The consequences of balanced harvesting of fish communities

Abstract

Balanced harvesting, where species or individuals are exploited in accordance with their productivity, has been proposed as a way to minimize the effects of fishing on marine fish communities and ecosystems. This calls for a thorough examination of the consequences balanced harvesting has on fish community structure and yield. We use a size- and trait-based model that resolves individual interactions through competition and predation to compare balanced harvesting with traditional selective harvesting, which protects juvenile fish from fishing. Four different exploitation patterns, generated by combining selective or unselective harvesting with balanced or unbalanced fishing, are compared. We find that unselective balanced fishing, where individuals are exploited in proportion to their productivity, produces a slightly larger total maximum sustainable yield than the other exploitation patterns and, for a given yield, the least change in the relative biomass composition of the fish community. Because fishing reduces competition, predation and cannibalism within the community, the total maximum sustainable yield is achieved at high exploitation rates. The yield from unselective balanced fishing is dominated by small individuals, whereas selective fishing produces a much higher proportion of large individuals in the yield. Although unselective balanced fishing is predicted to produce the highest total maximum sustainable yield and the lowest impact on trophic structure, it is effectively a fishery predominantly targeting small forage fish.

1. Introduction

The ecosystem approach to fishing has increased the interest in managing fisheries impacts at the level of fish communities and ecosystems [1,2], but so far, it is unclear exactly how this can be done. This relates not only to the proper definition and use of the concept of maximum sustainable yield (MSY) in a multi-species management context [2], but also to how exploitation patterns might be changed to better achieve fisheries and conservation objectives simultaneously. Recently, it has been suggested that both objectives can be achieved if species and individuals are exploited in relation to their natural productivity, the so-called balanced harvesting [3,4].

The theoretic rationale behind balanced harvesting is inspired by classic ‘surplus production’ theory, which predicts that the fishing mortality leading to MSY is proportional to natural mortality [5–7]. In an unexploited situation, natural mortality is equal to production per unit of biomass [8], and it is therefore expected to scale to body mass raised to the power of −0.25 [9]. The natural conclusion is that the largest yield from an ecosystem can be achieved if all species are fished proportional to their theoretical productivity, and hence that small species should be fished more intensely than large species. However, there are two complicating issues: first, most fish have large size differences between offspring and adults, and it is therefore relevant to ask whether productivity is measured at the stock level (as in classic surplus production theory) or at the level of individuals (as in [10,11]). Current single-species MSY management strives to enforce fishing mortality and size selectivity to maximize production in a manner that may be termed ‘balanced selective’ fishing, where juvenile fish are protected from fishing and adults are fished in proportion to their productivity. An exploitation pattern where each individual is exploited in proportion to its productivity challenges the predominant belief among fisheries scientists that MSY is achieved when juveniles are protected [12,13]. Second, changes in the abundance at higher trophic levels affect the productivity of lower trophic levels: when the fishing mortality on one trophic level goes up, the predation mortality on the lower trophic level goes down, and consequently that trophic level may tolerate a higher fishing mortality. This replacement of natural mortality by fishing mortality was actually one of the rationales behind balanced harvesting [14], and has the consequence that the fishing mortality producing MSY is different in the multi-species and single-species cases [15]. An evaluation of balanced versus unbalanced harvesting therefore requires a community perspective where changes in trophic interactions owing to fishing are fully accounted for [11].

A central piece of empirical evidence supporting balanced harvesting consists of studies of fisheries along African lakes demonstrating that non-regulated fisheries predominantly targeting smaller individuals had little impact on the size structure of the local fish communities [14,16]. Apart from these studies, the impact of balanced harvesting has mainly been explored through model predictions. Simulations with Ecosim models have shown that increasing the exploitation rate of small species generates less change in biomass distribution across trophic levels than increasing exploitation rate equally across species [17]. Garcia et al. [3] compiled results from a series of Ecosim and ATLANTIS models, and concluded that unselective harvesting was superior to selectively fishing fewer ecosystem components. The unselective pattern was better in terms of total yield and biomass, and produced fewer local population extinctions. Many of the models used in these simulations did not resolve the size structure of the species or did so using only few life stages and assuming food-independent growth from one stage to the next. It can therefore be questioned how well these simulations reflect the outcome of unselective harvesting.

