PDF《Blackwell Publishing LtdOxford,》

Tkalc , 1972) but largely disregarded. Previous research on bumble bee phylogeny was restricted in scope and utility by either: (1) the appli- u o u o cation of phenetic methods to infer phylogenetic rela- tionships;

(2) relatively sparse species sampling;

or (3) insuf?cient numbers of informative characters. Exam- ples of phenetic applications include the investigation by Ito (1985) of male genitalia and analyses of allo- zymes in regional taxa (Pekkarinen, 1979;

Pekkarinen, Varvio-Aho &

Pamilo, 1979;

Obrecht &

Scholl, 1981;

Pamilo, Pekkarinen &

Varvio, 1987). Williams (1985, 1994) applied parsimony to morphological data to address the higher-level relationships among subgen- era, but the available characters led to considerable lack of subgeneric resolution. DNA analyses of rela- tionships have been restricted to regional fauna (anal- yses of European taxa: Pedersen, 1996, 2002) or have included few species (assessment of Pyrobombus monophyly, 10C12 species overall: Koulianos, 1999;

analysis of subgenera based on

19 species: Koulianos &

Schmid-Hempel, 2000). Kawakita et al. (2003, 2004) estimated relationships among many of the subgenera using sequence data, but their sampling of only

76 species limits their ability to draw robust conclusions about the monophyly of groups and the evolution of traits. Comprehensive molecular studies include examina- tion of relationships within the large New World subgenus Fervidobombus, including DNA sequences and morphology, by Cameron &

Williams (2003) and sequence analysis of

37 of the

43 recognized Pyrobom- bus species by Hines, Cameron &

Williams (2006). In the present study, we report on the full hierarchy of Bombus relationships for most of the world fauna, including rare and potentially endangered species such as Bombus franklini and Bombus occidentalis (Thorp, 2005), exploiting the large number of charac- ters available from DNA sequences collected from multiple genes. MATERIAL AND METHODS TAXA EXAMINED A total of

218 Bombus taxa, representing most of the species from all

38 subgenera listed by Williams (1998), was sampled for analysis and their locality, collector, and voucher numbers recorded (Table 1). Intraspeci?c variants and taxa of controversial species status (Williams, 1998) were included when possible (Table 1). Outgroups were represented by exemplars selected from each of the........