Fertilization Regulates Grape Yield and Quality in by Altering Soil Nutrients and the Microbial Community

: Rational fertilization is a win-win strategy for rural incomes and environmental restoration in ecologically fragile regions. However, the long-term cumulative grape productivity response to soil fertility has rarely been quantiﬁed. Here, long-term fertilization experiments (over 15 years) in the desert–oasis transitional zone of Sinkiang, China, were used to evaluate the interactions among grape yield, quality, fertilization, soil nutrients, and microbial communities. There were ﬁve treatments, as follows: CK 0 (no planting and no fertilizing); NP (synthetic nitrogen and phosphorus); M (manure only); NPM 1 (0.25 times NP and 0.33 times M); and NPM 2 (NP and 0.5 times M). The grape yield increased with the application of total nitrogen. The soluble solids and reducing sugar contents had signiﬁcant positive linear correlations with grape yield, but the opposite trend was found between grape yield and titratable acidity and tannin contents. The redundancy analysis showed that fertilization, soil nutrients (soil organic carbon, available nitrogen, and dissolved organic nitrogen), and microbial communities (ratio of fungi to bacteria, ratio of Gram-negative bacteria to Gram-positive bacteria, and total phospholipid fatty acids) accounted for 31.9%, 19.7%, and 26.8% of the grape yield and nutritional ingredients, respectively. The path analysis identiﬁed that fertilization, soil nutrients, and the microbial communities were signiﬁcantly positively associated with the grape yield, soluble solids, and reducing sugars, while their associations with titratable acidity, tannins, and phenols were signiﬁcantly negative. These results suggested that fertilization is a viable strategy for regulating grape yields and grape quality because it alters soil fertility in ecologically fragile regions.


Introduction
The demand for fruit production is continuously increasing with improvements to human living standards.Globally, fruit production increased from 0.48 billion tons in 2000 to 0.72 billion tons in 2017, with an annual increase rate of 2.56% [1].Grapes, as the second most produced fruit, are becoming increasingly popular due to their nutritional and medicinal values [2,3].Another major use of grapes is in making wine, especially in Europe, Australia, and America.Wine is considered to be a healthy and hygienic drink [3], and the nutritional ingredients of grapes, including reducing sugars, titratable acidity, and tannin contents, determine the composition and quality of the wine [4].Obtaining high yields and superior nutritional ingredients requires large temperature differences, the avoidance of rain and dew, and a fine sandy loam soil, but regions with these characteristics are considered to be ecologically fragile [4,5].Furthermore, these regions are not suitable for grape growth without human interventions.Therefore, it is important to understand how to achieve high grape yields and good nutritional ingredients in ecologically fragile regions.
Rational fertilization is a win-win strategy that achieves high crop productivity and improves the soil quality in ecologically fragile regions [6].Fertilization with respect to crop requirements not only directly provides extra nutritional matter but also indirectly improves soil properties for better crop growth [7,8].Many researchers have explored the response of grape productivity to different fertilizers based on short-term experiments.For instance, Martínez, Ortega, Janssens, and Fincheira [8] reported that a combination of organic and synthetic fertilizers was better at increasing grape yields than no fertilizer, synthetic fertilizer, or organic fertilizer applications alone.Brunetto, et al. [9] showed that synthetic nitrogen fertilizer did not influence the nutritional contents of grapes, while manure decreased the soluble solids content and increased the total acidity content of grapes.Although the short-term responses of grape yields and nutritional ingredients to different fertilizers are well-known, the long-term cumulative grape productivity responses to different fertilizers may be complicated.To achieve high grape yields, a large amount of fertilizer is often applied [10].However, it is not clear whether grape yield increases are synchronized with an increase in nutritional ingredients over the long-term.
Fertilization can improve soil productivity, such as soil nutrients and microbial communities, which both play a vital role in determining crop productivity [11,12].For example, a combination of organic and synthetic fertilizer significantly increased the soil carbon (C) and nitrogen (N) concentrations compared to using a synthetic fertilizer or no fertilizer [7].The soil C and N concentrations represent the soil's ability to supply utilizable nutrients, and many studies have shown that high soil organic carbon (SOC) and N concentrations lead to high crop yields and quality [11,13].Soil biological fertility is a good indicator of the soil microbial parameter responses to different fertilizers and has been widely studied [12,14].A global meta-analysis reported that N addition significantly inhibited soil microbial growth, composition, and function based on 1408 paired observations [15].The ratio of fungi to bacteria usually decreased with N fertilization due to the low N demands of fungi [14].Compared to synthetic fertilizer, organic fertilizer affects soil microbial communities by providing C-rich organic compounds to the C-limited microbial communities in ecologically fragile regions.It also directly supplies nutrients [16].Bacteria and fungi are readily affected by available and complex C compounds, respectively [17,18].Furthermore, reducing C inputs could decrease Gram-negative and increase Gram-positive bacterial abundance [17,19].Soil fertility is associated with soil microbial communities because the mineralized nutrients at different soil fertility levels can actively change soil microbial communities [20].Overall, the combination of fertilization, soil nutrients, and soil microbial communities has complex effects (direct, indirect, and interacting) on crop productivity.Therefore, an increased understanding of how these variables directly and/or indirectly affect crop productivity is urgently needed if crop growth is to improve.
Sinkiang, China, is located in the hinterland of the Eurasian continent and is in a desertoasis transitional zone.This area is vulnerable to damage and has difficulty recovering due to the large temperature differences, high sunshine time, and low precipitation that are characteristic of the area.To prevent environmental degradation and ensure sustainable livelihoods, this natural and fragile ecological region was transformed into an area for growing grapes.Therefore, the objectives of this study were to (1) quantify the long-term cumulative responses of grape yields and nutritional ingredients to different fertilization treatments and (2) identify the links among grape productivity, soil nutrients, and soil microbial communities under different fertilization treatments.

