Bacteria are more sensitive to nitrogen fertilizer application in tea plantation soil while fungi are more correlated to tea yield and quality

Soil in tea plantations is characterised by severe acidication and high aluminium and uorine content. Applying excessive nitrogen (N) is a common strategy in tea plantations. Fungal and bacterial responses to N fertiliser addition in tea plantations, especially their relationship with tea growth, quality, and soil microbiome composition, remain unclear.


Introduction
Tea (Camellia sinensis) is mainly distributed in tropical and subtropical areas with acidic soils (Yan et al., 2018). The unique avour and health bene ts of tea have increased its demand, leading to an expansion of tea-farming area (FAO, 2019; Musial et al., 2020). As a leaf-harvest crop, tea requires more nitrogen (N) to meet the demand for leaf growth and metabolite synthesis than cereal crops, such as maize, rice, and wheat (Tang et al., 2020). Chemical fertiliser application is a practical method for improving tea yield and quality (Wang et al., 2020). In practice, excessive fertiliser (especially N fertiliser) is applied to tea plantations (Ni et  . Bacterial beta diversity and the stochasticity ratio rst increase and then decrease with an increasing N fertiliser rate, whereas fungal beta diversity is not in uenced and the stochasticity ratio decreases with an increasing N fertiliser rate in temperate steppes . However, a study on temperate grasslands suggested that the fungal community is more sensitive to N addition than the bacterial community (Widdig et al., 2020).
Tea plantation soil is highly acidic; the characteristics high aluminium and uorine content shape the unique soil microbial community of tea plantations (Shu et  Therefore, in the context of global soil acidi cation, tea plantation soil is optimal for studying the impact of acidi cation and soil microbial characteristics under highly acidic conditions (Guo et al., 2010). In a subtropical tea plantation, soil fungal community structure was shown to be signi cantly altered and fungal diversity decreased under higher N input due to a shift in soil and pruning characteristics Investigating the assembly process of the microbial community can provide new insights into microbial community succession and assist in exploring microbial function (Stegen et al., 2013;Zhou and Ning, 2017). The soil microbe assembly process is signi cantly correlated with soil properties, climate change, and land-use history (Zhang et al., 2016;Li et al., 2018;Shi et al., 2018). However, to our knowledge, the assembly process of the microbiome distributed in tea plantation soil, especially the effect of N fertiliser application on microbial assembly, has not been investigated.
In this study, a eld experiment with different N fertiliser application rates was commenced in 2016 in a tea-producing area of China. Tea and soil samples were collected in 2020 to evaluate the following: (1) bacterial and fungal community succession under N fertiliser application conditions, (2) factors driving microbial community, (3) consistency of assembly processes of bacteria and fungi under N fertiliser application, and (4) relationship between bacteria and fungi with tea yield and quality.

Field experiment design
The eld experiment was conducted in Shaoxing City, Zhejiang Province, China (29°5616″N, 120°4145″E) during 2016-2020. The experimental site is located in a subtropical monsoon climate region with an annual average temperature of 16.4°C and mean annual precipitation of 1,300 mm. The soil is Haplic Acrisol, according to the Food and Agriculture Organization standards. The initial physicochemical characteristics of soil were pH 3.98, 2.75% soil organic carbon (SOC), 0.17% total nitrogen (TN), 146.04 mg kg − 1 alkali-hydrolysable N (AN), 182.67 mg kg − 1 available phosphorus (AP), and 75.78 mg kg − 1 NH 4 OAc-K (AK). The experimental treatments were set as follows: N0 (N de ciency treatment: 0 kg N ha − 1 y − 1 ), N1 (low N application rate: 120 kg N ha − 1 y − 1 ), N2 (optimal N application rate: 360 kg N ha − 1 y − 1 ), and N3 (excessive N application rate: 600 kg N ha − 1 y − 1 ) (Ruan et al., 2020). P 2 O 5 and K 2 O application rates were 80 and 100 kg ha − 1 y − 1 , respectively, for all treatments. Urea, single superphosphate, and potassium sulphate were selected as N, P, and K fertilisers, respectively. N fertiliser was applied as a base fertiliser at the rate of 30% for 2 months after autumn tea harvest, 30% for 1 month before spring tea harvest, 20% for 1 month before summer tea harvest, and 20% for 1 month before autumn tea harvest. P and K fertilisers were applied as base fertilisers for 2 months after the autumn tea harvest. All fertilisers were applied at a soil depth of 15 cm between tea rows and covered with soil. The four treatments were conducted in a 2 × 18 m 2 experimental plot with three replicates for each treatment based on a randomised block design.

