Clay-organic matter interactions drive microbial necromass preservation in soils

Abstract

Microbial necromass is increasingly recognized as a major source of stable soil organic matter (SOM), and its persistence is often attributed to interactions with clay-sized minerals. However, the mechanisms underlying this mineral-mediated stabilization remain poorly understood. Here, we conducted an in situ dual-labeled (¹³C and ¹⁵N) microbial necromass experiment across a clay gradient to quantify how clay content and necromass origin (bacterial vs. fungal) regulate necromass persistence. We find that higher clay content markedly enhances necromass retention by strengthening mineral protection, suppressing microbial activity and diversity, and limiting leaching losses. NanoSIMS imaging shows that new necromass preferentially associates with organic matter coatings on the rough mineral surfaces, highlighting organo-organic interfaces as important stabilization pathways. Necromass origin exerts little effect on retention despite marked differences in C:N ratios and bulk chemical composition, indicating that finer-scale molecular features, rather than broad compositional differences, govern necromass stabilization in soils.

By incubating isotope-labeled microbial necromass in the field, the study shows that new microbial necromass preferentially associate with organic matter coatings on rough mineral surfaces rather than adhering to bare minerals.

Introduction

Soil organic matter (SOM) represents the largest actively cycling carbon pool in terrestrial ecosystems¹, and its persistence is vital for maintaining soil fertility and mitigating climate change^2,3. Along with plant inputs, microbial necromass has emerged as an important source of SOM^4–7. Microbial necromass consists of the dead microbial biomass, including intact cells, cell fragments, and intracellular components released upon lysis^8,9. Compared with plant-derived SOM, necromass is enriched in nitrogen-containing and polar functional groups (e.g., amide, carboxyl, and amino groups) that confer greater affinity for soil mineral surfaces^10,11. Thus, the long-term accumulation and persistence of necromass in soil is widely attributed to its associations with secondary minerals, which counteract its inherent chemical lability and limit microbial mineralization^12,13. Consequently, soils with higher clay content are expected to store more necromass^14,15. However, this expectation does not account for how variation in clay content shapes the multiple soil environments in ways that could increase necromass decomposition and thus constrain the magnitude of necromass accumulation.

Soil clay content is thought to influence microbial necromass persistence through multiple, sometimes competing, mechanisms^16,17. On one hand, clay particles possess high surface charge and specific surface area, promoting strong adsorption of low-molecular-weight necromass and forming abundant mineral–necromass associations that limit microbial access^10,18. The fine pores and microaggregates characteristic of clay-rich soils can further occlude necromass and physically restrict its contact with microbes and extracellular enzymes^19–22, while high water retention reduces leaching losses^23,24. Additionally, reduced oxygen diffusion in moist, clay-rich soils can enhance reducing conditions and promote partial Fe(III) reduction, weakening mineral sorption and altering necromass mobility²⁵. On the other hand, these same clay-rich environments can stimulate microbial activity that accelerates necromass mineralization²⁶. Abundant fine pores create microhabitats that support microbial growth and protect microbes from predation^21,22,27, while elevated nutrient and moisture availability alleviate resource and water limitations^23,28–30, thereby enhancing microbial activity and diversity. Clay content can also influence microbial community composition^31,32, which in turn shapes both necromass production and decomposition. Overall, clay can simultaneously promote necromass protection and stimulate necromass loss, underscoring the need to quantify necromass fate under in situ conditions where these protective and degradative pathways interact.

Independent of clay content, the chemical composition of microbial necromass strongly influences its decomposition^33,34. Necromass chemistry is associated with the composition of the living microbial community^9,13, with different microbial taxa producing necromass with contrasting recalcitrance and affinity for mineral surfaces^34,35. For instance, bacterial necromass, typically richer in nitrogen (N) and lower in chitin content than fungal necromass, may be preferentially mineralized by microorganisms^36,37. Yet its higher N content can also promote stronger association with minerals, as mineral surfaces preferentially adsorb N-rich organic substrates^38,39. In contrast, fungal necromass contains chitin and other complex polysaccharides that can degrade into carboxyl-rich compounds capable of forming strong inner-sphere complexes with Fe- and Al-oxide minerals^40,41. Necromass chemistry may further dictate its spatial distribution, thereby influencing its vulnerability to decomposition⁴². For instance, N-rich protein components may preferentially accumulate on mineral surfaces, whereas cellular lipids may contribute to more microbial-accessible organo-organic interactions stacked atop existing mineral-OM associations⁴³. Understanding whether necromass with different chemistry exhibits distinct mineralization rates is crucial for predicting its persistence⁴⁴. Likewise, the balance between preferential adsorption and mineralization of N-rich necromass substrates will influence the relationship between necromass carbon (C) and N mineralization. Nevertheless, the role of necromass chemistry in SOM persistence remains poorly understood, warranting further investigation.

Here, we established an experimental clay gradient to directly assess how variation in clay content regulates in situ microbial necromass decomposition and retention (Fig. 1). We constructed model soils with contrasting clay contents by mixing montmorillonite clay, quartz sand, and plant litter while maintaining identical organic matter input and clay mineral. Following a pre-incubation period, model soils were placed in a natural forest site and amended with dual isotope-labeled (¹³C and ¹⁵N) bacterial and fungal necromass. We hypothesized that (1) necromass retention would increase with clay content, as the mineral-protective effects of clay outweigh stimulation of microbial decomposition, and (2) bacterial necromass would be more strongly retained than fungal necromass owing to its higher affinity for clay minerals. To test these hypotheses, the model soils were incubated in situ for 386 days (Fig. 1), during which the fate of the labeled necromass was traced to quantify its mineralization, retention, and redistribution among SOM pools. This experimental framework enabled us to isolate the effects of clay content and necromass origin on the persistence of microbial necromass under realistic environmental conditions.

