Salinity reduces coastal marsh respiration more than photosynthesis

A paradigm in carbon cycling science predicts that sea-level rise will enhance carbon accumulation in an apparent negative carbon-climate feedback1,2. However, ecosystems exposed to combinations of stressors and subsidies – such as saltwater intrusion and sea-level rise – may adapt, transition to an alternative state, or experience a decline in functions, such as carbon storage, thereby altering their response trajectories to environmental changes3,4. Climate change is increasing salinity in coastal ecosystems worldwide yet the effects on ecosystem metabolism remain uncertain4-8. Here, we synthesized gross ecosystem productivity (GEP), ecosystem respiration [CO2 and CH4 (ERCO2 and ERCH4)], and net ecosystem productivity (NEP) from diverse coastal marshes exposed to experimental additions and observational gradients in salinity. Increases in salinity generally caused decreases in median GEP, ERCO2, and ERCH4 but increases in GEP and NEP from ~5 to 10 ppt. Increased saltwater intrusion can stimulate or stress wetlands based on relative exposure and acclimation to increased salinities, and we detected positive NEP where salinity increases had greater negative effects on ERCO2 and ERCH4 than GEP. Although increases in NEP are detectable at low salinities, saltwater intrusion and climate-driven disturbances may reduce carbon storage capacity of coastal ecosystems as productivity declines toward higher salinities. [as carbon dioxide (ER CO2 ) and methane (ER CH4 )] uxes among coastal marsh ecosystems?, (2) Are there threshold concentrations of salinity that emerge when comparing differential responses to ambient and elevated salinity exposures from experimental salinity additions and along observational salinity gradients?, and (3) What plant-soil mechanisms might explain differential ecosystem metabolic responses to elevated salinities? We predicted that marshes exposed to experimental salinity additions would have higher sensitivity to increases in salinities compared to marshes occurring along observational salinity gradients due to differences in exposure and acclimation. trend or a more complicated pattern. We used Quantile ‐ Quantile (Q ‐ Q) plots of the residuals to evaluate if the appropriate distribution, identity, and log ‐ link were used. The Durbin-Watson statistic, with the durbinWatsonTest function in the “car” package 36 , was used to determine the AR(1) correlation coecient. We then specied the AR(1) correlation coecient and used the gam.check function to evaluate the convergence of the smoothness selection optimization and run diagnostic tests to evaluate whether the dimension choices were adequate. We then proceeded by updating the original GAM model. A P-value of 0.05 was used to determine the inclusion of salinity as an independent

salinity concentrations from 5 to 10 ppt before reaching a threshold (~10 ppt) and declining above higher salinity concentrations (Figs. 3D, 3H). Although salinity decreases absolute marsh productivity, low-to-moderate increases in salinity may impact photosynthesis and respiration different.
When comparing percent differences in median ecosystem metabolic rates between ambient and elevated salinity conditions from across all studies, we found GEP may be stimulated at lower increases in salinity (Fig. 4A) compared to persistent or steady declines in ER CO2 and ER CH4 (Figs. 4B, 4C). We found weak evidence for overall effects of increases in salinity on NEP (Fig. 4D), which is likely explained by high variability in percent differences in NEP.

