Rice Root-Associated Endophytic Microbiome in Correspondence To Different Levels of Salinity At Indian Sundarban Areas


 The present study attempted to analyse rice root endogenous microbial diversity and their relationship with soil salinity and physicochemical factors in the salt stressed region of Sundarbans, India using amplicon metagenomics approaches. Our investigation indicates, the unique microbiome at slightly acidic nutrient enriched non-saline zone characterized by microbial genera that reported either having plant growth promotion (Flavobacterium, Novosphingobium and Kocuria) or biocontrol abilities (Leptotrichia) whereas high ionic alkaline saline stressed zone dominated with either salt-tolerant microbes or less characterized endophytes (Arcobacter and Vogesella). The number of genera represented by significantly abundant OTUs was higher at the non-saline zone compared to that of the saline stressed zone probably due to higher nutrient concentrations and the absence of abiotic stress factors including salinity. Physicochemical parameters like nitrogen, phosphorus and potassium were found significantly positively correlated with Muribaculaceae highly enriched at the non-saline zone. However, relative dissolved oxygen was found significantly negatively correlated with Rikenellaceae and Desulfovibrionaceae, enriched in the non-saline soil. This study gives a well resolved picture of microbial community composition impacted by salinity and other rhizospheric soil factors.


Introduction
Rice (Oryza sativa L.) being one of the major staple food for more than half of the world's population signi cantly contribute to the nation's food security (Kunda et al. 2021). In India, the state of West Bengal is the highest producer of rice (Biswas et al. 2020). Within West Bengal, rice is grown in about 1 million hectares of land under the coastal ecosystem comprising 'the Sundarbans' (Banerjee et al. 2018). The world's largest coastal wetland, the Sundarbans, is an ecological hotspot and recognized for its signi cant biodiversity (Pramanik et al. 2019). The socioeconomic status of the inhabitants of Sundarbans is mostly are dependant on farming where about 70% of farmers primarily grow rice crops (Ghosh and Mistri 2020). But Sundarbans facing the Bay of Bengal is a climatically vulnerable region exposed to severe impacts from climate change. Sea-level rise, increased frequency of cyclonic storms, erratic rainfall, and ine cient centuries-old river embankments result in saline ingression into the mainland thereby increasing the soil salinity (Nath et al. 2021). Salt stress generates reactive oxygen species (ROS), decreases osmotic potential, and reduces water availability to plant roots, followed by ion toxicity thereby hindering the growth and development of rice plants (Hussain et al. 2019). Moreover, during higher salt accumulation, Na+ ions replace other essential cations present in the soil, increasing the compactness of soil and decreasing oxygen availability in the root zone (Vaishnav et al. 2019). These changes in the soil cause nutrient de ciencies and sodium toxicity in plants thereby turning fertile paddy elds into barren lands and ultimately reducing rice yield (Vaishnav et Kumar et al. 2021). However, these studies were undergone with a limited number of samples and even lack comparative information about non-saline soil from the same region. Additionally, an elaborative and well depth knowledge about the role of rhizospheric soil parameters for shaping endophytic microbiome diversity at saline ecosystem has not been established yet.
Nonetheless, this will be the rst attempt to encounter all the above-mentioned problems with the following objectives: 1) Characterizing rice root endophytic microbial assemblages in the saline coastal zone of Sundarbans, West Bengal.
2) Identi cation of core and unique microbiota in rice plants grown in saline and non-saline region.
(3) Predicting the contributions of relevant soil properties on rice root endophytic community composition, if any. This study will expand the understanding of overall rice endophytes' role on the sustainability of cultivating the rice in the Sundarbans area concerning various environmental indices that may lead to prospective innovations in an agricultural eld.