Simulations of unselective balanced harvesting, where each individual is fished in proportion to its productivity, require a finely resolved or continuous size structure. Such simulations have only been performed with single-species size-spectrum models where it was found that targeting only small individuals caused less disruption to the size spectrum than targeting larger individuals, and that fishing according to productivity resulted in higher resilience [10,11]. Because unselective balanced harvesting implies a higher fishing intensity on small than on large individuals, we expect the size composition of the catch to differ from that of the traditional size-selective fisheries. A catch composed largely of small individuals may be desirable (e.g. as in some artisanal fisheries [14]), while it may be less so in industrial fisheries for human consumption, where larger individuals, regardless of species, typically return a higher price per weight. It is therefore relevant to extend the description of balanced harvesting beyond total yield and total biomass [3,10,18].

Our aim is to systematically examine how exploitation affects catch and community structure by comparing balanced versus unbalanced and selective versus unselective exploitation patterns. To this end, we use a size-based fish community model that fills the gap between the single-species size spectra and the community models [19–21]. The model provides a mechanistically based description of fish community dynamics, taking account of both intra- and interspecific competition and predation. Compared with traditional single-species models, predator–prey interactions and food-dependent growth are fully resolved, making the model suitable for long-term projections of an entire community. In contrast to previous publications on balanced harvesting, our model resolves the full life history of the individual fish from egg to adult. These attributes allow to us to explore community responses of exploitation of varying sizes and species in much greater resolution than previously carried out.

To precisely define the terms balanced/unbalanced and selective/unselective fishing, we use four exploitation patterns and use these to address total yield, size composition of yield and biomass distribution at the community level. The patterns differ with respect to whether or not fishing mortality is scaled with productivity (balanced or unbalanced), and whether or not juveniles are protected (selective and unselective). We thus compare the outcome of: (i) selective unbalanced fishing, a pattern that is a good approximation to the current exploi tation of the North Sea fishes, where juveniles are protected and adults are fished with approximately equal mortality across groups [22]; (ii) unselective unbalanced fishing, where all individuals are exploited with the same fishing mortality irrespective of whether they are juveniles or adults; (iii) selective balanced fishing, where juveniles are protected from fishing and adult fishing mortality scales with productivity at the population level; and (iv) unselective balanced fishing, which is the ‘ideal’ balanced pattern where each individual is exploited relative to its productivity. For a given total yield, we ask which exploitation pattern results in less change in the distribution of biomass across the community. We also compare the size composition of the catch and show that although balanced fishing causes less change in the structure of the community, it generates a substantial reduction in the average size of fish being caught.

2. Methods

(a). Trait-based size-spectrum model

We use the size-spectrum model developed by Andersen & Beyer [19] to estimate the equilibrium yield and biomass for each of the four exploitation patterns at different levels of fishing. The model is based on individual-level processes—encounters, growth, mortality and reproduction—and represents the fish community as a continuum of ‘species’ with increasing asymptotic body weights. In the computational implementation, the continuum is represented as a discrete number of asymptotic weight classes, which are referred to as ‘species’ for the sake of simplicity. A species i in the model is characterized by its asymptotic body weight, , and its population by its size distribution N_i(w). The results from the model are independent of the number of species as long as this is larger than 10. In our simulations, we used 20 species in the range from 10 to 30 000 g (figure 1).

Figure 1.

The fish biomass density spectrum in the unfished state. The thick line is the community spectrum, and the thin black lines show 20 ‘species’ with different asymptotic body weights.

The model rests on two main assumptions: (i) fish eat other fish that are smaller than themselves, and (ii) individual fish can be described exclusively by their individual and asymptotic body weights. The individual-level processes and the scaling from individuals to N_i(w) are described in detail in [21] and summarized in table 1 with parameters defined in table 2. Here, we provide a general overview of the model.

Table 1.

Equations governing the model. N_i(w) is the abundance density of individual fish with weight w from species i with asymptotic size .

description	equation	no.
community
conservation equation		M1
community spectrum		M2
food consumption
lognormal food preference curve		M3
encounter rate		M4
consumption		M5
somatic growth and reproduction
available energy		M6
allocation of energy		M7
somatic growth		M8
egg production		M9
density dependence		M10
max. recruitment		M11
mortality
predation mortality		M12
background mortality		M13
starvation mortality		M14
selective unbalanced fishing		M15
unselective unbalanced fishing		M16
selective balanced fishing		M17
unselective balanced fishing		M18
size-selection function		M19
resource spectrum
resource production		M20

Table 2.

