Bee Fossils and the Antiquity of the Bees
The fossil record can provide important information into the morphological diversity, the historical biogeographic distribution, and the antiquity of organisms. However, the fossil record has its limitations, the most important of which is that it can only provide us with minimum ages of groups and cannot necessarily establish actual ages. The fossil record of bees is relatively poor. Alexander & Michener (1995) commented as follows: "the fossil record [of bees] is extraordinarily fragmentary and biased toward taxa that collect resin for nesting purposes, and thus occasionally are trapped in it and fossilized in amber." The vast majority of bee fossils are in amber, and virtually all are Eocene or later in age (Rasnitsyn & Michener 1991, Engel 2001b).
Among the most important fossil bees is the presumed oldest fossil bee, Cretotrigona prisca, from New Jersey amber (Michener and Grimaldi 1988a,b; Engel 2000a). This is a fascinating but anomalous fossil for several reasons. First, it is a member of a relatively derived group of bees, the corbiculate tribe Meliponini nested well within the LT subfamily Apinae (Roig-Alsina & Michener 1993). Second, it is of uncertain origin and age, casting doubt on whether it can be adequately used to date the antiquity of bees. Because the precise stratigraphic location of the amber containing this bee is not known, dating of this fossil has been indirect (by comparison with the chemical composition of other New Jersey ambers [Grimaldi 1999]). While initially presumed to be 80 Ma in age (Michener & Grimaldi 1988a, b), it has since been estimated to be 70 Ma (Grimaldi 1999) and 65 Ma (Engel 2000a) in age. Furthermore, whether it is Cretaceous at all has been questioned by Rasnitsyn & Michener (1991). If this fossil is indeed from the late Cretaceous, it suggests that, far from being in their earliest stages of evolution, the bees had already undergone significant diversification by the end of the Cretaceous.
Diverse fossil faunas of bees also exist from the Dominican (approximately 23 myBP; Engel 2001b) and from the Baltic amber (approximately 42 myBP; Engel 2001b) as well. These two localities provide important insights into ancient bee diversity and biogeographic distributions. In the Dominican amber there are representatives of five families (Colletidae, Halictidae, Andrenidae, Megachilidae, and Apidae; summarized in Engel 2001b). The Baltic amber deposits include representatives of four extant families (Halictidae, Melittidae, Megachilidae, and Apidae) and one extinct family (Paleomelittidae). The Baltic fossil fauna is apparently biased toward resin-collecting corbiculate (LT) bees, with only a few ST bees known (Engel 2001b). The Baltic amber deposits provide important insights into the evolution of corbiculate bees and the transition from solitary to eusocial nesting (Engel 2001a).
The fossil record of spheciform wasps, a paraphyletic group within which bees arose, provides some additional information relevant to the age of bees. The oldest spheciform wasp fossils are mid Cretaceous in age (reviewed in Grimaldi [1999] and Bohart & Menke [1976]). However, as in bees, the oldest fossil spheciforms represent derived taxa. The oldest spheciform fossils have been placed in the Pemphredonini, a tribe which Melo's (1999) cladistic analysis places in a derived position within the Crabronidae (sensu Melo, = Sphecidae sensu Bohart and Menke [1976]), the putative sister group to bees. Thus, both bees and crabronids must have been diversifying in the mid Cretaceous, but the bee-crabronid split must have occurred considerably earlier than this. Overall, considering the bee and spheciform fossil record one is forced to the conclusion that the fossils currently available significantly underestimate the age of both these groups.
The antiquity of bees could be potentially estimated from the more complete fossil record of the angiosperm (flowering) plants. A substantial literature exists on the early evolution of angiosperm flowers (e.g., Crepet 1996, Friis & Crepet 1987, Thien et al. 2000) and the evolutionary history of flower-associated insects, such as flies, moths and butterflies, wasps, and bees (Grimaldi 1999, Labandeira 1997, Engel 2001b). Below, we review the literature on fossil angiosperms and conclude that there is little evidence from the fossil record alone that can establish an adequate estimate for the true age of the bees, the nature of early bee/angiosperm associations, and the timing of bee/angiosperm evolution.
