The Higher Classification of Bees

The following paragraphs serve to identify those groups recognized as families, as well as the relationships among the families. They concern only extant families; the extinct groups that have been given family-group status are ignored. They add detail to the information already provided by the dendrograms (Figs. 20-1, 20-2) and Table 16-1. They are not intended as descriptions of the families, but rather as indications of some of the reasons for decisions about the family-level classification. Keys to the families of bees will be found in Sections 33 to 35.

Stenotritidae. The two genera ofthis family, both Australian, are very similar to one another in most characters, although the male S7 is similar to that of most colletids in Stenotritus, much more simple in Ctenocolletes. In the phylogenetic studies by Alexander and Michener (1995), Ctenocolletes appeared in such diverse locations (see Fig. 20-1) that its phylogenetic position was uncertain. Stenotritidae therefore was given family status. In both cladistic and nearest-neighbor (phenetic) analyses of larval structures by McGinley (1981), stenotritrids fell among the Paracolletini in the Colletinae. These findings support the placement shown in Figure 20-1c and may indicate that stenotritids should be included in the Col-letidae. In the stenotritids, however, the glossa is rounded and lacks the familial synapomorphies listed for Colle-tidae in Section 20 and below. Also in stenotritids, the scopa is on the hind tibia, not developed on the tro-chanter and femer as in scopate Colletidae.

Colletidae. There is so much diversity within the Col-letidae that the subfamilies could easily be, and at various times have been, considered as separate families. No strong claim can be made for uniting them on the basis of general phenetic similarity; compare, for example, Hy-laeus and Caupolicana, or even Hylaeusand Colletes. There are, however, significant familial synapomorphies. The glossa of most colletids is recognizable by synapomorphies at least of females, that are not found in any other groups of Hymenoptera, i.e., the glossal lobes and brush and the preapical fringe. Moreover, if the Perkins-McGinley hypothesis (see Sec. 20) is correct, there are other colletid glossal synapomorphies, e.g., shape, and there is at least one synapomorphy other than the glossal characters that unites all colletid subfamilies. This is the very reduced disc of S7 of the male, with long apodemes extending basolat-erally and one to three pairs of usually hairy and elaborate apical processes or lobes (Fig. 13-2). Nothing of the sort is found in sphecoid wasps. In some other bees, such as some Stenotritidae, Melittidae, some panurgine Andrenidae, rophitine Halictidae, and even some Apinae, S7 of males has apical lobes or processes, but it is rarely so modified as those in Colletidae. A few colletids have probably lost the typical colletid S7; examples are the colletine Glossurocol-letes, the euryglossine Euryglossina (Euryglossella), and the hylaeines Hylaeus (Edriohylaeus and Metziella). In each case, however, closely related genera or subgenera have the usual colletid structure.

Interestingly, the colletids whose males have pointed glossae (Hemirhiza, Meroglossa, Palaeorhiza), and those like Amphylaeus, with a slightly pointed glossa, do not constitute a subfamily or other major group of their own. Their females and other male characters are typical of the subfamily Hylaeinae.

Given the fact that males of a few colletids have glos-sae that are in all features like those of acute-tongued, non-colletid, S-T bees, one may expect to discover bees that are in most features colletids but that have acute glos-sae in both sexes, or at any rate glossae that are not of the usual colletid style. Possibly such bees should be included in the Colletidae, since obviously the colletid-style glossa is not an essential feature of the family, at least in males. Candidates for possible inclusion in the Colletidae were the Oxaeinae and Stenotritidae. This idea is attractive because of the similarity of these families, in various features, to the colletid subfamily Diphaglossinae. The species of Stenotritidae, indeed, resemble Colletidae in many ways; they share with colletids certain glossal probable synapomorphies (no recognizable seriate hairs; annulate surface not broader than the disannulate surface) as well as the male S7 structure in Stenotritus but not in Ctenocolletes. The reasons for retaining a separate family, Stenotritidae, are indicated above. The Oxaeinae, however, have distinctive mouthparts that are more similar in some ways to those of Halictidae, and larvae that do not share the usual colletid characteristics (see below). In most of the analyses by Alexander and Michener (1995), Oxaeinae fell with Andreninae; in other analyses Ox-aeinae was the sister group to the Stenotritidae, but in no case did it appear among colletids. Apparently, there are no bees with a pointed glossa in both sexes that should be included in Colletidae. The monophyly ofthe Colletidae and exclusion of the Stenotritidae from the Colletidae are supported by Brady and Danforth (2004) who found a unique intron, not found elsewhere, in all colletid subfamilies in the study (Diphaglossinae was not included). Monophyly of the Colletidae was also shown by Dan-forth et al. (2006) on the basis of numerous molecular as well as morphological characters (see Fig. 20-6).

