OTU clustering and species annotation
The raw sequence data of all samples consisted of 5,060,529 reads prior to quality checking and assigning the reads to their respective samples. The average read length (± standard deviation) of reads before processing was 243.77 ± 11.20 bp. After quality trimming and assigning reads to different samples, 4,489,368 high-quality reads remained in the data set, with an average length (± standard deviation) of 246.45 ± 15.36 bp. A total of 3,753 OTUs were generated after clustering at a 97% similarity level. Representative sequence for each OTU was screened for further annotation. 7 phyla, 33 classes, 117 orders, 271 families, 480 genera, and 762 species were identified from these sequences.
Composition of fungal communities
At the phylum level, Ascomycota (83.77%), Basidiomycota (11.71%), and Zygomycota (3.45%) were the three dominant phyla (Fig. S1). At the class level, Dothideomycetes (52.8%) and Eurotiomycetes (32.67%) own a higher proportion compared to rhizosphere soil (11.41% and 6.95%, respectively) in all ages' leaf samples. Sordariomycetes and Leotiomycetes were more abundant in rhizosphere soil (23.76% and 14.40%, respectively) than that in leaf niches (1.25% and 1.45%, respectively) (Fig. 1). Notably, Cystobasidiomycetes and Microboiryomyceres were more abundant in the upper leaf than other leaf niches. Capnodiales (25.35%) and Chaetothyriales (13.93%) were the dominant orders in leaf niches (Fig. S1), while Mortierellales (12.10%), Hypocreales (11.47%) and Helotiales (8.53%) were the dominant orders in rhizosphere soil. The dominant families in the leaf niches were Davidiellaceae (14.65%), Incertae_sedis_Chaetothyriales (13.31%), Incertae_sedis_Dothideomycetes (12.71%) and Mycosphaerellaceae (9.86%). But Mortierellaceae (11.21%), Nectriaceae (4.86%) and Amphisphaeriaceae (4.27%) show a certain high proportion in the rhizosphere soil (Fig. S1). Cladosporium (12.73%), Zymoseptoria (9.18%), and Strelitziana (13.11%) were the dominant genera in the leaves (Fig. S1).
A Venn diagram was constructed to highlight the similarities and differences in communities among different ages of plants and leaf/soil niches. The communities in YS, ES and LS had 565 OTUs in common; the upper leaves, middle leaves, lower leaves, and rhizosphere soil had 487 OTUs in common (Fig. 2). Some OTUs appeared in the leaf endophytic fungi community were also detected in the rhizosphere soil, which showed the colonization probability of soil fungi in leaves. The large number of common OTUs among samples from different-aged trees indicated that colonization patterns may be conserved during long-term evolution. The leaf endophytic fungal communities differed between LS and YS/ES (Fig. 3) remarkably. According to the relative abundance, we found that the abundance of some families among the top 35 OTUs, such as Dothioraceae, Taphrinaceae, Wallemiaceae, Amanitaceae, Herpotrichiellaceae, Incertae_sedis_Capnodiales, Incertae_sedis_Sporidiobolales, Incertae_sedis_Dothideomycetes, Davidiellaceae, Teratosphaeriaceae, Incertae_sedis_Erythrobasidiales, Geoglossaceae, Incertae_sedis_Pleosporales, Rutstroemiaceae, and Pleosporaceae, decreased with the increasing age of the plant. In contrast, the abundance of Incertae_sedis_Chaetothyriales, Elsinoaceae, Mycosphaerellaceae, Tuberaceae, Glomeraceae, and Ramalinaceae increased with the tree age, and gradually became dominant in LS.
Alpha rarefaction curves and alpha diversity
The rarefaction curves approached the plateau phase, indicating that it would be unlikely for more fungal taxa to be detected with additional sequencing (Fig. 4). And these curves showed the endophytic fungi communities were less diverse in leaves than in the rhizosphere soil, as evidenced by differences in number of OTUs between these communities.
Community richness and diversity were analyzed using five alpha diversity indices: Chao1, Shannon’s, Simpson’s, ACE, and Goods_coverage (Table S1). The depth index (Goods_coverage) of each sample library was over 99% (99.2–99.8%), indicating that the sampling was reasonable. The Chao1 and ACE indices are indicative of fungal community richness, and Shannon’s and Simpson’s indices are indicative of fungal community diversity. The fungal richness and diversity were significantly higher in rhizosphere soil than in leaf samples (p < 0.0001) (Fig. 5a). Especially, the richness and diversity of fungi in the rhizosphere soil of the mother plant were highest among all samples (Table S1). The different leaf niches had similar fungal alpha diversity (Fig. 5a). The Shannon’s indices indicated that fungal diversity was greater in younger plants (YS and ES) than in the mother plant (LS) (p ≤ 0.01), but did not differ significantly between YS and ES (Fig. 5b).
Applying NMDS analysis, the degree of difference between groups or in-group can be reflected through a multidimensional space. Here, the NMDS analysis revealed clearly that the mycobiomes between rhizosphere soil samples and leaf samples were significantly distinguished (R2 = 0.23, p < 0.001) (Fig. 6, left).
Furthermore, when analyzing the mycobiome composition of all samples, no matter the different leaf niches or the different tree ages, the samples could be clearly separated (R2 = 0.022, p < 0.001) (Fig. 6, right). And independent analysis of YS, ES and LS revealed significant differences respectively in the composition of endophytic fungi in the upper, middle and lower niche (p < 0.001). Among them, the composition of endophytic fungi in the upper leaves of YS was not significantly different from that in the middle (R2 = 0.178, p = 0.092), but the differences between the upper and lower layers and the middle and lower layers were quite significant (R2 = 0.394, p = 0.007; R2 = 0.274, p = 0.006). This maybe the reason that plant height of annual plant was not high enough and endophytic fungi distributed in different height leaves evenly. In the comparative analysis of the endophytic fungi composition of the three leaf niches of ES, the pairwise differences between them were found to be significant (p < 0.05). This indicated that as the plant has grown for 20 years in this study, the colonization of leaf endophytic fungi at different heights has formed a significant difference. Growth time has a certain effect on the distribution of endophytic fungi. However, in the LS sample, the differences between the upper, middle, and lower layers were not significant. This may be the reason that the plant growth period was so long enough that the endophytic fungi distributed in the three niches widely.
A horizontal comparison of tree age based on different leaf niches showed that there was a significant difference between different samples in middle and lower leaf layers (p < 0.001). However, in upper leaf layer, although the fungi composition between YS, ES, and LS existed significant difference, the difference between YS and ES was not significant (p = 0.341). This indicated that the leaves of upper layer of the tea branch that have grown for 20 years were consistent with the growth situation and distribution of endophytic fungi of the same layer leaves of the one-year-old plant. This could also be explained by the number of shared OTUs among YS and ES (789 OTUs), which was much higher than that of YS and LS (678 OTUs) or ES and LS (704 OTUs) (Fig. 2).
All these analysis showed a certain discipline of the distribution of endophytic fungi in different leaf niches of the same plant. At the same time, the composition and distribution of endophytic fungi in plants also showed certain differences with the growth time. That is to say, the distribution of endophytic fungi in leaves of C. Sinensis (cv. Shu Cha Zao) has a certain spatiotemporal variation.