Microbial community of cultivated and uncultivated citrus rhizosphere microbiota in Brazil

Abstract

The rhizosphere microbiome is known to contain beneficial microorganisms that promote plant growth and increase tolerance to abiotic and biotic stresses. Understanding citrus microbiome diversity and the percentage of diversity that can be recovered in the laboratory is essential for developing innovative approaches to improve plant health and promote sustainable agricultural practices. However, information about the citrus root microbiome, especially in the context of exploring commercial citrus growing areas to identify beneficial plant growth-promoting rhizobacteria (PGPR), is scarce. Here, we present the microbiome data of healthy citrus trees sampled from geographical regions of São Paulo and Amazonas States, in Brazil. The resulting rhizosphere microbiome data comprise an average of 126,180 and 138,707 high-quality reads for the 16S rRNA V3–V4 and ITS1-5F regions, respectively. The taxonomic analysis of cultivated diversity revealed a total of 91 bacterial genera recovered in the laboratory. These data provide valuable information for understanding how the microbiome supports citrus plants in different environments and for developing new strategies to improve crop productivity by using PGPR.

Background & Summary

Citrus fruit production in Brazil, primarily sweet orange (Citrus sinensis), represents 57% of the total fruit production, amounting to 31.5 million tons. São Paulo State (SPS) accounts for 77% of the 17.1 million tons of sweet orange production¹. The citrus production sector significantly contributes to the economy, including the fresh fruit market and orange juice production, totaling USD 2.4 billion annually in SPS. Additionally, approximately 687,000 jobs are involved in the agricultural and industrial sectors related to citrus, underscoring the economic and social importance of citrus in Brazil¹.

To achieve high yields, citrus production in SPS is heavily dependent on advanced cultural practices, including optimised plant nutrition, pest management, and disease control. However, these practices require significant inputs of agricultural supplies and an increasing demand for irrigation water. In addition, there is a growing global discourse on the impacts of climate change—such as increased temperatures and drought—on crop yield stability^2–5, which is no different for citrus. This highlights the urgent need for innovative strategies to mitigate plant stress and enhance crop resilience.

Plant microbiomes from various plant habitats, including the rhizosphere, phyllosphere, and endosphere, play crucial roles in plant growth and health⁶. Among these, the rhizosphere microbiome is particularly critical for sustainable agriculture due to its beneficial microorganisms, which enhance plant growth and increase tolerance to both abiotic and biotic stresses^2,7. When introduced into crop production systems, plant growth-promoting bacteria (PGPB) and microorganisms with biocontrol potential can contribute to food health, as well as economic and environmental sustainability, while optimizing the use of soil plant nutrients^8–11.

In order to utilize rhizosphere microorganisms as bioinputs in agriculture, this study aims to elucidate the composition of citrus-rhizosphere microbial communities in different edaphoclimatic regions of Brazil (Fig. 1a) and compare the cultivable diversity recovered in the laboratory (Fig. 1b) with the diversity present in the citrus rhizosphere (Fig. 1c). To this end, a microbiota library from citrus roots, comprising a total of 2,880 isolates, was established from four distinct geographic regions in the São Paulo and Amazonas States (Fig. 2a). The regions in SPS include both varied responses to hydric stress (Fig. 2c) and organic and conventional orchard management systems (Fig. 2d). Rhizospheric soil samples were also collected under these conditions for comprehensive microbiome analysis (Fig. 2b).

Fig. 1.

