MEPS 560:223-235 (2016)  -  DOI: https://doi.org/10.3354/meps11901

ABSTRACT: Early studies within Algoa Bay, temperate South Africa, have predicted that coastal fish species with pelagic eggs spawn offshore. More recent studies on larval fish assemblages in the nearshore have shown that spawning may take place closer to shore based on the prevalence of preflexion larvae. The aim of the present study was to investigate whether species that are important from a fishery and ecological perspective are potentially spawning in Algoa Bay. Larval fish abundance data, from a mosaic of coastal habitats in Algoa Bay, were used to model (1) the spatial distribution of larvae and (2) the spawning location of species. A novel approach using a maximum entropy machine-learning algorithm to model larval distribution was coupled with reverse progressive vector plots (RPVPs) to predict potential spawning areas. The results demonstrated that larval fishes in the nearshore habitats of Algoa Bay exhibit species-specific spatial variability. Results from RPVPs indicate that larvae occurring at any location within Algoa Bay could be hatched and transported extensively within the bay before reaching a suitable size to swim independently of nearshore currents. The spawning of several important fishery species, namely Engraulis encrasicolus (Engraulidae), Sardinops sagax (Clupeidae) and Argyrosomus inodorus (Sciaenidae), was modelled in Algoa Bay. Our results indicate that spawning occurs closer to shore than what was previously thought. Eggs and larvae typically remained in close proximity to their spawning origin. Larval fish distributions, although variable amongst all study species, correspond to known adult distributions and potential reef habitats in the study region.

KEY WORDS: MaxEnt · Spawning origin · Species distribution modelling · Fish larvae · Reverse progressive vector plot · RPVP