Colletidae as here constituted is paraphyletic if the Australian subfamily Euryglossinae is the basal clade of bees, sister group to all other bees, as indicated in Figure

20-1a. In other analyses, however, Euryglossinae appeared as the sister group to all other colletids or to the Hylaeinae. Although the basal position of Euryglossinae is attractive, I am impressed by similarities to Hylaeinae, the large sclerite supporting the curved galeal comb (Figs.

21-1, 48-1) in both subfamilies, among other characters, being probably synapomorphic.

No doubt one reason for the impression that Eury-glossinae may be an ancestral group is the short proboscis and the very short, often almost rectangular, galeal blade, and frequently the reduction or absence of the galeal velum. The galea, however, is not at all like that of any sphecoid wasp known to me, although these wasps also lack a velum as distinctive as that of most bees. Moreover, euryglossines like Pachyprosopis have a well-formed galeal velum, as in other bees. The galeal structure is further discussed in Section 48 (Euryglossinae).

Andrenidae. Major maxillary types ofAndrenidae and Halictidae are diagrammed in Figure 21-2. In most ofthe analyses by Alexander and Michener (1995), the An-drenidae appeared as a monophyletic unit when the Ox-aeinae, frequently given family rank, was included. The Panurginae and the Andreninae have been associated in the past because of the presence of two subantennal sutures on each side of the face, a synapomorphy shared by few other bees. Less diagnostic characters are the presence, in most, of facial foveae (also found in some col-letids and, in less well-defined forms, in various other bees) and of a basal premental fragmentum, also found in most Melittidae. Indeed the proboscis as a whole closely resembles that of melittids (see Figs. 21-2c, 72-1a-c, 74-1b, c) except for the form of the mentum and lorum. Two subantennal sutures also are found in Oxaeinae, as well as in Stenotritidae and a few colletids. The presence of two subantennal sutures is a character that may have arisen independently in different apoid groups, and appears to have been lost in some species of Protandrena (Het-erosarus) in the subfamily Panurginae. The association of Andreninae and Panurginae is further supported by McGinley's (1981) phenetic study of larvae; all the an-drenids except Oxaeinae clustered close together in a nearest-neighbor analysis as well as in a phenogram. (In the phenogram, however, the genus Nomia in the Halic-tidae fell in the same group.)

As indicated in Sections 50 and 51, the genus Alocan-drena, hitherto placed in the Andreninae, cannot be retained in that subfamily and is placed in a separate sub-

Figure 21-1. Inner view of maxilla of Hylaeus basalis Smith, showing the large curved sclerite supporting the galeal comb. (See also Figure 47-1.) From Alexander and Michener, 1995.

family, Alocandreninae. This at least reduces the heterogeneity of the Andreninae. Further study may well support the hypothesis that Andreninae is a paraphyletic group, even after removal of Alocandrena.

The inclusion of the Oxaeinae in the Andrenidae was supported by Michener (1944), but most recent works gave oxaeines familial status until Alexander and Mich-ener (1995) reunited them with Andrenidae as a result of phylogenetic analyses. Rozen (1993b), in a careful study of andrenid larvae, found the Oxaeinae to be the sister group of the South American Euherbstia in the Andreni-nae, thus supporting a position in the Andrenidae. The Oxaeinae have many apomorphies, and phenetically they are very different from other Andrenidae (for maxillae, see Fig. 21-2). The possibility exists that a few characters like the two subantennal sutures overinfluenced the phylogenetic process, and that oxaeines are indeed an isolated

Figure 21-2. Diagrams of inner views of maxillae of three types of S-T bees. a, Halictus quadricinctus (Fabricius); b, Oxaea flavescens Klug; c, Andrena erythrogaster(Ashmead). (The membranous labiomaxillary tube is stippled.) From Michener, McGinley, and Danforth, 1994.

