All organisms exhibit substantial quantitative trait variation within populations. Such variation is important because it can affect fitness and serve as the substrate for adaptive evolution. Identifying the quantitative trait genes (QTGs) responsible for phenotypic variation is necessary to understand the mechanisms that generate trait variation and to determine the historical action of natural selection on quantitative traits and QTGs. However, in most complex organisms, the genetic mapping of QTGs is difficult and presently not feasible to do systematically at a gene-level resolution. Model organisms that are both tractable in the laboratory and complex developmentally can serve as trial systems for developing broadly applicable methods for QTG mapping. Using the plant genetic model Arabidopsis thaliana, I have attempted to map QTGs for ecologically-significant quantitative traits â shoot branching and flowering time â through a combination of forward and reverse genetic methods. Three main research projects are reported here: i) candidate gene association mapping and linkage mapping of shoot branching; ii) regulatory network-wide candidate gene association mapping of flowering time; and iii) a survey of intra- and interspecific genetic variation at nearly half of the microRNAs (miRNAs) and their binding sites in the genome. These studies have identified strong candidate QTGs for traits that are determinants of A. thaliana fitness in the wild. I synthesize my results with those of other researchers in this area to highlight the achievements, future promise, and looming challenges for statistical genetics in terms of elucidating the genetic basis of trait variation.

Kruppel homolog 1 (Kr-h1) is a zinc finger transcription factor whose brain expression levels are associated with foraging behavior and are regulated by queen mandibular pheromone (QMP) in honey bees. We used behavioral, physiological, and genomic approaches to investigate the factors which regulate expression of Kr-h1 in honey bees, and employed a comparative genomics approach to begin to characterize the molecular function of this protein. We demonstrated that brain expression of Kr-h1 is associated with permanent physiological changes that occur during behavioral maturation from nursing to foraging, rather than the acute expression of foraging behavior. Furthermore, we demonstrated that Kr-h1 expression is modulated by cGMP, a key regulator of behavior maturation, and identified a potential cGMP response element in the promoter of Kr-h1. We then characterized the interactions between cGMP and QMP. cGMP inhibited behavioral and physiological responses to QMP, and partially inhibited expression changes of QMP-responsive genes in the brain. Treatment with these factors specifically altered expression of genes associated with GTPase regulator activity, phototransduction, and positive regulation of antibacterial peptide biosynthetic process. However, cGMP did not affect the expression of AmOr11, an odorant receptor specific for a major chemical component of QMP, suggesting that cGMP might work centrally rather than peripherally to modulate the response to QMP. Finally, we elucidated changes in gene expression that resulted from the absence of Kr-h1 expression during D. melanogaster development and discovered an associated motif in the promoter of the majority of the significantly expressed genes, which may mediate the effects of Kr-h1. Further analysis, however, showed that this motif is unlikely to be biologically relevant. These studies have demonstrated the ability of physiological factors to modulate responsiveness to social cues, and provided further information about the role of Kr-h1 in regulating behavior in bees.

From prokaryotes to vertebrates, the use of chemical signals is widespread. However, the underlying mechanisms that have led to the diversification of chemical communication are poorly understood. Here, I focus on the pheromonal communication system of the honey bee, Apis mellifera, and describe some of the molecular and physiological mechanisms that underlie pheromone production and response in this species. The mandibular glands of queen honey bees produce a pheromone (QMP) which modulates many aspects of worker behavior and physiology, and is critical for colony social organization. Chapters 1 and 2 examine how the mating process in queens can produce changes in queen behavior, physiology, and pheromone production. Chapter 3 demonstrates that these changes in pheromone production appear to be linked to differences in ovary development, and that workers are most attracted to the pheromonal blend of queens with the most activated ovaries. Chapters 4 and 5 explore how variation in pheromone response can mediate queen-worker interactions. In chapter 4, I document extensive variation in worker attraction to QMP, show that this variation is linked to individual reproductive potential, and describe some of the molecular processes that are associated with this variation. Finally, in chapter 5, I test whether variation in a pheromone receptor for the main QMP component can explain the observed variation in worker attraction to the queen, and then take a molecular evolution approach to begin to elucidate the selection pressures acting on this receptor. In addition to the work presented in this dissertation, linkage mapping studies are currently underway to identify the genetic components underlying worker attraction to QMP as well as a set of behavioral and physiological manipulations to identify the epigenetic and environmental factors that can also contribute to this variation. The results of these studies demonstrate that the chemical communication system between honey bee queens and workers acts a dialog, rather than a simple, static signal-response system, and that variation in pheromone production and response both play a critical role in modulating queen-worker interactions within the hive.

