In this study, two cultivars of upland cotton with different levels of resistance were selected: Zhongzhimian 2 (Zhong 2, resistant to V. dahliae) and Xinluzao 36 (Xin 36, highly susceptible to V. dahliae). Two cultivars were cultivated in a random arrangement in the Verticillium wilt disease nursery at the Shihezi Academy of Agricultural Sciences, Xinjiang. The field has a continuous cotton growing history of more than 20 years, with a serious and uniform incidence of Verticillium wilt. In April 2021, the seeds of two cultivars were sown in the field, and 18 plants (9 healthy +9 diseased plants for each cultivar) were randomly uprooted with shovels when plants infected with V. dahliae showed obvious disease symptoms in August 2021 (Supplementary Figure 1). The disease index (DI) was used to evaluate the severity of cotton Verticillium wilt (Zhang et al., 2012) using the following formula: DI = [Σ (disease grades × number of infected plants)/(total checked plants×4)] × 100 (Zhang et al., 2017).

To collect the rhizosphere soil (1–2 mm–thick soil layer surrounding the root after shaking vigorously), the roots were transferred into a 50 mL centrifuge tube containing 15 mL of 1× phosphate buffer solution (PBS), rotated for 5 min and then removed. Next, the tubes were centrifuged at 4000 × g and 4°C for 10 min, and the supernatant was discarded. Then, the samples were centrifuged at 8000 × g for 5 min, the supernatant was discarded again, and the remaining part was regarded as the rhizosphere soil (Edwards et al., 2018).

Total DNA of soil was extracted from 36 rhizosphere soil samples according to the instructions using the DNeasy PowerSoil Kit (QIAGEN, Germany). The full-length bacterial 16S rRNA gene and fungal ITS were amplified by PCR using the bacteria-specific primer pair 27F (5′-AGAGTTTGATCMTGGCTCAG-3′)/1492R (5′-ACCTTGTTACGACTT-3′) (Thies, 2007) and the fungi-specific primer pair ITS1F (5′-CTTGGTCATTTAGAGGAAGTAA-3′)/LR3 (5′-CCGTGTTTCAAGACGGG-3′) (Kurtzman and Robnett, 1998), respectively. PCR procedure: 95°C for 2 min; 30 cycles of 95°C for 30 s, 55°C for 30 s, 72°C for 30 s with a final extension of 72°C for 5 min. PCR products were purified by Gel Extraction Kit (OMEGA, USA). Then, the entire 16S rRNA gene and ITS lengths of the community were determined using the PacBio Sequel platform at Personalbio, Inc. (Shanghai, China).

The sequence data were verified using Quantitative Insights Into Microbial Ecology 2 (QIIME2) and the R software package (version 3.2.0). The QIIME package¹ was used to extract the high-quality sequences, which were then clustered into amplicon sequence variants (ASVs). Taxonomic assignment of 16S rRNA gene and ITS fragment representative sequences was performed based on the Greengenes database (McDonald et al., 2012) and the UNITE database (Abarenkov et al., 2010). Alpha-diversity analyses included Shannon, Chao1, Simpson, Pielou_e and Observed_species. Beta diversity was calculated by the weighted UniFrac distance and then analyzed by principal coordinate analyses (PCoA) (Lozupone and Knight, 2005). The Kruskal–Wallis test and permutational multivariate analysis of variance (PERMANOVA) with 999 random permutations were used to analyze significant differences in alpha diversity and beta diversity (Anderson, 2001). Venn diagrams were implemented online to show unique and shared ASVs.² The abundances of healthy and diseased rhizospheres in two cotton cultivars were statistically compared at different taxonomic levels and visualized by histogram and heatmap. Two-sided analysis of variance with a t test was used for two-group comparison analyses using STAMP (v.2.0.0) (Parks et al., 2014). Co-occurrence network analysis was conducted at the genus level based on Spearman correlation with a threshold of |r| > 0.6 (p < 0.05).

A total of 433,283 high-quality bacterial 16S rRNA reads and 461,403 fungal internal transcribed spacer [ITS] reads were obtained via the PacBio Sequel platform from 18 healthy and 18 diseased samples, respectively. These reads were aggregated into 8,034 bacterial ASVs and 1,667 ITS fungal ASVs.

Venn diagrams show the unique and shared ASVs in the different samples in Figure 1. A total of 1742 bacterial ASVs and 169 fungal ASVs were common to all groups (Figures 1A,B). The healthy rhizosphere of Xin 36 had more unique bacterial (H36, 2072; D36, 1346) and fungal ASVs (H36, 392; D36, 372) than the diseased rhizosphere; the healthy rhizosphere of Zhong 2 had more unique bacterial ASVs than the diseased rhizosphere (H2, 2028; D2, 1562) but fewer unique fungal ASVs than the diseased rhizosphere (H2, 333; D2, 468).

