Abstract

Song Diversification and Speciation in the Empidonax difficilis-occidentalis-flavescens Complex

by

Andrew Christopher Rush

Doctor of Philosophy in Integrative Biology

University of California, Berkeley

Professor Rauri Bowie, Chair

Speciation can occur when barriers to gene flow form between populations. Phenotypic differences resulting from divergent selection can act as strong pre-zygotic barriers to gene flow. Birds provide excellent opportunities to examine the effectiveness of pre-zygotic barriers, because phenotypic traits, such as song and plumage, which affect species recognition and mate choice, are often conspicuous and relatively easy to observe and quantify. In the research outlined in this dissertation, I analyze the effects of divergence in song on lineage diversification in the six principal taxa that comprise the Empidonax-difficilis-occidentalis-flavescens complex, a clade of suboscine passerines. This dissertation is one of the broadest examinations of the interaction between genotype and innate song type yet conducted, and provides insights relevant to the role of divergent signals in speciation in a wide range of organisms.

In Chapter 1, I examine an area of geographic contact in southwestern Canada between two of these taxa, the Pacific-slope Flycatcher (E. d. difficilis) and the Cordilleran Flycatcher (E. o. hellmayri). Contact zones between recently diverged taxa offer unique opportunities to test whether the forms are reproductively isolated and therefore distinct species. I present the first analysis of genetic variation across this region, in order to determine whether hybridization and gene flow occurs between these taxa. I determine that parental populations of Pacific-slope and Cordilleran Flycatchers have distinct mitochondrial haplotypes, and that all of the individuals sampled in interior southwestern Canada have the Pacific-slope haplotype. In contrast, analysis of nuclear DNA (AFLPs) indicates a high level of population admixture between Pacific-slope and Cordilleran Flycatchers in this region, although I find no evidence of nuclear gene flow into core parental populations. I suggest that the discordance between the mitochondrial and nuclear markers most likely results from stochastic loss of Cordilleran mitochondrial haplotype lineages facilitated by asymmetries in mating due to earlier arrival and greater abundance of Pacific-slope Flycatchers in the contact zone. Although the discovery of hybridization between Pacific-slope and Cordilleran Flycatchers in southwestern Canada calls into question their status as distinct species, the lack of evidence of gene flow into parental populations indicated a need for genetic analyses of populations representing a broader sampling of the geographic range of each species.

In the remaining two chapters of the dissertation, I address two major questions:

1. Does innate song function as a particularly strong isolating mechanism between incipient suboscine species?

2. How do patterns of divergence in innate song differ from those of learned song, and how has song divergence affected lineage diversification across the E. difficilis-occidentalis-flavescens clade?

In Chapter 2, I perform an extensive study of song variation in Pacific-slope and Cordilleran Flycatchers to determine its effects on hybridization and gene flow. Suboscines offer interesting opportunities to investigate the effects of song divergence on lineage diversification, because songs develop without learning. When songs diverge between populations, they can create behavioral barriers to gene flow. Divergence in innate song in suboscine passerines could result in particularly strong behavioral barriers to gene flow because song type is more closely correlated with genotype, and thus a more direct marker of lineage affiliation than in passerine species with learned song. In this chapter, I demonstrate high levels of introgression in both mitochondrial and nuclear genetic markers, although the pattern of introgression is asymmetrical, with introgression limited into core Pacific-slope populations. Moreover, I highlight extensive geographic discordance between the frequencies of mitochondrial and nuclear markers. I demonstrate that the songs of the two taxa are distinct, and highly correlated with nuclear genotype, and that the songs of admixed individuals exhibit spectral characteristics intermediate to the parental species. Song playback experiments demonstrate that both species show some level of discrimination based on song, and highlight lineage-specific behavioral differences that have likely affected the outcome of secondary contact. Pacific-slope Flycatchers seem to rely more on song in territorial interactions, and may discriminate more among song types. Cordilleran Flycatchers exhibit higher levels of aggressiveness in response to playback. Based on the pattern of geographic variation in genetic and song characters, and on the results of the playback experiments, I propose a historical scenario of secondary contact in which asymmetrical introgression of nuclear alleles was facilitated by the social dominance of more aggressive Cordilleran Flycatchers. Finally, I predict that Cordilleran populations will become increasingly introgressed, while introgression into Pacific-slope populations may be limited by a combination of ecological and behavioral factors.

