The Yellow River Delta is located in Shandong province, China, with an area of 12,038 km². This area belongs to a warm temperate continental monsoon climate with a mean air temperature of 11.7°C–12.8°C, an annual evaporation of 1,900–2,400 mm and an annual precipitation of 530–630 mm. About 70% of the rain occurs between July and September. The soil types are Calcaric Fluvisols, Gleyic Solochaks, and Salic Fluvisols (FAO; Guan et al., 2019; Lu et al., 2021). The research site was located in the southern part of the YRD Nature Reserve, a nontidal wetland. The main vegetation species were Tamarix Chinensis Lour., Phragmites communis (Cav.) Trin. ex Stued. and Suaeda salsa (Linn.) Pall. At this site, the soil does not flood for most of the year, but it is easy to temporarily flood after heavy rainfall in summer and fall. According to the field investigation, dead reeds do not fall directly to the soil surface and their aboveground part can stand until the end of the next growing season, especially for the culm and sheath litter, resulting in a standing litter decomposition process.

At the end of October 2020, three sampling points were established in the reed growing area to collect the aboveground litter. In the laboratory, the reed sample was divided into leaf, sheath, and culm litter. The litter sample was cut to a length of approximately 5 cm and oven-dried to constant weight at 70°C after cleaning with a soft brush. Ten grams of the sample were placed in a nylon litter bag (20 cm × 20 cm) with a mesh size of 1 mm. In this experiment, 96 litterbags were prepared (three types of litter × four repetitions × two decomposition interfaces × four sample dates). In November 2020, litterbags were placed at the original sampling point. For each type of litterbag, sixteen litterbags were placed on the soil surface and fixed with nails, and the others were suspended in the air. According to Zhang et al. (2014a), the litterbags in the air were fixed on a horizontal nylon net at a height of 1 m, equivalent to 3/4 of the mean height of the reed.

Litter CO₂ emission was measured on the 90th (that is, in winter), 180th (that is, in spring), 270th (that is, in summer), and 360th (that is, in autumn) days. On each sampling date, the CO₂ emission rate was tested for a whole day at 14:00, 18:00, 24:00, next at 6:00, 10:00 and 14:00. Before sampling, four litterbags of each litter type were collected. The dust on the surface of the litter was removed with a soft brush and then the sample was placed in a new and clean litterbag (20 cm × 20 cm). A PVC pipe wrapped with thermal insulation was used to measure CO₂ emissions. This pipe has a diameter of 25 cm and a height of 30 cm, with one end closed and the other covered. A three-way valve and a temperature probe were installed on the cover of the PVC pipe. The new litterbag with the sample was placed in the PVC pipe and the pipe was sealed using a lid for 30 min. Gas samples with a volume of 50 ml were collected at the beginning and end of sealing, respectively. Each gas sample was stored in a vacuum bag. The CO₂ concentration was measured by gas chromatography (Agilent 7890A, United States). The difference in CO₂ concentration at the beginning and end of sealing is the CO₂ emitted by the litter sample. When CO₂ emissions were tested, air temperature and relative air humidity were measured in situ on the surface of the soil and in the air (that is, at a height of 1 m).

In the laboratory, the fresh litter was weighed and divided into two parts. A part of the fresh litter was oven-dried at 70°C to test the moisture content. The dry weight of each fresh sample was used to calculate the CO₂ emission rate. The other part of the fresh litter sample was cut to <2 mm and was used to test the β-glucosidase and invertase activities using the method of Guan (1986). The concentrations of labile (LOC) and recalcitrant (ROC) OC in the litter sample were measured using the sample at the beginning (i.e., day 0) and end of experiment (i.e., day 360) and an acid hydrolysis approach (Rovira and Vallejo, 2002). The initial content of OC was measured by the dry combustion method using a Multi N/C 2100 analyzer (Analytik Jena, Germany). The initial content of total phosphorus (TP) was measured by the ammonium molybdate method after H₂SO₄-H₂O₂ oxidation (Kuo, 1996). The initial content of total nitrogen (TN) was determined by Kjeldahl digestion using a Kjeltec Auto Analyzer (Foss 8,400, Denmark).

