Premium
Facilitating fishing decisions in an artificial reef area off southern Portugal: a case study using generalized additive models
Author(s) -
Ramos J.,
Santos M. N.
Publication year - 2015
Publication title -
journal of applied ichthyology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.392
H-Index - 62
eISSN - 1439-0426
pISSN - 0175-8659
DOI - 10.1111/jai.12948
Subject(s) - fishing , generalized additive model , reef , stratum , fishery , generalized linear model , artisanal fishing , statistics , mathematics , biology , paleontology
Summary Observations on fisheries activities off Vilamoura (southern Portugal) were conducted using a simple methodology with two objectives: (i) to determine fishing habitat preferences (presence or absence of artificial reefs), and (ii) to determine the importance of environmental and economic factors. For the first objective, the collected data were divided into three spatial strata: two identical strata consisting of either artificial reef presence or absence; a larger stratum further normalized in size for easier analysis and comparison with the remaining observed space (strata A, B and C, respectively). Results show that analyses of the spatial strata differed significantly with regard to the occurrence of fishing vessels, denoting by comparison a higher use associated with artificial reef areas, and that stratum A was statistically different from B (P‐values from Tukey's HSD test: B–A = 0.032, C–A = 0.115, C–B = 0.857). For the second objective, the observational data were allocated to only one spatial stratum. A methodology applying generalized additive models was used to understand the influence of environmental and economic factors on fishing decisions. The co‐variates incorporated into the model were: sea surface water temperature, wind speed, wave height, rainfall, fish price, fuel price (petrol and diesel), and recreational activity. By contrast, relative to effects on fishing, most of the co‐variates had a non‐linear association with fishing vessel occurrences. Best fit was achieved with GAM model 2, where the rainfall predictor was removed (adjusted r 2 = 0.582, deviance explained 66%, UBRE score = 2.28). It can be concluded that generalized additive models are useful for understanding the effects of multidimensional factors on fishing decisions.