In Chapter 3, I extend the examination of genetic and song variation to include the six principal taxa that comprise the Empidonax difficilis-occidentalis-flavescens clade, to examine whether song divergence has affected lineage divergence by acting as a species discrimination trait. Numerous studies have shown that the level of complexity can affect the efficacy of song in birds, but these studies have focused mainly on birds with learned song. In this study, I offer a novel approach for examining patterns of vocal repertoire evolution, by comparing within repertoire complexity (syllable diversity) across homologous vocalization types present in all six taxa. I find varying rates of song divergence across taxa and across latitude. Songs are distinct between some taxa, but not others. Song divergence is not correlated with mtDNA distance, but it is correlated with latitudinal distance between taxa. Song complexity seems to be higher in higher latitude migratory taxa, but this is due the extremely divergent song of one taxon (E. d. difficilis). Moreover, a high level of divergence in one particular vocalization (Song 2) is most responsible for the overall divergence of E. difficilis song. Song playback experiments show varying levels of discrimination among song types, and there is at least preliminary evidence that lower latitude species are able to use more subtle vocal cues in taxon recognition than higher latitude migratory taxa. This study provides a unique view into how vocal repertoires can evolve in birds, and how this relates to lineage diversification. Moreover, this study is consistent with other studies that have found elevated rates of signal diversification and song complexity in higher latitude migratory species.

Thus, innate song does seem to be able to function as a strong isolating mechanism, but this depends on ecological and behavioral contexts. That is, the abbreviated breeding seasons at higher latitudes might drive more extreme reproductive behaviors such as singing and territorial defense compared to lower latitude taxa. This can affect the effectiveness as song as a taxon discrimination trait, by affecting both the rate of evolution of song and the behavioral context in which the song is performed. Morevoer, I found that patterns of song divergence correlate closely to patterns of genetic divergence in the comparison of song divergence and admixture between Pacific-slope and Cordilleran Flycatchers, but that varying rates of song divergence relative to genetic divergence in the broader phylogenetic comparison make this correlation weaker. Song seems capable of a high level of consistency across relatively large genetic distances or a high level of divergence over relatively small genetic distances, indicating that in at least some of these taxa, song divergence has been driven by selection. Claiming that I have evidence that song differences drove lineage diversification would be premature, but song differences do seem to have important roles in maintaining taxon boundaries. Despite the high level of admixture between Pacific-slope and Cordilleran Flycatchers at interior sites, gene flow does not occur at any significant level into core Pacific-slope populations. Multiple lines of evidence suggest that Pacific-slope song has become highly derived, relative to the songs of other taxa in this clade and that it has a greater, or at least more varied, role in reproduction. Thus, attributing an important role to song differences in decreasing gene flow from Cordilleran populations into core Pacific-slope populations seems reasonable. Moreover, the high level of discrimination in lower latitude taxa among very similar song types suggests that song could be an effective cue for assortative mating in these taxa. This comes with the caveat that the context that often exists in secondary contact zones, such as low population density, can promote hybridization in taxa that would likely mate assortatively if the cost of mate searching were lower.

The Eastern Arc Mountains of Africa have become one of the focal systems with which to explore the patterns and mechanisms of diversification among montane species and populations. One unresolved question is the extent to which populations inhabiting montane forest interact with those of adjacent lowland forest abutting the coast of eastern Africa. The Tiny Greenbul (Phyllastephus debilis) represents the only described bird species within the Eastern Arc/coastal forest mosaic, which is polytypic across an altitudinal gradient: the subspecies albigula (green head) is distributed in the montane Usambara and Nguru Mountains whereas the subspecies rabai (grey head) is found in Tanzanian lowland and foothill forest. Using a combination of morphological and genetic data, we aim to establish if the pattern of morphological differentiation in the Tiny Greenbul (Phyllastrephus debilis) is the result of disruptive selection along an altitudinal gradient or a consequence of secondary contact following population expansion of two differentiated lineages.

