Browsing by Author "Cassar, Nicolas"

The Biological Carbon Pump (BCP) is a natural mechanism in the ocean that exports carbon in the deep ocean and is estimated to transfer between 5 and 12 Pg C from the surface to the deep ocean annually. While the underlying mechanisms of this process – primary producers create organic carbon from CO2 through photosynthesis, some of this organic carbon is recycled in the surface ocean, while some of it is exported to depth via physical or biological processes – have been identified for decades, this process remains difficult to quantify and predict. We estimate the carbon export potential from the surface ocean by estimating net community production (NCP) from continuously measured in situ O2/Ar ratios. In this dissertation, I aimed to assess the coherence of many methods of measuring NCP and determine factors, both physical and biological, that drive changes in NCP. Together, these goals allowed me to offer suggestions to improve modeling efforts to estimate the BCP from autonomous or remote sensing observations. To explore these topics, I used many different methods. In Chapter 2, I showed that measurements of NCP collected from different methods were consistent around Ocean Station Papa in the North Pacific after accounting for spatial heterogeneity. I compared estimates of NCP from shipboard O2/Ar measurements; O2, NO3-, particulate organic carbon (POC), and dissolved inorganic carbon (DIC) measurements from autonomous platforms, and shipboard incubations based on changes in Chl a and NO3-. I used a generalized additive mixed model to compare the datasets when spatial and temporal differences in the measurements were considered. In Chapter 3, I explored drivers of NCP by comparing how NCP related to various in situ biomarkers and biogeochemical rates measurements. I used moving Pearson’s correlations to assess how continuous measurements of biomarkers such as Chl a, POC, phytoplankton carbon, temperature, and community particle size distribution correlated to changes in continuous NCP. In addition, I showed that NCP was likely driven by changes in production, rather than respiration, in both the North Pacific and North Atlantic by comparing NCP with incubation-based estimates of gross primary production (GPP), net primary production (NPP), and microbial community respiration (mCR). Finally, I modeled NCP from the available biomarker data and determined that POC is a better proxy for estimating NCP than Chl a, in both locations. Finally, in Chapter 4, I examined how changes in the microbial community (from 16S and 18S amplicon sequencing) paired with changes in NCP in the North Pacific. I showed that at coarse taxonomic groupings, such as Phylum, Class, or plankton functional type, had no correlation to changes in NCP, while individual amplicon sequencing variants (ASVs) had strong correlations to changes in the surface ocean organic carbon balance. This indicates a need for increased granularity in microbial community composition estimates to effectively model NCP or carbon export from surface ocean microbial communities. Altogether, my research increases confidence in global NCP estimates from various platforms, presents potential improvements to biogeochemical modelling efforts, and suggests that respiration does not drive changes in NCP in the ocean.

Marine productivity is limited by nitrogen in a large portion of the global ocean. Marine nitrogen fixation, catalyzed by a select group of microorganisms called diazotrophs, converts nitrogen gas (N2) into bioavailable nitrogen that can support the growth of marine phytoplankton. By supplying new nitrogen to marine ecosystems, marine N2 fixation affects marine primary production, the uptake of carbon dioxide and ultimately the global climate. However, the environmental controls on N2 fixation and the physiologies of diverse diazotrophs remain elusive, in great part due to the limited number of observations. As part of this dissertation, I applied a variety of approaches including statistical modeling, high-resolution field measurements, and gene sequencing to characterize the biogeography of marine diazotrophy.

The first approach was to model marine N2 fixation and diazotrophs using machine learning methods. To that end, I conducted meta-analyses to update the global datasets of N2 fixation and diazotrophs. The number of observations in these updated datasets are ~80% and over 100% larger than previous datasets, respectively. Simple correlation analyses between N2 fixation rates and different environmental factors failed to identify a single factor explaining marine N2 fixation at a global scale. In contrast, individual diazotrophic phylotypes showed distinct relations to environmental properties. Machine learning methods including random forest (RF) and support vector regression (SVR) simulated the observed N2 fixation and diazotrophs fairly well by accounting for nonlinearities among multiple environmental factors. The estimated global N2 fixation fluxes from the two statistical models were within the range of other studies. However, the machine learning estimates and other simulations in some cases showed substantial disagreement in both the magnitude and distribution of N2 fixation and diazotrophs, especially in high latitudes and the eastern equatorial Pacific, where observations are scarce. The large uncertainties in simulated N2 fixation and diazotrophs emphasized the need for a better understanding of the factors regulating N2 fixation and the physiology of diazotrophs.

