We searched Web of Science¹ and China National Knowledge Infrastructure (CNKI)² databases for peer-reviewed articles. The literature search followed the procedure of Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA; Moher et al., 2009) (Supplementary Figure S1). Search terms were either “straw,” “straw incorporation,” “straw mulching,” “straw return,” “crop residue,” or “stover,” and either “enzyme,” “hydrolytic enzyme,” “glucosidase,” “β-1,4-glucosidase,” “α-1,4-glucosidase,” “EEA,” “cellulase,” β-1,4-D-cellobiohydrolase,” “soil carbon,” “soil microbial biomass,” “soil organic carbon,” or “β-1,4-xylosidase.” The dataset was established based on the following criteria: (1) the initial climatic conditions and soil physicochemical characteristics were same between no-straw-return and straw return treatments; (2) experimental duration must be clear, with field experiments (not surveys or pot experiments) lasting at least 1 year; (3) if an article contained results from multiple soil depths, we used data from the uppermost soil layer. Moreover, if any study contained duplicate results in different growing years for the same experiment, we included only the latest sampling time in this analysis; and (4) the experiment location was stated. For each article, we extracted mean values (Mean), replicate numbers (n), and standard deviation (SD) or standard error (SE) when possible. The missing SD values were calculated from reported SE or coefficient of variation (CV) as shown in Equations 1 and 2 (Jian et al., 2020; Zhang et al., 2026):

If data were only presented graphically, values were extracted using WebPlotDigitizer.³ When critical information was not provided in the article, we contacted the corresponding author to obtain this information. Figure 1 shows the geographical distribution of straw return experiments.

Our dataset comprised environmental and experimental variables, including latitude, longitude, elevation, mean annual precipitation (MAP), mean annual temperature (MAT), soil properties, straw characteristics, and field management information. Whenever MAP and MAT were not reported, these data were extracted from WorldClim 2.1.⁴ Google Earth⁵ was used to determine the unreported coordinates of experimental locations. For soil properties, the unreported initial soil parameters, such as, soil clay content, soil organic carbon, soil total nitrogen, and soil pH, were extracted using the website https://soilgrids.org/. Crop straw biochemical characteristics were extracted from the original articles. When these data were not reported in the article, mean values were collected from the straw quality dataset for the same species or group of species (Thiébeau et al., 2021). For field management, crop types were grouped into maize, rice, wheat, and other; straw management was grouped into incorporated and surface-applied; fertilizer form was grouped by urea and mixed; straw types were classified as green plant biomass, mature aboveground biomass, senescent plant biomass, and straw (Abalos et al., 2022). Moreover, straw rates were not considered in this study because of large variations in the amounts of straw used (Berhane et al., 2020).

Straw biochemical characteristics are cellulose, hemicellulose, neutral detergent soluble (soluble NDS fraction), water soluble carbon (WSC), and lignin (Zhang et al., 2026) in this meta-analysis. Cellulose, hemicellulose, and lignin relative contents, indicative of the composition of the insoluble residue fraction, were used; the difference between their sum and 100% was considered the soluble NDS fraction. The WSC determined after water extraction was expressed as a percentage of total carbon. The lignocellulose index (LCI) can be used as a criterion to show the recalcitrance of the plant cell wall (Herman et al., 2008). LCI was calculated following Li et al. (2025).

Soil C pool [i.e., soil microbial biomass C (MBC), soil organic C (SOC), soil dissolved organic C (DOC), soil easily oxidizable C (EOC), soil light fraction organic C (LFOC), and soil particulate organic C (POC)] and Hy-EEAs [α-1,4-glucosidase (AG, EC3.2.1.20), β-1,4-glucosidase (BG, EC3.2.1.21), β-1,4-xylosidase (BX, EC3.2.1.37), and β-d-cellobiosidase (CBH, EC3.2.1.91)] were included in this meta-analysis. Based on data availability criteria, HFOC was excluded from the meta-analysis as the number of comparable studies reporting this fraction was too limited to support meaningful statistical synthesis. If a study only reported soil organic matter (SOM) content, SOC was calculated as shown in Equation 3 (Gattinger et al., 2012):