Results:

We found significant biometric differences between the lowland (rabai) and montane (albigula) populations in Tanzania. The differences in shape are coupled with discrete differences in the coloration of the underparts. Using multi-locus data gathered from 124 individuals, we show that lowland and montane birds form two distinct genetic lineages. The divergence between the two forms occurred between 2.4 and 3.1 Myrs ago.

Our coalescent analyses suggest that limited gene flow, mostly from the subspecies rabai to albigula, is taking place at three mid-altitude localities, where lowland and montane rainforest directly abut. The extent of this introgression appears to be limited and is likely a consequence of the recent expansion of rabai further inland.

Conclusion:

The clear altitudinal segregation in morphology found within the Tiny Greenbul is the result of secondary contact of two highly differentiated lineages rather than disruptive selection in plumage pattern across an altitudinal gradient. Based on our results, we recommend albigula be elevated to species rank.

Social trait evolution can play a critical role in diversification and speciation. In birds, and especially oscine songbirds, the lability, species-specificity, and behavioral functions of song suggest that song divergence may lead or sustain the speciation process. This series of studies focuses on the evolution of song and its consequences in the Eastern Afromontane sky island sunbird species complex, which comprises the Eastern Double-collared Sunbird Nectarinia mediocris and its close relatives.

In Chapter 1, I examine the structure of one vocalization type, here termed perched song, for the members of the species complex. I show that song divergence matches molecular phylogenetic reconstructions for six major lineages of the species complex, and that song divergence is multi-dimensional. I use both unsupervised and supervised grouping techniques to assess the distinctness of song phenotypes as measured, with similar results. Model selection after unsupervised clustering analysis suggests that six clusters exist, and these correspond to a priori species distinctions at 92.8%. Model-based discriminant function analysis also correctly identifies songs to the six species designated a priori at 92.8%.

The discovery and characterization of the contact zone between Nectarinia moreaui and N. fuelleborni are detailed in Chapter 2. Molecular analyses reveal that there is hybridization, with little or no introgression, between the two species where they meet. The absence of an apparent ecological gradient closely corresponding with the abrupt spatial replacement of one species with the other suggested that interactions between the two species play a strong role in limiting each others' distribution. To test whether bioclimatic niche divergence is evident between species, I used ecological niche models developed from climate variables associated with species occurrence points. I show that the niches of the two species as characterized are conserved relative to a null distribution as generated by sampling the space within which each species' distribution is embedded. This test supports the hypothesis that the presence of each species inhibits the expansion of the others' distribution. Further, I characterize individual genotypes and song and bill length phenotypes sampled in a transect across the contact zone to estimate and compare the width of character clines. Cline analyses show that the genotype and song phenotype clines are roughly coincident and concordant, and very narrow, with the estimated cline width for these two characters ~1.5 km. The bill length cline is offset from the other two clines, and wider. I discuss the many processes that could contribute to the stability and narrowness of the genotype and song clines, including conspecific clumping.

In Chapter 3, I present the results of two playback experiments performed to examine male territorial responses to divergent signals in the parapatrically distributed N. moreaui and N. fuelleborni. These two species have diverged greatly in song phenotype, and subtly in morphometric and plumage traits. The first of the two experiments examines responses to multimodal signaling by assessing reactions to mount/song playback combinations, which are crossed for the two species. This experiment resulted in asymmetric heterospecific response between the two species, with N. moreaui responding consistent with predictions based on the species recognition hypothesis, and N. fuelleborni failing to respond more frequently to its own song than to N. moreaui song. Duration of response by treatment did not correspond to expectations based on species recognition, as there was a trend for individuals to respond longer to heterospecific mounts. In the second experiment, I investigated geographic variation in territorial response to the sibling species compared to own species' song and an ecologically relevant control stimulus, in the form of the song of a sunbird that co-occurs with both species for all tested populations. Responses to the control species were infrequent. Geographic variation in response differences to heterospecific (sibling taxon) and conspecific songs occurs. Strong responses to heterospecifics were not limited to the area of contact in N. fuelleborni, suggesting a role for stochastic processes in the evolutionary loss of response to the sibling taxon's song.