Site selection and sample collection
In the present investigation, two coastal saline rice elds named Dabu and Godkhali in the Sundarbans region, West Bengal, were carefully chosen in collaboration with the expertise of ICAR-Central Soil Salinity Research Institute (CSSRI), West Bengal, India. Four different sampling sites (S1, S2, S3 and S4) from these two closely associated rice eld stations representing both saline and non-saline soils were chosen so that we could avoid maximum disparity of location variation based microbial diversity (Fig S1). Total twenty-four independent rice plants (Oryza sativa; cultivar WGL20471, locally known as LalMiniket) from these four sites were collected at their vegetative stage during April 2019. The plants were randomly chosen and dug out carefully to prevent any damage to the roots. The rhizospheric soil (RS) for all samples was also carefully collected. Immediately sampled material was packaged in sterile bags (Himedia) placed on ice and brought back to the laboratory for further processing within 24 hours.
The salinity status of RS collected from these four sites were measured based on electrical conductivity (EC) of the saturated-paste extracts (EC e ) using Thermo Scienti c Orion Star 140 TM Series Multimeter (Thermo Fisher Scienti c, U.S). Based on the calculated result of soil salinity status, the sites were classi ed into two different zones: Sundarbans non-saline zone (SNSZ; S1, S2 and S3; 18 samples) and Sundarbans saline zone (SSZ; S4; 6 samples) ( Table   1). Soils were considered to be saline if EC e was greater than 4 dS/m at 25°C (Shrivastava and Kumar 2015).  (Bremner 1960). Soil organic matter (SOM) was analysed according to (Tabatabai 2015  were provided by Euro ns, Germany. The raw FASTAQ sequences were submitted to NCBI under BioProject PRJNA681119.
As we are only focused on the sequences of bacterial 16S rRNA gene, but there could be good chances of contamination with mitochondrial and chloroplast DNA sequences, other prominent strategies were adapted to exclude these contaminations.

Sequence data analysis
All the raw FastQ datasets) were processed following the protocol of (

Statistical analysis
Principal component analysis (PCA) was conducted to ordinate the sampling sites based on their measured RS physicochemical parameters. The measured environmental parameters and nutrient contents that signi cantly differ among the two zones were examined using the Wilcoxon rank sum test.
α diversity indices were calculated based on species richness estimators (average number of OTUs, Chao1 and ace) and the diversity index (average inverse Simpson Index, Shannon diversity) to assess overall microbial diversity within the zones after randomly rarefying the data set repeatedly to the minimum library size (28588 sequences). Signi cant differences in α diversity indices were assessed using nonparametric, unpaired p-value adjusted Wilcoxon tests. The relationship between α diversity and soil parameters (if any) was also assessed by the Spearman's Rank correlation test.
Prior to β-diversity calculation, the dataset was pruned to exclude the rare biosphere by retaining only those OTUs that were present in at least two sequences within more than 10% of samples. The β-diversity patterns were not changed by this reduction in the data sets (Mantel test, r = 0.99, P = 0.001). The change in the composition of the microbial communities (β-diversity) between two zones was explored by analysis of similarity (ANOSIM). This was con rmed by calculating Bray-Curtis dissimilarity coe cients, thereby generating cluster dendrogram as well as non-metric multidimensional scaling (NMDS). The env t function of the vegan package was further used to get the correlation of the most signi cant variable with overall microbial communities (Clarke and Ainsworth 1993; Oksanen et al. 2016).
The sequence counts were further clr-transformed using aldex.clr function of the R package ALDEx2, using median of 128 Monte Carlo Dirichlet instances (Fernandes et al. 2014). Differently abundant OTUs among SNSZ and SSZ were identi ed in the reduced data set of each size fraction at a parametric FDRadjusted and a non-parametric unadjusted signi cance threshold of 0.05 were plotted as Dot plot.
Correlation analyses of soil physicochemical factors and the microbial community at phylum level were carried out based on Spearman correlation coe cient using R package Hmisc (Miscellaneous and Yes 2021).
The core and unique microbial communities of the two different studied zones: SNSZ and SSZ were also identi ed in genus level using venny2. 1