Achieving this goal can be labor-intensive and difficult with current techniques, which are based on discrete sampling and long incubation time. To overcome some of the drawbacks of traditional methods, our laboratory developed a method for high-frequency underway N2 fixation measurements. This method provides better coverage of the spatial and temporal heterogeneity in N2 fixation. I deployed this method over large swaths of the western North Atlantic Ocean in the summers of 2015, 2016, and 2017, covering over 10,000 km cruise tracks. This extensive survey identified new hotspots of N2 fixation in the coastal waters of the mid-Atlantic Bight. By coupling high-resolution N2 fixation observations with underway estimates of net community production (NCP) derived from O2/Ar measurements, I revealed the heterogeneous contribution of N2 fixation to NCP and to the carbon cycle, with a surprisingly large contribution in coastal waters.

In addition to the spatial distribution of N2 fixation, I also characterized types of diazotrophs responsible for N2 fixation and how they responded to varying environmental conditions. By measuring diazotrophic diversity, abundance and activity at high-resolution using newly developed underway sampling and sensing techniques, I captured a shift between diazotrophs from Trichodesmium to UCYN-A from oligotrophic warm (25-29°C) subtropical Sargasso Sea to the relatively nutrient-enriched cold (13-24°C) eastern American coastal waters. Meanwhile, N2 fixation rates were significantly enhanced when phosphorus and Fe availabilities, and chlorophyll-a concentration increased across the Gulf Stream into the subpolar and coastal waters. Phosphorus limitation was confirmed with changes in the expression of phosphorus uptake genes in Trichodesmium and UCYN-A. While temperature was the major factor controlling the diazotrophic community, phosphorous was dominantly driving the changes of N2 fixation rates in the western North Atlantic.

Overall, this dissertation significantly improves our understanding of the distribution of N2 fixation and diazotrophs and their environmental controls in the western North Atlantic and in the global ocean.

The Arctic Ocean and Western Antarctic Peninsula (WAP) are the fastest warming regions on the planet and are undergoing rapid climate and ecosystem changes. Until we can fully resolve the coupling between biological and physical processes we cannot predict how warming will influence carbon cycling and ecosystem function and structure in these sensitive and climactically important regions. My dissertation centers on the use of high-resolution measurements of surface dissolved gases, primarily O2 and Ar, as tracers or physical and biological functioning that we measure underway using an optode and Equilibrator Inlet Mass Spectrometry (EIMS). Total O2 measurements are common throughout the historical and autonomous record but are influenced by biological (net metabolic balance) and physical (temperature, salinity, pressure changes, ice melt/freeze, mixing, bubbles and diffusive gas exchange) processes. We use Ar, an inert gas with similar solubility properties to O2, to devolve distinct records of biological (O2/Ar) and physical (Ar) oxygen. These high-resolution measurements that expose intersystem coupling and submesoscale variability were central to studies in the Arctic Ocean, WAP and open Southern Ocean that make up this dissertation.

Key findings of this work include the documentation of under ice and ice-edge blooms and basin scale net sea ice freeze/melt processes in the Arctic Ocean. In the WAP O2 and pCO2 are both biologically driven and net community production (NCP) variability is controlled by Fe and light availability tied to glacial and sea ice meltwater input. Further, we present a feasibility study that shows the ability to use modeled Ar to derive NCP from total O2 records. This approach has the potential to unlock critical carbon flux estimates from historical and autonomous O2 measurements in the global oceans.

