ABSTRACT
The Dectes stem borer, Dectes texanus LeConte (Coleoptera: Cerambycidae), can significantly reduce yields by causing significant lodging in soybean. While this stem borer has not been considered a major pest of soybean, damage from it is increasing in the United States Midwest region with no current elite cultivars found resistant. Our objective was to map quantitative trait loci (QTL) that reduce girdled stems caused by Dectes stem borer infection and infestation of Dectes stem borer. A genome‐wide association study (GWAS) using 50,000 single nucleotide polymorphisms was used to analyze data from a population of maturity group (MG) V to VII soybean accessions grown in North Carolina, which had been scored for Dectes stem borer larvae infestation and girdled stems caused by Dectes stem borer infestation. The GWAS identified 3 QTL with reduced larvae infestation and 4 QTL for reduced girdled stems. Allele effects ranged from 1% to 9% reduced larvae infestation or girdled stems. The QTL identified and germplasm containing the beneficial alleles can be used for improving resistance to the damage caused by the Dectes stem borer in elite soybean cultivars.
Keywords: Dectes stem borer, insect resistance, quantitative trait loci, soybean
1. Introduction
The Dectes stem borer, Dectes texanus Leconte (Coleoptera: Cerambycidae), is native to eastern North America and can infect soybeans, cultivated sunflowers, ragweed, and cocklebur (Buschman and Sloderbeck 2010). The Dectes stem borer has not been considered a major pest of soybean in the United States, but damage has been increasing in both frequency and geographic range (Jeschke 2018). In soybean, yield loss caused by Dectes stem borer mainly occurs through increased lodging and stem breakage due to internal girdling caused by infection of a single larva reaching the base of a soybean stem (Richardson 1975). Currently, no elite cultivars have been found resistant to the stem borer (Buschman and Sloderbeck 2010; Seiter 2018).
The life cycle of the Dectes stem borer has made it difficult to produce large‐scale cultivar screens and deploy effective biocontrol measures. In the soybean field, the overwintered adult beetle emerges from plant stubble around June to mate and oviposit on a mid‐canopy petiole (Rystrom 2015). The female D. texanus can mate in 5 days from emergence with a mating ritual that requires a contact female‐producing sex pheromone(s) (Crook et al. 2004). Larva tunnel inside the petiole and stem with the only visible indication of infestation being a small oviposition mark found on the petiole. The larvae are also cannibalistic, so by the end of the season, only one larva survives to travel down to the plant base, girdle the stem, and make a safe home for itself with a frass plug in preparation for diapause. Once girdled, slight pressure can cause the whole plant to topple over, which will reduce yield during harvest (Adkisson et al. 1960).
For biocontrol, aerial pesticide is not recommended since there is a long period of egg laying, larvae are protected inside the plant, and only one larvae needs to survive for plant lodging (Campbell and Van Duyn 1977; Michaud, Grant, and Jyoti 2007; Sloderbeck and Buschman 2011). Host resistance could potentially reduce infestation of Dectes stem borer or be used to reduced girdling caused by the larvae. A screen of 618 plant introduction (PI) accessions within maturity groups (MG) V–VII were previously screened for larvae infestation and girdling in the southeast United States (Richardson 1975). Four accessions were found with less than 10% girdled stems and 12 accessions with less than 30% larvae infestation. In this study, using the phenotype data from these 618 accessions along with recently obtained SNP data on these accessions, we aimed to identify regions within the soybean genome that provides resistance to infestation and girdled stems due to the Dectes stem borer. These associated genomic loci can be used to develop increased resistance in response to potential infestation by D. texanus .
2. Materials and Methods
2.1. Phenotype Data
Phenotype data for the late maturity groups (V–VII) were obtained from Richardson (1975). Briefly, 618 accessions were tested in 1972 for larva infestation and girdled stems in a historically infested North Carolina field. To ensure high insect pressure, D. texanus infested stubble was spread throughout the field and cage tests. The field design included one replication for each accession and plots 3‐m long row with approximately 40 plants per row. Each cage experiment contained 15 rows, each 0.5‐m long with approximately 10 plants per individual plot within the cages. After the initial screen, 185 accessions were chosen to plant in 1973, 3‐m long, three replications with approximately 30 plants each. Cage experiments in 1973 were similar size as 1972 but each plot was three replicates each 1‐m long. A few accessions were selected to test in 1974, 6 m, three replications. Of the 618 accessions, 379 had been genotyped with the SoySNP50k iSelect Beadchip (Song et al. 2015) and had sufficient replicated phenotype data from the field and cage screening. Best linear unbiased estimations (BLUEs) for each of accession was calculated using ASReml‐R (Butler et al. 2017).
