2019). In addition, complex results from these models can be difficult to translate into management decisions without specialist support (Primmer & Furman, 2012). The solution is to learn critical dynamics from these larger models but deploy more tractable models that can be more rapidly implemented and more transparently calibrated and interpreted.

Over the past decade, the development of multispecies ecosystem models that are relatively simple but can be complex enough to capture realistic outcomes of interactive mechanisms has progressed significantly (Blanchard et al., 2014; Plagányi et al., 2014). These models have been used to evaluate the ecological impacts and socio-economic fishery outcomes under different management strategies; especially the challenges in achieving ecosystem-based management objectives, including the need to account for fleet interactions (Pascoe et al., 2018). While these models have the capacity to represent species and fleet interactions simultaneously, their structural focus is most often concerned with either the ecological or the fishery component (Nielsen et al., 2017) as to capitalise on their strengths (simplicity) without overwhelming their potential by attempting to include too much. Size-based modelling approaches (Blanchard et al., 2017) form an excellent basis for taking a multispecies approach to exploring mixed fisheries. By being frugal but smart about the ecological simplifications made, they have the scope to connect the size-spectrum (trophic web) to environmental drivers at one end and economic drivers at the other without sacrificing ease of use.

Here, we integrate mixed-species fleet dynamics into a size-spectrum model (Blanchard et al., 2017; Scott et al., 2014) to simultaneously capture both economic and ecological dynamics. The ecological component is particularly useful for exploring the consequences of species- and size-selective fishing, which can change abundance and body sizes, with important feedbacks on ecosystem functioning (Law & Plank, 2018; Plank et al., 2017; Thorpe & De Oliveira, 2019; Thorpe et al., 2016). The fleet dynamics component is based on gear types, fleet profitability and management regulations, and it allows fleets to compete over species. We parameterise and calibrate the model to represent the Australian Southern and Eastern Scalefish and Shark Fishery, which is a multi-sector, multi-species fishery that covers almost half of the Australian Fishing Zone and is one of Australia's most complex fisheries (Novaglio et al., 2018, 2020). Using this model and the fishery, it represents as our case study, we evaluate a range of fleet competition scenarios and assess their performance using ecological and economic indicators.

2 MATERIALS AND METHODS

2.1 Integrating fleet dynamics into a multispecies size-spectrum model

Our multispecies size-spectrum model is based on the model framework detailed in the Mizer R package (modified from Scott et al., 2014; https://sizespectrum.org/mizer/; see Appendix S1 and Table S1 in Supporting Information). The model describes the population dynamics of each size-structured species as a function of an individuals' growth and mortality. Growth depends on food availability from prey within the explicitly modelled size-structured components and from a background resource representing small organisms not explicitly included in the model. Mortality is a combination of natural mortality, predation mortality, starvation mortality and fishing mortality. Fishing mortality is imposed by fishing fleets and depends on selectivity, catchability and fishing effort. Selectivity and catchability are defined by species and fleet-specific parameters (Appendix S3), while fishing effort is dynamic.

The dynamics of fishing fleets can be described as a function of economic and social factors, with the selection of factors considered and the assumptions underlying this component influencing the output of the model (Nielsen et al., 2017). Modelling approaches that focus either on the economic or social aspects are used under different circumstances. Economic models where profit is the main factor driving fishing fleets are most used to describe the dynamics of large-scale industrial fisheries, such as the Australian Southern and Eastern Scalefish and Shark Fishery (Nielsen et al., 2017; Novaglio et al., 2018). We used an extension of the classic Gordon-Schaefer economic model (Gordon, 1954; Schaefer, 1954) to describe the dynamics of fishing fleets as a function of profits. For each fleet, profits are the difference between the revenue and the cost of fishing. The revenue is the product of the fleet yield and the species market price which is given as input. The cost of fishing, also given as input, includes the fleet operational costs, such as fuel and salaries, and overhead costs, such as administration and vessels maintenance. Operational costs are proportional to fishing effort because the more a fleet fishes the higher is its expenditure on fuel and salaries, while overhead costs are fixed annual expenses that do not depend on effort (e.g. Bjørndal & Conrad, 1987). For each fleet, fishing effort increases if profits are positive and vice versa; it stays at the same level when the cost of fishing equals the revenue. This is a simplified model for change in fishing effort that includes some inertia: it stakes time for a fleet to adjust its effort in response to a change in revenue or cost (see Appendix S2).

