Saffron rhizosphere samples were collected from 10 locations including eight saffron fields from Kashmir district (Wuyan, Patalbagh_upper range, Patalbagh_lower range, Alchibagh, Galandar, Chandhara, Hatiwara, Samboora); one from Berwar saffron field from Kishtwar, Jammu district and one laboratory experiment conducted in Bengaluru (Figure 1). The details of all the saffron fields, geographical coordinates (latitude and longitude), fertilizers usage, total production, and total area of cultivation have been tabulated in Table 1. Composite sampling was performed from all the fields wherein the Crocus sativus rhizosphere samples were collected from the three corners of five fields of each location and all the 15 samples per location were pooled together to form composite samples (Ambardar and Vakhlu, 2013; Ambardar, 2014). Rhizosphere samples were collected by uprooting the plants and vigorously shaking by hand. The soil that remained adheres to the roots after vigorous shaking was taken as rhizosphere soil. The roots were washed in normal saline (0.85% NaCl) that was taken further for metagenomic DNA extraction.

A separate experiment was conducted in which saffron corms were collected from Kashmir, Pampore during the dormant stage of the saffron life cycle (July, 2016). These corms were then planted in 10 pots (5 corms/pot) filled with the garden soil at Trans-Disciplinary University, Bengaluru (Karnataka). These pots were kept in open under natural conditions (temperature 20°C) for a period of 3 months (August–November 2016). The rhizosphere samples from 15 plants were collected randomly and pooled together to form a composite sample during the flowering stage similar to field samples. This experiment was performed to address the question whether the root microbiome is soil driven or plant driven. Root microbiome was harvested using same design and method as in case of field samples.

Metagenomic DNA was extracted from rhizosphere samples using a MoBio Power soil DNA extraction kit (MoBio Laboratories Inc. Carlsbad, CA, USA) following the manufacturer's instructions. The DNA quality and quantity were determined by using a Nano Drop device (Thermo Scientific, Wilmington, DE) and electrophoresis on 0.8% agarose gel, with 1 kb plus ladder as molecular weight marker.

A 16S rDNA sequencing library was constructed targeting the V3 and V4 hyper-variable regions of the 16S rDNA gene, according to the 16S rDNA metagenomic sequencing library preparation protocol (Illumina, San Diego, CA, USA). The PCR was performed with 12 ng template DNA and region-specific primers with Illumina index and sequencing adapters (forward_primer: 5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGTCGTCGGCAGCGT CAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG-3′; reverse primer: 5′GTCTCGTG GGCTCGGAGATGTGTATAAGAGACAGGTCTCGTGGGCTCGGAGATGTGTATAAGAGAC AGGACTACHVGGGTATCTAATCC-3′) using KAPA HiFi Hot Start Ready Mix (KAPA Biosystems, Wilmington, MA, USA). The amplicons were purified using the Agencourt AMPure XP system (Beckman Coulter Genomics, Brea, CA, USA). Subsequently, purified PCR products were visualized after gel electrophoresis and quantified with a Qubit dsDNA HS Assay Kit (Thermo Scientific) on a Qubit 3.0 fluorometer. The PCR products of all the samples were pooled to 4 nM concentration, prior to sequencing, and were analyzed on an Agilent 2,200 Tape Station (Agilent Technologies, Santa Clara, CA, USA) for quantity and quality analysis. The pooled sample (4 nM) was denatured with 0.2 N NaOH, diluted further to 4 pM concentration, and combined with 20% (v/v) denatured 4 pM PhiX as control, following Illumina's guidelines. Samples were sequenced on the MiSeq sequencing platform (Illumina) using a 2 × 250 cycle V3 kit, following standard Illumina sequencing protocols.

