The co-management of fisheries, particularly when undertaken in an adaptive manner, is increasingly
promoted as an effective strategy to redress the commonly cited failures associated with 'top-down',
resource-orientated approaches to management. In spite of this re-orientation, analytical methods to help support adaptive management decision-making by local management bodies are poorly developed. Models of (co-) management performance based on comparisons of indicators of management performance and explanatory variables of a contrasting array of different co- or community managed fisheries are required in order to help guide local decision-making.
The project used the Institutional Analysis and Design (IAD) framework to compare interdisciplinary (resource, technical, socio-economic and political) explanatory and performance indicators of 119 artisanal (co-)managed fisheries in Africa, Asia and Melanesia.
Multivariate analysis of this dataset, together with a review of previous approaches, and hypotheses concerning co-management performance, were used to propose techniques for modeling management performance which can be used to feedback knowledge and advice to local managers to help them achieve their management objectives.
From the analyses, and review of existing approaches, two improved complementary techniques for modeling management performance were proposed:
- GLM (regression modelling) for identifying and assessing the effects of key attributes on outcomes; and
- Bayesian Network modelling for diagnosing strengths and weaknesses among co- management units and for exploring 'what if' scenarios.
GLM provides a powerful means of describing the more quantitative response elements of management systems. Bayesian Networks are able to model visually the more complex and intermediate pathways of causality. They are also able to learn, as more cases (evidence) become available, which is a particularly relevant feature for adaptive management applications. Managers can be readily trained in the skills needed for constructing network models, and the Netica software used for this project is very user-friendly and inexpensive.
Guidelines for field applications of the two modeling approaches are provided including practical advice on identifying sampling units, important variables, data levels and cleaning, minimum sample sizes, and sensitivity analysis.
By applying both methodologies to the dataset compiled, some key conditions for successful co-management were identified. Factors leading to higher production (catch-per-unit-area), sustainability (catch-per-unit-effort, CPUE), wellbeing (average annual household income), equity, compliance and perceived changes in CPUE were identified.