- Top of page
- LITERATURE CITED
- Supporting Information
Discordance between mitochondrial and nuclear DNA has been noted in many systems. Asymmetric introgression of mitochondria is a common cause of such discordances, although in most cases the drivers of introgression are unknown. In the yellow-rumped warbler, evidence suggests that mtDNA from the eastern, myrtle warbler, has introgressed across much of the range of the western form, the Audubon's warbler. Within the southwestern United States myrtle mtDNA comes into contact with another clade that occurs in the Mexican black-fronted warbler. Both northern forms exhibit seasonal migration, whereas black-fronted warblers are nonmigratory. We investigated the link between mitochondrial introgression, mitochondrial function, and migration using novel genetic, isotopic, biochemical, and phenotypic data obtained from populations in the transition zone. Isotopes suggest the zone is coincident with a shift in migration, with individuals in the south being resident and populations further north becoming increasingly more migratory. Mitochondrial respiration in flight muscles demonstrates that myrtle-type individuals have a significantly greater acceptor control ratio of mitochondria, suggesting it may be more metabolically efficient. To our knowledge this is the first time this type of intraspecific variation in mitochondrial respiration has been measured in wild birds and we discuss how such mitochondrial adaptations may have facilitated introgression.
Most well-studied hybrid zones between divergent taxa are characterized by steep, coincident clines at various genetic loci and phenotypic traits. These patterns are consistent with limited introgression across a hybrid zone and strong selection against hybrids (Barton and Hewitt 1985). In some instances, however, there is geographic discordance among genes or traits, where clines are strongly displaced in their spatial distribution compared to the rest of the genome (Barton 1993). Often this pattern is consistent with introgression following secondary contact, which can be promoted by various demographic and/or selective factors (Toews and Brelsford 2012). Hence the biogeographic patterns associated with these types of discordant clines can reveal novel insights into evolutionary processes (e.g., Brumfield et al. 2001).
One linked suite of genes that commonly shows biogeographic discordance with other genetic markers or phenotypic traits are the genes encoded in the mitochondrial genome (mtDNA; Takahata and Slatkin 1984; Barton 1993). Discordant biogeographic patterns in mtDNA and nuclear markers have been identified in numerous animal systems and most cases are likely due to introgression of mtDNA between taxa (approximately 90% of the studies reviewed by Toews and Brelsford 2012). Mitochondrial introgression can be a result of neutral diffusion and genetic drift or be driven by demographic asymmetries such as female-biased dispersal (mtDNA is maternally inherited in most animals) or geographically varied patterns of selection that differentially affect mitochondrial and nuclear genes (Rheindt and Edwards 2011; Irwin 2012). Although many authors have noted these biogeographic patterns and have speculated as to the potential drivers of mito-nuclear discordance, most have not collected additional evidence to test the various introgression hypotheses (Toews and Brelsford 2012).
Here we investigate the discordance between mtDNA and nuclear DNA in the yellow-rumped warbler (Setophaga [coronata] spp.) species complex, one of the most abundant and widespread warblers in North America. The species complex is composed of four currently recognized taxa (formerly of the genus Dendroica; designated as separate species by the International Ornithological Council but as subspecies of a single species by the American Ornithologists’ Union; Fig. 1A): Setophaga coronata, the myrtle warbler, which breeds in the boreal forest east of the Rocky Mountains and winters in eastern North America, Central America, and the Caribbean; Setophaga auduboni, the Audubon's warbler, which breeds west of the Rocky mountains and winters in the southwestern United States, Mexico, and central America; Setophaga nigrifrons, the black-fronted warbler, which is a resident year-round in Mexico; and finally, Setophaga goldmani, the Goldman's warbler, which consists of a small population of resident birds confined to Guatemala (Hubbard 1970).