The Gordon-Schaefer economic model describes an open access fishery, which means that, in the absence of management restrictions on effort, fleets increase their effort until the net profit drops to zero. This is consistent with how substantial proportions of the Australian Southern and Eastern Scalefish and Shark Fishery have operated historically when not constrained by management. However, between the late 1980s and early 1990s, this fishery became an output-controlled fishery (Novaglio et al., 2018). Therefore, we included a management component. We used the broken stick harvest control rule that defines a stock's biomass thresholds below which management rules are applied, reducing the effort allowed (Dichmont et al., 2016; Mackinson et al., 2018). While managing a multispecies fishery based on single species criteria that do not consider species and fleet interactions is a limitation, it reflects the current management framework of the Australian Southern and Eastern Scalefish and Shark Fishery as of many other fisheries worldwide (e.g. Pascoe et al., 2020). The target biomass thresholds that are commonly in place relate to the species' Biomass at Maximum Sustainable Yield (B_MSY). However, we adopted the harvest control rule as defined for this fishery, which also involves Maximum Economic Yield (B_MEY) reference points and limit reference points (below which fishing is stopped) set at 50% of Maximum Sustainable Yield (Biomass Limit, B_LIM; Figure 1; AFMA, 2019). The aim of the harvest control rule is to fish at levels that maintain species biomasses around B_MEY.

FIGURE 1

Changes in fleet's fishing effort according to the economic and management components. The coupled model evaluates whether the biomass of the fished stock/s is below reference points and whether the fleet is profitable. Based on this information, the model choses whether the economic or the management component determines changes in the fleet's fishing effort. Biomass reference points are Biomass Limit (B_LIM), Biomass at Maximum Sustainable Yield (B_MSY) and Biomass at Maximum Economic Yield (B_MEY). The dotted lines represent a situation when the change in effort is determined by either the economic or the management component based on which of the two rules results in a steeper decrease

The economic and the management components of our model are coupled. This means that the economic component determines changes in fishing effort based on profits, but management overrules profitability and forces fishing effort to stabilise, decrease or stop when stocks reach target reference points (Figure 1; Appendix S2). For each fishing fleet, if the biomass of all fished species is above B_MEY, fishing effort is determined by profits. If the biomass of one or more fished species is between B_MEY and B_MSY and profits are negative, fishing effort is still determined by profits. However, if profits are positive and hence would result in an increase in effort, management forces effort to stabilise and thus avoids further stock declines. Should the biomass of one or more fished species fall below B_MSY, fishing effort can only decrease. In this case, if profits are positive, management again determines the decrease in fishing effort, which is proportional to the decline in the species biomass relative to B_MSY. Given the multispecies structure of the model, the decrease in effort is also related to the species' biomass contribution to the fished community and the fleet's catch. Otherwise, if profits are negative, the decrease in effort is determined by either the economic or the management component based on which of the two rules results in a steeper decrease. Lastly, if the biomass of a species drops below B_LIM more constraining rules must come into play. In theory, the decline of even a single species should be sufficient to stop fleets catching the species. In practice, that is not possible and while targeted fishing is ceased, a small bycatch quota remains while the fleet keeps fishing for other target species.

As a simple approximation of the escalating public pressure that would pertain to a fishery where multiple fish stocks are depleted, we applied an additional rule in the model. If the biomass of five or more species is below B_LIM, management stops the fleet from fishing for a period of 5 years, after which fishing may begin again if less than five species have biomass levels below B_LIM. This limitation is an arbitrary setting but is informed by participatory management decision-making in the fishery and the stated pressure the resource assessment groups and management advisory committees are under when an increasing number of species are overfished. Moreover, the rule has proven sufficient to reproduce historical fishing-management interactions for management strategy evaluation purposes and has been previous successfully utilised in other fleet dynamics models in the region (Fulton et al., 2007).

