Phylogenetic relationships and host-plant evolution within the basal clade of Halictidae (Hymenoptera, Apoidea)

Bees are among the most important pollinators of angiosperm plants. Many bee species show narrow host-plant preferences, reflected both in behavioral and morphological adaptations to particular attributes of host-plant pollen or floral morphology. Whether bee host-plant associations reflect co-cladogenesis of bees and their host plants or host-switches to unrelated host plants is not clear. Rophitinae is a basal subfamily of Halictidae in which most species show narrow host-plant preferences (oligolecty). We reconstructed the phylogenetic relationships among the rophitine genera using a combination of adult morphology (24 characters) and DNA sequence data (EF-1a, LW rhodopsin, wingless; 2700 bp total). The data set was analyzed by parsimony, maximum likelihood and Bayesian methods. All methods yielded highly congruent results. Using the phylogeny, we investigated the pattern of host-plant association as well as the historical biogeography of Rophitinae. Our biogeographical analysis suggests a number of dispersal ⁄vicariance events: (1) a basal split between North America and South America (most likely a dispersal from South America to North America), and (2) at least two subsequent interchanges between North America and Eurasia (presumably via the northern hemisphere land bridges). Our analysis of host-plant associations indicates that Rophitinae specialized on a closely related group of angiosperm orders in the Euasterid I clade (mainly Gentianales, Lamiales and Solanales). However, there is little evidence of cocladogenesis between bees and plants and strong evidence of host switches to unrelated host plants. Based on our phylogenetic results we describe two new tribes of Rophitinae: Conanthalictini new tribe (including the genus Conanthalictus) and Xeralictini new tribe (including Xeralictus and Protodufourea). The Willi Hennig Society 2007. Bees constitute a monophyletic group of >16 000 described species feeding exclusively on the pollen and nectar of flowers (Michener, 2000). Like many groups of herbivorous insects (Schoonhoven et al., 1998), they show enormous variation in host-plant breadth. Some species, such as Apis mellifera and Halictus ligatus, are widespread and visit many different host plants. Many other species show much narrower host-plant associations. These oligoleges (host-plant specialists) tend to restrict their pollen foraging to phylogenetically closely related host plants (within the same family, tribe, or genus) (Robertson, 1925; Cane and Sipes, 2006). Bee species restricted to a single host-plant species are rare, but some examples exist (e.g., Andrena florea on Bryonia dioica) (Cane et al., 1996; Cane and Sipes, 2006; Schlindwein and Medeiros, 2006). Oligolecty in bees appears to be driven by several factors and is most common in solitary bees with short life cycles. On the contrary, eusocial species tend to be polylectic. Resource abundance is likewise closely related with the evolution of host-plant specificity in bees. Abundant and widespread plants [e.g., Larrea (Zygophyllaceae); Minckley et al., 1999) are known to support many specialist pollinators. Arid regions with highly seasonal rainfall patterns (Minckley et al., 1999) and Mediterranean climate regions (Pekkarinen, 1997) also host large proportions of oligolectic species. *Corresponding author: E-mail address: [email protected] The Willi Hennig Society 2007 Cladistics 10.1111/j.1096-0031.2007.00182.x Cladistics 23 (2007) 1–15 Oligolectic bees often show phenologies tightly coupled to host-plant flowering, suggesting that bees may be tracking the same abiotic factors (such as rainfall) as the host plants (Hurd, 1957; Danforth, 1999b). Host-plant specialization itself appears to be conserved within lineages of bees such that members of some bee genera, tribes or subfamilies are almost exclusively specialists. Such lineages include Melittidae, Fideliini (Megachilidae), Emphorini and Eucerini (Apidae), Panurginae and Andreninae (Andrenidae), Paracolletinae (Colletidae), and Rophitinae (Halictidae) (see Westrich, 1989a,b; Sipes and Wolf, 2001; Michez and Patiny, 2005; Sipes and Tepedino, 2005). Oligolectic bees often show morphological adaptations to collecting and manipulating pollen of their preferred host-plant (Müller, 1996a). Understanding evolutionary patterns of host-plant selection in bees requires that we know (1) the