Introduction

Biological soil crusts (BSCs) are a biological complex of microbiomes (cyanobacteria, heterotrophic bacteria, and fungi), cryptogams (algae, lichens, and mosses), and secretions including exopolysaccharides, proteins, and organic acids (Belnap 2003; Rossi et al. 2012). BSCs, known as ecosystem engineers, have great potential in ecological restoration and desertification control (Bowker et al. 2005, 2011; Zhang et al. 2021). BSCs can improve soil carbon and nitrogen content by fixing atmospheric carbon dioxide and nitrogen (Chamizo et al. 2012; Elbert et al. 2012; Kidron et al. 2015a, b). Furthermore, BSCs play important roles in improving water retention and regulating soil hydrology, especially in arid desert areas (Adessi et al. 2018; Belnap 2006). Finally, BSCs affect seed germination and plant growth (Li et al. 2005; Sedia and Ehrenfeld 2003).

BSCs are typically divided into algal, lichen, and moss crusts according to the composition of organisms (Redfield et al. 2002). Algal BSCs are the early developmental stage of BSCs and create conditions for the formation of lichen and moss BSCs. Cyanobacteria play a major role in the formation of algal BSCs, and are usually the dominant bacteria involved in carbon accumulation and nitrogen fixation (Wang et al. 2016). Polysaccharides that form an organic ‘cement’ with soil particles are produced by cyanobacteria, which improves soil water retention (Adessi et al. 2018). Lichen BSCs are a symbiotic complex of fungi and algae, where the algal cells are surrounded by fungal hyphae. Compared with algal BSCs, lichen BSCs have specific supporting structures such as rhizoids which could be tightly bound to soil particles (Belnap 1995). Microorganisms are important components of BSCs and participate in the formation, succession, and ecological functions (Blay et al. 2017; Liu et al. 2020; Zhao et al. 2020; Zhang et al. 2018). Microbial communities differ among different types of BSCs (Wang et al. 2020; Zhang et al. 2016). Moreover, microbial communities in biocrusts are affected by environmental factors such as soil characteristics, temperature, and precipitation (Abed et al. 2019; Nuttapon et al. 2020; Velasco et al. 2019). However, how climatic, and edaphic factors affect bacterial communities associated with algal and lichen BSCs at geospatial scales remain mostly unclear.

The distance-decay relationship, that describes the decrease in community similarity with an increase in geographic distance (Morlon et al. 2008), is one of the important diversity spatial distribution models and has been widely studied in the field of microbial ecology (Feng et al. 2019; Shi et al. 2018). Mechanisms that reveal variation patterns of microbial communities across spatial scales are the basis for understanding diversity formation and species co-existence (Dini-Andreote et al. 2015; He and Wang 2015). According to the niche theory, deterministic factors, such as environmental factors and interspecific relationships, govern community composition (Vellend 2010). In contrast, the neutral theory argues that community variations are mainly controlled by stochastic processes (Hubbell 2001). Even though both deterministic and stochastic processes govern community assembly, the relative importance of the two processes is environmentally dependent and varies across ecosystems and latitudes (Jiao et al. 2020; Ofiţeru et al. 2010; Shi et al. 2018). For example, long-term fertilization causes a shift from dominant deterministic processes to stochastic processes for soil diazotrophic communities in agroecosystems (Feng et al. 2018). However, the relative importance of deterministic and stochastic processes could be affected by spatial scales (Shi et al. 2018). Stochastic processes determine bacterial community assembly at smaller spatial scales (< 900 km distance between pairs of sites). On the contrary, deterministic processes determine the bacterial community assembly at broader scales (> 900 km).

