Long-Term Daytime Warming Rather Than Nighttime Warming Alters Soil Microbial Composition in a Semi-Arid Grassland

Simple Summary Global mean temperature has increased by 1.07 °C since the Industrial Revolution, which arouses people’s widespread concern about climate warming. Nighttime temperatures increase faster and higher than daytime temperatures around the world. Various microbial groups in the soil may respond differently to such asymmetrically diurnal warming and then change ecological functions. In this study, we used a ten-year experiment in a semi-arid grassland to examine the effects of daytime and nighttime warming on soil microbial composition. The results showed that short-term warming did not change soil microbial composition, but long-term daytime warming rather than nighttime warming decreased the fungi-to-bacteria ratio in soils. In addition, soil respiration enhanced with the decreasing fungi-to-bacteria ratio. This work implies the importance of soil microbial composition in regulating grassland C release under long-term warming, which may help us accurately assess the climate-C feedback in terrestrial ecosystems. Abstract Climate warming has profoundly influenced community structure and ecosystem functions in the terrestrial biosphere. However, how asymmetric rising temperatures between daytime and nighttime affect soil microbial communities that predominantly regulate soil carbon (C) release remains unclear. As part of a decade-long warming manipulation experiment in a semi-arid grassland, we aimed to examine the effects of short- and long-term asymmetrically diurnal warming on soil microbial composition. Neither daytime nor nighttime warming affected soil microbial composition in the short term, whereas long-term daytime warming instead of nighttime warming decreased fungal abundance by 6.28% (p < 0.05) and the ratio of fungi to bacteria by 6.76% (p < 0.01), which could be caused by the elevated soil temperature, reduced soil moisture, and increased grass cover. In addition, soil respiration enhanced with the decreasing fungi-to-bacteria ratio, but was not correlated with microbial biomass C during the 10 years, indicating that microbial composition may be more important than biomass in modulating soil respiration. These observations highlight the crucial role of soil microbial composition in regulating grassland C release under long-term climate warming, which facilitates an accurate assessment of climate-C feedback in the terrestrial biosphere.


Introduction
Climate warming has led to an increase in global mean temperature by 1.07 • C since 1850 [1] and can profoundly influence the carbon (C) balance between plants and soils in terrestrial ecosystems, resulting in an uncertain response of soil C storage to climate warming [2,3]. Soil microorganisms, an essential component in ecological communities, are extensively involved in soil organic matter decomposition and nutrient cycles, thus playing crucial roles in maintaining ecosystem stability [4,5]. Given the high sensitivity of microbial activities to rising temperatures, climate warming may greatly change the community structure and functions of soil microorganisms [6,7]. Previous studies have is classified as Haplic Calcisols according to the Food and Agriculture Organization of the United Nations (FAO) classification standard. Surface soil bulk density and pH are 1.31 g·cm −3 and 7.7 [24]. Plant species in this grassland are dominated by grasses (Agropyron cristatum, Cleistogenes squarrosa, Leymus chinensis, and Stipa krylovii), forbs (Artemisia frigida, Potentilla tanacetifolia, Heteropappus altaicus, and Potentilla acaulis) and legumes (Melilotoides ruthenica, Astragalus adsurgens, and Lespedeza bicolor).

Experimental Design
The experiment was established in March 2006 and conducted as part of a complete random block design with 6 treatments, including control (C), daytime warming (6:00-18:00; D), nighttime warming (18:00-6:00; N), diurnal warming (24 h; W), nitrogen fertilization, and diurnal warming plus nitrogen fertilization. There were 4 replicates for each treatment, and therefore 24 3 m × 4 m plots were assigned in a 4 × 6 matrix, with a 3 m distance between adjacent plots.
The infrared radiators (165 cm × 15 cm; MSR-2420, Kalglo Electronics, Bethlehem, PA, USA) were placed 2.25 m above the ground and set at 1600W as the power output in each warming plot [19,25]. The daily mean soil temperatures were elevated by 0.91 • C and 1.18 • C under daytime and nighttime warming, respectively ( Figure S1). In each control plot, a "dummy" heater with the same shape and size as the infrared heater was installed at the same height to imitate the shading effect. The asymmetric diurnal warming treatments began in April 2006, and the heaters ran continuously and automatically from 15 March to 15 November in each year. No other manipulations were performed in the experimental plots.