2.2 Parameterising species and fleet interactions

Our model includes 19 species (Figure 2), two background spectra, which represent pelagic and benthic sources of food for the smaller individuals, and five fishing fleets. These components represent the main target, bycatch and key ecological species of the Southern and Eastern Scalefish and Shark Fishery, and the main fleets accounting for more than 80% of the fishery's catches (Table S4).

FIGURE 2

Output of the baseline model used during the two-step calibration approach to inform our final model (Appendix S4). Modelled (lines) versus observed (dots) trends in relative yield for all species included in the model, except for lanternfish and perch as time-series data is not available for these species

Parameters characterising each species and fishing fleet (Tables S1 and S5 and Appendices S1–S4) describe the species' bioenergetic processes and ecological interactions, and the fleet's competitive interactions, economy and management. We either obtained parameters from the literature and online databases (ABARES, 2017, 2018; https://www.fishbase.se) or estimated them through statistical analysis of available data (see Table S3 for data details and sources, and Appendix S3). We assumed some parameters to be invariant among species or calculated them according to other species-specific parameters (Table S2).

In our model, species predatory interactions are defined by the preference of a predator for a prey species and its size (Appendix S1). These preferences are determined by a predator–prey interaction matrix and expressed in terms of the ratio between the body mass of the predator and the body mass of its prey. Fleets can be considered as predators that choose their prey based on a preference for a target species and its selectable size (Appendix S2). We used a species–fleet interaction matrix to determine how the catch of each species is allocated across fleets and thus whether fleets compete for target species. Species–fleet interaction terms are parameterised based on available catch and effort data (from the Australian Fisheries Management Authority database; Table S3). Each fleet in the model uses a gear and each gear is characterised by a selectivity function, which determines the size of individuals within a species that the gear can select and retain. In addition, each fleet is driven by economic factors and controlled by management. Input parameters describing this relationship are the market price of each species, the cost of fishing for each fleet and management target reference points (Appendix S2). The method we used to calculate this set of species and fleet parameters is described in Appendices S3 and S4.

Last, we took a two-step process to model calibration (described in Appendix S4). Our calibrated model represents the system between 2006, when economic data required to inform fleet dynamics became available (from the Australian Bureau of Agricultural and Resource Economics and Sciences database; Table S3) and 2017 (Figure 3). It is informed by a baseline model without fleet dynamics and that represents the system between 1995 and 2017 (Figure 2). Main model limitations, including parameter uncertainty, and model outputs given alternative values of the uncertain reproductive efficiency parameter are discussed in Appendix S5.

FIGURE 3

Output of the calibrated fleet dynamics and size-spectrum model. (a) Modelled (lines) versus observed (dots) trends in fishing effort by fleet. (b) Modelled (top) and observed (bottom) yield by fishing fleet and species. Fleets are the shelf (SET-SH) and upper-slope (SET-US) trawling fleets; the shark gillnet fleet (SSG); the deep-slope (SET-DS) trawling fleet; and the Danish-seine fleet (SED). Effort is in number of trawl or gillnet hauls for the area covered by the fishery. Yield is in tonnes for the same area and considering only species that contribute more than 50 tonnes to the total annual fleet's yield. In addition to the species listed in this figure, our model includes lanternfish which are currently unfished but form a key prey species for many of the modelled species

2.3 Scenario projections and socio-economic indicators

We considered a range of scenarios using the calibrated model. First, we ran our model from 2018 to 2040 using constant price and cost data at the latest available values. This was the status quo scenario against which all other scenarios were compared. Then we explored alternative scenarios by changing the species–fleet interaction matrix while keeping all other factors constant (e.g. price, cost and management regulations) to isolate the effect of fleet competition on the fishery's outcome. The set of scenarios considered were:

• No competition: each species is fished only by the fleet with the highest proportion of the species' yield. Fleets are highly specialised and do not compete.
• Maximal competition: all fleets diversify their catches and target all species. All fleets compete for a shared resource.
• Less bycatch: fleet competition is decreased, relative to the status quo, by reducing fishing pressure on the bycatch of each fleet. This means that one species is caught only by the two fleets with the highest proportion of yields of the species. For example, gummy and school sharks are caught only by the shark gillnet and the shelf trawling fleet. Such scenario simulates a situation where fleets fish only their most important species in terms of yields.
• More under-utilised: fleet competition is increased, relative to the status quo, by increasing fishing pressure on under-utilised species by 50% across all relevant fleets.
• More valuable: fleet competition is increased, relative to the status quo, by increasing fishing pressure on most valuable species by 50% across all relevant fleets.