All parameters in the model.

parameter	value	definition
foraging
β	100	preferred predator–prey weight ratio [23]
σ	1.3	width of selection curve [20]
γ	29.2 g−q yr−1	search volume factor [21]
q	0.8	search volume exponent [19]
physiology
h	20 g1 −n yr−1	max. consumption factor
n	0.75	max. consumption exponent [20]
	0.25	size at 50% maturation relative to W [20]
α	0.6	assimilation efficiency [20]
κs	2.4 g1 −p yr−1	standard metabolism factor
p	0.75	standard metabolism exponent [24]
ɛ	0.1	gonad efficiency [21]
a	0.66	expected mortality/growth ratio
μ0	2 g1 −n yr−1	background mortality factor
ξ	0.1	fraction of energy reserves used for reproduction
resource
κ,κr	0.005 gλ−1 m−3	carrying capacity of background spectrum
λ	2.05	resource community exponent
r0	4 g1 −p yr−1	resource community growth factor
fishing
ηF	0.05	fraction of W∞ to recruitment for trawl selectivity
wr	10 g	size at recruitment to fishery
wm	444 g	size at which for all fishing patterns
	free	fishing intensity

In the model, each individual is described by its asymptotic body weight, , and by its individual weight, w. Using mechanistically based submodels for growth, mortality and reproduction, the model calculates the number of individuals in each species's weight-group, N_i(w). All processes are subject to a conservation equation (M1), where growth and mortality accounts for the change in the number of fish and biomass within a given size class. In order to grow, an individual must encounter and ingest food. Food is selected from the entire size spectrum (M2) using a lognormal prey size preference (M3). The food encountered by a predator is the product of available food and a size-specific search rate (M4). The amount of encountered food consumed is limited by the maximum rate of food processing hwⁿ and modelled with a Holling type II functional response (M5).

The food consumed is used on standard metabolism, somatic growth and reproduction. The available energy is what remains after consumption is assimilated and standard metabolism is subtracted (M6). As a mature individual approaches its asymptotic weight, an increasing fraction of the available energy is allocated to reproduction (M7) and this reduces its somatic growth (M8). Growth and reproduction in the model are therefore linked and food-dependent. The particular allocation to reproduction (M7) is chosen such that a constant ratio between consumption and maximum consumption results in a growth pattern similar to a von Bertalanffy curve [21]. When the individual matures, energy is allocated to egg production (M9). Eggs hatch to become larvae using a Beverton–Holt-like density-dependent recruitment function (M10), where the maximum recruitment, , depends on the equilibrium solution of the model (M11) [19].

Natural mortality comes from three different sources: predation mortality, background mortality and starvation mortality. Predation mortality is proportional to encounters (M12). Background mortality is implemented to prevent the build-up of large individuals at zero fishing, and is assumed to scale with asymptotic weight (M13). Starvation mortality is only present when feeding is less than required to cover standard metabolism (M14).

To sustain the smallest individuals in the fish community, a dynamic background spectrum of edible plankton is provided. We model the spectrum as a semi-chemostat resource with a carrying capacity determining the total production of the system (M20).

(b). Fishing mortalities

Fishing is imposed as four conceptually different exploitation patterns (figure 2). The patterns differ with respect to their size selectivity within and across species. Further, the exploitation patterns are divided into ‘balanced’ and ‘unbalanced’ types of fishing. Selective fishing refers to exploitation patterns where juveniles are protected from fishing and individuals are recruited to the fishery as they approach some fraction of their asymptotic weight. Selection is modelled by a sigmoidal function that switches smoothly from 0 to 1 at the size (M19). In balanced fishing, individuals or species are exploited in accordance with their productivity as determined by metabolic scaling rules (i.e. scaling fishing mortality with individual or asymptotic body weights with exponent −0.25) [9,25]. Unbalanced fishing allocates F equally over all asymptotic weights or sizes. This gives four different fishing combinations: selective unbalanced fishing, (M15; figure 2a), unselective unbalanced fishing (M16; figure 2b), selective balanced fishing (M17; figure 2c) and unselective balanced fishing (M18; figure 2d). All exploitation patterns are subject to a lower cut-off size where the smallest individuals are recruited to the fishery. Each fishing pattern is scaled with a fishing intensity, F₀. As total fishing mortality at size is a product of selectivity and fishing intensity, comparison of the absolute fishing mortality, F, between the patterns is complicated and should be approached cautiously. To facilitate qualitative comparison, we have nevertheless scaled the fishing intensities such that the fishing mortality of an arbitrary ‘medium’-size, fully mature fish weighing w_m = 444 g is the same for all exploitation patterns (see dotted lines in figure 2).