Notably, there was no significant difference in alpha diversity between healthy and diseased rhizospheres of both bacteria and fungi in the two cotton cultivars (Supplementary Figure 3). Principal coordinate analysis (PCoA) was performed based on Bray–Curtis dissimilarity and revealed that bacterial and fungal communities showed no significant difference between healthy and diseased rhizospheres, especially for Xin 36 (Figure 2).

A total of 28 phyla, 72 classes, 133 orders, 266 families, and 505 genera were identified in the bacterial community. In Zhong 2 and Xin 36, the dominant bacterial phyla were Pseudomonadota (42.91%), Acidobacteriota (15.78%), Bacteroidota (6.34%) and Planctomycetota (6.18%) (relative abundance ≥5%), accounting for 71.22% (Figure 3A). At the genus level, most of the bacterial ASVs in Zhong 2 were assigned to Vicinamibacter (5.51%), Bacterium (4.69%), Pseudomonas (3.54%), Lysobacter (3.39%) and Novosphingobium (3.50%), whereas bacterial ASVs in Xin 36 were mainly classified into Pseudomonas (6.84%), Vicinamibacter (5.69%), Bacterium (4.48%), Bacillus (3.40%) and Lysobacter (3.24%) (Figure 3C).

A total of 19 phyla, 52 classes, 94 orders, 160 families, and 219 genera were identified in the fungal community. In Zhong 2 and Xin 36, the dominant fungal phyla were Ascomycota (23.67%), Basidiomycota (14.28%), Mortierellomycota (9.69%), and Rozellomycota (6.04%), accounting for 53.68% (Figure 3B). At the genus level, most of the fungal Zhong 2 ASVs were assigned to Vishniacozyma (7.10%), Olpidium (6.90%), Psathyrella (6.38%) and Cephalotrichum (4.19%), whereas fungal ASVs in Xin 36 were mainly classified into Vishniacozyma (3.06%), Cephalotrichum (4.27%), and Mortierella (3.42%) (Figure 3D).

Based on the heatmaps of the top 40 genera, the differences between the healthy and diseased rhizosphere microbiomes of the two cotton cultivars were compared (Figures 4A,B). According to the relative abundances of these genera, D2 and H36 were clustered together in the bacterial community, followed by D36 and H2, whereas H2 and D2, and H36 and D36 were clustered together in the fungal community.

We further identified the changes between two cultivars in the taxonomic composition of the rhizosphere microbiomes of the two cultivars. Specifically, the relative abundance of 31 and 18 bacterial orders differed significantly in Zhong 2 (Figure 5A) and Xin 36 (Figure 5B) between healthy and diseased rhizospheres, respectively. In Zhong 2, the relative abundances of bacterium WX65 and Fimbriimonadales significantly increased in D2, while the relative abundances of Caulobacterales, Myxococcales and Nevskiales were significantly reduced. Only Chitinophagales decreased significantly in D36. Furthermore, the relative abundance of Entomophthorales in the fungal order differed significantly in Xin 36 between the healthy and diseased rhizospheres (Figure 5C), and there was no significant change in Zhong 2.

The relative abundance of 40 and 32 bacterial families differed significantly in Zhong 2 (Figure 6A) and Xin 36 (Figure 6B) between healthy and diseased rhizospheres, respectively. In Zhong 2, the relative abundance of bacterium WX65 significantly increased in D2, whereas the relative abundances of Caulobacteraceae, Hyphomicrobiaceae and Steroidobacteraceae were significantly decreased. Only Chitinophagaceae decreased significantly in D36. Furthermore, the relative abundance of Ancylistaceae in the fungal family differed significantly in Xin 36 between the healthy and diseased rhizospheres (Figure 6C), and there was no significant change in Zhong 2. These results indicated that V. dahliae infection influenced the distribution and composition of the rhizosphere bacterial and fungal communities of the two cotton cultivars.

The relative abundance of Verticillium was significantly different between healthy and diseased rhizospheres of the two cotton cultivars (p < 0.05), with high enrichment in diseased rhizospheres compared with healthy rhizospheres. Notably, in both healthy and diseased rhizospheres, the relative abundance of Verticillium was higher in Xin 36 than in Zhong 2 (Supplementary Figure 2A). Verticillium dahliae and Verticillium albo-atrum are potential pathogens causing cotton Verticillium wilt according to previous studies (Qin et al., 2006, 2008). In the genus Verticillium, only Verticillium dahliae was identified, and the relative abundance was D36 (0.86%) > H36 (0.14%) > D2 (0.03%) > H2 (0%). This was consistent with the phenotypic results of the field Verticillium wilt investigation (Supplementary Figure 2B).