For the no competition, the maximal competition and the less bycatch scenarios, we changed species–fleet interaction terms but we kept the total fishing mortality for each species across all fleets at the start of the projections equal to that of the status quo scenario (Figure 4). Profitability and management restrictions for each fleet within each scenario are calculated based on the set of species that the fleet targets. In the no competition scenario, for example, each species is fished only by one fleet; therefore, each fleet targets a reduced set of species compared to the status quo (Figure 4a). Fleets might miss out much of the revenue associated with the mixed fishery, but they might counterbalance this loss with a gain in revenue due to higher catches of the species that they target. This is because being the only fleet targeting a species means being able to catch more of that species (as the allowed catch is not allocated across multiple fleets). It also means that the fleet is responsible for the species status and hence it is heavily and directly impacted by any management actions taken in response to the species potential depletion.

FIGURE 4

Scenarios and model projections to 2040. (a) Parameterisation of the species–fleet interaction matrix for each scenario (see Figure 3 for fleets' names). (b) Trends in fishing effort (top), where effort is in number of trawl or gillnet hauls, for each fishing fleet in each scenario, and relative trends in selected indicators (bottom; biomass of main target species, fishery's profit and yield) for each scenario. Shaded areas delimit modelled historical trends (Figure 3)

To quantify trade-offs across scenarios, we calculated indicators for 2040, the last year of the model run. We considered trade-offs after about 20 years to explore medium- to long-term outcomes and we also tested longer simulation runs (Appendix S6; Figure S12); the same approach could be used to support shorter-term tactical decision-making. Ecological indicators that we considered are the community biomass, the biomass of species that are sensitive to the effect of fishing or that have been historically depleted and are under a stock-rebuilding strategy, the biomass of target species, the slope of the size-spectrum and the number of species above fishery reference points. Socio-economic indicators that we considered are the number of active fleets, and fishing effort, yields and profits at the fishery level. The selected indicators are commonly used to track fishing-driven impacts on species and community structure and the socio-economic performance of the fishery (Blanchard et al., 2014; Thorpe et al., 2015, 2016). Each indicator focuses on a specific component of the ecosystem. For instance, the slope of the size-spectrum informs changes in community size-structure and composition and whether fishing leads to a decrease in large individuals or to an increase in small individuals due to trophic cascade effects (Blanchard et al., 2014). Fishing has other consequences on the community, such as a decrease in the abundance of target species, which biomass indicators will track. The trade-offs associated with each scenario are based on the full set of indicators.

3 RESULTS

3.1 Model validation

Our calibrated model reproduced observed trends in effort and yields for most fishing fleets as well as observed yield composition by fleet for the main target species (Figure 3). This indicates that profit-based changes in fishing effort, management and fleet competition can explain most of the dynamic of the fishery. Management and fleet competition played a key role in determining the fleets' effort and yields. At the start of the model run (2006), all fleets targeted some overfished species, therefore management forced effort to decrease. High numbers of target species led to the potential for a high number of management restrictions; and the higher the level of overfishing the steeper the mandated decrease in effort. For instance, the shelf trawling fleet caught many species and mostly targeted flathead and morwong, which were overfished and hence effort deceased. Also, fleet interactions intensified declines in effort. High fishing pressure on one species by one fleet hampers the species' ability to recover and forces all fleets catching even small quantities of the species to decrease their effort. Under these conditions, the profit-based rule for changes in fishing effort plays a marginal role in determining the fleets' effort. It overrules management only if profits are negative and the economic loss results in steeper decreases in effort than those dictated by management. As fishing effort decreased and species struggled to recover, fleets caught less and yields also decreased.

