1 Introduction
1.1 Fisheries Assessment and Management
One of the fundamental problems within fisheries management is that it is difficult to measure the status of different harvested stocks directly because surveys are expensive and, despite the cost, remain uncertain. Fortunately, it is generally possible to infer their status from samples, or models fitted to data based upon samples, although these also only provide an uncertain view of a stock. Happily, developing longer time-series of fishery observations (such as catches, catch rates, age- or size-structure data, and many other types of observation) can improve our understanding and modelling of events (assuming the quality and coverage of such data is good enough - a very big assumption). Nevertheless, there always remains a degree of uncertainty in any stock or fishery assessment. Such issues are of even greater importance in the many data-poor or data-limited fisheries and species globally (Vasconcellus and Cochrane, 2005; Pikitch et al, 2012). Despite the prevalence of uncertainty, fishery managers are still required to make decisions. This uncertainty, and the consequent difficulties it leads to, have not always been recognized (Smith, 1988) though now the most effective fisheries management jurisdictions attempt to account for uncertainty in explicit ways.
In the 19th century (and into the early 20th) many people believed that the exploitation of natural resources did not require management. This idea, which should now hopefully seem strange, has fortunately evolved into an acceptance that management of wild fisheries is essential, and a wide array of management approaches are now being used around the World (Smith, 1988; Hilborn, 2012). The objectives which systems of fisheries management attempt to achieve have also greatly changed through time. When declines in large fisheries were first identified at the end of the 19th century the focus mainly involved a combination of wanting to maintain catch rates (so as to fish economically) and to maximize the yields from different fisheries (Garstang, 1900). At that time the primary objective was to maximize yield, but it took some years before it was recognized that for many species applying more fishing effort did not necessarily lead to increased catches (the yield-per-recruit problem; Russell, 1931, Beverton & Holt, 1957). It is difficult now to grasp the limited and simplistic view of how fisheries ought to be managed that existed in the 1910s, and extended even up to the 1960s. Larger scale attention only began to be paid to fisheries dynamics and management after the late 1950s, with real progress only commencing in the 1980s onwards (Fournier and Archibald, 1982; Methot, 1989, 1990).
Prior to the late 1950s most thought was given to increasing catches and the efficiency of fishing gear and it still seemed contrary to intuition to recommend limiting catches. For example, at the second FAO conference in 1946, immediately following the second world war, the FAO was strongly urging the development of fisheries as a source of protein and food: “The fishing grounds of the world are teeming with fish of all kinds. Fisheries are an international resource. In underdeveloped areas especially, the harvest awaits the reaper.” (FAO, 1985). The consequences of uninhibited fishing were poorly conceived at that time and attempts to correct the outcomes of such misconceptions from that time are on-going.
Early deterministic stock assessment approaches effectively ignored uncertainty and tended to produce management advice based on the assumption that natural populations are in equilibrium with each other and with any fishing effort imposed on them (Schaefer, 1954, 1957; Gulland, 1965; Megrey, 1989). Assumptions of equilibrium and stability are clearly only an approximation and are invalid in many cases but nevertheless this approach led to concepts such as the Maximum Sustainable Yield (MSY), which related to catch levels, and \(F_{max}\), the fishing mortality which related to the effort expected to lead to the maximum yield. Unfortunately for many fisheries, catches relating to \(F_{max}\) could be larger than the MSY. Both these concepts were early fisheries targets or objectives, with fisheries legislation in many countries still including MSY as a primary aim of management. Sadly, the same legislation often neglects to define the concept of MSY (although this can be interpreted as an advantage). In the 1970s it became apparent, following the collapse of a number of fish stocks, that MSY, as it was then interpreted, was not the safest objective to adopt (Larkin, 1977) and more serious efforts were made to find safer alternatives.
Although the concept of MSY is still invoked it has evolved into use as an upper limit to fishing mortality or has been redefined to account for risks of alternative catch levels (Smith and Punt, 2001). In the 1970s and early 1980s, input controls relating to effort, gear, vessel numbers, and closed seasons were the management tools in most fisheries and some of the more successful management objectives focussed on defining an optimum fishing mortality rate. This work led to the concept of \(F_{0.1}\), which despite being ad hoc, was an advance over \(F_{max}\) in terms of sustainability as well as profitability. It usually led to a large reduction in fishing effort (reduction in fishing mortality and costs) but only led to a minor loss in yield (see Hilborn & Walters, 1992, for definitions of such classical fishery objectives). Even though this was an improvement over \(F_{max}\) or \(F_{msy}\) it was still based on the notion that fish stocks were able to achieve equilibrium with the fishing mortality imposed on them. While this was then assumed to be, at best, an approximation there was still a great deal of development needed to produce the methodologies required for taking uncertainty into account.
The importance of acting to provide management advice in the face of uncertainty was a growing theme in fisheries resource management through the late 1980s and early 1990s. The need to act before scientific consensus could be achieved rather than calling for more research was identified as a key problem for management (Ludwig et al., 1993). The precautionary approach in fisheries is based upon the notion that a lack of scientific certainty about the risk of serious environmental damage must not be used as an excuse for not acting to prevent that damage (FAO, 1995, 1996, 1997).
1.2 Formal Recognition of Harvest Strategies
As stock assessments became more sophisticated so were the management options that were developed. In the late 1980s and early 1990s the effects of variability, uncertainty, and associated risks began to be addressed in stock assessments (Francis, 1992) and the notion of presenting a decision table of management options with their associated risks was also developed. Hilborn & Walters (1992, p453) defined a harvest strategy as:
“…a plan stating how the catch taken from a stock will be adjusted from year-to-year depending upon the size of the stock, the economic or social conditions of the fishery, conditions of other stocks, and perhaps the state of uncertainty regarding biological knowledge of the stock.”
