The field experiment was conducted at the Xinzhou Experimental Station of the China National Maize Industry System (38°24′ N, 112°43′ E), located in a temperate continental monsoon climate zone. The soil at the site was classified as sandy loam. During the maize growing season, the mean air temperature was 20.38°C, and cumulative precipitation was 342.3 mm. Detailed climatic data for the experimental period are provided in Supplementary Figure S1.

This study was performed in the fourth year (2024) of a long-term experiment to assess the effects of different mulching practices on the spatiotemporal enrichment of rhizosphere microbes and to identify key microbial taxa. A randomized complete block design (RCBD) was employed, comprising three treatments: (i) film-side planting (FPM) using 60 cm-wide plastic film; (ii) on-film hole-sowing (UPM) using 80 cm-wide plastic film; and (iii) no mulching (CK) as the control. Each treatment was replicated three times, with plot dimensions of 7 m × 5 m (35 m²), resulting in a total of nine plots. The maize cultivar ‘Xianyu 335’ was sown at a plant spacing of 28 cm. A controlled-release compound fertilizer (28:15:8, N:P₂O₅:K₂O) was applied at a rate of 472.5 kg ha^-¹. Sowing was carried out on 2 May 2024, and the harvest took place on 28 September 2024. Standard local agronomic practices for pest, disease, and weed management were followed throughout the growing season.

Rhizosphere soil samples were collected from maize plants at the V12 (twelve-leaf) and R6 (physiological maturity) growth stages (Chen et al., 2019). These two stages were selected to represent the peak of vegetative growth with vigorous root activity (V12) and the final reproductive stage characterized by root senescence and cumulative treatment effects (R6). In each plot, sampling was conducted at three soil depths (10, 20, and 30 cm) using a standard five-point sampling method. At each sampling point, roots within each specific soil layer were carefully exposed. The plants were carefully uprooted, and loose soil was gently shaken off. The rhizosphere soil tightly adhering to the lateral roots at that specific depth was then collected using a sterile brush. To ensure representativeness and minimize heterogeneity, the rhizosphere soil from these five points at the same depth was thoroughly mixed to form one composite sample per plot. A total of 54 composite samples (3 treatments × 3 replicates × 2 stages × 3 depths) were obtained. From each composite sample, a 2 g subsample was homogenized for subsequent analysis, immediately frozen in liquid nitrogen, and stored at −80°C until further processing.

At harvest, the total number of plants, the number of prolific plants (i.e., those bearing double ears), and the number of lodged plants were recorded for each plot. In each plot, 20 representative ears were randomly collected from the two central rows. These ears were air-dried and transported to the laboratory for yield component analysis. The number of kernels per ear and the thousand-kernel weight (TKW) were subsequently measured. Grain yield of spring maize was calculated and standardized to a 14 % moisture content using the following formula (Guo et al., 2022):

At the 6-leaf stage (V6), 12-leaf stage (V12), tasseling(VT), milk ripening stage (R3), and mature stage (R6) of the maize, three representative maize plants with consistent growth were randomly selected from each plot. The aboveground part of the plant was sampled from the base of the maize plant near the ground surface, and the stems, leaves, sheaths, bracts, and other organs were packed in sampling paper bags, marked for processing, and placed in an oven. The samples were incubated at 105°C for 30 min to inactivate the enzymes and were then heated to a constant weight at 65°C. The dry samples were weighed and recorded using a balance that was accurate to 0.01 g.

At the V6, V12, VT, R3, and R6 stages of maize development, soil samples were collected from the 0–60 cm soil profile in 20 cm increments using a soil auger. Sampling points were located within the crop row between two adjacent plants. The collected samples were then oven-dried at 105°C to a constant weight to determine the gravimetric soil water content.

Soil temperature was monitored at fixed positions within each plot using bent-stem glass mercury thermometers. On the day of sowing, thermometers were installed in the center of each plot, adjacent to a seed between two maize plants, at depths of 5, 10, and 20 cm. From sowing to harvest, soil temperature readings at each depth were recorded at 5-day intervals at 08:00, 14:00, and 20:00 h. The daily mean soil temperature was calculated as the arithmetic mean of these three readings. For each developmental stage, the temperature difference between the air and the 5–20 cm mean soil temperature was also determined.

