4. DISCUSSION

In this study, we discovered that distinct variation trends of soil microbiomes during desertification regulated the changes in multifunctionality in vertical soil profiles. Notably, particular microbial taxa rather than microbial diversity better predicted the vertical profile variation in multifunctionality in desert ecosystems. Our results highlight the significance of deep soil microbiomes for buffering and regulating the multifunctionality of desert ecosystems.

4.1 Variation in multifunctionality in desert ecosystems

Deep soil multifunctionality was as important as superficial soil multifunctionality in the desert ecosystems (Fig. 1). The results of this study confirmed our first hypothesis. Desertification can lead to the loss of surface soil C, N and P and fine soil particles and decreases in the water-holding capacity and vegetation cover (D’Odorico et al., 2013; Ravi et al., 2010; Ward et al., 2018). Soil pores and water infiltration in superficial soil exhibited continuous increases due to the gradual loss of fine soil particles during desertification (Allington & Valone, 2010; D’Odorico et al., 2007), leading to enhanced nutrient accumulation in deep soils from plant litter and superficial soil nutrients. Furthermore, microbiomes play a vital role in regulating soil multifunctionality by supporting functional processes (i.e., soil nutrient cycling, litter decomposition, and N mineralization), which allow the transfer of materials and carbon energy between above- and belowground communities (Falkowski et al., 2008; Tedersoo et al., 2014). Thus, the simultaneous changes in and coupling of soil physical properties and vegetation and microbiomes jointly contributed to the equivalent multifunctionalities of superficial and deep soils (D’Odorico et al., 2019; Jiao et al., 2018; Ravi et al., 2010). In addition, soil nutrients are mainly derived from the decomposition of plant litter and root in both undisturbed and restored ecosystems (Barber et al., 2017; Lozano et al., 2014). In this study, the plant litter and root biomass gradually decreased during desertification (Figure S8), leading to reduced soil nutrient accumulation from litter and root decomposition, which was largely responsible for the significant decrease in multifunctionality in the vertical soil profiles along the desertification gradient (Fig. 1c).

4.2 Vertical variation in soil microbiomes in desert ecosystems

Our results showed that the desertification process drove distinct variation trends of the microbiomes in the vertical soil profiles. Microbial survival and growth may be severely limited by continuous abiotic stressors (i.e., limited bioavailability of water and C substrates), frequent disturbances (i.e., drying-rewetting events), and heterogeneous distributions of substrates across soil profiles (Fierer, 2017). In desert ecosystems, all these abiotic stressors often synchronously appear during desertification, which is characterized by decreasing plant biomass and fluctuating soil nutrients and soil structure, leading to enhanced environmental stress gradients that account for the distinct variation in microbiomes in vertical soil profiles (D’Odorico et al., 2013; D’Odorico et al., 2019; Neilson et al., 2017). Furthermore, the phyla Actinobacteria, Proteobacteria, Chloroflexi, Acidobacteria, Ascomycota, Basidiomycota, Thaumarchaeota and Euryarchaeota were the most abundant microbial taxa with distinct responses in the desert ecosystems. These phylum-level profiles were similar to those in other soils and environments (Jiao et al., 2018; Li et al., 2014; Tedersoo et al., 2014; Upton et al., 2020). We further found that bacteria are on average more resilient in the face of disturbances and perturbations because of their relatively fast intrinsic growth rates (Wardle, 2013), suggesting that they are more sensitive to the environmental filtering driven by desertification. As soil depth increased, the bacterial phyla Acidobacteria, Actinobacteria, and Chloroflexi typically declined, while Proteobacteria significantly increased (Figs. S4 and S5). The majority of Acidobacteria, Actinobacteria, Proteobacteria, and Chloroflexi have been suggested to be closely associated with organic substrates (Goldfarb et al., 2011). Except for the response of Proteobacteria, these observed changes were further confirmed by the findings of other studies (Jiao et al., 2018; Li et al., 2014), suggesting that soil pH and nutrient bioavailability are more likely to be the reasons for the decreased relative abundance in soil profiles. In this study, Alpha-, Delta-, and Gammaproteobacteria belonging to the class Proteobacteria were examined (Figure S2). Alpha- and Deltaproteobacteria have been suggested to be negatively associated with increased organic substrates, while Gammaproteobacteria are positively associated with increased organic substrates (Goldfarb et al., 2011). This contrasting pattern could reflect divergent ecological niches and microbial synergism, which are more likely to be the reasons for the enhanced abundance of Proteobacteria in the vertical soil profiles, as reported by Li et al. (2014) in farmland ecosystems. In addition, the most abundant fungal phyla (Ascomycota and Basidiomycota) were less mobile in vertical soil profiles than bacterial and archaeal phyla in the desert ecosystems, similar to previous findings (Tedersoo et al., 2014), suggesting that the richness of fungi and functional groups is not associated with plant productivity and that the plant-soil feedback loop does not typically reshape fungal diversity in different ecosystems. The archaeal phyla Thaumarchaeota and Euryarchaeota mainly drive soil N cycling (Haroon et al., 2013; Leininger et al., 2006). Thus, the gradual decrease in N substrate in vertical soil profiles is more likely to be the reason for the decreased relative abundance of Thaumarchaeota and Euryarchaeota.
Alpha-diversity can characterize the number of microbial taxa within sample sites, while beta diversity can describe the variation trend of microbial composition across sample sites (de Carvalho et al., 2016; Legendre & De Cáceres, 2013). Previous work has suggested that the alpha-diversity (i.e., Shannon and OTU richness indexes) of bacteria and fungi decreases while archaeal diversity typically increases with increasing soil depth in different systems (i.e., grassland, forest and farmland) (Eilers et al., 2012; Jiao et al., 2018). Our results further indicated that bacterial diversity decreased, archaeal diversity increased, and fungal diversity fluctuated with increasing soil depth along a desertification gradient (Fig. 2). These discordant patterns of soil microbiomes were due to their distinct ecological niches and differences in oxygen tolerance, showing that bacteria and fungi are mainly aerobic, while archaea are mainly anaerobic (Haroon et al., 2013; Upton et al., 2020). The oxygen content gradually decreased with soil depth in the desert ecosystems, given the increasing water content with increasing depth resulting from high soil water infiltration (D’Odorico et al., 2007). In addition, ecological restoration characterized by increased availability of soil C and N significantly enhanced the alpha- and beta-diversity of bacteria and archaea but not fungi (Barber et al., 2017; Jiao et al., 2018; Lozano et al., 2014). In contrast to ecological restoration, desertification is characterized by decreasing soil C and N, which may be the reason for the decreased alpha- and beta-diversity of bacteria and archaea in the vertical soil profiles as desertification proceeded. In this study, the indistinctive alpha- and beta-diversity of fungi suggested the stable performance of fungal communities in desert ecosystems (Figs. 2 and 3). Fungi are heterotrophic microorganisms that play fundamental ecological roles as decomposers, such as the decomposition of litter and senescence or death of roots (Tedersoo et al., 2014). In this study, the litter and root supplies decreased as desertification progressed, and this unfavourable and variable habitat was ineffective in completely restraining the growth and enrichment of fungi. The fungal kingdom contains a large proportion of various niche strategies ranging from saprotrophy through mutualism to parasitism across trophic levels (Nilsson et al., 2019). In addition, fungal communities with filamentous growth may show different interactions because of dispersal limitation and greater tolerance of desiccation (Austin et al., 2004; Fukami et al., 2010; Powell et al., 2015), leading to their co-enrichment and distinct vertical distributions.