2.2. Genome Wide Association Study (GWAS)
2.2.1. Best Linear Unbiased Estimates
The phenotypic values for larvae infestation and girdling were estimated by models incorporating known sources of variation in ASReml‐R, such as environment (field or cage and year), replications, and maturity group as a covariate (Butler et al. 2017). Since Richardson (1975) does not provide the raw phenotypic data, BLUEs were created by using line, year, and type (field or cage) as fixed effects and maturity group as a covariate.
Genome Wide Association of traits to markers on the SoySNP50k iSelect BeadChip.
The marker association for the best linear estimates for larvae infestation and girdle stems was conducted using the Fixed and random model Circulating Probability Unification (FarmCPU) method ran through the software Genomic Association and Prediction Integrate Tool (GAPIT) in R studio (Liu et al. 2016; Wang and Zhang 2021). Briefly, FarmCPU is a Mixed Linear Mixed Model (MLMM) but divided in two by a Fixed Effect Model (FEM), containing the testing markers one at a time and using multiple associated markers as covariates, and a random effect model (REM), which estimates the associated markers by using them to define kinship (Liu et al. 2016). The minor allele frequency threshold was set to MAF > 0.05 and heterozygotes along with markers mapping only to scaffolds were removed from the marker data. Scaffolds may not have a mapping location so they were eliminated from analysis. For the 379 later maturing accessions GWAS, 35,019 markers were included after filtering. Linkage disequilibrium around significant SNPs was calculated for these SNPs (Table S1) on a 25 SNP sliding window using Tassel 5.2.94 (Bradbury et al. 2007). Based on the principal component analysis using eigenvalue charts and data representation in 2D and 3D models, principal component analysis (PCA) was set to four on all analyses (Figure S1). Q‐Q plots were created for each GWAS analysis to help determine if population structure and kinship analysis were being controlled in the analysis (Figures S2 and S3). The effect is a standardized difference of the means. The Bonferroni correction for significance was applied using α = 0.05.
3. Results and Discussion
Statistical and computation tools have improved since the original publication of the Richardson (1975) data. The authors that originally collected and analyzed the phenotypic data noted a negative correlation with maturity and girdled stems. The original analysis observed an average infestation rate of 48.7% for the 1972 screen, which also included 26.1% girdled stems. The original analysis also demonstrated that infestation decreased as maturity of the line increased with average infestation for Maturity Group (MG) V, VI, and VII at 56.5%, 49.4%, and 41.2%, respectively. Girdled stems also showed a decrease with an increase in maturity with MG V, VI, and VII having girdled stem values of 36.9%, 27.4%, and 15.4%, respectively (Richardson 1975). Despite this observed negative correlation of infestation and girdling with maturity, the authors were not able to incorporate maturity into their statistical model.
For our reanalysis of the phenotype data, maturity was included as a covariate. When maturity was included as a covariate, the 618 lines had an average BLUE for larvae infestation of 60% and the percentage of girdled stems was 46% (Figure 1, Table S2). Using a significance threshold of p < 0.05, four accessions were found to have significantly less larvae infestation and five accessions with significantly less girdling (Table 1). PI 165989, PI 175195, and PI 319525 were significantly less for both traits while FC 03719 was significantly less for only larvae. PI 219698 and PI 171437 were only significant for fewer girdled stems. Both traits had moderate heritability at 0.51 for larvae infestation and 0.45 for girdled stems.
FIGURE 1.
Phenotypic histograms of the Best Linear Unbiased Estimates (BLUEs) of larvae count for (A) all lines and (B) lines genotyped for the SoySNP50k and of girdled stems for (C) all lines and (D) lines genotyped for the SoySNP50k.
TABLE 1.