Experimental Site
The research was conducted at the Shanshan monitoring site (43 • 06 N, 90 • 33 E), Sinkiang, China.The region is a desert-oasis transitional zone with an average temperature of 11.3 • C, an effective accumulated temperature of 5035 • C (>10 • C), and 3100 h of sunshine.The average annual precipitation is 25 mm and the evaporation is 3200 mm.The soil type is Haplic Cambisol and irrigation desert soil according to the Food and Agriculture Organization (FAO) [21] and the Chinese soil classification system, respectively.The initial (1987) topsoil (0-20 cm) at the Shanshan site had an SOC level of 7.88 g kg −1 , a bulk density of 1.15 kg m −3 , a soil pH of 8.10, and soil available N (AN), phosphorus (AP), and potassium (AK) levels of 78, 12.7, and 178 mg kg −1 , respectively.

Experimental Design
To determine differences in grape yields, grape nutritional ingredients, soil nutrients, and soil microbial communities, five different long-term fertilization regimes were established using a random design, as follows: (1) CK 0 (no planting and no fertilizing); (2) NP (synthetic nitrogen and phosphorus); (3) M (manure only); (4) NPM 1 (0.25 times NP and 0.33 times M); and (5) NPM 2 (NP and 0.5 times manure).The synthetic N and P were from urea and diammonium phosphate, respectively.The manure was pure sheep manure, with average C, N, and P contents (during the experiment) of 336, 20.1, and 4.96 g kg −1 dry weight, respectively.The CK 0 regime involved no planting and no fertilizing; NP was first applied in 1991, with synthetic N of 540 kg ha −1 year −1 and synthetic P of 300 kg ha −1 year −1 ; NPM 1 was first applied in 1996, with 0.25 times more synthetic fertilizer than of the NP treatment and manure at 12 × 10 3 kg ha −1 year −1 ; M was first applied in 2000, with 36 × 10 3 kg ha −1 year −1 ; and NPM 2 was first applied in 2001, with one application of synthetic NP fertilizer and manure at 15 × 10 3 kg ha −1 year −1 .This information is shown in Table 1.
Table 1.Description of the site information and the annual rates (kg ha −1 ) of synthetic nitrogen (Sy-N) with phosphorus (Sy-P) fertilizer additions and organic nitrogen (Or-N) with phosphorus (Or-P) fertilizer applied under the various fertilization treatments.Notes: CK 0 , no fertilizer and crops; NP, synthetic nitrogen and phosphorus; M, manure; NPM 1 , 0.25 times NP and 0.33 times M; NPM 2 , NP and 0.5 times manure; Sy-N was added as urea and diammonium phosphate, and Sy-P was added as diammonium phosphate.These experimental treatments were on land that had previously been a natural grape ecosystem (CK 0 ).Three years before the start of fertilization, the field was cleared to ensure uniform soil nutrients and microbial communities.The grape variety Thompson Seedless was planted at a density of 7500 plants per hectare (row spacing of 0.8 m × 1.7 m) and harvested in mid-August every year.Irrigation was applied because of the substantial evaporation and low amounts of precipitation.Flood irrigation was applied seven times during the germination, flowering, and berry expansion stages and during four periods after harvest (total irrigation amount was 27 × 10 3 m 3 per hectare).Lime sulfur was applied to prevent powdery mildew, and no herbicide was applied.The grapes were harvested manually and pruned in summer to leave 1-2 spikes on the fruiting branches.

