Three saltmarsh soil cores (T1–3) were retrieved ca. 30 m apart from the lower marsh zone from Skinflats (SK), an estuarine saltmarsh in Scotland (56°3^′34.04^′′ N, 3°43^′59.16^′′ W), as detailed in Houston et al. (2024a). Field methods and laboratory sub-sampling procedures are described in detail in Houston et al. (2024a). Briefly, the cores were split into 1 cm thick slices as follows: core T1 (0–1, 5–6, and 18–19 cm); T2 (0–1, 5–6, and 15–16 cm), and T3 (0–1, 5–6, and 19–20 cm) (with the deepest sample from each core being the deepest retrieved sample from the 20 cm length of the corer. On the occasions when a full core was not retrieved, the deepest retrieved soil was used). Each slice was subsequently divided to provide sample material for the RO procedure, and for aerobic laboratory incubations from which the biologically evolved CO₂ was collected for ¹³C and ¹⁴C analysis (Houston et al., 2024a).

The RO sub-samples were individually dried to constant mass before milling to a fine powder to homogenise and limit potential shielding effects from aggregates. Unlike most RO and RPO studies (e.g., Hemingway et al., 2017b), we did not remove carbonates from our samples. Acid treatment, which is required to remove carbonates from samples has been demonstrated to result in losses from the thermally labile OC fraction (Bao et al., 2019a), and affect the isotopic values of the lower temperature fractions (Rosengard et al., 2025). A loss of labile OC for our samples could seriously impact the interpretations in our study, and our ability to compare the ¹⁴C content of the CO₂ respired from bulk (untreated) soils in the incubation experiments (Houston et al., 2024a) to the ¹⁴C content of the RO thermal fractions.

The samples were sent to the NEIF Radiocarbon Laboratory for RO, which is described in Garnett et al. (2023). The RO procedure involved two stages, a first combustion to determine the relationship between the rate of CO₂ evolution and temperature (thermogram), and a second combustion where evolved sample gases were collected across defined temperature ranges, for subsequent isotope analysis. For the first combustion, ca. 200 mg of dried and homogenized sample material was weighed into a quartz vial which was inset into a quartz combustion tube, which was subsequently placed into a furnace set initially to room temperature. The furnace was progressively heated at a constant rate of 5 °C min⁻¹ to 800 °C in a stream of high purity oxygen (N5.5, BOC, UK). Heating caused combustion of the sample and the evolution of gas which was passed into a second quartz combustion tube containing platinised wool in a furnace set to a constant temperature of 950 °C. The platinised wool acted as a catalyst to ensure complete combustion of the evolved gases. Upon exiting the secondary combustion chamber the sample passed through a glass tube containing magnesium perchlorate desiccant to remove moisture and subsequently the CO₂ concentration of the gas was measured using a non-dispersive infrared CO₂ sensor (SprintIR^®-WF-5, Gas Sensing Solutions, UK). The sample was then passed out of the sensor unit and vented to the atmosphere.

The measured CO₂ concentration (normalised for sample mass) was plotted against temperature to produce thermograms which were used to identify temperature ranges, which defined C thermal reactivity pools for this study: 150–325, 325–425, 425–500, 500–650, and 650–800 °C.

For each sample, the required mass of material to evolve sufficient CO₂ (>3 mL) for ¹⁴C measurement was calculated based on the thermogram. A new sub-set from the original dried and homogenised sample was then re-run following the RO procedure outlined above, but instead of venting to atmosphere, after its measurement the evolved CO₂ was collected into foil gas bags based on the defined temperature ranges. CO₂ was collected for ¹³C analysis from 650–800 °C, but sufficient CO₂ was evolved for ¹⁴C analysis from this thermal fraction for only one sample (T1 0.5 cm, Table A1) and we do not consider this high temperature fraction further because it is likely dominated by carbonates and not relevant to the purpose of this study.

The foil gas bags (5 L Spout Pouch, https://www.pouchshop.co.uk/, last access: 20 October 2024) used for sample collection were sealed with one-hole rubber bungs into which a 0.6 cm diameter × 5 cm length stainless steel tube was inserted. Isoversinic tubing (Saint Gobain, France) was fitted over the stainless steel to connect it to a quick coupling (Colder Products Company, USA), which allowed connection to the RO kit.