Fig. 1. Experimental design of in situ incubation with ¹³C¹⁵N-labeled microbial necromass across a clay gradient.

a The microbial inoculum was prepared by suspending the field-collected soil in deionized water, then stirring and centrifuging the suspension. b The ¹³C¹⁵N-labeled bacterial and fungal necromass was produced by cultivating the microbial inoculum in ¹³C¹⁵N-labeled M9 salts medium with selective inhibitors, harvested by centrifugation, and then sterilized. c Model soils with low- (20%), medium- (40%), and high- (60%) clay content were constructed by mixing quartz sand, montmorillonite clay, and plant litter in PVC tubes, and then pre-incubated at 20 °C and 60% WHC for 120 days in the laboratory. d The PVC tubes containing the model soils were placed in a natural forest plot for a 30-day stabilization period. Subsequently, ¹³C¹⁵N-labeled necromass was injected and then incubated in situ for 386 days.

Results

Microbial necromass characteristics and mineralization

Bacterial necromass contained more N and exhibited a lower C:N ratio than fungal necromass (Table S1). Fourier-transform infrared spectroscopy (FTIR) analysis further showed that bacterial necromass was enriched in aliphatic compounds but depleted in aromatics and polysaccharides relative to fungal necromass (Table S1). Upon addition to the field soil, necromass was rapidly utilized by microorganisms during the initial stage of incubation (Fig. 2a, c), as reflected by the peak in ¹³C-CO₂ flux at 3–18 days. After this peak, the mineralization rate declined markedly (Fig. S1). By the end of in situ incubation (386 days), 9.7–22.9% of the added necromass ¹³C was mineralized to ¹³C-CO₂ (Fig. 2b, d). Cumulative ¹³C-CO₂ was significantly influenced by soil clay content (P < 0.001; Table S2), with the lowest values observed in high-clay soil. In contrast, no significant differences were found between bacterial and fungal groups (Table S2).

Fig. 2. Cumulative ¹³C-CO₂ release from microbial necromass mineralization.

Temporal cumulative ¹³C-CO₂ derived from a bacterial necromass and c fungal necromass during the 386-day in situ incubation in low-, medium-, and high-clay soil. The total respired ¹³C-CO₂ derived from b bacterial and d fungal necromass. Values represent means and standard error (n = 3). Different lowercase letters indicate significant differences (P < 0.05) among clay treatments.

The retention of the added necromass ¹³C and ¹⁵N, quantified as total recovery in the model soils, decreased rapidly during the first 10 days in low- and medium-clay soil (Fig. 3a–d). In contrast, the decline lasted for the first 30 days in high-clay soil, followed by a more gradual decrease thereafter (Fig. 3a–d). After 386 days of in situ incubation, 12.3–31.5% of the added necromass ¹³C remained in the model soils (Fig. 3a, b), significantly lower than the 24.6–45.4% retention of the added necromass ¹⁵N (P < 0.001, Fig. 3c, d). Accordingly, the ¹⁵N:¹³C retention ratios exceeded 1 across all clay contents (Fig. S2). Both ¹³C and ¹⁵N retention were significantly lower in low- and medium-clay soil than in high-clay soil (C: P < 0.001, N: P < 0.001; Table S3), with no significant differences between bacterial and fungal groups. Mineralization rate constants of the added necromass ¹³C (1.12–2.15 year⁻¹) were higher than those of the necromass ¹⁵N (0.76–1.36 year⁻¹; Fig. 3e, f). The mineralization rate constants of the added necromass ¹³C and ¹⁵N were significantly higher in low- and medium-clay soil compared to high-clay soil (C: P = 0.021, N: P = 0.032; Table S4), with no difference between microbial groups (Table S4).

Fig. 3. Retention and mineralization rate constant of microbial necromass.

The retention of a, c bacterial and b, d fungal necromass-derived ¹³C and ¹⁵N during the 386-day in situ incubation in low-, medium- and high-clay soil. Mineralization rate constant of e bacterial and f fungal necromass ¹³C and ¹⁵N. Values represent means and standard error (n = 3). Lowercase letters (P < 0.05) and asterisks (*P < 0.05; **P < 0.01; ***P < 0.001) indicate significant differences among clay treatments.

Partitioning of microbial necromass-derived C and N among soil pools

To investigate the fate of the added necromass, we traced its ¹³C and ¹⁵N signals in microbial biomass, inorganic N (IN), dissolved organic matter, remaining necromass in the model soils, and leached fractions in the underlying subsoil (5.5–14 cm depth). By day 10, 2.68% (±0.46%) of added necromass ¹³C and 0.77% (±0.15%) of ¹⁵N were recovered in microbial biomass, with both significantly influenced by clay content and microbial group (P < 0.001; Fig. 4; Figs. S3, S4; Tables S5, S6). Across all treatments and sampling times, less than 0.65% of the added necromass ¹⁵N was mineralized to IN, showing only a slight increase with clay content (P = 0.064; Fig. 4; Figs. S3, S4; Table S5). Similarly, recovery in dissolved organic matter was minimal, declining from 0.38% (±0.06%) of ¹³C and 0.10% (±0.01%) of ¹⁵N at day 10 to 0.09% (±0.02%) of ¹³C and 0.02% (±0.004%) of ¹⁵N at day 386, with ¹³C slightly increasing and ¹⁵N decreasing with clay content (P = 0.091 and 0.030; Fig. 4; Figs. S3, S4; Tables S5, S6). By day 10, 41.6% (±5.6%) of added necromass ¹³C and 51% (±6.3%) of added necromass ¹⁵N remained as necromass, declining rapidly thereafter. By the end of the incubation, 15.7% (±2%) of ¹³C and 31.2% (±2.7%) of ¹⁵N were retained in necromass, with ¹³C recovery significantly lower than ¹⁵N recovery (P < 0.001; Fig. 4). Both remaining necromass ¹³C and ¹⁵N were higher in high-clay soil than in low- and medium-clay soil (P < 0.001; Tables S5, S6). The added necromass leached rapidly from the model soils into the underlying subsoil (soil depth ~8.5 cm), with 2.97–51.39% of ¹³C and 2.54–60.37% of ¹⁵N recovered at day 10. Leaching losses were higher in the low- and medium-clay treatments (all P < 0.001; Fig. 4). Over the 386-day incubation, ¹³C and ¹⁵N leaching declined in low- and medium-clay soil but increased in high-clay soil (Fig. 4; Table S6). Overall, more ¹³C than ¹⁵N was leached into the subsoil (P = 0.033), indicating preferential mobilization of C relative to N.

