https://orcid.org/0000-0003-4345-5120
Figures
Abstract
Soybean is a globally important industrial, food, and cash crop. Despite its importance in present and future economies, its production is severely hampered by bruchids (Callosobruchus chinensis), a destructive storage insect pest, causing considerable yield losses. Therefore, the identification of genomic regions and candidate genes associated with bruchid resistance in soybean is crucial as it helps breeders to develop new soybean varieties with improved resistance and quality. In this study, 6 multi-locus methods of the mrMLM model for genome-wide association study were used to dissect the genetic architecture of bruchid resistance on 4traits: percentage adult bruchid emergence (PBE), percentage weight loss (PWL), median development period (MDP), and Dobie susceptibility index (DSI) on 100 diverse soybean genotypes, genotyped with 14,469 single-nucleotide polymorphism (SNP) markers. Using the best linear unbiased predictors (BLUPs), 13 quantitative trait nucleotides (QTNs) were identified by the mrMLM model, of which rs16_14976250 was associated with more than 1 bruchid resistance traits. As a result, the identified QTNs linked with resistance traits can be employed in marker-assisted breeding for the accurate and rapid screening of soybean genotypes for resistance to bruchids. Moreover, a gene search on the Phytozome soybean reference genome identified 27 potential candidate genes located within a window of 478.45 kb upstream and downstream of the most reliable QTNs. These candidate genes exhibit molecular and biological functionalities associated with various soybean resistance mechanisms and, therefore, could be incorporated into the farmers’ preferred soybean varieties that are susceptible to bruchids.
Citation: Mukuze C, Msiska UM, Badji A, Obua T, Kweyu SV, Nghituwamhata SN, et al. (2025) Genome-wide association mapping of bruchid resistance loci in soybean. PLoS ONE 20(1): e0292481. https://doi.org/10.1371/journal.pone.0292481
Editor: Karthikeyan Adhimoolam, Jeju National University, REPUBLIC OF KOREA
Received: September 20, 2023; Accepted: September 22, 2024; Published: January 10, 2025
Copyright: © 2025 Mukuze et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: Intra ACP-Mobility Scheme (CSAA) and Carnegie Cooperation (New York) through RUFORUM-Grant RU/2016/INTRA ACP/RG/013.
Competing interests: The authors have declared that no competing interests exist.
1. Introduction
Soybean (Glycine max (L.) Merr., 2n = 40) is an important food and cash crop worldwide with nitrogen-fixing ability and high oil and protein content [1]. Despite its importance, the crop’s grain storage is affected by bruchids (Callosobruchus chinensis L.) [1–3]. In most cases, the infestation starts in the field and spreads throughout the soybean value chain [3]. Bruchids have high fecundity and the ability to re-infest, in addition to causing irreversible damage to soybeans [4]. Bruchids damage to soybean grain has been estimated to cause 60–100% yield losses in Sub-Saharan Africa (Pawlowski et al., 2021). To minimize yield losses to bruchids, farmers have been employing various management strategies including insecticides, botanically active plants, and cultural practices [5, 6]. However, the majority of farmers tend to use insecticides for bruchid management [5]. Despite their effectiveness, insecticides have several drawbacks such as risks to human and environmental health, high cost to resource-constrained farmers, and the possibility of the development of insecticide resistance [3, 7, 8]. As a result, developing enhanced soybean cultivars with high resistance to C. chinensis is the most sustainable, cost-effective, strategically significant, and ecologically benign option.
For more effective development of soybean varieties that are resistant to C. chinensis, a thorough genetic dissection and understanding of the nature of this resistance is crucial. Recently, the study carried out by Msiska et al. [9] reported that soybean resistance to C. chinensis is due to the overproduction of secondary metabolites such as tannins and the high expression of enzymes such as peroxidases. A further genetic investigation of soybean resistance to C. chinensis revealed that the nature of gene action and mode of inheritance of bruchid resistance traits is quantitative and complex [10]. Similarly, quantitative inheritance for resistance to bruchids has been reported in other legumes such as common bean [5] and cowpea [11]. Breeding for such complex traits through conventional means is challenging due to environmental influences that slow down the selection process. Hence, the use of molecular markers flanking the trait of interest improves the selection process, although it necessitates a thorough genetic dissection and understanding of the association between the genotype and phenotypic variations [12]. To the best of our knowledge, the use of association mapping for identifying candidate genes and utilization in marker-assisted breeding (MAB) programmes for resistance to bruchids has not been conducted in soybean. This knowledge gap makes it difficult to implement marker-assisted selection (MAS) to combat the devastating impact of bruchids in Uganda and some parts of Sub-Saharan Africa that share the same genetic material. However, several studies have been carried out in other legumes aimed to determine quantitative trait loci (QTL) regions conferring bruchid resistance black gram (Vigna mungo (L.) Hepper) [13], mung bean (Vigna radiata (L) [14–17], wild Vigna spp [18–20], cowpea using both bi-parental [21] and diverse [6, 12] populations, and in common bean using bi-parental [22] and diverse [5] populations.
However, the use of bi-parental populations to identify QTLs has many drawbacks, including low resolution for QTL detection because only two parents contribute to the genetic information present in the mapping population, restricting its potential application in molecular plant breeding [6, 12, 23]. To address this issue, genome-wide association studies (GWASs), a more robust contemporary mapping technique for identifying genomic regions and candidate genes linked with important traits in both self-pollinated and cross-pollinated crops, have been widely used [12]. GWAS uses collections of genotypes with diverse genetic backgrounds and high historical recombination, resulting in more powerful QTL detection and better resolution [6, 12, 24]. In soybean, GWAS has been used to identify genomic regions associated with resistance to many insect pests, including beet armyworm, potato leaf hopper, soybean looper, Mexican bean beetle, velvet caterpillar, and soybean aphid [25]. Therefore, this study aims to identify quantitative trait nucleotides and candidate genes associated with resistance to C. chinensis in soybean using single-nucleotide polymorphism (SNP) molecular markers to uncover the genetic basis of bruchid resistance and facilitate MAS in soybean breeding.
2. Materials and methods
2.1. Sources of soybean germplasm
Only 100 genotyped (S1 Table) soybeans from various genetic backgrounds– 50 resistant (R), 30 moderately resistant (MR), and 20 susceptible (S)- out of 498 phenotyped were included in the experiment. These genotypes were acquired from AVRDC-Taiwan, IITA -Nigeria, Uganda, Japan, USA and SeedCo-Zimbabwe.
