Manure Application Affects Soil Bacterial Community Structure in the Maize-wheat Rotation System of North China Plain

A 2-year maize-wheat rotation experiment was conducted to clarify the characteristics of soil bacterial community under reduced fertiliser-N application rate on the North China Plain. Treatments were conventional fertilisation (CF), 80% fertiliser-N input (80%N) and 80% fertiliser-N plus manure input (80%NM) in the wheat season, while the fertilisation regimes of previous maize were consistent. Results showed that 80%NM increased wheat ear·m -2 and 1,000-grain weight by 5.1% and 7.5%, and it also increased soil C/N ratio (by 29.1%) and dissolved organic N content (by 41.6%), compared with CF, although the difference in crop yields between treatments was not signicant. Values of soil bacterial α-diversity parameters (observed species, chao1 estimator, evenness and Shannon’s diversity index) were improved under 80%NM, but not under 80%N. Compared with CF, 80%NM increased the abundance of phyla Firmicutes and Gemmatimonadetes but decreased the abundance of phylum Acidobacteria. Redundancy analysis indicated that organic C, dissolved organic N and C/N ratio were major environment variables affecting the distribution of bacterial phyla. These results suggest that manure application can improve soil fertility and productivity under reduced fertiliser-N application rate. and also increased For and increased the of copiotrophic but These results suggest that in the rotation application rate.


Introduction
Nitrogen fertiliser has been greatly used to enhance crop yields and plays a major role in global food security 1 . Less than 50% of applied N fertiliser are used by crops but 25% are emitted to the atmosphere and 20% are discharged to the aquatic system, which result in greenhouse effect and groundwater pollution 2 . Strategies for fertiliser-N reduction have been widely applied in the agricultural practices.
Manure application is the most important among fertilisation regimes, because it not only enhances nutrient use e ciency and the qualities of agricultural products, but also improves soil properties 3,4 .
Microorganisms play a critical role in driving nutrient cycling in the soils 5 , where their community are obviously affected by fertilisation regimes 6,7 . Generally, manure application usually increases microbial biomass and activities, whereas chemical fertiliser has relatively less effect on microbial biomass than manure 8,9 . Fertilisation regimes also result in microbial community shifts in the soils 8,10 . In addition, microbial community diversity and composition also vary with tillage measures, because these measures change the soil properties 11,12 . Therefore, the effects of agricultural practices on microorganisms still need further study in soils, especially fertilisation regimes.
The maize (Zea mays L.) and wheat (Triticum aestivum L.) rotation is a common rotation system on the North China Plain (NCP), where produces about 25% of China's food. Under this system, intensive tillage and excessive fertiliser-N use may not have a large impact on crop yields but can notably decrease N use e ciency 13 . Excessive fertiliser-N use has caused soil acidi cation and groundwater pollution 14,15 . With the maize-wheat rotation system, we consequently conducted a 2-year eld experiment under reduced fertiliser-N application rate. The objectives of this study were to: (i) ascertain the changes in crop yields and soil properties, (ii) quantify soil bacterial community diversity and composition, and (iii) determine the relationship between bacterial community composition and environment variables.