Raw data of 16S rDNA sequencing was analyzed for a quality check using the FastQc tool kit (Brown et al., 2017). The adapter and low quality reads (Q < 30 Phred score) were trimmed using the Cutadapt tool of Trimgalore (Jiang et al., 2014). The reads with phred score above 30 were selected for QIIME1 pipeline analysis (version 1.9.1). The QIIME 1 pipeline (Quantitative Insights into Microbial Ecology version 1.9.1) was used to process and filter multiplexed sequence reads. The sequencing reads were grouped into an operational taxonomic unit (OTUs) that were further clustered against GreenGenes_13_8 sequences to ascertain the taxonomical affiliation of bacteria (McDonald et al., 2012). Reads failing to hit the reference were subsequently clustered de novo at the 97% similarity level using the UCLUST greedy algorithm. Chimeric sequences were identified by the UCHIME algorithm of USEARCH61 and removed. OTU sequences were aligned using PYNAST. OTU taxonomy was determined using the Green Genes taxonomy database (Caporaso et al., 2010).

The bacterial community associated with each sample was compared for alpha and beta diversity analysis using Qiime 1 (Sinclair et al., 2015). For alpha diversity analysis, rarefaction curves and diversity indices like Chao1, Shannon, Simpson, phylogenetic diversity were calculated for estimating the richness and evenness in all the samples (Chao, 1987; Gotelli and Colwell, 2011). For rarefaction curves, the operational taxonomy units OTU table was rarified to an even depth of 1,00,000 sequences per sample, to avoid biases generated by the difference in sequencing depth. Rarefaction curves (97% identity) were plotted between the number of observed OTUs (cluster count) and the number of reads of the samples. In order to evaluate β diversity, PCOA plots, venn diagram and heat maps were constructed. In addition, a circos plot was generated for representing the core rhizo-bacteriome across all locations using the tool Circos Version 0.63-9 (Krzywinski et al., 2009).

Bacterial diversity associated with all the 10 rhizospheric samples was cataloged into 16 phyla, 261 genera and 73 species (Table 2). Out of 16 phyla, Proteobacteria was the dominant phyla in all the samples with higher relative abundance in Kashmir (avg 59.1%) followed by Bengaluru (49%) and Kishtwar (31.28%). However, Firmicutes and Planctomycetes were relatively abundant in Kishtwar (20, 10%) as compared to Kashmir (avg 4.9, 3.12%) and Bengaluru (5, 4%) respectively (Figure 2).

Out of total of 261 bacterial genera cataloged, the abundance of 4 bacterial genera namely Bacillus, Lysobacter, Rhodoplanes and Janthiobacterium were observed. Bacillus was dominant at four locations namely Kishtwar (54%), Hatiwara (24%), Patalbagh_lower range (22%) and Bengaluru (15%), (Figure 3). However, Bacillus was also found co-dominant with other genera at two locations namely Wuyan [Bacillus (16%) and Rhodoplane (17%)]; and Galandar [(Bacillus (12%) and Lysobacter (13%)]. The abundance of Bacillus was comparatively less in rest of the four locations i.e., Alchibagh (4%), Chandhara (7%), Samboora (8%), and Patalbagh_upper (11%). Janthiobacterium was dominant in Alchibagh (19%). Lysobacter was the most abundant bacterial genera in two locations namely Patalbagh_upper range (15%) and Samboora (18%), whereas Lysobacter (15%) shared dominance with Rhodoplanes (14%) in Chandhara (Figure 3).

53 genera out of 261 bacterial genera were common in all the samples, are considered as the core rhizo-bacteriome that has been discussed in detail under core rhizo-bacteriome section. on comparing all the 10 samples, unique genera specific to each location were also identified. The maximum number of unique bacterial genera were identified in Wuyan (19) followed by Alchibagh (10), Galandar and Kishtwar (6 each), Chandhara (4), Patalbagh_lower range and Samboora (2 each), Patalbagh_upper range, Hatiwara and Bengaluru (1 each), (Figure 4).