Figure 21-2. Diagrams of inner views of maxillae of three types of S-T bees. a, Halictus quadricinctus (Fabricius); b, Oxaea flavescens Klug; c, Andrena erythrogaster(Ashmead). (The membranous labiomaxillary tube is stippled.) From Michener, McGinley, and Danforth, 1994.

family-level group without close relation to the other An-drenidae. Cane (1983a, b) regarded the Colletidae, Ox-aeinae, and Halictidae as related because of the apomor-phic presence of macrocyclic lactones in their Dufour's glands. These lactones are rare chemicals, unique among bees, and they are lacking in the panurgines and an-drenines. They could have evolved twice or been lost in andrenines and panurgines, but clearly they make the cladograms shown (Fig. 20-1) more questionable, particularly as to the position of the Oxaeinae.

The Oxaeinae are also unusual in that, as in the Halic-tidae, the first (cardines) and second (prementum and stipites) segments of the proboscis are long and slender and the lorum and mentum are much simplified (Mich-ener, 1985a). Depending on one's interpretation of the structures, the mentum is either short and membranous, as in halictids, or sclerotized and indistinguishably fused to the lorum. I prefer the former interpretation, since such complete fusion is unknown in other bees. The loral apron is broad and sclerotized, as it is in some halic-tids, and the galea tapers gradually to a pointed base as in halictids (Fig. 21-2b) (Michener and Greenberg, 1985). The nearly hairless, sclerotized lacinia set against the inner anterior stipital margin, the broadly expanded, convex outer stipital margin, the stiff rather than flexible median region of the galea near the palpal base (the palpus is absent in Oxaea), the fused suspensoria of the premen-tum, and other unique synapomorphies emphasize the distinctness of the Oxaeinae.

In larval characters the Oxaeinae are not close to those of any other bees; they occupy an isolated position among the S-T bees in McGinley's (1981) nearest-neighbor (phenetic) study. Rozen (1964a, 1993b) also pointed out unique features of oxaeine larvae, the striking but probably convergent similarities to larvae of Nomadinae, and the sister-group relationship (based on larvae) to Euherb-stia (Andreninae), as noted above.

Halictidae. The Halictidae form a monophyletic unit that is easily characterized. The position of the lacinia, drawn high up on the anterior surface of the labiomaxil-lary tube, away from other skeletal structures of the mouthparts, is unique among bees and characteristic of all halictids (Fig. 21-2a) (Michener and Greenberg, 1985). Almost as characteristic is the fusion of the hy-postoma and tentorium almost to their anterior ends (Fig. 21-3j) (Michener, 1944, Alexander and Michener, 1995).

The subfamily Rophitinae has sometimes been given family rank (e.g., by Robertson, 1904, as Dufoureidae). Indeed, rophitine genera have traditionally been placed close to panurgine or melittid genera, and Cane (1983a) showed the similarity of Dufourea to those groups in the chemistry of the Dufour's gland secretions. Some rophi-tine features, such as cocoon spinning (except in Conan-thalictus) and associated larval structures (unlike those of other halictids), and the moderately large and usually api-cally lobate S7 of males, are plesiomorphies relative to other halictids. The relatively large labrum and low position of the antennae are apomorphies that, because of variability and the consequent difficulty of concisely describing the character states, were not exploited by Alexander and Michener (1995). Nonetheless, they might have made the Rophitinae a monophyletic group instead of the paraphyletic one that their analysis found. The position and form of the lacinia and the broad covering of the upper part of the middle coxa (Fig. 20-5b) are halictid synapomorphies that unify the family and argue against familial status for the Rophitinae, as does the extensive fusion of the tentorium and hypostoma. The membranous mentum, largely membranous or uniformly lightly sclerotized lorum, and other characters of the family Halictidae are also, of course, found in Rophitinae and were among the features that led Michener (1944) to include rophitines in the Halictidae.

Melittidae. No unique synapomorphy is known for this group, but it is well characterized by a combination of characters (Michener and Greenberg, 1980). The following characters are like those of many other S-T bees and unlike the L-T families: labral base elevated, galeal comb present, stipital comb and marginal concavity absent (concavity and marginal hairs present in Erema-phanta), basistipital process short, seriate hairs of glossa large and diverging, and glossa without specializations characteristic of L-T bees. At the same time, in the following characters, which are synapomorphies of the whole melittid/L-T bee clade, melittids are like most LT bees and generally unlike other S-T families: base of mentum uniformly tapering and curled to attach to lo-rum (Fig. 19-1b, c), loral apron reduced to slender baso-lateral arms so that the lorum is V-shaped, and lower ends of anterior conjunctival thickening of proboscis not separated as a distinct suspensorial sclerite of the premen-tum, as it is in all the families discussed previously and as shown beside the maxilla in Figure 21-2. Thus the members of the Melittidae are S-T bees with certain characters of the L-T bees.

