Molecular Phylogenetics and Evolution 50 (2009) 290–309
Contents lists available at ScienceDirect
Molecular Phylogenetics and Evolution
journal homepage: www.elsevier .com/locate /ympev
Phylogeny of colletid bees (Hymenoptera: Colletidae) inferredfrom four nuclear genes
Eduardo A.B. Almeida a,b,*, Bryan N. Danforth a
a Cornell Univ., Dept. Entomol., Ithaca, NY 14853, USAb Depto. de Zoologia, Universidade Federal do Paraná, Lab. Biol. Comparada Hymenoptera, Caixa Postal 19020, CEP 81531-980, Curitiba, Paraná, Brazil
a r t i c l e i n f o
Article history:Received 8 May 2008Revised 12 August 2008Accepted 23 September 2008Available online 22 October 2008
Keywords:ColletidaeBee phylogenyMolecular phylogenyMolecular evolutionGondwanaSystematics
1055-7903/$ - see front matter � 2008 Elsevier Inc. Adoi:10.1016/j.ympev.2008.09.028
* Corresponding author. Address: Depto. de ZooloParaná, Lab. Biol. Comparada Hymenoptera, Caixa PCuritiba, Paraná, Brazil. Fax: +55 41 3361 1763.
E-mail address: [emailprotected] (E.A.B. Alm
a b s t r a c t
Colletidae comprise approximately 2500 species of bees primarily distributed in the southern continents(only two colletid genera are widely distributed: Colletes and Hylaeus). Previously published studies havefailed to resolve phylogenetic relationships on a worldwide basis and this has been a major barrier to theprogress of research regarding systematics and evolution of colletid bees. For this study, data from fournuclear gene loci: elongation factor-1a (F2 copy), opsin, wingless, and 28S rRNA were analyzed for 122species of colletid bees, representing all subfamilies and tribes currently recognized; 22 species belongingto three other bee families were used as outgroups. Bayesian, maximum likelihood, and parsimony meth-ods were employed to investigate the phylogenetic relationships within Colletidae and resulted in highlycongruent and well-resolved trees. The phylogenetic results show that Colletidae are monophyletic andthat all traditionally recognized subfamilies (except Paracolletinae) are also strongly supported as mono-phyletic. Our phylogenetic hypothesis provides a framework within which broad questions related to thetaxonomy, biogeography, morphology, evolution, and ecology of colletid bees can be addressed.
� 2008 Elsevier Inc. All rights reserved.
1. Introduction
Colletidae are generally considered to constitute the most an-cient bee lineage, i.e., the sister-group to all other bees (e.g., Mich-ener, 1944, 1974, 1979; O’Toole and Raw, 1991; Engel, 2001). Thishypothesis is not ubiquitous, though, and its validity has beenquestioned (e.g., McGinley, 1980; Michener, 2007; Danforthet al., 2006a,b). The most comprehensive studies of bee phylogenywere those by Alexander and Michener (1995) and Danforth et al.(2006b). Alexander and Michener (1995) analyzed a morphologicalmatrix of adult and larval characters based on weighted and un-weighted parsimony. Their data firmly established the monophylyof many bee families, but their results were inconclusive about theoverall relationships among families. Tree topologies varied widelywith different analytical methods and nodal support was weak,especially at the base of the tree. Danforth et al. (2006b) incorpo-rated data from five nuclear genes and analyzed the data sepa-rately and in combination with Alexander and Michener’smorphological matrix. Their results strongly pointed to the rootof the bee clade being positioned among the lineages of ‘‘Melitti-dae” (sensu lato). Colletidae were placed far from the root nodeand statistical tests suggested that a basal placement of Colletidae
ll rights reserved.
gia, Universidade Federal doostal 19020, CEP 81531-980,
eida).
was significantly incongruent with the data (Danforth et al.,2006b).
1.1. Colletid monophyly
Some of Alexander and Michener’s (1995) analyses did notstrongly support the monophyly of Colletidae, even though thereare a number of unique traits, which appear to be synapomorphiesfor this family. The cellophane-like cell lining (e.g., Hefetz et al.,1979; Espelie et al., 1992; reviewed by Almeida, 2008a) has beenconsidered to be the strongest evidence for monophyly of the fam-ily because it is a unique and unreversed character found onlyamong colletid bees. McGinley (1980) presented a list of potentialsynapomorphies from mouthpart morphology. The most recentevidence for the monophyly of this family was given by a molecu-lar character: the presence of a unique intron in the F1 copy of thegene elongation factor-1a in all colletids sampled, but not inStenotritidae or any other bee family (Brady and Danforth, 2004).Additionally, Colletidae have been consistently recovered as astrongly supported monophyletic group based on several indepen-dent molecular data sets (Danforth et al., 2006a,b).
1.2. Diversity and classification of Colletidae
The Colletidae, as considered here, are currently divided into se-ven subfamilies: Colletinae, Euryglossinae, Hylaeinae, Paracolleti-
mailto:[emailprotected]
Stenotritidae
Diphaglossini
Caupolicanini
Dissoglottini
Colletinae s.str.
Euryglossinae
Hylaeinae
Xeromelissinae
Caupolicanini
Diphaglossini
Dissoglottini
Colletinae s.str.
Stenotritidae
Euryglossinae
Hylaeinae
Xeromelissinae
Diphaglossinae
Stenotritidae
DiphaglossiniCaupolicaniniDissoglottini
Colletinae s.str.Callomelitta (Paracolletinae)Paracolletinae (part)
EuryglossinaeScrapter (Paracolletinae)HylaeinaeXeromelissinae
other bee families
Paracolletinae
Diphaglossinae
Paracolletinae
Diphaglossinae
a
c d
bStenotritidae
Diphaglossini
Caupolicanini
Colletinae s.str.
Xeromelissinae
Euryglossinae
Hylaeinae
Paracolletinae
“Diphaglossinae”
Fig. 1. Historical account of attempts to describe relationships among subfamiliesand tribes of colletid bees as shown in summary trees of: (a) Michener (1944, p.230); (b) Michener (1974, p. 23); (c) Engel (2001, p. 156); (d) Alexander andMichener (1995, p. 398), consensus of four trees resulting from the analysis of thecomplete data set using implied weights parsimony. The African genus Scrapterused to be considered part of Paracolletinae when these four hypotheses werepresented. Stenotritidae is also included in the figures because, historically, it hasbeen considered to be part of Colletidae by several authors.
E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309 291
nae, Scrapterinae, and Xeromelissinae. These subfamilies corre-spond to the five subfamilies recognized by Michener (2007), ex-cept that the three tribes of Colletinae: Colletini, Paracolletini,and Scrapterini are treated as independent subfamilies. This classi-fication is also congruent with the classification by Melo and Gonç-alves (2005), except that those authors treat the subfamilies ofColletidae as tribes.
Colletid bees are diverse, ranging from small, slender, relativelyhairless bees (such as Euryglossinae) to large, robust, hairy bees(such as Diphaglossinae). Females carry pollen either externally,in a well-developed trochanteral and femoral scopa (Colletinaeand Diphaglossinae) or a scopa formed by sparse hairs on the hindlegs and long hairs primarily on the second abdominal sternum(Xeromelissinae); or internally in the crop (Euryglossinae, Hylaei-nae, and few Paracolletinae). There is only one known group of cle-ptoparasites (some Hawaiian species of the Hylaeus (Nesoprosopis)Perkins [Daly and Magnacca, 2003]). Floral relationships in Collet-idae range from polylectic (generalist) to oligolectic (specialist),with some taxa showing very narrow host-plant preferences(e.g., Wcislo and Cane, 1996). The highest diversity of colletid beesis observed in the temperate parts of southern South America andin Australia. Worldwide, there are 2485 available species names inColletidae (Ascher et al., 2008).
The Colletinae are widely distributed (they do not occur inAustralia, though) and comprise approximately 480 species. Col-letinae are morphologically hom*ogenous compared to theremainder of Colletidae, and their body size varies from 7 to16 mm in length. The Diphaglossinae include the largest colletidbees (up to 24 mm in length). These Neotropical bees are, ingeneral, fast flying, and many of them fly only early in the morn-ing or before and around dusk. There are 128 available speciesnames. This is the only colletid subfamily subdivided into tribes,and Michener (1986) resolved the tribal relationships as follows:(Caupolicanini, (Diphaglossini, Dissoglottini)). The Euryglossinaeare a subfamily occurring mostly in the Australian Region (onespecies has been introduced in South Africa) and comprise small,hairless bees. There are almost 400 species of Euryglossinae. TheHylaeinae comprise more species than any other subfamily ofColletidae, with about 900 available names. Morphological andtaxonomic diversity is highest in Australia where the group isthought to have originated; all species occurring outside theAustralian region belong in the genus Hylaeus. Their wasp-likeappearance and lack of external pollen-carrying structures madeHylaeinae candidates for the most primitive bees (e.g., Jander,1976; Michener, 1979; see also discussion in Michener, 2007,88–92). Females of Euryglossinae also lack pollen-carrying struc-tures (i.e., a scopa) and transport pollen internally. The Paracol-letinae are a morphologically diverse group of bees that rangesin size from 6 to 18 mm in length and there are over 400 de-scribed species, distributed throughout the Australian and Neo-tropical regions, mainly in subtropical and temperate drybiomes. The Scrapterinae are a monogeneric subfamily compris-ing 40 species of the endemic African genus Scrapter, most ofwhich are distributed in southern Africa, especially the Cape re-gion. One species was recently described from Kenya (Davieset al., 2005). Host-plant preferences and adult morphology arequite variable within this group of bees; body length varies from3.5 to 14 mm in length (Davies and Brothers, 2007). Research onthe taxonomy and biology of Scrapter (Rozen and Michener,1968; Eardley, 1996; Davies et al., 2005; Davies and Brothers,2007) makes it one of the best-studied groups of Colletidae. Un-til recently, Scrapter was considered as part of Paracolletinae. TheXeromelissinae are small, slender, and not very hairy, eventhough they have a small scopa on the hind leg and metasoma.Body length varies from 2.5 to 7.0 mm in length. There are al-most 120 described species with the highest diversity in temper-
ate regions of Chile and Argentina, but the group extendsthrough the Neotropical Region as far north as Mexico. Packer(2008) and Almeida et al. (2008) present results of two phyloge-netic studies of Xeromelissinae based on morphology and com-bined morphological and molecular data, respectively.
Although the composition of colletid subfamilies is largely set-tled, relationships among them are poorly understood. One of thefew points of agreement regarding the phylogenetic affinities with-in Colletidae is the grouping of Euryglossinae, Hylaeinae, and Xer-omelissinae. A graphical summary of four hypotheses of colletidphylogeny is presented in Fig. 1.
Alexander and Michener (1995) investigated the relationshipsamong the main lineages of bees based on a morphological dataset of adult and larval characters. Although a large number ofcolletid species were included in their analyses, the conclusionsone can draw from their results are limited by the extremelyvariable positioning of the groups depending on the way datawere analyzed. A summary of one of the consensus trees pre-sented by Alexander and Michener (1995) is shown in Fig. 1d.Examination of previous hypotheses of relationships among lin-eages of Colletidae based either on taxonomic experience andintuition of bee researchers (Fig. 1a–c), or on explicit data anal-ysis (Fig. 1d) reveals great uncertainty surrounding the relation-ships within this family.
The purpose of this study is to re-assess the phylogenetic rela-tionships within Colletidae. This was done using a novel source ofdata for this group: genetic sequence data from four nuclear geneloci. A better understanding of colletid phylogeny is essential forproviding the rationale of an improved classification of these bees,as well as for the reconstruction of their evolutionary and biogeo-graphical history.
292 E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309
2. Materials and methods
2.1. Taxon sampling
We assembled a data set that comprises a total of 144 terminalspecies; 122 species of Colletidae were sampled to represent alltraditionally accepted subfamilies and tribes of this family. Specialattention was paid to the Colletinae s.l. (sensu Michener, 2007; i.e.,Colletinae + Paracolletinae + Scrapterinae) because it is widely be-lieved to be para- or polyphyletic. Eighty species of Colletinae s.l.were included in the analysis (15 Colletinae s.str., 60 Paracolleti-nae, and 5 Scrapterinae). The remaining 42 species included repre-sentatives of the three tribes of Diphaglossinae [Caupolicanini (9spp.), Diphaglossini (4 spp.), and Dissoglottini (2 spp.)]; Eurygloss-inae (6 spp.); Hylaeinae (11 spp.); and Xeromelissinae (8 spp.).
Twenty-two representatives of three bee families were used asoutgroups for this analysis: Andrenidae (8 spp.), Halictidae (10spp.), and Stenotritidae (4 spp.). These represent the main lineagesof bees closely related to Colletidae according to the results of Dan-forth et al. (2006a,b). This choice of outgroups does not imply anagreement with a given placement of Colletidae relative to otherbee lineages. The analysis of a very similar molecular data set withtaxon sampling that intended to resolve the relationships amongbee families resulted in the following arrangement for the closestgroups to Colletidae: (Andrenidae (Halictidae (Colletidae, Stenotri-tidae))) (Danforth et al., 2006b). This means that the relationshipsamong bee families based on a data set of this sort are well-estab-lished. The purpose of outgroups is to establish the position of theingroup root node, and Andrenidae, Halictidae, and Stenotritidaeshould provide sufficient information for rooting Colletidae. Evenif Colletidae are indeed the earliest lineage of bees, the outgroupchoice still remains unproblematic. In Nixon and Carpenter’s(1993, 423) words: ‘‘[o]utgroup(s) need not be ‘‘primitive” relativeto the ingroup”.
Specimens used for sequencing were primarily preserved in 95%EtOH but recently collected pinned specimens and frozen speci-mens were also used. Pinned specimens older than 3–5 years werenot suitable for DNA extractions but those collected more recentlyprovided good quality, high-molecular weight DNA for PCR. Out-group and ingroup taxa included in this study, locality data, spec-imen voucher numbers, and GenBank Accession Nos. are listed inTable 1. The detailed generic classification adopted in this studyis presented in Appendix 1 (Supplementary materials). Voucherspecimens are housed in the Cornell University Insect Collectionor at the institutions that provided specimens for DNA extractions.Identification of various taxa included in the analyses was facili-tated by comparison to material deposited in entomological collec-tions and assistance of other bee taxonomists, especially J.S. Ascher(American Museum of Natural History, USA), T. Houston (WesternAustralian Museum, Australia), M. Kuhlmann (University of Muen-ster, Germany), G.A.R. Melo (Universidade Federal do Paraná, Bra-zil), and L. Packer (York University, Canada).
2.2. Data
Molecular data were collected from four nuclear loci that havebeen providing robust results for insect phylogenetic studies. Threeof them are nuclear protein-coding genes: EF-1a (elongation fac-tor-1a, F2 copy), opsin (long wavelength, green rhodopsin), andwingless (wg), and the fourth is a nuclear ribosomal RNA locus:the D1–D5 expansion regions and related core elements of thelarge subunit 28S rRNA (28S rRNA). Maps of the four gene lociand the relative positions of the primers used in this study areshown in Fig. 2, and primer information for each gene can be foundin Table 2. Each of these gene loci is referred to as a partition here-
after; introns and exons of EF-1a are treated as two separate par-titions. Five partitions are therefore recognized: exons and intronsof EF-1a, opsin (exons), wg (exons), and 28S rRNA.
Slowly evolving, nuclear genes are commonly used for phyloge-netic analysis in many groups of insects. They have been demon-strated to recover Cretaceous-age divergences (e.g., Danforthet al., 1999, 2004, 2006a,b; Wiegmann et al., 2000). Danforthet al. (2004, 310–311) reviewed the phylogenetic utility of EF-1a,opsin, and wg, and commented on the biological functions of theirgene products. Among these genes, EF-1a has been the mostwidely used nuclear protein-coding gene for insect phylogenetics.Despite alignment challenges of some ribosomal DNA regions,these genes cannot be distinguished from protein-coding genesregarding phylogenetic utility and nucleotide substitution patterns(Danforth et al., 2005). Among the various ribosomal gene loci, 18SrRNA tends to have the slowest rate of substitution (Hillis and Dix-on, 1991), while 28S rRNA has more variation and, therefore, morephylogenetic signal.
The three protein-coding genes used here comprised the dataset used by Danforth et al. (2004) to assess phylogenetic relation-ships within Halictidae, except that the exon of wg sampled for thepresent study is approximately 250 bp longer on the upstream (i.e.,50 end, Sipes, personal communication). The data set of Danforthet al. (2004) was robust enough to resolve relationships amonghalictid bees. Halictidae and the clade formed by Colletidae andStenotritidae are presumably roughly the same age, both havingoriginated between mid- and late-Cretaceous, are closely related,and have approximately the same species diversity. This providesthe basis for the initial assumption that this molecular data setshould have enough phylogenetic signal to resolve relationshipswithin either family.
