In 2022, soils with low fertility (LF) and high fertility (HF) were collected from the Hailun Agroecological Experimental Station, Chinese Academy of Sciences (126°38′E, 47°26′N), Heilongjiang Province, Northeast China. Soil samples from the 0–20 cm surface layer were collected from selected sites and transported to Nanjing University of Information Science and Technology. In the laboratory, soil samples from the same site were composited, manually cleared of visible plant roots and debris, air-dried, and sieved through a 2 mm mesh. The basic physicochemical properties of the two soils are presented in Table 1.

Values are represented as the mean ± standard error (n = 3). Different lowercase letters between the two soils represent significant differences (P < 0.05). SOC, soil organic carbon; Total N, total nitrogen; Total P, total phosphorus. LF, low-fertility soil; HF, high-fertility soil.

Considering that the average C:N requirement for microbial biomass is approximately 25:1, litter with a C:N ratio below 25 was classified as high-quality, whereas that above 25 was classified as low-quality [42]. Seven types of litter with markedly different C:N ratios (Table 2) were incorporated into the soil in this study. High-quality litters were collected from wheat (Triticum aestivum L.) and rice (Oryza sativa L.), separated into shoot and root fractions. Low-quality litters were collected from three representative tree species—Acronychia pedunculata, Cryptocarya chinensis, and Schima superba—growing in subtropical forests, consisting of freshly leaves. Labelled ¹³C litters, obtained in 2022 by performing multiple ¹³CO₂ in situ pulse labeling [43] during the plant growing season in the glasshouse and collected whole plants (high-quality litter) or leaves (low-quality litter). All litter samples were transported to the laboratory, washed, blanched at 105°C for 30 min, oven-dried at 80°C to a constant weight, ground to a fine powder, and passed through a 0.25 mm mesh sieve.

A completely randomized block experiment was designed with three factors: litter quality (seven litter types with different C:N ratios), soil type (LF and HF), and temperature (23°C and 33°C), along with additional “no-litter” controls, each with three replicates. After adjusting soil moisture to 60% water-holding capacity (WHC), all soils were pre-incubated in the dark at 25°C for 7 days. After pre-incubation, soil samples (30 g dry mass) were placed into specimen cups and thoroughly mixed with the designated litter powder. The amount of C from each added litter represented the rate of 5% of average SOC [14]. To maintain optimal soil moisture for microbial growth, it was adjusted to 60% WHC using deionized water. Specimen cups were incubated in 1 L Mason jars in the dark under two designated temperature conditions. The 2 mL deionized water was added to the each Mason jar to maintain moisture. All Mason jars were sealed with lids containing semi-permeable membranes to minimize water evaporation while permitting air exchange during incubation.

Values are represented as the mean ± standard error (n = 3). Different lowercase letters between seven litters represent significant differences (P < 0.05). Total C, total carbon; Total N, total nitrogen; Total P, total phosphorus.

CO₂ production was quantified on days 2, 7, 14, 28, and 42, and ¹³C-CO₂ abundance was measured simultaneously at each incubation time point. Before sampling, the lids with semi-permeable membranes were removed, the Mason jars were ventilated for 5 min, and then lids with butyl rubber stoppers were fitted onto each jar to ensure an airtight seal. At 0 and 12 h (days 1–7) or 24 h (days 7–42) after sealing, gas samples were collected using a gas-tight syringe and stored in pre-evacuated Exetainers (839 W, Labco, High Wycombe, UK). CO₂ concentration was determined using gas chromatography (7890 B, Agilent, CA, USA). δ¹³C abundance was determined using a PDZ Europa 2020 isotope ratio mass spectrometer (IRMS, Sercon Ltd., Cheshire, UK).

At the end of incubation, the activities of three extracellular hydrolytic enzymes—BG (β-1,4-glucosidase, C-acquiring enzyme), NAG (β-1,4-N-acetylglucosaminidase, N-acquiring enzyme), and AP (acid phosphatase, P-acquiring enzyme)—were simultaneously determined using a 96-well fluorescence plate assay on a microplate reader (Synergy H1, BioTek Instruments, VT, USA).