Soil Microclimate, PLFA Analysis, and Plant Cover Measurement
Soil temperature at the depth of 10 cm was automatically recorded every 10 min using a STM-01 Soil Temperature Measurement System Datalogger (Henan Electronic Institute, China), and soil temperatures during the growing season from May to October were used as the annual mean value [19]. Soil moisture (SM) at the same site and depth was measured by a portable soil moisture probe (Diviner-2000, Sentek, Australia). Soil moisture was analyzed three times per month from May to October each year.

Data Analysis
The effects of asymmetric daytime and nighttime warming and year, and their possible interactions on soil microclimate and microbial compositions were analyzed using two-way repeated ANOVA measures with SAS V8 software (SAS Institute). Linear regressions were investigated to determine correlations among soil microbial compositions, soil microclimate, and plant functional groups. Redundancy analysis (RDA) and random forest modeling were performed to reveal environmental factors influencing microbial composition in R (v 3.4.0). Structural equation modeling (SEM) exploring the effects of abiotic and biotic factors on soil microbial composition was constructed using AMOS 24.0 (Amos Development Corporation). The goodness of model fit was tested based on the p-value (>0.05), chi-square, and root mean square error of approximation (RMSEA; <0.05). The total, direct, and indirect effects of the abiotic and biotic factors on soil microbial composition were also recorded.

Soil Microclimate
Averaged across the 4 treatments, both soil temperature and moisture substantially fluctuated over the experimental durations (Figures 1 and S2, both p < 0.001, Table 1). Daily mean soil temperature was enhanced by 0.32 • C and 0.58 • C under daytime and nighttime warming averaged over the 5 years, respectively (Figure 1a, both p < 0.001). Soil moisture was decreased by 0.65% (absolute change, p < 0.01) and 0.55% (p < 0.05) under daytime and nighttime warming, respectively ( Figure 1b). There was no interaction between daytime and nighttime warming on soil temperature or moisture. Significance levels: ^ p < 0.1, * p < 0.05, ** p < 0.01, and *** p < 0.001.

Soil Microbial Composition
Averaged over the 5 years, daytime warming did not affect the relative abundance of bacteria, G+ bacteria, AMF, or the ratio of G− to G+ bacteria (Figure 2a,b,d,f, all p > 0.05, Table 1), but reduced fungi proportion by 4.47% (Figure 2c, p < 0.05) and the fungi-tobacteria ratio by 5.63% (Figure 2e, p < 0.01). Nighttime warming had no effect on the relative abundance of any of the microbial groups or the ratios of fungi to bacteria and G− to G+ bacteria (Figure 2, all p > 0.05). The responses of microbial compositions to daytime and nighttime warming varied with year (all p < 0.001). No interaction of daytime and nighttime warming on the relative abundance of any of microbial groups or the ratios of stages, daytime warming significantly increased soil temperature by 0.43 • (p < 0.001) and marginally reduced soil moisture by 0.53% (p < 0.1). Nighttime warming stimulated soil temperature by 0.50 • C (p < 0.001) but had no effect on soil moisture (p > 0.05) in the short term ( Figure 1 and Table 1). By contrast, daytime warming enhanced soil temperature by Biology 2023, 12, 699 5 of 12 0.23 • C (p < 0.05) and suppressed soil moisture by 0.70% (p < 0.05), and nighttime warming increased soil temperature by 0.64 • C (p < 0.001) and decreased soil moisture by 0.66% (p < 0.05) in the long term ( Figure 1).   Significance levels:ˆp < 0.1, * p < 0.05, ** p < 0.01, and *** p < 0.001.

