© 2004 by ICES/CIEM International Council for the Exploration of the Sea/Conseil International pour l'Exploration de la Mer
Management options for the Blackwater herring, a local spring-spawning stock in the Thames Estuary
CEFAS Lowestoft Laboratory, Pakefield Road, Lowestoft Suffolk NR33 0HT, England, United Kingdom
*Correspondence to B. A. Roel: tel: +44-1502-524358; fax: +44-1502-524511. e-mail: b.a.roel{at}cefas.co.uk.
Herring caught in the Thames Estuary sustain a small local commercial fishery (peak catch of 606 t during the 1972–1973 fishing season). Loss of local consumers' interest in the herring product has resulted in a gradual decline in catches and fishing effort for the stock. The stock is assessed using an age-structured model that relies on the information provided by a scientific trawl survey, and management advice is provided before the fishing season starts in October. Given its current low economic value, managers have requested evaluation of options for multi-annual Total Allowable Catches (TACs) in an attempt to reduce the frequency (and costs) of assessment and associated management advice. A tentative relationship between sea surface temperature and recruitment is used to predict the impact of increasing sea temperatures on future recruitment in the context of global warming. Hypotheses of auto-correlation and of an environmental effect on recruitment, together with trends in weight-at-age and the overestimation of spawning-stock biomass, form the basis for sensitivity tests of the management options considered. Implementation of a 3-year fixed TAC with 40% constraint in TAC variability and a slight reduction in target F would seem appropriate for the stock, given that it is within safe biological limits and compares well in terms of yield and risk with the current approach of annual TAC revision.
Keywords: age-based methods, fishery management, multi-annual TAC, recruitment, stock assessment
Received 21 July 2003; accepted 2 February 2004.
 |
1 Introduction |
|---|
 Top 1 Introduction 2 Methods3 Results3 DiscussionAppendixReferences |
|---|
Herring (Clupea harengus) have been caught in the Thames Estuary (Figure 1) for many years, but the so-called Blackwater stock, also referred to as Thames Estuary herring, was not recognized as separate from the North Sea stock until the early 1800s (Wood, 1981). Bolster (1954) first showed that Blackwater herring were spring spawners with "low vertebral counts" (mean 54.73). Mature fish were small compared with North Sea herring and, therefore, their catches could not compete successfully in the local markets. Only when the East Anglian herring fishery began to collapse in 1955 and there was a shortage of North Sea herring could Blackwater herring be sold without any difficulty. Fishing capacity grew and, by 1968, some 22 trawlers and driftnet vessels were targeting the stock; catches peaked at 606 t in the 1972–1973 season. A series of poor year classes followed and, as the stock declined, catch restrictions were implemented. However, the stock continued to decline, and eventually the fishery was closed during winter 1979. The fishery was reopened in winter 1980 after a research survey identified clear signs of stock recovery. At the start of the 1988–1989 fishing season, a redefined area for a licensed, driftnet-only herring fishery was introduced in the Thames Estuary. Herring caught inside this area were considered to belong to the Blackwater spring-spawning stock, and landings were monitored so as not to exceed the annual Total Allowable Catch (TAC). Herring caught outside the exclusive driftnet area are counted against the UK southern North Sea allocation of quota, and that fishery is subject to regulations set by the European Union pertaining to the combined ICES Division IVc and VIId. Hence, if the annual TAC is fully taken up in the driftnet area, any additional catches of Thames spring spawners taken outside the driftnet area will have adverse effects on stock biomass and on the fishing mortality. Although all catches taken in the Thames Estuary are taken into account in the assessment, there is still potential for overshooting the TAC and this has been flagged as a problem in managing this stock. Results from tagging experiments suggest that any fishing mortality generated outside the Thames Estuary is insignificant (Wood, 1981).
|
In recent years, the fishery has been regulated by an annual TAC based on the results from an age-structured stock assessment model (Extended Survivors Analysis, XSA; Darby and Flatman, 1994; Shepherd, 1999). The model settings are adjusted each year for the assessment performed in summer, before the fishing season starts in October. In recent years, the stock is reassessed in February, incorporating the results from an annual midwater trawl survey in November. The XSA settings of the previous assessment performed in summer are maintained. The assessment relies upon information provided by a scientific survey, because fishing effort has declined substantially, and consequently the commercial catch and effort data are sporadic. Reasons for the decline are related to the low value of the product because of the small size of Blackwater herring compared with North Sea herring, and because of changes in local consumer preference. The stock has recently been accredited by the Marine Stewardship Council (Project no. 6259GB), but this recognition has not raised the interest of consumers in the product. Given its low economic value, it would appear sensible to reduce the costs of management and assessment, and managers have requested that the consequences for the resource of implementing fixed TAC strategies be evaluated. This paper addresses this request.