Rapidly-evolving genes which exhibit an increased rate of amino acid substitution, resulting in greater amino acid sequence difference, often provide insight into the mechanisms of adaptation and speciation. In this dissertation research, the evolution of rapidly-evolving genes under selection was examined. Homologues to the <I>Arabidopsis APETALA3 (ASAP3/TM6)</I> and <I>APETALA1 (ASAP1)</I> floral regulatory genes and the <I>CHLOROPHYLL A/B BINDING PROTEIN9 (ASCAB9)</I> photosynthetic structural gene were isolated from species in the Hawaiian silversword alliance, a premier example of plant adaptive radiation. The two floral homeotic genes are found in duplicate copies within members of the Hawaiian silversword alliance and appear to have arisen as a result of interspecific hybridization between two North American tarweed species. Molecular phylogenetic analyses suggest that the interspecific hybridization event involved members of lineages that include <I>Carlquistia muirii</I> (and perhaps <I>Harmonia nutans</I>) and <I>Anisocarpus scabridus</I>. Next, rates of regulatory and structural gene evolution in the Hawaiian species were compared to those in related species of North American tarweeds. Molecular evolutionary analyses indicate significant increases in nonsynonymous relative to synonymous nucleotide substitution rates in the <I>ASAP3/TM6</I> and <I>ASAP1</I> regulatory genes in the rapidly evolving Hawaiian species. By contrast, no general increase is evident in neutral mutation rates for these loci in the Hawaiian species. Finally, a group of potential rapidly-evolving genes were identified in <I>Arabidopsis</I> using a powerful evolutionary expressed sequence tag (EST) approach. One indicator of adaptive selection at the molecular level is an excess of amino acid replacement fixed differences per replacement site relative to the number of silent fixed differences per silent site (<font face="symbol">w</font> = Ka/Ks). The evolutionary EST approach was used to estimate the distribution of <font face="symbol">w</font> among 304 orthologous loci between <I>Arabidopsis thaliana</I> and <I>Arabidopsis lyrata</I> to identify genes potentially involved in the adaptive divergence between these two Brassicaceae species. Twenty-one of 304 genes (7%) were found to have an estimated <font face="symbol">w</font> > 1 and are candidates for genes associated with adaptive divergence. A hierarchical Bayesian analysis of protein coding region evolution within and between species also indicates that the selection intensities of these genes are elevated compared to typical <I>Arabidopsis thaliana</I> nuclear loci.

The objective of this research has been to characterize the evolution of the GATA family of transcription factors through phylogenetic, molecular, and biochemical analyses. From a phylogenetic perspective, we address three major questions. First, does this protein family represent a monophyletic or polyphyletic group? Second, what methods of gene or modular duplication are utilized within different organisms to propagate and maintain this group of proteins? Third, what are the structural or functional constraints on evolution of the conserved DNA-binding domain? These questions are addressed through a combination of computational and molecular methods. Phylogenetic analyses provide evidence of monophyletic origin for the GATA family followed by gene duplication and modular evolution, accompanied by considerable divergence outside the conserved zinc finger DNA-binding domain. Genomic comparisons permit the tracing of GATA factor evolution and provide insight into mechanisms utilized by respective organisms. Finally, mutational and biochemical analyses enable the separation of phylogenetic and structural/functional constraints on the conserved zinc finger DNA-binding domain. The result of this research is a predictive motif for classifying potentially homologous proteins to be discovered in future studies.

Behaviors are complex traits, which exhibit continuous phenotypic variation in natural populations. The continuous variation is attributable to the segregation of multiple interacting loci with individually small effects on behavior, which are sensitive to the environment. In Drosophila, loci with small, environmentally sensitive effects on behavior can be identified by screening collections of P-element insertions that have been generated in a co-isogenic background. Here, we have used this approach to identify novel candidate genes affecting geotaxis. Drosophila melanogaster are negatively geotactic, i.e., flies move opposite the Earth?s gravitational vector when disturbed. We developed a rapid assay to quantify this geotactic behavior. Individual flies are placed in a 15cm tube, and lightly tapped to the bottom. The vertical distance traveled in 10s is the measure of behavior. Using this assay, we quantified the behavior of 475 co-isogenic P-element insertion lines, generated in co-isogenic Canton-S backgrounds as part of the Berkeley Drosophila Gene Disruption Project. The most extreme scoring lines were also assayed for locomotor activity to control for pleiotropic effects associated with this quantitative trait. We found 24 lines with increased, and 15 lines with decreased geotaxis. Four lines had sex-specific effects on geotactic behavior. Seventeen of the mutations are in known genes, many of which affect neurogenesis (e.g. Mushroom-body expressed and neuralized). The remaining are insertions in predicted genes of unknown function. We tested a subset of lines in the classic geotaxis maze. Of the ten lines chosen to be tested, eight lines showed a significant difference from the parental line, and of these, six lines showed a phenotype that corroborated our observations in the climbing assay. Thus, our approach identified new candidate genes that contribute to geotaxis in Drosophila melanogaster.

