Bacterial community composition
From 36 soil specimens, 3,370,643 high-quality 16S rDNA reads were obtained [(CK1, T1, CK2, and T2 treatments) × 9 replicates] with 74,955–103,931 sequencing reads (mean = 91,908) per sample. The maximum read length was 478 bp and the minimum average length was 341 bp for the 16S rDNA genes. All rarefaction curves for the bacterial samples revealed that the amount of recorded OTUs was generally 7,000 reads per plateau, indicating the assessment adequately covered the microbial variety (see Supplementary Fig. S2a). The bacterium richness (Chao1 and ACE), evenness indexes (Shannon and Simpson), and number of OTUs between CK1 and T1 as well as between CK2 and T2 were not significantly different (see Supplementary Table S1a).
The soil bacterial composition of the two treatment groups at the two growth periods were compared at the level of the phylum. A total of 26 bacterial phyla were identified, with the exception of 1.03% of unclassified sequencing reads. The main phyla of the sequenced bacteria were Proteobacteria, Actinobacteria, and Gemmatimonadetes, which occupied over 64.5% of the total bacterial populations in the sample sequences. Chloroflexi, Acidobacteria, Bacteroidetes, Parcubacteria, Verrucomicrobia, and Firmicutes were also identified at relatively elevated richness (average relative abundance > 1%) (Fig. 1a). Wheat straw addition significantly increased the relative abundances of Actinobacteria, Chloroflexi, and Saccharibacteria, while significantly decreasing the relative abundance of Parcubacteria at both moments of sampling (P < 0.05) in the consecutive watermelon monoculture system. The abundance of Gemmatimonadetes was visibly more elevated in T1 than in CK1 and no significant differences between T2 and CK2 (P > 0.05) were detected (see Fig. 1a and Supplementary Table S2a).
Major bacterial (a) and fungal phylum (b) relative abundance in the soil with (T1 and T2) and without (CK1 and CK2) wheat straw addition. Bacterial phyla with > 1% and fungal phyla with > 0.1% average relative abundances. Others included bacterial phyla below 1% relative abundance and unidentified bacterial and fungal phyla. According to the Student’s t-test (n = 9), * and ** represented P < 0.05 and P < 0.01 between soil with (T1 and T2) and without (CK1 and CK2) wheat straw addition, respectively.
Fungal community composition
For fungi, 4,991,373 high-quality fungi ITS gene sequence reads were derived from the 36 soil samples tested with 89,803–194,007 sequence reads (mean = 138,649) per sample. The maximum read length was 499 bp and the minimum average length was 201 bp for the ITS genes. At a 97% sequence similarity for all samples, the OTU rarefaction curves appeared to reach the saturation plateau (see Supplementary Fig. S2b), indicating the assessment has covered the microbial variety. The fungal ACE and Chao1 indexes were significantly higher in T1 than in CK1, whereas the differences were not significant between CK2 and T2. Moreover, the fungal Shannon index was significantly higher in T2 than in CK2, and no significant differences were detected between T1 and CK1. The two treatments and two watermelon growth stages did not differ in the amount of OTUs (see Supplementary Table S1b).
In all soil specimens, Ascomycota and Basidiomycota were the two prevalent fungal phyla, accounting for over 89% of the total reads. Wheat straw addition significantly enhanced the relative richness of Ascomycota while considerably reducing the relative abundance of Basidiomycota during the flowering stage in the consecutive watermelon monoculture system. On the contrary, the relative abundance of Ascomycota was greater without wheat straw compared to that with wheat straw, whereas the relative abundance of Basidiomycota was lower without wheat straw compared to that with wheat straw in the fruiting stage. The relative abundance of Chytridiomycota, Mortierellomy, Mucoromycota, and Glomeromycota was not significantly different with and without wheat straw (see Supplementary Fig. 1b and Table S2b).