2.2. Bruchid rearing
The methodology used for this experiment was adopted from [3]. Adult bruchid cultures were established in a laboratory at the Makerere University Agricultural Research Institute Kabanyolo (MUARIK) in 2018. The stock cultures of bruchids were obtained from National Crop Resources Research Institute (NaCCRI), Namulonge, Uganda. The cultures were continuously multiplied on seeds of three susceptible soybean varieties (Maksoy 2N, Maksoy 3N, and Maksoy 4N) kept in glass jars and plastic buckets under room temperature. The jars and buckets were covered with a muslin cloth for ventilation also to prevent bruchids escape. The bruchid populations were maintained by introducing them on a regular basis into new but susceptible soybean seeds.
2.3. Bruchid infestation and phenotypic data collection
The methodology used in this experiment was adopted from Msiska et al. [3]. A sample of 100 soybean seeds were taken and weighed from each of the 498 genotypes to obtain the baseline for the 100 seed weight. To determine the initial seed weight, 50 randomly selected seeds from each soybean genotype were planted in various plastic Petri dishes and weighed. Following that, the soybean seeds in each Petri dish were artificially infested with 20 randomly selected adult bruchids from the bruchid colony using the no-choice test method. The experiment was set up in a randomized complete block design, with insect infestation days serving as blocks, and it was replicated three times. The bruchids were kept in Petri dishes with soybean samples to allow for mating and oviposition before being removed from the soybean samples after 10 days. On day 11, the number of eggs laid on each of the 50 seeds were counted, and the number of emerged adult bruchids was counted and they were removed daily until no new insects emerged for 5 consecutive days. The final weight of the seed samples in each Petri dish was recorded. These observations were used for calculating the following variables: Percentage bruchid emergence:
The median development period (MDP) was calculated as the number of days from the middle of oviposition (day 5) to the first progeny’s emergence. At the end of the experiment, the DSI (Dobie susceptibility index) was determined for each accession according to [26]:
where Y = total number of emerging adults and MDP = median development period (days).
DSI = 0, if no insects emerged over the test period. In this study, the modified susceptibility index ranging from 0 to 9 was used to classify the soybean genotypes, where 0–1 = resistant, 2–3 = moderate resistant, 4–5 = susceptible, and ≥6 highly susceptible.
2.4. Phenotypic data analysis
To test soybean resistance to bruchids, the following parameters were used: number of eggs laid (NEL), percentage weight loss (PWL), Dobie susceptibility index (DSI), percentage adult bruchid emergence (PBE) and median development period (MDP). Using the GenStat Statistical Package 18th Edition, each resistance parameter was subjected to a one-way analysis of variance (ANOVA). The Pearson correlation analysis was performed to determine the nature of relationships between bruchid resistance variables [3]. All resistance parameters were analyzed using the GenStat Statistical Package 18th Edition following the linear model below:
with Y being the phenotype of the target trait and μ its grand mean.
The broad-sense coefficient of genetic determination (H2) (i.e., equivalent to broad-sense heritability) was estimated as described by Piepho and Möhring [27] for all the soybean resistance traits using variance components obtained from a mixed model that considered the effects of all factors present in the model as random. The following formula was used:
where nr is the number of replicates.
2.5. DNA isolation, SNP genotyping, LD decay, and multi-locus GWAS analysis
Seeds from 100 soybean genotypes were grown in a screen house at the Biosciences Eastern and Central Africa—International Livestock Research Institute (BecA-ILRI) Hub, Kenya, in 2019. Young, fresh leaf samples from each accession were collected twelve days after germination, and DNA was extracted using ZR Plant and Seed DNA Mini PrepTM according to the manufacturer’s protocol with minor modifications [28]. Prior to use, the DNA quality was determined on 0.8% (w/v) agarose gel in a 1 X Tris-acetate EDTA buffer and run at 80 V for 45 min, during which gel images were taken using a GelDoc-ItTM Imager (UVP), and the image was interpreted for DNA quality. Meanwhile, the DNA was quantified using a Thermo Scientific Nanodrop 2000C Spectrophotometer, Inqaba Biotech, South Africa and stored at 4°C [28].
The DNA samples were sent to Australia, where the soybean genotypes were genotyped using the Illumina HiSeq 2500, Illumina Inc, California, USA and Diversity Arrays Technology Sequencing (DArTSeqTM), DArT Laboratory, Australia. Subsequently, a genomic DNA library was constructed following an integrated DArT and genotyping-by-sequencing (GBS) methodology that involved complexity reduction of the genomic DNA, and repetitive sequences were eliminated using methylation-sensitive restrictive enzymes before sequencing on next-generation sequencing platforms [29].
The reads were aligned to the soybean reference genome Soybean_v7, which is publicly available at ftp://ftp.jgipsf.org/pub/JGI_data/phytozome/v7.0/Gmax [30], to identify single-nucleotide polymorphism (SNP) markers. Then, the HapMap genotypic data were determined for association analysis.
Prior to association analysis, the SNP data were filtered using a minor allele frequency (MAF) of 0.05 and a call rate of 80% of the sample size using TASSEL v5.2.88 software [31]. After filtering, a total of 14,469 SNPs were retained for subsequent analysis and imputed using the KNN imputation method, with all other parameters set to default. Principal components (PCs) (S2 Table) and the kinship matrix (S3 Table) to infer the population structure and cryptic relatedness were calculated as described by Badji et al. [32]. The kinship matrix was generated using the centered identity-by-state (Centered-IBS) function with other parameters set to default. The linkage disequilibrium (LD) decay, based on the squared Pearson correlation coefficient (r2) between pairs of SNPs was calculated. An LD decay graph was generated by plotting the r2 between pairs of SNPs against their pairwise physical distance between pairs of SNPs as described in [33, 34] based on Remington et al. [35]. The average pairwise distances at which LD decayed at r2 = 0.2 and 0.1 were also generated to visualize how LD decayed across the genome and allow the determination of the adequate LD blocks around markers.