Soil Microbial Composition
Averaged over the 5 years, daytime warming did not affect the relative abundance of bacteria, G+ bacteria, AMF, or the ratio of G− to G+ bacteria ( Figure 2a,b,d,f, all p > 0.05, Table 1), but reduced fungi proportion by 4.47% (Figure 2c, p < 0.05) and the fungi-tobacteria ratio by 5.63% (Figure 2e, p < 0.01). Nighttime warming had no effect on the relative abundance of any of the microbial groups or the ratios of fungi to bacteria and G− to G+ bacteria ( Figure 2, all p > 0.05). The responses of microbial compositions to daytime and nighttime warming varied with year (all p < 0.001). No interaction of daytime and nighttime warming on the relative abundance of any of microbial groups or the ratios of fungi to bacteria and G− to G+ bacteria was detected.
Averaged across the four treatments, the relative abundance of bacteria and the G− to G+ bacteria ratio were higher in the long-than in the short-term stage (Figure 3a, all p < 0.001). However, the relative abundance of G+ bacteria (p < 0.001) and fungi (p < 0.05) as well as the fungi-to-bacteria ratio (p < 0.001) in the long term were lower than those in the short term ( Figure 3a). Nighttime warming did not change the composition of any of the microbial groups in either short or long terms (all p > 0.05, Table 1), whereas daytime warming reduced the relative abundance of fungi by 6.28% (p < 0.05) and the ratio of fungi to bacteria by 6.76% (p < 0.01) in the long term only (Figure 3b). Averaged across the four treatments, the relative abundance of bacteria and th to G+ bacteria ratio were higher in the long-than in the short-term stage (Figure 3a < 0.001). However, the relative abundance of G+ bacteria (p < 0.001) and fungi (p < 0. well as the fungi-to-bacteria ratio (p < 0.001) in the long term were lower than those short term (Figure 3a). Nighttime warming did not change the composition of any microbial groups in either short or long terms (all p > 0.05, Table 1), whereas da warming reduced the relative abundance of fungi by 6.28% (p < 0.05) and the ratio of to bacteria by 6.76% (p < 0.01) in the long term only (Figure 3b).

Relationships of Microbial Composition with Soil Microclimate and Plant Cover
Grass cover was not affected by daytime warming but was significantly reduced by 19.2% ( Figure S3a, p < 0.05, Table S1) under nighttime warming in the long term. Forb cover was suppressed by 8.89% and 22.5% (both p < 0.05) under daytime warming in the

Relationships of Microbial Composition with Soil Microclimate and Plant Cover
Grass cover was not affected by daytime warming but was significantly reduced by 19.2% ( Figure S3a, p < 0.05, Table S1) under nighttime warming in the long term. Forb cover was suppressed by 8.89% and 22.5% (both p < 0.05) under daytime warming in the short and long term, respectively, but remained unchanged under nighttime warming ( Figure S3b). Legume cover was increased by 82.8% (p < 0.01) under daytime warming in the short term but was not changed under nighttime warming ( Figure S3c).

Controls of Abiotic and Biotic Factors on Soil Microbial Composition
Redundancy analysis (RDA) revealed the correlations among different treatments, soil microclimate, and plant functional groups, with the first and second canonical axis explaining 87.6% and 2.3% of the variance, respectively (Figure 5a). Grass cover, soil temperature, and moisture were the dominant factors influencing microbial variations, accounting for 8.91% (p < 0.01), 7.40% (p < 0.05), and 4.81% (p < 0.05) of the increase in mean square error (InMSE) based on random forest model analysis (Figure 5b).

