Evaluation of Microbial Diversity, Community Composition and Function in Mixed Cropping Systems Using Three Legume Species Under the Application of Biochar or Chemical Fertiliser

Mixed cropping systems involve utilising multiple crop species on the eld and diversifying aboveground plants. However, several contradicting results have been reported regarding their effects on soil microbial diversity. Therefore, to evaluate the effects of different leguminous species used in mixed cropping systems and the types of fertiliser on the diversity of soil microbes, a pot study was performed under maize/legume mixed cropping systems with one of three legumes, including cowpea [Vigna unguiculata (L.) Walp.], velvet bean [Mucuna pruriens (L.) DC.] and common bean (Phaseolus vulgaris L.) , and one of three types of fertiliser treatments, namely chemical fertiliser (CF), carbonised chicken manure (CM) or the lack of fertiliser (Ctr). 16S rRNA analyses were conducted using the soils sampled from each pot for soil bacterial diversity assessment, and Tax4Fun2 was used for bacterial functional prediction analysis. A decrease in microbial diversity after CM application was observed in the soil with velvet bean + maize (MM) compared to the Ctr treatment, whereas an increase in microbial diversity was observed in the soil with common bean + maize (PM) in the same condition. With CM application, the abundance of treatment-unique bacteria increased with PM treatment, whereas their decrease was observed with MM treatment. In contrast, the abundance of dominant microbes, including Thaumarchaeota, Chloroexi, Planctomycetes and Verrucomicrobia, was signicantly lower in PM but higher in MM after CM application. Functional prediction analysis indicated that the dominant bacteria were involved in CM decomposition processes and nitrication in MM treatment. Legume species-dependent factors, including nutrient absorption and root exudate composition, might be important concerning soil bacterial diversities.


Introduction
The expansion of agricultural lands has been the primary cause of biodiversity loss in terrestrial ecosystems (Bossio et al., 2005; Kehoe et al., 2017; Zabel et al., 2019). Particularly, modern agricultural practices, including monoculture cropping systems, the intensive use of inorganic fertilisers and pesticides, lead to soil degradation and loss of genetic diversity (Díaz et al., 2006;Rohr et al., 2019). Therefore, establishing e cient agricultural crop production systems compatible with biodiversity conservation is a global challenge for future food security.
Among different scales of biodiversity, the diversity of soil microorganisms is especially important for the stability of agricultural ecosystems. They can be considered the main drivers of biogeochemical reactions bene cial for soil health and crop productivity (Chaparro et al., 2012;Singh, 2015). For example, their involvement is notable in biogeochemical processes essential for plant health and growth, including nutrient absorption, immune function, pathogen prevention and stress tolerance (Loreau et al., 2001;Nannipieri et al., 2003). Thus, agricultural management systems to maintain or increase soil microbial diversity must be established.
Among various agricultural practices that can potentially diversify soil microbes, the use of mixed cropping systems is receiving heightened attention. Legume-based intercropping systems have been reported to enhance soil microbial diversity and bacterial functions, including the mineralisation of available phosphorus (P) and nitrogen (N) (Gao et al., 2010; Li et al., 2013;Lian et al., 2009;Gul et al., 2015;He et al., 2008). For example, the combined use of biochar composed of plant residues and legume-based intercropping systems could enhance the soil microbial functionality important for nutrient absorption by plants, such as N-xation and P-solubilisation in the rhizosphere (Duchene et al., 2017;Liao et al., 2019;Liu et al., 2017). Also, biochar application increases available N to plant because biochar increases soil pH and stimulates the access of ammonia-oxidising bacteria to NH 4 + and reduces gaseous N loss from soil by reducing the denitri cation potential (Lu . Furthermore, in mixed cropping systems, N mineralisation of biochar, the process of transformation of organic N to mineral N through soil microorganism activities, This study aimed to elucidate plant species-speci c effects of cereal/legume mixed cropping on soil microbial community structures, focusing on the interaction among legume species and fertiliser types. It was hypothesised that the combination of legume species and fertiliser types is important for bacterial diversity and community structure and the abundance of rare bacterial species. Further, the growth of a rare microbial community in diversi ed treatment was assumed to have an important role in the degradation process of complex chemicals, especially focusing on organic N degradation. Thus, a greenhouse experiment of legume-maize mixed cropping was performed using three legume varieties and chemical or biochar [carbonised chicken manure (CM)] to measure bacterial diversity and community structure and their functionality based on 16S rRNA analysis.