2.3. DNA extraction
Genomic DNA was extracted using phenol–chloroform proto-cols (Doyle and Doyle, 1990, adapted by Danforth, 1999), but with-out use of liquid nitrogen and RNase. Tissue was taken from thethoracic musculature and/or legs depending on the rarity and sizeof available specimens. Samples were (1) macerated in individual1.5 ml Eppendorf tubes with 2� CTAB extraction buffer and100 mg Proteinase K; (2) incubated for 2–4 h at 55 �C; (3) extractedwith 24:1 chloroform–isoamyl alcohol, (4) extracted again with25:24:1 phenol–chloroform–isoamyl alcohol; and (5) 24:1 chloro-form–isoamyl alcohol.
The phenol–chloroform–isoamyl alcohol stage was performedin Phase-Lock Gel� 2.0 ml Eppendorf tubes to facilitate the separa-tion of phenol from the remainder and thus increase the final DNAyield. DNA was (1) precipitated with 2.5 volumes of ice-cold 100%ethanol and 0.1 volume 3 M sodium acetate; (2) washed with 80%ethanol; and (3) resuspended in 50 ml Tris–EDTA (pH 7.6) buffer.
2.4. PCR and sequencing
PCR amplifications of the genes listed above were done for 35cycles under the following conditions: an initial denaturation at94 �C for 60 s, followed by 35 cycles of denaturation at 94 �C for60–90 s, annealing at 48–58 �C, 60–90 s, and extension at 72 �C,60–90 s. Specific conditions for each locus amplified are listed inTable 2. Prior to sequencing, most PCR products were gel-purifiedin low melting point agarose gels (FMC, Rockland, Maine) over-night at 4 �C. DNA was recovered from gel slices using the PromegaWizard PCR Preps DNA Purification kit. Gel purification was unnec-essary for PCR products that produced a single product: both frag-ments of 28S rRNA and the upstream (1100 bp) fragment of EF-1a.Automatic DNA sequencing was performed using the Applied Bio-systems Automated 3730 DNA Analyzer employing Big Dye Termi-
Table 1List of species included in this study, taxonomic information, locality data, and GenBank Accession nos.
Species [voucher code] Classification Collecting data EF-1a Opsin Wingless 28S rRNA
Colletes bicolor Smith, 1879 [EA0082] Colletidae:Colletinae
Argentina: Tucumán. Amaicha del Valle.24.x.2004
DQ884650 DQ884549 DQ884802 DQ768539
Colletes compactus Cresson, 1868 [EA0014] Colletidae:Colletinae
USA: New York. Tompkins Co., Ithaca.29.ix.1998
DQ884642 DQ884542 DQ884794 DQ768531
Colletes distinctus Cresson, 1868 [EA0013]b Colletidae:Colletinae
USA: California. SC: Berkeley Co., Honey Hill.27.ii.2000
DQ884641 — DQ884793 DQ768530
Colletes floralis Eversmann, 1852 [EA0076] Colletidae:Colletinae
Mongolia: Arkhangay Aimag. 10.vii. 2004 DQ884649 DQ884548 DQ884801 DQ768538
Colletes furfuraceus Holmberg, 1886 [EA0108] Colletidae:Colletinae
Argentina: Prov. Tucumán. Amaicha del Valle.25–26.x.2004
DQ884651 DQ884550 DQ884803 DQ768540
Colletes gilvus Vachal, 1909 [EA0073] Colletidae:Colletinae
Chile: Region I. Murmuntani (near Putre).Iv.2004
DQ884648 DQ884547 DQ884800 DQ768537
Colletes inaequalis Say, 1837 [Coin450] Colletidae:Colletinae
USA: NY. Tompkins Co., Ithaca AY585123 DQ115542 DQ884804 AY654484
Colletes pascoensis co*ckerell, 1898 [EA0015] Colletidae:Colletinae
USA: California. Contra Costa Co., El Cerrito.14.iv.2001
DQ884643 — DQ884795 DQ768532
Colletes seminitidus Spinola, 1851 [EA0045] Colletidae:Colletinae
Chile: Region IV. PanAm N Los Hornos. 9.x.2002 DQ884647 DQ884546 DQ884799 DQ768536
Colletes simulans Cresson, 1868 [EA0016] Colletidae:Colletinae
USA: New York. Tompkins Co., Ithaca.21.viii.1997
DQ884644 DQ884543 DQ884796 DQ768533
Colletes skinneri Viereck, 1903 [Cosk632] Colletidae:Colletinae
USA: AZ. Cochise Co., Chiricahua Monument.14.ix.1999
AY230130 AY227912 DQ884805 AY654485
Colletes thoracicus Smith, 1853 [EA0017] Colletidae:Colletinae
USA: Florida. Alachua Co., Gainsville.15.iii.2002
DQ884645 DQ884544 DQ884797 DQ768534
Hemicotelles ruizii (Herbst, 1923) [EA0032] Colletidae:Colletinae
Chile: Region IV. El Equi, Las Placetas.14.x.2001
DQ884638 DQ884539 DQ884790 DQ768527
Rhynchocolletes mixtus (Toro and Cabezas, 1977)[EA0033]
Colletidae:Colletinae
Chile: Region IV. Pejerreyes. 19.x.2001 DQ884639 DQ884540 DQ884791 DQ768528
Xanthocotelles sicheli (Vachal, 1909) [EA0056] Colletidae:Colletinae
Chile: Region VII. Curicó, Laguna de Teno. 1–5.ii.2003
DQ884640 DQ884541 DQ884792 DQ768529
Caupolicana bicolor Friese, 1899 [EA0041] Colletidae:Diphaglossinae
Chile: Region II. Huasco, 4 km N Domeyko.10.xi.2000
DQ884576 DQ884484 DQ884718 DQ768465
Caupolicana quadrifasciata Friese, 1898 [EA0040] Colletidae:Diphaglossinae
Chile: Region IV. Fray Jorge Pq. Ntl. 11.x.2001 DQ884575 DQ884483 DQ884717 DQ768464
Caupolicana vestita (Smith, 1879) [Cpve848] Colletidae:Diphaglossinae
Chile: Region I. Arica Playa las Machas. AY585124 DQ115543 DQ884726 DQ872758
Caupolicana yarrowi (Cresson, 1875) [Cpya654] Colletidae:Diphaglossinae
USA: NM. Hidalgo Co., 20 mi S. Animas.24.ix.1999
— DQ115544 DQ884727 AY654487
Ptiloglossa sp. [EA0018] Colletidae:Diphaglossinae
Costa Rica: Prov. Guanacaste, Carmona.15.i.2003
DQ884573 DQ884481 DQ884715 DQ768462
Ptiloglossa tarsata (Friese, 1900) [EA0085] Colletidae:Diphaglossinae
Argentina: Salta. General Güemes. 11.xi.2004 DQ884581 DQ884488 DQ884723 DQ768470
Ptiloglossa thoracica (Fox, 1895) [EA0078] Colletidae:Diphaglossinae
Mexico: Jalisco. Res. Chamela. 06.ix.2004 DQ884579 — DQ884721 DQ768468
Willinkapis chalybaea (Friese, 1906) [EA0118] Colletidae:Diphaglossinae
Argentina: Catamarca. 40 km N Andalgalá.15.ii.2003
DQ884582 DQ884489 DQ884724 DQ768471
Zikanapis clypeata (Smith, 1879) [EA0119] Colletidae:Diphaglossinae
Mexico: Jalisco. Res.Biosfera (Manantlan).11.ix.2004
DQ884583 DQ884490 DQ884725 DQ768472
Cadeguala albopilosa (Spinola, 1851) [EA0036] Colletidae:Diphaglossinae
Chile: Region VIII. Cordillera, El Manzano.xii.2001
DQ884574 DQ884482 DQ884716 DQ768463
Cadeguala occidentalis (Haliday, 1836) [EA0047] Colletidae:Diphaglossinae
Chile: Region V. Colliguay. 19.x.2002 DQ884577 DQ884485 DQ884719 DQ768466
Cadegualina andina (Friese, 1925) [EA0048] Colletidae:Diphaglossinae
Colombia: Boyacá. El Níspero. 12.xii.2001-19.i.2002
DQ884578 DQ884486 DQ884720 DQ768467
Diphaglossa gayi Spinola, 1851 [Diga850] Colletidae:Diphaglossinae
Chile: Region X. Aquas Calientes. AY585125 DQ115545 DQ884728 DQ872759
Mydrosoma aterrimum (Friese, 1925) [EA0156] Colletidae:Diphaglossinae
Bolivia: La Paz, Prov. Coroico, Cerro Uchumachi.05.iv.2004
EF032902 EF032903 EF032905 EF028342
Mydrosoma fallax (Moure, 1953) [EA0081] Colletidae:Diphaglossinae
Argentina: Salta. General Güemes. 11.xi.2004 DQ884580 DQ884487 DQ884722 DQ768469
Callohesma calliopsella (co*ckerell, 1910) [Euca688] Colletidae:Euryglossinae
Australia: Victoria. Yan yaen. 20.xi.1999 AY585126 DQ115550 DQ884809 DQ872768
Euhesma aff. crabronica (co*ckerell, 1914) [EA0155] Colletidae:Euryglossinae
Australia: WA; Eurardy Stat. 09.x.2005 DQ884654 — DQ884808 DQ768543
Euhesma platyrhina (co*ckerell, 1915) [EA0148] Colletidae:Euryglossinae
Australia: WA; Kalbarri Ntl.Prk. 08.x.2005 DQ884652 — DQ884806 DQ768541
Euhesma sp. [EA0149] Colletidae:Euryglossinae
Australia: WA. 15 km ESE Southern Cross.23.ix.2005
DQ884653 — DQ884807 DQ768542
Euryglossina globuliceps (co*ckerell, 1918)[Eugl692]
Colletidae:Euryglossinae
Australia: Victoria. Colquhuon State Forest.26.xi.1999
AY585127 DQ115551 DQ884810 DQ872769
Xanthesma furcifera (co*ckerell, 1913) [Xnfu709] Colletidae:Euryglossinae
Australia: Victoria. Patchewollock. 10.xii.1999. AY585140 DQ115552 DQ884811 DQ872770
Amphylaeus (Agogenohylaeus) obscuriceps (Friese,1924) [KM264]
Colletidae:Hylaeinae
Australia: Queensland. Kawana Waters.10.xii.2002
DQ884687 DQ884564 DQ884857 DQ768597
Hylaeus (Euprosopis) disjunctus (co*ckerell, 1905)[KM252]
Colletidae:Hylaeinae
Australia: Queensland. Somerset Dam.14.xii.2002
DQ884677 DQ884563 DQ884848 DQ768586
(continued on next page)
E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309 293
Table 1 (continued)
Species [voucher code] Classification Collecting data EF-1a Opsin Wingless 28S rRNA
Hylaeus (Euprosopis) elegans (Smith, 1853)[Hyel697]
Colletidae:Hylaeinae
Australia: South Australia. 10 km E Kimba.5.i.1999
AY585129 DQ115547 DQ884839 DQ872778
Hylaeus (Gnathoprosopis) amiculus (Smith, 1879)[Hyam698]
Colletidae:Hylaeinae
Australia: South Australia. 10 km E Kimba.5.i.1999
AY585128 DQ115546 DQ884838 DQ872777
Hylaeus (Macrohylaeus) alcyoneus (Erichson, 1842)[EA0129]
Colletidae:Hylaeinae
Australia: WA; Badgingarra Ntl.Prk. 11.x.2005 DQ884668 DQ884562 DQ884837 DQ768577
Hylaeus (Pseudhylaeus) aff. simplus Houston, 1993[EA0125]
Colletidae:Hylaeinae
Australia: WA; Boorabbin Ntl.Prk. 26.ix.2005 DQ884664 DQ884561 DQ884833 DQ768573
Hylaeus (Rhodohylaeus) proximus (Smith, 1879)[Hypr699]
Colletidae:Hylaeinae
Australia: South Australia. 10 km E Kimba.5.i.1999
AY585130 DQ115548 EF032906 DQ872779
Hyleoides concinna (Fabricius, 1775) [KM268] Colletidae:Hylaeinae
Australia: Queensland. South of Eukey.18.xii.2002
DQ884691 DQ884560 DQ884861 DQ768601
Meroglossa itamuca (co*ckerell, 1910) [KM271] Colletidae:Hylaeinae
Australia: Queensland. Noosa North Shore.10.i.2003
DQ884694 DQ884565 DQ884863 DQ768604
Palaeorhiza (Heterorhiza) sp. [KM275] Colletidae:Hylaeinae
Australia: Queensland. Kuranda. O4.i.2003 DQ884697 DQ884566 DQ884867 DQ768608
Palaeorhiza (Palaeorhiza) sp. [KM276] Colletidae:Hylaeinae
Australia: Queensland. Kawana Waters.10.xii.2002
DQ884698 DQ884567 DQ884868 DQ768609
Andrenopsis sp. [EA0094] Colletidae:Paracolletinae
Australia: WA; �18.4 km NE Menzies.24.ix.2005
DQ884624 DQ884529 DQ884772 DQ768513
Anthoglossa sp. [EA0115] Colletidae:Paracolletinae
Australia: WA; Boorabbin Ntl.Prk.25.ix.2005 DQ884585 DQ884492 DQ884730 DQ768474
Anthoglossa cfr. robustus (co*ckerell, 1929)[EA0116]
Colletidae:Paracolletinae
Australia: WA; Kalbarri Ntl.Prk. 08.x.2005 DQ884586 DQ884493 DQ884731 DQ768475
Baeocolletes minimus (Michener, 1965) [EA0095] Colletidae:Paracolletinae
Australia: WA; �72.6 km NE Menzies.27.ix.2005
DQ884608 DQ884514 DQ884756 DQ768497
Baeocolletes sp. [EA0096] Colletidae:Paracolletinae
Australia: WA; Eurardy Station. 09.x.2005 DQ884609 — DQ884757 DQ768498
Belopria nitidior Moure, 1956 [EA0027] Colletidae:Paracolletinae
Brazil: Paraná. 15 km S Bocaiúva do Sul.12.ix.2002
DQ884593 DQ884500 DQ884739 DQ768482
Brachyglossula communis Trucco-Aleman, 1999[EA0083]
Colletidae:Paracolletinae
Argentina: Catamarca. El Rodeo. 20.xi.2004 DQ884600 DQ884506 DQ884748 DQ768489
Callomelitta antipodes (Smith, 1853) [Cman687] Colletidae:Paracolletinae
Australia: NSW. Guyra, 74 km E. 5.xii.1999 AY585122 DQ115563 EF032907 DQ872767
Cephalocolletes isabelae Urban, 1995 [EA0012] Colletidae:Paracolletinae
Brazil: Santa Catarina. Laguna. 31.xii.2001. DQ884589 DQ884496 DQ884735 DQ768478
Cephalocolletes laticeps (Friese, 1906) [EA0086] Colletidae:Paracolletinae
Argentina: San Juan. 14 km W Media Agua.27.xi.2004
DQ884602 DQ884508 DQ884750 DQ768491
Chilicolletes delahozii (Toro, 1973) [Lesp568] Colletidae:Paracolletinae
Chile: Region IV. Llano de la Hignera. AF435392 AY227914 DQ884746 DQ872762
Colletellus aff. velutinus (co*ckerell, 1929) [EA0097] Colletidae:Paracolletinae
Australia: WA; Eurardy Stat. 09.x.2005 DQ884625 DQ884530 DQ884773 DQ768514
Edwyniana sp. [EA0071] Colletidae:Paracolletinae
Chile: Region IV. Guampulla SW Samo Alto.19.x.2001
DQ884598 DQ884504 DQ884745 DQ768487
Eulonchopria punctatissima Michener, 1963[EA0077]
Colletidae:Paracolletinae
Mexico: Jalisco. Carretera 200, km 55.02.ix.2004
DQ884599 DQ884505 DQ884747 DQ768488
Eulonchopria simplicicrus (Michener, 1989)[EA0009]
Colletidae:Paracolletinae
Brazil: Minas Gerais. Belo Horizonte. 06.v.2002 DQ884588 DQ884495 DQ884734 DQ768477
Euryglossidia sp.1 [EA0102] Colletidae:Paracolletinae
Australia: WA; Boorabbin Ntl.Prk. 25.ix.2005 DQ884611 DQ884516 DQ884759 DQ768500
Euryglossidia sp.2 [EA0103] Colletidae:Paracolletinae
Australia: WA; �6 km E Merredin 23.ix.2005 DQ884612 DQ884517 DQ884760 DQ768501
Euryglossidia sp.3 [EA0104] Colletidae:Paracolletinae
Australia: WA; Eurardy Station. 09.x.2005 DQ884613 DQ884518 DQ884761 DQ768502
Euryglossidia sp.4 [EA0151] Colletidae:Paracolletinae
Australia: WA; Badgingarra Ntl.Prk. 11.x.2005 DQ884618 DQ884523 DQ884766 DQ768507
Euryglossidia sp.5 [EA0152] Colletidae:Paracolletinae
Australia: WA; �5.8 km NE Menzies 24.ix.2005 DQ884619 DQ884524 DQ884767 DQ768508
