Rhizospheric soil physicochemical properties of sweet potato
After sweet potato continuous cropping two years, available Mn in the X18 and Y138 rhizospheric soil respectively decreased by 32.68% and 31.14% at the beginning of planting, and decreased by 27.35% and 31.10% in pre-harvest period. (Table 1). The soil pH of X18 and Y138 respectively decreased by 2.72% and 3.11% at the beginning of planting, the change was not significant in pre-harvest period. Available Ca in X18 and Y138 rhizospheric soil respectively increased by 29.80% and 38.97%, available Zn increased by 56.11% and 43.19% at the beginning of planting, and available Ca increased by 30.75% and 26.47%, respectively; available Zn increased by 29.46% and 30.81% in pre-harvest period. Available Fe of X18 and Y138 respectively decreased by 18.61% and 17.08% in pre-harvest period, the change was not significant at the beginning of planting. Available B in the X18 rhizospheric soil decreased by 20.63% at the beginning of planting, while the change of Y138 was not significant.
Rhizospheric soil bacterial α-diversity
The average coverage of all samples was 96.19% (Table 2). Rarefaction curves closed to plateau (Fig. 1), indicating that our sequencing depth was good. The reads ranged from 24695 - 37688 for samples, and the OTUs ranged from 3137 - 3734. Chao, Shannon and Simpson indexes were computed based on the bacterial OTU. Unlike Shannon index, the larger the Simpson value, the lower the community diversity. Chao and Shannon values of X18 and Y138 were higher in pre-harvest period than that at the beginning of planting, and they were the opposite of the Simpson index, indicating that two communities had higher species richness and diversity in pre-harvest period. At the same time, X18 had higher Chao and Shannon indices than Y138, which were contrary to the Simpson index. In other words, the bacteria richness and diversity in rhizospheric soil of X18 were higher than those of Y138.
Community composition analysis of rhizospheric soil bacteria
At the phylum level (Fig. 2), X18 and Y138 rhizospheric soil bacteria mainly belonged to Proteobacteria (28.5%-34.9%), Acidobacteria (10.4%-21.1%), Actinobacteria (11.3%-18.1%), Planctomycetes (5.2%-9.9%), Chloroflexi (4.6%-9.1%), Bacteroidetes (3.4%-6.1%), Gemmatimonadetes (3.0%-7.4%), and Firmicutes (1.4%-10.9%). Among them, Proteobacteria was the most prevalent, and Acidobacteria and Actinobacteria were next.
After X18 and Y138 continuous cropping, Proteobacteria decreased by 17.30% and 8.05% in pre-harvest period, respectively. Acidobacteria showed a decreasing trend and finally increased slightly, while Actinobacteria showed the opposite trend. Firmicutes was higher at the beginning of planting than that in pre-harvest period, while the change of Planctomycetes was opposite. Further, the content of Chloroflexi and Gemmatimonadetes showed an increasing trend. In X18 and Y138 rhizospheric soil, Chloroflexi respectively increased by 81.09% and 96.69%, and Gemmatimonadetes increased by 103.11% and 122.56%, respectively. In addition, compared with X18, Gemmatimonadetes in Y138 rhizospheric soil was higher, especially in 2016.
At the genus level (Fig. 3), Subgroup 6_norank (6.59% - 14.74%), Nitrosomonadaceae_uncultured (1.83%-6.40%), Anaerolineaceae_uncultured (1.75%-3.63%) were the top three dominant bacteria genus in all rhizospheric soils of X18 and Y138, other major genus included Bacillus (0.65%-4.14%), MSB-1E8_norank (0.87%-3.83%), Tepidisphaeraceae_norank (1.71%-2.56% ), Xanthomonadales_norank (0.62%-2.08%), and Lysobacter (0.55%-2.06%). After two years of continuous cropping, Subgroup 6_norank in the X18 and Y138 rhizospheric soil showed a decreasing trend, respectively decreased by 54.34% and 52.66%, and then increased slightly in pre-harvest period in 2016. However, Nitrosomonadaceae-uncultured and Anaerolineaceae-uncultured were present at low levels at the beginning of planting, while increased in pre-harvest period. Bacillus and Lysobacter showed the opposite trend. Moreover, in every sampling period, Lysobacter in rhizospheric soil of X18 was higher than that of Y138. Bacillus was the same as Lysobacter, except for the beginning of planting in 2015. In addition, in the second year of continuous cropping, Lysobacter in X18 and Y138 rhizosphere soil was 1.3 times and 2.4 times of the reduction in the first year.