The niche breadth of the rhizosphere microbial community was calculated using Levins’ formula. This index reflects both the number of distinct habitats a species occupies and the evenness of its distribution among them, thereby characterizing the extent to which an organism or population utilizes the available spectrum of resources (Hanski, 1978). All niche breadth calculations were performed in the R environment (version 4.3.1) using the spaa package.

Sequencing of the 16S rRNA and ITS amplicons yielded a total of 15,515 and 17,614 bacterial amplicon sequence variants (ASVs) and 1,988 and 3,009 fungal ASVs from samples at the V12 and R6 stages, respectively. The species accumulation curves indicated that the sequencing depth was sufficient to capture the majority of microbial taxa present in the samples (Supplementary Figure S3).

At the phylum level, the rhizosphere microbiome was dominated by a few key phyla, whose relative abundances showed pronounced spatiotemporal dynamics (Supplementary Figure S4). For bacteria, Proteobacteria and Actinobacteriota exhibited opposing depth preferences that shifted between growth stages. For fungi, a striking finding was the exceptionally high relative abundance of Basidiomycota in the mid-depth soil under UPM at V12, while both Ascomycota and Basidiomycota were co-enriched in the deep soil under UPM by R6. Notably, the symbiotic phylum Glomeromycota remained at a consistently low abundance across all mulched conditions.

At the genus level, the microbiome exhibited more fine-grained and functionally relevant shifts in response to treatments (Figure 2). During the V12 stage, key bacterial genera displayed distinct niche preferences (Figure 2A). Sphingomonas, a dominant genus with relative abundances ranging from 4.97% to 14.38%, clearly favored the shallower soil layers under mulching treatments. In stark contrast, the beneficial genus Streptomyces was most abundant in the deep, unmulched soil (CK at 30 cm), reaching a peak of 6.87%. This bacterial composition evolved by the R6 stage (Figure 2B). A functionally critical shift was the marked suppression of the nitrifying genus Nitrospira in the deep soil under both mulching treatments compared to the CK control, where it reached its highest relative abundance.

The fungal community at the genus level was also highly dynamic and treatment-sensitive. At the V12 stage, several genera showed highly localized enrichment, most notably the potential plant pathogen Fusarium, whose relative abundance was significantly elevated to 9.42% only under FPM in the topsoil (Figure 2C). By the R6 stage, the community structure continued to diverge (Figure 2D). Chaetomium was exclusively enriched in the deep soil of the FPM treatment, reaching 18.67%. The most remarkable trend was observed for Ceratobasidium, which displayed a dramatic increase in relative abundance with soil depth specifically under the UPM treatment, rising from 1.48% at 10 cm to a dominant 11.89% at 30 cm.

To elucidate the potential ecological consequences of the observed community shifts, we first identified the most influential microbial genera driving these changes using Random Forest models (Supplementary Figures S10, S11). We then performed a series of Spearman's correlation analyses using these key discriminant genera. First, we aimed to infer their functional roles by correlating their abundances with predicted functions (FAPROTAX and FUNGuild) (Figures 4, 5), and then directly linked them to maize productivity indicators to assess their relevance to crop performance.

The analysis identified associations between key taxa and their inferred ecological roles (Supplementary Figures S12, S13). For bacteria, the most notable finding was the strong positive correlation between Nitrospira and aerobic pathways, alongside its negative correlation with anaerobic N-cycling pathways. This aligns with its known role as an aerobic nitrifier and provides a functional rationale for its suppression under the likely hypoxic conditions of mulched soils. For fungi, the analysis supported the trophic modes of several key genera. For instance, the symbiotic role of Glomus was consistent with its strong positive correlation with the Arbuscular Mycorrhizal guild, while the abundance of Ceratobasidium at R6 was strongly and positively correlated with the Endomycorrhizal guild, suggesting a specific symbiotic strategy for this genus.