The species–fleet interaction matrix spreads fishing intensity on each species across fleets and thus determines how species' yields are allocated across fleets (Figure 4). Congruent with observations, in our model flathead and morwong made up the bulk of the shelf trawling fleet catch, blue grenadier and silver warehou were the most abundant species for the upper slope trawling fleet, gummy and school sharks were the main target of the shark gillnet fleet, and orange roughy was the main target of the deep slope trawling fleet. School whiting and flathead were commonly caught species for the Danish-seine fleet, but the model overestimated the fleet's mackerel and squid catches.

3.2 Evaluating extreme scenarios: Does increased or decreased competition among fleets lead to better ecological and socio-economic fishery outcomes?

Model projections showed that alternative fleet competition can give rise to alternative fishing patterns and can lead to different fishery outcomes (Figure 4). The initial reduction in effort at the start of the projections—common to all scenarios but particularly marked for the no competition and the maximal competition scenarios—was management driven. In both scenarios, some of the fleets were characterised by negative profits which could also lead to a decrease in effort (Figures S4 and S5). However, the impact of these fleets on their target species was more of concern than the fleets' economic loss, hence management determined the reduction in effort (see integrating fleet dynamics into a multispecies size-spectrum model section and Appendix S2).

No competition means that fleets are highly specialised and heavily impacted by any management actions taken in response to species depletion (Figure 4). For example, in our model, the shelf trawling fleet was the only fleet targeting flathead and was responsible for its depletion. High yield of a depleted species led to a steep decline in effort, which decreased until the fleet became almost inactive. After a period of inactivity, the target species recovered, and the fleet started fishing again. At low fishing effort, this fleet saw growing yields of the recovered target species and made high profits. At the whole fishery level, biomass, yields and profit increased. While increases in yields and profits were mostly due to the dynamics of the shelf trawl fleet, increases in biomass were due to decreases in effort by all fleets. In 2040, this resulted in higher community biomass, and biomass of the main target and sensitive species, and a higher slope of the size-spectrum compared to the status quo (Figure 5). More species were above the target reference point of B_MEY, more fleets were fishing, and the fishery's yields and profits were also higher compared to the status quo.

FIGURE 5

Changes in indicators across scenarios and relative to the status quo. (a) Percentage changes in community biomass, biomass of species under stock-recovery strategy, biomass of main target species, slope of the size-spectrum, and fishery effort, yield and profit. (b) Changes in the number of species above fishery reference points and active fleets. Values in blue mark no change or positive changes, while those in orange mark decreases

Maximal competition represents the opposite situation, meaning each fleet diversifies its catch by targeting all species and hence is impacted by management responses for each of those species (Figure 4). If the biomass of a species is below B_MSY, effort decreases for all fleets. Also, if the biomass of five or more species declines below B_LIM, all fleets are forced to stop fishing. In this scenario, few species closed the fishery, and this resulted in a steep increase in biomass for most species in the community. Nevertheless, some species remained below the fishery's reference points, which meant that fleets struggled to recommence operations, leading to lower yields and profits but higher community biomass and slope of the size-spectrum compared to the status quo (Figure 5).

These scenarios suggested that if the management rules used here apply and given our model assumptions (Appendix S5), catch diversification leads to greater fleet competition and to socio-economic loss as management forces fishing to decline or stop; though there can still be marked increases in biomass as the community recovers from fishing. In contrast, a fishery where fleets specialise sees positive economic and ecological outcomes.

3.3 Evaluating plausible scenarios: How effective are currently proposed management strategies for achieving better fisheries outcomes?

The less bycatch, more under-utilised and more valuable scenarios showed similar trends across fleets to those of the status quo (Figure 4). In general, fishing effort decreased for most fleets and biomass recovered. Towards the end of the projections, yields and profits increased despite effort remaining low, due to greater available biomass. Fishing effort remained low because the biomass of some species had not recovered above fishery reference points. There were, however, some differences across these scenarios.

