Kevin P. Smith

Genomic Selection in Plant Breeding: Knowledge and Prospects

Abstract “Genomic selection,” the ability to select for even complex, quantitative traits based on marker data alone, has arisen from the conjunction of new high-throughput marker technologies and new statistical methods needed to analyze the data. This review surveys what is known about these technologies, with sections on population and quantitative genetic background, DNA marker development, statistical methods, reported accuracies of genomic selection (GS) predictions, prediction of nonadditive genetic effects, prediction in the presence of subpopulation structure, and impacts of GS on long-term gain. GS works by estimating the effects of many loci spread across the genome. Marker and observation numbers therefore need to scale with the genetic map length in Morgans and with the effective population size of the population under GS. For typical crops, the requirements range from at least 200 to at most 10,000 markers and observations. With that baseline, GS can greatly accelerate the breeding cycle while also using marker information to maintain genetic diversity and potentially prolong gain beyond what is possible with phenotypic selection. With the costs of marker technologies continuing to decline and the statistical methods becoming more routine, the results reviewed here suggest that GS will play a large role in the plant breeding of the future. Our summary and interpretation should prove useful to breeders as they assess the value of GS in the context of their populations and resources.

Quantitative trait loci associated with resistance to Fusarium head blight and kernel discoloration in barley.

Abstract Resistance to Fusarium head blight (FHB), deoxynivalenol (DON) accumulation, and kernel discoloration (KD) in barley are difficult traits to introgress into elite varieties because current screening methods are laborious and disease levels are strongly influenced by environment. To improve breeding strategies directed toward enhancing these traits, we identified genomic regions containing quantitative trait loci (QTLs) associated with resistance to FHB, DON accumulation, and KD in a breeding population of F4:7 lines using restriction fragment length polymorphic (RFLP) markers. We evaluated 101 F4:7 lines, derived from a cross between the cultivar Chevron and an elite breeding line, M69, for each of the traits in three or four environments. We used 94 previously mapped RFLP markers to create a linkage map. Using composite interval mapping, we identified 10, 11, and 4 QTLs associated with resistance to FHB, DON accumulation, and KD, respectively. Markers flanking these QTLs should be useful for introgressing resistance to FHB, DON accumulation, and KD into elite barley cultivars.

Safety and efficacy of office-based surgery with monitored anesthesia care/sedation in 4778 consecutive plastic surgery procedures.

&NA; Office‐based surgery has several potential benefits over hospital‐based surgery, including cost containment, ease of scheduling, and convenience to both patients and surgeons. Scrutiny of office‐based surgery by regulators and state‐licensing agencies has increased and must be addressed by improved documentation of safety and efficacy. To evaluate the safety and efficacy of the authors’ officebased plastic surgery, a review was undertaken of 3615 consecutive patients undergoing 4778 outpatient plastic surgery procedures under monitored anesthesia care/sedation in a single office. The charts of 3615 consecutive patients who had undergone office‐based surgery with monitored anesthesia care/sedation between May of 1995 and May of 2000 were reviewed. In all cases, the anesthesia protocol used included sedation with midazolam, propofol, and a narcotic administered by a board‐certified registered nurse anesthetist with local anesthesia provided by the surgeon. Charts were reviewed for patient profile, types of procedures, multiple procedures, duration of anesthesia, American Society of Anesthesiologists class, and complications related to anesthesia. Outcomes measured included death, airway compromise, dyspnea, hypotension, venous thrombosis, pulmonary emboli, protracted nausea and vomiting lasting more than 24 hours, and unplanned hospital admissions. Statistical analyses were performed using the Microsoft Excel program and the SAS package. Results were as follows: 92.3 percent of the patients were female and 7.7 percent were male, with a mean age of 42.7 years (range, 3 to 83 years). Patients underwent aesthetic (95.6 percent) and reconstructive (4.4 percent) plastic surgery procedures. Same‐session multiple procedures occurred in 24.8 percent of patients. The vast majority of patients were healthy: 84.3 percent of patients were American Society of Anesthesiologists class I, 15.6 percent were class II, and 0.1 percent were class III. The operations required a mean of 111 minutes. There were no deaths, ventilator requirements, deep venous thromboses, or pulmonary emboli. Complications were as follows: 0.05 percent (n = 2) of patients had dyspnea that resolved, 0.2 percent (n = 6) of patients had protracted nausea and vomiting, and 0.05 percent (n = 2) of patients had unplanned hospital admissions (<24 hours). One patient had an emergent intubation. No prolonged adverse effects were noted. There was a 30‐day follow‐up minimum. Outpatient surgery is an important aspect of plastic surgery. It was shown that office‐based surgery with intravenous sedation, performed by board‐certified plastic surgeons and nurse anesthetists, is safe. Appropriate accreditation, safe anesthesia protocols, and proper patient selection constitute the basis for safe and efficacious office‐based outpatient plastic surgery. (Plast. Reconstr. Surg. 111: 150, 2003.)

