Interactive comment on “ Global change effects on decomposition processes in tidal wetlands : implications from a global survey using standardized litter

Mueller et al. conducted decomposition experiments using tea bags based on a standardized approach developed by Keuskamp et al. (2013), across different marsh and mangrove sites in order to cover a gradient in temperature, indundation regime, etc. While such cross-ecosystem studies have a high potential, I feel the impact of this dataset in terms of new insights is relatively limited. The dataset can be published but I feel the impact of the conclusions should be toned down somewhat – the manuscript does not really deliver what the title suggests. The datasset should be publishable, but it needs a more critical discussion and should provide the readers with a more complete overview of the caveats and assumptions used in the TBI approach, so that the readers can better assess what can and cannot be deduced from these data.


Introduction
Tidal wetlands, such as marshes and mangroves, provide a wide array of ecosystem services that have been valued at approximately US$ 10,000 per hectare and year, making them some of the most economically valuable ecosystems on Earth (Barbier et al., 2011;Kirwan and Megonigal, 2013).
Yet, tidal wetlands are threatened and vulnerable ecosystems, experiencing pronounced loss though 75 global-change impacts, such as land use (Pendleton et al., 2012), and accelerated sea-level rise (SLR;Craft et al., 2009;Crosby et al., 2016).In recent years, carbon sequestration has increasingly been recognized as an ecosystem service of tidal wetlands (e.g.Chmura et al., 2003;Mcleod et al., 2011).Tidal wetlands are efficient long-term carbon sinks, preserving organic matter (OM) for centuries to millennia.Here, high rates of OM input (from both autochthonous and allochthonous 80 production) co-occur with reducing soil conditions and thus slow rates of decomposition, leading to organic carbon sequestration rates that exceed those of most other ecosystem types by orders of magnitude (Mcleod et al., 2011).At the same time, suppressed decomposition and the preservation of OM is a primary process by which many tidal wetlands gain elevation and keep pace with rising sea level (Kirwan and Megonigal, 2013).Consequently, global changes that might increase OM 85 decomposition in tidal wetland soils not only affect carbon sequestration, but also decrease ecosystem stability against SLR.It is therefore critical to identify global change factors that affect the transformation of organic inputs to stable soil OM (SOM) in tidal wetlands and to assess the magnitude of their effects.
There are multiple methods for assessing factors that influence carbon sequestration, 90 including direct measurements of plant production, organic carbon stocks, accretion, and decomposition rates.Litter-bag techniques assessing the weight loss of plant material over time are probably the easiest way to measure decomposition rates in situ and have been widely used since the 1960s (Prescott, 2010).Global-scale assessments of litter decomposition have been conducted as both meta-analyses (e.g.Zhang et al., 2008) and as inter-site studies along latitudinal gradients 95 (Berg et al., 1993;Torfymow et al., 2002;McTiernan et al., 2003;Cornelissen et al., 2007;Powers et al., 2009) in order to assess effects of climate parameters, mostly with focus on temperature and moisture gradients, on decomposition rate.Besides abiotic or climate effects, these studies could also identify litter quality itself as an important predictor for decomposition rate (Zhang et al., 2008).