We provide here a description of the simulation framework used to evaluate the impact on the stock of fixed TAC strategies as well as other candidate management options. Robustness of the results is tested by applying the management options under alternative conditions regarding population dynamics and degrees of compliance. Finally, we discuss the performance of the strategies considered, taking into account conservation and utilization criteria.
|
2 Methods |
|---|
|
Top1 Introduction2 Methods3 Results3 DiscussionAppendixReferences |
|---|
The management options were evaluated by means of a simulation framework that incorporates uncertainties in initial population numbers, weight-at-age, and future recruitment and includes bias in the stock assessment model, based on forward projections of the stock numbers-at-age as estimated in February 2002 (Figure 2). The simulation framework consists of three main components: (1) an operating model representing the stock dynamics and the assessment process, (2) management options to be investigated, and (3) a selection of performance statistics.
|
2.1 Operating model
Stock numbers-at-age are projected forward, given future catches provided by the management option being tested and parameters such as natural mortality-at-age and maturity ogive, as adopted in the XSA routine assessment. Uncertainty in natural mortality and maturity-at-age is not taken into account in the current framework. The projection period is 20 years. Although ICES (2002) has indicated that 10 years was the longest period for which projections are sensible, this is not long enough to fully evaluate 5 years of fixed TAC strategies. Furthermore, initial conditions in 10-year projections are likely to be too influential. Starting numbers for projections are sampled from a lognormal distribution, with mean equal to the estimated stock numbers-at-age at the start of the season in 2001 and variance based on the XSA standard errors (s.e.) of the population estimates. In subsequent years, recruitment at age 1 is generated from a Ricker stock-recruitment (S-R) model based on estimates of spawning-stock biomass (SSB) and recruitment (R) at 1 year of age for the period 1962–2000. More details are presented in the Appendix.
In the framework, the assessment is simulated, rather than actually performed, by introducing bias and uncertainty in the numbers-at-age generated by the operating model. A positive bias has been present in consecutive estimates of population numbers-at-age for this stock. A retrospective bias of 3.4% was estimated by applying the measure of Jónsson and Hjörleifsson (2000) to the SSB estimates from the annual assessments performed in the years 1998–2002. For simplicity, and to be on the conservative side, a 5% positive bias was introduced in the perceived numbers-at-age. In the absence of a full simulation of the assessment, it is difficult to decide upon a realistic level of uncertainty in the perceived population numbers-at-age. A coefficient of variation (CV) of 30% in the numbers-at-age has been assumed here based on typical measures of uncertainty in the estimates of SSB and fishing mortality F (ICES, 1998). This value is likely to capture the uncertainty resulting from fitting the model to data, but observation errors such as those arising from landing statistics, insufficient biological sampling, and errors in the survey-tuning index, may not be accounted for. Based on the simulated assessment, the perceived numbers-at-age at the start of year y are given by
|
| (1) |
y,a is a normally distributed random variable with mean zero and 
=0.3 is a measure of uncertainty in the population numbers-at-age [denoted y,a
N(0,2)] and k=1.05 introduces positive bias. Estimates of stock and recruitment since 1962 suggest that R may be impaired when SSB is below approximately 200 t (Figure 3a). However, this interpretation is strongly influenced by three high recruitments in the 1960s, and conditions may have changed since then. In fact, if only data from the more recent period are considered, R appears to be reduced at values of SSB below 300 t (cf Figure 3b). A value of 250 t, a compromise between these two extremes, was identified in the management advice for the 1998/1999 fishing season as Blim, indicating drastic management action if SSB was estimated to be below this point.
|
The interpretation of the precautionary spawning biomass reference point (Bpa) is that if SSB is estimated between Bpa and Blim, F should be reduced to ensure that SSB recovers to a level above Bpa in the short or medium term. Following guidelines provided by ICES (1998), the precautionary biomass Bpa was derived from Blim as follows:
|
| (2) |
=0.3 is a measure of uncertainty in the estimate of SSB, resulting in a Bpa equal to 410 t. Based on an empirical distribution of the maximum of the fitted Ricker S-R curve (Table 1), the SSB below which R is impaired, is 275 t. The adopted Blim for the stock (250 t) corresponds to the first quartile (Q1) of the empirical distribution of that maximum, while Bpa (410 t) is situated above the 99th percentile.
|
If the maximum is estimated on the basis of the period 1970–1999, eliminating the three very high recruitments in 1963, 1964, and 1968, the estimate of the Ricker maximum is increased to 322 t. The Blim corresponds to the 5th percentile, while Bpa is now just above the 90th percentile (Table 1). A 10% risk of reaching a point of impaired R is considered acceptable, so the adopted Blim of 250 t, midway between the estimates for the two periods, seems appropriate. Similarly, the existing Bpa of 410 t would be compatible with a 95% upper confidence level.