Soybean accessions with significantly fewer larvae or girdled stems.
| Accessions | Larvae (mean ± SE, %) | Girdled stems (mean ± SE, %) | Phenotypic screens |
|---|---|---|---|
| Accessions with significantly fewer larvae | |||
| PI 165989 | 0 ± 0 | 0 ± 0 | 1972 Field, 1972 Cage, 1973 Field |
| PI 175195 | 5 ± 2 | 4 ± 2 | 1972 Field, 1972 Cage, 1973 Field |
| PI 319525 | 6 ± 4 | 4 ± 4 | 1972 Field, 1972 Cage, 1973 Field |
| FC 031719 | 14 ± 3 | 13 ± 3 | 1972 Field, 1972 Cage, 1973 Field |
| Accessions with significantly fewer girdled stems | |||
| PI 175195 | 5 ± 2 | 4 ± 2 | 1972 Field, 1972 Cage, 1973 Field |
| PI 319525 | 6 ± 4 | 4 ± 4 | 1972 Field, 1972 Cage, 1973 Field |
| PI 165989 | 0 ± 0 | 0 ± 0 | 1972 Field, 1972 Cage, 1973 Field |
| PI 219698 | 12 ± 9 | 9 ± 9 | 1972 Field, 1972 Cage, 1973 Field |
| PI 171437 | 21 ± 9 | 9 ± 9 | 1972 Field, 1972 Cage, 1973 Field |
Many of the late maturity plant introduction accessions screened by Richardson (1975) are still contained within the USDA soybean germplasm collection and available to breeders for use as a source of germplasm for breeding for resistance to Dectes stem borer. The USDA soybean germplasm collection has also been genotyped with the SoySNP50 iSelect Beadchip (Song et al. 2015). Out of the 618 accessions tested by Richardson 1975, there are 379 accessions with SoySNP50 SNP data that also had sufficient replicated phenotypic data from the 1975 screen. Since the Richardson (1975) thesis contained data from their original screen, this allowed us to take the historical phenotype data with our reanalysis and BLUE estimates and perform a GWAS with the SoySNP50 SNP data using the 379 accessions (Table S2).
With the 379 accessions, the SoySNP50 Beadchip provided 35,019 quality SNPs that were included in the GWAS. Using a Bonferroni correction for significance level, there were three significant markers (p‐value < 0.05) associated with reduced larvae infestation on chromosomes 7, 12, and 13 (Figure 2, Table 2). For reduced girdled stems, there were four significantly associated markers on chromosomes 1, 7, 11, and 12. The additive allelic effects ranged from 3% to 9% (Figure 3). The largest QTL located on chromosome 12 saw the minor allele reduced larvae infection from 67% down by 9% to 58% infestation (Figure 3).
FIGURE 2.
Manhattan plots for (A) larvae and (B) girdled stems from 379 accessions analyzed for the GWAS analysis. Green line is the Bonferroni correction at p = 0.01, and the red dashed line is at p = 0.05.
TABLE 2.
Significant markers for larvae and girdled stems.
| dbSNP name | Chromosome | Position Wm. a1 | Position Wm. a2 | p | MAF | Effect (%) | −log10 (p) |
|---|---|---|---|---|---|---|---|
| Larvae | |||||||
| ss715597107 | 7 | 308,906 | 313,803 | 1.34E‐06 | 0.49 | −3 | 5.87 |
| ss715612627 | 12 | 35,998,785 | 35,989,537 | 2.61E‐07 | 0.11 | 9 | 6.58 |
| ss715615024 | 13 | 29,524,129 | 30,724,301 | 1.88E‐06 | 0.21 | −4 | 5.73 |
| Girdled stems | |||||||
| ss715580443 | 1 | 53,365,928 | 54,250,600 | 1.86E‐06 | 0.09 | 5 | 5.73 |
| ss715597107 | 7 | 308,906 | 313,803 | 5.49E‐06 | 0.49 | −3 | 5.26 |
| ss715611045 | 11 | 6,890,057 | 6,900,048 | 3.06E‐06 | 0.45 | −3 | 5.51 |
| ss715612626 | 12 | 35,982,867 | 35,973,619 | 1.55E‐06 | 0.11 | 7 | 5.81 |
Note: Negative effects infer that the minor allele statistically has the lower larvae or girdled stem phenotype.
FIGURE 3.
Box and whisker plots of markers found significant in GWAS with (A) larvae and (B) girdled stems data. The solid line signifies the median while the dashed line is the mean. The boxes are notched as a comparison interval around the median value. The marker label is formatted as chromosome_position (soy genome a1) then major allele listed first and minor allele listed second. The major allele in the SNP name was originally determined using the 50 k SNP chip data genotyped with over 20,000 accessions.