Fig. 4. Allocation of added necromass ¹³C and ¹⁵N among soil pools.

Recovery of added a, b bacterial and c, d fungal necromass ¹³C or ¹⁵N in different pools in the model soils, including necromass C and N, microbial biomass C and N (MBC, MBN), dissolved organic C and N (DOC, DON), and inorganic N (IN), as well as subsoil total C and N (5.5–14 cm depth) and the proportion of ¹³C and ¹⁵N not recovered within the PVC soil incubations (assumed lost mainly as CO₂, N₂O, or via leaching) after a, c 10 and b, d 386 days of in situ incubation in low-, medium- and high-clay soil.

Drivers of microbial necromass retention

Despite clear chemical differences between bacterial and fungal necromass, their persistence did not differ (Fig. 3; Tables S2–S4), so the two types were combined for subsequent analyses. To identify the factors regulating necromass retention, we examined relationships between the retention of the added necromass ¹³C and ¹⁵N after 386 days of decomposition and a suite of potential influencing factors, including microbial community characteristics and soil properties (see “Methods” section; Fig. 5). Necromass ¹³C and ¹⁵N retention was closely related to the living microbial community characteristics and soil properties (Fig. 5a, b). Furthermore, bacterial community richness (Richness-B), soil moisture, total porosity, specific surface area (SSA), fungal community richness (Richness-F), and microbial metabolic quotient (qCO₂) were the most important predictors (Fig. 5b). Necromass ¹³C and ¹⁵N retention was positively correlated with soil moisture and SSA, but negatively correlated with richness-B, richness-F, and qCO₂ (Fig. 5c; Fig. S3). Moreover, qCO₂ increased with bacterial and fungal richness and with air-filled porosity (AFP), but declined with soil moisture (Fig. 5a). Both bacterial and fungal richness showed the same pattern, increasing with AFP and decreasing with soil moisture (Fig. 5a). Across the clay gradient, soil moisture, SSA, and total porosity increased with clay content, whereas AFP, qCO₂, bulk density, and microbial richness all decreased (Fig. 5a).

Fig. 5. Drivers of microbial necromass retention.

a Principal component analysis (PCA) of the factors influencing microbial necromass retention after 386 days of in situ incubation. Bacterial and fungal necromass results were combined, as their retention did not differ significantly. Points represent individual samples grouped by treatment, and arrows indicate variable loadings. The influencing factors include Richness-B and Richness-F (bacterial and fungal community richness), PCoA1-B and PCoA1-F (bacterial and fungal community compositions), qCO₂ (microbial metabolic quotient), soil moisture, total porosity, SSA (specific surface area), AFP (air-filled porosity), and bulk density. b Contribution of microbial community and soil properties to microbial necromass retention. c Linear regressions (n = 18) between the necromass retention and the five most important factors.

To resolve the microscale mechanisms underlying necromass stabilization, we examined the spatial arrangement and interactions between added labeled necromass, native OM, and mineral particles using scanning electron microscopy (SEM) and nanoscale secondary ion mass spectrometry (NanoSIMS) (Fig. 6). These complementary techniques allowed direct visualization of the co-localization between labeled necromass and soil constituents. Consistent with the bulk IRMS results, NanoSIMS showed higher ¹³C⁻:(¹²C⁻ + ¹³C⁻) and ¹²C¹⁵N⁻:(¹²C¹⁴N⁻ + ¹²C¹⁵N⁻) ratios in high-clay soil (Fig. 6d, g). Native OM, identified by the ¹²C⁻ and ¹²C¹⁴N⁻ secondary ion, was predominantly associated with minerals (79.5 ± 1.6%), as indicated by the ¹⁶O⁻ secondary ion (Fig. 6a¹–c²; Fig. S5). Mineral surfaces coated with OM were generally rougher, as reflected by a higher ¹⁶O⁻:area ratio (Fig. 6i). Newly added necromass ¹³C and ¹⁵N accumulated mainly via organo-organic interactions: labeled necromass preferentially attached to native OM already associated with mineral surfaces (79.8 ± 3.2% of ¹³C, 87.3 ± 1.9% of ¹⁵N), rather than bare mineral surfaces (Fig. 6a¹–c², e, h). High-clay soil exhibited a larger specific surface area covered by OM-mineral associations than low- and medium-clay soil (Fig. 6f). Density fractionation further confirmed that most retained microbial necromass was located within the mineral-associated organic matter fraction (96.2 ± 0.4% of ¹³C, 93.3 ± 0.5% of ¹⁵N; Fig. S6).

Fig. 6. Spatial distribution of retained necromass and native OM on mineral surfaces.

