Missing data and model selection in phylogenomics: A re-evaluation of Cicadomorpha (Hemiptera: Auchenorrhyncha) superfamily level relationships

Cicadomorpha, the most diverse infraorder of hemipteran insects, is a group of plant sap-sucking insects consisting of three monophyletic superfamilies: Cercopoidea, Cicadoidea, and Membracoidea. This study aims to resolve the controversy surrounding the phylogenetic relationships between these superfamilies, specifically whether Cercopoidea is more closely related to Cicadoidea (topology H1) or Membracoidea (topology H2). To tackle this issue, we assembled four matrices, including three amino acid datasets and one nucleotide dataset, and performed comprehensive phylogenomic analyses using both coalescent and concatenation methods. Our results showed that the amino acid matrix with low ratio of missing data-based analyses strongly supported topology H2, indicating that Cercopoidea is sister to Membracoidea. However, the other three matrices yielded conflicting results. Interestingly, when site-heterogeneous models were used, the results flipped, and the datasets that previously supported topology H1 supported topology H2. Further analyses, including model comparisons, tree topology tests, and phylogenetic analyses without an outgroup, also supported the conclusion that Cercopoidea is more closely related to Membracoidea. The study highlights the importance of modeling among site compositional heterogeneity in phylogenomic analyses to mitigate systematic errors. Additionally, we noted that datasets with high missing data ratios may lead to conflicting phylogenetic relationships. Our study provides strong evidence that Cercopoidea is sister to Membracoidea and highlights the importance of careful methodological considerations in phylogenetic analyses.