Overview of the workflows used to obtain and process the data. (a) Healthy citrus plants were collected from commercial orchards and individually processed to assess the rhizosphere microbiomes from five distinct environments. In the North region*, plants with varying responses to water stress were sampled; in the East region** plants from both conventional and organic management systems were collected. Microbial community DNA was extracted, and bacterial isolation was performed in the laboratory. All rhizobacteria isolated from citrus tree roots underwent PCR amplification to add individual-specific barcodes, enabling traceability of sequences back to their original plates. The amplicons were subsequently pooled into a single 96-well plate to form a mixed culture, which was sequenced on an Illumina MiSeq platform using 16S V4–V5 primers. The total DNA extracted from all samples was sequenced on an Illumina NovaSeq platform using 16S V4–V5 and ITS1–5 F primers. (b) Sequence demultiplexing was carried out using specific barcode plates, and the raw data were processed with DADA2. (c) The raw data from the rhizosphere microbial community were analysed using DADA2. Subsequent visualisation and community composition analysis were performed using the microbiome, phyloseq, and vegan packages.

Fig. 2.

The commercial citrus orchards collected. (a) The study was conducted in two states of Brazil, as depicted on the map. São Paulo is highlighted in yellow, while Amazonas is shown in green. The specific sites from which plants were collected are indicated on the left. (b) The sampling methodology for soil and roots involved collecting samples at a depth of 5–15 cm and within a 25 cm radius from the main trunk. This methodology was uniformly applied across all locations and seasons. (c) In the North region, orchards included plants with varying responses to water stress; exhibiting drought tolerance phenotype (left) and drought susceptible phenotype (right). (d) In the East region, orchards featured both conventional (left) and organic (right) management systems.

To assess the diversity of the rhizosphere microbiome, high-throughput sequencing of DNA fragments targeting the 16S V4-V5 and ITS1-5F regions was performed for 44 samples. For each primer set, 16S rRNA and ITS, 8,751,608 and 8,560,467 raw reads were obtained, respectively. After quality control, processing of these data yielded median read numbers of 126,180 and 138,707 per sample (Supplementary Table S1), resulting in 55,145 and 17,623 amplicon sequence variants (ASVs) from the 16S V4-V5 and ITS1-5F regions, respectively. The most abundant bacterial and fungal phyla were consistent across all areas and seasons, though differences in relative abundance were observed (Fig. 4). For the 16S V4–V5 amplicons, the predominant phyla were Proteobacteria, Actinobacteria, Acidobacteria, Firmicutes, and Chloroflexi (Fig. 4a). For the ITS1-5F amplicons, the most abundant phylum was Ascomycota (Fig. 4b). Additionally, it was found that prokaryotic diversity was generally higher than fungal diversity across all areas and seasons examined (Fig. 5a).

Fig. 4.

Rhizosphere community composition of regions collected in both seasons (dry and rainy) at the phylum level. Relative abundance of the most prevalent (a) bacterial and (b) fungal phyla retrieved from 16S V4–V5 and ITS1-5F amplicon sequencing, respectively. Each column represents a single sample, and samples were grouped according to the region from which the communities were collected. AM = Amazon region; CS = Center-South region; EA = East region; NO = North region; SO = South region.

Fig. 5.

Comparison of rhizosphere microbiome diversity vs. laboratory-recovered cultivable diversity. (a) Alpha diversity measured using the Shannon index of rhizosphere community composition across regions collected during both the dry and rainy seasons. AM = Amazon region; CS = Center-South region; EA = East region; NO = North region; SO = South region. (b) Venn diagram illustrating the proportion of genera from the core rhizosphere community that were successfully cultivated in the laboratory.

To assess the cultivated diversity, the approach of community-based culture collections (CBC) was employed¹². A total of 2,880 colonies were picked (Fig. 3a). After labelling with specific barcodes using a PCR technique (Fig. 3b,Supplementary Table S2), all DNA amplicons were pooled into a 96-well plate to create a mixed culture (Fig. 3c). Sequencing of the fragments targeting the 16S V4–V5 region was conducted using the Illumina MiSeq250 bp PE platform. In total, 10,924,291 raw reads were obtained, with median read numbers of 113,795 per sample. The taxonomic diversity analysis revealed a total of 91 bacterial genera recovered in the laboratory, representing 17.5% of the rhizosphere core community (Fig. 5b; Supplementary Table S3). Laboratory-isolated representatives of the most abundant bacterial phyla found in the citrus rhizosphere were successfully obtained, except the Acidobacteria phylum (Supplementary Table S3).