Figure 1. (A) The distribution of the four currently classified species in the yellow-rumped warbler (Setophaga spp.) complex. According to Brelsford et al. (2011) there are three distinct nuclear groups that align with the areas shaded blue (myrtle warbler), red/yellow (the Audubon's phenotype and black-fronted warbler phenotype), and violet (Goldman's warbler). The hatched areas distinguish the myrtle-type from black-fronted type mtDNA. (B) Sampling localities in the southwestern United States where we studied the cryptic mtDNA transition zone. Locations 1–7 indicate sites along the western transect, 8–15 in the eastern transect (see Table 1 for location information). (C) A simplified schematic adapted from Brelsford et al. (2011) to illustrate the discordances between plumage, morphometric, nuclear, and mitochondrial patterns in this system.
Download figure to PowerPoint
Studies of multilocus nuclear markers (AFLPs) suggest that there are three distinct nuclear clusters in this group of four recognized taxa, one consisting of the myrtle warbler, another that includes Audubon's and black-fronted warblers, and one distinct group corresponding to Goldman's warblers (Brelsford et al. 2011). Within the cluster that includes the Audubon's and black-fronted warbler there is a gradual latitudinal gradient in the nuclear genome that is also mirrored in some morphological traits (e.g., wing length; Brelsford et al. 2011). Darker plumage patterns and migratory behavior of the birds in Mexico (i.e., S. nigrifrons) separate these individuals from the rest of the Audubon's group in the United States (i.e., S. auduboni), however there is no known reproductive boundary between these taxa (Milá et al. 2011).
These patterns in phenotype and the nuclear genome contrast with the distribution of mtDNA: over much of the range of the Audubon's warbler, myrtle warbler mtDNA is fixed (see Fig. 1C for a simplified schematic of discordant clines). Near the border of Utah and Arizona there is a transition to a second, deeply divergent mitochondrial DNA clade (Brelsford et al. 2011), which was previously assumed to be geographically restricted to Mexico in the black-fronted warbler. This cryptic transition occurs within what are otherwise phenotypically and morphologically Audubon's warblers (Brelsford et al. 2011; Milá et al. 2011) and where there is no observed break in nuclear markers across this zone. Evidence suggests that this mito-nuclear discordance was generated from mitochondrial introgression that is estimated to have occurred relatively recently (divergence between northern Audubon's and myrtle mtDNA is estimated to have occurred roughly 16,000 years ago, although there is very large uncertainty in that estimate; Milá et al. 2011). Past hybridization and introgression seem possible given that myrtle and Audubon's warblers are known to currently interbreed extensively and form viable hybrids in the Rocky Mountains of British Columbia and Alberta, although this is far from the contemporary transition in mtDNA (Hubbard 1969; Brelsford and Irwin 2009).
Although all of the individuals in the Utah/Arizona mtDNA transition zone resemble Audubon's warblers in plumage and morphometric traits (Milá et al. 2011), Brelsford et al. (2011) suggest the possibility that this area may instead align with another important phenotypic trait—a shift in seasonal migratory behavior, from resident birds in the south to migratory birds in the north. Although this suggestion is based on limited observational data (Hunt and Flaspohler 1998), given the important role that mitochondria play in energy production during the metabolically demanding act of migration (e.g., Scott et al. 2009) there is an intuitive link between this behavior and variation in mitochondrial phenotype. This “migration adapted mitochondrion” hypothesis posits that in the past, natural selection favored myrtle mitochondrial variants in migratory Audubon's warblers compared to those with the ancestral black-fronted mitochondria, which presumably evolved in primarily resident or short-distance migrant populations.
Empirical examinations of whether selection may play a role in facilitating mtDNA introgression have been rare, although simulations have shown that local adaptation of mtDNA can occur with only a small selective advantage (Irwin 2012). This is likely due in part to the widespread assumption of neutrality and in part to the difficulty of detecting selection in the mitochondria, both at the molecular and biochemical level. Despite the long held assumption that variation in mtDNA is primarily neutral, a number of studies have identified intra- and interspecific variation in the proteins encoded by genes in the mitochondrial genome that authors have attributed to natural selection (Blier et al. 2001; Bazin et al. 2006; Meiklejohn et al. 2007; Ballard and Melvin 2010; Scott et al. 2011; Correa et al. 2012; Pichaud et al. 2012). In those studies that have tested for evidence of adaptive introgression (e.g., Blier et al. 2006; Boratynski et al. 2011), few have found evidence for functional differences in introgressed mitochondria. It has been suggested, however, that more sensitive techniques that test mitochondrial function directly, such as measuring potential variation in respiratory capacity in mitochondrial preparations (e.g., Pichaud et al. 2012), may be more appropriate to determine if introgressed mitochondria differ phenotypically from a native type. While collecting such functional data is useful, eventually connecting it with potential fitness consequences can be an important yet challenging additional step (Storz and Wheat 2010).