2.2 Scenarios
A number of scenarios related to the stock dynamics, environmental effects on R levels, and compliance were formulated. The base case represents the most likely scenario or the one that corresponds to historic conditions. The remaining scenarios were formulated by replacing a condition in the base case by an alternative but also plausible one. These scenarios form the basis for sensitivity tests to evaluate the performance of the management options if conditions depart from those assumed in the base case. An outline of the scenarios and the conditions that characterize them is presented in Table 2.
|
2.2.1 SSB in April 2002
Results are presented for two SSB scenarios in April 2002 (cf Figure 2), the spawning month in the first year of the projections, both assuming that the 105 t TAC was caught during the 2001–2002 fishing season: (1) SSB = 711 t, significantly above Bpa (based on stock projections from the revised February 2002 assessment, which incorporates the results from the November 2001 survey), and (2) SSB = 501 t, just above Bpa (based on stock projections from the September 2001 assessment). Scenario (1) was assigned to the base case because, although results from the routine summer assessment are not yet available, they are expected to either confirm the results from February or to give a more optimistic picture, given that only 53% of the Blackwater herring TAC was taken in the 2001–2002 fishing season.
2.2.2 Weight-at-age
Results are presented for two scenarios: (1) no trend in weight-at-age, and (2) a 1% decrease in weight-at-age per annum, with a lower bound of the minimum weight observed historically. Examination of the weight-at-age trends suggests that we may be at a historic minimum at present and that future trends are difficult to anticipate. Therefore, a scenario of constant weight-at-age equal to the average of the last 3 years was chosen for the base case.
2.2.3 Recruitment auto-correlation
The 1962–2000 data set of SSB and R values was split into two periods: 1962–1980, and 1981–2000. Both sets were fitted to a Ricker S-R model (Figure 3). The estimated auto-correlation in the residuals from the fits was 0.014 for the total data set, 0.13 for 1962–1980, and –0.20 for 1981–2000. Recruitment in the period 1962–1980 had a positive serial correlation, because both SSB and R were declining (Figure 3a). In the recent period (Figure 3b), when management aimed at trying to keep SSB above Bpa, SSB and R are fluctuating about the declining part of the curve, and hence are negatively correlated. The figures show two fitted Ricker curves, one derived from the fit to the total data set (1962–2000; continuous line), and one to data from each period. The curves practically superimpose and, because in future this stock is to be managed by keeping it above Bpa, we can use either curve and assume negative auto-correlation for the base case and the case of no auto-correlation as an alternative scenario.
2.2.4 Recruitment and sea surface temperature
A time-series of annual mean sea surface temperature (SST) for the first six months of the year was calculated from monthly means for 2° squares centred on 51–53°N and 1–3°E, derived from the Comprehensive Ocean Atmosphere Data Set (COADS) and provided by the National Center for Atmospheric Research (Boulder, Colorado). The six-month period was chosen by analogy with the Norwegian spring-spawning herring, whose year-class strength appears determined during the first six months of life (Toresen, 2001). Preliminary analysis, fitting a loess smoother (Cleveland and Devlin, 1988) to the SST and the R data for 1962–2000, suggested opposite trends in the two time-series (Figure 4).
|
There is a significant and positive correlation between R and SST for the Norwegian spring-spawning herring (Toresen, 2001). Southward et al. (1988) demonstrated that the abundance of herring and pilchard (Sardina pilchardus) off southwest England closely corresponded with fluctuations in water temperature. Pilchard were generally more abundant and extended farther east when climate was warmer, whereas herring were generally more abundant in cooler times. However, recruitment of the other herring stocks examined by Toresen (2001), including North Sea autumn-spawning herring, was not correlated with temperature.
ICES (2000, 2001, 2002b) investigated the implementation of S-R models mediated by hypothesized environmental influences. Following these studies, and supported by the apparent opposite trends of SST and R, temperature was modelled within a parametric function according to the Ricker S-R relationship as:
|
| (3) |
, ß, and 
are model parameters. This model was used in a scenario assuming a 2% increase in SST per annum, based on predictions (ICES, 2002b) related to a scenario of global warming.
