Phytoplankton are crucial in regulating global ocean biogeochemical cycles and marine food webs. Growing concerns exist regarding the effects of upper-ocean warming and increased stratification on phytoplankton production and carbon export. Observations show synchronous increases in surface ocean temperature and apparent global decreases in phytoplankton biomass (inferred from chlorophyll *a* concentrations). Warming is predicted to have both positive (impacting photosynthetic metabolism) and negative (increasing stratification, reducing nutrient inputs) effects on phytoplankton primary production. Warming is also projected to reduce biological carbon export to the ocean interior by decreasing nutrient upwelling and/or shifting phytoplankton communities towards smaller, less dense picophytoplankton. The magnitude of these changes depends on the stoichiometric relationship between carbon (C), nitrogen (N), and phosphorus (P), with the Redfield ratio (106 C:16 N:1 P) serving as a canonical reference. Fixed stoichiometric models suggest a potential 20% decline in global net primary production (NPP), while flexible stoichiometric models indicate a smaller (<10%) reduction. Both model types highlight the Arctic Ocean and subtropical gyres as particularly sensitive regions. These NPP reductions, in the models, are linked more to declines in phytoplankton growth rates than in biomass, emphasizing bottom-up controls (nutrients, temperature, light) on NPP. Flexible stoichiometric models predict smaller decreases in the biological carbon pump due to increased nutrient use efficiency, but this is partially offset by the reduction in mean phytoplankton cell size, which limits carbon export associated with smaller phytoplankton. In situ data resolving these competing processes remain scarce. The Sargasso Sea, with its highly flexible macronutrient stoichiometry and significant contributions of *Synechococcus* and *Prochlorococcus* to carbon export, presents an ideal case study. This research leverages three decades of data from the Bermuda Atlantic Time-series Study (BATS) site to examine the relationships between phytoplankton production, nutrients, ocean physics, and carbon export in the Sargasso Sea, focusing on whether warming inversely correlates with nutrient inventories, whether planktonic communities exhibit stable or variable compositions, and how these interactions impact observed carbon export.

Existing literature highlights the central role of phytoplankton in regulating global ocean biogeochemical cycles. Studies have raised concerns about the impact of ocean warming and increased stratification on phytoplankton productivity and carbon export. Observations suggest a global decline in phytoplankton biomass, linked to rising ocean temperatures. Theoretical models predict a reduction in net primary production (NPP) and carbon export due to enhanced stratification and nutrient limitation. These models vary in their predictions based on the assumed stoichiometric flexibility of phytoplankton. Some studies suggest a significant decline in NPP, while others predict a smaller decrease. The sensitivity of different ocean regions to these effects has also been investigated, with subtropical gyres, including the Sargasso Sea, identified as potentially vulnerable areas. The Sargasso Sea, however, has a unique phytoplankton community with highly flexible macronutrient stoichiometry and efficient carbon export mechanisms, making it a critical region for studying the impacts of climate change on marine ecosystems. Previous research on the Sargasso Sea has documented various aspects of its biogeochemical dynamics but a comprehensive long-term analysis of the combined effects of warming, nutrient limitation, and phytoplankton community change on carbon export has been lacking. This study bridges this gap.

The study used data from the Bermuda Atlantic Time-series Study (BATS) site in the Sargasso Sea, spanning nearly three decades (1990-2020). Hydrographic data (temperature, mixed layer depth) were obtained from CTD casts. Dissolved inorganic nutrients (nitrate, phosphate) were measured using standard autoanalyzer methods, with high-sensitivity phosphate measurements available from 2004 onwards. Particulate organic carbon (POC), nitrogen (PON), and phosphorus (PP) were determined using an elemental analyzer and an ash-hydrolysis method. Picoplankton abundance and biomass were measured using flow cytometry, differentiating between *Synechococcus*, *Prochlorococcus*, picoeukaryotes, and nanoeukaryotes. Phytoplankton carbon was also estimated independently using a POC:chlorophyll *a* regression. Particulate elemental fluxes were quantified using surface-tethered particle interceptor traps. Primary production was determined using in situ ¹⁴CO₂ incorporation methods. A trait-based phytoplankton stoichiometry model was used to predict phytoplankton C:P ratios as a function of satellite-derived growth rate, Chl:C ratio, and P limitation. Statistical analyses, including linear regressions and t-tests, were employed to examine trends and differences between periods with and without significant warming.

Surface ocean temperatures at the BATS site increased significantly by ~0.9 °C between 1990 and 2020, with the most rapid warming occurring in the 2010s. The increase in temperature and reduced seasonal mixing in the 2010s were correlated with significant declines in nitrate and phosphate inventories (0-140 m). Net primary production (NPP) decreased by ~30% during the 2010s. Contrary to model predictions, carbon export fluxes at 150 m did not decrease significantly. The export ratio (carbon export:NPP) increased, suggesting improved ecosystem efficiency in exporting carbon despite reduced nutrient availability and NPP. The decline in NPP was accompanied by changes in phytoplankton community composition, with a decrease in eukaryotic biomass and a stable prokaryotic biomass. The reduction in NPP appeared to be primarily driven by the decrease in phytoplankton biomass, not changes in growth rates. Significant increases in C:P and N:P ratios were observed in both exported particulate matter and seston, particularly in the upper euphotic zone. This increase in C:P ratios was linked to both shifts in the phytoplankton community composition towards cyanobacteria with higher C:P ratios and enhanced shallow remineralization of P. A trait-based model confirmed an increase in modeled phytoplankton C:P ratios during the 2010s, consistent with observational data. While some changes in zooplankton community structure were observed, their contributions to maintaining carbon export rates remained unclear.

The study's findings contradict predictions from global biogeochemical models, which suggest that warming-induced stratification would lead to reduced carbon export. The observed maintenance of carbon export, despite decreased nutrients and NPP, highlights the adaptive capacity of the Sargasso Sea ecosystem. The shift towards a phytoplankton community dominated by cyanobacteria with high C:P ratios, coupled with enhanced shallow phosphorus recycling, increased nutrient use efficiency and maintained carbon export. This adaptive response suggests that surface ocean ecosystems may be more resilient and adaptable to climate change impacts than previously thought. The observed decoupling of NPP and carbon export underscores the limitations of models that rely solely on bottom-up controls, highlighting the importance of incorporating trophic interactions and phytoplankton community dynamics into future models. The study's findings emphasize the need for improved ecosystem models that accurately capture these adaptive mechanisms and interactions.

This study reveals a counter-intuitive response of the Sargasso Sea ecosystem to warming. Despite reduced nutrients and NPP, carbon export was maintained through adaptive mechanisms, including a shift in phytoplankton community composition and enhanced phosphorus recycling. These findings challenge current earth system models and highlight the importance of incorporating adaptive capacity and trophic interactions into future climate change projections. Further research should focus on quantifying the relative contributions of different trophic levels to carbon export and improving model parameterizations to better capture ecosystem responses to environmental changes.

The study primarily focuses on a single location (BATS site) in the Sargasso Sea, limiting the generalizability of the findings to other regions. Some data, such as detailed zooplankton community composition and bacterial productivity, were limited, preventing a complete assessment of all factors influencing carbon export. The trait-based model used simplifications to represent phytoplankton communities, which could influence the accuracy of modeled C:P ratios. Future research could address these limitations by expanding spatial coverage, collecting more comprehensive data, and using more sophisticated model approaches.