100 Relationships between single climate or litter-quality parameters and decomposition rate often are not linear.Instead, complex interactions between litter quality and climate parameters seem to control litter decomposition (Zhang et al., 2008), creating challenges in separating climate from litter-quality effects and predicting the relevance of potential global-change drivers for decomposition rate.In order to separately assess environmental or climate effects on litter 105 decomposition at a global scale, it is therefore necessary to standardize litter quality in inter-site studies.However, implications of litter-decay data for carbon sequestration need to be considered cautiously, as the link among litter-decomposition rate, SOM formation, and ultimately carbon sequestration is not straightforward (Prescott, 2010;Cotrufo et al., 2013): Because plant tissues are not resistant to decay per se, it is critical to understand their biogeochemical transformation into 110 stable compounds that leads to the formation of SOM (i.e.stabilization) rather than understanding the pace at which early stages of decomposition proceeds (Prescott, 2010;Castellano et al., 2015;Haddix et al., 2016).Keuskamp and others (2013) developed an efficient approach for studying litter decomposition and OM transformation at a global scale, using commercially available tea as 115 standardized material.Their Tea Bag Index (TBI) approach is based on the deployment of two types of tea that considerably differ in their OM quality.The method allows for the determination of the decomposition rate constant (in the following referred to as decomposition rate or k), as in classic litter-bag approaches, and a stabilization factor (in the following referred to as stabilization or S), which describes the fraction of labile and decomposable OM that becomes stabilized during 120 deployment.
In the present study, we assessed the impacts of multiple global-change factors -warming, sea-level rise (SLR), salt-water intrusion, and coastal eutrophication -on both OM decomposition rate and stabilization in tidal wetland soils by conducting a worldwide field study using standardized litter.First, by covering a large temperature gradient of ΔT >15 °C across sites, we 125 aimed to capture temperature effects on OM decomposition and stabilization, thereby improving our process-level understanding on how global warming affects carbon turnover and ultimately sequestration in tidal wetlands.Second, by conducting paired measurements in both high-and lowelevated zones of tidal wetlands worldwide, we were aiming to gain insight into potential effects of accelerated relative SLR on carbon turnover.Despite the dominant paradigm that decomposition is 130 inversely related to flooding, the existing literature on hydrology and SLR effects on OM decomposition in tidal wetlands yields equivocal results, which is often due to the overriding effect of OM quality on decomposition rate (Hemminga and Buth, 1991;Kirwan et al., 2013;Mueller et al., 2016).Additionally, by expanding our study to include fresh and brackish sites, we anticipated to capture the effects of salt-water intrusion into brackish and fresh systems, which is likely to affect 135 decomposition processes in tidal wetlands (Morrissey et al., 2014).Specifically, high concentrations of dissolved sulfate in seawater, acting as an alternative terminal electron acceptor, can enhance anaerobic microbial metabolism in systems with lower salinity (Megonigal et al., 2004;Sutton-Grier et al., 2011).Lastly, we used the long-term ecological research site of the TIDE project in Massachusetts, US (Deegan et al., 2012) to experimentally assess both the effects of coastal 140 eutrophication and -with respect to SLR-driven increases in flooding frequency -the relevance of nutrient delivery through floodwater for the early stages of OM decomposition in tidal wetlands.

