Our Published Research

We derive formulas for the residual donor genome content during trait introgression via recurrent backcrossing and use these formulas to predict (without simulation) residual donor genome content for five future generations. Trait introgression is a common method for introducing valuable genes or alleles into breeding populations and inbred cultivars. The particular breeding scheme is usually designed to maximize the genetic similarity of the converted lines to the recurrent parent while minimizing cost and time to recover the near isogenic lines. Key variables include the number of generations and crosses and how to apply genotyping and selection. One form of trait introgression, which is our focus, involves an initial cross of an elite, homozygous recurrent parent line with a non-recurrent, homozygous donor line. The descendants of this cross are backcrossed with the recurrent parent for several generation before self-pollination in the final generation to recover lines with the alleles of interest. In this paper, we derive analytical formulas that characterize the stochastic nature of residual donor genome content during this form of trait introgression. The development of these formulas expands the mathematical methods one can integrate into breeding design. In particular, we show we can use our formulas in a novel mathematical program to allocate resources to optimize the reduction of residual donor genome content.

Bananas (Musa spp.) are one of the most highly consumed fruits globally, grown in the tropical and sub-tropical regions. We evaluated 856 Musa accessions from the breeding programs of the International Institute of Tropical Agriculture of Nigeria, Tanzania, and Uganda; the National Agricultural Research Organization of Uganda; the Brazilian Agricultural Research Corporation (Embrapa); and the National Research Centre for Banana of India. Accessions from the in vitro gene bank at the International Transit Centre in Belgium were included to provide a baseline of available global diversity. A total of 16,903 informative single nucleotide polymorphism markers were used to estimate and characterize the genetic diversity and population structure and identify overlaps and unique material among the breeding programs. Analysis of molecular variance displayed low genetic variation among accessions and diploids and a higher variation among tetraploids (p < 0.001). Structure analysis revealed two major clusters corresponding to genomic composition. The results indicate that there is potential for the banana breeding programs to increase the diversity in their breeding materials and should exploit this potential for parental improvement and to enhance genetic gains in future breeding efforts.

Using next-generation DNA sequencing, it was possible to clarify the genetic relationships of Avena species and deduce the likely pathway from which hexaploid oat was formed by sequential polyploidization events. A sequence-based diversity study was conducted on a representative sample of accessions from species in the genus Avena using genotyping-by-sequencing technology. The results show that all Avena taxa can be assigned to one of four major genetic clusters: Cluster 1 = all hexaploids including cultivated oat, Cluster 2 = AC genome tetraploids, Cluster 3 = C genome diploids, Cluster 4 = A genome diploid and tetraploids. No evidence was found for the existence of discrete B or D genomes. Through a series of experiments involving the creation of in silico polyploids, it was possible to deduce that hexaploid oat likely formed by the fusion of an ancestral diploid species from Cluster 3 (A. clauda, A. eriantha) with an ancestral diploid species from Cluster 4D (A. longiglumis, A. canariensis, A. wiestii) to create the ancestral tetraploid from Cluster 2 (A. magna, A. murphyi, A. insularis). Subsequently, that ancestral tetraploid fused again with another ancestral diploid from Cluster 4D to create hexaploid oat. Based on the geographic distribution of these species, it is hypothesized that both the tetraploidization and hexaploidization events may have occurred in the region of northwest Africa, followed by radiation of hexaploid oat to its current worldwide distribution. The results from this study shed light not only on the origins of this important grain crop, but also have implications for germplasm collection and utilization in oat breeding.

Recombination is a requirement for response to selection, but researchers still debate whether increasing recombination beyond normal levels will result in significant gains in short-term selection. We tested this hypothesis, in the context of plant breeding, through a series of simulation experiments comparing short-term selection response (≤20 cycles) between populations with normal levels of recombination and similar populations with unconstrained recombination (i.e., free recombination). We considered additive and epistatic models and examined a wide range of values for key design variables: selection cycles, QTL number, heritability, linkage phase, selection intensity and population size. With few exceptions, going from normal to unconstrained levels of recombination produced only modest gains in response to selection (≈11 % on average). We then asked how breeders might capture some of this theoretical gain by increasing recombination through either (1) extra rounds of mating or (2) selection of highly recombinant individuals via use of molecular markers/maps. All methods tested captured less than half of the potential gain, but our analysis indicates that the most effective method is to select for increased recombination and the trait simultaneously. This recommendation is based on evidence of a favorable interaction between trait selection and the impact of recombination on selection gains. Finally, we examined the relative contributions of the two components of meiotic recombination, chromosome assortment and crossing over, to short-term selection gain. Depending primarily on the presence of trait selection pressure, chromosome assortment alone accounted for 40-75 % of gain in response to short-term selection.

