5.6 Appendix B Statistics
6.1.1 Complexity of the integrated models
Modelling of both BoB and NPF fishery systems aimed at grouping various components of the fisheries, which are not usually brought together, into a single model. In confronting the classic
trade-off between increasing modelling complexity and maintaining the ability to derive general and robust conclusions about the systems considered, many features of the original BoB and NPF models have been simplified. However, key aspects of model structure have been maintained where these were considered crucial to the understanding of the bio-economic system under study.
Numerous features of both models (summarized in box 5) were similar. First both models are dynamic in order to capture the complex mechanisms, processes and drivers at play. In particular, the multi-species and multi-fleet (or fishing strategies) dimensions of mixed fisheries are accounted for in both the BoB and the NPF models, which was required in order to address the issue of tech- nical interactions. Regarding the biological processes, both models were structured in age (BoB) or in size (NPF) according to the biology of the key modelled species. Age- (or size-) structured bio-economic models provide a more accurate evaluation of the nature of technical interactions as well as of the gross incomes from landings (which are usually affected by the size structure of the landed catch). Another common complexity related to the non-linearities underlying the growth of species. Recruitment was assumed to be related to spawner stocks, however due to differences in assessments, data availability and biology, different assumptions were adopted in the two case stud- ies. Recruitment of BoB species was modelled assuming a segmented function where recruitment decreases for spawning stock biomasses below a certain level (this function is either called hockey stick or Ockham Razor function (Barrowman and Myers,2000;Mesnil and Rochet,2010)). This assumption is supported by empirical evidence, across taxonomic groups, that recruitment tends to be poorest when spawner abundance is low (Myers and Barrowman, 1996). In contrast, Ricker stock recruitment relationships were assumed in the NPF model as inDichmont et al.(2003) and Punt et al.(2010).
The decisional and management viewpoints adopted in the thesis, and the associated papers, focussed on fishing effort controls and constituted another common ingredient of the models and another aspect of their complexity as multi-fleet input controls were taken into account in both cases. In this same vein, the use of projections and scenarios over rather large temporal horizon (20-30 years), as compared to the unit-scales (1 year, 1 week) of the dynamics, brought another element of complexity to the modelling, and highlights the multi-scale (temporal) perspective of the work. Economic features were also incorporated in both models through production, price and
6.1. Important common features of the bio-economic models S. Gourguet
cost functions, which enabled the evaluation of economic indicators relating to fisher, industry and societal concerns, and further contributed to the common complexity of the models.
The analyses in this thesis also captured processes that were specific to each fishery (as sum- marized in box 5). In particular, the bio-economic models developed used different time-steps re- flecting the different biological characteristics of the two marine systems. Prawns are a short-lived species (i.e. 1-2 years life cycle), and the NPF model was expressed on a weekly basis. At a weekly time step, intra-annual and seasonal biological processes could be represented, including spawning and recruitment, and economic processes such as seasonal allocation of effort, as advocated inAn- derson and Seijo (2010). The BoB model structure and parameters were developed on an annual basis as inMacher et al.(2008) based on the ICES yearly1stock assessments. The life cycles of the species in this fishery are longer than those of tropical prawns and a shorter time step would have added unnecessary complexity to the model. An adaptive and rather complex effort allocation pro- cess was also included in the NPF model to represent the specific current fishing strategy between the predictable and less predictable prawn resources. This representation of the temporal effort al- location process between fishing strategies, allowed account to be taken of relevant effort allocation patterns at the scale at which the analysis was being performed. Analyses in the BoB fishery and the NPF included constraints representing the aim to conserve target species spawning stocks and to maintain economic profits. However, while the economic effects of by-catch and other targeted species were only accounted for indirectly in the BoB case study through the gross revenue from other species catches, sea snake catches related to fishing effort were explicitly included in the NPF model. Among all the threatened, endangered and protected (TEP) species caught by the NPF, sea snakes were chosen as a proxy to assess impacts of trawling on broader biodiversity, sea snake catches appearing to be significantly correlated to the effort of tiger and banana prawn sub-fisheries. There is increasing awareness of the need to more fully take the complexity of resource man- agement problems into account (Pahl-Wostl,2007). This thesis showed that this can be done using integrated modelling, in a way that includes only the necessary levels of complexity, while also being based on a close representation of reality (as displayed in the calibration graphs in appendix B). This approach of modelling only the necessary complexity is related to the work ofPlagányi
1Knowledge on BoB species intra-annual biological dynamic is poor, although a quarterly model in length has recently
et al. (2012) which described and reviewed ‘Models of Intermediate Complexity for Ecosystem assessments’ (MICE) that focused on management actions on short timescales. MICE are context- and question-driven and limit complexity by accounting for only the components of the ecosystem needed to address the main effects of the management question under consideration (Plagányi et al., 2012).