Scanning electron microscopy (SEM) and nanoscale secondary ion mass spectrometry (NanoSIMS) images of the a low-, b medium-, and c high-clay soil (solid line square indicates the region examined using NanoSIMS). The mineral is revealed by ¹⁶O⁻ secondary ion images, native OM is revealed by ¹²C⁻ and ¹²C¹⁴N⁻ secondary ion images, and new, necromass-derived (sub-panels a¹, b¹, and c¹) ¹³C or (a², b², and c²) ¹⁵N is indicated by areas where the ¹³C⁻:(¹²C⁻ + ¹³C⁻) or ¹²C¹⁵N⁻:(¹²C¹⁴N⁻ + ¹²C¹⁵N⁻) ratios were enriched compared with the ratios in the unlabeled soil. d ¹³C⁻:(¹²C⁻ + ¹³C⁻) and g ¹²C¹⁵N⁻:(¹²C¹⁴N⁻ + ¹²C¹⁵N⁻) ratios were quantified by examining the regions of interest of the low-, medium-, and high-clay soil. The natural abundance of ¹³C and ¹⁵N is indicated by the dashed line. Calculated proportions (%) of necromass-derived e ¹³C or h ¹⁵N spatial distribution on the OM-free mineral surfaces (i.e., mineral-¹³C or ¹⁵N) and mineral-OM associations (i.e., mineral-OM-¹³C or ¹⁵N) surfaces (n = 15). f The specific surface area covered by mineral-OM associations of the low-, medium-, or high-clay soil (n = 15). Area size is calculated by multiplying the proportion of the mineral surface covered by OM by the specific surface area of model soils. i ¹⁶O⁻:area ratio of the OM-free mineral and OM-covered mineral area (n = 45). Different lowercase letters indicate significant differences among clay treatments.

Discussion

Our in situ necromass incubation showed that higher clay content promoted the retention of necromass ¹³C and ¹⁵N in soils. This occurred despite the concurrent increase in microbial biomass, consistent with our first hypothesis. Comparable relationships between the mineralization of plant-derived organic matter and clay content have also been reported in previous studies^45,46. Our quantitative NanoSIMS imaging revealed that almost all newly accumulated necromass C and N were associated with OM already bound to these rough mineral surfaces, whereas only a minor fraction occurred on bare mineral surfaces. Compared with smooth mineral surfaces, rough mineral surfaces exhibited stronger ¹⁶O⁻ ion intensities, indicating a greater enrichment of hydroxyl groups, oxides, and hydrated layers that provide abundant reactive sites for organic matter association^18,47,48. High-clay soil contained more rough mineral surfaces, which promote the formation of mineral-OM associations and provide more reactive sites for new necromass preservation. These results imply that, although only a limited proportion of clay mineral surfaces was directly involved in necromass sequestration, they can promote the formation of multilayered, onion-like structures via the mineral-OM-necromass association, thereby enabling the stabilization of substantial microbial necromass. This mechanistic insight complements traditional sorption-based concepts by demonstrating that organo-organic interactions occurring on OM-coated mineral surfaces represent a key pathway for the accumulation of new necromass⁴³.

Microbial biomass increased with clay content; however, this did not result in greater necromass mineralization. The higher soil moisture (Fig. S7; Tables S7, S8) observed under high clay likely increased microbial nutrient use efficiency by reducing substrate diffusion constraints^49,50. Elevated moisture also sharply reduced air-filled porosity (Fig. S8), restricting oxygen diffusion and consequently suppressing microbial activity⁵¹, respiration⁵², and necromass decomposition (Figs. S9 and S10). Oxygen limitation was further associated with reduced bacterial diversity and lower abundance of copiotrophic taxa such as Firmicutes and Actinobacteria that are involved in the turnover of necromass-derived polysaccharides and proteins²⁸ (Figs. S11–S13). By exposing our constructed clay gradient to natural fluctuations in moisture and microbial community composition, we show that clay indirectly regulates necromass persistence through its strong effects on soil water status, oxygen availability, and microbial functioning.

Our in situ incubation further reveals that leaching represents an important necromass loss pathway. Within 10 days, up to 51.4% of the added necromass ¹³C and 60.8% of the added necromass ¹⁵N were leached into the subsoil following rainfall (Fig. S14), with leaching decreasing significantly as clay content increased. This substantial early leaching likely reflects the high mobility of low-molecular-weight microbial necromass⁵³. Supporting this, FTIR analysis showed that the microbial necromass contained more low-molecular-weight lipid compounds and fewer high-molecular-weight aromatic and polysaccharide compounds (Fig. S15) than plant litter⁵⁴. These findings suggest that the previously reported high proportion of necromass in subsoil organic matter¹⁴ partly originates from the leaching of microbial processing products of surface plant litter. The pronounced reduction in necromass in the subsoil of low-clay soil from day 10 to day 386 indicates ongoing vertical transport, further supporting our interpretation. These findings provide mechanistic support for the large contributions of necromass previously reported in subsoil organic matter and highlight the need to incorporate vertical transport into necromass and SOM models. By directly measuring necromass leaching in situ, we show that it is a significant loss pathway, strongly modulated by soil clay content.

Contrary to our second hypothesis, the source of microbial necromass did not strongly determine its retention in soils^44,55, as bacterial and fungal necromass exhibited similar recovery and mineralization rate constants. This indicates that broad compositional differences do not strongly govern necromass fate in this study. Instead, stabilization appears to depend on finer-scale molecular properties that are shared across necromass sources, such as low molecular weight and abundant polar functional groups^10,55,56, which facilitate association with OM-coated mineral surfaces^35,43. This result implies that shifts in microbial community composition are more likely to influence necromass persistence primarily through changes in activity or efficiency, rather than through inherent differences in the sorption potential of the necromass produced.

