Complete mitochondrial genomes from museum specimens clarify millipede evolution in the Eastern Arc Mountains Author Nielsen, Martin Author Margaryan, Ashot Author Nielsen, Tejs Lind Author Enghoff, Henrik Author Allentoft, Morten E. text Zoological Journal of the Linnean Society 2022 2022-10-01 196 2 924 939 https://academic.oup.com/zoolinnean/article/196/2/924/6697062 journal article 164204 10.1093/zoolinnean/zlac058 caf1f98b-3740-405f-bec6-9e7947a7c566 0024-4082 7184588 TROPOSTREPTUS PHYLOGENY AND EVOLUTION Twenty-one of our included mitochondrial genomes belong to Tropostreptus , allowing for a thorough investigation of the evolution of this genus in the Eastern Arc. We observe a clear genetic structure in Tropostreptus , with distinct lineages (both inter- and intraspecific) being defined by the mountain blocks ( Fig. 4 ). This is consistent with previous genetic results of Eastern Arc gene pools (e.g. cat snakes: Gravlund, 2002 ; chameleons: Tolley et al. , 2011 ; African violets: Dimitrov et al. , 2012 ), where forestadapted species inhabit the montane forests and are absent from the adjacent savannah lowlands. Today, the mountains capture the oceanic winds from the Indian Ocean, which maintains sufficient humidity for dense rain forest to grow, resulting in the forest ‘sky islands’ ( Lovett, 1993a , b; Burgess et al. , 2007 ). Until 30 Mya, the Eastern Arc region is thought to have been covered by rain forest ( Rodgers, 1998 ; Couvreur et al. , 2008 ), and an uplifting of the Eastern Arc Mountains is believed to have occurred within the last 7 Myr (although this is debated), changing the whole topography of East Africa ( Griffiths, 1993 ; Ring, 2014 ; Macgregor, 2015 ). Climatic and geological fluctuations through time have thus repeatedly affected the forest cover and, presumably, resulted in a multitude of vicariance events when species were isolated in patchy forest remnants ( Lovett, 1993a ; Sepulchre et al. , 2006 ; Couvreur et al. , 2008 ). For these reasons, the splitting order we observe in the Tropostreptus phylogeny might well reflect forest fragmentation in ancient times. We observe a general trend, whereby northern lineages appear to split off first. The earliest split in Tropostreptus separates the Tropostreptus austerus + Tropostreptus severus lineage from the rest, and the second split separates these two species, today occupying Nguru and Usambara Mountains in the north. A similar intraspecific pattern is evident in more recent splits in Tropostreptus hamatus and Tropostreptus sigmatospinus ( Fig. 4 ), suggesting a repeated pattern of vicariance events occurring first in the north. A separation of species between northern and southern mountains has also been observed in several other Eastern Arc taxa, including amphibians ( Blackburn & Measey, 2009 ), gastropods ( Tattersfield et al. , 1998 ) and reptiles ( Gravlund, 2002 ; Tolley et al. , 2011 ), but also in well-dispersing taxa, such as birds ( Fjeldså & Bowie, 2008 ). This indicates a forest retraction southwards during dry periods, resulting in vicariance events, followed by forest expansion and thus northward recolonization of species during periods with higher humidity. Northward migration is also observed in other Eastern Arc species, such as chameleons ( Tolley et al. , 2011 ; Ceccarelli et al. , 2014 ). A recent cycle of forest expansion/retraction can explain why Tropostreptus hamatus and Tropostreptus sigmatospinus exist across several of the mountains without having evolved into distinct species yet. Other events have isolated Tropostreptus kipunji in the forest on Mount Rungwe, the most south-westerly occurring species in the Eastern Arc region, in addition to Tropostreptus sigmatospinus in Zanzibar and, potentially, also the Rondo Plateau, from where Tropostreptus has been observed but for which molecular data are still lacking ( Enghoff, 2017 ). Regarding the timing of the species splits ( Fig. 5 ), several major events might have played a role. Around 30 Mya the Antarctic ice sheet started to form ( Couvreur et al. , 2008 ), along with rifting that started to occur in northern East Africa ( Ring, 2014 ), possibly initiating the fragmentation of the pan-African forest. Through millions of years, the rifting would continue southwards ( Ring, 2014 ), affecting the topology and possibly related to the forest fragmentation responsible for the divergence of Tropostreptus austerus and Tropostreptus severus observed ~22 Mya. The observed divergence of Tropostreptus hamatus and the split between Tropostreptus austerus and Tropostreptus severus correspond well to the closing of the Tethys Sea (17 Mya), which would have altered ocean currents and, probably, the climate of the area ( Couvreur et al. , 2008 ). Likewise, the isolation of the Tropostreptus kipunji lineage corresponds with the uplifting of Mount Rungwe from ~8 Mya ( Ring, 2014 ). Finally, between 5 Mya and today, we observe a radiation in Tropostreptus hamatus and Tropostreptus sigmatospinus ( Fig. 5 ). A reasonable explanation for this is the uplift of the Eastern Arc Mountains, shifting the precipitation from the lowlands to the mountains ( Lovett, 1993a , b), in combination with the Antarctic ice sheet forming, thus decreasing global humidity ( Polyak et al. , 2010 ). This would lead to the emergence of savannah in the lowlands between the mountains ( Sepulchre et al. , 2006 ; Ségalen et al. , 2007 ; Couvreur et al. , 2008 ), isolating the montane forest and limiting migration between populations of forest-restricted species. We emphasize that we have neither good fossil records nor mtDNA mutation rates estimated specifically for millipedes, which is why the split times of our millipede phylogenetic tree should be interpreted with caution. Moreover, comparable studies with dated phylogenies of Eastern Arc species are sparse, hence it is difficult to compare the split times we have estimated with those of other species in the region. Examining two separate studies of chameleons ( Kinyongia Tilbury, Tolley & Branch, 2006 and Trioceros Swainson, 1839 ) with dated phylogenies based on both mitochondrial and nuclear markers did show some correspondence with our dated splits ( Tolley et al. , 2011 ; Ceccarelli et al. , 2014 ). Tolley et al. (2011) dated the earliest split between the northern and southern Eastern Arc species to ~28 Mya, and both studies show several radiation events between 5 and 20 Mya, corresponding to the same overall time frame that we are discussing for the millipedes. In contrast, the chameleons display fewer speciation events during the last 5 Myr than the millipedes, perhaps suggesting that the latter have been more susceptible to vicariance during more recent climatic events.