Figure 2.

The four different exploitation patterns used in the model with F₀ = 0.8 yr⁻¹ at a size of 444 g (dotted line). (a,b) The unbalanced exploitation patterns for (a) selective fishing and (b) unselective fishing. (c,d) The two balanced fishing scenarios, for (c) selective and (d) unselective fishing. For all four exploitation patterns, a lower cut-off is imposed at 10 g.

We describe the state of the system using three indicators: total yield, spawning stock biomass relative to the unexploited situation and size composition of the yield. The yield in the system is calculated as, where w_r is the body size at recruitment to the fishery, and the sum runs over all species in the model. We calculate yield and ecosystem state when fishing intensity is gradually increased from 0 to 3 yr⁻¹ for each of the four exploitation patterns.

3. Results

The largest MSY from the community is generated at a fishing intensity around 1.0 in the unselective balanced fishing scenario, and is approximately 40% higher than the MSY produced by selective balanced harvesting, but only marginally larger than that of the two remaining scenarios (figure 3). These results are fairly robust to change in the lower cut-off size of fishing (see electronic supplementary material). At low fishing mortalities, all exploitation patterns cause trophic cascades owing to the decline in spawning stock biomass of the largest species (figure 4). The trophic cascades make species with an asymptotic weight of approximately 1 kg reach higher spawning stock biomasses than in the unexploited situation, while the largest species collapse (defined as a spawning stock biomass less than 20% of its unexploited value) already at relatively low fishing mortalities, except in the selective balanced fishing scenario where smaller species crash. The community structure generally responds differently to the selective (figure 4a,c) and unselective scenarios (figure 4b,d). Selective fisheries reduce the spawning stock biomass of fish with an asymptotic size around 200 g, most pronounced in the selective balanced harvesting case. For the unselective fisheries, this is less evident. In spite of the amount of species collapsing at the approximate same fishing intensity in all scenarios, further increases in fishing intensity generate a more gradual reduction in biomass in the selective scenarios than in the unselective. The unselective scenarios furthermore exhibit a ‘fishing down the food web’ type of response, where increasing overall fishing mortality gradually removes species from the large end of the size spectrum.

Figure 3.

The yield from the four different exploitation patterns. Grey (black) lines are selective (unselective) fishing. Dashed (solid) lines indicate that the fishing is balanced (unbalanced; figure 1). The horizontal dotted lines are 50% and 80% of the maximum yield of the unselected balanced exploitation pattern. The highest total yield is from unselective balanced fishing. The unbalanced fisheries have approximately equal yields, both peaking at high fishing intensities. Selective balanced fishing gives the lowest yield.

Figure 4.

Spawning stock biomass relative to the unfished spawning stock biomass on log₁₀ scale. White lines show 20% of unfished spawning stock biomass. (a) Selective unbalanced fishing removes the largest spawners at low fishing mortality, causing their main prey (asymptotic weight range ) to be less impacted by fishing. At high fishing intensities, only the smallest species remain. (b) Unselective unbalanced fishing fishes down the entire food web as fishing intensity increases. (c) Selective balanced fishing preserves community structure up to about F₀ ≈ 1 yr⁻¹, where the largest species are removed, causing a cascade through the community; the smallest individuals have a small refuge due to the mesh size used. (d) Unselective balanced fishing also fishes down the food web, but with a sharper decline in abundance of large species than the unbalanced fishery.

Considering the relative change in the spawning stock biomass for different species, unselective balanced fishing causes the least change from the unexploited situation at exploitation levels corresponding to 50% and 80% of the largest MSY (figure 5). For the other three types of fisheries, the spawning stock biomasses of the largest species are reduced more. Interestingly, the unselective fisheries do not produce lower total spawning stock biomasses (i.e. spawning stock biomass of all species) than the selective fisheries, even though juveniles are fished. The selective balanced fishery generates very low abundances for some groups at 50% of the largest MSY, and cannot produce 80% of the maximum yield. At the high exploitation rates, the size structure is truncated by all exploitation patterns, with unselective balanced fishing causing the least truncation.

Figure 5.