Prior to the RO CO₂ collection, all equipment was cleaned using a standardised procedure (Garnett et al., 2023). All glassware was combusted at 900 °C for a minimum of 2 h, and all couplings and connectors were washed in carbon-free detergent (Decon) and rinsed in Milli-Q water. The foil gas bags were cleaned by repeatedly (3 times) filling with ca. 1 L high purity nitrogen gas (Research Grade 99.9995 % purity, BOC, UK) and evacuating with an air pump, over a period of at least 24 h (to aid out-gassing of CO₂). The final evacuation, immediately before connecting to the RO rig, involved pumping out the bags with an SBA-5 CO₂ analyser (PPsystems, USA) to ensure that the bags did not contain significant contamination. Before commencing a sample combustion, the entire RO rig was checked for leaks and other potential sources of contamination by measuring the CO₂ concentration in the oxygen carrier gas exiting the kit, using the SBA-5 CO₂ analyser.

Within 3 d of combusting a sample, the evolved gas in each foil bag was connected to a vacuum rig for cryogenic recovery of pure sample CO₂ by passing it through slush (-78 °C; dry ice and industrial methylated spirits) and then liquid nitrogen (-196 °C) traps, under high vacuum (ca. 3×10-3 mbar). The sample CO₂ was then split into three aliquots: One for δ13C analysis using isotope ratio mass spectrometry (IRMS; Delta V, Thermo-Fisher, Germany), one for graphitisation and subsequent AMS ¹⁴C analysis, and one for an archive back-up. The graphitised AMS samples were measured for ¹⁴C content at the SUERC AMS Laboratory (see Ascough et al., 2024). The ¹³C content (δ13C-VPDB) was used to normalise the ¹⁴C results to a δ13C of -25 ‰ to correct for isotopic fractionation. Following convention, ¹⁴C results are presented as % Modern (fraction modern × 100) and conventional radiocarbon ages (years BP, where 0 BP = AD 1950 and age = -8033× Ln (%Modern / 100)).

Continuous activation energy distributions (p(o,E)) were modelled from thermograms using the “rampedpyrox” package in Python V3.8 (Hemingway, 2016; Hemingway et al., 2017b). For each temperature fraction which CO₂ was collected from, the distributed activation energy model calculates mean activation energies (µfE) and the standard deviation of activation energy (σfE), which is a measure of the heterogeneity of bond strength. Mean µE, σE and activation energy distribution (p(o,E)) are also calculated for each sample using the model. Further data analysis and visualisation of thermograms and isotopic data was undertaken using RStudio V4.2.2 (R Core Team, 2022). Statistical analyses (ANOVAs for parametric data; Kruskal-Wallis for non-parametric data) were undertaken in Sigmaplot V12.5 (Systat Software Inc, 2013).

The ¹⁴C content of the RO fractions (Fig. 1, Table 1) were statistically similar between the 0.5 cm, 5.5 cm, and deepest sample (T1 18.5 cm, T2 15.5 cm, T3 19.5 cm) depth increments for each of the temperature fractions (Kruskal-Wallis; p=0.83, 0.38, 0.66, 0.99, for 150–325, 325–425, 425–500, 500–650 °C, respectively). There were, however, clear differences in ¹⁴C contents between the temperature fractions, with ranges of 81.50 %–97.54 % Modern for 150–325 °C, 41.67 %–79.80 % Modern for 325–425 °C, 17.67 %–63.56 % Modern for 425–500 °C, and 15.69 %–53.96 % Modern for 500–650 °C (Fig. 1, Table 1).

The RO samples were not pre-treated with acid, but the samples for bulk soil-¹⁴C were (Houston et al., 2024a), so we cannot verify that the weighted RO-¹⁴C contents amassed to the bulk soil ¹⁴C content. However, previous work using this analytical set-up have done this for other samples and shown that the combined RO fractions do equal the bulk isotope values (Garnett et al., 2023).

There were no significant differences in the ¹³C content of the ramped oxidation fractions (Fig. 2, Table 2) between the depth increments (Kruskal-Wallis; p=0.66, 0.63, 0.63, 0.44, 0.17, for 150–325, 325–400, 425–500, 500–650, 650–800 °C respectively). ¹³C contents followed the opposite trend to ¹⁴C contents with temperature, with ranges of -28.0 ‰ to -24.7 ‰ for 150–325 °C, -26.6 ‰ to -22.3 ‰ for 325–425 °C, -25.4 ‰ to -20.2 ‰ for 425–500 °C, -24.4 ‰ to -13.9 ‰ for 500–650 °C, and -21.1 ‰ to -4.0 ‰ for 650–800 °C (Fig. 2, Table 2).

