Journal of Biogeography (J. Biogeogr.) (2017)
ORIGINAL Multilocus phylogeny of East African ARTICLE gerbils (Rodentia, Gerbilliscus) illuminates the history of the Somali-Masai savanna Tatiana Aghov a1,2* , Radim Sumbera3, Lubom ır Pi alek3, Ond rej Mikula1,4, Molly M. McDonough5,6, Leonid A. Lavrenchenko7, Yonas Meheretu8, Judith S. Mbau9 and Josef Bryja1,2
1Institute of Vertebrate Biology of the Czech ABSTRACT Academy of Sciences, 603 65 Brno, Czech Aim The rodent genus Gerbilliscus is widespread in savannas throughout sub- Republic, 2Department of Botany and Saharan Africa. The eastern clade comprises four species with distributions cen- Zoology, Faculty of Science, Masaryk University, 602 00 Brno, Czech Republic, tred in the Somali-Masai biogeographical region of East Africa. We investigated 3Department of Zoology, Faculty of Science, the genetic diversity of the group with a view to illuminating the historical University of South Bohemia, 370 05 Cesk e (Plio-Pleistocene) processes that formed contemporary biota of the understud- Bud ejovice, Czech Republic, 4Institute of ied Somali-Masai region. Animal Physiology and Genetics of the Czech Location Somali-Masai savanna, East Africa. Academy of Sciences, 602 00 Brno, Czech Republic, 5Center for Conservation Genomics, Methods We performed multilocus genetic analyses of 240 samples from 112 Smithsonian Conservation Biology Institute, localities, combining genotyping of recently collected samples (N = 145), 454- National Zoo, Washington, DC 20008, USA, pyrosequencing of museum material (N = 34) and published sequences 6Department of Vertebrate Zoology, National (N = 61). We used Bayesian and maximum likelihood approaches for phyloge- Museum of Natural History, Smithsonian netic reconstructions, and coalescent-based methods to delimit species. We also Institution, Washington, DC 20560-0108, estimated divergence times and modelled recent and past distributions to 7 USA, A.N. Severtsov Institute of Ecology and reconstruct the major evolutionary influences in the Somali-Masai region dur- Evolution RAS, 119081 Moscow, Russia, ing the Plio-Pleistocene. 8Department of Biology, College of Natural and Computational Sciences, Mekelle Results Genetic analyses provided evidence for six lineages, possibly corre- University, Mekelle, Tigray, Ethiopia, sponding to distinct species. The two main species groups (with two and four 9Department of Land Resource Management putative species, respectively) have overlapping distributions, but species within and Agricultural Technology, College of each group are distributed parapatrically. The origin of the eastern clade dates Agriculture and Veterinary Sciences, back to the Pliocene, while individual species diverged in the Pleistocene. The University of Nairobi, Nairobi, Kenya distribution of genetic diversity and ecological niche modelling point to the importance of the Rift Valley and the presence of unsuitable xeric habitats in the allopatric diversification of Gerbilliscus in the Somali-Masai savanna within *Correspondence: Tatiana Aghova, Institute of the last 5 Myr. Vertebrate Biology ASCR, Research Facility Studenec, 675 02 Studenec, Czech Republic. Conclusions This is the first detailed phylo(bio-)geographical study of ani- Current address: Department of Zoology, mals with predominant distribution in the Somali-Masai region. It revealed National Museum, 115 79 Prague, Czech Republic. currently underestimated diversity of eastern clade of Gerbilliscus and proposed E-mail: [email protected] a scenario of its evolution during Plio-Pleistocene. Conspicuous genetic struc- Dedication ture of these taxa can be now used to test detailed phylogeographical hypothe- We dedicate this paper to the late Bill Stanley ses related to Plio-Pleistocene history of gerbils and, to some extent, also biota who devoted much of his career to the study of Somali-Masai bioregion in general. of small mammals in East Africa. Bill facilitated acquiring numerous samples used in Keywords this study and the promise of future biogeography, Gerbillinae, historical DNA, murid rodents, phylogeography, collaborations was cut short as Bill passed Plio-Pleistocene climate change, pyrosequencing, Rift Valley, species delimita- away unexpectedly during his field trip in Ethiopia in 6 October 2015. tion, tropical Africa
ª 2017 John Wiley & Sons Ltd http://wileyonlinelibrary.com/journal/jbi 1 doi:10.1111/jbi.13017 T. Aghova et al.
