Advertisement

Conservation Genetics

, Volume 18, Issue 3, pp 553–572 | Cite as

Population genetics and geometric morphometrics of the Bombus ephippiatus species complex with implications for its use as a commercial pollinator

  • Michelle A. Duennes
  • Chris Petranek
  • Esteban Pineda Diez de Bonilla
  • Jorge Mérida-Rivas
  • Oscar Martinez-López
  • Philippe Sagot
  • Rémy Vandame
  • Sydney A. Cameron
Research Article

Abstract

Mexico and Central America are among the most biodiverse regions on Earth, harboring many species with high levels of interpopulation morphological and genetic diversity. The mountainous topography of this region contains isolated sky island habitats that have the potential to promote speciation. This has been studied in vertebrates, yet few studies have examined the phylogeographic and genetic structure of insect species encompassing this region. Here we investigate geographic patterns of genetic and morphological divergence and speciation among widespread populations of the highly polymorphic bumble bee Bombus ephippiatus and its closest relative B. wilmattae. We used DNA sequences from a fragment of cytochrome oxidase I (COI), genotypes for twelve microsatellite markers, and morphometric data from wings to construct a well-supported inference of the divergences among these taxa. We have found complex patterns of genetic isolation and morphological divergence within B. ephippiatus across its geographic range and present evidence that B. ephippiatus comprises multiple independent evolutionary lineages. The pattern of their diversification corresponds to geographic and environmental isolating mechanisms, including the Mexican highlands, the lowlands of the Isthmus of Tehuantepec in southern Mexico, the Nicaraguan Depression, the patchily distributed volcanic ranges in Nuclear Central America and Pleistocene glacial cycles. These results have important implications for the development and distribution of B. ephippiatus as a commercial pollinator in Mexico and Central America.

Keywords

Bumble bees Microsatellites Cytochrome oxidase I STRUCTURE GENELAND Species delimitation 

Introduction

Mexico and Central America are well known for their biological complexity (Mittermeier et al. 2000). The great biodiversity in this region is frequently attributed to its location between two large continents, arising from biotic interchange between North and South America. This region is also a transition zone between the northern Nearctic and southern Neotropical biogeographic regions (Heilprin 1887). Its volcanic topography has led to isolation and speciation in birds (Cracraft and Prum 1988; Roy et al. 1997; García-Moreno et al. 2006), mammals (Vrba 1993; Sullivan et al. 2000; León-Paniagua et al. 2007), and herpetofauna (Mulcahy et al. 2006; Castoe et al. 2009; Daza et al. 2010). Many endemic forms have been restricted within the last few million years to particular ecosystems, such as montane pine-oak or cloud forests (Escalante et al. 1993; León-Paniagua et al. 2007; Kerhoulas and Arbogast 2010; Barber and Klicka 2010). For example, many bird species restricted by ecological limits exist as a series of isolated populations in islands of suitable habitat (García-Moreno et al. 2004).

There are two particularly notable lowland regions within Mexico and Central America that have served as geographic barriers between these montane sky island habitats, preventing species movement. One of these barriers is the Isthmus of Tehuantepec (IT) (Fig. 1), a continental strait ending in two plains. This large, narrow area of low elevation with a hot, humid climate separates the Sierra Madre del Sur of southern Mexico from the highland regions of the Maya Block, a fault block encompassing Mexico south of the IT through central Guatemala (Gutiérrez-García and Vázquez-Domínguez 2013). The IT is proposed to be a major barrier to dispersal in toads (Mulcahy et al. 2006), snakes (Castoe et al. 2009; Daza et al. 2010), birds (Barber and Klicka 2010), and rodents (Sullivan et al. 2000). The other significant barrier is the Nicaraguan Depression (Fig. 1), which is a lowland expanse separating the Chortis Block highlands of Honduras and Nicaragua from the highlands of Costa Rica. The Nicaraguan Depression is an important isolating mechanism in snakes and rodents (Castoe et al. 2009; Daza et al. 2010; Gutiérrez-García and Vázquez-Domínguez 2013).
Fig. 1

Map illustrating the distinct mountain regions that B. ephippiatus and B. wilmattae inhabit. There are four major mountains chains north of the Isthmus of Tehuantepec (IT), three mountain ecoregions south of the IT and north of the Nicaraguan Depression (ND), which are further split in Mexico by the Central Depression (CD), and the Talamancan montane forests south of the ND in Costa Rica and Panama. All regions highlighted are World Wildlife Federation (WWF) recognized ecoregions (Olson et al. 2001). (Color figure online)

The highlands of Mexico are also important geographic barriers for many plant and animal taxa, especially during more recent Pleistocene glacial cycles (Ornelas et al. 2013; Mastretta-Yanes et al. 2015). North of the IT, there are four distinct mountain ranges that have shaped the genetic structure of the organisms that live within them: the Sierra Madre Occidental, the Sierra Madre Oriental, the Trans-Mexican Volcanic Belt and the Sierra Madre del Sur (Fig. 1). South of the IT, there are four mountain ecoregions: the mid-elevation Central American pine-oak forest that extend into Guatemala and Honduras and the high elevation Sierra Madre de Chiapas moist forest, Chiapas montane forest, and Central American montane forest (Fig. 1). These latter three ranges are partially separated by the Central American pine-oak forest and a lowland region called the Central Depression (CD) (Fig. 1), a barrier to hummingbirds and passerines (Ornelas et al. 2013).

This pattern of high endemism and diversification across Mexico and Central America via geological isolation has likely affected insect diversity in the region. Biogeographic patterns of some groups of beetles, for instance, suggest that the uplift of montane regions in Central America encouraged the southerly movement of Nearctic beetles, while the tropical lowlands of Central America enabled South American beetles to move northward after the Panamanian land bridge connection ~3 mya (Halffter 1987; Liebherr 1994; Lobo and Halffter 2000; Marshall and Liehberr 2000; Morrone 2006). These studies have generated numerous hypotheses to explain the diversification of Mexican and Central American beetles, but the explanations for how geological barriers could have shaped diversification are constrained by reliance on species distribution data alone. Multi-locus phylogenetic and population genetic analyses can clarify how evolutionary divergence is shaped by historical events using gene genealogies. Furthermore, to understand whether whole communities respond similarly to the same historical events and shared barriers, it is important to compare patterns across diverse groups of taxa. To date, a small amount of molecular data is available for insect species in Mexico and Central America (beetles: Morse and Farrell 2005; Anducho-Reyes et al. 2008; Ruiz et al. 2010; Baselga et al. 2011; Sánchez-Sánchez et al. 2012; true bugs: Dorn et al. 2009; stingless bees: May-Itzá et al. 2010).

In both the Old and New World, bumble bees (Bombus) have been the focus of intense molecular phylogenetic (Koulianos and Schmid-Hempel 2000; Kawakita et al. 2004; Ellis et al. 2005; Hines et al. 2006; Cameron et al. 2007; Williams et al. 2011, 2012, 2015; Lecocq et al. 2011; Hines and Williams 2012; Carolan et al. 2012; Bossert et al. 2016; Françoso et al. 2016; Sheffield et al. 2016) and population genetic analysis (Estoup et al. 1996; Widmer et al. 1998; Widmer and Schmid-Hempel 1999; Chapman et al. 2003; Shao et al. 2004; Ellis et al. 2006; Darvill et al. 2006; Schmid-Hempel et al. 2007; Herrman et al. 2007; Lozier and Cameron 2009; Kraus et al. 2009; Darvill et al. 2010; Charman et al. 2010; Lye et al. 2011; Cameron et al. 2011; Lozier et al. 2011; Kraus et al. 2011; Goulson et al. 2011; Carvell et al. 2012; Jha and Kremen 2013; Lozier et al. 2013; Maebe et al. 2013; Lecocq et al. 2013; Dreier et al. 2014; Moreira et al. 2015; Jha 2015; Santos Júnior et al. 2015; Huang et al. 2015; Lecocq et al. 2015a, b, c; Francisco et al. 2016). Despite the presence of nineteen different species (Labougle 1990) of bumble bees in Mexico and Central America, only a single study to date has explored the phylogeography of bumble bees in this region (Duennes et al. 2012). The phylogenetic patterns and known biogeographic distributions of Bombus provide a rich background for investigating the structure of inter- and intraspecific genetic diversity in Mexico and Central America and for examining whether Bombus diversification correlates temporally with historical events associated with the speciation of vertebrate and other insect taxa.

Bombus ephippiatus and B. wilmattae are a species complex especially relevant for the study of Mexican and Central American biodiversity. First, the estimated divergence of these southern species from their North American relative B. impatiens (~1 mya, Duennes et al. 2012) fits within the pertinent timescale of recent geological and climatic events in this region. Second, B. ephippiatus is distributed widely throughout Mexico and Central America and is found in diverse montane habitats (Duennes and Vandame 2015), while B. wilmattae, of uncertain species status (distinguished from B. ephippiatus primarily by the presence of a band of yellow or white hairs on the anterior pronotum near the head; Labougle et al. 1985; Labougle 1990; Williams 1998; Duennes et al. 2012), is restricted to a relatively small geographic range in southern Mexico and eastern Guatemala. These distributions allow comparative studies of genetic diversity in widespread and restricted populations. Third, B. ephippiatus is highly polymorphic across its large range, exhibiting both a gradation in color pattern and genetic diversity from north to south (Duennes et al. 2012). Preliminary studies of the genetic diversity of this group suggested five main lineages within B. ephippiatus and B. wilmattae: one lineage of B. ephippiatus north of the IT, two sympatric lineages of B. ephippiatus south of the IT through Honduras, one lineage of B. ephippiatus in Costa Rica and a fifth lineage comprising B. wilmattae, which was nested within B. ephippiatus (Duennes et al. 2012). Overall, the widespread geographic distribution of the B. ephippiatus-B. wilmattae species group over a topologically complex region, its highly polymorphic color pattern, and its genetic diversity suggest the possibility of additional cryptic diversity.

The B. ephippiatus complex is of conservation concern as it is currently under development as an alternative commercial pollinator that could replace the non-native B. impatiens sold extensively throughout Mexico and Guatemala for greenhouse and field crop pollination. The continued use of B. impatiens for commercial pollination in Mexico and Central America poses threats to the population health of native bumble bee species. These threats, which have been indicated in other parts of the world as a result of global intercontinental transport of non-native Bombus, include competition for floral resources (Morales et al. 2013) and spread of diseases (Arbetman et al. 2013; Sachman-Ruiz et al. 2015). Because B. impatiens is sister species to the B. ephippiatus-B. wilmattae complex (Duennes et al. 2012), the threat of reproductive disturbance (Kondo et al. 2009) and inter-specific hybridization (Yoon et al. 2009) is of particular concern. BioBest (http://www.biobestgroup.com/) and Koppert Biological Systems (https://www.koppert.com/), both pioneers in European commercial bumble bee production, have facilities in Mexico that produce and distribute B. impatiens throughout Mexico and Central America; both companies are interested in producing B. ephippiatus as a native alternative pollinator (Torres-Ruiz and Jones 2012). Moreover, twelve independent breeders of B. ephippiatus across five states in Mexico have formed an association with the goal of developing this species as a sustainable crop pollinator (Asociación Mexicana de Criadores de Abejorros Nativos, A.C.). While B. ephippiatus appears to be a promising native alternative to B. impatiens, its widespread commercial use without understanding the taxonomic status and population genetic structure could result in a homogenization of native population genetic diversity (Bourret et al. 2011; Williams et al. 2012; Lecocq et al. 2016).