Sample Collection
Samples were collected from the topsoil (0-20 cm) and subsoil (20-40 cm) in July, 2016.Five to ten random soil core samples were collected using the "S" curve method.Plant residues and rocks were removed to ensure the accuracy of the analysis indicators.Then, the soil samples were thoroughly mixed and divided into three parts.The first part was air-dried and crushed to pass through a 0.25 mm sieve to measure the soil nutrients, the second part was stored at 4 • C for the soil microbial biomass carbon (MBC) and nitrogen (MBN) analyses, and the third part was freeze-dried and stored at −70 • C for the soil microbial community analysis.
Additionally, five major nutritional ingredients in grapes were measured to represent grape quality.The soluble solids, reducing sugars, titratable acidity, tannins, and total phenols were measured using a refractometer, Fehling's reagent, point titration, potassium permanganate oxidation, and the Folin-Ciocalteu method, respectively.
The annual relative change rates method (ARs, %) for soil nutrients and the microbial communities was adopted to allow the datasets, which had different experimental durations, to be compared.The AR was calculated as follows: where S treatment represents the soil nutrients and microbial communities under the fertilization treatments, S control represents the soil nutrients and microbial communities under the CK 0 treatment, and T is the experimental duration (years).

Data Analyses
One-way ANOVA was used to explore the differences in grape yield, soil nutrients, and the microbial communities under the different fertilization treatments.A redundancy analysis (RDA) by Canoco 5 was used to select and quantify the factors driving the soil nutrients (SOC, TN, AN, AP, AK, DOC, and DON) and microbial communities (MBC, MBN, fungi, bacteria, ratio of fungi to bacteria, AMF, ACT, G+, G−, ratio of G−/G+, and PLFAs) effects on grape yield and nutritional ingredients.A path analysis using the Amos 17.0 package was employed to identify the relationships among yield, nutritional ingredients, soil nutrients, and the microbial communities.The driving factors and grape nutritional ingredients were divided into the following four latent variables: soil nutrients (SOC, AN, and DON), microbial communities (ratio of fungi to bacteria, G−/G+, and PLFAs), grape quality 1 (soluble solids and reducing sugars), and grape quality 2 (titratable acidity, tannins, and phenols).The following hypothetical paths were developed.First, fertilization, soil nutrients, and the microbial communities had a direct effect on the grape yield and quality, and second, fertilization indirectly affected the grape yield and quality via its effects on soil nutrients and the microbial communities.Finally, the soil nutrients that indirectly affected the grape yield and quality through their effects on the microbial communities were measured.

Grape Yield and Nutritional Ingredient Responses to Long-Term Fertilization
The grape yield increased with the application of TN (inorganic and organic) (NPM 2 > M > NP > NPM 1 , Figure 1).Significantly higher soluble solids and reducing sugar contents were recorded in the NPM 2 treatment (22 g L −1 and 199 g L −1 , respectively).A comparison of the M, NP, and NPM 1 treatments showed that NP led to the highest titratable acidity (7.20 g L −1 ), tannin (0.35 g L −1 ), and total phenol (0.52 g L −1 ) contents.The lowest titratable acidity, tannin, and total phenol contents occurred in the M treatment.The soluble solids and reducing sugar contents had a significant positive linear correlation with grape yield (Figure 2), but the titratable acidity and tannin contents significantly decreased with increasing grape yield.There was no significant linear relationship between grape yield and total phenol content.

Grape Yield and Nutritional Ingredient Responses to Long-Term Fertilization
The grape yield increased with the application of TN (inorganic and organic) (NPM2 > M > NP > NPM1, Figure 1).Significantly higher soluble solids and reducing sugar contents were recorded in the NPM2 treatment (22 g L −1 and 199 g L −1 , respectively).A comparison of the M, NP, and NPM1 treatments showed that NP led to the highest titratable acidity (7.20 g L −1 ), tannin (0.35 g L −1 ), and total phenol (0.52 g L −1 ) contents.The lowest titratable acidity, tannin, and total phenol contents occurred in the M treatment.The soluble solids and reducing sugar contents had a significant positive linear correlation with grape yield (Figure 2), but the titratable acidity and tannin contents significantly decreased with increasing grape yield.There was no significant linear relationship between grape yield and total phenol content.