Soil sampling
The soil used in this experiment was sampled from abandoned land for 30 years at the university farm located at the Field Science Centre for Northern Biosphere, Hokkaido University, Japan (43°04¢N, 141°20¢E). The properties of the soil are given in Table 1. The soil type was clay loam with 44.6% sand, 21.5% silt and 33.9% clay.

Experimental design
A pot experiment was performed in a greenhouse at the Graduate School of Hokkaido University. The sampled soil was air-dried and sieved with 2 mm mesh and subsequently lled into Wagner pots (surface area = 1/5000 a). Each pot contained 1.8 kg airdried soil. The experimental design was completely randomised, included three fertiliser treatments × four mixed cropping treatments, and conducted in trireplicates. The pots received one of the three types of fertiliser treatments, namely control ('Ctr'), chemical fertiliser containing P and K ('CF') or biochar made from CM (50 g pot −1 carbonised CM; 'CM'). The application rate for CF was 30 kg P ha −1 and 50 kg K ha −1 . The soil and the CM chemical property are described in Table 1. . Three replications were included in each treatment. During these treatments, maize and three legume seeds, including cowpea, velvet bean and common bean, were sprouted for 2 weeks in small pots lled with vermiculite before transplantation to Wagner pots. During the experiment, the temperature was maintained at 25°C to 30°C, and the plants were grown for 50 days after transplantation.

Chemical property analysis
After plant growth soil was sampled from each pot, soils were sampled and measured for pH and extractable NH 4 + and NO 3 − concentrations 50 days after transplantation. For soil pH, 6 g soil was shaken for 30 min with 30 mL Milli-Q water, and pH was subsequently measured by a pH sensor (AS800; ASONE Co., Japan). For extractable NH 4 + and NO 3 − , the samples were extracted with a KCl solution (2 mol L −1 ), followed by colorimetric analysis using a ow injection analyser system (ACLA-700; Aqualab Co., Ltd., Japan). A two-way analysis of variance (ANOVA) was then performed to investigate the interaction between environmental factors and experimental treatments ( Another PCR was performed with the utilisation of amplicon-obtained products to make them Ion Torrent sequence sample-speci c. To achieve this, the 515F forward primer with the Ion Xpress Barcode Adapters Kit sequence and the 806R reverse primer attached to the Ion Xpress sequence of the Ion P1 adaptor were used (Thermo Fisher Scienti c K.K.). The rst PCR products were diluted to 2000 ng mL −1 , and 1 μL of each product was subsequently mixed with 10 μL AmpliTaq Gold® 360 Master Mix, 0.4 μL of the forward primer, 0.4 μL of the reverse primer and 7.2 μL nuclease-free water. The second PCR cycle was set to 95°C for 10 min and then 5 cycles at 95°C for 30 s, 57°C for 30 s and 72°C for 1 min, followed by 72°C for 7 min. The second PCR products were puri ed in accordance with the same method outlined above. The nal length and concentration of the amplicons were con rmed using a Bioanalyzer DNA 1000 Kit (Agilent Technologies, USA). The library was subsequently diluted to 50 pM and loaded into the Ion 318 chip using Ion Chef Instruments with an Ion PGM Hi-Q Chef Solutions. The samples were sequenced on an Ion PGM Sequencer with Ion PGM Hi-Q View Sequence Solutions (Ion Torrent Life Technologies, USA). Sequence data were deposited in the Sequence Read Archive of the National Center for Biotechnology Information (NCBI) under accession number PRJNA743765.

Sequence processing
The barcoded 16S rRNA gene sequences were denoised, quality-ltered and assessed using the DADA2 algorithm implemented in Quantitative Insights Into Microbial Ecology (QIIME2) and with its work ow (Bolyen et al., 2019). Rarefaction was performed with minimal reads among all samples, and sequence data were subsampled to 41,095 sequences per sample. The R package Vegan (version 2.5.6) was used to access sample depth and generate the a-rarefaction curve plot (Fig. S1). The rarefaction curve was then evaluated using the interval of a step sample size of 1000.

