Multi-model inference applied to the total length-gutted weight relationship in sharks: The case of the bonnethead shark Sphyrna tiburo

Abstract

The modeling of the length-weight relationship (RLP) in sharks, the potential model has always been fitted {[}linear model (L) in logarithmic version], multi-model inference (IMM) has never been applied and constant growth has always been assumed in all phases of ontogenetic development. The bonnethead shark Sphyrna tiburo from the Gulf of Mexico is no exception. By modeling the RLP in females (H), males (M) and combined sexes (Comb) of S. tiburo, and applying the IMM {[}models (L), quadratic (Q), cubic (C), broken-stick (BS), two segments (DS), BS and DS with logistic transition between segments] based on information theory (TI), the hypothesis that the adjustment of L is insufficient to describe the RLP in sharks was tested. No model was selected. The IMM showed insufficient adjustment in L (0.86, 72.70 and 0.45\%, in H, M and Comb, respectively), and significant in the average model (99.14, 27.30 and 99.55\%, in H, M and Comb respectively); and showed that growth changes depending on the size and presents a marked morphological change in a certain specific size, manifested by the three types of growth registered in H and Comb {[}negative allometric (CAN) and positive (CAP), and isometric growth (CI)], and the two in M (CI and CAP), whose changes between growth types were coinciding with the change of juvenile-adult diet and the beginning of sexual maturity.