The experimental design consisted of planting M. arvensis in the earthen pots (12″ dia) that were supplemented with chemical (NPK at 37.5, 60, and 50 kg ha^–1, respectively)/organic fertilizers (VC at 19.76 t ha^–1 and FYM at 10 t ha^–1) (Reddy, 2005; Maji et al., 2017) in the month of February (first plantation). VC of the same origin (prepared from distilled waste of lemongrass) has been used for this study. Sterilized VC and FYM were used for all the treatments to minimize the impact/interference of microbiota present in VC and FYM. For the experimental pots where green manures/other crops followed after the first harvest, only basal dose chemical fertilizers (NPK at 37.5, 60, and 50 kg ha^–1, respectively) were given. The pots receiving chemical/organic fertilizers were maintained with Mentha throughout the experiment. The crop was harvested after 90 days in the month of May. The pots were replanted with Mentha and/or other crops in rotation and were harvested in August (second plantation). To check the influence of monocropping of Mentha, Mentha with some rotational crop and MA with ample use of organic fertilizers on its productivity, all the pots were replanted with Mentha (third plantation) that were harvested in the month of October (see Supplementary Material for Experimental plan). Mentha pots receiving only chemical fertilizer and a non-aromatic crop (maize) were also included in the study, but on the basis of preliminary results (TRFLP), only samples exhibiting prominent changes [Mentha (M), Mentha + VC (M + VC), Mentha_Sesbania (M_S), and Mentha_Chlorophytum (M_C) of the second plantation] were included for the NGS analysis.

Plant weight of above ground herb was recorded and essential oil extraction in the fresh herbage was performed at the time of first and third harvesting by using hydro-distillation in Clevenger’s apparatus (Maji et al., 2013). The oil yield (grams per pot) was calculated by multiplying the amount of oil obtained (ml) with a value of specific gravity (0.89) of Mentha oil.

The soil samples were collected in triplicate from the fields of CSIR-CIMAP continuously grown with MAPs, i.e., A. annua, A. paniculata, M. arvensis, O. basilicum, C. zizanioides, P. graveolens, B. monnieri, and C. citratus. To ensure that change in rhizospheric microbial diversity is a result of the cultivation of the selected crops, soil samples were taken from the field just before the harvesting of crops. The rhizospheric soil (soil firmly adhering to the root) was collected and stored at −20°C for further experimentations. To reduce the sample to sample variation, three to five samples from each of the three plots were drawn and mixed to form a plot sample and three such plot samples were mixed to form a composite sample for each treatment for NGS analysis.

Total DNA extraction was done by using the Power Soil DNA Extraction Kit (Mo Bio Laboratories) according to the manufacturer’s instructions. Isolated DNA from soil samples was checked qualitatively and quantitatively on 0.8% agarose gel electrophoresis and Nanodrop yield, respectively. PCR was performed using a standard protocol (Bharti et al., 2015). PCR mixtures containing 25 ml of Premix Taq (Takara Biotechnologies), 2 ml of the DNA template, and 1.5 ml of each primer were made up to a volume of 50 ml with sterilized Milli-Q water. The samples were amplified in an Eppendorf Thermal Cycler. Tetra base cutter restriction enzymes AluI and HhaI (Promega) were used to digest the fluorescently labeled PCR amplicons.

For TRFLP, 1 μl of the restriction-digested product was mixed with 12 μl of freshly deionized formamide containing 0.5 μl of LIZ 500 marker (PE Applied Biosystems). 3130xl Genetic Analyzer (Applied Biosystems, United States) of Gene Scan mode (Moeseneder et al., 1999) was used for detection as well as for the separation of DNA fragments. Raw data from Gene MapperTM were exported to T-REX, online software for the processing and analysis of T-RFLP data¹ (Culman et al., 2009). Raw files were analyzed by the Past 2.02 software package (Hammer et al., 2001).

Soil samples from experiment 1 were outsourced to Genotypic Technology Pvt. Ltd., and soil samples from experiment 2 were outsourced to Bionvid Technology Pvt. Ltd., for metagenomic studies. Samples outsourced were processed and sequenced by using Illumina MiSeq as per sequencer manufacturer’s protocol. Sequence data have been deposited in the NCBI Sequence Read Archive (SRA) database. The sequences were submitted to NCBI, and the accession numbers obtained are SUB2488937 and SUB2604678 for 16S and SUB2604678 and SUB2659621 for ITS for experiment 1 and SRP150923 for 16S and SRP150978 for ITS for experiment 2.

