ArticlePDF Available

Substantial genetic divergence and lack of recent gene flow support cryptic speciation in a colour polymorphic bumble bee (Bombus bifarius) species complex


Abstract and Figures

Phenotypic polymorphism can constitute an inherent challenge for species delimitation. This issue is exemplified in bumble bees (Bombus), where species can exhibit high colour variation across their range, but otherwise exhibit little morphological variation to distinguish them from close relatives. We examine the species status of one of the most abundant North American bumble bees, Bombus bifarius Cresson, which historically comprised two major taxa, bifarius s.s. and nearcticus. These lineages are recognized primarily by red and black variation in their mid‐abdominal coloration; however, a continuum from black (nearcticus) to red (bifarius s.s.) variation has led to their historic synonymization. Integrating mitochondrial and nuclear data and whole‐genome sequencing, we reveal a high level of both mitochondrial and nuclear divergence delimiting two morphologically cryptic species – the red bifarius s.s. and the colour‐variable (black to red) nearcticus. Population genomic analysis supports an absence of recent genomic admixture and a strong population structure between the two clades, even in sympatry. Species distribution models predict partially differentiated niches between the genetically inferred clades with annual precipitation being a leading differentiating variable. The bifarius s.s. lineage also occupies significantly higher elevations, with regions of sympatry being among the highest elevations in nearcticus. Our data also support a subspecies‐level divergence between the broadly distributed nearcticus and the island population vancouverensis. In this paper, we formally recognize the two species, Bombus bifarius Cresson and Bombus vancouverensis Cresson, the latter including the subspecies B. vancouverensis vancouverensis comb.n. and B. vancouverensis nearcticus comb.n., with vancouverensis the name bearer due to year priority. Mitochondrial and nuclear genomic data are used to assess and redefine the species status of the common widespread western North American bumble bee species Bombus bifarius Cresson. Bombus bifarius s.l. is determined to comprise two species that fail to admix in a hybrid zone: western widespread Bombus vancouverensis, and Bombus bifarius, confined to the Colorado plateau. These species are considered morphologically cryptic aside from somewhat overlapping differences in colour, with historic species confusion created by colour polymorphism related to mimicry.
Distribution and allelic designation of individuals sampled for nuclear genotyping, showing the close association between mitochondrial and nuclear haplotypes. Pie charts represent haplotype combinations at a locality; each one shows whether the specimen's haplotype was designated as being allied to the Bombus nearcticus vs. bifarius s.s. lineage (Fig. 1) using mitochondrial data on the left side. In comparison to mitochondrial data, on the right side of the pie, the four single nucleotide polymorphisms (SNPs) of the ATPase (ATP) and the three SNPs of the Serrate Effector (SE) are designated, using different colours for whether the allelic variant matches the fixed alleles for western nearticus (blue) vs. bifarius s.s. (red) from a previous study (Lozier et al., 2016a). A few individuals had heterozygosity at a single ATPase SNP, indicated in pink. Numbers next to pie charts indicate the number of individuals with that haplotype combination in localities where more than a single individual were sampled. Localities were combined if sites were very close with the exception of the hybrid zone in Utah, where all localities are indicated separately to show allelic patterns of individuals in sympatry. Three pie charts that also were all blue ('nearcticus') from Alaska and Yukon are not shown. An altitude layer is overlaid in the background with higher altitudes shaded darker. Data are further represented in Table S2 and specimen details in Appendix S1. [Colour figure can be viewed at].
Content may be subject to copyright.
Systematic Entomology (2020), DOI: 10.1111/syen.12419
Substantial genetic divergence and lack of recent gene
flow support cryptic speciation in a colour polymorphic
bumble bee (Bombus bifarius) species complex
1Laboratory of Zoology, Research Institute of Biosciences, University of Mons, Mons, Belgium, 2Department of Biological Sciences,
The University of Alabama, Tuscaloosa, AL, USA, 3Department of Biology, The Pennsylvania State University, University Park, PA,
USA, 4Department of Entomology, The Pennsylvania State University, University Park, PA, USA, 5United States Department of
Agriculture-Agricultural Research Services, Pollinating Insects-Biology, Management, and Systematics Research Laboratory, Logan,
UT, USA and 6Department of Entomology, The Ohio State University, Columbus, OH, USA
Abstract. Phenotypic polymorphism can constitute an inherent challenge for species
delimitation. This issue is exemplied in bumble bees (Bombus), where species can
exhibit high colour variation across their range, but otherwise exhibit little morpholog-
ical variation to distinguish them from close relatives. We examine the species status of
one of the most abundant North American bumble bees, Bombus bifarius Cresson, which
historically comprised two major taxa, bifarius s.s. and nearcticus. These lineages are
recognized primarily by red and black variation in their mid-abdominal coloration; how-
ever, a continuum from black (nearcticus)tored(bifarius s.s.) variation has led to their
historic synonymization. Integrating mitochondrial and nuclear data and whole-genome
sequencing, we reveal a high level of both mitochondrial and nuclear divergence delim-
iting two morphologically cryptic species the red bifarius s.s. and the colour-variable
(black to red) nearcticus. Population genomic analysis supports an absence of recent
genomic admixture and a strong population structure between the two clades, even in
sympatry. Species distribution models predict partially differentiated niches between
the genetically inferred clades with annual precipitation being a leading differentiating
variable. The bifarius s.s. lineage also occupies signicantly higher elevations, with
regions of sympatry being among the highest elevations in nearcticus. Our data also
support a subspecies-level divergence between the broadly distributed nearcticus and the
island population vancouverensis. In this paper, we formally recognize the two species,
Bombus bifarius Cresson and Bombus vancouverensis Cresson, the latter including
the subspecies B. vancouverensis vancouverensis comb.n. and B. vancouverensis
nearcticus comb.n., with vancouverensis the name bearer due to year priority.
The processes generating and maintaining genetic and pheno-
typic diversity in polymorphic organisms are of great inter-
est in evolutionary biology. Such polymorphic populations are
Correspondence: Department of Biology, The Pennsylvania State
University, University Park, PA 16801, USA. E-mail:
useful for revealing the complex interplay of drift, admixture
and selection in shaping phenotypic diversity. However, they
can present substantial challenges in species delimitation. For
example, some polymorphic species may lack genetic structure,
whereas others may contain multiple ‘cryptic’ species masked
by a lack of discrete morphological, behavioural and/or ecolog-
ical characters (Bickford et al., 2007; Pfenninger & Schwenk,
2007; Murray et al., 2008). Resolving the species status of such
© 2020 The Royal Entomological Society 1
2G. Ghisbain et al.
lineages is necessary for understanding the origins and evolution
of such variation, and is critical for assigning accurate taxonomy,
conducting biodiversity assessments and aiding their conserva-
tion (May, 1988; Bickford et al., 2007).
Bumble bees (Hymenoptera: Apoidea: genus Bombus
Latreille) are an emerging model system for understanding
patterns of complex species diversity (Woodard et al., 2015).
The 260 Bombus species recognized worldwide (Williams,
1998) occur mostly in temperate areas of the Northern Hemi-
sphere (e.g. Corbet et al., 1991; Memmott et al., 2004; Hegland
& Totland, 2008), where they contribute to ecosystem ser-
vices through pollination of both wild plants and crops. These
vital ecosystem services (Kremen et al., 2002; Velthuis & Van
Doorn, 2006) are under threat, as bumble bee declines have been
recognized worldwide (Williams & Osborne, 2009; Cameron
et al., 2011; Kerr et al., 2015). Ultimately, the conservation of
these important pollinators depends on accurate species-level
Species delimitation has been challenging in bumble bees,
as they have been considered ‘morphologically monotonous’
(Michener, 2000), often exhibiting only minor character dif-
ferences such as details of head puncturing and male genitalia
morphology to diagnose closely related species. By contrast,
coloration of thoracic and abdominal setae can be highly vari-
able, leading to a historical focus on colour traits for species
diagnosis. Colour patterns, however, can be poor traits for diag-
nosis in bumblebees (Carolan et al., 2012; Hines & Williams,
2012; Koch et al., 2018). One of the major reasons for this colour
variation is Müllerian mimicry, which has driven many Bombus
species towards nearly identical colour patterns within a geo-
graphical region, while promoting diverse colour forms within
species across their distributions (Williams, 2007). Such high
variation and convergence, the general unreliability of colour
traits, and the lack of morphological characters on which to base
species decisions has led to many cases of taxonomic misclas-
sication (Ellis et al., 2006; Bertsch, 2009; Koch et al., 2018),
contributing to bumble bees being one of the most highly syn-
onymized lineages (Williams, 1998).
Although denitions of species can vary (Mayr, 1961; De
Queiroz, 2007; Woodard et al., 2015), a practical denition
which we apply here is to consider them as independently evolv-
ing lineages (de Queiroz, 2007) with little to no evidence of
gene ow. In practice, examining multiple lines of evidence (e.g.
Lecocq et al., 2015; Martinet et al., 2019; Williams et al., 2019)
is optimal for recognizing whether character states are discrete
between lineages, and are thus evolving separately. For example,
mitochondrial DNA sequence data can yield more resolved
and discrete histories given a lack of recombination, haploidy
and maternal inheritance; however, mitochondrial genes can
lead to false interpretations because of introgression, selection,
sex-biased histories and historical population isolation. Nuclear
sequence data, although confounded by incomplete lineage sort-
ing, provide an independent signature of population history that
is better at revealing recent and ancient patterns of introgres-
sion. Genome-wide sequencing data are particularly informative
about the extent of admixture by revealing whether any possible
introgression is widespread throughout the genome, restricted
to certain gene regions, or largely absent. Integrated approaches
combining morphology, nuclear and mitochondrial sequences,
and/or chemical data, are common for bumble bee species delin-
eation (e.g. Lecocq et al., 2015; Martinet et al., 2019; Williams
et al., 2019).
In North America, one of the most taxonomically confound-
ing species is Bombus bifarius Cresson (subgenus Pyrobom-
bus Dalla Torre), a widespread and abundant taxon distributed
in western mountain ranges from the American Southwest to
Alaska. The species displays a striking range of colour patterns
across its distribution, resulting in various subspecic epithets
for its phenotypic variants (Stephen, 1957). The main morpho-
types were originally described as separate species during early
descriptions of North American fauna (Cresson, 1878; Han-
dlirsch, 1888) and have since been synonymized as subspecies
given their shared morphologies and continuous gradation in
colour variation (e.g. Stephen, 1957; Williams, 1998). Two pri-
mary subspecies are recognized: Bombus bifarius nearcticus
(henceforth referred to as nearcticus), having metasomal T2–3
ranging from all black to mostly red and occurring in the western
part of the distribution and Bombus bifarius bifarius (henceforth
referred to as bifarius s.s.), with T2–3 largely red and domi-
nating in the eastern part of the species range (Fig. 1; Stephen,
1957; Lozier et al., 2013; Ezray et al., 2019). Aside from dif-
ferences in colour, variation in the terminal male sternite was
reported to potentially discriminate these two taxa (Stephen,
1957). A third taxon, Bombus bifarius vancouverensis (hence-
forth referred to as vancouverensis), occurs on Vancouver Island
and surrounding islands of the Salish Sea (Fig. 1), exhibiting dis-
tinctive features of having whiter setae on the thorax, and largely
red T2+3 segments similar to disjunct eastern bifarius popula-
Previous studies have investigated population dynamics of
the B. bifarius species complex (i.e. including bifarius s.s.,
nearcticus and vancouverensis, henceforth referred to as Bom-
bus bifarius s.l.) using multiple genetic methods. Studies using
microsatellites and colour pattern have suggested the presence
of ongoing gene ow consistent with genetically heterogeneous,
but partially connected conspecic populations (Lozier et al.,
2011, 2013). Subsequent genomic analyses of transcriptome and
double digest restriction enzyme associated DNA sequencing
(ddRADSeq) have cast doubt on existing taxonomy, suggesting
substantial divergence without gene ow between red bifarius
s.s. and black to intermediately coloured nearcticus lineages
(Lozier et al., 2016a). These studies raise the possibility that
B. bifarius s.l. may represent a complex of phenotypically poly-
morphic cryptic species.
In this paper, we rigorously assess the specic status of
B. bifarius s.l. by examining patterns of gene ow and lineage
isolation among all three lineages using a combined mitochon-
drial gene and nuclear genomic approach. We sample the mito-
chondrial barcode (COI) marker from individuals across the
species range to determine patterns of population structure, and
utilize whole-genome sequencing of representative individuals,
along with more abundant taxon sampling of specic informa-
tive nuclear markers, to test for genomic admixture in regions
of allopatry and sympatry. We then use occurrence data and
© 2020 The Royal Entomological Society, Systematic Entomology, doi: 10.1111/syen.12419
Cryptic speciation in Bombus bifarius 3
Fig. 1. Distribution of colour forms in Bombus bifarius s.l. with colour patterns and regions assigned to historically recognized taxa names indicated.
Assignment of intermediate colour forms to species was not well-dened historically. % Red T2/T3 =the percentage of the second and third metasomal
tergites with red-orange setal coloration. Distribution data based on Ezray et al. (2019). [Colour gure can be viewed at].
bioclimatic variables to determine the role of ecological niche
in the inferred species distributions and assess whether dened
genetic lineages can be further demarcated with morphological
features. We nally assess the taxonomic status of the lineages
bifarius s.s., nearcticus,andvancouverensis considering avail-
able type material and their original descriptions.
Material and methods
Phylogenetic inference of mitochondrial COI
Sampling. Specimens were obtained from across the B. bifar-
ius s.l. distribution, including Alaska (AK, USA), Yukon (YT,
Canada), and the western United States (WA, OR, CA, ID, WY,
UT and CO) (N=119; Appendix S1), and were collected within
the last 10 years. Sampling intentionally covered a large part of
the colour forms displayed by the group, from the bright, fully
red-banded Eastern specimens (bifarius s.s.), through diverse
intermediate forms, to the extensively black ones (nearcticus),
and also included three exemplars of vancouverensis from the
San Juan Islands, WA. Most specimens used were directly trans-
ferred into 100% ethanol at 20C after collecting, whereas
others were mounted and dried.
DNA preparation, amplication and sequencing for COI.
Total DNA was extracted from the hind legs of the specimens
using E.Z.N.A. Tissue DNA Kits (Omega Bio-tek, Norcross,
GA, USA). Legs were crushed and digested with protease
K for 3.5 hours at 55C before column extraction, following
the manufacturer’s protocol with a single, nal elution in
50μL of elution buffer. Samples were individually checked for
both purity (NanoDrop Onec, Thermo Fischer Scientic) and
integrity (1% agarose electrophoresis gel) before amplication.
Polymerase chain reaction (PCR) amplications were carried
out (Hot Start Taq 2x Master Mix, NEB) on a commonly used
gene for species delimitation in bumble bees (e.g., Duennes
et al., 2012; Williams et al., 2012) and other insects: the mito-
chondrial barcode fragment of the cytochrome oxidase I (COI).