Study sites and experimental design
inundation regime across our study sites as these vary in absolute elevation and in elevation relative to mean high water.Finally, we included the long-term experimental site of the TIDE project in Massachusetts, US to assess effects of nutrient enrichment on litter-decomposition rate and stabilization.Through nitrate additions to the incoming tides on at least 120 days per year, nutrient enriched areas at the TIDE project site receive floodwater with 10-15 fold increased nitrogen (N) 160 concentrations compared to reference areas since 2004.From 2004-2010 also phosphate was added to the floodwater, however, this has been discontinued because creek water P concentrations are high enough to prevent secondary P limitation through N enrichment (details in Deegan et al., 2012;Johnson et al., 2016).
Decomposition rate and stabilization were measured by deploying tea bags in ten points per 165 zone or treatment within a site (n=10).Spacing between replicates within a zone or treatment was ≥2 m.However, as sites differed considerably in their areal extent, the distribution and thus spacing between points had to be adjusted to be representative for the given system.Temperature for the period of deployment was measured at site or temperature data was obtained from the online service of Accuweather (accuweather.com;accessed 12/25/2016) for locations within a distance of 15 km 170 to the site for most sites, but not further than 60 km for some remote sites.

Decomposition rate and stabilization measurements
Decomposition rate (k) and stabilization factor (S) were assessed following the Tea Bag Index protocol (Keuskamp et al., 2013).Briefly, at each point two nylon tea bags (200 µm mesh size), one 175 containing green tea (EAN: 8 722700 055525; Lipton, Unilever + PepsiCo, UK) and one containing rooibos (8 722700 188438, Lipton, Unilever + PepsiCo, UK), were deployed as pairs in ~8 cm soil depth, separated by ~5 cm.The initial weight of the contents was determined by subtracting the mean weight of 10 empty bags (bag + string + label) from the weight of the intact tea bag prior to deployment (content + bag + string + label).The tea bags were retrieved after an incubation time of 180 ~90 days.Upon retrieval, tea bags were opened, tea materials were carefully separated from clay particles and fine roots, dried for 48 h at 70°C, and weighed.
Calculations for k and S followed Keuskamp et al. (2013): Eq 1) W r (t) = a r e -kt + (1-a r ) 185 Eq 2) S = 1 -a g / H g Eq 3) a r = H r (1-S) 190 W r (t) describes the substrate weight of rooibos after incubation time (t in days), a r the labile and 1-a r the recalcitrant fraction of the substrate, and k is the decomposition rate constant.S describes the stabilization factor, a g the decomposable fraction of green tea (based on the mass loss during incubation) and H g the hydrolysable fraction of green tea.The decomposable fraction of rooibos tea is calculated in Eq 3 based on its hydrolysable fraction (H r ) and the stabilization factor S. With 195 W r (t) and a r known, k is calculated using Eq 1.
In accordance with Keuskamp et al. (2013), extractions for determination of the hydrolysable fractions of green and rooibos tea followed Ryan et al. (1990).However, instead of using Ryan's forest products protocol we conducted the alternative forage fiber protocol for the determination of the hydrolysable fraction.Briefly, 1 g of dried tea material (70°C for 24 h) was 200 boiled in cetyltrimethyl ammonium bromide (CTAB) solution (1 g CTAB in 100 ml 0.5 M H 2 SO 4 ) for 1 h (Ryan et al., 1990;Brinkmann et al., 2002).The extract was filtered through a 16-40-µm sinter filter crucible (Duran, Wertheim, Germany) using a water-jet vacuum pump and washed with 150 ml of hot water followed by addition of acetone until no further de-coloration occurred (Brinkmann et al., 2002).The remaining material was left in the sinter, dried for 12 h at 70°C, 205 cooled in a desiccator and weighed.20 mL of 72% H 2 SO 4 was added to the sinter and filtered off after an incubation of 3 h, followed by washing with hot water to remove remaining acid.The sinter was dried at 70°C for 12 h, cooled in a desiccator, and weighed to determine the non-hydrolysable fraction.Finally, the sinter containing the remaining sample was ignited at 450°C for 3 h in order to determine the ash content of the material.

210
In addition to the determination of the hydrolysable fraction, we measured total C and N contents of the tea material using an elemental analyzer (HEKAtech, Wegberg, Germany).The hydrolysable fraction of both green and rooibos tea was higher than reported in Keuskamp et al.
(2013) (Table 2).However, the determined C and N contents of the tea materials are in agreement with those reported in Keuskamp et al. (2013) (Table 2).Therefore, deviations from the 215 hydrolysable fraction as reported previously are likely due to the less conservative extraction assessment in the present study and not due to actual changes in the quality of the materials.