Consistent with previous observations⁵⁷, the added necromass ¹⁵N was more persistent than necromass ¹³C. The C:N ratio of our model soils exceeded the commonly recognized threshold of ~20 (Table S9), above which microbial growth typically shifts from C limitation to N limitation⁵⁸. Under such N-limited conditions, microorganisms preferentially retain N released from necromass decomposition to support biomass synthesis, leading to greater necromass-derived N accumulation⁵⁹. Consistently, only a small fraction of necromass N (0.23–0.65%) was recovered in the inorganic N pool (Fig. 4), further indicating microbial N limitation in our model soils. These results suggest that the contribution of necromass to SOM is partly modulated by soil nutrient status.

Although our study provides clear mechanistic insights, several limitations should be acknowledged. First, the injected necromass was not initially associated with minerals as naturally occurring necromass would be, making a portion more susceptible to early leaching. However, mineral adsorption appears to occur rapidly³⁵, and aggregate protection played only a minor role under our conditions, suggesting that this effect did not substantially bias the overall patterns of necromass retention and decomposition. Consistently, leaching in the high-clay treatment quickly declined to low levels, with only about 7% of necromass-derived ¹³C and 7.5% of ¹⁵N moving into the subsoil after 10 days, corresponding to a possible overestimation of the decomposition rate constant by 0.068 and 0.073 year⁻¹, respectively. Second, the fungal necromass used in this experiment originated from liquid cultures, producing diffuse necromass rather than rhizomorphic structures commonly formed in soils. Because diffuse necromass generally decomposes more rapidly than rhizomorphic necromass, the mineralization rate of fungal necromass in our study may be slightly overestimated. Despite these considerations, the main mechanisms identified here, including mineral-OM interfacial protection, moisture and oxygen constraints, and suppression of leaching, remain robust and provide a reliable basis for interpreting necromass dynamics across clay gradients.

Overall, our findings show that clay regulates necromass persistence through the combined influence of OM-mineral interfaces, microbial physiological constraints, and leaching losses. These interacting pathways reconcile previous contradictory observations regarding clay-necromass relationships and underscore the complex nature of necromass stabilization under natural soil conditions. By disentangling mineral protection, microbial processes, and leaching in an in situ setting, this study provides a mechanistic foundation for improving how microbial contributions to SOM stabilization are represented in soil carbon models and Earth System Models.

Methods

Study site and experimental design

The study was conducted in the Changbai Mountain National Nature Reserve, located in Jilin Province in northern China (42.70° N,127.63° E). The region has a typical temperate climate, with a mean annual temperature of 3.6 °C and a mean annual precipitation of 745 mm. The predominant coniferous species in the forest is Pinus koraiensis, and the broad-leaved species are Quercus mongolica, Tilia amurensis, and Fraxinus mandshurica. The soils are Albic Luvisols developed from volcanic ash⁶⁰. In October 2017, approximately 10 kg of fresh leaf litter from the forest floor, primarily consisting of Pinus koraiensis, Quercus mongolica, Fraxinus mandshurica, Tilia amurensis Rupr., and Sophora japonica Linn., was collected, dried, ground, and sterilized to use as the organic matter for building the model soils.

In May 2018, nine experimental plots (3 m × 3 m) were randomly established with >5 m buffer zones between adjacent plots. Six soil cores (5 cm in diameter, 20 cm in depth) were randomly collected from each plot after removing the litter layer. The cores were combined into one composite sample, immediately transported to the laboratory on ice, and homogenized for microbial inoculum extraction.

We created low-clay, medium-clay, and high-clay model soil with 20%, 40%, and 60% clay content, respectively, using quartz sand, montmorillonite, and plant litter (Fig. 1). The model soils were placed in PVC tubes (5 cm in diameter and 14 cm in height) with closed bottoms using PVC pipe caps, inoculated with the microbial inoculum, and pre-incubated for 120 days in the laboratory⁶¹. After pre-incubation, the PVC tubes were installed in the experimental plots for a stabilization period of 30 days. Subsequently, ¹³C¹⁵N-labeled bacterial and fungal necromass were injected into the model soils as tracers and incubated for 386 days under in situ conditions (Fig. 1). The experiment consisted of nine treatments: either (a) with ¹³C¹⁵N-labeled bacterial necromass, (b) ¹³C¹⁵N-labeled fungal necromass, or (c) without necromass (controls were injected with water-only); each of which was applied to soils of three different clay contents (low, medium, and high). In each of the 6 necromass plots, 15 PVC tubes with bacterial or fungal necromass were installed, while in each of the 3 control plots, 7 PVC tubes without added necromass were installed, resulting in a total of 111 PVC tubes.

Preparation of model soils

The quartz sand (composed of 30% silt- and 70% sand-sized particles; sourced from Aladdin) was cleaned in diluted acid and combusted at 500 °C to remove organic carbon, and then was ground and sieved to fine (50–250 μm) and coarse (250–2000 μm) particle sizes. Detailed properties of the model soil materials are listed in Table S9. A typical expandable 2:1 phyllosilicate clay (montmorillonite; sourced from Sigma-Aldrich) was mixed with quartz sand in varying proportions to form low-clay soil (sandy loam soil; 20% clay, 24% fine quartz, and 56% coarse quartz), medium-clay soil (clay loam soil; 40% clay, 18% fine quartz and 42% coarse quartz) and high-clay soil (clay soil; 60% clay, 12% fine quartz and 28% coarse quartz). The organic matter obtained from the forest dried leaf litter was mixed at 45 mg g⁻¹ model soil. The specific surface area of the low, medium, and high-clay soil was 27.1, 59.1, and 94.9 m² g⁻¹ soil, respectively. The water holding capacity (WHC) of low-, medium-, and high-clay soil was 0.60, 0.89, and 1.21, respectively, and increased significantly with clay content (Table S10).