Fig. 3.

Workflow of isolation and identification of culturable rhizobacteria. (a) Colonies with distinct morphologies from citrus roots were selected and transferred to a 96-well plate. (b) Each 96-well plate (n = 30) underwent PCR amplification to incorporate 9-bp individual-specific barcodes. (c) Following labelling, the amplicons were pooled into a single 96-well plate, creating a combined mixed culture. This plate was then sent for sequencing targeting the 16S V4–V5 amplicons.

In sum, this study provides valuable information about the citrus root microbiome in Brazil, particularly identifying beneficial plant growth-promoting (PGP) microorganisms and establishing a large collection of microbes for future exploration of their application in microbe engineering and bioproducts.

Methods

Study site and sampling

The citrus orchards were selected from commercial farms that supply juice processing plants. Fruit yields typically range from 36 to 55 tons per hectare under non-irrigated conditions, and harvesting occurs when the maturity ratio exceeds 10, ensuring the production of high-quality juice. Rhizospheric soil samples were collected from these citrus orchards located in the states of São Paulo (SPS) and Amazonas, Brazil (Fig. 2a), by sampling roots and soil at a depth of 5 to 15 cm and 25 cm from the trunk (Fig. 2b). In SPS samples were collected in the regions North, East, Center-South, and South adding samples from different responses to water stress (Fig. 2c) and management systems (Fig. 2d). Sampling was conducted at two distinct times: i) during the dry season (August 2022), characterised by severe water restriction in the sampled regions; ii) during the rainy season (March 2023), when water availability had been fully restored across all sampled regions. The only exception was the samples from Amazonas State, where sampling during the rainy season was not conducted (Table 1). These regions are characterised by distinct soil and climate conditions (Table 2). A total of 60 plants were sampled in SPS and 4 in the Amazonas.

Table 1.

Summary of sampling site description.

Sampling regions	Geographic coordinates	Sampling date	Host (scion / rootstock)	Plants years old
North1	20° 34′ 28.24″ S	Aug-22	Valencia/Swingle citrumelo	12
	49° 15′ 31.75″ W	Mar-23
South	23° 36′ 52.96″ S	Aug-22	Valencia/Swingle citrumelo	12
	48° 18′ 17.58″ W	Mar-23
East	22° 18′ 36.37″ S	Aug-22	Valencia/Swingle citrumelo	12
	47° 0′ 47.70″ W	Mar-23
East2	22° 18′ 18.524″ S	Aug-22	Siciano lemon/Swingle citrumelo	5
	47° 31′ 35.525″ W	Mar-23
Center-South	22° 38′ 52.40″ S	Aug-22	Valencia/Swingle citrumelo	13
	48° 18′ 45.14″ W	Mar-23
Amazonia State	2°42′ 50.44″ S	Aug-22	Valencia/Rungpur lime	13
	59°26′ 56.65″ W

¹Include plants with different drought susceptible

²Include orchards with conventional and organic management systems.

Table 2.

Soil chemical parameters of each citrus orchards at different periods of sampling. Values are mean.

