Hyperspectral ocean color encodes global eukaryotic phytoplankton community structure

Abstract

Phytoplankton community composition (PCC) impacts global ocean change across scales. Combining the taxonomic resolution of in situ methods with the spatiotemporal coverage of satellite ocean color remains a major challenge. Phytoplankton pigments link PCC to ocean color; past work has demonstrated that 4-6 pigment-based groups can be reconstructed from hyperspectral remote sensing reflectance (Rrs({lambda})). Alternately, metabarcoding approaches reveal thousands of taxa to the species level, but the links between optics and genes are less clear. Here, we compiled a global, coincident dataset of hyperspectral Rrs({lambda}), pigments, and 18S rRNA gene sequences. We used a bio-optical model to produce the relative contributions of 13 eukaryotic phytoplankton classes identified by metabarcoding with high confidence in most cases. Model coefficients were highly spectrally correlated between genes and pigments, revealing the strong PCC signal within Rrs({lambda}). This proof-of-concept study establishes a foundation for detecting eukaryotic PCC beyond pigments in the PACE era.