Linear responses of soil properties along moisture gradients
This study revealed significantly linear responses of soil properties among ecosystems from moisture-limited desert grassland gradually in transit to shrub steppe and forest, followed by water-saturated marsh (Fig. 2). Firstly, total carbon (TC) and total nitrogen (TN) levels raised in habitats with increasing soil moisture, which was similar to the results of studies conducted in habitats of forest ecosystems [87] as well as alpine meadow, marsh meadow, and marsh [88]. Similarly, we showed that for NH4+ and NO3− contents, the lowest levels were found in DG, while the highest levels were observed in MA. This pattern was attributed to stronger soil N mineralization but lower nitrification rates in MA, which converted organic N to NH4+, and this conversion was positively associated with soil moisture contents [89]. With increasing TC, and TN [90], elevated levels of NH4+ could provide substrates for microbial processes [91], which play important roles for plant growth. On the contrary, soil pH typically exhibits an opposite response to water availability [92] regulated by calcite (CaCO3) and gibbsite (Al(OH)3) equilibrium [93]. Specifically, when precipitation exceeded evapotranspiration, water leaching removed Ca2+ while retaining relatively immobile Al3+, leading to soil acidification. Conversely, at low leaching rates, Ca2+ accumulated and precipitated with dissolved CO2 as CaCO3, which consumed protons and made the soil alkaline. Thus, differences in soil moisture gradients among ecosystems result in linear responses of soil conditions and nutrient levels that affect microbial growth, resulting in distinct microbial community and potential functional diversity.
Linear responses of microbial characteristics along moisture gradients
Revealing the distribution of microbes in different habitats along water availability gradients was essential to enhance our understanding of the mechanisms on how soil microbiomes respond to climate and environmental changes [94], and their contribution to critical ecological processes [95]. The characteristics of microbial communities showing linear responses across ecosystems along water gradients in the current study includes microbial abundances (Fig. 2), alpha- and β-diversities, and major compositions (Fig. 3). More importantly, these soil microbial characteristics in the Altai region were associated with soil water content, total C and N, and pH (Fig. 3b), which were found to be the main factors influencing microbial distribution from both micropore to continental scales [96].
All the tested microbial abundances, including prokaryotes, fungi, and protists, showed increasing trends across ecosystems with moisture gradients (Fig. 2), which were consistent with the prior research across ecosystems such as desert, meadow, forest, and cropland [97]. The observed pattern was likely attributable to a combination of soil moisture with C and nutrient availability [98]. Specifically, soil moisture has a significant impact on microbial activity [99, 100], while C and nutrients in soils served as key energy sources for microbial growths and metabolisms [101]. Therefore, the rise of these resources ultimately led to an increase in microbial abundance.
At the continental scale, soil pH has been identified as the primary predictor of bacterial richness, of which the relationship has been found to be unimodal, with a plateau observed in soils with near-neutral pH [102]. However, our study reveals a negative correlation between prokaryotic richness and soil pH (Fig. 2, Fig. 3a), which might be due to the higher soil moisture, C and nutrient availability in FO and MA compared to DG [103], contributing to high prokaryotic richness in FO and MA. Similar findings has been reported in the Zoige peatlands of the Qinghai-Tibetan Plateau, that prokaryotic richness decreased with the decline of water table significantly associated with reducing soil TC and TN [104]. In research across vegetation types, a higher richness of prokaryotes was observed in flooded swamps compared to bare sandy land [105]. Moreover, our research showed that there was significant difference in the richness of fungi and protists among ecosystems, roughly corresponding to those reported in a national-scale metabarcoding analysis [106], indicating that fungal and protistan richness was driven by soil properties of soil pH, C:N ratios, and organic matter, potentially demonstrating the important roles of soil nutrient availability on microbial diversities. Unlike the tendency of microbial richness, there was no notable distinction in Shannon index for either prokaryotes or fungi. A similar result was reached by the study reporting that drying reduced the richness of prokaryotes and fungi, but had no significant effect on their evenness [107]. Since richness was the species numbers, while Shannon index represent both richness and evenness [108], one possible explanation was the presence of unique and rare populations characteristic of certain environment among ecosystems in Altai region, which simply affects the richness but not evenness. However, for protists, the Shannon index was notably lower in DG than others, potentially due to the strong impact of drought stress on survival of major protistan groups [109]. Because protists are similar to aquatic organisms requiring sufficient water to move, feed, and reproduce [110], which were regarded as biomarkers for drought along with Gram-positive bacteria [111]. Thus, we suggested that microbial diversity would be affected by soil moisture gradients, in which protists were overall more sensitive to soil water availability than fungi and prokaryotes.
