Ambigolimax, Lehmannia and other limacid slugs (Gastropoda: Stylommatophora); what their DNA tells us about phylogeny, taxonomy and phylogeography

ABSTRACT Phylogenetic trees of the slug family Limacidae were constructed using two nuclear markers—internal transcribed spacer 1 and the single-copy intron 8 of the embryonic lethality and abnormal visual system gene (ELAVI8)—and two mitochondrial ones—cytochrome c oxidase subunit I (COI) and 16S ribosomal rRNA. Two Eumilax species formed the most basal clade, supporting the distinction between subfamilies Eumilacinae and Limacinae. The next most basal clade consisted of Svanetia and a monophyletic Gigantomilax. Limax, Limacus and Turcomilax formed a clade that was the sister to the remaining genera considered. Of these, Lehmannia, Malacolimax and Weltersia formed a clade that was the sister to Ambigolimax and to Bielzia. ELAVI8 was particularly good at resolving the basal parts of the tree, but the mitochondrial markers were necessary for more distal parts. Despite their lacking a rectal caecum, Malacolimax wiktori and an undescribed North African species belong in Ambigolimax, as does A. melitensis. We argue for synonomizing Lehmannia islandica and L. janetscheki with L. marginata. Intraspecific variation in COI was often high (typically over 5%), and Malacolimax mrazeki appeared paraphyletic. We plotted the worldwide distributions of the 26 COI haplotypes of A. valentianus (based on 193 sequences) and 17 haplotypes of A. parvipenis (110 sequences). In both these invasive species, some haplotypes occur widely, but geographical restrictions are also apparent. North America is impoverished in haplotype diversity, suggesting a colonization via Europe. The low diversity of A. valentianus in Spain contradicts the assumption that it is a native there; probably Ambigolimax species derive from North Africa. Ambigolimax waterstoni and Lehmannia rupicola are newly recorded for Corsica based on reinterpretation of sequences in the BOLD database.