Genomic methods offer a promising tool for conservation genetics and are quickly becoming a common method for obtaining numerous loci for multiple individuals in phylogenetic and phylogeographic studies (Manthey et al. 2015). However, distinguishing the roles of dispersal, vicariance, and gene flow in explaining population structure continues to be a challenge. Here, we included evidence from mitochondrial and genome-wide variation, as well as environmental niche modeling, to elucidate the evolutionary history of Bell’s Vireo populations for management purposes. Our data indicate that gene flow is restricted in the Bell’s Vireo. However, restricted gene flow, in the form of isolation by distance, does not adequately explain the overall pattern. The mitochondrial data show a clear division of eastern and western birds, and analysis of the GBS dataset with a partial mantel test further established the presence of a barrier to gene flow between these regions. Additionally, the GBS data set identifies genetically cohesive units within the east and west that correspond to geography.
Species limits
Mitochondrial data divide the Bell’s Vireo complex into two well-supported clades of individuals from the eastern and western portions of their distribution that are approximately 3 % divergent from one another (Fig. 1). For comparison, Johnson and Cicero (2004) summarized data for 39 pairs of North American avian sister species and reported that, on average, mitochondrial DNA sequences of sister species are 1.86 % divergent. More recently, Arbeláez-Cortés et al. (2014), studying V. hypochryseus, identified internal sister clades that are approximately 1.5 % divergent for ND2. These studies suggest the degree of genetic structure within the Bell’s Vireo complex is large compared to birds in general, including other vireos. Furthermore, the division between eastern and western clades identified by mtDNA is concordant with genome-wide SNP data (Fig. 4), further establishing the evolutionary independence of the two lineages. Finally, ecological niche modeling revealed eastern clade refugia in northeast Mexico and western clade refugia in northwest Mexico with additional hotspots in California. The refugia in Mexico are allopatric, indicating long established and isolated habitat for the separate clades.
The difficulty in defining a concept of species that is universally applicable has long plagued biologists. Nevertheless, applying a variety of widely used species concepts to our data indicate that the Bell’s Vireo constitutes two species. Under a Phylogenetic Species Concept (PSC), species are defined as an irreducible, diagnosable group, with a pattern of ancestry and descent (Zink and McKitrick 1995). The eastern and western clades of Bell’s Vireo are clearly separate species based on the PSC because they can be diagnosed on the basis of 23 synapomorphies in ND2. Avise and Ball (1990) introduced a genealogical concordance perspective to the species concept debate, emphasizing the importance of multiple gene phylogenies. These ideas were further developed into a genealogical species concept (GSC) by Baum and Shaw (1995) whereby multiple concordant gene trees are indicative of exclusive groups and species status. Strictly speaking, this project did not examine individual SNP phylogenies in addition to the ND2 phylogeny; rather, we performed a concatenated analysis to determine the overall genomic signal within Bell’s Vireo. Therefore, the GSC would not technically be applicable. However, the results of the concatenated analyses indicate that most SNPs support the deep eastern/western division identified for the Bell’s Vireo. Thus, concordance between genomic and mitochondrial data observed in our study supports the division of Bell’s Vireo into two independently evolving lineages.
The Biological Species Concept (BSC) defines species based on reproductive isolation. Under the BSC, a group of populations is considered a species if they breed or could potentially breed with one another (Mayr 1993). The possible secondary contact between the east and west clades that was identified in the STRUCTURE analysis suggests caution in the recognition of two species under a traditional interpretation of the BSC. However, this pattern may also be the result of longer coalescent time needed for nuclear markers Even if there is some limited gene flow between the two clades, Gill (2014) showed how recent advances in understanding the role of gene flow in bird speciation forces a new approach to applying the BSC. Given that speciation may occur in the context of gene flow, complete reproductive isolation is deemphasized under this approach, and essential reproductive isolation (lack of free interbreeding) becomes a more appropriate criterion to apply in situations that involve potential gene flow (Gill 2014). Applying this perspective, the two clades qualify for separate species status under the BSC as well.
While the genetic data provide a solid basis for species delimitation regardless of species concept applied, additional lines of evidence identify distinct eastern and western populations. The eastern and western clades are considered field identifiable by their plumage characters (Sibley 2014). Additionally, tail length can differentiate eastern and western populations; western individuals range from 47 to 54 mm while eastern individuals vary from 41 to 47 mm (Ridgway 1904). Behavioral divergence has also been observed for populations of Bell’s Vireo. Western Bell’s Vireos wag their tails side-to-side, similar to a gnatcatcher, but eastern Bell’s Vireos flick their tails up and down like a Palm Warbler (Dendroica palmarum) (Greaves and Chadwick 2006; Sibley 2014).
