Effect of no-tillage and tillage systems on melon (Cucumis melo L.) yield, nutrient uptake and microbial community structures in greenhouse soils

Grafted melon seedlings were used in this study. The scion was a cultivated melon variety, ‘Cuimi’. This cultivar was suitable for greenhouse cultivation, and it could be available in the market in the near future. The rootstock was a pumpkin (Cucurbita moschata) cultivar (Wanzhen 2) and exhibited good graft compatibility and fruit quality. Now, it turned out to be one of the most popular rootstocks available for watermelon and melon grafting in Anhui Province, China.

The greenhouse experiment was accomplished in a planting base in Hexin County (118.35° E and 31.74° N), Anhui Province, China. Soil samples were collected thrice from January 1, 2018, to August 31, 2019. A total of 18 rhizosphere soil samples were collected from 10 points by using a sterile blade at a depth of 0–15 cm and composited together. Each sample was then divided into two portions. One portion was stored at 4°C for biogeochemical analysis and the other one at −80°C for DNA analysis.

A protected field experiment with two operation practices (i.e. UT and TL treatments) was conducted in Hexin County. Soil pH ranged between 7.1 and 7.3, and the organic matter (OM) ranged from 20 to 22 g · kg⁻¹ in two treatments. The complete basic soil properties are listed in Table 1. The soil was tilled with a depth of 20–35 cm in TL. Ridging ways and planting density were arranged the same way as UT. Grafted melon seedlings were transplanted at the beginning of January 2019 under greenhouse soils. Normal and regular management was followed to maintain melon seedling growth in UT and TL treatments. Before the fruit setting time, 20 kg · 667 m⁻² fertiliser (N:P:K ratio of 3:1:3) was added to plant soils. The experiment was terminated at the end of July 2019. The melon matured within 80 days from planting. The average yield from 667 m² was recorded immediately after harvest. Five fruits from each replicate were randomly selected to assess fruit weight (kg) and nutrient elements. Ten plants were used with three replicates.

Soil pH and OM content were detected following our previously reported methods (Zhang et al., 2019). The total N (TN) and alkaline N (AN) of the soil and the N content of the plant were determined following a reported method (Bremner, 1960). The total P (TP) in the melon plant was determined using the combustion and the molybdenum blue colorimetric methods (Sjösten and Blomqvist, 1997). Available P (AP) was extracted and detected following the procedures of a reported method (Li et al., 2004). Soil available K (AK) and plant total K were measured using flame atomic absorption spectrophotometry following a previous method (Zhang et al., 2019).

Approximately, 0.5 g of fresh rhizosphere soil was used to extract soil DNA in triplicate from each soil site by using the MoBio Power Soil DNA Isolation Kits (MoBio Laboratories, Carlsbad, CA, USA) following the manufacturer's instructions. The purity of the soil DNA extracts was determined using the NanoDrop™ 2000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). The obtained DNA was stored at −80°C for future analyses.

The amplification of 16S rRNA for sequencing was performed following a previously described method (Shen et al., 2013; Zhang et al., 2019). Briefly, the V4–V5 regions of the bacterial 16S rRNA genes were amplified using the primers F515 and R907. The primers ITS1 and ITS2 were used to amplify the ITS genes (Table 2). PCR was performed following a previous report (Zhang et al., 2019). The PCR products were purified, and sequencing was performed using the Illumina MiSeq platform from Biozeron Biotechnology Co., Ltd. (Shanghai, China).

The operational taxonomic units (OTUs) with identities of 97% were identified using the Mothur software (http://www.mothur.org). The sequence with the highest relative abundance in each OTU was chosen as the representative sequence. They were selected to search for similar sequences in the nucleotide nonredundant database from the National Center for Biotechnology Information (NCBI) following a previously reported method (Sun et al., 2015).

All data were analyzed using a one-way analysis of variance and considered statistically significant and highly significant at p < 0.05 (*) and p < 0.01 (**), respectively, by using the SPSS software (version 17.0). To determine the functional groups of microbes in the soils, we performed the LEfSe analysis by using a previously reported method (Zhang et al., 2013). The canonical correspondence analysis (CCA) and the redundancy analysis (RDA) were also performed to analyse the effect of soil factors (e.g. pH, AN, AP and AK) on the microbial communities (Zhou et al., 2018). All the sequence data of bacteria and fungi were deposited in the NCBI Sequence Read Archive database with the accession number of SRP241843.