4.3 Main predictors of ecosystem multifunctionality in desert ecosystems

Our results indicated that particular microbial phyla rather than total microbial diversity better predicted and explained the vertical profile variation in soil multifunctionality in desert ecosystems. The results of this study confirmed our second hypothesis. Experiments at the microcosm and global scales showed that microbial diversity variables (i.e., Shannon and phylogenetic diversity indexes and OTU richness) are important predictors of multifunctionality and are positively linked to superficial soil multifunctionality (Delgado-Baquerizo et al., 2016; Li et al., 2019; Wagg et al., 2014; Zheng et al., 2019), suggesting that microbial communities with higher richness perform better under varying conditions and better protect against the loss of taxa. However, our knowledge is largely based on microbial diversity and dominance in superficial soil, and less attention has been paid to deep soils of desert ecosystems. Our results suggested that individual bacterial and archaeal species are more important predictors of multifunctionality in desert soils, especially in deep soils (20-100 cm). These individual bacterial and archaeal species in deep soils may play a leading role in driving soil multifunctionality, which to some extent could explain why the process of desertification significantly decreased soil multifunctionality (Fig. 1) and dominant bacterial and fungal phyla maintained synchronous positive feedbacks (Figures. S3, S4, and S5). In contrast to those of bacterial and archaeal taxa, the links between individual species of fungi and ecosystem function are dependent on the presence of other species and a result of multiple interactions (i.e., positive and negative as well as direct and indirect) between the various species that as a whole regulate potential ecosystem functions (Tedersoo et al., 2014; Wagg et al., 2019).
Intriguingly, our results showed a disproportionate role of individual species (i.e., Acidobacteria or their strategic alliances) in multifunctionality, which seems counter-intuitive given their higher than expected ecological importance in soil microbial communities (Figs. 4 and 5). Acidobacteria in soil habitats are considered ubiquitous and physiologically active but are rarely cultured and consequently remain a poorly studied phylum (Goldfarb et al., 2011; Naether et al., 2012). The phylogenetic diversity and relative abundance of Acidobacteria in diverse habitats have suggested their vital role in driving biogeochemical processes and diverse metabolic functions (Naether et al., 2012). High C bioavailability is negatively associated with acidobacterial abundance in various soils (Fierer et al., 2007; Goldfarb et al., 2011), suggesting that Acidobacteria are adapted to habitats with poor substrates and often slow-growing oligotrophs. Indeed, the acidobacterial community can be energetically adapted to C-limited soils and may be predominant in oligotrophic habitats, where decreasing plant biomass results in a decrease in the availability of plant-derived C sources (Castro et al., 2010).
Microbial diversity can maintain and regulate multifunctionality in a variety of ways, suggesting that microbial communities with higher total richness perform better in progressively developed and less-stressed soils (i.e., forest, cropland, and wetland soils) (Delgado-Baquerizo et al., 2016; Jiao et al., 2018; Li et al., 2019; Wagg et al., 2014). Conversely, multifunctionality may be controlled by particular microbial taxa (relative abundance of phylotypes) but not the total richness and abundance of microbial communities in dryland soils (Delgado-Baquerizo et al., 2017). Interestingly, our results further support the notion that particular microbial phyla (microbial species index in Figs. 4 and 5) rather than total microbial diversity best predict and explain the vertical profile variation in soil multifunctionality in desert ecosystems.