Most studies on BSCs focus on dryland ecosystems including cold deserts and tropical deserts with extremely low precipitation, while little attention has been paid to BSCs in alpine grassland ecosystem at high altitude (Blay et al. 2017). In the context of global climate change, temperature increases at high latitudes are greater than in other regions (IPCC 2021). The Qinghai-Tibetan Plateau, the highest and largest plateau on Earth, is experiencing rapid warming and increases in precipitation (Shen et al. 2015). Considering its unique geographic location and significant ecological functions, the effects of climate change or human activities on biodiversity, vegetation composition, soil characteristics, and ecosystem functions have been studied in this area (Sun et al. 2020; Zhang et al. 2019, 2020), leaving BSCs mostly unexplored. In this study, we sampled algal BSCs and lichen BSCs from nine regions that varied in altitude, annual temperature, and precipitation on the Qinghai-Tibetan Plateau. The purpose of this study was to analyze spatial variation patterns of the bacterial communities in BSCs, and clarify driving factors and ecological processes that govern bacterial community turnover. Lichen BSCs are more mature and complex in composition than algal BSCs. As such, we hypothesized that: 1) The bacterial diversity of lichen BSCs is generally higher than that of algal BSCs. 2) The spatial distribution patterns and bacterial community assembly processes differed between algal and lichen BSCs. 3) The climate was a key factor influencing spatial distribution of bacterial communities in both algal and lichen BSCs. To the best of our knowledge, this research is the most extensive investigation on geographical scale of the microbial communities in the BSCs in alpine grassland ecosystem, which enriches our understanding of the driving factors and assembly processes of bacterial communities in the BSCs of alpine environments.

Materials and methods

Study area

Nine study areas that were characterized as alpine grassland ecosystems were selected for biocrust sampling (Supplemental Fig. S1, Supplemental Table 1). The geographical distance between sampling sites ranged between 93.6–664.9 km. Mean annual temperature (MAT) and mean annual precipitation (MAP) from 2014 to 2019 were obtained from the local meteorological stations. The study areas were experiencing an alpine continental climate with an MAT ranging from −3.12–5.09 °C and MAP ranging from 420 to 880 mm. The highest temperature and rainfall occurred during the growing season (June–September). The MAT and MAP were typically lower and higher, at the southern sites (Maqin, Dari, and Jiuzhi) than at the northern sites, respectively (Gonghe, Gangcha, and Qilian). Common plants in these areas were grasses including Stipa aliena Keng, Elymus nutans, sedges including Kobresia pygmaea, and forbs including Potentilla chinensis, and Leontopodium nanum. BSCs colonized plant interspaces and covered 5–15% of the grasslands.

Biocrust sampling

Two types of BSCs with different morphological characteristics and successional stages (algal BSCs with dark color and lichen BSCs with light color) were collected in August 2019 (Fig. 1a). The dark algal BSCs represented the early developmental stage, while the light lichen BSCs represented the late developmental stage. A 30–50 m2 grassland area was investigated, and regions containing both algal BSCs and lichen BSCs were selected. Soil blocks including BSCs and attached subsoils (> 5 cm depth) were collected with a shovel. Each soil block represented a single field replicate, and three replicates were obtained for each sampling location. Biocrusts can be highly heterogeneous in the field, and only BSCs with single type (> 95%) were sampled. Areas with growing plants in BSCs were avoided when sampling. The top 1-cm biocrust layer was scraped off with a sterilized blade and placed in sterile valve bags. These samples were transported in a cooler with dry ice and stored at −80 °C for 16S rRNA gene sequencing. Attached subsoils with depth of 5 cm were collected and stored in paper bags, before being air dried for determination of physicochemical properties.

Fig.1
figure 1

(a) Morphological characteristics of algal BSCs (dark color) and lichen BSCs (light color). (b) The Shannon diversity of bacterial community in algal BSCs and lichen BSCs across sampling sites (n = 3, SE = standard error). P < 0.05 for algal and lichen BSCs indicates significant effects of sampling locations on Shannon diversity. * indicates significant differences in Shannon diversity between algal and lichen BSCs according to the T test for each sampling site (P < 0.05). (c) Correlations between the difference degree of Shannon diversity between the two BSCs types (CV) and the aridity index. Each point represents one sampling location. (d) Correlations between bacterial diversity and aridity index for algal BSCs and lichen BSCs

Full size image

Three replicates were collected for each biocrust type, resulting in a total of 54 samples (nine regions × two types × three replicates) for 16S rRNA gene sequencing. Soil samples under the biocrusts were screened with a 2 mm sieve for homogenization. Soil pH was measured with a pH probe (Orion Star A215, ThermoFisher Scientific, USA), and the water-to-soil ratio was determined as 5:1. The soil total nitrogen (TN) and total carbon (TC) contents were measured using an elemental analyzer (Elementar, Hanau, Germany). The soil total phosphorus (TP) content was measured using molybdenum antimony colorimetry. Specifically, phosphorus was extracted using HClO4-H2SO4, and colorimetry was performed at a wavelength of 880 nm with molybdenum antimony resistance as an indicator.