Endophytic microbial diversity and community composition inhabiting in rice root samples:
In total, we obtained 50, 04,200 raw reads from the amplicon sequencing which were quality trimmed to procure 47, 33,273 (average 197220) number of high quality pair-end reads corresponding to 33, 09,331 (average 137889) merged sequences that led to 521843 swarm OTUs. Further elimination of singleton, mitochondria, chloroplast and unclassi ed OTUs at phylum level we obtained a total 40408 OTUs (Table S1).
Each of the bacterial species richness estimators (nOTUs, Chao1 and ace) were signi cantly higher in SNSZ compared to SSZ (Wilcoxon Test, p-value < 0.5). Bacterial species evenness indicated by abundance based coverage estimator invS and Shannon's diversity index measuring combined species richness and evenness were higher for SNSZ ( Fig S2).
Although bacterial genus Rhizobium (57.9% of all the sequences) was the most abundant in both the zones, SSZ showed higher relative abundance followed by Bacteroides. On the other hand, higher proportions of Muribaculaceae Incertae Sedis, Lachnospiraceae NK4A136 group and Pseudomonas were found in SNSZ (Fig: 2). Among 77 total common bacteria termed as core bacteria, the aforementioned bacterial genera were the most dominant common bacterial groups prevalent in both zones. Apart from them, 7 bacterial genera found exclusively at SSZ were Vogesella, Arcobacter, Saccharimonadaceae Incertae Sedis, Gmella, Woesearchaeia Incertae Sedis, Bergeyella and GCA-900066575 (Lachnospiraceae family) while 8 bacterial genera were found only in SNSZ: Undibacterium, Peptotococcaceae Incertae Sedis, Prolixibacteraceae Incertae Sedis, Flavobacterium, Kocuria, Novosphingobium, Leptotrichia and Shewanella (Fig S3).
To further identify the OTUs responsible for the patterns in endophytic bacterial community composition, the differences in the proportion of individual OTUs between the SNSZ and SSZ were tested using ALDEx2. Total  Gammaproteobacteria, Pantoea (OTU32) and Pseudomonas (OTU10) tended to be higher at SNSZ. Helicobacter (OTU7) belonging to Camphylobacteria, was signi cantly enriched at SNSZ (Fig S4).

Changes composition of endophytic rice root microbial community
At OTU level, variation in microbial communities between SNSZ and SSZ was explored using ANOSIM which didn't show any signi cant difference (ANOSIM, R= -0.08, p-value>0.05). This nding was supported by cluster dendrogram based on Bray-Curtis dissimilarities which also unable to show any distinctive pattern in bacterial community composition between SNSZ and SSZ (Fig. 2). The NMDS ordination didn't show any distinct separation between SNSZ and SSZ (Fig: 3). Env t function showed soil parameters EC e , EK, RDO, AP, ON and TDS appeared to be strongly correlated with the patterns in bacterial community composition (Table S2).

Connecting environmental parameters and microbial community
The correlation between the physicochemical parameters and relative microbial abundance at the family level was evaluated based on Spearman correlation coe cient analysis (Fig. 4). The abundance of the family Muribaculaceae at SNSZ was positively correlated with nutrient parameters ON, AP and EK (r > 0.5; p 0.05). In addition, AP was positively correlated with the abundance of Rikenellaceae and EK with Ruminococcaceae (r > 0.5; p 0.05). However, RDO concentration was signi cantly positively correlated with Rhizobiaceae (r >0.5; p <0.5) and signi cantly negatively correlated with Rikenellaceae and Desulfovibrionaceae (r<-0.5; p 0.05).
We also observed abundance of bacterial family Muribaculaceae at slightly acidic nutrient enriched non-saline soils positively correlated with nutrient parameters ON, AP and EK. SNSZ comprising Rikenellaceae was positively correlated with AP which was in line with the previous nding where the abundance of this family increased with higher phosphorus concentration (Latif et al. 2018). Moreover, RDO was found to be signi cantly positively correlated with one of the most abundant family Rhizobiaceae which was well compared with the previous investigation where their abundance was usually observed near the oxic surface of the polygon, primarily in the studied rim but was signi cantly negatively correlated with Rikenellaceae and Desulfovibrionaceae (Liebner et al. 2008). concentrations and dissolved solids, TDS at saline rice paddy eld, SSZ. Therefore, soils of SSZ can be referred to as high ionic alkaline saline soils. Thus, SSZ is characterized by ion toxicity and osmotic potential imbalance due to high salinity while non-saline SNSZ is marked by a balance of soil nutrients and organic matter.