In the less bycatch scenario, the decrease in fishing effort for the Danish-seine and the shelf trawling fleets was slightly steeper than in the status quo (Figure 4) because the two fleets caught more of their target and overfished species (and less of their bycatch species). Steeper declines in effort translated into higher abundance, size-spectrum slope and number of species above B_MEY in 2040 (Figure 5). However, recoveries were not marked enough to result in higher yields and profits at lower effort compared to the status quo. The more under-utilised and the more valuable scenarios saw an initial increase in yields, which was matched by an equal increase in profits for the latter scenario due to higher fishing pressure on species with higher market price (Figure 4). However, most of these preferentially targeted species were below B_MSY, and management decreased effort thus reversed the increasing trends in yields and profits. In 2040, the biomass of target species was equal or lower while the biomass of sensitive species was higher compared to the status quo (Figure 5). Under-utilised and high-value species are common targets and more fishing pressure on them meant that their abundance declined while the abundance of other species increased, due to predation release and more food becoming available. More effort on under-utilised species (typically smaller species) shifted the community size-distribution towards bigger sizes compared to the status quo and thus slightly increased the slope of the size-spectrum. On the other hand, more effort on high-value species, such as school and gummy sharks, which are often large-bodied species slightly lowered the slope of the community size-spectrum. The more valuable scenario also showed an increase in the number of overfished species. High number of overfished species triggered a strong management reaction which limited the fleets' effort. Lower effort corresponds to lower yields and profits if species do not recover fast.

Overall, the less bycatch scenario suggested that, within the constraints of our model, plausible management strategies that promote specialisation can have positive ecological but negative economic outcomes. Plausible strategies that promote diversification can increase competition on under-utilised or high-value species and can have some negative ecological effects and only negative economic effects. See also Figures S3–S8 for trends in species biomass and in fleets' yields and profits for all scenarios considered. Our model and scenarios assume that all yields are landed (see Appendix S5 on the limitations of not accounting for discards).

4 DISCUSSION

Competition among fishing fleets has important implications for the dynamics of mixed fisheries and their management. Here, we used a modelling approach to document the consequence of competition on the ecology and economy of a mixed fishery and to explore the effectiveness of management strategies that act to reduce or increase competition in achieving their objectives. We found that a fishery dependent on single species management structures is more likely to achieve ecosystem-based management objectives if fleets do not compete. This means that fleet interactions are removed and hence the fishery is closer to the assumptions behind the single species management approach. This modelled fishery led to higher biomass and number of species above management reference points, more active fleets and greater overall yields and profits (Figure 5). Other studies also show the gains of reducing competition in fishing which are mostly linked to more sustainable fish resource, better payoff, and a situation where more fishers or fleets coexist (Burgess, 2015; Mullon & Mullon, 2018). However, this and other studies also suggest that some fleets benefit from the lack of competition while other fleets suffer (Thorpe et al., 2016). The struggle for some fleets is due to a loss of yield and revenue from the multispecies community, or a reluctance to adopt strong management interventions when the biomass of target species is declining that would be followed by ecological and economic recoveries. The winners in our study were the Danish-seine fleet and the trawling fleet operating on the shelf and the loser was the trawling fleet operating on the upper continental slope. This implies that assessing the fishery as a whole means ignoring fleet-specific patterns, thereby potentially hindering the ability of management to fully meet implicit ecosystem-based management objectives, such as stable employment.

We also showed that the socio-ecological systems response to changing fleet interactions is nonlinear. Management measures that act to promote specialisation reduce fleet competition, but their effectiveness can vary depending on both ecological and fleet interactions. For instance, our extreme scenario where fleets specialise to the greatest degree possible (do not compete) resulted in the best fishery's ecological and socio-economic trade-offs; yet, the more plausible scenario where bycatch is reduced that also leads to some specialisation among fleets resulted in minimal ecological improvements (Figure 5). Ecological and fishing dynamics act together to give rise to complex systems where responses to changing conditions can be counterintuitive (Audzijonyte & Kuparinen, 2016). For example, in our model, the disproportional system response to different degrees of specialisation can be due to particularly influential predator–prey and species–fleet interactions that were affected only when specialisation was greatest. A useful management-oriented step forward would be to understand which predator–prey and species–fleet interactions are most influential and to set species' catch quotas that account for such understanding. This would require transitioning from single-stock catch-based management to multispecies fleet-based management (Thorpe et al., 2015; Ulrich et al., 2012).