Venn analysis of rhizospheric soil bacteria
Venn diagrams directly showed the overlapped and unique OTUs of all samples. (Fig. 4). After two years of continuous cropping, the OTUs shared by all samples was 507. In the four sampling periods, there were 95, 158, 127, and 202 unique OTUs in the rhizospheric soils of X18. However, the unique OTUs in the rhizospheric soils of Y138 were 89, 124,141, and 159, respectively. As continuous cropping year increased, the number of specific OTUs for X18 and Y138 tended to increase. The number of OTUs specific to X18 was more than that of Y138 (except for the beginning of planting in 2016), indicating that continuous cropping led to changes of bacterial communities in X18 and Y138 rhizosphere soil. Further, the differences were largest during the early harvest period of 2016.
Heatmap and clustering analysis and PCA of rhizospheric soil bacteria
The results of heatmap and clustering analysis for 40 phyla of all samples were illustrated in Fig. 5. The difference in rhizospheric soil bacterial composition between X18 and Y138 could be seen more clearly. Furthermore, the clustering results showed that the samples grouped into 2 clusters and samples from the same consecutive cropping time were gathered together. In addition, X18 and Y138 from the same sampling period grouped together.
The OTUs of X18 and Y138 were subjected to PCA. The extracted two principal components explained 72.48% of the variation in total (Fig. 6). With continuous cropping time increased, samples from different sampling time were far apart. However, at the same sampling time, X18 and Y138 samples were relatively close to each other. As continuous cropping year increased, the distance between these samples also gradually increased, which indicated that differences between their bacterial communities were also increasingly large. These results were consistent with the results of heatmap and cluster analysis in Fig. 5. Overall, the results suggested that (i) continuous cropping led to bacterial community structure changes in X18 and Y138 rhizosphere soil; (ii) rhizospheric soil bacterial community structures of X18 and Y138 were similar in the same sampling period.
Relationship between bacterialphyla and rhizospheric soil physicochemical characteristics of sweet potato
The results of RDA on top ten bacterial phyla and rhizospheric soil environmental factors of X18 and Y138 were showed in Fig. 7. RDA1 and RDA2 explained respectively 56.57% and 28.05% of the total variation. These effects of soil properties on bacterial community structure were found in the following order: soil pH > Ca > Mn > Zn > B > Fe. The results showed that soil pH (r2=0.9737, Pr=0.004), available Ca (r2=0.8815, Pr=0.011) were significantly correlative with the bacterial community. It indicated that pH was a strong predictor of rhizospheric soil bacterial community for X18 and Y138.
In addition, the results of Spearman’s correlation coefficient analysis were as follows (Fig. 8): pH was positively correlated with Planctomycetes (R = 0.97) and Acidobacteria (R = 0.93), but had a negative correlation with Actinobacteria (R = −0.79) and Firmicutes (R = −0.72); available Ca was positively related to Actinobacteria (R=0.89) and Gemmatimonadetes (R=0.86), and was inversely correlated with Acidobacteria (R = −0.79), Planctomycetes (R= −0.75) and Nitrospirae (R= −0.72). At the same time, it can be seen from the Fig.8 that the soil physicochemical properties were divided into two groups, with available Ca and available Zn clustered into one group and the rest clustered into another, indicating that available Ca and available Zn had a similar effect on the bacteria but different from the rest.