To conclude, song divergence accompanies molecular divergence in the Eastern Afromontane sky island sunbird species complex. Its role in the speciation process might be quite complex, and not as simple as the correlated evolution of female preference with male phenotype as in a Fisherian runaway process. I suggest that one role of song divergence during allopatric or parapatric speciation may be to reduce co-occurrence of individuals with dissimilar song phenotypes through anisotropic dispersal.

Phenotypic variation within species is an important area of evolutionary biology, for it is this variation which the process of selection directly acts upon. However, the biological factors that contribute to the generation and maintenance of this variability are still not well understood. Part of the ambiguity arises from the occurrence of multiple layers of natural and social selection factors which may be occurring simultaneously, and at different hierarchical levels within and between populations. How these factors interact with each other, and at what level, can generate highly variable traits in which the frequency of diversification differs from the individual to the species level. In addition, in conjunction with selective factors, stochastic genetic variation can impact the evolution and maintenance of phenotypic traits in ways that we do not fully understand, dependent on features such as population size, mutation rate, and the dispersal ability of individuals.

Plumage polymorphism, where multiple forms of a trait are maintained within a species, is a widespread occurrence in many families of birds, but is especially common in birds of prey. The high prevalence of variable plumage color and pattern makes raptors an excellent system in which to study the factors that may contribute to high phenotypic diversity in a species. A large number of polymorphic birds of prey have been identified and examined, but evidence for how multiple selection and stochastic factors interact to maintain these traits has not been well-studied. To explore this question, my research focused on the occurrence and overlap of natural and social selection factors, and how they may interact with genetic stochasticity, in the American Kestrel (Falco sparverius). The American Kestrel is an ideal subject for this study, for it represents one of the most dichromatic raptors with a range of plumage variation found across the New World. In order to investigate the evolutionary factors that originate and maintain this range of plumage variation, I integrated behavioral data collected within California, with museum studies, demographic data, and an analysis of the genetic structure and clinal variation of plumage from across North America.

The American Kestrel has been known to have highly variable plumage, but little work has been conducted to examine specific plumage traits in detail. Part of the difficulty in performing this task lies in the extensive range covered by the species. Found from the highest latitudes of North America to the southern tip of South America, the American Kestrel is almost continuous in its distribution across the New World. In my first chapter I use a combination of museum specimens from multiple collections to examine the full range of the variation found in one highly variable plumage trait of male American Kestrels, the black and white coloration on the tail. By doing so I was able to identify three repeatable morphs and four levels of variation in the color of the tail for males. In my first and subsequent chapters I then use this newly defined polymorphism to examine how the variation present in this trait relates to selective and stochastic factors at both the individual and population level.

Selection through social interactions, such as mate choice and competitive interactions, has often been associated with the evolution of plumage traits in birds. Using both a captive population of American Kestrels maintained in Maryland and free-living individuals from study sites in Northern California, my first chapter examines both the heritability and differences in reproductive fitness associated with the polymorphic tail color of male kestrels. I found that the tail morph appears to change little between molt for adults, and both the morph and level of variation are heritable in the captive population. However, there appeared to be little difference in the reproductive fitness of males with different morphs or variation level in the free-living groups. Males with larger amounts of black and white were shown to nest earlier, but this result was only recovered at one of four study sites. Overall, my research detected no consistent sign of selection on the black and white polymorphic tail of American Kestrels through social interaction for breeding birds.