savanna (Fig. 1). The major geomorphological feature here is INTRODUCTION the East African Rift system which has had a long-term Tropical grasslands, savannas and shrublands are the most influence on the climatic regime in Africa (Sepulchre et al., widespread terrestrial habitats in sub-Saharan Africa (Sayre 2006). While this region has relatively low floristic and fau- et al., 2013). These ecosystems, hosting some of the most nistic species diversity, a high level of endemism exists for abundant and diverse mammalian communities on Earth, plants (Thulin, 1993), reptiles (Burgess et al., 2007) and are threatened by increased human activities (Craigie et al., rodents (Varshavsky et al., 2007). This region is home to the 2010). Within the continent they can be divided into four Horn of Africa biodiversity hotspot, which ranks among the main bioregions (sensu Linder et al., 2012): South African, oldest and most stable arid regions of Africa (Kingdon, Zambezian, Sudanian and Somali-Masai, with the last one 1990). In addition, it is one of the least known African corresponding to the ‘Somalia-Masai’ region, originally pro- bioregions and possibly also one of the most threatened due posed by White (1983) as a distinct dry phytochorion, to the rapid increase in human populations and climatic extending from Eritrea to Tanzania (Fig. 1). International changes (Geist & Lambin, 2004). Vegetation Classification (IVC; naturserve.org; Faber-Langen- Despite the biogeographical uniqueness of the Somali- doen et al., 2014) identifies this division as Eastern Africa Masai region, genetic data on organisms living here are Xeric Scrub and Grassland with four ecoregions: (1) Somali, scarce and fragmentary – usually included as part of large- (2) Southern, (3) Northern Acacia-Commiphora bushlands scale phylogenetic reconstructions with under-representative and thickets and (4) Masai xeric grasslands and shrublands geographical sampling. Previous studies found that taxa from (Dixon et al., 2014), referred to herein as Somali-Masai this region are phylogenetically distinct, thereby providing
Figure 1 Distribution of species and intraspecific haplogroups of Gerbilliscus in East Africa. Species of the eastern clade are represented with different colours and intraspecific haplogroups are indicated by different symbol shapes. Empty squares represent genetically confirmed records of Gerbilliscus species from the western clade, empty circles those from the southern clade (sensu Granjon et al., 2012), and black crosses indicate localities where Gerbilliscus was not captured during our recent field trappings. The map is based on data from Colangelo et al. (2007, 2010), McDonough et al. (2015), this study and our unpublished data of DNA-barcoded specimens from western and southern clades. Orange background indicates the extent of Somali-Masai savanna, i.e. the distribution of Somali, Northern and Southern Acacia-Commiphora bushlands and thickets (based on WWF records http://www.eoearth.org/).
2 Journal of Biogeography ª 2017 John Wiley & Sons Ltd Phylogeny of Somali-Masai gerbils evidence that the Somali-Masai biota evolved in isolation for been studied in detail. Previously published genetic data an extended period of time (e.g. Sm ıd et al., 2013; Mikula have been limited to a few specimens genotyped at one et al., 2016). Noteworthy examples include the plant Wajira mitochondrial and one nuclear marker with low variability (Thulin et al., 2004), Grevy’s zebra (Kebede et al., 2016) and (Colangelo et al., 2007) and morphological analyses were warthogs (Randi et al., 2002). However, detailed phylogeo- geographically biased towards Tanzanian localities (Colan- graphical studies of the structure and evolutionary history of gelo et al., 2010). genetic lineages living predominantly in the Somali-Masai Species distributions based on morphological identifica- bioregion are lacking. tions (Monadjem et al., 2015) indicate that the eastern clade Eastern Africa experienced a turbulent climatic and geo- occurs within the borders of the Somali-Masai region and its morphological history during the Plio-Pleistocene. The evolu- genetic analysis may provide insights into the evolutionary tion of its biota was influenced by several major clima- processes that formed the contemporary biota of this region. tic transitions (Maslin & Christensen, 2007), including the We analysed how gerbil divergence dates reflect the geomor- intensification of the Northern Hemisphere glaciation (3.2– phology and climate of the area. More specifically, we tested 2.5 Ma), the development of the Walker circulation if (1) local geomorphological features such as mountain (2.0–1.7 Ma), and the early-middle Pleistocene transition ranges and the Great Rift Valley with its amplifier lakes and (1.2–0.8 Ma) (Maslin & Christensen, 2007; Trauth et al., (2) periods of pronounced climatic variation in the Plio- 2009). In particular, the tropical oscillations between warm Pleistocene influenced genetic structure and divergence and relatively wet pluvials and cooler