The dataset was analyzed using Qiime2 v 1.9.0 by using the Greengenes3 v 13.8 database at the backend. The dataset was clustered at 97% similarity and mapped to the reference using the UCLUST ref method. Abundance graphs were plotted based on the number of hits. Comparative taxa summary plots and heatmap were generated across the samples at phyla level considering relative abundance (RA) values using a cutoff greater than 0.1%. Krona charts were plotted using Krona5 tools for each sample. Pie charts were generated for each sample from phylum to species with a cutoff of 0.5% based on the absolute abundance count. Alpha diversity was calculated using Chao1, observed operating taxonomic units (OTUs), Shannon, and Simpson index. All the data and graphs were analyzed by R software (Vegan and ggPlot packages), SPSS 16.0, and PAST 2.02 (details in Supplementary Material).

Five grams of soil samples are used for analysis to check the signatures of essential oil/artemisinin in bulk and rhizosphere soil samples of maize, Mentha, and Artemisia using standard protocols (Gupta et al., 2002; Dundek et al., 2014).

The pH of collected soil samples was determined using Equiptronic’s pH meter as described by Jackson (1967). The electrical conductivity (EC) of a soil sample was determined by Equiptronic‘s digital EC bridge. Nitrogen (N) in the soil is mainly present in organic form together with small quantities of ammonium and nitrate ions and was determined by distilling soil with alkaline potassium permanganate according to Subbiah and Asija (1956). Soil-available phosphorus (P) is found as orthophosphate and was extracted by Olsen’s extract 0.5 M NaHCO₃ (pH 8.5) (Olsen et al., 1954) and measured calorimetrically by the ascorbic acid method (Murphy and Riley, 1962). The ammonium acetate method followed by flame photometric detection (Hanway and Heidel, 1952) was employed to estimate the available potassium (K) of samples. Organic carbon (OC) of all the soil samples was measured using the chromic acid titration method proposed by Walkey and Black (1934).

All statistical analyses [one-way ANOVA and Correspondence Canonical Analysis (CCA)] for physicochemical properties of soil, plant biomass, and oil yield were carried out using SPSS 16.0, PAST 2.02.

Plant residues are an important and decisive source of nutrients for the microbial community (Dilly et al., 2000; Schloter et al., 2003). However, the presence of some unique and structurally diverse bioactive molecules called secondary metabolites causes MAPs to harbor a distinctive and specific microbiome (Wu et al., 2015). Reduction in the total soil microbial diversity of Mentha and Artemisia as compared to maize was reflected by the diversity analysis, which could be attributed to the presence of such bioactive compounds with antimicrobial properties as revealed by GC-MS/HPTLC studies of soil. These bioactive volatile compounds may act as phytoncides for some group of microbes, but the community shift stabilizes the community structure by maintaining the evenness. Components of essential oil in this study, which were found in rhizospheric and bulk soil are known for their antimicrobial properties. While menthyl acetate and menthol are relatively more effective against the bacteria, neomenthol and limonene have been found to be more effective against fungi (Sokovic et al., 2008; Kuriata-Adamusiak et al., 2012; Vieira-Brock et al., 2017). The presence of limonene in both bulk and rhizospheric soil could have played a role in affecting fungal diversity, whereas the greater reduction of fungal diversity in bulk soil might be because of the presence of neomenthol, which was not detected in the rhizospheric soil. In case of bacterial diversity, the presence of menthol and menthyl acetate in both the soils (bulk and rhizosphere) could have resulted in greater reduction in bacterial diversity as compared to the control where no such peaks were detected.

Artemisinin mainly found in trichomes of leaves (Muranaka and Saito, 2010) and excessive leaf fall may release more of this compound in the soil post litter decomposition and its antimicrobial property might have affected both the bacterial and fungal diversity, which was more profound in bulk soil than in rhizospheric soil (Tajehmiri et al., 2014), which again could be correlated with the presence of relatively higher amounts in bulk soil. Moreover, cultivated plant type has a profound influence in determining the microbial species composition of the rhizosphere resulting in distinct community structures for different plant species grown in the same soil (Grayston et al., 1998; Miethling et al., 2000; Patkowska and Konopiński, 2014) explaining the significant difference found in the diversity pattern of Mentha and Artemisia as compared to maize and other MAPs included in the study.