Association Mapping of Agronomic QTLs in U.S. Spring Barley Breeding Germplasm

The use of genome‐wide association studies (GWAS) to detect quantitative trait loci (QTL) controlling complex traits has become a popular approach for studying key traits in crop plants. The goal of this study was to identify the genomic regions of barley (Hordeum vulgare L.) that impact five agronomic and one quality trait in U.S. elite barley breeding lines, as well as to identify markers tightly linked with these loci for further use in barley improvement. Advanced recombinant inbred lines submitted to the U.S. Barley Coordinated Agricultural Project (CAP) were genotyped using a platform of 3072 single nucleotide polymorphism (SNP) markers from the barley oligonucleotide pool assays (BOPAs) 1 and 2. In each of 4 yr, approximately 770 lines were evaluated in a replicated, randomized complete block design under both irrigated and dryland conditions. This gave an overall population size of >3000 lines, which we analyzed in a hierarchical fashion, including analyzing the lines in aggregate using a mixed model to account for population structure and relatedness among the lines. We identified 41 significant marker–trait associations, of which 31 had been previously reported as QTL using biparental mapping techniques; 10 novel marker‐trait associations were identified. The results of this work show that genes with major effects are still segregating in U.S. barley germplasm and demonstrate the utility of GWAS in barley breeding populations.

Genome-wide SNPs and re-sequencing of growth habit and inflorescence genes in barley: implications for association mapping in germplasm arrays varying in size and structure

BackgroundConsiderations in applying association mapping (AM) to plant breeding are population structure and size: not accounting for structure and/or using small populations can lead to elevated false-positive rates. The principal determinants of population structure in cultivated barley are growth habit and inflorescence type. Both are under complex genetic control: growth habit is controlled by the epistatic interactions of several genes. For inflorescence type, multiple loss-of-function alleles in one gene lead to the same phenotype. We used these two traits as models for assessing the effectiveness of AM. This research was initiated using the CAP Core germplasm array (n = 102) assembled at the start of the Barley Coordinated Agricultural Project (CAP). This array was genotyped with 4,608 SNPs and we re-sequenced genes involved in morphology, growth and development. Larger arrays of breeding germplasm were subsequently genotyped and phenotyped under the auspices of the CAP project. This provided sets of 247 accessions phenotyped for growth habit and 2,473 accessions phenotyped for inflorescence type. Each of the larger populations was genotyped with 3,072 SNPs derived from the original set of 4,608.ResultsSignificant associations with SNPs located in the vicinity of the loci involved in growth habit and inflorescence type were found in the CAP Core. Differentiation of true and spurious associations was not possible without a priori knowledge of the candidate genes, based on re-sequencing. The re-sequencing data were used to define allele types of the determinant genes based on functional polymorphisms. In a second round of association mapping, these synthetic markers based on allele types gave the most significant associations. When the synthetic markers were used as anchor points for analysis of interactions, we detected other known-function genes and candidate loci involved in the control of growth habit and inflorescence type. We then conducted association analyses - with SNP data only - in the larger germplasm arrays. For both vernalization sensitivity and inflorescence type, the most significant associations in the larger data sets were found with SNPs coincident with the synthetic markers used in the CAP Core and with SNPs detected via interaction analysis in the CAP Core.ConclusionsSmall and highly structured collections of germplasm, such as the CAP Core, are cost-effectively phenotyped and genotyped with high-throughput markers. They are also useful for characterizing allelic diversity at loci in germplasm of interest. Our results suggest that discovery-oriented exercises in AM in such small arrays may generate a large number of false-positives. However, if haplotypes in candidate genes are available, they may be used as anchors in an analysis of interactions to identify other candidate regions harboring genes determining target traits. Using larger germplasm arrays, genome regions where the principal genes determining vernalization sensitivity and row type are located were identified.

