INTRODUCTION

The honey bees (genus Apis Linnaeus, 1758) are perhaps the most recognizable of all insects. Together with a few others animals (e.g., Homo sapiens Linnaeus, Drosophila melanogaster Meigen, Caenorhabditis elegans [Maupas]), the common honey bee has become one of the most heavily studied of taxa in the animal kingdom. Despite this intense scrutiny, our understanding of honey bee systematics and evolution is remarkably confused and obscured. Apis is the only genus in the corbiculate bee tribe Apini (Engel, 1999; Michener, 2000), although some past authors have elevated the subgenera Megapis, Micrapis, and Synapis to generic rank (e.g., Ashmead, 1904; Maa, 1953; Wu and Kuang, 1986, 1987). Species of the corbiculate tribes are more frequently encountered as fossils than any other lineage of bees, particularly when the common Dominican amber species, Proplebeia dominicana (Wille and Chandler), is included in such a survey. Even if P. dominicana is excluded, however, the corbiculate apines presently outnumber other bee tribes in the fossil record. This preponderance is perhaps partly owing to the eusocial behavior of these species, with typically large colony sizes. Accordingly, the fossil record of these lineages is better understood than most other bee taxa. Despite this relative clarity, the geological history of Apini is obscured owing to the paucity of critical material from the Oligocene and Miocene of Asia, the region and period of time from which honey bees presumably originated and initially radiated, as well as the complete absence of data from the Pliocene.

Herein I present the description of a new Miocene species of honey bee from eastern Asia. The specimen described below originates from the middle Miocene Chôjabaru Formation of Iki Island, Japan. The fossil is not only significant because of its geographical location and age but also in that it is a representative of the giant honey bees and is quite similar to modern Apis dorsata Fabricius. Indeed, this is the first fossil record for crown-group Apis outside of relatively modern records of A. mellifera Linnaeus in Pleistocene copal (Foord, 1890; Cockerell, 1909; Engel, personal obs.) and Quaternary remains of honey bee combs (Stauffer, 1979). Morphological terminology for the description follows that of Engel (2001b).

Owing to the low specific diversity of the genus and the relative homogeneity among honey bee species, the recognition of subgenera in Apis should perhaps be abandoned, including the Recent subgenera, as has been advocated by some authors (e.g., Michener, 2000). I have, therefore, not formally placed the fossil as to subgenus, but for convenience have indicated in the diagnosis to which group it would belong in such a system (table 1). It should be noted that in the diagnoses presented for the subgenera of Apis (Engel, 1999) the shape of the third submarginal cell was mistakenly presented as “angle of posteroapical margin of first submarginal cell less than 45°”. In each diagnosis these should have read “angle of posteroapical margin of third submarginal cell less than 45°” (the standard wing metrics for Apis have been thoroughly explained by Ruttner [1988], Alexander [1991], and other authors). This correction should be kept in mind when attempting to identify honey bee subgenera and supraspecific clades (vide etiam Engel, 2001c).

SYSTEMATIC PALEONTOLOGY

DISCUSSION

Honey bees are ubiquitous animals today, being important in both agricultural and natural ecosystems. The tribe consists of distinctive bees that perhaps arose and first diversified in the Early Oligocene, after the Eocene–Oligocene climatic transitions (fig. 4). Certainly the lineage that gave rise to Apis had been around for millennia prior to this, as evidenced by the presence of numerous corbiculate bees allied to Apini in the Eocene (Meliponini alone is known from as far back as the Maastrichtian; Michener and Grimaldi, 1988; Engel, 2000, 2001b). However, stem-group Apini have not yet been discovered and bees with the synapomorphies we use to circumscribe the group first appear in the Late Oligocene. Interestingly, the middle Eocene Baltic amber fauna contains a group of taxa, placed in the extinct tribe Electrapini, which may represent a stem-group lineage relative to the living tribes Apini and Meliponini (Engel, 2001b). This group contains a mix of Bombus- or Apis-like bees (the genus Thaumastobombus in particular resembles primitive Apis species). The Baltic amber fossil figured by Poinar (2003) and attributed to Apis is a misidentification (the figure very clearly shows a nice specimen of an electrapine). It is perhaps from among taxa similar to the electrapines, which are unknown after the Eocene, that the honey bees as we understand them arose.