Marine microbes play key roles in driving patterns of important biogeochemical processes including primary production across the global ocean. Despite the importance of such interactions between the marine microbial community and ocean biogeochemistry, oceanographers have yet to attain a deep understanding of the ecological mechanisms underlying these connections. Due to the vast scale of ocean ecosystems, however, large-scale yet high-resolution surveys are necessary to uncover specific relationships between biology and elemental cycling for more detailed study.

With this need in mind, this dissertation takes advantage of recent advances in both underway techniques to measure in situ biogeochemical rates—most notably the dissolved O2/Ar method for measuring net community production (NCP)—as well as molecular sequencing methods to directly investigate relationships between marine microbial community structure, productivity, nitrogen (N2) fixation, and nutrient availability across large ocean regions. At the same time, this work also improves our understanding of the O2/Ar technique by evaluating its performance and key assumptions in a dynamic upwelling environment and by presenting recommendations to improve the accuracy of productivity estimates generated using this approach.

Presenting data and measurements from the most comprehensive survey of marine microbial community structure and patterns of productivity and N2 fixation in the western North Atlantic to date, this manuscript highlights intriguing connections between regional peaks in productivity and N2 fixation, the mixotrophic algae Chrysophyceae and Aureococcus anophagefferens, and Braarudosphaera bigelowii, a eukaryotic host organism for N2-fixing bacteria. In addition, we report a strong negative relationship between eukaryotic marine microbial diversity and productivity across the region. We further highlight the importance of considering diel cycles of productivity/respiration, other non-steady-state conditions, and vertical fluxes of O2/Ar when calculating and interpreting NCP rates obtained from surface O2/Ar measurements. Ultimately, these findings contribute to our ability to evaluate community production using surface ocean dissolved gas measurements and provide important insights into patterns of marine microbial activity and community structure into the western North Atlantic.

Oceanic net community production (NCP), defined as photosynthesis in excess of respiration, lowers the CO2 concentration at the ocean surface and in the process regulates atmospheric CO2 levels on seasonal to glacial-interglacial time scales. The magnitude of oceanic NCP, and the regulating factors are however poorly constrained. This dissertation aims to derive estimates of the large-scale distribution of NCP and to explore the mechanisms driving this variability, at regional scales (Western Antarctic Peninsula; Chapter 2), basin scales (Southern Ocean, Chapter 3), and global scale (world oceans, Chapter 4).

In Chapter 2, we use remotely sensed properties and in-situ observations of O2/Ar-NCP from 2008 to 2014 to explore the interannual variability in NCP at the Western Antarctic Peninsula. We find that annual NCP in the shelf and coastal regions is up to eight times higher than in offshore regions, with hotspots observed around canyons. The interannual variability in annual NCP observed in the region is likely controlled by the iron supply from subsurface or horizontal advection.

In Chapter 3, we use remotely sensed properties to investigate the impact of mixed-layer dynamics on NCP in the Southern Ocean. We find that, as expected, NCP is largely controlled by light availability on seasonal time scales. On intra-seasonal time scales, a deepening of mixed layer increases NCP which we attribute to increased nutrient availability. On interannual time scales, NCP correlates with a host of parameters (i.e., stratification, wind kinetic energy, and mixed layer depth), but not to mixed layer depth (MLD). Although we do not observe a secular trend in NCP for the entire Southern Ocean, NCP increases (decreases) in the Atlantic (Pacific) sector over the 1997-2014 period. Overall, our results show that the driving mechanisms behind the NCP distribution vary as a function of the temporal and spatial scales under study.

In Chapter 4, we derive two global satellite NCP algorithms using O2/Ar measurements and the machine-learning methods of genetic programming and support vector machine. Our new algorithms are comparable to other algorithms in their prediction accuracy and magnitude of global biological carbon fluxes at the ocean surface, but predict a more spatially uniform NCP distribution for the world’s oceans.

In Chapter 5, we develop a mechanistic model of the carbon export potential from the mixed layer as a function of light availability. We show that the model is remarkably consistent with in-situ observations of O2/Ar-derived NCP and export production estimates from 234Th and sediment traps. Our model suggests that carbon export production in the Southern Ocean is likely co-limited by light and nutrient availability.

We end with a discussion of future projects in the concluding chapter 6.