Changes in Soil Nutrients under Long-Term Fertilization
The soil nutrient levels under the CK0 treatment and the ARs under the NP, NPM1, NPM2, and M treatments are shown in Table 2.The different fertilization treatments improved the SOC, TN, AN, AP, DOC, and DON concentrations in the topsoil (0-20 cm) compared to those in the CK0 treatment, except for the DOC and DON concentrations

Grape Yield and Nutritional Ingredient Responses to Long-Term Fertilization
The grape yield increased with the application of TN (inorganic and organic) (NPM2 > M > NP > NPM1, Figure 1).Significantly higher soluble solids and reducing sugar contents were recorded in the NPM2 treatment (22 g L −1 and 199 g L −1 , respectively).A comparison of the M, NP, and NPM1 treatments showed that NP led to the highest titratable acidity (7.20 g L −1 ), tannin (0.35 g L −1 ), and total phenol (0.52 g L −1 ) contents.The lowest titratable acidity, tannin, and total phenol contents occurred in the M treatment.The soluble solids and reducing sugar contents had a significant positive linear correlation with grape yield (Figure 2), but the titratable acidity and tannin contents significantly decreased with increasing grape yield.There was no significant linear relationship between grape yield and total phenol content.

Changes in Soil Nutrients under Long-Term Fertilization
The soil nutrient levels under the CK0 treatment and the ARs under the NP, NPM1, NPM2, and M treatments are shown in Table 2.The different fertilization treatments improved the SOC, TN, AN, AP, DOC, and DON concentrations in the topsoil (0-20 cm) compared to those in the CK0 treatment, except for the DOC and DON concentrations  M treatments, but did significantly increased compared to the NP and NPM 1 treatments.The AN (2.64-3.55%)AR under the NP, NPM 2 , and M treatments was higher than that under the NPM 1 treatment.Finally, the NPM 2 and M treatments improved the DOC and DON concentrations compared to the CK 0 treatment, with AR values of 2.86-6.97%(DOC) and 2.29-4.29%(DON).

Mechanisms Driving Grape Yield and Quality
The RDA suggested that fertilization, SOC, AN, DON, F/B, G−/G+, and PLFAs were the driving factors regulating grape yield and nutritional ingredients (Figure 3).Fertilization had the most influential impact on grape yield and nutritional ingredients among the selected seven variables (31.9%).In addition, soil nutrients (SOC, AN, and DON) and the microbial communities (F/B, G−/G+, and PLFA) accounted for 19.7% and 26.8%, respectively.The driving factors and grape nutritional ingredients were divided into four latent variables (soil nutrients, microbial communities, grape quality 1, and grape quality 2; see Figure 4).The standardized loading scores suggested that SOC and G−/G+ were more powerful indicators of soil nutrients and the microbial communities, which was consistent with the RDA results.The path analysis explained 73-77% of the variance in grape yield and nutritional ingredients (Figure 5).Fertilization, soil nutrients, and the microbial communities directly affected grape yield and nutritional ingredients, and fertilization strongly, and positively affected soil nutrients.Soil nutrients directly affected grape yield and nutritional ingredients via their effects on microbial communities.Overall, fertilization, soil nutrients, and the microbial communities were significantly and positively associated with grape yield and quality 1, while their association with grape quality 2 was significantly negative.The standardized total effects on grape yield and quality occurred in the following order: fertilization > soil nutrients > microbial communities (Figure 6).

Mechanisms Driving Grape Yield and Quality
The RDA suggested that fertilization, SOC, AN, DON, F/B, G−/G+, and PLFAs were the driving factors regulating grape yield and nutritional ingredients (Figure 3).Fertilization had the most influential impact on grape yield and nutritional ingredients among the selected seven variables (31.9%).In addition, soil nutrients (SOC, AN, and DON) and the microbial communities (F/B, G−/G+, and PLFA) accounted for 19.7% and 26.8%, respectively.The driving factors and grape nutritional ingredients were divided into four latent variables (soil nutrients, microbial communities, grape quality 1, and grape quality 2; see Figure 4).The standardized loading scores suggested that SOC and G−/G+ were more powerful indicators of soil nutrients and the microbial communities, which was consistent with the RDA results.The path analysis explained 73-77% of the variance in grape yield and nutritional ingredients (Figure 5).Fertilization, soil nutrients, and the microbial communities directly affected grape yield and nutritional ingredients, and fertilization strongly, and positively affected soil nutrients.Soil nutrients directly affected grape yield and nutritional ingredients via their effects on microbial communities.Overall, fertilization, soil nutrients, and the microbial communities were significantly and positively associated with grape yield and quality 1, while their association with grape quality 2 was significantly negative.The standardized total effects on grape yield and quality occurred in the following order: fertilization > soil nutrients > microbial communities (Figure 6).

