Unravelling the dynamics of Indian mackerel abundance in the Malabar upwelling region: A comprehensive analysis of oceanographic influences using GLM, GAM, and BRT models

Abstract

The Indian mackerel stands as a pivotal small pelagic fish in the southeastern Arabian Sea, particularly along the coast of Kerala. Despite its significant commercial value, endeavours to forecast the abundance and availability of this species in the Indian waters have been notably limited. This study delves into the intricate relationship between diverse oceanographic parameters in the Malabar Upwelling Region (MUR) and the abundance of Indian mackerel. It employs non-parametric statistical models such as the Generalised Linear Model (GLM) and the Generalised Additive Model (GAM), alongside the machine learning methodology using the Boosted Regression Tree (BRT) model, to delineate the relationship between species landings. The response variable and satellite-derived parameters served as predictors. This study utilised 18 years (1995 – 2012) of fish landing data and environmental variables, encompassing rainfall, mixed layer depth, seawater temperature and salinity at 50 m depth; dissolved oxygen; chlorophyll-a and net primary productivity. Smooth terms were used to capture non-linear relationships between fish landing and the predictors. Rainfall, chlorophyll-a, net primary productivity, and dissolved oxygen have statistically significant effects (p-value < 0.05) on fish landing. Seawater temperature at 50 m depth showed a marginally significant effect (0.05 < p-value < 0.1). Mixed Layer Depth (MLD) and salinity at 50 m depth did not statistically affect fish landing (p-value > 0.1).