For GWAS analysis, six multi-locus methods of the mrMLM package https://cran.r-project.org/web/packages/mrMLM/index.html, corrected for population structure and inequal kinship [36], were applied to identify significant trait–marker associations using trait-phenotypic BLUPs and 14,469 imputed SNPs as described by Badji et al. [32]. GWAS analysis was performed on four bruchid resistance traits: percentage weight loss (PWL), Dobie susceptibility index (DSI), percentage bruchid emergence (PBE), and median development period (MDP). Furthermore, both the kinship and the first 5 PCs were included in each of the six multi-locus methods (mrMLM, FASTmrMLM, pKWmEB, pLARmEB, FASTmrEMMA, and ISIS EM-BLASSO) in order to control false marker–trait association due to stratification effects and hence increase accuracy and power in quantitative trait nucleotide (QTN) detection [37]. Additionally, the quantile–quantile (QQ) plots generated (S1 Fig) were used to visually assess the presence of spurious associations.
2.6. Annotation of candidate genes
The candidate genes underlying trait-associated SNPs involved in bruchid resistance functions were manually searched on the soybean reference genome hosted in Phytozome v13 https://phytozome next.jgi.doe.gov/geneatlas/jbrowse/index.html?data = genomes/Gmax_Wm82_a4_v1. Only QTNs that were detected by at least two multi-locus GWAS methods were considered for the candidate gene search. For this analysis, genes within a window of 478.45 kb upstream and 478.45 kb downstream of the physical positions of the significant reliable QTNs, based on the genome-wide LD block size characterizing the mapping population, were searched on soybean reference genome and considered as candidate genes. Their functional annotations were closely examined from the Gmax_Wm82_a4_v1 soybean reference genome hosted on Phytozome v13, and their association with bruchid resistance was used as a prioritization criterion.
3. Results
3.1. Phenotypic response of soybean accessions
Analysis of variance for soybean genotypes response to bruchid infestation showed significant differences (p < 0.05) for percentage weight loss (PWL), Dobie susceptibility index (DSI), percentage bruchid emergence (PBE), median development period (MDP), but not for number of eggs laid (NEL) (Table 1). Further, the broad-sense coefficient of genetic determination, which is equivalent to the broad-sense heritability, was relatively low to moderate for bruchid resistance traits: NEL (0.005), PWL (0.31), DSI (0.22), PBE (0.21), and MDP (0.39) (Table 1).
The Pearson correlation analysis performed to determine the nature of relationships between bruchid resistant variables showed that number of eggs (0.78), PBE (0.87), and PWL (0.55) were significantly positively correlated with DSI (p < 0.01), while weaker correlation with MDP (0.41) was observed as described by Msiska et al [3]. In addition, positive and significant (p < 00.1) correlations were observed among NEL, PBE, and PWL (S2 Table), suggesting that these bruchid resistance traits in soybean might be influenced by the same genetic basis.
3.2. Linkage disequilibrium (LD) decay
The genome-wide linkage disequilibrium (LD) computed for the 100 soybean lines using the 14,469 high-quality SNP markers is shown in Fig 1. The LD decay was relatively slow, with a distance of 478.45 kb at r2 of 0.2, decaying to an r2 of 0.1 at 1423.09 kb (Fig 1). This LD decay implies that the population is characterized by genomic LD blocks of 478.45, considering an LD threshold of r2 = 0.2.
3.3. Association mapping for bruchid resistance traits
The results for the evaluation of the model fit for six multi-locus random-SNP-effect mixed linear models for GWAS (mrMLM) with the kinship and the PC matrices showed the best fit model to consider Q for all the bruchid resistance traits (PWL, PBE, MDP, and DSI). Both the first 5 PCs and the kinship were included in the analysis to control false marker–trait association due to population stratification.
3.4. QTNs associated with the soybean resistance traits identified by multi-locus GWAS
A total of 13 significant QTNs (LOD > 3) associated with soybean bruchid resistance traits were detected by 6 multi-locus GWAS methods across 7 chromosomes (Table 2 and Fig 2). Of these QTNs, rs7_14971060 on chromosome 7, rs18_14972895 on chromosome 18, and rs16_22914864 and rs16_14976250 on chromosome 16 had an allelic effect of 3.50 2.88 3.69 and −0.0001 on PWL, respectively. The QTNs rs16_14976250 and rs16_14975721 on chromosome 16 and rs1_22916615 on chromosome 1 had an allelic effect of −29.63, 17.62, and 23.65 on PBE, respectively. Furthermore, the QTNs rs15_14981757 on chromosome 15 and rs18_14978590 on chromosome 18 had an allelic effect of −1.59 and −1.94 on MDP, respectively. The QTNs rs13_14978774 and rs13_14973454 on chromosome 13; rs15_14981838 on chromosome 15; and rs16_22916222, rs16_22915136, and rs16_14976250 on chromosome 16 had an allelic effect of −1.13, −3.00 × 10−4, 0.96, 0.80, 3.00 × 10−4, and −0.71 on DSI, respectively (Table 2). Three QTNs, rs16_14975721 and rs16_14976250 on chromosome 16 and rs1_22916615 on chromosome 1, associated with GI, had an allelic effect of 1.13 × 10−5, −0.566, and 1.5657, respectively (Table 2).
The horizontal dotted line indicates the genome-wide significance threshold (−log10 p ≥ 3.7). Association mapping of (A) PWL, (B) PBE, (C) MDP and (D) DSI.
The examination of QTN co-detection by multi-locus GWAS approaches revealed that three QTNs were simultaneously co-detected by more than two multi-locus methods (Table 2 and Fig 3). Among these QTNs, rs16_22914864 was associated with PWL and was simultaneously detected by ISIS EM-BLASSO, FastmrEMMA, and pLARmEB. The QTN rs16_14976250 was associated with three bruchid resistance traits, i.e., PWL, PBE and DSI, and simultaneously co-detected by FastmrMLM, ISIS EM-BLASSO, FastmrEMMA, and pLARmEB. The remaining QTN, rs1_22916615, was associated with PBE, and was simultaneously detected by ISIS EM-BLASSO and FastmrEMMA (Table 2 and Fig 3). All the other QTNs were detected by only one multi-locus GWAS method; for instance, the QTNs rs7_14971060 (LOD score of 4.24) and rs18_14972895 (LOD score of 3.66) were detected by mrMLM and explained 25.8% and 11.6% of the total phenotypic variation associated with soybean bruchid resistance trait PWL (Table 2 and Fig 3). The parameter PBE associated with the QTN rs16_14975721 (LOD score of 4.47) was detected by pLARmEB and explained 10.1% of the total phenotypic variation. The QTNs rs15_14981757 (LOD score of 3.49) and rs18_14978590 (LOD score of 3.03) accounted for 10.49% and 8.75% of the total phenotypic variation for MDP and were detected by mrMLM and ISIS EM-BLASSO, respectively. The QTNs rs13_14978774 (LOD score of 3.39), rs15_14981838 (LOD score of 4.15), and rs16_22916222 (LOD score of 4.65), detected by mrMLM, and rs13_14973454 (LOD score of 3.43) and rs16_22915136 (LOD score of 3.10), detected by FastmrMLM, were associated with DSI and accounted for 28.4%, 19.7%, 14.5%, 2.08 × 10−6%, and 2.70 × 10−6% of the total variation, respectively (Table 2 and Fig 3). Interestingly, in this experiment, the QTN rs16_14976250 on chromosome 16 was pleiotropic, and was associated with more than one bruchid resistance traits and it was detected by more than two multi-locus GWAS methods (Table 2).