DNA extraction, polymerase chain reaction (PCR) amplification, and sequencing

Samples of biocrust layer were used for bacterial community analysis. The sampled BSCs were homogenized with a mill in liquid nitrogen before DNA extraction. Total DNA was isolated from a 0.5 g biocrust sample using OMEGA: E.Z.N.A.® Soil DNA Kit (Omega Biotek, Norcross, GA, USA) following the manufacturer’s instructions. The integrity of the DNA was detected using 1% agarose gel electrophoresis, and NanoDrop One (NanoDrop Technologies, Wilmington, USA) was used for purity examination. Specific primers [(5’-GGACTACHVGGGTWTCTAAT-3′) and (5’-ACTCCTACGGGAGGCAGCA-3′)] were used to amplify the V3–V4 region of the 16S rRNA gene. The PCR procedure was performed as described by Wei et al. 2021. PCR products were detected by 1% agarose gel electrophoresis, mixed in equidensity ratios according to the GeneTools Analysis Software (Version 4.03.05.0, SynGene), and then purified using the EZNA Gel Extraction Kit (Omega, USA). Libraries were constructed according to the procedure of the NEBNext Ultra DNA Library Prep Kit for Illumina® (New England Biolabs, USA). The Qubit@ 2.0 Fluorometer (Thermo Scientific) and Agilent Bioanalyzer 2100 system were used for quality assessment, after which the libraries were sequenced with the Illumina Hiseq 2500 platform.

Sequencing data processing

High-quality paired-end (PE) clean reads were obtained by a Trimmomatic quality-controlled process (Bolger et al. 2014). Subsequently, the reads were merged into one sequence according to the overlapping relationship between PE reads using FLASH software (Chen et al. 2018). The allowed minimum overlap length was 10 base pairs (bp), and the nonconforming sequences with error ratios >0.2 in the overlapping region of the merged sequence were filtered. The spliced sequences were then filtered using the Trimmomatic software. The effective sequences were clustered into operational taxonomic units (OTUs) using Usearch (Edgar 2019), and sequences that shared ≥97% similarity were considered the same OTUs. Chimera sequences and singleton OTUs were removed using the UCHIME de novo algorithm and Usearch (Edgar et al. 2011; Edgar 2019), respectively. Finally, the taxonomic information of the representative out sequences was annotated using the GreenGene database (http://greengenes.secondgenome.com/) based on the Ribosomal Database Project (RDP) classifier algorithm and the assign_taxonomy.py script (http://qiime.org/scripts/assign_taxonomy.html) in QIIME1.

Data analysis

Shannon–Wiener index of bacterial communities in biocrust layers was analyzed using the QIIME (V1.9.1) software (alpha_diversity.py script) (Lozupone and Knight 2005). Dissimilarities in the bacterial community were shown using principal coordinates analysis (PCoA) plots based on the Bray–Curtis distance. Significant community dissimilarities were tested via permutational multivariate analysis of variance (PERMANOVA) using the “Adonis” function in R (3.6.1). Bacterial community assembly processes were quantified by the β-nearest taxon index (βNTI), which was calculated using the “picante” package in R (Shi et al. 2018), whereby a |βNTI| > 2 suggests that community assembly is mainly controlled by deterministic processes, and |βNTI| < 2 suggests that community assembly is predominately determined by stochastic processes (Stegen et al. 2012). The Mantel test was used to analyze correlations between bacterial community dissimilarity and the geographical distance of sampling sites, and between bacterial community composition and environmental factors using the “vegan” package in R. The actual geographic distances between the sample sites were calculated based on the latitudes and longitudes of sampling sites. Spearman’s correlation analysis was conducted to detect relationships between bacterial diversity and aridity index, as well as between relative abundance of bacterial taxa and environmental factors in SPSS 20.0. The De Martonne aridity index was calculated as follows: AI = average precipitation/(average temperature + 10) (Zhou et al. 2020).

