Figure 8: Experimental results suggest that the dominant
process through which hypoxia affects sediment organic carbon differs
between weekly and multiannual timescales. Left: in microcosm
incubations, short-term (weeks) periods of hypoxia led to increased DOC
and aqueous Fe, while decreasing sediment OC. On a whole-ecosystem
scale, hypolimnetic Fe was closely correlated with oxygen
concentrations, and short periods of hypoxia decreased both Fe-OC and OC
in sediment. Consequently, Fe-OC protection appears to be a more
dominant control on sediment OC sequestration than respiration on short
timescales. Right: two years of summer hypoxia in FCR led to increased
OC in sediment on a whole-ecosystem scale, suggesting that respiration
may be a more dominant control on OC sequestration than protection by Fe
on this timescale.
4.1 Short-term periods of hypoxia
lead to release of Fe-protected OC and decrease total sediment OC
Both whole-ecosystem and microcosm experiments suggest that short-term
periods (i.e., weeks) of hypoxia can dramatically alter coupled OC and
Fe cycling. Whole-ecosystem experiments revealed changes in hypolimnetic
DOC and Fe, sediment OC, and sediment Fe-OC associated with water-column
oxygenation (Figures 3, 4, 8), while microcosm incubations showed clear
differences in aqueous Fe, and DOC and sediment OC among treatments
(Figures 6, 7). The magnitude of these effects was substantial: on a
whole-ecosystem scale, two weeks of hypoxic conditions decreased both
sediment Fe-OC and total sediment OC by a mean of 30%. Declining Fe-OC
and total OC in sediment, as well as concomitant increases in Fe and OC
in overlying water, are consistent with the expectation that hypoxia
causes reductive dissolution of Fe(III) in Fe-OC complexes, releasing
soluble Fe and DOC on day to week scales (e.g., Carey et al., 2018; Pan
et al., 2016; Patzner et al., 2020; Peter et al., 2016; Skoog &
Arias-Esquivel, 2009; Figures 1, 8).
Our results contribute to an accumulating body of evidence that
short-term fluctuations in oxygen concentrations may have important
effects on OC storage in soils and sediment. Previous research has shown
that recently-formed Fe-OC associations may be particularly prone to
hypoxic release, and reduction of Fe(III) in Fe-OC compounds can
increase OC respiration rates during hypoxic conditions (Chen et al.,
2020). As a result of these and other processes, periodic fluctuations
in oxygen conditions may sustain or stimulate respiration rates relative
to both constant oxic and constant hypoxic conditions (Bastviken et al.,
2004; Huang et al., 2021). Here, we found substantial decreases in
sediment OC and Fe-OC following two weeks of hypoxia, with restored OC
and Fe-OC after two weeks of oxic conditions (Figure 3, S3), suggesting
that at least a fraction of the sediment Fe-OC pool is sensitive to
short-term changes in oxygen concentrations in overlying water on a
whole-ecosystem scale.
While sediment Fe-OC responded to oxygenation on a whole-ecosystem
scale, Fe-OC did not vary significantly among oxygen treatments in
experimental incubations. We expect that this difference derives at
least in part from the sediments we analyzed: whole-ecosystem samples
were composed of the top 1 cm of sediment from sediment cores, while
sediment for the experimental incubations was sampled using an Ekman
grab, and therefore included deeper layers of sediment. In soil, deeper
horizons are thought to have more stable Fe-OC aggregates (Rumpel &
Kögel-Knabner, 2011). Our results suggest that the same pattern may be
true in sediments, resulting in burial of stable Fe-OC compounds in
deeper sediments over time.
Formation and dissociation of Fe-OC complexes are two of many ways in
which Fe and OC are impacted by hypoxia; both Fe and OC can also respond
independently to changes in DO concentrations. Fe is oxidized from
Fe(II) (soluble) to Fe(III) minerals (insoluble) both biotically and
abiotically under oxic conditions (Kappler et al., 2021). Increased
microbial biomass may be partially responsible for the increase in
sediment OC under oxic conditions in experimental microcosms, as we
observed the formation of orange (likely Fe-oxidizing) biofilms on top
of the sediment layer in oxic microcosms (Figure S1). Fe reduction is
also often associated with an increase in pH, which may increase the
solubility of OC (Tavakkoli et al., 2015). However, pH did not differ
consistently between microcosm treatments in this study and remained
circumneutral on a whole-ecosystem scale (Figure S6 and S7). While these
alternative mechanisms likely play a role in Fe and OC release, the
decrease in Fe-OC and total OC following inactivation of the oxygenation
system in FCR suggests that, at least for some surficial sediments,
short (~2 week) periods of hypoxia can cause Fe-OC
complexes to dissociate and decrease sediment OC burial on a
whole-ecosystem scale.