Here we describe how both montane and lowland barriers across Mexico through Honduras (Fig. 1) have shaped the genetic structure of the B. ephippiatus complex. We add four additional microsatellite loci to the eight used by Duennes et al. (2012) and add nearly 2000 new specimens from across the range of this species complex, especially from northern Mexico. We explore whether morphological variation corresponds to genetic variation and isolation using wing morphometric data from over 600 specimens. We discuss possible geographic barriers and climatic events that could have caused the observed patterns of diversification within this group and provide an in-depth comparison of the efficacy of the Bayesian assignment programs STRUCTURE (Pritchard et al. 2000) and GENELAND (Guillot et al. 2012). We also discuss taxonomic consequences of this study and implications for the trade of B. ephippiatus for commercial pollination in Mexico and Central America.

Materials and methods

Taxa examined

Phylogenetic analysis

To resolve the basal relationships among the lineages in the B. ephippiatus-B. wilmattae species complex, sequence fragment data were collected from cytochrome oxidase I for 254 specimens spanning the group’s broad geographic distribution. One hundred and fifty one specimens from Mexico and Guatemala were added to the 103 specimens previously analyzed in Duennes et al. (2012). In total, 76 specimens of B. wilmattae and 167 specimens of B. ephippiatus were included in the analysis. Nine specimens of B. impatiens and one specimen each of B. huntii and B. vosnesenskii were selected as outgroup taxa. A list of all samples used for phylogenetic analyses can be found in Online Resource 1.

Microsatellite analysis

To delimit distinct genetic groups within the B. ephippiatus-B. wilmattae complex, extensive sampling was conducted throughout Mexico and Guatemala. Samples were collected from one to three sites approximately three kilometers apart within a single population. A maximum of 20 samples was collected from each population, and each population sampled was at least 30 km apart. This scheme was used to minimize repeated sampling of individuals from the same colony. Samples previously genotyped in Duennes et al. (2012) were also included and genotyped at an additional 4 loci. In total, 1917 female samples of B. ephippiatus and B. wilmattae were genotyped at 12 microsatellite loci (Online Resource 1).

Geometric morphometric analysis

To test whether genetic structure has shaped morphology in this group, geometric morphometric data were collected from dorsal left forewings of B. ephippiatus and B. wilmattae. The wings were removed from 606 worker caste specimens (Online Resource 1) across the distribution of the species complex. Following Aytekin et al. (2007), 20 landmarks on the forewing were mapped and analyzed (Online Resource 2, see details below).

Phylogenetic inference

Cytochrome oxidase I

An 811 base pair fragment of the cytochrome oxidase I (COI) gene was amplified from 254 specimens (Online Resource 1) using the primers RevmtR and FormtR (used in Duennes et al. 2012). These highly specific primers minimized amplification of mitochondrial insertions into the nuclear genome and span a variable region of the gene. Other mitochondrial (cytochrome oxidase B; 16S ribosomal RNA) and nuclear (Phosphoenolpyruvate carboxykinase, [PEPCK]; carbamoyl-phosphate synthetase 2, aspartate transcarbamylase, and dihydroorotase [CAD]) genes were explored but lacked informative nucleotide variation (Duennes et al. 2012 for nuclear genes; data not shown for mitochondrial genes).

DNA extraction, PCR, and DNA sequencing

A single foreleg was used for DNA extraction to generate template DNA for both COI amplification and microsatellite genotyping. Tissue was digested and DNA extracted in 150 μL of 5% Chelex® 100 resin (Bio-Rad, Hercules, CA) and 3 μL of Proteinase K (20 mg/μL) for 60 min at 55 °C, 15 min at 99 °C, 1 min at 37 °C, and then 15 min at 99 °C using a thermocyler. Standard conditions for PCR amplification were an initial denaturation step of 95 °C for 3 min; 35 cycles of denaturation for 60 s at 94 °C, annealing for 60 s at 48–57 °C, and elongation for 60 s at 72 °C, and a final extension of 3 min at 72 °C, 25 μL. PCR reactions were conducted in 5 μL of 5X GoTaq® reaction buffer (Promega, Fitchburg, WI), 1.875 mM MgCl2, 0.2 mM each dNTP, 10μL of each primer and 0.4 U of GoTaq® DNA polymerase (Promega) with 2.5 μL of genomic DNA. PCR products were purified using ExoSAP-IT® (Affymetrix, Santa Clara, CA). BigDye® Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA) was used for sequencing of PCR products with the corresponding primers. Sequencing was performed at the W.M. Keck Center for Comparative and Functional Genomics at the University of Illinois using an ABI 3730XL (Applied Biosystems) capillary sequencer. Tissue was obtained from the specimens by removing one of the forelegs. All samples are housed either at the University of Illinois (stored in 95–100% ethanol at 4 °C) or at El Colegio de la Frontera Sur (ECOSUR) (pinned and dried).

Alignment and phylogenetic analysis

DNA sequences were edited in Geneious 8.1.7 (Biomatters Ltd) and aligned using MUSCLE (Edgar 2004). The HKY + I model was selected for Bayesian phylogenetic inference based on Bayesian information criteria (BIC) calculated by jModelTest 2.1.7 (Darriba et al. 2012). For phylogenetic inference, the aligned COI sequences were analyzed using MrBayes 3.2.5 (Ronquist et al. 2012) with 10,000,000 generations, four chains, flat priors and sampling trees every 1,000 generations. Log-likelihood plots of the two parallel runs from one MrBayes analysis were compared using Tracer 1.6.0 (Rambaut et al. 2014) and discarded as burnin the first 2,500,000 generations (2500 trees). The two runs were combined and the posterior probability support for each node of the consensus tree was calculated. Pairwise FST values between clades and average number of nucleotide differences within and between clades from the Bayesian phylogeny were calculated with DnaSPv5.10.1 (Librado and Rozas 2009). In addition to using Bayesian methods to reconstruct relationships, a parsimony haplotype network was generated with TCS1.21 (Clement et al. 2000) using the default parsimony connection limit of the program.

Species delimitation

To establish a threshold for species determination with sequence data, the general mixed Yule-coalescent model (GMYC) was implemented in a Bayesian framework with the program bGMYC (Reid and Carstens 2012). First, unique haplotypes were identified with COLLAPSE 1.2 (accessed 2016: http://www.softpedia.com/get/Science-CAD/Collapse.shtml); individuals with identical haplotypes were subsequently removed so that only a single representative of each haplotype remained in the dataset (N = 48 remaining taxa). This reduced dataset was also run through jModelTest 2.1.7 (Darriba et al. 2012); Bayesian information criteria (BIC) selected HKY + I as the best fit model for the data. To build ultrametric trees for bGMYC, BEAST 1.8.3 (Drummond et al. 2012) was run with the uncorrelated lognormal clock model and constant-size coalescent process tree-speciation prior for 10,000,000 generations, sampling trees every 1000 generations. To account for phylogenetic uncertainty when examining the GMYC model, the last 100 trees generated were used to run bGMYC. The analysis was run for 50,000 generations with a burnin of 40,000 generations and a thinning interval of 100. The MCMC chain was visually inspected for proper mixing and stationarity to determine the number of generations to run and to set the burnin.

Microsatellite analysis

Microsatellite genotyping

To identify areas of restricted gene flow and genetic structure within this widespread species group, 1917 female specimens were genotyped at twelve microsatellite loci using the following published PCR primers: B10, B124, B126 (Estoup et al. 1995); B96, B100, B131, B132 (Estoup et al. 1996); BT10, BL13, BT30, BT28 (Reber-Funk et al. 2006); BTMS0125 (Stolle et al. 2009). A total of twenty microsatellite loci were tested for use with B. ephippiatus and B. wilmattae, but these twelve markers were selected for their consistent amplification across multiple Bombus species. The amplification of the twelve loci chosen was also tested randomly via re-genotyping a subset of samples to look for genotyping errors, none of which were detected. To examine the possible presence of null alleles in our dataset, GENEPOP 4.2 (Rousset 2008) was used to test for deviations from Hardy–Weinberg equilibrium (HWE) via heterozygote deficiency. Markov chain parameters for all tests were set to 1000 for dememorization at 100 batches with 1000 iterations per batch. We tested all populations with sample sizes greater than ten, as well as all loci. No populations deviated significantly from HWE (P > 0.01). Of the twelve loci selected, only locus B10 displayed heterozygote deficiency (P = 0.0011); its removal from the dataset did not affect population inference results. These HWE tests as well as the amplification consistency of the selected loci suggest the problem of null allleles is not a strong factor in this dataset. PCR reaction protocols and thermal cycling conditions are described in Lozier and Cameron (2009). Final PCR products with flouresecently-tagged probes were purified and genotyped at the high throughput DNA facility, W.M. Keck Center at the University of Illinois using ABI 3730xl capillary DNA analyzer (Applied Biosystems). Genotypes were scored manually with the Geneious Microsatellite Plugin 1.4 in Geneious 8.1.7 (Biomatters Ltd) using the same allele bin-set for all species. A random subset of samples was genotyped a second time to check accuracy of allele identification; no inconsistencies were observed. For consistency between datasets, all of the microsatellite genotypes for B. ephippiatus and B. wilmattae taken from Duennes et al. (2012) were re-scored with the Geneious Microsatellite Plugin 1.4 in Geneious 8.1.7 (Biomatters Ltd) for this study (Online Resource 1). The same thermocyclers and sequencing machines were used for amplification and sequencing in this study as in Duennes et al. (2012).

Population differentiation

All microsatellite data were tested for population differentiation using STRUCTURE 2.3.4 (Pritchard et al. 2000). Multiple analyses were run with different groups of samples because it is known that the assignment of individuals to genetic groups by STRUCTURE can be strongly influenced by sample size (Kalinowski 2011). The following groups of samples were assessed for the following K values: all samples (N = 1917; K = 2–7), only the samples that have also been sequenced for the COI fragment (N = 225; K = 2–7), equal sample sizes for each region (Sierra Madre Occidental, Sierra Madre Oriental, Trans-Mexican Volcanic Belt, Sierra Madre del Sur, two for the sympatric groups south of the IT through Honduras, Costa Rica; N = 119; K = 2–7), only the samples from the Mexican state of Chiapas, Guatemala, and Honduras (N = 664). All analyses were run with the model parameter defaults (admixture model with allele frequencies correlated among populations and no prior sample information) with a burnin of 150,000 generations followed by an additional 150,000 generations. Between three to six independent runs were conducted for each value of K tested. While the ΔK method (implemented in STUCTURE HARVESTER; Earl and vonHoldt 2012) was used to assess the optimal K value (Evanno et al. 2005) for each dataset, other values of K that corresponded to potential geographic barriers or biologically relevant factors were also considered and are reported below.