Controls of Abiotic and Biotic Factors on Soil Microbial Composition
Redundancy analysis (RDA) revealed the correlations among different treatments, soil microclimate, and plant functional groups, with the first and second canonical axis explaining 87.6% and 2.3% of the variance, respectively (Figure 5a). Grass cover, soil temperature, and moisture were the dominant factors influencing microbial variations, accounting for 8.91% (p < 0.01), 7.40% (p < 0.05), and 4.81% (p < 0.05) of the increase in mean square error (InMSE) based on random forest model analysis (Figure 5b).
The results of SEM analysis showed that under daytime warming, the increased soil temperature directly changed microbial composition as revealed by the fungi-to-bacteria ratio (standardized direct effect was 0.472, Figure 6a), and indirectly affected it through the changes in soil moisture and grass cover (standardized indirect effect was 0.024, Figure 6a).
Under nighttime warming, the ratio of fungi to bacteria was significantly correlated with nighttime soil temperature only (standardized total effect was 0.412, Figure 6b).
lines and shaded areas describe linear regressions and 95% CIs.

Controls of Abiotic and Biotic Factors on Soil Microbial Composition
Redundancy analysis (RDA) revealed the correlations among different treatments, soil microclimate, and plant functional groups, with the first and second canonical axis explaining 87.6% and 2.3% of the variance, respectively (Figure 5a). Grass cover, soil temperature, and moisture were the dominant factors influencing microbial variations, accounting for 8.91% (p < 0.01), 7.40% (p < 0.05), and 4.81% (p < 0.05) of the increase in mean square error (InMSE) based on random forest model analysis (Figure 5b). The results of SEM analysis showed that under daytime warming, the increased soil temperature directly changed microbial composition as revealed by the fungi-to-bacteria ratio (standardized direct effect was 0.472, Figure 6a), and indirectly affected it through the changes in soil moisture and grass cover (standardized indirect effect was 0.024, Figure 6a). Under nighttime warming, the ratio of fungi to bacteria was significantly correlated with nighttime soil temperature only (standardized total effect was 0.412, Figure 6b).

Short-and Long-Term Warming on Soil Microbial Composition
The obvious interannual variation in soil microclimate and some microbial group (e.g., G+ and G− bacteria) found in our study can be mainly attributed to the changes i precipitation amounts, the natural succession of plant communities, and the potential in teractions among different microbial components (Figures S2-S5). Our results that long term rather than short-term warming decreases fungal abundance and alters microbia composition revealed by the fungi-to-bacteria ratio agree with previous findings tha warming effects on soil microbial communities mainly depend on treatment duratio [9,26], with unchanged microbial biomass, diversity, and composition observing in sev eral short-term (e.g., 1-6 years) warming experiments [9,26,27]. This can be explained b several possible reasons: First, the moderate elevation of soil temperatures (<1 °C) durin the first two years found in our study might not be sufficient to induce a significant re sponse in the microbial community due to the wide range of microbial adaptation [9,26 Second, although soil moisture, forb, and legume cover vary with short-term warming they may have legacy effects on the soil microbial community because of the strong re sistance and functional redundancy [8,9]. Third, the assembly and stabilization of the so

