Copyright 2014 Maria Stager

   

MOLECULAR MECHANISMS UNDERLYING AVIAN METABOLIC VARIATION: A GENOMIC APPROACH

BY

MARIA STAGER

THESIS

Submitted in partial fulfillment of the requirements for the degree of Master of Science in Biology

with a concentration in Ecology, Ethology, and Evolution in the Graduate College of the

University of Illinois at Urbana-Champaign, 2014

Urbana, Illinois

Master’s Committee:

Assistant Professor Zachary A. Cheviron, Director of Research Professor Jeffrey D. Brawn Associate Professor Cory D. Suski Professor David L. Swanson, University of South Dakota

  ii    

ABSTRACT

Physiological adaptive acclimatization allows an organism to balance the competing

demands of its life cycle and environment. This thesis seeks to further our understanding of the

processes underlying physiological acclimation by characterizing two molecular mechanisms

that enable organisms to mount physiological responses to changing environmental conditions.

Comprised of two investigations, it focuses on the molecular mechanisms underlying variation in

avian aerobic performance. In the first, I explore regulatory changes that act on relatively short

time periods, enabling organisms to respond to rapid environmental changes within an

individual’s lifetime. I show that thermogenic flexibility manifested from simultaneous changes

in the expression of genes within hierarchical regulatory networks in response to temperature in

the Dark-eyed Junco (Junco hyemalis). This transcriptomic variation resulted in rapid

physiological modifications within individuals. In the second, I instead characterize coding

mutations that occur on the timescale of generations. I show that differences in the mitochondrial

sequences of the genes encoding the machinery of cellular metabolism could result in canalized

differences in whole-organism aerobic performance among Tachycineta swallow species. Sites

of positive selection were concentrated in species with ‘slow’ life histories that may be

associated with low metabolic rates. Together these investigations provide a concise survey of

the molecular tools birds are equipped with to metabolically respond and adapt to environmental

variation.

  iii    

TABLE OF CONTENTS CHAPTER 1: Introduction……....……………………………………………………………………. 1 CHAPTER 2: Regulatory mechanisms of metabolic flexibility in the

Dark-eyed Junco (Junco hyemalis)..…………………………………………………….. 4

Tables & Figures………………………………………………………………………...36 CHAPTER 3: Signatures of natural selection in the mitochondrial genomes of Tachycineta swallows and their implications for latitudinal patterns of the ‘pace of life’……………45 Tables & Figures.………………………………………………………………………. 64 CHAPTER 4: Conclusion….....…………………………………………………………………71 APPENDIX A.…………………………....…………...…...……………………………………73 APPENDIX B.…………………………....…………...…...…………………………………….74 APPENDIX C.…………………………....…………...…...…………………………………….75 APPENDIX D.…………………………....…………...…...…………………………………….76 APPENDIX E.…………………………....…………...…...…………………………………….77 APPENDIX F.…………………………....…………...…...…………………………………….78 APPENDIX G.…………………………....…………...…...……………………………………79 APPENDIX H.…………………………....…………...…...……………………………………80 APPENDIX I..…………………………....…………...…...…………………………………….81

  1    

CHAPTER 1

INTRODUCTION

Physiological adaptive acclimatization allows an organism to balance the competing

demands of its life cycle and environment. Variation in physiological responses can therefore

have dramatic fitness consequences and understanding physiological adaptation is thus critical

for understanding large-scale evolutionary processes. This thesis seeks to further our

understanding of the processes underlying physiological adaptation by characterizing the

molecular mechanisms that enable organisms to mount physiological responses to changing

environmental conditions. Specifically, I focus here on a physiological activity fundamental to

all eukaryotic life: aerobic metabolism.

Rates of energy expenditure vary tremendously among organisms— up to 200-fold

within a single class (e.g., Cephalopods; Seibel, 2007) — yet the underlying metabolic pathways

are highly conserved. The metabolic network that enables a sessile coral reef sponge to consume

as little as 0.02 μmol O2 g-1 min-1 (Hadas et al., 2008) is essentially the same as that which

enables a hummingbird to sustain hovering flight at a rate of 82 μmol O2 g-1 min-1 (Suarez, 1992).

These pathways are well characterized and, while we often think of metabolism in terms of

whole-organism performance, at lower levels, cellular respiration can be decomposed into a

series of enzyme-catalyzed reactions. Each component part feeds one into the next in a step-like

fashion that enables the conversion of biological molecules (carbohydrates, fats, and proteins)

into usable energy (ATP). Together, these vital reactions sum to form an organism’s metabolic

rate, a universal measure of energy exchange.

Metabolic performance varies both within and among individuals and broad-scale

patterns in metabolic performance exist (e.g., across seasons, elevations, and latitudes). Variation

within a single individual might represent an adaptive acclimatization response to short-term

environmental pressures. For instance, small endotherms exhibit seasonal changes in aerobic

capacity to meet the thermogenic demands of winter (e.g., Swanson, 2010). Variation among

populations may arise from differences in the abiotic selective pressures of their respective

environments, and could be indicative of adaptive evolution. For instance, basal (or resting)

metabolic rate has been shown to increase with increasing latitude for many groups of

  2    

vertebrates (e.g., birds, Wiersma et al., 2007a,b; fish, Naya & Bozinovic, 2012; mammals,

MacMillen & Garland, 1989; Lovegrove, 2000) and has been correlated with rates of

reproduction and survival (Wiersma et al., 2007a,b). Thus investigations of the molecular basis

of metabolic variation both within and among individuals can allow us to explore the underlying

mechanisms that enable organisms to respond differentially to changes in their environments.

At the biochemical level, metabolic variation can manifest from differences in the

regulatory mechanisms that govern aerobic respiration (e.g., Cheviron et al., 2012; 2014).

Chapter 2 investigates the molecular basis of metabolic flexibility in response to seasonally

changing environmental stimuli. We exposed Dark-eyed Juncos (Junco hyemalis) to two,

experimentally manipulated seasonal cues: photoperiod and temperature. We then tested for

associations between changes in gene expression profiles of skeletal muscle and the activities of

key enzymes, as well as whole-organism thermogenic performance. This integrative,

manipulative experiment allowed us to make explicit connections from each environmental

stimulus, to the transcriptomic and proteomic responses, to the resulting changes in organismal

aerobic capacity. In contrast, coding mutations in the genes encoding the machinery of cellular metabolism

can alter the functional properties of metabolic enzymes and result in canalized differences in

whole-organism metabolic performance. Genomic comparisons among closely related

populations or species are particularly useful for characterizing these genetic differences. To this

end, Chapter 3 explores the genetic basis of metabolic variation among Tachycineta swallows

within a phylogenetic framework. Avian life histories consistently vary with latitude, and many

of these differences can be attributed to underlying differences in rates of metabolic energy

expenditure. We therefore inferred differences in metabolic demands among Tachycineta species

using an index of life-history variation, and then tested for evidence of adaptive mitochondrial

divergence using complete mitochondrial genome sequences and suites of comparative genomic

analyses. Using this approach, we uncovered a number of amino acid substitutions that bear the

signature of positive selection across the mitogenome, and we make inferences regarding the

functional significance of this genetic variation on whole-organism metabolic performance.

These investigations serve to illuminate the molecular mechanisms underlying

physiological variation in non-model organisms. Separately, these chapters describe two distinct

processes underlying different levels of metabolic variation acting at disparate scales.

  3    

Collectively, they provide a concise survey of the molecular tools organisms are equipped with

to metabolically respond and adapt to environmental variation.

LITERATURE CITED

Cheviron, Z.A., Bachman, G.C., Connaty, A.D., McClelland, G.B., Storz, J.F., 2012. Regulatory

changes contribute to the adaptive enhancement of thermogenic capacity in high-altitude

deer mice. Proc. Natl. Acad. Sci. USA 109, 8635–8640.

Cheviron, Z.A., Connaty, A.D., McClelland, G.B., Storz, J.F., 2014. Functional genomics of

adaptation to hypoxic cold-stress in high-altitude deer mice: transcriptomic plasticity and

thermogenic performance. Evolution 68, 48–62.

Hadas, E., Ilan, M., Shpigel, M., 2008. Oxygen consumption by a coral reef sponge. J. Exp. Biol.

211: 2185-2190.

Lovegrove, B. G., 2000. Daily heterothermy in mammals: coping with unpredictable

environments. Pages: 29–40 in G. Heldmaier and M. Kinglenspor, eds. Life in the cold:

Eleventh International Hibernation Symposium. Springer, Berlin.

MacMillen, R. E., Garland Jr., T., 1989. Adaptive physiology. Pages 143–168 in G. L. Kirkland

and J. N. Layne, eds. Advances in the study of Peromyscus (Rodentia). Texas Tech

University Press, Lubbock.

Naya, D.E., Bozinovic, F., 2012. Metabolic scope of fish species increases with distributional

range. Evol. Ecol. Res. 14, 769-777.

Seibel, B. A., 2007. On the depth and scale of metabolic rate variation: scaling of oxygen

consumption and enzymatic activity in the Class Cephalopoda (Mollusca). J. Exp. Biol.

210, 1-11.

Swanson, D.L., 2010. Seasonal metabolic variation in birds: functional and mechanistic

correlates. Curr. Ornithol. 17, 75-129.

Suarez, R. K., 1992. Hummingbird flight: sustaining the highest mass-specific metabolic

rates among vertebrates. Experientia 48, 565-570.

Wiersma, P., Chappell, M., Williams, J.B., 2007a. Cold- and exercise-induced peak

metabolic rates in tropical birds. Proc. Natl. Acad. Sci. Biol. 104, 20866–20871.

Wiersma, P., Munoz-Garcia, A., Walker, A., Williams, J.B., 2007b. Tropical birds have

a slow pace of life. Proc. Natl. Acad. Sci. Biol. 104, 9340–9345.

  4    

CHAPTER 2

REGULATORY MECHANISMS OF METABOLIC FLEXIBILITY IN THE

DARK-EYED JUNCO (JUNCO HYEMALIS)

Maria Stager1, David L. Swanson2, and Zachary A. Cheviron1

1Department of Animal Biology, University of Illinois at Urbana-Champaign, Urbana, IL 61801 2Department of Biology, University of South Dakota, Vermillion, SD 57069

ABSTRACT

Small temperate birds can reversibly modify their aerobic performance to maintain

thermoregulatory homeostasis amidst winter conditions, and these physiological adjustments

may be attributable to changes in the expression of genes in the underlying regulatory networks.

Here we report the results of an experimental procedure designed to gain insight into the

fundamental mechanisms of metabolic flexibility in the Dark-eyed Junco (Junco hyemalis). We

combined genomic transcriptional profiles with measures of metabolic enzyme activities and

whole-animal thermogenic performance from juncos exposed to four experimental treatments

varying in exposure to temperature (cold=3°C; warm=24°C) and photoperiod (SD=8h light:16h

dark; LD=16h light:8h dark). Cold-acclimated birds increased thermogenic capacity compared to

warm-acclimated birds, and this enhanced performance was associated with upregulation of

genes involved in lipid transport and oxidation, and with catabolic enzyme activities. These

physiological changes occurred over ecologically relevant timescales, suggesting that birds make

regulatory adjustments to interacting, hierarchical networks in order to seasonally enhance

thermogenic capacity.

INTRODUCTION

Phenotypic flexibility is an organism’s ability to reversibly modify its phenotype with

respect to changes in its environment. Because it allows an organism to optimally match its

phenotype to prevailing biotic and abiotic conditions, phenotypic flexibility has profound

consequences for organismal fitness (Piersma & Drent, 2003; DeWitt & Scheiner, 2004).

  5    

Elucidating the mechanisms that underlie phenotypic flexibility therefore has important

implications for understanding how organisms cope with rapid environmental change.

Temporal variation in organismal bioenergetics is a prominent example of phenotypic

flexibility. Endotherms reversibly alter their metabolic rates across the annual cycle to meet

seasonal demands associated with migration, reproduction, and thermoregulation. Such

metabolic flexibility is especially important for small endotherms that are resident in seasonally

variable environments, as metabolic thermoregulatory performance has acute consequences for

survival during prolonged cold stress (Hayes & O’Conner, 1999). Birds that winter at high

latitudes can provide a valuable model for understanding the physiological mechanisms

underlying metabolic flexibility because they do not hibernate, and therefore must rely largely on

physiological adjustments that enhance thermogenic performance for survival. For instance, in

winter, temperate-zone birds elevate their capacities for shivering thermogenesis to maintain

thermoregulatory homeostasis in the face of falling temperatures, and this increased thermogenic

performance is associated with elevated basal and summit metabolic rates (e.g., Arens & Cooper,

2005). However, winter is also accompanied by reductions in productivity, decreased foraging

time, and increased fasting periods overnight (Marsh & Dawson, 1989), so birds must potentially

meet these increased thermoregulatory demands in the face of reduced energetic input (Swanson,

2010). Thus, short-term physiological adjustments that optimally balance competing energetic

demands are crucial for survival during the harsh winters in high-latitude environments.

At the biochemical level, variation in aerobic thermogenic performance is manifest

through differences in the expression and functional properties of the genes encoding the

machinery of cellular metabolism. For instance, high-elevation deer mice (Peromyscus

maniculatus) exhibit greater thermogenic capacities compared to their lowland counterparts, and

this enhanced performance has been linked to regulatory changes in the expression of genes that

influence metabolic fuel use and oxygen utilization (Cheviron et al., 2012; 2014). Because

changes in gene regulation allow the genome to rapidly and reversibly respond to environmental

variation, transcriptomic studies can provide key insights into the mechanistic underpinnings of

phenotypic flexibility and the adaptive modification of complex physiological traits, both of

which require concerted changes in hierarchical and interacting regulatory networks (Cheviron &

Brumfield, 2012). The hierarchical nature of complex physiological traits is well illustrated by

  6    

physiological adjustments that enhance aerobic thermogenic performance, which comprise three

broad levels of physiological organization (Figure 2.1):

(1) Enhancing the size or structure of thermogenic organs. Small birds rely on flight muscles for

facultative thermogenesis (Hohtola, 1982; Hohtola et al., 1998; Dawson & O’Connor, 1996;

Yacoe & Dawson, 1983), and several studies have demonstrated a prominent role for avian

pectoralis size on thermogenic capacity (e.g., Vézina et al., 2006; Vézina et al., 2007; Swanson

et al., 2013). Muscle hypertrophy can be achieved by altering the expression of genes in several

key growth signaling pathways. Specifically, the rapamycin (mTOR) signaling pathway is

recognized as the major player in protein synthesis regulation for adult mammalian skeletal

muscle (McCarthy & Esser, 2010). The mTOR pathway integrates signals from a number of

upstream pathways, including the insulin-like growth factor 1 (IGF1)-Akt signaling pathway

(also called PIK3-Akt signaling pathway), which plays a principal role in stimulating cell growth

and proliferation. Increased expression of the first component in this pathway (IGF1) has been

linked to muscle hypertrophy in birds and mammals (e.g., Rennie et al., 2004; Heinemeier et al.,

2007; Price et al., 2011). Inhibitory growth factors that interact with the IGF1-Akt3 and mTOR

pathways (e.g., myostatin) impede protein synthesis and induce muscle atrophy (Welle et al.,

2009). Blocking myostatin activity can therefore result in subsequent growth to myocytes (e.g.,

McPherron et al., 1997). In principle, modifications could also be made to the relative proportion

of fiber types within the muscle. This has been shown in mammals, where endurance training is

associated with the alteration of relative proportions of fast glycolytic fibers and fast oxidative-

glycolytic fibers in skeletal muscle (Holloszy & Coyle, 1984; Hudlicka, 1985); however, this has

not been documented in small birds. Instead, numerous studies have shown that the pectoralis of

small birds is composed almost exclusively of fast oxidative-glycolytic fibers, providing both an

intermediate contraction velocity and resistance to fatigue that allows for sustained flapping

flight (e.g., Lundgren & Kiessling, 1988; Welch & Altshuler, 2009). Therefore, while it is

possible that the avian pectoralis does grow and atrophy seasonally to meet changing

thermogenic demands, the fiber composition does not seem to be altered to any appreciable

degree.

  7    

(2) Enhancing oxygen delivery. Increased oxygen delivery can be accomplished by a number of

cardiovascular changes throughout the body. Within the pectoralis, oxygen diffusion distance

can be reduced through increased capillary density in the muscle, as has been shown in altitude-

acclimated pigeons (Mathieu-Costello & Agey, 1997). Angiogenesis is primarily stimulated by

the vascular endothelial growth factor (VEGF) signaling pathway, which mediates endothelial

cell proliferation and migration (Klagsbrun & D’Amore, 1996; Ferrara, 1997). The VEGF

pathway in turn can be activated by cellular hypoxia via the hypoxia-inducible factor 1(HIF-1)

signaling cascade (Bruick & McKnight, 2001). In addition to initiating angiogenesis, the HIF-1

cascade regulates oxygen homeostasis in a number of other ways, including eliciting

erythropoiesis via hormonal control (Rankin et al., 2007) and inducing vasodilation via nitric

oxide production (Beall et al., 2012). These many interacting mechanisms can serve to boost

oxygen delivery to the muscle, enabling individuals to sustain increased aerobic performance.

(3) Enhancing the cellular aerobic capacity of thermogenic organs. Increases in fuel supplies

and oxygen utilization can increase the cellular aerobic capacity of thermogenic organs like the

pectoralis. Although mammals shift between carbohydrate and lipid fuel supplies according to

exercise intensity (McClelland, 2004), energy-rich lipids are the primary fuel source for avian

skeletal muscle for all intensity levels — from rest to the prolonged exercise associated with

shivering and flight (Rothe et al., 1987; Suarez et al., 1990; Vaillancourt et al., 2005; Guglielmo,

2010; Kuzmiak-Glancy & Willis, 2014). Adipose stores have been shown to contribute 80% of

the energy required for avian shivering (Vaillancourt et al., 2005). Fuel availability is not only

dependent upon the size of adipose stores, but also upon an individual’s ability to rapidly

mobilize, catabolize, and oxidize these fuels. The first step, mobilization, occurs by modifying

expression to the many pathways that comprise the greater fatty acid supply pathway. Fats are

not water-soluble; therefore, carrier proteins are required to move fatty acids from the adipocyte

to the muscle and, ultimately, across the cellular membrane to the mitochondria. To that effect,

birds meet the heightened demands of migration by seasonally increasing fatty acid transporter

proteins in the flight muscle (Guglielmo et al., 1998, Pelsers et al., 1999, Guglielmo et al., 2002,

McFarlan et al., 2009). Moreover, increased metabolic intensity of the pectoralis can be achieved

by increasing mitochondrial density, which has been documented in premigratory birds (Gaunt et

al., 1990; Evans et al, 1992), or by elevating activities of catabolic enzymes involved in fatty

  8    

acid beta-oxidation, which has been documented in both wintering and migratory birds (e.g.,

Lundgren & Kiessling, 1985; Carey, 1989; O’Connor, 1995; Guglielmo et al., 2002). Birds also

exhibit corresponding seasonal changes in activities of oxidative enzymes that are involved in

the final steps of aerobic metabolism — the citric acid cycle and oxidative phosphorylation

pathways (e.g., Liknes, 2005; Liknes & Swanson, 2011; Zheng et al., 2008). Thus adjustments

can be made to fine-tune fuel supplies and oxidative utilization at many points along this

extensive cascade.