DSI = Dobie susceptibility index; PWL = percentage weight loss; PBE = percentage bruchid emergence and MDP = median development period (days).
3.5. Candidate gene identification and functional annotation
Based on the LD decay, this study identified 27 putative candidate genes spanning the window 478.45 kb upstream and downstream of the most reliable QTNs (Table 3). The functionalities of these identified genes are known to be involved in plants’ responses to biotic and abiotic stress. Some of their notable functions in response to plant stress are through their involvement in MATE efflux family proteins, F-box family proteins, LIM domain-containing proteins, nucleic acid binding transcription factors, 2-oxoglutarate (2OG) and Fe (II)-dependent oxygenase superfamily proteins, MYB domain proteins, leucine-rich repeat (LRR) protein kinase family proteins, and GDSL-like lipase/acylhydrolase superfamily proteins (Table 3). The genes, Glyma.10G025200, Glyma.01G025500, Glyma.01G026200, Glyma.01G026500, Glyma.01G027100, Glyma.01G027967, and Glyma.01G028700 found on chromosome 1 were associated with percentage adult bruchid emergence (PBE). Among these genes, Glyma.01G026200 was very close to QTN rs_22916615, notably at 272 bp. In addition, Glyma.16G091900, Glyma.16G092100, Glyma.16G092900, and Glyma.16G092600 found on chromosome 16, loci rs_22914864, were associated with percentage weight loss (PWL). Furthermore, on loci rs_14975721, the candidate genes Glyma.16G121000, Glyma.16G119200, Glyma.16G124000, Glyma.16G124200, Glyma.16G120500, Glyma.16G122100, Glyma.16G121300, Glyma.16G123900, and Glyma.16G121700 were associated with the bruchid resistance trait PBE. Moreover, on chromosome 16, Glyma.16G140800, Glyma.16G141600, Glyma.16G142500, Glyma.16G142700, Glyma.16G143100, Glyma.16G144000, and Glyma.16G143102, putative candidate genes involved in bruchid resistance traits, i.e., Dobie susceptibility index (DSI), PWL and PBE, were found in close proximity to the peak QTN rs_14976250 (Table 3).
4. Discussion
The phenotypic evaluation for bruchid resistance traits showed highly significant differences among the genotypes evaluated. This implies the presence of genetic variation among the soybean genotypes and the possibility to conventionally identify and select genotypes for bruchid resistance in soybean. However, environmental influences could slow down the selection process, since in many instances infestation begins in the field and continues to spread throughout the soybean value chain [3]. The moderate heritability estimates obtained in this study were consistent with previous studies for bruchid resistance traits in cowpea [6] and common bean [38]. Positive correlations were observed among bruchid resistance traits, suggesting that bruchid resistance in soybean might be influenced by the morphological (antixenosis) or biochemical (antibiosis) factors or both. Seed luster, color, texture, and hardness in mungbean have been found to affect bruchid oviposition behavior [39, 40]. Furthermore, antibiosis has been reported to influence bruchid resistance in soybean [2, 9]. These findings demonstrated the necessity of incorporating molecular techniques in the breeding of such complex traits to facilitate improvement.
Furthermore, the slow LD decay rate of 478.45 kb and the large LD block size observed in this study (Fig 1) could be attributed to the loss of genetic diversity caused by assortative mating during population improvement [41]. Previous studies have reported slow LD decay in soybean: 220 kb for cyst nematode resistance [42], 544.01 kb for seed hardness [43], 138 kb for plant height and number of primary branches [44], and 200 kb for seed shape [45]. Additionally, soybean is a strictly self-pollinated crop, which is expected to have a slower LD decay rate than outcrossing pollinated crops such as maize [41].
The present study detected a total of 13 QTNs using multi-locus GWAS methods. The presence of 13 significant QTNs associated with resistance to bruchids may indicate the predominance of additive gene action in conditioning soybean resistance to bruchids [10]. Previous studies reported the predominance of additive gene action conferring resistance to bruchids in cowpea [11, 46] and common bean [38]. This information could be useful for improving soybean resistance to bruchids through marker-assisted breeding. Furthermore, among the detected QTNs, rs16_14976250 was associated with more than one trait and also linked to at least two candidate genes. This polygenic association between a single locus and numerous phenotypes and candidate genes confers a polygenic nature of inheritance in the regulation of bruchid resistance traits in soybean, which is of significant interest for improving farmer-preferred varieties that are susceptible to bruchids. Pleiotropic nature of genetic control for resistance to bruchids has been reported in cowpea [6, 47, 48].
The identification of candidate genes near reliable QTNs associated with traits of interest is considered a crucial post-GWAS analysis [32]. In this study, 27 candidate genes associated with 3 bruchid resistance traits were identified within a window of 478.45 kb upstream and downstream of the 4 reliable QTNs on chromosomes 1 and 16. The functionalities of these identified genes are known to be involved in plants responses to biotic and abiotic stress. For instance, two upstream candidate genes, Glyma.01G026500, situated 20.9 kb of the peak QTN rs_22916615 on chromosome 1, and Glyma.16G123900, found at 363.4 kb of the peak QTN rs_14975721, are orthologous to AT1G70590.1 and AT4G35930.1, respectively, in Arabidopsis thaliana and Csa5M643280.1 in cucumbers, encoding for the F-box family protein, which plays a key role in jasmonate biosynthesis [49–51]. Jasmonates play a crucial role in plants defense against insects, pathogens, and abiotic stresses [49, 52, 53], suggesting that these genes could also be involved in enhancing soybean resistance to bruchids.