The harvest strategies discussed at that time revolved mainly around the classical three: ‘constant catch’ (e.g. TACs; output controls), ‘constant fishing mortality’ (e.g. \(F_{0.1}\); input controls), and ‘constant escapement’ (e.g. always leaving at least 75% of estimated Mackerel Icefish biomass in the Heard and McDonald Island fishery; mixed input and output controls).
Harvest strategies in the early 1990s focused mainly on setting out fishery objectives (defining biological reference points; Smith et al., 1993) and what constraints should be used. In more recent parlance, this was about determining how to assess each stock’s status and what limit and target reference points to put in place. These developments may have been encouraged, at least in part, by new legislation in the USA (the Magnuson & Stevens Act, 1976, 2007) that required definitions of overfishing that would explicitly guard against recruitment overfishing (Mace & Sissenwine, 1993).
A number of very influential documents were published by the FAO in the mid-1990s, including: the Code of Conduct for Responsible Fisheries (FAO, 1995), the Precautionary Approach to Capture Fisheries (FAO, 1996), and Fisheries Management (FAO, 1997); these latter two documents being parts of the Technical Guidelines for Responsible Fisheries series. The authors stated: “Long term management objectives should be translated into management actions, formulated as a fishery management plan or other management framework” (FAO, 1995, p 11). Giving more details, the Guidelines appear to be one of the first documents to describe the components of what are now termed Harvest Strategies.
The ‘Guidelines’ (FAO, 1996) identified the needs for:
targets, described as the desired outcomes for a fishery,
operational constraints or limits, described as the undesirable outcomes that are to be avoided, and
control rules which specify in advance what action should be taken when specified deviations from the operational targets and limits are observed.
Early work on simulation testing of management arrangements (now known as management strategy or management procedure evaluation) appears to have contributed to this approach to describing harvest or management strategies. Thus, in the FAO Guidelines it defines a management procedure as a description of the data to collect, how to analyze it, and how the analysis translates into actions. This is a standard way to describe a modern harvest strategy: define the data needed, the analysis of status relative to the target and limit reference points, and the control rules used to generate management advice from that status. However, in the FAO guidelines the emphasis given to management procedures was placed on the investigation of how uncertainties influenced the management process (which stemmed from how these management procedures were implemented in South Africa; Butterworth & Bergh, 1993).
The main difference to fisheries management brought about by the adoption of formal harvest strategies was the inclusion of explicit ‘decision rules’ or ‘harvest control rules’. Prior to the introduction of harvest strategies, the data required for stock assessments was certainly collected and the primary thrust of research was the development and articulation of improved stock assessment methodology. Unfortunately, what to do with those assessments to generate management advice sometimes varied from vague to completely unclear. With the addition of formal control rules, management responses become predetermined based on the outcome of the assessment. The use of a formal harvest strategy in a fishery represents a major change to management and constitutes the primary basis for improving the consistency or repeatability, predictability, and transparency of assessment and management.
1.3 Properties of Harvest Strategies
An advantage of using formal harvest strategies is that the outcome of their application should not be dependent upon who applies them to data from a fishery. Given the same data collected from a fishery, anyone who applies a given harvest strategy should produce exactly the same management advice. In order for this to be the case requires harvest strategies, and their implementation, to have the following properties:
1.3.1 Transparent
The information requirements of each harvest strategy and the objectives being aimed for should be fully documented and publicly available. Assessments invariably use summarized data so confidentiality should not become an issue. Where confidentiality might become an issue (e.g. cpue standardization should use raw data from individual fishers), then formal external, yet confidential, review can avoid this issue. Ideally, if software is used in the application of the harvest strategy that too should be publicly available. Being open to critical review is essential for there to be trust that the management being recommended is free from potentially conflicted outside influence. All reviews take time, effort, and usually costs, so, in potentially contentious cases, care should be taken to ensure that a review is not introduced to delay decisions. One objective of every review should be to assess and maintain transparency.
1.3.2 Repeatable
Given the same information and an understanding of the harvest strategy, anyone should be able to generate the required management advice, and that advice should be the same irrespective of who does the work. It should make no difference to the outcome who is in the room when the work is completed and decisions are made. Transparency is critical to achieving this goal, and the consequent repeatability of the outcomes from a formal harvest strategy should instill greater confidence in the management objectives and management actions.
1.3.3 Adaptable
A harvest strategy should not be seen as a static invention, it must be open to improvements while retaining the properties of being fully and openly documented, and of being repeatable. Fisheries management must have the capacity of learning and improving as understanding concerning each fishery increases. Being open to review means that if an improvement can be suggested, by anyone, then a change could be made. In addition, directional environmental changes that have occurred in a noticeable manner over the last 50+ years mean that if productivity changes with location, then changes in expectations and hence in the objectives, targets, and limits for each fishery are also likely to need changing.
1.3.4 Defensible
The details and effectiveness of each harvest strategy must be defensible. Ideally, a harvest strategy should be simulation tested for effectiveness at meeting their objectives (using management strategy evaluation, MSE). Even in the absence of MSE, a harvest strategy’s performance should be monitored and reviewed regularly. The defensibility of a harvest strategy is also highly dependent on the other three properties. If they are not transparent, repeatable, and adaptable, then they are less defensible.