To directly link these microbial shifts to crop performance, we performed a correlation analysis between key microbial taxa and maize productivity indicators (Figure 6). A striking pattern of negative correlations was observed, where the relative abundances of the nitrifying genus Nitrospira and the dominant genus Pseudomonas were significantly and negatively correlated with Grain Yield, Ear Number, and Dry Matter. In contrast, a suite of other key taxa exhibited significant positive correlations with productivity. The beneficial bacterial genus Streptomyces was positively correlated with Grain Yield, Kernel Number per Ear, and Dry Matter. Similarly, the fungal genus Chaetomium showed strong positive correlations with multiple yield components, including Grain Yield and Ear Number. These findings provide direct, albeit correlational, evidence that the mulching-induced shifts in specific microbial taxa—particularly those with important inferred functions in nutrient cycling and plant health—are closely linked to the final crop performance.

Agricultural ecosystems in arid and semi-arid regions face the dual challenges of water scarcity and high evaporative demand, which severely constrain crop productivity and resource-use efficiency (Zhang et al., 2022). Film mulching, a well-established agronomic practice, has been widely adopted to mitigate these constraints by improving the soil microenvironment, enhancing water-use efficiency (WUE), and suppressing weed growth (Gu et al., 2018; Xu et al., 2018). Consistent with these established benefits, both the FPM and UPM treatments significantly increased spring maize yield, accompanied by a higher number of effective ears. This result aligns with previous reports on the yield-enhancing effects of mulching in similar dryland regions (Zhang et al., 2022).

The underlying mechanism for this yield improvement is closely linked to the mulching-driven regulation of the soil hydrothermal environment. It is well-documented that during the early seedling stage, film mulching effectively reduces soil evaporation, elevates soil temperature, and improves water retention (Gao et al., 2014). These favorable conditions promote robust early root development and biomass accumulation, laying the foundation for higher final yields. Indeed, the UPM treatment accelerated biomass accumulation during the early to mid-vegetative stages (V12 and VT, likely due to its superior heat-trapping and moisture-preserving capabilities which create an “incubator effect” for early growth (Boatwright et al., 1976). However, this early advantage did not translate to the highest final biomass. Instead, the FPM treatment, despite a slower start, surpassed UPM in dry matter accumulation during the reproductive stages (R3 and R6). This suggests a critical trade-off: the aggressive early growth under UPM may lead to premature senescence, a phenomenon often linked to supra-optimal soil temperatures and hypoxic root conditions under full film coverage (Xu et al., 2015). High soil temperatures, particularly during the reproductive stage, can accelerate plant development but also impair root activity and nutrient uptake, leading to a shortened grain-filling period and reduced final biomass (Li et al., 2013). In contrast, FPM likely achieved this balance by facilitating gas exchange and moderating excessive soil heating. This strategy sustained plant vigor through to maturity, driving the highest final biomass and yield. These findings suggest that the advantage of FPM lies in mitigating late-season physiological stress, rather than merely maximizing early growth.

Film mulching fundamentally reshapes the rhizosphere microbiome by altering soil hydrothermal and aeration conditions, leading to pronounced spatiotemporal heterogeneity in community diversity, structure, and composition (Folman et al., 2001; Wang et al., 2021). We acknowledge that microbial community composition is also influenced by soil chemical properties (e.g., pH and organic matter), which were not systematically quantified in the present study. In dryland ecosystems, however, soil temperature and moisture constitute key environmental constraints, and the mulching-induced hydrothermal gradients captured here are therefore considered likely major drivers of the observed microbial patterns, while chemical co-variates may play a contributory role.

Our study revealed that the impact of mulching on microbial α-diversity was subtle and highly context-dependent, a finding consistent with some previous reports (Sun et al., 2021). For instance, at the V12 stage, the UPM treatment transiently promoted bacterial richness in the topsoil, likely due to early improvements in hydrothermal conditions. However, by the R6 stage, this effect diminished, and greater bacterial diversity was observed in the deep soil of the unmulched control, possibly because prolonged mulching created less favorable conditions (e.g., reduced aeration) that constrained diversity (Li et al., 2020).