Effect of Long-Term Fertilization on Grape Yield and Quality
Fertilization provides nutrients for plant growth when the existing soil nutrients are not sufficient, especially in ecologically fragile regions.Our results showed that grape yield was higher in the NPM2 treatment and M treatment than in the NP treatment and NPM1 treatment.One potential reason was the different amount and type of N input (Fertilizer TN:NPM2 (842 kg ha −1 year −1 ) > M (624 kg ha −1 year −1 ) > NP (540 kg ha −1 year −1 ) > NPM1 (kg ha −1 year −1 ) (Table 1).Another important potential reason may be that the manure also indirectly enhanced the soil properties, which led to improved crop growth, as the extent of soil quality improvement largely depended on the type of fertilizer used (Figure 5) [7,19].The grape yield increased with the application of TN (inorganic and organic) (Figure 1 and Table 2).Nitrogen is the main component of phospholipids, nucleic acids, and protein in plant tissues.Therefore, N additions have a strong effect on grape yields [19].Interestingly, there was a different relationship between grape yield and the nutritional ingredients, which meant that fertilization not only impacted the grape yield but also had a considerable effect on the grape nutritional ingredients.Reducing sugars, titratable acidity, soluble solids, tannins, and total phenols are the main nutritional ingredients in grapes [4].The short-term effects of different fertilization schemes on grape nutritional ingredients have been extensively studied.For example, Thomidis et al. [31] reported a large sugar content with high N input treatments, and the soluble solids content with a manure treatment was higher than that with a synthetic fertilizer treatment [9].These results were consistent with our findings.However, Brunetto, Ceretta, Melo, Miotto, Ferreira, Couto, Silva, Garlet, Somavilla, Cancian, and Ambrosini [9] found that a manure treatment produced a higher titratable acidity content than a synthetic fertilizer treatment.The tannin and total phenolic contents were also significantly higher under a manure treatment [32].This result is inconsistent with our findings and this difference was probably due to the amount of N fertilizer and different accumulative effects (shortterm and long-term).Overall, the soluble solids and reducing sugar contents had significant positive correlations with grape yield, but the opposite trend was found between grape yield and the titratable acidity and tannin contents in the different experimental fertilization treatments.Therefore, achieving higher grape yields can also lead to reductions in some nutritional ingredients.Fertilization provides nutrients for plant growth when the existing soil nutrients are not sufficient, especially in ecologically fragile regions.Our results showed that grape yield was higher in the NPM 2 treatment and M treatment than in the NP treatment and NPM 1 treatment.One potential reason was the different amount and type of N input (Fertilizer TN:NPM 2 (842 kg ha −1 year −1 ) > M (624 kg ha −1 year −1 ) > NP (540 kg ha −1 year −1 ) > NPM 1 (kg ha −1 year −1 ) (Table 1).Another important potential reason may be that the manure also indirectly enhanced the soil properties, which led to improved crop growth, as the extent of soil quality improvement largely depended on the type of fertilizer used (Figure 5) [7,19].The grape yield increased with the application of TN (inorganic and organic) (Figure 1 and Table 2).Nitrogen is the main component of phospholipids, nucleic acids, and protein in plant tissues.Therefore, N additions have a strong effect on grape yields [19].Interestingly, there was a different relationship between grape yield and the nutritional ingredients, which meant that fertilization not only impacted the grape yield but also had a considerable effect on the grape nutritional ingredients.Reducing sugars, titratable acidity, soluble solids, tannins, and total phenols are the main nutritional ingredients in grapes [4].The short-term effects of different fertilization schemes on grape nutritional ingredients have been extensively studied.For example, Thomidis et al. [31] reported a large sugar content with high N input treatments, and the soluble solids content with a manure treatment was higher than that with a synthetic fertilizer treatment [9].These results were consistent with our findings.However, Brunetto, Ceretta, Melo, Miotto, Ferreira, Couto, Silva, Garlet, Somavilla, Cancian, and Ambrosini [9] found that a manure treatment produced a higher titratable acidity content than a synthetic fertilizer treatment.The tannin and total phenolic contents were also significantly higher under a manure treatment [32].This result is inconsistent with our findings and this difference was probably due to the amount of N fertilizer and different accumulative effects (short-term and longterm).Overall, the soluble solids and reducing sugar contents had significant positive correlations with grape yield, but the opposite trend was found between grape yield and the titratable acidity and tannin contents in the different experimental fertilization treatments.Therefore, achieving higher grape yields can also lead to reductions in some nutritional ingredients.