Short-and Long-Term Warming on Soil Microbial Composition
The obvious interannual variation in soil microclimate and some microbial groups (e.g., G+ and G− bacteria) found in our study can be mainly attributed to the changes in precipitation amounts, the natural succession of plant communities, and the potential interactions among different microbial components (Figures S2-S5). Our results that long-term rather than short-term warming decreases fungal abundance and alters microbial composition revealed by the fungi-to-bacteria ratio agree with previous findings that warming effects on soil microbial communities mainly depend on treatment duration [9,26], with unchanged microbial biomass, diversity, and composition observing in several short-term (e.g., 1-6 years) warming experiments [9,26,27]. This can be explained by several possible Biology 2023, 12, 699 9 of 12 reasons: First, the moderate elevation of soil temperatures (<1 • C) during the first two years found in our study might not be sufficient to induce a significant response in the microbial community due to the wide range of microbial adaptation [9,26]. Second, although soil moisture, forb, and legume cover vary with short-term warming, they may have legacy effects on the soil microbial community because of the strong resistance and functional redundancy [8,9]. Third, the assembly and stabilization of the soil microbial community experienced considerable long-term environmental fluctuations, which partially explains the weak response of soil microorganisms to short-term warming [28].
In addition, the reductions in fungal abundance and consequent fungi-to-bacteria ratio under long-term warming can be attributed to the changes in soil microclimate and increased cover of grasses (Figures 4, 5, S6 and S8), which agrees with our Hypothesis 1. Previous studies have found the suppressed competitive ability of some grasses decreased microbial biomass and C use efficiency under warming in this temperate steppe [8,19], but no linkages between above-and below-ground components were examined. The dominant grass species in this grassland, such as Achnatherum Sibiricum, Agropyron Cristatum, and Stipa Krylovii, typically have taller stature but shallower root distributions than forbs and legumes; therefore, the presence of these species may lead to lower soil temperature because of the shading effect and have subsequent impacts on microbial activity in surface soils. Moreover, warming can reduce organic C inputs through grass litter [29] which decreases substrate availability, thereby decreasing fungal activity and saprotrophic fungal proportions [7]. Overall, our study not only suggests the close association between soil microbial composition and grass cover but also indicates that fungi respond to long-term warming more sensitively than bacteria in this semi-arid grassland.

Asymmetrically Diurnal Warming on Soil Microbial Composition
The observations of decreased fungal abundance and the fungi-to-bacteria ratio under daytime warming in the long term do not agree with our Hypothesis 2 and previous findings that nighttime warming increases fungal diversity and changes microbial composition and C use efficiency [8,19,21]. Two potential mechanisms may account for the daytime warming-induced reduction in fungal abundance. First, the enhancement in microbial metabolism and water consumption under daytime warming might lead to rapid exhaustion of C substrates, thus inducing an inhibition of microbial activity. Given that soil fungi, especially the saprotrophic lineages, are important decomposers of soil organic matter [30], the decreased water can significantly suppress the abundance of the fungal community. Second, because of the key role of plant growth in triggering microbial variations, the competition between root absorption and microbial activity for nutrients in soils may limit the growth of fungi which usually colonize plant rhizospheres [8,31]. Nighttime warming-induced overcompensation of plant photosynthesis also improves the substrates' utilization of plant roots in the daytime and further reduces fungal abundance [8,15,19,20].
The stronger impacts of warming on fungi than bacteria have been widely reported [32,33]. One possibility is that most fungi have higher C assimilation efficiency and metabolic ability than bacteria, hence the warming-induced decrease in soil water availability could significantly inhibit the growth and activity of fungi, showing greater suppression on the biomass and diversity of fungi, especially in our experimental grassland which is largely water limited [23]. A reduction in plant litter residues with higher cellulose contents under warming can also lead to a decline in saprotrophic fungal abundance [34,35]. In addition, although AMF is important in facilitating plant nutrient absorption and usually shows sensitive responses to warming [36], the unchanged AMF abundance observed in our study might contribute to their different temperature optima [37] and the direct nutrient acquisition process because they are obligate symbionts that can receive growth substrates from host plants.

Implications
Climate warming can result in soil C loss (release more CO 2 to the atmosphere) by stimulating microbial respiration and improving microbe-mediated mineralization of soil C stocks [35,38]. As the important component of soil respiration, microbial respiration is always thought to be closely associated with their biomass in most studies. However, the insignificant correlation between soil respiration and microbial biomass C, and the negative dependence of soil respiration on the fungi-to-bacteria ratio observed in our study suggest that microbial composition may be more important than biomass in regulating soil respiration under long-term warming (Figure 7). These findings indicate that changes in microbial composition in response to climate warming are critical for ecosystem C cycling in the semi-arid grassland and can provide new insights to predict the climate-C feedback in terrestrial ecosystems.

