PHYLOGENETIC RELATIONSHIPS OF THE MADAGASCAR PYGMY KINGFISHER (ISPIDINA MADAGASCARIENSIS)

Auk, The, Oct 2005 by Marks, Ben D, Willard, David E

Hierarchical likelihood ratio tests were run using MODELTEST (version 3.06; Posada and Crandall 1998) to choose the appropriate model of evolution for maximum-likelihood (ML) and Bayesian analyses. Maximum-likelihood searches were performed using the heuristic search option with 10 random-addition replicates and TBR branch-swapping. We estimated support for nodes on the likelihood tree using Bayesian posterior probabilities derived from MRBAYES (Huelsenbeck and Ronquist 2001). Bayesian analysis was performed with the following settings. The ML model employed six substitution types, with base frequencies set to the empirically observed values. Rate variation across sites was modeled using a gamma distribution. The Markov chain Monte Carlo search was run with four chains for 3 × 10^sup 6^ generations, with trees being sampled every 1,000 generations (the first 500 trees were discarded as "burnin"). Posterior probabilities for individual clades were based on a 50% majority rule consensus tree of the remaining 2,500 trees (Fig. 2).

Alternative tree topologies were developed using MACCLADE (version 4.0; Maddison and Maddison 2000) and evaluated with the Shimodaira and Hasegawa (SH) test (Shimodaira and Hasegawa 1999). The SH test was used because of its conservative nature in comparison with the similar SOWH test that has been shown to be misleading at times (Buckley 2002). The SH tests were run with full approximation, 1,000 bootstrap replicates, and model of evolution identified by MODELTEST.

RESULTS

An alignment of 755 bp of partial ND2 (362 bp) and complete ND3 sequences (393 bp, including tRNA and spacers) was obtained for all taxa. Levels of uncorrected sequence divergence (P distance) were generally low within the taxa represented by multiple individuals (0-0.6%), with the exception of /. madagascariensis. In this species, levels of divergence ranged from 0% to 1.4% between island localities. Homogeneity of base frequencies across taxa and codon positions could not be rejected for any comparison. All ILD tests failed to reject the null hypothesis of homogeneity between genes and codon partitions within genes, and there were no well-resolved nodes in conflict between separate analyses of the gene fragments.

Parsimony analysis of the combined data set resulted in 30 equally parsimonious trees (length 1,695, consistency index [CI] = 0.406, retention index [RI] = 0.652). Figure 1 shows the strict consensus of these trees with bootstrap support values from 100 replicates above the nodes. The Alcedinidae are supported as a monophyletic group. Within this clade, the species traditionally treated as Alcedo are not monophyletic. Alcedo cyanopecta is within the Ceyx clade, with a highly supported sister relationship to C. lepidus. The African A. quadribrachys, Eurasian A. atthis, and Southeast Asian A. meninting form a separate clade from the African leucogaster and cristata and the Malagasy vintsioides. Although weakly supported by bootstrap resampling, the Madagascar Pygmy Kingfisher is sister to these latter three species and not to Ceyx or Ispidina.