2.2.5 TAC compliance
Two scenarios were considered: (1) the TAC is fully taken (base case), and (2) the TAC is either exceeded or fully taken. The possibility of exceeding the TAC was simulated on the basis of the historic ratios of the herring landings in the Thames Estuary (Blackwater and southern North Sea herring caught both inside and outside the regulated area) and the total TAC (the figures are essentially the same if we only consider spring spawners) from 1988 to date (Figure 5). The uptake (Cy) in year y is equal to
|
| (4) |
is a value obtained by sampling with replacement from the historic set of landings/TAC ratios. Where the ratio was <1, it was set equal to 1. This represents a scenario where the TAC is taken in the driftnet area, but catches continue against the southern North Sea herring quota on the basis of what was seen in the past.
|
The driftnet area catches exceeded the TAC only twice (1988 and 1989), and only by relatively small amounts. Driftnet catches remained below the TAC in all other years. From a historic perspective, closing the driftnet area when the TAC was met would rarely have had a conservation effect, and the impact would have been small. Total landings also exceeded the TAC in 1990 and from 1993 to 1997. Therefore, the TAC could only be enforced by closing all the fisheries that exploit Blackwater herring once the TAC is met. Otherwise, the potential for overshooting the TAC will remain. The scenario does not take into account the possibility of undershooting the TAC (as in recent years), and therefore may be considered a worst-case scenario.
2.3 Management options
A number of management options (MOs) were considered taking into account stakeholder preferences, and the features of this particular stock. Given the low economic value of the fishery, a multi-annual TAC regime seems desirable for managers. On the other hand, industry has expressed a wish for some stability in quota, though there is a trade-off between stability and catch level if sustainability is also to be ensured. Furthermore, if the TAC is fixed for a number of years, there may be need to make big adjustments at the time a new TAC is put into place. Industry is likely to be averse to such large changes for understandable economic reasons, so options that constrain interannual variability in TAC need to be explored.
The management options evaluated and their main features are outlined in Table 3.
|
2.3.1 MO1: annual revision
Corresponds to the strategy currently in place. TAC is set annually, based on keeping F=Fpa. Constraint: if predicted SSB in April of the forthcoming season (based on the assessment) SSByp<Bpa, F is then reduced until SSByp
Bpa.
2.3.2 MO2: multi-annual, no constraints (3-year/5-year)
TAC is set based on F=Fpa and remains fixed for a period of either 3 or 5 years. At the end of the period, the stock is reassessed and a new TAC is computed and implemented. Constraint as in MO1.
2.3.3 MO3: multi-annual (10%/20%/40%), additional constraint
TAC as in MO2, but with an additional constraint on the maximum change from one period to the next (10, 20, and 40%). The constraint does not apply when the TAC is zero, which is equivalent to a fishery closure. The zero TAC remains in place for the duration of the period.
2.3.4 MO4: fixed (low/high)
TAC is set at a fixed level of either 60 or 120 t for a 20-year period.
2.4 Performance statistics
In selecting a management option, stakeholders need to decide how best to utilize the stock by maximizing catch and minimizing interannual catch variability while, at the same time, keeping risk levels low. The following performance statistics were computed to provide managers and stakeholders with the tools to make an informed decision between the management options presented:
Risk SSB<Bpa or <Blim: probability of the SSB falling at least once within the simulation period below one of the biomass reference points.
Frequency <Bpa: average over 1000 simulations of the number of times SSB fell below Bpa during the 20-year projection period.
Mean catch: median value over 1000 simulations of the average of 20 years of annual catch.
Mean SSB, end SSB: median values over 1000 simulations of the average of 20 years of SSB, and of the biomass at the end of the 20-year projection period.