In this study, two operation practices (i.e. UT and TL) were used. Under the UT and the TL treatments, the N, P and K were absorbed similarly by the melon plant, but the total amounts of absorbed nutrients were different. The differences in nutrient contents in leaves were significant. The most absorbed element was N. The N, P and K contents in the UT treatment were 2.71, 0.51, and 1.04 g · plant⁻¹ (Figure 1A), respectively. The N, P and K contents in the TL treatment were 1.94, 0.35 and 0.89 g · plant⁻¹, respectively. The N content was the highest, followed by K, and the P content was the lowest. The N content of the UT treatment was significantly higher than that of the TL treatment.

Figure 1

In the stem of the melon plant, the N, P and K contents in the UT treatment were 0.52, 0.09 and 0.65 g · plant⁻¹ (Figure 1B), respectively. The N, P and K contents in the TL treatment were 0.53, 0.08 and 0.97 g · plant⁻¹, respectively. K was the most absorbed, followed by N, and P was the least absorbed. The K content in the stems in the TL treatment was significantly higher than that in the UT treatment.

In melon fruit, the N, P and K contents in the UT treatment were 7.04, 1.37 and 6.45 g · plant⁻¹, respectively. The N, P and K contents in the TL treatment were 3.42, 0.73 and 4.21 g · plant⁻¹ (Figure 1C), respectively. The N, P and K contents in the UT treatment were significantly higher than those in the TL treatment. The total contents of the three elements in the UT treatment (20.41 g · plant⁻¹) were significantly higher than those in the TL treatment (13.14 g · plant⁻¹). The melon yields in the UT and TL treatments were 2813.34 and 2138.32 kg · 667 m⁻², respectively (Figure 1D), indicating that the UT treatment increased melon yield.

The distribution of the clustered bar showed that the bacterial species (A) was significantly higher than the fungal species (B). The distribution characteristics of the microbial communities in UT and TL treatments were similar. The main bacterial phyla included Proteobacteria, Acidobacteria, Actinobacteria, Gemmatimonadetes and Chloroflexi (Figure 2A), whereas that of fungi was Ascomycota (Figure 2B). The analysis of the bacterial distribution showed no significant difference in the distribution level of the bacterial species in the UT and the TL treatments even when the soil samples were measured at different periods (Figure 3). The thermogram of fungal distribution showed that UT and TL influenced the fungal community and showed differences (UT1 and TL1). The difference in species distribution between the two periods after planting was not significant (Figure 4). The bacterial OTUs were 1785, 1816, 1776, 1753, 1686 and 1767 in UT1, UT2, UT3, TL1, TL2 and TL3 soil samples (Figure 5A), respectively. A total of 689 OTUs were common in all samples, but the fungal OTUs were minimally observed. Rhodobiaceae, a bacterial genus, was the most significant community in cultivated TL soil. The functional communities in the UT treatment were higher than those of the TL treatment (Figure 5B).

Figure 2

Figure 3

Figure 4

Figure 5

The CCA correlation showed a significant negative correlation between the pH value and Patescibacteria and Armatimonadetes groups. Gemmatimonadetes, Bacteroidetes and Elusimicrobia were negatively correlated with OM and AN (Figure 6A). Verticillium was negatively correlated with pH, OM and AN in the fungal community. Ascobolus was negatively correlated with OM, AN, AP and AK. Acrophialophora was positively correlated with pH (Figure 6B). CCA showed that pH was the main factor affecting bacteria, followed by OM and AN content (Figure 7). CCA also showed that AN, OM and pH were the main factors affecting fungal communities in the UT and the TL treatments (Figure 8).

Figure 6

Figure 7

Figure 8

The correlation of the N, P and K contents absorbed by the leaves, stems and fruits of the melon plants with the soil microbial communities was further studied. The TN, TP, and TK contents of the leaves and fruits were significantly positively correlated with the bacterial genus Pirellula (p < 0.05). However, the TN and the TP in the stem were negatively correlated with Dongia, Gemmatimonas, Hamadaea and Nocardioides (Table 3). The bacterial phyla Patescibacteria and Armatimonadetes had a negative correlation with the TP in the stem. However, the bacterial phylum Planctomycetes was significantly positively correlated with the TP in the melon stem.