Despite this range of possibilities, it is still unclear how organisms effectively modify and

combine these modular and hierarchical trait components in order to rapidly respond to

exogenous factors, such as extreme weather events. Although regulatory adjustments could occur

at any step in a given pathway, theory predicts that flux control should be unevenly distributed

among pathway components (Wright & Rausher, 2010). Specifically, it is the upstream steps of

linear pathways that are expected to be most rate limiting (Wright & Rausher, 2010; Eanes,

2011). Indeed in Drosophila species, genes at the top of the glycolytic pathway exhibit

signatures of selection in the form of latitudinal allele clines and elevated rates of amino acid

substitution, suggesting that these positions have been targeted for increased pathway flux

(Sezgin et al., 2004; Flowers et al., 2007; Eanes, 2011). If flux control is also concentrated in the

upstream portions of pathways influencing thermogenic performance, we would expect to see

regulatory changes in the upstream components (e.g., the upper tier of the fatty acid supply

pathway). This hypothesis can be tested using transcriptomic pathway-level approaches that

allow for simultaneous analyses of many interacting regulatory networks to identify regional

patterns in regulatory change.

Here we report the results of an experimental procedure designed to gain insight into the

fundamental mechanisms of avian metabolic flexibility. We combine new data on genomic

transcriptional profiles with previously published data on thermogenic performance and enzyme

activities for the same individuals (Swanson et al., 2014). This work focuses on Dark-eyed

Juncos (Junco hyemalis), a small passerine that winters from Alaska to Arizona (Nolan et al.,

2002). Wintering juncos exhibit increases in thermogenic performance and cold tolerance

compared to summer-acclimatized birds (Swanson, 1991; 2001). To induce changes in junco

thermogenic performance, we exposed wild-caught juncos to synthetic photoperiod and

  9    

temperature regimes, two seasonal cues that have been previously implicated in altering avian

metabolic rate and thermogenic performance (i.e. Cooper & Swanson, 1994; Liknes & Swanson,

1996; Zajac et al., 2011). We then related variation in thermogenic performance to differences in

underlying gene expression profiles of a plastic, thermogenically-active tissue, the pectoralis. We

explicitly surveyed the genes associated with the cell-, tissue-, and organ-level responses

outlined above to test a priori hypotheses regarding the changes likely to occur at each of the

three levels of biological organization. By concurrently examining the effects of photoperiod and

temperature, we address how each environmental stimulus induces changes in metabolic gene

expression and how these transcriptomic responses, in turn, affect whole-organism thermogenic

performance.

METHODS Bird capture and acclimation treatments

We subjected juncos to a two-by-two experimental design using two relevant

environmental cues. Details of the sampling methods have been previously described by

Swanson et al. (2014). Briefly, we captured juncos overwintering near Vermilion, South Dakota

(~42°470 N, 97°W) in mid-December. We housed birds individually and, after a two-week

adjustment period, subjected individuals to one of four experimental temperature-photoperiod

regimes, each lasting six weeks. The treatments were 16 h light:8 h dark at 24°C (warm LD, n =

10), 8 h light:16 h dark at 24°C (warm SD, n = 10), 16 h light:8 h dark at 3°C (cold LD, n = 10),

or 8 h light:16 h dark at 3°C (cold SD, n = 9). To obtain appropriate sample sizes, procedures

were performed over the course of two consecutive winter seasons (2011-2012 and 2012-2013).

Measures of metabolic performance

We measured individual thermogenic performance before and after the treatments (at 6

weeks). We quantified summit metabolic rate (Msum), which is the maximum metabolic rate

attained by birds under cold exposure (Swanson, 2010) and is a proxy for thermogenic capacity,

using open-flow respirometry in a heliox atmosphere (21% oxygen/79% helium) with sliding

cold exposure (as per Swanson et al., 1996). Following the Msum trial, we measured cloacal

temperature to verify that individuals were hypothermic (body temperature ≤ 37°C). We

identified Msum as the period of peak oxygen consumption over a 5-minute window and report it

  10    

as ml O2 min-1. To control for existing differences in individual Msum before acclimation

treatments, we calculated the change in individual Msum before and after treatments (ΔMsum) for

use in subsequent analyses. We tested for differences among treatment groups after the

acclimation period using a one-way ANOVA on ΔMsum.

At the end of the acclimation treatment, immediately following the final Msum

measurement, we euthanized juncos via cervical dislocation, measured body mass, and rapidly

excised pectoralis tissue on ice. We weighed pectoralis tissue to the nearest 0.1 mg, then flash

froze it in liquid N2 for use in enzyme activity assays or stored it in RNAlater and froze it at -

60°C for transcriptomic analyses. We collected birds under appropriate state (11-7, 12-2) and

federal (MB758442) scientific collecting permits and all procedures were approved by the

University of South Dakota Institutional Animal Care and Use Committee (Protocol 79-01-11-

14C).

Functional genomic analyses

Generation and sequencing of RNA-seq libraries

We used massively parallel sequencing of pectoralis muscle transcriptomes (RNA-seq;

Wang et al., 2009) to quantify genome-wide patterns of gene expression for each of the 39

individuals. We isolated mRNA from pectoralis using Trizol reagent and prepared cDNA

libraries for sequencing using the Illumina TruSeq RNA Sample Preparation Kit. Libraries were

sequenced as 100nt single-end reads on the Illumina HiSeq2000 platform. We multiplexed 10

individuals in a single flow cell lane using Illumina index primers. Image analysis and base

calling were performed using Illumina pipeline software.

Read mapping, normalization and differential expression

We performed a series of sequential filtering steps to remove low quality reads and index

primer sequences. First, we removed all reads with mean Solexa quality scores less than 30.

Second, we trimmed low-quality bases (Solexa score < 30) from these remaining high-quality

sequences. Finally, we scanned reads for adaptor sequences, and if detected, they were trimmed

from the sequence read. The filtering steps resulted in a final dataset of nearly 743 million

sequence reads, with an average of 19 million reads per individual (range = 3.5 – 63.0 million

reads), and the trimming steps resulted in an average read length of 97.7 nt.

  11    

We estimated transcript abundance by mapping sequence reads to the Zebra Finch

(Taeniopygia guttata) genome (build taeGut3.2.4) using CLC Genomics workbench (v. 6.0.4). In

total, sequence reads mapped to 15,362 unique genes. However, we excluded genes with less

than an average of 5 reads per individual because genes with low count values are typically

subject to increased measurement error (Robinson & Smyth, 2007). This filtering step resulted in

a final dataset of 10,868 detected genes.

We performed normalization and differential expression analyses using the package

edgeR (Robinson et al., 2010) in Program R (ver. 3.0–R Development Core Team, 2013). We

first normalized read counts and controlled for differences in total library size (number of total

reads) among individuals using the function calcNormFactors (Robinson & Oshlack, 2010).

Following normalization, we tested for differences in transcript abundance among treatment

groups using a generalized linear model approach. We estimated model dispersion for each gene

separately using the function estimateTagwiseDisp (McCarthy et al., 2012). We tested for genes

that exhibited significant expression differences across treatments using a GLM likelihood ratio

test and controlled for multiple tests by enforcing a genome-wide false discovery rate (FDR) of

0.05 (Benjamini & Hochberg, 1995).

Connecting gene expression and phenotypic variation

By detecting suites of genes that exhibit correlated transcriptional patterns, previously

undetected modules of ostensibly co-regulated genes can be defined. These modules may expose

biologically meaningful associations between surprising assemblages of genes, and thus their use

in subsequent functional analyses can provide insight into the mechanistic underpinnings of

complex traits (Ayroles et al., 2009; Stone & Ayroles, 2009). To identify coordinated expression

changes among interacting genes, we assessed the degree of correlation in transcript abundance

among genes with significant treatment effects (FDR < 0.05) using Modulated Modularity

Clustering (MMC; Stone & Ayroles, 2009). This method calculates Pearson correlation

coefficients among transcript abundances for all pair-wise combinations then identifies suites

(modules) of highly intercorrelated genes (Stone & Ayroles, 2009).

For each of the modules, we then summarized overall module expression using a

principal component analysis and retained the first principal component axis (PC1), which

represented on average 84% of within module variation (49% - 100%). To determine if suites of

  12    

differentially expressed (DE) genes were related to whole-organism thermogenic performance,

we performed linear regressions of individual ΔMsum on module expression (PC1) and identified

correlations at a significance level of P ≤ 0.05. We repeated this procedure to identify

correlations between pectoralis size (Mp; expressed as a proportion of body mass) and module

expression. For modules that exhibited significant associations with either phenotype, we

performed functional enrichment analyses using the Zebra Finch GO Analysis (Wu & Watson,

2009) to identify functional categories that were over-represented among the genes that comprise

a given trait-associated module.

To isolate the independent contributions of photoperiod and temperature on gene

expression, we followed a procedure identical to that described above for the overall treatment

effects: we tested for genes that exhibited significant expression differences between the two

photoperiod treatments, identified suites of genes with correlated expression patterns,

summarized module expression, tested for associations between module expression (PC1) and

ΔMsum or Mp, and performed enrichment analyses. We repeated this procedure for the

temperature treatments. Overall-treatment-sensitive modules are denoted by A, photoperiod-

sensitive modules by P, and temperature-sensitive modules by T.

Targeted pathway and gene analyses

We performed analyses on targeted genes and pathways to test a priori hypotheses about

their influences on metabolic flexibility. These included all genes in the four major metabolic

pathways (glycolysis, fatty acid oxidation, citric acid cycle, oxidative phosphorylation), and in

muscle growth (mTOR signaling), angiogenesis (vascular endothelial growth factor), and fuel

supply (glycerolipid metabolism, glycerophospholipid metabolism, peroxisome proliferator-

activated receptor (PPAR) signaling, pyruvate metabolism) pathways. We identified these genes

using Taeniopygia guttata KEGG Database pathway maps (Kanehisa &Goto, 2000; Kanehisa et

al., 2014), downloaded from CytoKEGG (ver. 0.0.5). We also included select genes in the

insulin-like growth factor (IGF1) and hypoxia inducible factor 1 (HIF-1) signaling pathways for

which complete T. guttata maps were not available. We identified the presence/absence of these

genes among those differentially expressed (either among photoperiod or temperature

treatments) and noted fold changes with respect to the environmental stimuli where applicable.

  13    

Because phenotypic responses could manifest from large-scale changes in expression

across the entire pathway, we tested for concerted expression across these same pathways in

response to both photoperiod and temperature treatments. To do this we mapped expression

values onto T. guttata pathway maps and counted the number of genes exhibiting positive and

negative log fold-change values. We performed binomial tests to test for concerted expression

across each pathway with the null expectation that an equal number of genes would be up- and

down-regulated in given pathway simply by chance. We identified concerted expression at a

significance level of P ≤ 0.05.

Enzyme quantification and transcript abundance

Variation in transcript abundance does not consistently produce an equivalent or

proportional response in corresponding protein abundance (Abreu et al., 2009). To investigate

the relationship between the junco transcriptome and its corresponding gene products, we

quantified the activities of three metabolic enzymes that serve as biomarkers for the flux capacity

of important metabolic pathways. Enzyme activity is a product of both enzyme concentration and

kinetic efficiency, and consequently is not strictly equivalent to protein abundance. Nonetheless,

because we expect any potential variation in kinetic efficiency to be randomly distributed among

individuals that were assigned to acclimation treatments, we assumed that differences in enzyme

activities largely reflected differences in enzyme concentration, which in turn should be related

to differences in the transcript abundance of enzyme-encoding genes. The enzymes we surveyed

included: carnitine palmitoyl transferase (CPT), an indicator of fatty acid transport capacity

across the mitochondrial membrane; beta-hydroxyacyl Coenzyme-A dehydrogenase (HOAD), an

indicator of fatty acid oxidation capacity; and citrate synthase (CS), an indicator of mitochondrial

abundance and cellular metabolic intensity. We utilized pectoralis tissue to quantify enzymes

using standard assays (e.g., CPT as in Guglielmo et al., 2002; CS and HOAD as in Liknes &

Swanson, 2011) conducted at 39 ± 2°C with a Beckman DU 7400 spectrophotometer. Further

details have been previously described in Swanson et al. (2014). We report all enzyme activities

as mean mass-specific activity (μmoles · min-1 · g-1). To test whether differences in transcript

abundance translated into variation in enzyme activity, we performed linear regressions of

individual metabolic enzyme activity on the corresponding transcript abundances for each of the

  14    

enzyme-encoding genes. We identified associations between enzyme activity and transcript

abundance at a significance level of P ≤ 0.05.

RESULTS Phenotypic flexibility in thermogenic performance, not pectoralis size

We exposed juncos to two synthetic cues (temperature and photoperiod) using a two-by-

two experimental design in order to induce changes in thermogenic performance. As previously

reported (Swanson et al., 2014), these six-week acclimation treatments effectively altered junco

summit metabolic rate (ΔMsum; P = 1.43 x 10-5; Figure 2.2). Cold-acclimated birds increased

Msum by 1.19 ml O2 min-1 compared to warm-acclimated birds (P = 8.63 x 10-7), but the

photoperiod treatments did not result in changes to Msum (ΔMsum = 0.05 ml O2 min-1, P = 0.684).

Mass adjusted pectoralis size (Mp), however, did not vary among any of the acclimation

treatments (one-way ANOVA, Ptotal = 0.263; Ptemp = 0.133; Pphoto = 0.234; Appendix A).

Transcriptomic responses to overall treatment

We used a general linear model approach to identify genes that exhibited a significant

difference in transcript abundance due to acclimation treatment. A total of 1882 genes (17.3% of

the total number of transcripts after filtering) exhibited differential expression (DE) among

acclimation treatments (FDR < 0.05). Within this subset of DE genes, we identified modules of

coregulated genes using MMC (Stones and Ayroles, 2009). This analysis revealed 47

transcriptional modules composed of highly intercorrelated genes (Figure 2.3A). We then

employed principal component and regression analyses to determine which of these

transcriptional modules were associated with thermogenic performance (ΔMsum). Overall module

expression (summarized as PC1 scores) correlated with individual ΔMsum for 13 modules (Table

2.1). Gene enrichment analyses revealed that these modules were significantly enriched for terms

related to lipid metabolism (A18, A26, A27), fatty acid oxidation (A26), oxygen transport (A26),

angiogenesis (A23, A27, A28), and muscle growth (A28, A43) (see Appendix B for full list of

terms enriched in each module). Additionally, module expression (PC1) correlated with mass-

adjusted pectoralis size (Mp) for five modules. Gene enrichment analyses revealed that these

modules were enriched for terms largely related to ion transport (A7, A19).

  15    

Isolating transcriptomic responses to each cue

We used a similar approach to isolate the effects of photoperiod and temperature on junco

transcriptomic responses. A total of 1325 genes (12.2% of total transcripts) were differentially

expressed among the photoperiod treatments and 756 genes (7.0% of total transcripts) were

differentially expressed among the temperature treatments. Of these, 196 DE genes were shared

among the two treatment classes, and these were also detected in the overall treatment analyses

(Figure 2.4). Likewise, many of the DE genes particular to a single treatment class were also

detected in the overall treatment analysis. As a result, 252 novel DE genes were detected among

the photoperiod treatments alone and 92 novel DE genes were detected among the temperature

treatments. We again employed the combination of MMC, principal component analysis, and

regression analysis to determine which photoperiod-specific (or temperature-specific)

transcriptional modules were associated with ΔMsum and Mp.

Photoperiod-sensitive DE genes (those with significant photoperiod effects) clustered

into 51 transcriptional modules (Figure 2.3B). In regression analyses, three transcriptional

modules exhibited strong associations with ΔMsum (Table 2.1). These modules were enriched for

terms related to glycolysis (P36) and angiogenesis (P45; Appendix B). Similarly, five

transcriptional modules showed strong associations with Mp. These modules were enriched for

terms involved in ion transport (P8, P21) and lipid metabolism (P32, P51).

Temperature-sensitive DE genes (those with significant temperature effects) clustered

into 38 transcriptional modules (Figure 2.3C). In regression analyses, 23 transcriptional modules

exhibited strong associations with ΔMsum. These modules were enriched for terms related to lipid

metabolism (n = 5 modules), oxidative phosphorylation (n = 5), glycolysis (n = 2), angiogenesis

and oxygen transport (n = 4), and growth (n = 7; Table 2.2). Additionally, three transcriptional

modules showed strong associations with Mp. These modules was enriched for terms involved in

muscle contraction (T5) and growth and angiogenesis (T38).

Pathway-level expression patterns

Our dataset included the majority of unique genes (84-100%) from each of the ten

pathways that we examined in entirety. A high proportion of the genes involved in fatty acid

metabolism (27%), glycerolipid metabolism (27%), glycolysis (33%), and pyruvate metabolism

(29%) pathways exhibited significant differential expression among temperature treatments.

  16    

Photoperiod also had a strong effect on the expression of genes involved in fatty acid

metabolism; 27% of the genes in this pathway differed in expression between photoperiod

treatments. We used a binomial test to examine large-scale changes in expression across each

pathway. This analysis revealed evidence of concerted expression changes in response to

photoperiod across three integrated pathways (fatty acid metabolism, the citric acid cycle, and

oxidative phosphorylation; Table 2.3). A disproportionate number of genes exhibited trends

towards upregulation in SD across both the citric acid cycle and oxidative phosphorylation, while

the reverse pattern (upregulation in LD) was exhibited for the fatty acid oxidation pathways.

Additionally, the fatty acid metabolism pathway exhibited concerted expression in response to

temperature (upregulation in cold), and the citric acid cycle also exhibited a non-significant trend

in this same direction (P = 0.057).

Transcript abundance correlates to enzyme activity

Expression patterns were largely paralleled by changes in the activities of key enzymes in

relevant metabolic pathways (Figure 2.5). We quantified enzyme activities for enzymes involved

in fatty acid transport capacity (CPT), fatty-acid oxidation capacity (HOAD), and cellular

metabolic intensity (CS). Of these, only HOAD activity exhibited an overall effect of treatment

(one-way ANOVA, P = 0.002). Enhanced capacity for aerobic thermogenesis was associated

with elevated activities of CPT alone (P = 0.044), and CPT was also the only enzyme to reveal

temperature-sensitive activity patterns (P = 0.039). In contrast, CS and HOAD enzyme activities

exhibited photoperiod effects, such that LD individuals demonstrated increased activity of both

enzymes (PCS = 0.039, PHOAD = 0.001) but neither was associated with ΔMsum. We then verified

that differences in transcript abundance translated into variation at the protein level by examining

the relationship between transcript abundance and enzyme activity using linear regression.

Transcript abundance for the gene encoding CS correlated with CS activity (Figure 2.5).

Likewise, each of the three genes encoding HOAD subunits correlated with HOAD activity.

However, CPT activity was not correlated with CPT transcript abundance. The effect of

treatment on HOAD activity was mirrored by overall treatment effects on transcript abundance

for all three HOAD subunit genes, as well as photoperiod-sensitive expression patterns in

HADHA and HADHB, and temperature-sensitive expression patterns in HADHB. CS and CPT

transcript abundances did not exhibit effects of acclimation treatment.