Median interannual catch variability: median value over 1000 simulations of the average 20-year interannual catch variability (ICV):
|
| (5) |
Percentage change in target F: in relation to Fpa for a MO to present the same level of risk as that associated with the current annual strategy (MO1).
|
3 Results |
|---|
|
Top1 Introduction2 Methods3 Results3 DiscussionAppendixReferences |
|---|
Comparison between the performance of annual revision MO1 and multi-annual strategies MO2 and MO3 in terms of median values and 90th and 10th percentiles of yields and SSB (base case; Figure 6) suggests that the differences between the annual and the 5-year MOs considered are larger than that between the annual and the 3-year MOs. The relatively small differences in SSB suggest that the MOs considered are generally conservative and that built-in conditions such as reduction in F if SSB<Bpa are effective.
|
Mean yields and risk levels for the base case MOs are compared in Figure 7. The dotted line was generated by applying MO1 for increasing levels of target F, allowing the comparison of the other strategies against it. Comparison in terms of risk of falling below Bpa and mean yield over the period of the projections suggests that the annual strategy is more effective. However, variability in catches between years is highest in the case of the annual strategy (Table 4). A median value of 0.34 for the interannual catch variability doubles the corresponding value for the 3-year multi-annual/no-constraints strategy, which is the second most variable. Besides that, on average, this strategy yields catches similar to the annual one, but the associated risks are substantially higher. This is also manifest in the 90th percentile of the interannual variability, which shows a high value as a result of zero catches in some of the trajectories. Further, an associated high risk of falling below Blim makes the no-constraints strategy probably unacceptable from a conservation point of view.
|
|
The risk of SSB falling below reference points increases as the constraints vary from 10 to 40% (Figure 7; Table 4). This result is counter-intuitive, because bounds on TAC changes often prevent fast reduction in TAC and would result in increasing risk if a stock is heavily exploited and requires rebuilding (Punt, 1997). In this particular case, the TACs at the start of the simulation period are low and the stock is not overexploited. A highly constrained strategy is less risky because it results in a low and practically constant TAC, and comparatively lower catch rate (Figure 8a). Also, the 3-year 40% and no-constraints strategies attain higher yields on average and have lower associated risk than their 5-year counterparts (Figure 7). This is because, for some of the trajectories, maintaining a high and unsustainable TAC during 5 years, as opposed to three, resulted in fishing closures which were then enforced for the full duration of the fixed period (cf MO rules).
|
To allow fair comparison between MOs, mean yields were computed by tuning the target F to attain the risk level of 0.21 associated with MO1. The median, 90th, and 10th percentiles show small differences between the 3-year unbounded and the bounded strategies (Figure 8b). At the current level of risk, expected yields are clearly lower in the case of the 5-year unbounded MO where a 29% reduction in target F is required to keep the risk level the same as in MO1 (Table 4, last column).
The impact on recruitment of including SST in the Ricker S-R model is illustrated in Figure 3c. To put this figure into perspective, historic values of SST ranged between 6.3 and 9.8 °C. The thicker line corresponds to the model fit to R and SSB data only, and the thinner and broken lines incorporate SST values between 7 and 9 °C. In the context of global warming, increasing sea temperatures could cause a reduction in recruitment (Figure 4), thus increasing the risk of falling below reference points.
Results in terms of associated risk for the various management options considered are illustrated in Figure 9 for the base case and four alternative scenarios that constitute variations in the operating model and in the SSB level in the first year of the projections. The current Fpa strategy appears to be slightly less sensitive than the other ones when comparing performance against Bpa. The 60-t fixed TAC strategy also performs well and, although there is 11% risk associated with the scenario of low starting SSB, it only reflects the uncertainty around the starting values in the projections. When comparing risk of falling below Blim, neither the annual revision nor the fixed TAC strategies are sensitive to the alternatives considered. In contrast, the 5-year/no-constraints strategy is particularly sensitive.
|
Results for annual and multi-annual strategies with TAC constraints in the case of TAC overshoot are shown in Table 4b (results from multi-annual scenarios with no constraints are not tabled, but risk levels associated with these scenarios were very high). The overall risk of falling below Bpa is about 50% higher than for the base case. The average catches over a 20-year period are higher than corresponding ones for the base case. However, SSB at the end of the projection period is generally lower, the 10% percentile being below Bpa for the three most variable multi-annual strategies. The fixed TAC MOs considered remain viable. Further simulation work indicated that target F should be reduced to 0.25 to maintain a risk level comparable to the base case.
|
3 Discussion |
|---|
|
Top1 Introduction2 Methods3 Results3 DiscussionAppendixReferences |
|---|
Comparison of the tabulated performance statistics for the strategies considered illustrates the trade-offs between management objectives such as maximizing yield and minimizing risk while keeping interannual variability in catches to a minimum. Conflicting objectives is a common problem in fisheries, as stated by Cochrane et al. (1998) when discussing the merits of management procedures in the South African pelagic fishery, and our results show that Blackwater herring is no exception. The annual revision strategy results in better utilization of the stock: higher catches and lower associated risks than the other options. Interannual variability in catches is high, but imposing a TAC-change constraint could reduce it. However, the time spent in managing the stock on an annual basis is unlikely to be cost-effective. At the other end of the spectrum, the constant catch strategies considered offer stability and create low risk at the expense of average yields. Given the current difficulties in marketing the stock, which have resulted in catches of about 120 t in the past few fishing seasons, a constant catch strategy cannot be ruled out easily. Ultimately, managers will make the decisions by considering the effects of the various annual and multi-annual strategies presented. This could be done on the basis of performance statistics that participants in the fishery can relate to, and are of interest to them (such as average catch, variability, and associated risk for the resource), without having to understand complicated underlying models (De Oliveira et al., 1998).