Diversification, which is here intended as broader targeting across species, led to increased fleet competition and overall negative economic performance in the context of our model, where a fleet cannot avoid interaction with a species should the species become depleted (Figure 4). Taken at face value, this result is in contrast with empirical studies that propose diversification as a strategy to increase catches and stabilise revenues (Cline et al., 2017; Robinson et al., 2020; but see Ward et al., 2018). However, in real-world fisheries, diversification can also be achieved by targeting different areas or suits of species seasonally instead of expanding the set of species fished (as done here). Furthermore, diversification in previous studies has typically been explored through the lens of statistical analyses of catch and effort datasets and with a focus on socio-economic rather than ecological gains (Beaudreau et al., 2019). These analyses show that diversification decreases and revenues drop as management restrictions characterising output-controlled fisheries intensify but investigating the cause and effect of this relationship is typically beyond their scope. In contrast, our mechanistic model allows for exploration of socio-ecological feedbacks. We showed that diversification leads to increased fleet competition and overexploitation; therefore, management limits the fleet's ability to fish and profits decline.

This is one plausible system dynamic, and alternatives exist. They include the integration of spatial and seasonal fishing dynamics and of a greater capacity by fishing fleets to modify targeting in response to stock status. These alternatives may allow for diversified catches while avoiding constraining stocks and can be observed in real-world fisheries (Branch, 2008), but such fine scale targeting behaviour can be difficult to represent in coarsely resolved models (Fulton et al., 2007). Nevertheless, our results highlight that mixed fisheries will struggle to diversify while remaining environmentally and economically sustainable if they rely solely on single species management approaches without the capacity to call on a diversity of management approaches, or to coordinate activities across fleets and species fished.

Globally, single species output-control restrictions that have been shown to reduce overfishing promote specialisation (Holland & Kasperski, 2016; Ward et al., 2018). These regulations direct effort on the most productive and valuable stocks, endorse technological changes that increase efficiency, and bind the fleet's choice of target species by limiting fishing on species below management reference levels. However, these constraints can hinder the fleet adaptability to changing conditions and are often unsuited to the highly diverse nature of mixed fisheries (Kulanujaree et al., 2020). While management systems that lack such restrictions have the capacity to promote diversification, they may lead to unintended consequences including the overexploitation of unregulated and sensitive species and can ultimately result in ecological and economic failures (Beaudreau et al., 2019). Using models to explore the strengths and drawbacks of management options can be enlightening and adopting a flexible mix of management approaches suited to the system in question is to be recommended.

There are important benefits of catch diversification which can provide resilience for fishing communities facing changes (Cline et al., 2017). Target species can become less available due to overexploitation, changes in management regulations or environmental conditions that effect their distributions and productivity; they can also become less valuable due to market forces (Cline et al., 2017; Robinson et al., 2020). Changes in environmental and market conditions can be steady through time or can be abrupt and lead to sudden regime shifts, leaving fishing communities limited time to adapt if they are not already relying on a diverse set of species. However, catch diversification at an operational level can also bring risks (Beaudreau et al., 2019). Diversifying involves additional costs, such as the costs of gaining extra skills, equipment, permitting and quotas needed to target new species, work new gears and explore new fishing grounds, with the associated risks of venturing new opportunities. These additional costs are seldom factored in when considering the benefits of catch diversification, but they can deter fishers from pursuing diversified strategies (Fulton et al., 2007).

Catch diversification can be achieved in many ways and balanced harvest is a diversification strategy that has been widely debated (Garcia et al., 2015). It refers to targeting species and/or sizes according to their productivity (Garcia et al., 2012) and it may be effective in maintaining overall fishery yield while reducing the impact on ecosystem structure and functioning (Law & Plank, 2018; Plank et al., 2017). Our results reinforce the argument that such a radically different fishing style would also require new management approaches; diversifying while relying on single species management structures would not work. In addition, market forces would need to be considered because balanced harvest, or any diversification strategies, can pay off in both ecological and economic terms only if there is a market for less preferred and under-utilised species (Garcia et al., 2015). In countries such as Australia, where consumer demand is concentrated on a few species, this may require challenging shifts in consumer choices (van Putten et al., 2019).