Saltwater subsidies and stressors in coastal marshes
Despite contrasting observations from previous, site-speci c experiments, we nd that the response of coastal marsh ecosystem productivity to salinization is predictable and generalizable across fundamentally different marshes around the globe. Overall, we nd a nonlinear response where productivity is maximized at an increase in salinity of 10 ppt. Our synthesis has identi ed that NEP increases up to a salinity threshold range, beyond which ecosystem productivity decreases despite the combinations of plant species and hydroperiods surveyed. Given the diversity of sites selected in our synthesis (climate, geography, hydrology, dominant vegetation), site-speci c responses to elevated salinities were variable in magnitude and direction (Table S1). The subsidy-stress framework of perturbation theory predicts stress causes immediate and sustained decreases in ecosystem functions and subsidies cause an increase in function at lower levels of a subsidy until a threshold is reached 3 . Our synthesis -using median of long-term rates across a range of salinities -indicates that saltwater intrusion into coastal marsh ecosystems can act as a both a stress and a subsidy, stress to GEP and NEP may be overcome at low levels of salinity with apparent subsidies of saltwater, whereas salinity stress to ER CO2 and ER CH4 is sustained. Our synthesis also revealed that theoretical expectations for how climate drives metabolism, such as the temperature sensitivity ecosystem respiration and photosynthesis 11 , must be further expanded to better predict how effects of temperature-driven increases in sea-level rise and saltwater intrusion will impact ecosystem metabolic rates. We expected marshes exposed to experimental salinity additions would have higher sensitivity to changes in salinities compared to marshes occurring along observational salinity gradients due to differences in exposure and acclimation, but we observed sensitivity to increases in salinity across all marshes. Our ndings suggest that coastal wetland ecosystems may exhibit some adaptive responses to increases in salinity at low-to-moderate salinity concentrations.
Increased salinity appears to stimulate wetland net ecosystem productivity through greater negative effects on ecosystem respiration than photosynthesis (Figs. 3,4). Saltwater intrusion increases electron acceptors that oxidize reduced wetland conditions and likely stimulates ecosystem productivity in lower salinity wetlands, despite the negative effect of salinity stress on plant primary productivity 8 (Fig. 5). For example, sulfate increases with saltwater intrusion, and sulfate reduction is more energetically favorable that CO 2 reduction to CH 4 (methanogenesis) 8 . Ecological theory as well as results from our synthesis suggest that methanogens are displaced by sulfate reducers as the main heterotrophs degrading organic matter under low redox anaerobic conditions 8,12 (Figs. 3,4). Saltwater exposure increases soil porewater salinity, as well as alkalinity and concentrations of ions that can release nutrient subsidies in coastal wetlands 8,12,13 . As changes in ionization and alkalinization occur along with both subsidies and stressors, this likely explains our observations of apparent acclimation of GEP and ER CO2 to low-to-moderate increases in saltwater concentrations (Figs. 3, 4). Overall, salinity inhibited ER (both CO 2 and CH 4 ) more than GEP in both experimental salinity additions and observational salinity gradients, which resulted in stimulated NEP at lower salinity concentrations (Figs. 3,4).
Ecosystems that become salinized or oxygen-depleted from prolonged inundation can experience plant dieback, leading to loss of soil aggregates, increased erosion, and loss of production 14,15 . Wetland plant roots exposed to elevated salinity can lose biomass, especially of ne roots, which also results in soil elevation loss and reduced dissolved and particulate carbon export 16,17 . Further acceleration of soil carbon loss can occur from the production of labile detrital carbon from dying plant material, which can be exacerbated by concomitant changes in porewater and surface water nutrients and redox conditions 8,14 . Saltwater intrusion may alter dissolved organic carbon concentrations through physical interactions between cations and dissolved organic matter 18 or through loss of ne root biomass, soil compression, and release of dissolved organic carbon from soil 16,17 . Changes in carbon and nutrient concentrations may result in saltwater-induced changes in land-ocean biogeochemical uxes, as well as shifts in the landatmosphere exchange of greenhouse gases 4,19,20 . Soil organic matter processing is altered by exposure to seawater. For example, higher surface litter breakdown rates may occur with increased nutrient concentrations, release of labile forms of carbon, and increased surface and porewater oxidation 14,21 . Saltwater intrusion releases soil-bound nutrients through cation exchange, desorption, or increased rates of litter decomposition 4,8,13 , and nutrient subsidies may alleviate short-term, pulsed salinity stress associated with saltwater intrusion 38 (Fig. 5), but continued saltwater exposure decreases plant root biomass and leads to soil elevation loss 16 .

Adaptations to saltwater intrusion
How local ecosystems respond to increasing salinization will be in uenced by prior salinity exposure and salinity tolerance of existing plant and soil microbial communities. Plant and microbial compositional shifts can have larger effects on carbon cycling than the direct in uence of seawater.
Temperate tidal freshwater marshes are expected to transition to brackish or salt marshes where plant communities can migrate upstream with sea-level rise and saltwater intrusion 6,22 . In the transition from tidal freshwater herbaceous plants to salt-tolerant marsh graminoids, the lignin-rich grasses slow decay, despite a positive direct effect of salinity 21,22 . Marsh salinization generally decreases plant and microbial species diversity 21 , and the capacity and type of community shifts depend upon the relative amount of saltwater intrusion, the degree of sul dation (iron-sulfur buffering) and sensitivity to salinity, and legacies of land use 8,22-24 . In subtropical and tropical coastal wetlands, mangroves are expanding -latitudinally 25,26 and landward 27 with rising temperatures and rising seas -into marsh-dominated wetlands. This functional shift from herbaceous and succulent marshes to woody mangrove swamps is a major and global transformation that can provide stability and, in the case of landward migration, may provide a pathway toward enhanced accretion and carbon storage compared to coastal freshwater marshes 28 . A similar transition in abandoned, salinized agricultural lands, initially to woody shrublands 29 , could also provide greater opportunity for carbon storage. Where tidal freshwater forested wetlands transition to herbaceous marsh, the opposite trend is expected, particularly if there is not a landward movement of forested wetland habitat upstream 29 . In the face of uncertain effects on carbon storage with increasing saltwater intrusion and the potential for freshwater wetland peat collapse, coastal wetland persistence may increasingly depend on how quickly salt-tolerant plants (i.e., mangroves and salt marsh graminoids) are able to move landward into freshwater and upland ecosystems.