Careful studies of fossil flowers have provided important insights into the morphology and evolution of early angiosperms. The earliest unambiguous angiosperm fossils are early Cretaceous in age (Crane et al. 1995). Fruiting axes of Archaefructus from China were originally inferred to be from the late Jurassic (Sun et al. 1998), but more recently have been dated as early Cretaceous (probably Barremian; Barrett 2000). Sun et al. (2002) described a complete fossil Archaefructus, established that this group forms the sister group to all other angiosperms (Archaefructaceae) and estimated the age of these fossils to be 124.6 myBP. Oddly, these plants appear to be aquatic. Fossils indicate that flowers from the lower and mid Cretaceous were very small (1mm to 2 cm diameter), had small or no petals, and offered no reward other than pollen to prospective pollinators. A fossil flora from Portugal (Friis et al. 2000,) and one from New Jersey (Crepet et al. 1991) indicate that both bi- and unisexual angiosperm flowers existed in the early Cretaceous, suggesting that insect pollination characterized the earliest angiosperms (Dilcher 2000). By the Albian (112 Ma) and Cenomanian (97 Ma) flowers began to show traits indicative of insect visitation, including well developed sepals and petals, filamentous stamens, bisexuality, and pollen and nectar rewards (Crepet 1996, Dilcher 2001). By mid-Cretaceous (Turonian [90 Ma] and Cenomanian [97 Ma]) flowers began to show more specialized traits associated with insect (especially bee) pollination, including well developed corolla tubes, clawed petals, pollen clustered into polyads with viscin threads, elongate styles, hidden nectaries, and bilateral symmetry (Nixon & Crepet 1993; Herendeen et al. 1994; Crepet 1996; Crepet & Nixon 1998; Gandolfo et al. 1998; Dilcher 2000). These Turonian flowers have been attributed to a diversity of angiosperm taxa, including Magnoliidae, Hamamelididae, and higher taxa such as Rosidae, Dilleniidae, and Asteridae (Ericales, sensu lato).
Previous authors have generally emphasized the importance of non-bee pollinators prior to the Turonian (90 Ma) (Crepet et al. 1991, Grimaldi 1999, Labandeira 1997, Thien et al. 2000, Dilcher 2000). Important flower visiting groups whose fossil records extend well into the Cretaceous include vespid wasps, flies in the families Stratiomyidae, Nemestrinidae, Acroceridae, Bombyliidae, Asilidae, and Syrphidae, and numerous families of basal Lepidoptera (Grimaldi 1999). However, few authors have provided strong evidence that bee pollination could not have existed prior to this time. Today, many bees of medium to large size visit and pollinate small, shallow, relatively simple flowers (e.g. Andrena, Colletes, and Apis on Salix and Acer to name just a few examples, Krombein et al. 1979), and most bees that are generalist feeders, regardless of body size, visit flowers having a large range of sizes. Numerous members of the ST bee families have short tongues used for drinking nectar from shallow, cup-shaped flowers (including Hylaeinae, Euryglossinae, Colletinae, some Melittinae and numerous Halictidae). Furthermore, contemporary pollination relationships are diffuse (rarely involving a one-to-one relationship between plant and pollinator) and exceptions to "floral syndromes" abound (Robertson 1928; Tepedino and Stanton 1980; Janzen 1980; Herrera 1988; Waser et al. 1996; Johnson and Steiner 2000) making it difficult to predict from flower morphology alone the types of insect visitors. Exceptions to these same rules surely existed in the Mesozoic.
Additional insights into the timing of bee pollination can be gained from mapping extant pollinator relationships onto the phylogeny of the angiosperms (Soltis et al. 1999, Qiu et al. 1999; see the Missouri Botanical Garden’s Angiosperm Phyogeny web site for the most up-to-date information on this subject). Alone, this approach is fraught with problems. First, pollination syndromes are highly variable within higher categories and it is difficult to assign a state unambiguously to angiosperm families or orders. Second, there is no guarantee that plant/pollinator relationships have remained the same from the origins of the angiosperms to the present. Nevertheless, specialized relationships between bees and angiosperms are not likely to have existed prior to the common ancestor of the eudicots because extant magnoliids (monocots, Winterales, Laurales, Magnoliales, Chloranthales, Piperales, etc.) are, for the most part, not bee pollinated (Thien et al. 2000). The eudicots have recently been estimated to be between 147 and 131 million years old based on combined fossil and DNA evidence (Wikström et al. 2001).
Finally, molecular estimates of angiosperm origins push the date of the angiosperm/gymnosperm split back to approximately the Triassic/Jurassic boundary (210 Ma; Wolfe et al. 1989, Laroche et al. 1995, Goremykin et al. 1996, and Sanderson 1997). Needless to say, this is considerably earlier than the oldest fossil angiosperms (early Cretaceous, 145 Ma) and casts even more doubt on the belief that bees necessarily arose in the mid- to late Cretaceous.
We are using the existing bee fossils in combination with the DNA sequence data sets for the bee families and recently described Bayesian methods (e.g., Thorne et al. 1998, Kishino et al. 2001) to estimate the antiquity of important branches in the phylogeny of bees (Danforth et al. 2004). These methods are pushing back the date of bee origins and we hope to have a better understanding of bee antiquity in the near future. Establishing the antiquity of bees will help us understand more about the importance of bee pollination in the origin and diversification of the angiosperm plants.