The Ability of Different Tea Tree Germplasm Resources in South China to Aggregate Rhizosphere Soil Characteristic Fungi Affects Tea Quality

Abstract

It is generally recognized that the quality differences in plant germplasm resources are genetically determined, and that only a good "pedigree" can have good quality. Ecological memory of plants and rhizosphere soil fungi provides a new perspective to understand this phenomenon. Here, we selected 45 tea tree germplasm resources and analyzed the rhizosphere soil fungi, nutrient content and tea quality. We found that the ecological memory of tea trees for soil fungi led to the recruitment and aggregation of dominant fungal populations that were similar across tea tree varieties, differing only in the number of fungi. We performed continuous simulation and validation to identify four characteristic fungal genera that determined the quality differences. Further analysis showed that the greater the recruitment and aggregation of Saitozyma and Archaeorhizomyces by tea trees, the greater the rejection of Chaetomium and Trechispora, the higher the available nutrient content in the soil and the better the tea quality. In summary, our study presents a new perspective, showing that ecological memory between tea trees and rhizosphere soil fungi leads to differences in plants' ability to recruit and aggregate characteristic fungi, which is one of the most important determinants of tea quality. The artificial inoculation of rhizosphere fungi may reconstruct the ecological memory of tea trees and substantially improve their quality.

Keywords: Camellia sinensis; deep machine learning; ecological memory; nutrient transformation; quality.

Figure 1

Rhizosphere soil fungal abundance analysis of 45 tea tree germplasm resources. F1~F45 denote the numbers of the 45 tea tree germplasm resources (Table S1). Only fungal genera with relative abundance of 1% or more presented in the rhizosphere soils of the 45 tea tree germplasm resources are listed in the figure, and those that did not reach 1% or more are categorized as others. The variation range of microbial abundance indicates the range of minimum to maximum abundance of the rhizosphere soil fungi of the 45 tea tree germplasm resources.

Figure 2

Classification and determination of tea tree germplasm resources based on the abundance of rhizosphere soil fungi. F1~F45 denote the numbers of the 45 tea tree germplasm resources (Table S1). (A) Unsupervised K-means clustering was used for automatic classification based on the abundance of soil fungi in the rhizosphere of tea tree. The three categories were defined as group A, group B, and group C, and their corresponding tea tree germplasm resources are shown in Table S5. (B) Simulated classification of key fungi based on rhizosphere soil fungal abundance of tea tree using the supervised OPLS-DA model (Figure S4).

Figure 3

Validation of classification accuracy of key microorganisms and screening of characteristic microorganisms. F1~F45 indicate the numbers of the 45 tea tree germplasm resources (see Table S1 for the tea tree varieties corresponding to the numbers). Groups A, B, and C denote the three categories into which the 45 tea tree germplasm resources were classified after unsupervised K-means clustering (see Table S5 for the tea tree varieties corresponding to groups A, B, C). KNN, BPNN, SVM, RF, and XGboost denote five types of machine and deep learning, namely K-nearest neighbors, support vector machine, back-propagation neural network, random forest, and extreme gradient boosting. (A) Five machine and deep learning simulations were used to validate the accuracy of the classification based on the abundance of 181 key fungal genera. The overall accuracy of the classification was obtained through the confusion matrix, and the accuracy for groups A, B, and C was obtained through ROC curves (Figure S6). (B) Based on RF and XGboost, used to obtain the importance eigenvalues of different fungi when distinguishing between groups A, B, and C, 25 genera of characteristic fungi were obtained (Figure S7). (C) Based on the abundance of these 25 characteristic fungal genera, five machine and deep learning simulations were used to validate the accuracy of the classification. The overall accuracy of the classification was obtained through the confusion matrix, and the accuracy for groups A, B, and C was obtained through ROC curves (Figure S8). (D) Analysis of differences in abundance of 25 characteristic fungal genera (* indicates that the abundance of the fungus differs at the p < 0.05 level in groups A, B, and C; Figure S9). (E) Differential analysis of the four fungal genera with the highest relative abundance in groups A, B, and C.

