<p> Stem rust, caused by the macrocyclic fungal pathogen <i> P. graminis</i> (<i>Pg</i>), is one of the most devastating diseases of wheat and other small grains globally; and the emergence of new stem rust races virulent on deployed resistance genes brings urgency to the discovery of more durable sources of genetic resistance. Given its intrinsic durability and effectiveness across a broad range of pathogens, non-host resistance (NHR) presents a compelling strategy for achieving long-term rust control in wheat. However, NHR to <i>Pg</i> (<i>Pg</i>-NHR) remains largely unexplored as a protection strategy in wheat, in part due to the challenge of developing a genetically tractable system in which <i>Pg</i>-NHR segregates. In this dissertation, an investigation of <i>Pg</i>-NHR is undertaken via the pathogen's alternate (sexual) host, barberry (<i> Berberis</i> spp.). Within the highly diverse <i>Berberis</i> genus, numerous species function as alternate hosts to <i>Pg</i> but others are non-hosts. European barberry (<i>B. vulgaris</i> L.), for example, is susceptible to <i>Pg</i> infection but Japanese barberry (<i>B. thunbergii</i> DC.) is a non-host. In this study, the nothospecies <i>B. &times;ottawensis</i> C.K. Scheid, an inter-specific hybrid between <i>Pg</i>-susceptible <i>B. vulgaris</i> and <i>Pg</i>-resistant <i>B. thunbergii</i>, is explored as a possible means of mapping the gene(s) underlying the apparent <i> Pg</i>-NHR exhibited by <i>B. thunbergii</i>. The overall goal of this research is to contribute to the global search for novel sources of potentially durable stem rust resistance genes. </p><p> The first chapter describes a field study conducted in western Massachusetts, in which a natural population of <i>B. &times;ottawensis</i> was characterized to determine if the hybrid can be used to genetically dissect the <i>Pg</i>-NHR exhibited by <i>B. thunbergii</i>. A population of 63 <i>B. &times;ottawensis</i> individuals were clonally propagated, phenotyped for disease response to <i>Pg</i> via controlled inoculation using overwintered telia of <i>Pg</i> found on naturally infected <i>E. repens</i>, and genotyped using the <i>de novo </i> genotyping-by-sequencing (GBS) pipeline GBS-SNP-CROP. Controlled inoculation of a subset of 53 <i>B. &times;ottawensis</i> accessions, verified via GBS to be true, first-generation hybrids, revealed 51% susceptible, 33% resistant, and 16% intermediate phenotypes. Although such variation in disease response within a natural population of F₁ hybrids could be explained by non-nuclear (cytoplasmic) inheritance of resistance, a similar pattern of segregation was observed in a population of <i>B. &times;ottawensis </i> full-sibs, developed via controlled crosses. The results of this first chapter demonstrate not only that the <i>Pg</i>-NHR observed in <i>B. thunbergii</i> segregates among F₁ interspecific hybrids with <i>Pg</i>-susceptible <i>B. vulgaris</i> but that the resistance is likely nuclearly inherited. Therefore, at least in principle, the gene(s) underlying <i>Pg</i>-NHR in <i>B. thunbergii</i> should be mappable in an F₁ population derived from the controlled hybridization of the two parental species. </p><p> Building on the results of first chapter, the second chapter of this dissertation details the generation and use of a bi-parental <i>B. &times;ottawensis </i> mapping population to develop genetic linkage maps for both parental species and begin mapping the gene(s) underlying <i>Pg</i>-NHR in <i> B. thunbergii</i>. Using 162 full-sib F₁ hybrids and a total of 15,411 sequence variants (SNPs and indels) identified between the parents via GBS, genetic linkage maps with 1,757 and 706 markers were constructed for <i>B. thunbergii</i> accession 'BtUCONN1' and <i>B. vulgaris </i> accession 'Wagon Hill', respectively. In each map, the markers segregated into 14 linkage groups, in agreement with the 14 chromosomes present in these <i> Berberis</i> spp. The total lengths of the linkage maps were 1474 cM (<i>B. thunbergii</i>) and 1714 cM (<i>B. vulgaris</i>), with average distances between markers of 2.6 cM and 5.5 cM. QTL analysis for <i>Pg</i> resistance led to the identification of a single QTL, dubbed Q<i>Pg</i>r-3S, on the short arm of chromosome 3 of <i> B. thunbergii</i>. The peak LOD score of Q<i>Pg</i>r-3S is 28.2, and the QTL spans 13 cM, bounded by the distal SNP marker M411 and proximal SNP marker M969. To gain further insight into the Q<i>Pg</i>r-3S region, a chromosome-level 1.2 Gb draft genome for <i>B. thunbergii</i> was assembled using long PacBio reads and Hi-C data. By anchoring the <i> B. thunbergii</i> linkage map to the draft genome, the 13 cM Q<i> Pg</i>r-3S region was found to correspond to ~3.4 Mbp, represented by 10 contigs. Using a 189.3 Mb transcriptome assembled from a multiple tissue library of RNA-seq data, the Q<i>Pg</i>r-3S region was found to contain 99 genes. To help narrow this list to candidate genes of highest priority for subsequent investigation, a combination of approaches was taken. Specifically, annotation of the QTL region and differential gene expression analysis led to the identification of 12 candidate genes within the region. (Abstract shortened by ProQuest.) </p><p>