PHYLOGENETIC RELATIONSHIPS OF THE MADAGASCAR PYGMY KINGFISHER (ISPIDINA MADAGASCARIENSIS)

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

Here, we investigate the phylogenetic relationships of the Madagascar Pygmy Kingfisher using sequences from the mitochondrial NADH dehydrogenase subunit II (ND2) and NADH dehydrogenase subunit III (ND3) genes. In the process, we hope to solve one piece of the much larger puzzle of Madagascar endemic bird relationships.

METHODS

Sibley and Monroe (1990) split the kingfishers into three families: Alcedinidae, Cerylidae, and Dacelonidae. We designed the taxon sampling here to allow us to investigate the generic placement of the Madagascar Pygmy Kingfisher as well as to assess generic limits of some other controversial groupings within the Alcedinidae. Following Sibley and Monroe (1990), we sampled 4 of the 11 species of Ceyx, 7 of 10 Alcedo species, and all 3 Ispidina species (including Myioceyx). Genera representing Cerylidae and Dacelonidae were included as outgroups to the Alcedinidae, as were several other Coraciiform species (Table 2).

We sequenced 755 base pairs (bp) of mitochondrial DNA (mtDNA) from two genes (complete ND3 and a fragment of ND2) from 1-6 individuals of each of the 28 study taxa using the primer sets L5215 and H5578 for ND2 (Hackett 1996) and L10755 and H11151 for ND3 (Chesser 1999). Sequences obtained using the ND2 primers were 362 bp long, and those obtained with the ND3 primers were 393 bp long. Total genomic DNA was extracted from muscle or toe pads using the reagents and protocols provided with the QIAamp Tissue Kit (Qiagen, Valencia, California). All polymerase chain reactions (PCR) followed the protocols outlined in Marks et al. (2002); automated sequencing protocols followed manufacturer recommendations (ABI Big Dye, version 2.0, Applied Biosystems, Foster, California) and were run on an ABI 377 automated DNA sequencer. We sequenced DNA in both directions and verified and aligned using SEQUENCHER (version 3.1.1, Gene Codes, Ann Arbor, Michigan). All of the mtDNA sequences collected for the study have been deposited in GenBank (accession numbers AY998882-AY998927 for ND2 sequences and AY998928-AY998974 for ND3 sequences).

Phylogenetic analyses were conducted with PAUP* (version 4.0b4a; Swofford 2000) and MRBAYES (version 2.1; Huelsenbeck and Ronquist 2001). Prior to phylogenetic analysis, we used PAUP* to evaluate base composition of the ND2 and ND3 sequences for variability among taxa and among codon positions using a chi-square analysis of base frequencies across taxa, examining each codon position separately. We performed partition homogeneity tests (ILD statistic of Farris et al. 1994, 1995) to determine whether the two gene fragments could be combined in tree searches. All ILD tests were run excluding uninformative characters (Cunningham 1997), using a heuristic search with 10 random-addition replicates and tree bisection-reconnection (TBR) branch-swapping with 100 null-distribution replicates, and equal weights among characters. Barker and Lutzoni (2002) suggested that the ILD test can be misleading, resulting in false rejection of homogeneity across partitions and lacking power to detect certain types of heterogeneity. Therefore, we further assessed combinability of the two gene fragments by analyzing each fragment separately under a parsimony framework, looking for conflict between topologies at wellsupported nodes. Each individual gene analysis was run using the heuristic search option with 10 random-addition replicates and TBR branchswapping. Nodal support was assessed by nonparametric bootstrapping (100 pseudoreplicates, each with 10 random-addition replicates and TBR branch-swapping). After data exploration, parsimony tree searches with the combined data were run as described above.