We amplied 640 bp of COI using universal primers LCO1490
et al., 1994) with the amplication conditions as follows: initial
denaturing at 94C for 2 min, a rst cycle of 30 s at 94C, 40 s
at 45C, 1 min at 72C, followed by 35 cycles of 30 s at 94C,
40 s at 49C, 1 min at 72C, before nal elongation of 10 min at
72C. DNA from all amplied samples was enzymatically puri-
ed using ExoSAP-ITTM (Thermo Fischer Scientic) followed
by Sanger sequencing (Genomic Core Facility, Pennsylvania
State University, PA, USA).
COI phylogenetic inference and haplotype network. We
constructed a COI Bayesian phylogeny of 119 newly sequenced
individuals along with sequences of 48 additional B. bifarius
s.l. individuals from all three lineagesobtained from GenBank
( Sequences were
selected from GenBank to include representative individuals
from each sampled locality and only sequences with nearly
complete COI data in the sequenced interval. To examine
the placement of each lineage relative to the closest extant
relatives, we also included exemplar COI sequences of the
closest relatives (Cameron et al., 2007) to B. bifarius s.l.,
including Bombus ephippiatus,Bombus impatiens,Bombus
huntii,Bombus ternarius and Bombus vosnesenskii, along with
the slightly more distant Bombus melanopygus as an outgroup,
using either newly sequenced individuals or sequences from
GenBank (Appendix S1). Comparative ‘Percent Identity’ values
in GenBank were used to conrm that these exemplars were
typical representatives of the diversity of COI sequences in
their respective species clades. COI sequences were edited
manually and aligned in G v8.1.9 (Biomatters, http://, with ends trimmed to remove missing
© 2020 The Royal Entomological Society, Systematic Entomology, doi: 10.1111/syen.12419
4G. Ghisbain et al.
data to yield a 562-bp aligned fragment. The best available
model for our COI data according to the Bayesian Information
Criterion (BIC) obtained from MEGA X (v10.0.5; Kumar
et al., 2018) is GTR+I. We performed Bayesian inference
under this model using MB 3.2.6 (Ronquist & Huelsen-
beck, 2003) analyzed in the CIPRES Science Gateway v3.3
servers ( with four independent runs,
four chains, 15 million generations and sampling trees every
1000 generations. We assessed convergence through examining
likelihood plots (conrming stationarity) and convergence
statistics in MB 3.2.6 and ESS values in T 1.7.1
(Rambaut et al., 2018), which led us to conservatively discard
the rst 25% of the trees as burn-in for all runs to obtain a
majority-rule 50% consensus tree. This tree was run through
the Poisson-Tree-Process procedure (PTP; Zhang et al., 2013),
which applies coalescent models to designate optimal species
assignments from an input tree, using the bPTP server (https:// and applying default PTP options. We also
constructed a haplotype network for the 119 newly sequenced
individuals using a 584-bp fragment of COI with no missing
data. The network was constructed using the median-joining
method in N (http://www.uxus-engineering
.com/) and haplotypes were mapped geographically. Sequence
alignments are available at Dryad (
.7d7wm37r9), and newly obtained sequences on GenBank
(Accession numbers MN781411-MN781539).
Genomic tests of admixture
Sampling. To assess population structure and admixture, we
performed whole genome sequencing on 21 bee specimens
(Appendix S1) that represent the two major lineages inferred
from our mitochondrial data (‘nearcticus-clade’ and ‘bifar-
ius-clade’), thus enabling comparison between mitochondrial
and genomic signatures. Sampling included ve mostly black
nearcticus-clade haploid males from different locations in the
western United States (CA, OR; nearcticus-West), one diploid
queen representing the historical vancouverensis lineage from
within the nearcticus-clade (San Juan Island, WA), four mostly
red bifarius s.s.-clade individuals from separate locations in CO
and eastern UT (three diploid workers, one haploid male), and
11 mostly red nearcticus-clade males from several geograph-
ically intermediate locations in the central Rockies (UT and
WY; nearcticus-Central). By sampling nearcticus populations
both distant and close to bifarius s.s. populations, we can test
for signatures of admixture along a geographical gradient. The
more abundant sampling of central nearcticus allows more indi-
viduals to be examined for signs of admixture in the possible
contact zone. This sampling included individuals from both lin-
eages (one each) collected together at the same site in the eastern
Uinta Mountains (Utah, USA) thus allowing direct assessment
of whether gene ow is occurring in sympatry.
Genomic sequencing and variant calling. Leg or thoracic
muscle tissue was extracted from the male samples, previously
kept frozen or in 90% ethanol at 20C, using an E.Z.N.A.
DNA extraction kit run with standard protocols. Samples were
prepared for sequencing using an Illumina TruSeq DNA Nano
library construction kit following standard protocols. DNA
from the diploid Colorado samples was extracted with Qiagen
(Venlo, the Netherlands) DNeasy kits (with RNase treatment)
following modications in Lozier (2014). Dual-indexed Illu-
mina libraries for the diploid samples were prepared by Hud-
sonAlpha Institute for Biotechnology Genome Services Lab
(Huntsville, AL USA) for CO bifarius s.s.-clade bees or Pso-
magen, Inc. (Rockville, MD) for the vancouverensis specimen.
Whole-genome sequencing for haploid samples (n=17) was
performed using an Illumina HiSeq 2500 sequencer to generate
2×150-bp paired-end read libraries at Pennsylvania State Uni-
versity Genomics Core Facility. Diploid samples (n=4) were
sequenced (2 ×150-bp paired end reads) on an Illumina HiSeq X
(HudsonAlpha) or a NovaSeq6000 S4 (Psomagen). After initial
quality control assessment of raw sequencing reads performed
with FQC v0.11.7 (Andrews, 2010), appropriate quality
control on the raw reads was conducted using T
v0.38 (Bolger et al., 2014) to remove low-quality bases (SLID-
INGWINDOW:4:30 LEADING:3 TRAILING:3), clip adapters
(ILLUMINACLIP:adapters.fa:2:30:5) and discard reads <36 bp
(MINLEN:36). Post-QC reads were aligned to the published
genome assembly of closely related B. impatiens (NCBI Gen-
Bank Assembly GCA_000188095.3, BIMP_2.1; Sadd et al.,
2015) using BWA v0.7.17, utilizing the BWA-mem algorithm
with default parameters (Li & Durbin, 2009). Post-processing of
aligned reads was conducted in SAM v1.8 (Li et al., 2009)
and P  v1.119 (available at: http://broadinstitute
Multi-sample variant calling for haploid samples (n=17) was
performed using GATK U G v3.6 (McKenna
et al., 2010; DePristo et al., 2011) following best practices
(Van der Auwera et al., 2013) with specic parameters for
haploidy [-ploidy 1 -glm single nucleotide polymorphism (SNP)
-stand_call_conf 25.0]. The variant calling procedure for diploid
(n=4) samples was conducted using the same protocol except
for diploid designation (-ploidy 2). Variant calling datasets from
both procedures were merged using the C V
utility in GATK.
We generated two ltered datasets for different genomic
analyses. One dataset was used for statistical analyses of
individual bee genotypes (e.g. principal components analy-
sis, neighbour-joining analysis, population structure inference),
hereafter referred to as the ‘individual-level’ dataset, that
included bifarius s.s., nearcticus-Central, nearcticus-West and
the single San Juan Island vancouverensis bee (n=21). Filtering
of this dataset included allowing a minimum minor allele count
of one, requiring a maximum of two alleles, minimum depth per
SNP per bee of four, minimum genotype quality 20; we also
excluded a small number of sites (n=50 645) which had any
heterozygous sites in haploid samples (erroneous by denition)
and any sites with missing data. Finally, we restricted our analy-
ses to the 78 scaffolds with length >1 Mb (total of 154 Mb, 63%
of total B. impatiens genome length). The nal individual-level
dataset had 322 605 SNPs.
© 2020 The Royal Entomological Society, Systematic Entomology, doi: 10.1111/syen.12419
Cryptic speciation in Bombus bifarius 5
The second dataset was used for population-level analysis
(‘population-level’ dataset, used to analyze genetic diversity,
population differentiation and admixture). This differed from the
individual-level dataset in excluding the single vancouverensis
specimen (n=20). Filters for SNPs included requiring a maxi-
mum of two alleles, minimum depth per SNP per bee of four,
minimum genotype quality 20, minimum root mean squared
mapping quality 40, missing data rate of maximum 20%, and
a minimum minor allele count of two, resulting in 426 358
total SNPs.
Individual-level analysis of genomic relationships. To infer
the relationship among all individuals (n=21), a principal
component analysis (PCA) was conducted on the ltered SNP
dataset (322 605 SNPs) using the PCA option from the Anal-
ysis menu in TASSEL 5 (Bradbury et al., 2007). The rst two
principal components were plotted using the  (Wick-
ham, 2016) package in R. A neighbour-joining distance-based
tree was constructed from the ltered variant calling data using
TASSEL 5 (Bradbury et al., 2007). To infer population structure,
sNMF analysis was implemented in R/ (Frichot & François,
2015). We removed mixed ploidy by phasing all diploid (n=4)
individuals using B v4.1 (Browning & Browning, 2007,
parameters impute =false window =10000 overlap =1000
gprobs =false), effectively treating them as two haploid individ-
uals each. We ran ve independent runs at each K-value (K=1
to 10) and obtained the optimal number of ancestral popula-
tion assignments by assessing minimum cross-entropy across all
Population-level analyses of diversity, population structure
and admixture. We used custom bioinformatics scripts by
Simon Martin implementing methods from (Martin et al., 2015;
Van Belleghem et al., 2017;
genomics_general) to perform genome-wide population
genetic analyses on 20 bees and 426 358 SNPs represent-
ing population-level samples (bifarius s.s., nearcticus-West,
nearcticus-Central). These methods have the advantage of
allowing a mixture of haploids and diploids in the sample
set. Allele frequencies were estimated at each site using
B. impatiens as a reference. We calculated nucleotide diversity
(𝜋) within nearcticus-West, nearcticus-Central and bifarius
s.s., and the xation index FST and absolute divergence (dXY)
between group pairs across 100-kb sliding windows across the
genome containing at least 100 SNPs, slid by 25 kb per step.
Mean and condence intervals for genome-wide 𝜋,FST and dXY
were determined by jack-kning over the 100-kb blocks using
Analyses of introgression were conducted using the
ABBA-BABA framework (Green et al., 2010) with scripts
by Simon Martin as above. Most analyses employed a
three-population model {[(P1,P2),P3],O} where nearcti-
cus-West is specied as P1,bifarius s.s. as P3and central
nearcticus-Central as the hypothesized admixed lineage (P2)
(3-Pop model; see Figure S1 for schematic of models examined
in this study). Under a history of pure divergence and drift,
alleles at polymorphic sites should only be shared by these
populations due to incomplete lineage sorting, which should
result in an equal number of SNPs with the derived allele (‘B’,
vs. ancestral allele ‘A’) shared by bifarius s.s. (P3) and either
nearcticus population (P1or P2) and thus ABBA sites equal
BABA sites. By contrast, with post-divergence gene ow, it is
expected that P3and P2(the neighbouring Utah nearcticus and
bifarius s.s. populations) would share an excess of nonreference
alleles, resulting in an excess of ABBA over BABA sites.
Several statistics have been developed to quantify this excess,
including Patterson’s D(Green et al., 2010), which is most
suitable for examining introgression at the whole-genome level,
and the admixture proportion fD(Martin et al., 2015), which
is more suitable for examining the proportion of admixture at
narrower genome windows to identify potentially introgressed
chromosomal regions. We report Dand fdM (Malinsky et al.,
2015), a modication of fDthat is bounded between [-1,1] and
symmetrically distributed around zero under a null hypothesis
of no gene ow, although all ABBA– BABA statistics were
highly correlated. In all cases, positive values represent potential
introgression between P3and P2(ABBA >BABA), whereas
values of zero are consistent with a model of divergence and
drift alone (ABBA =BABA). As above, genome-wide esti-
mates were calculated in nonoverlapping 100-kb blocks, as well
as averaged for each scaffold, and signicance (95% condence
intervals) of whole-genome estimates was tested using block
jack-kning. In a second analysis (Hybrid-zone model) we
used the single nearcticus-Central bee with a red phenotype
(conrmed by COI sequence) collected from within the range
of bifarius s.s. in the Uinta Mountains of Utah, and tested
for hybridization by specifying this sample as P2, with other
nearcticus-Central bees as P1and bifarius s.s. as P3.
We also examined ancestry and potential for recent admix-
ture of populations within the putative hybrid population
(nearcticus-Central) using an ancestry painting approach (Der
Sarkissian et al., 2015; Runemark et al., 2018; scripts from were used to conduct this anal-
ysis). First, we extracted SNPs with alternate alleles xed
between bifarius s.s. and nearcticus-West, using only SNPs
with no missing data. The genotypes were then extracted for
each haploid bee from the nearcticus-Central population at these
SNPs and scored according to parental allele state (allowing
10% missing data, thinned to one SNP per 100bp). This anal-
ysis is facilitated by the haploidy of all nearcticus-Central bees,
which allows visualization of fully phased haplotypes across
each scaffold. If recent hybridization had occurred, this would
be apparent as large blocks of admixed parental sequence.
Supplementary les for bioinformatic analyses are avail-
able on Dryad (, Sequenc-
ing data available for utilized genomic samples are available on
NCBI BioProject PRJNA592825.
Genotyping for nuclear markers
Genomic data sampling includes only a single inferred
nearcticus in sympatry with a single bifarius s.s. To examine for
signs of genomic admixture in regions of sympatry with larger
sample sizes, we examined variation in two nuclear markers
© 2020 The Royal Entomological Society, Systematic Entomology, doi: 10.1111/syen.12419
6G. Ghisbain et al.
across 66 individuals representing the geographic range of the
lineages of interest, including six nearcticus and seven bifar-
ius s.s. specimens from three regions of sympatry (Book Cliffs,
Uinta and Manti-La Sal mountain ranges in Utah), and two indi-
viduals of vancouverensis (Table S1). These markers were cho-
sen as they have multiple synonymous SNPs (i.e. less likely to
be under selective pressure than nonsynonymous ones) that were
xed between disjunct nearcticus-West and bifarius s.s. popu-
lations in a previous study (Lozier et al., 2016a), and are thus
informative for detecting allele sharing between sympatric indi-
viduals with minimal complications from incomplete lineage
sorting. We designed primers for these markers that are com-
plete matches to sequences of both forms by considering the
alignment of SNPs for these populations against their respec-
tive homologues in B. impatiens (NCBI GenBank Assembly
GCA_000188095.3, BIMP_2.1). We amplied 924bp of the
serrate RNA effector (BIMP21906, Drosophila ortholog Ars2),
a gene involved in RNA transport and processing, using the
designed primers SerF-a (5’ CGAAAGCAGATCCTCGTAAGT
3’) and SerR-ab-4 (5’ GGCATTTCGTCTTCGTTTGG 3’), and
610 bp of the sodium/potassium-exchanging ATPase subunit
alpha (BIMP10181, Drosophila Atpalpha), a gene involved
in salt ion homeostasis, using ATPF-b (5’ TGCTGCAAA-
CTTCCCATTAC 3’). Amplication of both loci were per-
formed using the following cycle parameters: a rst cycle
of 30 s at 94C, 40 s at 45C, 1 min at 72C, followed by
35 cycles of 30 s at 94C, 40 s at 55C, 1 min at 72C,
before nal elongation of 10 min at 72C. Previously identi-
ed xed-divergence SNPs in these amplicons were compared
among samples along with their mitochondrial haplotypes to
assess for signs of introgression. Sequences are available on
GenBank (Accession numbers MN781540-MN781579 for the
ATPase; MN788377-MN788416 for the serrate RNA effector).