A lack of evidence for the use of the tail color in behavioral interactions does not mean that the trait is not under other forms of selection. A wide-ranging species could encounter a large scope of different microhabitats and ecological factors. Natural selection, such as specialization in habitat type or differing fitness under divergent climatic variables, could potentially drive the maintenance of multiple morphs if different types were specialized for different ecological areas. Such a pattern is often thought to result in clinal distribution of forms across large areas. In my fourth chapter, I used plumage data from 633 male kestrels collected during the breeding season from across North America to construct a map of the frequencies of the three morphs and four variation levels for the continent. A geographic cline was discovered, where birds along the eastern Atlantic Coast showed higher frequencies of both the pattern and level of variation with the greatest degree of black and white. Extending to the west and ending at the Pacific Coast, birds showed a much higher frequency of morphs and patterns with smaller amounts of the two colors. The geographic division in morph frequencies suggested that broad scale climatic or ecological variables might play a role in the distribution of American Kestrels with different tail morphs. However, no correlation was found between a range of bioclimatic variables and tail color. There was also found to be no relationship between morph and variation level, and broad habitat type. Thus, it appears that the highly variable tail pattern of male American Kestrels is not strongly affected by either natural or social selection factors.

The inability to discover any strong relationship between selection factors and the polymorphic tail color of American Kestrels suggests that the trait may be under weak or no selection at this point in time. In such cases, the major influences on the tail color may instead be from stochastic genetic variation, and changes in population size, genetic population structure, and the movement of birds across the landscape. In chapter three, I used demographic and genetic data from a long-term study program of American Kestrels in Saskatchewan, Canada to show that kestrels do experience rapid changes in population size. Findings also suggested that the species appears to be very fluid with its movement into local breeding areas, and is readily affected by changes in land-use and habitat. In chapter four, I turn back to the continent wide data set and compared the different morph frequencies seen with population structure constructed from 375 genetic samples of American Kestrels collected during the breeding season. The genetic structure detected separated out three populations, one in the southeast corresponding with the subspecies F. s. paulus, one along the Atlantic Coast, and a third that appeared to encompass the rest of the continent. This does not correspond exactly with the geographic cline of morph frequencies, but both the genetic and plumage data highlight that the Atlantic Coast is a unique area for the American Kestrel.

Broadly, my dissertation used a combination of research at both the individual and population level to identify a new polymorphic plumage trait in American Kestrels and to determine that the trait appears to exist in different frequencies among populations on the North American continent. While no direct association was found between tail color and natural selection factors at the broad geographical level, or social selection factors at the individual level, the findings do not rule out these areas completely. Overall, the work conducted here suggests that the factors influencing plumage polymorphism in birds of prey are complex, and that stochastic genetic variation may have a larger impact on the maintenance of plumage polymorphism then originally recognized.

We report here the assembly of a northern spotted owl (Strix occidentalis caurina) genome. We generated Illumina paired-end sequence data at 90X coverage using nine libraries with insert lengths ranging from approximately 250 - 9,600 nucleotides and read lengths from 100-375 nucleotides. The genome assembly is comprised of 8,108 scaffolds totaling 1.26 × 109 nucleotides in length with an N50 length of 3.98 × 106 nucleotides. We calculated the genome- wide fixation index (FST) of S. o. caurina with the closely related barred owl (S. varia) as 0.819. We examined nineteen genes that encode proteins with light-dependent functions in our genome assembly as well as in that of the barn owl (Tyto alba). We present genomic evidence for loss of three of these in S. o. caurina and four in T. alba. We suggest that most light-associated gene functions have been maintained in owls and their loss has not proceeded to the same extent as in other dim-light-adapted vertebrates.