Excolletes sp. [EA0098] Colletidae:Paracolletinae
Australia: WA; Kalbarri Ntl.Prk. 08.x.2005 DQ884626 — DQ884774 DQ768515
Glossurocolletes bilobatus (Michener, 1965)[EA0093]
Colletidae:Paracolletinae
Australia: WA; Kalbarri Ntl.Prk. 06.x.2005 DQ884623 DQ884528 DQ884771 DQ768512
Goniocolletes fimbriatinus (co*ckerell, 1910)[Lefm702]
Colletidae:Paracolletinae
Australia: Victoria. 12 km E Hattah. 6.i.1999 AY585131 DQ115554 DQ884786 DQ872763
Goniocolletes perfasciatus (co*ckerell, 1906)[Lepr704]
Colletidae:Paracolletinae
Australia: Victoria. 12 km E Hattah. 9.i.1999 AY585134 DQ115557 DQ884787 DQ872764
Halictanthrena malpighiacearum Ducke, 1907[EA0088]
Colletidae:Paracolletinae
Brazil: Minas Gerais. Serra do Salitre. 11.xi.2004 DQ884604 DQ884510 DQ884752 DQ768493
Hexantheda missionica Ogloblin, 1948 [EA0124] Colletidae:Paracolletinae
Brazil: Minas Gerais. Brumadinho. 12.i.2001 DQ884615 DQ884520 DQ884763 DQ768504
Hoplocolletes ventralis (Friese, 1924) [EA0021] Colletidae:Paracolletinae
Brazil: Minas Gerais. Florestal. 03.xii.2001 DQ884590 DQ884497 DQ884736 DQ768479
Kylopasiphae pruinosa (Michener, 1989) [EA0084] Colletidae:Paracolletinae
Argentina: Mendoza. 37 km SSE Uspallata. 29–30.xi.2004
DQ884601 DQ884507 DQ884749 DQ768490
294 E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309
Table 1 (continued)
Species [voucher code] Classification Collecting data EF-1a Opsin Wingless 28S rRNA
Lamprocolletes chalybeatus (Erichson, 1851)[EA0099]
Colletidae:Paracolletinae
Australia: WA. Boorabbin Ntl.Prk. 25.ix.2005 DQ884627 — DQ884775 DQ768516
Leioproctus conospermi Houston, 1989 [EA0110] Colletidae:Paracolletinae
Australia: WA; Kalbarri Ntl.Prk. 08.x.2005 DQ884632 DQ884534 DQ884780 DQ768521
Leioproctus irroratus (Smith, 1853) [Leir705] Colletidae:Paracolletinae
Australia: NSW. Hilltop. 2.xii.1999 AY585132 DQ115555 DQ884788 DQ872765
Leioproctus lanceolatus Houston, 1990 [EA0111] Colletidae:Paracolletinae
Australia: WA; �90 km E Leonora. 28.ix.2005 DQ884633 DQ884535 DQ884781 DQ768522
Leioproctus megachalcoides Michener, 1965[EA0112]
Colletidae:Paracolletinae
Australia: WA Eurardy Station. 09.x.2005 DQ884634 DQ884536 DQ884782 DQ768523
Leioproctus pappus Houston, 1989 [EA0109] Colletidae:Paracolletinae
Australia: WA; Badgingarra Ntl.Prk. 11.x.2005 DQ884631 — DQ884779 DQ768520
Leioproctus platycephalus (co*ckerell, 1912)[EA0113]
Colletidae:Paracolletinae
Australia: WA; North Tarin Rock Nat. Res.02.x.2005
DQ884635 DQ884537 DQ884783 DQ768524
Leioproctus plumosus (Smith, 1853) [Lepl706] Colletidae:Paracolletinae
Australia: Victoria. Torquay. 19.xi.1999 AY585133 DQ115556 DQ884789 DQ872766
Paracolletinae sp. [EA0123] Colletidae:Paracolletinae
Australia: WA. Badgingarra Ntl.Prk. 11.x.2005 DQ884637 DQ884538 DQ884785 DQ768526
Lonchopria (Biglossidia) robertsi Michener, 1989[EA0080]
Colletidae:Paracolletinae
Argentina: Mendoza. 6 km SSE Uspallata. 29–30.xi.2004
DQ884622 DQ884527 DQ884770 DQ768511
Lonchopria (Biglossidia) sp. [EA0070] Colletidae:Paracolletinae
Chile: Region I. Zapahuira (near Putre). Iv.2004 DQ884621 DQ884526 DQ884769 DQ768510
Lonchopria (Lonchopria) similis (Friese, 1906)[EA0069]
Colletidae:Paracolletinae
Chile: Region II. Pqe. Nacional Llanos de Challe.13.x.2000
DQ884620 DQ884525 DQ884768 DQ768509
Neopasiphae mirabilis Perkins, 1912 [EA0100] Colletidae:Paracolletinae
Australia: WA. �18.4 km NE Menzies.27.ix.2005
DQ884628 DQ884531 DQ884776 DQ768517
Niltonia virgilii Moure, 1964 [EA0087] Colletidae:Paracolletinae
Brazil: Santa Catarina. Maracajá. 04.x.2004 DQ884603 DQ884509 DQ884751 DQ768492
Nomiocolletes jenseni Friese, 1906 [EA0090] Colletidae:Paracolletinae
Argentina: Mendoza. 18 km SE Potrerillos. 29–30.xi.2004
DQ884606 DQ884512 DQ884754 DQ768495
Odontocolletes aff. asper (Maynard, 1997) [EA0105] Colletidae:Paracolletinae
Australia: WA. Eurardy Station. 09.x.2005 DQ884629 DQ884532 DQ884777 DQ768518
Odontocolletes pachyodontus (co*ckerell, 1915)[EA0106]
Colletidae:Paracolletinae
Australia: WA; �13 km NNE Eurardy Station.09.x.2005
DQ884630 DQ884533 DQ884778 DQ768519
Paracolletes cfr. crassipes Smith, 1853 [EA0019] Colletitidae:Paracolletinae
Australia: Queensland, Stanthorpe. 16.xii.2002 DQ884584 DQ884491 DQ884729 DQ768473
Perditomorpha laena (Vachal, 1909) [EA0024] Colletidae:Paracolletinae
Brazil: Minas Gerais. Santana do Riacho. 20/ii/2001.
DQ884592 DQ884499 DQ884738 DQ768481
Perditomorpha leucostoma (co*ckerell, 1917)[EA0091]
Colletidae:Paracolletinae
Argentina: Catamarca. Aconquija. 18.xi.2004 DQ884607 DQ884513 DQ884755 DQ768496
Perditomorpha neotropica (Friese, 1908) [EA0147] Colletidae:Paracolletinae
Argentina: Tucumán. 19 km SE Amaicha delValle. 17.ii.2003
DQ884616 DQ884521 DQ884764 DQ768505
Perditomorpha rufiventris (Spinola, 1851) [EA0046] Colletidae:Paracolletinae
Chile: Region IV. El Tofo. 23.x.2002 DQ884595 DQ884502 DQ884742 DQ768484
Perditomorpha stilborhina (Moure, 1954) [EA0107] Colletidae:Paracolletinae
Argentina: Prov. Tucumán. Amaicha del Valle.26.x.2004
DQ884614 DQ884519 DQ884762 DQ768503
Perditomorpha sp. [EA0068] Colletidae:Paracolletinae
Argentina: Chubut. 8 km N Sarmiento Hwy.24.26–27.xi.2003
DQ884597 DQ884503 DQ884744 DQ768486
Phenacolletes mimus co*ckerell, 1905 [EA0101] Colletidae:Paracolletinae
Australia: WA; �12 km SSE Dongara. 10.x.2005 DQ884610 DQ884515 DQ884758 DQ768499
Protomorpha aff. alloeopus (Maynard, 1991)[EA0114]
Colletidae:Paracolletinae
Australia: WA; Kalbarri Ntl.Prk. 08.x.2005 DQ884636 — DQ884784 DQ768525
Reedapis bathycyanea (Toro, 1973) [Lesp851] Colletidae:Paracolletinae
Chile: Region II. Santa Juana, E of Vallenar AY585141 DQ115553 DQ884741 DQ872761
Spinolapis caerulescens (Spinola, 1851) [EA0034] Colletidae:Paracolletinae
Chile: Region IV. Parque Nacional Talinay10.x.2001
DQ884594 DQ884501 DQ884740 DQ768483
Spinolapis sp. [EA0089] Colletidae:Paracolletinae
Argentina: Mendoza. 37 km SSE Uspallata. 29–30.xi.2004
DQ884605 DQ884511 DQ884753 DQ768494
Tetraglossula anthracina (Michener, 1989)[EA0023]
Colletidae:Paracolletinae
Brazil: Minas Gerais. Santana do Riacho.15.iv.2001
DQ884591 DQ884498 DQ884737 DQ768480
Trichocolletes (Trichocolletes) aff. venustus (Smith,1862) [EA0117]
Colletidae:Paracolletinae
Australia: WA; Boorabbin Ntl.Prk. 25.ix.2005 DQ884587 DQ884494 DQ884732 DQ768476
Trichocolletes (Trichocolletes) sp. [Trsp708] Colletidae:Paracolletinae
Australia: NSW. 53 km S Oberon. 30.xi.1999 AY585139 DQ115562 DQ884733 DQ872760
Scrapter algoensis (Friese, 1925) [Scal899] Colletidae:Scrapterinae
South Africa: NCP. 90 km ENE Sprinbok.10.ix.2001
EF032901 EF032904 DQ884812 DQ872771
Scrapter erubescens (Friese, 1925) [Scer901] Colletidae:Scrapterinae
South Africa: WCP. Pakhuis pass. Sept. 8.ix.2001 AY585135 DQ115558 DQ884813 DQ872772
Scrapter heterodoxus (co*ckerell, 1921) [Scht903] Colletidae:Scrapterinae
South Africa: WCP. 31 km S Clanwillian.7.ix.2001
AY585136 DQ115559 DQ884814 DQ872773
Scrapter niger Lepeletier and Serville, 1825[Scng905]
Colletidae:Scrapterinae
South Africa: WCP. 21 km N Hermanus.28.ix.2001
AY585137 DQ115560 DQ884815 DQ872774
Scrapter ruficornis (co*ckerell, 1916) [Scrc937] Colletidae:Scrapterinae
South Africa: WCP. Kunje Farm, near Citrusdal.23.ix.2001
AY585138 DQ115561 DQ884816 DQ872775
(continued on next page)
E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309 295
Table 1 (continued)
Species [voucher code] Classification Collecting data EF-1a Opsin Wingless 28S rRNA
Chilicola (Anoediscelis) herbsti (Friese, 1906)[EA0140]
Colletidae:Xeromelissinae
Chile: Region IV. Liman, Chañar. 04.ix.2004 DQ884663 DQ884559 DQ884831 DQ768572
Chilicola (Oediscelis) vicugna Toro and Moldenke,1979 [EA0136]
Colletidae:Xeromelissinae
Chile: Region IV. Elqui, Pangue. 11–30.ix.2004 DQ884661 DQ884557 DQ884828 DQ768569
Chilicola (Pseudiscelis) rostrata (Friese, 1906)[EA0137]
Colletidae:Xeromelissinae
Argentina: Tucumán. 19 km SE Amaicha delValle. 17.ii.2003
DQ884662 DQ884558 DQ884829 DQ768570
Geodiscelis longiceps Packer, 2005 [EA0049] Colletidae:Xeromelissinae
Argentina: Tucumán. 19 km SE Amaicha delValle. 17.ii.2003
DQ884655 DQ884551 DQ884817 DQ768544
Xenochilicola mamigna Toro and Moldenke, 1979[EA0055]
Colletidae:Xeromelissinae
Chile: Region II. Aguas Blancas (SSE San PedroAtacama)
DQ884660 DQ884556 DQ884827 DQ768558
Xeromelissa australis (Toro and Moldenke, 1979)[EA0051]
Colletidae:Xeromelissinae
Chile: Region II. Panamerican Hwy., km 1005,NE Chanaral.
DQ884656 DQ884552 DQ884818 DQ768545
Xeromelissa irwini (Toro and Moldenke, 1979)[EA0053]
Colletidae:Xeromelissinae
Chile: Region I. 83.5 km ESE Pozo Almonte. 8–20.iv.2004
DQ884658 DQ884554 DQ884821 DQ768547
Xeromelissa nortina (Toro and Moldenke, 1979)[EA0052]
Colletidae:Xeromelissinae
Argentina: Santa Cruz. 20 km E Los Antiguos.17.xi.2003
DQ884657 DQ884553 DQ884820 DQ768546
Xeromelissa rozeni (Toro and Moldenke, 1979)[Chrz857]
Colletidae:Xeromelissinae
Argentina: Santa Cruz. 25 km E Los Antiguos22.xi.2003
AY585120 DQ115549 DQ884826 DQ872776
Xeromelissa sp. [EA0138] Colletidae:Xeromelissinae
Chile: Region I. 83.5 km ESE Pozo Almonte.20.iv.2004
DQ884659 DQ884555 DQ884824 DQ768554
Alocandrena porteri Michener, 1986 [Alpo49] Andrenidae:Andreninae
Peru: Lima Dept. St. Bartholome. 21.x.1997 AY585099 DQ113659 — AY654473
Andrena brooksi Larkin, 2004 [Ansp643] Andrenidae:Andreninae
USA: NM. Hidalgo Co., 20 mi S Animas.17.ix.1999
AY230129 AF344618 AY222551 AY654474
Orphana wagenknechti Rozen, 1971 [Orph56] Andrenidae:Andreninae
Chile: Region IV. 7 km S Pisco Elqui DQ884568 DQ884476 DQ884709 DQ872755
Protoxaea gloriosa (Fox, 1893) [Pxgl226] Andrenidae:Oxaeinae
USA: AZ. Cochise Co., Portal AY585106 DQ113658 — AY654480
Nolanomelissa toroi Rozen, 2003 [Noto74] Andrenidae:Panurginae
Chile: Region II. 4 km N Domeyko DQ884569 DQ884477 DQ884710 DQ872756
Calliopsis (Nomadopsis) fracta (Rozen, 1952)[Cafr515]
Andrenidae:Panurginae
USA: CA. Santa Cruz Co., San Antonio junction.28.v.1999
AY585101 AF344587 — AY654476
Panurgus (Panurgus) calcaratus (Scopoli, 1763)[Pnca514]
Andrenidae:Panurginae
Italy: Rome. 07.vi.1998 AY585105 AF344612 — AY654479
Melitturga (Melitturga) clavicornis (Lattreille, 1806)[Mtcl959]
Andrenidae:Panurginae
France: Herault. Causse de la Selle. 17.vi.2002 AY585104 DQ116703 — AY654478
Agapostemon tyleri (co*ckerell, 1917) [Agty230] Halictidae:Halictinae
USA: AZ. Cochise Co., Portal AF140320 AY227940 AY222577 AY654506
Augochlorella pomoniella (co*ckerell, 1915)[Aupo592]
Halictidae:Halictinae
USA: CA. Inyo Co., Big Pine. 15.vi.1999 AF435373 AY227935 AY222572 AY654507
Mexalictus arizonensis Eickwort, 1978 [Mxaz98] Halictidae:Halictinae
USA: AZ. Santa Cruz Co. AF140322 AY227959 AY222595 —
Halictus (Halictus) rubicundus (Christ, 1791)[Haru32]
Halictidae:Halictinae
USA: MT. Missoula Co., Missoula AF140335 DQ116674 AY222592 AY654510
Dieunomia (Epinomia) nevadensis (Cresson, 1874)[None207]
Halictidae:Nomiinae
USA: AZ. Cochise Co., 1 mi E. Apache, 22.ix.1999 AF435396 AY227931 AY222568 AY654512
Conanthalictus wilmattae co*ckerell, 1936[Cowi351]
Halictidae:Rophitinae
USA: CA. Riverside Co., 10 mi S Palm Desert.15.iii.1997
AF435378 AY227934 AY222553 AY654511
Dufourea mulleri (co*ckerell, 1898) [Dumu233] Halictidae:Rophitinae
USA: Michigan AF435383 AY227918 AY222555 AY654509
Penapis penai Michener, 1965 [Pnpe572] Halictidae:Rophitinae
Chile: Region II. N Vallenar AF435401 AY227921 AY222558 AY654513
Rophites (Rophites) algirus Perez, 1895 [Roal968] Halictidae:Rophitinae
France: Var. 5 km S Entrecasteaux. 14.vi.2002 AY585144 DQ116675 — AY654515
Systropha curvicornis (Scopoli, 1770) [Sycu350] Halictidae:Rophitinae
Austria: Vienna AF435411 AY227925 AY222562 AY654516
Ctenocolletes nigricans Houston, 1985 [EA0120] Stenotritidae Australia: WA; �18.4 km NNE Eurardy Station.05.x.2005
DQ884570 DQ884478 DQ884711 DQ768459
Ctenocolletes rufescens Houston, 1983 [EA0121] Stenotritidae Australia: WA. Boorabbin Ntl.Prk. 25.ix.2005 DQ884571 DQ884479 DQ884712 DQ768460Ctenocolletes smaragdinus (Smith, 1868) [EA0122] Stenotritidae Australia: WA. Boorabbin Ntl.Prk. 25.ix.2005 DQ884572 DQ884480 DQ884713 DQ768461Stenotritus sp. [Stsp1015] Stenotritidae Australia: WA. 23 km SW Coorow. 17.xi.1997 DQ141115 DQ115564 DQ884714 DQ872757
296 E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309
nator chemistry and AmpliTaq-FS DNA Polymerase at Cornell Uni-versity Life Sciences Core Laboratories Center.