Morphological analysis
We qualitatively examined barcoded specimens of both sexes
covering a large part of the B. bifarius s.l. range (Alaska to Col-
orado) for morphological differentiation, focusing on cuticular
characters (e.g. shape and puncturing of the clypeus, labrum,
cheek, abdominal segments, ocellar area, corbiculae) that are
generally considered as reliable characters for species-level dis-
crimination in bumble bees (Williams et al., 2010, 2014, 2019).
For males, we examined the genitalia in more detail (shape
and/or pile of the volsella, penis valve, gonostylus, gonocox-
ite and gonobase). As the shape of the medio-apical part of
the eighth ventral plate of the males was argued to be dis-
tinct between nearcticus and bifarius s.s. by Stephen (1957), we
more closely examined and imaged this sternite across speci-
mens. For imaging, the 8th ventral plates were dissected from
specimens relaxed in an ethanol:water dilution series. These
were mounted in glycerol between two cover slips and imaged
with an Olympus BX43 compound microscope with an attached
Olympus DP73 digital camera. Image series were aligned and
stacked using Z S (v1.04 Build T201404082055)
and exposure was corrected in Adobe P CC 2019.
Species distribution modelling
To examine whether the two lineages inferred with genetic
data occupy distinct niche spaces, we estimated habitat suit-
ability by species group using the maximum entropy approach
applied in ME v3.4.1 (Phillips et al., 2004; Phillips
& Dudík, 2008). This program is especially well-suited for
presence-only datasets and can create accurate predictions from
relatively small sample sizes (Phillips & Dudík, 2008). However,
as MaxEnt is susceptible to overtting and spatial autocorrela-
tion, care should be taken to sample occurrence records evenly
across space as well as to select appropriate model parame-
ters. We combined specimen occurrence records conrmed as
nearcticus or bifarius s.s. using COI, with additional locali-
ties of specimens determined to be nearcticus vs. bifarius s.s.
using genetic data by Lozier et al. (2016a, 2016b). Occurrence
records were distributed throughout the range extent and were
limited to unique sites to minimize spatial bias prior to niche
modelling (bifarius s.s., n=22; nearcticus,n=81). Previous
ecological niche analyses on bifarius s.s. determined that eight
climatic variables (annual precipitation, maximum temperature
of the warmest month, mean temperature of the wettest quar-
ter, precipitation of the driest month, annual mean temperature,
precipitation of the wettest month, mean temperature of the dri-
est quarter and minimum temperature of the coldest month)
aptly describe the climatic trends pertinent to montane bum-
ble bee habitat suitability (Lozier et al., 2013), thus we built
our models using these variables. Contemporary climatic (aver-
aged across the years 1970–2000) raster layers were down-
loaded at a resolution of 2.5 arc-minute from WorldClim v1.4
(Hijmans et al., 2005), clipped to our study extent (-150, -103,
30, 65), and converted to an ASCII le in R/ (Hijmans,
2018). Species-specic niche models were created using the
default parameters applied in ME v3.4.1 averaged across
15 cross-validated replicates. Model performance was assessed
using the AUC statistic and variable permutation importance
was determined using jack-knife analysis (Phillips et al., 2004;
Phillips & Dudík, 2008; Phillips et al., 2017). The plot combin-
ing the predicted distribution of both bifarius s.s. and nearcti-
cus was visualized in QGIS (Quantum GIS 2018), whereas the
species-specic plots were visualized in R/ (Hijmans,
2018) and R/ (Wickham, 2016) packages. In addition,
we determined using the Student’s t-test whether parameters dif-
fered between bifarius s.s. and nearcticus.Wealsorant-tests on
altitude for each variant for all specimens but also on Utah speci-
mens alone, to determine whether bifarius s.s. occupies different
altitude around contact zones.
Mitochondrial phylogeny and haplotype network
The Bayesian inference conducted on the mitochondrial COI
gene highlighted a species-level divergence between the taxa
bifarius s.s. and nearcticus with strong support, as each comes
out as a clade of lower sequence divergence (<2.0%, 0.5% on
© 2020 The Royal Entomological Society, Systematic Entomology, doi: 10.1111/syen.12419
Cryptic speciation in Bombus bifarius 7
average) subtended by long branches (Fig. 2A) with a 6.9%
pairwise COI sequence divergence between them. bifarius s.s. +
nearcticus were resolved as sister lineages with B. ternarius as
the closest relative. Bombus ternarius has less COI-divergence
with bifarius s.s. (3.0% divergence) than bifarius s.s. shares
with nearcticus (6.9% divergence). No evidence for population
structure was found within bifarius s.s., whereas some regional
taxon clustering could be observed within nearcticus, although
global population structure in the latter remained weak rela-
tive to species-level differences. The taxon vancouverensis was
contained as a derived lineage within nearcticus. The three
newly sampled vancouverensis specimens were a sister lineage
to three British Columbia barcoded specimens from the same
island chain, which are also likely vancouverensis, and these
are derived from a clade that spans the Pacic coastal states and
provinces. The PTP partition with the most support recognized
seven candidate species (outgroup B. melanopygus not included
in the analysis) as most likely (see the grey values accompanying
the posterior probabilities; Fig. 2A), including all currently rec-
ognized species and separate species assignments for nearcticus
and bifarius s.s., but not for vancouverensis.
The haplotype network analysis (Fig. 2B) shows that the two
haplotypes for the bifarius lineage only occur in the Southern
Rocky Mountains (eastern Utah and Colorado), and the 29 hap-
lotypes for the nearcticus lineage occur west of this region, with
haplotypes for the lineages only overlapping in a narrow con-
tact zone (Fig. 2C). Although most satellite haplotypes diverged
from the core haplotypes by a single base pair only, individu-
als from more distant sampling zones (e.g. Alaska) showed a
stronger divergence from other relatives. Although weak popula-
tion structure was highlighted by mapping nearcticus haplotypes
(e.g. northern California/southern Oregon), specimens coming
from close areas also can possess diverse mitochondrial haplo-
types (e.g. northern Oregon, Utah).
Genomic analysis
Individual-level analysis of genomic relationships. The
PCA on the ltered individual-level SNP dataset of genomic
samples exhibit clustering by population nearcticus-Wes t,
nearcticus-Central, vancouverensis and bifarius s.s. in the rst
two principal components, with bifarius s.s. distinguished from
nearcticus +vancouverensis in the rst principal component
(15.13% of variation) and geographical separation of nearcticus
+vancouverensis samples on the second (5.71% of varia-
tion) (Fig. 3B). This result is consistent with the phenogram
(Fig. 3C), and with sNMF analysis results (Fig. 3D), where
K=2 is the optimal number of population clusters (Figure S1)
and bifarius s.s. is designated as a separate group from a com-
bined nearcticus-West, nearcticus-Central and vancouverensis
population (Fig. 3D). The nearcticus and bifarius s.s. samples
from the same location in the Uinta mountains remained dis-
tinct to their respective populations across all analyses (Fig. 3,
marked with a ‘U’). The vancouverensis specimen consistently
was placed in the nearcticus clade across analyses, although
with a slightly longer branch length in the phenogram and in
between western and eastern nearcticus clades along axis 2 in
the PCA.
Population-level divergence and admixture across
the genome. Differentiation and divergence across the genome
were signicantly higher between bifarius s.s. and both nearcti-
cus-West (mean FST =0.14, dXY =0.46) and nearcticus-Central
(mean FST =0.13, dXY =0.46) than for any intra-nearcticus
window (mean FST =0.01, dXY =0.32) (Fig. 4; Table S1).
The differences between bifarius s.s. and either nearcticus
lineage across the genome were highly correlated, consis-
tent with strong bifarius s.s.–nearcticus differentiation and
weak intra-nearcticus differentiation. Divergence (dXY) for the
nearcticus population pair can be almost perfectly explained
by mean 𝜋(simple regression slope =0.98, R2=0.96), as
expected if samples actually come from a single population and
divergence simply reects diversity. By contrast, dXY in bifarius
s.s.nearcticus pairs greatly exceeded their mean 𝜋(Fig. 4).
Alongside the overall signature of strong divergence between
bifarius s.s. and nearcticus, there was some heterogeneity
across the genome, particularly noticeable in blocks of sharply
elevated FST and dXY between bifarius s.s. and nearcticus
(Fig. 4). Three blocks within scaffold JH157950 showed some
notable intra-nearcticus elevation (Fig. 4, S2A). This region is
also unusual in that dXY exceeded mean 𝜋within nearcticus
across much of the scaffold, especially in low diversity regions
(Fig. 4). Notably, this scaffold corresponds to NT_176739.1
from the BIMP 2.0 (Sadd et al., 2015) RefSeq reference assem-
bly, which was implicated in a previous transcriptome study
as a potential target of selection in nearcticus (Pimsler et al.,
Consistent with the strong interlineage population struc-
ture, genome-wide evidence for widespread introgression
between bifarius s.s. and nearcticus via geographically inter-
mediate populations (UT+WY) was very low for the three
population ABBABABA analysis. Condence intervals from
block-jack-kning did not overlap zero (D=0.005, 95% CI:
0.002–0.008; z-score =3.1) so some weak historical admixture
cannot be fully ruled out (Fig. S1; Table S1); however, average
Dacross whole scaffolds ranged from – 0.045 to 0.033, and
none were signicantly different from zero. The low fdM value
of 0.3% (0.1–0.5% 95% CI) likewise indicated little evidence
for gene ow between lineages (Figure S1A). There is some
evidence for more restricted introgression at certain portions of
the genome that might explain the subtle excess of ABBA over
BABA sites (Fig. 4). For example, one window in JH157950
(275-375kb), discussed above, reaches D=0.43 with an fdM of
15.5%, the largest in the dataset, indicating an excess of shared
polymorphism between bifarius s.s. and nearcticus-Central
in this region (Figure S2A). However, this window contains
Xanthine dehydrogenase/oxidase-like, a selection target pre-
viously identied from RNA-Seq SNPs (Pimsler et al., 2017).
Like the full three-population comparison, the model involving
specication of a single nearcticus bee collected from a site of
geographic overlap with bifarius s.s. in the Uinta mountains
in UT also showed Dand fdM values surrounding zero (Figure
© 2020 The Royal Entomological Society, Systematic Entomology, doi: 10.1111/syen.12419
8G. Ghisbain et al.
Fig. 2. Relationships of Bombus bifarius s.l. specimens inferred with cytochrome oxidase I (COI). (A) Bayesian tree of B. bifarius s.l. and close
relatives based on the barcode fragment of the COI. Clade support values are Bayesian posterior probabilities (black) and PTP values in support of all
daughter alleles belonging to a single species (grey). The right zoomed-in part of the gure highlights some population structure within the specimens
of vancouverensis. The map shows the sampling locations of the specimens of B. bifarius s.l. used in the phylogenetic analysis with the original
sampling of the study displayed in purple and the previously published sequences in black. (B) Haplotype network for the newly sequenced individuals,
highlighting the connections between the haplotypes shown on the map (right). To avoid colour overloading, very closely related haplotypes (i.e. within
1 bp from a main core haplotype) share the same colour. (C) Visualization of the major haplotypes from the same sampling, mapped at their respective
geographical locations. The sizes of the circles and pie charts are proportional to the number of haplotypes that they contain. [Colour gure can be
viewed at].
© 2020 The Royal Entomological Society, Systematic Entomology, doi: 10.1111/syen.12419
Cryptic speciation in Bombus bifarius 9
Fig. 3. Population structure inferred from whole-genome sequence data: (A) A map of sampled specimens, coloured to match respective clades in
(B)– (D). (B) Principal component analysis (PCA) of the whole-genome single nucleotide polymorphism (SNP) data, showing the separation of bifarius
s.s. from the nearcticus and vancouverensis populations. (C) Neighbour-joining unrooted phylogeny of SNP data. (D) sNMF analysis results for K=2,
with colours reecting different population assignment. Carets (^) above the plots indicate differently phased samples from the same diploid individual.
‘U’s represent two specimens obtained from the same locality and collecting event in the Uinta Mountains for (B)–(D). [Colour gure can be viewed
S2B). Overall, such weak evidence for introgression should be
considered with caution, with most of the genome suggesting a
lack of ongoing gene ow between bifarius s.s. and nearcticus
even in sites of sympatry.
Ancestry painting reveals that of the 5160 xed divergent
SNPs between the more distant bifarius s.s. and nearcticus-West
lineages, 9–10% of each nearcticus-Central genome repre-
sented the bifarius s.s. allele (Figure S2C). However, these
regions were haphazardly scattered across the genome, and there
were no blocks of bifarius s.s. ancestry, so it is unlikely that
any of these genomes are produced by recent introgression
from bifarius s.s. Much of the shared variation can thus likely
be explained by incomplete lineage sorting, as suggested by
ABBA– BABA analysis. We repeated the ancestry analysis by
specifying ve haploid nearcticus-Central populations (to hold
sample size equivalent to nearcticus-West) and examined SNPs
with xed alternative states between bifarius s.s. and nearcti-
cus-Central, using nearcticus-West samples instead as the puta-
tive hybrids. Under a pure divergence history with incomplete
lineage sorting, the same fraction of SNPs should show ancestry
with bifarius s.s. in this test dataset as in the previous dataset.
However, we saw a slightly smaller proportion of SNPs with
bifarius s.s. ancestry in nearcticus-West (Wilcoxon rank sum
test: P<0.001) compared to nearcticus-Central, which is con-
sistent with the small but signicant D>0.
Nuclear genotyping
Clustering based on the genotyping across seven SNPs of
both nuclear markers matched the clustering based on the
mitochondrial COI for all individuals, including all specimens
coming from sympatric areas (Fig. 5; Table S2). A single SNP
among four for the ATPase was heterozygous for a few of the
B. bifarius s.s. Although this could be considered as a sign of
gene ow, given the alleles of linked SNPs, it is likely indicative
of a lack of complete historic lineage sorting at this particular
SNP. The vancouverensis bees do not differ from the other
nearcticus at either nuclear locus.
Morphological analyses
Using qualitative approaches, no obvious external morpholog-
ical character could be found to reliably discriminate between
© 2020 The Royal Entomological Society, Systematic Entomology, doi: 10.1111/syen.12419
10 G. Ghisbain et al.
bifarius vs. nearcticus-west bifarius vs. nearcticus-central nearcticus-west vs. central
00.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Scaffold JH157950
y = 0.98x - 0.002
R = 0.96
Average π
nearcticus-west vs. central
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
Average π
bifarius vs. nearcticus-west
bifarius vs. nearcticus-central
Fig. 4. Genome-wide estimates of FST, absolute divergence dXY, and the introgression statistics fdM (ABBA –BABA topology: P1 =nearcticus-West;
P2 =nearcticus-Central; P3 =bifarius s.s.) for 100-kb sliding windows (25kb slide/step) with at least 100 single nucleotide polymorphisms (SNPs) in
the three Bombus bifarius s.l. lineage comparisons. Data are shown for 78 B. impatiens scaffolds with length >1Mb,where x-axis tick markers each
represent 500 kb. In the FST and dXY plots, both bifarius vs. nearcticus plots overlap nearly perfectly so they are largely indistinguishable. Scaffold
JH157950 is demarcated with a dotted rectangle. The bottom plots show the relationship between absolute divergence dXY between population pairs
and mean nucleotide diversity (𝜋) for the pair being considered (left =within-nearcticus divergence, right =bifarius s.s. vs. nearcticus divergence).