Concerning fungal genus communities, the TN, TP and TK contents in melon leaves and fruits were negatively correlated with Chaetomium and positively correlated with Dendrostilbella. In the stem, TN was positively correlated with fungal genera Acrophialophora, Cephaliophora and Scedosporium. Fungal communities, such as Polyschema, Cercospora, Verticillium, Fusarium, Dendrostilbella, and Ascobolus, were negatively correlated with the TN, TP, and TK contents (Table 3). Although the nutrients in the melon plant were correlated with bacterial or fungal genera in facility soils, no direct evidence in this work proved that those microbial groups increased or affected the absorption of elements in the melon plant.

The TL and the UT treatments are important for improving soil structures, which are beneficial for plant growth and physical soil properties (Pöhlitz et al., 2018). TL treatment affects corn grain nutrient composition and yield (Houx et al., 2016). In the present work, the two tillage styles showed different effects on the nutrient uptake in melon plants under facility conditions. Our data showed that in the UT treatment, additional N was absorbed in the leaves and fruit (Figures 1A and 1B), and the N content was kept constant in the melon stem under the TL treatment. Previous findings have discovered that the autotrophic nitrification rate in the topsoil layer of UT is significantly higher than that of conventional TL (Liu et al., 2017), which provides a reasonable explanation on the absorption of additional N by the melon plant. TL and P fertilization showed no significant effects on root P contents (Li et al., 2017). By contrast, the P contents in leaves and fruits in the UT treatment in our study were higher than the values in TL treatment especially the P content in fruits (*p < 0.05). Although the effect of TL on the dynamics of the P element is not significant (Bünemann et al., 2006), our study showed that different tillage practices can improve the P uptake in the melon plant. Compared with the TL treatment, additional N and P were detected in melon plant leaves and fruits under the UT treatment. However, the N and the P contents were constant in the melon stems under the two treatments (Figure 1C), thereby suggesting that nutrients were absorbed differently in the parts of the melon plant under the two treatments.

Here, additional K in the fruits under UT treatment was found, whereas high K content was observed in melon stem under TL treatment. K can affect fruit quality (Lester et al., 2010). Following previous studies, our data showed that the K accumulation in melon plants had different amounts under the two treatments. However, in a previous study, TL has produced no-effect on seed K concentrations (Farmaha et al., 2011). Soil management and rotations can affect the P and K in soil and increase fertiliser use efficiency (Rosolem and Calonego, 2013). In the present study, under UT treatment, the proper addition of nutrients can increase the N, P, and K contents in leaves and fruits. The total amount of the three elements in the UT treatment (20.41 g · plant⁻¹) was significantly higher than that in the TL treatment (13.14 g · plant⁻¹), thereby suggesting that nutrient uptake can also be improved by adding fertiliser in later management without TL. Melon yield demonstrated that the UT treatment produced more fruits than the TL treatment, thereby indicating that UT treatment can increase melon yields. Similar results have been reported that UT can ensure equivalent or even higher yields than conventional TL (Ruisi et al., 2016). Hence, the results of the present study indicated that the UT treatment exhibited the same positive effect on the yield of the melon crop compared with TL treatment.

Bacterial and fungal diversities changed differently under UT and TL operations. The OTUs in bacteria were higher than those in fungi, and the functional communities in the UT treatment were higher than those in the TL treatment. Results revealed that the predominant bacterial phyla were Proteobacteria, Acidobacteria, Actinobacteria, Gemmatimonadetes and Chloroflexi (Figure 2A). In our previous study, these bacterial phyla are the main groups in greenhouse soils (Zhang et al., 2019). These phyla have also been detected in plateau freshwater lakes or loess plateau (Zeng et al., 2016; Zhang et al., 2015). In the present study, bacterial communities were dominated by the bacterial groups mentioned above in facility soils.