Species diversity is a community parameter, which is related to the degree of stability of that community (Bardgett et al., 2014). In the rhizosphere, the plant selects those microbes that can utilize the nutrients efficiently from total microbial diversity (Hartmann et al., 2009; Berendsen et al., 2012), thereby creating a selection pressure. The higher number of OTUs in the maize rhizosphere was because of the presence of high amounts of simple sugars in the root exudates that attracted bacterial and fungal communities in a manner similar to the establishment of mycorrhizal associations (Broeckling et al., 2008; Berg and Smalla, 2009).

In bulk soil of both Mentha and Artemisia, the presence of leaf fall with antimicrobial activity caused a reduction in the number of OTUs as compared to rhizospheric soil. Mentha was less severely affected probably because it is a short-duration (3–4 months) crop and the leaf fall with antimicrobial essential oil is volatile in nature, whereas Artemisia is a long-duration crop with continuous leaf fall continuously leaving artemisinin into the soil, which is not volatile and less degradable as compared to the Mentha essential oil. As the leaf fall covers more of bulk soil, artemisinin content was found to be higher in bulk soil, so the results are more marked in AB than in MB.

Bacterial commune responded less severely than the fungal commune to the presence of bioactive compounds in the soil. The unicellular and smaller size of bacteria make them exposed only to their immediate surroundings, i.e., microniche (Vos et al., 2013) with very specific conditions separating them from the direct influence of plant root, and this could be the reason why bacterial communities are not significantly different under different plants (Urbanova et al., 2011). In the case of fungi, the root-associated mycelia extending from the rhizosphere to the bulk makes them more sensitive to plantation type and other factors (Urbanova and Baldrian, 2015). Similar observations were made in our study where bacterial OTUs were not considerably affected in Mentha and Artemisia, whereas the decrease in fungal OTUs was drastic.

In case of Mentha, high leaf fall at the time of maturity and their decomposition probably improved the growth of degrading bacteria, which may attract diverse communities capable of utilizing sugars (Jiang et al., 2017) as well as secondary metabolites, making it more diverse and richer than rhizospheric bacterial community as observed in MB soil. The outcome of these complex processes is the development of a rhizospheric microbial community that differs noticeably from the source communities of bulk soil (Fan et al., 2017). However, the above condition reversed in AB soil, where considerable defoliation occurred in the bulk soil increasing the signatures of secondary metabolites with little input of sugar source, decreasing both bacterial and fungal community richness/diversity.

At the phylum level, known bacterial phyla were detected, which accounted for 99.5%, while the unknown ones occupied only about 0.5% of the total population. Among all the phyla identified, the members of Proteobacteria, Actinobacteria, Firmicutes, and Acidobacteria were highly abundant in all the soils. Proteobacteria is a functionally diverse group of fast-growing Gram-negative bacteria, especially plant growth promoters (Gupta, 2000) and possess the ability to utilize a broad range of root-derived carbon substances (Philippot et al., 2013a, b), making their population highest in each soil type. Actinobacteria have an important role in the decomposition of organic material (Baldrian et al., 2012) and thus play an important role in organic matter turnover and nutrient cycling. Increased abundance of Actinobacteria in MB soil may replace or reduce the abundance of other microbes, including those found useful in improving productivity and reducing the population of phytopathogens. Firmicutes are spore-forming phyla that live in the soil in small pockets of anaerobic habitats created by them, thereby reducing the effect of secondary metabolite (fallen litter of Mentha) firmicutes, making them highly abundant in MB soil samples. Members of phylum Acidobacteria are oligotrophic and enriched in soils with very low resource availability, whereas Chloroflexi is an anaerobic heterotroph that plays an important role in litter decomposition (Purahong et al., 2016). Acidobacteria and Chloroflexi are slow-growing entities (Jones et al., 2009) but were substantially higher in abundance in MB and AR, probably because of the sufficient amount of litter availability and nature of root exudates, respectively. Besides this, a higher abundance of Cyanobacteria, Planctomycetes, Bacteroidetes, and Verrucomicrobia were found in rhizospheric soil. All the phyla members decreased drastically in both bulk and rhizospheric soil of Artemisia except Chloroflexi and Cyanobacteria, which were found to be higher in AR as compared to CR. In all the cases, bacterial diversity was higher than fungal diversity, and bacterial communities were more even.

