Constraining the mechanisms controlling organic matter (OM) reactivity and,
thus, degradation, preservation, and burial in marine sediments across
spatial and temporal scales is key to understanding carbon cycling in the
past, present, and future. However, we still lack a detailed quantitative
understanding of what controls OM reactivity in marine sediments and,
consequently, a general framework that would allow model parametrization in
data-poor areas. To fill this gap, we quantify apparent OM reactivity (i.e.
OM degradation rate constants) by extracting reactive continuum model (RCM)
parameters (
Organic matter (OM) buried in marine sediments represents the largest
reactive reservoir of reduced carbon on Earth (e.g.
Hedges, 1992). The majority of OM that reaches modern global ocean seafloors
originates from contemporary primary productivity (PP) in terrestrial
(global net primary productivity, NPP
Environmental factors influencing organic matter reactivity in marine sediments.
Traditionally, apparent (i.e. estimated representation of) OM reactivity has been determined through laboratory-based incubation experiments in shallow (few centimetres deep) and mixed sediments (e.g. Dai et al., 2009; Grossi et al., 2003; Westrich and Berner, 1984) or integrated model–data approaches in deeply buried sediments (hundreds of centimetres to metres deep and hundreds to thousands and millions of years old) (e.g. Boudreau and Ruddick, 1991; Middelburg, 1989; Wehrmann et al., 2013), using reaction–transport models (RTM). The latter are powerful tools since they allow the extraction of quantitative measures of OM reactivity from comprehensive multi-species porewater datasets integrated over greater sediment depths and ages (e.g. Middelburg, 2019). However, due to our limited quantitative understanding of the environmental controls on OM reactivity, the application of RTM approaches to data-poor settings is still complicated by the difficulty in constraining model parameters.
The need for predictive frameworks that constrain OM degradation model parameters has been recognized since the early days of RTM. Based on the rationale that the reactivity of OM settling onto the sediment is controlled by the degree of degradation, and thus its residence time in the oxygenated water column, early efforts have explored correlations between OM reactivity and water depth, sedimentation rate, or OM fluxes (Boudreau, 1997; Boudreau and Ruddick, 1991; Müller and Suess, 1979; Tromp et al., 1995). However, the derived relationships were based on limited datasets, and a more recent compilation of global data did not reveal statistically significant correlations between OM reactivity and single depositional environmental descriptors (Arndt et al., 2013). However, the latter combined OM degradation model parameters derived from structurally different models and used observational data of different complexity, thus somewhat compromising the comparability of their findings. Nevertheless, it emphasized the role of a complex interplay of environmental controls in reactivity (Kayler et al., 2019; LaRowe et al., 2020b; Mayer, 1995; Schmidt et al., 2011; Zonneveld et al., 2010).
The efforts by Arndt et al. (2013) to develop a general predictive framework highlighted two critical issues: OM reactivity is variable within and across settings with no significative global trends, and there is a crucial need to resolve the lack of comparative studies. Therefore, we here use a common, integrated model–data approach to inversely determine the model parameters that control apparent OM reactivity from contemporary sediment depth profiles (Sect. 3) at 14 different sites from five different depositional environments (Sect. 2). OM degradation is formulated according to the reactive continuum model (RCM) (Boudreau and Ruddick, 1991), which represents the most suitable degradation model type to explore links between reactivity parameters and environmental controls (see Sect. 3.2 for details). Derived RCM parameters are discussed in the environmental context of depositional settings. We further complement the newly derived parameter set with a compilation of previously published RCM parameter values to explore the distribution of model parameters and provide first-order constraints on model parametrization in data-poor areas. Finally, we evaluate the control of OM reactivity on benthic cycling and fluxes by linking RCM parameters to estimated OM degradation rates and relative contributions of metabolic pathways, anaerobic oxidation of methane (AOM) coupled to sulfate reduction, and fluxes of dissolved species across the sediment–water interface (SWI) at each of the 14 studied sites.
Depositional environment geographic locations.