Approximately 80 g of the model soil was placed in a PVC tube (5 cm in diameter, 14 cm in height) for each sample. The height and bulk density of the model soils were ~5.5 cm and 0.75 g cm⁻³, respectively. A total of 37 samples were prepared for each clay treatment. We created the microbial inoculum by suspending 10 g of the field-collected soil in 90 mL of deionized water⁴⁵. After 3 days, each sample was inoculated with the microbial inoculum. All samples were pre-incubated at 20 °C and 60% WHC for 120 days to establish a stable soil structure and microbial community. Litter water extract was added biweekly (1.4 mg C, 0.027 mg N g⁻¹ soil). The litter water extract was created by mixing 150 g dried forest leaf litter with 2 L of deionized water and shaking for 24 h before filtering⁶².

Preparation of labeled microbial necromass

The ¹³C¹⁵N-labeled bacterial and fungal necromass was prepared by cultivating the microbial inoculum in a minimal M9 salts medium with 99 atom% ¹³C-glucose and 99.5 atom% ¹⁵NH₄Cl in a rotary shaker incubator. The bacterial culture was grown in 500 mL batches for 3 days at 37 °C with the addition of a fungal inhibitor (cycloheximide). The fungal culture was grown in 500 mL batches for 4 weeks at 28 °C with the addition of bacterial inhibitors (ampicillin and streptomycin). The bacterial and fungal cultures were then collected by centrifugation, sterilized by ultraviolet irradiation, mixed thoroughly, and ground into powder. This approach yielded bacterial and fungal necromass with a much broader species composition than necromass generated from pure cultures. The atom % ¹³C and atom % ¹⁵N of microbial necromass exceeded 71.95% and 92.08%, respectively (Table S1). The phospholipid fatty acid (PLFA) analysis confirmed that bacterial necromass predominantly consisted of bacterial species, while fungal necromass predominantly consisted of fungal species (Table S11).

Field experiment with dual-labeled microbial necromass

After laboratory pre-incubation of the model soils, on 19 August 2018, the PVC tubes were transported to the established field plots. Plots were randomly assigned to each treatment. In each necromass plot, we first filled the top ~8.5 cm of the PVC tubes with natural subsoils (5.5–14 cm), then turned the PVC tubes over and removed the PVC caps. We then inserted 15 PVC tubes of the same clay treatment into the soil to a depth of 14 cm after removing the natural soil. During the experiment, both ends of the PVC tubes were open, and the model soils were separated from the surrounding natural soil by a 1 mm nylon mesh, which allowed physical contact while preventing direct mixing. In each control plot, 7 PVC tubes were installed using the same process. The PVC tubes were placed in the plots for about one month prior to the addition of necromass to minimize potential disturbance effects associated with long-distance transport of the model soils.

On 18 September 2018, the necromass (10 mg dry weight) was suspended in deionized water and injected with a side-port needle in a 10-point grid pattern at a depth of 3 cm in the model soils within the PVC tubes. The added microbial necromass corresponded to an average of 9% and 22% of model soil microbial biomass C and N, respectively. Within each necromass plot, 15 PVC tubes were injected either with ¹³C¹⁵N-labeled bacterial or fungal necromass, while in each control plot, 7 PVC tubes were injected with deionized water.

Three PVC tubes with bacterial necromass or three PVC tubes with fungal necromass from each necromass plot, and one control PVC tube from each control plot were collected on 28 September 2018 (10 days), 18 October 2018 (30 days), 24 April 2019 (218 days), and 9 October 2019 (386 days). The soils from the collected PVC tubes were divided into model soils and natural subsoils. A portion of model soils collected at 10 and 386 days was stored at 4 °C before measuring soil microbial biomass C and N, dissolved organic C and N, inorganic N, and water content. The natural subsoils and a portion of the model soils collected at the four sample times were air-dried for soil organic C and total N analyses. The remaining model soils collected at 386 days were stored at −80 °C for microbial community DNA extraction. To estimate necromass-derived CO₂-C production, three PVC tubes of each necromass and control treatment were attached to a 1 L PVC chamber on 21 September 2018 (3 days), 28 September 2018 (10 days), 6 October 2018 (18 days), 12 October 2018 (24 days), 21 October 2018 (33 days), 28 October 2018 (40 days), 22 April 2019 (216 days) and 9 October 2019 (386 days). Then, a 150 mL gas sample was taken from each chamber at 0 h and 2 h using a syringe and stored in air bags for CO₂ concentration and isotope measurement. The CO₂ concentration and isotope ratio were analyzed using a CO₂ isotope analyzer (CCIA-36d-EP, LGR, USA).

Physical fractionation

The model soils at day 386 were separated into four distinct OM fractions using a combined size-density fractionation protocol⁶³. Air-dried model soil (18–20 g) was gently capillary-saturated with sodium iodide solution (NaI; 1.8 g cm⁻³) after 12 h, and the free-floating particulate organic matter (fPOM) was collected using a vacuum pump. Occluded POM (oPOM) was released from soil aggregates by ultrasonic dispersion (Bandelin Sonopuls HD 2200; 440 J mL⁻¹)⁶⁴, enabling its separation from the heavier mineral fractions. Both fPOM and oPOM fractions were washed several times with deionized water to remove excess salt. Coarse and fine fractions were separated by wet sieving (53 μm), but because the coarse fraction contained negligible necromass, the fractions were pooled and treated as mineral-associated organic matter fraction (MAOM).

