Presentation Detail

SSB Symposium: Ernst Mayr Symposium

Narula, Nitish [1], Houde, Peter [1].

Utility and Limitations of Indels as Phylogenomic Characters.

Insertion/deletion events (indels) should be less susceptible than nucleotide characters to homoplasy because they exist in greater character space. Therefore, they should be superior as phylogenetically informative characters and for documenting hemiplasious character conflict due to incomplete lineage sorting (ILS). In collaboration with the Avian Phylogenome Project led by BGI and G10K, we report the phylogenetic analysis of 48 avian genomes using insertion-deletion characters. We scored indels in orthologous introns, exons, and ultra-conserved elements (UCEs), and aligned chromosomes using simple gap coding. We performed parsimony (using TNT) and likelihood (using RAxML) analyses on each of the partitions and on concatenated introns, exons, and UCEs (4.6 million indels). Resulting parsimony trees were better resolved and more strongly supported than likelihood trees despite fewer parsimony informative characters than total characters. Long indels were less informative than inclusive data sets. In general, phylogenetic results were more highly supported by bootstrap than those obtained from independent analyses of nucleotide characters of like partitions (e.g., intron indels compared to intron nucleotides), with which they were mostly congruent. However, some strongly supported indel differences were found to be incorrect, so we mapped indels greater than 100 bp onto indel-optimized trees to investigate their macroevolutionary pattern. Character conflict was evident at all nodes; but, surprisingly, homoplasious indels were distributed in disparate branches of the tree, seemingly uncharacteristic of hemiplasy. Although real hemiplasy is evident, character conflict probably results from error in automated alignment and inadequacies of simple gap coding. Manual curation can improve on automated alignment and indel scoring, but it is impractical for genome-scale analyses. Thus, while genome-scale data provide unprecedented opportunities for phylogenetic analysis, better automated procedures to handle massive data sets will be required before their potential can be fully realized.

1 - New Mexico State University, Department of Biology, 1200 S. Horseshoe, Foster Hall, MSC 3AF, Las Cruces, NM, 88003, USA

Keywords:
indels
Gaps
phylogenomics.

Presentation Type: Symposium Presentation
Session: 210
Location: Ballroom 2/Cliff Lodge
Date: Sunday, June 23rd, 2013
Time: 8:45 AM
Number: 210002
Abstract ID:210
Candidate for Awards:Ernst Mayr Award