Within nearcticus interpopulation divergence is essentially equivalent to intrapopulation diversity for each region of the genome (except for regions in
scaffold JH157950 highlighted in red), whereas divergence greatly exceeds mean diversity for bifarius s.s. vs. nearcticus comparisons. The dashed line
shows equal diversity-divergence in each panel. [Colour gure can be viewed at].
© 2020 The Royal Entomological Society, Systematic Entomology, doi: 10.1111/syen.12419
Cryptic speciation in Bombus bifarius 11
Fig. 5. Distribution and allelic designation of individuals sampled for nuclear genotyping, showing the close association between mitochondrial and
nuclear haplotypes. Pie charts represent haplotype combinations at a locality; each one shows whether the specimen’s haplotype was designated as
being allied to the Bombus nearcticus vs. bifarius s.s. lineage (Fig. 1) using mitochondrial data on the left side. In comparison to mitochondrial data,
on the right side of the pie, the four single nucleotide polymorphisms (SNPs) of the ATPase (ATP) and the three SNPs of the Serrate Effector (SE)
are designated, using different colours for whether the allelic variant matches the xed alleles for western nearticus (blue) vs. bifarius s.s. (red) from
a previous study (Lozier et al., 2016a). A few individuals had heterozygosity at a single ATPase SNP, indicated in pink. Numbers next to pie charts
indicate the number of individuals with that haplotype combination in localities where more than a single individual were sampled. Localities were
combined if sites were very close with the exception of the hybrid zone in Utah, where all localities are indicated separately to show allelic patterns of
individuals in sympatry. Three pie charts that also were all blue (‘nearcticus’) from Alaska and Yukon are not shown. An altitude layer is overlaid in
the background with higher altitudes shaded darker. Data are further represented in Table S2 and specimen details in Appendix S1. [Colour gure can
be viewed at].
the females of nearcticus and bifarius s.s. For males, we found
that the shape of the medio-apical ridge of the eighth ventral
plate, previously reported as distinct between lineages, showed
high variability and thus is not a reliable interspecic discrimi-
nant character (Figure S3). Thus, these two lineages can only be
discriminated outside of DNA sequences based on coloration,
in part, combined with information on geographical range (see
Species distribution prediction
Species distribution models predict that bifarius s.s.
(AUC =0.981, SD =0.014) and nearcticus (AUC =0.884,
SD =0.057) occupy somewhat different niche spaces, with
bifarius s.s. occupying a narrower spatial projection of the
environmental niche. Despite limited realized overlap in
geographical ranges, predicted niches overlap in distribu-
tion throughout portions of the Rocky Mountains (Fig. 6). For
example, the niche model predicts that bifarius s.s. is distributed
at mainly high elevation locales throughout the eastern portions
of the Rocky Mountain chain, whereas available specimen
data suggest that it is largely conned to the Southern Rocky
Mountains. When comparing values for the niche variables
between each form, only four variables differed signicantly
(Table S3). The taxon nearcticus tends to occupy wetter areas
on average (P=0.021), but the range of precipitation fully
spans that of bifarius s.s. (Figure S4). The nearcticus clade
also varies in temperature in the driest quarter (P=0.0022) and
mean temperatures of the wettest quarter (P=0.041), likely
because the wet season is different between the Pacic west
and the Rocky Mountains. Furthermore, bifarius s.s. occupies
higher elevations than nearcticus (P<0.001; Figure S3). In
Utah, where zones of sympatry occur, bifarius s.s. also differs
in altitude (P =0.033) occurring in the highest elevation zones
of nearcticus.
Species-level differentiation
We utilized an integrative approach gathering mitochon-
drial COI barcode data, targeted nuclear gene sequencing and
© 2020 The Royal Entomological Society, Systematic Entomology, doi: 10.1111/syen.12419
12 G. Ghisbain et al.
Fig. 6. ME predictions based on unique occurrence data for the lineages Bombus bifarius s.s. (n=22) and nearcticus (n=81), using only
specimens identied based on nuclear and mitochondrial sequence data. Left: nearcticus and bifarius s.s. niche models. Triangles represent localities
from Lozier et al. (2013, 2016a), and circles represent individuals identied herein with barcode data. Darker colours are more suitable niches. Right:
Combined model for both lineages nearcticus (black) and bifarius s.s. (red). [Colour gure can be viewed at].
whole-genome sequencing to achieve a robust understanding
of patterns of divergence and gene ow in B. bifarius s.l. Our
high spatial resolution analysis of COI across the geographi-
cal range of this species complex revealed strongly supported
species-level differentiation that roughly corresponds to the his-
torically recognized lineages bifarius s.s. and nearcticus.These
two lineages exhibit high interlineage divergence (6.9%) and
low intralineage divergence, a pattern of phylogenetic differen-
tiation that is typically indicative of species-level differentiation
(Hebert et al., 2003). The divergence between bifarius s.s. and
nearcticus is concordant for COI and a set of diagnostic nuclear
SNPs, further supporting a lack of ongoing gene ow between
the two lineages. Finally, we nd consistent evidence for the two
major lineages from whole-genome analyses of differentiation
and divergence. Individuals belonging to these two respective
lineages are largely separated geographically, with nearcticus in
the western part of the range and bifarius s.s. conned to the
Southern Rocky Mountains surrounding the Colorado Plateau.
This analysis revealed three fairly narrow regions of sympatry
in the Uinta, Book Cliffs and Manti-La Sal mountain chains of
Utah (Fig. 5; Table S2). Despite the opportunity for allele shar-
ing in these regions, no recent gene ow was inferred from the
genomes of individuals in sympatry or from targeted nuclear
markers with increased taxon sampling at these zones.
Analyses utilizing whole-genome sequencing data suggest
some possibility of weak introgression between bifarius s.s.
and nearcticus via geographically proximate nearcticus-Central
populations (Figure S1). However, we see very little evidence
for large blocks of shared ancestry in genome scans (Fig. 4)
or ancestry paining (Figure S1); given the perfect phasing of
haploid nearcticus samples, introgressed haplotypes should be
readily apparent if recent ongoing gene ow was contributing
to shared variation. Thus, it seems more likely that any admix-
ture is infrequent or has occurred well into the past. Further-
more, inferring gene ow from ABBA–BABA statistics relies
on assumptions of the underlying demographic model, and sim-
ilar values can be produced under both introgression and ances-
tral population structure (Martin et al., 2015). Finally, we sus-
pect that some of the positive Dand fdM signatures may result
from selection. One scaffold, JH157950 (GenBank assembly,
BIMP 2.1), showed notable intra-nearcticus differentiation that
exceeded other regions of the genome, and was the only scaffold
where FST within nearcticus exceeded the lowest observed FST
between bifarius and nearcticus. The sharp discrepancy between
FST and dXY in this interval (Fig. 4; Figure S1) potentially sug-
gests the action of a selective sweep in the ancestral population
(Cruickshank & Hahn, 2014).
Altogether, our different types of genetic data, combining
high-resolution spatial sampling at small numbers of mitochon-
drial and nuclear markers with high-resolution genomic sam-
pling of spatially representative populations, reject the presence
of common recent gene ow between the two taxa, hereby lead-
ing us to designate these lineages as separate species.
Population-level distinctions
Within-lineage divergence is low, although the more
widespread nearcticus demonstrates some clustering of eastern
(intermediate or mostly red coloured) individuals from west-
ern (mostly black forms) using genomic data. Lozier et al.
(2016a) observed similar population structure among eastern
© 2020 The Royal Entomological Society, Systematic Entomology, doi: 10.1111/syen.12419
Cryptic speciation in Bombus bifarius 13
and western populations from transcriptome and reduced
representation (ddRAD) sequencing. The COI data showed
minor geographical structure in nearcticus in more peripheral
populations, but haplotype variants as a whole tend to be
broadly admixed across much of the western United States. In
the whole-genome data, FST was near zero and dXY was not
substantially different from 𝜋for the nearcticus populations, as
expected if samples are drawn from a single population. These
data are in line with previous genetic data which suggest that
landscape heterogeneity likely restricts gene ow leading to
genetic differentiation following an isolation-by-distance model
shaped by the arrangement of suitable habitat (Lozier et al.,
2013; Jackson et al., 2018).
The lineage vancouverensis, once recognized as a separate
species, is here supported in all sequencing results as a derived
clade within nearcticus that is closely related to other speci-
mens from the Pacic coastal states and provinces. The branch
length of this clade is a bit higher in genomic analyses and there
is some COI divergence (0.6%) from sister clades of nearcti-
cus, an expected pattern given that it is an island population
with restrictions on gene ow. In the light of these ndings,
and given their distinct colour pattern from neighbouring main-
land populations, restricted distribution and previous population
genetic results indicating high levels of differentiation for the
San Juan Island populations (Lozier et al., 2011), we consider
the latter specimens as belonging to a subspecies that is dis-
tinct from the one occurring on the continent, but not sufciently
distinct to warrant species-level separation from the remaining
Crypticism and mimicry
We did not nd any reliable cuticular characters to diagnose
individuals belonging to either species. Stephen (1957) pre-
viously considered the terminal male sternite to be different
between the two, but we found this character to be highly vari-
able and also not diagnostic. In the absence of clearly diagnostic
external characters, and until more detailed quantitative anal-
yses of internal and external characters are performed, these
can be considered a case of cryptic speciation. The most reli-
able way found to diagnose them aside from genetic sequenc-
ing, is through assessing their coloration and locality. However,
colour also can fail as a diagnostic character in regions of sym-
patry where nearcticus individuals exhibit colour most similar
to bifarius s.s. (Figure S6). This lack of discriminant cuticular
characters is not surprising within the subgenus Pyrobombus,
for which many identication keys rely on pile colour to allow
species-level identication (Stephen, 1957; Thorp et al., 1983;
Koch et al., 2012; Williams et al., 2014).
The presence of numerous intermediate colour forms in
the B. bifarius s.l. species complex between the westernmost
Pacic region (black-banded) and the southeastern specimens
(red-banded) previously led to the conclusion that continu-
ous gene ow was occurring across this lineage, thus leading
to synonymization of previously described species (Stephen,
1957; Williams et al., 2014). In this case, colour traits have led
taxonomists astray. Coloration has often led to false inferences
of species delimitation in bumble bees. Some species histori-
cally described by discrete colour differences have been revealed
to be conspecic, as discrete morphologies can be generated
from diallelic Mendelian traits (Owen & Plowright, 1980; Ras-
mont et al., 2005; De Meulemeester et al., 2011). Failure to
fully sort ancestral colour polymorphisms by species also has
led to false inferences of species boundaries (Hines, 2008; Car-
olan et al., 2012; Koch et al., 2018; Williams et al., 2019). In
this instance, bifarius s.s. is a monomorphic red-banded bee
species and nearcticus is a polymorphic species that follows
a continuum of variation from black to red from west to east
(excepting island populations). At the root of this confusion are
issues with using colour traits inuenced by Müllerian mimicry
(Hines & Williams, 2012). North American mimicry complexes
involve darker colour forms in the Pacic coastal regions and
red colour forms in the Rocky Mountains. Higher delity pat-
terns can be found in the far west and east, respectively, of
these ranges, with lower delity patterns in between (Ezray
et al., 2019). Thus, bifarius s.s. is seemingly converging on the
abundant high-delity pattern in the southern Rocky Mountains,
whereas nearcticus is converging on the most abundant patterns
in its respective regions, showing delity matching the colour
continuum that occurs across its range.
Convergence upon the high-delity Rocky Mountain pat-
tern in the East could alone explain colour similarities in
areas of sympatry. However, it also is possible that simi-
larities in coloration between eastern nearcticus and bifar-
ius s.s. could result from historical patterns of allele sorting
and/or gene ow. The examination of admixture across the
genome revealed a single genomic interval with high differ-
entiation between the western black and eastern mostly-red
nearcticus, a region that also has a relatively high percent-
age of ABBA– BABA allele sharing between bifarius s.s. and
mostly red ‘central’ nearcticus populations. This interval falls
in the same contig as the one inferred to potentially harbour
a gene for coloration in this species inferred from transcrip-
tome sequencing (Xanthine dehydrogenase/oxidase-like; Pim-
sler et al., 2017). Alleles for red coloration may have under-
gone adaptive introgression from bifarius s.s. to nearcticus
to generate its red mimetic phenotype, similar to what was
observed in Heliconius mimicry (Pardo-Diaz et al., 2012), or
could result from sorting of colour variation from a polymor-
phic ancestor. Alternatively, black-associated alleles may be
derived from ancestral red-associated alleles within nearcticus,
with polymorphism in nearcticus maintained through selection,
thus driving low effective population sizes and corresponding
genetic diversity in this interval. Haplotypes of the Xanthine
dehydrogenase/oxidase-like gene show unusual relationships
compared to the genome as a whole that is more consistent
with ancestral polymorphism than recent introgression (Pimsler
et al., 2017), supporting the selective sweep hypothesis. Indeed,
the entire JH157950 contig shows somewhat unusual patterns
in nearcticus, suggesting that there could be some interest-
ing genomic architecture relating to phenotypic variation in the
region. By resolving species status and taxonomy in this study,
we now open the door for a more informed investigation into
© 2020 The Royal Entomological Society, Systematic Entomology, doi: 10.1111/syen.12419
14 G. Ghisbain et al.
evolutionary forces that may drive polymorphism and mimicry
in this complex and a closer examination of the genome regions
that may be driving speciation.
Niche occupation
The narrow region of sympatry observed in these two species
suggests that some factor must maintain these lineages sepa-
rate from one another and thus be important for their specia-
tion. For our two taxa of interest, species distribution modelling
previously has been found to effectively explain genetic dif-
ferentiation at the population level (Lozier et al., 2013; Jack-
son et al., 2018). Species distribution modelling performed on
bifarius s.s. and nearcticus separately, as dened by barcode
sequences, inferred that bifarius s.s. occurs in drier areas and at
higher elevations than nearcticus. Although altitude could ulti-
mately be a primary factor differentiating them, this variable
may be inuenced by its more southern geographical location
(e.g. Williams et al., 2016) and the effects of drier climate on
altitudinal distribution. With the variables tested, these species
distribution models also predict that the species should over-
lap. Models thus raise the question regarding factors that pre-
vent more substantial co-occurrence of these species. Given
that occasional darker forms have been recorded in Colorado
and some unusually red forms have occurred further west in
museum specimen data (Fig. 1), this may suggest that the region
of overlap may be broader than identied with our sampling. Our
improved taxonomy, together with more extensive geographic
sampling, will be of value for resolving discrepancies between
the realized and predicted ranges of these species and identi-
fying the factors that have contributed to their spatially limited
coexistence and potentially to the maintenance of their genetic
Revised species status
Given the strong congruence of our results, also supported
by previous analyses at the RNA (RNA-Seq), DNA (ddRAD-
Seq) and phenotypical levels (Lozier et al., 2016a), we resur-
rect both lineages bifarius s.s. and nearcticus to species status.
