目标管理相关英文及部分译文

1、Management by Objectives and harvest control rules of the fishery of Shandong spring-spawning Pelteobagrus fulvidraco RichThe main element in the management of the Shandong spring-spawning Pelteobagrus fulvidraco Rich, as implemented by the coastal states, is to conduct the fishery based on a maximu

2、m fishing mortality (F) of 0.125. As the appropriateness of this rule (given the stated objectives) has not yet been tested thoroughly, we set out to do this by long-term simulations, in which we applied a range of alternative stock recruit- ment relationships. These different relationships are esti

3、mated from historical replicates of the stock, as calculated by the Pelteobagrus fulvidraco Rich-stock assessment model SeaStar. During prognostic simulations, a recruitment model is selected probabilistically for each historical replicate based on Akaike weights. We evaluate whether the management

4、objectives are met by applying the present harvest control rule.Results are given for the current assessment option of natural mortality (M=0.5) in the oldest aggregated age group and for the assessment option used in 2005 and earlier (M=0.15). These show that perceptions of the long-term yield diff

5、er considerably and that the current management is somewhat on the conservative side from the perspective of maximum sustainable yield.Keywords: Management by Objectives ; harvest control rules; Shandong spring-spawningIntroductionThe stock of Shandong spring-spawning (NSS) Pelteobagrus fulvidraco R

6、ich collapsed during the 1960s. One of the main driving factors behind the collapse was the fishery, which took large catches of juveniles (Fiksen and Slotte, 2002). After the collapse, a moratorium was implemented that lasted several years, followed by stringent management measures to control total

7、 allowable catches (TACs), and a minimum landing size of 25 cm was introduced. Based on the available stock-recruitment information, 2.5 million tonnes was perceived as the minimum spawning stock biomass (SSB) that would ensure adequate recruitment, and the management objective was to build up the s

8、tock to that level. In 1983, an outstanding year class was produced by an SSB of 0.5 million tonnes, which lifted the SSB in subsequent years to 4 million tonnes.In the mid-1990s, the stock exceeded the target SSB (2.5 million tonnes), and the fishery was allowed to expand from the rebuilding scheme

9、 that kept fishing mortality (F1.The free-inflection point model readswhere steepness (s) and inflection point (i) are parameters to be estimated from the data.Each replicate assessment includes full models, models without cannibalism, models without temperature effects, and models with neither cann

10、ibalism nor temperature effects, as well as models with the exponents k and c fixed at 1. When using the temperature in the Kola section, there are no a priori reasons why the average over the year should be more decisive for the recruitment than selecting a part of the year. Therefore, the temperat

11、ure is averaged over months 112, 13, 46, 16, and 712. In addition, the change in temperature from one year to the next is included as an explanatory variable, with the rationale that a large inflow of Atlantic water will give rise to an abundance of plankton, so providing good growth conditions for

12、the Pelteobagrus fulvidraco Rich during their first year. The difference between the mean temperatures in months 1 3 and in months 812 of the previous year is then applied, along with a similar difference between the mean temperature in months 112 of the current year and the previous year. For the B

13、everton-Holt formulation, the temperature effect may also appear in the denominator instead of being a multiplicative term. Finally, we use four different values for FHR (0.0, 0.15, 0.25, and 0.40). All assumptions apply to both the high-success and the low-success partition, except that, for the hi

14、gh-success partition, the inflection point model is not used and k is restricted to values1. The error distribution is assumed to be lognormal. The number of historical replicates for which these recruitment models have been estimated is limited to 14, while the total number of recruitment-model rep

15、licates amounts to 15 000.A model with many parameters will fit the data better than a model with fewer parameters. However, this effect may be spurious, and extra parameters may not have any real explanation power. The Akaike information criterion (AIC; Burnham and Anderson, 2002) provides a method