The CO₂-C emission rate during incubation was calculated with the following equation:

where R was the CO₂-C emission rate (mg C kg⁻¹ soil day⁻¹); c₁ and c₂ represented the CO₂ concentrations (ppm) measured at two different time points: c₁ was the concentration at time zero, and c₂ was the concentration in either 12 hours (1–7 d) or 24 hours (14–42 d); V was the effective volume of the Mason jar used in the experiment (L); M is the molar mass of C (12 g mol⁻¹); T was the incubation temperature (23°C or 33°C); W_soil was the dry weight of the soil (g); t was the time of CO₂ accumulation (days).

Cumulative CO₂-C emission (C, mg C kg⁻¹ soil) was estimated by linear interpolation of daily CO₂-C emission rates. Cumulative CO₂ emissions were calculated as follows:

where R_i and R_i+1 were respiration rates at i^th and (i + 1)^th incubation time, t_i+1−t_i signified the interval between the i^th and (i + 1)^th incubation times (days); and n was the number of incubation times.

A mass balance equation [44] was used to separate the amount of CO₂-C derived from soil and litter:

where C_total, C_litter, and C_soil represented the total CO₂-C emissions from the soil-litter mixture samples, the CO₂-C derived from the litter, and the CO₂-C derived from the soil after litter addition, δ_total was the δ¹³C value of CO₂ emitted from the soil-litter mixture samples, and δ_litter and δ_soil were the δ¹³C values of litter material and soil, respectively.

The proportion of litter decomposition (L_d, %) was calculated using the following equation:

where C_litter was the cumulative CO₂ (mg C kg⁻¹ soil) efflux from litter during the incubation period, and M_litter was the amount of litter C added to the soil (mg C kg⁻¹ soil).

The PE induced by litter was calculated as the method of:

where PE (mg C kg⁻¹ soil) was defined as the difference of CO₂-C emission from the soil with litter addition and without litter control soil; C_soil represented the mineralized C from the soil with litter; C_control represented the mineralized C from without litter soil.

Microbial resource limitation was measured using vector analysis of ecoenzymatic stoichiometry [40, 45]. The vector length and vector angle are calculated following Equations:

The longer Vector L, the higher the C limitation. A vector angle lower than 45^◦ indicates N limitation, otherwise P limitation dominates. When vector angles < 45^◦, a greater vector angle indicates a smaller N limitation. When vector angles > 45^◦, a greater vector angle indicates greater P limitation.

Net C balance (mg C kg⁻¹ soil) was calculated as the method of [46]:

where C_remained indicated the retention of added litters-C in the soil (mg C kg⁻¹ soil), which was obtained by subtracting litters-C respired from litters-C initially addition; C_primed was the primed soil C loss by litters (mg C kg⁻¹ soil).

The response ratio (RR) of microbial variables was calculated at the end of the incubation as the ratio of the variable in the litter addition treatment to that in the corresponding control under identical incubation conditions. An RR value greater than 1 indicates a positive effect of litter addition, whereas a value less than 1 indicates a negative effect [47].

Litter decomposition, CO₂ emission, PE, soil microbial resource limitations, and soil net C balance were analyzed using three-way analysis of variance (ANOVA). Significant differences among treatments were determined using the least significant difference (LSD) test at α = 0.05. Statistical significance was set at P < 0.05. All statistical analyses were performed using SPSS Statistics 26 (IBM Corp., Armonk, NY, USA). Linear regressions and graphical representations were generated using Origin 2019b (OriginLab, Northampton, MA, USA). The mantel test was used to evaluate the relationships between litter properties, ecoenzymatic activity, ecoenzymatic stoichiometry, and microbial resource limitations with litter decomposition and PE, using the ggcor R package.

