Repeated colonization and hybridization in Lake Malawi cichlids

Through adaptive radiation, ancestral species rapidly diversify into multiple species with different ecological adaptations. The haplochromine cichlid fishes of the East African Great Lakes are considered classic examples of adaptive radiation, but our understanding of the evolutionary origins of these radiations has been limited by inadequate taxonomic and genomic sampling [1,2]. Perhaps the largest of these radiations is from Lake Malawi, estimated to contain between 500 and 800 endemic species. Surprisingly, its monophyly — the origin from a single ancestral species — has never been critically tested. This is because river populations which could have seeded the radiation, with one very limited exception [3], have never been included in phylogenetic reconstructions. Moreover, phylogenies have relied heavily on mitochondrial DNA (mtDNA), which can be a misleading phylogenetic marker for species capable of hybridization [4,5] because its nonrecombining nature means that transfer to other species can occur via asymmetric introgression and ‘allele surfing’ [6]. Here, we used broad taxonomic sampling and nuclear DNA markers with wide genomic coverage and find that the Lake Malawi radiation is not monophyletic, but instead contains genetic material from divergent riverine ancestors indicating multiple invasions and hybridization. The Lake Malawi radiation consists of six deeply divergent mitochondrial lineages broadly classed as ‘mbuna’dominated, ‘sand-dweller’-dominated (benthic), Astatotilapia calliptera, Copadichromis virginalis-dominated, Diplotaxodon/Pallidochromis and Rhamphochromis [2]. We chose taxa representing these mitochondrial clades, including species that are ecologically and morphologically typical of each clade (for example, Metriaclima thapsinogen and Cynotilapia afra are typical mbuna) as well as atypical species such as deep water Lethrinops spp., which are morphologically typical sanddwellers, but have mtDNA haplotypes that belong to the mbuna-dominated mtDNA clade. Additionally, we sampled 17 haplochromine populations from river systems inside and outside the Lake Malawi catchment (Figure 1; Supplemental information). We sequenced the mitochondrial control region (876 bp) and generated AFLP data from thirteen primer pairs. Our analysis reveals that the Lake Malawi radiation is not monophyletic as previously thought, but contains genetic material from divergent riverine lineages which indicates multiple invasions and hybridization. The ‘mbuna-dominated’ mtDNA haplogroup was shared with an ‘eastern’ lineage of Indian Ocean drainage riverine haplochromines. This lineage includes an undescribed Astatotilapia from the Rovuma river, A. swynnertoni from the Buzi River in Mozambique, A. cf. calliptera from Lake Chilwa near the head of the Rovuma, and A. calliptera from the lower Zambezi River. A. calliptera