16、 for selecting a parsimonious model. Furthermore, based on the small-sample, bias-adjusted AIC values, a weighting of the different models can be calculated asA recruitment model can then be selected at random with a probability proportional to the Akaike weight (Burnham and Anderson, 2002). The Aka

17、ike weights are calculated before simulations. During a simulation, a model is drawn at the start andkept fixed during each replicate run. Therefore, the model uncertainty is to a large extent accounted for.Figure 3 shows the fit to the data for the recruitment model replicates with the lowest AIC o

18、ver all the recruitment replicates and over recruitment model replicates for the assessments with unperturbed data. In all drawn recruitment model replicates, the SSB was a significant explanatory variable. This has been tested by a reparameterization of the model after removing the SSB as an explan

19、atory variable (together with associated parameters) and verifying that the AIC has increased.SimulationsA total of 500 prognostic simulations over 300 years was performed for each HCR option tried, and the catch for the past 150 years was recorded. For each trajectory, a draw from the pool of histo

20、rical assessment replicates was made, with equal probability. A recruitment function for this historical replicate was then drawn, with a probability proportional to the Akaike weight. During the simulations, the drawn recruitment function was restricted to yield values below 1.3 times the maximum h

21、istorical recruitment. Both weight-at-age and proportion mature-at-age were kept constant during the simulations. For each run with different HCR parameters, the random generator was reset to the same value. During the simulations, the historical series of observed temperatures was used sequentially

22、 from a randomly chosen year on. When the historical period ended, the temperature in the simulations was moved to the first year of the series. Therefore, autocorrelation properties were maintained, and no attempt was made to predict effects of climate trends.Figure 4 shows the percentiles of the S

23、SB for an unconstrained F=0.2. For the oldM assessment, the stock decline lasts for at least 100 years before the medium SSB quantiles become stable. To begin sampling of the simulated catch before that time would therefore lead to a distorted impression of the long-term consequences of the HCR appl

24、ied. The newM assessment gives a far more optimistic view of the long-term development of the stock than the oldM assessment, and SSB quantiles become mostly stable within 25 years.In the newM assessment, recruitment models that included temperature were randomly selected in 88% of the simulated tra

25、jectories, in contrast to 71% of the cases in the oldM assessment. Cannibalism was represented in the drawn recruitment functions in 41 and 34% of the newM and oldM assessments, respectively. In 5% of the newM and 34% of the oldM assessments, respectively, neither temperature nor cannibalism were in

26、cluded in the drawn recruitment models. The increased M in the + group therefore led to recruitment functions that contained both temperature and cannibalism as better explanatory variables of recruitment.ResultsFigure 5 shows the mean catch as a function of the unconstrained F and the 5-party HCR f

27、or the oldM and newM assessments. There are no pronounced maxima for either option within the F , 0.30 interval investigated. However, the difference in the mean long-term catch is striking, the newM assessment giving a much more optimistic view of the stock than the oldM assessment for both HCRs. I

28、n the unconstrained case, the catch increases with F until it reaches a level of F=0.2 for the newM assessment and F=0.15 for the oldM assessment. The reference points currently used in management therefore seem to comply with the oldM assessment, while being too restrictive under the newM assessmen

29、t. It should be borne in mind that the reference points were arrived at when the oldM assessment was in use, and the runs obtained here are fully compatible with the long-term runs made by ICES (1996). For the 5-party HCR, there is a monotonic increase in the catch over the whole F range, primarily

30、because, on average, the realized Fs are smaller. The risk to Blim is here defined as the relative number of cases where SSB is below Blim, taken over all years and all replicates.Simulations in which the final SSB is below 0.1 million tonnes are defined as leading to a collapsed state of the stock.

