Gene tree discordance, rapid diversification and convergence impact phylogenetic inference in Zygophylloideae
The Zygophyllaceae subfamily Zygophylloideae includes seven genera and around 180 species distributed worldwide particularly in arid areas where they show adaptations to drought including C4 and CAM photosynthesis. Phylogenetic relationships of Zygophylloideae have been analysed in several studies using Sanger sequenced markers, but none have been able to resolve all deeper relationships between genera in the subfamily. Here, we use ion semi-conductor sequencing of chloroplast and nuclear ribosomal regions to analyse these relationships and assess why they have proved so challenging to resolve. We sequenced 39 chloroplast genes and nuclear ribosomal 18S, 5.8S, and 26S genes and internal transcribed spacers (ITS) for key species of Zygophyllum sensu stricto, Roepera, Tetraena, Fagonia, Zygophyllum stapffii and monotypic Augea capensis; and Tribulus of the subfamily Tribuloideae as outgroup. We aligned these with data from published chloroplast genomes followed by phylogenetic analyses using Maximum Likelihood and Bayesian approaches, summarising phylogenetic results separately for non-coding introns and intergenic spacers, and for genes that are, or are not, involved in photosynthesis. We combined the chloroplast data with sequences of three highly variable noncoding chloroplast regions used in previous studies of greater numbers of taxa, and similarly the nuclear ribosomal data with more densely sampled ITS sequences and assessed whether adding such data resulted in improved support for consistent topologies. We used a chloroplast sequence matrix to infer a dated phylogeny. Adding data did not always improve phylogenetic resolution which was apparently impacted by non-informative variation in chloroplast photosynthesis related genes as well as by conflicting chloroplast and nuclear signals. Chloroplast genes not related to photosynthesis provide robust support for a topology that is consistent with independent data and may therefore represent a useful species tree hypothesis. The cause of the gene tree conflicts remains unclear: potential explanations include functional constraints in photosynthesis related genes and introgressive hybridization and/or incomplete lineage sorting during an ancient rapid divergence. Further testing will require a broader sampling of lineages with different photosynthetic modes and multiple informative, independent, and selectively neutral gene trees.