Asian-Pacific Aquaculture 2019

June 19 - 21, 2019

Chennai Tamil Nadu - India

GENERAL ADDITIVE MODELLING APPROACH TO PREDICT INDIAN MACKEREL HABITAT ALONG MUMBAI COAST USING SATELLITE DERIVED DATA

Sahina Akter1*, Ajay Nakhawa2, Santosh Bhendekar2,B.B.Nayak1,Anulekshmi Challapan2, Karan K Ramteke1, Dhanya M Lal1, and Zeba Jaffer Abidi1
1.ICAR-Central Institute of Fisheries Education (CIFE), Mumbai - 400 061, India  
2.ICAR-Central Marine Fisheries Research Institute, Seven Bungalows, Mumbai - 400061
 Presenting author:aktersahina1@gmail.com
 

Indian Mackerel (Rastrelliger kanagurta) is widely distributed along the coasts of the Indian and West Pacific oceans. It has high commercial value with 26% contribution to the pelagic fish landing of Maharashtra and about 7.5% contribution (i.e.2.88 lakh t) to the total Indian marine landings. The major objectives of undertaken study were to assess the impact of oceanographic variables on the catch fluctuations of Indian Mackerel and to investigate the Generalized Additive Model (GAM) for predicting Indian Mackerel habitat preference along with the estimation of sea surface temperature (SST), chlorophyll-a concentration (CC), sea surface height (SSH), sea surface salinity (SSS), mixed layer depth (MLD) and Ocean Currents as oceanographic parameters. These data were derived from satellite data of MODIS (Moderate Resolution Imaging Spectroradiometer) and Copernicus Marine Environment Monitoring Service. To test normality, homogeneity of variance and co-linearity the data diagnosis was performed using different packages in R platform after transferring them into log to reduce the skewness and kurtosis. All predictor variables in a GAM model necessarily contribute substantially to explain the variation in the response variables. Redundancy analysis (RDA) was applied to provide a preliminary view of the relationships between Indian Mackerel and environmental variables. Model selection was performed using Multi-Model Inference in the R platform to fit all the combinations of models and then rank them on the basis of Akaike Information Criterion (AICc). The model with the lowest (=best) AICc and high weight were considered for the GAM. Chlorophyll (p<0.001), SSH (p<0.001), SSS (p<0.001) and SST (p<0.1) were having significance in prediction of Indian Mackerel habitat. The constructed models were applicable and therefore they were suitable for predicting the potential fishing grounds of Indian Mackerel inhabiting Mumbai coast. The results from this study highlighted the use of multispectral satellite images for describing the potential fishing grounds.