Parameters	São Paulo State								Amazonia State
	Center-South		East		North		South
	S1	S2	S1	S2	S1	S2	S1	S2	S1
OM (g/dm3)	14	10	18	21	15.5	10	26.5	23.5	35
pH	5.1	4.95	5.2	5.2	5.8	5.1	5.1	5.15	5.5
P (mg/dm3)	48	45	138	139	46	25.5	27.5	68	73
K (mmolc/dm3)	1.8	1.45	1.7	3.5	2.75	1.6	2.45	3.65	4.9
Ca (mmolc/dm3)	21	13.5	27	44	26.5	17	33	38	80
Mg (mmolc/dm3)	9	6.5	15	18	11.5	6	14.5	13.5	19
B (mg/dm3)	0.52	0.44	1.05	1.2	0.76	1.24	0.96	1.36	0.47
Cu (mg/dm3)	11.8	9.3	18.8	24.1	7.3	4.5	3.3	8.3	30.4
Fe (mg/dm3)	31.5	26	52	46	14.5	31.5	22	39	109
Mn (mg/dm3)	10.95	6.9	4.9	4.1	5.95	8	4.6	5.05	13.9
Zn (mg/dm3)	4.55	3.5	8.3	12.2	9.85	2.8	3.25	7.75	60.7
Pt (mm)	35.0	822.8	44.4	1099.4	33.4	755	82	595.6	—
Pm (mm/month)	11.7	274.7	14.8	366.7	11.3	251.7	27.3	198.5	—
Tma (°C)	27.6	30.8	28.1	31.2	29.9	31.7	24.7	29.2	—
Tmi (°C)	17.8	20.1	8.9	19.0	8.9	19.7	9.3	17.8	—

OM – organic matter; P – phosphorus; K – potassium; Ca – calcium; Mg – magnesium; B – boron; Cu – cuprum; Fe – iron; Mn – manganese and Zn – zinc; Pt - Accumulated rainfall in the 3 months prior to collection; Pm - Average precipitation rainfall in the 3 months prior to collection; Tma: Average maximum temperature of the last 3 months prior to collection; Tmi: Average minimum temperature of the last 3 months prior to collection.

S1 – Sampling in August 2022 - Dry season.

S2 – Sampling in March 2023 - Raining season.

Sampling parameters were determined based on the availability of citrus plants on the farms. In commercial citrus production areas of São Paulo, 12-year-old ‘Valencia’ sweet orange (Citrus sinensis) plants grafted onto Swingle citrumelo rootstocks were selected from orchards across the four geographic regions. In addition to the aforementioned collection, in the North region, plants with varying responses to water stress were collected (Fig. 2c), and for the East region, 5-year-old ‘Siciliano’ lemon (C. lemon) plants grafted on Swingle citrumelo rootstock samples were taken from both conventional and organic management systems (Fig. 2d). In Amazonas State, sampling was conducted on 12-year-old ‘Valencia’ sweet orange plants grafted on ‘Cravo’ rangpur lime rootstocks.

Regardless of citrus varieties and plant age, the same sampling methodology was consistently applied across all locations and seasons. Firstly, i) all selected plants exhibited no visual symptoms of common diseases in SPS orchards, such as Citrus Variegated Chlorosis (CVC), huanglongbing (HLB/greening), Citrus Canker, or Gummosis, and no pest infestations were observed; ii) five plants were sampled from each condition; iii) the same plants were sampled during different seasons; and iv) preference was given to roots with a maximum thickness of 3 mm. After sampling, roots were shaken to remove excess soil and placed in sterilised plastic bags, followed by spraying with phosphate-buffered saline (PBS) to maintain humidity. The samples were kept in containers with ice during transport to the laboratory. A fraction of bulk soil was sampled from each experimental unit (plant) for chemical analysis. In the laboratory, the samples were preserved at 4 °C for subsequent microbial isolation, and the soil extracted from the rhizoplane was preserved at –80 °C for DNA extraction.

Chemical characterization of soil and edaphoclimatic conditions of each region

For all samples, the chemical properties of soil were determined at the Agronomy Institute (IAC), in Campinas (Instituto Agronômico de Campinas (iac.sp.gov.br) following standard methods¹³. Briefly, the pH was measured using CaCl₂ (0.01 M) and exchangeable acidity was determined by dilution of the samples in SMP buffer solution. Organic matter (OM) content was determined through dichromate oxidation followed by colorimetry. The element phosphorus (P) was extracted using ion exchange resins and quantified by colorimetry. Magnesium (Mg), and calcium (Ca) were extracted using ion exchange resins and quantified by atomic absorption spectrophotometry. Potassium (K) was extracted using ion exchange resins and quantified by flame photometry. Copper (Cu), iron (Fe), manganese (Mn), and zinc (Zn) were extracted using the diethylene triamine penta acetic acid (DTPA) method and quantified with inductively coupled plasma optical emission spectrophotometry. Boron (B) was extracted using hot water method and followed by colorimetry.