If a study reported more than one type of Hy-EEAs, their sum values were considered the overall responses of Hy-EEAs (see Supplementary Materials and Methods). If a study reported multiple straw return responses (i.e., more than one straw return treatment), each treatment was included separately in our dataset. We also recorded soil pH for both no-straw-return and straw return treatments when these data were reported. All original data used in this meta-analysis are available from the figshare (Li S. L. et al., 2022).

A meta-analysis was used to evaluate the effects of straw return on Hy-EEAs and soil C pool, and other ancillary variables (Hedges et al., 1999). We calculated the logarithmic response ratio (lnR) and its variance for each observation to synthesize the effect of straw return on Hy-EEAs and soil C pool as shown in Equation 4:

where X_S and X_C are the arithmetic mean values in the straw return and no-straw-return treatments, respectively.

The variances (ν) of lnR were calculated as shown in Equation 5 (Chen et al., 2018; Li et al., 2025):

where n S and n C refer to the number of replicates and S_S and S_C are the SD for straw return and no-straw-return treatments, respectively.

We calculated effect sizes using the escalc function in metafor package (Viechtbauer, 2010). Overall effect size in a weighted mixed-effects model was calculated using rma.mv function from the metafor package (Viechtbauer, 2010). The overall effect size was transformed into percentage change, that is, (e ^lnR  − 1) × 100%. The overall effect of straw return on each response variable was considered significant if the p-value was < 0.05.

A random-forest model selection in the context of meta-analysis was used to determine the most important predictors of the effect of straw return on the studied variables. We trained a random-forest meta-analysis with preselected predictors and calculated variable importance with metaforest (Van Lissa, 2017). Model selection analysis in the glmulti R package was used to determine the important predictors of the lnR of Hy-EEAs and soil C pool (Calcagno and de Mazancourt, 2010). Possible combinations of the environmental and experimental variables (e.g., latitude, MAT, MAP, soil pH, soil clay content, soil C:N ratios, straw management, and fertilizer form) were incorporated into the model selection analysis. Model selection was based on the Akaike Information Criterion. The relative importance of each variable for a certain model was estimated as the sum of Akaike weights of all predictors in this model. A threshold of 0.8 was used to differentiate the important and unimportant predictors.

Straw return-induced increase in soil Hy-EEAs reveals that soil microorganisms help decompose additional C inputs by stimulating the production of extracellular enzymes (Sinsabaugh, 2010; Zhang et al., 2016). Indeed, several studies have reported that straw return favors soil microbial functional communities degrading the additional C inputs (Henriksen and Breland, 1999; Zhao X. M. et al., 2016). In this study, we propose four possible explanations (Figure 4).

First, shifts in C substrate availability due to straw return may be the explicit mechanism to explain this change. This is because soil Hy-EEAs could hydrolyze labile C substrates (Ma et al., 2019; Zuo et al., 2022). Root exudates and plant litter are important sources of hydrolyzable C (Zechmeister-Boltenstern et al., 2015). Straw return can increase crop growth and thus increase the production of root exudates and plant litter (Liu et al., 2014; Xia et al., 2018), which in turn increases the availability of labile C substrates. Therefore, to support microbial metabolic activities, soil microbes might increase soil Hy-EEAs to use existing pools of labile C substrates (Chen et al., 2020; Cotrufo et al., 2013).

Second, a shift in soil nutrient contents with straw return could help explain the changes in Hy-EEAs. Straw return can enhance soil nutrient contents, such as soil nitrogen and soil C, promoting crop growth (Xia et al., 2018). Straw return stimulation of crop growth would also change root community composition, forming more soil microorganisms and soil organic matter (Huang et al., 2021; Li et al., 2020; Wingeyer et al., 2012). Besides, straw return-stimulation of soil temperature and soil water would enhance the soil organic matter mineralization and soil nutrient availability (Jarvis et al., 1996; Stemmer et al., 1999). Increased soil nutrient availability with straw return could alleviate plant–microbial competition for nutrients and increase nutrient storage capacity to fuel the production of Hy-EEAs (Liu et al., 2014; Xia et al., 2018).

