Modeling long-term fluctuations in the distribution and abundance of Engraulis anchoita eggs and larvae in the southeastern Brazilian bight

Marine ecology progress series

Jana M. del Favero¹  *, Mario Katsuragawa¹  , Maria de L. Zani-Teixeira¹ 

[1] Instituto Oceanográfico/Universidade de São Paulo, Praça do Oceanográfico, 191, São Paulo, SP 05508-120, Brazil

关键词: Zero-inflated models;    Engraulidae;    Ichthyoplankton;    Brazil;   

DOI  :  10.3354/meps12375

学科分类：海洋学与技术

来源: Inter-Research

【 摘 要 】

Most data sets of ichthyoplankton contain a high frequency of zeros, and, not considering the possible inflation of zero counts, as in most studies, may result in incorrect model predictions. Thus, we modeled abundance of Engraulis anchoita eggs and larvae sampled during 18 oceanographic cruises conducted between 1974 and 2010 using zero-inflated (ZI) models. ZI models are mixture models with 2 components: (1) containing false zeros (due to design, survey or observer errors) modeled using a binomial generalized linear model (GLM); (2) containing abundance data that may produce zeros (true zeros), and modeled with a negative binomial GLM (ZINB). Although ZINB has been used in other areas of research, we are not aware of its previous use for ichthyoplankton. Common to larvae and eggs, the probability of false zeros was lowest in the southern area of the southeastern Brazilian Bight and higher at deeper station depths. The probability of false zeros was higher in 0.505 mm mesh-size samples than in 0.333 mm mesh only for eggs. Egg and larval abundance was negatively related to temperature in the count portion of the model; only larval abundance was negatively related to salinity. Egg abundance was higher in years sampled with 0.333 mm mesh under conditions of moderate/strong El Niño. The high interannual variability in the abundance and distribution of eggs emphasizes the importance of long-term studies to better understand patterns of fluctuations in the occurrence of ichthyoplankton that are related to environmental conditions.

【 授权许可】

Unknown