As previously discussed, definitive Apis likely originated in the Early Oligocene and radiated throughout Eurasia at that time. The distribution of the group at this time was perhaps limited by temperature extremes since species were presumably all open-nesting (i.e., building exposed combs; cavity-enclosed combs are an apomorphic feature of the mellifera group; Engel and Schultz, 1997; Engel, 1998). The clade of living species perhaps originated in the uppermost Oligocene or lowermost Miocene (revised from the Late Miocene hypothesis of Engel, 1998, 1999, based on the current fossil) (fig. 4). The development of cavity-nesting behavior in the latest Miocene or earliest Pliocene enabled this lineage to invade more cool temperate areas (e.g., Europe and northern Asia). Competition among species perhaps led to relatively monospecific honey bee cultures in any given area and resulted in a low, modern specific diversity.

Contrary to my earlier assertion (Engel, 1998), specific diversity in Apis may have indeed slightly declined over geological time as originally suggested by Culliney (1983) (table 1; fig. 4). At present, such a decline is barely evident, but assuming that we have scarcely begun to sample the past diversity of Apis from Asia (most fossils come from Europe), then additional fossil taxa will undoubtedly be discovered there. Thus, Apis of the past would become more diverse than the genus at present (presently the number of species are nearly equal). If such a pattern of declining diversity can eventually be further supported by new paleontological discoveries, then such a decline is similar to that which has been noted generally among the highly eusocial corbiculate bee tribes (Engel, 2001a, 2001b). Aggressive competition in honey bees may have contributed to a decline in specific diversity over geological time. Certainly advanced eusocial species such as A. mellifera are known to influence the foraging patterns, abundance, and fecundity of other species (e.g., birds and other bees, particularly other eusocial bees) through competition for resources (Eickwort and Ginsberg, 1980; Paton, 1993; Buchmann, 1996; Sugden et al., 1996; Paini and Roberts, 2005). Indeed, A. mellifera workers are aggressive enough to steal pollen directly from foragers of native bee species—so-called cleptolecty (Thorp and Briggs, 1980; Jean, 2005). In fact, introduced populations of A. mellifera have been observed to aggressively displace native populations of A. cerana (e.g., Sakagami, 1959). Intensely competitive interactions among honey bee species over expanses of geological time is possibly one factor that, in combination with factors such as global climate change, may have altered Apis diversity.

Nel et al. (1999)2 have presented excellent documentation of wing variation in what appears to be the common Oligocene–Miocene European species A. henshawi Cockerell (A. aquisextana Nel et al. and A. aquisextusensis Nel et al. are both synonyms of A. henshawi; Engel, 1999; vide appendix 1) and the Miocene species A. armbrusteri Zeuner. From their article one would be impressed by a drastic decline in honey bee species since the Oligocene–Miocene. These authors have, however, followed the morphometric system of Ruttner (1988) and thereby arrived at several “species” denoted with letters as “A– J”.3 As has been thoroughly demonstrated by the work of Hepburn and Radloff (1998) and Hepburn et al. (1998) for honey bees in Africa, such morphometric measures of wing venation do not characterize natural entities, and clustering systems based on other characters produce entirely incongruent groups of subspecies or species. Thus, morphometric clusters in Apis generally do not correlate with other morphological attributes, mitochondrial DNA haplotypes (Moritz et al., 1994, 1998), or general biological attributes such as presence versus absence of thelytokous parthenogenesis or ovariole number (Hepburn and Crewe, 1991). The recognition of taxa, specific or subspecific, in Apis based solely upon morphometric clusters of wing venation is highly suspect (summarized in Hepburn and Radloff, 1998; Hepburn et al., 1998; Hepburn, 2000). Indeed, the use of highly variable characters (e.g., drone flight times, metasomal banding coloration) for recognition of honey bee species in the absence of fixed, diagnostic apomorphic traits (or diagnostic combination of apomorphies) is not recommended. Remarkably, only a few traits of wing venation appear to be truly diagnostic for species-group taxa in Apis.

The recognition of species in living Apis is rife with difficulties, making paleontological studies all the more challenging for this group of bees. Honey bees exhibit dramatic ranges of variation over huge geographical distributions, and peripheral isolation may be the primary mechanism of speciation in the genus (thereby leaving large paraphyletic, ancestral populations difficult to characterize relative to peripheral derivatives). A picture of relatively recent honey bee history (e.g., affects of Pleistocene climatic and oceanic changes) is coming into focus simply through the study of present-day populations. A more ancient perspective will certainly reveal a rich and complicated evolutionary history. Further paleomelittological work will be required, particularly in southern Asia, to further resolve the origin, radiation, and possible post-Miocene decline (if any) of the honey bees.