Materials And Methods
Experiment site. The experiment was conducted from 2017 to 2018 at the Duqu Agricultural Technology Extension Station (33°47′N, 113°49′E), Luohe city, Henan province, China. The altitude of experiment site is 80 m. Annual rainfall is 750-850 mm, and mean annual temperature is 14.6℃. Soil taxonomy is a Typic Dystruderts, which belongs to Vertisols in USDA. The tillage layer is concentrated at 0-20 cm depth band. Before this study, the soil properties at 0-20 cm depth band were: organic C (OC) 8.6 g·kg -1 , total N (TN) 0.64 g·kg -1 , available P (AP) 5.24 mg·kg -1 , available K 143.1 mg·kg -1 , pH 7.23. Experiment design. Treatments were conventional fertilisation (CF), 80% fertiliser-N input (80%N) and (iii) 80% fertiliser-N plus manure input (80%NM) in the wheat season, while the fertilisation regimes of previous maize were consistent. Each treatment was replicated four times randomly distributed in the eld, with a plot size of 5×6 m 2 . Detailed fertilisation and eld management were shown in Table 1.
Manure application rate was calculated based on 20% of net N rate. In the maize season, like local farmers, fertilisation and sowing were simultaneously implemented with an integrated machine. Before wheat sowing, 70% of fertiliser-N rate and manure were applied under 80%NM, and 70% of fertiliser-N rate was applied under both treatments of 80%N and CF. The reminding fertiliser-N rate was evenly broadcasted on the soil surface as top dressing at the jointing stage. The sowing densities of maize and wheat were 0.07 and 2.32 million plants per hectare, respectively. Agricultural practices kept consistent with local farmers. Sample collection and analyses. During maize harvest, 2 rows of plant samples from each plot were collected and air-dried to determine ear·hm -2 , grain number per ear and 100-grain weight. During wheat harvest, 2 m 2 plant samples from each plot were collected and air-dried to determine ear·m -2 , grain number per ear and 1,000-grain weight. Crop yield in each plot was determined at a 12% moisture content. After wheat harvest in 2019, ve samples at 0-20 cm soil depth from each plot were collected with an auger and mixed to give a bulk sample. Stone-free samples were thoroughly homogenized and divided into two-part subsample: some subsamples were air-dried and screened (0.25 mm mesh) to determine nutrient contents and pH, and the others were stored at -20℃ for microbial analysis.
The contents of OC, TN, inorganic N (including -N and -N) and AP in soils were determined according to the methods described by Bao 16 . Concentration of total dissolved N (TDN) was determined using an ultraviolet spectrophotometer (U-3900H, Hitachi, Japan) by an alkaline potassium persulfate oxidation method. Dissolved organic N (DON) was calculated as the difference between TDN and IN readings 17  Statistical analyses. Sequence data were processed using QIIME2. All sequence reads were trimmed and assigned to each sample based on their barcodes. Similar sequences were clustered into operational taxonomic units (OTUs) using a 97% identity threshold with mothur v.