73 bacterial species were present in all the samples, out of which, 3 bacterial species i.e., Bacillus flexus, Lysobacter brunescens and Janthiobacterium lividum were dominant; but their abundance varied (Figure 5). Bacillus flexus was abundant in seven locations namely Kishtwar (49%), Patalbagh_lower range (38%), Hatiwara (35%), Bengaluru (31%), Galandar (28%), Wuyan (22%) and Patalbagh_upper range (20%). However, Janthiobacterium lividum (19%) was abundant in Alchibagh (Figure 5). Lysobacter brunescens was abundant in rest of the two locations namely Samboora (14%) and Chandhara (40%). Out of 73 bacterial species, 8 bacterial species were considered as the core rhizo-bacteriome (common in all the samples) that has been discussed in core rhizo-bacteriome section. In addition to core species, each sample has unique bacterial species that were specific to that particular location and absent in all other samples. Kishtwar had four unique bacterial species namely Bacillus marisflavi, Macrococcus caseolyticus, Myroides odoratimimus and Staphylococcus scuiri whereas two species were unique in Alchibagh namely Methylotenera mobilis and Sphingomonas suberifaciens and Galandar namely Paenibacillus barengoltzii and Paracoccus marcusii. Only one unique species was cataloged from Patalbagh_upper range (Vellonellas dispar), Wuyan (Paenibacillus ginsengarvi) and Samboora (Roseomonas mucosa). However, Bengaluru, Hatiwara, Chandhara and Patalbagh_lower range did not have any unique bacterial species (Figure 6).

The bacterial diversity of each sample was also analyzed using the alpha diversity indices i.e., Chao1, Simpson, phylogenetic diversity and Shanon (Table 3), rarefaction curves (Figure 7), PCOA plots (Figure 8) generated by Qiime software. Rarefaction curves (97% identity) of rhizosphere samples did not reach plateau indicating bacterial diversity was well-represented but could increase on repetitive sampling (Figure 7). The bacterial community in saffron rhizosphere from Samboora field was diverse as compared to other sample, as depicted by higher number of different species (OTUs) in the rarefaction curve. This result was further complemented by diversity indices such as Chao1, phylogenetic diversity, Shannon and Simpson indexes were also higher in the case of Samboora (Table 3). Beta diversity analysis using PCOA plots was done to evaluate the significant variation in bacterial diversity among all the samples. Beta bacterial diversity of Kishtwar was significantly different from Kashmir and Bengaluru, as it does not cluster with the rest of the samples in PCOA plots (Figure 8).

Despite the variation in the rhizobacterial community associated with saffron plant across different locations, the core rhizo-bacteriome that remains constant in the saffron growing across different geographical locations was identified. Core rhizo-bacteriome in saffron consisted of 11 phyla, out of total of 16 phyla i.e., Proteobacteria, Chloroflexi, Firmicutes, Acidobacteria, Cyanobacteria, Gemmatimonadetes, Bacteroidetes, Nitrosiprae, Planctomycetes, Verrucomicrobia and Actionobacteria. At genera level, it constituted 53 out of 261 genera and eight bacterial species out of 73 bacterial species. The bacterial species that constituted the core comprised of Bacillus flexus, Bacillus muralis, Edaphobacter modestum, Lysobacter brunescens, Pseudoxanthomonas mexicana, Psychrobacter celer, Roseateles depolymerans and Variovorax paradoxus. The core rhizo-bacteriome in saffron is also well-represented in heat maps at genus and species level and with circos plot (Figures 9–11).

