Water and Heat-Triggered Changes of Bacterial Community Structure and Function in Biological Soil Crusts

Background and Aims(cid:0)The outstanding ability of biological soil crusts (BSCs) in soil microenvironments regulation is mainly attribute to microorganisms that colonizing in biocrusts. We aimed to investigate the changes of bacterial community structure and function with biocrust succession, as well as their responses to climatic changes across large geographical scales. Methods: Algal BSCs and lichen BSCs were sampled along an aridity gradient on alpine grasslands. Bacterial communities in biocrusts were measured using high-throughput sequencing, and soil underlying biocrusts (0-5 cm) was collected for nutrients determination. Results: Our results indicated that compared with algal BSCs, bacterial community in lichen BSCs was characterized by lower diversity, more complex co-occurrence network and mutually benecial relationships. The bacterial community assembly was governed mainly by stochastic processes for lichen BSCs, which was different from the almost equally important roles of stochastic and deterministic processes for algal BSCs. Geographical location had a signicant effect on bacterial communities in both algal and lichen BSCs, while had a greater effect on lichen BSCs. It is noteworthy that the bacterial diversity of algal BSCs was positively correlated with aridity index, while that of lichens was negatively correlated with aridity index. Moreover, we determined lower soil pH and higher soil phosphorus content underlying lichen BSCs, implying their advantages in soil improvement. Conclusions: Aridity index was one of important driving factors of bacterial community in biocrusts, and its effects were biocrust type dependent. Lichen BSCs had greater effects on soil improvement than that of algal BSCs.

In the context of global climate change, temperatures at high latitudes and elevations have increased signi cantly more than in other regions (IPCC 2007;Benito et al. 2011). The Qinghai-Tibetan Plateau, the highest and largest plateau on Earth (Shen et al. 2015) is experiencing rapid warming and precipitation regime changes. Considering its unique geographic location and signi cant ecological and production functions, the effects of climate change or human activities on biodiversity, vegetation composition, soil characteristics, and ecosystem functions have largely been studied in this area Zhang et al. 2019. In this study, we sampled two types of BSCs with different succession stages from nine regions on the Qinghai-Tibetan Plateau that varied in altitude, annual temperature, and precipitation.
Bacterial community diversity and assembly in the crust layer were measured using high-throughput sequencing, and soil nutrients underlying biocrusts were determined. The nine selected locations were across the north to south gradient along the eastern boundary of the Qinghai Province, covering the main grassland types (alpine steppe and alpine meadow) and showed signi cant climatic differences. To the best of our knowledge, this research is the most extensive investigation on a geographical scale of the microbial communities in the BSCs in the Qinghai-Tibet Plateau, which has signi cant application in the utilization of biocrust resources for ecosystem restoration. We hypothesized that 1) microbial community diversity increases with biocrust development, 2) microbial communities in biocrusts with different successional stages responds differently to climatic changes, 3) the community assembly process and co-occurrence network structure changes with biocrust succession, and 4) the ability to improve the underlying soil quality differs between the two biocrust types.

Study area
Along the eastern edge of Qinghai Province, China, nine regions experiencing different annual mean temperatures and precipitation (AMT and AMP, respectively) were selected for biocrust sampling (Supplemental Fig. S1, Table 1). The AMT and AMP data from 2014 to 2019 were obtained from local meteorological stations. The study areas were characterized by an alpine continental climate with an AMT ranging from -3.12 to 5.09 °C and AMP ranging from 420 to 880 mm. Rain and heat over the same period with the highest temperature and rainfall occurred during the growing season (June-September). The AMT and AMP were typically lower and higher, respectively, at the southern sites (Maqin, Dari, and Jiuzhi) than at the northern sites (Gonghe, Gangcha, and Qilian). The de Martonne aridity index (AI) (AI=average precipitation/average temperature + 10) was used to determine the water and heat conditions of the study sites (Zhou et al. 2020). The main vegetation types in the sampling areas were alpine steppe and alpine meadows, dominated by Stipa aliena Keng, Kobresia pygmaea, Elymus nutans, Potentilla chinensis, and Leontopodium nanum. BSCs colonized plant interspaces (Supplemental Fig.  S1).  DNA extraction, PCR ampli cation, and sequencing The sampled BSCs were homogenized with a mill and cooled 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 1.0% agarose gel electrophoresis, and NanoDrop One (NanoDrop Technologies, Wilmington, USA) was used for purity examination. Speci c primers [(5'-GGACTACHVGGGTWTCTAAT-3') and (5'-ACTCCTACGGGAGGCAGCA-3')] were used to amplify the V3-V4 region of the 16S rRNA. The PCR procedure was performed as described by Wei