Figure 1 presents a comparison of the ¹⁴C content of the RO temperature fractions and respired CO₂ from the same soils during aerobic laboratory incubations (Houston et al., 2024a). These comparisons show that for each of the 0.5 cm depth samples, the ¹⁴C content of the respired CO₂ was greater than the ¹⁴C content of any of the RO temperature fractions in the same soils (Fig. 1). For the 5.5 cm depth samples, the ¹⁴C content of the CO₂ respired in the incubations was approximately equivalent to the ¹⁴C content of the 150–325 °C RO temperature fraction (Fig. 1). For T2 15.5 cm, the ¹⁴C content of the respired CO₂ was also closest to the 150–325 °C RO temperature fraction (Fig. 1). For the T1 18.5 cm and T3 19.5 cm samples, the ¹⁴C contents of the incubation CO₂ were depleted relative to the 150–325 °C RO temperature fraction for both samples, and instead, were closest to the 325–425 °C and 425–500 °C RO temperature fractions, respectively (Fig. 1).

µE ranged from 157.50–170.97 kJ mol⁻¹ for the 0.5 cm depth samples, 159.97–165.32 kJ mol⁻¹ for the 5.5 cm depth samples, and 154.38–160.44 kJ mol⁻¹ for the deepest samples (T1 18.5 cm, T2 15.5 cm, T3 19.5 cm. Table 3). σE ranged from 23.16–35.83 kJ mol⁻¹ for the 0.5 cm depth samples, 22.16–25.25 kJ mol⁻¹ for the 5.5 cm depth samples, and 21.43–23.51 kJ mol⁻¹ for the deepest samples (Table 3). Between the three depth increments, there were no significant changes in µE or σE (Table 1; ANOVA; p=0.47 and 0.37, respectively).

Table 4 shows µfE and the associated σfE for each thermal fraction. µfE ranged from 131.04–133.23 kJ mol⁻¹ for 150–325 °C, 156.83–157.78 kJ mol⁻¹ for 325–425 °C, 176.14–177.79 kJ mol⁻¹ for 425–500 °C, 185.44–199.19 kJ mol⁻¹ for 500–650 °C, and 213.06–247.75 kJ mol⁻¹ for 650–800 °C (Table 4). σfE ranged from 7.33–8.71 kJ mol⁻¹ for 150–325 °C, 9.83–10.23 kJ mol⁻¹ for 325–425 °C, 6.88–8.83 kJ mol⁻¹ for 425–500 °C, 3.68–16.04 kJ mol⁻¹ for 500–650 °C, and 1.83–10.94 kJ mol⁻¹ for 650–800 °C (Table 4). µfE and σfE both varied significantly between the thermal fractions, increasing sequentially (Kruskal-Wallis, p = 0.001 and 0.001, respectively).

The first three RO temperature fractions (150–325, 325–425, 425–500 °C) were derived solely from OC sources, as IC begins to breakdown from ca. 550 °C (Hemingway et al., 2017b). CO₂ from the 500–650 and 650–800 °C fractions may, however, have been evolved from a mix of OC and IC sources. The IC contents of the studied soils (0.11 %–0.48 %) were low relative to OC contents (4.18 %–7.71 %), and IC makes only 1.95 %–10.48 % of the total soil C pool for these samples (Table A2).

IC could have been removed from our saltmarsh soil samples to allow complete analysis of the soil OC pool, and many R(P)O studies have taken this approach (Bao et al., 2019b; Hemingway et al., 2017b; Luk et al., 2021; Stoner et al., 2023; Williams and Rosenheim, 2015). However, our samples have low IC contents (Table A2), and acid-treatment, which is required to remove IC from samples, can cause losses of labile OC (Bao et al., 2019a). Indeed, in Hemingway et al. (2017b), acid treatment of samples prior to RO resulted in a shift of 4 % Modern ¹⁴C, which could change one of our samples from having a pre-bomb ¹⁴C content to a post-bomb ¹⁴C content, or vice-versa. A similar shift in ¹⁴C content for our samples could seriously impact the interpretations in our study, and our ability to compare the ¹⁴C content of the CO₂ respired from bulk (untreated) soils in the incubation experiments (Houston et al., 2024a) to the ¹⁴C content of the RO fractions. The soils in the incubation experiments were also not decarbonated as the acid-treatment would have affected soil respiration processes and made the results incomparable to in-situ soil degradation processes (Houston et al., 2024a).