Results

Alpha diversity of bacterial communities in the biocrusts

Following final quality filtering, the read numbers were rarefied to the minimum size (7796) to ensure the same sequencing depth, resulting in a total of 11,298 bacterial OTUs. Of these, 4245 OTUs (37.6% of the total OTUs) were specific for the algal BSCs, 1327 (11.7%) OTUs were specific for the lichen BSCs, and 5726 OTUs (50.7%) were shared between both (Supplemental Fig. S2). The Shannon diversity in the algal BSCs was significantly higher than that in the lichen BSCs at all sampling sites except for Gonghe and Qilian (Fig. 1b). Furthermore, the variation degree between the two biocrust types increased as the AI increased (Fig. 1c). Notably, we found that the Shannon index was negatively correlated with the AI and the MAP for the lichen BSCs, while it was positively correlated with the AI for the algal BSCs (Fig. 1d, Supplemental Fig. S4).

Beta diversity and taxonomic composition of the bacterial communities

Results of PERMANOVA suggested that the bacterial community composition in the algal BSCs significantly differed from that in the lichen BSCs (F = 11.9, R2 = 0.186, P = 0.001) (Fig. 2a). Furthermore, there were significant differences in bacterial composition among the sampling sites for the lichen (F = 4.77, R2 = 0.679, P = 0.001) and algal (F = 4.18, R2 = 0.650, P = 0.001) BSCs (Supplemental Fig. S5). The AI of the nine sampling sites ranged from 27 to 71. Additionally, the bacterial community diversity increased for the algal BSCs, but decreased sharply for the lichen BSCs when the AI value exceeded 48 (Fig. 1b). Based on this result, we divided the nine sample sites into low aridity index group (Low AI) and high aridity index group (High AI). According to the PCoA plots, sampling sites with higher AI (Maqin, Maduo, Dari, and Jiuzhi) were grouped together and separated from those with lower AI for both the algal and lichen BSCs (Fig. 2b).

Fig. 2
figure 2

(a) The PCoA (principal co-ordinates analysis) plot showing the differentiation of bacterial community composition between algal and lichen BSCs. (b) The PCoA plot showing the differentiation of bacterial community composition between aridity scales. AI-aridity index. High AI-sampling sites with aridity index above 48, low AI-sampling sites with aridity index lower than 48. (c) The taxonomic composition of bacterial community in biological soil crusts. Each bar represents an average of three replicates. GN-Guinan, MQ-Maqin, DR-Dari, JZ-Jiuzhi, MD-Maduo, YS-Yushu, GH-Gonghe, QL-Qilian, GC-Gangcha

Full size image

At the phylum level, proteobacteria was dominant, with relative abundance ranging from 26 to 44% in the algal BSCs and 31–88% in the lichen BSCs across sampling sites (Supplemental Fig. S6). The relative abundance of Actinobacteria was higher in areas with low AI (Supplemental Fig. S6). The relative abundance of cyanobacteria was 1.77–10.89% in algal BSCs, and 0.13–3.32% in lichen BSCs across sampling sites (Supplemental Fig. S6). At the genus level, variations in the bacterial taxonomic composition across sampling sites were more evidently observed in the lichen BSCs than in the algal BSCs. We found that Burkholderia was the dominant genus for lichen BSCs in regions showing higher AI (Guinan, Maqin, Dari, and Jiuzhi) (Fig. 2c). The relative abundance of Microcoleus across sampling location ranged from 0.008–0.449% in lichen BSCs, and 0.197–4.079% in algal BSCs (Supplemental Fig. S7). In addition, it showed a higher relative abundance of Microcoleus in algal BSCs than in lichen BSCS at all sampling sites except for Gonghe.

