Thermal responses of dissolved organic matter under global change

Dissolved organic matter (DOM) is a major active carbon reservoir in aquatic ecosystems and strongly influences biogeochemical cycles. Temperature increases are expected to accelerate DOM decomposition and CO2 release, but the observed temperature sensitivity reflects both intrinsic molecular traits (e.g., thermodynamic favorability, bioavailability) and environmental constraints such as nutrient enrichment. DOM consists of thousands of molecules with distinct traits, yet most models rely on coarse classes (labile/recalcitrant) and lack quantitative approaches to capture molecule-specific temperature responses and how these scale to whole-assemblage behavior. The authors hypothesize that climate change reorganizes DOM functional composition because individual molecules respond differently to temperature, and that nutrient enrichment can further modulate these responses. They develop a compositional-level indicator (iCER) that integrates molecule-specific environmental responses (MERs) to quantify how DOM assemblages respond to temperature, test the spatial transferability of MERs across contrasting climate zones, and examine links to thermodynamics and ecosystem processes including greenhouse gas emissions.

Prior work indicates that warming accelerates decomposition via kinetic and metabolic theory and that temperature sensitivity may vary with baseline temperature and DOM quality. Studies often classify DOM into broad pools (labile vs recalcitrant; active/slow/passive) and use bulk indices, which obscures molecular diversity. Evidence from soils suggests warming drives compositional changes in organic matter. Thermodynamic analyses relate molecular traits (e.g., H/C ratio, aromaticity, Gibbs free energy, NOSC) to bioavailability and persistence, with environmental conditions (e.g., oxic vs anoxic) further constraining decomposition. Nutrient enrichment has been shown to interact with temperature, altering productivity and DOM composition. However, there has been no quantitative, transferable metric to profile molecule-specific temperature responses and upscale them to compositional responses across spatial gradients.

Study design included replicated field microcosm experiments across natural temperature gradients at five to six elevations on three mountains spanning subtropical wet (Laojun, China), temperate arid (Dangjin, China), and subarctic (Balggesvarri, Norway) climate zones, yielding 480 microcosms (16 elevations total × 10 nutrient levels × 3 replicates). Each 1.5 L bottle contained 15 g of sterilized common lake sediment (from Taihu Lake) and 1.2 L of sterilized artificial lake water with nitrate and phosphate additions to achieve 10 nutrient levels (0 to 36 mg N L−1; N:P ≈ 14.93). Bottles were open to airborne microbial colonization; water temperature was measured before sampling. After one month, sediment DOM was extracted and characterized by FT-ICR MS. Laboratory microcosms (84 total; 3 inoculum sources × 7 temperatures × 4 replicates) used the same sterilized Taihu sediments inoculated with microbial communities from lakes representing subtropical, transitional, and temperate zones, incubated anoxically in sealed vials at 5–35 °C for 33 days with serial headspace measurements of CO2 and CH4 by gas chromatography to quantify fluxes and release rates. DOM extraction used sonication, filtration (0.45 μm), SPE (Oasis HLB), and FT-ICR MS analysis (Bruker solariX XRT 15T, negative ESI, m/z 150–1,200, resolving power ~800,000 at m/z 400). Molecular formulae (C, H, O, N, S, P) were assigned using Formularity with stringent criteria and blank subtraction. Molecules were grouped into seven compound classes via van Krevelen diagram and modified aromaticity index (Almod). Molecular traits calculated included mass, elemental ratios (H/C, O/C, N/C, P/C, S/C), DBE-based indices, Almod, NOSC, Kendrick defect, and standard Gibbs free energy (GFE) for half-reaction of carbon oxidation (GFE = 60.3 − 28.5 × NOSC). Development of iCER: (1) Compute molecule-specific environmental responses (MERs) as Spearman correlation coefficients (ρ) between each molecule’s relative abundance and water temperature using an 80:20 split of the dataset to ensure independence between MER estimation and iCER calculation. Molecules occurring in at least one-third of samples were retained. (2) Compute compositional-level environmental response (iCER) for each sample in the held-out set as the abundance-weighted mean of MERs across molecules: iCER = Σ(MERi × li) / Σ(li). iCERs were also calculated using only molecules with statistically significant MERs (P ≤ 0.05), which were strongly correlated with all-molecule iCERs and emphasized temperature sensitivity. Spatial transferability of MERs was assessed via Pearson correlations and linear regression slopes between climate zones (inter-regional) and between randomized nutrient groupings within zones (intra-regional). At the compositional level, relationships of iCER with elevation (proxy for temperature), nutrient enrichment, DOM traits, sediment TOC, and greenhouse gas fluxes/rates were analyzed using linear models and correlations.