host-plant associations of the bee taxa; (2) the phylogenetic relationships among the bee taxa; and (3) the phylogenetic relationships among the host-plant taxa. Previous studies have analyzed phylogenetic patterns in hostplant usage in two other bee taxa: Müller (1996b) on western Palearctic Anthidiini (Megachilinae: Megachilidae) and Sipes and Wolf (2001) and Sipes and Tepedino (2005) on Diadasia (Emphorini: Apidae). The Sipes and Tepedino (2005) study is exemplary because their study combined accurate, quantitative estimates of host-plant preferences (derived from the analysis of bee pollen loads). Their study demonstrated both that host-plant usage may be highly constrained (i.e., host-plant associations may persist over several speciation events) and that when host-switching occurs, the bees do not necessarily switch to phylogenetically related host plants. Results of the Müller (1996b) study are more difficult to interpret because the western Palearctic Anthidiini are not necessarily a monophyletic group. Because few studies of host-plant evolution in bees have been conducted, it is difficult to know if the patterns detected by Sipes and Wolf (2001) and Sipes and Tepedino (2005) are general patterns applicable to other oligolectic bee clades. In this study we analyzed phylogenetic relationships and host-plant associations among a monophyletic group of oligolectic bees: Rophitinae (Halictidae). Rophitinae is one of the four subfamilies of Halictidae (Michener, 2000; Danforth et al., 2004). Rophitinae appears to be the sister group to the remaining three subfamilies (Nomiinae, Nomioidinae, Halictinae) based on morphology (Pesenko, 1999) and molecular data (Danforth et al., 2004). All species are solitary and they occur primarily in warm xeric regions. The 208 described species of Rophitinae are unique among halictids in that most species are narrow oligoleges (Ebmer, 1984, 1993, 1994; Westrich, 1989a; Baker, 1996; Rozen, 1997; Patiny and Michez, 2006). Many species restrict their pollen collecting to closely related species or genera of host plants [e.g., Rophites algirus is a specialist on Stachys recta and Clinopodium vulgare (closely related within the family Lamiaceae)], whereas other species are apparently monolectic [e.g., Conanthalictus conanthi is a specialist on Nama hispidum (Hydrophyllaceae)]. We know of no species that would be described as a ‘‘generalist’’ or ‘‘polylege’’. Another particularity of the subfamily is its large number of genera (13; 17.5% of the Halictidae) but a rather small number of species (208; 6% of the species), mostly grouped in two genera (Dufourea and Systropha). This pattern may be due to the antiquity of the group, which is estimated to have arisen over 90 My bp (Danforth et al., 2004). A similar pattern is also observed in Andrenidae (for example), in which generic diversity is far higher in Panurginae than in the other two subfamilies (Andreninae and Oxaeinae). In order to elucidate the evolutionary history of host-plant usage in Rophitinae (as well as to analyze the historical biogeography of the group) we conducted a phylogenetic analysis of the 13 recognized genera based on a combined analysis of morphological and molecular data. We analyzed a data set of three singlecopy, nuclear genes that have proven phylogenetically informative in a previous study of generic relationships in Halictidae: EF-1a, LW-rhodopsin and wingless (Danforth et al., 2004). The phylogeny provides the basis for an historical analysis of floral host use in Rophitinae. Materials and methods Our analyses are based on a data set including four partitions corresponding to the sequences of three single-copy nuclear genes plus a set of morphological characters for 34 species (Table 1; Appendix 2). The phylogeographic analysis and the mapping of floral choices refer additionally to an area matrix (describing spatial distributions), a distribution block (describing bee ⁄host-plant associations) and a topology of the plant relationships at the ordinal level (Fig. 5 adapted from Stevens, 2001). A list of the taxa studied is given in Table 1. Apis mellifera was used to determine the reading frame for the sequence alignment and for the delimitation of intron ⁄exon boundaries. This taxon was included in the outgroup together with nine other bee species belonging to the genera Andrena (Andrenidae), Hesperapis (Dasypodaidae), Nomioides, Dieunomia, Curvinomia, Hoplonomia, Lipotriches and Pseudapis (Halictidae) (Table 1).