4.2 Over multiannual timescales,
OC respiration rates play a greater role than Fe-OC in determining the
net effect of hypoxia on sediment OC content
Over multiannual timescales (2019–2021), exposure to seasonal hypoxia
increased the amount of OC in sediments from FCR by 57% without
changing the amount of Fe-OC (Figure 5). This clearly contrasts with
short-term experimental results, which showed decreased OC content
following short periods of hypoxia (section 4.3). While many factors
could affect OC and Fe-OC over multiannual timescales, the fact that no
comparable effects were seen in the unoxygenated reference reservoir
(BVR) suggests that these changes may be attributed to changes in DO
concentrations in overlying water. Over two years of summer hypoxia, OC
levels in sediment from FCR increased to the extent that they were no
longer significantly different from the hypoxic reference reservoir
(Figure 5), suggesting that legacy effects of oxygen conditions on total
sediment OC may diminish after a two-year interval.
Increases in sediment OC content with increased hypoxic duration are
consistent with a reduction in sediment OC respiration rates under
hypoxic conditions (Carey et al., 2018; Hargrave, 1969; Walker &
Snodgrass, 1986). OC respiration rates decrease under hypoxic conditions
because OC must be broken down using alternative electron acceptors,
which produce a lower energy yield (Bastviken et al., 2003, 2004).
Previous work conducted in BVR found that CH4 was the
dominant terminal electron acceptor in the hypolimnion during hypoxic
summer conditions, and CH4 has one of the lowest energy
yields of alternate electron acceptors (McClure et al., 2021).
Consequently, as less OC is respired in hypoxic hypolimnetic water and
sediments, OC can accumulate more quickly in surficial sediments. Our
results suggest that over multiannual timescales, this process
(decreased respiration under hypoxic conditions) outweighs the
counteracting decrease in Fe protection of OC that we observed during
short periods of hypoxia.
Sediment Fe-OC content (per g sediment mass) did not significantly
change after two years of hypoxic conditions in FCR (Figure 5),
indicating that at least a fraction of these compounds are able to
withstand fluctuations in environmental redox conditions. Long-term
stability of Fe-OC complexes can be promoted by the formation of strong
chemical bonds between OC and mineral surfaces, and these bonds continue
to form over time (Kaiser et al., 2007). Likewise, weathering of Fe
increases the porosity of mineral surfaces and allows for OC protection
within pore spaces (Kaiser & Guggenberger, 2003; Kleber et al., 2005).
Decreased accessibility to microbial decomposition (e.g., through
burial) may further increase the ability of Fe-OC compounds to persist
over time (Kaiser & Guggenberger, 2003; Kleber et al., 2015). In FCR,
the history of oxic conditions (2013–2019) may have contributed to the
formation of particularly stable Fe-OC complexes in sediment, which were
then able to withstand two summers of hypoxia.
Importantly, much of the OC that accumulates under hypoxic conditions
does not end up being bound to Fe. This result may be due to Fe
oxidation state, as sorptive associations between DOC and Fe in sediment
are much less likely to form if Fe is in a reduced state (Fe(II); Nierop
et al., 2002). Because total OC increased following hypoxia and Fe-OC
did not change, Fe-OC as a percentage of sediment OC was significantly
lower after two years of hypoxia than before this hypoxic period.
Declines in the percentage of OC that is bound to Fe may have important
implications for ecosystem-scale carbon cycling, as OC that is
associated with Fe is comparatively more protected from respiration
(e.g., Chen et al., 2018, 2020; Hemingway et al., 2019; Kleber et al.,
2005). Increased stocks of OC that are not associated with Fe may
increase rates of methane production and OC release from the sediment to
the water column (e.g., Hounshell et al., 2021), and could increase
aerobic respiration rates under subsequent oxic periods (e.g., Chen et
al., 2020; Huang et al., 2021).