Modeling Dose-Response Relationships in Biological Control: Partitioning Host Responses to the Pathogen and Biocontrol Agent

ABSTRACT Breeding plants to improve the effectiveness of biocontrol agents is a promising approach to enhance disease suppression by microorganisms. Differences in biocontrol efficacy among cultivars suggest there is genetic variation for this trait within crop germplasm. The ability to quantify host differences in support of biological control is influenced by variation in host response to the pathogen and the dose of pathogen and biocontrol agent applied to the host. To assess the contribution of each of these factors to successful biocontrol interactions, we measured disease over a range of pathogen (Pythium) and biocontrol agent (Bacillus cereus UW85) inoculum doses. We fit dose-response models to these data and used model parameter estimates to quantify host differences in response to the pathogen and biocontrol agent. We first inoculated eight plant species separately with three species of Pythium and evaluated three dose-response models for their ability to describe the disease response to pathogen inoculum level. All three models fit well to at least some of the host-pathogen combinations; the hyperbolic saturation model provided the best overall fit. To quantify the host contribution to biological control, we next evaluated these models with data from a tomato assay, using six inbred tomato lines, P. torulosum, and UW85. The lowest dose of pathogen applied revealed the greatest differences in seedling mortality among the inbred lines, ranging from 40 to 80%. The negative exponential (NE) pathogen model gave the best fit to these pathogen data, and these differences corresponded to model parameter values, which quantify pathogen efficiency, of 0.023 and 0.091. At a high pathogen dose, we detected the greatest differences in biocontrol efficacy among the inbred lines, ranging from no effect to a 68% reduction in mortality. The NE pathogen model with a NE biocontrol component, the NE/NE biocontrol model, gave the best fit to these biocontrol data, and these reductions corresponded to model parameter values, which quantify biocontrol efficiency, of 0.00 and 0.038, respectively. There was no correlation between the host response to the pathogen and biocontrol agent for these inbred lines. This work demonstrates the utility of epidemiological modeling approaches for the study of biological control and lays the groundwork to employ manipulation of host genetics to improve biocontrol efficacy.

Assessing genomic selection prediction accuracy in a dynamic barley breeding population

Prediction accuracy of genomic selection (GS) has been previously evaluated through simulation and cross‐validation; however, validation based on progeny performance in a plant breeding program has not been investigated thoroughly. We evaluated several prediction models in a dynamic barley breeding population comprised of 647 six‐row lines using four traits differing in genetic architecture and 1536 single nucleotide polymorphism (SNP) markers. The breeding lines were divided into six sets designated as one parent set and five consecutive progeny sets comprised of representative samples of breeding lines over a 5‐yr period. We used these data sets to investigate the effect of model and training population composition on prediction accuracy over time. We found little difference in prediction accuracy among the models confirming prior studies that found the simplest model, random regression best linear unbiased prediction (RR‐BLUP), to be accurate across a range of situations. In general, we found that using the parent set was sufficient to predict progeny sets with little to no gain in accuracy from generating larger training populations by combining the parent set with subsequent progeny sets. The prediction accuracy ranged from 0.03 to 0.99 across the four traits and five progeny sets. We explored characteristics of the training and validation populations (marker allele frequency, population structure, and linkage disequilibrium, LD) as well as characteristics of the trait (genetic architecture and heritability, H2). Fixation of markers associated with a trait over time was most clearly associated with reduced prediction accuracy for the mycotoxin trait DON. Higher trait H2 in the training population and simpler trait architecture were associated with greater prediction accuracy.