Self-fertilization (selfing) is commonly used for population development in plant breeding, and it is well established that selfing increases genetic variance between lines, thus increasing response to phenotypic selection. Furthermore, numerous studies have explored how selfing can be deployed to maximal benefit in the context of traditional plant breeding programs (Cornish in Heredity 65:201-211,1990a, Heredity 65:213-220,1990b; Liu et al. in Theor Appl Genet 109:370-376, 2004; Pooni and Jinks in Heredity 54:255-260, 1985). However, the impact of selfing on response to genomic selection has not been explored. In the current study we examined how selfing impacts the two key aspects of genomic selection-GEBV prediction (training) and selection response. We reach the following conclusions: (1) On average, selfing increases genomic selection gains by more than 70 %. (2) The gains in genomic selection response attributable to selfing hold over a wide range population sizes (100-500), heritabilities (0.2-0.8), and selection intensities (0.01-0.1). However, the benefits of selfing are dramatically reduced as the number of QTLs drops below 20. (3) The major cause of the improved response to genomic selection with selfing is through an increase in the occurrence of superior genotypes and not through improved GEBV predictions. While performance of the training population improves with selfing (especially with low heritability and small population sizes), the magnitude of these improvements is relatively small compared with improvements observed in the selection population. To illustrate the value of these insights, we propose a practical genomic selection scheme that substantially shortens the number of generations required to fully capture the benefits of selfing. Specifically, we provide simulation evidence that indicates the proposed scheme matches or exceeds the selection gains observed in advanced populations (i.e. F 8 and doubled haploid) across a broad range of heritability and QTL models. Without sacrificing selection gains, we also predict that fully inbred candidates for potential commercialization can be identified as early as the F 4 generation.

The Churchill-Doerge approach toward constructing empirical thresholds has received widespread use in the genetic mapping literature through the past 16 years. The method is valued for both its simplicity and its ability to preserve the genome-wide error rate at a prespecified level. However, the Churchill-Doerge method is not designed to maintain the local (comparison-wise) error rate at a constant level except in situations that are unlikely to occur in practice. In this article, we introduce the objective of preserving the local error rate at a constant level in the context of mapping quantitative trait loci in linkage populations. We derive a method that preserves the local error rate at a constant level, provide an application via simulation on a Hordeum vulgare population, and demonstrate evidence of the relationship between recombination and location bias. Furthermore, we indicate that this method is equivalent to the Churchill-Doerge method when several assumptions are satisfied.

Researchers in the life sciences (i.e., healthcare and agriculture) commonly use heuristics to process and interpret the vast amount of available DNA sequence data. The application of discrete optimization techniques, such as mixed-integer programming (MIP), remains largely unexplored and has the potential to transform the field. This paper reports on the successful use of MIP to optimize experimental design in a practical genetics application. More generally, our results illustrate the potential benefits of using MIP for subset selection problems in genetics.

Basic statistical theory implies that genotypic class cardinalities play a strong role in determining power to detect QTL, but the classes do not contribute equal information to the model. For example, while it is generally accepted that homozygotes contribute more to the detection of additive effects, heterozygotes are necessary to detect dominance effects. The literature on QTL detection often mentions the importance of genotypic class sizes in passing (Belknap (1998); Belknap et al. (1996); Jin et al. (2004); Kliebenstein (2007); Kao (2006); Martinez et al. (2002)), but no rigorous study of their relative values appears to exist. The purpose of this paper is to quantify the relative contribution of the heterozygous class. Researchers can use these results in evaluating the tradeoff between gain in statistical power and the cost of developing populations with specified genotypic class sizes. In addition, we arrive at the surprising conclusion that a misspecified additive model often outperforms a full model that incorporates dominance. This result is significant because standard software packages normally use the full model by default.