Plumage-Based Phylogenetic Analyses of the Merops Bee-Eaters
Total Page：16
File Type：pdf, Size：1020Kb
Ibis (2004), 146, 481–492 Blackwell Publishing,Plumage-based Ltd. phylogenetic analyses of the Merops bee-eaters D. BRENT BURT* Department of Ecology and Evolutionary Biology, University of Arizona, Tucson, AZ 85712, USA I review previous systematic work on the family Meropidae and present phylogenetic hypotheses derived from my analyses of colour, pattern and shape variation in 30 plumage regions among species and subspecies in this family. Consistent patterns are seen across shallow portions of the trees. Uncertainty remains concerning the placement of several deep branches within this group’s phylogeny. In particular, the phylogenetic placement of Meropogon forsteni and Merops breweri, M. ornatus, M. hirundineus and M. boehmi remains uncertain. The biogeographical patterns in the resultant trees are similar with either a Southeast Asian or African origin for the family, with most of the early diversification occurring in Africa, and with multiple independent subsequent invasions of non-African areas. The bee-eaters, family Meropidae, are a group of and further behavioural observations. The major areas 26 species of brightly coloured coraciiform birds. of disagreement between the conclusions of these They are distributed throughout the palaeotropics two researchers concern the placement of four forest and southern Eurasia (Fry 2001). Bee-eaters display species. Fry’s Merops gularis and M. muelleri are placed a great range of diversity in several ecological and in a separate genus (Meropiscus) in Boetticher’s clas- behavioural traits, including breeding systems, migratory sification. Fry’s Merops breweri and Meropogon forsteni habits, foraging behaviours and nest-site preferences are placed into the genus Nyctyornis in Boetticher’s (Fry 1969, 1984, Fry & Fry 1992, Burt 1996). These classification. Furthermore, Fry (1969) places the African traits could be subjects of productive comparative forest-dwelling M. breweri among African savanna studies. However, to make a valid comparative study Merops species whereas Boetticher places this species of any of these traits, we must first know the phylo- with Asian forest species. Fry later revised his views, genetic relationships within the group. agreeing that M. breweri’s closest relative may be The relationships among families in the Order Meropogon forsteni (Fry 1984, p. 203). Relationships Coraciiformes are uncertain because of early and rapid among bee-eater species from these previous systematic diversification of these lineages (Cracraft 1981, Sibley studies were based primarily on each author’s intuitive & Ahlquist 1990). Few systematic studies of the family feel for the characters considered. The analyses have been completed previously and no studies of reported here are the first to reconstruct the phylo- the family have utilized molecular data. Von Boetticher genetic relationships within the family using explicit, (1951) primarily used plumage colour and shape cladistic methods. characters to derive a classification for the family. Fry In this study, I reconstruct the evolutionary rela- (1969) used behavioural, biogeographical, ecological tionships among the species in the family Meropidae and plumage colour/shape characters to derive a using data on variation in plumage colour, pattern phenetic classification and tree. Fry (1984) later modi- and shape (see Appendix). Results are presented fied the placement of certain lineages using informa- from two maximum parsimony-based analyses with tion from geographical distributions of subspecies either species or subspecies as terminal taxa. I then examine the biogeographical distribution of each lineage to uncover potential patterns of geographical *Present address: Department of Biology, Stephen F. Austin State University, Box 13003 SFA Station, Nacogdoches, TX 75962- origin and dispersal. Finally, I compare the results of 3003, USA. my analyses with the previous phenetic studies of Email: [email protected] this group by Fry (1969, 1984). © 2004 British Ornithologists’ Union 482 D. B. Burt that transitions between similarly coloured character METHODS states are easier than they are to other coloured character states (e.g. orange-red to red easier than Specimens and plumage characters either is to blue). Two data matrices were used in the study: one with Three basic levels of transition rules were used in species as terminal taxa and one with subspecies as construction of step matrices. It is assumed that each terminal taxa. Single specimens from each of the level represents an increasingly difficult transition nominate subspecies were first examined to docu- between colours. The easiest level of transition was ment the basic plumage character patterns of each that between