Bombus bifarius Cresson was originally described from a type
series including males and females as being a bee with red in
T2 and T3 except for a black anteromedial v-shaped notch in
T2; however, individuals from regions where we only found
darker forms of nearcticus, such as in the region around British
Columbia, are included in the description. We have observed
the lectotype of B. bifarius Cresson at The Academy of Natu-
ral Science of Drexel University (Philadelphia, U.S.A.), a queen
that was labelled and formally designated as the lectotype by
Cresson (Cresson, 1916) after the original description (Cres-
son, 1878). This lectotype (Figure S5A,B), carrying the labels
(i) white, printed ‘Col.’ and (ii) red, printed ‘LECTOTYPE’ /
handwritten ‘2628’ corresponds to the typical red-banded indi-
viduals of the bifarius s.s. from Colorado sequenced here. There-
fore, this gives the eastern lineage the name B. bifarius Cresson.
In the same paper, Cresson (1878) also described the species
B. vancouverensis Cresson from a type series including several
males and a female available in the same collection (Figure
S5C,D). Cresson (1916) had designated a male from Vancouver
Island, British Columbia as the lectotype, carrying the labels:
(1) white, printed ‘Van.’ and (2) red, printed ‘LECTOTYPE’ /
handwritten ‘2644’. Although the colour has faded on many of
these specimens, including the lectotype, the morphology and
colour pattern match that corresponding to the lineage herein
referred as vancouverensis fromtheSanJuanIslands,witha
paler yellow on the thorax evident in some of the specimens, and
mostlyredonT2+T3. Finally, we requested from the Naturhis-
torisches Museum Wien (Vienna, Austria) individuals from the
nearcticus type-series. Only one individual, identied by Han-
dlirsch himself, phenotypically corresponding to his original
description and agreeing with the collection data referred to by
Handlirsch (1888), was located (Figure S5E,F). To reduce uncer-
tainty in the identity and application of the name, we assigned
a lectotype status to this female individual carrying the labels:
(i) white, printed ‘Brit. Col’; (ii) white, handwritten ‘nearcti-
cus’ / printed ‘det. Handlirsch’; (iii) white, printed ‘Bombus’
/ handwritten ‘bifarius / var. nearcticus’ / printed ‘det. Babiy’;
(iv) white, printed ‘Pyrobombus / b. nearc- / ticus (Handl.) / Det.
Milliron 1962 [female]; (v) white, printed ‘NHM’; (vii) white,
printed ‘NHMW’; (vii) red, handwritten ‘Rasmont det. 2019’ /
printed ‘LECTOTYPE’ [female] / handwritten ‘Bombus nearcti-
cus / Handlirsch’; (viii) printed ‘det. P. Rasmont 2019 / Bombus
(Pyrobombus) / vancouverensis nearcticus / Handlirsch’. The
colour pattern of the specimen (with the T2 –3 fully covered with
black pile) also distinctly corresponds to most western speci-
mens herein referred to as nearcticus. As we nd the lineage
vancouverensis conspecic with nearcticus based on sequencing
data, the name B. vancouverensis Cresson consequently holds
precedent by 10 years over nearcticus Handlirsch. Therefore,
we recognize two distinct species within the B. bifarius s.l.:
B. bifarius Cresson and B. vancouverensis Cresson. In addi-
tion, because of the previously mentioned divergence of the
island populations near Vancouver with the continental vancou-
verensis, we consider two subspecies within B. vancouverensis:
B. vancouverensis vancouverensis comb.n. (occurring on Van-
couver Island and the surrounding islands of the Salish Sea,
phenotypically similar to the types described by Cresson in
1878) and B. vancouverensis nearcticus comb.n. (occurring in
the western part of the continent, phenotypically similar to the
specimens described by Handlirsch in 1888).
These species are best diagnosed given their barcode
sequences (Appendix S1; e.g. B. bifarius NCBI MN781432,
B. vancouverensis nearcticus NCBI MN781514). They also can
be diagnosed to some degree by colour (Fig. 1) and geographical
range (Fig. 6). Bombus bifarius in all cases examined was fully
red in females in T2/T3 aside from an anteromedial V-shaped
black region in T2 for some females and is typically fully red
in males. In B. vancouverensis nearcticus comb.n. females
typically have all black in T2 and T3 in western Oregon and
Washington and throughout California. Some red occurs on seg-
ments T2 and T3 east of the Pacic coastal mountains starting
in the westernmost Rocky Mountains and increasing in amount
© 2020 The Royal Entomological Society, Systematic Entomology, doi: 10.1111/syen.12419
Cryptic speciation in Bombus bifarius 15
eastward (Fig. 1), but the amount of black is nearly always more
than B. bifarius. Females in regions of sympatry tend to exhibit
<80% red in T2+T3 in B. vancouverensis nearcticus comb.n.,
exhibiting a more pronounced black ‘V’ extending from the
anteromedial border of T2, as opposed to >80% red and a
smaller ‘V’ in B. bifarius (Figure S6; cf. Fig. 1 top vs. middle
right bee). Males in the region of sympatry are often fully red
for both forms but can exhibit some black in B. vancouverensis
nearcticus comb.n. Finally, B. vancouverensis vancouverensis
comb.n. females present a distinctly paler thoracic coloration
(cream-white, by contrast with the yellowish coloration of bifar-
ius and vancouverensis nearcticus comb.n.) with a majority red
coloration in T2+T3, involving black in the anterior portion
of T2 and extending posteriorly into T3 in a V-shaped notch.
The distinction between cream and yellow colours is reduced
in males, which tend to be mostly red in T2+T3. Males and
to a lesser extent, females, of B. bifarius,B. vancouverensis
nearcticus,andB. vancouverensis vancouverensis can have a
yellow arch in place of the V-shaped notch in T2. The amount of
red in B. vancouverensis vancouverensis is considerably greater
than neighbouring mainland population of B. vancouverensis
nearcticus comb.n., which are almost exclusively black.
Species concepts and the conservation of cryptic diversity
This study highlights how multi-trait approaches including
powerful genomic tools and ecological modelling constitute a
robust framework to uncover and explain patterns of popula-
tion structure and inconspicuous speciation in cryptic organ-
isms. Lineages such as B. bifarius/B. vancouverensis that lack
gene ow even in sympatry, and thus adhere to most species
concepts, may not yet have evolved sufcient xed morpho-
logical variation. Criteria that disallow morphologically cryptic
lineages may thus be prone to false negatives.
Taxonomic incompleteness constitutes a fundamental obsta-
cle to invertebrate conservation worldwide, limiting understand-
ing of species distribution and ecology and the ability to effec-
tively conserve taxa (Cardoso et al., 2011). Underestimations
of cryptic biodiversity can now be unveiled by the use of
integrative molecular taxonomic approaches (e.g. Fujita et al.,
2012; Fontaneto et al., 2015) such as those used herein. Bum-
ble bees are highly valuable pollinators in natural and agricul-
tural ecosystems (Kremen et al., 2002; Velthuis & Van Doorn,
2006), as well as an emerging model system for evolutionary
genetics (Tian et al., 2019), and thus have a robust history of
taxonomic and phylogenetic revision. However, as this study
reveals, there is still a need to improve our understanding of
species delimitation in these bees, even in some of the most com-
mon species. The formerly recognized Bombus bifarius s.l. was
regarded as one the most common and widespread species in
North America. This study highlights a need to reassess the rel-
ative abundance and species status of the lineage (B. bifarius
vs. B. vancouverensis) as newly circumscribed. Eco-climatic
perturbations of conned niches including the high mountains
involved in these species, are likely to have a substantial impact
on bumble bee diversity in the future decades (Kerr et al., 2015;
Rasmont et al., 2015; Jackson et al., 2018) which makes it ever
more imperative to adequately recognize these species.
Supporting Information
Additional supporting information may be found online in
the Supporting Information section at the end of the article.
Appendix S1. Specimen information including locality,
haplotype from COI data, identication from nuclear and
mitochondrial (mt) data, and accession number.
Tabl e S1. Population genetics statistics for three population
whole-genome sequencing data.
Tabl e S2. Nucleotides obtained from ATPase and Serrate
Effector genes compared to mitochondrial haplotypes.
Tabl e S3. Contribution of the bioclimatic variables to the
distribution models of bifarius s.s. and nearcticus.
Figure S1. Optimal numbers of populations for the sNMF
analysis inferred from minimal cross-entropy.
Figure S2. Tests of admixture between sampled genomic
populations in Figure 4: (A) A detailed depiction of scaf-
fold JH157950 (highlighted with box in above plots). (B)
Two introgression scenarios tested with the ABBA-BABA
approach (Dand fDM statistics). (C) Ancestry painting anal-
ysis of xed divergent SNPs between two specied ‘parental’
lineages (bifarius s.s. =black and nearcticus-West =grey)
and the putative ‘hybrid’ (nearcticus-Central).
Figure S3. Morphological variation in the medio-apical
ridge of the 8th ventral plate from randomly selected indi-
Figure S4. The lineages bifarius s.s. and nearcticus yielded
signicant differences in the leading explanatory variable,
annual precipitation (left). Altitude also signicantly differs
both overall and in Utah, where zones of sympatry are found
Figure S5. Type specimens examined: (A,B) Lectotype of
Bombus bifarius Cresson from The Academy of Natural Sci-
ence of Drexel University (Philadelphia, U.S.A.) (Photos L.
Tian). (C,D) Lectotype of Bombus vancouverensis Cresson
from The Academy of Natural Science of Drexel University
(Philadelphia, U.S.A.), revised here as Bombus vancouveren-
sis vancouverensis comb.n. (Photos L. Tian). (E,F) Lecto-
type of Bombus nearcticus Handlirsch from the Naturhis-
torisches Museum Wien (Vienna, Austria), revised here as
Bombus vancouverensis nearcticus comb.n. (Photos P. Ras-
Figure S6. The percent red (as opposed to black) coloration
in metasomal tergites 2 and 3 combined across all speci-
mens sampled from the three regions of sympatry (Manti-La
© 2020 The Royal Entomological Society, Systematic Entomology, doi: 10.1111/syen.12419
16 G. Ghisbain et al.
Sal site, Uinta Mountain sites, and the Western Book
Cliff site).
We would like to thank Baptiste Martinet for providing spec-
imens from Alaska and Canada, Seth Davis for providing
specimens from Colorado, Elyse McCormick for technical
support during laboratory work, and Andy Deans for the use
of microscope facilities. We also thank Paul H. Williams
and an anonymous reviewer for their constructive comments
on the manuscript. Funds for this research were provided by
National Science Foundation grants DEB-1457645 to JDL, NSF
CAREER DEB-1453473 to HMH, and F.R.S.-FNRS (Fonds de
la recherche scientique) in support of GG. The authors declare
that they have no conicts of interest.
Andrews, S. (2010) FastQC: a quality control tool for high throughput
sequence data [WWW document]. URL http://www.bioinformatics [accessed in 2016]
Bertsch, A. (2009) Barcoding cryptic bumblebee taxa: B. lucorum,B.
crytarum and B. magnus, a case study. Beitraege zur Entomologie,
59, 287– 310.
Bickford, D., Lohman, D.J., Sodhi, N.S. et al. (2007) Cryptic species
as a window on diversity and conservation. Trends in Ecology and
Evolution,22, 148– 155.
Bolger, A.M., Lohse, M. & Usadel, B. (2014) Trimmomatic: a exible
trimmer for Illumina sequence data. Bioinformatics,30, 2114– 2120.
Bradbury, P.J., Zhang, Z., Kroon, D.E. et al. (2007) TASSEL: software
for association mapping of complex traits in diverse samples. Bioin-
formatics,23, 2633– 2635.
Browning, S.R. & Browning, B.L. (2007) Rapid and accurate haplotype
phasing and missing-data inference for whole-genome association
studies by use of localized haplotype clustering. The American
Journal of Human Genetics,81, 1084– 1097.
Cameron, S.A., Hines, H.M. & Williams, P.H. (2007) A comprehensive
phylogeny of the bumble bees (Bombus). Biological Journal of the
Linnean Society,91, 161– 188.
Cameron, S.A., Lozier, J.D., Strange, J.P. et al. (2011) Patterns of
widespread decline in North American bumble bees. Proceedings of
the National Academy of Sciences USA,108, 662– 667.
Cardoso, P., Erwin, T.L., Borges, P.A.V. & New, T.R. (2011) The seven
impediments in invertebrate conservation and how to overcome them.
Biological Conservation,144, 2647– 2655.
Carolan, J.C., Murray, T.E., Fitzpatrick, Ú. et al. (2012) Colour patterns
do not diagnose species: quantitative evaluation of a DNA barcoded
cryptic bumble bee complex. PLoS ONE,7, e29251.
Corbet, S.A., Williams, I.H. & Osborne, J.L. (1991) Bees and the
pollination of crops and owers in the European Community. Bee
Wor l d ,72, 47–59.
Cresson, E.T. (1878) Descriptions of new species of North American
bees. Proceedings of the Academy of Natural Sciences of Philadel-
phia,30, 181– 221.
Cresson, E.T. (1916) The Cresson types of Hymenoptera. Memoirs of
the American Entomological Society 1. American Entomological
Society, Lanham, MD
Cruickshank, T.E. & Hahn, M.W. (2014) Reanalysis suggests that
genomic islands of speciation are due to reduced diversity,not reduced
gene ow. Molecular Ecology,23, 3133– 3157.
De Meulemeester, T., Aytekin, A.M., Cameron, S. & Rasmont, P.
(2011) Nest architecture and species status of the bumble bee Bombus
(Mendacibombus)shaposhnikovi (Hymenoptera: Apidae: Bombini).
Apidologie,42, 301– 306.
De Queiroz, K. (2007) Species concepts and species delimitation.
Systematic Biology,56, 879– 886.
DePristo, M.A., Banks, E., Poplin, R. et al. (2011) A framework
for variation discovery and genotyping using next-generation DNA
sequencing data. Nature Genetics,43, 491– 498.
Der Sarkissian, C., Ermini, L., Schubert, M. et al. (2015) Evolutionary
genomics and conservation of the endangered Przewalski’s horse.
Current Biology,25, 2577–2583.
Duennes, M.A., Lozier, J.D., Hines, H.M. & Cameron, S.A. (2012) Geo-
graphical patterns of genetic divergence in the widespread Mesoamer-
ican bumble bee Bombus ephippiatus (Hymenoptera: Apidae). Molec-
ular Phylogenetics and Evolution,64, 219– 231.
Ellis, J.S., Knight, M.E., Carvell, C. & Goulson, D. (2006) Cryptic
species identication: a simple diagnostic tool for discriminating
between two problematic bumblebee species. Molecular Ecology
Notes,6, 540– 542.
Ezray, B.D., Wham, D.C., Hill, C. & Hines, H.M. (2019) Unsupervised
machine learning reveals mimicry complexes in bumble bees occur
along a perceptual continuum. Proceedings of the Royal Society of
London B,286, 20191501.
