Integrated Bayesian network frameworks for modelling complex ecological issues

摘要: Ecological problems are typically multi faceted and need to be addressed from a scientific management perspective. There is wealth of modelling simulation software available, each designed address particular aspect the issue concern. Choosing appropriate tool, making sense disparate outputs, taking decisions when little or no empirical data everyday challenges facing ecologist environmental manager. Bayesian Networks provide statistical framework that enables analysis integration information in its own right as well variety models addressing different aspects common overall problem. has been increased interest use BNs model systems issues However, development more sophisticated BNs, utilising dynamic object oriented (OO) features, still at frontier ecological research. Such features particularly appealing an context, since underlying facts often spatial temporal nature. This thesis focuses on integrated BN approach which facilitates OO modelling. Our research devises new heuristic method, Iterative Bayesian Network Development Cycle (IBNDC), for within multi-field multi-expert context. Expert elicitation popular method used quantify sparse, but expert knowledge abundant. The resulting substantiated validated this uncertainty into account. demonstrates application IBNDC support these modelling. The complex nature makes them ideal case studies proposed Moreover, they lend themselves series sub-networks describing components, combining perspectives, pooling similar contributions developed locations by groups. In southern Africa two largest free-ranging cheetah (Acinonyx jubatus) populations Namibia Botswana, where majority cheetahs located outside protected areas. Consequently, conservation countries focussed primarily mitigation conflict between humans cheetahs. In contrast, neighbouring South Africa, found fenced reserves. Nonetheless, remains here. Conservation effort also managing geographically isolated one large meta-population. Relocation option among suite tools resolve human-cheetah Africa. Successfully relocating captured problem cheetahs, maintaining viable population, forming first study thesis. The second involves initiation blooms Lyngbya majuscula, blue-green algae, Deception Bay, Australia. L. majuscula toxic algal bloom severe health, economic impacts community vicinity bloom. Bay important tourist destination with proximity Brisbane, Australia’s third city. several algae considered Harmful Algal Bloom (HAB). This group includes other widespread such red tides. occurrence not local phenomenon, weed occur coastal waters worldwide. With increase frequency extent HAB blooms, it gain better understanding factors contributing sustenance blooms. will contribute practices identification those actions could prevent diminish severity