Decision tree analysis for the determination of relevant variables and quantifiable reference points to establish maturity stages in Enteroctopus megalocyathus and Illex argentinus

CRESPI ABRIL, AUGUSTO CÉSAR; NICOLÁS ORTIZ; GALVÁN, DAVID

ICES JOURNAL OF MARINE SCIENCE

OXFORD UNIV PRESS

Lugar: Oxford; Año: 2015 vol. 1 p. 1 - 16

1054-3139

Determining the maturity condition of a large number of individuals is crucial for stock assessment and management of cephalopod populations,but this task is difficult to conduct in practice. We propose a novel approach for maturity stage classification using observer-independent criteria.Relevant morphological variables for classification are determined via decision tree (DT) analysis. Using Illex argentinus and Enteroctopus megalocyathusas case studies, individuals were sexed and assigned to a maturity stage defined by specific macroscopic maturity scales. Also, for each individual,the weight of the gonad, accessory glands/ducts, mantle length, and total weight were recorded and maturity indices were calculated(Hayashi index and gonadosomatic index). Two different DT models were fitted: one considering all maturity stages and the other consideringonly intermediate maturity stages since these are the most difficult to determine in practice. For the classification of I. argentinus among allstages, theweights of the nidamental gland and oviductswere the most relevant variables for females (misclassification 23%), while spermatophoriccomplex and testis weights were the key variables for males (misclassification 23%). For classification of intermediate stages only, the nidamentalgland and spermatophoric complex weights were the most relevant variables to classify females (misclassification 19%) and males (misclassification21%), respectively. For E. megalocyathus, the oviducts and ovary weights of females and the terminal organ weight of males were the most relevantvariables for classification among all maturity stages (misclassification 16% and 18%, respectively). For intermediate maturity stages, the same variableswere most important and misclassification improved to 13% for both sexes. Gonadosomatic and Hayashi?s indices were not relevant in eithermodel. DTs based on measurements of cephalopod reproductive systems revealed a simple classification systemfor maturity stages using only a fewvariables that are easy to measure in the field and are independent of observer training. Moreover, the present approach could be easily applied toother cephalopod species using biological data that already exist.