For β-diversity, microbial communities of ecosystems along water gradients were distributed linearly (Figs. 3b, 5a). There was growing evidence that diverse ecosystems hold their unique microbial populations. For example, a field survey in the alpine permafrost regions of the Qinghai-Tibet Plateau has shown that an increase in microbial β-diversity was observed positively correlated with vegetation biomass differences [112]. Additionally, it has been found that the ecosystem types of grasslands, shrubland, and woodlands exhibit unique fungi communities in Australian forests [113]. A global scale survey has revealed that the significant difference in microbial β-diversity existed among tropical forests, prairies, and cold deserts, especially between the desert and non-desert soils [114]. These apparent differences in the distribution patterns of soil microorganisms are mainly driven by environmental factors, in particular soil pH and precipitation, the key predictors of soil bacterial, fungal, and protist community diversities [115]. Besides, soil organic C [116] and vegetation types [117, 118] also played an important role in microbial community construction [110]. In our research, the db-RDA and DistLM analysis identified TC, NH4+, and pH as the key drivers of below-ground microbial β-diversity of the ecosystems in Altai region. Similar results were found in the deserts and steppes, in which soil moisture and pH [105], soil C and N [119], and NH4+ contents [120] explained the most microbial variance among ecosystems along water gradients. In our result, soil properties such as TC, NH4+, and pH, shaped the unique microbial communities of ecosystems in Altai region.
Major microbial community compositions, including prokaryotes, fungi, and protists at the phylum levels, varied significantly among ecosystems along water gradients (Fig. 3c). Soil environment with limited amounts of water and available nutrients, like DG and SS, favors oligotrophic microorganisms while copiotrophs mainly predominate in soils of sufficient water and nutrients, like FO and MA [121]. As the most dominant phyla, Proteobacteria was significantly higher in FO and MA than DG and SS, implying these groups had relatively faster growth rates with the capability to use various substrates [122, 123]. Thus, our results suggested that significant increase in the relative abundance of Alphaproteobacteria, including Bradyrhizobium and Rhodomicrobium [124], and Deltaproteobacteria, including Haliangium and Geobacter, with soil water gradients could be attributed to copiotrophs (Fig. S1). Similarly, Acidobacteria was regarded as strongly predicted by the C mineralization rates, which was an index of organic C availability to microorganisms [125]. We showed that relative abundance of Acidobacteria, in particular Candidatus Koribacter, Koribacteraceae, Bryobacter, and RB41 at the genus level, was significantly greater in FO and MA than DG and SS (Fig. 3c, Fig. 5b), which can be explained by the lower soil pH but more sufficient substrate in FO and MA [126].
On the contrary, we found that Actinobacteria, an oligotrophic bacteria [127], especially Nakamurella, Acidothermus, Knoellia, Microlunatus, Gaiella) at the genus levels, dominated in DG and SS than MA. Since Actinobacteria was found to be positively correlated with pH [126] higher nutrient availability but more acidic soil conditions in FO and MA partially explained their lower relative abundance. Surprisingly, as copiotrophic bacteria, we showed that Bacteroidetes enriched in DG and SS had relatively lower abundance in MA (Fig. 5b), with opposite trends to Proteobacteria in responses to ecosystems of moisture gradients. Thus, our results suggest that it was insufficient to explain the linear responses of microbial major compositions to water gradients simply based on the oligotrophic-copiotrophic theory.