Out of 16 bacterial phyla, Proteobacteria was the most abundant phylum in all the samples (Figure 2). This is in accordance with a previously published report on saffron rhizosphere by our group (Ambardar and Vakhlu, 2013; Ambardar et al., 2014, 2016). Proteobacteria have been also reported to be the dominant phyla in the rhizosphere of other plants such as, Gossypium hirsutum, Artemisia argyi, Ageratum conyzoides, Erigeron annuus, Bidens biternata, Euphorbia hirta and Viola japonica (Qiao et al., 2017; Lei et al., 2019). Proteobacteria are known to regulate nutrients (carbon, nitrogen, sulfur) cycling in the environment that enhance plant growth (Rampelotto et al., 2013; Mukhtar et al., 2020). In our recent publication, Proteobacteria was also found to be the most abundant phylum in the cormosphere of saffron from Kashmir and Kishtwar, as well as in Morocco (Bhagat et al., 2021). On comparing all the samples, the rhizo-bacterial diversity of saffron growing in Kishtwar was found distinct from the rest of the samples as the relative abundance of Proteobacteria (PRO-31.28%) was comparatively less whereas the relative abundance of Firmicutes (FIR-20%) and Planctomycetes (PLA-10%) were more as compared to Kashmir (PRO-59.1%, FIR-4.9%, PLA-3.12%) and Bengaluru (PRO-49%, FIR-5%, PLA-4%), (Figure 2). This was further complemented by beta diversity analysis wherein rhizo-bacterial diversity of saffron grown in Kishtwar did not cluster with the rest of the samples based on the PCOA plots and heat maps at genus and species level (Figures 8–10) respectively. Similar to rhizo-bacteriome, cormo-bacteriome of Kishtwar was found significantly different than that of Kashmir and surprisingly, relatively similar to Morocco in our recent report (Bhagat et al., 2021). The significant variation in the rhizo-bacterial diversity of saffron growing in Kishtwar from rest of the samples can be preliminarily attributed to corms (plant) as Kishtwar is closer to Kashmir geographically than Morocco. Though saffron is reported to be a sterile monogenetic triploid plant, with no genetic variation and reproduces vegetatively by corms (Nemati et al., 2019), however, the variation in the yield and quality of saffron has been reported worldwide (Cardone et al., 2021). Climate and soil are thought to be two major abiotic factors (Cardone et al., 2019, 2020) and epigenetic influences have been reported to be one of the biotic factors to affect the quality and yield of saffron (Chen et al., 2021). In the preliminary observation, it seems that Cormo/rhizo-bacteriome has a role to play and is “plant driven” because rhizo-bacteriome of saffron grown in different climates and soil at Bengaluru was similar to that of Kashmir. The reason was the corms that were collected from Kashmir and grown in Bengaluru indicating that rhizo-bacteriome in saffron could be plant driven.

Comparison of rarefaction curve (97% identity) of all the samples indicated that the bacterial community in saffron rhizosphere from Samboora field in Kashmir was more diverse as compared to other samples, as the number of different species (OTUs) was higher here. This result was further complemented by diversity indices like Chao1, phylogenetic diversity, Shannon and Simpson indexes which were also higher in the case of Samboora (Table 3). However, higher bacterial diversity in the rhizosphere of saffron growing in Samboora field does not affect the saffron production positively as saffron yield in Samboora (2.5 kg/hectare) was less than that of Wuyan fields (5 kg/hectare) that had less diversity than Samboora.

The dominance of three bacterial species, Bacillus flexus, Lysobacter brunescens and Janthiobacterium lividium were observed in the rhizo-bacteriome of saffron grown in all the 10 locations (Figures 3, 5). Bacillus flexus was dominant in the rhizosphere of saffron grown in seven locations namely Kishtwar, Bengaluru, Wuyan, Galandar, Hatiwara, Patalbagh_upper range and Patalbagh_lower range. Lysobacter brunescens was dominant in Chandhara and Samboora; and Janthobacterium lividium in Alchibagh (Figure 5). Dominant bacterial genera identified in the saffron rhizosphere (Bacillus, Lysobacter, Janthobacterium and Rhodoplanes) have been reported from plant rhizospheres such as saffron (Ambardar and Vakhlu, 2013; Ambardar, 2014; Ambardar et al., 2014, 2016), green pepper (Liu et al., 2019), grapevine (Sacca et al., 2019) and oilseed rape (Gkarmiri et al., 2017), etc. However, only Bacillus flexus have been reported in Limoneum sinense and rice (Roy et al., 2020; Xiong et al., 2020) whereas Lysobacter brunescens and Janthobacterium lividium have not been reported from any plant so far.

In our previous studies, Pseudomonas was found dominant in the saffron rhizospheres by cloning based 16S rDNA metagenomic approach wherein full length 16S rDNA was amplified, TA cloned and sequenced using Sanger sequencing (Ambardar et al., 2014). Surprisingly, in the present study, Pseudomonas was only present in the saffron field from Alchibagh and absent in all other samples. The absence of Pseudomonas could be attributed to DNA extraction protocol and PCR primers and cloning approaches that were different from the present study. In a previous study, the metagenomic DNA was extracted manually using four protocols and pooled together and full length 16S rDNA primers followed by TA cloning were used to cataloge the diversity (Ambardar et al., 2014), whereas, in the present study, a commercial kit was used for DNA extraction followed by amplification of only V3-V4 region of 16S rDNA. In addition to this, the previous sampling was done in the year 2009 and present sampling was done after about 5 years in 2016, therefore climate change and floods in between in Kashmir can also be probable reasons, though this needs further investigation.

