We aim to construct a spatial multi-species ecosystem model focused on commercially exploited fish in the northern European shelf (47 to 62 degrees N, and -12 to 10 degrees E). Species or functional groups within the model will be placed within a hierarchy of biological detail. At the highest, or 'structured', level of detail the target species will be represented by interacting length-structured populations in which the processes governing individual growth, predation, reproduction, and mortality are fully specified. At an intermediate level will be 'unstructured' ecosystem elements, such as benthic invertebrates and zooplankton, that are represented dynamically but without life history detail. At the lowest level of detail we will have the 'drivers' of top predators (cetacean and seabird) and fishing mortality, which are not modelled but provide forcing functions to the rest of the ecosystem. The model will be run at various spatial scales, from a series of linked regional models through to medium-scale grid covering the entire domain. The physical environment will be obtained from an ocean circulation model, with temperatures used to drive temperature-dependent process such as growth rate, and flows determining the transport of plankton zooplankton and fish larvae. The model will be parameterised using fishing survey data, and estimates of zooplankton and benthic production, and estimates of consumption by top predators. Due to the potential for ecologically and commercially important impacts by top predators a joint proposal (for a tied studentship) will aim to produce improved spatial estimates for these drivers. Free parameters will be estimated by Markov chain Monte Carlo (MCMC) methods. We will draw on a range of computational innovations developed at Strathclyde University for dealing with explicitly spatial models of populations with life history structure and growth variability. We will use the Ecopath software as an aid to constructing regional budgets for mass transfer between the species in our system, and which can then be compared to those predicted by our model. Our modelling approach differs substantially from alternative ecosystem modelling methods, such as Ecopath with Ecosim (EwE), that are frequently advocated as the way forward for ecosystem-based management in fisheries. A key part of our proposal is therefore a comparative study involving EwE regional ecosystem models, with a view to identifying the strengths and weaknesses of alternative methodologies.
A multi-species model of the North Sea fish community in which each species is represented by length classes from egg to adult was constructed. The model was highly successful in replicating historical stock biomasses, as well as time series of recruitment and landings, population length distributions, and diet. Key results suggested that herring predation on early life history stages of cod is dynamically important, and that high herring abundance may play a role in the decline of stocks even during periods of declining fishing pressure. The maximum sustainable yield of cod was strongly dependent on herring abundance, and it was found that current levels of cod exploitation may become unsustainable if herring recruitment returns to historical high levels. The model used the model to explore the effects of different fishing strategies on a size-based community index. Currently there is a management target that the proportion by weight of fish caught in scientific surveys that are over 40cm in length - known as the Large Fish indicator (LFI) -should be 0.3. The Marine Strategy Framework Directive, which applies to all marine seas under the jurisdiction of the European Union, requires 'good environmental status' be achieved in respect of maintaining marine biodiversity and food web structure by 2020. The LFI has been explicitly proposed for use as a food web indicator to monitor change in the proportion of top predators, and there is clear potential for it to be used as a biodiversity indicator to monitor change in the relative proportions of ecosystem components. From its peak of 0.3 in 1983, the LFI declined to a minimum of 0.05 in 2001, before recovering to around 0.15 by 2010. These changes have been linked to changes in fishing pressure on the demersal fish community, which peaked in the mid-1980s and has subsequently been reduced by approximately 60%. With the MSFD requiring the target of 0.3 to be reached by 2020, there is now an urgent need for specific advice regarding the exact management action required to achieve this. The model was used to forecast the effects of different fisheries management scenarios on the future trajectory of the LFI. A community-wide reduction in fishing mortality of approximately 60% from 2008 values was required to reach the target LFI within a seven year forecast period. The projected increase in the LFI was mainly produced by an increase in the abundance of cod and to a lesser extent saithe, the two largest-bodies species in the modelled community. The LFI target can be reached by a 70% reduction in fishing mortality on cod alone, and that a reduction to meet the maximum sustainable harvesting rate is projected to exceed the LFI target (Figure 4). The model was also used to explore the consequences of varying the effort of six fishing metiers. A 75% reduction in otter trawl effort alone is sufficient to reach the LFI target.