Effects of Long-Term Fertilization on Soil Nutrients and Microbial Communities
Soil fertility is a well-studied subject that affects human living standards and environmental quality.Fertilization is the most effective management method for improving soil fertility, especially soil nutrients and microbial communities.Previous studies have shown that synthetic fertilizer and manure increased the SOC and AN concentrations compared to CK 0 , and that this was due to exogenous nutrients and C inputs [33,34].The SOC, AN, and DON under the NPM 1 treatment were lower than those under the NP, NPM 2 , and M treatments (Tables 1 and 2).The fertilizer types were as follows: NPM 2 (Sy-N 540, Or-N 302), M (Sy-N 0, Or-N 624), NP (Sy-N 540, Or-N 0), and NPM 1 (Sy-N 135, Or-N 240) (kg ha−1 year−1).A large amount of synthetic fertilizer could promote the growth of plant roots, bringing more root exudates to the soil [35].Although manure can directly add exogenous substances to soil, the effect of a small amount of manure on soil fertility was not as significant as adding a large amount of synthetic fertilizer.Several studies have explored the effects of N addition on soil microbial communities at local and global scales [15,36].Most studies have reported that excessive application of nitrogen fertilizer (urea) will reduce the number of soil microorganisms, which may be related to the toxicity of ammonia and excess microbial nutrition [14,15].Previous studies have suggested that the ratio of fungi to bacteria decreased with the application of N fertilizer because the N requirement of bacteria is higher than that of fungi [37].However, our results were inconsistent with this finding.Two potential reasons are possible.First, our experiment was carried out using gray desert soil, which has a high sand content [38].Most nitrogen fertilizer is leached and lost when external irrigation is applied, which also meant that there were no toxic effects on the microorganisms [39].Another reason is that our site was located in a desert-oasis transitional zone that is extremely deficient in soil nutrients and has a low soil microbial biomass [40].Following soil N enrichment, the lower N requirement of plants reduced the AMF content (Table 3).Furthermore, a high AMF content promoted more C to be transferred from plant roots to soil, which increased the G+ bacterial abundance (Table 3) [41,42].Overall, the effects of fertilization on soil nutrients and microbial communities depended on the amount and type of fertilization.

Grape Yield and Quality Responses to Soil Nutrient, Microbial Community and Soil Fertility
The standardized loading scores suggested that SOC and G−/G+ were powerful indicators of soil fertility (Figure 4).The SOC content represents the capacity of soil to improve nutrient levels [16].Furthermore, G− and G+ bacteria rely on readily available and recalcitrant C sources [43].Therefore, the G−/G+ ratio indicates the level of newly formed SOC, namely soil fertility.As expected, fertilization significantly influenced grape yield and nutritional ingredients compared to soil nutrients and microbial communities (Figure 3).
Fertilization not only directly provides extra nutritional matter but also indirectly improves the soil properties, soil microbial communities, and soil microenvironment, which regulate plant growth [44].For example, based on a 25 year fertilization experiment, Cai, Xu, Wang, Zhang, Liang, Hou, and Luo [7] reported that fertilization increased the crop yield by enhancing soil nutrients.Using 223 Arctic and Antarctic soil samples, Siciliano, Palmer, Winsley, Lamb, Bissett, Brown, van Dorst, Ji, Ferrari, Grogan, Chu, and Snape [20] found that soil nutrients were associated with soil microbial communities.Based on previous studies, a relationship among grape yields and nutritional ingredients, fertilization, soil nutrients, and soil microbial communities has been hypothesized.Our results showed that fertilization, soil nutrients, and the microbial communities were significantly and positively associated with grape yield, soluble solids, and reducing sugars, while their association with titratable acidity, tannins, and phenols was significantly negative.Soil nutrients directly affect grape yield and nutritional ingredients via microbial communities.Soil nutrients provide energy for soil microbial activity and determine the microbial composition, and this might be an important mechanism by which soil fertility regulates grape productivity [20].Additionally, there was a greater direct effect between the soil microbial communities and grape nutritional.Overall, our results further verified that the interplay among fertilization, soil nutrients, and soil microbial communities, and their interaction jointly affected grape yield and nutritional ingredients (Figure 5).These results suggested that fertilization was a viable strategy for regulating grape yields and grape quality because it altered soil fertility in ecologically fragile regions.