Here, risk was computed as the probability of the stock falling, at least once, below limit and precautionary reference points. To be cautious, reductions in spawner abundance must be assumed to reduce productivity directly (Rosenberg and Restrepo, 1996). Using the change point (segmented) regression to estimate explicitly the SSB of Blackwater at which R is impaired, O'Brien et al. (2002) arrived at a tentative value of 82.7 t. Also Wood (1981), in a biological study of Blackwater herring, stated that recruitment from a spawning stock below about 80 t had not been measured, but was likely to be very poor. Reference points derived from segmented regression are less model-dependent than those derived from the Ricker S-R model, and show a decline in R only at a relatively low SSB. There is some evidence that supports the use of a Ricker model (Fox, 2001) and, although a large degree of uncertainty was estimated about the SSB level beyond which recruitment overfishing would take place, this uncertainty was taken into account when defining precautionary and limit reference points.
Performance of the multi-annual strategies was evaluated in relation to their associated risk of falling below precautionary reference points. Strategies that could lead to a stock size below the limit reference points were considered unsustainable and, therefore, were ruled out. Another source of uncertainty with respect to the precautionary application of limit reference points for fisheries relates to the likelihood that environmental conditions may change and thus influence productivity (Rosenberg and Restrepo, 1996). For the Blackwater herring, the reference points implicitly take into account the effect of the environment, because different values of SST will only shift the S-R curve up or down, thus maintaining the relative stability of the reference points in relation to SSB, regardless of changes in mean SST. The biological interpretation of this S-R model is that environmental conditions do not affect the spawning potential (number of eggs produced), but rather the survival of eggs to age 1 recruits.
In a review of "risk" in fisheries management, Francis and Shotton (1997) examined types of uncertainty, and some of them are relevant to our study. The framework used takes into account process uncertainty related to natural variability in recruitment and in weight-at-age, uncertainty in some of the model parameters, and implementation uncertainty. However, the model of the dynamics is based only on the results from the XSA stock assessment, which means that there is only one "model" of the past. Therefore, the uncertainty related to lack of information about the correct structure is not taken into account. Parameters such as natural mortality, maturity-at-age, and selectivity-at-age are considered to be known perfectly and sampling error associated with the survey index was not taken into account explicitly. Overlooking structural and parameter uncertainty would lead to a conservative estimate of the total associated risk, and pose problems at the time of assessing the merits of a particular strategy against management objectives expressed in absolute terms, i.e. attaining certain catch levels over a given period of time. Notwithstanding, the approach adopted remains appropriate to assess and to compare the performance of the strategies investigated against each other.
In a study on the development of scientific advice based on incomplete information, Hilborn and Peterman (1996) consider major sources of uncertainty in fish stock assessment. They state that any stock projections must make assumptions about future environmental conditions. We are aware of a large number of research initiatives aimed at understanding and quantifying the impact of the environment on recruitment, particularly in the case of pelagic (e.g. Uriarte et al., 2002) and gadoid stocks (e.g. O'Brien et al., 2000). Toresen (2001) found a positive significant relationship between temperature and recruitment for Norwegian spring-spawning herring. However, the same author states that the large variation in recruitment level observed for various stocks at high temperatures suggests mechanisms other than a direct influence. The mechanisms controlling year-class strength of Blackwater herring are not yet fully understood, although Fox's (2001) results suggest that the overall shape of the S-R curve is determined by density-dependent mechanisms operating after hatching. Basson (1999), in evaluating the importance of environmental factors in the design of management procedures, concluded that in the context of global warming, where predictions may indicate a greater frequency of warm years than observed in the past, it would be interesting to consider ways of incorporating such information into a management procedure. Given forecasts of increasing SST and weak evidence of a negative impact of SST on recruitment, we conclude on the basis of robustness tests that, if this scenario were true, the risks associated with multi-annual strategies would increase seriously. Other scenarios of future environmental change might be considered, such as periodic changes or jumps in conditions. However, estimating the probabilities associated with such scenarios is not a trivial matter (Hilborn and Peterman, 1996).