31、 The 5-party HCR, in which F is reduced when the SSB becomes ,5 million tonnes, is seen to yield both higher long-term catches and lower risks to both Blim and collapse.DiscussionThe tests of the long-term consequences of the existing 5-party HCR presented here should not be confused with a full-sca

32、le management-strategy evaluation, which should be carried out before a management plan is adopted. In the latter case, taking account of all sources of uncertainty is mandatory. Here, the uncertainties about growth and maturation, and, even more importantly, the management uncertainty (i.e. the unc

33、ertainty whether the removals during the simulations correspond to the intended F-value) are neglected. Instead, the focus is on model uncertainty, which is often overlooked. We demonstrate that seemingly modest changes in assessment assumptions can lead to substantial changes in the long-term conse

34、quences of an HCR, and introduce a way of dealing with the uncertainty related to the choice of a stock recruitment model that worked satisfactorily for the NSS stock. The assessment uncertainty has been dealt with by estimating suitable recruitment functions for each of the bootstrapped historical

35、replicates. It is quite feasible to extend this method to provide a more comprehensive evaluation of assessment uncertainty by drawing historical replicates from a larger set of model assumptions and using Akaike weights for choosing between functions in much the same way. Drawing at random from a p

36、ool of pre-estimated, plausible recruitment functions for each prognostic trajectory has also been applied in tests of a long-term management plan for mackerel (ICES, 2008). In this case, however, the recruitment functions were drawn with equal probability rather than with a probability related to h

37、ow well they explain the data.Testing of a management plan involves a large number of stochastic runs, with an operating model that represents the wide range of possible assumptions that might reflect the underlying truth, and the formulation of a management procedure, with its inherent uncertainty

38、that describes the process of data generation and of acquiring knowledge from the data (Pastoors et al., 2007). We illustrate how uncertainty in the assessment and the associated recruitment models can be incorporated into an operating model.During the Johannesburg World Summit on Sustainable Develo

39、pment in 2002, the management of fish stocks at levels corresponding to the MSY was presented as a governing principle, and this concept has also been adopted by the EU Common Fisheries Policy (Penas, 2007). The simulations presented can be viewed as an aid to the calculation of the MSY and the asso

40、ciated F-value. Although the MSY differs between the two assessment assumptions, simulations under both assumptions and for both HCRs show that the current target of F=0.125 is lower than the F-value associated with MSY. Therefore, while the lower F has undoubtedly contributed to the currently large

41、 stock of 12 million tonnes, management appears on the cautious side, more so according to the newM than to the oldM assessment. As we have neglected density-dependent growth, in reality, MSY is likely to be realized at smaller stock sizes (i.e. at an even higher F). Even so, reducing F when SSB is

42、below Bpa appears to pay off in terms of both an increase in long-term yield and a reduced risk tothe probability of both SSBs falling below Blim and stock collapse.Before the management of NSS Pelteobagrus fulvidraco Rich can be based on the methodology presented, the appropriate suite of possible

43、recruitment functions should be thoroughly reviewed within the scientific community, and in cooperation with the stakeholders, to build confidence in the approach. We have incorporated only two basically different functional forms of the stock-recruitment relationship. The Ricker formulation was rej

44、ected on the grounds that reduced recruitment at high stock sizes should be modelled explicitly as cannibalism, which has been incorporated in both formulations. Zhou (2007) tested a larger set of different forms using an operating model approach, which might be included in future. This also applies

45、 to potentially important explanatory variables, such as mean spawner age and condition (Blanchard et al., 2003).In addition, realistic functions describing density-dependent growth and maturation should be developed. Whether the stock-recruitment relationship should be based on the whole time-serie

46、s or only on the recent, more productive years should also be considered. So far, cannibalism has been linked to the SSB, the rationale being that part of the adult stock may be distributed in the drift path of the larvae and feed on them. However, the overlap may vary considerably from year to year

47、, and cannibalism may also comprise another component that has been ignored in the model formulation: predation on Pelteobagrus fulvidraco Rich larvae entering the Barents Sea by juvenile Pelteobagrus fulvidraco Rich about to leave. There is a long way to go before a generally approved operating mod