Variation in microbial richness and evenness across the studied zones
In this study, we found a signi cant decrease in species richness and evenness estimators along SSZ compared to SNSZ indicating bacterial community richness decreases with increasing salinities which support the previous ndings (Zhao et al., 2020). However, the differences in these estimators within SSZ and SNSZ are probably not only because of salinity as well as various other environmental and nutrient parameters. Since other physicochemical factors also changed along with the salinity of the Sundarbans, the observed bacterial community composition may be a result of parameters that correlate with salinity.

Relative abundance of microbial taxa found in two zones
In this study, among seven bacterial genera which are unique to SSZ, Arcobacter was previously reported as rice endophytes from the same region (Kunda et al. 2018). Gemella and Saccharimonadaceae were reported to be present in heavy metal contaminated saline soil (Marzan et al., 2017). The abundance of Vogesella is frequently observed in alkaline and rhizospheric soil of saline tolerant pokkali rice plant, therefore its presence as an endophyte in rice plants at SSZ is quite familiar (Rameshkumar et al., 2016). Archaeal family Woesearchaeia Incertae Sedis and Bergeyella were previously observed in the saline marine environment but recognized as rice root endophytes for the rst time in this study (Cho and Hwang 2011;Fernandes et al. 2020).
In contrast, among eight unique bacterial genera found at SNSZ, Flavobacterium strain, Kocuria and Novosphingobium well studied for plant growth promoting (PGP) capabilities were reported to promote rice growth by encoding several plant microbe interacting genes, regulating phytohormone levels, nutrient acquisition, redox potential, ion homeostasis, Shewanella and Undibacterium identi ed as rice endophytes responsible for biogeochemical Fe and S cycles in the tested rice paddy soils ( other common genera found abundantly in both the zones, endophytes Bacteroides, Lachnospiraceae NK4A136 group dominating in lower pH SNSZ is according to nding by the author (Huang et al. 2018) who identi ed this genus as the key contributor to decrease in pH for reductive soil disinfestation.
Among the 17 most differently abundant OTUs, SNSZ associated bacterial communities mainly belonged to genera Pantoea and Pseudomonas. Endophytic diazotrophic Pantoea promote growth of rice plants even in salinity stress by making biologically xed nitrogen and present in almost all the samples of rice collected from all around the world (Prakamhang et al., 2009;Kunda et al., 2018). They vigorously colonize rice and exhibit useful properties like phytohormone production viz. abscisic acid, gibberellic acid, cytokinin and indole-3-acetic acid and phosphate solubilization in rice plants (Prakamhang et al., 2009). Pseudomonas sp. found in the rhizospheric region of rice plants have Ca-P solubilizing and IAA production ability, act as biocontrol agents which produce HCN, pyoleutorin, pyrrolnitrin, 2, 4-diacetylphloroglucinol and phenazines chie y phenazine-1-carboxylic acid and phenazine-1-carboxamide (Prakamhang et al. 2009). Ammonia oxidizing Odoribacter and Bacteroides endophytes were also enriched in SNSZ. The number of genera represented by signi cantly abundant OTUs was higher at the non-saline zone compared to that of the saline stressed zone probably due to higher nutrient concentrations and the absence of abiotic stress factors including salinity may responsible for hampering rice growth. This data corroborates the result of differences in αdiversity indices where the species diversity was higher at the non-saline than the saline prone zone.
Some of the aforementioned endophytes have been reported as prominent candidates for biofertilizers in salinity affected rice paddy elds which include Rhizobium sp., Pantoea sp and Pseudomonas sp (Gupta et al. 2012). These endophytic diazotrophs are reported to have better N 2 xing capability than their counterparts i.e., rhizospheric diazotrophs because they escape competition with soil microbes for nutrients and achieve close contact with the plant tissues ( Prakamhang et al., 2009). Also, the diversity and the spatio-temporal distribution of diazotrophs is essential to improve our understanding of biogeochemical cycles in the mangrove ecosystem. Thus, rice endophytes could provide an immense agricultural bene t which is of de nite ecological and economic signi cance.