The dynamics of major fungal phyla among ecosystems were perhaps due to habitat-related functional properties, such as their wind-dispersed spores [113]. Ascomycota was the most abundant fungi in various ecosystems, such as temperate grasslands [128], global drylands [129], and soil worldwide [130], implying their extensive capability to use resources and withstand environmental pressure. Unlike Ascomycota that were relatively steady among ecosystems along water gradients, Basidiomycota showed a clear preference for humid soil environments [113]. It has been shown that soil nutrient availability [131], especially NH4+ levels [132] was positively related to increased biomass of Basidiomycota. Thus, our study showed that Basidiomycota and their genus, Solicoccozyma, also exhibited higher relative abundance in FO and MA with sufficient soil water and nutrient availability (Fig. 2, Fig. 3c). Additionally, our results found that relative abundances of Mortierellomycota and Rozellomycota were relatively higher in FO and MA, because it has been reported that both Mortierellomycota [133] and Rozellomycota [134] responded positively to soil nutrients availability.
As the most abundant phylum within protists, Cercozoa played a critical role in the soil microbial food webs, material circulation, energy flows, and information transfers [135]. In our study, we observed a significant decrease in relative abundance of Cercozoa along soil water gradients (Fig. 3c, P < 0.05), in which their genera, Cercomonas, Neocercomonas, Paracercomonas, and Spongomonas, were more abundant in DG and SS compared to FO and MA (Fig. S3, P < 0.05). Unsurprisingly, Cercozoa and Ciliophora were the most abundant protists lineages in the four ecosystems and together accounted for 73% of all protists in a study spanning 180 sites across six continents [110]. As Cercozoa usually tended to be more abundant in arid soils [136], its decline among the ecosystems along an ascending moisture gradient in our results was in line with expectations. Conversely, Evosea (its genus-Schizoplasmodiopsis) [137] and Endomyxa (its genera-Leptophrys and Vampyrella) [138] as heterotrophic protists, indicated the opposite pattern. In summary, different prokaryotes, fungi, and protists at the phylum level responded differently to soil properties and showed disparity trends in ecosystems along the moisture gradient.
Microbial metagenomic functions and metabolomics revealed a strong functional distinction among ecosystems
The β-diversity pattern of metagenomic function (Fig. 5a) was nearly identical to the patterns of microbial taxonomy showing linear responses across the ecosystems with soil moisture gradients (Fig. 3b, Fig. 5a), significantly associated with soil properties of TC, NH4+, and pH. Previous studies showed that broad metabolic functions were enriched in ecosystems with low microbial diversity [114], whereas microbial specialized functions showed a higher level relative abundance in ecosystems with high microbial diversity [139, 140]. Our results showed that metagenomic genes related to broad metabolic functions acquired by all microbes were more abundant in DG and SS than in FO and MA. These functions included carbohydrates metabolism, clustering-based subsystems, cofactors, vitamins, prosthetic groups, pigments, and DNA metabolism, etc., which supported microbial cell survival and replication [141]. However, the genes related to specialized substrate and nutrient cycling, such as the metabolisms of nitrogen, potassium, and aromatic compounds showing higher relative abundances in resource-rich ecosystems of FO and MA (Fig. 5b). At the meantime, certain metagenomic functional categories seem to be directly linked to the gradients of soil moisture and nutrient conditions across ecosystems. For instance, the functional genes related to carbohydrates metabolisms could improve microbial survivals in nutrient-limiting conditions, such as DG and SS [142]. The relatively lower relative abundance of carbohydrates metabolisms in MA was probably due to the anaerobic conditions in MA, where microbes metabolized C more slowly than in aerobic conditions [127], leading to soil organic C accumulation [143, 144]. Amino acids, acting as xeroprotectants, were produced by desiccation-tolerant microorganisms for protection of cells from drought stress [145], and thus we showed that drought stress conditions led to an increased demand for genes related to the metabolisms of Amino Acids and Derivatives in DG and SS. Besides, it has been observed that a direct and significant positive effect of soil TC on microbial respiration [146], suggesting that soil respiration was directly regulated by resource availability [147]. Supporting these studies, our results further evidenced that both soil TC and respiration-related genes were significantly higher in MA (Fig. 2, Fig. 5). Additionally, we observed greater relative abundance of genes involved in RNA metabolism in MA, implying the potential for rapid growth of soil microbes through a more aggressive expression of the overall genetic information encoded in DNA [148]. This mechanism was one of the life strategies of copiotrophs (r-strategists), which tended to become dominant in the resource-rich environments, such as FO and MA. Furthermore, we showed that metagenomic genes involved in the metabolism of aromatic compounds were more abundant in FO and MA, which were likely related to a dense vegetation cover [149], as plants typically represent major sources of aromatics to soil organisms [150]. In conclusion, unique soil physicochemical conditions and microbial characteristics could influence microbial metagenomic functions related to broad metabolisms and specialized processes divergently across ecosystems with soil water gradients.
