In this study, we present the first genetic evidence of the phylogenetic position of Tarsius pumilus, the mountain tarsier of Sulawesi, Indonesia . This mysterious primate is the only Eastern tarsier species that occurs exclusively in cloud forests above 1800 m.a.s.l. It exhibits striking morphological peculiarities—most prominently its extremely reduced body size, which led to the common name of ‘pygmy tarsier’. However, our results indicate that T. pumilus is not an aberrant form of a lowland tarsier, but in fact, the most basal of all Sulawesi tarsiers. Applying a Bayesian multi-locus coalescent approach, we dated the divergence between the T. pumilus lineage and the ancestor of all other extant Sulawesi tarsiers to 9.88 Mya. This is as deep as the split between the two other tarsier genera Carlito (Philippine tarsiers) and Cephalopachus (Western tarsiers), and predates further tarsier diversification on Sulawesi by around 7 Myr. The date coincides with the deepening of the marine environment between eastern and western Sulawesi, which likely led to allopatric speciation between T. pumilus or its predecessor in the west and the ancestor of all other Sulawesi tarsiers in the east. As the split preceded the emergence of permanent mountains in western Sulawesi, it is unlikely that the shift to montane habitat has driven the formation of the T. pumilus lineage.

The basal splits within Syndermata probably date back hundreds of millions of years, and extant syndermatans greatly differ in reproduction modes (parthenogenesis, metagenesis, heterosexuality) and lifestyles (free-living, epizoic, endoparasitic). Against this background, the present investigation aims to reconstruct genome and transcriptome metrics in the last common ancestor (LCA) of Syndermata. We especially annotated nuclear genome assemblies of one representative, each, of monogononts (Brachionus plicatilis), bdelloids (Adineta vaga), seisonids (Seison nebaliae), and acanthocephalans (Pomphorhynchus laevis), S1 and additionally considered previously published data. Present results suggest that the syndermatan LCA possessed a compact nuclear genome of 50–100 Mb with a coding portion of ≥ 40%. Its ~ 17,000 protein-coding genes probably had about 4–6 exons and 3–5 introns on average. Long interspersed nuclear elements (LINEs) should have occupied 4–5% of the ancestral genome. Based on this, independent evolutionary trajectories involving gene loss, genome duplication, and repeat proliferation resulted in greatly varying genomes within the clade. Nevertheless, we identified 2114 clusters containing proteins of all four syndermatan taxa. About 200 of these clusters lacked sufficiently similar counterparts in the platyhelminths Macrostomum lignano and Schmidtea mediterranea. Gene ontologies assigned to these clusters might reference the muscular pharynx, light perception, and transposition, amongst others.