and drier interpluvials within the eastern clade of Gerbilliscus. during the Pleistocene resulted in repeated expansions and contractions of either savanna-like or forest-like habitats MATERIALS AND METHODS (e.g. Cowling et al., 2008). These oscillations, which roughly correspond to interglacial and glacial periods in the northern DNA extraction, PCR amplification and sequencing and southern latitudes (Dupont, 2011), significantly affected diversification and speciation processes. Beside these global Our genetic dataset includes sequences from recent collec- transitions, several additional local factors, such as climate- tions (145 individuals/72 localities), museum samples (34/14) driven vegetation changes (Sepulchre et al., 2006; Cerling and georeferenced sequences from Genbank (61/26) (Fig. 1; et al., 2011; Uno et al., 2016), or the presence of the amplifier for details see Table S1 in Appendix S1 in Supporting Infor- lakes in the bottom of the Rift Valley (Trauth et al., 2010), mation). DNA from 96% ethanol-preserved tissue samples probably influenced the evolution of taxa adapted to eastern was extracted using a DNeasy Blood & Tissue kit (Qiagen, African savannas, woodlands and grasslands. Unfortunately, Hilden, Germany). For phylogenetic analysis we selected four without studies of detailed organismal genetic structure, it is genetic markers; all specimens were sequenced for one mito- difficult to evaluate the roles of climatic and geomorphologi- chondrial gene, cytochrome b (CYTB), and a subset of indi- cal factors (e.g. Trauth et al., 2010) in forming current viduals (1–5 individuals per mitochondrial lineage), for three biological diversity in the area. Carefully selected model spe- nuclear markers that have previously been used successfully cies can provide valuable information for understanding our to resolve intrageneric phylogenies in mammals: Intron 7 of own history, as numerous crucial localities for studies of the gene for b-fibrinogen (FGB) and exons of breast cancer human evolution are located in eastern Africa (Maslin et al., susceptibility gene (BRCA1) and Interphotoreceptor Binding 2015). Protein gene (IRBP). For genotyping protocols see Small mammals such as rodents are often good models Appendix S2. Museum samples (taken mostly from dry for phylogeographical reconstruction because they are habi- skins) were pyrosequenced on a GS Junior (Roche, Basel, tat specialists, exhibit low dispersal ability, and have rela- Switzerland) using the CYTB mini-barcode protocol (Galan tively high genetic substitution rates. Gerbils of the genus et al., 2012; see Appendix S2). This method allows for the Gerbilliscus (Thomas, 1897) are widespread rodents living separation of focal sequences in samples contaminated by in the savannas, woodlands, grasslands and semi-deserts of distantly related organisms (e.g. human DNA; Bryja et al., sub-Saharan Africa. The phylogeny of the genus was 2014). recently investigated using karyology, morphology and DNA sequences (e.g. Granjon et al., 2012). The mono- Phylogenetic analysis phyletic eastern Gerbilliscus clade (also called ‘robustus group’, but referred to here as the ‘eastern clade’) repre- Sequences were aligned using Mafft 7 (Katoh & Standley, sents the first cladogenetic split within the genus and com- 2013) and the supermatrix (4473 bp) was created in Sea- prises four currently recognized species: G. nigricaudus View 4 (Gouy et al., 2010). The final dataset for phyloge- (Peters, 1987), G. vicinus (Peters, 1987), G. robustus (Cret- netic analyses consisted of 185 unique sequences of CYTB, zschmar, 1826) and G. phillipsi (deWinton, 1989) (sensu 25 sequences of BRCA1, 27 sequences of FGB and 26 Monadjem et al., 2015). Unlike the western (Granjon et al., sequences of IRBP. The remaining 55 CYTB sequences (iden- 2012) and southern (McDonough et al., 2015) clades, the tical and/or shorter sequences from the same or neighbour- phylogeny and distribution of the eastern clade have never ing localities) were unambiguously assigned to particular
Journal of Biogeography 3 ª 2017 John Wiley & Sons Ltd T. Aghova et al. lineages in RAxML 7.2.8 (Stamatakis, 2006). These data were 10 9 106 generations with sampling every 1000 generations used to increase the precision with which we mapped the in beast. We discarded first 25% as burn-in and the result- geographical distribution of clades and assigned type material ing parameter and tree files were examined for convergence to particular genetic groups. As outgroups we used two and effective sample sizes (> 200) in Tracer 1.6 (Rambaut sequences of G. giffardi from western clade, which is the sis- et al., 2014). The two runs were combined in LogCombiner ter lineage to eastern Gerbilliscus (see Table S1 in 1.8.2 and the species tree was visualized in DensiTree Appendix S1 for the GenBank numbers). (Bouckaert, 2010). The best partitioning scheme and substitution models were determined using PartitionFinder 1.1.1 (Lanfear et al., Molecular dating 2014). Evolutionary relationships were estimated using