Relative spawning stock biomasses at (a) 50% and (b) 80% of the MSY from the unselective balanced exploitation pattern (see dashed lines in figure 3). Grey (black) lines are selective (unselective) fishing. Dashed (solid) lines indicate that the fishing is balanced (unbalanced; figure 1). The unselective unbalanced fishing (solid black) causes the most severe trophic cascades in the community. The other exploitation patterns also cause some cascades, but they are relatively smaller. Unselective balanced fishing (black dashed) is the only fishery that protects the largest species. For all fisheries, fish with an asymptotic weight in the range 500–1000 g have the largest biomass. The selective balanced fishery (grey dashed) does not reach 80% of the maximal yield and does therefore not figure in (b).

The size composition of the yield depends strongly on the exploitation pattern used (figure 6). At a fishing intensity producing half the largest MSY, selective exploitation produces the highest fraction of large individuals in the yield. Unselective unbalanced fishing also produces approximately the same yield over the 10–1000 g body weight range, but for the larger fish, the yield gradually declines with body weight. The yield from the unselective balanced fishing is dominated by individuals weighing between 10 and 100 g, and few large fish are caught. Increasing the total fishing mortality to produce 80% of the maximum yield provides a similar picture, but with a much reduced yield of large individuals. At the higher fishing mortalities producing 80% of the maximum total yield, the unselective fishery includes only small fish in the catch.

Figure 6.

The relative size composition of the yield at (a) 50% and (b) 80% of the MSY from the unselective balanced exploitation pattern. Grey (black) lines are selective (unselective) fishing. Dashed (solid) lines indicate that the fishing is balanced (unbalanced; figure 1). Under moderate fishing pressure (a), the selective unbalanced fishing (solid grey) has the highest fraction of large individuals in the yield (≈1–10 kg), whereas the unselective balanced fishing (dashed black) has the highest fraction of individuals less than 100 g. The unselective unbalanced fishing (solid black) distributes the yield equally over a large size range, but also with few large individuals. The selective balanced fishery (dashed grey) predominantly catches a combination of many small and also some large individuals. At higher exploitation levels (b), the selective unbalanced fishery (solid grey) has the highest fraction of the yield between 100 and 1000 g, but also some individuals at 10 g. The unselective balanced fishing (dashed black) has the same pattern as with lower fishing intensities, with a high proportion of the yield being small fish. The selective balanced fishery does not reach 80% of the maximum yield.

4. Discussion

The simulations show how maximum yield and fish community structure respond to different exploitation patterns. The response is driven by the direct effects of fishing mortality on the exploited populations, and by compensatory changes in growth, recruitment and predation within the community. When fishing reduces the abundance of predators, their prey will suffer less predation mortality. This will, relatively speaking, increase prey abundance and may lead to trophic cascades if the effect spreads across several trophic levels. Such cascades occur in all four scenarios, but are most pronounced when fishing is selective. As the abundance of predators is reduced, the release from predation makes it possible to fish the remaining individuals harder than expected from single-species fish stock assessments, where the fishing mortality producing MSY typically would be around 0.3 yr⁻¹ for larger species [26]. Due mainly to reduction in predation mortality, the MSY is generated at much higher levels of fishing mortality, where a high yield is achieved at the expense of a collapse of the largest species.

Traditional single-species fisheries models often predict that fishing on juveniles should be reduced or eliminated in order to better exploit the growth potential of the individuals and thus increase yield [1,12]. Our results show that selective fishing reduces the MSY from the community and does not preserve the distribution of biomass across species as well as unselective fishing. If the management objectives are to achieve MSY and maintain the unexploited distribution of biomass across species as well as possible, then unselective balanced fishing is more effective than the other fishing strategies (figures 3–5). However, compared with selective unbalanced fishing the yield from unselective balanced fishing is composed of small individuals and species, which may not find as profitable a market as the larger fish caught by selective harvesting (figure 6).

Our results are in line with other recent modelling studies which also predict balanced unselective fishing to generate higher yields than traditional fishing patterns [3,4,27]; however, our analysis reveals that several fishing patterns (including unbalanced fishing) have almost the same maximum yield and that the yield in the balanced case will consist of fairly small fish. In all fishing patterns, low-trophic-level individuals account for a large part of the catch, which explains similarities in total yield. The targeting of small species (‘forage fisheries’) is in conflict with recent publications where it is argued that small fish species should be exploited at levels well below those producing MSY [28–30]. Those results emerge because exploitation of forage fish species not only affects the species themselves, but also affects top predators such as the seabirds and marine mammals predating upon them [29,31].