This phenomenon was further supported by the determination of functional genes at a higher resolution (Fig. 6), showing that within functional categories related to C and nutrient cycling, DG had higher relative abundance in carbohydrates and iron metabolisms, including CO2 fixation, one-carbon metabolism, and di- and oligosaccharides, iron acquisition in Vibrio, Campylobacter iron metabolism, etc. As Fe-minerals can be used as a source of energy for microorganisms [151, 152], higher relative abundance of genes related to iron acquisition and metabolism in DG possibly reflected that iron was the main driver for microbial growth in nutrient-poor environments. On the contrary, we observed that N metabolism-related functional genes were the most abundant in MA, including N fixation, ammonia assimilation, and denitrification, etc. The enrichment of prokaryotic phyla, such as Proteobacteria and Acidobacteria, in MA (Fig. 3) likely led to higher abundance of functional genes linked to N metabolism (Fig. 6), because several species in Proteobacteria were associated with N cycles, such as ammonia oxidizers, rhizobia [139], and burkholderia [153], and Acidobacteria were regarded as important players involved in nitrate and nitrite reduction as well as plant cell wall polysaccharide degradation [154]. Therefore, the linear responses of microbial diversity and composition across ecosystems along water gradients could lead to diverging of soil metagenomic functions related to C and nutrient metabolisms.
We also found that the relative abundance of genes associated with microbial biological interaction and environmental information processing was greater in FO and MA than in DG (Fig. 5b). Metagenomic genes related to Virulence, disease and defense (VDD) increased by nearly 90% as the ecosystems in transition from nutrient-poor deserts to eutrophic forests and marsh (Fig. 5b). Another study has also shown that VDD increased during natural vegetation restoration [155]. Specifically, the functional categories within VDD, such as resistance to antibiotics and toxic compounds [156], bacteriocins, ribosomally synthesized antibacterial peptides [157], toxins and superantigens [155], were often associated with microbial competition. Thus, our study further suggests that an increase in the relative abundance of VDD is indicative of the enhancement of microbial interactions during ecosystem transition from moisture-limited DG to water-saturated MA. Additionally, we showed that the relative abundance of genes related to cell wall and capsule was greater in FO and MA than in DG (Fig. 5b), within which murein hydrolases, capable of cleaving covalent bonds in bacterial peptidoglycan, were considered relevant to bacterial cell lysis and fratricide occurring in bacterial colonization and population [158]. Besides, in our study, FO and MA had greater relative abundances of genes involved in membrane transport (Fig. 5b), including ABC transporters, transporters in the sugar phosphotransferase systems, and transporters in the protein secretion system. It has been found that these genes were related to soil conditions of high moisture levels [159], where sufficient soil water availability can promote the liquid diffusion rates providing vital substrates for soil microbes [160]. These genes related to membrane transporters could import or export a broad range of compounds and further facilitate copiotrophic microbes to uptake readily available extracellular compounds from the soil [122]. Besides, we also showed that the enrichment of genes encoding the two-component system in FO and MA, belonging to regulation and cell signaling function, might facilitate copiotrophs to sense and respond to changes in environmental conditions [161]. These observations validate the hypothesis that when microbial proliferation was limited by water and resources, such as in DG and SS, the adaptation process to adverse environmental conditions for microorganisms may prioritize resistance and resilience over microbial competition [114], but the intensity of competition and the capacity for environmental information processing gained growing significance when soil water and resources are sufficient in ecosystems, like FO and MA.