  17    

DISCUSSION Phenotypic flexibility allows an organism to optimally match its phenotype to prevailing

environmental conditions. This reversible modification of complex physiological traits can be

driven by changes in the regulation of genes at several levels of hierarchical organization

(Cheviron & Brumfield, 2012). Such regulatory adjustments could occur at any step in the

underlying linear pathways, but theory predicts that flux control is disproportionately allocated

among the upstream steps (Wright & Rausher, 2010; Eanes, 2011). Our acclimation treatments

successfully altered junco thermogenic performance, and we found that these changes in

performance were associated with underlying transcriptomic changes in genetically independent

but interacting hierarchical pathways. Although juncos in our acclimation treatments only

utilized a subset of the potential physiological strategies for elevating aerobic thermogenic

performance, regulatory modifications were concentrated in upstream portions of integrated

pathways. Additionally, seasonal cues associated with the onset of winter (changes in

temperature and photoperiod) induced distinct transcriptomic responses, highlighting the

importance of temperature and photoperiod cues in mediating distinct, seasonal responses in

avian physiology. Cold-acclimated individuals exhibited elevated thermogenic capacities and

transcriptomic signatures indicative of enhanced O2 delivery as well as enhanced transport and

beta-oxidation of fatty acids, suggesting that enhancement of thermogenic capacity is mediated

through the changes to upstream, fuel supply portions of the metabolic pathways. These

differences in transcript abundances were largely associated with the activities of enzymes that

influence flux through the fatty-acid oxidation and citric acid cycle pathways, suggesting that

changes in gene regulation in these pathways are translated to functional variation at the

proteomic level. Conversely, changes in photoperiod were associated with differential regulation

of downstream pathways that influence cellular oxidative capacities. However, photoperiod

manipulation was not associated with significant changes in thermogenic performance. Taken

together, these results provide important clues into the differential regulation of hierarchal,

multistep step pathways that influence complex physiological adjustments to abiotic stressors.

Transcriptomic signatures of muscle growth

  18    

As in other bird species (e.g., Swanson, 2010), winter increases in junco thermogenic

performance and metabolic rate are often accompanied by pectoralis muscle hypertrophy

(Swanson, 1991). However, we found that pectoralis size was not affected by acclimation

treatment in this study, thus increases in thermogenic performance were not accompanied by

corresponding increases in pectoralis size (Appendix A). Seasonal increases in Msum without

corresponding increases in pectoralis size have also been documented in populations of

American Goldfinches (Spinus tristis; Carey et al. 1978; Liknes et al. 2002). The lack of

pectoralis size changes as a driver of improved thermogenic performance further emphasizes the

importance of cellular adjustments that enhance lipid oxidation.

Despite the lack of changes in pectoralis mass across the acclimation treatments, we still

found transcriptomic responses suggesting the induction of muscle growth. Both metabolic

performance and pectoralis size were correlated with the expression of transcriptomic modules

containing genes related to muscle development. Temperature-sensitive modules associated with

thermogenic capacity included collagen (two in T28), myoblast proliferation (T22), skeletal

muscle cell differentiation (T23), transforming growth factor (T29 & T33), and regulation of

growth (T38) terms. Considering that pectoralis sizes were similar among treatments, differential

expression among these genes could reflect a potential lag between gene expression signaling

and ensuing changes in muscle growth.

Very few genes (6%) in the mTOR signaling pathway were differentially regulated

among temperature treatments, and those that were differentially expressed were

counterintuitively upregulated in warm-acclimated birds. Upstream regulators of the mTOR

pathway, myostatin (MSTN) and insulin-like growth factor 1 (IGF1), are major components of

muscle growth regulation in mammals (Rennie et al., 2004). These same proteins are central in

regulating avian embryonic muscle development (Guernec et al, 2003; Duclos, 2005; Sato et al.,

2006; Kim et al., 2007; McFarland et al., 2007), but their role in seasonal regulation of avian

adult muscle size is unclear. Swanson et al. (2009) found that increases in pectoralis size of

wintering House Sparrows (Passer domesticus) were associated with decreased MSTN

expression, but increases in flight muscle size of photo-stimulated White-throated Sparrows were

conflictingly accompanied by increased expression of both IGF1 and MSTN (Price et al., 2011).

Although IGF1 expression was not detected in our dataset, we found that MSTN expression

differed significantly among the acclimation treatments. Contrary to our expectations however,

  19    

individuals in the winter-like treatment (cold SD) upregulated MSTN expression relative to other

treatment groups (FDR = 0.016), though MSTN was not associated with pectoralis size. Taken as

a whole, this indicates that cold-acclimated individuals were not employing any of the major

muscle growth elements to enhance thermogenic performance.

Transcriptomic signatures of angiogenesis

We found some evidence to indicate that juncos may be enhancing oxygen delivery to the

muscle via angiogenesis. Although the temperature treatments were not associated with

differential expression of either hypoxia inducible factors (HIF), and were only associated with

differential expression of a single gene (paxillin, PXN) across the vascular endothelial growth

factor (VEFG) pathway, VEGFC was upregulated in cold-acclimated individuals and included in

Msum-associated module T28. VEGFC is a VEGF family member largely associated with

lymphangiogenesis (Joukov et al., 1996). Although it is not clear how increased lymphatic

vessels might specifically aid cold individuals, VEGFC has also been shown to increase

angiogenesis under certain experimental conditions (Cao et al., 1998; Witzenbichler et al.,

1998), including in avian embryonic development (Eichmann et al., 1998). Similar function in

cold-acclimated individuals could lead to decreased oxygen diffusion distances at the muscles,

enabling increased oxygen delivery for use in oxidative phosphorylation, and ultimately aerobic

thermogenesis.

Modifications to fuel supply pathways

In addition to enhancements of O2 delivery, increased thermogenic performance could

also result from regulatory changes that enhance the supplies of metabolic fuels to thermogenic

organs. Indeed, enhancements to lipid mobilization have been documented in both migrating and

winter-acclimatized birds (Swanson, 2010). In particular, the processes of circulatory transport

and muscular uptake have been previously implicated as limiting steps in the avian fatty acid

supply pathway (Guglielmo et al., 2002; Jenni-Eiermann & Jenni, 1992; McFarlan et al., 2009;

Vock et al., 1996; Weber, 1992). The fatty acid supply pathway has been well characterized for

birds (Figure 2.1C; for a thorough review see Price, 2010). To briefly summarize the main steps

in this pathway: Before transport, lipids — largely stored as triacylglycerols in both muscle and

adipose tissues, but also as phospholipids in the cell membrane — are hydrolyzed by hormone-

  20    

sensitive lipases (Ramenofsky, 1990). The resulting fatty acids then bind to albumin in the blood

for transport to the muscle or, alternatively, to the liver where they are repackaged as very low

density lipoproteins (VLDL) before continuing on to the muscle (Jenni-Eiermann & Jenni,

1992). Muscle lipoprotein lipases can then hydrolyze VLDL in the capillary endothelium to

enable uptake of fatty acids by the tissue (Nilson-Ehle et al., 1980; Oscai et al., 1990).

Conversion to VLDL is optimal because it allows improved fatty acid transport rate (via directed

uptake at the tissues) without altering blood viscosity and subsequently affecting blood oxygen

transport (Robinson, 1970; Jenni-Eiermann & Jenni, 1992). Once at the myocyte, transport

proteins are similarly required to cross the cellular membrane including plasma membrane-

bound fatty acid binding protein (FABPpm) and fatty acyl translocase (FAT/CD36; Luiken et al.,

1999; Bonen et al. 2004; McFarlan et al., 2009). Within the cell, transport to the mitochondria

occurs with the aid of cytosolic fatty acid binding proteins (FABP; Guglielmo et al., 1998). Here

fatty acids are converted to acyl-CoA by acyl-CoA synthetase, and then to acyl-carnitine via

carnitine palmitoyl tranferase I (Price, 2010). In the final step of fat transport, translocase (or

carnitine acyl transferase) moves acyl-carnitine across the mitochondrial membrane and carnitine

palmitoyl tranferase II reforms acyl-CoA for beta-oxidation, ultimately resulting in ATP

production (Price, 2010). For these reasons, we targeted the genes in the glycerolipid,

glycerophospholipid, and fatty acid metabolism pathways, as well as the many carrier protein

genes in the peroxisome proliferator-activated receptor (PPAR) signaling pathway in order to

determine if seasonal increases in thermogenic capacity could in part be driven by increased fuel

supplies to the pectoralis mitochondria.

Juncos have been shown to seasonally increase muscle lipoprotein lipase during

migration (Ramenofsky, 1990; but see Savard et al., 1991), but its role in enhancing circulatory

lipid transport capacity during winter remains inconclusive (Swanson, 2010). We did not find

evidence of differential regulation in hepatic, endothelial or lipoprotein lipases in response to our

acclimation treatments, suggesting that these lipases did not contribute to elevated thermogenic

performance. Similarly, increases in the expression of sarcolemmal fatty acid transporters in

migrating White-throated Sparrows (Zonotrichia albicollis) suggest that these could also mediate

seasonal variation in fatty acid transport capacities (McFarlan, 2007), but fatty acyl translocase

(CD36) was not differentially expressed in juncos across the acclimation treatments. Seasonal

increases in the abundance of heart-type FABP, which correlates with flight muscle fatty acid

  21    

oxidation capacity (Haunerland, 1994), have been documented in the pectoralis of a number of

avian species during winter [e.g., Black-capped Chickadees, Poecile atricapillus; White-breasted

Nuthatches, Sitta carolinensis (Liknes et al., 2014)]. This points to the central role of H-FABP in

accelerating fatty acid uptake from the circulation (Guglielmo et al., 2002). Although we did not

detect H-FABP expression in our dataset, brain-type FABP was upregulated in the pectoralis of

cold-acclimated individuals, suggesting that it may also play a critical role in the uptake of fats

by skeletal muscle myocytes in juncos. Finally, seasonal increases in the enzymes facilitating fat

transport across the mitochondrial membrane (carnitine acyl CoA transferases) have also been

shown in migrating Semipalmated Sandpipers (Calidris pusilla; Driedzic et al., 1993) and

Western Sandpipers (Guglielmo et al., 2002) but their role in avian metabolic performance

outside of migration (i.e. for winter birds) is unknown (Swanson, 2010). Though we did not find

differential expression among any of the genes encoding the three enzymes that facilitate

transport across the mitochondrial membrane (CPT1, CPT2, and carnitine-acylcarnitine

translocase), we did find a trend towards increased CPT enzyme activity in cold-acclimated

individuals (P = 0.039). Taken together, these results indicate that increased junco thermogenic

performance was not enabled by wholesale changes in gene expression across the lipid fuel

supply pathway. Rather, while upstream components of the supply pathway did not exhibit

differential regulation (i.e. mobilization of fatty acids and circulatory transport), downstream

components were apparently modified to enhance lipid uptake in response to temperature (i.e.

myocyte and mitochondrial uptake).

Apart from limitations to the machinery for fatty acid transport and oxidation, there is

also evidence that rates of mobilization and utilization are strongly influenced by fatty acid type.

Differential mobilization of fatty acids from the adipocyte occurs due to differences in the

molecular structure of fatty acid species (i.e. length, saturation, and relative location of double

bonds), and is thought to be a conserved feature of adipose tissue (Raclot et al., 1995; Raclot,

2003). Accordingly, selective mobilization has been shown in vitro from the adipocytes of Ruff

(Philomachus pugnax) and White-crowned Sparrows, regardless of migratory or exercise state

(Price et al., 2008; Price, 2009). Fatty acid type also influences lipid oxidation rates in avian

muscle (Price, 2009). Deposition of fatty acids in the adipose tissues is largely dependent on the

types of fats consumed in the diet (Pierce & McWilliams, 2005), and experimental manipulation

of dietary fats has been shown to influence whole-animal aerobic performance in birds (Pierce et

  22    

al., 2005; Price & Guglielmo, 2009; McWilliams & Pierce, 2006). While it is possible that

selective mobilization and/or utilization of fatty acids contributed to this variation in

performance, the mechanisms underlying this effect are yet unclear (Price, 2010). Because

juncos were fed a uniform diet for the 8-week duration of the experiment, we do not expect that

adipose stores differed greatly in fatty acid content among treatment groups. Therefore selective

mobilization and/or utilization of fatty acids are not likely to explain the differences in

thermoregulatory performance shown here.

Enhanced cellular aerobic capacity

Regulatory changes to genes involved in lipid oxidation and oxidative phosphorylation

are known to be important for thermogenic capacity in mammals (Cheviron et al. 2012, 2014).

We found similar patterns in juncos, although our results emphasize the importance of the

former, suggesting that regulatory changes that enhance thermogenic performance in juncos are

made to upstream portions of these integrated pathways. Module T32 (negatively correlated with

ΔMsum, Table 2.1) included 1 subunit of NADH dehydrogenase (Complex I – NDUFB9), which

was downregulated in cold-acclimated individuals, but this is the only member of the oxidative

phosphorylation pathway to exhibit temperature-sensitive differential expression. However, it is

unclear how differential regulation of a single component of NADH dehydrogenase — a large

complex composed of 41 avian subunits — could influence the function of Complex I and

subsequently the electron transport chain as a whole. Increased metabolic performance in cold-

acclimated juncos was instead associated with heightened expression of genes related to the beta-

oxidation of fatty acids, the major source of fuel for sustained avian thermogenesis. For instance,

temperature-sensitive transcriptional module T31 was enriched for fatty-acyl-CoA biosynthesis

and HOAD activity GO terms, while the expression of T33 was negatively correlated with

thermogenic capacity and included terms related to fatty acid elongation. Roughly one third of

the genes involved in the fatty acid metabolism pathway were differentially expressed among

temperature treatments. In combination with trends in CPT activity, temperature-sensitive

expression patterns in the genes involved in fatty acid oxidation are largely indicative of

enhanced lipid oxidative capacity in cold-acclimated individuals.

Cold-acclimated birds also downregulated 25% of the genes involved in glycolysis,

indicating a preference for lipid over carbohydrate fuels to enhance thermogenic capacity. This

  23    

preference has similarly been shown in Ruff and in rodents under conditions of prolonged

shivering (Vaillancourt et al., 2005; 2009; Vaillancourt & Weber, 2007). Recent in vitro studies

of House Sparrow flight muscle mitochondria suggest that selection of fatty acid over glycolytic

fuels is likely due to suppression of pyruvate oxidation rather than relative lipid catalytic

potentials (Kuzmiak-Glancy & Willis, 2014). Although we found temperature-sensitive

expression patterns in 29% of the genes in the pyruvate metabolism pathway, cold-acclimated

birds downregulated the expression of four of these genes while upregulating the expression of

the remaining two genes. Thus these regulatory changes were not wholly indicative of pyruvate

oxidation suppression in cold-acclimated individuals.

Enhanced catabolic enzyme activity

For a number of birds, winter increases in Msum are associated with increased cellular

metabolic and lipid oxidation capacities of the pectoralis, as inferred from citrate synthase (CS),

cytochrome c oxidase (COX), and beta-hydroxyacyl Co-enzymeA dehydrogenase (HOAD)

enzyme activities (e.g., Guglielmo et al., 2002; Vézina & Williams, 2005; Zheng et al., 2008;

Liknes & Swanson, 2011). Of the three muscle metabolic enzyme activities that we measured,

enhanced capacity for aerobic thermogenesis was only associated with elevated activities of

CPT. CPT was also the only enzyme to reveal temperature-sensitive activity patterns, while CS

and HOAD enzyme activities exhibited photoperiod-sensitive patterns. Transcript abundances of

the enzyme-encoding genes were correlated with enzyme activities for CS and HOAD but not for

CPT.

Enzyme activity patterns largely paralleled concomitant changes in expression across the

corresponding metabolic pathways. Increased thermogenic capacity among populations of deer

mice is associated with increased HOAD and COX activities and concerted expression patterns

across both the fatty acid beta-oxidation and oxidative phosphorylation pathways (Cheviron et

al., 2012; 2014). Similarly, CPT is considered to be the rate-limiting step in fatty acid oxidation,

and we found that increased CPT activity in cold-acclimated individuals was accompanied by

concerted expression across the fatty acid oxidation pathway in response to temperature.

Likewise, in response to photoperiod we found concomitant changes across fatty acid oxidation,

the citric acid cycle, and oxidative phosphorylation pathways, and these patterns corresponded

with increased CS and HOAD activities among the same treatments. Together, these results

  24    

suggest differential effects of temperature and photoperiod on the two ends of the pathway:

supply is manipulated in response to temperature with seemingly little effect further downstream,

while photoperiod has less of an effect to upstream supply steps and largely modifies

downstream oxidative capacity.

CONCLUSIONS

The mechanisms underlying seasonal phenotypic flexibility remain uncertain, but recent

advances in genomic technologies have enabled high-throughput characterization of the distinct

and concurrent physiological adjustments an organism concurrently employs when responding to

changes in abiotic stressors. Previous studies that surveyed a limited number of parameters

suggest that individuals can increase aerobic performance by making modifications to three

broad levels of physiological organization – enhancements to thermogenic organ size, oxygen

delivery, and cellular aerobic capacity (Figure 2.1). We found evidence that juncos are using

only a small subset of these potential mechanisms to enhance thermogenic performance in

response to experimental manipulations of their abiotic environment. This limited response could

be due, in part, to the strength of the stressor that we employed and the subsequent strength of

the physiological response that was required to offset it. The magnitude of increase by cold-

acclimated individuals is representative of seasonal changes to Msum documented in free-living

passerines (usually 10-50% from summer to winter; Swanson, 2010). However, changes in

thermogenic performance shown here are likely less dramatic than those juncos sustain in the

wild. Msum varies with environmental temperature (Swanson, 2010) and the temperature

employed for cold-acclimation (3°C) was milder than what juncos likely experience throughout

much of their winter range. We suggest that, in the face of maintaining energetically expensive

metabolic machinery, juncos may only engage the minimum degree of change that is necessary

to meet their specific aerobic demands. Given the modular structure of trait components that

influence aerobic performance, efficient regulation of metabolic flexibility may involve

differential regulation of trait components according to the relative costs and benefits of

particular adjustments. Although we successfully increased thermogenic performance through

exposure to relatively mild temperatures, it seems likely that lower temperatures would stimulate

an even more pronounced metabolic response, perhaps uncovering additional transcriptomic

signatures as well. Subsequent investigations to tease apart the relationship between stimulus

  25    

strength and pathway-level responses will help to elucidate which physiological changes might

be more or less costly to induce, and how more severe stresses are mediated.

Changes in thermogenic performance were elicited within a relatively short time period,

with cold-acclimated individuals already showing incremental increases in thermogenic capacity

when measured midway through the acclimation period (at 3 weeks; Swanson et al., 2014),

indicating that temperature is a proximal cue that can be integrated to make rapid physiological

adjustments. This is not surprising as temperatures can be highly variable in the temperate

zone—in contrast to daylength—and birds must quickly respond to precipitous drops in

temperature that could serve as selective events (e.g., polar vortices; Bumpus, 1898). It therefore

follows that these changes would occur at the level of fuel mobilization and oxygen utilization,

representing a rapid vehicle for response. In comparison, relatively immediate (and temporary)

responses in the oxygen cascade come in the form of accelerated breath rate and enhanced

cardiac output, which require no physiological reorganization, while changes that increase

vasculature and mitochondrial abundance represent more persistent changes that are achieved

over prolonged time periods. We found that temperature-sensitive transcriptional modules

associated with increased thermogenic performance showed upregulation of genes related to

beta-oxidation and oxygen utilization. These modules also contain genes related to muscle

development, possibly reflecting longer-term changes in muscle size and structure; however,

individuals did not exhibit corresponding pectoralis hypertrophy. Taken together these results

suggest differential regulation of pathway components that are associated with short-term and

long-term phenotypic adjustments. It remains to be seen how quickly birds can mount responses

to temperature and photoperiod manipulations, and which pathway components respond most

rapidly to these changes. The unpredictable nature of adverse weather events suggests that

temperate endotherms can achieve metabolic flexibility quite rapidly, but further work is needed

to characterize the metabolic — and corresponding transcriptomic — responses associated with

such rapid physiological adjustments in order to determine whether the same physiological

mechanisms are employed.