Unlike the nuanced changes in α-diversity, β-diversity revealed pronounced, treatment-specific shifts in community structure. This disparity underscores that mulching primarily reshapes the identities and relative abundances of microbial taxa, rather than merely altering species richness (Lozupone and Knight, 2007). Initially, treatment effects were restricted to the topsoil. By the R6 stage, however, these structural divergences had propagated into the subsoil. This shift was particularly pronounced for bacteria, with the strongest differentiation occurring at 30 cm (P = 0.001). This pattern likely reflects a lagged response of the deep-soil microbiome to the cumulative alteration of soil conditions and the changing inputs from senescing roots. Conversely, the fungal community exhibited an attenuating response pattern, with treatment effects diminishing in deeper soil layers at the later stage. This marked divergence from bacteria underscores that these two kingdoms employ distinct strategies to navigate mulching-induced gradients. Bacteria, as rapid-responding r-strategists, likely track localized physicochemical shifts. In contrast, fungi may buffer against deep-soil heterogeneity through their extensive mycelial networks, allowing them to integrate environmental signals across larger spatial scales and maintain greater stability in the subsoil (Bahram et al., 2018).

Consistent with previous studies on gramineous crops, the maize rhizosphere was dominated by Proteobacteria, Actinobacteriota, Ascomycota, and Basidiomycota (Peiffer et al., 2013; Coleman-Derr et al., 2016). More importantly, the results revealed dynamic responses of key functional taxa, exemplified by the nitrifying genus Nitrospira, which was markedly suppressed in deeper soil layers under mulched treatments; this pattern in likely microaerobic soil layers at the R6 stage strongly points to oxygen limitation as a key driver, a finding that contrasts with hypotheses emphasizing substrate availability as the primary control in some agricultural systems (Zhang et al., 2012). This suggests that under long-term film mulching, the physical barrier to gas exchange may override the biochemical benefits of enhanced mineralization, fundamentally altering the nitrogen cycle. Similarly, the consistent suppression of the symbiotic arbuscular mycorrhizal fungi (Glomeromycota) under both mulching treatments is a critical finding, indicating a potential ecological trade-off where the physical benefits of mulching (e.g., improved water status) may come at the cost of weakening a key plant-microbe mutualism (Treseder, 2004). The sustained suppression of Glomeromycota observed in this study suggests potential longer-term ecological implications, as AMF contribute to phosphorus acquisition and soil aggregation through glomalin production (Rillig, 2004). However, these implications cannot be directly inferred from the present short-term observations, and long-term studies are needed to evaluate whether management strategies, such as crop rotation or organic amendments, can mitigate these effects. Furthermore, the enrichment of the potential plant pathogen Fusarium under FPM at the V12 stage, and the accumulation of the root-associated genus Ceratobasidium under UPM at the R6 stage, further illustrate how specific mulching regimes create distinct ecological niches that favor different functional guilds.

Shifts in community structure ultimately translate into a reconfiguration of the rhizosphere microbiome’s functional potential. While amplicon sequencing primarily reveals community structure, function prediction based on marker genes can still provide critical hypotheses and directions for understanding how mulching impacts the potential functions of the ecosystem (Ling et al., 2022). While we acknowledge that marker-gene-based predictions are inferential, the functional profiles derived from PICRUSt2, FAPROTAX, and FUNGuild demonstrated remarkable consistency with our taxonomic observations and environmental data. This multi-evidence convergence robustly indicates a systematic reconfiguration of rhizosphere carbon and nitrogen cycling and pathogen dynamics under film mulching.

Our study indicates that the two mulching strategies appeared to foster fundamentally different rhizosphere ecosystems. During the early V12 stage, the UPM treatment created a high-temperature, high-moisture, and poorly aerated microenvironment that strongly selected for anaerobic metabolic guilds, particularly those involved in denitrification (Yu et al., 2021). We hypothesize that this hypoxic stress imposed physiological burdens on the root system, potentially contributing to the premature senescence often observed in on-film sowing. In contrast, the FPM treatment, likely by providing superior gas permeability, established a more balanced rhizosphere dominated by aerobic heterotrophs and nitrogen-fixing bacteria. Meanwhile, the unmulched CK treatment represented a baseline state, supporting a community focused on degrading complex native organic matter. By the R6 stage, the primary driver of functional differentiation shifted from purely environmental selection to a complex interplay between environmental stress and substrate availability from senescing roots (Gunina and Kuzyakov, 2015). Under UPM, the abundance of plant residues fueled a massive enhancement of anaerobic metabolism and decomposition pathways, but also increased the potential for plant pathogens. The microbial community under UPM was characterized by the enrichment of pathways associated with quorum sensing and environmental adaptation, indicating a community with intensified microbial interactions under relatively favorable resource conditions. In contrast, the FPM treatment was associated with higher predicted functional potentials in core metabolic pathways, including carbon metabolism, oxidative phosphorylation, and macromolecule biosynthesis. These patterns are consistent with a microbial community exhibiting enhanced metabolic capacity under FPM, although such inferences are based on functional predictions rather than direct measurements of microbial activity.