The Bayesian assignment program GENELAND 4.0.5 was also used to assess population structure (Guillot et al. 2012). Six analyses (Table 1) designed to test the influence of geographic and morphometric data on the program’s assignment of individuals to distinct genetic groups were run with the following parameters for 1,000,000 iterations with every 1,000th iteration saved: maximum rate of Poisson process at 100; uncertainty on spatial coordinates of 1; uncorrelated allele frequency, null allele and spatial models. Ten independent runs were computed for each analysis for K values from one to ten. The iteration with the highest log likelihood after successive burnin values of 100, 200, 300, 400 and 500 was chosen as the best fit to the data and the MCMC plot for that iteration was independently examined to determine stationarity and burnin for the results.
Table 1

Summary of the six analyses run in GENELAND

 

DNA

GPS

MORPH

N MEX

NUC CA

N

burnin

K (logLk)

1

+

+

+

+

+

606

300

8 (81,607.2106)

2

+

+

 

+

+

606

200

8 (−24,524.6806)

3

+

+

+

+

 

281

200

3 (40,686.0608)

4

+

+

 

+

 

1229

450

5 (−42,727.4268)

5

+

+

+

 

+

310

200

5 (40,414.5477)

6

+

+

  

+

624

300

6 (−26,724.7870)

If a “+” is indicated under “GPS,” a spatial dataset was included in the analysis that consisted of the GPS coordinates for each specimen in UTM format. If “+” is indicated under “MORPH,” a phenotype dataset was included in the analysis that consisted of the regression residuals of the 20 Procrustes superimposed landmark coordinates for each sample. “N” indicates the total sample size for each analysis and “burnin” indicates the number of iterations discarded before the MCMC chain reached stationarity (we assessed this individually for each analysis). Under “K (logLk)” are the K values for the replicates with the highest log likelihood values with their corresponding log likelihood values (after burnin) in parentheses

Geometric morphometric analysis

Imaging and landmark mapping

The left forewing was removed and mounted in glycerol on glass slides with glass slipcovers. All slide-mounted wings were imaged at 10X with a Leica microscope camera (DFC425) in Leica Application Suite 3.8.0. After the order of specimen images was randomized using tpsUtil 1.64 (Rohlf 2015b), the coordinates of twenty landmarks on the forewing (Online Resource 2) were mapped on the images using tpsDIG2.22 (Rohlf 2015a). C. Petranek removed, mounted and imaged all wings and mapped all landmark coordinates to ensure that landmarks were placed in the same location and to avoid researcher bias in the mapping of the landmarks.

MorphoJ analyses

The program MorphoJ1.06d (Klingenberg 2011) was used for all post-processing analyses of the landmark coordinate dataset. A Procrustes fit was performed to scale, rotate and superimpose landmark data across specimens, thereby eliminating all variation due to non-shape differences among individuals. A regression was then conducted with Procrustes coordinates (representing shape) as the dependent variable and log centroid size (an approximation of size) as the independent variable. To control for any size-based shape variation (allometry), the residuals from this regression were used for further analysis. To test whether wing shape variation corresponded to genetic differentiation, a canonical analysis of variance (CVA) was conducted. Each specimen was assigned to a haplotype group based on its genotypic assignment in the STRUCTURE analysis (>50% assignment to a STRUCTURE-determined K group using the K = 6 result from the analysis of the entire population genetic dataset [N = 1917]) and these groups were used as the classifier variables for the CVA. A permutation test for 1000 iterations was run to test the significance of the difference in mean shape between haplotype groups. Leave-one-out cross validation tests were also conducted on all pairwise comparisons of groups via discriminant function analysis (DFA) to assess the reliability of assignment of individuals to a haplotype group based on wing shape. CVA and DFA differ in that the CVA examines the relative separation of groups using the pooled variance -covariance matrix, while DFA examines the separation of each pair of groups using only the covariance matrices of the two groups in question for each comparison. CVA therefore focuses on the degree to which groups differ from one another, while DFA focuses on the degree of difference between groups and the probability of correct group membership for each specimen. Although CVA and DFA are robust to unequal sample size when the number of variables is high (in this case, 20 landmarks), the sample size of the Costa Rican lineage (N = 17) was much smaller than the sample size for all other haplotype groups, so CVA was also run on a randomly reduced dataset in which each haplotype group had an equal sample size of N = 17.

Results

COI phylogenetic relationships and concordance with genotypic divergences

Bayesian analysis of Bombus ephippiatus and B. wilmattae COI sequence data (Fig. 2a; Online Resource 3) recovers six major polytomous clades. All B. ephippiatus from north of the IT and a subset of B. ephippiatus from south of the IT to Honduras form one clade (Fig. 2a; clade i), with the Honduran B. ephippiatus and Mexican B. ephippiatus north of the IT basal to the Mexican samples south of the IT and Guatemala (Fig. 2a; clade ii). Five additional clades are recovered (Fig. 2a; clades iii–vii), each structured by species and geographic distribution.
Fig. 2

a Bayesian phylogeny of 811 bp of the COI locus for 254 samples of B. ephippiatus, B. wilmattae and their sister species. All nodes in the phylogeny are colored according to their assignment in b STRUCTURE analysis of 12 microsatellite loci from the same samples. Individuals in the STRUCTURE graph are ordered and labeled by species designation and geographic location. (Color figure online)

The parsimony haplotype network of the COI sequence fragment data (Fig. 3) is highly congruent with the Bayesian phylogeny. All strongly supported clades from the phylogeny correspond with distinct haplotype groups in the network (Fig. 3). The most divergent lineage in the network is the group from Costa Rica, with four steps separating it from a haplotype of B. wilmattae (Fig. 3). Clade iii, iv and v from the phylogeny are also distinct in the haplotype network (Fig. 3). Differences between groups in the network are also reflected in the pairwise FST (Online Resource 4) between clades and the average number of nucleotide differences between and among clades in the phylogeny (Online Resource 5). Both distance measures demonstrate high divergence of Costa Rican individuals. High divergences also separate the different sympatric clades from south of the IT to Honduras (Online Resource 4; Online Resource 5).
Fig. 3

Parsimony haplotype network of 811 bp of the COI locus for samples of B. ephippiatus and B. wilmattae. The size of the circles is relative to the number of samples possessing each haplotype and black dots along connection lines represent steps. Each haplotype is colored to the assignment of each sample in the STRUCTURE analysis of 12 microsatellite loci from the same samples presented in Fig. 2. Grey-shaded areas represent samples with COI sequence data that lack microsatellite data. (Color figure online)

A STRUCTURE analysis of microsatellite genotype data for samples that were also sequenced for COI (Fig. 2b) was also conducted for a direct comparison of the COI phylogeny (Fig. 2a) and haplotype network (Fig. 3). Using ΔK criteria, K = 4 is the best fit to the genotype data (Online Resource 6). The genetic groups recovered from this analysis are largely congruent with the results of the Bayesian phylogeny and the haplotype network, and correspond to geographic subclades (Figs. 2, 3). The only strong incongruence between the datasets is the mixed assignment of the Costa Rican individuals in the STRUCTURE analysis. While STRUCTURE could not assign the Costa Rican samples to a single K cluster, they all belong to the well-supported clade vi in the phylogeny (Fig. 2) and fell out as a distinct group in the haplotype network (Fig. 3).

Species delimitation with bGMYC

Outgroups (B. huntii, B. vosnesenskii, and B. impatiens) are assigned low posterior probabilities of belonging to the same species (PP = 0.00–0.50; Online Resource 7). Phylogenetic clades ii–vii receive low/moderate probability of being conspecific (PP = 0.50–0.90; Online Resource 7) when compared to each other. Within clade i not all taxa are assigned to the same species (Online Resource 7). B. ephippiatus from Mexico North of the IT are all assigned to the same species as each other, but eight taxa from that region area also assigned to the same species as clade ii, while the other seven taxa area assigned low/moderate probability of belonging to the same species as clade ii (Online Resource 7). The uncertain species determination of the taxa in clade i coincides with their polytomous placement within the clade in the Bayesian phylogeny (Fig. 2) as well as their multiple connections within the haplotype network (Fig. 3).

Genetic groups inferred from STRUCTURE analyses of the microsatellite genotype data

Using the ΔK criteria, K = 5 was the best fit for the complete genotype dataset (n = 1917) (Online Resource 8). The K groups assigned are highly congruent with geographic location. Cluster A (Fig. 4) predominates throughout the Sierra Madre Occidental, but also occurs in high frequency in samples from the western region of the Trans-Mexican Volcanic Belt where these two mountain ranges overlap (Fig. 4). Cluster B (Fig. 4) is found throughout the eastern side of Mexico in the Sierra Madre Oriental, the Sierra Madre del Sur and the eastern region of the Trans-Mexican Volcanic Belt. While the samples from the Trans-Mexican Volcanic Belt share genotypes with mountain ranges in the far western and eastern portions of the region, a distinct cluster C (Fig. 4) occurs in the middle of the range. The genetic structure presented in this analysis corresponds to the mountains of Mexico, but there are signatures of gene flow in samples where these mountain chains connect and overlap (Fig. 4).
Fig. 4

Results from the K = 5 STRUCTURE analysis of the microsatellite genotype data for all samples. Samples are separated by mountain range and ordered either North to South or East to West depending on the orientation of the mountain range. The inset map has WWF (Olson et al. 2001) ecoregions colored according to the genetic cluster assignment of the majority of samples present within each region. (Color figure online)

The genotypic composition of B. ephippiatus and B. wilmattae in Nuclear Central America (Nuc CA) and south of the IT through Honduras was distinct from populations north of the IT (Fig. 4). Two distinct sympatric clusters occur within Nuc CA: one that includes B. ephippiatus from south of the IT through Honduras (Fig. 4, cluster D) and another that includes B. ephippiatus from Mexico south of the IT and Honduras and all samples of B. wilmattae from south of the IT (Fig. 4, cluster E).

Cluster E (Fig. 4) separates into two groups in the smaller STRUCTURE analysis of samples with COI data (Fig. 2b) and is placed in multiple lineages in the COI phylogeny (Fig. 2a); the larger analysis of all samples identifies one panmictic cluster at higher values of K. While ΔK criteria selected K = 5 for this dataset, K = 6 suggests further genetic structure corresponding to the mountain ranges of Mexico (Online Resource 9); a new cluster (cluster F) is revealed that corresponds to the Sierra Madre Oriental and the eastern sides of the Trans-Mexican Volcanic Belt and the Sierra Madre del Sur. The net nucleotide distances between the clusters at K = 6 (Online Resource 10) imply that cluster E (B. wilmattae and B. ephippiatus from Nuc CA) and cluster D (B. ephippiatus from Nuc CA) are most divergent from the populations north of the IT. The northwestern Sierra Madre Oriental (cluster A) and the southern Sierra Madre del Sur (cluster B) also exhibit high differentiation (Online Resource 10).

Both the large analysis and the smaller analysis using samples with COI sequence data (Fig. 2b) did not always assign Costa Rican individuals to a single cluster, so a separate analysis with all sample sizes equal to that for Costa Rica (N = 17 for each of the seven main regions; see Methods; Online Resource 11) was run. At successive K values from 2 to 7, Costa Rica is defined as a distinct lineage from K = 5–7 (Online Resource 11). For K = 2–4, Costa Rica is grouped with the samples from Nuc CA (Online Resource 11).