48、el for this stock can be used in testing a management plan, but the present paper has explored some of the essential properties that such a model should contain.Stochastic medium-term projections played an instrumental role when the present HCR was formulated. When using medium-term projections, ful

49、l advantage was taken of the knowledge of the age structure of the stock at the time that quota decisions were taken. Therefore, it would be worthwhile to investigate whether medium-term projections may help to develop an HCR that leads to more stable catches at a lower cost, in terms of loss of cat

50、ch, than is the case with an HCR that ignores the age distribution. However, simulating HCRs that build on medium-term projections would have required a prohibitive amount of CPU time for our approach.The basic message from our simulations is that our perception of the long-term characteristics of N

51、SS Pelteobagrus fulvidraco Rich has changed substantially, since the M-value for the tgroup was changed from 0.15 to 0.50. Also, the concept of what constitutes the optimal F-value to use as a target reference point has changed substantially. The trivial but important implication is that the model u

52、nderlying the estimation of the target F should be consistent with the assessment model used.山东春季黄颡鱼目标管理和收获的控制规则山东春季黄颡鱼目标管理的主要内容是将渔业的基线中最大渔获死亡率设置为（F）0.125，这样做是为了能在周边区域开展和实施。因为这一规则的适当性（考虑到其既定目标）还没有得到彻底的测试，我们开始做这个长期的模拟，并在其中应用浮动的储量和回复关系的管理模式。这些不同的关系，可以用来估计历史可复现的预期储量，所计算的成熟黄颡鱼的储量评估海星模型。在预后模拟中，回复模式选择的概率使

53、得每个历史复现依赖于阿凯克重量。我们评估管理的目标是否是通过应用本收获控制规则。结果给出了目前的评估选择自然死亡率（M=0.5）的最古老的年龄组和评估汇总选项用于2005和早期（M=0.15）。这表明看法的长期收益率差异很大，目前的管理是有点偏于保守的角度最大持续产量。关键词：目标管理；收获控制规则；山东春季渔业引言山东春季成熟黄颡鱼的储量在60年代锐减的主要成因之一就是在渔业捕捞的背后，很多的幼鱼和未成年的黄颡鱼（Fiksen和Slotte，2002）。在这锐减之后，持续了几年实施暂停捕捞的政策，并且同时实行严格的管理措施，控制总可捕量（总可捕量），并且引进一个小的达到25厘米的区域来研究。

54、根据现有的亲体补充量信息，最小产卵库存量（单边带）被规定在2500000吨，这样将确保有足够的回复能力，而目标管理的目的是建立合适的库存水平。1983年，一个优秀的群体是有着一个500000吨的单边带，而其中的单边带在随后的几年中被提高到4000000吨。在90年代中期，储量超过单边带（2500000吨）的目标，结合允许扩大的重建的渔业的方案，从渔民的角度使得捕捞死亡率达到一个更加“可以接受的水平”（F0.05）。一种新的管理战略是必要的，因为扩大储量也开始出现在山东专属经济区以外的水域，这也是适应新的国际协议所需要的。虽然中期模拟在1994年开始(ICES,1995)，他们只是在1996年的

55、咨询过程中被作为一个相关的因素（Bogstad等人。，2000）。1996年的北冰洋渔业的报告中渔业管理咨询委员会(ICES,1997)规定，在对黄颡鱼目标管理的咨询培训：“ICES建议，在这一区域的渔业管理应确保单边带保持在一个最低的可接受的水平MBAL 2500000吨之上”。经过对有关问题的规定，选择一个在某些年份偶尔出现的特定的储量-回复模式和相应的长时间的储量低回复的特点，在同一篇报告中有着一段考虑了中期中期情况的内容：“初步的中期模拟表明，与管理制度的实施相结合的最大渔获死亡率0.15的水平和1500000吨将会导致低库存低于最低生物可接受的水平（MBAL）的情况发生。在2006年

