Genome-wide DNA polymorphisms in two peatland adapted Coffea liberica varieties

Abstract

Objectives

Coffea liberica is one of the species within the Coffea genus known for its distinctive flavor and resistance to leaf rust disease. Through breeding approaches, two superior varieties of C. liberica, designated as Liberoid Meranti 1 (Lim 1) and Liberoid Meranti 2 (Lim 2), were introduced in 2015. These varieties are known for their high adaptability in peatlands. The genetic basis of plant adaptability to peatlands remains largely unknown. It is therefore essential to identify genome-wide DNA polymorphisms in Lim 1 and 2 in order to gain insights into its capacity for adaptation in peatlands.

Data description

Whole genome sequencing was performed on three plants from each variety (Lim 1 and 2), resulting in 430 million sequencing reads. The mean depth of sequencing for each sample was 36.90x. The reads were mapped to the Coffea canephora genome, with an average mapping rate of 96.34%. The sequencing data revealed the presence of 3,766,805 single-nucleotide polymorphisms (SNPs) and 1,123,683 insertion-deletions (indels) in all six plants. Among the SNPs, there was a notable prevalence of transitions, with a ratio of approximately twofold compared to transversions. The generated data offers invaluable genomic resources for marker development, with significant implications for understanding peatlands adaptability.

Objective

The majority of coffee types cultivated globally is arabica (Coffea arabica), followed by robusta (Coffea canephora) [1, 2]. In addition to these two species, the genus Coffea comprises several other species that produce coffee beans. Coffea liberica is a species from Liberia, which was then introduced to the rest of the world during the nineteenth century, offering the advantage of resistance to the leaf rust disease caused by the fungus Hemileia vastatrix [3, 4]. Following further development, several genotypes of C. liberica were identified as being well adapted to peatlands. In 2015, the Indonesian Ministry of Agriculture released two superior varieties of C. liberica that are well-suited to cultivation in wetland, which are Liberoid Meranti 1 (Lim 1) and Liberoid Meranti 2 (Lim 2) [5]. The estimated coffee production potential for these two varieties is 1.69 and 1.98 tons per hectare for Lim 1 and Lim 2, respectively [5].

Tropical peatlands represent a distinctive ecosystem, occurring across a vast expanse from Southeast Asia, Central and South America to Africa [6, 7]. Plants in peatlands must adapt to a number of challenges, including lower nutrient levels and high water saturation [8]. Many plants can flourish naturally on peatlands, some of which have high economic value [9, 10]. The genetic underpinnings of this high adaptability in tropical peatlands have yet to be explored, which is essential for the development of other peat-adapted plant varieties. A genomic approach has been employed in a number of crops to identify DNA mutations associated with environmental adaptations, including variations in flower color in Mimulus lewisii [11], flowering time in Arabidopsis thaliana [12], and drought resistance in Cedrus atlantica [13]. In this study, we employed genomic approaches to identify DNA mutations in Lim 1 and Lim 2 with the aim of gaining insight into its adaptability in peatlands.

Data description

Six young C. liberica plants were planted in the Botanical Garden of the Biology Department at Universitas Riau, Indonesia. The six plants were divided into two varieties: three Lim 1 and three Lim 2. These plants were obtained from a certified coffee breeding program on Meranti Island, overseen by the Indonesian Ministry of Agriculture. The plants were propagated from seeds collected from select parent trees within their tree improvement program. Specimens from all six plants were deposited in a herbarium (Herbarium Riauensis) and assigned the following voucher identifiers: CL-NNW-L1-202501 for Lim 1 plant 1, CL-NNW-L1-202502 for Lim 2 plant 2, CL-NNW-L1-202 503 for Lim 1 plant 3, CL-NNW-L2-202504 for Lim 2 plant 1, CL-NNW-L2-202505 for Lim 2 plant 2, and CL-NNW-L2-202506 for Lim 2 plant 3.

A sample of young leaves, comprising 50–100 mg, was taken from each of the six plants. The DNA was extracted using the Genomic DNA Mini Kit (Plant) Geneaid (catalog no. GP100), in accordance with the instructions provided in the kit. The quality of the extracted DNA was evaluated using a 1% agarose gel and a Nanodrop 2000 (Thermo Scientific, MA, USA). Subsequently, the extracted DNA was sent to an external service provider for library preparation and Illumina sequencing. The sequencing process yielded 430,270,734 151-bp paired-end reads (Table 1), with an average of 71.71 million paired-end reads per sample. In the absence of a sequenced genome for C. liberica, we assume that its genome size is similar to that of other diploid Coffea sp. genomes, which is approximately 600 Mb. Our assumption is consistent with the estimated average total nuclear DNA content of C. liberica (1.59 picograms (pg)), which is similar to C. canephora at 1.46 pg, compared to the tetraploid C. arabica at 2.47 pg [14]. This allows us to conclude that the average depth of our sequencing per sample is 36.09, which is sufficient to discover DNA mutations [15]. Subsequently, all raw reads underwent quality control analysis using FastQC v0.11.8 [16], wherein 96.16% of the reads exhibited an average per-base quality score of at least Q30.

Table 1 Overview of data files/data sets

The raw sequencing reads from each sample were mapped to the C. canephora genome [1] using BWA-MEM [21] with the default parameters. The mapping of sequencing reads to the C. canephora genome was successful for an average of 96.34% of the reads per sample, which is indicative of the high degree of similarity between the genomes of C. canephora and C. liberica. Prior to utilizing the mapped reads for the identification of DNA mutations, we employed the MarkDuplicates tool to remove duplicated reads that were likely the result of PCR [22]. Subsequently, we followed the recommended workflow to identify germline short variants using GATK [19]. Furthermore, low-quality SNPs were removed based on several criteria, specifically QD < 5, QUAL < 30, SOR > 3, FS > 10, and MQ < 50. For indels, those with QD < 2, QUAL < 30, FS > 200, and ReadPosRankSum < -20 were removed. Following the removal of SNPs and indels that were not present in all samples, a total of 3,766,805 SNPs and 1,123,683 indels were identified (Table 1). 1.02% (38,354) of SNPs and 10.10% (113,439) of indels are multiallelic, indicating the presence of at least three alleles at each position. Among the SNPs, there were almost twice as many transitions (2,503,452) as transversions (1,302,755), which is consistent with the expectation that transitions require fewer changes to the DNA structure than transversions [23]. The number of transitions and transversions were cumulative across all six samples, and therefore may exceed the total number of SNPs. A total of 22.6% of the SNPs (853,209) were found to be overlapping with genes, with 5.5% (207,478) of the SNPs located within exon regions. A total of 223 and 428 SNPs were identified as overlapping with the start and stop codons, respectively. The aforementioned overlapping statistics were calculated based on genome annotation with the NCBI accession number GCA_900059795.1, which was subsequently transferred to the GCA_036785865.1 genome using Liftoff with the default parameters [24].

Limitations

The present dataset is limited to three samples for each variety (Lim 1 and Lim 2), resulting in a total of six samples (plants). This may be a relatively small number for the purpose of representing the genetic variations inherent to a given variety. Our dataset was sequenced using the Illumina paired-end short reads technology (150 bp), and as a consequence, we anticipate that larger DNA mutations, such as those involving large DNA rearrangements, may be undetected.