For each significant marker, the surrounding interval was checked using Soybase (www.soybase.org) to look for any previous identified QTL that might be related to the novel QTL discovered in this study. To determine the interval for QTL, the nearest marker found on the Glyma.1.0 physical map was identified and 1 cM on each side of that marker using the Composite 2003 genetic map was considered the surrounding ‘interval’. The nearest marker information on the genetic and physical maps along with previously known QTL listed in Soybase, which fall within that interval are listed in Table S3. The QTL that overlap with the significant markers for larvae infestation include QTL for P. sojae resistance, sudden death syndrome (SDS) resistance, lodging resistance, drought tolerance, flood tolerance, and pubescence density (Table S2). For girdling, significant markers overlap with whitefly resistance, drought tolerance, and lodging resistance. Lodging resistance was a common QTL across larvae infestation and girdled stem resistant QTL. The infestation and girdling QTL being associated with lodging resistance QTL are especially of interest since the main loss for yield from Dectes stem borer is due to increased lodging and stem breakage (Richardson 1975).
The marker on chromosome 7, position 313,803 bp, was significant for both Dectes stem borer traits but did not overlap any known QTL for insect resistance or other traits of potential interest. The marker on chromosome 12 for larvae is located 15,918 bp from the significant marker on chromosome 12 for girdled stems and both intervals had overlap with drought tolerance related QTL.
Significant QTL identified here and overlapping QTL for insect and disease resistance along with architectural traits could give further insight into the mechanism of resistance to D. texanus . As this pest becomes more prevalent in the Midwest, there is renewed interest in breeding for resistant lines through finding germplasm and QTL that can increase resistance. Future work should entail phenotyping in more environments to confirm QTL discovered in this study and fine mapping of highly resistant vs susceptible accessions to provide a better understanding of the genetics involved in the resistance to Dectes stem borer. As future work begins to find additional resistance to this emerging pest, the ability to gain insight into resistance using phenotype data collected 50 years ago on accessions that have been genotyped with SNPs along with these accessions being still available to breeders today demonstrates the utility and value of having public resources such as the USDA germplasm collection characterized with molecular resources. The soybean accessions identified as having low infestation and/or stem breakage, along with the putative QTL identified here, are potential germplasm sources and genomic targets for improving resistance to the damage caused by the stem borer in elite soybean cultivars.
Author Contributions
S.J. and D.H. designed the research, S.J. performed the research, S.J. and D.H. analyzed data, S.J. wrote the manuscript and D.H. edited the manuscript.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Data S1. Peer Review.
Figure S1. Principal component analysis. Top row includes 2‐D graph of principal components one and two. Bottom left is a 3‐D graph of principal components one, two, and three. Bottom right is Eigen values for the SNPs included for the accessions used for GWAS. The first four principal components were chosen since the variance line at that point starts to level and the variance percentage does not change much for the later principal components.
Figure S2. QQ plots from the FARMPU‐GWAS.
Figure S3. Kinship analysis for the accessions and marker data.
Table S1. Linkage Disequilibrium results for the SoySNP50K filtered for SNPs with a minor allele frequence >0.05 and using a 25 SNP sliding window.
Table S2. List of means ± SE (%) and Best Linear Unbiased Estimates (BLUEs) (%) for each line produced by our model with data derived from Richardson et al 1975 within MG V‐VII. Tests performed and whether that line is included on the SoySNP50k iSelect BeadChip information is included. Bolded accessions were not included in the GWAS since they were only tested for one year and the BLUEs were both less than 10%.
Table S3. List of D. texanus significant markers and previously known QTL on GmComposite2003 on Soybase.
Acknowledgments
The research reported in this publication was supported by the Nebraska Soybean Board project #1744. This work was completed utilizing the Holland Computing Center of the University of Nebraska, which receives support from the Nebraska Research Initiative.
Funding: This work was supported by Nebraska Soybean Board (NSB) (1744).
Data Availability Statement
Genotype data for all accessions is available at www.soybase.org. Original phenotypic data is available from Richardson, 1975. All other data supporting the findings are contained within the manuscript.