Conclusions
Significantly different grape yields, nutritional ingredients, soil nutrients, and soil microbial communities were found among the long-term different fertilization treatments.The soluble solids and reducing sugar contents had a significantly positive linear correlation with grape yield, but the opposite trend was found between grape yield, and titratable acidity and tannin contents.The effects of fertilization on soil nutrients and the microbial communities depended on the amount of fertilization.Fertilization accounted for 31.9% of the variance in grape yield and nutritional ingredients, followed by microbial communities at 26.8% and soil nutrients at 19.7%.Fertilization, soil nutrients, and the microbial communities were significantly and positively associated with grape yield, soluble solids, and reducing sugars, while their association with titratable acidity, tannins, and phenols was significantly negative.Our results suggested that fertilization was a viable strategy for regulating grape yields and quality because it alters soil fertility.The results also showed that a high grape yield implied that all nutritional ingredients would also be high.

Figure 1 .
Figure 1.Grape yields (a), soluble solids (b), reducing sugars (c), titratable acidity (d), tannins (e), and total phenols (f) under the various fertilization treatments based on four fertilization experiments.Notes: See Table 1 for the abbreviations of the fertilization treatments; bars represent standard deviations; and different letters indicate significant differences (p < 0.05) among the various fertilization treatments.

Figure 2 .
Figure 2. Relationship between the grape yield and grape quality in terms of soluble solids (a), reducing sugars (b), titratable acidity (c), tannins (d), and total phenols (e).

Figure 1 .
Figure 1.Grape yields (a), soluble solids (b), reducing sugars (c), titratable acidity (d), tannins (e), and total phenols (f) under the various fertilization treatments based on four fertilization experiments.Notes: See Table 1 for the abbreviations of the fertilization treatments; bars represent standard deviations; and different letters indicate significant differences (p < 0.05) among the various fertilization treatments.

Figure 1 .
Figure 1.Grape yields (a), soluble solids (b), reducing sugars (c), titratable acidity (d), tannins (e), and total phenols (f) under the various fertilization treatments based on four fertilization experiments.Notes: See Table 1 for the abbreviations of the fertilization treatments; bars represent standard deviations; and different letters indicate significant differences (p < 0.05) among the various fertilization treatments.

Figure 2 .
Figure 2. Relationship between the grape yield and grape quality in terms of soluble solids (a), reducing sugars (b), titratable acidity (c), tannins (d), and total phenols (e).

Figure 2 .
Figure 2. Relationship between the grape yield and grape quality in terms of soluble solids (a), reducing sugars (b), titratable acidity (c), tannins (d), and total phenols (e).

3. 2 .
Changes in Soil Nutrients under Long-Term FertilizationThe soil nutrient levels under the CK 0 treatment and the ARs under the NP, NPM 1 , NPM 2 , and M treatments are shown in Table2.The different fertilization treatments improved the SOC, TN, AN, AP, DOC, and DON concentrations in the topsoil (0-20 cm) compared to those in the CK 0 treatment, except for the DOC and DON concentrations using the NPM 1 treatment.Compared to the NPM 1 treatment, the M treatment significantly increased the SOC (18.05%),TN (31.48%),AN (3.91%), DOC (19.06%), and DON (12.03%)ARs.The ARs for AP and AK were not significantly different among the NP, NPM 1 , NPM 2 , and M treatments.The ARs of all soil nutrients in the subsoil (20-40 cm) were lower than in the topsoil.Compared to the CK 0 treatment, the NP and NPM 1 treatments improved the SOC, TN, AN, and AP concentrations, but reduced the AK, DOC, and DON concentrations.The ARs for SOC and TN did not significantly differ between the NPM 2 and

Figure 3 .
Figure 3. Redundancy analysis (RDA) for the multivariate effects of fertilization, soil nutrients, and the soil microbial community on the grape yield and quality.The soil nutrients included soil organic carbon (SOC), soil available nitrogen (AN), and dissolved organic nitrogen (DON); the soil microbial community included phospholipid fatty acids (PLFAs), the ratio of fungi to bacteria (F:B), and the ratio of Gram-positive to Gram-negative bacteria (G−:G+); the grape quality included soluble solids (SS), reducing sugars (RS), titratable acidity (TA), tannins, and phenols.

