Trade-oﬀ between stochastic and deterministic processes shifts from soil to leaf microbiome of tea plant

Background: Plant microbiome is thought to play an important role in promoting plant health and production. However, even though the microbiomes in various compartments have been widely investigated, the association between above and belowground compartments of plants remain unclear. Tea is a globally popular beverage due to its ﬂavor and health beneﬁts associating with secondary metabolites; the microbiomes of tea plant ( Camellia sinensis ) play a signiﬁcant role in the production of these secondary metabolites. Here, we investigated the microbiomes of bulk and rhizosphere soils, roots and leaves of C. sinensis collected from tea plantations across over 2000 km to investigate the association and driving mechanisms for microbiomes in the compartments. Results: Camellia sinensis microbiomes diﬀered between the compartments with α -diversity gradually decreasing from soils to roots and leaves. The core leaf microbiome comprised Bacilli, Sphingobacteriia and α -Proteobacteria, which we suggest might ascendingly migrate from soils to leaves. Microbial community assembly processes were dominated by deterministic processes in bulk and rhizosphere soils; these assembly processes were dominated by stochastic processes in roots and leaves. Dispersal limitation was stronger in old leaves than in other compartments. Amino acids were also critical drivers for environmental selection. The microbiomes in C. sinensis roots and leaves possessed a lower intensity of microbial associations and more negative microbial associations than in bulk and rhizosphere soils, suggesting that the contribution of microbial interactions varied in diﬀerent compartments. Conclusion: there is a trade-oﬀ between stochastic and deterministic processes in microbiomes community assembly along from soil to leaf of C. sinensis . These results provide valuable information for understanding the associations and driving mechanism of microbiomes in various C. sinensi compartments, which could be used to predict C. sinensis microbiome and harness its power to improve tea production and quality. for community assembly allows us to use community as- sembly mechanisms to predict C. sinensis microbiomes. In addition, the association between amino acids and endophytic leaf microbiomes provide insight into the roles of tea microbiomes in improving the productivity and quality of tea this provides a comprehensive analysis of associations across C. sinen- sis microbiome in compartments from to leaves, our these provide insight into the driving mechanisms for community assembly, the roles of tea microbiomes in improving the produc- tivity and quality of tea production can only be fully understood by deciphering the underlying relationships between endophytic leaf microbiomes and C. sinensis ’s secondary metabolites.

of other compounds such as α-dicarbonyl, glyoxal, methylglyoxal, and diacetyl [11]. 32 The antibacterial activity of flavonoids and caffeine could therefore regulate endo-33 phytic bacterial communities [12]. In contrast, endophytic bacteria can also degrade 34 flavonoids via deglycosylation [13] and degrade caffeine via demethylation and ox-35 idation pathways [14]. The microbiomes of C. sinensis are also a key factor in the 36 fermentation processes of various tea products [15]. Accordingly, understanding the 37 assembly processes for C. sinensis microbiomes is essential for improving the pro-38 duction and quality of tea products. 39 Endophytic microbiomes are transmitted through seed dispersal or recruited  To capture biogeographical differences in C. sinensis microbiomes, we collected 68 samples from 45 locations spanning all 15 tea planting provinces in China (Fig. 1a).

69
The samples were collected from diverse C. sinensis varieties, soil types and climate 70 types (Table S1); soil characteristics are listed in Data file S1. Five compartments 71 (bulk soil, rhizosphere, roots, young leaves and old leaves) were collected at each     Community similarity geographic decay was tested by estimating the relationship 132 between community similarity and geographic distance. Dispersal limitation was 133 tested using correlations between community similarity and geographic distance un-134 der a partial correlation condition for environmental properties using partial mantel

