The process of nitrogen-adaptation root endophytic bacterial rather than phosphorus-adaptation fungal subcommunities construction unveiled the tomato yield improvement under long-term fertilization

Abstract

Interactions between endophytes (endophytic bacteria and fungi) and plants are crucial in maintaining crop fitness in agricultural systems, particularly in relation to abundant and rare subcommunities involved in community construction. However, the influence of long-term fertilization on heterogeneous rhizosphere nitrogen and phosphorus environments and how these conditions affect the key subcommunities of root endophytes and their community assembly mechanisms remain unclear. We studied the 26th year of a field experiment conducted in a greenhouse with varying levels of nitrogen and phosphorus (CKP0, CKP1, CNP0, CNP1, ONP0, and ONP1) to assess the composition of tomato root endophytes and their impact on yield. We employed 16S rRNA and fungal ITS region amplicon sequencing to investigate the assembly mechanisms of abundant and rare endophytic subcommunities, network correlations, core subcommunity structures, and key species that enhance crop yield. The results indicated that organic manure and phosphorus fertilizers significantly increased the rhizosphere soil nitrogen content, phosphorus content, and phosphorus availability (labile P, moderately labile P, and non-labile P). These fertilizers also significantly affected the composition (based on Bray-Curtis distance) and community assembly processes (βNTI) of endophytic microbial subcommunities. The assembly of both bacterial and fungal subcommunities was primarily governed by dispersal limitation, with community structures being significantly regulated by the content of rhizosphere soil available nitrogen (AN) and moderately labile P (MLP). Rare bacterial and fungal subcommunities complemented the ecological niches of abundant subcommunities in the co-occurrence network, supporting community functions and enhancing network stability. Nitrogen-adapting abundant and rare bacterial subcommunities provided a stronger predictive correlation for tomato yield than phosphorus-adapting fungal subcommunities. Additionally, three core genera of rare endophytic bacteria such as Arthrobacter, Microbacterium, and Sphingobium were identified as potentially involved in improving crop yield improvement. These findings revealed the distinct assembly mechanisms of endophytic microbial subcommunities affected by fertilization, enhancing our understanding of better management practices and controlling endophytes to improve crop yield in intensive agricultural ecosystems.

Keywords: abundant microbial taxa; community assembly; endophyte; rare microbial taxa; solar greenhouse.

Figure 1

Shannon and richness indexes of root endophytic abundant bacteria (A,E), rare bacteria (B,F), abundant fungi (C,G), and rare fungi (D,H) under different treatments. CKP0, Unfertilized; CKP1, Chemical phosphorus applied; CNP0, Chemical nitrogen applied; CNP1, Chemical nitrogen plus phosphorus; ONP0, Organic manure supply nitrogen; ONP1, Organic manure supply nitrogen plus phosphorus. *p < 0.05, as tested by multiple comparisons using the Kruskal–Wallis test, and only significant differences are displayed. The treatment denotations are the same as those in Figure 1.

Figure 2

Subcommunity composition at the phylum level and non-metric multidimensional scaling (NMDS) plots of root endophytic abundant bacteria (A,E), rare bacteria (B,F), abundant fungi (C,G), and rare fungi (D,H) under different treatments. Unclassified bacterial and fungal OTUs were labeled as “Others.” The treatment denotations are the same as those in Figure 1.

Figure 3

Root endophytic bacterial and fungal subcommunities assembly processes (based on βNTI and RCBray). βNTI of root endophytic abundant bacteria (A), rare bacteria (B), abundant fungi (C), and rare fungi (D) subcommunities. Proportions of deterministic and stochastic assembly processes in governing root endophytic abundant bacteria (E), rare bacteria (F), abundant fungi (G), and rare fungi (H) subcommunities under different treatments.

Figure 4

Co-occurrence networks of bacterial and fungal coexistence communities under different fertilizations (A–F). The networks were established by calculating correlations among abundant, rare, and other OTUs.

Figure 5

Distributions of network roles by analyzing module features, within-module connectivities (Zi) and among-module connectivities (Pi), in the endophytic abundant bacterial (A), rare bacterial (B), abundant fungal (C), and rare fungal (D) co-occurrence networks of root, respectively.

Figure 6

Changes in tomato yield (A). The random forest regression model shows the root endophytic subcommunity structures as drivers of yield. A correlation between the structure of microbial subcommunities and soil nutrients was identified using Spearman’s test (B). The random Forest regression model identifies the top 10 most important taxa of abundant (C) and rare (D) subcommunities at the genus level, along with their respective phylum, as key drivers of tomato yield. The bars illustrate the increase in mean squared error (%IncMSE) scores for the contribution of root endophytes at the genus level among six fertilization treatments, affecting the tomato yield. Significance levels are denoted with *p < 0.05, **p < 0.01, and ***p < 0.001.

Figure 7

Structural equation model (SEM) illustrating how the root endophytic bacterial subcommunities influenced yield by responding to soil AN, pH, and MLP (A). The total standardized effects of these factors on yield, α-diversity (Shannon diversity), and β-diversity are shown (B). Red arrows, black arrows, and dashed arrows indicate the positive significant, negative significant, and nonsignificant relationships between different variables. The width of the arrows indicates the strength of the standardized path coefficient. Adjacent values near the arrows indicate path coefficients. r² values indicate the proportion of variance explained by each variable. Significance levels are denoted with *p < 0.05, **p < 0.01, and ***p < 0.001.