The CO₂ emission rate and the cumulative CO₂ emission were calculated using the method of Tao et al. (2022). Cumulative CO₂ emissions from 18:00 to 6:00 the next day were defined as CO₂ emissions at night, and cumulative CO₂ emissions from 6:00 to 18:00 were defined as CO₂ emissions in the daytime. R = P × V × ∆ c / ∆ t × r T − 1 × M × m − 1 (1) Where R represents the CO₂ emission rate, mg kg⁻¹ h⁻¹. P represents the standard atmospheric pressure, Pa. V represents the volume of the PVC pipe, cm³. c represents the difference in CO₂ concentration at the beginning and end of sealing, ppm. t represents the sealing time, 0.5 h. r represents the universal gas constant. T represents the absolute air temperature, K. M is the molecular mass of CO₂, g mol⁻¹ m represents the dry weight of the litter sample, kg. C u m u l a t i v e   C O 2   e m i s s i o n   m g   k g − 1 = ∑ 0.5 × R n + R n + 1 × T n + 1 − T n (2) Where R _n and R _n+1 represent the CO₂ emission rate of any two adjacent sampling times, mg kg⁻¹ h⁻¹. (T _n+1 − T _n) represents the time intervals between any two adjacent samples, h.

The temperature sensitivity (Q ₁₀) of CO₂ emission was calculated following the method of Luo et al. (2001). R = A   e k T (3) Q 10 = e 10 k (4) Where R represents the CO₂ emission rate, mg kg⁻¹ h⁻¹. T represents the air temperature, °C. A and k represent constants.

The LOC loss ratio (%) and the residual ROC ratio (%) were calculated as follows. L O C   l o s s   r a t i o   % = 100 × M 0 × L 0 − M 360 × L 360   /   M 0 × L 0 (5) r e s i d u a l   R O C   r a t i o   % = 100 × M 360 × R 360   /   M 0 × R 0 (6) Where M ₀ and M ₃₆₀ represent the litter mass on days 0 and 360, g. L ₀ and L ₃₆₀ represent the LOC concentration on days 0 and 360, mg g⁻¹. R ₀ and R ₃₆₀ represent the ROC concentration on days 0 and 360, mg g⁻¹.

The Shapiro-Wilk test was used to test the normality of data, and nonnormal data were logarithmically transformed before analysis. The difference in cumulative CO₂ emission, enzyme activities, Q ₁₀ value, LOC loss ratio, and residual ROC ratio were compared separately using a one-way analysis of ANOVA with Tukey’s HSD test (Tamhane’s test when equal variances were not assumed) at a 95% confidence level. Data on CO₂ production were also analyzed using a three-way analysis of variance (ANOVA) with sample date, decomposition interface and litter types as independent factors. Data on air temperature and relative air humidity were analyzed using a two-way analysis of variance (ANOVA) with sample date and decomposition interface as independent factors. Pearson’s correlation coefficients between CO₂ emission and factors (that is, air temperature, relative air humidity, enzyme activities, and LOC loss ratio) and the relationship coefficients between the Q ₁₀ value and residual ROC ratio were also calculated. All statistical analyzes were conducted using SPSS 25.0 software (SPSS Inc. United States) and the figures were drawn using Origin 9.0 (OriginLab, Northampton, MA, United States).

Litter types, decomposition interfaces, and sample date significantly affected cumulative CO₂ emission (Supplementary Table S1; p <0.001). Cumulative CO₂ emissions from standing litter (that is, leaf, culm, and sheath) were less than those on the surface of the soil (Supplementary Figure S1; Figure 1A; p <0.001). The cumulative CO₂ emissions of the standing leaf, culm, and sheath were 56.76%, 66.67%, and 69.19% of that on the soil surface. For standing litter, the cumulative CO₂ emission from sheath litter is 1.33 and 1.36 times that of leaf and culm litter (p <0.05). On the surface of the soil, the cumulative CO₂ emission of the sheath litter was 1.30 times the culm litter (p <0.05) and was similar to the leaf litter (Supplementary Figure S1; Figure 1A).