Folmer, O., Black, M., Hoeh, W., Lutz, R. & Vrijenhoek, R. (1994) DNA
primers for amplication of mitochondrial cytochrome c oxidase
subunit I from diverse metazoan invertebrates. Molecular Marine
Biology and Biotechnology,3, 294– 299.
Fontaneto, D., Flot, J.F. & Tang, C.Q. (2015) Guidelines for DNA
taxonomy, with a focus on the meiofauna. Marine Biodiversity,45,
433– 451.
Frichot, E. & François, O. (2015) LEA: an R package for landscape and
ecological association studies. Methods in Ecology and Evolution,6,
925– 929.
Fujita, M.K., Leaché, A.D., Burbrink, F.T., McGuire, J.A. & Moritz,
C. (2012) Coalescent-based species delimitation in an integrative
taxonomy. Trends in Ecology and Evolution,27, 480–488.
Green, R.E., Krause, J., Briggs, A.W. et al. (2010) A draft sequence of
the Neandertal genome. Science,328, 710– 722.
Handlirsch, A. (1888) Die Hummelsammlung des k. k.
naturhistor-ischen Hofmuseums. Annalen des Naturhistorischen
Museums in Wien,3, 209–250.
Hebert, P.D.N., Ratnasingham, S. & deWaard, J.R. (2003) Barcoding
animal life: cytochrome c oxidase subunit 1 divergences among
closely related species. Proceeding of the Royal Society B,270(suppl),
S96– S99.
Hegland, S.J. & Totland, O. (2008) Is the magnitude of pollen limitation
in a plant community affected by pollinator visitation and plant
species specialization levels? Oikos,117, 883–891.
Hijmans, 2018. raster: Geographic Data Analysis and Modeling. R
package version 2.7-15.
Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G. & Jarvis, A. (2005)
Very high resolution interpolated climate surfaces for global land
areas. International Journal of Climatology: A Journal of the Royal
Meteorological Society,25, 1965–1978.
Hines, H.M. (2008) Historical biogeography, divergence times, and
diversication patterns of bumble bees (Hymenoptera: Apidae: Bom-
bus). Systematic Biology,57, 58–75.
Hines, H.M. & Williams, P.H. (2012) Mimetic colour pat-
tern evolution in the highly polymorphic Bombus trifasciatus
(Hymenoptera:Apidae) species complex and its comimics. Zoological
Journal of the Linnaean Society,166, 805– 826.
© 2020 The Royal Entomological Society, Systematic Entomology, doi: 10.1111/syen.12419
Cryptic speciation in Bombus bifarius 17
Jackson, J.M., Pimsler, M.L., Oyen, K.J. et al. (2018) Distance, eleva-
tion and environment as drivers of diversity and divergence in bumble
bees across latitude and altitude. Molecular Ecology,27, 2926– 2942.
Kerr, J.T., Pindar, A., Galpern, P. et al. (2015) Climate change
impacts on bumblebees converge across continents. Science,349,
177– 180.
Koch, J., Strange, J.P. & Williams, P. (2012) Bumble Bees of the Western
United States. A product of the USDA Forest Service Pollinator
Partnership, Washington, D.C. and the Pollinator Partnership, San
Francisco, California .owers/pollinators/
Koch, J.B., Rodriguez, J., Pitts, J.P. & Strange, J.P. (2018) Phylogeny and
population genetic analyses reveals cryptic speciation in the Bombus
fervidus species complex (Hymenoptera: Apidae). PLoS ONE,13,
Kremen, C., Williams, N.M. & Thorp, R.W. (2002) Crop pollination
from native bees at risk from agricultural intensication. Proceedings
of the National Academy of Sciences, USA,99, 16812–16816.
Kumar, S., Stecher, G., Li, M., Knyaz, C. & Tamura, K. (2018)
MEGA X: molecular evolutionary genetics analysis across computing
platforms. Molecular Biology & Evolution,35, 1547– 1549.
Lecocq, T., Brasero, N., Martinet, B., Valterová, I. & Rasmont, P. (2015)
Highly polytypic taxon complex: interspecic and intraspecic inte-
grative taxonomic assessment of the widespread pollinator Bombus
pascuorum Scopoli 1763 (Hymenoptera: Apidae). Systematic Ento-
mology,40, 881– 888.
Li, H. & Durbin, R. (2009) Fast and accurate short read alignment with
Burrows-Wheeler Transform. Bioinformatics,25, 1754–1760.
Li, H., Handsaker, B., Wysoker, A. et al. (2009) The sequence align-
ment/map format and SAMtools. Bioinformatics,25, 2078–2079.
Lozier, J.D. (2014) Revisiting comparisons of genetic diversity in
stable and declining species: assessing genome-wide polymorphism
in North American bumble bees using RAD sequencing. Molecular
Ecology,23, 788– 801.
Lozier, J.D., Jackson, J.M., Dillon, M.E. & Strange, J.P. (2016a)
Population genomics of divergence among extreme and intermediate
color forms in a polymorphic insect. Ecology and Evolution,6,
1075– 1091.
Lozier, J.D., Jackson, J.M., Dillon, M.E. & Strange, J.P. (2016b)
Data from: population genomics of divergence among extreme and
intermediate colour forms in a polymorphic insect. Dryad Digital
Lozier, J.D., Strange, J.P. & Koch, J.B. (2013) Landscape heterogene-
ity predicts gene ow in a widespread polymorphic bumble bee,
Bombus bifarius (Hymenoptera: Apidae). Conservation Genetics,14,
1099– 1110.
Lozier, J.D., Strange, J.P., Stewart, I.J. & Cameron, S.A. (2011) Patterns
of range-wide genetic variation in six North American bumble bee
(Apidae: Bombus) species. Molecular Ecology,20, 4870– 4888.
Malinsky, M., Challis, R.J., Tyers,A.M. et al. (2015) Genomic islands of
speciation separate cichlid ecomorphs in an East African crater lake.
Science,350, 1493– 1498.
Martin, S.H., Davey, J.W. & Jiggins, C.D. (2015) Evaluating the use
of ABBA– BABA statistics to locate introgressed loci. Molecular
Biology and Evolution,32, 244– 257.
Martinet, B., Lecocq, T., Brasero, N. et al. (2019) Integrative taxonomy
of an arctic bumblebee species complex highlights a new cryptic
species (Apidae: Bombus). Zoological Journal of the Linnean Society,
187, 599– 621.
May, R.M. (1988) How many species are there on earth? Science,241,
1441– 1449.
Mayr, E. (1961) Cause and effect in biology. Science,134, 1501– 1506.
McKenna, A., Hanna, M., Banks, E. et al. (2010) The Genome Analysis
Toolkit: a MapReduce framework for analyzing next-generation DNA
sequencing data. Genome Research,20, 1297–1303.
Memmott, J., Waser, N.M. & Price, M.V. (2004) Tolerance of pollination
networks to species extinctions. Proceeding of the Royal Society B,
271, 2605– 2611.
Michener, C.D. (2000) The Bees of the World. Johns Hopkins Press,
Baltimore, MD.
Murray, T.E., Fitzpatrick, Ú., Brown, M.J.F. & Paxton, R.J. (2008)
Cryptic species diversity in a widespread bumble bee complex
revealed using mitochondrial DNA RFLPs. Conservation Genetics,
9, 653– 666.
Owen, R.E. & Plowright, R.C. (1980) Abdominal pile color dimorphism
in the bumble bee, Bombus melanopygus.Journal of Heredity,71,
241– 247.
Pardo-Diaz, C., Salazar, C., Baxter, S.W. et al. (2012) Adaptive intro-
gression across species boundaries in Heliconius butteries. PLoS
Genetics,8, e1002752.
Pfenninger, M. & Schwenk, K. (2007) Cryptic animal species are
homogeneously distributed among taxa and biogeographical regions.
BMC Evolutionary Biology,7, 121.
Phillips, S.J. & Dudík, M. (2008) Modeling of species distributions with
Maxent: new extensions and a comprehensive evaluation. Ecography,
31, 161– 175.
Phillips, S.J., Dudík, M. & Schapire, R.E. (2004) A maximum entropy
approach to species distribution modeling. Proceedings of the 21st
International Conference on Machine Learning, pp. 655–662. ACM
Press, New York, NY.
Phillips, S.J., Dudík, M. & Schapire, R.E. (2017) Maxent software for
modeling species niches and distributions v. 3.4.1 [WWW document].
URL http://biodiversityinformatics.amnh. org/open_source/maxent
[accessed in 2017].
Pimsler, M.L., Jackson, J.M. & Lozier, J.D. (2017) Population genomics
reveals a candidate gene involved in bumble bee pigmentation.
Ecology and Evolution,7, 3406– 3413.
Rambaut, A., Drummond, A.J., Xie, D., Baele, G. & Suchard, M.A.
(2018) Posterior summarisation in Bayesian phylogenetics using
Tracer 1.7. Systematic Biology,67, 901–904.
Rasmont, P., Franzén, M., Lecocq, T. et al. (2015) Climatic risk and
distribution Atlas of european bumblebees. BioRisk,10, 1– 246.
Rasmont, P., Terzo, M., Aytekin, A.M. et al. (2005) Cephalic secre-
tions of the bumble bee subgenus Sibiricobombus Vogt suggest
Bombus niveatus Kriechbaumer and Bombus vorticosus Gerstaecker
are conspecic (Hymenoptera, Apidae, Bombus). Apidologie,36,
571– 584.
Ronquist, F. & Huelsenbeck, J.P. (2003) MrBayes 3: Bayesian phyloge-
netic inference under mixed models. Bioinformatics,19, 1572– 1574.
Runemark, A., Trier, C.N., Eroukhmanoff, F. et al. (2018) Variation and
constraints in hybrid genome formation. Nature Ecology & Evolution,
2, 549– 556.
Sadd, B.M., Barribeau, S.M., Bloch, G. et al. (2015) The genomes of two
key bumblebee species with primitive eusocial organization. Genome
Biology,16, 76.
Stephen, W.P. (1957) Bumble Bees of Western America (Hymenoptera:
Apoidea). Technical Bulletin 40. Agricultural Experiment Station,
Oregon State College, Corvallis, OR.
Thorp, R.W., Horning, D.S. & Dunning, L.L. (1983) Bumble bees and
cuckoo bumble bees of California (Hymenoptera:Apidae). Bulletin of
the California Insect Survey,23, 1– 79.
Tian, L., Rahman, S.R., Ezray, B.D., Franzini, L., Strange, J.P.,
Lhomme, P. & Hines, H.M. (2019) A homeotic shift late in devel-
opment drives mimetic color variation in a bumble bee. Proceedings
of the National Academy of Sciences,116, 11857– 11865.
© 2020 The Royal Entomological Society, Systematic Entomology, doi: 10.1111/syen.12419
18 G. Ghisbain et al.
Van Belleghem, S.M., Rastas, P., Papanicolaou, A. et al. (2017) Com-
plex modular architecture around a simple toolkit of wing pattern
genes. Nature Ecology & Evolution,1, 0052.
Van der Auwera, G.A., Carneiro, M.O., Hartl, C. et al. (2013) From
FastQ data to high-condence variant calls: the genome analysis
toolkit best practices pipeline. Current Protocols in Bioinformatics,
43, 11.10.1– 11.10.33.
Velthuis, H.H.W. & Van Doorn, A. (2006) A century of advances
in bumble bee domestication and the economic and environmental
aspects of its commercialization for pollination. Apidologie,37,
421– 451.
Wickham, H. (2016) ggplot2: Elegant Graphics for Data Analysis.
Springer, New York, NY.
Williams, P.H. (1998) An annotated checklist of bumble bees with
an analysis of patterns of description (Hymenoptera: Apidae,
Bombini). Bulletin of the Natural History Museum of London,67,
79– 152.
Williams, P.H. (2007) The distribution of bumble bee colour patterns
worldwide: possible signicance for thermoregulation, crypsis and
warning mimicry. Biological Journal of the Linnaen Society,92,
97– 118.
Williams, P.H., Berezin, M.V., Cannings, S.G. et al. (2019). The arctic
and alpine bumblebees of the subgenus Alpinobombus revised from
integrative assessment of species’ gene coalescents and morphology
(Hymenoptera, Apidae, Bombus). Zootaxa,4625, 1– 68.
Williams, P.H., Brown, M.J.F., Carolan, J.C. et al. (2012) Unveiling
cryptic species of the bumble bee subgenus Bombus s. str.worldwide
with COI barcodes (Hymenoptera: Apidae). Systematics and Biodi-
versity,10, 21– 56.
Williams, P.H., Huang, J., Rasmont, P. & An, J. (2016) Early-diverging
bumblebees from across the roof of the world: the high-mountain
subgenus Mendacibombus revised from species’ gene coalescents and
morphology (Hymenoptera, Apidae). Zootaxa,4204, 1–72.
Williams, P.H., Ito, M., Matsumura, T. & Kudo, I. (2010) The bumble-
bees of the Nepal Himalaya (Hymenoptera: Apidae). Insecta Mat-
sumurana,66, 115– 151.
Williams, P.H. & Osborne, J.L. (2009) Bumble bee vulnerability and
conservation world-wide. Apidologie,40, 367–387.
Williams, P.H., Thorp, R.W., Richardson, L.L. & Colla, S.R. (2014)
Bumble Bees of North America: An Identication Guide. Princeton
University Press, Princeton, NJ.
Woodard, S.H., Lozier, J.D., Goulson, D. et al. (2015) Molecular tools
and bumble bees: revealing hidden details of ecology and evolution in
a model system. Molecular Ecology,24, 2916– 2936.
Zhang, J.J., Kapli, P., Pavlidis, P. & Stamatakis, A. (2013) A general
species delimitation method with applications to phylogenetic place-
ments. Bioinformatics,29, 2869– 2876.
Accepted 6 December 2019
© 2020 The Royal Entomological Society, Systematic Entomology, doi: 10.1111/syen.12419
... However, despite the long history of classical taxonomic work (i.e. not involving molecular tools) on bumblebees compared to other bees (Williams, 1998), their effective conservation has remained challenging due to particularly low levels of interspecific morphological differentiation (Michener, 2007;Williams et al., 2012Williams et al., , 2020 associated with highly variable intraspecific colour patterns (Williams, 2007;Hines & Williams, 2012;Ezray et al., 2019;Tian et al., 2019;Williams et al., 2019;Ghisbain et al., 2020a), making species level identification difficult. Delineation based on regular taxonomic tools (i.e. ...
... cryptic taxa) within widespread species . In this context, the use of genetic and semiochemical characters to delineate species in integrative frameworks has led to profound changes in the accepted taxonomy of species and their associated distributions Ghisbain et al., 2020a;Lhomme et al., 2021). In Europe, for instance, the consequences of such reassessments are crucial, with an increasing need to revise and update the assessments and conclusions presented in the last European Red List of Bees (Nieto et al., 2014). ...
... This gene has been shown to accurately predict bumblebee species delineation in many large-scale studies (e.g. Williams et al., 2012Williams et al., , 2019Williams et al., , 2020 and has been recently shown as a useful proxy of gene flow in a widespread cryptic bumblebee species complex (Ghisbain et al., 2020a). It is, however, widely acknowledged that the use of COI must be always combined with other traits given that its high substitution rate can excessively separate taxa that are conspecific but with a strong population structuring (e.g. ...
