Microbial metabolite fluxes in a model marine anoxic ecosystem

Abstract

Permanently anoxic regions in the ocean are widespread and exhibit unique microbial metabolic activity exerting substantial influence on global elemental cycles and climate. Reconstructing microbial metabolic activity rates in these regions has been challenging, due to the technical difficulty of direct rate measurements. In Cariaco Basin, which is the largest permanently anoxic marine basin and an important model system for geobiology, long‐term monitoring has yielded time series for the concentrations of biologically important compounds; however, the underlying metabolite fluxes remain poorly quantified. Here, we present a computational approach for reconstructing vertical fluxes and in situ net production/consumption rates from chemical concentration data, based on a 1‐dimensional time‐dependent diffusive transport model that includes adaptive penalization of overfitting. We use this approach to estimate spatiotemporally resolved fluxes of oxygen, nitrate, hydrogen sulfide, ammonium, methane, and phosphate within the sub‐euphotic Cariaco Basin water column (depths 150–900 m, years 2001–2014) and to identify hotspots of microbial chemolithotrophic activity. Predictions of the fitted models are in excellent agreement with the data and substantially expand our knowledge of the geobiology in Cariaco Basin. In particular, we find that the diffusivity, and consequently fluxes of major reductants such as hydrogen sulfide, and methane, is about two orders of magnitude greater than previously estimated, thus resolving a long‐standing apparent conundrum between electron donor fluxes and measured dark carbon assimilation rates.

Transport‐limited reactions in microbial systems

“…To test the applicability of the FRT limit to realistic microbial systems, we constructed an FRT model for the Cariaco Basin water column, depths 180–900 m. The model describes the spatiotemporal dynamics of multiple dissolved substances involved in microbial energy metabolism, and transported across space via turbulent (eddy) diffusion. Since lateral transport in the Cariaco Basin is generally much faster than vertical transport, we assume lateral homogeneity, consistent with previous models for this system (Li et al, 2012; Louca, Astor, et al, 2019; Louca, Scranton, et al, 2019; Scranton et al, 1987). The model thus exhibits a single spatial dimension representing depth, though it accounts for changes in the basin's lateral surface area with depth (details in Experimental Procedures).…”

“…Our model focuses on microbially catalyzed reactions thought to be largely responsible for the nitrogen and sulfur cycles in the oxygen‐deficient part of the Cariaco Basin (Louca, Scranton, et al, 2019) and other similar marine anoxic zones (Chronopoulou et al, 2017; Hawley et al, 2014; Lam et al, 2009; Louca, Hawley, et al, 2016; Ulloa et al, 2012), including anaerobic ammonium oxidation to N 2 (anammox), oxidation of H 2 S to using (reduced to ) or using (reduced to N 2 ), aerobic oxidation of H 2 S to , oxidation of H 2 S to via dissimilatory nitrate reduction to ammonium (DNRA), the two nitrification steps (), and oxidation of methane using either O 2 , (reduced to ) or (reduced to N 2 ). To parameterize this model we only need the time‐ and depth‐dependent vertical eddy diffusivity (Figure S3C, taken from Louca, Astor, et al, 2019), the boundary concentrations at the top and bottom of the considered depth interval, the reaction stoichiometries, and the initial concentration profiles in January 2001 (if brief transients are of interest), all of which would be required in conventional biogeochemical models as well; in contrast to conventional models, however, our FRT model does not require any knowledge of microbial species composition, population dynamics or reaction kinetics. The model also makes no assumption regarding proximity to steady state.…”

“…To determine the chemical boundary concentrations at the top and bottom and to compare model predictions to data, chemical concentration data were bilinearly interpolated on a regular spatiotemporal grid (Figure 3A–F). Previously estimated vertical eddy diffusivities were obtained from (Louca, Astor, et al, 2019). Temperature, salinity and diffusivity are shown in Figure S3.…”

“…Lateral transport was assumed to be much faster than vertical transport, thus leading to a laterally homogeneous model with a single spatial axis corresponding to depth, similarly to previous models of this system (Li et al, 2012; Louca, Astor, et al, 2019; Louca, Scranton, et al, 2019; Scranton et al, 1987). Vertical transport of dissolved metabolites was assumed to be largely via eddy (turbulent) diffusion; hence, the advection speed was set to zero.…”

Anomalous δ13C in Particulate Organic Carbon at the Chemoautotrophy Maximum in the Cariaco Basin

“…Modeling geochemical balances that are consistent with measured rates of dark carbon fixation has been challenging (Li et al, ). However, recent biogeochemical flux models yield much better agreement between supply and demand for reactants (Louca, Scranton, et al, , Louca, Astor, et al, ).…”