A separate analysis with the samples from Nuc CA was also run (N = 664; Online Resource 12). Across K values, cluster D from the larger analysis remains the same, but cluster E separates into smaller groups (Online Resource 12). The results from K = 2 mirror those of the analysis containing all (N = 1917) samples; the samples separate into the same two clusters from the larger dataset. At K = 3, cluster E is separated into one group of B. ephippiatus and B. wilmattae from Chiapas and a second group of B. wilmattae from Chiapas and Guatemala and B. ephippiatus from Honduras (Online Resource 12). At K = 4, cluster E split into one group of B. ephippiatus from Chiapas, a group of B. wilmattae from Chiapas and a third group of B. wilmattae from Chiapas and Guatemala and B. ephippiatus from Honduras (Online Resource 12). At K = 5, cluster E is split into four groups: a group of B. ephippiatus from Chiapas, a group of B. wilmattae from Chiapas, a group of B. wilmattae from Chiapas and Guatemala and a fourth group of B. ephippiatus from Honduras (Online Resource 12; Online Resource 13).

Genetic groups inferred from the GENELAND analyses of the microsatellite genotype data

The analysis of all samples with wing morphometric data and genotype data (scenario 1 in Table 1) reveals eight distinct genetic clusters (Fig. 5). Two groups are differentiated north of the IT; the samples from the Sierra Madre Occidental separate into a cluster (8) distinct from the rest of Mexico north of the IT (4; Fig. 5). A single distinct group is assigned to the individuals from Costa Rica (7; Fig. 5). In Nuc CA, five groups are assigned to the samples (1, 2, 3, 5, 6; Fig. 5). When compared to the STRUCTURE analysis of all samples, GENELAND differentiates less genetic groups north of the IT and more genetic groups south of the IT (Online Resource 14).
Fig. 5

Map displaying the geographic distribution of each genetic lineage identified by GENELAND for scenario 1 (Table 1). Grey-shaded areas of the map represent WWF (Olson et al. 2001) ecoregions in which the species can be found. (Color figure online)

When the morphometric data are excluded from the analysis (scenario 2 in Table 1; Online Resource 15), the results are largely congruent with those that include morphometric data, with some minor differences (Online Resource 16). Because the results of scenario 1 and 2 are similar, the pairwise comparison of scenario 2 to the STRUCTURE analysis (Online Resource 17) yield similar results, with STRUCTURE identifying more groups north of the IT and fewer groups south of the IT (Online Resource 17).

GENELAND analyses of the samples from north of the IT with morphometric data (scenario 3) and without morphometric data (scenario 4) reveal three and five groups, respectively (Table 1), which correspond to geography. In scenario 3 (Online Resource 18), Sierra Madre Oriental, the Trans-Mexican Volcanic Belt and the eastern edge of the Sierra Madre del Sur form a group, the Sierra Madre del Sur and southeastern Trans-Mexican Volcanic Belt form a group, and the Sierra Madre Occidental is a distinct group. These groups also correspond to the same regions identified by STRUCTURE (Online Resource 19). When morphometric data are excluded (scenario 4), five genetic groups are identified by GENELAND (Online Resource 20). Group 1 occurs in the Sierra Madre del Sur, group 2 occurs in the Sierra Madre Oriental through to the eastern Sierra Madre del Sur, group 3 occurs in the Sierra Madre Occidental, group 4 occurs in the western Trans-Mexican Volcanic Belt and southern Sierra Madre Occidental and group 5 occurs in the Trans-Mexican Volcanic Belt (Online Resource 20). These groups are also highly congruent with the STRUCTURE clusters assigned to these samples (Online Resource 21).

When the samples from Nuc CA are analyzed separately, five groups are identified with morphometric data (scenario 5; Online Resource 22) and 6 groups are identified without morphometric data (scenario 6; Online Resource 23). With the exception of a few outlier samples, the five groups identified with morphometric data largely correspond to the five STRUCTURE groups identified in the analysis of the Nuc CA samples (Online Resource 24; Online Resource 12). When morphometric data are excluded (scenario 6; Online Resource 23), a sixth group is added and the distribution of samples among groups corresponds less so to the STRUCTURE results (Online Resource 25).

Morphological divergence inferred from wing morphometric data

A CVA of all morphometric data (N = 606) with samples classified by their genotype assignment reveals that wing shape difference between each pair of haplotype groups is statistically significant (Online Resource 26) and 67.36% of the variation in the data is explained in the first two canonical variates (Online Resource 27). All samples are correctly identified to their pre-assigned group via pairwise DFA except when cluster B (Sierra Madre del Sur) is compared to cluster F (Sierra Madre Oriental) (Online Resource 26). Costa Rica is most divergent in wing shape from all other haplotype groups (Online Resource 26; Online Resource 27). The Nuc CA lineage with both B. ephippiatus and B. wilmattae (cluster E) is the second most different from all groups (Online Resource 26; Online Resource 27). Of the three haplotype groups south of the IT, the Nuc CA group (cluster D) containing B. ephippiatus is the least different in wing shape from the samples north of the IT (Online Resource 26; Online Resource 27), which also reflects its close relationship to the group north of the IT, as shown in the COI phylogeny (Fig. 2a). The Sierra Madre Oriental and the Sierra Madre del Sur are the least different in wing shape (Online Resource 26). This close relationship is also reflected in the STRUCTURE analyses (Online Resource 10).

The CVA with equal samples sizes (N = 17 for each haplotype group; N = 119 total) demonstrates the same patterns seen in the larger dataset, with 72.55% of the variation in the data explained by the first two canonical variates (Table 2; Fig. 6). The only substantial difference between the results of the analyses is that no pairwise comparison between haplotypes north of the IT has a greater Mahalanobis distance than the comparisons between haplotypes south of the IT (Table 2). In the larger analysis, the Mahalanobis distance between the most northern (Sierra Madre Occidental) and the most southern (Sierra Madre del Sur) haplotypes north of the IT is greater than any comparison of the Nuc CA group (cluster D) containing B. ephippiatus to any haplotype group north of the IT (Table 2).
Table 2

Pairwise Mahalanobis distances between haplotypes groups calculated by MorphoJ for equal sample size haplotype groups (N = 17 for each haplotype group, N = 119 total)

Mahalanobis distance

A

B

8.4982a

Sierra Madre del Sur

Costa Rica

8.3784a

Sierra Madre Oriental

Costa Rica

8.1556a

Trans-Mexican Volcanic Belt

Costa Rica

7.2092a

Sierra Madre Occidental

Costa Rica

6.3474a

Nuc CA (eph/wilm)

Costa Rica

6.1282a

Sierra Madre del Sur

Nuc CA (eph/wilm)

6.0663a

Nuc CA (eph)

Costa Rica

5.9101a

Trans-Mexican Volcanic Belt

Nuc CA (eph/wilm)

5.7076a

Sierra Madre Oriental

Nuc CA (eph/wilm)

5.3348a

Sierra Madre Occidental

Nuc CA (eph/wilm)

5.165a

Nuc CA (eph)

Nuc CA (eph/wilm)

4.9026a

Sierra Madre Occidental

Nuc CA (eph)

4.7405a

Sierra Madre del Sur

Nuc CA (eph)

4.6114a

Trans-Mexican Volcanic Belt

Nuc CA (eph)

4.3754a

Sierra Madre Oriental

Nuc CA (eph)

4.3493a

Sierra Madre del Sur

Sierra Madre Occidental

3.9552a

Trans-Mexican Volcanic Belt

Sierra Madre del Sur

3.8866a

Sierra Madre Oriental

Sierra Madre Occidental

3.3786a

Trans-Mexican Volcanic Belt

Sierra Madre Occidental

3.2525a

Trans-Mexican Volcanic Belt

Sierra Madre Oriental

2.3874b

Sierra Madre del Sur

Sierra Madre Oriental

Comparisons are ordered from most to least different

Nuc CA Nuclear Central America, eph, B. ephippiatus, wilm, B. wilmattae

aThe permutation test P value was < 0.0001

bThe permutation test P value was < 0.05

Fig. 6

Graph of the canonical analysis of variance conducted on equal samples sizes from each region (N = 119). All samples are color-coded according to their genotypic assignment by the K = 6 analysis of all samples (N = 1917). Nuc CA (eph) Nuclear Central America B. ephippiatus, Nuc CA (eph/wilm) Nuclear Central America B. ephippiatus and B. wilmattae. (Color figure online)

Discussion

Utility of methods for species delimitation

Mitochondrial and nuclear sequence data have been widely used to examine Bombus genetic diversity worldwide (Koulianos and Schmid-Hempel 2000; Kawakita et al. 2004; Hines et al. 2006; Cameron et al. 2007; Williams et al. 2012, 2015), and also regionally in Europe (Ellis et al. 2005; Lecocq et al. 2011; Carolan et al. 2012; Lecocq et al. 2013, 2015a, b, c; Bossert et al. 2016), Asia (Williams et al. 2011; Hines and Williams 2012; Huang et al. 2015), South America (Santos Júnior et al. 2015; Françoso et al. 2016),) and North America (Cameron and Williams 2003; Sheffield et al. 2016). Population, landscape, and conservation genetic studies of Bombus (Europe: Chapman et al. 2003; Ellis et al. 2006; Darvill et al. 2006, 2010; Herrman et al. 2007; Kraus et al. 2009, 2011; Charman et al. 2010; Goulson et al. 2011; Carvell et al. 2012; Maebe et al. 2013; Dreier et al. 2014; United States: Lozier and Cameron 2009; Cameron et al. 2011; Lozier et al. 2011, 2013; Jha and Kremen 2013; Jha 2015) as well population genetics research on invasive Bombus species (New Zealand: Schmid-Hempel et al. 2007; Lye et al. 2011) have utilized microsatellite genotype data extensively to examine patterns of gene flow and isolation. In comparison, few studies have used sequence data and microsatellites in concert to examine species-level and population-level patterns of genetic structure (Europe: Estoup et al. 1996; Widmer et al. 1998; Widmer and Schmid-Hempel 1999; Moreira et al. 2015; Asia: Shao et al. 2004; South America: Francisco et al. 2016); a single study to date has investigated bumble bee genetic diversity in Mexico and Central America (Duennes et al. 2012).

In this study, one of our aims was to balance evidence of older divergence patterns based on sequence data with more recent patterns of gene flow and isolation using faster-evolving microsatellite genotypes to provide a more robust inference of species delineation in the B. ephippiatusB. wilmattae complex. The COI phylogeny and haplotype network presented here are largely congruent with the patterns present in the microsatellite genotype data, but there is evidence of contemporary gene flow in clusters D and E and contemporary genetic isolation in Mexico north of the IT that would not have been discerned from COI sequence data alone. The species delimitation model bGMYC assigned low posterior probabilities of conspecificity to the major clades in the COI phylogeny, but could not discern sublineages that, in separate analyses, display a complete lack of gene flow at microsatellite loci. Previous research has reported similar assignment problems with bGMYC and attributed it to introgression and incomplete lineage sorting within sequence data of butterfly taxa (Talavera et al. 2013). The recent diversification of B. ephippiatusB. wilmattae (~1 mya; Duennes et al. 2012) and the conflicting inferences between sequence data and microsatellite data suggest that incomplete lineage sorting and/or introgression are confounding sequence data in this group as well. While sequence data can be useful for delimiting bumble bee species with older divergences and species delimitation models can aid in choosing a threshold for species identification, future studies of recently diverging groups should consider independent lines of evidence (in addition to sequence data) that can provide a robust examination and estimation of contemporary gene flow between lineages (Schlick-Steiner et al. 2010).