Finally, many previous investigations of avian seasonal metabolic flexibility have

focused on the migratory period. Avian migration is an extreme athletic feat associated with

dramatic reorganization of multiple physiological systems (van Gils & Piersma, 2011), yet it

represents an acute stressor that lasts for a relatively short period of time (on the order of days).

  26    

This is in stark contrast to the chronic stress of winter, which lasts six months or more in the

temperate north. Therefore, while it may be possible for birds to concurrently maintain many

physiological responses for the duration of a migratory event, it could prove too costly to sustain

these same responses for an entire winter season. Further investigations are thus needed to

determine how free-living birds employ these mechanisms of metabolic flexibility amidst winter

conditions.

ACKNOWLEDGEMENTS

We are very appreciative of Jennifer Jones, for her patience and guidance in the

laboratory, and Nathan Senner, for his comments made to an earlier version of this manuscript.

We also thank Marisa King, Jin-Song Liu, Christopher Merkord, and Yufeng Zhang for their

contributions quantifying metabolic rates and enzyme activities; Ming Liu, Kyle Kirby,

Stephanie Owens, Tyler Tordsen, Jake Johnson, and Travis Carter for technical assistance in the

laboratory and field; and Sol Redlin and Kathie Rasmussen for access to their property for

collection of juncos. This work was supported by grants from the National Science Foundation

[DLS IOS-1021218] and the Illinois Ornithological Society to MS, startup funds from the

University of Illinois at Urbana-Champaign (UIUC) to ZAC, and funds from the UIUC School

of Integrative Biology and Department of Animal Biology to MS.

LITERATURE CITED

Abreu, R.D., Penalva, L.O., Marcotte, E.M., Vogel, C., 2009. Global signatures of protein and

mRNA expression levels. Mol. BioSystems 5, 1512-1526.

Arens, J.R., Cooper, S.J., 2005. Metabolic and ventilatory acclimatization to cold stress in House

Sparrows (Passer domesticus). Physiol. Biochem. Zool. 78, 579-589.

Ayroles, J.F., Carbone, M.A., Stone, E.A., Jordan, K.W., Lyman, R.F., Magwire, M.M. et al.,

2009. Systems genetics of complex traits in Drosophila melanogaster. Nature Gen. 41,

299-307.

Beall, C.M., Laskowski, D.L., Erzurum, S.C. Nitric oxide in adaptation to altitude. Free

Radic, Biol. Med. 52, 1123-1134.

Benjamini, Y., Hochberg, Y., 1995. Controlling the false discovery rate: a practical and

powerful approach to multiple testing. J. Royal Statistical Soc. Series B 57, 289–300.

  27    

Bonen, A., Campbell, S.E., Benton, C.R., Chabowski, A., Coort, S.L.M., Han, X-X., et al., 2004.

Regulation of fatty acid transport by fatty acid translocase/CD36, Proc. Nutr. Soc. 63,

245–249.

Bruick, R., McKnight, S.L., 2001. A conserved family of prolyl-4-hydroxylases that modify

HIF. Science 294, 1337-1340.

Bumpus, H.C., 1898. Eleventh lecture. The elimination of the unfit as illustrated by the

introduced sparrow, Passer domesticus. (A fourth contribution to the study of variation.)

Biol. Lectures: Woods Hole Marine Biological Laboratory, 209–225.

Cao, Y., Linden, P., Farnebo, J., Cao, R., Erikkson, A., Kumar, V. et al., 1998. Vascular

endothelial growth factor C induces angiogenesis in vivo. Proc. Natl. Acad. Sci. USA 95,

14389-14394.

Carey, C., Dawson, W.R., Maxwell, L.C., Faulkner, J.A., 1978. Seasonal acclimatization to

temperature in cardueline finches. II. Changes in body composition and mass in relation

to season and acute cold stress. J. Comp. Physiol. 125, 101–113.

Carey, C., Marsh, R.L., Bekoff, A., Johnston, R.M., Olin, A.M., 1989. Enzyme activities in

muscles of seasonally acclimatized House Finches. In: Bech, C. & Reinertsen, R.E. (eds).

Physiology of cold adaptation in birds. Plenum Life Sciences, New York, pp 95–104.

Cheviron, Z.A., Bachman, G.C., Connaty, A.D., McClelland, G.B., Storz, J.F., 2012.

Regulatory changes contribute to the adaptive enhancement of thermogenic capacity in

high-altitude deer mice. Proc. Natl. Acad. Sci. USA 109, 8635-8640.

Cheviron, Z.A., Brumfield, R., 2012. Genomic insights into adaptation to high-altitude

environments. Heredity 108, 354-361.

Cheviron, Z.A., Connaty, A.D., McClelland, G.B., Storz, J.F. 2014. Functional genomics of

adaptation to hypoxic cold-stress in high-altitude deer mice: transcriptomic plasticity and

thermogenic performance. Evolution 68, 42-62.

Cooper, S.J., Swanson, D.L., 1994. Seasonal acclimatization of thermoregulation in the Black-

Capped Chickadee. Condor 96, 638-646.

Dawson, W.R., O’Connor, T.P., 1996. Energetic features of avian thermoregulatory responses.

In: Carey, C. (ed). Avian energetics and nutritional ecology. Chapman & Hall, New

York, pp 85–124.

DeWitt, T.J., Scheiner, S.M., 2004. Phenotypic plasticity: functional and conceptual

  28    

approaches. Oxford University Press, New York, NY.

Driedzic, W.R., Crowe, H.L., Hicklin, P.W. & Sephton, D.H., 1993. Adaptations of pectoralis

muscle, heart mass, and energy metabolism during premigratory fattening in

Semipalmated Sandpipers (Calidris pusilla). Can. J. Zool. 71, 1602–1608.

Duclos, M. J., 2005. Insulin-like growth factor-I (IGF-1) mRNA levels and chicken muscle

growth. J. Physiol. Pharmacol. 56 Suppl. 3, 25-35.

Eanes, W.F., 2011. Molecular population genetics and selection in the glycolytic pathway. J.

Exp. Biol. 214, 165-171.

Eichmann, A., Corbel, C., Jaffredo, T., Bréant, C., Joukov, V., Kumar, V. et al., 1998. Avian

VEGF-C: cloning, embryonic expression pattern and stimulation of the differentiation of

VEGFR2-expressing endothelial cell precursors. Development 125, 743-752.

Evans, P.R., Davidson, N.C., Uttley, J.D., Evans, R.D., 1992. Premigratory hypertrophy of flight

muscles: an ultrastructural study. Ornis. Scand. 23, 238–243.

Ferrara, N., 1997. The biology of vascular endothelial growth factor. Endocrine Rev. 18, 4-25.

Flowers, J., Sezgin, E., Kumagai, S., Duvernell, D., Matzkin, L., Schmidt, P. et al., 2007.

Adaptive evolution of metabolic pathways in Drosophila. Mol. Biol. Evol. 24, 1347-

1354.

Gaunt, A.S., Hikida, R.S., Jehl, J.R. Jr, Fenbert, L., 1990. Rapid atrophy and hypertrophy of an

avian flight muscle. Auk 107, 649–659.

Guernec A., Berri, C., Chevalier, B., Wacrenier-Cere, N., Le Bihan-Duval, E., Duclos, M.J.

2003. Muscle development, insulin-like growth factor-I and myostatin mRNA levels in

chickens selected for increased breast muscle yield. Growth Horm. IGF Res. 13, 8–18.

Guglielmo, C.G., 2010. Move that fatty acid: fuel selection and transport in migratory birds and

bats. Integr. Comp. Biol. 50, 336-345.

Guglielmo, C.G., Haunerland, N.H., Williams, T.D., 1998. Fatty acid binding protein, a major

protein in the flight muscle of migrating Western Sandpipers. Comp. Biochem. Physiol.

B 119, 549–555.

Guglielmo, C.G., Haunerland, N.H., Hochachka, P.W., Williams, T.D., 2002. Seasonal dynamics

of flight muscle fatty acid binding protein and catabolic enzymes in a migratory

shorebird. Am. J. Physiol. 282, R1405–R1413.

Hayes, J. P., O’Connor, C.S., 1999. Natural selection on thermogenic capacity of high-altitude

  29    

deer mice. Evolution 53, 1280–1287.

Heinemeier, K. M., Olesen, J. L., Schjerling, P., Haddad, F., Langberg, H., Baldwin, K. M., et

al., 2007. Short-term strength training and the expression of myostatin and IGF-I

isoforms in rat muscle and tendon: differential effects of specific contraction types. J.

Appl. Physiol. 102, 573-581.

Hohtola, E., 1982. Thermal and electromyographic correlates of shivering thermogenesis in the

pigeon. Comp. Biochem. Physiol. 73A, 159–166.

Hohtola, E., Henderson, R.P., Rashotte, M.E., 1998. Shivering thermogenesis in the pigeon: the

effects of activity, diurnal factors, and feeding state. Am. J. Physiol. 275, R1553–R1562.

Holloszy, J.O., Coyle, E.F., 1984. Adaptations of skeletal muscle to endurance exercise and

their consequences. J. Appl. Physiol. 56, 831-838.

Hudlicka, O., 1985. Development and adaptability of microvasculature in skeletal muscle. J.

Exp. Biol. 115, 215 - 228.

Jenni-Eiermann, S., Jenni, L., 1992. High plasma triglyceride levels in small birds during

migratory flight: a new pathway for fuel supply during endurance locomotion at very

high mass-specific metabolic rates. Physiol. Zool. 65, 112–123.

Joukov, V., Pajusola, K., Kaipainen, A., Chilov, D., Lahtinen, I., Kukk, E., et al., 1996. A novel

vascular endothelial growth factor, VEGF-C, is a ligand for the Flt-4 (VEGFR3) and

KDR (VEGFR2) receptor tyrosine kinases. EMBO J. 15, 290-298.

Kanehisa, M., Goto, S., 2000. KEGG: Kyoto Encyclopedia of Genes and Genomes. Nucleic

Acids Res. 28, 27-30.

Kanehisa, M., Goto, S., Sato, Y., Kawashima, M., Furumichi, M., Tanabe, M., 2014. Data,

information, knowledge and principle: back to metabolism in KEGG. Nucleic Acids Res.

42, D199–D205.

Kim, Y.S., Bobbili, N.K., Lee, Y.K., Jin, H.J., Dunn, M.A., 2007. Production of a polyclonal

anti-myostatin antibody and the effects of in ovo administration of the antibody on

posthatch broiler growth and muscle mass. Poult. Sci. 86, 1196–1205.

Klagsbrun, M., D’Amore, P., 1996. Vascular endothelial growth factor and its receptors.

Cytokine Growth Factor Rev. 7, 259-270.

Kuzmiak-Glancy, S., Willis, W.T., 2014. Skeletal muscle fuel selection occurs at the

mitochondrial level. J. Exp. Biol. 217, 1993-2003.

  30    

Liknes, E.T., 2005. Seasonal acclimatization patterns and mechanisms in small, temperate-

resident passerines: phenotypic flexibility of complex traits. Ph.D. dissertation,

University of South Dakota, Vermillion.

Liknes, E.T., Guglielmo, C.G., Swanson, D. L., 2014. Phenotypic flexibility in passerine birds:

seasonal variation in fuel storage, mobilization and transport. Comp. Biochem. Physiol. A

174, 1-10.

Liknes, E.T., Scott, S.M., Swanson, D.L., 2002. Seasonal acclimatization in the American

goldfinch revisited: to what extent to metabolic rates vary seasonally? Condor 104, 548–

557.

Liknes, E.T., Swanson, D.L., 1996. Seasonal variation in cold tolerance, basal metabolic rate,

and maximal capacity for thermogenesis in White-Breasted Nuthatches Sitta carolinensis

and Downy Woodpeckers Picoides pubescens, two unrelated arboreal temperate

residents. J. Avian Biol. 27, 279-288.

Liknes, E.T., D.L. Swanson., 2011. Phenotypic flexibility in passerine birds: seasonal variation

of aerobic enzyme activities in skeletal muscle. J. Therm. Biol. 36, 430-436.

Luiken, J.J.F.P., Schaap, F.G., van Nieuwenhoven, F.A., van der Vusse, G.J., Bonen, A., Glatz,

J.F.C. ,1999. Cellular fatty acid transport in heart and skeletal muscle as facilitated by

proteins. Lipids 34, S169–S175.

Lundgren, B.O., Kiessling, K-H., 1985. Seasonal variation in catabolic enzyme activities in

breast muscle of some migratory birds. Oecologia 66, 468–471.

Lundgren, B.O., Kiessling, K-H., 1988. Comparative aspects of fibre types, areas, and capillary

supply in the pectoralis muscle of some passerine birds with differing migratory

behavior. J. Comp. Physiol. B 158, 165–173.

Marsh, R.L., Dawson, W.R., 1989. Avian adjustments to cold. In: Wang LCH (ed) Advances in

comparative and environmental physiology 4: animal adaptation to cold. Springer, Berlin,

pp 206–253.

Mathieu-Costello, O., Agey, P.J., 1997. Chronic hypoxia affects capillary density and geometry

in pigeon pectoralis muscle. Respir. Physiol. 109, 39-52.

McCarthy, J.J., Esser, K.A., 2010. Anabolic and catabolic pathways regulating skeletal muscle

mass. Curr. Opin. Clin. Nutr. Metab. Care. 13, 230-235.

McCarthy, D., Chen, Y. & Smyth, G. 2012. Differential expression analysis of multifactor RNA-

  31    

Seq experiments with respect to biological variation. Nucl. Acids Res. 40, 4288–4297.

McClelland, G.B., 2004. Fat to the fire: the regulation of lipid oxidation with exercise and

environmental stress. Comp. Biochem. Physiol. B 139, 443–60.

McFarland, D. C., Velleman, S. G., Pesall, J. E., Liu, C., 2007. The role of myostatin in chicken

(Gallus domesticus) myogenic satellite cell proliferation and differentiation. Gen. Comp.

Endocrinol. 151, 351-357.

McFarlan, J.T., 2007. Seasonal upregulation of fatty acid uptake proteins in flight muscles of

Zonotrichia sparrows during migration. M.S. Thesis, University of Western Ontario,

London, ON.

McFarlan, J.T., Bonen, A., Guglielmo, C.G., 2009. Seasonal up-regulation of protein mediated

fatty acid transport in flight muscles of migratory white-throated sparrows (Zonotrichia

albicollis). J. Exp. Biol. 212, 2934–40.

McPherron, A.C., Lawler, A.M., Lee, S-J., 1997. Regulation of skeletal muscle mass in mice by

new TGF-β superfamily member. Nature 387, 83-90.

McWilliams, S.R., Pierce, B., 2006. Diet, body composition, and exercise performance: why

birds during migration should be careful about what they eat (abstract). In: Comparative

Physiology 2006: integrating diversity. The American Physiological Society, Virginia

Beach.

Nilson-Ehle, P.A., Garfinkel, A.S., Schotz, M.C., 1980. Lipolytic enzymes and plasma

lipoprotein metabolism. Annu. Rev. Biochem. 49, 667-693.

Nolan, Jr., V., Ketterson, E.D., Cristol, D.A., Rogers, C.M., Clotfelter, E.D., Titus, R.C., et al.,

2002. Dark-eyed Junco (Junco hyemalis), The Birds of North America Online (A. Poole,

Ed.). Ithaca: Cornell Lab of Ornithology; Retrieved from the Birds of North America

Online: http://bna.birds.cornell.edu/bna/species/716

O’Connor, T.P., 1995. Seasonal acclimatization of lipid mobilization and catabolism in House

Finches (Carpodacus mexicanus). Physiol. Zool. 68, 985–1005.

Oscai, L.B., Essig, D.A., Palmer, W.K., 1990. Lipase regulation of muscle triglyceride

hydrolysis. J. Appl. Physiol 69, 1571-1577.

Pelsers, M.M.A.L., Butler, P.J., Bishop, C.M., Glatz, J.F.C., 1999. Fatty acid binding protein in

heart and skeletal muscles of the migratory Barnacle Goose throughout development.

Am. J. Physiol. 276, R637–R643.

  32    

Pierce, B.J., McWilliams, S.R., 2005. Seasonal changes in composition of lipid stores in

migratory birds: causes and consequences. Condor 107, 269–279.

Pierce, B.J., McWilliams, S.R., O’Connor, T.P., Place, A.R., Guglielmo, C.G., 2005. Effect of

dietary fatty acid composition on depot fat and exercise performance in a migrating

songbird, the red-eyed vireo. J. Exp. Biol. 208, 1277-1285.

Piersma, T., Drent, J., 2003. Phenotypic flexibility and the evolution of organismal design.

Trends Ecol. Evol. 18, 228-233.

Piersma, T., van Gils, J.A., 2011. The flexible phenotype: a body-centred integration of

ecology, physiology, and behaviour. Oxford University Press, New York, NY.

Price, E.R., 2009. The Effects of Dietary Fatty Acids on Avian Migratory Performance,

Biology. University of Western Ontario, London, ON.

Price, E.R., 2010. Dietary lipid composition and avian migra- tory flight: development of a

theoretical framework for avian fat storage. Comp. Biochem. Physiol. A 157, 297-309.

Price, E.R., Guglielmo, C.G., 2009. The effect of muscle phospholipid fatty acid composition

on exercise performance: a direct test in the migratory white- throated sparrow

(Zonotrichia albicollis). Am. J. Physiol. Regul. Integr. Comp. Physiol. 297, R775–R782.

Price, E.R., Krokfors, A., Guglielmo, C.G., 2008. Selective mobilization of fatty acids from

adipose tissue in migratory birds. J. Exp. Biol. 211, 29–34.

Price, E.R., Bauchinger, U., Zajac, D.M., Cerasale, D.J., McFarlan, J., Gerson, A.R. et al., 2011.

Migration- and exercise-induced changes to flight muscle size in migratory birds and

association with IGF1 and myostatin mRNA. J. Exp. Biol. 214, 2823-2831.

Raclot, T., 2003. Selective mobilization of fatty acids from adipose tissue triacylglycerols.

Prog. Lipid Res. 42, 257-288.

Raclot, T., Mioskowski, E., Bach, A.C., Groscolas, R., 1995. Selectivity of fatty-acid

mobilization – a general metabolic feature of adipose-tissue. Am. J. Physiol. 269, R1060-

R1067.

Ramenofsky, M., 1990. Fat storage and fat metabolism in relation to migration. In: Gwinner E

(ed) Bird migration: physiology and ecophysiology. Springer, Berlin, pp 214–231.

Rankin, E.B., Biju, M.R., Liu, Q., Unger, T.L., Rha, J., Johnson, R.S., et al., 2007. Hypoxia-

inducible factor-2 regulates hepatic erythropoietin in vivo. J. Clin. Invest. 117, 1068-

1077.

  33    

Rennie, M. J., Wackerhage, H., Spangenburg, E. E., Booth, F. W., 2004. Control of the size of

the human muscle mass. Annu. Rev. Physiol. 66, 799-828.

Robinson, D.S., 1970. The function of the plasma triglycerides in fatty acid transport. P. 51-105

in M. Florkin and E.H. Stotz, ed. Comprehensive biochemistry. Vol. 18. Elsevier, New

York.