Several characters have been proposed as synapomor-phic for melittids. The almost complete loss of the epis-ternal groove, both above and below the scrobal groove, is such a character. But being also found in a few other ST bees and in megachilids among L-T bees, this is not a strong character. The paraglossae are unusually small, even absent or fused to the suspensoria in some Sambini. But they are larger in the Meganomiinae. Roig-Alsina and Michener (1993) considered a minute sclerite in the cardo-stipital articulation of the maxilla to be characteristic of melittids and absent in other bees. It is also absent in Hesperapis (Dasypodainae), and Alexander and Michener (1995) did not find it a useful characteristic ofmelit-tids. These authors, therefore, did not consider the melittids to have synapomorphies and therefore regarded them as a paraphyletic group from among the members of which the L-T bees arose. Michener (1981a), in a review of Melittidae, also found no synapomorphies for the entire group.

Alexander and Michener (1995) considered the para-phyly of Melittidae s. l. to be established and, like Dan-forth et al. (2006), recognized three families, as follows: Melittidae s. str., Dasypodaidae, and Meganomiidae. They were all characterized as subfamilies of Melittidae by Michener (1981a). I believe, however, that it is premature to recognize these families of melittoid bees in a general work such as this, because further changes in familial classification might well be made in the near future.

Melittidae is already a small family. Melittid phylogeny, using exemplars of all or at least more genera, should be investigated. Alexander and Michener's analysis was inconclusive about the relationship of Macropis and Melitta and did not include Rediviva; it may be that a cladistic classification would require more than three families (or union of melittids, megachilids, and apids into one family, Apidae). I think it is best to await further studies before breaking up the Melittidae s. l., in order to avoid more unnecessary vacillation in family-level classification.

Megachilidae. The phylogeny of the L-T bees was dealt with by Roig- Alsina and Michener (1993). A summary is presented in Figure 20-2. As a group, they are monophyletic but could all be put in one family, the Ap-idae. The group characters that unite the Megachilidae and Apidae are those of the L-T bees, enumerated in Section 19.

The common Megachilidae, i.e., the subfamily Mega-chilinae, constitute a cohesive and easily recognized unit. Its characters include the broad labral articulation, the labrum nearly as long as or longer than broad and more or less rectangular, the presence of a dististipital process, the presence of the metasomal scopa (Fig. 8-7b) in nonparasitic forms, the lack of scopa on the hind legs, the lack of basitibial and usually of pygidial plates, and the lack of pygidial and prepygidial fimbriae. All these are synapo-

morphies not shared with ancestral members of the Ap-idae, although the losses have arisen more than once and are shared with some derived Apidae such as the Apini.

Fideliini has been shown to be a basal megachilid group (Rozen, 1970a); it shares various characters (listed above) with Megachilinae, but has some plesiomorphies relative to both Megachilinae and Apidae, such as the presence of well-developed free volsellae and of apical lobes of S7 of the male arising from a slender base, i.e., from a small disc. S7 thus resembles that of colletids and some other S-T bees, not that ofother megachilids. Compared to two in Megachilinae, the presence of three submarginal cells is a conspicuous plesiomorphy. Fideliini also has apomorphies, such as the short second abscissa of M-Cu of the hind wing, the presence of alar papillae, and the long hairs (not functioning as a scopa) on the hind tibiae of females. In larval characters the fideliines are similar to megachilines. McGinley's (1981) nearest-neighbor analysis showed larvae of Fideliini to be most similar to those of Lithurgini, a connection that accords with the latter's being the megachiline tribe with the most ancestral characters.

Similarities of Pararhophitini to Fideliini are apparent in such characters as the loss of basitibial and pygidial plates (or it may be that, in both tribes, the pygidial plate is extremely expanded) and loss of pygidial and prepy-gidial fimbriae. A remarkable feature of the Pararhophi-

Figure 21-3. Head structures of certain bees. a-d, Ventral views of heads with labrum, mandibles, and proboscis removed; the lateral black areas are the mandibular sockets; the clypeus is shaded, and shows the lower lateral areas bent back in d, not at all in c, and slightly bent back in a and b. a, Andrena mimetica Cockerell; b, Halictus farinosus Smith; c, Xy-locopa tabaniformis orpifex Smith; d, Anthophora edward-sii Cresson.