We describe here the successful assembly of the complete mitochondrial genomes of the northern spotted owl (Strix occidentalis caurina) and the barred owl (S. varia). We utilized sequence data from two sequencing methodologies, Illumina paired-end sequence data with insert lengths ranging from approximately 250 nucleotides (nt) to 9,600 nt and read lengths from 100-375 nt and Sanger-derived sequences. We employed multiple assemblers and alignment methods to generate the final assemblies. The circular genomes of S. o. caurina and S. varia are comprised of 19,948 nt and 18,975 nt, respectively. Both code for two rRNAs, twenty-two tRNAs, and thirteen polypeptides. They both have duplicated control region sequences with complex repeat structures. We were not able to assemble the control regions solely using Illumina paired-end sequence data. By fully spanning the control regions, Sanger-derived sequences enabled accurate and complete assembly of these mitochondrial genomes. These are the first complete mitochondrial genome sequences of owls (Aves: Strigiformes) possessing duplicated control regions. We searched the nuclear genome of S. o. caurina for copies of mitochondrial genes and found at least nine separate stretches of nuclear copies of gene sequences originating in the mitochondrial genome (Numts). The Numts ranged from 226-19,522 nt in length and included copies of all mitochondrial genes except tRNAPro, ND6, and tRNAGlu. Strix occidentalis caurina and S. varia exhibited an average of 10.74% (8.68% uncorrected p- distance) divergence across the non-tRNA mitochondrial genes.

We analyzed low-coverage, whole-genome sequence data from fifty-one barred owl (Strix varia) and spotted owl (S. occidentalis) individuals to investigate recent introgression between these two species in western North America. Although we obtained genomic confirmation that these species can hybridize and backcross, we found no evidence of widespread introgression. Plumage characteristics of western S. varia that suggested admixture with S. occidentalis appeared unrelated to S. occidentalis ancestry and may instead reflect local selection.

This dissertation aims to understand the patterns and processes involved in the origin, maintenance, and diversification of montane birds in Nuclear Central America. Nuclear Central America, the region between the Isthmus of Tehuantepec and the Nicaraguan Depression, is an understudied tropical, but highly biodiverse region. We assemble genetic, morphological, and climatic datasets to uncover biogeographic and evolutionary processes across a spatial and temporal continuum. We report genomic data from 850 individuals representing ten bird species from the mountains of Nuclear Central America, and using these data we detect several new biogeographic breaks. This dissertation constitutes the first study incorporating genomic data to understand evolutionary processes underpinning bird diversity in Nuclear Central America.

Chapter 1 reveals that historical and contemporary processes have fueled in situ diversification within Nuclear Central America. We study Lampornis viridipallens and L. sybillae integrating phylogeography, landscape genetics, ecological niche modeling, and morphological measurements to understand diversification across the species-population interface. Mitochondrial and microsatellite data indicate that L. viridipallens and L. sybillae diverged in allopatry during the early Pleistocene on either side of the Honduras Depression. This divergence was followed or accompanied by ecological niche divergence. Within L. viridipallens, dry valleys that separate mountain ranges are geographic barriers promoting isolation in this hummingbird species restricted to cloud forests. Moreover, environmental variables are much more critical for facilitating gene flow and population connectivity than straight-line geographic distance.

Chapter 2 highlights the importance of assessing multiple processes to understand the origin and maintenance of biodiversity, especially in topographically complex regions such as the Neotropics. We investigate the phylogeography and genetic basis of phenotype-genotype discordance in Atlapetes albinucha across its geographic range, from Mexico to Colombia. We assemble mitochondrial and genomic data of 73 individuals, including the yellow and gray phenotypes of the species. Our analyses show that allopatric divergence results in high geographic genetic structure. Additionally, we included two individuals with intermediate phenotypes. Ours is the first report of hybridization between the two plumage color phenotypes. We describe the environment where the barrier between the yellow and gray phenotypes occurs (biogeographic barrier east to Altos de Chiapas). We find that high precipitation and low elevation limit the geographic range of both phenotypes and provide a different environment in the hybrid zone. Furthermore, our genomic data suggests UDP-glucuronosyltransferase as a candidate gene explaining the presence-absence of yellow color in the ventral feathers of A. albinucha.