Tea sampling and analysis
Tea leaves with one bud and two young expanding leaves were harvested in April, July, and September 2020 as spring, summer, and autumn teas, respectively. Tea yield was determined using a 30 × 30 cm stainless-steel frame, and adjusted to 75% of water content. The tea samples were processed for enzyme inactivation immediately after harvest. Subsequently, the samples were oven-dried at 70°C for 48 h before being sieved through a 250-µm screen. Finally, dry samples were extracted with boiling water to evaluate tea quality. Samples were incubated with ninhydrin to calorimetrically quantify the total amount of free amino acids (AAs) in tea leaves at 570 nm using a spectrophotometer (UV-1780, Shimadzu, Japan).

Soil sampling and analysis of physicochemical characteristics
Soil samples from topsoil (0-25 cm) and subsoil (25-50 cm) were collected after the spring tea harvest in April 2020. The soil samples were divided into two parts: the rst was air-dried to determine its physicochemical properties and the second was stored at − 80°C for the analysis of soil bacteria and fungi. The pH of soil extracted with deionised water (soil:deionised water = 1:5) was measured using the Delta 320 pH meter (Mettler-Toledo Instruments Co., Shanghai, China).

Statistical analysis
All data analyses were performed using R (version 4.0.5). The Shapiro-Wilk test was used to assess normality prior to testing signi cance. Differences in soil and tea characteristics between treatments were analysed using one-way analysis of variance (ANOVA) and the least signi cant difference test (P < 0.05).
Interactions between harvest seasons and treatments were investigated using two-way ANOVA. Random forest analysis was performed using the 'randomForest' package in R. Differences in bacterial and fungal community structures were analysed using non-metric multidimensional scaling (NMDS) based on the Bray-Curtis dissimilarities. Canonical redundancy analysis (RDA) and Mantel ' s test were implemented using the 'vegan' package in R to investigate the contribution of soil properties to bacterial and fungal communities. Partial least squares path modelling (PLS-PM) was performed using the 'plspm' package in R to investigate the relationships between N fertiliser application rate, soil characteristics, and bacterial and fungal community diversity. The contributions of neutral processes to the assembly of bacterial and fungal communities in soil were evaluated using the neutral community model 3 Results

Effect of N fertiliser application on the physicochemical characteristics of soil
N addition altered the physicochemical characteristics of topsoil and subsoil (Table S1)

Tea yield and quality under different N fertiliser application rates
Compared with under N0 conditions, tea yields signi cantly increased in spring, summer, and autumn under N2 and N3 conditions (P < 0.05) (Fig. 1A). In all harvest seasons, tea yield increased with N application rate increasing until the N application rate > 360 kg ha − 1 y − 1 . Annual yield was de ned as the total yield of spring, summer, and autumn tea. Under N1, N2, and N3 conditions annual tea yield increased by 16.4%, 37.0%, and 33.6%, respectively, compared with under N0 conditions. The results of two-way ANOVA analysis indicated that the N fertiliser application rate and the harvest season had signi cant effects on tea yield. Figure 1. Tea yield and quality under different N fertiliser rates in spring, summer, and autumn. Standard deviations of three replicates are represented as error bars. Different letters above the bars indicate signi cant differences (P < 0.05) between N fertiliser treatments in the same harvest season. N0, N1, N2, and N3 refer to N fertiliser application rates of 0, 120, 360, and 600 kg ha − 1 y − 1 for 5 years.
Under N1, N2, and N3 conditions, AA content signi cantly increased (P < 0.05) by 11.6%, 17.2%, and 17.2%, respectively, for spring tea compared with that under N0 conditions, whereas the N fertiliser application rate did not have a signi cant effect on the AA content in summer or autumn tea (Fig. 1B).
When the N fertiliser application rate was > 360 kg ha − 1 , the AA content no longer increased with an increase in the N fertiliser application rate. For all harvest seasons, the highest TP content was observed under N0 conditions (Fig. 1C)