Data Analyses
For all across-site analyses, mean values of each site by elevation zone (or site by salinity class) 220 combination were used (N=51).Relationships between single parameters and litter decomposition are often not linear.Instead, critical thresholds seem to exist at which a certain predictor (e.g.mean annual temperature) becomes influential (Rothwell et al., 2008;Prescott, 2010).
In the first step of our data analysis, we therefore used classification and regression tree analysis (CRTA) to identify important predictors for k and S. CRTA is a non-parametric procedure 225 for the step-wise splitting of the data set with any number of continuous or categorical and correlated or uncorrelated predictor variables (Breiman et al., 1984;Rothwell et al., 2008), and it has been recommended to identify thresholds and to handle large-scale decomposition data sets (Rothwell et al., 2008;Prescott, 2010).We conducted CRTA separately for k and S using temperature, salinity class, tidal amplitude, ecosystem type, soil type, and relative elevation as predictor variables (Table 1).V-fold cross validation was set at 5 (as commonly conducted, compare Rothwell et al. (2008)), and the minimum number for observations per child node was set at n = 4, corresponding to at least two sites or 8% of the total data set.
To test for correlations between the variables salinity class, temperature, latitude, tidal amplitude, k and S, Spearman rank correlations were used (Table 3).Mann-Whitney U tests were 235 conducted to test for differences in k and S between marshes and mangroves and between mineral and organic soil types.
We tested for linear effects of temperature on k and S across sites, using simple linear regression analyses (Fig. 2).Two-tailed paired t-tests were used to test for effects of relative elevation as proxy for relative sea level on k and S (Fig. 3).Subsequent one-tailed paired t-tests 240 were conducted to test for the same effect within mineral, organic, marsh, and mangrove systems separately.
In 21 of our 22 sites where tea bags were deployed in both high and low elevation zones, replication was sufficient to conduct one-way ANOVA to test for differences in k and S between zones for each site separately (Fig. S2).We tested for effects of nutrient enrichment on k and S in 245 the data from the TIDE site (Massachusetts, US) using two-way ANOVA with enrichment treatment and marsh zone as predictors.All analyses were conducted using STATISTICA 10 (StatSoft Inc., Tulsa, OK, USA).

Temperature effects
We found no linear (Fig. 2a) or monotonic (Table 3) relationships between temperature and k.Also, CRTA revealed temperature only as a minor predictor for k (Figure S1a).Specifically, temperature seems to positively affect k in meso-tidal systems only (amplitude >2.1m; Fig. S1a; node 5) with sites ≥14.5°C during deployment supporting considerably higher rates of decomposition than sites 255 characterized by lower temperatures.However, this apparent temperature effect was inconsistent within the group of observations with tidal amplitude >2.1m (Fig. S1a; nodes 13-15).Furthermore, the majority of sites (65%) are characterized by tidal amplitudes <2.1 m and show no temperature effect on k.
In contrast to the temperature response of k, S was strongly affected by temperature (Fig. 260 2b).The significant negative correlation between S and temperature (p < 0.01; r 2 = 0.287; Fig. 2b) agrees well with the CRTA (Fig. S1b).However, CRTA also identified a narrow temperature range (21.9-23.6°C) in which increasing temperature led to higher stabilization (Fig. S1b; node 11).This group of observations (n = 5) from the general pattern is also clearly visible in Fig. 2b.

Effects of relative sea level
Paired comparisons of high vs. low elevated zones indicate no consistent effect of relative sea level on k across sites (p > 0.1; Fig. 3a + c), whereas S was significantly reduced by 29% in low compared to high elevated zones (p < 0.01; Fig. 3b).This significant reduction of S in low vs. high elevated zones was consistent across mineral and organic, as well as marsh and mangrove systems 270 (Fig. 3d).Testing for effects of relative sea level within each site separately revealed that S is significantly reduced by 28-87% in the lower elevated zone in 15 of 21 sites.A significant increase of S in low vs. high elevated zones was found in none of these 21 sites (Fig. S2).In ten of the sites, we found a significant effect of relative sea level on k; with significantly higher k in low vs. high zones in seven sites and significantly lower k in low vs. high zones in three sites (Fig. S2). 275

Effects of salinity and nutrient enrichment
We found no significant relationship between salinity class and k or S (Table 3).Also, CRTA did not reveal salinity class as an important factor for k and S (Fig. S1), and no consistent salinity effect on k and S was found when comparing sites of different salinities within single estuarine regions 280 (Chesapeake, Ebro Delta, Long Marsh, San Francisco Bay; Fig. S3).
The nutrient enrichment treatment at the TIDE project site decreased S by 72% in the high marsh.S in the low marsh likewise was similarly low as in the fertilized high marsh and not further reduced by fertilization (Fig. 4).In contrast, k was not responsive to the fertilization treatment in neither low nor high marsh (Fig. 4).