In uence of environmental parameters and nutrient concentrations on microbial community composition
Bray-Curtis dissimilarity analysis revealed no signi cant differences for endophytic microbiome pro les between rice plants grown in SSZ and SNSZ in a particular geographical location which is also supported by ANOSIM indicator. This nding may support the idea that the endophytic community composition is dependent not only on the surrounding environment but also on the host genotype (Hlholm et al. 2002). The change in microbial community composition was not signi cant could be due to the same host genotype selected from the same region. In this study, a single variety of cultivar (Oryza sativa; cultivar WGL20471, locally known as LalMiniket) has been selected for both the zones thereby conferring the same genotype of rice plant harbouring the endogenous microbial communities. However, variation in endophytic microbial diversity was observed between the saline stressed and non-saline zones because of differences among their physicochemical parameters.
It is generally accepted that the growth of rice plants are strongly in uenced by their endophytic bacteria and soil health status. Thus, an elaborate study on rice endophytic communities associated with soil characteristics on saline stressed and non-saline paddy elds is helpful to predict any relationship between them. From NMDS env t, it was found that ECe, EK, RDO, AP, ON and TDS coincided most strongly with microbial community composition in the studied zones. Thereby proving the role of environmental and nutrient parameters in shaping microbial communities at the studied zones.
As we are the very rst one to report about these observations, the present study could be a pillar to new approaches in the eld of interdisciplinary methods to know about endophytic communities in rice roots along with coastal saline soil and their trait based relationships.

Conclusion
This study provides the rst insight into the diversity of rice root endophytic microbes in the saline stressed zone of Sundarbans, West Bengal along with the non-saline zone from the same region. From our ndings, the saline zone could be represented as high ionic alkaline saline soils while non-saline zone is slightly acidic nutrient enriched soils. Although the variation in endophytic microbial community composition between saline and non-saline zones was not signi cant possibly due to the same variety of cultivar chosen for this study conferring similar plant based genotypic makeup from the same regions.
However, this study reports, the unique microbiome at non-saline zone characterized by microbial genera that reported either having PGP abilities or acting as potential biocontrol agents whereas saline stressed SSZ dominated with either salt-tolerant microbes or less characterized endophytes. An attempt was also made to correlate the association of endophytic microbial diversity and rhizospheric soil environmental and nutrient parameters in the studied zones. The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Ethics approval and consent to participate: Not applicable. Consent for publication: Not applicable.

Figure 1
Principal component analysis (PCA) of measured environmental parameters at Site1 (S1), Site2 (S2), Site3 (S3) and Site4 (S4). Based on ECe; electrical conductivity of the saturated-paste extracts, samples were grouped into two distinct zones, shown as shaded hulls: Sundarbans non-saline zone (violet) and Sundarbans saline zone (cyan). TDS, total dissolved solids; SO4, sulphate; OC, organic carbon; SOM, soil organic matter; AP, available phosphorus; EK, exchangeable potassium; RDO, relative dissolved oxygen; ON, organic nitrogen Bacterial community composition on class and genus level. Order of plots from top to bottom: hierarchical cluster dendrogram based on Bray-Curtis dissimilarity, bar plot of relative sequence proportions of dominant bacterial class, heatmap of dominant genera map (white: no sequences) and total sequence proportion of genera displayed in heatmap Correlation Heatmap indicating spearman rank correlation between the relative abundance of bacterial phyla and soil physicochemical parameters. Horizontal ordinate represents soil physicochemical properties, vertical ordinate represent bacterial community abundance information. Negative correlation and positive correlation were represented by blue color and red color respectively, * indicates p<0.05, r<=-0.5 and r>=0.5

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