Análise comportamental e distribuição da atividade pesqueira no Arquipelágo de Fernando de Noronha (Nordeste, BR) baseada em dados de GPS

• Autor(a) principal: COSTA, Tatiana Beltrão Alves da
• Data de Publicação: 2019
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da UFRPE
• Texto Completo: http://www.tede2.ufrpe.br:8080/tede2/handle/tede2/8424
• topic: Frota pesqueira; Atividade pesqueira; Fernando de Noronha (PE); Geolocalização; CIENCIAS AGRARIAS::RECURSOS PESQUEIROS E ENGENHARIA DE PESCA

Information on the temporal dynamics of the fishing fleets has been widely used to infer many aspects of fisheries science, such as the evaluation of species distribution patterns, investigate impacts on habitats due to fishing effort, in distribution of fishing vessels, among others. In this work we described the spatial distribution and catch composition of the artisanal and recreational fleets of the Fernando de Noronha Archipelago, based on GPS data. For the description of the spatial distribution, a Hidden Markov model was applied in order to segment the trajectories in different activities, called behavioral states. Onboard observer’s data from 21% of the trips monitored via GPS were used to validate the prediction of the models. Values of accuracy over and underestimation of fishing activity estimated by the modeling were calculated through confusion matrixes. In addition, random forest models were applied to define which variables (subset, interpolation period, number of states, step distribution family and angular distribution family) were most important in the accuracy, over and underestimation. According to distribution results, both fleets occupy similar areas, tending to perform fishing at points traditionally known by fishermen. However, although sharing similar fishing zones the composition and structure of catches differ among fishery fleet. The artisanal fleet concentrates its catch on medium-sized individuals, mainly barracudas (Sphyraena barracuda) and rainbow runner (Elagatis bipinnulata), while the recreative catches fish of more varied sizes, mainly barracudas and tunas. Regarding the modeling of the fishing trajectories, the models generally obtained good values of accuracy between 58% and 79%. In addition, the mean overestimation and mean underestimation of fishing activity were approximately 21% and 6%, respectively. According to results from the random forests, the subset, number of states and period of interpolation were considered the most influential variables for accuracy, overestimation and underestimation of catching state. It was observed that the models tended to overestimate fishing events in high sinuosity and high-speed segments. In addition, models also underestimated fishing in portions of the trajectory where boats sailed straight and at moderate speed. In relation to the number of states, the addition of a third behavioral state resulted in better accuracy results, but it did not mean the increment of a new behavioral state serving only to refine the estimation of the fishing state. In general, the results obtained in this work can help to understand the spatial dynamics of the fishing fleets of Fernando de Noronha, highlighting important fishing areas that mostly surround the limits of the Marine National Park. The information presented here may serve to better clarify the particularities of the artisanal and recreational fishermen of the archipelago and also in the forecast the impacts that changes in the conservation units could cause in the distribution of the vessels.