Necromass and soil chemical analyses

The specific surface area (SSA) of model soil materials and bulk model soils was determined by N₂ adsorption at 77 K and subsequent desorption of nitrogen with a Tristar 5-point BET-instrument on freeze-dried samples. The total soil porosity was estimated from the measured bulk density and the particle density. The air-filled porosity was calculated as the difference between the total porosity and volumetric water content. The N and OC contents of the necromass and soil samples were determined by combustion using an elemental analyzer (Vario EL III; Elementar, Hanau, Germany). Inorganic N, including ammonium (NH₄⁺-N) and nitrate (NO₃^–-N), was extracted from fresh model soils using 2 M potassium chloride at a soil:water ratio of 1:5 and analyzed using a continuous flow injection analyzer (Autoanalyzer 3 SEAL, Bran and Luebbe, Norderstedt, Germany). The N isotope ratios of NH₄⁺-N and NO₃⁻-N were measured using the ammonium diffusion method⁶⁰. Microbial biomass C (MBC) and N (MBN) were determined by a fumigation-extraction method⁶⁵. Two subsamples (fumigated and non-fumigated) of fresh model soils were extracted with 0.5 M K₂SO₄. The non-fumigated soil extracts were used to measure dissolved organic C (DOC) and N (DON). The C and N concentrations of the extracts were determined by a TOC-500 analyzer (Shimadzu Corporation). The C or N isotope ratios of necromass C and N, soil organic C and total N, MAOM fractions, dissolved organic C and N, microbial biomass C and N, and inorganic N were analyzed using an Elementar Vario EL Cube (Elementar Analysis System GmbH) interfaced with an isotope ratio mass spectrometer (IsoPrime100, IsoPrime Limited, Stockport, UK).

To characterize the chemical composition of bacterial and fungal necromass, Fourier-transform infrared spectroscopy (FTIR) analysis was performed. Briefly, freeze-dried bacterial and fungal necromass were finely ground, mixed with KBr at a ratio of 1:100 (wt:wt), and analyzed using a Thermo Nicolet 6700 infrared spectrometer (Thermo Electron Scientific Instruments Corp., Madison, WI, USA). Absorbance spectra (4000–400 cm⁻¹) were recorded with 64 scans and a resolution of 4 cm⁻¹. Absorbance intensity at 700, 860, and 975 cm⁻¹ corresponded to aromatic groups; 2855, 2930, and 3300 cm⁻¹ were assigned to aliphatic groups; and 1075 cm⁻¹ corresponded to the polysaccharide group⁶⁶. The relative absorbance intensity (rA, %) of each of the seven bands was normalized to the percent absorbance intensity of the sum of absorbance intensities of the seven bands.

To characterize the microbial community composition of bacterial and fungal necromass, PLFA analysis was performed. Briefly, 80 mg freeze-dried necromass was extracted with a chloroform:methanol:phosphorus-buffer (volume ratio: 1:2:0.8). After the organic phase was separated, dried, purified, and subjected to methyl esterification, lipids were analyzed using an Agilent 7890B GC (Agilent Technologies, Santa Clara, CA, USA) equipped with MIDI software. Microbial composition was reflected by the biomarker PLFAs of different microbial groups⁶⁷.

SEM and NanoSIMS micro-spectroscopy

To directly visualize the distribution of added microbial necromass on mineral surfaces, we used SEM and NanoSIMS to analyze the model soils at 386 days. Dried model soil samples (<0.15 mm) were dispersed in deionized water, and 10 µL of the suspension was deposited onto a germanium wafer and dried overnight in a desiccator. To avoid the charging phenomena, samples were gold-coated prior to SEM analysis by physical vapor deposition under an argon atmosphere (Emitech Sputtercoater SC7620, Gala Instrumente, Bad Schwalbach, Germany). The microscale structures of mineral–necromass associations were first examined by SEM (Jeol JSM 5900LV, Freising, Germany). Representative regions were then analyzed using a Cameca NanoSIMS 50L (Cameca, Gennevilliers, France). For NanoSIMS analysis, a high Cs⁺ primary ion beam current was used to remove surface contamination and gold coating, and to implant Cs⁺ ions into the sample to enhance the yield of secondary ions. Secondary ions of ¹²C⁻, ¹³C⁻, ¹²C¹⁴N⁻, ¹²C¹⁵N⁻, ¹⁶O⁻, ²⁸Si⁻, and ²⁷Al¹⁶O⁻ were collected simultaneously using electron multipliers with a fixed dead time of 44 ns. Ion images were acquired with a dwell time of 2 ms pixel⁻¹, at a resolution of 512 × 512 pixels over a 35 × 35 µm area in a single scan cycle. For each sample, five regions of interest were analyzed to obtain a representative dataset for quantifying ¹³C and ¹⁵N distribution.

NanoSIMS image data were processed using the OpenMIMS plugin in Fiji. All images were corrected for electron multiplier dead time (44 ns). The ion images of ¹²C⁻ and ¹²C¹⁴N⁻ were merged for an indicator of native OM⁴⁸. Regions of interest (ROIs) on these images were selected using the threshold option with the default method. All ROIs with an area greater than 10 pixels² were used for further calculations. The ¹³C⁻:(¹²C⁻ + ¹³C⁻), ¹²C¹⁵N⁻:(¹²C¹⁴N⁻ + ¹²C¹⁵N⁻) ratios, and the area of the ROIs were extracted from all images and used for further calculations. The new, necromass-derived OM is defined as ROIs in the ¹³C⁻:(¹²C⁻ + ¹³C⁻) and ¹²C¹⁵N⁻:(¹²C¹⁴N⁻ + ¹²C¹⁵N⁻) ratio image with pixel values greater than natural abundance levels.