In addition, two genes, Glyma.01G025200 and Glyma.01G026200, were found in the vicinity of the peak QTN rs_22916615 (approximately 131.2 kb and 0.27 kb downstream, respectively) on chromosome 1. These genes encode the MATE efflux family protein transporters, which facilitate the intracellular transport of isoflavones in soybean [54]. Isoflavones are major specialized secondary metabolites in soybean that play a crucial role in the plant’s response against pathogens and insect attacks [54, 55]. Furthermore, we found four putative genes, Glyma.01G027100, Glyma.16G122100, Glyma.16G141600, and Glyma.16G144000, that encode protein kinase superfamily proteins (Table 3). In plants, the protein kinase superfamily proteins activates the biosynthesis of phytohormones including ethylene and jasmonic acid which all plays a key role in defense against insect pests [56]. Moreover, the other detected genes that could have contributed to bruchid resistance in soybean were Glyma.01G02700, Glyma.16G092600, and Glyma.16G140800. These genes are orthologous to AT1G60800.1, AT5G48740.1, and AT3G50690.1 in Arabidopsis thaliana, encoding leucine-rich repeat (LLR) family proteins (Table 3). Leucine-rich repeat receptor-like kinase (LLR-RLK) is a family protein that plays a crucial role in the regulation of plant growth, morphogenesis and signal transduction which in-turn results in plants’ responses to biotic and abiotic stress tolerance [57, 58].
Also, Glyma.16G142700 was detected on locus rs_14976250 just 34.9 kb downstream of the peak QTN. This gene encodes for ubiquitin-mediated regulated proteolysis, structural LIM domain-containing proteins, and the biosynthesis of brassinosteroid (BR), which plays a major role in seed coat development [59, 60]. A hard seed coat would aid soybean defense against boring insect pests such as bruchids. Although seed hardening will be useful to avoid insect attack, it would not be useful for the soybean food industry; therefore, candidate gene prioritization is crucial. Interestingly, two candidate genes, Glyma.16G121000 and Glyma.16G142500, detected on chromosome 16, encode calmodulin-like protein (IQ-domain 20 and 41, respectively) [61]. Calmodulin-like proteins are known to play an essential role in signalling pathways to the regulation of gene expression for glucosinolate accumulation which involves in plants defense against insects and pathogenic bacteria [61–63]. Furthermore, some of the identified candidate genes are known to be involved in encoding for GDSL-like lipase/acylhydrolase superfamily proteins, nucleic acid binding transcription factors, basic chitinase, cysteine-rich RLK, MYB domain protein 9, and Syntaxin/t-SNARE family proteins and in the biosynthesis of phenolic acids and lignin, which all play a crucial role in the plant’s defense against biotic stress [64–72].
Given the plethora of candidate genes identified in this study, validation procedures are required to determine which genes are directly responsible for soybean resistance to bruchids. Furthermore, QTNs associated with the candidate genes identified in this study need to be validated on different populations using gene expression analysis methods such as qPCR or RNA-Seq to determine the expression levels of candidate genes across resistant, moderately resistant, and susceptible genotypes, with the expectation of higher expression in resistant genotypes. In addition, functional validation using CRISPR/Cas9 gene editing or RNA interference (RNAi) can be performed to test the impact of knocking out or silencing these genes, with the expectation that their loss would result in a reduction in resistance. Marker-assisted selection (MAS) can be subsequently be used to incorporate validated QTNs into breeding programs, with the goal of producing novel genotypes with increased insect resistance. Finally, field trials will be conducted to assess the performance of these genotypes under natural insect pressure, with the expectation that genotypes with validated QTNs will exhibit greater resistance in real-world conditions. This comprehensive approach is intended to ensure the rigorous validation of insect resistance QTNs and candidate genes, hence facilitating their use in the development of bruchid resistance in soybean varieties, particularly in Sub-Saharan Africa.
5. Conclusions
In the present study, 6 multi-locus methods of the mrMLM for GWAS identified 27 candidate genes that are closely associated with bruchid resistance in soybean. The identified candidate genes could be targeted for introgression in the development of new soybean varieties with improved bruchid resistance. Furthermore, the QTN rs_14976250, which was closely linked with major bruchid resistance traits, could be used for the fast and efficient breeding of soybean genotypes with a formidable bruchid resistance through targeted marker-assisted selection. The validation and implementation of the genetic information obtained in this study could be helpful in reducing postharvest loss due to bruchids in soybean.
Supporting information
S1 Table. List and origin of 100 diverse soybean genotypes used in this study.
https://doi.org/10.1371/journal.pone.0292481.s001
(TXT)
S2 Table. Principle components used for GWAS analysis.
https://doi.org/10.1371/journal.pone.0292481.s002
(TXT)
S3 Table. The Kinship matrix used for GWAS analysis.
https://doi.org/10.1371/journal.pone.0292481.s003
(TXT)
S4 Table. Correlations among bruchid resistance traits.
https://doi.org/10.1371/journal.pone.0292481.s004
(XLSX)
S5 Table. Summary of GWAS output for bruchid resistance traits.
https://doi.org/10.1371/journal.pone.0292481.s005
(XLSX)
S1 Fig. QQ plots for GWAS for bruchid resistance traits.
Note, A = percentage weight loss, B = percentage bruchids emergence, C = median development period, and D = Dobie susceptibility index.
https://doi.org/10.1371/journal.pone.0292481.s006
(DOCX)
Acknowledgments
The authors would like to thank the Makerere University Centre for Soybean Improvement and Development for acquisition and provision of soybean germplasm from different regions. We are also grateful to the Biosciences Eastern and Central Africa—International Livestock Research Institute (BecA-ILRI) Hub for genotyping of the soybean population.
References
- 1. Pawlowski ML, Lagos-Kutz DM, Santos MDF, Lee N, Chigeza G, Nachilima C, et al. Potential Threat of Bruchids on Soybean Production in Sub-Saharan Africa. Plant Heal Prog. 2021;22(2):86–91.
- 2. Sharma S and Thakur DR. Biochemical basis for bruchid resistance in cowpea, chickpea and soybean genotypes. J. Food Technol. 2014; 6, 318–324. https://doi.org/3923/ajft.2014.318.324.