e-h, Side views of the heads of the same species in the same sequence, showing the protuberance of the clypeus associated with the degree to which the lower lateral areas are bent backward.

i-l, Dissections of the heads of the same species in the same sequence, showing lateral view of the tentorium and the extent of its fusion with the hypostoma of the proboscidial fossa. The location of the articulation of the maxillary cardo in each is indicated by a guide line. From Michener, 1944.

tini is the lack of a distinct scopa; to judge by dry specimens, pollen sticks to various parts of the body, although the hind tibiae do bear possible scopae. Pararhophites is the only nonparasitic megachilid without a sternal scopa. Indeed, inclusion of Pararhophites in the Megachilidae is a relatively new idea and still in need of reexamination (McGinley and Rozen, 1987).

Apidae. This family is here used in the same broad sense that I employed in 1944. Subsequently, and unfortunately, I agreed with various authors that the An-thophoridae warranted familial recognition as distinct from the Apidae s. str., i.e., the corbiculate Apidae. My decision published in 1944 was based on traditional systematic methods and intuitional phylogenetic trees. Roig-Alsina and Michener (1993) came to the same viewpoint after cladistic analysis. It is convenient, however, to have some shorthand terms for groups of tribes within the Apidae. Distinguishing the subfamilies Nomadinae, Xy-locopinae, and Apinae provides for part of this need. Ap-inae, however, is an especially large and diverse group, containing groups that have elsewhere been given family status, resulting in such names as Anthophoridae, Bombidae, Emphoridae, Euceridae, Meliponidae, Xylo-copidae, etc. It is therefore useful to divide the Apinae into (1) the corbiculate Apinae (usually actually called the corbiculate Apidae), for the Euglossini, Bombini, Meliponini, and Apini, i.e., Apidae in the sense of Michener (1990a), and (2) the noncorbiculate Apinae, for all the other tribes of Apinae. The noncorbiculate Apinae, which are equivalent to the Anthophoridae or An-thophorinae of some older classifications, are united by no synapomorphies, are very diverse morphologically, and consist of those L-T bees that do not fall into some other recognized group. These are the reasons why recognition of Anthophoridae is no longer justified.

Finding synapomorphies for the whole of Apidae is difficult. The characters listed in the following paragraphs are synapomorphies for at least the more primitive Apidae; most of them fail in some of the more derived tribes. They separate most Apidae from the Megachilidae except as noted above for the megachilid tribes Fideliini and Pararhophitini.

In most Apidae, the lower lateral parts of the clypeus and often also the lateral parts of the labrum are bent back (Fig. 21-3d). This is also true of the Fideliini. This feature, associated with the protuberance ofthe clypeus (Fig. 21-3e-h), making more space for the folded proboscis, is completely lost in some apids like Xylocopa that have a flat clypeus. There is usually a preapical concavity on the stipes containing a comb, except in most parasitic forms and some others. The episternal groove is absent below the scrobal groove, i.e., it is neither complete nor entirely absent, but forms an arc from the upper episternal groove to the posterior scrobal groove (Fig. 20-5c). Exceptionally, a few Nomadinae (e.g., some Ammobatini) have a long episternal groove as illustrated by Grutte (1935). The feature may not be homologous to the episternal groove of many S-T bees. (Also in the megachilid tribe Fi-deliini the episternal groove is short, as in most Apidae; it is entirely absent in Megachilinae and Pararhophitini.) In apids that have been examined there are only five pairs of ostia in the metasomal part of the dorsal vessel (Wille, 1958); in Megachilinae there are six, and the same is likely to be true of Fideliini and Pararhophitini.

A group that has often been given family status but is here included in the Apidae, subfamily Apinae, is the Ctenoplectrini. Michener and Greenberg (1980) considered it to be a family between the melittids and the L-T bees, i.e., as the sister group to L-T bees, for it has nearly all the features of L-T bees except that the labial palpi are like those of S-T bees. Roig-Alsina and Michener (1993), however, in their phylogenetic study of L-T bees, found that Ctenoplectra fell in the midst of the Apinae, as near to the Eucerini as to any other tribe. They therefore regarded Ctenoplectrini as a tribe of Apinae. Alexander and Michener (1995) showed it once more as the sister group to L-T bees, but their analysis (intended for S-T bees) is probably not meaningful in this respect, for it included few L-T bees and none of those found to be close to Ctenoplectra by Roig-Alsina and Michener.

+1 0

Responses

  • sophie
    Where is Episternal Groove found in Bee?
    10 months ago

Post a comment