Chapter 3 investigates that the evolutionary and biogeographic history of Saucerottia hummingbirds and sets the biological stage for understanding the genetic basis of structural color variation in hummingbird plumage. We integrate mitochondrial and genomic data of 128 individuals, including representation from Mexico to Guyana, to understand phylogenetic relationships and the geographic distribution of genetic variation. Additionally, we examine and describe the physical structure of rufous, blue, and intermediately colored tail feathers. Our results demonstrate that shallow divergence and mixed ancestry in Saucerottia provide similar genetic backgrounds across species. Evolutionary labile tail coloration occurs across multiple species pairs differing mainly in structural coloration. Geographically detailed sampling within Nuclear Central America uncovers divergence consistent with the geographic distribution of tail color variants. We also identify various candidate genes related to the melanin biosynthesis pathway that may underlie variation in structural coloration.

Chapter 4 reports that the current mountain bird assemblage in Nuclear Central America results from common biogeographic history and species idiosyncratic evolutionary paths. Through a comparative phylogeographic approach, we examine mitochondrial and genomic data of 676 individuals comprising the bird species: Lampornis viridipallens, L. sybillae, Henicorhina leucophrys, Basileuterus belli, Myioborus miniatus, Chlorospingus flavopectus, and Arremon brunneinucha. Phylogenetic analyses support the role of Nuclear Central America as the interconnection between the avian fauna of North and South America. Additionally, high levels of population genetic structure among different mountain ranges within Nuclear Central America suggest in situ diversification. We show that the isolating effect of the Motagua-Polochic-Jocotán Fault System is not homogenous along its extent and that the biogeographic barrier east to Altos de Chiapas is essential to isolating montane populations from one another.

Microbes occur in extraordinary abundance and diversity in vertebrates, and a growing body of work, particularly on humans and mice, has highlighted the pervasive role that microbiota may play in host physiology and health. Yet factors underlying the assembly of microbiota and the potential for microbiota to mediate host fitness in free-living vertebrate populations remain understudied. Investigating host-microbial interactions in natural systems is necessary to understand the extent to which ecologically relevant levels of environmental heterogeneity shape host-microbial relationships and to gain insight into the role that microbiota may play in the ecology and evolution of non-model host taxa. The neonatal vertebrate life stage is of particular interest during which substantial physiological development occurs and through which the initial microbial colonization can have longlasting effects on metabolic programming. To better understand mechanisms underlying microbial variation and associations with host condition during early periods of vertebrate development, this study characterized the gut microbiota of nestlings and their mothers in a natural population of Western Bluebirds Sialia mexicana. A novel method based on solid phase reversible immobilization beads was developed to extract metagenomic DNA from minimally invasive avian fecal and swab specimens. Illumina sequence analysis of bacterial 16S rRNA amplicons was performed on oral, cloacal, and fecal samples collected longitudinally from nestlings throughout the nestling period across multiple nests and over three consecutive breeding seasons (2011-2013). Environmental selection appeared to drive variation in microbiota across gut sites within individuals as well as between nests among individuals. Ambient temperature and sampling year correlated substantially with variation in the gut microbiota. In contrast, surprisingly no association was recovered between host age and alpha diversity, and increased similarity of microbiota across gut sites with host age was observed. Also unexpectedly, shared nest environment and geographical distance were weak predictors of microbial similarity among individuals. Variation in the gut microbiota significantly associated with morphological metrics of nestling growth, even after accounting for the influence of several temporal and environmental variables known to mediate nestling development. However, gut microbiota comprised a less robust predictor of nestling morphology than host age, sampling year, and ambient temperature. Furthermore, perturbation in the form of a viral immune challenge caused measurable shifts in the bacterial gut microbiota, which has previously been undocumented, but no measured effect of the immune challenge was observed on morphological metrics. These data inform intriguing hypotheses for future research on the drivers of avian host-microbial associations and the extent to which microbiota may impact host growth in natural systems. Discerning how microbiota may generate phenotypic or fitness differences among vertebrate host organisms is essential to understanding the evolution and maintenance of biodiversity.