Third, soil clay content could account for shifts in Hy-EEAs (Supplementary Figures S2, S3). Soil texture usually reflects soil aeration, with soil clay content having an important role in soil water-holding capacity (Liu et al., 2014; Yu et al., 2018). Previous studies have shown that soil clay content is a key soil property affecting the soil’s capacity to store C (Jagadamma and Lal, 2010; Six et al., 2004). Importantly, straw return is more conducive to decreasing clay dispersion (Huang et al., 2021; Liu et al., 2014), which will prevent structurally complex macromolecules and decrease their accessibility to soil microbes, potentially increasing the production of Hy-EEAs (Fenner and Freeman, 2020).

Fourth, shifts in soil pH may be one of the predictors of the variation in the response of Hy-EEAs to straw return (Supplementary Figure S1). The predominant role of soil pH has been reported in published studies at local sites and/or large spatial regions (Bahram et al., 2018; Shu et al., 2022; Zheng et al., 2019). There are associations between soil pH and other soil characteristics (Högberg et al., 2007). For example, soil pH is a major factor influencing the structure of the soil microbial community (Fierer and Jackson, 2006; Nilsson et al., 2007; Powlson et al., 2011; Zak et al., 2003). Straw return affects soil pH (Islam et al., 2022), making the soil more suitable for microbial community composition and growth (Zak et al., 2003). Therefore, straw return-induced changes in enzyme and microbial activities could increase soil Hy-EEAs (Fierer and Jackson, 2006; Nilsson et al., 2007; Powlson et al., 2011).

Our results indicated that straw return significantly increased soil microbial biomass carbon (MBC) content (p < 0.05; Figure 2a). This increase can be explained by several interconnected mechanisms: (1) Straw provides a direct labile carbon source, stimulating microbial growth and activity. Subsequent production of microbial by-products promotes the formation and stabilization of soil macroaggregates, which can physically protect microbial biomass and reduce its turnover, leading to net MBC accumulation (Jastrow, 1996; Kalbitz et al., 2000). (2) By altering soil physical conditions, surface straw may limit oxygen diffusion and slow the decomposition rate. This suppression of aerobic mineralization can shift the microbial community toward groups with higher biomass efficiency or slower turnover, thereby increasing MBC (Zhu et al., 2015). (3) Straw incorporation rapidly modifies the pool sizes of labile soil organic carbon fractions. This shift in resource availability stimulates microbial growth and can result in a short-term elevation of MBC (Roper et al., 2010; Singh et al., 2007). (4) Straw amendment can enhance crop root growth and rhizodeposition. The increased input of readily available root-derived carbon further supports microbial growth in the rhizosphere, contributing to higher MBC (Li S. L. et al., 2022; Maarastawi et al., 2019).

In our study, straw-induced increases in Hy-EEAs were positively correlated with soil MBC content (Table 1). Why does the effect of straw return on MBC content increase with increasing Hy-EEAs? The degradation of labile C substrates requires less energy than the degradation of structurally complex macromolecule substrates (Mooshammer et al., 2017; Sinsabaugh, 2010). In addition, soil microbes can adjust their community composition or change their C utilization strategies to adapt to straw return (Huang et al., 2021; Jensen et al., 1997; Ocio et al., 1991; Xia et al., 2018). Alternatively, straw return could help restructure the microbial community and change litter quality, promoting the microbial production of Hy-EEAs (Liu et al., 2014; Su et al., 2020).