Marsh had a larger and more complex range of metabolites than other ecosystems
Combined with metagenomic data, soil metabolomic profiling could provide a more holistic understanding of potential shifts in ecosystem functioning [162] and microbial processes in response to environmental variations [163]. Our study observed a linear response of the metabolomics (Fig. 7a) across ecosystems with soil water gradients, which were also largely explained by TC, NH4+, and pH.
The enrichment of nutrients such as C, N, and P usually caused the accumulation of small organic molecules, i.e., metabolomic compositions [164], most of which were accumulated in resource-riched MA (Fig. 2), especially the lipids and lipid-like molecules and organic oxygen compounds (Fig. 7b). Lipids as cell membrane elements play a crucial role in regulating membrane fluidity [165], so microorganisms could alter their membrane lipid quantity and composition in response to environmental changes [166]. Moreover, lipids and lipid-like molecules have been implicated in membrane transport [167] and cell signaling [168]. Notably, both the metagenomic functional profiles (Fig. 5b) and molecular components (Fig. 7a) exhibited a consistent pattern of linear responses across ecosystems in transition from DG and SS to FO and MA. Specifically, the enrichment of lipids in MA, such as erucic acid, kojibiose, and petroselinic acid, was likely due to the acidic environment and thicker O horizon of the marsh [169]. As the most chemically reduced constituents of soil organic C, lipids required greater activation energy for oxidation than other compounds [170], so anaerobic conditions in the marsh might further promote lipid accumulation. We also found significantly higher relative abundance of organic oxygen compounds in MA (Fig. 7b), mainly Maltotetraose, Sucrose, Trehalose, and Quinone (Fig. 7c). These soil metabolites in MA, including maltotetraose [171], sucrose [172], and trehalose [173], have been found to reflect the complexity of metabolism for microbial growth, whereas quinone, the redox signal of the bacterial two-component system [174], was an important component of membrane-bound electron transport systems [175]. Thus, our findings suggest that higher abundance of these metabolites determined in MA might be related to the greater demands of soil microbes for regulation and cell signaling functions.
In our study, there were a total of 9 metabolites enriched in DG, three of which belonged to lipids and lipid-like molecules, including 28-Norcyclomusalenone, Epoxyganoderiol A, and Methyl (2E,6Z)-dodecadienoate. However, they generally differed in their biological roles. For example, Methyl (2E,6Z)-dodecadienoate could provide nutrients to organisms [176]; Epoxyganoderiol A acts as a membrane stabilizer while also providing nutrients to microbes [177]; 28-Norcyclomusalenone was regarded as a signaling molecule as well as a type of nutrient and membrane stabilizer [178]. Besides, N’-Hydroxymethylnorcotinine, a primary in vitro metabolite of cotinine [179], was reported in the process of biotransformation of nicotine in mammalian systems [180, 181]. Another study on mammalian lignan metabolism discovered a major peak within metabolomics corresponding to 5-Hydroxyenterolactone [182]. For the first time, our study further revealed that these two metabolites, N’-Hydroxymethylnorcotinine and 5-Hydroxyenterolactone, were enriched in the desert with limited water and nutrients. Interestingly, Phaseollidin was a type of metabolite selectively toxic to gram-positive bacteria [183] that often survived better than gram-negative bacteria in water-deficient and nutrient-poor deserts due to their thicker peptidoglycan cell wall layer [36, 184]. Moreover, Lithocholic acid glycine conjugate also had potent toxic properties, such as membrane disruption [185], so we suggested that both of them have an inhibitory effect on the growth of certain microorganisms due to their dominance in DG affected by water and nutrient stress. In conclusion, our metabolomic analyses yield valuable insights into our knowledge of the adaptive mechanisms of microorganisms within the soil environment of limited versus sufficient moisture and resources, suggesting that sufficient water and substrates greatly promote microbial metabolism and intercommunication.