Various bacterial species, being reported in the present study, have been also reported by our group earlier by cultivation based method and were characterized in-vitro and in-vivo as well (Ambardar and Vakhlu, 2013; Ambardar et al., 2014; Kour et al., 2018; Bhagat et al., 2021). However, in the present study, six bacterial phyla, 212 bacterial genera and 70 bacterial species have been reported from the rhizosphere of saffron for the first time. Though, the rhizo-bacteria cataloged in the present study have been reported from other plants, but not reported from the cormosphere and rhizosphere saffron by our group earlier.

Comparative analysis of rhizo-bacteriome of saffron across different sites revealed the presence of various common bacterial phyla, genera and species representing the core rhizo-bacteriome of saffron. Core rhizo-bacteriome in saffron was represented by 53 bacterial genera and eight bacterial species namely Bacillus flexus, Bacillus muralis, Edaphobacter modestum, Lysobacter brunescens, Pseudoxanthomonas mexicana, Psychrobacter celer, Roseateles depolymerans, Variovorax paradoxus. These bacteria mostly belonged to 11 bacterial phyla namely Proteobacteria, Chloroflexi, Firmicutes, Acidobacteria, Cyanobacteria, Gemmatimonadetes, Bacteroidetes, Nitrosiprae, Planctomycetes and Actionobacteria. Core rhizo-bacteriome in saffron has been illustrated in the heat maps at genus & species level and with circos plot (Figures 9–11) respectively. Recently, core bacteriome associated with the corms of saffron were compared across different geographical locations i.e., Kashmir, Kishtwar and Morocco. 24 bacteria genera belonging to the phylum Proteobacteria, Actinobacteria, Bacteriodetes and Verrumicrobia made the core of cormo-bacteriome in saffron (Bhagat et al., 2021). Core rhizo-bacteriome in saffron represents 20.3% (53 out of 261) of the whole rhizo-bacteriome which was found comparatively less than the core cormo-bacteriome i.e., 32.8% of total cormo-bacteriome, however the number of bacterial genera constituting the core was more in rhizosphere (53 out of 261) as compared to cormosphere (24 out of 73), (Bhagat et al., 2021). Core microbiome of various plants have been studied such as wheat (Kuzniar et al., 2020), tomato (Cordero et al., 2020), Gymnadenia conopsea (Lin et al., 2020), coffee (Fulthorpe et al., 2020), switch grass (Grady et al., 2019), Phaseolus vulgaris (Pérez-Jaramillo et al., 2019), Oryza sativa (Eyre et al., 2019), Vineyards soil (Coller et al., 2019), Dalbergia spruceana (Skaltsas et al., 2019), and Salvia miltiorrhiza (Chen et al., 2018), etc. Core microbiome of 21 Salvia miltiorrhiza seeds represented 54% of the whole microbiome cataloged from seven different geographic origins (Chen et al., 2018). In the case of Phaseolus vulgaris rhizosphere, the core microbiome represented 25.9% of the total microbiome in native and agricultural soils (Pérez-Jaramillo et al., 2019). The core microbiome of 30 phylogenetically diverse angiosperm plants constitutes 40% of the whole microbiome (Fitzpatrick et al., 2018). Core rhizo-bacteriome in saffron was 20.3% of the total microbiome, which was comparatively less than Salvia miltiorrhiza, Phaseolus vulgaris and Angiosperm plants. In rice seeds, the core microbiome was enriched in Rhizobium, Pantoea, Sphingomonas, Methylobacterium, Xanthomonas, Paenibacillus, Alternaria, and Occultifur (Wang et al., 2020). Saffron's core rhizo-bacteriome was enriched with Bacillus, Rhizobium, Sphingomonas, Agrobacterium, and Methylobacterium which was similar to the core microbiome in rice.