On the daily scale, cumulative CO₂ emission from the culm and sheath litter at night was greater than in the daytime at the two decomposition interfaces (p <0.05). Cumulative CO₂ emissions from standing leaf litter at night were also higher than in the daytime (p <0.05). However, cumulative CO₂ emission from leaf litter on the soil surface did not differ between daytime and night (Figure 1B). On the seasonal scale, the CO₂ emission from standing litter on the 270th day was greater than on other sample dates (p <0.05). Similarly, CO₂ emission on the 270th and 360th day was higher than on other sample dates for litter on the soil surface (Figure 1A; p <0.05).

The litter on the ground had a larger temperature sensitivity (Q ₁₀) of CO₂ emission than the standing litter (p =0.004). For standing litter, the Q ₁₀ value of CO₂ emission ranged from 1.34 to 1.60, with an average of 1.44. However, the Q ₁₀ value of CO₂ emission from the litter on the ground ranged from 2.03 to 2.24, with an average of 2.16 (Figure 2).

Leaf litter had larger β-glucosidase and invertase activities than non-leaf litter at both decomposition interfaces (p <0.05), and these enzyme activities of litter on the soil surface were greater than those in the air (Supplementary Figure S2; p <0.05). β-glucosidase and invertase activities were positively correlated with cumulative CO₂ emission (Figure 3; p <0.05).

For standing litter, the LOC loss ratio of sheath litter was greater than that of leaf and culm litter (Figure 4A; p <0.05). Sheath litter had a higher LOC loss ratio than culm litter (Figure 4A; p <0.05), but had a similar LOC loss ratio to leaf litter on the soil surface. Overall, the litter on the soil surface had a higher LOC loss ratio than standing litter (Figure 4A; p <0.05). The LOC loss ratio was positively related to cumulative CO₂ emission (Figure 4B; p <0.001). For leaf and culm litter, the residual ROC ratio on the soil surface was higher than in the air (p < 0.05). Sheath litter on the soil surface had a similar residual ROC ratio to standing litter (Figure 5A). Furthermore, the residual ROC ratio was positively related to the Q ₁₀ value (Figure 5B; p =0.02).

The sample date significantly affected the air temperature and relative air humidity (Supplementary Table S2; p <0.001). Air temperature and relative air humidity on days 270 and 360 were higher than those on days 90 and 180. The decomposition interface did not influence the air temperature and relative air humidity. The air temperature in the daytime was higher than at night, but the relative air humidity was higher at night than in the daytime (Supplementary Figure S3; p < 0.05).

On the daily scale, the CO₂ emission rate was negatively related to air temperature and positively associated with relative air humidity on the 90th, 180th and 270th day for standing litter, and similar relationships were observed on the 180th and 270th day for litter on the soil surface (Table 1). However, the CO₂ emission rate had a positive relationship with the air temperature and a negative relationship with the relative air humidity on day 360 (Table 1). On the seasonal scale, air temperature and relative air humidity were positively related to cumulative CO₂ emissions (Table 2; p <0.001).

Leaf litter had higher content of TN and TP than non-leaf litter, while non-leaf litter had higher content of total OC than leaf litter (Supplementary Table S3; p <0.05). Culm litter had the highest C/N, C/P ratios and the smallest N/P ratio among three types of litter (Supplementary Table S3; p <0.05). Cumulative CO₂ emission was not related to the contents of OC, TN, TP and stoichiometric ratios of C/N, C/P, and N/P (Supplementary Table S4).

Standing litter plays a vital role in CO₂ emission from litter. However, the contribution of standing litter to litter CO₂ emissions is still uncertain. Standing litter can release more (Gliksman et al., 2018), or similar (Kuehn and Suberkropp, 1998), or less (Zhang et al., 2014b; Mao et al., 2021) CO₂ than litter on the ground. In this study, although standing litter released less CO₂ than litter on the soil surface (Supplementary Figure S1; Figure 1), it was equivalent to 56%–70% of that on the soil surface, indicating that CO₂ emission from standing reed litter should not be ignored in the YRD.