Influence of Tomato Genotype on Growth of Inoculated and Indigenous Bacteria in the Spermosphere

ABSTRACT We previously demonstrated a genetic basis in tomato for support of the growth of a biological control agent, Bacillus cereusUW85, in the spermosphere after seed inoculation (K. P. Smith, J. Handelsman, and R. M. Goodman, Proc. Natl. Acad. Sci. USA 96:4786–4790, 1999). Here we report results of studies examining the host effect on the support of growth of Bacillus andPseudomonas strains, both inoculated on seeds and recruited from soil, using selected inbred tomato lines from the recombinant inbred line (RIL) population used in our previous study. Two tomato lines, one previously found to support high and the other low growth ofB. cereus UW85 in the spermosphere, had similar effects on growth of each of a diverse, worldwide collection of 24 B. cereus strains that were inoculated on seeds and planted in sterilized vermiculite. In contrast, among RILs that differed for support of B. cereus UW85 growth in the spermosphere, we found no difference for support of growth of the biocontrol strainsPseudomonas fluorescens 2-79 or Pseudomonas aureofaciens AB254. Thus, while the host effect on growth extended to all strains of B. cereus examined, it was not exerted on other bacterial species tested. When seeds were inoculated with a marked mutant of B. cereus UW85 and planted in soil, RIL-dependent high and low support of bacterial growth was observed that was similar to results from experiments conducted in sterilized vermiculite. When uninoculated seeds from two of these RILs were planted in soil, changes in population levels of indigenousBacillus and fluorescent Pseudomonas bacteria differed, as measured over time by culturing and direct microscopy, from growth patterns observed in the inoculation experiments. Neither RIL supported detectable levels of growth of indigenousBacillus soil bacteria, while the line that supported growth of inoculated B. cereus UW85 supported higher growth of indigenous fluorescent pseudomonads and total bacteria. The vermiculite system used in these experiments was predictive for growth of B. cereus UW85 inoculated on seeds and grown in soil, but the patterns of growth of inoculated strains—bothBacillus and Pseudomonas spp.—did not reflect host genotype effects on indigenous microflora recruited from soil to the spermosphere.

The Genetics of Winterhardiness in Barley: Perspectives from Genome-Wide Association Mapping

Winterhardiness is a complex trait that involves low temperature tolerance (LTT), vernalization sensitivity, and photoperiod sensitivity. Quantitative trait loci (QTL) for these traits were first identified using biparental mapping populations; candidate genes for all loci have since been identified and characterized. In this research we used a set of 148 accessions consisting of advanced breeding lines from the Oregon barley (Hordeum vulgare L. subsp vulgare) breeding program and selected cultivars that were extensively phenotyped and genotyped with single nucleotide polymorphisms. Using these data for genome‐wide association mapping we detected the same QTL and genes that have been systematically characterized using biparental populations over nearly two decades of intensive research. In this sample of germplasm, maximum LTT can be achieved with facultative growth habit, which can be predicted using a three‐locus haplotype involving FR‐H1, FR‐H2, and VRN‐H2. The FR‐H1 and FR‐H2 LTT QTL explained 25% of the phenotypic variation, offering the prospect that additional gains from selection can be achieved once favorable alleles are fixed at these loci.

Effect of population size and unbalanced data sets on QTL detection using genome-wide association mapping in barley breeding germplasm

Over the past two decades many quantitative trait loci (QTL) have been detected; however, very few have been incorporated into breeding programs. The recent development of genome-wide association studies (GWAS) in plants provides the opportunity to detect QTL in germplasm collections such as unstructured populations from breeding programs. The overall goal of the barley Coordinated Agricultural Project was to conduct GWAS with the intent to couple QTL detection and breeding. The basic idea is that breeding programs generate a vast amount of phenotypic data and combined with cheap genotyping it should be possible to use GWAS to detect QTL that would be immediately accessible and used by breeding programs. There are several constraints to using breeding program-derived phenotype data for conducting GWAS namely: limited population size and unbalanced data sets. We chose the highly heritable trait heading date to study these two variables. We examined 766 spring barley breeding lines (panel #1) grown in balanced trials and a subset of 384 spring barley breeding lines (panel #2) grown in balanced and unbalanced trials. In panel #1, we detected three major QTL for heading date that have been detected in previous bi-parental mapping studies. Simulation studies showed that population sizes greater than 384 individuals are required to consistently detect QTL. We also showed that unbalanced data sets from panel #2 can be used to detect the three major QTL. However, unbalanced data sets resulted in an increase in the false-positive rate. Interestingly, one-step analysis performed better than two-step analysis in reducing the false-positive rate. The results of this work show that it is possible to use phenotypic data from breeding programs to detect QTL, but that careful consideration of population size and experimental design are required.