A new species taxonomy is warranted for the Bell’s Vireo, based on reciprocally monophyletic sister populations, essential reproductive isolation, and morphological and behavioral evidence. Following rules of taxonomic priority, the name Vireo bellii can be used for eastern populations (Central United States, Texas, Chihuahua Desert), and Vireo pusillus can be used for western populations (Arizona, California, Baja California, Sonoran Desert). For common names, we suggest the use of Bell’s Vireo (Vireo bellii) and Least Vireo (Vireo pusillus). The retention of Bell’s Vireo and use of Least Vireo returns to the original common names used for each group (Coues 1866, 1890).
Subspecific limits
The category of subspecies is important for conservation of North American vertebrates under the U.S. Endangered Species Act because subspecies, and even smaller units (i.e., distinct population segments, evolutionarily significant unit) can be recognized as the taxonomic units in need of protection independent of the status of the overall species (Haig et al. 2010). For the Bell’s Vireo complex, genome-wide SNPs provided additional resolution within the two proposed species (Bell’s Vireo and Least Vireo) that can be used to guide subspecies designations. In particular, these data provide insight into the genetic distinctiveness of the endangered Least Bell’s Vireo. All eight Least Bell’s Vireos formed an exclusive, well-supported cluster in our ML analysis. This group includes individuals sampled from distant parts of the range of the subspecies (San Diego and San Bernardino Counties); thus, monophyly is not the result of a geographically small sample, or sampling only closely related individuals. Our STRUCTURE analysis of a 3 or 4-population model and the sub-clade analysis of western populations corroborate the ML result for Least Bell’s Vireo. Taking this into consideration, the overall signal from the SNP dataset supports maintaining the Least Bell’s Vireo as a distinct evolutionary unit that qualifies as a subspecies, evolutionarily significant unit, and distinct population segment. In addition, all individuals of V. b. bellii form a clade, with 62 % bootstrap support. Two individuals of V. b. medius from south Texas (LSU B43284, LSU B30896) received marginal support as grouping with V. b. bellii individuals rather than V. b. medius. Nevertheless, none of the four currently recognized subspecies (including V. b. medius) are polyphyletic with respect to each other, and thus the tree structure does not show widespread gene flow among subspecies. Monophyly of Bell’s Vireo subspecies was not recovered in our ND2 phylogenies, likely due to the small amount of the genome represented by ND2 relative to our SNP data set. Nevertheless, ND2 still showed significant Fst values between subspecies. Therefore, given our current data and analyses, we recommend retaining the current subspecies. Thus, under our proposed classification, the Bell’s Vireo complex includes two species, the Bell’s Vireo (Vireo bellii) that includes two subspecies Vireo b. bellii and Vireo b. medius, and the Least Vireo (Vireo pusillus) that includes two subspecies, Vireo p. pusillus and Vireo p. arizonae. The endangered subspecies V. p. pusillus can be identified as California Least Vireo and V. p. arizonae can be referred to as the Arizona Least Vireo, as in previous works (Coues 1866, 1890).
Biogeographic history of the Bell’s vireo
Physical barriers may facilitate diversification/speciation by either dispersal or vicariance. In dispersal, a pre-existing barrier does not completely restrict gene flow, and colonists may move to a new area and form a new population. With time, the separated populations become distinct. The second pattern depends on vicariant events to explain taxon distributions. Under this model of diversification, the ancestor is widespread, and isolating (vicariant) events divide the ancestral population into sub-populations that can then differentiate in isolation (Ronquist 1997). These two explanations are not mutually exclusive, especially in highly vagile organisms like birds; therefore, teasing apart their relative influence on any one species can be problematic because either explanation can be equally parsimonious. However, comparative phylogeography across groups of taxa can help to provide insight into the most likely scenario of speciation.