The microorganism is a vital factor driving litter decomposition and CO₂ emission (Evans et al., 2020; Logan et al., 2021), and fungal decomposers had colonized standing litter even at the beginning of plant senescence (Newell, 2002; Chimney and Pietro, 2006). Enzymes (e.g., β-glucosidase) were involved in the decomposition and transformation of OC (Sinsabaugh, 2010; Song et al., 2017; Miao et al., 2020; Chen et al., 2022). Additionally, labile organic substrates stimulated microbial activity and accelerated litter decomposition (Kuzyakov et al., 2000; de Graaff et al., 2010). Higher β-glucosidase activity (e.g., Song et al., 2017; Tao et al., 2022) and LOC concentrations (Don and Kalbitz, 2005; Wang L. et al., 2015) resulted in faster litter decomposition or CO₂ emission. Compared to litter on the surface of soil or sediment, standing litter could not absorb water and immobilize nutrients from the soil by microbes (He et al., 2013; Zhang et al., 2021), reducing microbial growth and its activity. In this study, the litter on the surface of the soil had larger activities of β-glucosidase and invertase than the standing litter, especially on days 180 and 360 (Supplementary Figure S2). Such enzyme activities were positively related to CO₂ emission (Figure 3), resulting in rapid CO₂ emission. Moreover, the LOC loss ratio of the litter on the soil surface was higher than that in the air (Figure 4), which further explained the rapid release of CO₂ from the litter on the soil surface.

The types of litter had a significant effect on CO₂ emission rates. Standing sheath litter had the highest cumulative CO₂ emission per unit mass of dry biomass among the three types of litter. The leaf and sheath litter had higher cumulative CO₂ emissions per unit mass of dry biomass than the culm litter on the soil surface (Figure 1). Usually, LOC decomposed more quickly than ROC (see, e.g., Kuzyakov et al., 2000; de Graaff et al., 2010; Tao et al., 2013). In this experiment, the sheath had the most considerable LOC loss ratio among the three types of standing litter. Leaf and sheath litter had a greater LOC loss ratio than the culm litter on the soil surface. The LOC loss ratio was positively related to cumulative CO₂ emission (Figure 4), indicating that rapid LOC decomposition of leaf and sheath litter accelerated CO₂ emission. Previous studies found that litter stoichiometry did not constrain the litter decomposition (Aerts et al., 2012), and carbon quality rather than stoichiometry controlled the litter decomposition (Hättenschwiler and Jørgensen, 2010). In this experiment, non-significant relationships between cumulative CO₂ emission and stoichiometric ratios were observed (Supplementary Table S4). Thus, we concluded that OC quality rather than stoichiometry was the major factor adjusting CO₂ emission from litter in the YRD.

A study in the adjacent area showed that the biomass of the reed leaf, culm, and sheath was 1264.32, 3667.58, and 1123.78 g m⁻² (Zan et al., 2011). Based on the results of Zan et al. (2011), we estimated that cumulative CO₂ emissions of standing culm and sheath litters were 2.97 and 1.21 times that of leaf litter, while cumulative CO₂ emission from culm and sheath litter was 2.54 and 1.01 times that of leaf litter on the soil surface in the YRD. In other words, the reed litter without leaves in the YRD contributed about 80% of total CO₂ emissions from the litter regardless of the decomposition interface. Therefore, non-leaf litter (e.g., culm and sheath) of reeds may be the main contributor to CO₂ emissions from litter in the YRD. This study highlights the importance of CO₂ emission from non-leaf litter, and subsequent studies should focus on the decomposition of non-leaf litter.

Air temperature and relative air humidity were vital factors in adjusting litter CO₂ emission (Zhang et al., 2014b; Wang et al., 2017), but their effects on litter CO₂ emission varied at different time scales. On the daily scale, we found that the litter released more CO₂ emissions at night than in the day, especially for the standing litter (Figure 1B). The previous study observed that, in Mediterranean grasslands, water vapor from the atmosphere stimulated microbial activity and litter decomposition at night (Dirks et al., 2010; Gliksman et al., 2017), and similar results were found in a semi-arid grassland ecosystem due to outstanding absorption of water from the atmosphere overnight for standing litter (Wang et al., 2017). Microbes are mainly r-strategy organisms with short lifespans, responding rapidly to changes in water supply (Jacobson et al., 2015). High air humidity at night was positively associated with microbial activity, especially for standing litter (Wang et al., 2017). Similarly, high relative air humidity at night and dew condensation were observed to adjust microbial activity (McHugh et al., 2015) and led to rapid CO₂ emission (Wang et al., 2017).