GenBank Accession Nos. (Table 1) starting with ‘‘DQ7”, ‘‘DQ8”,and ‘‘EF0” are novel sequences. The remaining sequences were gen-erated by B.N. Danforth and co-workers (Danforth et al., 2004,2006a,b). The wingless sequence of Halictus quadricinctus (Fabricius)was used to represent that of Halictus rubicundus (Christ).
2.5. Alignment
Alignments for the individual gene data matrices were gener-ated using similarity calculated at the nucleotide level (‘‘�n’’)
with DIALIGN 2.2 (Morgenstern, 1999) and with the ClustalW(Lasergene DNA Star software package). The resulting alignmentswere then corrected manually for obvious alignment errors usingMacClade v 4.08 OSX (Maddison and Maddison, 2005) and WinC-lada 1.00.08 (Nixon, 2002). Regions of 28S rRNA and introns ofEF-1a where alignment was overly ambiguous were excludedfrom the phylogenetic analyses. Despite losing some information,the removal of problematic regions of the alignments potentiallyincreases the actual phylogenetic signal (see discussion by Talave-ra and Castresana, 2007). For the protein-coding genes, honey-bee(Apis mellifera Linnaeus) sequences were used to establish readingframes and intron/exon boundaries. Introns of the three protein-
(b) long wavelength (green) rhodopsin
620 bp 275 bp 250 bp
Intron 1(~210 bp)
HaF2For1 For3
F2Rev1 Cho10
(a) elongation factor-1 alpha (F2 copy)
1 2 3 6 7a 7b 8 10 1254
A-28S-For(3118F)
Bel28S-For(D2-3665F)
Mar28S-Rev(D3-4283R)
28SD4-28S(D5-4749R)
(c) wingless
(d) large subunit ribosomal RNA: 28S rRNA
length in bees = ~ 1500 bp
10008006004002001
8006004002001
120010008006004002001 1400 1600
Intron 2(~230-500 bp)
Intron 1(~70-480 bp)
Intron 2(~60-1080 bp)
Intron 3(~60-750 bp)
Intron(~100-1400 bp)
OpsRevOpsRev4/4a
Ops-Rev1b, (mod)
Ops-For4, 5Ops-For
Ops-For3, 3b, 3(mod)
210 bp 260 bp 170 bp 70 bp
Bee-wgFor2
175 bp 670 bp
Lep-Wg2a-Rev
Wg-Collet-For
Fig. 2. Maps of the gene loci sampled for this study. Primer sites (see Table 2 for a complete list of primer sequences), introns, and exons are indicated. Introns, when variable inlength, are represented on their lower limit of the range; length variation reported in the figure is observed in the sample of colletid bees included in this study; (a) elongationfactor-1a, F2 copy (adapted from Danforth et al., 2004), numbering based on Apis mellifera coding sequence (Walldorf and Hoverman, 1990, GenBank Accession AF015267); (b)long wavelength rhodopsin (adapted from Danforth et al., 2004), numbering based on A. mellifera coding sequence (Chang et al., 1996; GenBank Accession U26026); (c)wingless, numbering based on Drosophila melanogaster coding sequence (Uzvölgyi et al., 1988; GenBank Accession J03650); (d) 28S rRNA (adapted from Caterino et al., 2000),nucleotide numbering and region names based on D. melanogaster ribosomal genes (Hanco*ck et al., 1988; Linares et al., 1991; Tautz et al., 1988, GenBank Accession M21017).
E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309 297
coding genes sampled were initially aligned for all taxa. Introns ofopsin and wingless were subsequently excluded from the analysesbecause length variation and sequence variability observed inthese regions made them unalignable. Both EF-1a introns couldonly be aligned for colletid and stenotritid sequences (i.e., intronsfrom all other outgroup terminals were discarded to allow for theremaining to be alignable). The second EF-1a intron is more var-iable than intron 1 (Fig. 2), so was aligned separately for two setsof species: (1) representatives of Stenotritidae, Diphaglossinae,and Paracolletinae (excluding Callomelitta antipodes); and (2) ofColletinae s.str., Euryglossinae, Hylaeinae, Scrapterinae, Xerome-lissinae, plus Callomelitta antipodes. Selection of these blocks oftaxa was made based on groupings found in preliminary analysesof exon data plus the 28S sequences, and on sequence similarityfound among introns of the terminals. Although unusual, thisprocedure will not introduce phylogenetic artifacts. We are sim-
ply attempting to extract the maximum amount of informationfrom the largest regions of the non-coding partitions of the dataset.
In cases where multiple sequences were available for the samespecies, the sequences were merged after being resolved in thesame clade in preliminary analyses and resulting partial polymor-phisms were kept as such. Individual gene data sets were concate-nated with WinClada.
The combined data set in Nexus-format used in phylogeneticanalyses is available for download as Supplementary information.
2.6. Phylogenetic analyses
Bayesian phylogenetic inference was used to estimate the treetopology. Metropolis-coupled MCMC, as implemented in the serialversion of MrBayes 3.1.2 (Altekar et al., 2004; Huelsenbeck and
Table 2Primer sequences for EF-1a (F2 copy), opsin, wingless, and 28S rRNA used for PCR assays of bees.
Locus Primer Sequence Position Reference
EF-1aa HaF2For1 50-GGG YAA AGG WTC CTT CAA RTA TGC-30 511 Danforth et al. (1999)For3rho 50-GGY GAC AAY GTT GTT TTY AAY G-30 1496 Danforth et al. (1999)F2-rev1 50-A ATC AGC AGC ACC TTT AGG TGG-30 1600 Danforth et al. (1999)Cho10-Rev(mod) 50-AC RGC VAC KGT YTG HCK CAT GTC-30 1887 Danforth et al. (1999)
Opsinb Opsin-For [=LWRhFor)] 50-AAT TGC TAT TAY GAR ACN TGG GT-30 398 Mardulyn and Cameron (1999)
Opsin-For3 [=LWRhFor3] 50-AGA TAC AAC GTR ATC GTS AAR GGT-30 512 Danforth et al. (2004)
Opsin-For3(mod) 50-TTC GAY AGA TAC AAC GTR ATC GTN AAR GG-30 506 Danforth (unpublished)
Opsin-For3b 50-AGA TAC AAC GTR ATY GTN AAR GGT-30 512 Almeida (unpublished)Opsin-For4 50-GAG AAR AAY ATG CGB GAR CAA GC-30 803 Danforth et al. (2004)Opsin-For5 50-ATG CGN GAR CAR GCN AAR AAR ATG AA-30 812 Danforth (unpublished)Opsin-Rev (=LWRhRev) 50-ATA TGG AGT CCA NGC CAT RAA CCA-30 946 Mardulyn and Cameron (1999)
Opsin-Rev1b 50-RTA YGG RGT CCA NGC CAT RAA CCA-30 946 Almeida (unpublished)Opsin-Rev(mod) 50-ATA NGG NGT CCA NGC CAT GAA CCA-30 946 Danforth (unpublished)Opsin-Rev4 50-GGT GGT GGT RCC GGA RAC GGT G-30 1147 Danforth et al. (2004)Opsin-Rev4b 50-GGT RCC GGA RAC GGT GGA DGT NGC RTC-30 1147 Danforth (unpublished)
Winglessc Bee-wg-For2 50-GGC AGC ATY CAG TCS TGY TCC TGC GA-30 445 Sipes (unpublished)Wg-Collet-For 50-CAC GTG TCB TCB GRG ATG MGR SAG GA-30 670 Almeida (unpublished)Lep-Wg2a-Rev 50-ACT ICG CAR CAC CAR TGG AAT GTR CA-30 1336 Brower and DeSalle (1998)
28S rRNAd A-28S-For 50-CCC CCT GAA TTT AAG CAT AT-30 3318 Ward and Brady (2003)Bel28S-For (D2-3665F) 50-AGA GAG AGT TCA AGA GTA CG TG-30 3665 Belshaw and Quicke (1997)Mar28S-Rev (D3-4283R) 50-TAG TTC ACC ATC TTT CGG GTC CC-30 4283 Mardulyn and Whitfield (1999)28SD4-Rev (D5-4749R) 50-GTT ACA CAC TCC TTA GCG GA-30 4749 Danforth et al. (2006a)
a PCR conditions. HaF2For1/F2-rev1: 94 �C for 1 min, 48–52 �C for 1 min, 72 �C for 1.5 min (35 cycles); For3rho/Cho10-Rev(mod): 94 �C for1 min, 54–56 �C for 1 min, 72 �Cfor 1 min (35 cycles). Positions based on the 50 end of the primer in Apis mellifera (Walldorf and Hoverman, 1990; GenBank Accession No. AF015267).
b PCR conditions. Opsin-For/Opsin-Rev: 94 �C for 1 min, 50–54 �C for 1 min, 72 �C for 1.5 min (35 cycles); Opsin-For3/Opsin-Rev: 94 �C for 1 min, 52–54 �C for 1 min, 72 �Cfor 1 min (35 cycles); Opsin-For3(mod)/Opsin-Rev(mod): 94 �C for 1 min, 52–54 �C for 1 min, 72 �C for 1 min (35 cycles); Opsin-For3 b/Opsin-Rev1 b: 94 �C for 1 min, 50–54 �C for 1 min, 72 �C for 1 min (35 cycles); Opsin-For4/Opsin-Rev4: 94 �C for 1 min, 55–58 �C for 1 min, 72 �C for 1 min (35 cycles); Opsin-For5/Opsin Rev4b: 94 �C for 1 min,55 �C for 1 min, 72 �C for 1 min (35 cycles). Positions based on the 50 end of the primer in Apis mellifera (Chang et al., 1996; GenBank Accession No. U26026).
c PCR conditions. Bee-wg-For2/Lep-wg2a-Rev: 94 �C for 1 min, 54–58 �C for 1 min, 72 �C for 1.5 min (35 cycles); Wg-Collet-For/Lep-wg2a-Rev: 94 �C for 1 min, 56 �C for1 min, 72 �C for 1 min (35 cycles). Positions based on the 50 end of the primer in Drosophila melanogaster (Uzvölgyi et al., 1988; GenBank Accession No. J03650).
d PCR conditions. A-28S-For Mar28S-Rev: 94 �C for 1 min, 58 �C for 1 min, 72 �C for 1.5 min (35 cycles); Bel28S-For/28SD4-Rev: 94 �C for 1 min, 58 �C for 1 min, 72 �C for1.5 min (35 cycles). Positions based on the 50 end of the primer in Drosophila melanogaster (Tautz et al., 1988; Linares et al., 1991; GenBank Accession No. M21017).Numbering corresponds to the expansion regions as proposed by Hanco*ck et al. (1988).
298 E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309
Ronquist, 2005), was used to estimate the posterior probability dis-tribution. The gamma distribution of rate variation across sites wasapproximated by a discrete distribution with four rate categories,each category being represented by its mean rate. All chains, includ-ing coupled chains in the same run, were started from different, ran-domly chosen trees. Searches were run for 2 � 106 generations ontwo sets of 10 chains each, through the Computational Biology Ser-vice Unit at the Cornell Theory Center (http://cbsu.tc.cornell.edu/).Convergence was assessed by the standard deviation of split fre-quencies of the two independent MrBayes runs, by the convergencediagnostic for individual parameters employing potential scalereduction factor (Gelman and Rubin, 1992, uncorrected), and bythe achievement of stationarity of the log likelihood values of thecold chain. Trees were saved every 100 generations. The initial1000–3000 trees were discarded after examining the variation inlog likelihood scores over time. Models for the five partitions of theconcatenated data set (EF-1a [exons], EF-1a [introns], 28S rRNA, op-sin [exons], and wg [exons]) were individually selected and imple-mented. Partitioned analyses were run with MrBayes applyingmodels selected by different methods (see below), in order to testthe effect of various levels of model complexity.
The best-fit model of evolution for each of the five partitionswas statistically tested. Model selection is one of the most contro-versial fields of statistics (Burnham and Anderson, 2002), and thisis not different in the field of phylogenetics (e.g., Posada andCrandall, 2001; Nylander et al., 2004; Posada and Buckley, 2004;Sullivan and Joyce, 2005). The four approaches most commonlyused for model selection in phylogenetics are hierarchical likeli-hood ratio test (hLRT), Akaike information content (AIC), Bayesianinformation content (BIC), and decision theory (DT) (Posada andBuckley, 2004; Sullivan and Joyce, 2005). Four programs were used
to shed light on the best-fit model(s) for the data: (1) DT-ModSel(Minin et al., 2003—DT model selection); (2) ModelTest 3.7 (Posadaand Crandall, 1998—hLRT, AIC, BIC; a set to 0.05); (3) MrModelTest(Nylander, 2004b—hLRT, AIC); and (4) MrAIC.pl 2.2 (Nylander,2004a—AIC and BIC). In addition to statistical tests for model selec-tion, highly complex models were tested in MrBayes for the com-bined data matrix, to check the effects of higher model-realismin the phylogenetic results: GTR + I + C and GTR + SSR for each par-tition, with all parameters unlinked across partitions.
For comparison with the Bayesian analysis, the data set wassubjected to maximum likelihood analysis and equal weights par-simony analyses. Maximum likelihood searches were done usingGarli v0.951 (Zwickl, 2006). Garli does not allow for data set parti-tioning, so the complete concatenated matrix was analyzed undera GTR + I + C model as a means to deal with the heterogeneitywithin the data. When analyzing individual genes, simpler modelswere used when that appeared to be a reasonable alternative, butvarious levels of complexity were tested. The termination condi-tion was set to 1 � 105 generations without significantly betterlikelihood scoring topology being found (‘‘genthreshfortopo-term = 100,000”); remaining settings were left unchanged fromthe defaults. Non-parametric bootstrap proportions were calcu-lated based on 1000 pseudo-replicates generated with Garli. Thefrequency of occurrence of each group present in the most likelytree topology was calculated using WinClada.
Parsimony analyses were conducted in PAUP� v4.0b10 (Swof-ford, 2002) using heuristic searches with tree bisection–reconnec-tion (TBR), 1000 random-taxon-addition replicates holding 50trees per replicate, and treating sequence indels as missing data.Branch support was assessed with 1000 bootstrap pseudo-repli-cates (Felsenstein, 1985). Each resampled matrix was searched
Table 3Overview of the partition and combined data sets.