Microsatellite genotype data provide detailed information on genetic diversity and gene flow within this complex, but we have demonstrated that caution should be taken when using genotype assignment programs to assess genetic diversity with microsatellites. The difficulty STRUCTURE had in assigning the Costa Rican samples to a single cluster, except when samples from all regions were reduced to equal those of Costa Rica, illustrates a pitfall in using the STRUCTURE algorithm to assign unequal population samples to genetic groups (Kalinowski 2011). However, GENELAND, which incorporates geospatial data, readily identifies Costa Rica as a unique genetic group in an analysis of all samples with morphometric data included and excluded. The results from GENELAND do not change notably when sample sizes are changed or samples are excluded, but fewer populations are identified by GENELAND when morphometric data are included in the analysis.

The analyses of the Nuc CA region show that STRUCTURE and GENELAND can differ substantially in their assignments. STRUCTURE consistently identifies cluster D as distinct from all other samples from Nuc CA, regardless of the K value being tested; this suggests that there is little to no gene flow between these lineages. In contrast, GENELAND does not identify cluster D as a unique lineage, suggesting admixture throughout Nuc CA. When a map of the groups assigned by GENELAND is compared to a map of the STRUCTURE assignments, it is clear that GENELAND weights geographic proximity much greater than genotype when assigning samples to groups. The ability to include spatial and phenotypic information suggests that GENELAND provides a more holistic picture of evolutionary history, yet here GENELAND appears to prioritize geospatial data over genetic data when multiple cryptic, sympatric lineages are included (i.e. populations deviate from Hardy–Weinberg equilibrium).

This study also aimed to include non-molecular methods as additional evidence towards the delineation of unique groups. Discriminant function analysis of landmark data indicates significant differences in wing shape between all but one pair of genetic lineages and returned high hit ratios for cross-validation tests. While one can use discriminant analyses such as CVA and DFA to reliably differentiate between user-specified groups, these tests can detect significant differences even when the magnitude of morphological shape difference is miniscule, potentially supplying weak information for phylogenetic inference. Lecocq et al. (2015b) implemented a non-discriminant analysis (Schlick-Steiner et al. 2010) of bumble bee wing morphometric data to avoid the bias implicit to designating a priori species hypotheses, but the results suggested that all lineages within the Bombus lapidarius species complex were not distinguishable by wing shape. The results of Lecocq et al. (2015b) could therefore be explained by convergent or stabilizing selection on wing shape (Dockx 2007). However, the efficacy of non-discriminant methods should be further explored by examining wing shape divergence in other well-supported groups of bumble bees, as analyzing landmark data without a priori hypotheses may lack the statistical rigor to tell groups apart (Mutanen and Pretorius 2007).

Biogeographic patterns

When the results of our three independent datasets are considered as a whole, the B. ephippiatus lineage in Costa Rica is highly divergent from the rest of the complex. Analyses of all datasets identified the Costa Rican samples as a distinct group, with the exception of some STRUCTURE analyses. The strong differentiation of the Costa Rican lineage suggests that the Nicarguan Depression is an important isolating mechanism for this species complex, as has been demonstrated in several vertebrate taxa (Castoe et al. 2009; Daza et al. 2010; Gutiérrez-García and Vázquez-Domínguez 2013).

For such a small region, Nuclear Central America (south of the IT and north of the ND) contains a remarkable amount of genetic diversity. The volcanoes and mountains chains in this region have generated a mosaic of unique habitats across elevations, which in turn led to high levels of diversification for many taxa (Strecker et al. 2004; Crawford and Smith 2005; Gutiérrez-García and Vázquez-Domínguez 2012; Suárez-Atilano et al. 2014; Pérez and Vázquez-Domínguez 2015). The separation of clade ii Chiapas and Guatemala samples from the Honduras samples at the base of clade i, despite being placed into the same K cluster in the STRUCTURE analysis of the genotype data, suggests that this might be a case of secondary contact in which the ancestral distribution arose south of the IT, dispersed into northern Mexico and then dispersed south, back across the IT, introgressing with the ancestral population in the mid-elevation Central American pine-oak forests (Fig. 1). The large amount of COI genetic diversity (also present in the microsatellite data) in the four groups comprising cluster E provides further support for the hypothesis that these bees have been in Nuc CA much longer than the taxa comprising cluster D. These lineages, which correspond to the high elevation Chiapas montane forest, Sierra Madre de Chiapas, and Central American montane forest ecoregions (Fig. 1), might even represent the ancestral range of the group with retention of ancestral polymorphism. The cluster E subgroups also show that the Central Depression might restrict gene flow between populations, a result that could not have been discerned from the smaller sampling of Duennes et al. (2012). Divergence estimates from Duennes et al. (2012) suggest that both of these possible scenarios could have occurred during Pleistocene glacial cycles; studies across animal and plant taxa inhabiting montane regions of Mexico and Central America suggest multiple dispersal events across the IT have occurred during the late Pliocene and Pleistocene (Barber and Klicka 2010; Ornelas et al. 2013).

Bombus ephippiatus in Mexico north of the IT also contain substantial genetic differentiation that corresponds with the elevation barriers imposed by the mountain ranges of this region. In contrast to Nuclear Central America, pine-oak forests are continuous throughout north Mexican mountains and converge with other ranges via subalpine forests. Gene flow between these distinct populations is greatest where the separate mountain chains meet each other; this pattern has been found in other taxa within the region (Ornelas et al. 2013; Mastretta-Yanes et al. 2015). While geographic distance seems to have a strong impact on differentiation in northern Mexico where mountain chains and habitat are more continuous, habitat and elevation seem to be a more important isolating mechanism in Nuclear Central America.

While some population genetic studies of Bombus have found less genetic variation across much larger geographic areas (Lozier et al. 2011; Moreira et al. 2015; Francisco et al. 2016), other research reveals population structure within bumble bee species. Multiple studies suggest that oceanic barriers between island and mainland populations have facilitated genetic differentiation within bumble bees (Widmer et al. 1998; Shao et al. 2004; Darvill et al. 2006, 2010; Kraus et al. 2009; Charman et al. 2010; Lye et al. 2011; Goulson et al. 2011; Lozier et al. 2011; Hines and Williams 2012; Jha 2015; Moreira et al. 2015; Lecocq et al. 2015c; Williams et al. 2015; Francisco et al. 2016). Like the lineages in this complex (Duennes et al. 2012), many bumble bee species also exhibit genetic differentiation corresponding to unique eco-regions across latitudinal/longitudinal and elevational space, implying that mountainous regions in concert with glacial refugia are responsible for generating this diversity (Widmer and Schmid-Hempel 1999; Hines and Williams 2012; Lecocq et al. 2013, 2015; Lozier et al. 2013; Santos Júnior et al. 2015; Françoso et al. 2016; Sheffield et al. 2016).

Conservation implications

Upon the discovery of cryptic diversity, species once thought to have wide distributions and to be of least concern for conservation can actually comprise multiple species with contracted ranges and small population sizes that are in need of conservation (Bickford et al. 2007; Funk et al. 2012; Niemiller et al. 2012, 2013). Currently, B. ephippiatus is listed as “least concern” by the International Union for the Conservation of Nature (IUCN) Red List of Threatened Species (B. wilmattae was considered conspecific for this assessment; Duennes and Vandame 2015). In light of the genetic and morphometric patterns we see here, a taxonomic revision of this species complex will be important, as well as a subsequent conservation status assessment: a reassessment for the IUCN Red List could result in different risk categorizations for each species within this complex. We strongly encourage phylogenetic and population genetic studies of the other seventeen species (Labougle 1990) present in Mexico and Central America to develop more accurate and effective conservation assessments for these important native pollinators inhabiting fragile cloud forest habitat (Ornelas et al. 2013).

Within Mexico alone, twelve independent companies across five different states are rearing B. ephippiatus for commercial pollination of greenhouse crops (Asociación Mexicana de Criadores de Abejorros Nativos, A.C.). Although raising B. ephippiatus as a native species is a more sustainable alternative to the non-native North American B. impatiens currently used for commercial pollination in Mexico and Central America, the inter-regional movement of B. ephippiatus colonies with unique population structure could be detrimental to the other three un-described species in this complex as well as to native population diversity within northern Mexico. Not only does their movement pose the threat of facilitating the spread of potential diseases (the same B. ephippiatus specimens used for this study demonstrated relatively high pathogen infection compared to other Mexican Bombus spp.; Gallot-Lavallée et al. 2016), but commercial B. ephippiatus could outcompete native populations for resources and even cause genetic pollution of the native populations through interbreeding (Goulson 2010; Kraus et al. 2011; Williams et al. 2012). The results presented here demonstrate that B. ephippiatus exhibits substantial population structure across its range and we suggest that bombiculture companies restrict the movement of colonies to within the distinct mountain ranges of Mexico and not move colonies between ranges, thereby preserving the native diversity across the species’ range.

Consequences for species delimitation

The genetic and morphometric results presented here demonstrate that a taxonomic revision of the B. ephippiatusB. wilmattae complex is needed. Mexico north of the IT is clearly a distinct taxonomic unit with genetically unique populations in the Sierra Madre Occidental, Sierra Madre Oriental, Sierra Madre del Sur and the Trans-Mexican Volcanic Belt. There are at least two sympatric taxonomic units from Mexico south of the IT to Honduras, with one of these lineages containing the previously recognized B. wilmattae as well as substantial population structure corresponding to the Central Depression. A fourth distinct taxonomic unit inhabits Costa Rica, and presumably also Panama based on the consistent color pattern phenotype present through these regions. The results presented here highlight the importance of thorough investigations of cryptic species diversity for the conservation of native species.

Notes

Acknowledgements

We are grateful to Zachary Cheviron, Kevin Johnson, Ripan Malhi, Zoi Rapti and especially Kyle Parks for helpful comments on the manuscript and Eunice Enríquez for her support in collecting samples from Guatemala.

Funding

This study was funded by the Illinois Chapter of Achievement Rewards for College Scientists (ARCS), Inc. to M.A. Duennes, Comisión Nacional para el Uso y el Conocimiento de la Biodiversidad (CONABIO Grant JE016) to R. Vandame, Consejo Nacional de Ciencia y Tecnología (CONACYT Grant 106043) to R. Vandame and the National Science Foundation U.S.-Mexico International Program (award no. 0503834) to S.A. Cameron and J. Nieh.