Amplicon sequencing and data processing

Soil genomic DNA was extracted from 0.25 g freeze-dried soil collected on the last sampling time using PowerSoil DNA Isolation Kits (MoBio Laboratories, Carlsbad, CA, U.S.A.). Bacterial and fungal communities were assigned by amplifying and sequencing the V4–V5 region of the 16S rRNA gene and the ITS1 region, respectively. After PCR amplification and purification, 16S rRNA and ITS gene sequencing were performed using the Illumina platform. Raw sequences shorter than 200 bp or with a quality score below 20 (Q < 20) were removed prior to analysis using the QIIME 2 pipeline. Quality-filtered sequences of 16S rRNA and ITS gene amplicons were clustered into operational taxonomic units (OTUs) at 97% sequence similarity using UPARSE. Taxonomic classification of representative bacterial and fungal ASVs was conducted using the SILVA database and the UNITE database, respectively. Observed OTUs and phylogenetic diversity were calculated to assess microbial alpha diversity.

Data analysis

To determine the fraction of necromass-derived C or N (C_frac or N_frac) in the model soils, subsoil, microbial biomass, dissolved organic matter, inorganic nitrogen, and headspace gas, we calculated the fraction of each pool that was necromass-derived by the following isotope mixing model:

$${C_{frac}\mathrm{or}N}_{frac}=\frac{{\mathrm{atom}\%}_{\mathrm{Sample}}-{\mathrm{atom}\%}_{\mathrm{Control}}}{{\mathrm{atom}\%}_{\mathrm{Necromass}}-{\mathrm{atom}\%}_{\mathrm{Control}}}$$ 1

where atom%_Sample is the atom% of ¹³C or ¹⁵N value of the C or N pools (i.e., MBC, MBN, DOC, DON, CO₂-C, IN, model soil and subsoil C or N) of a given sample; atom%_Control is the atom% of ¹³C or ¹⁵N in the equivalent C or N pool from the control; atom%_Necromass is the atom% of ¹³C or ¹⁵N in the labeled microbial necromass added to soils.

The ¹³C or ¹⁵N recovery was calculated as the percentage of labeled C or N remaining in each C or N pool (MBC, MBN, DOC, DON, IN, bulk soil, and subsoil C or N) relative to initial necromass inputs. The ¹³C recovery in CO₂-C was calculated using linear interpolation between sample points. The ¹³C recovery in necromass C was calculated by subtracting the ¹³C recovery in MBC and DOC from the ¹³C recovery in model soil C. The ¹⁵N recovery in necromass N was calculated by subtracting the ¹⁵N recovery in MBN, DON, and IN from the ¹⁵N recovery in model soil N.

Due to the variability and limited sampling times for ¹³C and ¹⁵N recovery of labeled necromass, the commonly used one- or two-pool exponential decay models could not accurately simulate our data. Therefore, we determined the mineralization rate constant of necromass ¹³C or ¹⁵N (k_C or k_N) using a linear regression on the natural log transformation of ¹³C or ¹⁵N recovery⁶⁸.

$$k_C\mathrm{or}{k}_N=\ln \left({\mathrm{S}}_{\mathrm{t}}/{\mathrm{S}}_{{\mathrm{t}}_0}\right)/\left(\mathrm{t}-{\mathrm{t}}_0\right)$$ 2

where S_t is the recovery of necromass ¹³C or ¹⁵N in bulk soil at the end of the experiment (t); S₀ is the recovery of necromass ¹³C or ¹⁵N in bulk soil at the beginning of the experiment (t₀).

A two-way analysis of variance (ANOVA) was conducted to assess the effects of clay content (low-, medium-, and high-clay) and microbial groups (bacteria and fungi) on the mineralization rate constant and cumulative gaseous emissions of necromass. A three-way ANOVA was conducted to assess the effects of clay contents, microbial groups, and time on necromass fates in different soil C or N pools. A repeated measures ANOVA model (MANOVA) was used to compare the effects of clay contents, microbial groups, and time on the retention and gaseous emissions of necromass. Statistical findings were considered significant if the confidence level was in excess of 95% (P < 0.05). The above statistical analyses were conducted using SPSS 19.0 (SPSS Inc., Chicago, IL, USA). Additionally, linear regressions were used to analyze the relationship between the necromass retention and bacterial community diversity (richness-B), soil moisture content, specific surface area (SSA), microbial metabolic quotient (qCO₂), and air-filled porosity (AFP) at the end of incubation. Principal component analysis (PCA) was conducted to explore relationships between necromass retention and environmental variables. Principal coordinates analysis (PCoA) with Bray-Curtis distance was conducted to explore differences in microbial community composition using amplicon data. Statistical differences in the microbial community composition were tested using permutational multivariate analysis of variance (PERMANOVA). Both PCA and PCoA analyses were conducted in R (v. 4.4.3) using the vegan package⁶⁹. The hierarchical partitioning method was employed to quantify the contribution of microbial community and soil properties to necromass retention in R (v. 4.4.3) using the “rdacca.hp” package⁷⁰.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

X.W., C.W., and E.B. designed the study. X.W., C.W., and P.J. performed the experiment. X.W., C.P.S., and L.F.S. performed the laboratory analyses. X.W., Y.L., and C.W. analyzed the data. X.W., C.M.K., M.A., K.G., E.B., and C.W. wrote the manuscript. All authors contributed to the article and approved the submitted version.

Peer review

Peer review information

Nature Communications thanks Javier Cuadros and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

Microbial amplicon sequencing data have been deposited in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) data repository under accession no. PRJNA1191639. All aggregated data supporting the findings of this study are available in the Figshare repository: 10.6084/m9.figshare.27896424.

Code availability

The main R code used in this study is available in the Figshare repository: 10.6084/m9.figshare.27896424.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-70156-1.