4.3 Substantial OC and Fe cycling
occurs at the sediment-water interface
Notably, the OC content of the top 1 cm of sediment was significantly
lower than that of settling particulate material in both FCR and BVR,
and nearly three times as much of this OC was bound to Fe in sediments
compared to settling material (Figure 2). These results imply a
substantial level of OC and Fe processing at the sediment-water
interface, and they reinforce previous research in suggesting that the
sediment-water interface is a hotspot for biogeochemical cycling
freshwater lakes and reservoirs (e.g., Dadi et al., 2017; Hanson et al.,
2015; Krueger et al., 2020).
From a mass-balance perspective, the difference in Fe-OC between
settling material and surficial sediments suggests that Fe-OC complexes
are either formed or preferentially preserved (compared to OC that is
not associated with Fe) in sediments. Preferential preservation of Fe-OC
is well-supported, as complexation with Fe has been shown to decrease OC
turnover rates across multiple ecosystems (Eusterhues et al., 2014;
Kaiser & Guggenberger, 2003; Kalbitz et al., 2005; Kleber et al., 2005;
Lalonde et al., 2012; Mikutta & Kaiser, 2011). However, the difference
in Fe-OC between settling material and surficial sediments likely also
results in large part from Fe-OC associations formed in sediment (e.g.,
through adsorption of organic matter onto existing Fe minerals and Fe-OC
complexes), as Fe(III) levels are much higher in sediments (e.g.,
Davison et al., 1991) and the composition of OC in sediments may be more
preferable for complexation with Fe. While we did not measure OC quality
in this study, we anticipate that settling material may have higher
autochthonous OC levels and be more rapidly respired, while sediment OC
may be enriched in allochthonous aromatic OC, which preferentially
associates with Fe (Kramer et al., 2012; Riedel et al., 2013; Shields et
al., 2016). Documenting changes in Fe-OC throughout the process of
sediment diagenesis enhances our understanding of OC sequestration, as
few if any previous studies have quantified the difference between Fe-OC
inputs and stocks in aquatic sediments.
4.4 High Fe-OC levels reflect
site-specific characteristics
On average, nearly one-third of surficial sediment OC was bound to Fe
(dithionite-extractable) across two years in FCR and BVR (Figure 5).
This percentage is far greater than that documented by Peter et al.
(2018), where Fe-OC comprised ≤11% of total sediment OC across five
boreal lakes. Furthermore, the levels of Fe-OC recorded here are higher
than the mean of 21.5±8.6% reported for a broad range of marine
sediments (Lalonde et al. 2012). With few other studies analyzing Fe-OC
in freshwater lakes and reservoirs to date, our analysis provides new
evidence that Fe-OC may play an important role in carbon sequestration
in some freshwaters.
Differences in the percentage of organic matter that is bound to Fe may
result from differences in overlying DO concentrations, as described
throughout this study, as well as a number of other factors. For
example, increasing ratios of Fe:OC and increasing absolute
concentrations of Fe and OC can all increase the amount of Fe-OC
coprecipitation (Chen et al., 2016; Kleber et al., 2015 and references
therein). Likewise, differing Fe forms, OC quality, and pH may also
impact the formation and stability of Fe-OC complexes (Curti et al.,
2021; Kaiser et al., 2007; Kaiser & Guggenberger, 2003; Kleber et al.,
2015); these differences may derive from contrasting geology, catchment
vegetation, and trophic status, among many other factors. However, none
of these factors fully explain differences in Fe-OC content between
sites studied to date (e.g., Lalonde et al., 2012; Peter & Sobek,
2018).