We investigated the potential link between mitochondrial introgression, mitochondrial function, and migration in yellow-rumped warblers using novel genetic, isotopic, biochemical, and phenotypic data obtained from several natural populations in the mtDNA transition zone. We made two major predictions. First, if there is any link between mitochondrial introgression and migratory behavior, we predicted that the mtDNA transition would also be located at the transition in migratory phenotype. Although stochastic processes alone could generate such a correlation, preliminary studies suggest that no other phenotypic trait (e.g., color), environmental factors (e.g., habitat characteristics), or demographic parameters (e.g., population size) covary with mtDNA. To determine the location and shape of the mtDNA cline, we genotyped individuals across two transects (depicted in Fig. 1B) in the mtDNA transition zone and used a maximum-likelihood cline-fitting procedure to estimate the cline center and width. To estimate the migratory movements of individuals, we analyzed stable hydrogen isotopes in feathers to estimate the distance that each individual travels between its breeding and predicted wintering grounds. We expected that individuals in the south would show little difference in the isotopic composition of feathers grown on the breeding and wintering grounds (i.e., year-round residents), whereas individuals in the north would show a larger difference, consistent with moving between disparate localities (i.e., migrants).
Second, we predicted that variation in mitochondrial genotype explains some variation in mitochondrial phenotype. Such a finding is important in that it suggests that there is functional variation present that selection could act upon, although it is important to note that documenting variation in biochemical phenotype does not necessarily imply any effect on fitness (Storz and Wheat 2010). We first sequenced protein coding mtDNA genes involved in the electron transport complexes (ETC) from multiple Audubon's warblers with either myrtle or black-fronted mitochondrial types, examining whether these proteins differ in amino acid sequence between the two groups. To test if these amino acid differences had functional consequences, we directly assayed maximal enzyme activity in one of the electron transport chain complexes (complex I) that had a number of amino acid substitutions. We predicted we would observe higher maximal activities in this enzyme from the northern, myrtle-type mitochondria. We also assayed the respiratory capacity of mitochondria in permeabilized muscle fibers from wild-caught individuals of both mtDNA types in the contact zone. If individuals in the north of the transition zone, with myrtle-type mtDNA, have mitochondria better adapted for migratory movements, this might be reflected in the ability of their mitochondria to consume oxygen and produce ATP more efficiently. Specifically, we predicted that a mitochondrion associated with a more migratory lifestyle would (1) have a higher maximum capacity for respiration as measured by a higher state III (ADP-stimulated) oxygen consumption rate and/or (2) have a more efficient mitochondria as a result of increased coupling, which can be estimated by calculating the ratio of state III-to-state II consumption rates, also known as the acceptor control ratio (ACR; Nicholls and Ferguson 2002).
- Top of page
- LITERATURE CITED
- Supporting Information
In an effort to understand the drivers of mitochondrial introgression in the yellow-rumped warbler system, we have assembled a diverse data set that includes genetic, biochemical, and phenotypic variation obtained from several natural populations. We sampled individuals in an area in the southwestern United States that is a cryptic transition zone between the myrtle and black-fronted mtDNA (i.e., introgressed vs. ancestral mitochondrial types, respectively). We found that this transition in mtDNA is broadly coincident with a shift in migratory behavior and also with some aspects of mitochondrial phenotype. We discuss and interpret the findings for this novel data set in detail below.