Other factors influencing decomposition rate and stabilization
CRTA revealed tidal amplitude as an important predictor for k (Fig. S1a).However, this result needs to be interpreted cautiously because no linear (p > 0.68; r 2 = 0.004) or monotonic relationship (Table 3) existed between tidal amplitude and k, and effects of tidal amplitude are not independent 290 from other factors because strong correlations existed with ecosystem and soil type, temperature, and latitude (Table 3).
Soil (mineral vs. organic) and ecosystem type (marsh vs. mangrove) did not affect k (Table 3, Fig. S1a).In comparison, S was lower in mangroves than in marshes and lower in organic than in mineral systems (Table 3), presumably caused by temperature effects because ecosystem and soil 295 type did not show up as predictors in CRTA (Fig. S1b).

Discussion
The findings of the present study cannot demonstrate consistent effects of either temperature or relative sea level on the decomposition rate of recent OM inputs (commonly assessed as k in litter 300 bag studies) in tidal wetlands.With respect to C sequestration, however, litter-decay data need to be considered cautiously, as the link among decomposition rate, SOM formation, and ultimately C sequestration is not straightforward.That is, plant tissues and other fresh OM inputs into an ecosystem are not resistant to decay per se, and as a consequence, it is critical to understand their biogeochemical transformation into stable compounds that leads to the formation of SOM (i.e. 305 stabilization) rather than understanding the pace at which decomposition proceeds (Prescott, 2010;Castellano et al., 2015;Haddix et al., 2016).Here, we also assessed OM stabilization, and in contrast to decomposition rate, stabilization decreased with temperature and was consistently lower in low vs. high elevated zones of tidal wetlands.Our study therefore provides indirect evidence that rising temperature and accelerated SLR could decrease the capacity of tidal wetlands to sequester C 310 by affecting the initial transformations of recent OM inputs to SOM (i.e.stabilization).

Temperature effects on decomposition processes
Surprising in the context of basic biokinetic theory, the temperature response of decomposition rate was weak or not present.Following typical Q10 values for biological systems of 2-3 (Davidson & 315 Janssens, 2006), k should have at least doubled over a gradient of ΔT >15°C.However, findings from studies conducted at single-marsh to regional scales are not conclusive either, ranging from no or small (Charles & Dukes, 2009;Kirwan et al., 2014;Janousek et al., 2017) to strong seasonallydriven temperature effects with a Q10 >3.4 as found within a single site (Kirwan & Blum, 2011).
Page 12 of 30  et al., 2016), temperature sensitivity of the used TBI materials was sufficiently demonstrated (Keuskamp et al., 2013).We therefore conclude that other parameters exerted overriding influence on k, mainly masking temperature effects and have not been captured by our experimental design.For instance, we do not have data on plant-biomass parameters that are thought to exert strong control on decomposition in tidal wetlands through priming effects (Wolf et 325 al., 2007;Mueller et al., 2016;Bernal et al., 2017).Likewise, large differences in site elevation and hydrology could have induced high variability in k across sites and masked potential temperature effects.Indeed, we demonstrate significant but mixed effects of relative sea level on k for some sites; however, we do not have sufficient data on actual site elevation or hydrology to control for these factors as covariates affecting the temperature effect on k.

330
In contrast to missing or subtle effects of temperature on k, OM stabilization was strongly affected by temperature.Overall, S decreased by ~90% over our temperature gradient from 10.9 to 28.5°C, corresponding to a decline of ~5% over a 1°C-temperature increase (Figure 2b).Thus, we demonstrate a considerable temperature effect on the initial steps of biomass decomposition in tidal wetlands.This effect, however, is not driven by changes in decomposition rate per se, but -more 335 importantly -by affecting the transformation of fresh and decomposable organic matter into stable compounds, with implications for C sequestration (e.g.Cotrufo et al., 2013).
In their global-scale assessment, Chmura et al. (2003) report a negative relationship of soil organic C density and mean annual temperature within both salt marshes and mangroves.Indeed, Chmura and colleagues hypothesized stimulated microbial decomposition at higher temperatures to 340 be the responsible driver for this relationship.Plant production and thus OM input is known to increase with latitude and temperature in tidal wetlands (Charles & Dukes, 2009;Gedan & Bertness, 2009;Kirwan et al., 2009;Baldwin et al., 2014), but this increase seems to be more than compensated by higher microbial decomposition.Working on the same spatial scale as Chmura et