In the case of a fungal community, at the phylum level, Ascomycota and Basidiomycota are known to produce various organic matter digestive enzymes, thereby occupying the highest abundance. Glomeromycota forms arbuscular mycorrhizal association with roots of higher plants, so they were found to be higher in rhizosphere as compared to bulk soil of Mentha and Artemisia. Members of Zygomycota and Chytridiomycota (known to utilize dead plant material) were higher in bulk soil, as bulk soil contained more organic matter due to high litterfall, which served as an effective nutrient source for them. More unidentified phyla were found in bulk soil in comparison to rhizospheric soil samples in all the plant type. This suggests that although a vast number of microbes have been identified, there is still a significant proportion of unidentified microbial genera (Uroz et al., 2010).

Biodiversity includes species richness (the number of species) and evenness (how well distributed abundance or biomass among species), which are considered among the most important measures (Wittebolle et al., 2009). The community evenness confers the functional stability of an ecosystem and highly uneven communities or extreme dominance of one or a few species results in an ecosystem vulnerable to environmental stress (Chapin et al., 2000; Wittebolle et al., 2009). In our study, higher evenness in MA indicates the existence of stable microbial community, which may provide functional stability during environmental fluctuations (Balvanera et al., 2005; Awasthi et al., 2014).

Monocropping damages the soil ecology and crop productivity by depleting the soil nutrient, resulting in an unbuffered niche (Marais et al., 2012). The same crop when grown incessantly causes a continuous loss of the similar type of nutrients, creating a nutrient shock for the microbial community, thereby decreasing the soil microbial diversity and increasing the abundance of only some specific candidates (Marais et al., 2012). In the case of Mentha, a considerable leaf fall of essential oil-bearing leaves could have greatly affected the microbial diversity because of the antimicrobial nature of these oils. This was clearly reflected in our experiments where the soil that was continuously cropped with Mentha had low microbial diversity. The other objective of the study was to look at the ways that could restore the microbial diversity losses by the introduction of some organic/green manures and selected crops which have shown some promise in improving microbial diversity and abundance in our preliminary studies (unpublished data). Sesbania, a nitrogen fixer (Dreyfus and Dommergues, 1980), may have supported the abundance of phosphate solubilizers secreting higher amounts of organic acids, lowering the soil pH, and thereby influencing the microbial diversity. As green manure, Sesbania might have increased the organic matter of soil supporting the microbial growth and increasing microbial diversity as reflected in our study. In our experiment, higher numbers of OTUs were observed in M_S, and M + VC cultivated soil as compared to other plants. Chlorophytum borivilianum (safed musli), an eminent medicinal plant, also improved the evenness of the microbial community, which could be functionally more efficient, thereby playing a role in enhancing productivity. Amendment of organics (VC/FYM) helps the soil to mitigate the imbalance caused by monocropping (Xiong et al., 2015). The higher water holding capacity and organic matter content of VC makes it highly suitable for replenishment of microbial diversity (Lim et al., 2015). It indicates that after the leaf fall of Mentha in the soil (which starts after about 70 days), secondary metabolite/essential oils are immediately released to the soil environment because of the presence of oil glands on the surface of the leaves. These essential oils may cause a drastic reduction in the microbial richness including degraders initially, which may cause a nutrient shock for the microbes. On the other hand, the presence of organic matter and abundant nutrients in VC may alleviate microbes from such stress. Later, with the loss of volatile essential oils from the soil, there could be a revival of many degraders, which may start the degradation of defoliated leaves and provide nutrients to the plants. Though samples were not collected at the initiation of leaf fall in Mentha, it is assumed that the loss of diversity could be much higher at that time than at the time of harvesting, which occurs after 90–100 days. Our results suggest that the effect of MAPs cultivation on soil microbial diversity is difficult to generalize under negative or positive impacts. However, it induces the shifts in microbial community structure, which may affect soil health and productivity. Some communities, by using organic and green manure as well as other MAPs, rescue itself or resilience takes place, which will be beneficial for soil health. Higher abundance of decomposers and growth-promoting microbes may minimize the effect of essential oil-bearing leaves by quickly decomposing the leaves aiding in faster volatilization of oil, and the resultant manure may support the rest of the microbiome to colonize efficiently there.