Taxonomic characteristics of bacterial and fungal communities
LEfSe was further used to assess the association of the bacterial genera (average relative abundances > 0.1%) in the microbial community in the soil with (T1 and T2) and without (CK1 and CK2) wheat straw addition in the flowering and fruiting stages. The bacterial genera, Sporosarcina, Nitrospira, Truepera, Acidovorax, Woodsholea, Polycyclovorans, Actinomadura, and Chryseolinea were more abundant in CK1 than in T1 (Fig. 2a). Sphingobium, Phaselicystis, Flavobacterium, and some Actinobacteria (Sporichthya, Aeromicrobium, Nonomoricola, Marmoricola, Ilumatobacter, Streptomyces, Agromyces, Patulibacter, Amycolatopsis, and Nocardioides) were more abundant in T1 than in CK1 (Fig. 2a). Actinobacteria was overdominant in T1, taking up 4.30% of the relative abundance with > 0.1% bacterial reads in T1 soil (only 0.04% in CK1 soil) (see Supplementary Table S3a). In addition, 14 bacterial genera with relative abundance > 0.1% were more prevalent in CK2 samples, including Spororosarcina, Chryseollinea, Nitrososporia, Truepera, Actinomadura, two Planctomycetes (SM1A02 and I-8), and some Ptoteobacteria (Sulfurifustis, Polycyclovorans, Woodsholes, and H16) (Fig. 2b). In contrast, the Actinobacteria (Aeromicrobium, Nonomured, Nocardioides, Dactylosporangium, and Ilumatobacter) group and Proteobacetia (brachysporum_group, Ramlibacter, Dongia, Hyphomicrobium, Rhizobium, Sphingobium, Parablastomonas, Pseudoxanthononas, Dokdonella, and Pseudohoniella) group were more abundant in the T2 soil than in CK2 (see Fig. 2b and Supplementary Table S3b).
LDA histogram scores for bacterial genera with different abundance for the flowering stage (a) and fruiting stage (b) in the watermelon monoculture system.
With respect to the fungal genera, LEfSe analysis showed that there was higher relative abundance of Schizothecium, Entoloma, Preussia, Lecanicillium, and Bipolairs in T1, whereas there was a higher relative abundance of Thanatephorus, Scopulariopsis, Fusarium, and Conocybe in CK1 (Fig. 3a) for the flowering stage. In the fruiting stage, Psathyrella, Filobasidium, Aphanoascus, Cladosporium, Microascus, and Scopulariopsis were found in CK2, while only Schizothecium was found in T2 (Fig. 3b). Furthermore, the second prevailing genus, Fusarium, accounted for 9.70% of all fungal genera in CK1 (only 0.64% in T1). The relative abundance of Fusarium was higher in CK2 than in T2 (see Supplementary Table S4).
LDA histogram scores for fungal genera with different abundance for the flowering stage (a) and fruiting stage (b) in the watermelon monoculture system.
Microbial community variety and the link between genera abundance and environmental conditions
Non-metric multidimensional scaling (NMDS) clearly indicated that there were considerable variations in the composition of the soil bacterial populations between the samples with (T1 and T2) and without (CK1 and CK2) wheat straw addition in the consecutive watermelon monoculture system in the two growth stages evaluated (Fig. 4a). In the NMDS plot, the nine replicates in the same groups were not closely located for the fungal communities in all the samples, which indicated that there was no distinct difference in fungal community composition between two treatments for the two growth stages (Fig. 4b). RDA revealed that the relative abundances of Planctomyces, Pirellula, and Exiguobacterium were positively correlated with the DI for bacteria (Fig. 5a and Supplementary Table S5a). The relative abundances of Aspergillus, Fusarium, Sopulariopsis, Cladosporium, and Aphanoascus were positively correlated with the DI for fungi (Fig. 5b and Supplementary Table S5b).
Non-metric multidimensional scaling (NMDS) according to the Euclidean distance plot of bacterial (a) and fungal (b) microbiota in the flowering stage (CK1 and T1) and fruiting stage (CK2 and T2).
Plots of redundancy analysis (RDA) ordination displaying the interactions between the top 10 bacterial (a) and fungal genera (b) and soil environmental variables. AP denotes available phosphorus; pH denotes the solar pH; EC denotes electrolyte conductivity; the disease index (DI) denotes healthy plants as “0”and Fusarium wilt plants as “1”. CK1 represents the soil without wheat straw addition at the watermelon flowering stage while T1 represents the soil with wheat straw addition at the watermelon flowering stage; CK2 represents the soil without wheat straw addition at the watermelon fruiting stage and T2 represents the soil with wheat straw addition at the watermelon fruiting stage.
Fungal community network analysis
Soil microbial network analysis is widely performed to understand the taxonomic and functional relations within complex microbial communities13. With respect to fungi, the top 300 OTUs of T1 and CK1 soil at the watermelon flowering stage were chosen for pMEN analysis (Fig. 6). The T1 network consisted of 180 nodes (OTUs), 1,036 connections, and 12 modules, with an average connectivity of 11.511, average path length of 2.999, and clustering coefficient of 0.278. The CK1 network consisted of 166 nodes, 741 connections, and 18 modules, with an average connectivity of 8.927, average path length of 2.920, and clustering coefficient of 0.155. The modularity proportion was higher in the T1 network, although fewer total modules were recognized (Table 1). Strikingly, there were more links in T1 soil (1,036 links) than in CK1 soil (741 links). The positive link/negative link ratio (P/N) was higher in T1 soil (P/N = 0.333) than in CK1 soil (P/N = 0.211), demonstrating that the T1 soil had more complex and positive co-occurrence correlations than the CK1 soil.