In each of four geographic regions in Sao Paulo State the precipitation rainfall (mm) and atmospheric temperature (°C) dataset were taken in the period of three months before each sampling period by the meteorology stations placed in the respective farms.

DNA extraction and amplicon sequencing

Total soil DNA was extracted from all samples using a FastDNA^TM spin kit for Soil (MPbio®, Brazil), according to the manufacturer’s protocol. Extracted DNA quality was assessed by a NanoDrop spectrophotometer (Thermo Fisher Scientific Inc., MA, USA) and was assessed on a 1% agarose gel electrophoresis, stained with Gel Red (Biotium, USA), and visualised in a transilluminator (Applied Biosystems, MA, USA). Extracted DNA was stored at −80 °C for subsequent analysis.

The metagenomic sequencing was done at Novogene Corporation Inc. (http://www.novogene.com/, Sacramento, USA) using the Illumina NovaSeq6000 250 bp PE platform. For the bacterial sequencing, the 16S rRNA V4–V5 region was amplified from total DNA using 515 F (5′-GTGCCAGCMGCCGCGGTAA3-3′) and 907 R (5′-CCGTCAATTCCTTTGAGTTT-3′) primers. For the fungal sequencing, the ITS1–5 F region was amplified using ITS5-1737F (5′-GGAAGTAAAAGTCGTAACAAGG3-3′) and ITS2–2043 R (5′-GCTGCGTTCTTCATCGATGC3-3′) primers were used.

Metataxonomy sequencing data processing and taxonomic annotation

The quality of raw reads was assessed using FastQC (version 0.11.7), with results compiled into an HTML report via MultiQC (version 1.6). Quality control filtering, paired read assembly, Amplicon Sequence Variants (ASVs) inference, chimera removal and taxonomic assignment for both 16S and ITS libraries were conducted using the DADA2 denoising algorithm¹⁴. Initially, the forward and reverse reads of the 16S rRNA gene sequences were truncated to 245 bp and 220 bp, respectively. In contrast, reads from the ITS sequences were not truncated to a fixed length due to the substantial length variation across fungal species. Subsequently, reads were filtered to exclude those with more than two expected errors (maxEE = 2) and any ambiguous bases (maxN = 0). The error models, along with dereplicated reads pooled from all samples, were used as input to obtain denoised sequences. Finally, pairs of forward and reverse reads were merged to obtain ASVs, and chimeric sequences identified via the consensus method were removed. Based on the naive Bayesian classifier method¹⁵, taxonomic assignment of the 16S and ITS ASVs was performed using SILVA database release 138.1¹⁶ and UNITE database version 10.0¹⁷, respectively.

Isolation and diversity of culturable rhizobacteria

To access the cultivated diversity, the methodology proposed by Armani et al.¹² was followed with modifications. To enable the recovery of a broader range of cultivable diversity in the laboratory, for each sampling area, rhizobacteria were isolated from citrus tree roots (see details in the section “Study site characteristics and sampling”) using four culture media: i) TWYE (Tap Water Yeast Extract); ii) R2A (Reasoner’s 2A Agar); iii) PSM (Pseudomonas syringae Medium); and iv) TSA 10% (Tryptic Soy Agar). The TWYE and TSA 10% media are rich and non-selective, supporting the growth of a wide range of bacterial species^18–22. The R2A medium is selective for slow-growing bacteria^21,23,24, while the PSM medium specifically enhances the growth of Pseudomonas species, a genus recognized for its potential as a plant growth promoter. The isolates were cryopreserved at -80 °C, following the concept of community-based culture collections (CBC)^12,25. A total of 2880 colony communities were picked and stored in 96-well plates (Fig. 3a). To identify and annotate the rhizobacteria, PCR amplifications were designed to add individual specific barcodes for the identification of each plate using primers 515 F (5′- GTGCCAGCMGCCGCGGTAA-3′) and 907 R (5′- CCGTCAATTCCTTTGAGTTT-3′) with modifications to include specific barcodes for each plate in reverse primer (Supplementary Table S2) (Fig. 3b). Subsequently, the amplicons were pooled into a single 96-well plate, forming a pooled mixed culture (Fig. 3c). The amplicon products were then sent for sequencing at LaCTAD (Universidade Estadual de Campinas – Unicamp, Campinas, SP, Brazil) using the Illumina MiSeq250 bp PE platform. This methodology allows tracing individual sequences back to their original plates. The software Fgbio (https://github.com/fulcrumgenomics/fgbio) was used to perform sequence demultiplexing using specific barcode plates. The raw data was then processed in the same way as described in the section “Metataxonomy sequencing data processing and taxonomic annotation”, above.

