2014-01-012024-05-18https://scholars.lib.ntu.edu.tw/handle/123456789/698927Abstract: Fisheries management is a pressing concern for both food security and marine conservation issues. Sustainable fisheries require wise management strategies, which call for incorporation of ecosystem knowledge into the management of ocean resources. This is in contrast to most existing management methods, which are based on classical single-stock assessments in most part of world. Criically, any marine fish species coexists and interact with other species. As such, multi-species management approaches have long been advocated to replace the current single-species management scheme. However in practice, how to implement multi-species assessment methods remain elusive. A critical knowledge gap is “how to identify and quantify species interactions in natural ecosystems?” Here, we propose to employee a novel method that can effectively quantify species interactions given time series data. This method, known as Convergent Cross Mapping (CCM), was developed by my research group and has been published in Science (2012). This method is gaining attention and considered as a very effective method to study causal relationships among components in complex ecosystems. In this research, we aim to apply this new method to improve fisheries management. Our objective is to 1) identify a functional-coupling group (a group of interacting species in an ecosystem), and 2) for each target fish species, we incorporate the information of interacting species into fisheries assessment. To achieve these tasks, first, we develop a global database by compiling multi-species time series data, including fishery-dependent catch, CPUE, and spawning biomass data and fishery-independent survey data. Secondly, using nonlinear methods, we determine dimensionality (system complexity), nonlinearity (existence of nonlinear complex behavior and interactions), and predictivity (uncertainty) of each time series. Thirdly, we use CCM to determine species interactions and define the functional-coupling group. Finally, we use multivariate embedding to do multi-species management.