Number ofcharacters
Informativecharacters
Totalinformation
Information/numberof characters
EF-1a (exons 1151 427 7966 6.92EF-1a (introns) 1425 601 6174 4.33Opsin (exons) 707 324 5824 8.24wg (exons) 692 218 3862 5.5828S rRNA 1523 333 3352 2.32
Combined 5498 1903 27358 4.98
E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309 299
100 times and consensus trees found at each iteration were savedand used to calculate node support (percentage count) for eachclade present in the strict consensus of the most parsimonioustrees using WinClada.
2.7. Rate of substitution and parameters of molecular evolution
MrBayes was also used to explore some properties of the molec-ular data sets. Summarized parameters after discarding the initialtrees were used as estimates of base frequencies, rates of substitu-tion, and shape parameter (a) of the gamma distribution. Datawere analyzed with GTR + C models in which all parameters wereunlinked across partitions. This allowed base composition of eachgene and within partition rate variation (a) to be estimated. In or-der to estimate the rates of substitution, the data set was analyzedwith a site-specific rates model (GTR + SSR) with rate categoriescorresponding to each codon position within the protein-codinggene exons, to the intron of EF-1a, and to 28S rRNA.
2.8. Topological congruence between partition trees and the combinedtree
The phylogenetic signal of each partition was assessed by com-paring the Bayesian topology obtained for each gene analyzedalong with the topology obtained based on an analysis of all fivepartitions combined. Nodes present in the tree topology withinthe ingroup resulting from the five partitioned analyses were com-pared in a pair-wise manner to the combined tree and classified aseither congruent or not. A congruent node was one underpinning amonophyletic group of species found in the analysis of the com-bined data set or not contradicted by the results of the latter. ‘Con-gruent topological information’ (CTI) was a measure used here toapproximate concordance between the partitions and the com-bined tree topologies. CTI was calculated as the ratio betweennumber of congruent nodes and maximum number of resolvednodes for a given number of terminals.
2.9. Hypothesis testing
In order to compare alternative tree topologies, Kishino–Hase-gawa and Shimodara–Hasegawa tests (Kishino and Hasegawa,1989; Shimodara and Hasegawa, 1999) were performed with1000 bootstrap replicates using PAUP�. Likelihood was estimatedwith the RELL (resampling of estimated log-likelihood) method(Kishino et al., 1990), as an approximation of a computationallyintensive bootstrap (Felsenstein, 2004). Significance of likelihoodscore differences was assessed in a pair-wise manner using aone-tailed test, as suggested by Felsenstein (2004, 369). Addition-ally, a Bayesian approach to hypothesis testing was used for com-parisons among alternative topologies. Constrained tree topologieswere generated in MrBayes under a fixed model and their har-monic means were compared (Kass and Raftery, 1995; Nylanderet al., 2004). From the harmonic means, 2loge(B10) is calculatedby doubling the difference in the means of two topologies beingcompared and this quantity can be interpreted with the aid ofthe table from Kass and Raftery (1995, 277; a useful discussionabout the application of Bayes factors specifically to phylogeneticquestions is given by Nylander et al., 2004).
3. Results
3.1. Molecular data sets
The final combined data matrix contains 144 taxa and 5498aligned base pairs. Among those, 1903 are parsimony-informativecharacters. Information of a character is a quantity defined as its
maximum number of steps minus its minimum number of steps(denominator for the retention index [Farris, 1989]). The matrix’sinformation is simply the sum of the information of each of itscomponent characters. The information of the combined matrixwas equal to 27,358 steps (calculated with WinClada). The result-ing concatenated data matrix includes the exons of the three pro-tein-coding genes (EF-1a, opsin, and wg), both introns of EF-1a,and most of the sequences of 28S rRNA. After the 28S sequenceswere aligned, regions where alignment was judged to be ambigu-ous were deactivated. The total of excluded 28S aligned characterswas 371 steps, representing an information loss of 1928 steps. Thetotal number of characters and of informative characters per gene,and the information content of each partition are presented in Ta-ble 3.
3.2. Model selection and comparisons among genes
A summary of the results found with the different model selec-tion programs is presented in Table 4. The most complex modeltested by those programs, GTR + I + C was indicated as the mostappropriate one for each locus by at least one model selectionstrategy. Except for wg, all other genes had a model indicated tobe superior by some of the model selection methods, which wassimpler than GTR + I + C and implemented in MrBayes. The com-bined and unpartitioned data set was tested for one overall best-fit model and it was invariably GTR + I + C. MrBayes allows modelpartitioning and this is a desirable feature for allowing for a morerealistic phylogenetic approach (Nylander et al., 2004). Empiricaltests of evaluating the effects of using models with variable levelsof complexity to the Bayesian phylogenetic analyses (partitionedand combined) did not yield significantly different results; topolo-gies found for the data sets tended to be largely resilient to varia-tion in model complexity, with slight changes being observed inthe posterior probability values. All phylogenetic results shownare based on the simplest models considered adequate for the databy the model selection strategies tested.
The result of a characterization of the gene data is presented inFig. 3. Wingless and EF-1a (both exons and introns) present a devi-ation from equal base composition more noticeable than in theother two genes—the former has a slight G–C bias and the latteran A–T bias (Fig. 3a). This is reflected in the models found to bethe best-fit for each gene (Table 4). The simplest models shownto be adequate for opsin and 28S rRNA have equal base composi-tion (K80 and SYM, respectively), whereas wg and EF-1a requiremore complex models (Table 4).
Relative rates of substitution of each of the gene loci are shownin Fig. 3b and an approximation of how variable these rates arewithin each locus is given by the a parameter of the gamma distri-bution (Fig. 3c). High values of a (e.g., EF-1a introns) indicate rel-atively hom*ogeneous rates across sites, whereas low values of a(e.g., wingless exons) indicate high heterogeneity in among-siterate variation (Fig. 3c). Introns would be expected to show hom*og-enous rates relative to protein-coding exons because there se-
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
A C G T A C G T A C G T A C G T A C G T
opsin(exons)
wingless(exons)
EF-1a(exons)
EF-1a(introns)
28S rRNA
prop
ortio
n
0.1
0.2
0.3
0.4
0.5
0.6
0.7
alph
a
opsin
wingles
sEF
-1a (
exon
s)EF
-1a (
intro
ns)
28S
rRNA0
0.5
1
1.5
2
2.5
3
rela
tive
rate
nt1 nt2 nt3
opsin (exons)
nt1 nt2 nt3
wingless (exons)
nt1 nt2 nt3
EF-1a (exons) EF-1
a(in
trons
)28
S rR
NA
a
b c
Fig. 3. Comparisons among gene loci sampled for this study based on molecular parameters estimated by MrBayes: (a) frequencies of the nucleotide bases for each genelocus; (b) relative rates of substitution; (c) a parameter of the gamma distribution of rates.
Table 4Models for each gene as selected by different computer programs. Models in bold and italicized are implemented in MrBayes 3.1.2.
Gene locus Program: model selection test Model [number of free parameters]*
28S rRNA DT_ModSel; modeltest: BIC TVMef+I+C [4]MrAIC: AIC, AICc, BIC SYM+I+C [5]modeltest: hLRTs TrN+I+C [5]modeltest: AIC; MrModelTest: hLRTs, AIC GTR+I+C [8]
EF-1a (exons) DT_ModSel; modeltest: BIC; MrAIC: AIC*, AICc*, BIC* HKY+I+C [4]modeltest: hLRTs, AIC TrN+I+C [5]MrModelTest: hLRTs, AIC GTR+I+C [8]
EF-1a (introns) DT_ModSel; modeltest: AIC, BIC HKY+I+C [4]modeltest: hLRTs TVM+I+C [7]MrModelTest: hLRTs, AIC; MrAIC: AIC, AICc, BIC GTR+I+C [8]
Opsin (exons) DT_ModSel; modeltest: AIC, BIC K80+I+C [1]modeltest: hLRTs TrN+I+C [5]MrModelTest: hLRTs, AIC; MrAIC: AIC, AICc, BIC GTR+I+C [8]
wingless (exons) DT_ModSel; modeltest: AIC, BIC TrN+I+C [5]modeltest: hLRTs; MrModelTest: hLRTs, AIC; MrAIC: AIC*, AICc*, BIC* GTR+I+C [8]
* Number of free parameters for the model of substitution alone, this number does not include parameters associated with the gamma distribution (C) and the proportion ofinvariant sites (I).
300 E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309
quences should be under less selection than coding regions. In rel-ative terms, rates of substitution of first and second positions ofopsin are higher and of third positions are lower than those ob-
served in EF-1a and wg (Fig. 3b). The relative rate of the intronsof EF-1a fits in the range of rates of third positions of the exonsand that of 28S rRNA is similar to 1st positions of exons (Fig. 3b).
E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309 301
3.3. Phylogenetic relationships
Results of the Bayesian phylogenetic analysis are shown in Figs.4 and 5. Overall, topologies obtained with maximum likelihoodand parsimony analyses were very similar to those of the Bayesian
100100/100
100100/100
100/100100100/100
10087/92100
100100/100
100100/95
10087/71
100100/100
95n.r./66
81n.r./12
10066/n.r.
100100/90
“Paracolletinae”(except Callomelitta and Paracolletes)
100/100100
1
87/83100
100/100100
99/99100
65/6197
100/98100
96/89100
80/71100
100/92100
100/100100
100/100100
n.r./n.r.60
100/100100
Colletidae
Stenotritidae
Diphaglossinae
Andrenidae
Halictidae
Colletinae
Euryglossinae +Scrapterinae
Hylaeinae
71/5799
Xeromelissinae
Fig. 4. Majority-rule consensus tree resulting from mixed-model Bayesian analysis of 549(K80 + I + C), exons of wingless (GTR + I + C), and 28S rRNA (SYM + I + C) from 122 specidashed lines were not recovered in the Bayesian consensus and are based on the maxprobabilities; numbers below internodes indicate non-parametric bootstrap proportionsrecovered in one of the three analyses are indicated by ‘‘n.r.”. (a) Andrenidae, HalictidaCallomelitta and Paracolletes).
result (Figs. S1–S3, Supplementary materials). Because of this topo-logical congruence among the three analytical methods, the Bayes-ian tree is used to show node support values obtained usingbootstrap proportions of parsimony and ML, as well as Bayesianposterior probabilities (Fig. 4). As expected, Bayesian posterior
Alocandrena porteriAndrena brooksiOrphana wagenknechti
Nolanomelissa toroiCalliopsis fractaMelitturga clavicornisPanurgus calcaratus
Protoxaea gloriosa
Agapostemon tyleriMexalictus arizonensisHalictus rubicundusAugochlorella pomoniellaDieunomia nevadensisConanthalictus wilmattaeDufourea mulleriSystropha curvicornisRophites algirusPenapis penai
Ctenocolletes nigricansStenotritus sp.Ctenocolletes rufescensCtenocolletes smaragdinus
Zikanapis clypeataPtiloglossa tarsataPtiloglossa sp.Ptiloglossa thoracica
Willinkapis chalybaeaCaupolicana yarrowi
Caupolicana bicolorCaupolicana vestita
Caupolicana quadrifasciata
Cadegualina andinaCadeguala occidentalisCadeguala albopilosaDiphaglossa gayiMydrosoma fallaxMydrosoma aterrimumParacolletes cfr. crassipes
Hemicotelles ruizii
Colletes compactus
Colletes floralisColletes simulans
Colletes thoracicusColletes inaequalis
Colletes skinneriColletes pascoensis
Colletes distinctus
Colletes gilvusColletes furfuraceusColletes bicolor
Colletes seminitidusXanthocotelles sicheliRhynchocolletes mixtusCallomelitta antipodesEuhesma sp.Callohesma calliopsella
Euhesma platyrhinaEuhesma aff. crabronicaEuryglossina globulicepsXanthesma furciferaScrapter heterodoxusScrapter ruficornisScrapter nigerScrapter algoensisScrapter erubescens
Xeromelissa australisXeromelissa sp.Xeromelissa irwiniXeromelissa nortinaXeromelissa rozeniGeodiscelis longiceps
Xenochilicola mamignaChilicola (Anoediscelis) herbstiChilicola (Pseudiscelis) rostrataChilicola (Oediscelis) vicugnaHyleoides concinna
Meroglossa itamucaAmphylaeus (Agogenohylaeus) obscuricepsPalaeorhiza (Heterorhiza) sp.Palaeorhiza (Palaeorhiza) sp.
Hylaeus (Macrohylaeus) alcyoneusHylaeus (Euprosopis) disjunctusHylaeus (Euprosopis) elegansHylaeus (Rhodohylaeus) proximusHylaeus (Gnathoprosopis) amiculus
Hylaeus (Pseudhylaeus) aff. simplus
n.r.
n.r.
n.r.
100/100
99 100
100
100
99
100100
100
100/100
96/9783/72
100/100100/95
100/100
84/73
10098/98
98/97100
100/100100100/100
10098/95100
98/95100
100/100100
100/100100100/100
100
100/100100
100/100100
36/n.r. 9685/44
100100/97
100100/100
9981/80
100100/100
100100/100
00/100100
n.r. 60/4590
100/100100
100/100100
78/5899
100/100100
100/100100
94/76100
67/5083100/100
100 100/100100
n.r./n.r.74
100/10010059/61
99
53/n.r.8595/78
100
100/99100
100/100100
53/10089
100/100100
100/100100
80/7996
100/98100 100/100
100
100/100100
100/100100
85/76100
n.r./n.r.73 n.r./n.r.
76
44/n.r.
44/40
32/n.r.
8 aligned bp of the exons and introns of EF-1a (model = HKY + I + C), exons of opsines of colletid bees and 22 species of other bee families. Branches represented withimum likelihood topology. Numbers above internodes indicate Bayesian posterior
for the for the likelihood analysis (left) and parsimony analysis (right). Nodes note, Stenotritidae, and most of the colletid subfamilies; (b) ‘‘Paracolletinae” (except
Anthoglossa sp.Anthoglossa cfr. robustus
Trichocolletes (T.) aff. venustusTrichocolletes (T.) sp.
Phenacolletes mimusEuryglossidia sp.4Euryglossidia sp.2
Euryglossidia sp.3Euryglossidia sp.1
Euryglossidia sp.5
Neopasiphae mirabilisAndrenopsis sp.Paracolletinae sp.Leioproctus megachalcoidesLamprocolletes chalybeataLeioproctus lanceolatusExcolletes sp.
Colletellus aff. velutinusGlossurocolletes bilobatusLeioproctus conospermiLeioproctus pappus
Goniocolletes fimbriatinusGoniocolletes perfasciatus
Protomorpha aff. alloeopus
Odontocolletes aff. asperOdontocolletes pachyodontus
Leioproctus platycephalusLeioproctus irroratusLeioproctus plumosus
Baeocolletes minimusBaeocolletes sp.
Halictanthrena malpighiacearumPerditomorpha neotropicaCephalocolletes laticepsCephalocolletes isabelaeReedapis bathycyanea
Nomiocolletes jenseni
Eulonchopria punctatissimaEulonchopria simplicicrusHoplocolletes ventralisSpinolapis sp.Spinolapis caerulescensBelopria nitidiorEdwyniana sp.Perditomorpha leucostomaPerditomorpha laenaTetraglossula anthracinaBrachyglossula communisHexantheda missionicaPerditomorpha rufiventrisPerditomorpha stiborhinaKylopasiphae pruinosaPerditomorpha sp.Niltonia virgiliiChilicolletes delahozii
Lonchopria (Biglossa) robertsiLonchopria (Biglossa) sp.Lonchopria (Lonchopria) similis
n.r.
n.r.
100/10010042/44
78100/100
100
44/306141/n.r.
100
100/100100100/89
100
100/100100
100/100100
100/100100
100/100100 77/61
98
84/89100
100/100100
99/91100
96/91100
100/100100
91/8099
77/5797
100/100100
100/100100
100/100100
58/n.r.90
82/53100
85/8899
100/100100
100/100100
100/100100
100/100100
100/100100
100/100100
100/100100
100/100100
54/22100
66/1999
79/35100
97/92 100
99/95100
77/63100
59/n.r.100
78/8698
77/7399
83
62 98/90100
80 99/99100
61
41/3786 96/95
100
58/6686
95/88100
100/99100
63/n.r.83
100/100100
n.r./n.r.53/n.r.
n.r./n.r.
44/n.r.
99/97
30/n.r.
outgroupsDiphaglossinaeColletinaeParacolletinae
EuryglossinaeScrapterinaeXeromelissinaeHylaeinae
b
Fig. 4 (continued)
302 E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309
probabilities were higher or equal to bootstrap proportions foundwith ML and with parsimony.