Measurement of bacterial abundance
To measure bacterial abundance, quantitative PCR (qPCR) was performed using the extracted DNA, diluted 50 times with nucleasefree water. The 515F/806R primer pairs described above were used to amplify the V4 region of the 16S rRNA. For the standard curve, the PCR products from the DNA extracted from the Ctr pots were used, which were puri ed with AMPure XP and further diluted to ve stages of different concentrations. The samples were prepared with 10.4 μL KAPA SYBR Fast qPCR kit (Kapa Biosystems, USA), 0.08 μL of the forward primer, 0.08 μL of the reverse primer and 2 μL diluted DNA extract. Nuclease-free water was added to achieve the nal volume of 20 μL. CFX96 TouchÔ Real (Bio-Rad Laboratories, Inc., USA) was used, and the cycling condition was set to 95°C for 30 s and then 35 cycles at 95°C for 30 s, 58°C for 30 s and 72°C for 1 min, followed by 95°C for 1 min and subsequently 55°C to 95°C by 1°C increment for 10 s. Ct values (threshold cycle) were calculated after quantifying the ampli cation results using qpcR R package version 1.4.1.

Statistical analysis
To quantify the diversity of soil microbial communities, the Shannon index (Shannon, 1948) and the Simpson index (Simpson, 1949), an estimation of community a-diversity, were calculated. For each diversity index, two-way ANOVA was performed using fertiliser treatments and plant species as factors with the emmeans R package version 1.4.7 (Lenth et al., 2020). Multiple comparisons were subsequently performed using the Tukey-Kramer method. Also, permutational ANOVA (PERMANOVA; permutation = 9999) was conducted, and signi cant phyla (p < 0.001) were subsequently identi ed for the interactions between fertiliser treatments and plant types using two-way ANOVA.

Diversity indices and abundance of bacterial communities
The interaction effects between plant species and fertilisation were shown by soil microbial diversity indices. Within PM treatment, a signi cantly lower Simpson's diversity indicator was observed with no fertilisation (Ctr) compared to CM treatment ( Fig. 1). In contrast, a relatively lower diversity was shown by MM treatment with CM application compared to Ctr treatment.
Bacterial absolute abundance, based on qPCR analyses, showed a signi cant effect of plant treatments when averaged across fertiliser types (Table S2). The bacterial abundance of SM and VM treatments was signi cantly different. However, there was no correlation between bacterial absolute abundance and diversity indicators.
The analysis of the soil 16S rRNA gene sequence provided 738 to 1196 OTUs per sample. Venn diagrams showed that soils under Ctr, CF and CM treatments had 297, 293 and 283 core OTUs (the overlapped area across plant treatments), respectively (Fig. 2) and increasing numbers of unique OTUs with increasing diversity indices, for example, unique OTU numbers increased from 28 to 84 for PM compared to Ctr and CM, whereas it decreased from 52 to 35 in MM. The relative abundance of these unique bacteria within the whole bacterial abundance increased from 1.9% (Ctr) to 5.2% (CM) in PM treatment but decreased from 4.3% (Ctr) to 1.9% (CM) in MM treatment (Fig. 3). The abundance of Acidobacteria and Verrucomicrobia was suggested by the community structures of these unique bacterial taxa to dominate within unique microbial communities and was correlated to the increase of unique OTU numbers during comparison of different treatments. In contrast, the abundance of Gemmatimonadates, Chloro exi and Planctomycetes was not correlated to the increase in unique OTU numbers, although they were considered to be the dominant phyla within unique OTUs.

Environmental factors and dominant OTU composition
The relative abundance of Thaumarchaeota, Armatimonadetes, Chloro exi, Planctomycetes, Verrucomicrobia and Proteobacteria contributed to the changes in community structures, and their abundance was in uenced by the interactions between plant species and fertiliser treatments ( Table 3; Fig. S2). Within CM treatment, Thaumarchaeota, Chloro exi, Planctomycetes and Verrucomicrobia were signi cantly higher in MM treatment than PM treatment (p < 0.05; Table 4).
The signi cant increase in soil pH in CM treatment was due to the content of salt-based ions, such as potassium, sodium and calcium, which can reduce the exchangeable hydrogen ions in the soil (Table 2). Also, CM application signi cantly increased NO 3 − -N concentration (p < 0.001), but NH 4 + -N concentration had no signi cant difference (p = 0.90).

Functional prediction by Tax4Fun2
Two-way ANOVA analysis indicated that gene abundance coding ammonium oxidation, carbamate kinase, glutamate dehydrogenase, nitrate reductase, nitrite oxidoreductase and nitrite reductase (NO-forming) was signi cantly in uenced by the interaction effects of plant type and fertiliser (Table S1). In Ctr treatment, gene abundance coding ammonium oxidation and nitrite reductase (NO-forming) were higher in mixtures of legume and maize (Fig. 4). In contrast, in CF, there was no difference in gene abundance coding ammonium oxidation and nitrite reductase (NO-forming) between single and mixed cropping treatments. CM treatment most altered gene abundance among other fertilisation treatments. Especially, with CM application, MM treatment showed the highest abundance in glutamate dehydrogenase, carbamate kinase, ammonium oxidation and nitrate reductase.