Our results show that straw return significantly enhanced SOC storage. Importantly, straw return-induced increases in Hy-EEAs were positively correlated with straw return responses of SOC storage. This is primarily due to the fact that straw return increased C allocation for microbial growth by enhancing soil enzyme activity, which would increase SOC storage (Diacono and Montemurro, 2011; Guo et al., 2018). The increase in C allocation for microbial growth will increase the microbial C residual available for soil C (Martens et al., 1992; Wang et al., 2021). Furthermore, straw return could increase the microbial C use efficiency, i.e., the ratio of C allocated for growth to C allocated for respiration (Allison et al., 2007; Manzoni et al., 2012; Rath and Rousk, 2015). Another explanation may be that straw return will likely decrease the accessibility of newly generated microbial C residue and formerly protected SOM by stimulating Hy-EEAs to degrade less chemically complex macromolecules (Chen et al., 2020; Luo et al., 2018; Mueller et al., 2020; Sinsabaugh and Shah, 2012). The decomposition of these uncomplicated macromolecules would release some associated physically and/or chemically related N and phosphorus (P), which may amplify the effect of straw return on SOM decomposition due to increased nutrient availability to microbial decomposers (Chen et al., 2018; Lavallee et al., 2020).

The positive correlation between straw return effects on SOC storage and Hy-EEAs (Table 1) indicates that straw return effects on SOC storage can be explained by soil enzyme responses. Indeed, soil C stocks are affected by the balance between crop litter that is decomposed and transformed into soil organic matter versus the amount that is mineralized (Cotrufo et al., 2015; Kallenbach et al., 2016). Thus, other soil processes, for example, straw return-induced shifts in root exudation, litter input, the formation of stable soil organic matter from microbial products, and C leaching, would also contribute to shifts in SOC storage with straw return (Liang et al., 2019; Luo et al., 2017; Pausch and Kuzyakov, 2018), while those potential processes are not evaluated in this meta-analysis. Besides, even though the enzymes analyzed in this meta-analysis could indirectly affect soil C decomposition in bulk soil, they are involved in the decomposition of particulate soil organic matter and plant litter (Lavallee et al., 2020; Kuzyakov, 2010; Soong et al., 2020; Zechmeister-Boltenstern et al., 2015). Therefore, although straw return-induced changes in Hy-EEAs play a critical role in determining soil C dynamics with straw return, the microbial mechanisms underlying shifts in enzyme activities remain unclear.

Understanding the effects of straw return on C in soils and Hy-EEAs and revealing the key mechanisms between them will help improve predictions of below-ground C dynamics of the future in agricultural systems. Our meta-analysis provided important information for the test and development of biogeochemistry models, further predicting the potential mechanisms for soil C cycling and turnover. Several critical uncertainties with respect to further studies still exist. First, since the differences in straw amount and straw type differ among cropping systems and vary substantially across different experiment sites (Berhane et al., 2020), this heterogeneity might pose a major challenge to the prediction of microbial EEAs, microbial processes, and soil C cycling under straw return.

Second, few studies simultaneously measured soil enzyme activities and soil C content under straw return in the same experimental platform. Therefore, to advance our understanding of the relationships between soil enzyme activities and C in soils under straw return, we strongly encourage agricultural researchers to observe enzyme and soil C stocks concurrently.

Third, the microbial mechanisms’ potential shifts in soil enzyme activities remain unclear, hampering the incorporation of microbial mechanisms in the models (Chen et al., 2019). Although incorporating microbial processes can promote the performance of Earth System Models (Allison et al., 2010; Wieder et al., 2013), the representation of microbial processes in these models varies and is disputed (Treseder et al., 2012). Therefore, we suggest that future research quantifies the associations between gene expression, microbial community composition, and soil enzyme activities.

Fourth, straw return might interactively and simultaneously change the soil characteristics (e.g., soil pH, soil moisture, and microbial biomass) (Liu et al., 2014; Xia et al., 2018), which could potentially modify Hy-EEAs and further mediate soil C cycling. These changes may render the impacts of straw return on soil C dynamics and Hy-EEAs are highly complex but deserve to be further explored and studied.