Full-text available
• Against the context of global wildlife declines, targeted mitigation strategies have become critical to preserve what remains of biodiversity. However, the effective development of conservation tools in order to counteract these changes relies on unambiguous taxonomic determination and delineation. • In this study, we focus on an endemic bumblebee species recorded only from the highest altitudes of the Sierra Nevada (Spain), Bombus reinigiellus (Rasmont, 1983). The species has the smallest range of any European bumblebee, along with a restricted diet and an inability to disperse because of its isolated montane distribution, making it an appropriate conservation target. However, through an integrative taxonomic approach including genetics, morphometrics and semio-chemistry, we demonstrate the conspecificity of this taxon with one of the most common and widespread bumblebee species of Europe, Bombus hortorum (L. 1761). We assign a subspecies status to this endemic taxon (Bombus hortorum reinigiellus comb. nov.) shown to be different in colour and morphology but also in wing shape and relative wing size compared to the other conspecific subspecies. • Following our taxonomic revision, we reassessed the IUCN conservation status of Bombus hortorum both at the continental and Spanish scale. We then propose how historic climatic oscillations of the last Ice age could explain such a phenotypic divergence in a post-glacial refugium and highlight the critical role of establishing unambiguous taxonomic revision prior to any conservation assessment.
... Convergence in coloration due to Müllerian mimicry results in highly similar morphologies among bumblebee species (Williams 2007;Ezray et al. 2019), which rely mainly on chemical signaling for mate recognition (Goulson 2003). These factors make species recognition particularly difficult in bumblebees and studies utilizing multiple genetic loci have recently resulted in the discovery of several previously undescribed species (Martinet et al. 2019;Ghisbain et al. 2020). The number of bumblebee species is likely underestimated with many cryptic species living in sympatry, which may have experienced gene flow during their formation (Bertsch et al. 2004;Murray et al. 2008;Bossert 2015). ...
... Surprisingly, the 284 samples identified as B. sylvicola were split into two distinct clusters, containing 217 and 67 samples, respectively, with no observations of intermediates between the two clusters. The B. bifarius and B. vancouverensis samples also formed two distinct clusters, consistent with their assignment as two separate species by Ghisbain et al. (2020). A neighbor-joining tree also strongly supported the division of the B. sylvicola samples into two clusters with the B. bifarius-B. ...
... None of these traits could be used to distinguish between the two species. Bombus sylvicola samples were significantly larger on average based on measurements of intertegular distance, which is a proxy of body size (mean ¼ 3.84 mm and 3.59 mm for B. sylvicola and B. incognitus, respectively; Wilcoxon rank sum test, W ¼ 7673, Pyrobombus bees were sampled for this study, as well as the sampling locations of Bombus bifarius and B. vancouverensis from a previous study (inset) (Ghisbain et al. 2020). Bombus melanopygus was collected widely across western USA in a previous study (Tian et al. 2019). ...
Full-text available
Evidence is accumulating that gene flow commonly occurs between recently-diverged species, despite the existence of barriers to gene flow in their genomes. However, we still know little about what regions of the genome become barriers to gene flow and how such barriers form. Here we compare genetic differentiation across the genomes of bumblebee species living in sympatry and allopatry to reveal the potential impact of gene flow during species divergence and uncover genetic barrier loci. We first compared the genomes of the alpine bumblebee Bombus sylvicola and a previously unidentified sister species living in sympatry in the Rocky Mountains, revealing prominent islands of elevated genetic divergence in the genome that co-localize with centromeres and regions of low recombination. This same pattern is observed between the genomes of another pair of closely-related species living in allopatry (B. bifarius and B. vancouverensis). Strikingly however, the genomic islands exhibit significantly elevated absolute divergence (dXY) in the sympatric, but not the allopatric, comparison indicating that they contain loci that have acted as barriers to historical gene flow in sympatry. Our results suggest that intrinsic barriers to gene flow between species may often accumulate in regions of low recombination and near centromeres through processes such as genetic hitchhiking, and that divergence in these regions is accentuated in the presence of gene flow.
... The degree of gene flow in bumblebee populations is determined by the distances that reproductive individuals (males and queens) travel in order to mate and establish new nests (Heinrich, 2004;Woodard et al., 2015). Genetic methods have been used to investigate population structure of several species of bumblebees across a variety of habitats in the UK, continental Europe and continental USA Ellis et al., 2006;Ghisbain et al., 2020;Jackson et al., 2018;Jha, 2015;Koch et al., 2017;Woodard et al., 2015). A general finding is that populations of common species tend to exhibit very little structure in the absence of geographical barriers, even at continental scales. ...
... at this scale. Foragers inferred to belong to the same nest (i.e., with identical parents) were always found in the same or adjacent sites, consistent with the limited foraging distance (on average <110 m) inferred previously byGeib et al. (2015) for these species.Studies of several bumblebee species across continental USA, including B. bifarius and B. vosnesenkii, also found in Colorado and across western North America, indicate weak geographical differentiation even at distances over 1000 km(Ghisbain et al., 2020;Lozier et al., 2011). However, the complex topography of mountain ranges F I G U R E 5 Genome-wide association Manhattan plot. ...
Over the last six decades, populations of the bumblebees Bombus sylvicola and Bombus balteatus in Colorado have experienced decreases in tongue length, a trait important for plant‐pollinator mutualisms. It has been hypothesized that this observation reflects selection resulting from shifts in floral composition under climate change. Here we used morphometrics and population genomics to determine whether morphological change is ongoing, investigate the genetic basis of morphological variation, and analyze population structure in these populations. We generated a genome assembly of B. balteatus. We then analyzed whole‐genome sequencing data and morphometric measurements of 580 samples of both species from seven high‐altitude localities. Out of 281 samples originally identified as B. sylvicola, 67 formed a separate genetic cluster comprising a newly‐discovered cryptic species (“incognitus”). However, an absence of genetic structure within species suggests that gene flow is common between mountains. We found a significant decrease in tongue length between bees collected between 2012‐2014 and in 2017, indicating that morphological shifts are ongoing. We did not discover any genetic associations with tongue length, but a SNP related to production of a proteolytic digestive enzyme was implicated in body size variation. We identified evidence of covariance between kinship and both tongue length and body size, which is suggestive of a genetic component of these traits, although it is possible that shared environmental effects between colonies are responsible. Our results provide evidence for ongoing modification of a morphological trait important for pollination and indicate that this trait likely has a complex genetic and environmental basis.
... In this study, we compare patterns of intraspecific trait variation for two bumble bee species, Bombus vosnesenskii and Bombus vancouverensis nearcticus (Ghisbain et al. 2020), sampled across latitude (36.5-48.6° N) and altitude (49-2,293 m) in the Sierra-Cascade Mountain region of California, Oregon, and Washington, United States (Fig. 1). ...
... The species complex to which B. vancouverensis nearcticus belongs (together with Bombus bifarius) tends to be associated with high-elevation habitats throughout much of the United States (Lozier et al. 2011(Lozier et al. , 2016Jackson et al. 2018;Ghisbain et al. 2020). As seen in other montane Bombus (Duennes et al. 2012, Hines and Williams 2012, Williams et al. 2018, this drives much of the phylogeographic history of the group, and it is possible that morphology in such specialists is adapted more for high-elevation challenges than those imposed by temperature alone. ...
Biogeographic clines in morphology along environmental gradients can illuminate forces influencing trait evolution within and between species. Latitude has long been studied as a driver of morphological clines, with a focus on body size and temperature. However, counteracting environmental pressures may impose constraints on body size. In montane landscapes, declines in air density with elevation can negatively impact flight performance in volant species, which may contribute to selection for reduced body mass despite declining temperatures. We examine morphology in two bumble bee (Hymenoptera: Apidae: Bombus Latreille) species, Bombus vancouverensis Cresson and Bombus vosnesenskii Radoszkowski, across mountainous regions of California, Oregon, and Washington, United States. We incorporate population genomic data to investigate the relationship between genomic ancestry and morphological divergence. We find that B. vancouverensis, which tends to be more specialized for high elevations, exhibits stronger spatial-environmental variation, being smaller in the southern and higher elevation parts of its range and having reduced wing loading (mass relative to wing area) at high elevations. Bombus vosnesenskii, which is more of an elevational generalist, has substantial trait variation, but spatial-environmental correlations are weak. Population structure is stronger in the smaller B. vancouverensis, and we find a significant association between elevation and wing loading after accounting for genetic structure, suggesting the possibility of local adaptation for this flight performance trait. Our findings suggest that some conflicting results for body size trends may stem from distinct environmental pressures that impact different aspects of bumble bee ecology, and that different species show different morphological clines in the same region.
... Bees were identified to species following diagnostic characteristics in Thorp et al. (1983), Stephen (1957), and Williams et al. (2014), and were retained by the Woodard Lab at UC Riverside. We refer to some species listed in Thorp et al. (1983) (Ghisbain et al., 2020). We used a combination of morphological characters and geography to distinguish between two very similar species, B. vosnesenskii and B. caliginosus. ...
Full-text available
Bumble bees (genus Bombus) are important pollinators with more than 260 species found worldwide, many of which are in decline. Twenty-five species occur in California with the highest species abundance and diversity found in coastal, northern, and montane regions. No recent studies have examined California bumble bee diversity across large spatial scales nor explored contemporary community composition patterns across the state. To fill these gaps, we collected 1740 bumble bee individuals, representing 17 species from 17 sites (~100 bees per site) in California, using an assemblage monitoring framework. This framework is intended to provide an accurate estimate of relative abundance of more common species without negatively impacting populations through overcollection. Our sites were distributed across six ecoregions, with an emphasis on those that historically hosted high bumble bee diversity. We compared bumble bee composition among these sites to provide a snapshot of California bumble bee biodiversity in a single year. Overall, the assemblage monitoring framework that we employed successfully captured estimated relative abundance of species for most sites, but not all. This shortcoming suggests that bumble bee biodiversity monitoring in California might require multiple monitoring approaches, including greater depth of sampling in some regions, given the variable patterns in bumble bee abundance and richness throughout the state. Our study sheds light on the current status of bumble bee diversity in California, identifies some areas where greater sampling effort and conservation action should be focused in the future, and performs the first assessment of an assembly monitoring framework for bumble bee communities in the state.
... Species revisions for bumblebees now routinely adopt the concept of species as 'evolutionarily Source Smith, 1854von Dalla Torre, 1896Skorikov, 1923Krüger, 1951-1958Williams, 1998Williams et al., 2012Williams, 2021 Combined 1896Skorikov, 1923Reinig, 1935Williams, 1998Williams et al., 2020 (Tables 1-4; e.g. Lecocq et al., 2015;Williams et al., 2016bWilliams et al., , 2019Williams et al., , 2020Martinet et al., 2018;Ghisbain et al., 2020). One source of support is often from coalescent analyses in fast-evolving genes, which often provide the most direct evidence of evolutionarily independent lineages (Monaghan et al., 2005(Monaghan et al., , 2009Zhang et al., 2013). ...
Splitting or lumping of species is a concern because of its potential confounding effect on comparisons of biodiversity and on conservation assessments. By comparing global lists of species reported by previous authors to lists of the presently recognized species that were known to those authors, a simple ratio can be used to describe their relative splitting or lumping of species. One group of 'model' organisms claimed for the study of what species are and how to recognize them is bumblebees. A comparison of four bumblebee subgenera shows: (1) an early phase (up to and including 1931) showing splitting, in which taxonomy was dominated by a typological concept of invariant species with heavy reliance on colour-pattern characters; (2) a middle phase (1935-98) showing lumping, associated with a shift to a polytypic concept of species emphasizing morphological characters, often justified with an interbreeding concept of species, but only rarely associated directly with process-related characters; and (3) a recent phase (after 2000), using a concept of species as evolutionarily independent lineages, as evidenced by corroboration from integrative assessment, usually including evidence for coalescents of species in fast-evolving genes compared with morphology. Analysis of splitting or lumping should help to improve biodiversity comparisons and conservation.
... Stars mark sampling locations Index (LAI) and slope. The importance of temperature did not come as a surprise, as it is a predictor of the distribution of Bombus species [21,77], and known to govern the emergence time of queens from hibernation [78]. Hence, queens hibernating in warmer areas might emerge earlier than those in colder areas. ...
Full-text available
Background The environment is a strong driver of genetic structure in many natural populations, yet often neglected in population genetic studies. This may be a particular problem in vagile species, where subtle structure cannot be explained by limitations to dispersal. Consequently, these species might falsely be considered quasi-panmictic and hence potentially mismanaged. A species this might apply to, is the buff-tailed bumble bee (Bombus terrestris), an economically important and widespread pollinator, which is considered to be quasi-panmictic at mainland continental scales. Here we aimed to (i) quantify genetic structure in 21+ populations of the buff-tailed bumble bee, sampled throughout two Eastern European countries, and (ii) analyse the degree to which structure is explained by environmental differences, habitat permeability and geographic distance. Using 12 microsatellite loci, we characterised populations of this species with Fst analyses, complemented by discriminant analysis of principal components and Bayesian clustering approaches. We then applied generalized dissimilarity modelling to simultaneously assess the informativeness of geographic distance, habitat permeability and environmental differences among populations in explaining divergence. Results Genetic structure of the buff-tailed bumble bee quantified by means of Fst was subtle and not detected by Bayesian clustering. Discriminant analysis of principal components suggested insignificant but still noticeable structure that slightly exceeded estimates obtained through Fst analyses. As expected, geographic distance and habitat permeability were not informative in explaining the spatial pattern of genetic divergence. Yet, environmental variables related to temperature, vegetation and topography were highly informative, explaining between 33 and 39% of the genetic variation observed. Conclusions In contrast to previous studies reporting quasi-panmixia in continental populations of this species, we demonstrated the presence of subtle population structure related to environmental heterogeneity. Environmental data proved to be highly useful in unravelling the drivers of genetic structure in this vagile and opportunistic species. We highlight the potential of including these data to obtain a better understanding of population structure and the processes driving it in species considered to be quasi-panmictic.
... Species with ranges that cross more than one mimicry complex often converge onto distinct mimicry patterns as a result of direct selection for specific phenotypic color patterns in different geographic regions Hines & Williams, 2012;Owen & Plowright, 1980;Williams, 2007). The resulting color pattern diversity has generated taxonomic confusion on species composition, which has motivated several studies to assess species status (e.g., Bossert et al., 2016;Duennes et al., 2012;Ghisbain et al., 2020;Hines & Williams, 2012;Koch et al., 2018;Martinet et al., 2018Martinet et al., , 2019Williams et al., 2020). It also has resulted in ample intraspecific polymorphisms that meet in mimicry transition zones Williams, 2007). ...