Robinson, M., McCarthy, D., Smyth, G., 2010. edgeR: a Bioconductor package for differential

expression analysis of digital gene expression data. Bioinformatics 26, 139–140.

Robinson, M., Oshlack, A., 2010. A scaling normalization method for differential expression

analysis of RNA-seq data. Genome Biol. 11, R25.

Robinson, M., Smyth, G., 2007. Moderated statistical tests for assessing differences in tag

abundance. Bioinformatics 23, 2881–2887.

Rothe, H. J., Biesel, W., Nachtigall, W. 1987. Pigeon flight in a wind-tunnel. 2. Gas-exchange

and power requirements. J. Comp. Physiol. B Biochem. Syst. Environ. Physiol. 157, 99-

109.

Sato, F., Kurokawa, M., Yamauchi, N., Hattori, M-a., 2006. Gene silencing of myostatin in

differentiation of chicken embryonic myoblasts by small interfering RNA. Am. J.

Physiol. Cell Physiol. 291, C538-C545.

Savard, R., Ramenofsky, M., Greenwood, M.R.C., 1991. A north-temperate migratory bird: a

model for the fate of lipids during exercise of long duration. Can. J. Physiol. Pharmacol.

69, 1443–1447.

Sezgin, E., Duvernell, D. D., Matzkin, L. M., Duan, Y., Zhu, C. T., Verrelli, B. C. et al., 2004.

Single-locus latitudinal clines and their relationship to temperate adaptation in metabolic

genes and derived alleles in Drosophila melanogaster. Genetics 168, 923-931.

Stone, E. A., Ayroles, J.F., 2009. Modulated modularity clustering as a tool for functional

genomic inference. PLoS Genet. 5, e1000479.

Suarez, R.K., Lighton, J.R.B., Moyes, C.D., Brown, G.S., Gass, C.L., Hochachka, P.W., 1990.

Fuel selection in rufous hummingbirds: Ecological implications of metabolic

biochemistry. Proc. Natl. Acad. Sci. USA 87, 9207-9210.

Swanson, D.L., 1991. Substrate metabolism under cold stress in seasonally acclimatized dark-

eyed juncos. Physiol. Zool. 64, 1578–1592.

Swanson, D.L., 2001. Are summit metabolism and thermogenic endurance correlated in winter

  34    

acclimatized passerine birds? J. Comp. Physiol. B 171, 475–481.

Swanson, D.L., 2010. Seasonal metabolic variation in birds: functional and mechanistic

correlates. Curr. Ornithol. 17, 75-129.

Swanson, D.L., Drymalski, M.W., Brown, J.R., 1996. Sliding vs static cold exposure and the

measurement of summit metabolism in birds. J. Therm. Biol. 21, 221-226.

Swanson, D.K., Liknes, E.T., 2006. A comparative analysis of thermogenic capacity and cold

tolerance in small birds. J. Exp. Biol. 209, 466-474.

Swanson, D.L., Sabirzhanov, B., VandeZande, A., Clark, T.G., 2009. Down-regulation of

myostatin gene expression in pectoralis muscle is associated with elevated thermogenic

capacity in winter acclimatized house sparrows (Passer domesticus). Physiol. Biochem.

Zool. 82, 121–128.

Swanson, D.L., Zhang, Y., King, M., 2013. Individual variation in thermogenic capacity is

correlated with muscle flight size but not cellular metabolic capacity in American

Goldfinches (Spinus tristis). Physiol. Biochem. Zool. 86: 421-431.

Swanson, D.L., Zhang, Y., Liu, J-S., Merkord, C.L., King, M., 2014. Relative roles of

temperature and photoperiod as drivers of metabolic flexibility in dark-eyed juncos. J.

Exp. Biol. 217, 866-875.

Vaillancourt, E., Haman, F., Weber, J-M., 2009. Fuel selection in Wistar rats exposed to cold:

shivering thermogenesis diverts fatty acids from re-esterification to oxidation. J. Physiol.

587, 4349-4359.

Vaillancourt, E., Prud’Homme, S., Haman, F., Guglielmo, C.G., Weber, J-M., 2005. Energetics

of a long- distance migrant shorebird (Philomachus pugnax) during cold exposure and

running. J. Exp. Biol. 208, 317–325.

Vaillancourt, E., Weber, J-M., 2007. Lipid mobilization of long-distance migrant birds in vivo:

the high lipolytic rate of ruff sandpipers is not stimulated during shivering. J. Exp. Biol.

210, 1161-1169.

Vézina, F., Jalvingh, K.M., Dekinga, A., Piersma, T., 2006. Acclimation to different thermal

conditions in a northerly wintering shorebird is driven by body mass-related changes in

organ size. J. Exp. Biol. 209, 3141–3154.

Vézina, F., Jalvingh, K.M., Dekinga, A., Piersma, T., 2007. Thermogenic side effects to

migratory disposition in shorebirds. Am. J. Physiol. Regul. Integr. Comp. Physiol. 292,

  35    

1287–1297.

Vézina, F., Williams, T.D., 2005. Interaction between organ mass and citrate synthase activity

as an indicator of tissue maximal oxidative capacity in breeding European Starlings:

implications for metabolic rate and organ mass relationships. Funct. Ecol. 19, 119–128.

Vock, R., Weibel, E.R., Hoppeler, H., Ordway, G., Weber, J-M., Taylor, C.R., 1996. Design of

the oxygen substrate pathways. V. Structural basis of vascular substrate supply to muscle

cells. J. Exp. Biol. 199, 1675–88.

Wang, Z., Gerstein, M., Snyder, M., 2009. RNA-Seq: a revolutionary tool for transcriptomics.

Nat. Rev. Genet. 10, 57–63.

Weber, J-M., 1992. Pathways for oxidative fuel provision to working muscles: ecological

consequences of maximal supply limitations. Experientia 48, 557–64.

Welch, K., Altschuler, D.L., 2009. Fiber type homogeneity of the flight musculature in small

birds. Comparative Biochemistry and Physiology B 152, 324-331.

Welle, S., Burgess, K., Thornton, C.A., Tawil, R., 2009. Relation between extent of myostatin

depletion and muscle growth in mature mice. Am. J. Physiol. Endocrinol. Metab. 297,

E935-940.

Witzenbichler, B., Asahara, T. Murohara, T., Silver, M., Spyyridopoulos, I., Magner, M. et al.,

1998. Vascular endothelial growth factor-C (VEGF-C/VEGF-2) promotes angiogenesis

in the setting of tissue ischemia. Am. J. Pathol. 153, 381-394.

Wright, K. M., Rausher, M. D., 2010. The evolution of control and distribution of adaptive

mutations in a metabolic pathway. Genetics 18, 483-502.

Wu, X., Watson, M., 2009. CORNA: testing gene lists for regulation by microRNAs.

Bioinformatics 25, 832-833.

Yacoe, M. E., Dawson, W.R., 1983. Seasonal acclimatization in American goldfinches: the role

of the pectoralis muscle. Am. J. Physiol. 245, R265-R271.

Zajac, D.M., Cerasale, D.J., Landman, S., Guglielmo, C.G., 2011. Behavioral and physiological

effects of photoperiod-induced migratory state and leptin on Zonotrichia albicollis: II.

Effects on fatty acid metabolism. Gen. Comp. Endocr. 174, 269–275.

Zheng, W-H., Li, M., Liu, J-S., Shao S-L., 2008. Seasonal acclimatization of metabolism in

Eurasian Tree Sparrows (Passer montanus). Comp. Biochem. Physiol. A 151, 519–525.

  36    

CHAPTER 2 TABLES & FIGURES

TABLE 2.1. Associations between transcriptional modules and thermogenic performance (ΔMsum) or pectoralis size (Mp). Number of genes and proportion of variance explained by PC1 are shown for each module. Modules are illustrated in Figure 2.3. Direction of correlation (positive or negative), r-squared, and p-values from linear regressions. NS = uncorrected P > 0.05. Values in bold are significant after Bonferroni correction from multiple tests. Modules A19 and P21 are comprised of the same genes.

Module No.

genes Prop.

variance Thermo. performance Pectoralis size +/- R2 P +/- R2 P

A3 3 0.99 - 0.13 0.014 NS NS A5 4 0.91 NS NS + 0.12 0.017 A7 4 0.96 NS NS + 0.09 0.034 A18 6 0.59 + 0.09 0.038 NS NS

A19/P21 6 0.98 NS NS + 0.09 0.033 A23 4 0.76 - 0.09 0.033 NS NS A26 15 0.83 + 0.09 0.037 NS NS A27 10 0.98 - 0.16 0.007 NS NS A28 5 0.95 + 0.11 0.022 NS NS A31 6 0.60 + 0.15 0.009 NS NS A34 65 0.55 - 0.12 0.016 NS NS A38 9 0.70 + 0.16 0.006 NS NS A41 144 0.50 - 0.09 0.037 + 0.08 0.048 A43 43 0.85 + 0.10 0.030 NS NS A45 407 0.75 - 0.25 <0.001 + 0.08 0.05 A47 116 0.99 - 0.29 <0.001 NS NS P8 4 0.96 NS NS + 0.09 0.034 P32 8 0.81 NS NS + 0.10 0.030 P36 20 0.67 - 0.12 0.017 NS NS P41 164 0.68 NS NS + 0.09 0.033 P45 11 0.91 - 0.09 0.040 NS NS P50 49 0.50 0.13 0.014 + NS NS P51 15 0.49 NS NS + 0.10 0.025 T1 2 0.99 + 0.19 0.003 NS NS T3 3 0.99 NS NS + 0.09 0.034 T4 4 0.98 - 0.13 0.013 NS NS T5 3 0.99 NS NS - 0.12 0.019 T6 3 1.00 + 0.11 0.023 NS NS T10 3 0.80 + 0.09 0.036 NS NS T12 4 0.81 - 0.10 0.032 NS NS T17 5 0.99 - 0.11 0.025 NS NS T19 7 0.95 - 0.09 0.035 NS NS T20 18 0.56 - 0.17 0.005 NS NS T21 12 0.87 - 0.10 0.031 NS NS T22 9 0.98 - 0.16 0.007 NS NS

  37    

TABLE 2.1 (continued).

No. genes

Prop. variance

Thermo. performance Pectoralis size Module +/- R2 P +/- R2 P

T23 12 0.82 - 0.20 0.003 NS NS T24 11 0.70 - 0.21 0.002 NS NS T25 19 0.96 - 0.30 <0.001 NS NS T27 47 0.68 + 0.16 0.007 NS NS T28 41 0.50 + 0.24 <0.001 NS NS T29 17 0.90 - 0.15 0.008 NS NS T30 16 0.54 - 0.18 0.004 NS NS T31 48 0.64 - 0.34 <0.001 NS NS T32 41 0.97 - 0.09 0.033 NS NS T33 33 0.60 - 0.20 0.002 NS NS T35 193 0.94 - 0.21 0.002 NS NS T37 41 0.82 + 0.14 0.010 NS NS T38 15 0.91 - 0.16 0.007 + 0.10 0.026

  38    

TABLE 2.2. Summarized enrichment results for temperature-sensitive modules of differentially expressed genes associated with thermogenic capacity. Individual modules were significantly enriched for terms related to categories indicated in black.

Module Lipid metabol. OXPHOS Glyco-

lysis Angio-

genesis Muscle growth

Muscle contract.

O2 transport

Cell growth

Cell commun.

Immune response

Transcript factor

T4

T6

T12

T17

T20

T21

T22

T23

T24

T25

T27

T28

T29

T30

T31

T33

T37

T38

  39    

TABLE 2.3. Targeted pathway analyses. Number of unique genes in each pathway, number detected within our dataset, and number significantly differentially expressed (DE) across temperature and across photoperiod treatments. P-values shown from binomial tests performed to evaluate the presence of concerted expression across the pathway (significant values in bold).

Pathway No. unique

genes No. genes in dataset

Temperature Photoperiod DE P DE P

Citric acid cycle 14 14 1 0.057 0 0.002 Fatty acid metabolism 15 15 4 0.035 4 0.035

Glycerolipid metabolism 16 15 4 0.302 6 0.119 Glycerophospholipid metabolism 28 25 4 1.000 5 0.690

Glycolysis 26 24 8 0.152 4 0.152 mTOR signaling 33 31 2 0.281 4 0.720

Oxidative phosphorylation 97 84 1 1.000 3 <0.001 PPAR signaling 38 32 4 0.215 5 0.377

Pyruvate metabolism 21 21 6 1.000 5 0.189 VEGF signaling 24 24 1 0.839 1 0.152

  40    

FIGURE 2.1. (A) Modifications to enhance avian metabolic performance can occur via increases to the size of thermogenic organs (hypertrophy of pectoralis), to oxygen delivery (angiogenesis), or to cellular aerobic capacity. These modifications are achieved by altering the expression of genes in associated pathways, bulleted below. (B) Each pathway is a network of many interacting genes; those comprising the citric acid cycle are shown in the inset. (C) The trajectory of lipids (gray circles) from adipocyte to myocyte and into the mitochondria with the aid of enzymes (in italics) and transporter proteins (white boxes). Solid lines represent enzymatic conversion of lipid forms; dotted lines represent transport. Acyl carnitine (AC); carnitine-acylcarnitine translocase (CACT); fatty acid translocase (CD36); carnitine palmitoyl transferase (CPT); fatty acid (FA); fatty acid binding protein (FABP); plasma-membrane bound fatty acid binding protein (FABPpm); hormone-sensitive lipase (HSL); triacylglyceride (TAG); very low density lipid (VLDL).

Angiogenesis Muscle growth

2. Oxygen delivery

1. Thermogenic organ size

3. Cellular aerobic capacity

myocyte

Glycolysis

mitochondria

Fatty acid β-oxidation

Citric acid cycle

Oxidative phosphorylation

ATP

Lipid fuel supply

•  Citric acid cycle •  Fatty acid β-oxidation •  Glycerolipid metabolism •  Glycerophospholipid metabolism •  Glycolysis •  Oxidative phosphorylation •  PPAR signaling •  Pyruvate metabolism

•  mTOR signaling •  PI3K-Akt signaling

•  HIF-1 signaling •  VEGF signaling

O2

A

B

TAG FA HSL

bloodstream Albumin

VLDL

Lipoprotein lipase

FA

β-oxidation

FA

FA

Acyl synthetase CPT1 Acyl

CoA

CPT2

sarcolemma

cytosol

mitochondrial membrane

mitochondrial matrix

FABPpm CD36

FABP C

ACLY ACO2 ACO2 IDH1 IDH3A IDH1 OGDH

MDH1 FH SDHA SUCLA DLST DLD OGDH

VLDL FA

FA AC

AC Acyl CoA

adipocyte liver

CACT

  41    

FIGURE 2.2. Change in thermogenic performance (ΔMsum) from before and after six-week acclimation treatments.

Cold LD Cold SD Warm LD Warm SD

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

ΔMsum

  42    

FIGURE 2.3. Correlated transcriptional modules. (A) Clustering of 1882 transcripts with significant treatment effects into 47 modules. (B) Clustering of 1325 transcripts with significant photoperiod treatment effects into 51 modules. (C) Clustering of 756 transcripts with significant temperature treatment effects into 38 modules. For each, ordered genes listed along x- and y-axes with pair-wise Pearson correlation coefficients shown in color (red = highly correlated; blue = highly uncorrelated). Red blocks along diagonal indicate clusters of genes that are highly co-regulated in expression pattern.

A ! Total treatment effects! B ! Photoperiod effects!

C ! !Temperature effects!

Correlation (r)!

  43    

FIGURE 2.4. Number of genes differentially expressed among all treatments, temperature treatments, and photoperiod treatments.

92 252468

196877

339Temperature Photoperiod

Treatment

  44  

FIGURE 2.5. Metabolic enzyme activity and corresponding transcript abundance for the enzyme-encoding genes. (A) CPT activity and CPT1α abundance. (B) CS activity and ACLY abundance. (C) HOAD activity and HADH (blue), HADHA (black), and HADHB (gray) abundance. Lines fit with linear regression for significant associations, P-values shown.

10 15 20 25 30 35 40

510

15

CPT transcript abundance

CP

T ac

tivity

(uni

ts m

in−1)

p = 0.180

20 40 60 80 100 120 140

200

300

400

500

CS transcript abundance

CS

act

ivity

(uni

ts m

in−1)

p = 0.022

500 1000 1500 2000 2500

80100

120

140

160

180

200

HOAD transcipt abundance

HO

AD

act

ivity

(uni

ts m

in−1)

p = 0.002p = 0.004p = 0.003

A! B!

C!

  45  

CHAPTER 3

SIGNATURES OF NATURAL SELECTION IN THE MITOCHONDRIAL GENOMES OF

TACHYCINETA SWALLOWS AND THEIR IMPLICATIONS FOR LATITUDINAL

PATTERNS OF THE ‘PACE OF LIFE’*

Maria Stager1, David J. Cerasale2, Roi Dor3, David W. Winkler4, and Zachary A. Cheviron1

1 Department of Animal Biology, University of Illinois at Urbana-Champaign, Urbana, IL 61801 USA 2 WestLand Resources, Inc. 4001 E. Paradise Falls Drive, Tucson, AZ 85712, USA 3 Department of Zoology, Tel Aviv University, Tel Aviv 6997801, Israel 4 Cornell University Museum of Vertebrates, Department of Ecology & Evolutionary Biology and Laboratory of

Ornithology, Cornell University, Ithaca, NY 14853 USA

ABSTRACT

Latitudinal variation in avian life histories can be summarized as a slow-fast continuum,

termed the ‘pace of life’, that encompasses patterns in life span, reproduction, and rates of

development among tropical and temperate species. Much of the variation in avian pace of life is

tied to differences in rates of long-term metabolic energy expenditure. Given the vital role of the

mitochondrion in metabolic processes, studies of variation in the mitochondrial genome may

offer opportunities to establish mechanistic links between genetic variation and latitudinal ‘pace

of life’ patterns. Using comparative genomic analyses, we examined complete mitochondrial

genome sequences obtained from nine, broadly distributed Tachycineta swallow species to test

for signatures of natural selection across the mitogenome within a phylogenetic framework. Our

results show that although purifying selection is the dominant selective force acting on the

mitochondrial genome in Tachycineta, three mitochondrial genes (ND2, ND5, and CYTB)

contain regions that exhibit signatures of diversifying selection. Two of these genes (ND2 and

ND5) encode interacting subunits of NADH dehydrogenase, and amino residues that were

                                                                                                               *  This chapter was published in its entirety in the journal Gene. Stager, M., Cerasale, D.J., Dor, R., Winkler, D.W. & Cheviron, Z.A. 2014. Signatures of natural selection in the mitochondrial genomes of Tachycineta swallows and their implications for latitudinal patterns of the ‘pace of life’. Gene 546: 104-111. It is reproduced here with permission from the publisher (Elsevier license #3406400803111 to M. Stager) and can be found online at: http://dx.doi.org/10.1016/j.gene.2014.05.019. DJC, RD, and DWW previously published the mitochondrial sequences used here. ZAC contributed to writing and experimental design.

  46  

inferred to be targets of positive selection were disproportionately concentrated in these genes.

Moreover, the positively selected sites exhibited a phylogenetic pattern that could be indicative

of adaptive divergence between “fast” and “slow” lineages. These results suggest that functional

variation in cytochrome b and NADH dehydrogenase could mechanistically contribute to

latitudinal ‘pace of life’ patterns in Tachycineta.

