The Western Antarctic Peninsula (WAP) is experiencing rapid warming and significant sea ice loss, impacting its marine ecosystem. Eukaryotic plankton, forming the base of the food web, are crucial for biogeochemical cycles and carbon uptake. Understanding the effects of these changes on plankton community structure, biodiversity, and carbon flux is critical. This study utilizes five years of high-resolution net community production (NCP) and high-throughput DNA sequencing data to explore the contribution of polar eukaryotic plankton to biological carbon fluxes in the WAP. The WAP system is characterized by a short, highly productive growing season, with NCP reflecting the amount of organic carbon available for export from the surface mixed layer. The researchers aimed to determine the influence of environmental factors, particularly sea surface temperature (SST) and sea ice conditions, on plankton community structure, biodiversity, and carbon fluxes.

Previous research has documented significant changes in the WAP, including rising air temperatures, warming and freshening of the upper ocean, and a rapid decrease in sea ice extent. These changes have been linked to observed alterations throughout the Antarctic marine food web. Studies have shown a link between elevated NCP or primary production and high sea-ice conditions, attributed to ice-melt-enhanced water column stability, increased light availability, and potential iron supplied by sea ice. Bloom-favorable conditions have been associated with negative phases of the Southern Annular Mode (SAM), leading to increased ice extent in winter and enhanced ice melt in spring/summer. Existing literature highlights the importance of eukaryotic plankton in driving carbon fluxes at the WAP, but further investigation into the interplay between environmental changes and plankton community dynamics is needed.

The study used five years of data from the Palmer Long-Term Ecological Research (LTER) cruises, encompassing oceanographic and biological surveys. Net community production (NCP) was measured in situ using the O2/Ar method, providing estimates of carbon fluxes in the mixed layer. High-throughput DNA sequencing of the 18S rRNA gene marker was used to assess eukaryotic plankton community structure and diversity. Environmental data, including sea surface temperature (SST), mixed layer depth (MLD), salinity, and sea ice concentration, were collected concurrently. Canonical correspondence analysis (CCA) was used to examine relationships between community composition and environmental variables. Alpha diversity indices (Chao1, Pielou’s evenness, and Shannon) were calculated to assess biodiversity changes in relation to SST. A weighted gene correlation network analysis (WGCNA) was employed to identify clusters of highly interconnected plankton ASVs (amplicon sequence variants). Finally, Genetic Programming (GP), a machine learning technique, was used to build predictive models for NCP using the WGCNA modules and environmental factors as input variables.

The study found substantial spatial heterogeneity and interannual variability in summer NCP. Years with late sea-ice retreat showed significantly higher NCP than years with early retreat. Four eukaryotic plankton groups dominated the WAP surface water: diatoms, cryptophytes, dinoflagellates, and haptophytes. Community composition differed significantly between years with high and low sea ice extent. CCA revealed that sea-ice conditions and SST were the primary drivers of community differentiation at the ASV level. All three diversity indices (Chao1, Pielou’s evenness, and Shannon) showed significant negative correlations with SST, indicating lower richness and evenness in warmer waters. The GP models explained up to 80% of NCP variability, with the sea-ice-associated plankton assemblage ('turquoise' module – MET) being the most important predictor. This assemblage was dominated by diverse diatoms and dinoflagellates, with central taxa including *Thalassiosira*, *Odontella*, *Porosira*, *Actinocyclus*, *Proboscia*, *Chaetoceros*, and *Gyrodinium*. The MET module was positively correlated with low SST, low salinity, shallow MLD, and elevated sea-ice melt, and was also positively correlated with primary production and bacterial production. Two other modules (MER and MEG) also contributed to NCP prediction but to a lesser extent.

The findings highlight the strong influence of sea ice extent and SST on the structure, diversity, and productivity of eukaryotic plankton communities in the WAP. The negative relationship between biodiversity and SST contrasts with global-scale trends observed in other studies. This difference might be attributed to the unique characteristics of ice-associated diatom communities in the WAP, which exhibit high diversity at lower temperatures. The study’s predictive models successfully capture the complex, non-linear relationships between environmental factors and NCP. The identification of specific plankton assemblages as key drivers of NCP provides valuable insights into ecosystem functioning and its response to environmental change. The three patterns identified – ice-associated diatoms dominating in high-ice scenarios, cryptophytes in warmer waters, and heterotrophic protists in deeply mixed waters – reveal how differing plankton communities lead to different carbon fluxes.

This study demonstrates that sea-ice conditions are a critical driver of eukaryotic plankton diversity, community structure, and carbon cycling in the WAP. The strong predictive power of the machine-learning models underscores the importance of considering both environmental and biological factors when modeling NCP. Future research should focus on extending these findings to larger spatial and temporal scales, integrating other omics data, and conducting field and lab experiments to validate the findings. Continued monitoring of the WAP ecosystem is crucial for understanding the long-term consequences of climate change on Antarctic marine biodiversity and carbon sequestration.

The study's observations are limited to seasonal snapshots (summer) of the WAP system, limiting the ability to generalize findings across all seasons. The lack of direct iron measurements introduces uncertainty in assessing the role of iron availability in regulating NCP. The statistical models, while robust, require further validation with field experiments and lab studies. The focus on eukaryotic plankton might overlook the contributions of other microbial groups. Finally, despite utilizing a long-term dataset (five years), more data would enhance the reliability and generality of the trends detected.