Our data–model study covers a wide range of depositional environments, from
coastal to shelf and slope settings characterized by widely different
environmental conditions (Fig. 1, Table 2). It comprises the following
benthic provinces (Seiter et al., 2004): northern European
margin (EUR1), NW Mediterranean margin off Rhone delta (EUR2), western
Arabian Sea (WARAB), Bering Sea (NWPAC), and SW Atlantic off La Plata River
(RIOPLATA). Most of those regions are characterized by predominantly
temperate climate regimes, except for the Arabian Sea region (arid) and the
Bering Sea area (snow/polar) (Chen and Chen, 2013). OM
characteristics also cover a broad spectrum, ranging from organic-rich
(TOC
Oceanographic context of depositional environments studied in the large-scale organic matter reactivity quantification.
NA: not available.
Here, we adopt an integrated model–data approach to determine inversely OM
reactivity parameters from contemporary total organic carbon (TOC) and
sulfate (
We benefit from previously published datasets to develop our large-scale OM
reactivity assessments. TOC and porewater
The Biogeochemical Reaction Network Simulator (BRNS)
(Aguilera et al., 2005; Regnier et
al., 2002) is an adaptive simulation environment that has been successfully
employed to reproduce and quantify diagenetic processes in marine sediments
across a wide range of depositional environments and timescales
(Dale et al.,
2008a; Thullner et al., 2009; Wehrmann et al., 2013). The BRNS is suitable
for large, mixed kinetic–equilibrium reaction networks
(Dale et al., 2009; Jourabchi,
2005; Thullner et al., 2009). The concentration depth profiles of solid and
dissolved species in marine sediments are calculated according to the
vertically resolved mass conservation equation of solid and dissolved
species in porous media (Berner, 1980; Boudreau,
1997):
Briefly, the implemented reaction network encompasses the most pertinent
primary and secondary redox reactions found in the upper layers of marine
sediments (e.g.
Aguilera
et al., 2005; Thullner et al., 2009; Van Cappellen and Wang, 1996; Wang and Van Cappellen, 1996). It
explicitly accounts for the heterotrophic degradation of OM coupled to the
consumption of oxygen (aerobic OM degradation), nitrate (denitrification),
sulfate (organoclastic sulfate reduction), and methanogenesis.
Additionally, the reaction network accounts for secondary redox reactions,
i.e. re-oxidation of reduced species produced during primary redox
reactions. It explicitly resolves nitrification, sulfide re-oxidation by
The model transport and reaction equations were solved sequentially
according to Regnier et al. (1998). The size of the model
domain was fixed at 1000 cm for all sites, except for the Bering Sea, in
which the model domain was extended to 1500 cm, due to the low
sedimentation rate assumed for this site (Table 2). This choice is based on
initial tests and ensures that the model domain covers the diagenetically
most active zone, thus reducing the influence of biogeochemical dynamics in
underlying sediments on biogeochemical dynamics within the model domain. The
model equations were solved on an uneven grid and with a time-adapting
time step (
OM is composed of a complex and dynamic mixture of compounds that are
distributed over a wide, continuous spectrum of reactivities. Thus, OM
degradation is described by the reactive continuum model (RCM)
(Boudreau and Ruddick, 1991), which assumes a continuous
distribution of OM compounds over the entire reactivity spectrum. The RCM
assumes a gamma distribution to describe the probability density function of
OM distribution,
Due to the rapid depletion of the most reactive compounds, the reactivity of
the bulk material decreases during degradation, reflecting the widely
observed reactivity decrease with burial time, depth, and age
(Boudreau and Ruddick, 1991; Middelburg, 1989).