Relative-sea-level effects on decomposition processes
Flooding and thus progressively lower oxygen availability in soil is supposed to be a strong suppressor of decomposition (Davidson & Janssens, 2006).Despite this dominant paradigm, we 350 clearly demonstrate that k is not reduced in low vs. high elevated zones of tidal wetlands (Fig. 3a).
This finding is in accordance with an increasing number of studies demonstrating negligible direct effects of sea level on decomposition rate in tidal wetland soils (Kirwan et al., 2013;Mueller et al., 2016;Janousek et al., 2017).A SLR-induced reduction in decomposition rate with positive feedback on tidal wetland stability seems therefore to be an unlikely scenario.Furthermore, we

355
show that S is strongly reduced in low vs. high elevation zones, suggesting that the conversion of recent OM inputs to stable compounds and eventually SOM is in fact lower in more flooded zones of tidal wetlands.Accelerated SLR consequently seems to yield the potential to decrease SOM formation and with that C sequestration.
This finding and its implication may occur counterintuitive with respect to the often sharp 360 redox gradients along tidal wetland zonations and with flooding (Davy et al., 2011;Kirwan et al., 2013;Langley et al., 2013), and the mechanism by which S is decreased in the more flooded zones is unknown.Because we did not observe consistent salinity effects on S and k in our data (Figs.S1, S3), we do not suppose that regular exposure of litter to salt water explains the unexpected finding.
Instead, more favorable soil moisture conditions in low vs. high elevated zones could have 365 decreased OM stabilization if higher flooding frequencies did not induce redox conditions low enough to suppress microbial activity in the top soil.In support of this, flooding-frequency induced changes in moisture conditions have been reported as primary driver of surface litter break down, leading to more than four-fold increased litter mass loss in low vs. high marsh zones of a New Jersey salt marsh (Halupa & Howes, 1995).

370
Additionally, greater nutrient availability and less nutrient-limited microbial communities in more frequently flooded zones could have contributed to this effect (Deegan et al., 2012;Kirwan et al., 2013).Strong effects of both high quality marine-derived OM and nutrient amendments on microbial structure and activity have been reported (Deegan et al., 2012;Keuskamp et al., 2015a;Kearns et al., 2016;Mueller et al., 2017), suggesting that regular marine OM and nutrient inputs in 375 more frequently flooded zones can positively affect decomposition.

Nutrient enrichment reduces stabilization -insights from the TIDE project
In addition to our global survey of early-stage decomposition processes in tidal wetlands worldwide, we included the long-term ecological research site of the TIDE project in 380 Massachusetts, US to experimentally assess both the effects of coastal eutrophication and the relevance of nutrient delivery through floodwater for OM decomposition in tidal wetlands.
Important for our argument that decomposition may be favored by higher nutrient availability in low elevated, more frequently flooded zones, we observed a strong reduction (>70%) of S by nutrient enrichment in the high marsh, whereas S in the low marsh likewise was low as in the 385 fertilized high marsh and not further reduced by fertilization (Fig. 4).Johnson et al. (2016) demonstrate that nutrient enriched high-marsh plots of the TIDE project receive 19±2 g N m -2 yr 1 , approximately 10-times the N load of reference high-marsh plots (2±1 g N m -2 yr -1 ; mean±SE), thus explaining the strong treatment effect observed in the high marsh.In accordance with low stabilization in the reference low marsh, which is equally low as the nutrient enriched high marsh, 390 reference plots of the low marsh receive 16±4 g N m -2 yr 1 , the same high N load as the enriched high-marsh plots.Surprisingly, however, N loads of 171±19 g N m -2 yr 1 in the enriched low-marsh plots do not result in additional reduction of S compared to the reference low marsh (Fig. 4).These findings suggest that microbial communities of the high marsh are N limited, and that N additions to a certain level can stimulate early OM decomposition and thus reduce stabilization.The missing 395 effect of N loads exceeding 16 g m -2 yr 1 on stabilization in the low marsh indicates that microbial communities are less or not N limited due to the naturally greater nutrient availability.The findings of the TIDE project therefore support our concept that higher nutrient availability and less nutrientlimited microbial communities in more frequently flooded zones could have contributed to the observed reduction of OM stabilization in low vs. high elevated zones of tidal wetlands in our 400 global assessment.
Although our conclusions on effects of nutrient enrichment on OM decomposition are based on the findings of a single field experiment only, our study adds to a growing number of reports illustrating the impact of coastal eutrophication on tidal wetland C cycling (Morris & Bradley, 1999;Deegan et al., 2012;Kirwan & Megonigal, 2013;Keuskamp et al., 2015b).At the same time, 405 however, we highlight the need to improve our understanding of coastal eutrophication in interaction with other global changes, particularly accelerated SLR and concomitant changes in flooding frequency, on the cycling of both labile and refractory C pools in order to predict future stability of tidal wetlands.