In terms of fungi, the predominant fungal phylum was Ascomycota (Figure 2B). Various studies have demonstrated that Ascomycota is common in soils (Gorfer et al., 2011; Klaubauf et al., 2010). The results of the present study demonstrated that TL treatment resulted in changes in microbial communities, but these changes were not significant. The heatmap analysis exhibited that TL treatment showed less effect on the distribution or shift in bacterial groups than UT treatment (Figure 3). However, only the effect on the fungal groups was detected in initial samples (Figure 4; UT1 and TL1), thereby suggesting that the TL operation can change the fungal distribution before planting. The long-term application of UT caused a significant increase in bacterial diversities compared with that of TL (Dong et al., 2017). UT with organic input management can enhance soil microbial diversity (Wang et al., 2017b). However, in the present study, the microbial structures under the two treatments were similar, thereby suggesting that the UT treatment had the same effect on the bacterial groups as the TL treatment.

A significant negative correlation was observed between the pH value and Patescibacteria and Armatimonadetes groups. Gemmatimonadetes, Bacteroidetes and Elusimicrobia were negatively correlated with OM and available N (AN; Figure 6A). These bacterial phyla are considerably influenced by soil properties (Kim et al., 2014; Wolińska et al., 2017). In the fungal community, Verticillium was negatively correlated with pH, OM and AN. Ascobolus was negatively correlated with OM, AN, AP and AK. Acrophialophora was positively correlated with pH (Figure 6B). pH-controlled bacterial distribution was followed by OM and AN contents (Figure 7). The soil chemical properties especially soil pH can remarkably shape microbial communities and control the distribution of microbial structures (Shi et al., 2019; Tian et al., 2017). In the present study, AN and OM were the main factors affecting the fungal community in the UT and the TL treatments (Figure 8). Soil fungal communities are linked in the long-term fertilised grassland (Cassman et al., 2016). Endophytic fungal communities are also associated with plants in organic and conventional farming systems, including their environmental growth conditions (Xia et al., 2019). Long-term UT increases soil organic carbon content, which influences fungal group distribution (Desanctis et al., 2012).

Microbial communities are important for soil quality, and plant species and soil type shape the structure and function of the microbial communities in the plant rhizosphere (Berg and Smalla, 2009; Yao et al., 2003). The correlation of the N, P, and K contents that were absorbed by melon plant leaves, stems and fruits with the soil microbial communities was further analysed. However, although the nutrients in different melon parts were correlated to microbial groups, these bacterial or fungal phyla did not decide the nutrient uptake in melon plants. By contrast, the nutrient uptake in melon leaves, stems and fruits may collect those microbial groups. The efficient use of plant nutrient can be enhanced by beneficial bacteria or fungi (Adesemoye et al., 2008). Rhizosphere interactions between microorganisms and plants can also stimulate the P acquisition in plant roots (Marschner et al., 2011). Here, TN, TP, and TK were significantly positively correlated with Pirellula (p < 0.05) in melon leaves and fruits (Table 3). However, TN and TP were negatively correlated with bacterial genera and phyla, such as Gemmatimonas and Nocardioides, in the melon stem. Additional bacterial genera were positively correlated with TP, whereas Patescibacteria and Armatimonadetes were negatively correlated with TP in melon stem (Table 3). Our results suggested that these bacterial genera may promote nutrient uptake. However, precise mechanisms by which these microbial groups can improve the growth of the melon plant should be further investigated. Concerning fungal genera, TN, TP and TK contents in melon leaves and fruits were negatively correlated with Chaetomium and positively correlated with Dendrostilbella. Additional fungal communities were negatively correlated with TN, TP, and TK (Table 3). Previous findings have discovered that chemical fertilisation significantly decreases the diversity of the arbuscular mycorrhizal fungal community structures and is negatively correlated with increased soil nutrients (Liu et al., 2020). As noted above, the data in the present study proved that these bacterial and fungal genera may be correlated with the nutrient uptake in the melon plant. In previous studies, bacteria and fungi have potential abilities to improve plant growth by increasing the uptake of nutrients such as N and P (Rashid et al., 2016; Xia et al., 2019). The results of the present study indicated that UT can also have significant positive effects on the microbial diversity in soils planted with melon crop under facility conditions.