α-diversity of bacterial communities in the biocrusts
After nal quality ltering, the read numbers were rare ed to the minimum size (7,796) to ensure the same sequencing depth, resulting in a total of 1,1298 bacterial OTUs. Of these, 4245 OTUs (37.6% of the total OTUs) were speci c for the algal BSCs, 1327 (11.7%) OTUs were speci c for the lichen BSC, and 5726 OTUs (50.7%) were shared between both (Supplemental Fig. S2). The rarefaction curves reached a saturation plateau, suggesting that the sequencing depth was su cient to cover most bacterial species (Supplemental Fig. S3). Our results indicated that sampling locations signi cantly affected the Shannon index of the bacterial community for both algal BSCs and lichen BSCs. Moreover, the Shannon index of the bacterial communities in the algal BSCs were signi cantly higher than that in the lichen BSCs ( Fig. 1b). Further, the variation degree between the two biocrust types increased as the AI increased (Fig. 1c). To further clarify the variation rules of bacterial diversity in the two biocrusts across large geographical scales, Spearman's correlation analysis was performed between the AI and the Shannon diversity index. Interestingly, we found that the Shannon index was negatively correlated with the AI for the lichen BSCs, while it was positively correlated with the AI for the algal BSCs (Fig. 1d).
β-diversity, taxonomic composition, and functional genes of the bacterial communities Variations in the bacterial community composition were tested using PERMANOVA and visualized through PCoA plots. The results suggested that the bacterial community composition in the algal BSCs signi cantly differed from that in the lichen BSCs (F = 11.9, R 2 = 0.186, P = 0.001) (Fig. 2a). Furthermore, sampling locations signi cantly affected the bacterial community composition in the lichen (F = 4.77, R 2 = 0.679, P = 0.001) and algal (F = 4.18, R 2 = 0.650, P = 0.001) BSCs (Supplemental Fig. S4). According to the PCoA plots, sampling sites with higher humidity (Maqin, Maduo, Dari, and Jiuzhi) were grouped together and separated from those with lower humidity for both the algal and lichen BSCs (Fig. 2b).
Differences in the taxonomic composition were also analyzed. At the phylum level, proteobacteria was the dominant phylum, with relative abundance ranging from 26-44% in the algal BSCs and 31-88% in the lichen BSCs along the climatic gradients (Supplemental Fig. S5). The relative abundance of Actinobacteria was higher in areas with low AI (Supplemental Fig. S5). At the genus level, variations in the bacterial taxonomic composition across the different geographical locations were more evidently observed in the lichen BSCs than in the algal BSCs. We also found that Burkholderia was the absolute dominant genus for lichen BSCs in regions that showed high AI (Guinan, Maqin, Dari, and Jiuzhi) (Fig. 2c). In addition, the relative abundance of functional genes associated with ferric uptake, phosphate transport, and secretion proteins in the lichen BSCs was much higher than those in the algal BSCs, especially in areas with high AI (Fig. 3).
Spatial distribution patterns, assemblage processes, and co-occurrence network of the bacterial communities The distance-decay relationship, which re ects spatial distribution pattern of biodiversity, has been widely studied in the eld of microbial ecology. It describes the reduced similarity in species composition between two communities as the geographic distance increases. Our results indicated that the bacterial communities in both lichen and algal BSCs followed the distance-decay pattern (Fig. 4a). The slope of this linearity curve for the lichen BSCs was higher than the algal BSCs, which indicated a faster turnover rate of the lichen BSCs. βNTI was calculated to determine the relative importance of the deterministic and stochastic assembly processes in the bacterial community assembly. The results indicated that the dominant assemblage processes of the bacterial communities changed with the succession of BSCs. The stochastic processes accounted for 53% for the algal BSCs, whereas it increased to 90% for the lichen BSCs (Fig. 4b). The co-occurrence network of the bacterial community in the lichen BSCs was characterized by higher link number, average degree, clustering coe cient, and network density compared with that in the algal BSCs (Fig. 5). Further, the percentage of positive linkages in the network increased with biocrust succession.

