An in-depth analysis of the impact of environmental drivers on the variability of phytoplankton community in the Arabian Sea during 2010-2021.

This study examines the long-term changes in phytoplankton size classes (PSCs) in the Arabian Sea (AS) using the remote sensing reflectance () data collected over 12 years (2010-2021) from the Moderate Resolution Imaging Spectroradiometer (MODIS). The spectra were inverted to chlorophyll-a (Chl-a) concentrations using a non-linear optimisation method, which were then used to estimate the PSC using a region specific three-component model. The analysis is carried out for all four seasons, , winter (December-February), pre-monsoon (March-May), monsoon (June-September) and post-monsoon (October-November). A machine learning random forest (RF) model is employed to predict the seasonal and long-term variability in PSCs and to quantify the influence of environmental drivers. The seasonal climatology of three size classes - micro (larger), nano (medium-sized), and pico (smaller) - reveals that micro-phytoplankton predominantly occupy the northern AS during winter and pre-monsoon seasons, contributing over 50% to the total Chl-a. During the monsoon season, a significant rise in micro-phytoplankton contribution (60-80%) is noted off the coasts of Somalia, Oman and Kerala due to strong upwelling. In contrast, nano-phytoplankton contributions are minimal during the pre-monsoon season but remain fairly consistent in other seasons, and pico-phytoplankton dominates the oligotrophic waters of the central and southern AS during pre- and post-monsoon. The analysis of PSCs from 2010 to 2021 shows a strong decreasing trend in micro-phytoplankton concentration (-0.13 ± 0.19 mg m year), accompanied by a steady increase in pico-phytoplankton (0.0009 ± 0.0005 mg m year) and nano-phytoplankton (0.001 ± 0.0009 mg m year). To elucidate these long-term trends, RF model was instrumental in identifying key environmental drivers, with sea surface temperature (SST) emerging as the most influential factor affecting pico- and micro-phytoplankton. The feature importance scores for SST are highest during winter and pre-monsoon for both pico-phytoplankton and micro-phytoplankton, underscoring the sensitivity of these classes to temperature changes. RF model also highlights the role of mixed layer depth (MLD) and wind speed (WS) in driving the seasonal shifts in PSCs, particularly during the monsoon and post-monsoon periods. These findings suggest that the rise in SST, coupled with changes in vertical mixing and stratification, drives the shift towards smaller cells, mainly pico-phytoplankton in the AS. This shift towards smaller cells indicates a possible decline in marine food chain efficiency, reduced carbon export rates and declining primary productivity-a real concern for food security in the region.