First, our data support and extend previous research (e.g. Milá et al. 2011) suggesting that this mtDNA transition occurs within what are otherwise phenotypically and morphometrically Audubon's warblers in Arizona and Utah and also in previously unsampled sites in Colorado and New Mexico. However, the results of the cline fitting analysis suggest that the evolutionary and/or ecological processes shaping mtDNA variation differ somewhat between eastern and western transects, with the width of the eastern transition being approximately twice that of the west. Although many factors may be involved, including the fact that the intensity of selection between the two transects may vary, one of the most striking differences between the two transects is that the eastern (CO/NM) transect has a more contiguous matrix of suitable forest between sampling locations (green areas in Fig. 1B). More habitat continuity could increase population connectivity and mtDNA gene flow, thereby generating wider clines. In contrast, along the western (UT/AZ) transect, populations are concentrated in high-elevation coniferous forests, such as the Coconino and Kaibab National Forest (AZ) and the Dixie National Forest (UT), that are separated by wider areas of unsuitable habitat, such as the Grand Canyon, potentially inhibiting dispersal and generating narrower clines.
The mtDNA clines can be compared to a model of neutral diffusion following secondary contact using the equation (Endler 1977; Barton and Gale 1993) where w is cline width, σ is root-mean-squared dispersal, and t is number of generations since contact. Assuming a dispersal distance of 20 km and a generation time of 2 years (Brelsford and Irwin 2009), it would take approximately 400 years following contact for a neutral mtDNA cline to form that is as wide as that observed along the eastern transect (less than half that time is necessary for the narrower cline in the west). This is not long, considering the biogeographic history of western North America, and it seems likely that these populations have been in contact much longer.
Both transitions in the mtDNA contact zone are wider than another well-studied hybrid zone in this species complex, between the myrtle and Audubon's warbler in the northern Rocky Mountains of British Columbia and Alberta, which is estimated to be 132 km wide (Brelsford and Irwin 2009). The narrowness of that hybrid zone in the Rockies suggests that there is moderately strong selection against hybrids (Brelsford and Irwin 2009). In the southern contact zone studied here, however, the observation that no other phenotypic trait studied to date, such as plumage, morphometrics (Milá et al. 2011), and song (unpubl. data) shows a transition in the same location as mtDNA, combined with the fact these clines in mtDNA are much wider, suggests there is not likely a strong reproductive barrier between individuals with myrtle-type and black-fronted-type mtDNA. This result could be confirmed with additional nuclear markers using next generation technologies. For instance, are there small portions of the nuclear genome that covary with mtDNA, consistent with a pattern of cryptic genomic regions of isolation between individuals with the two mitochondrial types? Such high-resolution genomic data could also be useful in asking whether there is any reproductive isolation between birds with the black-fronted mtDNA in the southern United States and those in Mexico (i.e., across the traditional taxonomic boundary between S. auduboni and S. nigrifons).
Brelsford et al. (2011) suggests one possible factor driving the introgression of the myrtle-type mitochondria to high frequency in Audubon's warblers: it may be better adapted for the energetic demands of long-distance migration. Supporting this suggestion, our isotopic analysis suggests that there is a broad transition in migratory behavior, from individuals that do not move large distances between seasons at the southern end of the mtDNA transition zone, to individuals that display behaviors associated with a fully migratory phenotype in the north. The ANCOVA analysis suggests that mtDNA type may explain a small amount of variation in individual movement (Table 2) but that breeding latitude in the contact zone is a much better predictor of migratory distance. Although finer-scale sampling along both transects would have been ideal, our estimates suggest that the shift in migratory behavior occurs between 35oN and 36oN (Table 1), very near the centers of the mtDNA transitions (35.94–37.4oN; Fig. 2). The fact that this transition in migratory behavior is broadly coincident with the shift in mitochondrial DNA is consistent with migration-adapted mtDNA hypothesis suggested by Brelsford et al. (2011). However, another possibility is that the mtDNA contact zone is still moving and it is by chance located in the current location that parallels the shift in migration.