- 3. Msiska UM, Odong TL, Hailay M, Miesho B, Kyamanywa S, Rubaihayo PR, et al. Resistance of Uganda soybean germplasm to Adzuki bean bruchid. African Crop Sci J. 2018 Aug 30;26(3):399.
- 4. Kesho A. A Review on Integrated Management of Callosobruchus chinensis (Coleoptera: Bruchidae) on Faba bean. Int J Nov Res Life Sci. 2019;6(4):1–9.
- 5. Tigist SG, Melis R, Sibiya J, Amelework AB, Keneni G, Tegene A. Population Structure and Genome-Wide Association Analysis of Bruchid Resistance in Ethiopian Common Bean Genotypes. Crop Sci. 2019; 1504–1515. https://doi.org/10.2135/cropsci2018.09.0559.
- 6. Kpoviessi AD, Agbahoungba S, Chougourou DC, Agoyi EE, Badji A, Assogbadjo AE, et al. Application of multi-locus GWAS for the detection of bruchid resistance loci in cowpea (Vigna unguiculata). Plant Breed. 2022; 439–50. https://doi.org/10.1111/pbr.13014.
- 7. Miesho WB, Gebremedhin HM, Msiska UM, Mohammed KE, Malinga GM, Sadik K, et al. New sources of cowpea genotype resistance to cowpea bruchid Callosobruchus maculatus (F.) in Uganda. Int. J. Agron. Agric. Res. 2018;12(4):39–52.
- 8. Kpoviessi AD, Agbahoungba S, Agoyi EE, Chougourou DC, Assogbadjo AE. Resistance of cowpea to Cowpea bruchid (Callosobruchus maculatus Fab.): Knowledge level on the genetic advances. J. Plant Breed. Crop Sci. 2019; 11, 185–195,
- 9. Msiska UM, Miesho BW, Hailay MG, Kyamanywa S, Rubahaiyo P. Biochemicals associated with Callosobruchus chinensis resistance in soybean. Afr. J. Rural. Dev. 2018; 3, 859–868.
- 10. Msiska UM, Hailay MG, Miesho BW, Ibanda AP, Tukamuhabwa P, Kyamanywa S, et al. Genetics of Resistance in F2 Soybean Populations for Adzuki Bean Bruchid (Callosobruchus chinensis). J Agric Sci. 2018; Nov 15;10(12):171.
- 11. Miesho WB, Msiska UM, Gebremedhn HM, Malinga M, Ongom PO, Rubaihayo P, et al. Inheritance and combining ability of cowpea resistance to bruchid (Callosobruchus maculatus F.). J. Plant Breed. Crop Sci. 2018; 10, 228–238. https://doi.org/10.5897/JPBCS2018.0747.
- 12. Miesho B, Hailay M, Msiska U, Bruno A, Malinga GM, Obia P, et al. Identification of candidate genes associated with resistance to bruchid (Callosobruchus maculatus) in cowpea. Plant Breed. 2019; 138, 605–613. https://doi.org/10.1111/pbr.12705.
- 13. Somta P, Chen J, Yundaeng C, Yuan X, Yimram T, Tomooka N, et al. Development of an SNP-based high-density linkage map and QTL analysis for bruchid (Callosobruchus maculatus F.) resistance in black gram (Vigna mungo (L.) Hepper). Sci Rep. 2019; 1;9(1). pmid:30850726
- 14. Young ND, Kumar L, Menancio-Hautea D, Danesh D, Talekar NS, Shanmugasundarum S, et al. RFLP mapping of a major bruchid resistance gene in mungbean (Vigna radiata, L. Wilczek). Theor Appl Genet. 1992; 84:839–844. pmid:24201484
- 15. Mei L, Cheng XZ, Wang SH, Wang LX, Liu CY, Sun L, et al. Relationship between bruchid resistance and seed mass in mungbean based on QTL analysis. Genome. 2009;52(7):589–96. pmid:19767890
- 16. Wang L, Wu C, Zhong M, Zhao D, Mei L, Chen H, et al. Construction of an integrated map and location of a bruchid resistance gene in mung bean. Crop J. 2016;4(5):360–6.
- 17. Chen T, Hu L, Wang S, Wang L, Cheng X, Chen H. Construction of High-Density Genetic Map and Identification of a Bruchid Resistance Locus in Mung Bean (Vigna radiata L.). Front Genet. 2022;13. pmid:35873485
- 18. Somta P, Kag A A, To Moo Ka N, Ka Shiwab A K, Isemura T, Chaitieng B, et al. Development of an SNP-based high-density linkage map and QTL analysis for bruchid (Callosobruchus maculatus F.) resistance in black gram (Vigna mungo (L.) Hepper). Sci Rep. 2019; 9(1). pmid:30850726
- 19. Somta P, Kaga A, Tomooka N, Isemura T, Vaughan DA, Srinives P. Mapping of quantitative trait loci for a new source of resistance to bruchids in the wild species Vigna nepalensis Tateishi & Maxted (Vigna subgenus Ceratotropis). Theor Appl Genet. 2008;117(4):621–8.
- 20. Venkataramana PB, Gowda R, Somta P, Ramesh S, Mohan Rao A, Bhanuprakash K, et al. Mapping QTL for bruchid resistance in rice bean (Vigna umbellata). Euphytica. 2016; 1; 207(1):135–47.
- 21. Boukar O, Abberton M, Oyatomi O, Togola A, Tripathi L, Fatokun C. Introgression Breeding in Cowpea [Vigna unguiculata (L.) Walp.]. Front. Plant Sci. 2020; 1–11. https://doi.org/10.3389/fpls.2020.567425.
- 22. Blair MW, Muñoz C, Buendía HF, Flower J, Bueno JM, Cardona C. Genetic mapping of microsatellite markers around the arcelin bruchid resistance locus in common bean. Theor. Appl. Genet. 2010; 121, 393–402. pmid:20358173
- 23. Korte A, Farlow A. The advantages and limitations of trait analysis with GWAS: a review. Plant Methods. 2013; 9, 29. pmid:23876160
- 24. Ongom PO. Association mapping of gene regions for drought tolerance and agronomic traits in sorghum. ResearchGate. 2018; Available online: https://www.researchgate.net/publication/325880384.
- 25. Chang H xun, Hartman GLCannon SB. Characterization of Insect Resistance Loci in the USDA Soybean Germplasm Collection Using Genome-Wide Association Studies. Front. Plant Sci. 2017; 8, 670. pmid:28555141
- 26. Dobee P. The laboratory assessment of the inherent susceptibility of maize varieties to post- harvest infestation by Sitophilus zeamais. J. Stored Prod. Res.1974;10:183–97.