The key questions pertinent to any fishing strategy are what management framework will foster socio-ecological resilience under changing climatic and economic conditions, and how we can test it before its implementation to avert unintended risks (Beaudreau et al., 2019). Dissatisfaction with the performance of single species-oriented approaches in a multispecies context has led many to consider the benefits of adopting ecosystem-oriented approaches (Patrick & Link, 2015). This paper emphasises the need for a rethink in management approaches for mixed fisheries, where either technological improvement removes fleet interactions or fishery management avoids the serial application of single species harvest control rules. For instance, most of our modelling experiments showed that management forces effort to decrease because species' biomass is below maximum sustainable yield and hence forces fleets into an inefficient state (Figures S3–S8). In a multispecies context, biomass at maximum sustainable yield can exceed biomass at maximum economic yield for some species (Hoshino et al., 2018); thus, the forced decrease in effort is in contrast with the harvest control rule which aims at maintaining species biomass around biomass at maximum economic yield. Also, achieving this target reference point across most species may be unfeasible for mixed fisheries (Hoshino et al., 2018; Pascoe et al., 2020). While transitioning towards multispecies or ecosystem-oriented management rules has proven challenging, research on multispecies management frameworks is progressing fast and proposing new options, with a balanced consideration of the benefits and risks of such approaches (e.g. Pascoe et al., 2020; Thorpe & De Oliveira, 2019). Our future work aims at considering some of these options to allow comparison of fisheries outcomes across both fleet competition patterns and management frameworks to identify ecological and socio-economic trade-offs, including those emerging from a much-needed improvement to the management of mixed fisheries.

Our study can also be extended to help further explore these key management questions under changing climatic and economic conditions. This would involve the integration of recently developed protocols linking size-spectrum models to changes in oceanographic variables (Woodworth-Jefcoats et al., 2019) and more work to understand how competitive fleet interactions may change as climate and economy alter species availability and market value. This approach would be well suited to test management regulations, including multispecies frameworks, and an expanded set of fishing strategies. For example, some studies on balanced harvest have received criticism for their limited realism of multispecies fleets and economics (Froese et al., 2016) and our modelling approach (or modifications to it) could be used to address these perceived shortcomings and to test the potential of balanced harvest as fisheries evolve.

Ecological and socio-economic trade-offs are inherent to mixed fisheries, and any other ocean economic activity, and choosing among them is ultimately a societal call (Novaglio, Bax, et al., 2021). Here we provide a tool that is appropriate for strategic considerations and that can be used in conjunction with other models to inform such decisions. We showed, as others have (Ulrich et al., 2012), that fleet dynamics and interactions can play an important and often unexpected role in management effectiveness and overall fishery outcomes, thus influencing trade-offs. Management plans for mixed fisheries and the models that inform them seldom take a fleet-centred approach and explicitly acknowledge the complexities of these fisheries. Despite the many challenges, we showed that it is feasible to integrate mixed fisheries complexities, including both fleet and species interactions, into relatively simple models and thus to assess the effectiveness of alternative management rules more readily. It is through such improved representation of fishing activity that we are more likely to identify management benefits and to avert ecological and socio-economic risks.

ACKNOWLEDGEMENTS

We are grateful to Dr Jemery Day, Dr Robin Thomson, Dr Miriana Sporcic and Dr Catherine Bulman for their assistance in gathering and interpreting fisheries data for the Australian Southern and Eastern Scalefish and Shark Fishery. Thank you to two anonymous reviewers for their constrictive reviews and improvements to the manuscript.

CONFLICTS OF INTEREST

The authors have no conflicts of interest to declare.

AUTHORS' CONTRIBUTIONS

C.N., J.L.B. and E.I.v.P. conceived the ideas; C.N., J.L.B., M.J.P., A.A. and E.A.F. performed the modelling and C.N. led the writing of the manuscript. All authors contributed critically to the drafts and gave final approval for publication.

Open Research

DATA AVAILABILITY STATEMENT

Raw data available at https://www.afma.gov.au/ and https://www.awe.gov.au/abares/. Modelled data and code available via the Zenodo Repository https://doi.org/10.5281/zenodo.5715717 (Novaglio, Blanchard, et al., 2021).