It is interesting to note that inferred wintering latitude did not differ between individuals breeding at different locations. For instance, 95% of the individuals have winter deuterium ratios between −74‰ and −27‰ δD (Fig. S3; deuterium ratios increase toward the south), which would put most of the individuals wintering between approximately 37oN and 33oN latitude (Hobson et al. 2012). This is consistent with field observations of wintering yellow-rumped warblers in the southwestern United States (Hunt and Flaspohler 1998). However, it seems likely that at least some of the southern populations still move a short distance from their breeding locations. For instance, most sites where we captured individuals on their breeding territories across this mtDNA transition are at high elevations and likely still experience harsh winters. It may be that individuals in the south are moving out of the coniferous forests and into low-lying areas, performing a short altitudinal migration. Currently this is speculation, but future studies employing other, more sensitive techniques may be able to address these hypotheses, although it would not alter the conclusions of this study.
The mtDNA sequencing data suggests that there are numerous amino acid differences between the two mitochondrial types (Fig. 3); it is possible that some evolved via selection in myrtle warblers and were subsequently the targets of selective introgression following secondary contact with Audubon's warblers. Consistent with this, data from ND2, which has been sequenced in a number of Setophaga warblers, suggests that three of the four amino acid substitutions we identified in this gene (Fig. 3) are derived in the myrtle-type mtDNA and are retained in the ancestral-state in the black-fronted mtDNA. However, distinguishing directional selection from genetic drift or purifying selection as causes of mitochondrial DNA sequence differences is difficult (Hudson and Turelli 2003). In addition, the genes in mtDNA are inherited as a linked group, therefore testing the phenotypic effects of certain mutations is impossible without experimental work. Most of the amino acid changes are present in genes that code for protein subunits present in complex I, suggesting it may be a target of directional selection, however there are a number of reasons why this interpretation should be treated with caution. First, complex I is the most poorly understood complex in the electron transport chain, primarily because of its “L-shaped” membrane structure (Efremov and Sazanov 2011), so it is currently not possible to predict if amino acid changes occur in important active sites of the enzyme. Second, the genes coding for proteins involved in complex I have been found to evolve at a high rate in birds and reptiles as compared to other mitochondrial genes, so it is not necessarily surprising that these genes show a number of amino acid differences (Eo and DeWoody 2010). Third, we did not find significant differences in maximal enzyme activity of complex I between individuals with myrtle versus black-fronted type proteins, providing insufficient evidence that these changes affect enzyme function via changes in catalytic efficiency (kcat) or enzyme amount. We note that there are other kinetic properties of complex I that we did not measure which may be influenced by the amino acid differences, such as the binding affinity for its substrates (NADH and ubiquinone) that may vary between the mitochondrial types and could be assayed in future studies. More generally, other nuclear genes involved with mitochondrial products could instead be the target of selective introgression resulting in mitochondrial discordance. Although a previous multilocus nuclear study using AFLPs (Brelsford et al. 2011) did not find sharp changes in the nuclear genome along the mtDNA transition zone, this does not rule out the potential contribution of nuclear-encoded mitochondrial products. We suggest that future studies employing next generation sequencing technologies could address this latter alternative more conclusively.
The data from measures of mitochondrial respiration suggest there is a small but statistically significant difference between how the mitochondrial types consume oxygen. Here we predicted that a mitochondrion associated with a more migratory lifestyle would (1) have a higher maximum capacity for respiration as measured by a higher state III consumption rate and/or (2) be more efficient as a result of increased coupling, measured here as the ratio of state III to state II consumption rates or the ACR. Coupling is defined as the amount of inorganic phosphate that is incorporated into ATP per unit of O2 consumed by the mitochondria. The uncoupling of oxidative phosphorylation describes any process that decreases this phosphate/oxygen ratio and generally leads to a waste of redox energy (Nicholls and Ferguson 2002). In this case, we find no significant difference in the maximal efficiency of mitochondrial respiration (state III) between myrtle and black-fronted mito-types, but we do find a significant difference in ACR between them. Although this difference is small, it is in the expected direction of more migratory myrtle-type birds exhibiting a higher ACR and a potentially more efficient production of ATP, due to less proton leak and uncoupling, compared to the southern and sedentary black-fronted type individuals. This ratio in ACR was driven mostly by differences in state II, as it differs most strongly between the types, although the difference is not statistically significant (Fig. 5A).