Spatial distribution patterns and assemblage processes of the bacterial communities

The spatial variations of bacterial communities followed the distance-decay pattern for both lichen (R = 0.38, P < 0.001) and algal BSCs (R = 0.31, P < 0.001) (Fig. 3a). The slope of this linearity curve for the lichen BSCs was higher than for the algal BSCs, which indicated a faster turnover rate of bacterial community in the lichen BSCs. βNTI was calculated to determine the relative importance of the deterministic and stochastic assembly processes of bacterial community turnover across geographical scales. Results indicated that the dominant processes governing spatial variations of bacterial community differed between algal and lichen BSCs. The stochastic processes accounted for 53% for the algal BSCs, whereas they increased to 90% for the lichen BSCs (Fig. 3b).

Fig. 3
figure 3

(a) The distance-decay relationship of bacterial communities in algal and lichen BSCs. The x axial represents the geographical distance between pairwise sampling sites. (b) Assembly processes of bacterial community in algal and lichen BSCs across spatial scales. Red lines represent the average value of Beta NTI

Full size image

Soil characteristics underlying biocrusts and their associations with bacterial community structure

Soil pH ranged from 6.0–7.8 across sampling sites under lichen BSCs, and from 6.3–8.0 under algal BSCs (Fig. 4). According to the paired T test, soil pH under the lichen BSCs was slightly but significantly lower than under the algal BSCs at all sampling sites except for in Maduo (P < 0.05). TN ranged from 3 to 7 g·Kg−1 across sampling sites, and showed marginal differences between the two biocrust types (P > 0.05). TC ranged from 35 to 100 g·Kg−1 across sampling sites, and showed significant differences between the two biocrust types in Gonghe, Yushu, Maduo, and Jiuzhi (P > 0.05). TP ranged from 0.41–0.77 g·Kg−1 under lichen BSCs, and from 0.33–0.63 g·Kg−1 under algal BSCs, and there were no significant differences between the two biocrust types except for in Gangcha (Fig. 4).

Fig. 4
figure 4

Soil properties underlying algal BSCs and lichen BSCs at each sampling sites (means ± SD, n = 3). * indicates significant differences between algal and lichen BSCs according to the T test (P < 0.05)

Full size image

The Mantel test revealed that bacterial community composition in algal BSCs was significantly correlated with AI, TC, and TP. Soil pH and TN were significantly correlated with bacterial community composition in lichen BSCs but not in algal BSCs (Table 1). The relative abundance of bacterial taxa in algal BSCs was mainly correlated with climatic factors rather than with soil properties. While, for lichen BSCs, it was not only related to climatic factors, but also associated with soil pH (Fig. 5). In lichen BSCs, soil pH was positively correlated with the relative abundance of Chthoniobacter, Sphingomonas, Rubrobacter, Microvirga, and Methylobacterium. The relative abundance of Stenotrophomonas was negatively correlated with TC, TN, and TP. Furthermore, TP was negatively correlated with the relative abundance of Actinoplanes, Bryobacter, and Microcoleus; for algal BSCs, soil pH was positively correlated with the relative abundance of Rubrobacter, while negatively correlated with the relative abundance of Microcoleus.

Table 1 Correlations between bacterial community composition in BSCs and environmental factors based on the Mantel test. P < 0.05 indicates significant correlations
Full size table
Fig. 5
figure 5

Correlations between the relative abundance of the top 20 genera and environmental factors based on the Mantel test. AI-aridity index, MAT-mean annual temperature, MAP-mean annual precipitation, TC-soil carbon content, TN-soil nitrogen content, TP-soil phosphorus content, pH-soil pH value. Only genera that are significantly correlated with at least one of the environmental factors are listed. The red color represents positive correlations, and the green color represents negative correlations. The value indicates the correlation coefficient. * indicates significant correlations at P < 0.05, ** indicates significant correlations at P < 0.01