After 42 days of incubation, litter decomposition ranged from 4.97–33.59% (Fig. 1), which varied with the C: N ratio. Furthermore, litters with lower C:N ratios (CN1, CN2, CN3, CN4) exhibited lower decomposition rates than those with higher C:N ratios (CN5, CN6, CN7). The highest decomposition rate was observed in the CN5 treatment (C:N ratio of 28) across all soil fertility and temperature conditions, and it was significantly higher than that of other C:N ratio treatments (Fig. 1, P < 0.05). For CN5 litter, the decomposition rate in HF was 7.94% and 16.27% higher than in LF under 23°C and 33°C conditions, respectively (P < 0.05). However, in HF soils, the decomposition rate of CN5 litter was not significantly affected by temperature, whereas in LF soils it decreased under warming (Fig. 1). For litters with other C:N ratios, temperature had no significant effect on decomposition rates (Table 3). Nevertheless, at the same temperature, decomposition rates in HF soils were significantly higher than those in LF soils (Fig. 1, P < 0.05). A three-way ANOVA indicated that litter decomposition was significantly affected by the interactive effects of incubation temperature, soil fertility, and litter C:N ratio (Table 3).

The cumulative litter-derived CO₂ increased progressively over the course of incubation (Fig. 2a–d). In CN1, CN2, CN3, and CN4 treatments, cumulative litter-derived CO₂ increased sharply during the first week and then plateaued, whereas in CN5, CN6, and CN7 treatments, it increased sharply over the first two weeks before stabilizing (Fig. 2a–d). Notably, the largest cumulative litter-derived CO₂ was observed in the CN5 treatment across all soil fertility and temperature conditions (Fig. 2i–l). Cumulative litter-derived CO₂ was significantly lower in LF soils than in HF (Table 3, P < 0.001), however, within the same fertility level, temperature had little effect on cumulative litter-derived CO₂ (Table 3, P > 0.05).

The variation trend in cumulative soil-derived CO₂ was similar to that of litter-derived CO₂ with incubation across soil fertility and temperature (Fig. 2e–h). During the first two weeks, cumulative soil-derived CO₂ accounted for 56.18–95.42% of the total CO₂ emissions. Compared with the control (CK) without litter addition, the cumulative soil-derived CO₂ was higher in treatments with low C:N ratio litter (CN1–CN4) and relatively lower in treatments with high C:N ratio litter (CN5–CN7) (Fig. 2i–l). A three-way ANOVA showed that cumulative soil-derived CO₂ emissions were significantly influenced by litter C:N ratio, incubation temperature, and soil fertility, as well as by the interactive effects among these factors (Table 3).

PE induced by low C:N ratio litters (CN1–CN4) was positive, whereas that induced by high C:N ratio litters (CN5–CN7) was negative (Fig. 3). Regardless of soil fertility or temperature, the transition from positive to negative PE occurred at a C:N ratio of approximately 25 (Fig. 3 and Fig. 4). Furthermore, the magnitude of PE in low C:N ratio litter treatments was significantly greater than that in high C:N ratio litter treatments (Fig. 4a, P < 0.05). In general, cumulative PE during the first two weeks accounted for the majority of the total PE, after which its magnitude remained stable throughout the incubation period. For the same C:N ratio litter treatments, cumulative PE was higher in HF than in LF (Fig. 4a, Table 3, P < 0.001). Cumulative PE was not significantly affected by temperature (Table 3, P > 0.05), except for the enhancement observed in high C:N ratio (CN5–CN7) litter treatments under HF conditions (Fig. 4a). Overall, cumulative PE ranged from − 294.45 to 497.69 mg C kg⁻¹ soil. Overall, a significant negative correlation was observed between the cumulative PE and the litter C:N ratio across the 4 incubation environments (Fig. 4b). The HF-33°C treatment exhibited the largest positive and the smallest negative PE. With increasing litter C:N ratio, this treatment also displayed the greatest magnitude of change in PE (steepest slope).

Litter addition significantly increased the activities of β-1,4-glucosidase (BG), β-1,4-N-acetyl-glucosaminnidase (NAG), and Acid phosphatase (AP) (Fig. 5a–c, P < 0.05), These effects were greater in LF than in HF except for AP, irrespective of incubation temperature. In the LF treatment, incubation temperature significantly affected the activities of BG and NAG, with the most pronounced difference in NAG activity observed in the CN5 litter addition treatment (Fig. 5b).