Out of 53 core rhizosphere bacteria identified in the present study, 19 bacteria were also reported previously from the rhizosphere and cormosphere of saffron grown in Kashmir, by our group using culture dependent and independent approaches (Ambardar and Vakhlu, 2013; Ambardar, 2014; Ambardar et al., 2014, 2016; Kour et al., 2018; Bhagat et al., 2021). Interestingly, the core rhizo-bacteriome of saffron also harbored some common bacteria as in previously published core cormo-bacteriome. 11 bacterial genera namely Agrobacterium, Chitinophaga, Dyadobacter, Mesorhizobium, Methylobacterium, Nocardioides, Opitutus, Rhizobium, Sphingomonas, Streptomyces and Varivorax were common in both the underground organs of saffron that is corm (Bhagat et al., 2021) and roots. In addition, Agrobacterium, Bacillus, Delftia, Rhizobium, and Variovorax have also been reported from rhizosphere and cormosphere of saffron, grown in Morocco using cultivation dependent methods (Chamkhi et al., 2018). Though we could not capture the genera reported in Morocco cormo/rhizosphere by earlier used techniques but with the present technique they were captured and found to be present in Kashmir. However, 34 bacterial genera of the core rhizo-bacteriome of saffron grown in India were reported for the first time in the saffron, in the present study.

Eight core bacterial species identified in the present study was also reported from other plants such as Bacillus flexus from rice (Roy et al., 2020), Bacillus muralis from peanut (Jiang et al., 2016), Edaphobacter modestum from Magnolia grandiflora (Stone and Jackson, 2016), Pseudoxanthomonas mexicana from maize and peanut rhizosphere (Geng et al., 2018; Youseif, 2018), Roseateles depolymerans from ginger (Chen et al., 2014), Variovorax paradoxus from mustard roots (Belimov et al., 2005). However, Lysobacter brunescens and Psychrobacter celer has not been reported from any plant so far.

64.1% of total core rhizo-bacteriome i.e., 34 out of 53 bacteria has been reported as PGPBs from saffron and other plants (Supplementary Table 1). While exploring the core rhizo-bacteriome for PGPR, it was found that 21 bacteria have been reported as PGPB for the first time from saffron in the present study as previously used methods failed to capture them. The core rhizo-bacteria that are being reported as PGPRs for the first in saffron are Azospirillum, Agromyces, Agrobacterium, Bradyrhizobium, Burkholderia, Dokdonella, Edaphobacter, Flavobacterium, Flavisolibacter, Frankia, Leuconostoc, Luteimonas, Lysobacter, Phenylobacterium, Pseudoxanthomonas, Mesorhizobium, Nocardia, Novosphingobium, Rhizobium, Rubellimicrobium, Variovorax etc.

In addition to core rhizo-bacteriome, saffron rhizo-bacteriomes of each location were enriched with some specific bacteria that were unique to that particular location. An inventory of unique PGPBs of all these bacteria across different locations was made based on the previous reports of PGPBs from saffron and other plants. The location specific unique PGPBs details have been summarized in Supplementary Table 2. Wuyan had the maximum number of unique genera i.e., 19 out of which 5 are reported as PGPB i.e., Actinoplanes, Beijerinckia, Desulfotomaculum Dethiobacter and Ureibacillus. In other locations unique number of genera was 4 in Chandhara, 2 in Samboora, 1 in Hatiwara and 1 in patalbagh_upper range but none of them have been reported as PGPB. Kishtwar had six unique genera and three genera namely Desemzia, Macrococcus and Myroides have been reported as PGPB (Supplementary Table 2). In addition, Alchibagh had 10 unique genera and four have been reported as PGPB. Galandar had six unique genera and three PGPB reported, Patalbagh_upper had two and Bengaluru has one unique genera, both locations have one PGPB reported (Supplementary Table 2).

In the present study, Bacillus flexus, Lysobacter brunescens and Janthiobacterium lividium were the dominant bacterial species identified in saffron rhizo-bacteriome. Bacillus genus dominant in the rhizosphere microbiome is predominantly used as plant growth promoting bacteria, due to their ability to colonize the roots rapidly, competitive colonization potential, production of various phytohormone and conversion of complex nutrients such as phosphorous and nitrogen into simple absorbable forms (Aloo et al., 2019; Enebe and Babalola, 2019; Hashem et al., 2019; Kashyap et al., 2019). Bacillus flexus KLBMP 4941 in Limoneum sinense (Xiong et al., 2020) and B. flexus U8 in rice (Roy et al., 2020) have been reported as PGPR. However, Lysobacter brunescens and Janthobacterium lividium have not been reported so far from any plant.