The ability to respond to and interact with the chemical environment is fundamental to the survival of many species. It governs predator-prey relationships, kin and mate selection, food localization, maternal behaviors and avoidance of environmental toxins. Olfactory behavior is determined by the concerted action of multiple genes that interact with one another and with the environment, be it external, genetic or sexual. Like a number of other quantitative traits, odor-guided behavior shows significant sex-specificity in its phenotypic expression. However, the molecular basis of sexual dimorphism remains poorly defined. This study provides the first example of an autosomal pleiotropic gene that undergoes sex-specific transcriptional regulation to provide the potential for sexually dimorphic olfactory behavior. The phenotypic and molecular characterization of a P-element tagged locus, smi97B, reveals that the multiple PDZ (PSD-95, Discs-large, Zo-1) and LRR (Leucine-Rich Repeat) domain protein, Scribble (Scrib) is responsible for olfactory behavior in adult and larval stages of Drosophila melanogaster. In the adult, scrib is alternatively spliced to generate sex-specific transcripts that are correlated with sexually dimorphic olfactory phenotypes. Head-derived scrib splice variants differ in the number and positions of protein-interaction (PDZ and LRR) domains they encode. Since, Scrib is a synaptic scaffolding protein, these differences may direct the organization of sexually dimorphic synaptic signaling assemblies that contribute to odor-guided behavior.

<p>Association based tests are designed to capitalize on evolutionaryforces and population history in order to localize genes affecting thetraits of interest to within very small regions. In the case-controltest, a sample of affected individuals (the cases) and a matched setof unaffected individuals (the controls) are collected, and markerallele frequency differences between the two groups are compared. Ifa significant difference between allele frequencies is found, it isdetermined that there is an association between the marker and adisease susceptibility locus. One shortcoming of this test is that ifthe cases and controls are not well matched, or if the controls arechosen from different subpopulations than the cases, spuriousassociations may be detected within the samples which do not reflectactual population values. Additionally, it is possible that genotypeinformation on a set of controls is simply not available. We explorethe relationship between Hardy-Weinberg disequilibrium among affectedindividuals at a marker locus and linkage disequilibrium between themarker and a disease susceptibility locus and show that there is aconnection between these disequilibrium measures which may be usefulfor detecting association using affected individuals only. As part ofthis work, we introduce two summary disequilibrium terms, one allelicand one genotypic, which appear as factors in variousassociation-based measures. Following up on several suggestive equations which led to the summarydisequilibrium terms, we examine the relationship between phenotypeand marker genotypes through the perspective of classical quantitativegenetics. Within this framework, we show that in a randomly matingpopulation there is a simple connection between the additive effectsof a marker locus and the additive effects of an associated traitlocus. An equivalent relationship holds between the dominancedeviations at the marker and the dominance deviations at the traitlocus. These relationships are captured by the summary disequilibriumterms introduced earlier. Using these results, we characterize the genetic properties that lociaffecting a quantitative trait must express in order for common testsof association to be able to detect them. We examine the case-controltest and the basic form of the transmission/disequilibrium test (TDT),and show that by focusing on alleles rather than on genotypes, thesetests are sensitive mainly to additive genetic effects at thesusceptibility loci. We offer several illustrations of theeffectiveness of these tests in detecting association under variousgenetic models.<P>