The most unanticipated results were the placement of Para-colletes and Callomelitta as two lineages independent of theremaining Paracolletinae. To facilitate communication, the cladeformed by all Paracolletinae except Paracolletes and Callomelittawill be referred to as ‘‘Paracolletinae” henceforth. For the mostpart, all three analytical methods found the same arrangementof subfamilies and tribes, with one noteworthy exception: MLplaced Callomelitta antipodes as sister-group to ‘‘Paracolletinae”,whereas Bayesian and parsimony analyses placed it as sister ofColletinae s.str.
Alternative phylogenetic arrangements regarding the place-ment of Paracolletes and Callomelitta were statistically tested. Thetopology favored by both parsimony and Bayesian analyses isshown in Fig. 6a; pair-wise comparisons of this tree to five alterna-tives (Fig. 6b–f) were done using Bayesian and likelihood-basedstatistics (Table 5). The only alternative topology considered tobe significantly different by both statistical measures was the sis-
ter-group relationship (or inclusive relationship) of Paracolletesand ‘‘Paracolletinae” (Fig. 6c).
The strict consensus of the most parsimonious trees (Fig. S3)was less resolved than the Bayesian and ML topologies. Relation-ships between Colletinae plus Callomelitta, ‘‘Paracolletinae”, andthe clade formed by Euryglossinae, Hylaeinae, Scrapterinae, andXeromelissinae form an unresolved trichotomy. The relationshipbetween ‘‘Paracolletinae” and its sister-clade was strongly sup-ported by the Bayesian analysis (100% posterior probability) butonly moderately supported by ML (66% bootstrap support). All col-letid subfamilies, except Paracolletinae, were very well-supportedby the data as monophyletic. Furthermore, the grouping of Eury-glossinae, Scrapterinae, Hylaeinae, and Xeromelissinae, as well asthe sister-group relationships between [Euryglossinae + Scrapteri-nae], and [Hylaeinae + Xeromelissinae] all received fairly highbootstrap and Bayesian posterior probability endorsem*nt. Thelevels of support for groupings within Colletidae based on the dif-ferent gene partitions and the combined results can be comparedin Table 6.
E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309 303
The sister-group relationship between Colletinae (with or with-out Callomelitta) and the clade formed by Euryglossinae, Hylaeinae,Scrapterinae, and Xeromelissinae is weakly supported (Figs. 4a, S1
0.05 changes
AAndrena br
Orphana wagenknec
ConanthalictDufoure
Systropha curvicoRophites algirus
Penapis
Ctenocolletes nigricansStenotritus sp.
Ctenocolletes rufescensCtenocolletes smaragdinus
Ca
CaupC
Caupolicana quad
Cadegualina aCadeguala oCadeguala a
Diphaglossa gayiMydro
MydParacolletes cfr. crassipes
CCall
EuhesmaEuhes
XanSc
ScrScrapter niger
Scrapter aScrapter erubesce
Anthoglossa sp.Anthoglossa cfr. robustus
Trichocolletes (T.) aff. venuTrichocolletes (T.) sp.
Phenacolletes mEuryglossidia sp.4Euryglossidia sp.2
Euryglossidia sp.3Euryglossidia sp.1Euryglossidia sp.5
NeopAndPar
LeioLamLeio
Excollet
Colletellus affGlossuroLeio
OdontocolletesOdontocollete
Leioproctus platyceLeioproctus irror
Leioproctus pl
Baeocolletes mBaeocolletes
Halictanthrena malpPerditomorpha neotropCephalocolletes latice
Cephalocolletes isaReedapis bathycyanea
Nomiocolletes
EulonHoplocolletes vent
Spinolapis sp.Spinolapis caerulesBelopria nitidior
Edwyniana sp.Perditomorpha leuco
PerditomorpTetraglos
BrachyHexanthe
Perditomorpha ruPerditomorpha stiborhiKylopasiphae pruinosa
Perditomorpha sp.Niltonia virgilii
Chilicolletes delahozii
LoncLonc
L
Colletidae
Stenotritidae
Andrenidae
Halictidae
Fig. 5. Majority-rule consensus phylogram resulting from mixed-model Bayesian analysisand 28S rRNA genes from 122 species of colletid bees and 22 species of other bee famil
and S3: 81% posterior probability; ML bootstrap = 50; parsimonybootstrap = 12). The Bayesian posterior probability value is lowerthan that for the majority of other clades, which are in general sup-
locandrena porteriooksihti
Nolanomelissa toroiCalliopsis fractaMelitturga clavicornis
Panurgus calcaratus
Protoxaea gloriosa
Agapostemon tyleriMexalictus arizonensis
Halictus rubicundusAugochlorella pomoniellaDieunomia nevadensis
us wilmattaea mullerirnis
penai
Zikanapis clypeataPtiloglossa tarsata
Ptiloglossa sp.Ptiloglossa thoracica
Willinkapis chalybaeaupolicana yarrowi
olicana bicoloraupolicana vestitarifasciata
ndinaccidentalislbopilosa
soma fallaxrosoma aterrimum
Hemicotelles ruizii
Colletes compactus
Colletes floralisColletes simulans
Colletes thoracicusColletes inaequalis
Colletes skinneriColletes pascoensis
Colletes distinctus
Colletes gilvusColletes furfuraceus
Colletes bicolor
Colletes seminitidusXanthocotelles sicheli
Rhynchocolletes mixtusallomelitta antipodes Euhesma sp.ohesma calliopsella platyrhinama aff. crabronica
Euryglossina globulicepsthesma furciferarapter heterodoxusapter ruficornislgoensisns
Xeromelissa australisXeromelissa sp.
Xeromelissa irwiniXeromelissa nortina
Xeromelissa rozeniGeodiscelis longiceps
Xenochilicola mamignaChilicola (Anoediscelis) herbsti
Chilicola (Pseudiscelis) rostrataChilicola (Oediscelis) vicugnaHyleoides concinna
Meroglossa itamucaAmphylaeus (Agogenohylaeus) obscuriceps
Palaeorhiza (Heterorhiza) sp.Palaeorhiza (Palaeorhiza) sp.
Hylaeus (Macrohylaeus) alcyoneusHylaeus (Euprosopis) disjunctus
Hylaeus (Euprosopis) elegansHylaeus (Rhodohylaeus) proximus
Hylaeus (Gnathoprosopis) amiculus
Hylaeus (Pseudhylaeus) aff. simplus
stus
imus
asiphae mirabilisrenopsis sp.acolletinae sp.proctus megachalcoidesprocolletes chalybeataproctus lanceolatuses sp.
. velutinuscolletes bilobatusproctus conospermi
Leioproctus pappusGoniocolletes fimbriatinus
Goniocolletes perfasciatusProtomorpha aff. alloeopus
aff. aspers pachyodontus
phalusatusumosus
inimus sp.
ighiacearumicapsbelae
jenseni
chopria punctatissimaEulonchopria simplicicrusralis
cens
stomaha laenasula anthracinaglossula communisda missionica
fiventrisna
hopria (Biglossa) robertsihopria (Biglossa) sp.onchopria (Lonchopria) similis
OutgroupsDiphaglossinaeColletinaeParacolletinaeEuryglossinaeScrapterinaeXeromelissinaeHylaeinae
of 5498 aligned bp of the EF-1a (exons and introns), opsin (exons), wingless (exons),ies.
Table 6Support for the monophyly of Colletidae, colletid subfamilies, and clades formed by Euryglossinae, Hylaeinae, Scrapterinae, and Xeromelissinae, assessed using Bayesian posteriorprobabilities of individual gene loci (‘‘n.r.” is used to indicate clades not recovered by individual loci).
EF-1a, exons EF-1a introns Opsin Wingless 28S rRNA Combined
Colletidae 100 100 100 93 100 100Colletinae s.str. 100 100 100 97 100 100Diphaglossinae 85 93 100 82 63 100Euryglossinae 98 77 96 n.r. n.r. 100Hylaeinae 100 100 100 100 n.r. 100‘‘Paracolletinae”* n.r. n.r. 78 n.r. n.r. 100Scrapterinae 100 100 100 98 98 100Xeromelissinae 100 100 100 100 n.r. 100Euryglossinae + Scrapterinae 74 57 99 n.r. n.r. 100Hylaeinae + Xeromelissinae 93 89 n.r. n.r. n.r. 99Eurygl. + Scrapt. + Hylaei. + Xerom. 80 n.r. 100 n.r. n.r. 100
* Clade formed by all representatives of Paracolletinae except Callomelita antipodes and Paracolletes cfr. crassipes.
Diph
aglo
ssin
aePa
raco
llete
sPa
raco
lletin
aeCo
lletin
aeC
allo
mel
itta
Eury
glos
sina
eSc
rapt
erin
aeHy
laei
nae
Xero
mel
issi
nae
Diph
aglo
ssin
aePa
raco
llete
sPa
raco
lletin
aeCo
lletin
aeC
allo
mel
itta
Eury
glos
sina
eSc
rapt
erin
aeHy
laei
nae
Xero
mel
issi
nae
Diph
aglo
ssin
aePa
raco
llete
sPa
raco
lletin
aeCo
lletin
aeCa
llom
elitt
aEu
rygl
ossi
nae
Scra
pter
inae
Hyla
eina
eXe
rom
elis
sina
e
Diph
aglo
ssin
aePa
raco
llete
sPa
raco
lletin
ae
Colle
tinae
Callo
mel
itta
Eury
glos
sina
eSc
rapt
erin
aeHy
laei
nae
Xero
mel
issi
nae
Diph
aglo
ssin
aePa
raco
llete
s
Para
colle
tinae
Colle
tinae
Callo
mel
itta
Eury
glos
sina
eSc
rapt
erin
aeHy
laei
nae
Xero
mel
issi
nae
Diph
aglo
ssin
aePa
raco
llete
sPa
raco
lletin
aeCo
lletin
aeCa
llom
elitt
aEu
rygl
ossi
nae
Scra
pter
inae
Hyla
eina
eXe
rom
elis
sina
ea
c
e f
d
b
Fig. 6. Alternative tree topologies statistically compared using likelihood-based tests and Bayesian information content. The resulting topology of an unconstrained Bayesiananalysis of the combined data set (a) was compared in a pair-wise manner to five alternative topologies (b–f) with various placements of Callomelitta antipodes andParacolletes cfr. crassipes.
Table 5Results of comparisons among alternative topologies for Colletidae using maximum likelihood (Kishino–Hasegawa and Shimodaira–Hasegawa tests) and Bayesian approaches.ML tests, as implemented in PAUP� using a one-tailed test of significance (p-value for both tests was identical in all pair-wise comparisons made). Harmonic means from MrBayesanalyses and Bayes factor comparisons: 2loge(B10) stands for twice the log of the Bayes factor in the comparison between two topologies. All comparisons were made between theunconstrained tree topology and the alternative constrained topologies.
Tree topology �lnL Diff. �lnL p-Value Harmonic mean 2loge(B10)
Unconstrained tree topology (Fig. 7a) 85524.62207 �78986.35Paracolletes constraint 1 (Fig. 7b) 85531.55656 6.94 0.258 �79007.35 42.00***
Paracolletes constraint 2 (Fig. 7c) 85558.56353 33.93 0.020* �79023.13 73.56***
Callomelitta constraint 1 (Fig. 7d) 85527.1818 2.56 0.264 �78990.34 7.98**
Callomelitta constraint 2 (Fig. 7e) 85521.8022 2.82 0.335 �78990.34 7.98**
Callomelitta constraint 3 (Fig. 7f) 85524.10473 0.52 0.481 �78992.79 12.88***
* Statistically significant difference of likelihood scores at p = 0.05.** Strong evidence against null hypothesis, i.e., tree topologies equally likely—as proposed by Kass and Raftery (1995, 277).
*** Very strong evidence against null hypothesis (Kass and Raftery ibid).
304 E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309
E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309 305
ported by values equal or higher than 90%. The three basal-mostnodes of ‘‘Paracolletinae” in the Bayesian and ML phylograms re-veals very short internodes (Figs. 5 and S2). These groupings areweakly supported by parsimony and ML bootstrap proportions,but are well-supported by Bayesian posterior probabilities. NeitherML nor parsimony analyses yield alternative arrangements of thetaxa shown as basal in the ‘‘Paracolletinae” clade.
The majority of the Paracolletinae were shown to form a well-supported clade, which comprises most groups traditionally com-prised in this subfamily, except Paracolletes, Callomelitta, Anthoglos-sa, Trichocolletes, and Lonchopria. This clade roughly corresponds toLeioproctus s.l. (e.g., Michener, 1965, 1989, 2007), although a num-ber of paracolletine genera recognized by Michener (2007) renderLeioproctus s.l. paraphyletic. Three main lineages recovered withinthis clade: (1) an entirely Australian clade which comprises Baeo-colletes, Goniocolletes and relatives; (2) a Neotropical clade that in-cludes Eulonchopria, Niltonia and related genera; and (3) anotherNeotropical group formed by Halictanthrena, Reedapis and relatedgenera. Relationships among these three groups are not firmlyestablished by any of the analytical methods employed for phylo-genetic estimation. The internodes on this part of the tree are veryshort (Figs. 5 and S2), making the recovery of unambiguous phylo-genetic information difficult. The above mentioned clade formedby Australian paracolletine bees contains a well-supported subc-lade which corresponds to Leioproctus s.str., Goniocolletes and re-lated genera. This unnamed clade not only contains enoughsequence synapomorphies to render high support values(Fig. 4b), but also has exceptionally long introns in the non-codingregions of EF-1a (intron 2) and opsin (intron 3) (Danforth and Al-meida, unpublished data).
Five partition trees, resulting from Bayesian analyses of theindividual partition data sets (introns and exons of EF-1a, exonsof opsin and wg, and 28S rRNA), are shown in Figs. S4–S8 (Supple-mentary materials). All five partitions, independently (Table 6), orcombined recovered colletid monophyly. The relationships be-tween Colletidae and outgroups agree with the molecular resultsobtained by Danforth et al. (2006a,b). Moreover, relationshipswithin Andrenidae and within Halictidae fully agree with more de-tailed studies of these two bee families conducted previously(Ascher, 2004; Danforth et al., 2004, respectively).
Additional investigation of the resolution power provided byeach gene showed a different (and more fundamental) role of EF-1a and opsin compared to wg and 28S rRNA in the phylogeneticanalyses (Fig. 7). The former two were particularly effective inresolving the earliest divergences among colletid lineages (Fig. 7:
0.2
0.4
0.6
0.8
1all nodes
Paracolletinaedeep nodes
opsin
(exo
ns)
wingles
s(e
xons
)
EF-1
alpha
(exo
ns)
EF-1
alpha
(intro
ns)
28S
rRNA
Fig. 7. Congruent topological information (CTI) of partition Bayesian trees incomparison to the combined Bayesian tree topology (see Section 2 for CTIcalculation). CTI was computed for (1) the complete set of possible relationshipswithin Colletidae (‘‘all nodes”); (2) 17 deeper nodes within the ingroup (list ofnodes in Supplementary Appendix 2); and (3) nodes within ‘‘Paracolletinae”, toinfer on the capacity of the individual loci to recover more recent relationships.
‘‘deep nodes”). Opsin and EF-1a contained information to resolvemore recent divergences as well, as measured by the congruencebetween paracolletine nodes in each of the gene partition treesand the combined tree (Fig. 7: ‘‘Paracolletinae”). These two genesare highly informative as compared to the other data partitions(Table 3), and high information content appears to translate intophylogenetic signal needed to resolve different regions of the col-letid tree topology.
4. Discussion
Stable classifications are ultimately dependent on establishing awell-corroborated phylogenetic hypothesis for groups of organ-isms. For many taxonomically difficult groups of bees, much ofthe current classifications may be unsatisfying primarily becauseit is often not based on such a robust phylogenetic hypothesis.Molecular data can substantially clarify difficult and confusingphylogenetic matters. Colletidae and, specifically, Colletinae s.l.are examples of this. In Michener (2007, 136) words: ‘‘[i]t is clearthat the Colletinae includes diverse elements. A needed step is aphylogenetic study of the forms here placed in Colletinae”.
The continued addition of ideas and information to the discus-sion of relationships among lineages of Colletidae seems to haveprogressed in ways congruent with the results found by the cur-rent study. Rozen (1984) conjectured that Diphaglossinae shouldconstitute the sister-group to all other colletid bees. Rozen’shypothesis was based on diphaglossine bees being the only collet-ids known to spin a cocoon, and the interpretation that this behav-ior (and morphology associated with it) could hardly have re-evolved from non-cocoon-spinning ancestors (Rozen, 1984). Thesister-group relationship between Diphaglossinae and the remain-ing colletid subfamilies is strongly supported by the molecularphylogenetic results.