Full size image

Discussion

Differences of bacterial community between algal BSCs and lichen BSCs

Bacterial communities between algal BSCs and lichen BSCs were compared with respect to alpha diversity, beta diversity, taxonomic composition, and assembly processes. Previous studies failed to obtain a confirmed conclusion on microbial diversity change with biocrust succession. Some studies indicated that there were no significant changes in bacterial community diversity with biocrust development (Chilton et al. 2018; Nuttapon et al. 2020). Another study also indicated that the Shannon diversity of bacterial communities was similar among algal, lichen, and moss biocrust in the Gurbantunggut Desert (Zhang et al. 2016). Chen et al. (2020) analyzed microbial community in different development stages of algal biocrusts (formation stage, initial stage, intermediate stage, and maturity stage) and concluded that bacterial community diversity initially increased and later decreased with algal development. Shannon diversity of bacterial community in lichen BSCs was much lower than that in algal BSCs in the present study, which failed to support our hypothesis that bacterial diversity increased with biocrust development. Differences of bacterial diversity between BSCs types might be influenced by ecosystem types and climatic factors. In the present study, the difference between algal BSCs and lichen BSCs was closely associated with AI. Furthermore, the difference degree was more significant in areas with higher AI (Maqin, Maduo, Dari, and Jiuzhi). Therefore, changes in bacterial community diversity are not only related to the development of BSCs but also to the climate. Most of previous studies on BSCs were conducted on desert ecosystems (with rainfall less than 200 mm), while BSCs were sampled from grassland ecosystems with annual precipitation ranging from 420 to 880 mm in this study, which is much higher than that in desert ecosystems.

Cyanobacteria, the primary producers of BSCs (Xu et al. 2021), are usually the dominant bacterial taxon in arid areas including the coastal and central deserts of Oman (Abed et al. 2019) and cold desert of the Colorado Plateau, USA (Kuske et al. 2012), while actinobacteria were reported to be the dominant bacterial taxon of BSCs in cold steppe ecosystems especially in cooler and wetter climates (Blay et al. 2017). Proteobacteria were the dominant bacterial community in BSCs for dry steppes in northern Mongolia and a citrus orchard in Central Florida, USA (Kemmling et al. 2012; Nevins et al. 2021). Proteobacteria and actinobacteria were more abundant than cyanobacteria in alpine grasslands in this study. The differences in relative abundance of bacteria among studies may be caused by the varying climatic conditions in study areas. The annual precipitation ranges from 250 mm (the Colorado Plateau) to >800 mm (the Qinghai-Tibet Plateau). Moreover, nonlinear changes were observed with AI increasing for relative abundance of cyanobacteria, with the highest in areas with AI of 42 for both algal and lichen BSCs. Burkholderia, belonging to proteobacteria, has been reported to be capable of nitrogen fixation (Caballero-Mellado et al. 2007; Coenye and Vandamme 2003) and phosphate solubilization (Chen et al. 2022; Lin et al. 2006). The relative abundance of Burkholderia in lichen BSCs was much higher than that in algal BSCs in wetter areas, and became the dominant genus in Maqin (66.2%), Dari (53.8%), and Jiuzhi (32.4%). Besides, significant positive correlations between relative abundance of Burkholderia and annual precipitation were found in this study. Therefore, changes in microbial community composition caused by climate change may affect ecological functions of BSCs (Delgado-Baquerizo et al. 2014).

Assembly processes governing variations of bacterial community across geographical scales

The distance-decay pattern, which describes the effect of geographical distance on community dissimilarities (Morlon et al. 2008), is critical for understanding the driving factors of community turnover at spatial scales and beta-diversity studies (Ferrier et al. 2010), and has been widely studied in soil microbial biogeography (Feng et al. 2019; Shi et al. 2018). Bacterial communities in both lichen BSCs and algal BSCs show significant distance-decay patterns (Fig. 4a), which was consistent with previous studies that involved three biocrust types (alga, lichen, and moss-dominated BSCs) across northern China (Li and Hu 2021; Su et al. 2020). Hanson et al. (2012) proposed four processes (selection, drift, dispersal, and mutation) that interplay the distance-decay pattern; selection and drift increase the slope of the distance-decay curve, while dispersion weaken this relationship. In this study, the slope of the linear curve for lichen BSCs was greater than that of algal BSCs. Therefore, selection and drift might play greater roles in spatial variations of bacterial community in lichen BSCs, and dispersal was more important in algal BSCs. Environmental selection is one of deterministic processes, and diffusion and genetic drift are stochastic processes (Chase and Myers 2011; Vellend 2010). Larger contributions to stochastic processes were found in lichen BSCs in this study, which indicated that drift had stronger effects on spatial variations of bacterial community in lichen BSCs than that in algal BSCs. Moreover, drift is more likely to occur in small communities with lower diversity (Hanson et al. 2012), which further confirmed the results. Despite the primary role of stochastic processes, some deterministic factors, including environmental variables (AI, soil pH, TC, TN, and TP), were still detected (Fig. 5).