- 27. Piepho HP, Möhring J. Computing heritability and selection response from unbalanced plant breeding trials. Genetics. 2007;177(3):1881–8. pmid:18039886
- 28. Obua BT, Sserumaga JP, Opiyo SO, Tukamuhabwa P, Odong TL, Mutuku J, et al. Genetic diversity and population structure analysis of tropical soybean (Glycine Max (L.) Merrill) using single nucleotide polymorphic markers. Glob. J. Sci. Front. Res. Agric. Vet. 2020; 6, 2249–4626.
- 29. Kilian A, Wenzl P, Huttner E, Carling J, Xia L, Caig V, et al. Diversity Arrays Technology: A Generic Genome Pro fi ling Technology on Open Platforms. Data Prod. Anal. Popul. Genom. Methods Protoc. Methods Mol. Biol. 2012, 888, 67–89. https://doi.org/10.1007/978-1-61779-870-2.
- 30. Goodstein DM, Shu S, Howson R, Neupane R, Hayes RD, Fazo J, et al. Phytozome: a comparative platform for green plant genomics. Nucleic Acids Res. 2012; 40, 1178–1186. pmid:22110026
- 31. Bradbury PJ, Zhang Z, Kroon DE, Casstevens TM, Ramdoss Y, Buckler ES. TASSEL: software for association mapping of complex traits in diverse samples. Bioinformatics 2007; 23, 2633–2635. pmid:17586829
- 32. Badji A, Kwemoi DB, Machida L, Okii D, Mwila N, Agbahoungba S, et al. Genetic basis of maize resistance to multiple insect pests: Integrated genome-wide comparative mapping and candidate gene prioritization. Genes. 2020;11(6):1–27. pmid:32599710
- 33. Marroni F, Pinosio S, Zaina G, Fogolari F, Felice N, Cattonaro F, et al. Nucleotide diversity and linkage disequilibrium in Populus nigra cinnamyl alcohol dehydrogenase (CAD4) gene. Tree Genetics & Genomes. 2011; 7, 1011–1023. https://doi.org/10.1007/s11295-011-0391-5.
- 34. Nyaga C, Gowda M, Beyene Y, Muriithi WT, Makumbi D, Olsen MS, et al. Genome-Wide Analyses and Prediction of Resistance to MLN in Large Tropical Maize Germplasm. Genes. 2020;11(16):1–16.
- 35. Remington DL, Thornsberry JM, Matsuoka Y, Wilson LM, Whitt SR, Doebley J, et al. Structure of linkage disequilibrium and phenotypic associations in the maize genome. Proc. Natl. Acad. Sci. 2001; 98(20):11479–11484. pmid:11562485
- 36. Zhang YM, Jia Z, Dunwell JM. Editorial: The applications of new multi-locus gwas methodologies in the genetic dissection of complex traits. Front Plant Sci. 2019;10:1–6.
- 37. Zhang Y wen, Tamba CL, Wen Y jun, Li P, long Ren W. mrMLM v4. 0: An R Platform for Multi-locus Genome-wide Association Studies. Genom. Proteom. Bioinform. 2020; 18, 481–487. https://doi.org/10.1016/j.gpb.2020.06.006.
- 38. Acrey G, Kananji D. A Study of Bruchid Resistance and Its Inheritance in Malawian Dry Bean Germplasm. Diss Birmingham, Uk. 2007; (May).
- 39. Somta P, Ammaranan C, Ooi PAC, Srinives P. Inheritance of seed resistance to bruchids in cultivated mungbean (Vigna radiata, L. Wilczek). Euphytica. 2007;155(1–2):47–55.
- 40. Aidbhavi R, Pratap A, Verma P, Lamichaney A, Bandi SM, Nitesh SD, et al. Screening of endemic wild Vigna accessions for resistance to three bruchid species. J Stored Prod Res. 2021 Sep 1;93.
- 41. Wen Z, Boyse JF, Song Q, Cregan PB, Wang D. Genomic consequences of selection and genome-wide association mapping in soybean. BMC Genomics. 2015;1–14. http://dx.doi.org/10.1186/s12864-015-1872-y
- 42. Zhang J, Wen Z, Li W, Zhang Y. Genome-wide association study for soybean cyst nematode resistance in Chinese elite soybean cultivars. Mol. Breed. 2017; 37, 60. https://doi.org/10.1007/s11032-017-0665-1.
- 43. Zhang X, Zhao J, Bu Y, Xue D, Liu Z, Li X, et al. Genome-Wide Association Studies of Soybean Seed Hardness in the Chinese Mini Core Collection. Plant Mol. Biol. Report. 2018; 36, 605–617.
- 44. Borah J, Singode A, Talukdar A, Yadav RR, Sarma RN. Genome-wide association studies (GWAS) reveal candidate genes for plant height and number of primary branches in soybean [Glycine max (L.) merrill]. Indian J Genet Plant Breed. 2018;78(4):460–9.
- 45. Zhao B, Zhang S, Yang W, Li B, Lan C, Zhang J, et al. Multi-omic dissection of the drought resistance traits of soybean landrace LX. Plant Cell Environ. 2021; 44, 1379–1398. pmid:33554357
- 46. Kpoviessi AD, Adoukonou-sagbadja H, Agbahoungba S, Agoyi EE, Assogbadjo AE, Chougourou DC. Ecological Genetics and Genomics Inheritance and combining ability estimates for cowpea resistance to bruchid (Callosobruchus maculatus Fab.) in Benin cowpea. Ecol Genet Genomics. 2021;18:100082. https://doi.org/10.1016/j.egg.2021.100082
- 47. Visscher PM, Yang J. A plethora of pleiotropy across complex traits. Nat Publ Gr. 2016;48(7):707–8. pmid:27350602
- 48. Hassani-pak K. Knowledge Discovery in Biological Databases for Revealing Candidate Genes Linked to Complex Phenotypes. J. Integr. Bioin-form. 2017; 14, 20160002. pmid:28609292
- 49. Wang Z, Dai L, Jiang Z, Peng W, Zhang L, Wang G, et al. GmCOI1, a Soybean F-Box Protein Gene, Shows Ability to Mediate Jasmonate-Regulated Plant Defense and Fertility in Arabidopsis. Am. Phytopathol. Soc. 2005, 18 (12), 1285–1295. pmid:16478048
- 50. Liang D, Chen M, Qi X, Xu Q, Zhou F, Chen X. QTL mapping by SLAF-seq and expression analysis of candidate genes for aphid resistance in cucumber. Front Plant Sci. 2016;7:1–8.