The ¹⁴C-RO content decreased over the four thermal fractions (150–325, 325–425, 425–500, 500–650 °C, Fig. 1), implying that ¹⁴C-depleted OC had a greater thermal recalcitrance than ¹⁴C-enriched OC for these saltmarsh soil samples. Since the ¹⁴C content of each RO fraction was <100 % Modern (Table 1), each of the OC reactivity pools were likely to be predominantly composed of carbon sequestered from the atmosphere before the 1963 ¹⁴C bomb-spike caused by atmospheric nuclear weapons testing, although we cannot completely discount some contributions from post-bomb carbon (Hajdas et al., 2021). Nevertheless, using ¹⁴C content as an estimate of the age of the OC we can infer that the older (¹⁴C-depleted) OC has a greater thermal recalcitrance than young OC for these samples, which is consistent with previous studies on the thermal reactivity of carbon stored in soils and sediments (e.g., Bao et al., 2019b; Luk et al., 2021; Plante et al., 2013; Stoner et al., 2023).

The results suggest inhomogeneity within at least one of the temperature fractions for each sample because, while there were no post-bomb ¹⁴C contents for the incubation or RO samples (Table 1), there is likely to be a fraction of post-bomb (post-AD1955) OC in at least one of the temperature fractions. Autochthonous OC sequestration (post-bomb) at this accreting saltmarsh (Hajdas et al., 2021; Smeaton et al., 2024) may become obscured by contributions from pre-bomb (pre-AD1955) OC. Observing the decline in ¹⁴C content with increasing temperature (Fig. 1), we hypothesise that, if present, this mixing of pre- and post-bomb C most likely occurred in the 150–325 °C fraction.

As the oldest (most ¹⁴C-depleted) C had the greatest thermal recalcitrance (Fig. 1), this emphasises that saltmarshes accumulating greater amounts of older (¹⁴C-depleted) OC will likely provide the most thermally recalcitrant OC stores, and saltmarshes accumulating greater proportions of contemporary OC, either through in-situ production or young allochthonous components, contain soil OC stores which are of greater thermal lability (Komada et al., 2022; Van de Broek et al., 2018). However, the ¹⁴C contents of the lowest temperature RO fraction (81 %–98 % Modern; Table 1) highlight that although ¹⁴C content decreases with decreasing thermal reactivity (Fig. 1), thermally labile OC can still be aged (at least hundreds of years old) for these soils. Due to the often anaerobic and non-eroding conditions of buried sediments, saltmarshes can therefore be stores of old, but thermally labile carbon. Of course, the thermal recalcitrance of OC is not necessarily related to biological turnover time, as this is also related to the prevailing environmental conditions (Schmidt et al., 2011; Spivak et al., 2019).

¹³C-RO increased sequentially with the thermal fractions (Fig. 2), due to greater contributions from relatively ¹³C-enriched C sources from the higher temperature thermal fractions. The ¹³C-RO contents of the 150–650 °C fractions were each typical of OC sources (Leng and Lewis, 2017), whereas the ¹³C-RO contents of the 650–800 °C fraction were mostly typical of at least a partial contribution from an IC source, with the exception of T2 5.5 cm and T3 5.5 cm (Table 2) (Brand et al., 2014; Ramnarine et al., 2012). As IC can begin to evolve from 550 °C, it is possible that a mix of OC and IC sources was present in the 500–650 °C thermal fractions.