Characteristics of unique bacteria
The diversity index and unique bacterial abundance in PM were signi cantly higher in CM treatment, whereas CM negatively affected the diversity and unique bacterial abundance in MM ( Fig. 1 and 2). There was no universal effect of mixed legume species and fertiliser types on soil bacterial diversities. There was also no consistent statement about microbial diversity in multiple cropping systems using cereals and legumes in previous studies.

Enzyme activity related to the N cycle
Functional prediction analysis with Tax4Fun2 indicated that without fertiliser (Ctr) the presence of legume facilitated nitri cation and denitri cation with higher NO 3 − concentrations in the soil ( Fig. 4;  Sangakkara et al., 1996), and this study showed more nodulation in MM and PM with CF treatment (Table  S3), the low gene abundance in the mixed treatment indicated that N de ciency occurred between maize and legume through CF application. Moreover, compared to Ctr and CF, CM treatment dynamically affected N cycling in each cropping system because of their organic N contents. In MM with CM treatment, the abundance of ammonia oxidation and nitrite reductase (NO-forming) was higher. The abundance of carbamate kinase and glutamate dehydrogenase, which mediate amino acid metabolism, was higher than other plant treatments. A higher abundance of those genes might be related to the fast decomposition of organic N in CM and the supply of more N to plants and the greater growth of coplanted maize in MM (Table S3). In contrast, the lower abundance of those enzymes in PM might be related to the slower growth of maize. In previous research, intercropped maize (Z. mays L.) with wheat (Triticum aestivum L.) or faba bean (Vicia faba L.) showed maize growth suppression in the initial cultivation period to avoid nutritional competition with neighbour species (Li et al., 2011). In this study, a similar pattern appeared in PM, whereas little suppression was observed in MM. It is also consistent with a previous study with intercropping using velvet bean, which showed fewer maize growth suppressions or even better growth than monocropping (Akobundu et al., 2000;Correia et al., 2014). These studies on weed reduction by cover crop effects of velvet bean were signi cant with maize growth, but they lacked focus on the mechanism to avoid N competition between velvet bean and maize. Thus, this study provided a better understanding of how velvet bean maintains greater growth of maize than other legumes in terms of the bacterial community driving N cycling.
While comparing community structures at the phylum level within CM-applied soils, the abundance of Thaumarchaeota, Chloro exi, Planctomycetes and Verrucomicrobia was signi cantly higher in MM compared to PM (Table 4). Especially, Thaumarchaeota, an archaea ubiquitously present in a wide variety of ecosystems, would contribute to the increase of ammonium-oxidising function

Conclusion
Mixed cropping systems diversify soil bacterial communities only under a speci c combination of legume species and fertilisers. Soil bacterial diversity was increased when biochar was used for the common bean-based mixed cropping system with an increase in the number of unique OTUs (treatment-speci c OTUs). Especially, plant root associated bacteria such as Verrucomicrobia and Actinobacteria were increasing in diversi ed treatment.
With biochar application, bacterial functionalities, such as ammonia oxidation and denitri cation, were higher in the velvet beanbased mixed cropping system with a low diversi ed bacterial community. Some dominant bacterial phyla, such as Thaumarchaeota, Planctomycetes, Verrucomicrobia and Chloro exi, contributed higher enzyme abundance related to organic N decomposition in the mixed cropping system with velvet bean and biochar.
Further efforts are needed to con rm the effects of plant root exudates on soil microbial community and diversity in association with biochar application. Also, as a limitation of the pot study, only bacterial diversity and functionality were evaluated early on cultivation. Therefore, eld research to evaluate the long-term effects of mixed cropping systems on the bacterial diversi cation process and functionality associated with the variability of legume species and fertiliser types is required.

Declarations
Funding This work was supported by JSPS KAKENHI grant numbers 18KK0183 and 18H02310.

Con icts of interest/Competing interests
To the best of our knowledge, the named authors have no con ict of interest, nancial or otherwise.

Availability of data and material
Sequence data were deposited in the Sequence Read Archive of the National Center for Biotechnology Information (NCBI) under accession number PRJNA743765 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA743765). The other datasets during and/or analysed during the current study available from the corresponding author on reasonable request.

Code availability
The code used in this study are available from the corresponding author on reasonable request.

Ethics approval
Not applicable Consent to participate

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Supplementary.docx