There were significant increases in BG:AP and NAG:AP, whereas there was a significant decrease in BG:NAG following litter addition (Fig. 5d–f, P < 0.05). Following the addition of low C:N ratio litters (CN1–CN4), significant differences in BG:NAG were detected between soil fertility, with BG:NAG significantly higher in HF than in LF (Fig. 5d). After CN5 litter addition, the lowest BG:NAG were observed across all 4 incubation conditions (Fig. 5d). Interestingly, in contrast to BG:NAG, both BG:AP and NAG:AP were higher in LF than in HF following litter addition (Fig. 5e–f).

Soil microbial activity was consistently limited by C and P across all litter C:N ratios, soil fertility levels, and temperatures (Fig. 6a). Vector analysis (Fig. 6b, c) further indicated that these limitations were alleviated following litter addition relative to the CK. Compared with the addition of low C:N ratio litter (CN1–CN4), the addition of high C:N ratio litter (CN5–CN7) significantly increased soil microbial C limitation (Fig. 6b), and microbial P limitation showed a similar trend (Fig. 6c). In LF soils, P limitation after high C:N ratio litter addition was significantly greater than that in HF soils (Fig. 6c). Furthermore, within soils of the same fertility level, warming generally increased both microbial C and P limitations to some extent (Fig. 6b, c). A three-way ANOVA (Table 3) revealed that microbial C limitation was significantly influenced by litter C:N ratio, incubation temperature, and their interaction, whereas microbial P limitation was significantly affected by litter C:N ratio, soil fertility, incubation temperature, and the pairwise interactions among these factors.

Correlation analysis revealed that, following the addition of low C:N ratio litter (Fig. 7a), AP exhibited a significant positive correlation with litter P content, but a significant negative correlation with other litter properties. Additionally, the BG:AP ratio was significantly positively correlated with litter C content (P < 0.05). Microbial C limitation showed a significant positive correlation with BG but a significant negative correlation with NAG (P < 0.05), whereas microbial P limitation displayed significant negative correlations with both BG and NAG (P < 0.05). Mantel test analysis showed that the positive PE induced by low C:N ratio litter was significantly associated with litter properties (e.g., litter C:N ratio), BG, and microbial C limitation (Mantel’s P < 0.05). Moreover, litter decomposition was also significantly associated with litter C content, BG, and microbial C limitation (Mantel’s P < 0.05).

Following the addition of high C:N ratio litter (Fig. 7b), the activities of BG, NAG, and AP were all significantly and negatively correlated with litter C content (P < 0.05). Furthermore, NAG showed significant positive correlations with litter N and P contents (P < 0.05). Both microbial C and P limitations were significantly correlated with litter properties, and each exhibited significant negative correlations with the activities of soil enzymes (BG, NAG, and AP) (P < 0.05). Mantel test results further revealed that the decomposition of high C:N ratio litter was significantly associated with litter properties, the activities of soil enzymes (particularly AP), and microbial resource limitations. The negative PE induced by the addition of high C:N ratio litter was significantly associated with the activities of soil enzymes and microbial resource limitations (Mantel’s P < 0.05).

A three-way ANOVA (Table 3) indicated that litter C:N ratio, soil fertility, and incubation temperature each had a significant effect on the soil net C balance (P < 0.001), whereas no significant interaction was observed among the three factors (P > 0.05). As shown in Fig. 8a, the addition of high C:N ratio litter significantly increased the soil net C balance. Under the same litter C:N ratio treatment, HF contained greater SOC content compared with LF. Moreover, under high C:N ratio litter addition, elevated incubation temperature further increased SOC content in HF. Overall, a significant negative correlation was observed between the PE and the soil net C balance (Fig. 8b).