Correlations between bacterial community in biocrusts and environment factors

Soil characteristics and plant communities are the most important factors influencing the bacterial communities in BSCs at local scales (Rivera-Aguilar et al. 2009), while climate is the main driving factor at large geographical scales (Zedda et al. 2011). Bacterial communities in biocrusts are fragile and sensitive to climate changes (Belnap and Eldridge 2001; Steven et al. 2015). Cold and wet environments could lead to a dramatic decline in the microbial diversity in biocrusts (Blay et al. 2017). Warming strongly reduced the diversity and changed the composition of lichen-dominated BSCs, which eroded the positive impacts of BSCs on ecosystem processes (Escolar et al. 2012). However, the direction and magnitude of climate effects differs among the biocrust types. Previous studies have indicated that precipitation is positively correlated with the species richness of mosses and negatively correlated with species richness of cyanobacteria and algae (Li et al. 2017). In this study, wetter climate had negative effects on bacterial community diversity for lichen BSCs and positive effects for algal BSCs. Decreasing biodiversity has a significantly adverse impact on ecosystem functions (Isbell et al. 2017). Therefore, climate changes in the future might accelerate functional differentiation of different biocrust types due to different responses of bacterial community.

State-and-transition Models proposed that thresholds exist for different successional stages of ecosystems (Bowker 2010). In desert ecosystems, organic carbon is considered a threshold of biocrust successions from cyanobacterial BSCs to lichen BSCs, and nitrogen and phosphorus availability determines biocrust succession from lichen to moss (Deng et al. 2020). In the present study, the bacterial community diversity and composition changed significantly at the site with aridity index of 48, especially for lichen BSCs. Therefore, we believe that this value was a critical transition point for changes in the bacterial microbial community. In addition, relative abundance of bacterial taxa in different BSCs types respond differently to AI. The relative abundance of Microcoleus, the most common cyanobacteria in this study, was negatively correlated with AI in lichen BSCs; however, this relationship was not significant in algal BSCs. Species belonging to Microcoleus genus respond differently to environmental factors. For example, a survey across North America indicated that two Microcoleus species (M. vaginatus and M. steenstrupii) have different environmental adaptations and responses to temperature (Garcia-Pichel et al. 2013). Muñoz-Martín et al. (2019) also reported the different sensitivity of cyanobacteria species to increasing temperature. Algal and lichen BSCs might have diverse species of Microcoleus with distinct adaptabilities to temperature or precipitation, which could explain the different AI pattern in both types of biocrust.

Conclusions

Spatial variations of bacterial community composition associated with algal and lichen BSCs were explored in high altitude areas that are more sensitive to climate changes. Bacterial diversity in algal BSCs was much higher than that of lichen BSCs. Additionally, bacterial community structure and the spatial distribution pattern also differed between algal BSCs and lichen BSCs. A wetter climate stimulated increases in bacterial diversity for algal BSCs but caused a decrease for lichen BSCs, which led to the increased difference of bacterial community diversity and composition between the two BSCs types. Stochastic processes are much more important in shaping spatial variations of bacterial community in lichen BSCs than in algal BSCs. These results highlight the influence of climate on the diversity and composition of BSCs, and provide an important reference for studies of bacterial community in BSCs in alpine ecosystems. Future long-term controlled experiments are required to explore the effects of climate changes on structures and functions of BSCs in different ecosystems to enrich our knowledge regarding the response of terrestrial ecosystems to climate changes.