Network plots of fungal community at the order level from soil without (CK1) (a) and with (T1) (b) wheat straw addition at the watermelon flowering stage. The size of the node corresponds to the relative abundance of the OTUs. The node colors show various phylogenetic associations. Node (edge) connection lines represent co-occurrence with positive (blue) and negative (red) correlations.
The top 300 OTUs of T2 and CK2 soil at the watermelon fruiting stage were also chosen for pMEN analysis (see Supplementary Fig. S3). The T2 network consisted of 202 OTU nodes, 1,040 connections, and 14 modules, with an average connectivity of 12.545, average path length of 2.727, and clustering coefficient of 0.214. The CK2 network consisted of 181 nodes, 739 links and 15 modules, with an average connectivity of 8.166, average path length of 3.186, and clustering coefficient of 0.264 (Table 1). Strikingly, there were more links in T2 soil (1,040 links) than in CK2 soil (739 links), which indicated that the T2 soil had more complex and stable microbial networks than the CK2 soil. In T2 soil, the P/N (P/N = 0.218) was lower than that in CK2 soil (P/N = 0.451), indicating that the T2 soil had more negative co-occurrence relationships in the microbial community than those in CK2 soil.
In addition, CK1 and T1 networks shared 49 nodes (Fig. 7). Nodes of the Sordariales, Onygenales, Microascales, Hypocreales, Eurotiales, Agaricales, and Arachnomycetales genera dominated in both networks. The relative abundance of Sordariales and Hypocreales was higher in the two networks. Furthermore, there was a higher proportion of Sordariales-affiliated OTUs and a lower proportion of Hypocreales-affiliated OTUs in T1 compared to CK1 (Fig. 7a). However, 72 nodes were shared between CK2 and T2 networks. Nodes belonging to the Sordariales, Pleosporales, Onygenales, Microascales, Hypocreales, Eurotiales, and Agaricales genera dominated in both networks. Sordariales, Hypocreales, and Agaricales were relatively more abundant in these two networks. However, the relative abundance of Sordariales and Hypocreales had more significant differences in CK2 compared to T2. There was also a higher proportion of Sordariales-affiliated OTUs and a lower proportion of Hypocreales-affiliated OTUs in T2 compared to CK2 (Fig. 7b). These network analysis results suggested that Sordariales dominated in the T1 and T2 soils, which were treated with wheat straw, while Hypocreales dominated in the soil (CK1 and CK2) without wheat straw addition.
Relative abundance of nodes at the order level in modules inside the fungal network created from the flowering stage (a) and fruiting stage (b). Venn diagrams display the amount of shared and unshared network nodes in the soil sample with and without wheat straw addition.
Bacterial community network analysis
T1 and CK1 soil bacterial community analysis revealed similar sized networks with 224 and 221 nodes, respectively (see Supplementary Fig. S4 and Table S6). The average connectivity for the T1 and CK1 networks was 7.482 and 7.493, with an average path length of 4.273 and 3.557, respectively. The average clustering coefficient value (0.323 or 0.326) was comparable in the T1 and CK1 soil networks, while modularity was somewhat lower in the T1 network (0.403) than in the CK1 network (0.440). However, the number of modules in T1 (27) was higher than that in CK1 (22). In T1 and CK1 soil, the total number of links was 828 and 838 (P/N = 2.33 for T1 soil and P/N = 2.37 for CK1 soil), respectively. However, T2 and CK2 soil had different nodes (228 and 201, respectively) at the watermelon fruiting stage. Furthermore, the average connectivity was higher in T2 (4.263) than in CK2 (3.562) networks. The average path length, average clustering coefficient value, and modularity were higher in CK2 than in T2 networks (see Supplementary Fig. S5 and Table S6). In T2 and CK2 soils, the total number of links was 486 and 358 (P/N = 1.612 for T2 soil and P/N = 1.732 for CK2 soil), respectively. The data suggested that wheat straw addition did not affect the bacterial co-occurrence relationship in both the watermelon flowering and fruiting stages.
Inside the T1 versus the CK1 network, a greater percentage of OTUs associated with Proteobacteria, and a reduced OTU ratio for Chloroflexi, Bacteroidetes, and Acidobacteria were found. (see Supplementary Fig. S6a). However, a higher proportion of Proteobacteria and Actinobacteria– affiliated OTUs and a lower proportion of Gemmatimonadetes, Euryarchaeota, Chloroflexi, and Bacteroidetes-affiliated OTUs inside the modules were identified in the T2 versus CK2 network (see Supplementary Fig. S6b).
Source: Ecology - nature.com