Figure 3 .
Figure 3. Redundancy analysis (RDA) for the multivariate effects of fertilization, soil nutrients, and the soil microbial community on the grape yield and quality.The soil nutrients included soil organic carbon (SOC), soil available nitrogen (AN), and dissolved organic nitrogen (DON); the soil microbial community included phospholipid fatty acids (PLFAs), the ratio of fungi to bacteria (F:B), and the ratio of Gram-positive to Gram-negative bacteria (G−:G+); the grape quality included soluble solids (SS), reducing sugars (RS), titratable acidity (TA), tannins, and phenols.

Figure 4 .
Figure 4. Latent variables with their indicators that were considered in the path analysis.The numbers in parentheses show the loading scores.(a) The soil nutrients included soil organic carbon (SOC), soil available nitrogen (AN), and dissolved organic nitrogen (DON); (b) the soil microbial community (MC) included the phospholipid fatty acids (PLFAs), ratio of fungi to bacteria (F:B), and ratio of Gram-positive to Gram-negative bacteria (GP:GN); (c) grape quality 1 included soluble solids (SS) and reducing sugars (RS); (d) and grape quality 2 included titratable acidity (TA), tannins, and phenols.The data of DON were logarithmically converted.

Figure 5 .
Figure 5. Path analysis results regarding the direct and indirect effects of fertilization, soil nutrients, and the soil microbial community (MC) on the grape yield and quality (chi/df = 1.3, p = 0.43).The numbers show the path coefficients.The gray path and number indicate that the effect is statistically significant, and the dashed paths and associated numbers indicate that the effect is negative.See Figure 4 for the indicators of the four latent variables (nutrients, microbial community, quality 1, and quality 2).

Figure 4 .
Figure 4. Latent variables with their indicators that were considered in the path analysis.The numbers in parentheses show the loading scores.(a) The soil nutrients included soil organic carbon (SOC), soil available nitrogen (AN), and dissolved organic nitrogen (DON); (b) the soil microbial community (MC) included the phospholipid fatty acids (PLFAs), ratio of fungi to bacteria (F:B), and ratio of Gram-positive to Gram-negative bacteria (GP:GN); (c) grape quality 1 included soluble solids (SS) and reducing sugars (RS); (d) and grape quality 2 included titratable acidity (TA), tannins, and phenols.The data of DON were logarithmically converted.

Figure 4 .
Figure 4. Latent variables with their indicators that were considered in the path analysis.The numbers in parentheses show the loading scores.(a) The soil nutrients included soil organic carbon (SOC), soil available nitrogen (AN), and dissolved organic nitrogen (DON); (b) the soil microbial community (MC) included the phospholipid fatty acids (PLFAs), ratio of fungi to bacteria (F:B), and ratio of Gram-positive to Gram-negative bacteria (GP:GN); (c) grape quality 1 included soluble solids (SS) and reducing sugars (RS); (d) and grape quality 2 included titratable acidity (TA), tannins, and phenols.The data of DON were logarithmically converted.

Figure 5 .
Figure 5. Path analysis results regarding the direct and indirect effects of fertilization, soil nutrients, and the soil microbial community (MC) on the grape yield and quality (chi/df = 1.3, p = 0.43).The numbers show the path coefficients.The gray path and number indicate that the effect is statistically significant, and the dashed paths and associated numbers indicate that the effect is negative.See Figure 4 for the indicators of the four latent variables (nutrients, microbial community, quality 1, and quality 2).

Figure 5 .
Figure 5. Path analysis results regarding the direct and indirect effects of fertilization, soil nutrients, and the soil microbial community (MC) on the grape yield and quality (chi/df = 1.3, p = 0.43).The numbers show the path coefficients.The gray path and number indicate that the effect is statistically significant, and the dashed paths and associated numbers indicate that the effect is negative.See Figure 4 for the indicators of the four latent variables (nutrients, microbial community, quality 1, and quality 2).

Figure 6 .
Figure 6.Standardized total effects of fertilization, soil nutrients, and the microbial community on the grape yield and quality.See Figure 4 for the indicators of the four latent variables (nutrients, microbial community, quality 1, and quality 2).

Figure 6 .
Figure 6.Standardized total effects of fertilization, soil nutrients, and the microbial community on the grape yield and quality.See Figure 4 for the indicators of the four latent variables (nutrients, microbial community, quality 1, and quality 2).

1 .
Effect of Long-Term Fertilization on Grape Yield and Quality