As ¹³C-RO increased with temperature (Fig. 2, Table 2), ¹³C-enriched OC had a greater thermal recalcitrance than ¹³C-depleted OC for these samples. Previous work has demonstrated that >80 % of the OC accumulating at Skinflats saltmarsh is autochthonous/terrestrial in origin (Miller et al., 2023), with limited contributions from marine OC. The thermally recalcitrant OC was potentially composed of a greater amount of OC which has undergone microbial decomposition as this process tends to enrich the degraded OC in ¹³C (Boström et al., 2007; Etcheverría et al., 2009; Luk et al., 2021; Sanderman and Grandy, 2020; Soldatova et al., 2024; Stoner et al., 2023). The thermally recalcitrant OC may instead/also have been composed of different OM compounds (e.g., lignins, aromatics) than the more thermally labile OC (e.g., carbohydrates, lipids) (Sanderman and Grandy, 2020). It is also possible that methodological artefacts, such as kinetic fractionation, influenced the ¹³C-RO contents. Kinetic fractionation is explained by different carbon isotopes evolving as CO₂ from the soil sample at different rates during the ramped heating (Hemingway et al., 2017a). Kinetic fractionation would cause the ¹³C content of the evolved CO₂ to increase linearly with temperature (Hemingway et al., 2017a), and we cannot rule out this artefact. Hemingway et al. (2017a) determined that kinetic fractionation was not an important factor in their RPO procedure, but we used a different set-up (described in Garnett et al., 2023).

The isotopic composition of the evolved CO₂ did not vary significantly with depth for any of the temperature fractions. The lack of an increase in the age (¹⁴C-depletion) of soil C with sample depth is unusual, as typically C undergoes a burial process, and previous work has shown diagenetic ageing of saltmarsh soils with depth as young OC is turned over faster than old OC (Komada et al., 2022; Van de Broek et al., 2018).

Compared to other UK saltmarshes, Skinflats has relatively high C accumulation rates (Miller et al., 2023; Smeaton et al., 2024). Depleted ¹⁴C contents of the OC accumulating at the Skinflats saltmarsh (Houston et al., 2024a) imply that a proportion of the OC being buried may already have been aged at the time of deposition on the marsh surface, as the marsh formed in the 1930s (Miller et al., 2023). The combination of high carbon accumulation rates and depleted soil ¹⁴C contents implies that the Skinflats saltmarsh accumulates a high proportion of old, most likely allochthonous OC. Some of the aged, allochthonous OC may have undergone significant microbial processing and degradation prior to its accumulation in the saltmarsh soil. As the OM is degraded, and the energetically favourable components are consumed, the resulting OM becomes increasingly thermally recalcitrant (Luk et al., 2021; Sanderman and Grandy, 2020; Soldatova et al., 2024). The accumulation of a high proportion of degraded OC on the Skinflats saltmarsh may therefore explain the lack of observed change in the isotopic composition of the soil OC pools with depth. This interpretation is supported by the lack of change in both the amount and the proportion of CO₂ evolved from each temperature fraction with depth (ANOVAs, p>0.05; Table A3).

Not all old OC is degraded or thermally recalcitrant, and our results show that the Skinflats saltmarsh is also a store of old (¹⁴C-depleted), thermally labile OC (Fig. 1). Old OC can be thermally labile if it “ages” (is stored) in an environment with low decomposition rates, e.g., a peatland (Dean et al., 2023), prior to transport and accumulation into the saltmarsh. There are extensive peatlands in the Skinflats catchment, many of which are degrading (Lilly et al., 2012). Regardless of the age and degradation state of the OC deposited onto the marsh surface, as it gets buried it will undergo a degree of microbial processing and degradation in the saltmarsh soil (Luk et al., 2021), but that process is potentially less prevalent at Skinflats than saltmarshes accumulating younger, less degraded OC.

Through isotopic analysis of saltmarsh soils partitioned using ramped oxidation, we have determined that increased thermal recalcitrance is related to older (¹⁴C-depleted; Fig. 1), more degraded/microbially derived (¹³C-enriched; Fig. 2) soil C. These findings are consistent with previous research on the thermal reactivity of soil and sediment C, which have found that in most cases, more energy is required (higher temperature/µE) to decompose older (¹⁴C-depleted), degraded/microbially derived (¹³C-enriched) C than younger (¹⁴C-enriched), less processed (¹³C-depleted) C (e.g., Bao et al., 2019b; Plante et al., 2013; Stoner et al., 2023), including one saltmarsh study (Luk et al., 2021).

As the biological turnover time of OC is related to the prevailing environmental conditions as well as thermal reactivity, the isotopic composition of the most biologically- and thermally-reactive saltmarsh soil OC pools may not be the same. To determine if this is the case, or not, we compared the isotopic composition of the RO thermal reactivity fractions to the isotopic composition of the CO₂ that was evolved biologically during incubations of equivalent samples (Houston et al., 2024a) (Fig. 1).