Figure 4

Quantitative validation of characteristic microorganisms and determination of classification accuracy. F1~F45 indicate the numbers of the 45 tea tree germplasm resources (see Table S1 for the tea tree varieties corresponding to the numbers). Groups A, B, and C denote the three categories into which the 45 tea tree germplasm resources were classified after unsupervised K-means clustering (see Table S5 for the corresponding tea varieties in groups A, B, C). KNN, BPNN, SVM, RF, and XGboost denote five types of machine and deep learning, namely K-nearest neighbors, support vector machine, back-propagation neural network, random forest, and extreme gradient boosting. (A) Analysis of variance of quantitative data for 20 characteristic fungal genera (* indicates that the number of these fungi that differed at the p < 0.05 level in groups A, B, and C; Figure S10). (B) Based on the quantitative data of the 20 characteristic fungal genera, five machine and deep learning simulations were used to verify the accuracy of the classification. The overall accuracy of the classification was obtained through the confusion matrix, and the accuracy for groups A, B, and C was obtained through ROC curves (Figure S11). (C) Differential analysis of the four fungal genera with the greatest numbers in groups A, B, and C.

Figure 5

Classification of tea tree germplasm resources based on available nitrogen, phosphorus, and potassium content of soil. F1~F45 indicate the numbers of the 45 tea tree germplasm resources (see Table S1 for the tea tree varieties corresponding to the numbers). KNN, BPNN, SVM, RF, and XGboost denote five types of machine and deep learning, namely K-nearest neighbors, support vector machine, back-propagation neural network, random forest, and extreme gradient boosting. (A) Based on the available nitrogen, available phosphorus, and available potassium content of the tea tree rhizosphere soils (Figure S12), unsupervised K-means clustering was used for automatic classification. The three categories were defined as groups A1, B1, and C1, and their corresponding tea tree germplasm resources are shown in Table S6. (B) Five machine and deep learning methods were used to simulate and verify the accuracy of the classification based on the available nitrogen, available phosphorus, and available potassium content of the tea tree rhizosphere soil (Figure S12). The overall accuracy of the classification was obtained through the confusion matrix, and the accuracy for groups A1, B1, and C1 was obtained through ROC curves (Figure S13). (C) Analysis of available nitrogen, available phosphorus, and available potassium content in groups A1, B1, and C1 (* indicates that the difference in the content of this index in groups A1, B1, and C1 reaches the p < 0.05 level; Figure S14).

Figure 6

Classification of tea germplasm resources based on tea polyphenol, theanine, and caffeine content. F1~F45 indicate the numbers of the 45 tea tree germplasm resources (see Table S1 for the tea tree varieties corresponding to the numbers). KNN, BPNN, SVM, RF, and XGboost denote five types of machine and deep learning, namely K-nearest neighbors, support vector machine, back-propagation neural network, random forest, and extreme gradient boosting. (A) Based on the tea polyphenol, theanine, and caffeine content of the tea leaves (Figure S15), unsupervised K-means clustering was used for automatic classification. The three categories were defined as groups A2, B2, and C2, and their corresponding tea germplasm resources are shown in Table S7. (B) Based on the tea polyphenol, theanine, and caffeine content of the tea leaves (Figure S15), five machine and deep learning methods were used to simulate and verify the accuracy of the classification. The overall accuracy of the classification was obtained through the confusion matrix, and the accuracy of groups A1, B1, and C1 was obtained through ROC curves (Figure S16). (C) Analysis of tea polyphenols, theanine, and caffeine content in groups A2, B2, and C2 (Figure S17).

Figure 7

PLS-SME equations were constructed based on the quantitative data of four characteristic fungi genera; the content of soil available nitrogen, available phosphorus, and available potassium; and the content of tea polyphenols, theanine, and caffeine in tea leaves from the rhizosphere soil of 45 tea germplasm resources. Note: *** indicates that the effect value reaches the p < 0.001 level.

Figure 8

The 45 different varieties of tea germplasm resources.