References
- Adkisson, P. , Vanderzant E. S., Bull D., and Allison W.. 1960. “A Wheat Germ Medium for Rearing the Pink Bollworm.” Journal of Economic Entomology 53, no. 5: 759–762. [Google Scholar]
- Bradbury, P. J. , Zhang Z., Kroon D. E., Casstevens T. M., Ramdoss Y., and Buckler E. S.. 2007. “TASSEL: Software for Association Mapping of Complex Traits in Diverse Samples.” Bioinformatics 23, no. 19: 2633–2635. [DOI] [PubMed] [Google Scholar]
- Buschman, L. L. , and Sloderbeck P. E.. 2010. “Pest Status and Distribution of the Stem Borer, Dectes Texanus, in Kansas.” Journal of Insect Science 10, no. 1: 198. 10.1673/031.010.19801. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Butler, D. , Cullis B., Gilmour A., Gogel B., and Thompson R.. 2017. ASReml‐R Reference Manual Version 4. Hemel Hempstead, HP1 1ES, UK: VSN International Ltd. [Google Scholar]
- Campbell, W. , and Van Duyn J.. 1977. “Cultural and Chemical Control of Dectes texanus Texanus on Soybeans.” Journal of Economic Entomology 70, no. 2: 256–258. [Google Scholar]
- Crook, D. J. , Hopper J. A., Ramaswamy S. B., and Higgins R. A.. 2004. “Courtship Behavior of the Soybean Stem Borer Dectes texanus Texanus (Coleoptera: Cerambycidae): Evidence for a Female Contact Sex Pheromone.” Annals of the Entomological Society of America 97, no. 3: 600–604. [Google Scholar]
- Jeschke, M. 2018. Dectes Stem Borer in Soybeans. https://www.pioneer.com/us/agronomy/stem‐borer‐dectes.html.
- Liu, X. , Huang M., Fan B., Buckler E. S., and Zhang Z.. 2016. “Iterative Usage of Fixed and Random Effect Models for Powerful and Efficient Genome‐Wide Association Studies.” PLoS Genetics 12, no. 2: e1005767. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Michaud, J. , Grant A. K., and Jyoti J.. 2007. “Impact of the Stem Borer, Dectes texanus, on Yield of the Cultivated Sunflower, Helianthus annuus .” Journal of Insect Science 7, no. 1: 21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Richardson, L. G. 1975. Resistance of Soybeans to a Stem Borer Dectes texanus Texanus LeConte. Raleigh North Carolina, USA: North Carolina State University. [Google Scholar]
- Rystrom, Z. D. (2015). Seasonal Activity and Sampling Methods for the Dectes Stem Borer, Dectes texanus LeConte, In Nebraska Soybeans. Lincoln Nebraska, USA: University of Nebraska. [Google Scholar]
- Seiter, N. (2018). Dectes Stem Borer and Lodged Soybeans. Department of Crop Sciences, University of Illinois at Urbana‐Champaign. https://farmdoc.illinois.edu/field‐crop‐production/insects/dectes‐stem‐borer‐and‐lodged‐soybeans.html.
- Sloderbeck, P. , and Buschman L.. 2011. “Aerial Insecticide Treatments for Management of Dectes Stem Borer, Dectes texanus, in Soybean.” Journal of Insect Science 11, no. 1: 49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Song, Q. , Hyten D. L., Jia G., et al. 2015. “Fingerprinting Soybean Germplasm and Its Utility in Genomic Research.” G3 (Bethesda) 5, no. 10: 1999–2006. 10.1534/g3.115.019000. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wang, J. , and Zhang Z.. 2021. “GAPIT Version 3: Boosting Power and Accuracy for Genomic Association and Prediction.” Genomics, Proteomics & Bioinformatics 19, no. 4: 629–640. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data S1. Peer Review.
Figure S1. Principal component analysis. Top row includes 2‐D graph of principal components one and two. Bottom left is a 3‐D graph of principal components one, two, and three. Bottom right is Eigen values for the SNPs included for the accessions used for GWAS. The first four principal components were chosen since the variance line at that point starts to level and the variance percentage does not change much for the later principal components.
Figure S2. QQ plots from the FARMPU‐GWAS.
Figure S3. Kinship analysis for the accessions and marker data.
Table S1. Linkage Disequilibrium results for the SoySNP50K filtered for SNPs with a minor allele frequence >0.05 and using a 25 SNP sliding window.
Table S2. List of means ± SE (%) and Best Linear Unbiased Estimates (BLUEs) (%) for each line produced by our model with data derived from Richardson et al 1975 within MG V‐VII. Tests performed and whether that line is included on the SoySNP50k iSelect BeadChip information is included. Bolded accessions were not included in the GWAS since they were only tested for one year and the BLUEs were both less than 10%.
Table S3. List of D. texanus significant markers and previously known QTL on GmComposite2003 on Soybase.
Data Availability Statement
Genotype data for all accessions is available at www.soybase.org. Original phenotypic data is available from Richardson, 1975. All other data supporting the findings are contained within the manuscript.