INTRODUCTION

Over the past century, biologists have accumulated a multitude of comparative data on

latitudinal patterns in avian life histories. One of the strongest generalizations emerging from this

large body of work is that tropical birds tend to exhibit long life spans and low reproductive and

developmental rates compared to their temperate counterparts (Lack, 1947; Ricklefs, 1969;

James, 1970; Ricklefs, 1976; Skutch, 1976). More recently, these general patterns have been

summarized as a “slow-fast” continuum in the ‘Pace of Life’ (POL) hypothesis of latitudinal

variation in avian life histories (Ricklefs & Wikelski, 2002; Wiersma et al., 2007b). The

mechanistic underpinnings of latitudinal pace of life patterns are poorly understood, but recent

comparative studies suggest that much of this variation may be linked to the rate of long-term

metabolic energy expenditure (Wiersma et al., 2007a). In general, temperate-zone species tend to

have higher basal metabolic rates, and higher aerobic capacities than closely related tropical

species (Wiersma et al., 2007a,b), both of which may be necessary to support an increased pace

of life in the temperate zone. For example, increased energetic investment may be necessary to

support faster developmental rates or greater provisioning rates for larger clutches in temperate

zone birds. Increased rates of aerobic metabolism can also increase rates of cellular oxidative

damage, which may contribute to the reduced life span of temperate zone species (Vleck et al.,

2007; Monaghan et al., 2008). Given the central role of the mitochondrion in cellular metabolic

processes, mitochondrial variants that are segregating within and among closely related species

may affect both cellular and whole-organism metabolic performance (Ballard & Melvin, 2010;

Toews et al., 2013), and therefore, studies of mitochondrial variation may offer opportunities to

begin to link genetic variation to latitudinal ‘pace of life’ patterns.

The vertebrate mitochondrial genome contains 13 protein-coding genes, all of which are

central to the process of oxidative phosphorylation and aerobic metabolism. Moreover, in

animals, the mitochondrial genome exhibits an elevated substitution rate, smaller effective

  47  

population sizes, and reduced recombination rates compared with the nuclear genome, all of

which facilitate rapid evolution and selective pressures that favor compensatory evolutionary

changes in interacting nuclear genes (Rand et al., 2004; Burton & Barreto, 2012; Osada &

Akashi, 2012). As a result, the mitochondrial genome represents an important potential target of

natural selection in taxa that are distributed across environmental gradients. Indeed, recent

empirical studies have documented a number of putative examples of local adaptation in

mitochondrial variants of species that are distributed along latitudinal and elevational gradients

(Fontanillas et al., 2005; Cheviron & Brumfield, 2009; Hassanin et al., 2009; Ribeiro et al.,

2010; Scott et al., 2011; Toews et al., 2013). In birds specifically, mitochondrial variants have

been linked to differences in kinetic properties of cytochrome c oxidase (Scott et al., 2011), and

the degree of coupling efficiency between substrate oxidation and ADP phosphorylation (Toews

et al., 2013), both of which can influence overall cellular aerobic capacity. Similarly, a wide

variety of mitochondrial variants have been linked to metabolic disorders and the rate of

production of reactive oxygen species in humans (Wallace, 2005).

Lineages with broad latitudinal distributions are well suited for examining the

mechanistic underpinnings of latitudinal POL patterns. Tachycineta is a monophyletic genus of

nine swallow species (Dor et al., 2012) that are continuously distributed from Alaska to Cape

Horn (Figure 3.1A). This exceptionally broad latitudinal distribution makes it possible to make

multiple comparisons of temperate and tropical species (Figure 3.1B). Multiple aspects of

latitudinal variation in the life histories of Tachycineta species conform to POL predictions:

species that occur nearer to the tropics tend to have smaller clutches, higher investment per

offspring, slower developmental rates, and longer nestling periods (Appendix C). Additionally,

preliminary evidence suggests that tropical Tachycineta species also have lower adult

provisioning rates (Ardia et al., unpub. data) and generally higher adult survival rates (Winkler

et al., unpub. data).

Here, we utilized a suite of comparative genomic approaches to analyze complete

mitochondrial genome sequences from each of the nine Tachycineta species. Using a

phylogenetic framework, we specifically tested for evidence of adaptive mitochondrial

divergence among lineages along a “slow-fast” POL continuum. We summarized interspecific

life-history variation as a multivariate POL score to test the prediction that mitochondrial

variation is associated with latitudinal patterns of POL variation in Tachycineta. First, we

  48  

predicted that substitutions at codon positions that are under positive diversifying selection

would disproportionately occur along branches that separate taxa that exhibit the greatest

differences in POL scores (i.e., adaptive divergence at mitochondrial genes should be associated

with divergence between sister taxa that differ strongly in their life histories). Second, we

predicted that the “slowest”, most derived species would also exhibit the largest accumulation of

derived, positively selected substitutions. We show that although purifying selection is the

dominant selective force acting on the mitochondrial genome in Tachycineta, 3 of 13

mitochondrial genes contain regions that exhibit signatures of diversifying selection. These three

genes (ND2, ND5 and CYTB) encode two enzymes (NADH dehydrogenase and cytochrome b)

that are central to cellular metabolic processes. Moreover, the number of positively selected sites

was positively correlated with POL score, demonstrating that derived species with slow life

histories (high POL scores) are characterized by relatively high levels of adaptive variation at

these mitochondrial genes. Taken together, these results suggest that mitochondrial variation

could potentially contribute to metabolic differences that are related to latitudinal pace of life

patterns in Tachycineta.

METHODS

Taxon sampling and mitogenomic data

We downloaded complete mitochondrial genome sequences for each of the nine

Tachycineta species and one outgroup, Progne chalybea, from GenBank (Table 3.1). From these

mitogenomic sequences, we extracted protein-coding regions, and constructed gene-specific

alignments using Sequencher (ver. 5.0).

Summarizing life-history variation

To determine where each Tachycineta species fell on the POL continuum, we gathered

life-history data for traits that were measured in all nine species from the literature (clutch size,

egg size, growth rate, nestling period, adult body mass, and absolute latitude; Appendix C).

Because the POL hypothesis assumes a continuous axis from “slow” to “fast” life histories, we

chose to summarize variation among traits using a principal component analysis, rather than

classifying each species as either temperate or tropical. We performed a principal component

analysis on the z-normalized values for each trait and retained PC components with eigenvalue

  49  

>1 (Kaiser, 1960), which proved true of PC1 alone. We used the resulting PC1 scores, which we

refer to as the POL score, as a continuous variable with which to perform ancestral state

reconstruction. To do this, we reconstructed ancestral states using R (ver. 2.15.0–R Development

Core Team, 2012) with ancestral character estimation (ace) based on restricted maximum

likelihood and a model of Brownian motion (Paradis, 2006). We calculated absolute changes in

POL score between the terminal branches and their ancestral nodes for use in determining

whether substitutions occur disproportionately along branches with greater POL differences. We

then used the POL scores to predict which species are the most derived from the temperate, basal

lineage (Appendix D). Species with the highest POL score are considered “slow” and derived,

and thus are likely to exhibit the highest rate of positive selection; while species with the lowest

POL scores are considered to be “fast” and are likely to exhibit low rates of positive selection.

To test this prediction, we used a generalized linear model with Poisson distribution for count

data (i.e. number of substitutions) and verified that overdispersion was not present. We chose not

to test for phylogenetic signal among these variables as estimates of phylogenetic signal among

datasets of such a limited size (e.g., 9 species) lack power and can therefore be inaccurate

(Boettinger et al., 2012). We therefore did not employ phylogenetically independent contrasts

(PICs), as use of PICs without proper assessment of phylogenetic signal can be inappropriate or

misleading (Revell, 2010).

Detecting selection

We examined each of the thirteen protein-coding genes for evidence of negative selection

using the single likelihood ancestor counting (SLAC) method performed with the HyPhy

software package (Kosakovosky Pond et al., 2005). We estimated non-synonymous to

synonymous substitution (dN/dS) ratios conditioned using neighbor-joining trees, and we

identified gene-specific nucleotide substitution models using a model selection procedure

performed in HyPhy that tests 203 hierarchical time-reversible substitution models. The most

appropriate rate matrix was selected using both likelihood ratio tests (LRTs) and AIC selection.

We identified sites under purifying selection at a significance level of P ≤ 0.05. Although SLAC

can also be used to detect signatures of positive selection, it does not allow for variation in dN/dS

ratios across branches and individual codons. We therefore employed two additional methods

with greater sensitivity for detecting positive selection.

  50  

First, we implemented branch-site models, which allowed us to test for positive selection

at individual codons using a modified branch-site model A test 2 (Zhang et al., 2005) in the

program CODEML (PAML 4.6; Yang, 2007). For each of the thirteen genes, we used LRTs to

compare the null neutral model (model = 2, NSsites = 2, fix_omega = 1, omega = 1) against

alternate models of branch-specific positive selection (model = 2, NSsites = 2, fix_omega = 0,

omega = 1.5). Each branch (terminal and internal) served independently as the foreground

lineage and we tested sites within the foreground lineage for selection. To control for

phylogenetic uncertainty, we conditioned substitution models using four different trees for each

of the models. Input trees were modified from the consensus mitochondrial tree (Cerasale et al.,

2012), the consensus nuclear tree comprised of 16 nuclear loci (Dor et al., 2012), and two

nuclear trees for which we coded polytomies at each of two nodes with posterior probability less

than 0.85 (Figure 3.1B). The mitochondrial tree differs from the nuclear tree in that it is

composed of two distinct clades—a North American/Caribbean clade and a Central/South

American clade—whereas the nuclear tree does not have this distinction, and thus some of the

sister relationships, such as that of T. euchrysea, vary between the nuclear and mitochondrial

phylogenies. Branch lengths were estimated for each gene in CODEML (Yang, 1996). Positive

selection was inferred if the P-value derived from the LRT between the neutral and branch-

specific models was significant (at P ≤ 0.05) based on the chi-squared asymptotic distribution.

Bayes empirical Bayes (BEB; Yang et al., 2005) procedures were performed for branches

exhibiting dN/dS >1 to identify individual sites subject to diversifying selection.

Branch-site models of this kind can be limiting due to the necessary specification of

foreground lineages and the assumption that dN/dS = 1 for all background lineages

(Kosakovosky Pond et al., 2011). We therefore also examined each of the thirteen genes for

signatures of episodic diversifying selection using a mixed effects model of evolution (MEME;

Murrell et al., 2012) performed with HyPhy. This method allows the distribution of the estimated

dN/dS ratio to vary from site to site and from branch to branch and, in so doing, identifies

individual episodes of positive selection that affect only a subset of lineages, which are often

overlooked using traditional methods that assume dN/dS is shared by all sites in the alignment or

that selective pressures are constant throughout time (Murrell et al., 2012). We identified

positively selected sites at a significance level of P ≤ 0.05.

  51  

Ancestral reconstruction

Following detection of selection, we identified amino acid substitutions for each of the

ten species at all sites exhibiting signatures of positive selection and mapped substitutions to the

gene tree (the consensus mitochondrial tree; Cerasale et al., 2012). To do this, we reconstructed

ancestral states with ancestral character estimation (ace) using R (ver. 2.15.0–R Development

Core Team 2012) with three standard models: equal rates, symmetric, and all rates different

(Paradis, 2006). We compared alternative models using LRTs at a significance level of P ≤ 0.05.

We calculated the degree of physicochemical similarity of each substitution using Grantham’s

difference (D), a measure of amino acid exchangeability based on volume, weight, polarity, and

carbon composition (Grantham, 1974). Using D, we then classified amino acid substitutions as

conservative (0-50), moderately conservative (50-100), moderately radical (100-150), or radical

(>150) according to the criteria proposed by Li et al. (1984). Due to our specific interest in sites

that were inferred to be under positive selection and the prohibitively large number of negatively

selected sites, we did not repeat this procedure for sites exhibiting signatures of negative

selection.

Mapping selected residues

To provide insight regarding the physical location of sites under selection within the

protein, we mapped all sites to topologies of the transmembrane segments of electron transport

proteins. We predicted topologies based on the consensus of multiple models (Granseth et al.,

2006; Hessa et al., 2007; Bernsel et al., 2009) using the software TOPCONS

(http://topcons.cbr.su.se/). Topology of these proteins is largely conserved across eukaryotes

(Degli Eposti et al., 1993; Brandt 2006) and so we anticipated that the overall architecture of the

proteins would be conserved at the species level. We therefore randomly chose T. leucorrhoa as

the query sequence. We then mapped selected sites to predicted topologies for genes exhibiting

signatures of both positive and negative selection.

RESULTS

Pace of life (POL) scores, representing life-history variation among Tachycineta species,

show a continuous distribution from “slow” (high POL values) to “fast” (low POL values) with

T. stolzmanni and T. bicolor at each end of the spectrum, respectively (Table 3.1). As predicted,

  52  

we found a significant correlation between the POL score and the number of amino acid

substitutions for a given taxon (Poisson regression, P = 0.05), such that “slow”, derived species

with high POL scores (T. albilinea, T. albiventer, and T. stolzmanni) exhibited the highest degree

of positive selection (Figure 3.2). However, we found equivocal evidence that positively selected

substitutions occur disproportionately along branches separating taxa with the greatest

differences in POL score (Appendix D).

Signatures of negative selection were pervasive across the mitogenomes of Tachycineta

species. We identified 320 negatively selected sites across all 13 protein-coding genes. The

percentage of codons under negative purifying selection within a single mitochondrial gene

averaged 7.5% with COX1 exhibiting the highest (11%) and ATP8 the lowest (3%) rates of

purifying selection (Table 3.2).

Signatures of positive diversifying selection were much less prevalent across the

mitogenome than those of purifying selection and were concentrated in three genes: ND2, ND5

and CYTB. Using branch-specific models, we detected signatures of positive selection in 2 genes,

both of which are subunits of NADH dehydrogenase (Appendix E). Two codons in ND5 were

inferred to be under positive selection within the T. stolzmanni lineage, site 16 (92.5%

probability) and site 545 (55.2%), and this result was robust across tree topologies (Table 3.3).

ND5 also exhibited signatures of positive selection within branch B at site 27 (86.8-88.7%), and

this result was also robust across the three tree topologies to which branch B is common

(Phylogenies 2-4). Additionally, two codons in ND2 exhibited signatures of positive selection

along branch A: sites 222 (99.4%) and 327 (79.5%). This model could only be tested with

Phylogeny 1 as this branch is not common to the other three trees.

Using MEME, we identified episodic positive selection in 2 of 13 genes: CYTB (site 161)

and ND5 (sites 27, 67, 190, and 519; Table 3.2). Thus, 1 of the 10 sites (ND5 27) was identified

by both methods. Additionally, one site each in ND2 and ND4 exhibited signatures of positive

selection that were marginally significant (sites ND2 222, P = 0.07; ND4 192, P = 0.06).

The equal rates model proved best for all instances of ancestral state reconstruction

(Appendix F). Positively selected substitutions produce polarity changes in the resulting amino

acids for 3 of the 5 sites identified by branch-site models and in 2 additional sites of episodic

selection; the remaining 3 substitutions do not result in polarity changes (Figure 3.3). The

physicochemical dissimilarity of the amino acid change was categorized as conservative for 4

  53  

substitutions, moderately conservative for 3 substitutions, and moderately radical for 1

substitution (ND5 519, D = 103; Figure 3.3). When mapped to the phylogeny, the outgroup alone

exhibited differences for sites ND4 192 and ND5 67 (Appendices G & H), and they are therefore

not considered for the remainder of this work.

Consistent with other vertebrate studies (e.g., Whitehead, 2009; Birrell & Hirst 2010),

predicted protein topologies contained 8, 11, and 14 transmembrane regions in CYTB, ND2, and

ND5, respectively. However, mammalian ND5 has been shown to contain an additional one to

two transmembrane helices (da Fonseca et al., 2008; Bridges et al., 2011; Lemay et al., 2013).

All of the sites that exhibited signatures of episodic positive selection in the MEME analyses

were restricted to the predicted loop regions of their respective proteins (Figure 3.4). However, 2

of the sites identified by branch-site models were located wholly within transmembrane regions

(ND2 327 and ND5 16). Sites of negative selection occurred in both loop and transmembrane

regions for all three genes that also exhibited positive selection (Figure 3.4).

DISCUSSION Mitochondrial variation and latitudinal patterns in the ‘pace of life’

Latitudinal variation in ‘pace of life’ (POL) has been well documented in a wide variety

of taxa (Ricklefs & Wikelski, 2002; Wiersma et al., 2007b). However, little is known about the

mechanistic underpinnings of POL variation within and among species. Several lines of evidence

suggest that much of this variation may be functionally linked to differences in rates of metabolic

energy expenditure. Given the vital role of the mitochondrion in cellular metabolism, the

mitochondrial genome therefore represents a suitable target to begin investigations of the genetic

basis underlying life-history variation. Using a genus of broadly distributed passerines, we

identified mitogenomic regions that may contribute to a mechanistic understanding of this

macroecological pattern. Although purifying selection is the dominant selective force acting on

the mitochondrial genome in Tachycineta, we found evidence of diversifying selection in three

mitochondrial genes.

The positively selected sites exhibited a phylogenetic pattern that is consistent with

adaptive divergence between “fast” temperate and “slow” tropical lineages. Derived substitutions

at both positively selected sites in ND2 were shared by the three equatorial species whose life-

history variation is at the “slow” end of the spectrum (T. albilinea, T. albiventer, and T.

  54  

stolzmanni; Figure 3.3). Moreover, substitutions at two positively selected sites within ND5 were

unique to T. stolzmanni alone, which is likely to be the “slowest” of the Tachycineta species

(Table 3.1), exhibiting both the lowest fecundity and lowest growth rate of the genus (Stager et

al., 2012). The remaining sites under positive selection were distributed across the phylogeny

resulting in an overall pattern consistent with the POL hypothesis, (i.e., “slow” lineages were

characterized by more positively selected substitutions than “fast” lineages; Figure 3.2).

The high rate of positively selected substitutions within T. stolzmanni could alternatively

be explained by the narrow distribution and relatively small effective population size of this

lineage. However, this pattern does not hold across the genus: the two island species, T.

euchrysea and T. cyaneoviridis, are characterized by even more restricted range sizes and smaller

populations — listed as vulnerable and endangered, respectively (IUCN, 2013) — yet they do

not show the same elevation in positively selected sites. In fact, T. cyaneoviridis does not exhibit

a single positively selected substitution, yielding this an unlikely mechanism for the interspecific

patterns we have shown here. Likewise, interspecific divergence levels across the mitochondrial

genome are similar among Tachycineta species with both small and large distributions (Cerasale

et al., 2012). This finding is supported by the work of Bazin and colleagues (2006) who have

demonstrated across ~3000 animal species that mitochondrial DNA diversity cannot be predicted

by effective population size, but rather is better explained by adaptive evolution.

The three positively selected genes encode subunits of two separate complexes of the

electron transport chain. Seven of the eight positively selected sites occurred within genes (ND2

and ND5) that encode subunits of NADH dehydrogenase, a large, multimeric enzyme comprising

respiratory chain complex I. This complex has been previously implicated in the adaptive

evolution of the mitogenome in other taxa (41 placental mammals, da Fonseca et al., 2008;

salmon, Garvin et al., 2011; monkeys, Yu et al., 2011; wild donkeys, Luo et al., 2012; herring,

Teacher et al., 2012; hares, Melo-Ferreira et al., 2014). Located in the arm of L-shaped complex

I, subunits ND2, ND4, and ND5 collectively serve as a proton pump for H+ ions across the inner

membrane (Brandt, 2006; Radermacher et al., 2006). These three subunits are linked by the arm

of ND5, allowing for coordinated shifts in proton pumping that in turn drive ATP production

(Efremov et al., 2010; Hunte et al., 2010).