4.1. Molecular data and the phylogeny of Colletidae
The degree of phylogenetic resolution and the relatively strongsupport at many nodes obtained with the various methods testedin this study are very encouraging. This is the first phylogeneticstudy of Colletidae with extensive taxon sampling to provide awell-resolved hypothesis for the relationships on a worldwidescale. It is noticeable, however, that some regions of the trees arecharacterized by low support values. Future research may demon-strate that these weakly supported nodes are inherently hard to re-solve, instead of having been caused by insufficient data.Parametric branch lengths (Figs. 5 and S2) reveal very short inter-nodes at various parts of the tree of Colletidae, which suggest rapiddiversification among the early-diverging colletid lineages. Consid-ering that a relatively large amount of data was used and that thosedata came from genes appropriate for resolving the relationshipsamong colletid subfamilies, low support may indeed derive froma rapid radiation (Whitfield and Kjer, 2008).
The improved phylogenetic resolution obtained in the com-bined analysis of all gene data partitions as compared with theindividual partitioned analyses is readily appreciated when Figs.S4–S8 and Figs. 4, S1, and S3 are contrasted. In general, individualdata partitions are expected to provide lower resolution than asimultaneous analysis. Moreover, phylogenetic noise present inindividual partitions may obscure the true signal that is amplifiedwhen all data are combined (Nixon and Carpenter, 1996; Baker andDeSalle, 1997).
The phylogenetic signal present in EF-1a alone recovered therelationships among the major lineages of Halictidae (Danforth,2002), which largely agreed with the morphological results of Pes-enko (1999). Later, addition of genetic data from opsin and wg did
306 E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309
not alter the deeper relationships for an analysis of basically thesame set of taxa, but resulted in increased support values for mostinternodes (Danforth et al., 2004). Phylogeny inferred from EF-1awas, among the genes sampled by Danforth et al. (2004), the mostresolved tree. Likewise, EF-1a alone was sufficient to resolve rela-tionships within Andrenidae (Ascher in Rozen, 2003, Fig. 37). Thisresult is strengthened by the high degree of observed congruencewith the morphological data set collected by Ascher (2004). And-renidae comprise approximately 2650 species, a wide distributionand a morphological diversity comparable to Colletidae. Ascher’stree based on this one locus was well-resolved and relationshipswere also well-supported by bootstrap proportions and Bremersupport values (Ascher, 2004, 43).
In our study too, EF-1a seem to provide most of the meaningfulphylogenetic signal for the analysis of colletid relationships (Fig. 7and Table 6), followed by opsin. The parameter a of the gammadistribution is a correlate of data set quality in empirical and sim-ulation studies (Yang, 1998; Lin and Danforth, 2004). It is higher inthe exons of EF-1a and opsin compared to those of wg and in 28SrRNA, reflecting less heterogeneity in among-site rate variation inEF-1a and opsin as compared to wingless and 28S (Fig. 3c).
Differences in the diverse measures of clade support assessmentwere expected given their differing natures. Parsimony relies onabsolute number of changes, whereas maximum likelihood andBayesian phylogenetics depend on the likelihood function. Eventhough phylogenetic estimates are expected to converge in certainsituations (Steel and Penny, 2000; Swofford et al., 2001), it is alsotrue that different phylogenetic methods may extract differentinformation from the same data set. Overall, the topologies foundby the three methods are almost identical, most of the differencebeing on support for certain clades.
Bootstrap and posterior probabilities are not easily compara-ble, despite the efforts of some researchers to find ways to linkthem (e.g., Wilcox et al., 2002; Alfaro et al., 2003; Huelsenbeckand Ranalla, 2004). The bootstrap has long been used as anassessment of support of clades in phylogenetic analyses; poster-ior probabilities are more recent and less well-understood (Huel-senbeck et al., 2002, 683–684 for a discussion). Huelsenbeck andRanalla (2004) used simulations to demonstrate that Bayesianposterior probabilities can be a useful measure of phylogeneticreliability as long as the substitution model is not violated.Empirical studies found Bayesian posterior probabilities to beuseful measures of support as well (e.g., Wilcox et al., 2002).Comparisons among various models of different levels of com-plexity in this study showed the topology to be very resilientto these changes. Posterior probabilities changed in some cases,usually varying by less than 10%. This serves as evidence thatthe Bayesian analysis of the current data set was not very sensi-tive to model misspecification and renders credible the posteriorprobability values obtained.
4.2. Monophyly of Colletidae
Colletid monophyly is well-established. Sequence data fromthis study and larger analyses of bee relationships (Danforthet al., 2006a,b) consistently support it. All colletids whose nestshave been studied line the brood cells with a polyester materialunique among bees (Hefetz et al., 1979; Espelie et al., 1992; Almei-da, 2008), and morphological characteristics of the colletid glossamay be associated with this lining (McGinley, 1980). Results alsoconfirm that Strenotridae do not belong within the Colletidae.Stenotritids are different from colletids in terms of their nest con-struction and bionomy (Houston and Thorp, 1984), as well as mor-phology (McGinley, 1980). Based on larval morphologicalcharacters alone, Paracolletinae and Stenotritidae were groupedtogether by McGinley (1981). Stenotritidae constitutes the sister-
group of Colletidae according to results of this analysis and in Dan-forth et al. (2006a,b).
4.3. Subfamilies of Colletidae
Historically, Colletes and other hairy colletid groups (generallyincluding Diphaglossinae) have been placed separately from Hyla-eus and relatives (e.g., Ashmead, 1899). It had been supposed thatHylaeinae, Xeromelissinae, and Euryglossinae formed a distinctgroup within the Colletidae, despite the enormous variation inways colletid subfamilies were imagined to relate to each other(Fig. 1). The presumed phylogenetic proximity of Hylaeinae, Xer-omelissinae, and Euryglossinae was encountered in analyses ofmorphological data (Alexander and Michener, 1995; McGinley,1981), but, despite their proximity, they do not form a naturalgroup unless Scrapter is also included. Accumulation of evidencethat Scrapter does not belong with the remaining Paracolletinaelead to the recent recognition of an independent monogeneric unit:Scrapterinae (Melo and Gonçalves, 2005; Ascher and Engel in En-gel, 2005; Ascher and Engel, 2006). A sister-group relationship be-tween Scrapter and Euryglossinae had been suggested by a variablebut interesting morphological character: some species of both taxapossess a unique (among bees) basitibial plate not delimited by acarina, ‘‘but by a ring of large tubercles probably which are a resultif subdivision of the normally elevated plate” (McGinley, 1981,160). Additionally some larval characters support the sister-grouprelationship between Scrapterinae and Euryglossinae (McGinley,1981; see comments by Engel, 2005; Ascher and Engel, 2006; Da-vies and Brothers, 2007), even though the analysis of the entire lar-val data set does not support this finding (McGinley, 1981). Resultspresented here unambiguously corroborate previous proposals ofplacing Scrapter in a separate subfamily and its close relationshipto the Euryglossinae, despite the morphological disparity betweenthem.
The results do not provide evidence for the grouping of Colleti-nae s.str. and Paracolletinae, contrary to the common and longused practice of uniting them (e.g. Michener, 1944, 1965, 1989,2007 treated these two groups as tribes comprised into Colletinaes.l.). The proposal of keeping the two subfamilies as separate taxa(e.g., Silveira et al., 2002; Engel, 2005; Melo and Gonçalves, 2005)is compatible with the phylogenetic findings presented here(Fig. 4).
The need for additional investigation to support the separationof Callomelitta and Paracolletes from Paracolletinae becomes evi-dent from the phylogenetic results. Although the phylogeneticaffinities of Callomelitta and Paracolletes to the colletid lineages re-main uncertain, there is strong evidence to assert that neither ar-ose from within one of the formally recognized subfamilies.According to the phylogenetic analysis and the statistical topolog-ical tests (Table 5), it is possible that Callomelitta should be placedas sister-group to all other Paracolletinae, but this is not necessar-ily the case. Based on these phylogenetic results and on the obser-vation of the distinctive morphology of these bees (Michener,2007), the recognition of a separate subfamily to harbor Call-omelitta would be a reasonable solution for the placement of thisgenus (Almeida, 2008b).
Colletinae s.str. is a clade subtended by a fairly long branch, rel-ative to other internodes of Colletidae (Figs. 5 and S2). It is possiblethat the attraction of Callomelitta to the proximity of Colletinae wasartifactual, a long branch effect. Other kinds of data are needed toelucidate this matter: morphological data should not be affected bya possible long branch effect in the same way as molecular data(Bergsten, 2005).
The placement of Paracolletes is another serious issue, because itentails the nomenclatural modification needed for the clade provi-sionally referred to by ‘‘Paracolletinae”. Paracolletes is the type
E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309 307
genus of Paracolletinae and, therefore, its name-bearer taxon. Fam-ily-group names with the root ‘‘Paracollet-” must, as a result, in-clude Paracolletes.
4.4. Genus-level classification of the ‘‘Paracolletinae”
Michener (2007, 145) states his awareness that Leioproctus s.l.most certainly does not constitute a natural group, but he acknowl-edges its usefulness as a diagnosable taxonomic unit. Leioproctuss.l. is a group recognizable on the basis of superficial similarity ofits subgenera and the lack of synapomorphies present in othergroups, such as Neopasiphae or Eulonchopria. Michener (2007,141–142) resorted to caution in changing the classification:‘‘[u]ntil more species are known and cladogeny is better studied,especially in the diverse Australian fauna, there is no point inattempting a new classification, for new combinations that wouldlast only until the next revision would be the result”.
Leioproctus s.l. (e.g. Michener, 2007) is rendered paraphyletic bya number of lineages that arise from within it. Its distinction hasbeen traditionally made on the basis of symplesiomorphies, andthe remaining groups being characterized by synapomorphies. Itis not surprising, then, to find highly modified (morphologicallyautapomorphic) lineages arising from within it and being recog-nized as distinct genera, e.g., Brachyglossula, Eulonchopria, Neopa-siphae, Phenacolletes. Part of the problem is resolved by elevatingthe subgenera of Leioproctus to the status of genus. This view hasbeen long adopted by most South American bee taxonomists,who have traditionally treated all supra-specific taxa of SouthAmerican Paracolletinae as separate genera (e.g., Moure et al.,2007; Silveira et al. 2002). In this case, the genus Leioproctus is re-stricted to the Australian region. However, the classification ofParacolletinae will require future adjustments at the generic levelbecause the phylogenetic results very clearly show that at leasttwo large genera constitute artificial assemblages: Perditomorphaand Leioproctus s.str. even when many subgenera are raised to gen-eric rank.
4.5. Biogeography
Colletidae are distributed primarily in the southern hemisphereand comprise various endemic groups in Australia, South America,and Africa. As discussed elsewhere, the current distribution ofcolletid bees suggests the occurrence of various events of Austra-lian-South American interchanges in the past (Almeida et al.,unpublished results). Paracolletinae is formed by various clades,some endemic to the Australian region, some South American. Ina similar fashion, Hylaeinae and Xeromelissinae form a clade inwhich the former is primarily Australian and the latter occurs fromsouthern South America up to Central America.
Note added on proof
Publication of the article by Almeida (2008b) while the presentpaper was in press made available the name Andrenopsis micheneri-anus Almeida for the species referred to as ‘‘Paracolletinae sp.” inTable 1, and Figs. 4b, 5, S1–S8.
Acknowledgments
Our thanks for comments on earlier versions of this paper toLaurence Packer, James K. Liebherr, and Richard G. Harrison. Dis-cussions with Charles Michener, John Ascher, Terry Houston, GlynnMaynard, Gabriel Melo, and Gregg Davies contributed to the devel-opment of this research. Seán Brady, Jenn Fang, Karl Magnacca, andSedonia Sipes provided valuable advice concerning the molecular
work. We are indebted to all who supplied specimens used forDNA extraction: Antonio Aguiar, Isabel Alves dos Santos, JohnAscher, Torsten Dikow, Martin Hauser, Terry Houston, Karl Magna-cca, Gabriel Melo, Claus Rasmussen, Jerome Rozen, Andrew Short,Lorenzo Zanette, and Amro Zayed. We are mostly indebted to Lau-rence Packer for his generosity in providing samples of a largenumber of rare colletid species. We are grateful for the bee taxon-omists listed in the methods section for assistance with the identi-fication of taxa included in the present study. Assistance for fieldtrips was given by Paulo Almeida, Ricardo Ayala, Ricardo Botelho,Guillermo Debandi, Mariano Devoto, Eduardo Dominguez, ConnalEardley, Dimitri Forero, Terry Houston, Laurence Packer, CharlotteSkov. Collecting permits in Argentina and Australia were facilitatedby the competent work of Roberto Salinas (Servicio de Fauna Sil-vestre, Catamaca, Argentina); Gustavo Solá (Dirección de RecursosNaturales Renovables, Mendoza, Argentina); Ing. Ana Inés Arce(Secretaría de Medio Ambiente y Desarollo Sustentable, Salta,Argentina); Lic. Claudia Perez Miranda (Dirección de RecursosNaturales y Suelos, Tucumán, Argentina), and Mr. Danny Stefony(CALM Wildlife Branch, Western Australia). The majority of fund-ing for this research came from a National Science Foundation Re-search Grant in Systematic Biology to B.N. Danforth and E.A.B.Almeida (DEB-0412176). Additional funding was provided by theA.C.Rawlins Endowment (Cornell University); Grace Griswold Fund(Cornell University); Tinker Foundation & Latin American StudiesProgram (Cornell University). E.A.B. Almeida is grateul to Fundaçãode Coordenação de Aperfeiçoamento de Pessoal de Nível Superior(CAPES: Brazilian Ministry of Education) for a Ph.D. scholarship(BEX-0967/01-7) and to CNPq for a post-doctoral fellowship(PDJ-150626/2007-0).
Appendix A. Supplementary data
Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.ympev.2008.09.028.
References
Alexander, B.A., Michener, C.D., 1995. Phylogenetic studies of the families of short-tongued bees (Hymenoptera: Apoidea). Univ. Kans. Sci. Bull. 55, 377–424.
Alfaro, M.E., Zoller, S., Luzzoni, F., 2003. Bayes or bootstrap? A simulation studycomparing the performance of Bayesian Markov chain Monte Carlo samplingand bootstrapping in assessing phylogenetic confidence. Mol. Biol. Evol. 20,255–266.
Almeida, E.A.B., 2008a. Colletidae nesting biology (Hymenoptera: Apoidea).Apidologie 39, 16–29.
Almeida, E.A.B., 2008b. Revised species checklist of the Paracolletinae(Hymenoptera, Colletidae) of the Australian Region, with the description ofnew taxa. Zootaxa 1891, 1–24.
Almeida, E.A.B., Packer, L., Danforth, B.N., 2008. Phylogeny of the Xeromelissinae(Hymenoptera: Colletidae) based upon morphology and molecules. Apidologie39, 75–85.
Altekar, G., Dwarkadas, S., Huelsenbeck, J.P., Ronquist, F., 2004. Parallel Metropolis-coupled Markov chain Monte Carlo for Bayesian phylogenetic inference.Bioinformatics 20, 407–415.
Ashmead, W.H., 1899. Classification of the bees. Trans. Am. Entomol. Soc. 26, 49–100.
Ascher, J.S., 2004. Systematics of the Bee Family Andrenidae (Hymenoptera:Apoidea). Ph.D. Dissertation, Cornell University, Ithaca.
Ascher, J.S., Engel, M.S., 2006. On the availability of family-group names based onScrapter (Hymenoptera: Colletidae). Entomol. News 117, 117–119.
Ascher, J.S. et al., 2008. Apoidea Species Guide. Available from: <http://www.discoverlife.org/mp/20q?guide=Apoidea_species> (accessed 22.04.08).
Baker, R.H., DeSalle, R., 1997. Multiple sources of character information and thephylogeny of Hawaiian drosophilids. Syst. Biol. 46, 654–673.
Belshaw, R., Quicke, D.L.J., 1997. A molecular phylogeny of the Aphidiinae(Hymenoptera: Braconidae). Mol. Phylogenet. Evol. 7, 281–293.
Bergsten, J., 2005. A review of long-branch attraction. Cladistics 21, 163–193.Brady, S.G., Danforth, B.N., 2004. Recent intron gain in elongation factor-1a of
colletid bees (Hymenoptera: Colletidae). Mol. Biol. Evol. 21, 691–696.Brower, A.V.Z., DeSalle, R., 1998. Patterns of mitochondrial versus nuclear DNA
sequence divergence among nymphalid butterflies: the utility of winglessas a source of characters for phylogenetic inference. Insect Mol. Biol. 7,73–82.