The fungal community displayed parallel, yet distinct, functional trajectories. At the V12 stage, the FPM treatment selectively promoted beneficial guilds, including symbiotic AMF and predatory nematophagous fungi, indicating the potential establishment of a disease-suppressive rhizosphere. The UPM treatment, in contrast, tended to enrich fungal parasites and epiphytes. As plant senescence progressed to the R6 stage, the UPM treatment became a hotbed for both saprotrophic and pathogenic fungi, likely fueled by nutrients from decaying roots. Intriguingly, it also harbored the highest potential for AMF at this stage, suggesting a complex trade-off between pathogenic and symbiotic interactions under these specific conditions. It is noteworthy that while different prediction tools pointed to different treatments as having higher pathogenic potential, the overarching conclusion is that the warm, moist environment created by any form of mulching can increase the risk of proliferation for certain thermophilic or hygrophilous pathogens.

A key observation was the significant correlation between specific microbial taxa and maize productivity. A striking, albeit counter-intuitive, observation was the significant negative correlation between the relative abundances of the nitrifying genus Nitrospira and the versatile genus Pseudomonas with multiple yield components, including grain yield and dry matter. However, we argue that this does not necessarily imply a direct suppressive effect of these microbes on plant growth. Instead, this correlation likely reflects their role as bio-indicators of different ecosystem states. It is well-established that at larger scales, microbial community structure is predominantly governed by abiotic factors (Fierer and Jackson, 2006). In our system, Nitrospira, a known aerobe, thrived in the unmulched (CK) treatment, which, due to water and heat stress, was also the lowest-yielding system. Thus, the high abundance of Nitrospira is an indicator of the well-aerated but less productive CK environment, rather than a direct cause of low yield. This underscores the primacy of physicochemical drivers in our arid system, where the powerful water- and heat-conserving effects of mulching are the dominant determinants of yield, potentially overriding the subtler effects of individual microbial taxa. Conversely, the significant positive correlations between the beneficial genus Streptomyces and the fungal genus Chaetomium with grain yield and other productivity indicators provide compelling, albeit correlational, evidence for their contribution to a high-yielding rhizosphere. These taxa were more abundant in the FPM treatment, which ultimately produced the highest yield. This suggests that the functionally balanced, aerobic microbial community fostered by FPM is a key feature of a high-productivity agroecosystem.

To further explore the functional underpinnings of these microbe-yield relationships, we correlated these key taxa with their predicted ecological functions (detailed in Supplementary Figures S12, Supplementary Figures S13). This analysis provided strong support for their inferred roles. For instance, the positive correlation of Nitrospira with aerobic chemoheterotrophy and its negative correlation with anaerobic N-cycling pathways confirmed its aerobic nature, reinforcing the hypothesis that its suppression under mulching is driven by oxygen limitation (Daims et al., 2015). Similarly, the symbiotic role of Glomus (Berruti et al., 2016) and the multi-trophic nature of Mycosphaerella (Crous et al., 2009) were also consistent with their correlations to respective functional guilds. A particularly important finding was the strong positive correlation of Ceratobasidium with the Endomycorrhizal guild, suggesting its potential role as a key root-associated symbiont under late-season UPM conditions (Weiß et al., 2011). These taxon-function linkages, while inferential, provide a mechanistic basis for understanding how the specific microbial players shaped by mulching could contribute to the overall functional trajectory of the rhizosphere and, ultimately, its association with crop productivity.