To determine the effects of V. dahliae on rhizosphere microbial co-occurrence patterns between healthy and diseased samples in two cotton cultivars, the genera with the top 20 relative abundances were screened to construct the networks based on correlation relationships (Figure 7). For bacterial communities, the microbial networks in Zhong 2, neither healthy nor diseased rhizospheres, were denser and had a more complex network than those of Xin 36. In contrast, Xin 36 had a more complex network than Zhong 2 in the fungal communities. The results clearly show that the resistance of cotton cultivars has an effect on the complexity of the rhizosphere microbiome. In addition, the influence of V. dahliae infection on the complexity of rhizosphere bacterial communities was greater than that of fungal communities.

After V. dahliae infection, the unique bacterial and fungal ASVs in the healthy rhizosphere of Xin 36 were higher than those in the diseased rhizosphere, whereas the healthy rhizosphere of Zhong 2 had more unique bacterial ASVs and fewer unique fungal ASVs than the diseased rhizosphere. The results suggested that V. dahliae infection may alter the structure of the rhizosphere microbiome, which is consistent with the result that pathogen infection disrupts host control over the rhizosphere microbiome (Wei et al., 2018; Wen et al., 2020).

The occurrence of diseases is usually accompanied by diversity changes in the rhizosphere microbiome (Wei et al., 2018; Yuan et al., 2018; Shi et al., 2019). However, the analysis results of the alpha and beta diversity showed that there were no significant differences between the resistant cultivar and susceptible cultivar, consistent with a previous study (Fernández-González et al., 2020). Such results may be due to differences in plant host, pathogen, soil, agricultural practices, or environmental conditions (Kwak et al., 2018; Gu et al., 2020; Jiang et al., 2021).

This study demonstrated that the relative abundance of many rhizosphere microorganisms in the healthy rhizosphere differed from that in the diseased rhizosphere of each cultivar. The taxonomic composition of the rhizosphere bacterial communities showed that Pseudomonadota, Acidobacteriota, Bacteroidota and Planctomycetota were dominant in two cotton cultivars, with Pseudomonadota members accounting for 42.91% of the community composition (Trivedi et al., 2020). Zhong 2 was characterized by a higher relative abundance of Actinomycetota than Xin 36, which is known as a biocontrol microorganism (Lee et al., 2021). In addition, the rhizosphere fungal community is mainly composed of Ascomycota and Basidiomycota, which are the most abundant phyla observed in previous studies (Bálint et al., 2015; Coleman-Derr et al., 2016).

Compared with the rhizosphere microbial communities in the healthy rhizosphere of Zhong 2 and Xin 36, Pseudomonas and Acidobacteria bacterium LP6 in the bacterial communities and Cephalotrichum and Mortierella in the fungal communities were both highly enriched in the diseased rhizosphere. Pseudomonas abundance in diseased rhizospheres was significantly higher than that in the healthy rhizosphere regardless of the cultivar type. Consistently, a previous study found that Rhizoctonia solani invasion alters the rhizosphere microbial community and specifically accumulates beneficial Pseudomonas (Yin et al., 2021). Therefore, it is reasonable to speculate that the increase in Pseudomonas in diseased rhizospheres may contribute to the potential resistance of their host plants to Verticillium wilt. These results suggest that plants may increase the enrichment of specific microbiomes in response to pathogen infection (Busby et al., 2016), which can be used as antagonistic candidates for Verticillium wilt and need to be confirmed by further culture-based experiments.

Different microorganisms are recruited by plants to shape their rhizosphere microbiome, and the rhizosphere microflora community structure of the same species changes due to genotype differences (Bressan et al., 2009; Zhang et al., 2021; Yue et al., 2023). The growth and development of different blueberry cultivars were enhanced by recruiting specific rhizosphere microflora based on genotype (Jacoby et al., 2017). Our results showed that the different genotypes of cotton recruited specific rhizosphere microbiomes, suggesting that the rhizosphere microbial community was regulated by host genotypes, consistent with the results in soybean, rice, and barley (Bulgarelli et al., 2015; Singh et al., 2022; Qu et al., 2023). In addition, resistant cultivars may resist pathogen invasion by enriching specific bacterial or fungal groups in the rhizosphere (Mendes et al., 2017; Kwak et al., 2018). The analysis of co-occurrence networks also showed that the resistant cultivar Zhong 2 has a more complex network than the susceptible cultivar Xin 36 in the bacterial communities, and V. dahliae has a significant impact on the bacterial community structure compared with fungal communities.