This indicates that degradation of OM proceeds at different rates in
parallel. Interactions between different compounds or transformations of
compounds can change the reactivity of a given compound. While the RCM does
not explicitly account for such interactions and transformations, the overall OM
profiles take these interactions implicitly into account. Thus, OM compounds
are continuously and dynamically distributed over a range of reactivities
that capture the decrease in apparent reactivity with burial age and depth as
the most reactive compounds are successively degraded. The interplay of
Boundary conditions place the BRNS in the environmental context of each of
the study sites (Table 4). The boundary conditions are constrained based on
either site-specific measurements or alternatively published data if
direct observations were not available. Here, we assume a fixed boundary
concentration of OM at the SWI. Concentrations of
We use an inverse model approach to extract the optimal OM reactivity
parameter set
These limitations can be alleviated by using comprehensive, multi-component
observational datasets, because the inclusion of additional information
adds further constraints. However, such datasets are often not available.
For instance, the first efforts to quantify
The optimal parameter set was then determined for each site by assuming that
the rank of a parameter set depends on the similarity between simulated and
measured data. Best-fit parameter couples were found by minimizing the
misfit between model results and simulations. We quantified the similarity
between the simulated and observed TOC and
The best-fit RCM parameters were inversely determined by first running the
model for each site with a set of
The apparent reactivity of bulk OM
The integrated data–model analysis yielded a comprehensive picture of OM
reactivity parameters (
Total organic carbon data–model best fit:
We inversely determined the set of RCM parameters
Sulfate data–model best fit:
Summary of model elements incorporated in the BRNS.
The inversely determined OM reactivity parameters
Parameter
Frequency histogram of organic matter reactivity parameter
distribution on a global scale:
Site-specific transport parameters and upper boundary conditions implemented in the reaction–transport model.
Nevertheless, our results also reveal exceptions. Particularly high
Parameter
Summary of statistical tests for the best-fit
Inversely determined parameters
Apparent organic matter reactivity
The Arabian Sea region (Fig. 5j–l) is characterized by the deposition of
highly reactive OM, the concentration of which rapidly decreases with
sediment depth (high
Benthic organic matter dynamics derived from the coupled model–data assessments across the depositional settings.
n.d.: not determined (oxygen and nitrate benthic fluxes not determined for locations where SWI concentrations of those TEA were assumed to be zero).
Between those two endmember cases (high-reactivity Arabian Sea and low-reactivity Skagerrak), sediments from other sites generally reveal small
changes in the continuous OM–
Parameter
Depositional environment controls on organic matter reactivity
parameters
Despite the large scatter, broad trends emerge that generally agree with
previously proposed relationships
However, in agreement with the previous global assessment
(Arndt et al., 2013), we do not find any statistically
significant global or regional correlation between the reactivity parameters
(
However, Fig. 6b indicates that water depth seems to be a useful and easily
accessible first-order proxy for this complex interplay of environmental
controls on OM reactivity. Thus, it may serve as an acceptable predictor for
first-order global reactivity patterns in the absence of more suitable
information. At the global scale, inversely determined apparent OM
reactivity for the deep, well-oxygenated sites (e.g. Bering Sea and
Argentine Basin) fall on the lower end of the reactivity spectrum, whereas
shallow coastal environments, such as the northern European region, reveal
generally higher apparent OM reactivities. In addition, within the northern
European region and along the Rhone delta transect, parameter
The weak water depth–reactivity trends and numerous exceptions at both
global and regional level caution against the uncritical use of simplistic
Both quantity and quality of the deposited OM exerts an important control on benthic–pelagic coupling processes with potential implications for global biogeochemical cycles and climate. While the influence of OM quantity on benthic–pelagic exchange along the global hypsometry has been previously investigated (Krumins et al., 2013; Soetaert et al., 1996, 1998; Thullner et al., 2009), the role of reactivity on benthic–pelagic fluxes remains less constrained.
In agreement with previous findings (Henrichs, 1992),
depth-integrated OM degradation rates (
Benthic organic matter degradation dynamics.