Saltwater intrusion and coastal carbon storage
Saltwater intrusion is increasing worldwide with rapidly rising sea levels, requiring a broader understanding of the consequences for coastal communities and ecosystems. Site-speci c studies have recorded varying productivity responses to elevated salinity despite decreases in soil carbon storage in coastal wetlands 4,[12][13][14][15][16][17] . Reduced plant biomass and growth, and increased soil organic matter processing and export will contribute to carbon losses. Saltwater intrusion can decrease carbon storage in non-acclimated coastal wetland ecosystems and can increase both carbon and nutrient export. To further identify how vulnerability to carbon loss will vary in coastal wetlands requires a stronger mechanistic understanding of how climate change and land-use legacies will interact with salinizing conditions across that vary in prior exposure and sensitivity to saltwater intrusion. As productivities decline above a threshold of salinity exposure as our synthesis reveals, coastal ecosystems become vulnerable to carbon losses. For example, coastal wetlands with low iron-sul de buffering and dominant plant species with low sul de tolerance, as well as those with high stores of carbon and nutrients that are readily exported and transported by land-atmosphere and land-ocean uxes, are likely among the most vulnerable to sealevel rise and saltwater intrusion 4,23 . Climate may also drive differential metabolic responses to saltwater intrusion and carbon uxes. Further understanding of how plant-soil-water chemistry drive variable responses to salinity in coastal wetlands is needed.
Continued exposure to saltwater intrusion and climate-driven disturbances may reduce carbon storage capacity of coastal wetland ecosystems as productivity declines at higher salinities. Development of vast belowground stores of carbon in coastal wetlands will depend upon continued production of plant biomass, growth, and accretion and burial of carbon despite changing abiotic stressors (e.g., salinity, sul dation, inundation, and disturbances) that otherwise may lead to non-linear responses characterized by abrupt and unpredictable degradation of vegetated habitat to open-water states 9,10,30 .
Forecasting whether coastal ecosystems continue to develop or transition to another vegetated state (e.g., freshwater marshes to mangroves, tidal freshwater forests to marshes) or to slowly or abruptly decline to open water requires a mechanistic understanding of how saltwater intrusion affects carbon production, storage, and export as sea levels rise 31,32 . This information is necessary to evaluate the longevity of carbon-climate feedbacks and to simplify the processes implemented in the next generation of Earth Systems models 33

Data Collection
We collated published and unpublished data from studies that measured net ecosystem exchange as carbon dioxide (CO 2 ) uxes (and methane, CH 4 , where available) from marshes that varied in surface water and porewater salinity concentrations (both manipulative experiments and observations).
We selected studies to minimize the in uence of exogenous covariates such as land-use, hydrologic, and water quality changes, but we included studies that had at least a year of data. We calculated minimum, maximum, and median values from time series of gross ecosystem productivity (GEP), ecosystem respiration (ER CO2 , ER CH4 ), and net ecosystem productivity [NEP = gross ecosystem productivity (GEP) -ecosystem respiration [as carbon dioxide (ER CO2 ) and methane (ER CH4 )]. We chose median metabolic rates from time series from each study as the most conservative metric to compare across the range of diverse marsh ecosystems. We included data that were measured at daily and monthly scales 34 and converted to daily rates (Table   S1). Whenever possible, we used measured porewater salinities. If porewater data were not available, we used measured surface water salinities. We calculated the change in salinity as the difference between ambient and elevated concentrations from experimental salinity additions, or as the differences in observed salinities between freshwater and brackish marshes or freshwater and saltwater marshes. We calculated the % difference of change in median values of GEP, ER CO2 , ER CH4 , and NEP using the following equation:

Data Analyses
To understand salinity effects on ecosystem metabolism (GEP, ER CO2 , ER CH4 , and NEP) across diverse marshes, we used generalized additive models (GAM) with the gam function in the "mgcv" package 35 in R. Generalized additive models use smooth functions to model non-linear functional relationships between predictors and the response. This approach separates linear trends from any general non-linear trends to determine if the signi cance of a smoothed variable is associated with a simple linear trend or a more complicated pattern. We used Quantile-Quantile (Q-Q) plots of the residuals to evaluate if the appropriate distribution, identity, and log-link were used. The Durbin-Watson statistic, with the durbinWatsonTest function in the "car" package 36 , was used to determine the AR(1) correlation coe cient. We then speci ed the AR(1) correlation coe cient and used the gam.check function to evaluate the convergence of the smoothness selection optimization and run diagnostic tests to evaluate whether the dimension choices were adequate. We then proceeded by updating the original GAM model. A P-value of 0.05 was used to determine the inclusion of salinity as an independent variable, and the explanatory powers of the nal models were compared using the R 2 statistic and the percent deviance explained. All statistical analyses were performed in R v.3.6.1 and RStudio v.   Table S1 Locations of studies and data used to synthesize cross-ecosystem productivity responses to elevated experimental salinity additions and observational salinity gradients. Median and % differences were calculated from published and unpublished values (indicated by *) of gross ecosystem productivity (GEP), ecosystem respiration [as CO 2 (ER CO2 ) and CH 4 (ER CH4 ) uxes], and net ecosystem productivity (NEP).     Predicted mechanisms of saltwater intrusion effects on ecosystem metabolism. Conceptual illustration of predicted mechanisms driving observed response patterns of gross ecosystem productivity (GEP), ecosystem respiration (ER) as CO2 (ERCO2) and CH4 (ERCH4), and net ecosystem productivity [NEP = gross ecosystem productivity (GEP) -ecosystem respiration [as carbon dioxide (ERCO2) and methane (ERCH4)] in marshes exposed to increasing saltwater.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.