Supplementary material

10592_2016_903_MOESM1_ESM.docx (81.7 mb)
Supplementary material 1 (DOCX 83630 kb)

References

  1. Anducho-Reyes MA, Cognato AI, Hayes JL, Zúñiga G (2008) Phylogeography of the bark beetle Dendroctonus mexicanus Hopkins (Coleoptera: Curculionidae: Scolytinae). Mol Phylogenet Evol 49:930–940PubMedCrossRefGoogle Scholar
  2. Arbetman MP, Meeus I, Morales CL, Aizen MA, Smagghe G (2013) Alien parasite hitchhikes to Patagonia on invasive bumblebee. Biol Invasions 15:489–494CrossRefGoogle Scholar
  3. Aytekin MA, Terzo M, Rasmont P, Çağatay N (2007) Landmark based geometric morphometric analysis of wing shape in Sibiricobombus Vogt (Hymenoptera: Apidae: Bombus Latreille). Annales de la Société entomologique de France (N.S.) 43:95–102CrossRefGoogle Scholar
  4. Barber BR, Klicka J (2010) Two pulses of diversification across the Isthmus of Tehuantepec in a montane Mexican bird fauna. Proc R Soc Lond B 277:2675–2681CrossRefGoogle Scholar
  5. Baselga A, Recuero E, Parra-Olea G, García-París M (2011) Phylogenetic patterns in zopherine beetles are related to ecological niche width and dispersal limitation. Mol Ecol 20:5060–5073PubMedCrossRefGoogle Scholar
  6. Bickford D, Lohman DJ, Sodhi NS, Ng PKL, Meier R, Winker K, Ingram KK, Das I (2007) Cryptic species as a window on diversity and conservation. Trends Ecol Evol 22:148–155PubMedCrossRefGoogle Scholar
  7. Bossert S, Gereben-Krenn BA, Neumayer J, Schneller B, Krenn HE (2016) The cryptic Bombus lucorum complex (Hymenoptera: Apidae) in Austria: phylogeny, distribution, habitat usage and a climatic characterization based on COI sequence data. Zool Stud 55:1–15Google Scholar
  8. Bourret V, O’Reilly PT, Carr JW, Berg PR, Bernatchez L (2011) Temporal change in genetic integrity suggests loss of local adaptation in a wild Atlantic salmon (Salmo salar) population following introgression by farmed escapees. Heredity 106:500–510PubMedPubMedCentralCrossRefGoogle Scholar
  9. Cameron SA, Williams PH (2003) Phylogeny of bumble bee in the New World subgenus Fervidobombus (Hymenoptera: Apidae): congruence of molecular and morphological data. Mol Phylogenet Evol 34:321–331Google Scholar
  10. Cameron SA, Hines HM, Williams PH (2007) A comprehensive phylogeny of the bumble bees (Bombus). Biol J Linn Soc 91:161–188CrossRefGoogle Scholar
  11. Cameron SA, Lozier JD, Strange JP, Koch JB, Cordes N, Solter LF, Griswold TL (2011) Patterns of widespread decline in North American bumble bees. Proc Natl Acad Sci 108:662–667PubMedPubMedCentralCrossRefGoogle Scholar
  12. Carolan JC, Murray TE, Fitzpatrick Ú, Crossley J, Schmidt H, Cederberg B, McNally L, Paxton RJ, Williams PH, Brown MJF, Steinke D (2012) Colour patterns do not diagnose species: quantitative evaluation of a DNA barcoded cryptic bumblebee complex. PLoS ONE 7(1):e29251PubMedPubMedCentralCrossRefGoogle Scholar
  13. Castoe TA, Daza JM, Smith EN, Sasa MM, Kuch U, Campbell JA, Chippindale PT, Parkinson CL (2009) Comparative phylogeography of pitvipers suggests a consensus of ancient Middle American highland biogeography. J Biogeogr 36:88–103CrossRefGoogle Scholar
  14. Chapman RE, Wang J, Bourke AFG (2003) Genetic analysis of spatial foraging patterns and resource sharing in bumble bee pollinators. Mol Ecol 12:2801–2808PubMedCrossRefGoogle Scholar
  15. Charman TG, Sears J, Green RE, Bourke AFG (2010) Conservation genetics, foraging distance and nest density of the scarce Great Yellow Bumblebee (Bombus distinguendus). Mol Ecol 19:2661–2674PubMedCrossRefGoogle Scholar
  16. Clement M, Posada D, Crandall K (2000) TCS: a computer program to estimate gene genealogies. Mol Ecol 10:1657–1660CrossRefGoogle Scholar
  17. Cracraft J, Prum RO (1988) Patterns and processes of diversification: speciation and historical congruence in some Neotropical birds. Evolution 42:603–620CrossRefGoogle Scholar
  18. Crawford A, Smith E (2005) Cenozoic biogeography and evolution in direct-developing frogs of Central America (Leptodactylidae: Eleutherodactylus) as inferred from a phylogenetic analysis of nuclear and mitochondrial genes. Mol Phylogenet Evol 35:536–555PubMedCrossRefGoogle Scholar
  19. Darriba D, Taboada GL, Doallo R, Posada D (2012) jModelTest 2: more models, new heuristics and parallel computing. Nat Methods 9:772PubMedPubMedCentralCrossRefGoogle Scholar
  20. Darvill B, Ellis JS, Lye GC, Goulson D (2006) Population structure and inbreeding in a rare and declining bumblebee, Bombus muscorum (Hymenoptera: Apidae). Mol Ecol 15:601–611PubMedCrossRefGoogle Scholar
  21. Darvill B, O’Connor S, Lye GC, Waters J, Lepais O, Goulson D (2010) Cryptic differences in dispersal lead to differential sensitivity to habitat fragmentation in two bumblebee species. Mol Ecol 19:53–63PubMedCrossRefGoogle Scholar
  22. Daza JM, Castoe TA, Parkinson CL (2010) Using regional comparative phylogeographic data from snake lineage to infer historical processes in Middle America. Ecography 33:343–354Google Scholar
  23. Dockx C (2007) Directional and stabilizing selection on wing size and shape in migrant and resident monarch butterflies, Danaus plexippus (L.), in Cuba. Biol J Linn Soc 92:605–616CrossRefGoogle Scholar
  24. Dorn PL, Calderon C, Melgar S, Moguel B, Solorzano E, Dumonteil E, Rodas A, de la Rua N, Garnica R, Monroy C (2009) Two distinct Triatoma dimidiata (Latreille, 1811) taxa are found in sympatry in Guatemala and Mexico. PLoS Neglect Trop Dis 3:e393CrossRefGoogle Scholar
  25. Dreier S, Redhead JW, Warren IA, Bourke AFG, Heard MS, Jordan WC, Sumner S, Wang J, Carvell C (2014) Fine-scle spatial genetic structure of common and declining bumble bees across an agricultural landspace. Mol Ecol 23:3384–3395PubMedPubMedCentralCrossRefGoogle Scholar
  26. Drummond AJ, Suchard MA, Xie D, Rambaut A (2012) Bayesian phylogenetics with BEAUti and BEAST 1.7. Mol Biol Evol 29:1969–1973PubMedPubMedCentralCrossRefGoogle Scholar
  27. Duennes MA, Vandame R. 2015. Bombus ephippiatus. The IUCN Red List of Threatened Species. e.T21215149A21215217. doi: 10.2305/IUCN.UK.2015-4.RLTS.T21215149A21215217.en. Downloaded on 29 March 2016
  28. Duennes MA, Lozier JD, Hines HM, Cameron SA (2012) Geographical patterns of genetic divergence in the widespread Mesoamerican bumble bee Bombus ephippiatus (Hymenoptera: Apidae). Mol Phylogenet Evol 64:219–231PubMedCrossRefGoogle Scholar
  29. Earl DA, vonHoldt BM (2012) STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conserv Genet Resour 4:359–361CrossRefGoogle Scholar
  30. Edgar RC (2004) MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucl Acids Res 32:1792–1797PubMedPubMedCentralCrossRefGoogle Scholar
  31. Ellis JS, Knight ME, Goulson D (2005) Delineating species for conservation using mitochondrial sequence data: the taxonomic status of two problematic Bombus species (Hymenoptera: Apidae). J Insect Conserv 9:75–83CrossRefGoogle Scholar
  32. Ellis JS, Knight ME, Darvill B, Goulson D (2006) Extremely low effective population sizes, genetic structuring and reduced genetic diversity in a threatened bumblebee species, Bombus sylvarum (Hymenoptera: Apidae). Mol Ecol 15:4375–4386PubMedCrossRefGoogle Scholar
  33. Escalante P, Navarro-Sigüenza AG, Peterson AT (1993) A geographic, ecological and historical analysis of land bird diversity in Mexico. In: Ramamoorthy TP, Bye R, Lot A, Fa J (eds) Biological diversity in Mexico: origins and distributions. Oxford University Press, New York, pp 281–307Google Scholar
  34. Estoup A, Scholl A, Pouvreau A, Solignac M (1995) Monoandry and polyandry in bumble bees (Hymenoptera: Bombinae) as evidenced by highly variable microsatellites. Mol Ecol 4:89–93PubMedCrossRefGoogle Scholar
  35. Estoup A, Solignac M, Cornuet JM, Goudet J, Scholl A (1996) Genetic differentiation of continental and island populations of Bombus terrestris (Hymenoptera: Apidae) in Europe. Mol Ecol 5:19–31PubMedCrossRefGoogle Scholar
  36. Evanno G, Regnaut S, Goudet J (2005) Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Mol Ecol 14:2611–2620PubMedCrossRefGoogle Scholar
  37. Francisco FO, Santiago LR, Mizusawa YM, Oldroyd BP, Arias MC (2016) Genetic structure of island and mainland populations of a Neotropical bumble bee species. J Insect Conserv 20:383–394CrossRefGoogle Scholar
  38. Françoso E, de Oliveira FF, Arias MC (2016) An integrative approach identifies a new species of bumblebee (Hymenoptera: Apidae: Bombini) from northeastern Brazil. Apidologie 47:171–185CrossRefGoogle Scholar
  39. Funk WC, Caminer M, Ron SR (2012) High levels of cryptic species diversity uncovered in Amazonian frogs. Proc R Soc B 279:1806–1814PubMedCrossRefGoogle Scholar
  40. Gallot-Lavallée M, Schmid-Hempel R, Vandame R, Vergara CH, Schmid-Hempel P (2016) Large scale patterns of abundance and distribution of parasites in Mexican bumblebees. J Invertebr Pathol 133:73–82PubMedCrossRefGoogle Scholar
  41. García-Moreno J, Navarro-Sigüenza AG, Peterson AT, Sánchez-González LA (2004) Genetic variation coincides with geographic structure in the common bush-tanager (Chlorospingus ophthalmicus) complex from Mexico. Mol Phylogenet Evol 33:186–196PubMedCrossRefGoogle Scholar
  42. García-Moreno J, Cortés N, García-Deras G, Hernández-Baños B (2006) Local origin and diversification among Lampornis hummingbirds: a Mesoamerican taxon. Mol Phylogenet Evol 38:488–498PubMedCrossRefGoogle Scholar
  43. Goulson D (2010) Impacts of non-native bumblebees in Western Europe and North America. Appl Entomol Zool 45:7–12CrossRefGoogle Scholar
  44. Goulson D, Kaden JC, Lepais O, Lye GC (2011) Population structure, dispersal and colonization history of the garden bumblebee Bombus hortorum in the Western Isles of Scotland. Conserv Genet 12:867–879CrossRefGoogle Scholar
  45. Guillot G, Renaud S, Ledevin R, Michaux J, Claude J (2012) A unifying model for the analysis of phenotypic, genetic and geographic data. Syst Biol 61:897–911PubMedCrossRefGoogle Scholar
  46. Gutiérrez-García T, Vázquez-Domínguez E (2012) Biogeographically dynamic genetic structure bridging two continents in the monotypic Central American rodent, Ototylomys phyllotis. Biol J Linn Soc 107:593–610CrossRefGoogle Scholar
  47. Gutiérrez-García T, Vázquez-Domínguez E (2013) Consensus between genes and stones in the biogeographic and evolutionary history of Central America. Quatern Res 79:311–324CrossRefGoogle Scholar
  48. Halffter G (1987) Biogeography of the montane entomofauna of Mexico and Central America. Annu Rev Entomol 32:95–114CrossRefGoogle Scholar
  49. Heilprin A (1887) The geographical and geological distribution of animals. Kegan Paul, Trench and Co, LondonCrossRefGoogle Scholar
  50. Hines HM, Williams PH (2012) Mimetic colour pattern evolution in the highly polymorphic Bombus trifasciatus (Hymenoptera: Apidae) species complex and its comimics. Zool J Linn Soc 166:805–826CrossRefGoogle Scholar
  51. Hines HM, Cameron SA, Williams PH (2006) Molecular phylogeny of the bumble bee subgenus Pyrobombus (Hymenoptera: Apidae: Bombus) with insights into gene utility for lower-level analysis. Invertebr System 20:289–303CrossRefGoogle Scholar
  52. Huang J, Wu J, An J, Williams PH (2015) Newly discovered colour-pattern polymorphism of Bombus koreanus females (Hymenoptera: Apidae) demonstrated by DNA barcoding. Apidologie 46:250–261CrossRefGoogle Scholar
  53. Jha S (2015) Contemporary human-altered landscapes and oceanic barriers reduce bumble bee gene flow. Mol Ecol 24:993–1006PubMedCrossRefGoogle Scholar
  54. Jha S, Kremen C (2013) Urban land use limits regional bumble bee gene flow. Mol Ecol 22:2483–2495PubMedCrossRefGoogle Scholar
  55. Kalinowski ST (2011) The computer program STRUCTURE does not reliably identify the main genetic clusters within species: simulations and implications for human population structure. Heredity 106:625–632PubMedCrossRefGoogle Scholar
  56. Kawakita A, Sota T, Ito M, Ascher JS, Tanaka H, Kato M, Roubik DW (2004) Phylogeny, historical biogeography, and character evolution in bumble bees (Bombus: Apidae) based on simultaneous analysis of three nuclear gene sequences. Mol Phylogenet Evol 31:799–804PubMedCrossRefGoogle Scholar
  57. Kerhoulas NJ, Arbogast BS (2010) Molecular systematics and Pleistocene biogeography of Mesoamerican flying squirrels. J Mammal 91:654–667CrossRefGoogle Scholar
  58. Klingenberg CP (2011) MorphoJ: an integrated software package for geometric morphometrics. Mol Ecol Resour 11:353–357PubMedCrossRefGoogle Scholar
  59. Kondo NI, Yamanaka D, Kanbe Y, Kunitake YK, Yoneda M, Tsuchida K, Goka K (2009) Reproductive disturbance of Japanese bumblebees by the introduced European bumblebee Bombus terrestris. Naturwissenschaften 96:467–475PubMedCrossRefGoogle Scholar
  60. Koulianos S, Schmid-Hempel P (2000) Phylogenetic relationships among bumble bees (Bombus, Latreille) inferred from mitochondrial cytochrome b and cytochrome oxidase I sequences. Mol Phylogenet Evol 14:335–341PubMedCrossRefGoogle Scholar
  61. Kraus FB, Wolf S, Moritz RFA (2009) Male flight distance and population substructure in the bumblebee Bombus terrestris. J Anim Ecol 78:247–252PubMedCrossRefGoogle Scholar
  62. Kraus FB, Szentgyörgyi H, Rożej E, Rhode M, Moroń D, Woyciechowski M, Moritz RFA (2011) Greenhouse bumblebees (Bombus terrestris) spread their genes into the wild. Conserv Genet 12:187–192CrossRefGoogle Scholar
  63. Labougle JM (1990) Bombus of México and Central America (Hymenoptera, Apidae). Univ Kansas Sci Bull 54:35–73Google Scholar
  64. Labougle JM, Ito M, Okazawa T (1985) The species of the genus Bombus (Hymenoptera: Apidae) of Chiapas, Mexico and Guatemala; with a morphometric and altitudinal analysis. Folia Entomólgica Mexicana 64:55–72Google Scholar
  65. Lecocq T, Lhomme P, Michez D, Dellicour S, Valterová Rasmont P (2011) Molecular and chemical characters to evaluate species status of two cuckoo bumblebees: Bombus barbutellus and Bombus maxillosus (Hymenoptera, Apidae, Bombini). Syst Entomol 36:453–469CrossRefGoogle Scholar