OM reactivity does exert an important impact on the relative contributions
of metabolic pathways (Fig. 7). Overall, rates of OM degradation coupled to
consumption of oxygen (aerobic OM degradation pathway) and nitrate
(denitrification) make a small relative contribution (
It has been suggested that anoxic metabolic pathways represent significant
pathways of OM degradation
(e.g. Bowles et al.,
2014; Jørgensen and Kasten, 2006; Thullner et al., 2009). Our results
show that sulfate reduction is the main oxidative pathway in regions
characterized by high apparent OM reactivity (
We do not consider the degradation pathways mediated by metal oxides due to a lack of SWI data that would allow constraint of boundary conditions of those processes. This might result in a slight overrepresentation of sulfate reduction (e.g. at the Skagerrak; Canfield et al., 1993; Rysgaard et al., 2001). It has been demonstrated that metal oxide pathways represent a minor contribution to total OM heterotrophic degradation on a global scale (Thullner et al., 2009), although in continental margin sediments iron hydroxide reduction may represent an important OM degradation pathway (Beckler et al., 2016).
AOM occurs at the sulfate–methane transition zone (SMTZ), which can vary in
sediment depth from centimetres to hundreds of metres as a function of
different environmental controls, such as OM quantity and quality,
sedimentation rate, active fluid flow, or microbial growth dynamics
(Puglini et al.,
2020; Regnier et al., 2011). Reflecting the diversity of the depositional
environments studied here, we observe a broad range of SMTZ depths (Fig. 8a–c). Generally and in agreement with previous findings
(Egger et al., 2018; Regnier et al.,
2011), we observe a deepening of the SMTZ with decreasing sedimentation
rates in addition to an important control of OM reactivity on the depth of
the SMTZ (Fig. 8b) and the depth-integrated AOM rates (
Anaerobic oxidation of methane (AOM) patterns as a function of
apparent organic matter reactivity and depositional rates.
Benthic fluxes of dissolved species
The lowest
Model results suggest that apparent OM reactivity also influences nutrient
recycling at the seafloor. Thus, we estimate benthic–pelagic fluxes of
dissolved species (
Apparent OM reactivity broadly controls the spatial patterns of
benthic–pelagic fluxes, as well as the relative importance of different
transport processes (Glud, 2008; Bourgeois et al.,
2017). Overall, molecular diffusion is the main transport pathway (Fig. 9f–j). However, the relative importance of bioirrigation increases at
environments characterized by low deposition of OM and less reactive OM.
This occurs because of the relative decrease in diffusive fluxes as a
consequence of weak concentration gradients across the SWI. Bioturbation
fluxes are generally low and reflect our assumption of depth-dependent
Overall, high apparent OM reactivity drives high sedimentary uptake fluxes
of dissolved TEA (Table 6; Fig. 9a–c). Sulfate largely dominates the
benthic TEA uptake, particularly at sites where oxygen and nitrate are
unavailable (Fig. 9c). At the Rhone proximal zone, unusually high
sedimentation rates (Pastor et al., 2011) deliver fresh
and highly reactive OM
(Cathalot et
al., 2010; Pruski et al., 2015), supporting high OM fluxes and degradation
rates (Fig. 7a; Table 6). Consequently, intense oxygen consumption occurs at
the SWI (Rassmann et al., 2016)
driven by both aerobic degradation of OM and re-oxidation of reduced
species. Similarly, in the Arabian Sea sediments below the OMZ the high
input of fresh phytoplankton debris
(Cowie,
2005; Koho et al., 2013; Rixen et al., 2019; Vandewiele et al., 2009) and
potentially chemoautotroph biomass (Lengger et al.,
2019) associated with intense degradation rates in the uppermost sediment
layers (Luff et al., 2000) result in a high relative
contribution of aerobic degradation of OM and denitrification to total OM
oxidation. Consequently, both benthic oxygen and nitrate uptake fluxes are
comparably high. In contrast, at regions characterized by deposition of less
reactive OM (
Benthic fluxes play a crucial role in nutrient recycling. In the Arabian Sea region, large phytoplankton blooms are associated with monsoon conditions, which result in upwelling of nutrient-rich bottom water (SW monsoon) and deepening of the mixed layer (NE monsoon) (Cowie, 2005; Luff et al., 2000; Rixen et al., 2019). Similarly, across the northern European margin, spring and summer diatom blooms are common (Fleming-Lehtinen and Laamanen, 2012; Jensen et al., 1990; Karlson et al., 1996; Lomstein et al., 1990; Wiltshire and Manly, 2004) and maintained by benthic nutrient fluxes. Additionally, bottom water upwelling and nutrient recycling are important mechanisms sustaining spring and summer PP at the Bering Sea (Coyle et al., 2008). Benthic ammonium and phosphate recycling fluxes mirror the reactivity and OM degradation patterns (Fig. 9d–e). The largest fluxes are determined for the Arabian Sea region, as well as the shallow Aarhus Bay and Arkona Basin (Table 6). In contrast, the lowest nutrient recycling fluxes are observed in regions characterized by the deposition of less reactive OM (phosphate fluxes are negligible at Skagerrak sites S10 and S13), where heterotrophic degradation rates are slow.
