Phytoplankton size structure in a subtropical area from ocean colour and its applications

Abstract

Ocean colour remote sensing provides information on phytoplankton biomass at the global scale, indexed as chlorophyll-a concentration (Chl-a). Several models have also been developed to estimate phytoplankton size classes (PSCs) from ocean colour data. Here we evaluate an abundance-based (AB) model that relies on the total Chl-a as input and the spectral-based (SB) model that relies on the spectrally-resolved phytoplankton absorption coefficient ($a_{ph}$). The models were regionally adjusted using in situ data from a coastal time-series station in the South Brazil Bight, Southwestern Atlantic Ocean (ANTARES-Ubatuba). The regionally-adjusted models were applied to MODIS/Aqua images from 2002 to 2021, using the Generalized Inherent Optical Properties model (GIOP) to estimate Chl-a and $a_{ph}$. The satellite time series was used to analyse the spatio-temporal variability of phytoplankton size classes. A correlation analysis was then performed with annual mean sea surface temperature (SST), Chl-a, and micro and nanophytoplankton size fractions ($F_{m,n}$). The Multivariate El Niño-Southern Oscillation Index (MEI) was used to investigate the influence of the El Niño-Southern Oscillation (ENSO) in the study region. Both PSC models performed reasonably well when tested against an independent in-situ dataset collected in the study region, yielding a correlation coefficient $\rho >$0.6 and a $p$-value $<$0.01. The AB model underestimated the Chl-a associated with the micro- and nano-sized classes ($C_{m,n}$) by 20%, while the SB model underestimated it by 48%. The AB model underestimated $F_{m,n}$ by 12% and the SB by 6%. The satellite validation for the $F_{m,n}$ showed an underestimation of 5% by the AB model and an overestimation of 18% by the SB model. A seasonal pattern was observed, with $F_{m,n}$ dominating the inner shelf during the austral summer and spring, and extending to the entire shelf during the autumn and winter. The micro- and nano-sized fractions showed a significant negative correlation with SST, whereas MEI was positively correlated with Chl-a and $F_{m,n}$ in some areas near the shelf-break, and offshore. These results suggest a potential influence of ENSO on the phytoplankton community structure of the Southwestern Atlantic, off the Southeastern coast of Brazil. The satellite-derived PSC using both models proved to be a reliable resource for ecological and climate studies, and could contribute to the study of the impact of other climate phenomena in the region. Further studies to improve the accuracy of the PSC models are also essential, especially using hyperspectral ocean colour missions that have become available recently.