Previous study found that non-rainfall moisture, such as humidity and dew, was a key driver of microbial respiration from standing litter (Evans et al., 2020). When there was no precipitation, the maximum rate of CO₂ emission occurred in the evening and early morning when dew condensed (Kuehn et al., 2004). Moreover, the total PLFAs in litter were positively related to relative humidity at night, especially for the standing litter (Wang et al., 2017). In the coastal wetland of this research area, relative air humidity at night exceeded 70% and even approached 100%, higher than in the daytime. The air temperature at night was lower than in the daytime (Supplementary Figure S3). Higher air humidity and lower air temperature at night may help dew condensation, thereby promoting microbial respiration. Our investigation found that the maximum rate of CO₂ emission occurred from 18:00 to 6:00 (Supplementary Figure S1). Our results also showed that, on the daily scale, the CO₂ emission rate had a negative relationship with air temperature and a positive relationship with relative air humidity (Table 1). Therefore, we speculate that lower air temperature and higher relative air humidity at night may help dew formation, providing moisture for microorganisms, and thus accelerating CO₂ emission.

It should be noted that, on the daily scale, the rate of CO₂ emission from soil litter on day 360 was positively related to the air temperature and negatively associated with relative air humidity, in conflict with the results on the 180th and 270th days (Table 1). When collecting samples on day 360, precipitation caused temporary flooding (about 0–3 cm depth) on the soil surface, which may alleviate the limitation of moisture on microbial respiration. Moreover, an incubation experiment found that increasing temperature accelerated CO₂ emission from water-saturated litter (Zhang et al., 2014b). Thus, we reasoned that, under the condition of sufficient water, a higher air temperature in the daytime might stimulate microbial activity and microbial respiration compared to that at night on day 360.

Fungi made up the majority of the microorganisms in the standing litter (Findlay et al., 2002). Temperature was an important factor affecting seasonal dynamics of fungal biomass on standing litter (Verma et al., 2003). The biomass of fungi on litter increased exponentially with temperature (Suberkropp and Weyers, 1996). On the seasonal scale, cumulative CO₂ emission and β-glucosidase activity were positively correlated with air temperature and relative air humidity (Table 2; p <0.001). Higher mean air temperature and relative air humidity (Supplementary Figure S3) in summer (i.e., the 270th day) and autumn (i.e., the 360th day) can be conducive to stimulating enzyme activity and CO₂ emission. Overall, this study emphasized the importance of microenvironment fluctuations that influence litter CO₂ emissions at different time scales.

The quality of OC can affect the Q ₁₀ of CO₂ emissions, but their relationships are still uncertain. The earlier results showed that the ROC decomposition rate did not vary with temperature (Giardina and Ryan, 2000) or had a similar response to varied temperatures as the LOC decomposition (Fang et al., 2005). Since then, many studies have observed that ROC decomposition was more sensitive to temperature increase than LOC decomposition (see, e.g., Davidson and Janssens, 2006; Wang et al., 2018). We found that the soil litter had a higher Q ₁₀ value than the standing litter (Figure 2; p =0.004).

Compared to standing litter, the higher microbial activity and fungal biomass of litter on the soil surface triggered rapid decomposition (He et al., 2013; Zhang et al., 2015; Wang et al., 2017), which can result in an abundant accumulation of litter ROC fractions. A previous study in the YRD also found an increased proportion of ROC in reed litter after 2 years of in situ decomposition (Tao et al., 2019). In this study, the ground litter accumulated more ROC than standing litter (Figure 5A). According to the hypothesis of ‘carbon quality temperature’ (Davidson and Janssens, 2006), the abundant accumulation of ROC in soil litter may require considerable activation energy for the decomposition of OC, thus increasing the temperature sensitivity. In addition, the positive relationship between residual ROC ratio and the Q ₁₀ value (Figure 5B) further testified to the speculation mentioned above.