Bacterial and fungal community succession under N fertiliser application
The alpha diversities of the bacterial communities decreased with increasing N application rates ( Fig. 2A and 2B). Excessive N fertiliser application signi cantly decreased fungal Chao1 index of topsoil (P < 0.05) (Fig. 2C). Two-way ANOVA analysis suggested that N application signi cantly altered the alpha diversities of bacterial and not fungal communities (Fig. 2F). The soil sampling layer had a signi cant effect on the Chao1 index of the fungal community. Pearson's correlation analysis results also indicated that the alpha diversities of soil bacteria were more sensitive to soil physicochemical characteristics than soil fungi. Chao1 and Shannon indices of soil bacteria were positively correlated with soil pH, SOC, TN, and AK contents (P < 0.05). However, only soil NH 4 + -N and AN content were negatively correlated with the Chao1 and Shannon indices of the fungal community, respectively. properties and alpha diversity of bacteria and fungi. Red and blue boxes indicate positive and negative correlations between alpha diversity index and soil properties. (F) Two-way ANOVA analysis results; *P < 0.05, **P < 0.01, and ***P < 0.001. SOC: soil organic carbon; TN: total nitrogen; AN: alkali-hydrolysable N; AP: available phosphate; AK: available potassium.
N addition altered the relative abundances of the dominant bacterial and fungal phyla present in soil (Fig. 3). The phyla Proteobacteria, Acidobacteria, Chloro exi, and Actinobacteria accounted for up to 75% of soil bacteria (Fig. 3A) Basidiomycota, Mortierellomycota, and Ascomycota contributed to up to 35% of soil fungi (Fig. 3C). The relative abundance of Proteobacteria increased with increasing N application rates. The top 10 phyla of soil bacteria were more sensitive to N fertiliser application and the sampling layer than those of soil fungi ( Fig. 3B and 3D). Results from Pearson's correlation analysis suggested that the dominant bacterial phyla were more sensitive to soil physicochemical properties than dominant fungal phyla ( Fig. 3B and 3D). The phyla Acidobacteria, Actinobacteria, Planctomycetes, and Firmicutes were positively and Proteobacteria and WPS-2 were negatively correlated with soil nutrient content. The fungal phyla Cercozoa and Zoopagomycota were negatively and Mortierellomycota and Rozellomycota were positively correlated with soil nutrient content. NMDS results indicated that soil bacterial and fungal community structures were signi cantly altered after N addition ( Fig. 4A and 4B). Bacterial and fungal community structures distributed in the topsoil and subsoil were signi cantly different (P < 0.05), and were signi cantly correlated with soil pH (P < 0.05) ( Fig. 4C and 4D).

Driving factors and assembly of soil bacterial and fungal communities
RDA results revealed that soil pH, SOC, AN, AK, NH 4 + -N, and NO 3 − -N were the main factors driving the bacterial community (Fig. 5A). The soil fungal community was signi cantly correlated with soil pH, SOC, C/N, AK, NH 4 + -N, and NO 3 − -N (Fig. 5B). Mantel's analysis showed that soil pH, SOC, TN, AN, and AK had stronger correlations with the bacterial community than with the fungal community (Fig. 5C).
However, the soil fungal community was more sensitive to soil NH 4 + -N and AP contents. Results from PLS-PM analysis also suggested that bacterial and fungal diversity was negatively correlated with N fertiliser application (P < 0.05), whereas bacterial and fungal communities were positively correlated with the N fertiliser application rate (P < 0.05) (Fig. 6). N addition in uenced the bacterial community by indirectly altering the soil conditions (SOC, TN, and AK) and pH.
A higher R 2 of subsoil suggested that bacteria and fungi distributed in subsoil tted the NCM better than that in the topsoil, which indicated that the stochastic processes of bacterial and fungal communities increased with the increase in the soil depth (Fig. S2). The migration rates of bacterial and fungal communities in the subsoil were higher than those in the topsoil. Overall, compared with the fungal community, the bacterial community was shaped more by stochastic processes. The migration of the fungal community was more restricted than the bacterial community. Cluster analysis classi ed N0 and N1 samples of topsoil and subsoil into one category and N2 and N3 of topsoil into another category, according to soil physicochemical properties (Fig S3). To investigate the impact of N on bacterial and fungal community assemblies, N0 and N1 samples were clustered into a low N fertiliser application rate group, whereas N2 and N3 samples were clustered into a high N fertiliser application rate group (Figs. S4 and S5). The migration rates of bacterial and fungal communities were restricted after high N fertiliser application compared to low N fertiliser application.