Within NADH subunits, sites exhibiting positive selection largely occurred in the

predicted loop regions of the protein. Regions lacking sites of positive selection may reflect the

  55  

presence of conserved functional domains that are subject to purifying selection (Whitehead,

2009), and we do indeed find evidence of purifying selection in many of the regions that are

devoid of codons under positive selection (Figure 3.4). The low incidence of positively selected

sites in these particular regions of complex I is consistent with previous studies of fish

(Whitehead, 2009) and mammals (da Fonseca et al., 2008). Although one positively selected site

in each of ND2 (site 327) and ND5 (site 27) occurred within the transmembrane region, these

substitutions are physiochemically conservative. The most radical substitutions were restricted to

loop regions (Figures 2.3 & 2.4).

Cytochrome b is an integral component of respiratory chain Complex III (cytochrome bc1

complex), which converts ubiquinol to cytochrome c, thereby generating an electrochemical

gradient across the inner mitochondrial membrane that is utilized for ATP production (Mitchell,

1976; Trumpower, 1990; Hunte et al., 2003). In this process, ubiquinol is first converted to

ubiquinone via oxidation at the Qo binding site and ubiquinone is then reduced to cytochrome c

at the Qi binding site (Gao et al., 2003; Kolling et al., 2003; Crofts, 2004). Both of these sites

are located within the cytochrome b subunit. This process ultimately relies on the capacity of the

Qi site to bind water molecules for the reduction of ubiquinone to ubiquinol (Crofts, 2004).

Because amino acid replacements can result in regional changes to hydrophobicity and structure

within the protein, amino acid replacements within CYTB have the potential to alter the

efficiency of this biding site. Replacements at cytbI7T characteristic of human mitochondrial

haplogroup H likely enhance the binding of water at Qi, and have been linked to patterns of

increased longevity for this group (Beckstead et al., 2009). Similarly, mutations to key residues

at Qo in yeast alter this binding site and have resulted in decreased catalysis efficiency, as well as

increased damaging, oxygen radical-producing bypass reactions (Wenz et al., 2007). However,

the isoleucine to leucine substitution at CYTB 161 is among the most conservative amino

replacements that occur at positively selected residues (D = 5), making the functional

significance of this substitution in Tachycineta unclear.

Model assumptions and consistency of inferences across methods

Our results differ from a previous investigation of mitogenomic variation in Tachycineta

in which positive selection was not detected (Cerasale et al., 2012). This disparity is likely due to

our use of methods that are more sensitive to detecting lineage-specific selection. Cerasale et al.

  56  

used a two-rate fixed effects likelihood (FEL) approach, whereas we have employed a mixed

effects model of evolution (MEME). The two methods differ in that MEME allows dN/dS to

vary across both sites and branches (such that some branches may exhibit positive selection at

the same time that others exhibit negative selection). Conversely, FEL assumes constant

selection across time (prohibiting variation across branches), thereby limiting its power to detect

selection occurring in only a single or few lineages (Murrell et al., 2012). We also implemented

a branch-specific model of positive selection in CODEML to control for phylogenetic

uncertainty. Cerasale et al. did not employ branch-site models.

CODEML inferences were largely robust to tree topology but only a subset of the sites

inferred to be under selection by CODEML were also identified by MEME: one site was

detected by both methods (ND5 27) and another (ND2 222) was found to be marginally

significant. The differences can likely be attributed to differences in the underlying assumptions

of the two methods (namely the use of one vs. multiple dN/dS ratios across branches), and also

to the constraint of testing one focal branch at a time using branch-site tests. The latter issue

prevented detection of sites under positive selection at CYTB 161, which exhibits a parallel

amino acid change in two non-sister taxa (Figure 3.3).

CONCLUSIONS & FUTURE DIRECTIONS

Taken together, the results presented here suggest that the mitochondrial genome may be

a promising candidate locus to begin to tie molecular genetic variation to latitudinal patterns in

the pace of life. We have found evidence of positive selection acting on DNA sequence variation

in the mitogenome of a widespread avian genus that exhibits latitudinal patterns of life-history

variation that are consistent with POL expectations. While we found weak support for our

prediction that branches separating taxa with greater POL variation also exhibit higher rates of

positive diversifying selection, “slow” species did harbor the largest accumulation of derived,

positively selected substitutions. We speculate that these substitutions are associated with

functional metabolic differences among species and hope that our work will motivate future

studies to characterize whole-organism metabolic rates, energy expenditure, and cellular

metabolic processes within Tachycineta. Because the metabolic machinery is also encoded by

the nuclear genome, similar investigations should be performed as sequences become available

for the adjacent nuclear subunits: these could likely be targets of intergenomic coevolution

  57  

(Burton & Barreto, 2012) and responsible for some of the remaining POL variation among

species.

At the biochemical level, variation in metabolic rates is manifest through differences in

the expression of genes that encode the machinery of cellular metabolism and the kinetic

properties of these metabolic enzymes. Thus, although the positively selected amino acid

substitutions we have identified here may contribute to functional variation in NADH

dehydrogenase and cytochrome b, variation in whole-organism metabolic rate can also be driven

through regulatory mechanisms. Indeed, recent studies of metabolic performance in deer mice

(Peromyscus maniculatus) have demonstrated that elevated thermogenic performance is

associated with large-scale upregulation of genes that participate in the beta-oxidation of lipids

and oxidative phosphorylation, and these transcriptomic changes paralleled differences in the

activities of metabolic enzymes that influence flux through these same pathways (Cheviron et

al., 2012; 2014). Therefore, studies of regulatory variation in relation to pace of life expectations

are likely to be particularly fruitful, and would complement our analyses of DNA sequence

variation. Similarly, detailed studies of mitochondrial respiration rates coupled with

measurements of whole-organism metabolic rates could provide insights into the functional

consequences of putatively adaptive amino acid substitutions (Storz & Wheat, 2010; Toews et

al., 2013). Collectively, these studies may begin to elucidate the genetic and physiological

mechanisms that underlie an important broad-scale macrophysiological pattern – latitudinal

variation in the ‘pace of life’ (Gaston et al., 2009).

ACKNOWLEDGEMENTS

We are thankful to Al Roca and two anonymous reviewers for comments made to an

earlier draft of this manuscript. This research was supported by start-up funds from the

University of Illinois to ZAC.

LITERATURE CITED Ballard, J.W.O., Melvin, R.G., 2010. Linking the mitochondrial genotype to the organismal

phenotype. Mol. Ecol. 19, 1523-1539.

Bazin, E., Glemin, S., Galtier, N., 2006. Population size does not influence mitochondrial genetic

diversity in animals. Science 312, 570-572.

  58  

Beckstead, W.A., Ebbert, M.T.W., Rowe, M.J., McClellan, D.A., 2009. Evolutionary pressure on

the mitochondrial cytochrome b is consistent with a role of CytbI7T affecting longevity

during caloric restriction. PLoS One 4, e5836.

Bernsel, A., Viklund, H., Hennerdal, A., Elofsson, A., 2009. TOPCONS: consensus prediction of

membrane protein topology. Nucleic Acids Res. 37, W465-W468.

Birrell, J.A., Hirst, J., 2010. Truncation of subunit ND2 disrupts the threefold symmetry of the

antiporter-like subunits in complex I from higher metazoans. FEBS Lett. 584: 4247-4252.

Boettinger, C., Coop, G., Ralph, P., 2012. Is your phylogeny informative? Measuring the

power of comparative methods. Evolution 66, 2240-2251.

Brandt, U., 2006. Energy converting NADH: Quinone oxidoreductase (Complex I). Ann. Rev.

Biochem. 75, 69-92.

Bridges, H.R., Birrell, J.A., Hirst, J., 2011. The mitochondrial-encoded subunits of respiratory

complex I (NADH:ubiquinone oxidoreductase): identifying residues important in

mechanism ���and disease. Biochem. Soc. Trans. 39, 799-806.

Burton, R.S., Barreto, F.S., 2012. A disproportionate role for mtDNA in Dobzhansky-Muller

incompatibilities? Mol. Ecol. 21, 4294-4957.

Cerasale, D., Dor, R., Winkler, D.W., Lovette, I.J., 2012. Phylogeny of the Tachycineta genus of

New World swallows: insights from complete mitochondrial genomes. Mol. Phylogenet.

Evol. 63, 64-71.

Cheviron, Z.A., Bachman, G.C., Connaty, A.D., McClelland, G.B., Storz, J.F., 2012. Regulatory

changes contribute to the adaptive enhancement of thermogenic capacity in high-altitude

deer mice. Proc. Nat. Acad. Sci. U.S.A. 109, 8635-8640.

Cheviron, Z.A., Brumfield, R.T., 2009. Migration-selection balance and local adaptation of

mitochondrial haplotypes in Rufous-collared Sparrows (Zonotrichia capensis) along an

elevational gradient. Evolution 63, 1593-1605.

Cheviron, Z.A., Connaty, A.D., McClelland, G.B., Storz, J.F., 2014. Functional genomics of

adaptation to hypoxic cold-stress in high-altitude deer mice: transcriptomic plasticity and

thermogenic performance. Evolution 68, 48-62.

Crofts, A.R., 2004. The cytochrome bc1 complex: function in the context of structure. Annu.

Rev. Physiol. 66, 689-733.

da Fonseca, R.R., Johnson, W.E., O'Brien, S.J., Ramos, M.J., Antunes, A., 2008. The adaptive

  59  

evolution of the mammalian mitochondrial genome. BMC Genomics 9, 119.

Degli Eposti, M., De Vries, S., Crimi, M., Ghelli, A., Patarnello, T., Meyer, A., 1993.

Mitochondrial cytochrome b: evolution and structure of the protein. Biochim. Biophys.

Acta 1143, 243-271.

Dor, R., Carling, M.D., Lovette, I.J., Sheldon, F.H., Winkler, D.W., 2012. Species trees for the

tree swallows (genus Tachycineta): an alternative phylogenetic hypothesis to the

mitochondrial gene tree. Mol. Phylogenet. Evol. 65, 317-322.

Efremov, R.G., Baradaran, R., Sazanov, L.A., 2010. The architecture of respiratory complex I.

Nature 465, 441-447.

Fontanillas, P., Depraz, A., Giorgi, M.S., Perrin, N.R., 2005. Nonshivering thermogenesis

capacity associated with mitochondrial DNA haplotypes in the greater white-toothed

shrew, Crocidura russula. Mol. Ecol. 14, 661–670.

Gao, X., Wen. X., Esser, L., Quinn, B., Yu, L., Yu, C. et al., 2003. Structural basis for the

quinone reduction in the bc1 complex: a comparative analysis of crystal structures of

mitochondrial cytochrome bc1 with bound substrate and inhibitors at the Qi site.

Biochemistry 42, 9067-9080.

Garvin, M.R., Bielawski, J.P., Gharrett, A.J., 2011. Positive Darwinian selection in the piston

that powers proton pumps in complex I of the mitochondria of Pacific salmon. PLOS One

6, e24127.

Gaston, K.J., Chown, S.L., Calosi, P., Bernardo, J., Bilton, D.T., Clarke, A. et al., 2009.

Macrophysiology: A conceptual re-unification. Am. Nat. 174, 595-612.

Granseth, E., Viklund, H., Elofsson, A., 2006. ZPRED: Predicting the distance to the membrane

center for residues in alpha-helical membrane proteins. Bioinformatics 22: E191-E196.

Grantham, R., 1974. Amino acid difference formula to help explain protein evolution. Science

185, 862-864.

Hassanin, A., Ropiquet, A., Couloux, A., Cruaud, C., 2009. Evolution of the mitochondrial

genome in mammals living at high altitude: new insights from a study of the tribe Caprini

(Bovidae, Antilopinae). J. Mol. Evol. 68, 293-310.

Hessa, T., Meindl-Beinker, N.M., Bernsel, A., Kim, H., Sato, Y., Lerch-Bader, M. et al., 2007.

Molecular code for transmembrane-helix recognition by the Sec61 translocon. Nature

450, 1026-U1022.

  60  

Hunte, C., Palsdottir, H., Trumpower, B.L., 2003. Protonmotive pathways and mechanisms in

the cytochrome bc1 complex. FEBS Lett. 545, 39-46.

Hunte, C., Zickermann, V., Brandt, U., 2010. Functional modules and structural basis of

conformational coupling in mitochondrial complex I. Science 329, 448–451.

IUCN, 2013. The IUCN Red List of Threatened Species. Version 2013.2. http://www.

iucnredlist.org (Downloaded on 24 March 2014).

James, F. C., 1970. Geographic size variation in birds and its relation to climate. Ecology 51,

365-390.

Kaiser, H. F. ,1960. The application of electronic computers to factor analysis. Educational and

Psychological Measurement, 20, 141-151.

Kolling, D.R., Samoilova, R.I., Holland, J.T., Berry, E.A., Dikanov, S.A., Crofts, A.R., 2003.

Exploration of ligands to the Qi site semiquinone in the bc1 complex using high-

resolution EPR. J. Biol. Chem. 278, 39747–39754.

Kosakovosky Pond, S.L., Frost, S.D.W., Muse, S.V., 2005. HyPhy: hypothesis testing using

phylogenies. Bioinformatics 21, 676–679.

Kosakovosky Pond, S.L., Murrell, B., Fourment, M., Frost, S.D.W., Delport, W., Schleffer, K.,

2011. A random effects branch-site model for detecting episodic diversifying selection.

Mol. Biol. Evol. 28, 3033-3043.

Lack, D., 1947. The significance of clutch-size. Ibis 89, 302-352.

Lemay, M.A., Philippe, H., Lamb, C.T., Robson, K.M., Russello, M.A., 2013. Novel genomic

resources for a climate change sensitive mammal: characterization of the American pika

transcriptome. BMC Genomics 14, 311.

Li, W.H., Wu, C.I., Luo, C.C., 1984. Nonrandomness of point mutation as reflected in nucleotide

substitutions in pseudogenes and its evolutionary implications. J. Mol. Evol. 21, 58–71.

Luo, Y., Chen, Y., Liu, F.Y., Gao, Y.Q., 2012. Mitochondrial genome of Tibetan wild ass

(Equus kiang) reveals substitutions in NADH which may reflect evolutionary adaptation

to cold and hypoxic conditions. Asia Life Sci. 21, 1-11.

Melo-Ferreira, J., Vilela, J., Fonseca, M., da Fonseca, R., Boursot, P., Alves, P., 2014. The

elusive nature of adaptive mitochondrial DNA evolution of an arctic lineage prone to

frequent introgression. Genome Biol. Evol. 6, 886-896.

Mitchell, P., 1976. Possible molecular mechanisms of the protonmotive function of cytochrome

  61  

systems. J. Theor. Biol. 62, 327-367.

Monaghan, P., Charmantier, A., Nussey, D.H. & Ricklefs, R.E. 2008. The evolutionary ecology

of senescence. Funct. Ecol. 22, 371-378.

Murrell, B., Wertheim, J.O., Moola, S., Weighill, T., Scheffler, K., Kosakovosky Pond, S.L.,

2012. Detecting diversifying sites subject to episodic diversifying selection. PLoS Genet.

8, 1002764.

Osada, N. & Akashi, H. 2012. Mitochondrial-nuclear interactions and accelerated compensatory

evolution: evidence from the primate cytochrome c oxidase complex. Mol. Biol. Evol.

29, 337-346.

Paradis, E. 2006. Analyses of Phylogenetics and Evolution with R. Springer, New York.

Radermacher, M., Ruiz, T., Clason, T., Benjamin, S., Brandt, U., Zickermann, V., 2006. The

three-dimensional structure of complex I from Yarrowia lioplytica: a highly dynamic

enzyme. J. Struct. Biol. 154, 269-279.

Rand, D.M., Haney, R.A., Fry, A.J., 2004. Cytonuclear coevolution: the genomics of

cooperation. Trends Ecol. Evol. 19, 645-653.

Revell, L. 2010. Phylogenetic signal and linear regression on species data. Methods Ecol. Evol.

1, 319-329.

Ribeiro, A., Lloyd, P., Bowie, R.C.K., 2011. A tight balance between natural selection and gene

flow in a southern African arid-zone endemic bird. Evolution 65, 3499-3514.

Ricklefs, R.E., 1969. The nesting cycle of songbirds in tropical and temperate regions. Living

Bird 8, 165-175.

Ricklefs, R.E., 1976. Growth rates of birds in the New World tropics. Ibis 118, 179-207.

Ricklefs, R.E., Wikelski, M., 2002. The physiology/life-history nexus. Trends Ecol. Evol. 17,

462-468.

Scott, G.R., Schulte, P.B., Egginton, S., Scott, A.L.M., Richards, J.G., Milsom, W.K., 2011.

Molecular evolution of cytochrome c oxidase underlies high-altitude adaptation in the

Bar-headed Goose. Mol. Biol. Evol. 28, 351-363.

Skutch, A.F. 1976. Parent birds and their young. Austin: University of Texas Press.

Stager, M., Lopresti, E., Angulo, F., Ardia, D.R., Caceres, D., Cooper, C.B. et al., 2012.

Reproductive biology of a narrowly endemic Tachycineta swallow in dry, seasonal forest

in coastal Peru. Ornitol. Neotrop. 23, 95-112.

  62  

Storz, J. F., Wheat, C.W., 2010. Integrating evolutionary and functional approaches to infer

adaptation at specific loci. Evolution 64, 2489-2509.

Teacher, A.G.F., Andre, C., Merila, J., Wheat, C., 2012. Whole mitochondrial genome scan for

population structure and selection in the Atlantic herring. BMC Evol. Biol. 12, 248.

Toews, P.L., Mandic, M., Richards, J.G., Irwin, D.E., 2013. Migration, mitochondria, and the

yellow-rumped warbler. Evolution:  doi:10.1111/evo.12260.

Trumpower, B.L., 1990. The protonmotive Q cycle. Energy transduction by coupling of proton

translocation to electron transfer by the cytochrome bc1 complex. J. Biol. Chem. 265,

11409-11412.

Vleck, C. M., Haussmann, M. F., Vleck, D., 2007. Avian senescence: underlying mechanisms.

J. Ornithol. 148, S611-S624.

Wallace, D.C., 2005. A mitochondrial paradigm of metabolic and degenerative diseases, aging,

and cancer: a dawn for evolutionary medicine. Annu. Rev. Genet. 39, 359–407.

Wenz, T., Covian, R., Hellwig, P., Macmillan, F., Meunier, B., Trumpower, B.L., Hunte, C.,

2007. Mutational analysis of cytochrome b at the ubiquinol oxidation site of yeast

complex III. J. Biol. Chem. 282, 3977–3988.

Whitehead, A., 2009. Comparative mitochondrial genomics within and among species of

Killifish. BMC Evol. Biol. 9, 11.

Wiersma, P., Chappell, M., Williams, J.B., 2007a. Cold- and exercise-induced peak metabolic

rates in tropical birds. P. Natl. Acad. Sci-Biol. 104, 20866-20871.

Wiersma, P., Munoz-Garcia, A., Walker, A., Williams, J.B., 2007b. Tropical birds have a slow

pace of life. P. Natl. Acad. Sci-Biol. 104, 9340-9345.

Yang, Z., 1996. Maximum-likelihood models for combined analyses of multiple sequence data.