Figure 1 shows that for each of the 0.5 cm depth samples, the ¹⁴C content of the CO₂ respired in the aerobic laboratory experiments was ¹⁴C-enriched relative to any of the RO temperature fractions, which was also the case for the T3 5.5 cm sample (Table 3). The relative ¹⁴C-enrichment of the biologically respired CO₂ compared to the thermally evolved CO₂ was likely caused by inhomogeneity in the OC thermal reactivity pools, as each defined thermal reactivity pool may be composed of multiple OC sources of variable age and composition. As thermal recalcitrance is related to ¹⁴C-depletion for these samples (Fig. 1), we hypothesise that for saltmarsh soil samples producing respired CO₂ that was ¹⁴C-enriched relative to any of the RO fractions (T1 0.5 cm, T2 0.5 cm, T3 0.5 cm, T3 5.5 cm; Table 1, Fig. 1), that this CO₂ was biologically-produced from an OC pool within the most thermally labile RO fraction (150–325 °C). Thus, we suggest that even within the 150–325 °C RO fraction there are pools of even younger OC, but that they are masked by older, ¹⁴C-depleted OC. Similar findings of mixing within thermal fractions has been reported in previous RPO work (e.g., Rosengard et al., 2025; Rosenheim et al., 2008).

The ¹⁴C content of respired CO₂ from the 5.5 cm depth samples tended to be closer to the ¹⁴C content of the lowest temperature (150–325 °C) RO fraction (Fig. 1), implying that for these samples the biologically evolved CO₂ was from a thermally labile OC pool. The T2 15.5 cm respired CO₂ sample was also similar in ¹⁴C content to the lowest temperature RO fraction, whereas respired CO₂ from the slightly deeper T1 18.5 cm and T3 19.5 cm samples was ¹⁴C-depleted relative to the 150–325 °C RO fraction, instead aligning closer to the higher temperature RO fractions (Fig. 1). The biologically evolved CO₂ from T1 18.5 cm and T3 19.5 cm were therefore potentially derived from less thermally labile OC pools than the other samples, although it is possible that the thermally labile pools were composed of multiple OC sources with different ¹⁴C contents. The ¹⁴C content of the CO₂ evolved from the aerobic incubations of T1 18.5 cm and T3 19.5 cm was hypothesized to have been derived from an inorganic C source due to the enriched ¹³C contents of -6.1 ‰ and -3.7 ‰, respectively (Houston et al., 2024a). As IC biological turnover times are controlled by different factors than OC (Van Dam et al., 2021), and the remainder of the samples were determined to evolve from OC substrates, this is likely to explain why the ¹⁴C content of the CO₂ evolved from the aerobic incubation experiments for T1 18.5 cm and T3 19.5 cm did not align with the lowest temperature (most thermally labile) RO fraction (Fig. 1). Therefore, there was a clear depth trend in the relationship between the ¹⁴C content of CO₂ respired in the aerobic incubation experiments and the ¹⁴C content of RO fractions of the same bulk soils. Degradation of some of the thermally labile OM components during burial may reduce the range of differently aged OC sources within the most thermally labile RO fraction for the deeper samples in this study.

For seven out of nine samples (T1 18.5 cm and T3 19.5 cm being the outliers), the ¹⁴C content of the CO₂ evolved from the aerobic laboratory incubations was closest to the ¹⁴C content of the 150–325 °C RO temperature fraction. Therefore, even though the CO₂ evolved from the aerobic incubation experiments was determined to be from a predominantly aged, allochthonous OC source (Houston et al., 2024a), it can now also be shown to be derived from a predominantly thermally labile OC pool (Fig. 1).

We did not attempt to relate the ¹³C-RO to the ¹³C content of the CO₂ respired in the incubation experiments, due to the potential for microbial fractionation during the incubation experiments which can change the ¹³C content of the respired CO₂ and the resulting soil OC (Soldatova et al., 2024; Werth and Kuzyakov, 2010). In contrast, ¹⁴C results are normalised using the measured δ13C values and are therefore immune to such isotopic fractionation effects.

