Do contemporary age-growth models overlook life-history complexities in protandrous fishes? A case study on the large protandrous polynemid, the giant African threadfin Polydactylus quadrifilis

Abstract

Sagittal otolith pairs were extracted from two-hundred-and-twenty-six giant African threadfin Polydactylus quadrifilis collected from the Kwanza Estuary in Angola between 2016 and 2018 ranging in size from 160 to 1360 mm fork-length (FL) and from one to 23 years of age (26 juvenile, 170 male, nine intersex and 27 female). An additional 85 otolith pairs (24 male, six female and 55 unsexed) were extracted in the same region between 2007 and 2014 (475−1525 mm FL, 2–23 years) and were utilised solely for modelling growth. Otolith aging revealed rapid growth and early maturation (L50 = 399.2 mm FL, A50 = 1.50 years). Sex change occurred over a wide size (790−1125 mm FL) and age (3–8 years) range and was initiated at 1082.3 mm FL and 6.2 years and completed at 1127.4 mm FL and 7.06 years. Strong evidence was provided for partial protandry in P. quadrifilis via the presence of several extremely old male fish, including one individual of 22 years and noticeable differences in length-at-age were observed between older male and female fish. To investigate potential life-history variability, three approaches were taken to group individuals and assess for improvements in model fit using the von Bertalanffy Growth Function (VBGF). Subgroups were determined by sex in Approach 1 and by predicted life-history pathway (‘changers’ & ‘non-changers’) in Approaches 2 (individuals on sex change pathways have increased growth following sex change) & 3 (individuals on sex change pathways have increased growth throughout ontogeny). In all approaches, differences were observed in the L∞ and k values between fish presumed to change sex and those presumed not to change sex. When compared to the conventional model produced for the entire population, all approaches produced considerably better model fit (Approach 1 = ΔAkaike Information Criterion (ΔAIC) = 64.7, Approach 2 – ΔAIC = 56.1 and Approach 3 – ΔAIC = 136.3) and Approach 3 produced better fit than Approach 1. This suggests that utilising conventional modelling techniques may be inappropriate for the stock assessment and management of sequential hermaphrodites and indicates that a revision of age-growth modelling techniques is required for species with multiple developmental pathways in order to maintain intrapopulation diversity and preserve resilience.