Despite having higher Fe-OC levels (as a percentage of total sediment
OC) than most aquatic sediments studied to date, other sediment
characteristics in FCR and BVR are within the range of those measured in
other locations. FCR and BVR have much lower sediment OC content than
the boreal lakes analyzed by Peter and Sobek (2018; 14–38% of sediment
mass), but higher sediment OC than the primarily marine sediments
analyzed by Lalonde et al. (2012; 0–7% of sediment mass). Fe
concentrations are high in sediment from FCR and BVR, with a mean of
53,466 mg/kg dry weight (Krueger et al., 2020). However, Peter and Sobek
(2018) observed low Fe-OC as a percentage of sediment OC (µ=6.7%) in
one extremely high-Fe lake (Övre Skärsjön; 226,172 mg/kg reducible Fe in
sediment). Likewise, pH in FCR and BVR is circumneutral (Figure S7),
well within the range of 5.4–7.6 reported by Peter and Sobek (2018),
and both Peter and Sobek (2018) and Lalonde et al. (2012) included a
range of oxic and hypoxic sediments in their analyses. These
observations from a range of aquatic sediments suggest that
site-specific characteristics associated with catchment geology, water
residence time, OC and Fe input rates, OC quality, and Fe mineral forms
play an important role in determining the percentage of sediment OC that
is bound to Fe. Understanding the controls on Fe-OC in freshwater
sediment will require Fe-OC characterization at a greater number and
diversity of lakes and reservoirs. Such research will be essential to
understanding how freshwater OC sequestration may be affected by global
changes in Fe concentrations (Weyhenmeyer et al., 2014), water
temperatures (Dokulil et al., 2021; O’Reilly et al., 2015), and pH
(Garmo et al., 2014; Gavin et al., 2018; Stoddard et al., 1999), among
other factors.
5. Conclusions
Results from this study help reconcile previous Fe-OC research and shed
light on how declining oxygen concentrations may impact the role of
lakes and reservoirs in the global carbon cycle. Research across
terrestrial soils and marine sediments has provided contradictory
evidence that Fe-OC complexes are (1) readily dissociated under hypoxic
conditions and (2) capable of promoting sediment OC burial into deeper
(hypoxic) layers over the course of decades to millennia. Here, we find
that the timescale of analysis plays a critical role in understanding
the net effect of hypoxia on sediment OC and Fe-OC. Specifically, a
portion of the Fe-OC pool in surficial sediment is highly responsive to
hypoxia in overlying water on a weekly timescale, resulting in decreased
sediment OC. However, over longer timescales, the decrease in OC that
results from dissociation of Fe-OC complexes is outweighed by the
increase in sediment OC that results from slower respiration rates under
hypoxia. At both timescales, our results reinforce that Fe may serve as
an important control over OC cycling and sediment preservation of OC in
some freshwater ecosystems. As the duration of hypoxia increases in
lakes and reservoirs (Jane et al., 2021; Jenny et al., 2016), our
results suggest that OC dynamics will respond non-linearly. While short
periods of hypoxia may decrease OC burial, increasing prevalence and
duration of hypoxia over multiannual timescales has the potential to
increase OC burial in freshwater sediment, intensifying the role of
freshwaters as sinks in the global carbon cycle.
Acknowledgements
We are grateful for ongoing partnerships with the Western Virginia Water
Authority that facilitated this research. Additionally, we thank the
Virginia Tech Reservoir Group for helping to collect these data,
particularly Adrienne Breef-Pilz for organizing the 2021 field season
and helping to collect sediment cores in 2021, James Maze and Dexter
Howard for helping to collect sediment cores in 2019, Ryan McClure for
leading CTD data collection before 2019, and Alex Hounshell for
organizing the 2019 field season. We thank Chip Frazier and Jeb Barrett
for providing access to laboratory equipment and Jeff Parks in the VT
ICP-MS lab for analyzing Fe samples. This work is supported by the
Institute for Critical Science and Applied Technology at Virginia Tech
and the U.S. National Science Foundation (NSF) foundation grants
DGE-1651272, DEB-1753639, SCC-1737424, DBI-1933016, and
DBI-1933102.
Author Contributions
ASL conceptualized this study with CCC and MES. ASL performed chemical
Fe-OC extractions, analyzed data, developed figures, and wrote this
manuscript with MES, MEL, and CCC. BRN led analytical chemistry methods
development and performed DOC analyses. NWH and MES oversaw
whole-ecosystem Fe sampling, analysis, and data collation. AD helped to
design and run the microcosm experiment and process sediment samples.
HLW collated and processed DOC and YSI data and reviewed code for this
manuscript. CCC oversaw whole-ecosystem experiments. All authors edited
and approved the final manuscript.
Open Research
All data used in this study are available in the Environmental Data
Initiative (Carey et al., 2021; Carey, Lewis, Gantzer, et al., 2022;
Carey, Lewis, McClure, et al., 2022; Carey, Wander, McClure, et al.,
2022; Lewis et al., 2022; Lewis, Schreiber, et al., 2022; Schreiber et
al., 2022). Code to reproduce results in this manuscript is available in
a Zenodo repository (Lewis, 2022).