Methodological considerations
The quality of OM (i.e. its chemical composition) is a key parameter affecting its decomposition.
As the quality of the TBI materials differ from that of wetland plant litters, we did not expect to capture precise and absolute values for wetland litter break down in this study.Instead, we used the Tea-Bag Index to characterize the decomposition environment by obtaining a measure for the 415 potential to decompose and stabilize the deployed standardized material.Standardized approaches like this, or also the cotton-strip assay (e.g.Latter and Walton, 1988), are useful to separate the effects of environmental factors other than OM quality on decomposition processes and to assess their relative importance.Otherwise, complex interaction effects of the abiotic environment and OM quality make it difficult to predict the relevance of certain environmental factors for 420 decomposition processes, potentially masking the effects of important global-change drivers (reviewed in Prescott, 2010).
Stabilization is thought to be a key parameter for capturing changes in decomposition with consequence for C sequestration.Indeed, Keuskamp et al. (2013) demonstrate that S, as determined by the TBI, is significantly related with the C sequestration potential of an ecosystem as defined by 425 FAO (2000).In the present study, however, a large percentage of observations showed relatively low values for S, although tidal wetlands are known to act as particularly effective C sinks (Mcleod et al., 2011).Based on the S values obtained from initial calculations using the hydrolysable fractions suggested by Keuskamp et al. (2013), a large number of observations in fact yielded a negative S (data not shown).S becomes negative when the mass loss from green tea is greater than 430 the predicated maximum loss based on its hydrolysable fraction.At least two processes could have caused or contributed to this result: First, we demonstrate that S is indeed reduced in low vs. high elevated zones across our study sites, indicating that redox conditions in the top soil of tidal wetlands are at least often not low enough to hamper decomposition processes.As a consequence, the relatively high top-soil moisture of tidal wetlands provide favorable conditions for 435 decomposition, following typical moisture-decomposition relationships as demonstrated for terrestrial ecosystems (e.g.Curiel Yuste et al., 2007), and S should at least not expected to be high in the top-soil environment of tidal wetlands.Potentially, moisture and nutrient supply are even high enough to allow for considerable break down of non-hydrolysable compounds within three months of deployment, such as lignin (Berg & McClaugherty, 2003;Knorr et al., 2005;Feng et al., 440 2010; Duboc et al., 2014).Second, different protocols and methods to determine hydrolysable and non-hydrolysable fractions of plant materials exist and lead to variable results.The hydrolysable fraction of plant materials can consequently be over-or underestimated depending on method, protocol, and type of sample material.The use of the slightly higher hydrolysable fractions we determined for calculations of the TBI parameters effectively eliminated negative S values.In that 445 regard, using the values obtained from the alternative protocol given in Ryan et al. (1990) seemed more reasonable in our study.Although direction and size of reported effects on S and k in the present study are almost independent from the hydrolysable fraction used for calculations, further research is required to improve our estimates of the hydrolysable fractions in TBI materials.

