Unraveling the recruitment problem: A review of environmentally-informed forecasting and management strategy evaluation

Highlights

• Environmental drivers may improve forecasts for species with short pre-recruit survival windows.

• Risks may outweigh benefits for species with long pre-recruit survival windows.

• Simulations and MSEs should be used to improve the robustness of future advice.

• Future research should focus on species-specific spatio-temporal scale processes.

Abstract

Studies describing and hypothesizing the impact of climate change and environmental processes on vital rates of fish stocks are increasing in frequency, and concomitant with that is interest in incorporating these processes in fish stock assessments and forecasting models. Previous research suggests that including environmental drivers of fish recruitment in forecasting is of limited value, concluding that forecasting improvements are minimal while potential spurious relationships were sufficient to advise against inclusion. This review evaluates progress in implementing environmental factors in stock-recruitment projections and Management Strategy Evaluations (MSEs), from the year 2000 through 2017, by reviewing studies that incorporate environmental processes into recruitment forecasting, full-cycle MSEs, or simulations investigating harvest control rules. The only successes identified were for species with a short pre-recruit survival window (e.g., opportunistic life-history strategy), where the abbreviated life-span made it easier to identify one or a limited set of key drivers that directly impact dynamics. Autoregressive methods appeared to perform as well, if not better, for species with a longer pre-recruit survival window (e.g., seasonal, inter-annual) during which the environment could potentially exert influence. This review suggests that the inclusion of environmental drivers into assessments and forecasting is most likely to be successful for species with short pre-recruit survival windows (e.g., squid, sardine) and for those that have bottlenecks in their life history during which the environment can exert a well-defined pressure (e.g., anadromous fishes, those reliant on nursery areas). The effects of environment may be more complicated and variable for species with a longer pre-recruit survival window, reducing the ability to quantify environment-recruitment relationships. Species with more complex early life histories and longer pre-recruit survival windows would benefit from future research that focuses on relevant species-specific spatio-temporal scales to improve mechanistic understanding of abiotic-biotic interactions.

Introduction

Interest in incorporating environmental processes in fish stock assessments and forecasting models is increasing in step with studies describing and hypothesizing impacts of climate change and environmental drivers on vital rates of fishes. However, much of the research on the importance of environmental factors in the design of management procedures found that including environmental indices in recruitment forecasting models results in little or no improvement with respect to fishery management performance. This lack of improvement is largely based on research related to gadoid-like life histories. More specifically, since 1999, thirty eight studies have cited Basson (1999), which we consider to be a benchmark analysis on the topic of using environmental factors for short-term recruitment predictions for a gadoid-like life history. This incongruity makes it difficult to justify calls for increased complexity that potentially increase risk without increasing benefits.

In this study we highlight the simulation study of Basson (1999) and then trace developments in the two decades since its publication. Basson (1999) found no benefits to either stock conservation or average yield from including an environmental factor in a stock assessment model. Furthermore, reductions in fishery management uncertainty took place only given a strong environment-recruitment relationship. Basson (1999) recommended that future work explicitly incorporate the assessment procedure within the simulation framework and that the stock-recruit relationship be refitted during every assessment, both with and without environmental factors. A more general recommendation was to explore additional life histories to determine which may benefit from inclusion of environmental covariates in stock assessment models.

The implicit nature of the relationship assumed between the environmental driver and recruitment, and the linkage with the harvest strategy, may have contributed to the negative result in Basson’s study. For example, Basson (1999) modeled a gadoid-like periodic life history, based largely on the strong negative relationship between Irish Sea cod, Gadus morhua (ICES sub-area 7a; ICES, 1999), age-0 recruitment and sea-surface temperature (SST; Planque and Fox, 1998). Drinkwater (2005) also predicted a reduction in Irish Sea cod recruitment with increased SST with climate change. Planque and Fox (1998) hypothesized several mechanisms to explain how warmer water might lower recruitment, such as decreased fecundity, negative effects on egg or larval physiology, or food availability. The most important effect of the gadoid life history was likely on the simulated management strategy, where higher predicted temperatures, implying lower recruitment, dictated a lower fishing mortality (F) reference point and subsequent total allowable catch (TAC) for that year. Recruits (age-0) were not available to the fishery until age one, so setting a lower TAC during a year of poorer environmental conditions would not have an immediate effect on fishing mortality on recruits from that year. However, temperature was similar in consecutive years because of modelled autocorrelation in the time series, so the TAC set during the next year would also be relatively low, relieving some pressure on the previous year’s recruits. The overall effect of the management tactic on Basson’s performance measures, yield and spawning stock biomass (SSB), was delayed and indirect because age-1 fish made up a small proportion of the catch and were not fully mature until age 3. These management tactics might be more effective for stocks that matured earlier, had higher natural mortality, or grew more quickly, with more of the spawning biomass at younger ages. An alternative management approach would be to delay the implementation of stricter TACs to more closely match the timing of the entry into the fishery for cohorts spawned during years of poor environment. This approach might be more effective at lower stock sizes and at a steeper part of the stock-recruit relationship because reduced mortality on spawning age classes should have a stronger positive effect on recruitment. Basson (1999) does not report the parameter values for the stock-recruit curve used, but when analyzing the same data, Planque and Fox (1998) found recruitment to be essentially independent of SSB.