As expected, the higher litter decomposition percentage observed in HF compared with LF indicated that fertile soil provided a more favorable environment for litter decomposition. This finding supports part of Hypothesis 1 regarding the effect of soil nutrient availability on litter decomposition. This could be attributed to the abundant nutrients in HF, which supplied sufficient resources to support a large soil microbial community [48], thereby stimulating microbial activity [49] and accelerating litter decomposition. This result was consistent with previous findings reported [50–52]. Collectively, these findings indicate that soil fertility directly influences litter decomposition (Table 3). In addition, the present study found a higher decomposition percentage for high C:N ratio litter (CN5–CN7) compared with low C:N ratio litter (CN1–CN4), possibly due to differences in litter origin, with CN1–CN4 derived from crop residues and CN5–CN7 from forest litter. This may be explained by the findings of Yan [53] et al., which suggested that litters with different compositions can induce distinct soil microbial communities associated with litter decomposition patterns, thereby influencing the mineralization process. Forest litter with a higher C:N ratio, when added to farmland soil, exhibited a greater decomposition percentage than crop litter, indicating that the soil environment plays a key role in litter decomposition. For the same litter source, the percentage of litter decomposed decreased as the C:N ratio increased. This may be attributed to the strong relationship between litter quality and early-stage litter decomposition dynamics [32, 54, 55]. Although CN5 and CN4 had similar C:N ratios close to 25—the optimal ratio for bacterial activity during litter decomposition—the decomposition percentage of CN5 was higher than that of CN4. Previous studies have demonstrated that a C:N ratio of 25:1 is optimal for stimulating the activities of lignin peroxidase, polyphenol oxidase, and peroxidase, thereby maximizing lignin and cellulose degradation rates [56]. We speculate that the relatively high P content in CN5 (Table 2) contributed to its higher mineralization percentage, likely due to P limitation across all litter-amended treatments (Fig. 6). Previous studies support our findings, as P -rich exogenous litter has been shown to increase microbial P content or accelerate microbial turnover [57], thereby enhancing litter decomposition. Thus, soil microorganisms likely maintain a homeostatic C:N ratio in their biomass, with P availability regulating their turnover and activity. Consistent with these findings, the pronounced decrease in the BG:NAG in the CN5 treatment compared with other litter-amended treatments (Fig. 5d) clearly indicates a shift in decomposer community composition [22].

Overall, incubation temperature did not significantly affect litter decomposition (Table 3), except under the CN5 treatment. This observation partially contradicts Hypothesis 1, which posited that incubation temperature would influence litter decomposition. One possible explanation is the presence of a temperature-adaptation mechanism in soil enzymes. This mechanism suggests that the capacity of soil microbial communities to adapt to temperature fluctuations may modulate soil enzyme activity. For example, although microbial community composition may converge across different incubation temperatures, the overall level of soil enzyme activity may remain largely unchanged [58]. Therefore, our results suggest that the nutrient requirements of decomposers are a critical driver of litter mineralization.

PE shifted from positive to negative at a litter C:N ratio of approximately 25, consistent with our second hypothesis. The positive PE induced by low C:N ratio litter in this study confirms the classical “co-metabolism” and “N mining” theories. This phenomenon occurs because available fresh C rapidly activates copiotrophic microbial communities, thereby increasing soil microbial biomass and activity [59, 60]. Such strong stimulation likely generates an “apparent” PE in HF, due to accelerated microbial metabolism [6](Bastida et al., 2019) and litter decomposition (Fig. 1). The minimal influence of litter addition on enzyme activities further supports this idea (Fig. 5a–c). Conversely, the “triggering effect” described by De Nobili [61] et al. in LF contributes to soil N mining, as evidenced by elevated BG and NAG enzyme activities (Fig. 5a–c). As the C:N ratio increases, microbial N limitation promotes SOC decomposition to satisfy N demand [62], ultimately leading to a positive PE. This conclusion is supported by the finding that microbial C limitation was most alleviated following the addition of low C:N ratio litter (Fig. 6b and Fig. 7a). The negative PE observed with high C:N ratio litter additions can be explained by the “preferential substrate utilization” theory [63] and the “stoichiometric decomposition” theory [10]. As mentioned above, forest litter additions induce specific microbial communities that enhance litter decomposition, supported by the relatively higher mineralization percentages and lower soil-derived CO₂ emissions (Fig. 1 and Fig. 2). These findings indicate that shifts in microbial communities employ various mechanisms in response to resources differing in elemental composition and stoichiometry. The pronounced correlation observed between the negative PE and soil enzyme activity in this study further substantiates our conclusion (Fig. 7b). The intensity of PE depends on litter C:N ratio, as reported in previous studies [52, 64, 65]. It is negatively correlated with the C:N ratio of farmland litters, consistent with our second hypothesis, whereas the magnitude of negative PE increases with the C:N ratio of forest litters, indicating a strong dependence on litter source. Farmland litters, characterized by lower C:N ratios, provide high-quality C sources for soil microorganisms, activating them to induce soil-derived CO₂ emissions indicative of an “immediate and real” PE. Conversely, higher litter-derived CO₂ and lower soil-derived CO₂ observed in forest litter treatments may represent “apparent” PE due to enhanced microbial turnover.