Full-text available
As hybrid zones exhibit selective patterns of gene flow between otherwise distinct lineages, they can be especially valuable for informing processes of microevolution and speciation. The bumble bee, Bombus melanopygus, displays two distinct color forms generated by Müllerian mimicry: a northern “Rocky Mountain'’ color form with ferruginous mid-abdominal segments (B. m. melanopygus) and a southern “Pacific'’ form with black mid-abdominal segments (B. m. edwardsii). These morphs meet in a mimetic transition zone in northern California and southern Oregon that is more narrow and transitions further west than comimetic bumble bee species. To understand the historical formation of this mimicry zone, we assessed color distribution data for B. melanopygus from the last 100 years. We then examined gene flow among the color forms in the transition zone by comparing sequences from mitochondrial COI barcode sequences, color-controlling loci, and the rest of the nuclear genome. These data support two geographically distinct mitochondrial haplogroups aligned to the ancestrally ferruginous and black forms that meet within the color transition zone. This clustering is also supported by the nuclear genome, which, while showing strong admixture across individuals, distinguishes individuals most by their mitochondrial haplotype, followed by geography. These data suggest the two lineages most likely were historically isolated, acquired fixed color differences, and then came into secondary contact with ongoing gene flow. The transition zone, however, exhibits asymmetries: mitochondrial haplotypes transition further south than color pattern, and both transition over shorter distances in the south. This system thus demonstrates alternative patterns of gene flow that occur in contact zones, presenting another example of mito-nuclear discordance. Discordant gene flow is inferred to most likely be driven by a combination of mimetic selection, dominance effects, and assortative mating.
Full-text available
Background The environment is a strong driver of genetic structure in many natural populations, yet often neglected in population genetic studies. This may be a particular problem in vagile species, where subtle structure cannot be explained by limitations to dispersal. These species might falsely be considered panmictic and hence potentially mismanaged. Here we analysed the genetic structure in an economically important and widespread pollinator, the buff-tailed bumble bee ( Bombus terrestris ), which is considered to be quasi-panmictic at mainland continental scales. We first quantified population structure in Romania and Bulgaria with spatially implicit Fst and Bayesian clustering analyses. We then incorporated environmental information to infer the influence of the permeability of the habitat matrix between populations (resistance distances) as well as environmental differences among sites in explaining population divergence. Results Genetic structure of the buff-tailed bumble bee was subtle and not detected by Bayesian clustering. As expected, geographic distance and habitat permeability were not informative in explaining the spatial pattern of genetic divergence. Yet, environmental variables related to temperature, vegetation and topography were highly informative, explaining between 33 and 39% of the genetic variation observed. Conclusions Where in the past spatially implicit approaches had repeatedly failed, incorporating environmental data proved to be highly beneficial in detecting and unravelling the drivers of genetic structure in this vagile and opportunistic species. Indeed, structure followed a pattern of isolation by environment, where the establishment of dispersers is limited by environmental differences among populations, resulting in the disruption of genetic connectivity and the divergence of populations through genetic drift and divergent natural selection. With this work, we highlight the potential of incorporating environmental differences among population locations to complement the more traditional approach of isolation by geographic distance, in order to obtain a holistic understanding of the processes driving structure in natural populations.
Full-text available
Bombus glacialis Friese, 1902 is an Arctic bumblebee that was thought to have a disjunctive range on the Arctic Ocean Islands, i.e., on Novaya Zemlya and Wrangel Island, with a 3600 km gap between these isolates. While the species status of the Novaya Zemlya’s population was recently confirmed using a molecular approach, its close affinities to the Wrangel’s isolate were proposed by a morphological similarity alone. Here, we report that B. glacialis samples from Wrangel share three unique COI haplotypes, which are closely related to each other and to those inferred from samples from Novaya Zemlya. Based on a combination of diagnostic morphological and molecular characters, the Wrangel’s population is described here as a new subspecies, Bombus glacialis marinae ssp. nov. Our novel data indicate that this species can be considered a polar relict that had had a much broader distribution along the continental margin of Eurasia during the Pleistocene. The two subspecies of B. glacialis were likely separated in the Middle Pleistocene (mean age = 0.27 Myr, 95% CI = 0.12–0.46 Myr) due to increasing of the sea level during an interglacial. The Novaya Zemlya Archipelago supported a B. glacialis population in the Late Pleistocene. Hence, this insular land was not covered completely by ice sheet during the Last Glacial Maximum. Our findings highlight that relict cold-adapted Pleistocene lineages still survive on remote islands in the Arctic Ocean and that such lineages are highly threatened by recent climate warming.
Full-text available
The bumblebees of the subgenus Alpinobombus of the genus Bombus are unusual among bees for specialising in many of the most northerly vegetated arctic habitats on Earth. Most named taxa in this group (37 available names from a total of 67 names) were described originally from differences in the colour patterns of the hair. Previous revisions have shown unusually little agreement, recognising a range of 6‒9 species, in part because of pronounced intraspecific variation in both skeletal morphology and in the colour patterns of the hair. Here we examine variation among 4622 specimens from throughout the group’s global range. Bayesian inference of the gene tree for the fast evolving mitochondrial COI gene combined with Poisson-tree-process analysis of this tree shows support for 10 gene lineages as candidates for being putative species lineages. Integrative assessment shows that the interpretation of these results is not straightforward. Evidence from the fast evolving mitochondrial 16S ribosomal RNA gene supports two of the COI gene alleles (from the samples B. kluanensis s. str. and ‘unnamed2’) as being associated with just one 16S allele. Double COI bands on the PCR gels for these individuals and double peaks on sequence traces (in one case with both COI alleles sequenced from one individual) identifies this as a likely case of COI paralogy that has resulted in mitochondrial heteroplasmy. Evidence from morphology also supports only the remaining nine lineages as separate. Evidence from extracts of cephalic labial gland secretions (CLGS, with components believed to function as sex pheromones) reported by others shows small diagnostic differences between all of the candidate species examined (although B. kluanensis s. l. was not examined) and shows larger differences between all of the species pairs that we find are likely to have co-occurred at least in the past, revealing a likely limitation to the CLGS approach in cases of recent and continuously allopatric species. Consequently we infer nine species in the subgenus Alpinobombus (so that B. kluanensis s. str. and ‘unnamed2’ are interpreted as conspecific, as B. kluanensis s. l.). We provide distribution maps and identification keys for the nine species. The morphology of the male of B. kluanensis is described for the first time, including a unique, unusually dense pad of short hair on the mandible that may have a function involving CLGS in mate-searching behaviour. In seeking to identify the valid names for these species, seven new lectotypes are designated and support is provided for synonymizing 10 names as proposed in a recent summary table of names. The prevailing usage of Bombus balteatus Dahlbom is maintained as valid by proposing Bombus nivalis Dahlbom and Bombus tricolor Dahlbom as nomina oblita and by proposing Bombus balteatus Dahlbom as a nomen protectum. The prevailing usage of Bombus hyperboreus Schönherr is maintained as valid by supporting Apis arctica Quensel as a nomen oblitum and by supporting Bombus hyperboreus Schönherr as a nomen protectum. We then use sequence data from COI and 16S together with nuclear PEPCK and opsin genes to estimate dated phylogenetic relationships among the nine species, allowing for incongruent gene trees with *BEAST. If crown-group divergence within the subgenus Alpinobombus coincided with the global climate cooling and with the growth of the northern ice sheets at the end of the Miocene at ca 7.2 Ma, then divergences between each of the three pairs of sister species are likely to have coincided with fluctuations in vegetated land connections across the Bering Strait after ca 2.5 Ma.
Full-text available
Natural phenotypic radiations, with their high diversity and convergence, are well-suited for informing how genomic changes translate to natural phenotypic variation. New genomic tools enable discovery in such traditionally nonmodel systems. Here, we characterize the genomic basis of color pattern variation in bumble bees (Hymenoptera, Apidae, Bombus), a group that has undergone extensive convergence of setal color patterns as a result of Müllerian mimicry. In western North America, multiple species converge on local mimicry patterns through parallel shifts of midabdominal segments from red to black. Using genome-wide association, we establish that a cis-regulatory locus between the abdominal fate-determining Hox genes, abd-A and Abd-B, controls the red–black color switch in a western species, Bombus melanopygus. Gene expression analysis reveals distinct shifts in Abd-B aligned with the duration of setal pigmentation at the pupal–adult transition. This results in atypical anterior Abd-B expression, a late developmental homeotic shift. Changing expression of Hox genes can have widespread effects, given their important role across segmental phenotypes; however, the late timing reduces this pleiotropy, making Hox genes suitable targets. Analysis of this locus across mimics and relatives reveals that other species follow independent genetic routes to obtain the same phenotypes.
Full-text available
Natural phenotypic radiations, with their high diversity and convergence, are well-suited for informing how genomic changes translate to natural phenotypic variation. New genomic tools enable discovery in such traditionally nonmodel systems. Here, we characterize the genomic basis of color pattern variation in bumble bees (Hymenoptera, Apidae, Bombus), a group that has undergone extensive convergence of setal color patterns as a result of Müllerian mimicry. In western North America, multiple species converge on local mimicry patterns through parallel shifts of midabdominal segments from red to black. Using genome-wide association, we establish that a cis-regulatory locus between the abdominal fate-determining Hox genes, abd-A and Abd-B, controls the red–black color switch in a western species, Bombus melanopygus. Gene expression analysis reveals distinct shifts in Abd-B aligned with the duration of setal pigmentation at the pupal–adult transition. This results in atypical anterior Abd-B expression, a late developmental homeotic shift. Changing expression of Hox genes can have widespread effects, given their important role across segmental phenotypes; however, the late timing reduces this pleiotropy, making Hox genes suitable targets. Analysis of this locus across mimics and relatives reveals that other species follow independent genetic routes to obtain the same phenotypes.
Full-text available
Bumble bees (Bombus Latrielle) are significant pollinators of flowering plants due to their large body size, abundant setae, and generalist foraging strategies. However, shared setal coloration patterns among closely and distantly related bumble bee species makes identification notoriously difficult. The advent of molecular genetic techniques has increased our understanding of bumble bee evolution and taxonomy, and enables effective conservation policy and management. Individuals belonging to the North American Bombus fervidus species-complex (SC) are homogenous in body structure but exhibit significant body color phenotype variation across their geographic distribution. Given the uncertainty of the genea-logical boundaries within the SC, some authors have synonymized all members of the B. fer-vidus SC within a single taxon, while others propose an alternative two taxa hypothesis. Operating under the phylogenetic species concept, our analysis supports the hypothesis that there are two independent lineages of bumble bees within the B. fervidus SC. With the current evidence, however, it is not possible to assign valid names to either of them, because both lineages include the color phenotypes found in the original species descriptions of B. fervidus and B. californicus. Cryptic speciation does not seem to be the product of Mü llerian mimicry between the clades, because diverging coloration patterns are observed when the distribution of the clades overlaps. Furthermore, within each lineage there is evidence for strong population differentiation that is correlated with geographic distribution rather than color phenotype. In our study, we demonstrate the importance of obtaining a broad sample of multiple populations when conducting lower-level phylogenetic analyses. In addition to improving our knowledge of bumble bee diversification patterns, characterizing the evolutionary history of these pollinators provides the foundation needed to guide contemporary conservation assessments and management strategies.
Full-text available
The molecular evolutionary genetics analysis (Mega) software implements many analytical methods and tools for phylogenomics and phylomedicine. Here, we report a transformation of Mega to enable cross-platform use on Microsoft Windows and Linux operating systems. Mega X does not require virtualization or emulation software and provides a uniform user experience across platforms. Mega X has additionally been upgraded to use multiple computing cores for many molecular evolutionary analyses. Mega X is available in two interfaces (graphical and command line) and can be downloaded from free of charge.
Full-text available
Bayesian inference of phylogeny using Markov chain Monte Carlo (MCMC) (Drummond et al., 2002; Mau et al., 1999; Rannala and Yang, 1996) flourishes as a popular approach to uncover the evolutionary relationships among taxa, such as genes, genomes, individuals or species. MCMC approaches generate samples of model parameter values - including the phylogenetic tree -drawn from their posterior distribution given molecular sequence data and a selection of evolutionary models. Visualising, tabulating and marginalising these samples is critical for approximating the posterior quantities of interest that one reports as the outcome of a Bayesian phylogenetic analysis. To facilitate this task, we have developed the Tracer (version 1.7) software package to process MCMC trace files containing parameter samples and to interactively explore the high-dimensional posterior distribution. Tracer works automatically with sample output from BEAST (Drummond et al., 2012), BEAST2 (Bouckaert et al., 2014), LAMARC (Kuhner, 2006), Migrate (Beerli, 2006), MrBayes (Ronquist et al., 2012), RevBayes (Höhna et al., 2016) and possibly other MCMC programs from other domains.
Müllerian mimicry theory states that frequency-dependent selection should favour geographical convergence of harmful species onto a shared colour pattern. As such, mimetic patterns are commonly circumscribed into discrete mimicry complexes, each containing a predominant phenotype. Outside a few examples in butterflies, the location of transition zones between mimicry complexes and the factors driving mimicry zones has rarely been examined. To infer the patterns and processes of Müllerian mimicry, we integrate large-scale data on the geographical distribution of colour patterns of social bumblebees across the contiguous United States and use these to quantify colour pattern mimicry using an innovative, unsupervised machine-learning approach based on computer vision. Our data suggest that bumblebees exhibit geographically clustered, but sometimes imperfect colour patterns, and that mimicry patterns gradually transition spatially rather than exhibit discrete boundaries. Additionally, examination of colour pattern transition zones of three comimicking, polymorphic species, where active selection is driving phenotype frequencies, revealed that their transition zones differ in location within a broad region of poor mimicry. Potential factors influencing mimicry transition zone dynamics are discussed.
While bumblebees have been the focus of much research, the taxonomy of many species groups is still unclear, especially within circum-polar species. Delimiting species based on multisource datasets provides a solution to overcome current systematic issues of closely related populations. Here, we use an integrative taxonomic approach based on novel genetic and eco-chemical datasets to resolve the taxonomic status of Bombus lapponicus and B. sylvicola. Our results support the conspecific status of B. lapponicus and B. sylvicola and that the low gradual divergence around the Arctic Circle between Fennoscandia and Alaska does not imply speciation in this species complex. Therefore, based on our molecular and morphological analyses, we propose to assign the sub-specific status: B. lapponicus lapponicus comb. nov. from Fennoscandia and W-Siberia and B. lapponicus sylvicola comb. nov. from Alaska and Yukon. Moreover, our analyses reveal a new cryptic arctic species in the B. lapponicus complex: B. (Pyrobombus) interacti sp. nov. Martinet, Brasero & Rasmont from Alaska.
Identifying drivers of dispersal limitation and genetic differentiation is a key goal in biogeography. We examine patterns of population connectivity and genetic diversity using Restriction‐site Associated DNA sequencing (RADseq) in two bumble bee species, Bombus vosnesenskii and Bombus bifarius across latitude and altitude in mountain ranges from California, Oregon, and Washington, U.S.A. Bombus vosnesenskii, which occurs across a broader elevational range at most latitudes, exhibits little population structure while B. bifarius, which occupies a relatively narrow higher elevation niche across most latitudes, exhibits much stronger population differentiation, although gene flow in both species is best explained by isolation with environmental niche resistance. A relationship between elevational habitat breadth and genetic diversity is also apparent, with B. vosnesenskii exhibiting relatively consistent levels of genetic diversity across its range, while B. bifarius has reduced genetic diversity at low latitudes, where it is restricted to high elevation habitat. The results of this study highlight the importance of the intersect between elevational range and habitat suitability in influencing population connectivity and suggest that future climate warming will have a fragmenting effect even on populations that are presently well connected, as they track their thermal niches upward in montane systems. This article is protected by copyright. All rights reserved.