In reviewing the work that has built upon Basson (1999), we use the term ‘recruitment’ to designate the single age class at the end of the interval where population dynamics are governed by a stock-recruit function. In this case, the stock-recruit function generally models the early-stage mortality prior to recruitment as being density-dependent, whereas the mortality rates of individuals older than the recruitment age are typically modeled as being density-independent. We distinguish this use of ‘recruitment’ from earlier definitions where it was defined as the age when fish become fully vulnerable to the fishery (Ricker, 1954, 1975). In contemporary stock assessment models, vulnerability to fishing gear at all ages can be directly modeled as part of the assessment, as either a function of size or age, and potential temporal changes in vulnerability can be accommodated as well (Quinn and Deriso, 1999). Thus, the recruitment definition that we adopt places the emphasis on processes that exert influence on survival between spawning and the resultant recruits, rather than focusing on older, fully vulnerable ages where survival is less likely to be influenced by the environment and more likely to vary with fishing pressure.

Since Basson (1999), the inclusion of environmental variables in fisheries assessments remains rare. A recent examination of 1250 assessments worldwide identified only 24 that included ecosystem information regarding stock productivity, of which 14 were environmental factors (Skern-Mauritzen et al., 2016). However, a review with a wider scope, which summarized the use of all ecosystem information used in stock assessments in the U.S. rather than just recruitment, found that 24% of all U.S. stock assessments included ecosystem information (Marshall et al., 2018). Since Basson (1999), several studies have made recruitment forecasts (e.g., Ward et al., 2014), some of which include environmental drivers. Most likely, the combination of life-history traits and management tactics will affect the utility of environmentally based recruitment forecasting. We undertook a review of literature from the year 2000 through 2017, to evaluate progress in implementing environmental factors in Management Strategy Evaluations (MSEs) and stock-recruitment forecasts. We identified over sixty peer-reviewed studies that address questions of environmentally-driven recruitment forecasting and related MSEs. Just over one-third of these studies were found to be relevant during initial screening, where relevance was determined as a study going beyond mere correlation to model an environmental driver explicitly or implicitly and the stock assessment and/or management implications were considered. The relevant studies we consider below are broadly categorized as either primarily focused on recruitment forecasting or else on simulations of full-cycle MSEs or harvest control rules (HCRs; Table 1).

The goals of this review are 1) to examine successes and challenges across studies that include environmental considerations in MSEs or stock recruitment forecasts and 2) to highlight study characteristics that result in different outcomes. This review is loosely structured by topics related to how the relevant studies address or fulfill issues raised in, or recommended by, Basson (1999). Basson’s main recommendations included: 1) the consideration of a wide range of life-history types, exploitation patterns, and stock-recruitment relationships, 2) identification of strong interactions between environmental drivers and recruitment, 3) increased prediction skill for environmental factors, 4) increased realism with respect to management actions and implementation, and 5) the consideration of long-term trends in environmental time series. Section 2 reviews studies across a wide range of different stock-specific life histories or general scenarios. Section 3 reviews the need for simulation and forecasting studies with more realistic assessment, harvest, and management implementation modules. Section 4 summarizes the use of a) either univariate or multivariate mechanistic drivers, b) historical data, c) spatial scale, and d) temporal forecast scale. Finally, sections 5,6, and 7 provide a synthesis of successes, challenges, and future research recommendations to improve relevance to fisheries management.