Our results show that aged (presumed allochthonous), thermally labile OC stored in saltmarsh soils remains vulnerable to loss to the atmosphere upon habitat drainage. Saltmarsh soils usually exist in low-oxygen, tidally-inundated conditions which slow decomposition of OC (Chapman et al., 2019), but many saltmarshes globally have been drained (and their soils subsequently oxidised) to convert them for land uses such as housing developments and agriculture (Bromberg and Bertness, 2005; Campbell et al., 2022; Morris et al., 2012). In the Forth Estuary, where the Skinflats saltmarsh is located, as much as 50 % of the intertidal area has been converted to agricultural land since 1600, often involving the drainage of saltmarshes (Hansom and McGlashan, 2004).

Protecting saltmarshes from degradation following drainage is listed as an eligible activity for generating carbon credits for blue carbon ecosystem (BCE) projects (VERRA, 2023) and there is significant potential for climate mitigation by avoided emissions from protecting vulnerable stocks of soil OC in BCEs (Goldstein et al., 2020; Griscom et al., 2017; Kwan et al., 2025; Sasmito et al., 2025). Similarly, the re-creation of saltmarsh habitat through managed realignment (rewetting by tidal inundation) of historic saltmarsh habitats which were previously reclaimed for land use purposes (e.g., agriculture) could reduce (and possibly reverse) the emissions of aged OC to the atmosphere, both locally to Skinflats, and globally.

The evidence for the respiration of thermally labile, allochthonous OC from saltmarsh soils in a drainage degradation scenario demonstrates that at least this fraction of allochthonous OC could be counted as additional in carbon crediting projects and National GHG Inventories. Because allochthonous OC can account for up to 90 % of saltmarsh soil carbon (Komada et al., 2022), the inclusion of allochthonous OC (or even a fraction of it) would significantly increase the climate mitigation awarded to blue carbon projects (as carbon credits, or contributions to National GHG Inventories) (Houston et al., 2024b).

As the bioavailable OC respired in the experiments of Houston et al. (2024a) was (in most cases) from a predominantly thermally labile OC pool, and ¹⁴C-RO decreased (C became older) with increasing temperature (thermal recalcitrance), RO measurements could be useful for characterising the turnover times of OC pools for saltmarsh soils exposed to oxic conditions (drainage degradation scenario). The use of thermally defined OC pools to characterize OC turnover times for saltmarsh soils would require a modelling advancement to constrain degradation rates and residence times. Such efforts are not within the scope of this study but could inform additionality/permanence in these saltmarsh systems. Experimentally defined turnover times of OC thermal reactivity pools could, for example, provide a more robust approach than inclusion/exclusion of allochthonous OC from saltmarsh “blue carbon” projects (Houston et al., 2024b).

Further research is needed to determine if the relationship between biological and thermal lability exists for different degradation scenarios such as nutrient enrichment, as OC turnover time is related to the environmental conditions as well as the thermal lability of the OC pools. Similarly, these experiments would need to be replicated for a wider range of saltmarshes (high and low latitude saltmarshes, different typologies), as there are likely to be differences in OC turnover in different systems.

The samples used for this study were from the low marsh zone only, but it is likely that the thermal reactivity of the Skinflats saltmarsh soil C will vary spatially across the marsh, as the proportion of OC sources has been shown to be variable across saltmarshes (Middelburg et al., 1997). Given our findings that old (¹⁴C-depleted) OC has greater thermal recalcitrance than young (¹⁴C-enriched) OC (Fig. 1), we anticipate that higher marsh zones, which typically have greater proportions of autochthonous OC than lower marsh zones (Spohn et al., 2013), would contain a greater proportion of thermally labile OC. However, it is important to recognise that some of the young (¹⁴C-enriched), autochthonous OC in saltmarsh soils can also be thermally recalcitrant. As well as marsh zonation, we expect that the proportion of OC sources (and associated mix of thermal reactivities) would also vary with proximity to marsh creeks which redistribute autochthonous and allochthonous C across the saltmarsh habitat (Middelburg et al., 1997; Reed et al., 1999). In previously published work we showed that Skinflats accumulates OC of a much greater “age” (depleted soil ¹⁴C contents) than two other saltmarshes in Scotland (Houston et al., 2024a).

In this paper we have determined that age (¹⁴C-content) is related to the thermal recalcitrance of saltmarsh soil OC. We therefore speculate that sites accumulating younger OC would have more thermally labile soil OC than sites accumulating older OC, like Skinflats, with wider implications for the risks to these vulnerable stores of soil carbon from human disturbances.