Random forest analysis of tea yield and quality
Random forest analysis was applied to predict the driving factors of tea yield and quality (Fig. 7). The results suggested that the variation in tea yield was mainly driven by beta and alpha diversities of bacterial and fungal communities (Fig. 7A). The most correlated factor for tea yield was the beta diversity of the fungal community. Tea quality was most associated with soil AK; the correlation between tea quality and fungal beta diversity was higher than that with bacterial beta diversity (Fig. 7B). Moreover, Pearson's correlation analysis results suggested that microbial alpha diversity was signi cantly negatively correlated with tea yield, whereas no signi cant correlation between microbial alpha diversity and tea quality was observed (Fig. 7).

Effect of N input on soil and tea characteristics
Soil physicochemical properties were markedly altered after N addition (Table S1). N2 and N3 conditions affected soil properties more signi cantly than N1 conditions (Fig. S3). In addition, the effects of N input on topsoil were more apparent than those on subsoil. The decrease of soil pH after N input was related to the release of Al ions (Ruan et al., 2006). Interestingly, after excessive N input, SOC and soil TN decreased simultaneously to maintain a stable C/N ratio, since N addition caused a stoichiometric imbalance and decreased the microbial C utilization e ciency, which stimulated SOC decomposition (Li et al., 2021a; Li et al., 2021b). The soil microbiome regulated stoichiometric stability after nutrient addition (Ma et al., 2021).
Excessive N addition decreased the yield and quality of tea (Fig. 1). These results suggested that highest tea yield and quality could be achieved under N2 conditions. In practice, the N fertiliser application rate under N2 conditions is recommended for green tea production (Ruan et al., 2020). N1 conditions could not meet the N requirement for tea growth. Overuse of N fertiliser decreases soil pH and causes adverse soil conditions (Table S1), which inhibit the synthesis of biochemical components, thereby decreasing tea yield (Ruan et al., 2007;Ma et al., 2013). For instance, arginine and not theanine forms when excessive N fertiliser is applied, resulting in a bitter taste and decreased quality of tea (Ruan et al., 2007). TP/AA is a comprehensive indicator, negatively correlated with green tea quality (Li et al., 2020a). N fertiliser application contributed to the synthesis of AA and decreased the TP content, thereby decreasing TP/AA ( Fig. 1). Spring tea was of a higher quality than summer or autumn tea (Fig. 1D), owing to the optimal spring climate accompanied with suitable temperature and su cient rain that favour tea growth (Wang et al., 2011); long-term nutrient accumulation from basal fertiliser application to spring tea harvest also contributes to the synthesis of quality ingredients (Sun et al., 2019).