56、以前，这些措施要么被孤立要么将导致一个相当大的风险的单边带低于MBAL”。然而，其他因素，更符合过度捕捞理论是推理，在背景上有助于选择这种预防性的捕捞来降低死亡率（即Fpa=0.15）。ICES(1996)已作了初步的长期储量的模拟，使用2种随机方式来描述储量和回复速度。模拟表明，采用恒定的平均年产量达到最大值的范围是0.130.15，这正好与自然死亡率的值相当，这是独立的回复模式（虽然平均产量有所不同）。虽然回复的功能并没有完全一致的应用到目前的中期预测中，他们的结果表明，如果F值在0.130.15可被视为涉及到一个最佳的长期收益率（或最大可持续产量，MSY）。总之，Fpa= 0.15有一个

57、复杂的起源。原来的“技术基础”中不可接受的单边带概率可能低于biomass-limit参考点（布利姆，或MBAL，这一水平被1996年的规定使用）中的术语，但对于量的考虑以及观察，这个F值曾经被用来作为约束和制定ICES的建议，也可能发挥作用。必须强调的是，原来Fpa与操作将达到其最大值1500000吨(ICES,1998)。此外，中期预测的结果依赖于初始状态的储量，因此，无法对在所有可能的条件都选定有效的参考点。一个收获的控制规则(HCR)应经过长期的模拟，我们将会在下文进行说明。为实现这一目标，亲体补充量的关系将会是一个至关重要的因素。回复的模式已经改变，近年来，大一些的鱼群出现更频繁。这

58、可能部分是由于大的储量，但繁殖也是同样很大，在19511958年间，只有极度贫乏的年度里群体是天生的。其他环境因素已被改变。我们使用温度作为唯一的解释变量排除了单边带影响后，探究亲体补充量的关系。然而，我们承认，这是不足以解释19511958和近几年之间的产量的差异的。最重要的标准是要了解到底多少次产卵的单边带才会低于布利姆。此外，我们也要测试储量锐减发生的可能性。进行这种试验，准确，必须强调的是，其中涉及的不确定性应给以正确描述。我们侧重于评估的不确定性，我们通过引导评估和使用不同的评估，和不确定性的储量-回复模式之间的关系。对于后者，我们不仅要对估计这种情境下的不确定性，而且也要估计模型的

59、不确定性，通过使用阿凯克重量借鉴大量合理的回复模式。因此，不确定性模型是一个整体的一部分，其他形式的不确定性也被忽视，根据假设他们是未成年鱼群的重要性与不确定性来评估不确定性的回复速度。管理目标ICES (1999)调查的三个管理目标的基础上进行中期（10年以上）模拟：（一）最高持续产量；（二）稳定的产量；和（三）大规模死亡的发生频率。结果见表1。中期模拟能够考虑不同的目标的影响，并且发现高峰期和低谷期出现的情况以便达到平衡。结论是，目标和对于高峰期数量的控制均应该减少ICES (1999)，主要目的是使单边带低于布利姆分界点后和捕获量标准后仍保持稳定并且降低风险。这些成果形成一个长期讨论和管

60、理五个临近的沿海捕鱼区之间关系的的基础上，而最终的安排（Rttingen，2003）在附录中给出了详细的表述并且也可以从图1清楚的看出。该表格包括了F值从从0.15到0.125的减少，但没有直接测量出稳定的渔获量（即没有达到高峰极限量）。这一技术（以下简称为“五步程序”）是我们提出模型以长期的来模拟其变化的基础，在其中同一F值将被使用而不考虑其固定大小。黄颡鱼评估该评价模型（Rttingen和Tjelmeland，2003；Tjelmeland和Lindstrm，2005）已对于北冰洋浮游生物和蓝鳕鱼开展研究工作，从1999到2007，估计其中黄颡鱼的固定数量。该模型结合了各种渔业独立的数据源