  66. Lecocq T, Dellicour S, Michez D, Lhomme P, Vanderplanck M, Valterová I, Rasplus J, Rasmont P (2013) Scent of a break-up: phylogeography and reproductive trait divergences in the red-tailed bumblebee (Bombus lapidarius). BMC Evol Biol 13:263PubMedPubMedCentralCrossRefGoogle Scholar
  67. Lecocq T, Brasero N, Martinet B, Valterová I, Rasmont P (2015a) Highly polytypic taxon complex: interspecific and intraspecific integrative taxonomic assessment of the widespread pollinator Bombus pascuorum Scopoli 1763 (Hymenoptera: Apidae). Syst Entomol 40:881–890CrossRefGoogle Scholar
  68. Lecocq T, Dellicour S, Michez D, Dehon M, Dewulf A, De Meulemeester T, Brasero N, Valterová I, Rasplus J, Rasmont P (2015b) Methods for species delimitation in bumblebees (Hymenoptera, Apidae, Bombus): towards an integrative approach. Zool Scripta 44:281–297CrossRefGoogle Scholar
  69. Lecocq T, Brasero N, De Meulemeester T, Michez D, Dellicour S, Lhomme P, de Jonghe R, Valterová I, Urbanová K, Rasmont P (2015c) An integrative taxonomic approach to assess the status of Corsican bumblebees: implications for conservation. Anim Conserv 18:236–248CrossRefGoogle Scholar
  70. Lecocq T, Coppée A, Michez D, Brasero N, Rasplus J, Valterová I, Rasmont P (2016) The alien’s identity: consequences of taxonomic status for the international bumblebee trade regulations. Biol Conserv 195:169–176CrossRefGoogle Scholar
  71. León-Paniagua L, Navarro-Sigüenza AG, Hernández-Baños BE, Morales JC (2007) Diversification of the arboreal mice of the genus Habromys (Rodentia: Cricetidae: Neotominae) in the Mesoamerican highlands. Mol Phylogenet Evol 42:653–664PubMedCrossRefGoogle Scholar
  72. Librado P, Rozas J (2009) DnaSP v5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics 25:1451–1452PubMedCrossRefGoogle Scholar
  73. Liebherr JK (1994) Biogeographic patterns of montane Mexican and Central American carabidae (Coleoptera). Can Entomol 126:841–860CrossRefGoogle Scholar
  74. Lobo JM, Halffter G (2000) Biogeographical and ecological factors affecting the altitudinal variation of mountainous communities of coprophagous beetles (Coleoptera: Scarabaeoidea): a comparative study. Ann Entomol Soc Am 93:115–126CrossRefGoogle Scholar
  75. Lozier JD, Cameron SA (2009) Comparative genetic analyses of historical and contemporary collections highlight contrasting demographic histories for the bumble bee Bombus pensylvanicus and B. impatiens in Illinois. Mol Ecol 18:1875–1886PubMedCrossRefGoogle Scholar
  76. Lozier JD, Strange JP, Steward IJ, Cameron SA (2011) Patterns of range-wide genetic variation in six North American bumble bee (Apidae: Bombus) species. Mol Ecol 20:4870–4888PubMedCrossRefGoogle Scholar
  77. Lozier JD, Strange JP, Koch JB (2013) Landscape heterogeneity predicts gene flow in a widespread polymorphic bumble bee, Bombus bifarius (Hymenoptera: Apidae). Conserv Genet 14:1099–1110CrossRefGoogle Scholar
  78. Lye GC, Lepais O, Goulson D (2011) Reconstructing demographic events from population genetic data: the introduction of bumblebees to New Zealand. Mol Ecol 20:2888–2900PubMedCrossRefGoogle Scholar
  79. Maebe K, Meeus I, Maharramov J, Grootaert P, Michez D, Rasmont P, Smagghe G (2013) Microsatellite analysis in museum samples reveals inbreeding before the regression of Bombus veteranus. Apidologie 44:188–197CrossRefGoogle Scholar
  80. Marshall CJ, Liehberr JK (2000) Cladistic biogeography of the Mexican transition zone. J Biogeogr 27:203–216CrossRefGoogle Scholar
  81. Mastretta-Yanes A, Moreno-Letelier A, Piñero D, Jorgensen TH, Emerson BC (2015) Biodiversity in the Mexican highlands and the interaction of geology, geography and climate within the Trans-Mexican Volcanic Belt. J Biogeogr 42:1586–1600CrossRefGoogle Scholar
  82. May-Itzá W, Quezada-Euán JJG, Medina Medina LA, Enríquez E, De la Rúa P (2010) Morphometric and genetic differentiation in isolated populations of the endangered Mesoamerican stingless bee Melipona yucatanica (Hymenoptera: Apoidea) suggest the existence of a two species complex. Conserv Genet 11:2079–2084CrossRefGoogle Scholar
  83. Mittermeier R, Myers N, Goettsch Mittermeier C (2000) Hotspots: earth’s biologically richest and most endangered terrestrial ecoregions. Conservation International, Washington, DCGoogle Scholar
  84. Morales CL, Arbetman M, Cameron SA, Aizen M (2013) Rapid ecological replacement of a native by invasive bumble bee species. Front Ecol Environ 11:529–534CrossRefGoogle Scholar
  85. Moreira AS, Horgan FG, Murray TE, Kakouli-Duarte T (2015) Population genetic structure of Bombus terrestris in Europe: isolation and genetic differentiation of Irish and British populations. Mol Ecol 24:3257–3268PubMedCrossRefGoogle Scholar
  86. Morrone JJ (2006) Biogeographic areas and transition zones of Latin America and the Caribbean Islands, based on panbiogeographic and cladistic analyses of the entomofauna. Annu Rev Entomol 51:467–494PubMedCrossRefGoogle Scholar
  87. Morse GE, Farrell BD (2005) Interspecific phylogeography of the Stator limbatus species complex: the geographic context of speciation and specialization. Mol Phylogenet Evol 36:201–213PubMedCrossRefGoogle Scholar
  88. Mulcahy DG, Morrill BH, Mendelson JR (2006) Historical biogeography of lowland species of toads (Bufo) across the Trans-Mexican Neovolcanic Belt and the Isthmus of Tehuantepec. J Biogeogr 33:1889–1904CrossRefGoogle Scholar
  89. Mutanen M, Pretorius E (2007) Subjective visual evaluation vs. traditional and geometric morphometrics in species delimitation: a comparison of moth genitalia. Syst Entomol 32:371–386CrossRefGoogle Scholar
  90. Niemiller ML, Near TJ, Fitzpatrick BM (2012) Delimiting species using multilocus data: diagnosing cryptic diversity in the southern cavefish Typhlichthys subterraneus (Teleostei: Amblyopsidae). Evolution 66:846–866PubMedCrossRefGoogle Scholar
  91. Niemiller ML, Graening GO, Fenolio DB, Godwin JC, Cooley JR, Pearson WD, Fitzpatrick BM, Near TJ (2013) Doomed before they are described? The need for conservation assessments of cryptic species complexes using an amblyopsid cavefish (Amblyopsidae: Typhlichthys) as a case study. Biodivers Conserv 22:1799–1820CrossRefGoogle Scholar
  92. Olson DM, Dinerstein E, Wikramanayake ED, Burgess ND, Powell GVN, Underwood EC, D’amico JA, Itoua I, Strand HE, Morrison JC, Loucks CJ, Allnutt TF, Ricketts TH, Kura Y, Lamoreux JF, Wettengel WW, Hedao P, Kassem KR (2001) Terrestrial ecoregions of the world: a new map of life on Earth. Bioscience 51:933–938CrossRefGoogle Scholar
  93. Ornelas JF, Sosa V, Soltis DE, Daza JM, González C, Soltis PS, Gutiérrez-Rodríguez C, de los Monteros AE, Castoe TA, Bell C, Ruiz-Sanchez E (2013) Comparative phylogeographic analyses illustrate the complex evolutionary history of threatened cloud forests of northern Mesoamerica. PLoS ONE. doi: 10.1371/journal.pone.0056283 Google Scholar
  94. Pérez S, Vázquez-Domínguez E (2015) Mitochondrial diversification of the Peromyscus mexicanus species group in Nuclear Central America: biogeographic and taxonomic implications. J Zool Systematics Evol Res 53:300–311CrossRefGoogle Scholar
  95. Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using multilocus genotype data. Genetics 155:949–959Google Scholar
  96. Rambaut A, Suchard MA, Xie D and Drummond AJ (2014) Tracer v1.6, Available from http://beast.bio.ed.ac.uk/Tracer
  97. Reber-Funk CR, Schmid-Hempel R, Schmid-Hempel P (2006) Microsatellite loci for Bombus spp. Mol Ecol Resour 6:83–86CrossRefGoogle Scholar
  98. Reid NM, Carstens BC (2012) Phylogenetic estimation error can decrease the accuracy of species delimitation: a Bayesian implementation of the general mixed Yule-coalescent model. BMC Evol Biol 12:196PubMedPubMedCentralCrossRefGoogle Scholar
  99. Rohlf FJ (2015) tpsDIG2: a program for digitizing landmarks and outlines from image files, scanner or video. Department of Ecology and Evolution, State University of New York at Stony Brook. Available at https://life.bio.sunysb.edu/morph/
  100. Rohlf FJ (2015) tpsUtil: a tps file utility program. Department of Ecology and Evolution, State University of New York at Stony Brook. Available at https://life.bio.sunysb.edu/morph/
  101. Ronquist F, Teslenko M, van der Mark P, Ayres DL, Darling A, Höhna S, Larget B, Liu L, Suchard MA, Huelsenbeck JP (2012) MrBayes 3.2: efficient Bayesian phylogenetic inference and model choice across a large model space. Syst Biol 61:539–542PubMedPubMedCentralCrossRefGoogle Scholar
  102. Rousset F (2008) GENEPOP’007: a complete re-implementation of the GENEPOP software for Windows and Linux. Molecular Ecology Resources 8:103–106PubMedCrossRefGoogle Scholar
  103. Roy MS, da Silva JMC, Arctander P, García-Moreno J, Fjeldså J (1997) The speciation of South American and African birds in montane regions. In: Mindell D (ed) Avian molecular evolution and systematics. Academic Press, San Diego, pp 325–343CrossRefGoogle Scholar
  104. Ruiz EA, Rinehart JE, Hayes JL, Zuñiga G (2010) Historical demography and phylogeography of a specialist bark beetle, Dendroctonus pseudotsugae Hopkins (Curculionidae: Scolytinae). Environ Entomol 39:1685–1697PubMedCrossRefGoogle Scholar
  105. Sachman-Ruiz B, Narváez-Padilla V, Reynaud E (2015) Commercial Bombus impatiens as reservoirs of emerging infectious diseases in central México. Biol Invasions 17:2043–2053CrossRefGoogle Scholar
  106. Sánchez-Sánchez H, López-Barrera G, Peñaloza-Ramírez JM, Rocha-Ramírez V, Oyama K (2012) Phylogeography reveals routes of colonization of the bark beetle Dendroctonus approximatus Dietz in Mexico. J Hered 103:638–650PubMedCrossRefGoogle Scholar
  107. Santos Júnior JE, Santos FR, Silveira FA (2015) Hitting an unintended target: phylogeography of Bombus brasiliensis Lepeletier, 1836 and the first new Brazilian bumblebee species in a century (Hymenoptera: Apidae). PLoS ONE 10:e0125847PubMedPubMedCentralCrossRefGoogle Scholar
  108. Schlick-Steiner BC, Steiner FM, Seifer B, Stauffer C, Christian E, Crozier RH (2010) Integrative taxonomy: a multisource approach to exploring biodiversity. Ann Rev Entomol 55:421–438CrossRefGoogle Scholar
  109. Schmid-Hempel P, Schmid-Hempel R, Brunner PC, Seeman OD, Allen GR (2007) Invasion success of the bumblebee, Bombus terrestris, despite a drastic genetic bottleneck. Heredity 99:414–422PubMedCrossRefGoogle Scholar
  110. Shao ZY, Mao HX, Fu WJ, Ono M, Wang DS, Bonizzoni M, Zhang YP (2004) Genetic structure of Asian populations of Bombus ignitus (Hymenoptera: Apidae). J Hered 95:46–52PubMedCrossRefGoogle Scholar
  111. Sheffield CS, Richardson L, Cannings S, Ngo H, Heron J, Williams PH (2016) Biogeography and designatable units of Bombus occidentalis Greene and B. terricola Kirby (Hymenoptera: Apidae) with implications for conservation status assessments. J Insect Conserv 20:189–199CrossRefGoogle Scholar
  112. Stolle E, Rohde M, Vautrin D, Solignac M, Schmid-Hempel P, Schmid-Hempel R, Moritz RFA (2009) Novel microsatellite DNA loci for Bombus terrestris. Mol Ecol Resour 9:1345–1352PubMedCrossRefGoogle Scholar
  113. Strecker U, Faúndez V, Wilkens H (2004) Phylogeography of surface and cave Astyanax (Telostei) from Central and North America based on cytochrome b sequence data. Mol Phylogenet Evol 33:469–481PubMedCrossRefGoogle Scholar
  114. Suárez-Atilano M, Burbrink F, Vázquez-Domínguez E (2014) Phylogeographic structure of Boa constrictor imperator with emphasis of diversification across the lowlands and mountains of Central America and Mexico. J Biogeogr 41:2371–2384CrossRefGoogle Scholar
  115. Sullivan J, Arellano E, Rogers DS (2000) Comparative phylogeography of Mesoamerican highland rodents: concerted versus independent response to past climate fluctuations. American Naturalist 155:755–768PubMedCrossRefGoogle Scholar
  116. Talavera G, Dincã V, Vila R (2013) Factors affecting species delimitations with the GMYC model: insights from a butterfly survey. Methods Ecol Evol 4:1101–1110CrossRefGoogle Scholar
  117. Torres-Ruiz A, Jones RW (2012) Comparison of the efficiency of the bumble bees Bombus impatiens and Bombus ephippiatus (Hymenoptera: Apidae) as pollinators of tomato in greenhouses. J Econ Entomol 105:1871–1877PubMedCrossRefGoogle Scholar
  118. Vrba E (1993) Mammal evolution in the African Neogene and a new look at the Great American Interchange. In: Goldblatt P (ed) Biological relationships between Africa and South America. Yale University Press, New Haven, pp 393–434Google Scholar
  119. Widmer A, Schmid-Hempel P (1999) The population genetic structure of a large temperate pollinator species, Bombus pascuorum (Scopoli) (Hymenoptera: Apidae). Mol Ecol 8:387–398PubMedCrossRefGoogle Scholar
  120. Widmer A, Schmid-Hempel P, Estoup A, Scholl A (1998) Population genetic structure and colonization history of Bombus terrestris s.l. (Hymenoptera: Apidae) from the Canary Islands and Madeira. Heredity 81(5):563–572CrossRefGoogle Scholar
  121. Williams PH (1998) An annotated checklist of bumble bees with an analysis of patterns of description (Hymenoptera: Apidae, Bombini). Bull Nat History Museum Entomol Ser 67:79–152Google Scholar
  122. Williams PH, An J, Huang J (2011) The bumblebees of the subgenus Subterraneobombus: integrating evidence from morphology and DNA barcodes (Hymenoptera, Apidae, Bombus). Zool J Linn Soc 163:813–862CrossRefGoogle Scholar
  123. Williams PH, An J, Brown MJF, Carolan JC, Goulson D, Huang J, Ito M (2012) Cryptic bumblebee species: consequences for conservation and the trade in greenhouse pollinators. PLoS ONE 7:32992CrossRefGoogle Scholar
  124. Williams PH, Byvaltsev AM, Cederberg B, Berezin MV, Ødegaard F, Rasmussen C, Richardson LL, Huang J, Sheffield CS, Williams ST (2015) Genes Suggest Ancestral Colour polymorphisms are shared across morphologically cryptic species in Arctic bumblebees. PLoS ONE 10:e0144544PubMedPubMedCentralCrossRefGoogle Scholar
  125. Yoon HJ, Kim SY, Lee KY, Lee SB, Park IG, Kim I (2009) Interspecific hybridization of the bumblebees Bombus ignitus and B. terrestris. Int J Ind Entomol 18:45–52Google Scholar