We developed and applied an inverse modelling approach to quantify apparent
OM reactivity (i.e. parameters
Our findings corroborate previous results
(Arndt et al., 2013;
Boudreau and Ruddick, 1991; Forney and Rothman, 2012; Middelburg, 1989) that
the RCM parameter
Additionally, results indicate that the large variability in apparent OM
reactivity is linked to a combination of multiple environmental drivers that
control the quality of OM delivered to sediments and the timescale of
settling and burial. Therefore, our findings contribute to the notion that
apparent OM reactivity is controlled by a complex and dynamic interplay of
environmental drivers, which are measurable and allow the quantification of
Finally, results also show that OM reactivity exerts a dominant control on the redox zonation of the sediment, the depth of the SMTZ, depth-integrated AOM rates, and benthic–pelagic exchange fluxes, indicating that these processes could serve as predictor variables for OM reactivity. In contrast, depth-integrated OM degradation rates are largely controlled by the magnitude rather than the quality of OM deposition to the sediment.
The model code is available at the GitHub repository
(
The supplement related to this article is available online at:
FSF, SA, and RDP designed this study. PAP developed and implemented changes to model description (multi-G approximation of the RCM). SK, BBJ, JR, CR, ST, and HS provided porewater and sediment data to inform model experiments and contributed to discussing and interpreting the data and model results. FSF compiled data, ran all model simulations, and performed data–model integration. FSF wrote the manuscript with significant contributions from SA. All co-authors edited and approved the manuscript.
The authors declare that they have no conflict of interest.
Publisher’s note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Felipe S. Freitas was supported by a PhD scholarship from the Science without Borders programme (Ciência sem Fronteiras), sponsored by the CAPES Foundation within the Ministry of Education, Brazil, grant number 9999.009541/2013-06. Felipe S. Freitas also thanks the UKRI Natural Environment Research Council (NERC) for funding, grant number NE/P006493/1 (Changing Arctic Ocean Seafloor – ChAOS). Richard D. Pancost acknowledges the advanced ERC grant “The greenhouse earth system” (T-GRES, project reference 340923) and the Royal Society Wolfson Research Merit Award. Sandra Arndt and Philip A. Pika acknowledge funding from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement number 643052 745 (C-CASCADES). Shaun Thomas was funded by a President's Scholarship, Cardiff University. Christophe Rabouille and Jens Rassmann were supported by the INSU/EC2CO programme MissRhoDia and by the French state programme “Investissement d'avenir” run by the National Research Agency (AMORAD project ANR-11-RSNR-0002). The authors also thank Bernard Boudreau and the anonymous reviewer as well as associate editor Marilaure Grégoire for their constructive comments on early versions of the paper. Finally, we thank co-editor in chief Steve Bouillon for granting us a discount on the APC.
This research has been supported by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (grant no. 9999.009541/2013-06).
This paper was edited by Marilaure Grégoire and reviewed by Bernard Boudreau and one anonymous referee.