- 51. Zhang X, Gonzalez-Carranza ZH, Zhang S, Miao Y, Liu CJ, Roberts JA. F-box proteins in plants. Annu Plant Rev Online. 2019;2(1):307–27.
- 52. Zhou SM, Kong XZ, Kang HH, Sun XD, Wang W. The involvement of wheat F-box protein gene TaFBA1 in the oxidative stress tolerance of plants. PLoS One. 2015;10(4):1–16. pmid:25906259
- 53. Stefanowicz K, Lannoo N, Zhao Y, Eggermont L, Hove J Van, Al Atalah B, et al. Glycan-binding F-box protein from Arabidopsis thaliana protects plants from Pseudomonas syringae infection. BMC Plant Biol. 2016;1–13. http://dx.doi.org/10.1186/s12870-016-0905-2
- 54. Matsuda H, Nakayasu M, Aoki Y, Yamazaki S, Nagano AJ, Yazaki K, et al. Diurnal metabolic regulation of isoflavones and soyasaponins in soybean roots. Plant Direct 2020; 4, e00286. pmid:33241173
- 55. Subramanian S, Graham MY, Yu O, Graham TL. RNA Interference of Soybean Isoflavone Synthase Genes Leads to Silencing in Tissues Distal to the Transformation Site and to Enhanced Susceptibility to Phytophthora sojae 1. Plant Physiol. 2005; 137, 1345–1353. https://doi.org/10.1104/pp.104.057257.
- 56. Hettenhausen C, Schuman MC, Wu J. MAPK signaling: A key element in plant defense response to insects. Insect Sci. 2015;22(2):157–64. pmid:24753304
- 57. Mitra SK, Chen R, Dhandaydham M, Wang X, Blackburn RK, Kota U. An autophosphorylation site database for leucine-rich repeat receptor-like kinases in Arabidopsis thaliana. Plant J. 2015; 82, 1042–1060. pmid:25912465
- 58. Dievart A, Perin C, Hirsch J, Bettembourg M, Lanau N, Artus F, et al. Plant Science The phenome analysis of mutant alleles in Leucine-Rich Repeat Receptor-Like Kinase genes in rice reveals new potential targets for stress tolerant cereals. Plant Sci. 2016;242:240–9. http://dx.doi.org/10.1016/j.plantsci.2015.06.019
- 59. Jiang W bo, hui Lin W, bo Jiang W, hui Lin W Brassinosteroid functions in Arabidopsis seed development Brassinosteroid functions in Arabidopsis seed development. Plant Physiol. 2013; 2324, e25928. https://doi.org/10.4161/psb.25928.
- 60. Du J, Wang S, He C, Zhou B, Ruan YL, Shou H. Identifcation of regulatory networks and hub genes controlling soybean seed set and size using RNA sequencing analysis. J Exp Bot. 2017;68(8):1955–72.
- 61. Abel S, Savchenko T, Levy M. Arabidopsis thaliana and Oryza sativa. BMC Evol. Biol. 2005; 25, 72. https://doi.org/10.1186/1471-2148-5-72.
- 62. Fischer C, Kugler A, Hoth S, Dietrich P. An IQ Domain Mediates the Interaction with Calmodulin in a Plant Cyclic Nucleotide-Gated Channel. Plant Cell Physiol. 2013; 54, 573–584. pmid:23385145
- 63. Faulkner C, Chiasson D, Olsson TSG, Shirasu K, Faulkner C, Gilliham M. Rapid report A calmodulin-like protein regulates plasmodesmal closure during bacterial immune responses. New Phytol. 2017; 215, 77–84. https://doi.org/10.1111/nph.14599.
- 64. Kwon SJ, Jin HC, Lee S, Nam MH, Chung JH, Kwon S Il, et al. GDSL lipase-like 1 regulates systemic resistance associated with ethylene signaling in Arabidopsis. Plant J. 2009;58(2):235–45. pmid:19077166
- 65. Zhang F, Fang H, Wang M, He F, Tao H, Wang R, et al. APIP5 functions as a transcription factor and an RNA-binding protein to modulate cell death and immunity in rice. Nucleic Acids Res. 2022;50(9):5064–79. pmid:35524572
- 66. Lv P, Zhang C, Xie P, Yang X, El-sheikh MA, Hefft DI, et al. Genome-Wide Identification and Expression Analyses of the Chitinase Gene Family in Response to White Mold and Drought Stress in Soybean (Glycine max). Life. 2022; 12, 1340. pmid:36143377
- 67. Wrzaczek M, Brosché M, Salojärvi J, Kangasjärvi S, Idänheimo N, Mersmann S, et al. Transcriptional regulation of the CRK / DUF26 group of Receptor-like protein kinases by ozone and plant hormones in Arabidopsis. BMC Plant Biol. 2010; 10(1):1–19. pmid:20500828
- 68. Du H, Yang S si, Liang Z, run Feng B, Liu L, Huang Y, et al. Genome-wide analysis of the MYB transcription factor superfamily in soybean. BMC Plant Biol. 2012; 12, 106. pmid:22776508
- 69. Tombuloglu H. Genomics Genome-wide identification and expression analysis of R2R3, 3R- and 4R-MYB transcription factors during lignin biosynthesis in flax (Linum usitatissimum). Genomics. 2019;1–14. https://doi.org/10.1016/j.ygeno.2019.05.017
- 70. Bock JB, Matern HT, Peden AA, Scheller RH. A genomic perspective on membrane compartment organization. Nature. 2001; 409, 839–841. pmid:11237004
- 71. Carter C, Pan S, Zouhar J, Avila EL, Girke T, Raikhel N V. The Vegetative Vacuole Proteome of Arabidopsis thaliana Reveals Predicted and Unexpected Proteins. Plant Cell. 2004; 16, 3285–3303. pmid:15539469
- 72. Lu L, Wang J, Zhu R, Lu H, Zheng X, Yu T. Transcript profiling analysis of Rhodosporidium paludigenum -mediated signalling pathways and defense responses in mandarin orange. Food Chem. 2015;172:603–12.: pmid:25442597