Copyright information

© Springer Science+Business Media Dordrecht 2016

Authors and Affiliations

  • Michelle A. Duennes
    • 1
    • 4
  • Chris Petranek
    • 1
    • 5
  • Esteban Pineda Diez de Bonilla
    • 2
    • 6
  • Jorge Mérida-Rivas
    • 2
  • Oscar Martinez-López
    • 2
    • 3
  • Philippe Sagot
    • 2
  • Rémy Vandame
    • 2
  • Sydney A. Cameron
    • 1
  1. 1.Department of Entomology, School of Integrative BiologyUniversity of IllinoisUrbanaUSA
  2. 2.Departamento Agricultura, Sociedad y AmbienteEl Colegio de la Frontera SurSan Cristóbal De Las CasasMexico
  3. 3.Unidad para la Conservación, Uso y Valoración de la Biodiversidad, Centro de Estudios Conservacionistas, Facultad de Ciencias Químicas y FarmaciaUniversidad de San Carlos de GuatemalaCuidad De GuatemalaGuatemala
  4. 4.Department of Entomology, College of Natural and Agricultural SciencesUniversity of California, RiversideRiversideUSA
  5. 5.Department of Zoology & PhysiologyUniversity of WyomingLaramieUSA
  6. 6.Instituto de Ciencias BiológicasUniversidad de Ciencias y Artes de ChiapasTuxtla GutiérrezMexico

Personalised recommendations