308 E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309
Burnham, K.P., Anderson, D.R., 2002. Model Selection and Multimodel Inference. APractical Information-Theoretic Approach, second ed. Springer-Verlag, NewYork.
Caterino, M.S., Cho, S., Sperling, F.A.H., 2000. The current state of insect molecularsystematics: a thriving Tower of Babel. Annu. Rev. Entomol. 45, 1–54.
Chang, B.S.W., Ayers, D., Smith, W.C., Pierce, N.E., 1996. Cloning of the geneencoding honeybee long-wavelength rhodopsin: a new class of insect visualpigments. Gene 173, 215–219.
Daly, H.V., Magnacca, K.N., 2003. Insects of Hawaii. Volume 17: Hawaiian Hylaeus(Nesoprosopis) Bees (Hymenoptera: Apoidea). University Hawai‘i Press,Honolulu.
Danforth, B.N., 1999. Phylogeny of the bee genus Lasioglossum (Hymenoptera:Halictidae) based on mitochondrial cytochrome oxidase. Syst. Entomol. 24,377–393.
Danforth, B.N., 2002. Evolution of sociality in a primitively eusocial lineage of bees.Proc. Natl. Acad. Sci. USA 99, 286–290.
Danforth, B.N., Brady, S.G., Sipes, S.D., Pearson, A., 2004. Single-copy nuclear genesrecover Cretaceous-age divergences in bees. Syst. Biol. 55, 309–326.
Danforth, B.N., Fang, J., Sipes, S., 2006a. Analysis of family-level relationships inbees (Hymenoptera: Apiformes) using 28S and two previously unexplorednuclear genes: CAD and RNA polymerase II. Mol. Phylogenet. Evol. 39, 358–372.
Danforth, B.N., Sipes, S., Fang, J., Brady, S.G., 2006b. The history of early beediversification based on five genes plus morphology. Proc. Natl. Acad. Sci. USA103, 15118–15123.
Danforth, B.N., Lin, C.P., Fang, J., 2005. How do insect nuclear ribosomal genescompare to protein-coding genes in phylogenetic utility and nucleotidesubstitution patterns? Syst. Entomol. 30, 549–562.
Danforth, B.N., Sauquet, H., Packer, L., 1999. Phylogeny of the bee genus Halictus(Hymenoptera: Halictidae), based on parsimony and likelihood analyses ofnuclear EF-1 sequence data. Mol. Phylogenet. Evol. 13, 605–618.
Davies, G.B.P., Brothers, D.J., 2007. Morphology of Scrapter (Hymenoptera:Anthophila: Colletidae), with description of three new species and taxonomicstatus of five co*ckerell taxa. Afr. Invert. 47, 135–183.
Davies, G.B.D., Eardley, C.D., Brothers, D.J., 2005. Eight new species of Scrapter(Hymenoptera: Apoidea: Colletidae), with description of S. albifumus and S.albitarsus females and a major range extension of the genus. Afr. Invert. 46,141–179.
Doyle, J.J., Doyle, J.L., 1990. A rapid total DNA preparation procedure for fresh planttissue. Focus 12, 13–15.
Eardley, C., 1996. The genus Scrapter Lepeletier and Serville (Hymenoptera:Colletidae). Afr. Entomol. 4, 37–92.
Engel, M.S., 2001. A monograph of the Baltic amber bees and evolution of theApoidea (Hymenoptera). Bull. Am. Mus. Nat. Hist. 259, 1–192.
Engel, M.S., 2005. Family-group names for bees (Hymenoptera: Apoidea). Am. Mus.Novit. 3476, 1–33.
Espelie, K.E., Cane, J.H., Himmelsbach, D.S., 1992. Nest cell lining of the solitary beeHylaeus bisinuatus (Hymenoptera: Colletidae). Experientia 48, 414–416.
Farris, J.S., 1989. The retention index and the rescaled consistency index. Cladistics5, 417–419.
Felsenstein, J., 1985. Confidence limits on phylogenies: an approach using thebootstrap. Evolution 39, 783–791.
Felsenstein, J., 2004. Inferring Phylogenies. Sinauer, Sunderland.Gelman, A., Rubin, D.B., 1992. Inference from iterative simulation using multiple
sequences. Stat. Sci. 7, 457–472.Hanco*ck, J.M., Tautz, D., Dover, G.A., 1988. Evolution of the secondary structures and
compensatory mutations of the ribosomal RNAs of Drosophila melanogaster.Mol. Biol. Evol. 5, 393–414.
Hefetz, A., Fales, H.M., Batra, S.W.T., 1979. Natural polyesters: Dufour’s glandmacrocyclic lactones form brood cell laminesters in Colletes bees. Science 204,415–417.
Hillis, D.M., Dixon, M.T., 1991. Ribsomal DNA: molecular evolution andphylogenetic inference. Q. Rev. Biol. 66, 411–453.
Houston, T.F., Thorp, R.W., 1984. Bionomics of the bee Stenotritus greavesi andethological characteristics of Stenotritidae (Hymenoptera). Rec. West. Aust.Mus. 11, 375–385.
Huelsenbeck, J.P., Larget, B., Miller, R.E., Ronquist, F., 2002. Potential applicationsand pitfalls of Bayesian inference of phylogeny. Syst. Biol. 51, 673–688.
Huelsenbeck, J.P., Ranalla, B., 2004. Properties of Bayesian posterior probabilities ofphylogenetic trees under simple and complex substitution models. Syst. Biol.53, 904–913.
Huelsenbeck, J.P., Ronquist, F., 2005. MrBayes: Bayesian inference of phylogeny.Bioinformatics 17, 754–755.
Jander, R., 1976. Grooming and pollen manipulation in bees (Apoidea): the natureand evolution of movements involving the foreleg. Physiol. Entomol. 1, 179–194.
Kass, R.E., Raftery, A.E., 1995. Bayes factors. J. Am. Stat. Assoc. 90, 773–795.Kishino, H., Hasegawa, M., 1989. Evaluation of the maximum likelihood estimate of
the evolutionary tree topologies from DNA sequence data, and the branchingorder in Hominioidea. J. Mol. Evol. 29, 170–179.
Kishino, H., Miyata, T., Hasegawa, M., 1990. Maximum likelihood inference ofprotein phylogeny and the origin of chloroplasts. J. Mol. Evol. 31, 151–160.
Lin, C.P., Danforth, B.N., 2004. How do insect nuclear and mitochondrial genesubstitution patterns differ? Insights from Bayesian analyses of combineddatasets. Mol. Phylogenet. Evol. 30, 686–702.
Linares, A.R., Hanco*ck, J.M., Dover, G.A., 1991. Secondary structure constraints onthe evolution of Drosophila 28S ribosomal RNA expansion segments. J. Mol. Biol.219, 381–390.
Maddison, D.R., Maddison, W.P., 2005. MacClade 4.08. Sinauer Associates,Sunderland.
Mardulyn, P., Cameron, S., 1999. The major opsin in bees (Insecta: Hymenoptera): apromising nuclear gene for higher level phylogenetics. Mol. Phylogenet. Evol.12, 168–176.
Mardulyn, P., Whitfield, J.B., 1999. Phylogenetic signal in the COI, 16S, and 28Sgenes for inferring relationships among genera of microgastrinae(Hymenoptera: Braconidae): evidence of a high diversification rate in thisgroup of parasitoids. Mol. Phylogenet. Evol. 12, 282–294.
McGinley, R.J., 1980. Glossal morphology of the Colletidae and recognition of theStenotritidae at the family level. J. Kans. Entomol. Soc. 53, 539–552.
McGinley, R.J., 1981. Systematics of the Colletidae based on mature larvae withphenetic analysis if apoid larvae (Hymenoptera: Apoidea). Univ. Calif. Publ.Entomol. 91, 1–307 + vxi.
Melo, G.A.R., Gonçalves, R.B., 2005. Revised bee classifications (Hymenoptera,Apoidea, Apidae sensu lato). Rev. Bras. Zool. 22, 153–159.
Michener, C.D., 1944. Comparative external morphology, phylogeny, and aclassification of the bees (Hymenoptera). Bull. Am. Mus. Nat. Hist. 82, 151–326.
Michener, C.D., 1965. A classification of the bees of the Australian and South Pacificregions. Bull. Am. Mus. Nat. Hist. 130, 1–362. pls. 1–15.
Michener, C.D., 1974. The Social Behavior of the Bees: A Comparative Study. TheBelknap Press of Harvard University Press, Cambridge, USA.
Michener, C.D., 1979. Biogeography of the bees. Ann. MO Bot. Gard. 66, 277–347.Michener, C.D., 1986. A review of the tribes Diphaglossini and Dissoglottini. Univ.
Kans. Sci. Bull. 53, 183–214.Michener, C.D., 1989. Classification of the American Colletinae. Univ. Kans. Sci. Bull.
53, 622–703.Michener, C.D., 2007. The Bees of the World, 2nd ed. The John Hopkins University
Press, Baltimore.Minin, V., Abdo, Z., Joyce, P., Sullivan, J., 2003. Performance-based selection of
likelihood models for phylogeny estimation. Syst. Biol. 52, 674–683.Morgenstern, B., 1999. DIALIGN 2: improvement of the segment-to-segment
approach to multiple sequence alignment. Bioinformatics 15, 211–218.Moure, J.S., Graf, V., Urban, D., 2007. Paracolletini co*ckerell, 1934. In: Moure, J.S.,
Urban, D., Melo, G.A.R. (Eds.), Catalogue of Bees (Hymenoptera, Apoidea) in theNeotropical Region. Sociedade Brasileira de Entomologia, Curitiba, pp. 723–743.
Nixon, K.C., Carpenter, J.M., 1993. On outgroups. Cladistics 9, 413–426.Nixon, K.C., Carpenter, J.M., 1996. On simultaneous analysis. Cladistics 12, 221–241.Nixon, K.C., 2002. WinClada v1.00.08. Computer program distributed by the author.
Available from: <http://www.cladistics.com/wincDownload.htm>.Nylander, J.A.A. 2004a. MrAIC.pl. Computer program distributed by the author.
Evolutionary Biology Centre, Uppsala University. Available from: <http://www.ebc.uu.se/systzoo/staff/nylander/MrAIC142.zip>.
Nylander, J.A.A., 2004b. MrModeltest v2. Computer program distributed by theauthor. Evolutionary Biology Centre, Uppsala University. Available from:<http://www.ebc.uu.se/systzoo/staff/nylander/MrMod22.sit>.
Nylander, J.A.A., Ronquist, F., Huelsenbeck, J.P., Nieves-Aldrey, J.L., 2004. Bayesianphylogenetic analysis of combined data. Syst. Biol. 53, 47–67.
O’Toole, C., Raw, A., 1991. Bees of the World. Blanford, London, UK.Packer, L., 2008. Phylogeny and classification of the Xeromelissinae (Hymenoptera:
Apoidea, Colletidae) with special emphasis on the genus Chilicola. Syst. Entomol.33, 72–96.
Pesenko, Y.A., 1999. Phylogeny and classification of the family Halictidae revised(Hymenoptera: Apoidea). J. Kans. Entomol. Soc. 72, 104–123.
Posada, D., Buckley, T.R., 2004. Model selection and model averaging inphylogenetics: advantages of Akaike information criterion and Bayesianapproaches over likelihood ratio tests. Syst. Biol. 53, 793–808.
Posada, D., Crandall, K.A., 1998. Modeltest: testing the model of DNA substitution.Bioinformatics 14, 817–818.
Posada, D., Crandall, K.A., 2001. Selecting the best-fit model of nucleotidesubstitution. Syst. Biol. 50, 580–601.
Rozen Jr., J.G., 1984. Nesting biology of diphaglossine bees (Hymenoptera:Colletidae). Am. Mus. Novit. 2786, 1–33.
Rozen Jr., J.G., 2003. A new tribe, genus, and species of South American panurginebee (Andrenidae: Panurginae), oligolectic on Nolana (Nolanaceae); with andappendix by J.S. Ascher: ‘‘Evidence for the phylogenetic position ofNolanomelissa from nuclear EF-1a sequence data”. In: Melo, G.A.R., Alves-dos-Santos, I. (Eds.), Apoidea Neotropica: Homenagem aos 90 Anos de Jesus SantiagoMoure. Editora da UNESC, Criciúma, pp. 3–19.
Rozen Jr., J.G., Michener, C.D., 1968. The biology of Scrapter and its cuckoo bee,Pseudodichroa (Hymenoptera: Colletidae and Anthophoridae). Am. Mus. Novit.2335, 1–13.
Shimodara, H., Hasegawa, M., 1999. Multiple comparisons of log-likelihoods withapplications to phylogenetic inference. Mol. Biol. Evol. 16, 1114–1116.
Silveira, F.A., Melo, G.A.R., Almeida, E.A.B., 2002. Abelhas Brasileiras. Sistemática eIdentificação. Published by the authors, Belo Horizonte, Brazil.
Swofford, D.L., 2002. PAUP�. Phylogenetic Analysis Using Parsimony (�and OtherMethods), v4.0b10. Sinauer Associates, Sunderland, MA.
Swofford, D.L., Waddell, P.J., Huelsenbeck, J.P., Foster, P.G., Lewis, P.O., Rogers, J.S.,2001. Bias in phylogenetic estimation and its relevance to the choice betweenparsimony and likelihood methods. Syst. Biol. 50, 525–539.
Steel, M., Penny, D., 2000. Parsimony, likelihood, and the role of models inmolecular phylogenetics. Mol. Biol. Evol. 17, 839–850.
E.A.B. Almeida, B.N. Danforth / Molecular Phylogenetics and Evolution 50 (2009) 290–309 309
Sullivan, J., Joyce, P., 2005. Model selection in phylogenetics. Annu. Rev. Ecol. Evol.Syst. 36, 445–466.
Talavera, G., Castresana, J., 2007. Improvement of phylogenies after removingdivergent and ambiguously aligned blocks from protein sequence alignments.Syst. Biol. 56, 564–577.
Tautz, D., Hanco*ck, J.M., Webb, D.A., Tautz, C., Dover, G.A., 1988. Completesequences of the rRNA genes of Drosophila melanogaster. Mol. Biol. Evol. 5,366–376.
Uzvölgyi, E., Kiss, I., Pitt, A., Arsenian, S., Ingvarsson, S., Udvardy, A., Hamada, M.,Klein, G., Sumegi, J., 1988. Drosophila hom*olog of the murine Int-1protooncogene. Proc. Natl. Acad. Sci. USA 85, 3034–3038.
Walldorf, U., Hoverman, B.T., 1990. Apis mellifera cytoplasmatic elongation factor 1alpha (EF-1a) is closely related to Drosophila melanogaster EF-1a. FEBS 267, 245–249.
Ward, P.S., Brady, S.G., 2003. Phylogeny and biogeography of the antsubfamily Myrmeciinae (Hymenoptera: Formicidae). Invertebr. Syst. 17,361–386.
Wcislo, W.T., Cane, J.H., 1996. Floral resource utilization by solitary bees(Hymenoptera: Apoidea) and exploitation of their stored foods by naturalenemies. Annu. Rev. Entomol. 41, 257–286.
Whitfield, J.B., Kjer, K.M., 2008. Ancient rapid radiations of insects: challenges forphylogenetic analysis. Annu. Rev. Entomol. 53, 449–472.
Wiegmann, B.M., Mitter, C., Regier, J.C., Friedlander, T.P., Wagner, D.M., Nielsen, E.S.,2000. Nuclear genes resolve Mesozoic-aged divergences in the insect orderLepidoptera. Mol. Phylogenet. Evol. 15, 242–259.
Wilcox, T.P., Zwickl, D.J., Heath, T.A., Hillis, D., 2002. Phylogenetic relationships ofthe dwarf boas and a comparison of Bayesian and bootstrap measures ofphylogenetic support. Mol. Phylogenet. Evol. 25, 361–371.
Yang, Z., 1998. On the best evolutionary rate for phylogenetic analysis. Syst. Biol. 47,125–133.
Zwickl, D.J., 2006. Genetic algorithm approaches for the phylogenetic analysis oflarge biological sequence datasets under the maximum likelihood criterion.Ph.D. Dissertation, University Texas, Austin. [Computer program distributed bythe author at: <http://www.bio.utexas.edu/faculty/antisense/garli/Garli.html>].