Visualisation, filtering and community analysis

The visualisation and filtering were conducted using R (version 4.2.2) with the Phyloseq (version 1.42.0) (https://github.com/joey711/phyloseq) and Microbiome (version 1.23.1) (https://github.com/microbiome/microbiome) packages. To eliminate spurious and undesirable ASVs, all ASVs underwent a filtering process: 16S ASV sequences assigned to mitochondrial or chloroplast taxa and those with a prevalence lower than 5% of the total number of samples were discarded. Subsequently, the ASV abundance table was transformed into relative abundance percentages (%) for further analyses. Finally, the Shannon index was calculated to infer sample diversity.

Data Records

The raw reads of both the 16S rRNA and ITS amplicon sequencing data (.fastq format) were deposited in NCBI Sequence Read Archive (PRJNA1130990)²⁶. Sample description and NCBI accessions of each of the sequencing libraries generated in this study are available in Supplementary Table S4.

Technical Validation

To ensure unbiased data production, the orchards from four regions in São Paulo State were standardised in terms of citrus variety (‘Valência’ sweet orange as scion grafted onto Swingle citrumelo as rootstocks) and tree age (12 to 13 years old). Among these orchards the environmental conditions (soil and regional climate) were the variability. For areas with different management strategies (organic and conventional), the orchards are composed of ‘Siciliano’ lemon as scion grafted onto Swingle citrumelo as rootstock. Both orchards are placed next to each other, however, under the same type of soil and climate. In the Amazonas State, the orchards consisted of ‘Valencia’ scion grafted onto Rangpur lime rootstock. Randomization principles were applied during sample collection in each area but selecting the plants with no disease or pests symptoms. For the picking colonies were prioritised those with different morphology and as individually possible. The maximum of 10 colonies with the same morphology was picked from each sample. The quality and purity of the extracted DNA was assessed by NanoDrop spectrophotometer (Thermo Fisher Scientific Inc., MA, USA), inspection of the 260/280 and 260/230 wavelength (nm) ratios and analysis by electrophoresis agarose gel 1%. Amplicon sequencing raw data went through multiple steps of rigorous quality control, which included removing low-quality reads, host-associated sequences and chimeras.

Author contributions

H.D.C.F. coordinated the project. H.S.L., H.D.C.F., C.S.F., R.S.C.S., and J.S.L.A. contributed to the study conception and design. Samples collections were performed by N.M., G.B.P., C.S.F. and H.D.C.F. Extraction of total DNA from the microbiome and supervision of sequencing were performed by H.S.L. The Bioinformatic analysis and data presentation were performed by H.S.L. Isolation of rhizobacteria was performed by N.M and G.B.P. The first draft of the manuscript was written by H.S.L. and H.D.C.F. The final draft of the manuscript was reviewed by H.S.L., H.D.C.F., N.W., and J.S.L.A. All authors read and approved the final version of the manuscript.

Code availability

All software and codes used in this study were published in peer-reviewed journals. Additional information was described in detail in the Material and Methods section.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41597-024-04141-y.