Based on the analysis presented, the existing annual revision strategy is the most efficient, but it is expensive relative to the commercial value of the stock. A fixed TAC strategy requires little management effort, but needs to be inherently conservative and well monitored. A multi-annual (3- or 5-year) fixed TAC strategy will result in higher catches. However, such a strategy is more risk-prone. Adding additional constraints on maximum TAC change provides a compromise solution. A 3-year fixed TAC with 40% constraint and a slight reduction in target F compares well with the current strategy: predicted TACs are lower, but only by 5%, and the associated risks are similar. Moreover, a multi-annual strategy that limits variability in TAC meets the objective of reducing the frequency of assessment and management advice (and hence reduces the cost of management). This approach could be successful in the case of a stock that is within safe biological limits as long as the management system is dynamic enough to allow a quick response if signs of rapid stock decline are detected by routine annual monitoring, which can then trigger an assessment.
|
Appendix |
|---|
|
Top1 Introduction2 Methods3 Results3 DiscussionAppendixReferences |
|---|
Population abundance (N) at age a+1 in year y+1 is calculated as:
|
| (A1) |
|
| (A2) |
yN(0,r2), r2 represents the log-residual variance from fitting the parametric stock-recruitment function f(SSBy). Because the shape of the stock-recruitment curve for Blackwater herring is determined by density-dependent mechanisms acting at some stage after hatching (Fox, 2001), a Ricker curve was used to model the relationship. Fishing mortality in the year starting in October 2001 was estimated by solving Gulland's catch equation assuming that the uptake for the fishing season was equal to TAC2001–2002 (105t) and that selectivity-at-age was equal to the average for the period 1998–2000.
|
| (A3) |
Examination of annual weight-at-age data from both catches (Figure A1) and stock (not shown) shows a temporal trend for most age groups as well as strong year effects. A loess smoother (Cleveland and Devlin, 1988), with a default span of 0.25, was fitted to the data for ages 2–8 to remove the temporal trend. The residuals from each year were saved as a set and randomly selected values were added to the average weight-at-age for the period 1999–2001 to generate future weight-at-age in both the stock and catches (independently), maintaining the year effect.
|
The TAC was split into numbers-at-age caught according to
|
| (A4) |
|
| (A5) |
|
Acknowledgements |
|---|
A first draft of this paper was presented at the 2002 ICES ASC in Copenhagen, Theme Session on Pelagic Fish Responses to Climate Variability – Consequences for Fisheries and Ecosystem Advice. Thanks are due to the co-conveners, Maria-de-Fatima Borges, Dankert Skagen, Carmela Porteiro, and Brian Rothschild, to Jose de Oliveira and Colin Bannister for helpful comments on the manuscript during its several phases of drafting, to an anonymous reviewer for a perceptive review, to Clive Fox for facilitating our access to SST data, and to Niels Daan for his valuable input to the final version. The study was funded by the Department for Environment, Food and Rural Affairs (Defra), UK.
|
Footnotes |
|---|
1 Current address of M. Basson: CSIRO Division of Marine Research, PO Box 1538, Hobart, Tasmania, 7001, Australia.
|
References |
|---|
|
Top1 Introduction2 Methods3 Results3 DiscussionAppendixReferences |
|---|
-
Basson M. (1999) The importance of environmental factors in the design of management procedures. ICES Journal of Marine Science 56:933–942.
Bolster G.C. (1954) Strange herring on the East Anglian coast. Annales Biologiques (Copenhagen) 10:168–169.
Cleveland W.S. and Devlin S.J. (1988) Locally-weighted regression: an approach to regression analysis by local fitting. Journal of the American Statistical Association 83:596–610.[CrossRef][Web of Science]
Cochrane K.L., Butterworth D.S., De Oliveira J.A.A., Roel B.A. (1998) Management procedures in a fishery based on highly variable stocks and with conflicting objectives: experiences in the South African pelagic fishery. Reviews in Fish Biology and Fisheries 8:177–214.[CrossRef][Web of Science]
Darby C. D. and Flatman S. (1994) Virtual Population Analysis: Version 3.1 (Windows/DOS) User Guide. Information Technology Series, MAFF Directorate of Fisheries Research, Lowestoft, 1: 85 pp.