One aspect of harvesting that we do not address is to which degree fishing creates or amplifies temporal variability in abundance of fish. Law et al. [10] show that balanced harvesting reduces variability in a single-species context. Community-level simulations have shown that selective fishing increases variability [32]; however, this was done with a model that did not resolve life-history diversity. It has since then been shown that excluding life-history diversity creates artificially high temporal variability [33], so these results should be revisited, preferably in the context of balanced harvesting.

(a). Modelling approach

The model contains the most important biological processes and interactions taking place in a fish community and predicts the consequences of changing its trophic structure. However, like many other fish stock models, it relies on a stock–recruitment function with a weak mechanistic foundation. The function is included to stabilize the model and prevent competitive exclusion [34]. Ecosystem models need one or several controlling processes to generate coexistence of a larger number of species. Specific examples are ratio-dependent functional response [35], stock–recruitment relationships [36], prey switching [37] or food webs with more weakly linked components [21,38]. The implemented stock–recruitment relationship is characterized by two things: the slope at the origin that controls the degree of density dependence, and maximum recruitment. The slope at the origin is an emergent property of the model as it is determined by the food-dependent egg production [26]. Maximum recruitment is defined as a declining function of (M11), and the specific functional form is determined from the equilibrium solution of the model [19]. This functional form has been independently verified from food-web model simulations, which do not use a stock–recruitment function [21]. The only free parameter is thus the total recruitment in the system , but the dynamics of the system are fairly insensitive to changes in this parameter [39].

We used metabolic scaling as an indicator of relative productivity in our calculations rather than as a dynamically calculated productivity. Productivity is essentially a rate (dimension = time⁻¹), and specific ways to calculate it have been proposed [25]. Law et al. [10] defined productivity at the individual level by making fishing mortality proportional to the somatic growth rate of an unexploited population. Another measure of individual productivity would be to consider the biomass production rate, which depends on consumption and predation mortality [40]. In both cases, productivity is a dynamic property that is difficult to measure in nature because it depends on the density of predators and prey. Theoretically, the unexploited productivity would obey metabolic scaling rules, which we have applied in our simulations.

Our implementation of the Andersen and Beyer model uses parameter values reflecting the life history of an average fish, without differentiating between fast- and slow-growing species, or between species with high and low reproductive output. It is not suited to predict the consequences of fishing on slow-growing, late-maturing species with large offspring size and low fecundity, such as sharks and rays, which are likely to be particularly sensitive to fishing [41]. The model describes the response of fish communities to perturbations and we expect similar responses if the simulated exploitation patterns were applied to concrete species and systems. Further progress is needed in the parametrization of size-structured models to investigate consequences of balanced harvesting in applied management of marine ecosystems.

5. Conclusion

We have applied a trait- and size-based model to assess the impacts of fishing on fish communities and yield. Our results show that unselective balanced harvesting is more likely to preserve the structure of fish communities than unbalanced fisheries, but with a major reduction in the average size of the fish being caught.

Implementing balanced harvesting in traditional industrial fisheries would require increased targeting of smaller ecosystem components and elimination of current minimum mesh size regulations [42]. The exploitation patterns used in this study reflect fishing mortalities that should be enforced at the management level. The resulting fisheries could, however, turn out to be less profitable in markets where large fish often return a higher price per kilogram than small fish. In many richer countries, the demand for small fish for human consumption is low. In these countries, balanced harvesting may result in a change towards industrial fisheries for production of fish meal and oil or for providing feed to the aquaculture industry, at the same time as the yield of wild caught high value fish for human consumption is reduced. In developing countries dominated by artisanal fisheries, and where consumers are willing to buy and use small fish for human consumption, balanced harvesting may be easier to implement [14].

Although balanced harvesting is an interesting concept, we agree with Maxwell et al. [43] when they conclude that its practical implementation and ecological and socio-economic consequences need to be further studied before it can be used as a general principle to guide the rational exploitation of fish communities in the context of ecosystem-based management.

Funding statement

N.S.J. and K.H.A. are supported through the VKR Centre of Excellence: Ocean Life. The research leading to these results has received funding from the European Community's Seventh Framework Programme (FP7/2007–2013) under grant agreements MYFISH no. 289257 (see Article II.30. of the Grant Agreement) and VECTORS no. 266445.