• MERs spanned strong negative to positive responses (ρ from −0.90 to 0.89) with median negative and positive values of −0.44 and 0.30, respectively, across all three climate zones. Warm-depleting molecules (negative MER) were more thermodynamically favorable for oxidation (lower GFE), generally exhibited higher aromaticity (Almod) and lower H/C, and were enriched in aromatic-like and highly unsaturated high-O compound classes (43–52% of formulae with MER < −0.25). Warm-accumulating molecules (positive MER) tended to have lower aromaticity, higher H/C, and included aliphatic, peptide, and low-O highly unsaturated compounds; their GFEs were generally higher (75–90 kJ (mol C)−1) in field sediments, consistent with reduced thermodynamic favorability under anoxic sediment conditions. • Spatial transferability: MERs were consistent among regions. Inter-regional Pearson r values between zones ranged 0.73–0.82 with slopes 0.30–1.34 (P ≤ 0.05). Intra-regional random splits showed higher consistency (mean r = 0.80, 0.98, 0.90; slopes = 0.84, 0.98, 0.91 in subtropical wet, temperate arid, and subarctic zones, respectively; P ≤ 0.05). • iCER patterns: Using molecules with significant MERs (P ≤ 0.05), mean iCERs differed among zones: subtropical wet −0.027, subarctic −0.120 (over four-times lower; P ≤ 0.001), temperate arid mean −0.211 with the largest variability (SD 0.326). iCER decreased with elevation (warmer sites had higher iCER) across nutrient levels in all zones (P ≤ 0.05). • Nutrient interactions: Temperature sensitivity of iCER (slope of iCER vs temperature) increased with nutrient enrichment in all zones. Per 1 mg N L−1 increase, sensitivity rose by up to 21.8% (subtropical wet), and by 11.4% and 7.0% in temperate arid and subarctic zones, respectively. • Ecosystem links (field): iCER positively correlated with sediment total organic carbon after incubation (subtropical wet r = 0.41; temperate arid r = 0.53; P ≤ 0.05), consistent across nutrient levels. • Laboratory experiments: Molecular-level patterns were broadly consistent; warm-accumulating molecules were thermodynamically more favorable (lower GFE, 62–82 kJ (mol C)−1) than in field context and were less dominated by aliphatics (4–13% of significant MERs) and aromatics (21–36%) compared to warm-depleting molecules (50–75% aromatics). MERs showed spatial transferability among inoculum sources. iCER increased with temperature in all inoculum groups (r = 0.51, 0.68, 0.64; P ≤ 0.005) and correlated with GFE-derived traits. iCER was positively associated with CO2 and CH4 fluxes (Pearson r = 0.43–0.50; P ≤ 0.05 in subtropical and temperate groups). Release rates ranged 0.72–4.15 μg C g−1 d−1 for CO2 and 0–0.20 μg C g−1 d−1 for CH4. • Overall, warming reorganized DOM composition toward molecules with higher GFE (less thermodynamically favorable) in field sediments, while in controlled lab incubations without new production, higher iCER aligned with enhanced decomposition and greenhouse gas release.

The study demonstrates that molecule-specific temperature responses are consistent across distinct climate zones, supporting the concept of intrinsic, transferable MERs governed by molecular traits. Aggregating MERs into the iCER indicator reveals that DOM assemblages exhibit stronger positive thermal responses at warmer sites, indicating functional reorganization of DOM composition. In field microcosms, higher iCER coincided with greater sediment organic carbon stocks, likely because warming and nutrient enrichment stimulated primary production and inputs of fresh DOM, shifting assemblages toward less thermodynamically favorable compounds (higher GFE) and potentially lengthening persistence of certain aliphatic/peptide-like molecules under anoxic conditions. In laboratory incubations isolating decomposition, higher iCER corresponded to greater CO2 and CH4 emissions, indicating that warming enhanced biodegradation of thermodynamically favorable compounds (including some aromatic-like molecules) when new production was absent. Nutrient enrichment amplified the temperature sensitivity of compositional change, implying that eutrophication can intensify warming effects on DOM dynamics. These findings bridge molecular thermodynamics and ecosystem process responses, suggesting that iCER can inform predictive models by mechanistically linking DOM chemistry to carbon cycling outcomes across spatial scales.

The authors introduce iCER, a simple and robust indicator that upscales molecule-specific thermal responses (MERs) to quantify compositional-level DOM responses to temperature. They show that MERs are spatially transferable across climate zones and that iCER captures warming-driven functional reorganization of DOM, modulated by nutrient enrichment. Field and laboratory experiments jointly link iCER to carbon stock and decomposition processes, including greenhouse gas emissions. The work provides an MER database and open-source tools to compute iCER from FT-ICR MS datasets and suggests incorporating thermal responses into substrate-explicit biogeochemical models. Future research should extend iCER to multivariate environmental drivers (e.g., pH), include longer-term and time-series designs, broaden sample types and initial chemistry (including different sediments and mineralogy), and integrate observations along natural gradients to improve generality and model integration.

Findings are based on microcosms using standardized sediments from a single lake (Taihu), which may not represent global DOM chemistry. Incubations lasted one month, potentially underestimating responses of more resistant compounds and not capturing seasonal dynamics or microbial phenology. Field microcosms were open systems with concurrent production and decomposition, complicating attribution. FT-ICR MS introduces known biases, including incomplete extraction, ionization efficiency differences, and uncharacterized fractions. Transferability of MERs may vary across laboratories, sample types, and regions. Initial dissolved nutrients and DOC at experiment start were not measured due to mixing, and only end-state conditions were analyzed.