<p>MYBURG, ALEXANDER ANDREW. Genetic Architecture of Hybrid Fitness andWood Quality Traits in a Wide Interspecific Cross of Tree Species. (Under thesupervision of Ross W. Whetten and Ronald R. Sederoff.)The genetic architecture of interspecific differentiation plays a key role in the evolution ofreproductive isolating barriers in plants, and has important implications for hybrid breedingprograms. Interspecific hybridization is an important approach towards the genetic improvementof tree plantations. However, incompatibilities between diverged eucalypt genomesoften lead to reduced fitness of interspecific hybrids and decrease the efficiency of hybridbreeding programs. Furthermore, very little information is available on the genetic control ofquantitative traits in interspecific hybrids of tree species.The aim of this project was to obtain a detailed description of the genetic architecture ofhybrid fitness and wood property traits in an interspecific cross between two commerciallyimportant hardwood tree species, . This cross combinesthe superior growth and adaptability of E.globulus</I>. The two species are members of different sections of the subgenus Symphyomyrtus andhybrid generations of this wide cross are characterized by large amounts of F hybrid inviabilityand F hybrid breakdown.Two interspecific backcross families of were each genotyped withmore than 800 amplified fragment length polymorphism (AFLP) markers using a new high-throughputprotocol for AFLP analysis on automated DNA sequencers. A pseudo-backcrossmapping approach was used to generate detailed comparative genetic maps of a single superior Fhybrid tree and of two backcross parents. The genetic maps of the two pure species parents andthe F hybrid were colinear and 11 comparative synteny groups were characterized. Overallrecombination rates did not differ significantly, although heterogeneity in recombination rateswas observed in several map regions.Approximately 30% of the AFLP markers were significantly distorted from expectedsegregation ratios. The distorted markers were located in specific map regions and distortion washighly directional in these regions. A Bayesian Markov chain Monte Carlo approach was used toestimate the position and effect of genetic factors that cause segregation distortion. At least sixsegregation distorting loci (SDL) were located in the genetic maps of the F hybrid. Two SDLwere detected in the genome of the E.globulus</I> backcross parent. Donor alleles were found to be favored in the recurrent geneticbackground at several SDL in the maps of the F hybrid. Marker-assisted breeding based ondetailed, whole-genome genotypes of hybrids may be useful to minimize the effect of hybridbreakdown factors in further generations of this cross.Near-infrared (NIR) analysis was used to predict wood property trait values for approximately270 individuals of each backcross family after two years of growth in a field site. The trait dataand AFLP genotypes were used for QTL detection and analysis in the parental genetic maps. Inaddition, a principal component analysis was performed on the NIR spectral data of eachbackcross family and the data used to map QTLs for NIR spectral variation. A total of 18 QTLsfor NIR predicted wood properties and NIR spectral variation were characterized in the parentalmaps of the BC family.Many of the QTLs had effects on multiple, correlated wood property traits and on raw NIRspectral variation. Individual principal components of NIR spectral variation were correlated withgroups of chemical and physical wood properties, and shared several large-effect QTLs with theindividual wood properties. These QTL may represent key genetic loci that are involved in thegenetic differentiation of wood properties between these two species.<P>

<p>PHELAN, THOMAS JOSEPH, Genetic and Molecular Analysis of Plant Nuclear Matrix Proteins. (Under the direction of Steven L. Spiker.)The eukaryotic nucleus is composed of DNA, RNA and protein, encapsulated by a nuclear envelope. DNA is compacted up to ten thousand times in order to be packaged into the nucleus. The nucleus must maintain order in the presence of a very high density and variety of protein and RNA. The nuclear matrix is a proteinaceous network thought to provide structure and organization to the nucleus. We believe that relatively stable interactions of nuclear molecules with the nuclear matrix are key to organization of the nucleus. Numerous "Matrix Attachment Region" DNA elements (MARs), have been isolated from plants, animals, and fungi. Evidence suggests that these MARs attach to the nuclear matrix, delimiting loops of chromosomal DNA. In studies of transgenic plants and animals, MARs have been shown to give important advantages to organisms transformed with genes flanked by these elements. Unlike most DNA elements, no specific sequence elements have been identified in MAR DNAs. Partly due to the insolubility of the matrix, and to the heterogeneity of MAR DNA, very few of the protein components of the nuclear matrix have been identified. This work presents analysis the proteins of the plant nuclear matrix. We have characterized a set of related proteins from the model plant Arabidopsis that associate with MAR DNA in vitro. These proteins appear to be similar to the NOP56/NOP58 family of proteins previously identified in several eukaryotic organisms. The NOP56/NOP58 proteins are thought to be involved in modifications of ribosomal RNA. Binding studies presented in this work suggest that these plant proteins may participate in RNA/DNA/protein complexes in the nucleus.<P>