Previous studies have shown that proton leak decreases with body size in birds and other endotherms (Brand et al. 2003). This suggests that small birds, such as wood warblers studied here, have a high mitochondrial membrane permeability and proton leak and therefore have potentially a greater scope for evolving a more coupled system. Although the molecular basis of the proton leak is currently unclear, we suggest that future studies assessing membrane permeability of the different mitochondrial types could further test our finding of higher ACR and potentially lower state II respiration rate in myrtle-type mitochondria.
To our knowledge this is the first time mitochondrial respiration has been measured in permeabilized muscle fibers in a wild caught bird. Compared to captive geese (Scott et al. 2009), yellow-rumped warblers have higher state III and state II respiration rates and a lower ACR. This is consistent with the allometric relationship of these variables as measured in isolated mitochondria in other birds (Brand et al. 2003). The sensitivity of this method suggests that it could be useful for other applications. For instance, a broader comparative study of migratory versus sedentary species could be used to test whether mitochondrial adaptation to a migratory lifestyle is common. This would be especially useful in cases where past introgression may not have left a distinct biogeographic pattern such as observed in this system. Beyond a migratory phenotype, these types of assays would be ideal for testing hypotheses of mitochondrial adaptation to conditions of hypoxia, such as at high elevation, as highlighted by studies of the rufous-collared sparrow (Cheviron and Brumfield 2009) and deer mice (Cheviron et al. 2012).
Given the current extent of introgression and the functional differences we have identified, it is unclear why the myrtle-type mtDNA has not swept to fixation throughout the entire Audubon's/black-fronted warbler range. One possibility is that it has swept to frequencies at or near fixation in populations for which it is adaptive (i.e., migratory populations) and then, in those populations for which the myrtle mitochondrial-type no longer presents a selective advantage, more local demographic processes such as population size and dispersal may dominate. Although speculative, there may also be a tradeoff between the two mitochondrial types: increased coupling may be important for migration in the north, whereas an increased proton leak may be advantageous in the south due to other environmental factors that we have not measured. There are also scenarios that could generate such discordant mtDNA clines that do not involve selective introgression of mtDNA. For instance, perhaps the original contact between the divergent forms is close to the current transition zone in the southwestern United States. If so, it is possible that the zone may be trapped in a population sink or in an ecological transition that is not currently obvious. To explore these alternatives, future studies should consider collecting additional environmental and genetic data and also sample other populations further east, west, and along various elevations. This would allow a more robust test of whether the patterns observed here are likely a result of selection, demographics, or simply stochastic variation.
In conclusion, our study adds to the small number of cases where the proposed drivers of mitochondrial and nuclear discordance have been rigorously tested. Indeed, of the 35 studies reviewed by Toews and Brelsford (2012) that invoke adaptive explanations to explain mitochondrial introgression, very few have rigorously tested those explanations (e.g., Aubert and Solignac 1990; Blier et al. 2006; Boratynski et al. 2011). By combining molecular biogeography, stable isotopes, and mitochondrial biochemistry, we have examined the correlation between migratory phenotype and mitochondrial genotype in this system. We suggest that data presented here are consistent with the migration-adapted mitochondrion hypothesis. However, given the correlative nature of this study, other processes, including stochastic shifts in the frequency and distribution of mtDNA that may have produced this mtDNA-nuDNA discordance (and any correlated phenotypic differences), should not be excluded. In the future, a more direct test of the role of selection driving mtDNA introgression in this system will require a combination of the functional data presented here with measures of fitness that these different phenotypes may confer. Although challenging, especially for a complex phenotype such as migration, such cases of mitochondrial introgression present the exciting opportunity to link underlying genetic changes with phenotypic variation at the level of the mitochondria and whole-organism performance (Storz and Wheat 2010). This is a shift from the past, where many previous studies assumed that mtDNA evolves neutrally and where potential differences associated with selection were often disregarded (Ballard and Whitlock 2004; Irwin 2012). Indeed, studies that examine possible adaptive causes of genetic introgression can highlight the role that hybridization plays in providing an important source of adaptive alleles between partially reproductively isolated taxa.