In addition to Bacillus flexus other bacterial species of core rhizo-bacteriome reported as PGPB are Bacillus muralis, Pseudoxanthomonas mexicana, Roseateles depolymerans and Variovorax paradoxus (Supplementary Table 3). Bacillus muralis strain HS4 has been reported as PGPR in peanut (Jiang et al., 2016). Pseudoxanthomonas Mexicana has been reported as PGPR from rhizosphere of maize and peanut (Geng et al., 2018; Youseif, 2018). Roseateles depolymerans has been isolated as an endophyte from the seedling stage of ginger and reported to have PGPR properties (Chen et al., 2014) and Variovorax paradoxus has been reported as PGPR from Indian mustard roots (Belimov et al., 2005). Lysobacter brunescens, Edaphobacter modestum and Psychrobacter celer have not been reported as PGPB so far (Supplementary Table 3).

Among different locations, Kishtwar had four unique species (Bacillus marisflavi, Macrococcus caseolyticus, Myroides odoratimimus and Staphylococcus scuiri) and all are reported as PGPB in different plant. In addition, Galandar from Kashmir had only one PGPB (Paracoccus marcusii) reported. All other locations have no unique species reported PGPB so far (Supplementary Table 4).

Altogether, 110 bacterial genera (out of total 261) and 21 bacterial species (out of total 73), cataloged in present study, have also been reported to be plant growth promoting bacteria from saffron and other plants (Supplementary Table 5). Plant growth promoting bacterial species cataloged in saffron are Acidovorax facilis, Bacillus cereus, Bacillus firmus, Bacillus flexus, Bacillus horikoshii, Bacillus marisflavi, Bacillus muralis, Brevundimonas diminuta, Burkholderia bryophila, Burkholderia tuberum, Luteibacter rhizovicinus, Macrococcus caseolyticus, Myroides odoratimimus, Paracoccus marcusii, Pseudoxanthomonas mexicana, Rhizobium leguminosarum, Roseateles depolymerans, Staphylococcuss epidermidis, Staphylococcus equorum, Staphylococcus sciuri, Staphylococcus succinus and Variovorax paradoxus.

On comparison of PGPB from each sample, Wuyan had maximum number of PGPBs (76 out of 154) followed by Galandar (75 out of 132), Alchibagh (73 out of 147), Chandhara (72 out of 143), Hatiwara (73 out of 140), Kishtwar (66 out of 139), Bengaluru (63 out of 143), Samboora (62 out of 124), Patalbagh_lower (61 out of 111) whereas Patalbagh_upper (60 out of 122) with the minimum number of PGPBs (Supplementary Table 6). The highest number of PGPB in the Wuyan sample could be correlated to the yield as the production in Wuyan is two times as compared to other fields. In the previous report by our group, PGP Bacillus amyloliquefaciens W2 for growth promotion of saffron and corm rot inhibition were isolated from the Wuyan by cultivation based method (Gupta and Vakhlu, 2015). However, in present study a well-reported PGPR Rhizobium leguminosarum and Luteibacter rhizovicinus was relatively abundant in all the nine locations, including Wuyan and completely absent in Kishtwar. Rhizobium leguminosarum possess various PGP characteristics such as phosphorous solubilization, siderophore, IAA, HCN production and nitrogen fixation which enhance the growth of host plants in pea plant, lettuce and carrot (Mishra et al., 2009; Tank and Saraf, 2010; Flores-Félix et al., 2013; Gopalakrishnan et al., 2015). Luteibacter rhizovicinus MIMR1 have been reported as plant growth promoting bacteria that promotes root development in barley (Guglielmetti et al., 2013). Absence of Rhizobium leguminosarum and Luteibacter rhizovicinus may be attributed to the low production status of Kishtwar as compared to Kashmir, but the real confirmation will come only after their evaluation in-vivo. The presence of unique location specific PGPB at places other than Wuyan, with a lower yield than Wuyan indicates that the potential of PGP bacteria varies and maybe all the PGPB are not as effective as that of Wuyan fields. The PGPB cataloge in the present and previous studies in saffron by metagenomic approaches will be captured by culture based technique and evaluated for PGP potential in the future.