Conclusions
Climate warming has both direct and indirect effects on soil microbial composition with implications for soil respiration in the temperate steppe. Our study shows that longterm rather than short-term warming has significant impacts on soil microbial composition, which can be attributed to the changes in soil microclimate and grass cover. The decreases in fungal abundance and the fungi-to-bacteria ratio under long-term daytime warming instead of nighttime warming observed in this study are also mainly driven by the increased soil temperature, reduced soil moisture, and improved grass cover. Our findings indicate that microbial composition may be more important than microbial biomass in regulating soil respiration under long-term climate warming, which will facilitate an accurate assessment of climate warming-C feedback in terrestrial ecosystems.
Supplementary Materials: The following supporting information can be downloaded at: www.mdpi.com/xxx/s1, Figure S1: Warming-induced changes in soil temperature from 0:00-23:00 on each day (a) and from 1 May to 31 October during the growing season in each year (b) at a depth of 10 cm. Mean soil temperature (M ± 1 SE, n = 4) during short term, long term, and all 5 years (c). C, control; D, daytime warming; N, nighttime warming; and W, diurnal warming; Figure S2 2012, and 2015). The solid lines and shaded areas describe linear regressions and 95% CIs; Figure S8: Dependences of warming-induced changes in the relative abundance of fungi (a-e) and

Conclusions
Climate warming has both direct and indirect effects on soil microbial composition with implications for soil respiration in the temperate steppe. Our study shows that longterm rather than short-term warming has significant impacts on soil microbial composition, which can be attributed to the changes in soil microclimate and grass cover. The decreases in fungal abundance and the fungi-to-bacteria ratio under long-term daytime warming instead of nighttime warming observed in this study are also mainly driven by the increased soil temperature, reduced soil moisture, and improved grass cover. Our findings indicate that microbial composition may be more important than microbial biomass in regulating soil respiration under long-term climate warming, which will facilitate an accurate assessment of climate warming-C feedback in terrestrial ecosystems.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/biology12050699/s1, Figure S1: Warming-induced changes in soil temperature from 0:00-23:00 on each day (a) and from 1 May to 31 October during the growing season in each year (b) at a depth of 10 cm. Mean soil temperature (M ± 1 SE, n = 4) during short term, long term, and all 5 years (c). C, control; D, daytime warming; N, nighttime warming; and W, diurnal warming; Figure S2 Figure S4: Dependences of the relative abundance of Gram-positive (G+; a-d) and Gram-negative (G−; e-h) bacteria on soil moisture (a,e) and plant cover of grasses (b,f), forbs (c,g), and legumes (d,h). Each data point represents the mean annual value in each plot. The solid lines and shaded areas describe linear regressions and 95% CIs; Figure S5: Dependences of the relative abundance of fungi on bacteria (a), G+ bacteria (b), G− bacteria (c), and G−/G+ ratio (d). Each data point represents the mean annual value in each plot. The solid lines and shaded areas describe linear regressions and 95% CIs; Figure S6: Dependences of the relative abundance of fungi (a,b) and the ratio of fungi to bacteria (c,d) on soil temperature (a,c) and moisture (b,d). Each data point represents the mean annual value in each plot in the short (2006 and 2007) and long term (2011, 2012, and 2015). The solid lines and shaded areas describe linear regressions and 95% CIs; Figure S7: Dependences of grass (a,b), forb (c,d), and legume cover (e,f) on soil temperature (a,c,e) and moisture (b,d,f). Each data point represents the mean annual value in each plot in the short (2006 and 2007) and long term (2011, 2012, and 2015). The solid lines and shaded areas describe linear regressions and 95% CIs; Figure S8: Dependences of warming-induced changes in the relative abundance of fungi (a-e) and the ratio of fungi to bacteria (f-j) on warming-induced changes in grass cover (a,f), forb cover (b,g), legume cover (c,h), soil temperature (d,i), and moisture (e,j). Each data point represents the mean annual value in each plot in the short (2006 and 2007) and long term (2011, 2012, and 2015). The solid lines and shaded areas describe linear regressions and 95% CIs; Table S1: Results (

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available upon request from the corresponding author.