green and green with blue tips, chestnut species. Multiple individuals of each sex from each and red chestnut, and between dark greenish-brown of the geographically defined subspecies (Fry 2001) and light greenish-brown. The next level of transition were then examined to document the extent of char- linked all greens together and these to greens mixed acter variation within and among taxa. with other colours. Also in this transition level were Thirty plumage regions were scored for colour, similar rules for blues, oranges and reds. The most pattern and shape using museum study skins and spirit difficult level of transition was that between more specimens (see Appendix). Data from spirit specimens distant colours. No attempt was made to construct were excluded when specific plumage colours had more detailed step matrices for relationships between evidently faded. Evidence of fading was derived by colours of more distant spectral affinity. Although comparison of plumage colours from study skins and complex, these step matrices are actually quite illustrations in Fry and Fry (1992). Detailed structural conservative while still utilizing much of the informa- components of feathers consistently associated with tion available in each multistate plumage character. carotenoid pigments suggest homology of these colours Character step matrices can be accessed in the data (Fry 1969). Plumage characters were retained only files at the TreeBASE website (www.treebase.org/ if they were invariant within subspecies in analyses. treebase. S 1021). Subspecies identified as distinct by Fry (2001) were An alternative method of extracting the same combined if they did not differ in plumage characters information is to break each multistate plumage (Nyctyornis athertoni athertoni = N. a. brevicaudata, character into multiple, individual characters. That Merops hirundineus hirundineus = M. h. furcatus, M. method was not used in this study because it would pusillus meridionalis = M. p. argutus, M. bullockoides artificially inflate the number of independent char- bullockoides = M. b. randorum, M. orientalis cyanophrys acters in each data matrix. That is, certain characters = M. o. muscatensis, M. o. viridissimus = M. o. flavoviridis, would contain character states that could not exist, M. superciliosus superciliosus = M. s. alternans). M. given any character state in certain other characters. nubicus and M. nubicoides are considered as distinct An example illustrates how this problem could occur. phylogenetic species owing to their disjunct breeding The step matrix for the forecrown colour has three distributions and plumage differences. Sexual dichro- basic transition classes. The first transition class matism is rare in bee-eaters, but in cases where it does represents slight changes among species with green in occur, the dichromatic characters were coded as the forehead. In some cases the basic green forehead polymorphic for that taxon. Data matrices can be is only slightly modified by the addition of faint accessed at the TreeBASE website (www.treebase.org/ blue tips and this slight difference is indicated by a treebase. S 1021). The subspecies data matrix and cost of only a single step. Other transitions from the character descriptions are documented in the basic green involve more complex changes such Appendix. as the addition of other colours (e.g. orange, brown) throughout the forehead region. These more com- plex changes result in an increased cost of two Character step matrices steps. Finally, some species have either a chestnut or Each plumage region typically has a large array of reddish-chestnut forehead, probably indicating that colours when examined across all geographically these species are closely related, and therefore transi- distinct subspecies. Each plumage region is repre- tions between these two colours cost only a single sented here as a single character with several colour step in the matrix. All other transitions are between or shape character states. Each multistate character colours of more distant spectral affinity and therefore is linked to character-specific step matrices to retain cost three steps. It is possible to represent these additional information. Character step matrices assume basic colour changes in the forehead as different © 2004 British Ornithologists’ Union, Ibis, 146, 481–492 Plumage-based phylogenetic analyses of Merops bee-eaters 483 characters (e.g. ‘basic green or not’, ‘green with other branch swapping in each analysis. The starting trees colour or not’, ‘chestnut or not’), but the characters are for branch swapping were derived by stepwise addi- not then independent. That is, if the forehead is chestnut, tion of taxa. The number of TBR islands represented it cannot be green. Although the independence of in the MPT set and the number of replicates in