Succession of bacterial and fungal community under N fertiliser application
Microbial diversity is related to the soil nutrient cycle and contributes to the maintenance of soil function (Zheng et al., 2019; Jiao et al., 2021a). The decrease in bacterial diversity was more apparent than that of fungal diversity with N fertiliser application so as to bene t fungi. This could be explained by the niche differentiation between bacteria and fungi related to their different responses to soil character changes induced by N fertiliser application (Fig. 2E), which has also been demonstrated in previous studies ( . Generally, the Chao1 index was more sensitive than the Shannon index under N fertiliser application conditions (Fig. 2). The Chao1 index accounts for species richness and re ects rare species change, whereas the Shannon index suggests both evenness and abundance of species (Shannon, 1948;Chao, 1984). Therefore, the decrease in the fungal Chao1 index indicated that more rare species disappeared, whereas the evenness and abundance of species distributed in the fungal community were not altered under N fertiliser application conditions.
The distribution of the major phyla of both bacteria and fungi was altered by N fertiliser application (Fig. 3). Overall, disturbance of the microbiome was more apparent in the topsoil than in the subsoil; topsoil is more susceptible to farmer practice, animal disturbance, and climate change than subsoil (Jiao et al., 2021a). The bacterial phyla Proteobacteria and Actinobacteria and fungal phylum Ascomycota bene tted from N fertiliser application, whereas the relative abundances of Acidobacteria and Basidiomycota decreased after N addition ( Fig. 3A and 3C). These results are consistent with the oligotroph-copiotroph theory (Fierer et al., 2007;Yao et al., 2017). Proteobacteria, Actinobacteria, and Ascomycota are classi ed as copiotrophic taxa, whereas Acidobacteria and Basidiomycota are oligotrophic taxa according to the oligotroph-copiotroph theory, which is based on the net carbon mineralisation rate of soil (Fierer et al., 2007). Changes of oligotrophic and copiotrophic taxa under N fertiliser application conditions contributed to SOC mineralisation, which led to a decrease in the SOC content (Table S1). The shift in the relative abundance of the dominant phyla was attributed to the increase in N supply, which met the higher N demand of copiotrophic bacteria than that of oligotrophic bacteria (Fierer et al., 2007). Moreover, under N fertiliser application conditions, tea production and organic carbon levels increased, which might also have contributed to changes in the dominant phyla (Fierer et al., 2007;Fierer et al., 2012). The microbial community was signi cantly in uenced by N fertiliser application and soil sampling layer (P < 0.05); soil bacteria were more sensitive to the sampling layer ( Fig. 4A and 4B). Soil pH was the most important factor regulating both bacterial and fungal communities ( Fig. 4C and 4D). Soil acidi cation induced by N fertiliser application imposes strong environmental ltering, which leads to microbial community assembly through deterministic processes  (Fig. 3B and 3D),the results shown in Fig. 5C indicate that bacterial communities were more associated with soil properties than fungal communities. Studies have shown that fungi are more tolerant to adverse environments, including soil acidi cation (Rousk et al., 2010) and drought (de Vries et al., 2018) than bacteria. Compared with fungi, the pH range suitable for bacterial growth is narrower (Rousk et al., 2010).
Overall, the NCM results indicated that the fungal community migration rate (m = 0.025) was much lower than that of the bacterial community (m = 0.459) (Fig. S2). Previous studies also demonstrated that dispersal limitation has a greater impact on fungi than bacteria because of their larger size (Schmidt et al., 2014;Chen et al., 2020;Liu et al., 2021). Moreover, the increase in the stochastic process and migration rate of both bacteria and fungi from topsoil to subsoil may be partly explained by the alleviation of subsoil acidi cation, which affects microbial community distribution (Tripathi et al., 2018).
In this study, we found that the fungal stochastic process increased, whereas richness decreased, with an increasing rate of N application (Figs. 2 and S5). Jiao et al. (2021b) also reported that fungal richness decreased with increasing stochastic process, and thus in uenced ecosystem functions driven by biodiversity (Jiao et al., 2021a). In tea plantation, assembly process of fungi was more susceptible to N fertiliser application than that of bacteria (Figs. S4 and S5).

Relationship between bacteria, fungi, and tea
The soil microbiome participates in the nutrient cycle and contributes to plant growth (Saleem et al., 2019). In this study, fungal and bacterial communities and diversity were the most relevant indicators of tea yield. Fungi were more associated with tea yield and quality than bacteria ( Fig. 7A and 7B). K plays a vital role in the synthesis of caffeine, AAs, and water-extractable dry matter; thus, soil AK is closely related to tea quality ( . However, we found that microbial alpha diversity is negatively correlated with tea yield (Fig. 7D and 7F). Although N fertiliser application supplied nutrients that contribute to tea growth, it caused an increase in soil acidi cation, thereby decreasing microbial alpha diversity. A study also reported that microbial diversity is not always positively correlated with . This decomposition-niche differentiation between fungi and bacteria contributes to nutrient cycling. Moreover, fungi were more tolerant to adverse environments, including N-induced acidi cation, than bacteria, thus playing a more stable role under N fertiliser application conditions (Figs. 2, 3, and 5).

Conclusion
In this study, the response of tea plants, bacteria, and fungi under 5-year N fertiliser application conditions were systematically investigated. Excessive N fertiliser application decreased tea yield and quality. Moreover, N fertiliser application signi cantly decreased bacterial and fungal diversity and altered bacterial and fungal compositions and communities. N fertiliser application favoured the growth of copiotrophic taxa (Proteobacteria, Actinobacteria, and Ascomycota) and inhibited oligotrophic taxa (Acidobacteria and Basidiomycota). Bacteria are more sensitive to environmental variations than fungi, and soil pH was the most important factor driving bacterial and fungal community succession.
Stochastic processes contributed more to bacterial community assembly than to fungal community assembly in tea plantation soil. Tea yield and quality were more associated with fungi than bacteria based on random forest analysis. In conclusion, considering the extreme environment of tea plantation soil (i.e., low pH and high F and Al contents) and the tolerance of fungi for extreme environments, future studies should investigate the association between fungi and tea growth. Declaration of interests The authors declare that they have no known competing nancial interests or personal relationships that could have appeared to in uence the work reported in this paper.