Although a dispersal scenario cannot be entirely excluded, evidence from our niche modeling analyses and comparative phylogeographic studies point to a more important role for vicariance. The area of potential contact of the eastern and western Bell’s Vireo clades (Fig. 1) has also been identified as a contact zone in previous studies of various organisms, many with similar dispersal abilities (Riddle and Hafner 2006; Shepherd and Burns 2007; Smith et al. 2011; Weyandt and Van Den Bussche 2007; Zink et al. 2001). A comparative study of co-distributed birds in the arid southwest of North America discovered that both the Canyon Towhee (Melozone fusca) and Curve-billed Thrasher (Toxostoma curvirostre) show sister relationships between the Sonoran and Chihuahuan deserts (Zink et al. 2001). Within the Northern Cardinal (Cardinalis cardinalis), Smith et al. (2011) also found a division of east and west sister taxa with a contact zone in eastern Arizona. Additional work on bats (Weyandt and Van Den Bussche 2007) provides further evidence of a similar pattern of differentiation in the dry deserts of southwest North America. Uplift in the late Neogene, from approximately 15–2 mya, produced the SMO, potentially isolating populations into Sonoran and Chihuahuan Desert sister taxa (Wilson et al. 2011). In the Pliocene and Pleistocene, after major landforms such as SMO had formed, three major filter-barriers in the warm deserts of North America had strong influences on the evolutionary history of organisms (Hafner and Riddle 2011). Among the three is the Cochise filter-barrier (Morafka 1977); it is located in the vicinity of the Deming Plains, an area in northwest Mexico near the Arizona/New Mexico border. The Cochise filter-barrier, like the SMO, may have isolated Bell’s Vireo into eastern and western forms. Vicariance resulting from uplift and SMO formation in the late Neogene would require a deeper divergence than the 1 - 2 mya divergence identified for the western vs. eastern clades of Bell’s Vireo. Instead, diversification due to the Cochise filter-barrier better aligns with the molecular dating, niche modeling results, and comparative phylogeography.
The two clades of Bell’s Vireo correspond geographically and temporally with the same pattern of divergence as other highly vagile species in this region (Smith et al. 2011; Weyandt and Van Den Bussche 2007; Zink et al. 2001). In addition, niche modeling identified allopatric refugia for east and west populations during the LGM. During this time frame, riparian corridors (critical habitat for Bell’s Vireo) would have been at their broadest and most connected (Hafner and Riddle 2011), yet two allopatric refugia were recovered. Therefore, the most likely explanation for the divergence in east and west populations is a result of glacial cycling and the Cochise filter-barrier.
Conservation implications
In this study, we present evidence for multiple species within the Bell’s Vireo, and suggest a new taxonomy that better reflects the species’ evolutionary history. Our findings reveal that within the context of conservation, Bell’s Vireo is more appropriately considered not as one species with four subspecies, but as two species, each with two subspecies. This new information has implications for the federal, state and other listing status of Bell’s Vireo throughout its range. Our results support the distinctiveness of the Least Bell’s Vireo as a taxon and its recognition as state and federally endangered. A re-assessment of the conservation status of Bell’s Vireo in the rest of the U.S. is warranted given that the new taxonomy will change the contexts within which abundance and distribution, threats, and management options for each species and subspecies are considered. Such a re-assessment, which might elevate the listing status of particular taxa or populations, is necessary to ensure that the full breadth of taxonomic diversity within Bell’s Vireo is protected.
Genetic studies will continue to play a prominent role in providing necessary information for effective management of Bell’s Vireo. Additional sampling both within and between Bell’s Vireo and Least Vireo populations will provide clarification on the area of contact for the two species and allow for a better assessment of subspecies status. In particular, additional information is needed from the potential contact zone in New Mexico, where Bell’s Vireo habitat is sparse and patchy. Similarly, further sampling to determine the boundary separating the Least and Arizona subspecies is necessary to resolve the taxonomic status of populations in eastern California, western Arizona and southern Nevada and thus identify which fall under the protection of the Endangered Species Act afforded to Least Bell’s Vireo. Within California, where vireos occupy a highly urbanized and fragmented landscape, genetic analyses are needed to evaluate gene flow among drainages and the degree of genetic diversity within populations, an important determinant of their potential to respond to environmental change. Such analyses would establish whether Least Bell’s Vireo populations are functioning as an inter-connected metapopulation, and reveal areas genetically isolated that would benefit from management action to re-connect them to other populations. Determination of the extent to which Least Bell’s Vireo in coastal California is isolated by the Transverse Ranges from desert populations in eastern California would similarly provide information regarding genetic structuring and improve our ability to integrate evolutionary dynamics of populations into conservation strategies.