Implications
While awareness about potential global-warming impacts on OM preservation and their resulting threat to future tidal wetland stability has been raised (Kirwan & Mudd, 2012), concepts on the vulnerability of tidal wetlands to accelerated SLR mainly focus on plant-productivity responses and their biophysical feedbacks (Kirwan et al., 2016).Potentially negative effects of accelerated SLR on 455 OM preservation were thus far overlooked, probably because stimulation of decomposition processes through increasing flooding is counterintuitive (Mueller et al., 2016).Here, we provide evidence that accelerated SLR is unlikely to slow down the decomposition rate of fresh OM inputs and additionally may strongly decrease OM stabilization and thus potentially the fraction of net primary production and other OM inputs to stable SOM.

460
This study addresses the influence of temperature, relative sea level, and coastal eutrophication on the initial transformation of biomass to SOM, and it does not encompass their effects on the existing SOM pool.However, aspects of S and k are key components of many tidal wetland resiliency models (Schile et al., 2014;Swanson et al., 2014) that have highlighted the critical role of the organic contribution to marsh elevation gain.Thus, combined negative effects of 465 temperature, relative sea level, and coastal eutrophication on OM stabilization may yield the potential to strongly reduce OM accumulation rates and increase wetland vulnerability to accelerated SLR.
Our findings imply that particularly the vulnerability of organic systems might increase with global change because in these systems soil volume is almost exclusively generated by the 470 preservation of OM.At the same time, however, mineral dominated systems, such as temperate European salt marshes, experience large amounts of easily decomposable allochthonous-OM input that relies on substantial stabilization in order to become sequestered (Middelburg et al., 1997;Allen, 2000;Khan et al., 2015).Thus, future rates of C sequestration could be substantially reduced in mineral dominated tidal wetland systems.

Figure captions
Figure 1 Overview map of study sites.Notes: See Table 1 for site details.(blue) elevated zones of tidal marsh and mangrove sites (compare Table 1).High and low elevated

Figure 1
Figure   Table 1 Overview of study regions, site names, and site properties.Sites in which tea bags were deployed in zones of different elevation and flooding frequency are marked (x).Different salinity classes are indicated as 'S' (salt water), 'B' (brackish water), and 'F' (fresh water).Tidal amplitude (Ampl.) is given in meters.Table 2 Hydrolysable (H) and mineral fractions of green tea (n = 5 batches) and rooibos tea (n = 3 batches) and C and N contents (n = 2 batches).Samples of each batch were analyzed as duplicates. Green Biogeosciences Discuss., https://doi.org/10.5194/bg-2017-533Manuscript under review for journal Biogeosciences Discussion started: 15 December 2017 c Author(s) 2017.CC BY 4.0 License.
Biogeosciences Discuss., https://doi.org/10.5194/bg-2017-533Manuscript under review for journal Biogeosciences Discussion started: 15 December 2017 c Author(s) 2017.CC BY 4.0 License.al. (2003), our study strongly supports this hypothesis and provides the mechanistic insight into the 345 temperature control of OM decomposition as an important driver of C sequestration tidal wetlands. 410 450

490Figure 2
Figure 2 (a) Decomposition rate (k) and (b) stabilization factor (S) versus mean atmospheric temperature during deployment period.Shown are results of linear regression analyses across and within elevation zones and organic and mineral soils.

Figure 3 (
Figure 3 (a + c) Decomposition rate (k) and (b + d) stabilization factor (S) in high (orange) and low

495
zones are characterized by distinct plant-species assemblages or by different stature of mangroves along the flooding gradient within each site.Shown are means ± SE for all sites (a + b) and for mineral, organic, marsh, and mangrove systems separately (c + d).P-values refer to results of paired t-tests (ns, P > 0.05; * P ≤ 0.05; ** P ≤ 0.01).

Figure 4
Figure 4Effects of marsh elevation (zone) and nutrient enrichment on both decomposition rate (k) Figure2