162
Core microbiota of C. sinensis microbiomes from soil to leaves 163 A total of 225 samples from 45 tea plantations in prominent tea-producing regions 164 of China (Fig. 1a, Table S1) were collected to explore the microbiomes of bulk 165 soil, rhizosphere soil (hereafter 'rhizosphere'), roots, young leaves and old leaves.  The SourceTracker analysis revealed substantial similarities within the two leaf com-184 partments and within the two soil compartments, and also indicates that there are 185 rare exchanges between roots and leaves, indicating partitioning between above and 186 belowground microbiomes (Fig. 2a). This transfer bottleneck can be observed as a 187 34% similarity between the rhizosphere and roots but only a 5% similarity between 188 roots and old leaves. When genera are compared sequentially across the compart-189 ments, each ascension in a compartment (e.g. from soil to rhizosphere or old leaf 190 to young leaf) lead to substantially more depleted OTUs than enriched OTUs (Fig   191   2b). Depletion ratios (depletion/enrichment) ranged from 0.5 in rhizosphere/bulk 192 soil to 7.5 for both old leaf/root and root/rhizosphere. Sphingomonas,  terium and Burkholderia were frequently associated with these changes (Fig. 2c).

194
These results show strong filtering effects from rhizosphere samples to roots and 195 from roots to leaves.

196
Community assembly mechanisms of C. sinensis microbiomes from soils to leaves 197 Community assembly normalized stochasticity ratio (NST) increased from 40% in 198 bulk soil to more than 70% in leaves (Fig. 3a). Nestedness was significant for rhi-199 zosphere (t-tests, P < 0.05) and bulk soil microbiomes (t-tests, P < 0.05) but 200 non-significant for root and leaf microbiomes (t-tests, P > 0.05) at all ranks from 201 genus to phylum (Fig. 3b). Microbiomes in soils, roots and leaves were nested as 202 subsets of rhizosphere microbiomes and show distinct patterns in species richness 203 (Fig 3c). When considered using Bray-Cutis similarity and after controlling the 204 impact of the environmental matrix using Partial Mantel tests, all compartments 205 were negatively correlated with geographic distance (Fig. 3d). This response was 206 particularly strong in old leaves; Mantel correlograms show that only this com-207 partment linearly changed with geographic distance (Fig. 3e). The environmental 208 selection effect of physiochemical properties in rhizosphere and bulk soils, amino 209 acids and catechin in roots and leaves, and heavy metals on microbial communi-210 ties was assessed using constrained corresponding analysis (Fig. 3f). Explanation  interactions, was the largest in bulk and rhizosphere soils, followed by roots and 224 young leaves, and was the smallest in old leaves (Fig. 4). The network diameter, 225 representing the longest path in a network, was the longest in young and shortest in old leaves; transitivity, representing network modularity, was highest in roots and 227 lowest in young leaves (Table S2). Although plant tissues had fewer links, negative 228 links in tissue microbiomes were more abundant than in rhizosphere and bulk soils 229 (Fig. 4). The co-occurrence networks also displayed different taxon assortativity 230 (Table S2) stochastic and deterministic processes in above and belowground C. sinensis micro-242 biomes (Fig. 5). This finding allows us to use community assembly mechanisms to 243 predict C. sinensis microbiomes, which offers fundamental information to explore 244 the role of microbiomes in improving tea productivity and quality.

245
The NST values suggest that microbial community assembly processes were dom-  The associations between microbial communities and geographic distance after con-  In summary, the present study provides an overview of C. sinensis microbiome in 340 compartments from soils to leaves. With this study, we have shown that C. sinen-341 sis microbiomes gradually changed along the compartments from soils to leaves in 342 term of α-and β-diversity. Moreover, we find that microbial community assembly 343 processes were dominated by deterministic processes in bulk and rhizosphere soils; 344 these assembly processes were dominated by stochastic processes in roots and leaves.

345
The driving mechanisms for community assembly allows us to use community as-346 sembly mechanisms to predict C. sinensis microbiomes. In addition, the association 347 between amino acids and endophytic leaf microbiomes provide insight into the roles 348 of tea microbiomes in improving the productivity and quality of tea production.

349
While this study provides a comprehensive analysis of associations across C. sinen-350 sis microbiome in compartments from soils to leaves, our understanding is still in 351 its infancy. Although these analyses provide insight into the driving mechanisms 352 for community assembly, the roles of tea microbiomes in improving the produc-    Figure 1 Compositions and core microbiota of microbiomes in different C. sinensis compartments.      Data file S1. The physicochemical properties of bulk soils. 515 Data file S2. The physicochemical properties of rhizosphere soils. 516 Data file S3. The physicochemical properties of roots. 517 Data file S4. The physicochemical properties of old leaves. 518 Data file S5. The physicochemical properties of young leaves. 519