Soil characteristics underlying biocrusts
The physicochemical properties of soils underlying the biocrusts were determined (Fig. 6). Results indicated that soil pH under the lichen BSCs was much lower than that under the algal BSCs at most sampling sites. Soil carbon and nitrogen content showed marginal differences between the two biocrust types, while soil phosphorus content underlying the lichen BSCs was much higher than that underlying the algal BSCs. We also found that the bacterial community in the lichen BSCs was signi cantly correlated with soil pH, and soil carbon, nitrogen, and phosphorus contents. Further, the bacterial community in the algal BSCs was signi cantly correlated with soil nitrogen and phosphorus contents ( Table 2).  Previous studies also indicated a signi cant effect of climate changes on biocrust succession by monitoring changes in biocrust coverage at different developmental stages (Ferrenberg et al. 2015).
Microbial communities of different crusts types respond differently to climate changes. For example, precipitation was positively correlated with the species richness of mosses and negatively correlated with the species richness of cyanobacteria and algae (Li et al. 2017). We found that the AI was negatively and positively correlated with the bacterial diversity for the lichen BSCs and the algal BSCs, respectively. In addition, the variations in the diversity between the two biocrust types were more evident in wetter regions. Decreasing biodiversity has a signi cantly adverse impact on ecosystem functions (Isbell et al.  2017). Therefore, climate changes in the future might accelerate functional differentiation of different biocrust types.
State-and-Transition Models propose that thresholds exist for different successional stages of ecosystems (Bowker 2010). In desert ecosystems, organic carbon is considered as the 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). We predicted that the AI might be an important threshold of bacterial community diversity in the BSCs. 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. Based on this result, we divided the nine sample sites into low aridity index group and high aridity index group. Signi cant differences in the bacterial community composition were observed between the two groups. Therefore, we believed that this value was a critical transition point for changes in the bacterial microbial community. In addition, the dominant Burkholderia genus in regions with high rainfall (Guinan, Maqin, Dari, and Jiuzhi) for the lichen BSCs might restrict the growth of other microorganisms and decrease the microbial community diversity.

Effects of the biocrusts on the underlying soil properties
Biocrusts contribute signi cantly to soil nutrient accumulations through microbial activities (Mager and Thomas 2011; . Moreover, BSCs affect nutrient cycling by modifying subsoil microbial community composition (Nevins et al. 2021). Our results revealed that the effects of biocrust on the underlying soil properties were associated with biocrust types and sampling locations. Biocrusts could decrease their underlying soil pH by releasing H + or organic acids (Belnap 2011). The soil pH underlying the lichen BSCs was lower than that underlying the algal BSCs, which proved the greater role of biocrusts in the late succession stages on soil improvement than biocrusts in the early succession stages. Organic phosphorus compounds secreted by microorganisms colonizing in biocrusts also increase phosphorus levels in soil underlying biocrusts (Baumann et al. 2018). The higher abundance of functional genes related to phosphorus uptake and transport, and secreted proteins might explain the higher ability of phosphorus accumulation at the late succession stage of biocrust (lichen BSCs). Previous studies reported soil carbon and nitrogen accumulation in the top-layer of soil with crust development (Chamizo et al. 2012); however, we detected marginal differences in the soil carbon and nitrogen contents between the two biocrust types compared with total phosphorus content. Phosphorus transformation in soils is an open-loop process, which is different from carbon and nitrogen cycles. Autotrophs in terrestrial ecosystems can x carbon and nitrogen in the atmosphere, and heterotrophic microbes release the xed carbon and nitrogen into the atmosphere as gases by decomposing organic matter. However, phosphorus barely returns to the atmosphere from terrestrial ecosystems, and only transforms in different forms within terrestrial ecosystems, such as labile phosphorous, relatively stable phosphorous, and organic phosphorous.

Conclusions
In this study, we explored the variations in the bacterial community composition with biocrust succession stages across large geographical scales. The effect of biocrusts on the underlying soil improvement was also evaluated. We found that bacterial communities in the BSCs were affected signi cantly by the biocrust types and geographical locations, and geographical locations had greater effects than biocrust types at large geographical scales. With biocrust development, bacterial diversity decreased, cooccurrence network became complicated, and the proportion of coexistence edges increased.
Furthermore, we found that stochastic process governed bacterial community assembly for the lichen    Functional genes with signi cant differences in relative abundance between the two biocrust types at sampling sites with higher aridity index (A), as well as between higher and lower aridity index sites for the lichen BSCs (B). The distance-decay relationship and community assembly processes of bacterial community in algal and lichen BSCs.
Page 22/23 Figure 5 Co-occurrence networks of microbiomes colonizing in biocrusts. The nodes size was proportional to the degree, and the nodes color indicated bacterial phylum taxa. The edges with green color indicated positive correlations, and that with red color indicated negative correlations. The width of edges was proportional to the correlation coe cient.