Results And Discussion
Crop yields and yield components. The results of maize yields and yield components were shown in Supplementary Table S1, but no signi cant differences were observed between fertilisation treatments.
The differences in wheat yields between treatments were not signi cant, but wheat yield components were obviously affected by fertilisation treatments (Table 2). Compared with CF, ear·m -2 and 1,000-grain weight under 80%NM were increased by 5.1% (by 5.9% in 2018 and 4.2% in 2019) and 7.5% (by 7.6% in 2018 and 7.3% in 2019), respectively. This result may be due to manure application improves root growth environment, thus increasing total wheat tillers and nutrient absorption 21 . However, 1,000-grain weight under 80%N was decreased by 5.4% in 2018 and 4.7% in 2019, respectively ( Table 2). Note: Values (means±SD, n = 4) within the same column followed by different letters represent signi cant differences (P < 0.05).
Soil chemical properties. Among the measured indicators, those were no signi cant differences between different fertilisation treatments, except C/N ratio, DON and AP contents (Table 3). Soil C/N ratio and DON content were increased by 29.1% and 41.6% under 80%NM than those under CF, respectively. The result can be explained as manure application increased crop residues into the soil, and hence promote C accumulation and the concentration of dissolved nutrients 22 . For AP content, it was an 8.4% increase under 80%N when compared with CF. This result has been described by Akinnifesi et al. 23 , who attributed to a synergistic effect between N and P absorbed by crops. However, no signi cant difference in AP was found between 80%NM and CF. Note: Values (means±SD, n = 4) within the same row followed by different letters represent signi cant differences (P < 0.05) between treatments.
Bacterial community diversity and structure. A total of 98,948 high quality and chimera-free reads were obtained by MiSeq sequencing of 16S rRNA gene amplicon from 12 soil samples. After equalizing the sampling effort, 8,376 sequences from each sample were retained for analyses. Both sequencing and qPCR data showed that the relative abundance of bacterial community was more than 98.5%. The bacterial α-diversity parameters were listed in Table 4. In general, the reduction in fertiliser-N application rate increased bacterial α-diversity, compared with CF. Meanwhile, 80%NM signi cantly increased obvious species, chao1 estimator and Shannon's diversity index. This is related to manure application increase exogenous-C input, such as ne non-protected C, light fraction C and heavy fraction C, which improve C availability and stimulate microbial growth 24,25 . Shannon's diversity index 6.69±0.02b 6.71±0.03b 6.78±0.02a Note: Values (means±SD, n = 4) within the same row followed by different letters represent signi cant differences (P < 0.05).
PCoA and HCA were used to reveal differences in bacterial community structure between treatments (Fig.  1). PCoA showed that fertilisation treatments contributed 44.7% of total variation in the rst principal coordinate axis (Fig. 1A), where bacterial community structure was similar under CF and 80%N but became signi cantly (P < 0.01) different between 80%NM and CF. The result was con rmed by HCA, such as the treatments of 80%NM were roughly separated from those of CF, whereas the treatments of 80%N and CF were grouped together (Fig. 1B). This result can be explained that manure application increases phospholipid fatty acid content for bacteria 8 . Meanwhile, the increases in C quantity and availability have a priming effect on shifts in microbial community structure 26 .
Bacterial community composition. High-quality 16S rRNA gene sequences were clustered into 13,420 OTUs, with 1,054-1,156 OTUs each sample. The predominant phyla across all samples were Proteobacteria, Actinobacteria, Acidobacteria, Chloro exi, Firmicutes, Bacteroidetes, Gemmatimonadetes and Rokubacteria (> 1%), accounting for 88.7% of gene sequences in each sample ( Fig. 2A). Compare to CF, the abundance of phyla Firmicutes and Gemmatimonadetes under 80%NM were increased by 62.4% and 50.6%, respectively ( Fig. 2A). The phyla Firmicutes and Gemmatimonadetes belong to typical copiotrophic microorganisms and tend to grow fast in environments with rich nutrients, particularly organic nutrients 20 , and they play an important role in soil C and N cycle 27,28 . However, the abundance of phylum Acidobacteria under 80%NM was decreased by 39.3% when compared with CF ( Fig. 2A). This result may be due to the fact that most members of Acidobacteria were suggested to be slow-growing oligotrophic microorganisms 29 , and they are more adapted to nutrient-limited soil environment 30 . No signi cant differences in bacterial community composition were observed between 80%N and CF ( Fig.   2A).
Relationship between environment variables and bacterial community composition. Fig. 3 showed the relationship between environment variables and the distribution of bacterial phyla. RDA1 (horizontal) and RDA2 (vertical) axis explained 82.0% of total variability. RDA1 axis was signi cantly (P < 0.05) correlated with OC, DON and C/N ratio, which explained 67.9% of total variability. RDA2 axis was strongly associated with AP and IN, which explained 8.8% of total variability. According to the result of RDA, OC, DON and C/N ratio were major environment variables affecting the distribution of bacterial phyla.
Cluster heatmap was used to show the correlation between environment variables and the abundance of bacterial phyla (Fig. 4). A signi cant (P < 0.05) positive correlation could be found between phyla Firmicutes and Gemmatimonadetes and OC or C/N ratio, and between phylum Gemmatimonadetes and DON. However, there was a signi cant (P < 0.05) negative correlation between phylum Acidobacteria and OC.
Changes in environment variables regulate shifts in microbial community composition in soils 20 . In our study, OC, DON and C/N ratio were the major environment variables affecting bacterial community composition (Figs. 3 and 4). Nutrient availability, particularly C and N have been considered as the main factors driving shift in microbial community, since soil attributes may select some keystone species over others 20,31 .

Conclusions
In the maize-wheat rotation system, fertiliser-N plus manure input increased wheat ear·m -2 and 1,000grain weight under reduced fertiliser-N application rate, and it also increased the soil C/N ratio and dissolved organic N content, although it did not affect crop yields. For bacterial community, manure application improved bacterial diversity and increased the abundance of copiotrophic bacteria but decreased the abundance of oligotrophic bacteria. These results suggest that in the maize-wheat rotation system, manure application is more conducive to promoting soil fertility and productivity under reduced fertiliser-N application rate. Declarations