De Oliveira J.A.A., Butterworth D.S., Roel B.A., Cochrane K.L., Brown J.P. (1998) The application of a management procedure to regulate the directed and bycatch fishery of South African sardine Sardinops sagax. South African Journal of Marine Science 19:449–469.
Fox C.J. (2001) Recent trends in stock-recruitment of Blackwater herring (Clupea harengus L.) in relation to larval production. ICES Journal of Marine Science 58:750–762.
Francis R.I.C.C. and Shotton R. (1997) "Risk" in fisheries management: a review. Canadian Journal of Fisheries and Aquatic Sciences 54:1699–1715.
Hilborn R. and Peterman R.M. (1996) The development of scientific advice with incomplete information in the context of the precautionary approach. FAO Fisheries Technical Paper 350:278–101.
ICES. (1998) Report of the Study Group on the Precautionary Approach to Fisheries Management. ICES CM 1998/ACFM: 10.
ICES. (2000) Report of the Study Group on Incorporation of Process Information into Stock Recruitment Models. ICES CM 2000/C: 01.
ICES. (2001) Report of the Study Group on Incorporation of Process Information into Stock Recruitment Models. ICES CM 2001/C: 02.
ICES. (2002a) Report of the Working Group on Methods on Fish Stock Assessments. ICES CM 2002/D: 01.
ICES. (2002b) Report of the Study Group on Incorporation of Process Information into Stock Recruitment Models. ICES CM 2002/C: 01.
Jónsson S. T. and Hjörleifsson E. (2000) Stock assessment bias and variation analyzed retrospectively and introducing the PA-residual. ICES CM 2000/X: 09.
O'Brien C.M., Fox C.J., Planque B., Casey J. (2000) Climate variability and North Sea cod. Nature 404:142.[Medline]
O'Brien C. M., Maxwell D. L., Roel B. A., Basson M. (2002) A segmented regression approach to the Precautionary Approach—the case of the Thames Estuary (or Blackwater) herring. Working document presented to the ICES Study Group on the Further Development of the Precautionary Approach to Fishery Management4–8 March 2002Lisbon, Portugal.
Punt A.E. (1997) The performance of VPA-based management. Fisheries Research 29:217–243.[CrossRef][Web of Science]
Rosenberg A.A. and Restrepo V.R. (1996) Precautionary management reference points and management strategies. FAO Fisheries Technical Paper 350:2129–158.
Shepherd J.G. (1999) Extended survivors analysis: an improved method for the analysis of catch-at-age data and abundance indices. ICES Journal of Marine Science 56:584–591.
Southward A.J., Boalch G.T., Maddock L. (1988) Fluctuations in the herring and pilchard fisheries of Devon and Cornwall linked to change in climate since the 16th century. Journal of the Marine Biological Association of the United Kingdom 68:423–445.[Web of Science]
Toresen R. (2001) Spawning stock fluctuations and recruitment variability related to temperature for selected herring (Clupea harengus) stocks in the North Atlantic. In Funk F., Blackburn J., Hay D., Paul A. J., Stephenson R., Toresen R., Witherell D. (Eds.). Herring: Expectations for a New MillenniumUniversity of Alaska Sea Grant pp. 315–334 AK-SG-01-04, Fairbanks. 800 pp.
Uriarte A., Roel B., Borja A., Allain G., O'Brien C. M. (2002) Role of environmental indices in determining the recruitment of the Bay of Biscay anchovy. ICES CM 2002/O: 25.
Wood R. J. (1981) The Thames Estuary herring stock. MAFF: Fisheries Research Technical Report, 64.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. P. Groger, R. A. Rountree, M. Missong, and H.-J. Ratz A stock rebuilding algorithm featuring risk assessment and an optimization strategy of single or multispecies fisheries ICES J. Mar. Sci., September 1, 2007; 64(6): 1101 - 1115. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Kimoto, T. Mouri, and T. Matsuishi Modelling stock-recruitment relationships to examine stock management policies ICES J. Mar. Sci., July 1, 2007; 64(5): 870 - 877. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. A. Roel and J. A. A. De Oliveira Harvest control rules for the Western horse mackerel (Trachurus trachurus) stock given paucity of fishery-independent data ICES J. Mar. Sci., May 7, 2007; (2007) fsm016v1. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. T. Kell, I. Mosqueira, P. Grosjean, J-M. Fromentin, D. Garcia, R. Hillary, E. Jardim, S. Mardle, M. A. Pastoors, J. J. Poos, et al. FLR: an open-source framework for the evaluation and development of management strategies ICES J. Mar. Sci., May 1, 2007; 64(4): 640 - 646. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||