J. Mol. Evol. 42, 587-596.

Yang, Z., 2007. PAML 4: a program package for phylogenetic analysis by maximum likelihood.

Mol. Biol. Evol. 24, 1586-1591.

Yang, Z., Wong, W.S.W., Nielsen, R., 2005. Bayes empirical Bayes inference of amino acid

sites under positive selection. Mol. Biol. Evol. 22, 1107-1118.

Yu, L., Wang, X., Ting, N., Zhang, Y., 2011. Mitogenomic analysis of Chinese snub-nosed

monkeys: evidence of positive selection in NADH dehydrogenase genes in high-altitude

adaptation. Mitochondrion 11, 497-503.

  63  

Zhang, J., Nielsen, R., Yang, Z., 2005. Evaluation of an improved branch-site likelihood method

for detecting positive selection at the molecular level. Mol. Biol. Evol. 22, 2472-2479.

  64  

CHAPTER 3 TABLES & FIGURES

TABLE 3.1. Pace of life variation among the nine Tachycineta species. Variation in life history traits was summarized using a principal component analysis on the z-normalized values. The resulting PC1 scores (POL score) represent the degree of variation among species; positive scores represent “slow” species and negative scores represent “fast”. GenBank accession numbers of mitochondrial sequences for each species and one outgroup used in this study are also indicated (these data were originally reported by Cerasale et al., 2012). Life history data are presented in Appendix C.

Species Accession # POL score T. bicolor JQ071614 -2.81 T. leucorrhoa JQ071621 -1.95 T. meyeni JQ071622 -0.76 T. thalassina JQ071615 -0.69 T. cyaneoviridis JQ071617 0.38 T. albilinea JQ071619 0.82 T. euchrysea JQ071616 0.90 T. albiventer JQ071620 1.21 T. stolzmanni JQ071618 2.90 Progne chalybea JQ071623

  65  

TABLE 3.2. Proportion of sites undergoing positive episodic (Pos) or negative (Neg) selection (identified with MEME and SLAC, respectively) for each of thirteen Tachycineta mitochondrial genes. Sites identified at P ≤ 0.05.

Gene Pos. Neg. ATP6 0 0.044 ATP8 0 0.036 COX1 0 0.112 COX2 0 0.070 COX3 0 0.080 CYTB 0.003 0.097 ND1 0 0.111 ND2 0 0.058 ND3 0 0.034 ND4 0 0.089

ND4L 0 0.071 ND5 0.007 0.091 ND6 0 0.076

  66  

TABLE 3.3. Genes inferred to be under positive selection using branch-site models with CODEML. Trees and branches correspond to Figure 3.1B. Degrees of freedom = 1.

Gene Branch Tree lnLNeutral lnLSelection LRT P

ND2 A 1 -3488.26 -3486.06 4.41 0.0357

ND5 B 2 -5791.16 -5787.08 8.15 0.0043

ND5 B 3 -5849.67 -5845.53 8.27 0.0040

ND5 B 4 -6005.78 -6000.92 9.72 0.0018

ND5 T. stolzmanni 1 -5767.91 -5770.08 4.33 0.0374

ND5 T. stolzmanni 2 -5789.65 -5787.52 4.25 0.0392

ND5 T. stolzmanni 3 -5849.29 -5847.01 4.56 0.0327

ND5 T. stolzmanni 4 -6004.54 -6003.22 2.64 0.1040

  67  

FIGURE 3.1. (A) Color-coded distributions of the nine Tachycineta species; colors correspond to those used in phylogenies. Dots indicate approximate localities of specimen origin for each sequence. Gray line indicates the equator. (B) The four phylogenies employed in CODEML analyses are shown: the consensus mitochondrial (Phylogeny 1; Cerasale et al., 2012), consensus nuclear (Phylogeny 2; Dor et al., 2012), and nuclear trees with polytomy at node 17 (Phylogeny 3) and at node 14 (Phylogeny 4). Nodes 14 and 17 indicated by black dots in Phylogeny 2. Branches A and B are highlighted in red. Branch lengths not to scale.

A!

B

  68  

FIGURE 3.2. Number of substitutions exhibited by each species as a function of their respective pace of life (POL) scores. Line fit by Poisson regression, p-value shown. Colors correspond to Figure 3.1.

-3 -2 -1 0 1 2 3

01

23

45

POL score

Num

ber o

f sub

stitu

tions

p = 0.046

Fast Slow

  69  

FIGURE 3.3. Ancestral state reconstruction of polarity changes resulting from amino acid substitutions for 8 residues exhibiting positive selection: sites (A) CYTB 161; (B) ND2 222; (C) ND2 327; (D) ND5 16; (E) ND5 27; (F) ND5 190; (G) ND5 519; (H) and ND5 545. Amino acids are indicated by color and their polarity specified: nonpolar (np) or polar (p). Pie charts represent the proportional likelihood of particular state changes at each of the ancestral nodes. Grantham’s difference (D) shown for each amino acid pair within Tachycineta (Grantham, 1974).

T. stolzmanniT. albiventer

T. albilinea

T. meyeniT. leucorrhoa

T. bicolor

T. cyaneoviridisT. euchrysea

T. thalassina

Progne chalybeaCYTB 161a.

D = 5

Isoleucine (np)Leusine (np)

T. stolzmanniT. albiventer

T. albilinea

T. meyeniT. leucorrhoa

T. bicolor

T. cyaneoviridisT. euchrysea

T. thalassina

Progne chalybeaND2 222b.

D = 69

Threonine (p)Valine (np)

T. stolzmanniT. albiventer

T. albilinea

T. meyeniT. leucorrhoa

T. bicolor

T. cyaneoviridisT. euchrysea

T. thalassina

Progne chalybeaND2 327c.

D = 29

Isoleucine (np)Valine (np)

T. stolzmanniT. albiventer

T. albilinea

T. meyeniT. leucorrhoa

T. bicolor

T. cyaneoviridisT. euchrysea

T. thalassina

Progne chalybeaND5 16d.

D = 69

Methionine (np)Threonine (p)Valine (np)

T. stolzmanniT. albiventer

T. albilinea

T. meyeniT. leucorrhoa

T. bicolor

T. cyaneoviridisT. euchrysea

T. thalassina

Progne chalybeaND5 27e.

D = 5

Isoleucine (np)Leucine (np)

T. stolzmanniT. albiventer

T. albilinea

T. meyeniT. leucorrhoa

T. bicolor

T. cyaneoviridisT. euchrysea

T. thalassina

Progne chalybeaND5 190f.

D = 37

Tryptophan (np)Tyrosine (p)

T. stolzmanniT. albiventer

T. albilinea

T. meyeniT. leucorrhoa

T. bicolor

T. cyaneoviridisT. euchrysea

T. thalassina

Progne chalybeaND5 519g.

D = 103

Proline (np)Arginine (p)

T. stolzmanniT. albiventer

T. albilinea

T. meyeniT. leucorrhoa

T. bicolor

T. cyaneoviridisT. euchrysea

T. thalassina

Progne chalybeaND5 545h.

D = 56

Serine (p)Glycine (np)

A" B"

C"

E"

G

D

F"

H

  70  

FIGURE 3.4. Predicted topologies of transmembrane domains for genes exhibiting signatures of selection. The x-axis represents amino acid sites for each protein, and transmembrane regions are shown in gray while loop regions are shown in white. Sites under negative selection identified with SLAC are indicated with triangles. Sites that are inferred to be targets of positive selection identified with MEME are indicated with circles, and closed circles represent amino acid substitutions resulting in a change in polarity, whereas open circles represent substitutions that do not change polarity. Sites inferred to be under positive selection in the CODEML analysis are indicated with squares (closed = polarity change; open = no change in polarity). Colors correspond to the different branches on which they were identified: red = T. stolzmanni; green = branch A; blue = branch B (branches correspond to Figure 3.1).

CYTBMethod

SLACMEMEPAML

BranchT. stolzmanniAB

PolarityChangeNo change

ND2

0 100 200 300 400 500 600

ND5

Amino Acid Sites

  71  

CHAPTER 4

CONCLUSION

Together, these investigations elucidate two distinct molecular mechanisms underlying

variation in avian metabolic performance. Regulatory changes can act on relatively short time

periods, enabling organisms to respond to rapid environmental changes within the lifetime of an

individual. Junco thermogenic flexibility manifested from simultaneous changes in the

expression of genes within hierarchical regulatory networks in response to temperature. This

transcriptomic variation resulted in rapid physiological modifications within individuals. In

comparison, amino acid substitutions occur over the course of generations. Differences in the

mitochondrial sequences of the genes encoding the machinery of cellular metabolism could

result in canalized differences in whole-organism aerobic performance among Tachycineta

swallow species. Here sites of positive selection were concentrated in species with ‘slow’ life

histories that may be associated with low metabolic rates.

While our investigations focused on each mechanism in isolation, these two molecular

mechanisms could in fact act in concert to generate phenotypic variation among individuals. An

individual possesses a single genome; therefore phenotypic flexibility cannot be driven by

sequence variation. In contrast, it is likely that regulatory changes also contribute to interspecific

differences in aerobic performance, though we did not examine the presence of transcriptomic

variation among Tachycineta species here. Future studies investigating the mechanisms

underlying physiological adaptation among individuals should therefore integrate methodologies

to simultaneously characterize both regulatory and sequence variation. Quantifying

transcriptomic variation among Tachycineta species, for instance, would allow us to determine

the degree to which metabolic variation results from standing sequence variation in

mitochondrial protein-coding genes vs. differential regulation of metabolic genes. Likewise,

further studies among junco populations could illuminate whether these regulatory mechanisms

are conserved across evolutionary lineages, and genomic sequencing could help to uncover

mechanisms of potential phenotypic canalization.

Given that we find evidence for each of these mechanisms here, it begs the question:

Which is a stronger driver of interindividual differences in avian aerobic performance? This

  72  

question is particularly interesting in light of the diverse habitats that avian lineages inhabit

around the globe, encompassing a wide range of environmental conditions that can exert an

equally wide breadth of selective pressures on metabolic performance. With the aid of recent

advances in genomic technology, the answer is within our reach.

  73  

APPENDIX A. Effect of acclimation treatment on pectoralis size (Mp).

Cold LD Cold SD Warm LD Warm SD

1.2

1.3

1.4

1.5

1.6

1.7

Mp

(g)

  74  

APPENDIX B. Significant enrichment results for all transcriptional modules significantly associated with thermogenic performance or pectoralis size (P ≤ 0.05). See supplemental file “Appendix_B.xls”.

  75  

APPENDIX C. Life-history traits for each of the nine Tachycineta species. Data for clutch size (Clutch), ratio of egg volume to adult body mass (Vegg:Mad), growth rate (Growth), and nestling period (Nest Per.) were gathered from the literature with corresponding absolute latitudes.

Species Latitude Clutch Vegg:Mad Growth Nest Per. T. bicolor1,2,3,4 42.44 5.4 0.09 0.50 21.0 T. leucorrhoa5,6 35.57 4.9 0.08 0.46 23.3 T. meyeni7,8 54.80 3.8 0.09 0.43 26.0 T. thalassina9,10,11 48.70 4.3 0.11 0.41 24.5 T. cyaneoviridis12 26.67 3.0 0.10 0.36 22.8 T. euchrysea13 18.85 3.0 0.12 0.51 25.6 T. albilinea14,15,16 9.16 4.0 0.11 0.41 25.0 T. albiventer17,18 7.45 3.0 0.10 0.37 24.7 T. stolzmanni19 6.48 2.7 0.12 0.36 28.3

1 Ardia, D.R., Wasson, M.F., Winkler, D.W., 2006. Individual quality and food availability determine yolk and egg mass and egg composition in tree swallows Tachycineta bicolor. J. Avian Biol. 37, 252-259. 2 McCarty, J. P., 2001. Variation in growth of nestling Tree Swallows across multiple temporal and spatial scales. Auk 118, 176–190. 3

Palacios, M.G., Winkler, D.W., Klasing, K.K., Hasselquist, D., Vleck, C.M., 2011. Consequences of immune system aging in nature: a study of immunosenescence costs in free-living Tree Swallows. Ecology 92, 952-966. 4

Winkler, D.W., Hallinger, K.K., Ardia, D.R., Robertson, R.J., Stutchbury, B.J., Cohen, R.R., 2011. Tree Swallow (Tachycineta bicolor), The Birds of North America Online (A. Poole, Ed.). Ithaca: Cornell Lab of Ornithology; retrieved February 27, 2014 from Birds of North America Online: http://bna.birds.cornell.edu/bna/species/011. 5

Barrionuevo, M., Bulit, F., Massoni, V., 2010. Variables que afectan el peso de los huevos en la golondrina ceja blanca (Tachycineta leucorrhoa). Hornero 25, 1-7. 6 Massoni, V., Bulit, F., Reboreda, J.C., 2007. Breeding biology of the White-rumped Swallow Tachycineta leucorrhoa in the Buenos Aires province, Argentina. Ibis 149, 10-17. 7

Liljesthröm, M., 2011. Biología reproductiva de la Golondrina Patagónica Tachycineta meyeni en Ushuaia, Tierra del Fuego. Tesis Doc., Univ. Buenos Aires, Buenos Aires, Argentina.8 Liljesthröm, M., Schiavini, A., Reboreda, J.C., 2012. Time of breeding and female condition affect chick-growth in the Chilean Swallow (Tachycineta meyeni). Emu 112, 157-161. 9 Edson, J. M., 1943. A study of the Violet-green Swallow. Auk 60, 396–403. 10 Turner, A., Rose, C. 1989. A handbook to the swallows and martins of the world. Christopher Helm, London, UK. 11 Winkler, D.W., Ringelman, K., Dunn, P., Whittingham, L., Hussell, D., Clark, R., et al., 2014. Latitudinal variation in clutch-size laydate regression in Tachycineta swallows: effects of food supply or demography? Ecography 37. DOI: 10.1111/j.1600-0587.2013.00458.x 12 Allen, P. E., 1996. Breeding biology and natural history of the Bahama Swallow. Wilson Bull. 108: 480–495. 13 Proctor, C. J., 2013. Golden Swallow (Tachycineta euchrysea), Neotropical Birds Online (T. S. Schulenberg, Ed). Ithaca: Cornell Lab of Ornithology; retrieved February 27, 2014 from Neotropical Birds Online: http://neotropical.birds.cornell.edu/portal/species/overview?p_p_spp=523596. 14 Dyrcz, A., 1984. Breeding biology of the Mangrove Swallow Tachycineta albilinea and the Grey- breasted Martin Progne chalybea at Barro Colorado Island, Panama. Ibis, 126, 59-66. 15 Ricklefs, R. E., 1968. Patterns of growth in birds. Ibis 110, 419–451. 16 Strauch, J.C., 1977. Further bird weights from Panama. Bull. Br. Orn. Club 97, 61-65. 17 Stager, M., Rakhimberdiev, E., Ardia, D.R., Cooper, C.B., Rivers, J., Liljesthrӧm, M. et al., in revision. Interspecific variation in growth rates of nestling Tachycineta. 18 Struve, S., Pérez-Emán, J., Stager, M., Taylor, N., Ardia, D.R., Cooper, C.B. et al., 2012. First nest-box study of Tachycineta albiventer in the llanos of Venezuela: preliminary results. Proceedings of the IX Congreso de Ornitología Neotropical, Cusco, Peru. 19 Stager, M., Lopresti, E., Angulo, F., Ardia, D.R., Caceres, D., Cooper, C.B. et al., 2012. Reproductive biology of a narrowly endemic Tachycineta swallow in dry, seasonal forest in coastal Peru. Ornitol. Neotrop. 23, 95-112.

  76  

APPENDIX D. Pace of life (POL) score ancestral state reconstruction using each the consensus mitochondrial tree (A) and the consensus nuclear tree (B). Numbers indicate the POL score for the associated node. Taxa names are color-coded in accordance with Figure 3.1. A

B

T stolzmanni

T albiventer

T albilinea

T meyeni

T leucorrhoa

T bicolor

T cyaneoviridis

T euchrysea

T thalassina

2.898

1.216

0.823

-0.757

-1.954

-2.812

0.378

0.9

-0.693

-0.252

0.09

0.667

0.917

-1.015

-0.386

-0.218

-0.064

T albilinea

T albiventer

T bicolor

T cyaneoviridis

T euchrysea

T leucorrhoa

T meyeni

T stolzmanni

T thalassina

0.823

1.216

-2.812

0.378

0.9

-1.954

-0.757

2.898

-0.693-0.39

-0.135

-0.003

0.509

0.947

0.52

-0.918

-0.141

b.

  77  

APPENDIX E. Branch-specific models of positive selection for each of the 4 trees. All branches examined for positive selection using branch-site model A test 2 are shown by gene with log-likelihoods for both neutral (lnLneutral) and branch-specific models (lnLselection), likelihood ratio test (LRT) values, and P-values (P). Terminal branches denoted by species name; internal branches denoted by letter, corresponding to Appendix I. Degrees of freedom = 1. See supplemental file “Appendix_E.xls”.

  78  

APPENDIX F. Ancestral reconstruction methods results. Log-likelihood values for each of the three models used: equal rates (ER), symmetric (SYM), and all rates different (ARD). The P-value represents the chi-square distribution for the LRT between the ER and ARD models.

Gene Site ER SYM ARD P CYTB 161 -4.808 -4.808 -4.143 0.249 ND2 222 -4.737 -4.737 -4.746 0.892 ND2 327 -4.737 -4.737 -4.746 0.892 ND4 192 -2.284 -2.284 -1.364 0.175 ND5 16 -5.651 -5.395 -4.663 0.160 ND5 27 -3.961 -3.961 -3.765 0.532 ND5 67 -2.284 -2.284 -1.364 0.175 ND5 190 -4.190 -4.190 -4.044 0.589 ND5 519 -5.302 -5.302 -4.311 0.159 ND5 545 -3.539 -3.539 -2.558 0.161

  79  

APPENDIX G. Ancestral state reconstruction of polarity changes resulting from amino acid substitutions for ND4 site 192. Amino acids are indicated by color and their polarity specified: nonpolar (np) or polar (p). Pie charts represent the proportional likelihood of particular state changes at each of the ancestral nodes. Grantham’s difference (D) shown for each amino acid pair within Tachycineta (Grantham, 1974).  

T. stolzmanni

T. albiventer

T. albilinea

T. meyeni

T. leucorrhoa

T. bicolor

T. cyaneoviridis

T. euchrysea

T. thalassina

Progne chalybea

ND4 192

D = 59

Threonine (p)Glycine (np)

  80  

APPENDIX H. Ancestral state reconstruction of polarity changes resulting from amino acid substitutions for ND5 site 67. Amino acids are indicated by color and their polarity specified: nonpolar (np) or polar (p). Pie charts represent the proportional likelihood of particular state changes at each of the ancestral nodes.  Grantham’s difference (D) shown for each amino acid pair within Tachycineta (Grantham, 1974).    

 

     

T. stolzmanni

T. albiventer

T. albilinea

T. meyeni

T. leucorrhoa

T. bicolor

T. cyaneoviridis

T. euchrysea

T. thalassina

Progne chalybea

ND5 67

D = 96

Alanine (np)Leucine (np)

  81  

APPENDIX I. Branch labels corresponding to Appendix E. (A) Nuclear phylogeny with nodes and branches labeled. Branch labels identical for shared branches in Phylogeny 3 and 4. (B) Mitochondrial phylogeny with branches labeled. A

B