HF exhibit greater sensitivity to litter addition than LF, as evidenced by a wider range of cumulative PE values ranging from − 300 mg C kg⁻¹ soil to 500 mg C kg⁻¹ soil in HF, compared to -60 mg C kg⁻¹ soil to 250 mg C kg⁻¹ soil in LF (Fig. 4, Table 3). This finding partially confirmed Hypothesis 3, specifically with respect to the influence of soil nutrient availability on the PE. This finding is consistent with Zhu [66] et al., who reported that NPKM-fertilized soils exhibited the largest cumulative PE following ¹³C-labeled glucose addition among four long-term fertilized paddy soils. This may be because HF soils provide sufficient C and nutrients to meet microbial stoichiometric demands [67, 68]. Furthermore, PE in HF soils responds to temperature, whereas PE in LF does not, confirming that microbial communities associated with PE differ between these soils, with those in HF being more temperature-sensitive. This finding aligns with previous reports indicating that a 10°C temperature increase leads to a comparable proportion of accessible and decomposable SOC in HF. Collectively, these results indicate a significant interaction between temperature and soil fertility on PE (Table 3). The greater negative PE at higher C:N ratios may be explained by the greater temperature sensitivity of recalcitrant organic compounds compared to labile substrates [36, 69]. In this study, PE in HF responded to temperature without corresponding changes in enzyme activity or nutrient limitation, whereas in LF, enzyme activities and nutrient limitations were temperature-sensitive without a PE response. The reasons and evidence mentioned above support our hypothesis three. Therefore, future studies should carefully consider the adaptation of microbial communities and enzyme activities to temperature changes.

Regardless of the direction of the PE, litter addition increased SOC content, as evidenced by the positive net C balance (Fig. 8), which is consistent with previous findings [46]. These results indicate that PE plays a critical role in regulating soil C storage, although this effect is highly dependent on both soil type and litter characteristics [70, 71]. In this study, the net C balance was significantly higher in HF than in LF, and it increased with the litter C:N ratio regardless of the PE direction (Fig. 8a). The observed increase in SOC under positive PE can be explained by the “Microbial Carbon Pump” (MCP) theory [72], which posits that soil microbes function not only as decomposers but also as contributors to SOC. During the decomposition of litter and SOC, soil microbes utilize C and nutrients to increase their biomass [1]. As microbial turnover accelerates, dead microbial biomass becomes incorporated into the soil C pool. Negative PE reduce the decomposition of native SOC, thereby leading to a positive net C balance in the soil—an outcome that is conceptually straightforward. Moreover, an increase in PE consistently resulted in a decline in the net C balance (Fig. 8b), confirming that litter addition can stimulate additional C emissions to the atmosphere [8, 73]. Nevertheless, our results support the notion that litter addition induces greater C losses in HF than in LF soils, largely due to the higher absolute SOC content in HF [70]. Therefore, soil C turnover and storage in HF soils may be more sensitive to exogenous C inputs. It is therefore reasonable to hypothesize that fertile soils could make a substantial contribution to climate change. Future investigations into the fate of litter will provide valuable insights into its long-term contribution to C management.