maxi- mum likelihood (ML) in RAxML 7.2.8 (Stamatakis, 2006) The times to most recent common ancestors (TMRCA) were and Bayesian inference (BI) in MrBayes 3.2.2 (Ronquist inferred in BEAST 1.8.2 from a three-locus dataset (CYTB, et al., 2012) (see Appendix S2). For visualization of BRCA1, IRBP); sequences of FGB were excluded because of a intraspecific structure we also constructed haplotype net- lack of sequenced outgroup taxa. The final dataset consisted works for the two most numerous species, G. vicinus and G. of 17 species of Gerbilliscus as well as Desmodillus auricularis nigricaudus. Haplotype files were produced in DnaSP 5 and Tatera indica, their closest relatives in the phylogeny of (Librado & Rozas, 2009) and median-joining networks in gerbillines (Granjon et al., 2012). Sequences analysed are Network 4.6.1.3 (Bandelt et al., 1999). listed in Table S2 in Appendix S1. We used partition-specific substitution models (see Table S4 in Appendix S2), gene-spe- cific lognormal relaxed clocks (Drummond et al., 2006) and Species delimitation birth–death tree priors (Gernhard, 2008), as well as three fos- Our results indicated six lineages that might represent sepa- sil calibrations: (1) Abudhabia pakistanensis from the Potwar rate species. We tested the distinctiveness of their gene pools Plateau, Pakistan (Flynn & Jacobs, 1999) for the whole Ger- using the three nuclear loci in a Bayesian framework and billiscus-Tatera-Desmodillus clade (Denys & Winkler, 2015); BP&P 3 (Yang & Rannala, 2014), which samples shared phy- (2) Gerbilliscus sp. from Lemudong0o, Kenya (Manthi, 2007) logenetic space and possible mergers of predefined candidate for Gerbilliscus and (3) Gerbilliscus sp. from Hadar, Ethiopia species under assumptions of a multispecific coalescent (Sabatier, 1982) for the eastern clade. The calibration density model (Degnan & Rosenberg, 2009). The analysis was run was always lognormal (Mean in real space = 1.6, SD = 0.9) with three combinations of gamma priors for root age (s0) with an offset corresponding to the particular fossil age, and effective population size (h). First, we used G(10,1300) which is 8.6 million years (Ma) for Abudhabia (Flynn & 0 as an informative prior for s0, based on an assumption that Jacobs, 1999), 6.1 Ma for the Lemudong o Gerbilliscus deeply divergent and broadly sympatric G. nigricaudus and (Deino & Ambrose, 2007) and 3.3 Ma for the Hadar Gerbil-
G. vicinus are heterospecific and s0 is approximately equal to liscus (Reed & Geraads 2012). For more details see half of their mean divergence in terms of Jukes-Cantor dis- Appendix S2. tance (see Table S6-S8, Appendix S3). We have no specific information about h and hence our prior was either diffuse, Species distribution modelling G(2,300), and hence uninformative, or suggestive of large values, G(10,1300) and hence conservative (tending to lump The present distribution of the East African Gerbilliscus clade the species). Finally, we used the diffuse prior for h in com- was predicted by maximum entropy (MaxEnt) modelling bination with diffuse prior for s0, G(0.5,25). Dirichlet distri- (Phillips et al., 2006) interfaced with R computing environ- bution with a = 1 is fixed in the software as a prior for ment by packages ‘dismo’ (Hijmans et al., 2016) and ‘ENMe- internal branch lengths and we also used it as a prior for val’ (Muscarella et al., 2014). Only CYTB-barcoded between-gene variability in evolutionary rates. Each analysis specimens were included to avoid taxonomic confusion was run twice to check if it converged on a similar posterior resulting in 66 unique presence records (in 0.5° resolution). distribution. We focused on climatic conditions as predictors of relative occurrence rates (RORs). More specifically, we used 19 BIO- CLIM variables downloaded from the WorldClim website Species tree (Hijmans et al., 2005). Following corrected Akaike informa- The species tree was calculated under the Bayesian frame- tion criterion (AICc) based model selection (Warren & Sei- work implemented in *BEAST package (Heled & Drum- fert, 2011) we used linear and quadratic features mond, 2010), an extension of BEAST 1.8.2 (Drummond (transformations of original variables) and a regularization et al., 2012). Alignments for each of the four genes were coefficient of 1.68 (see Appendix S2 for details). Under these imported into BEAUti 1.8.2 where they were assigned sepa- parameters, RORs were predicted for all background sites as rate and unlinked substitution, clock and tree models. well as the corresponding sites in palaeoclimatic layers pro- Sequences were assigned to species as delimited by the best vided by WorldClim for climatic conditions during the Last model in BP&P. Two independent runs were carried out for Glacial Maximum (21 ka) and during the last interglacial
4 Journal of Biogeography ª 2017 John Wiley & Sons Ltd Phylogeny of Somali-Masai gerbils
(140–120 ka). These layers were produced as estimates from Divergence time estimates global circulation models built by Braconnot et al. (2007) for the last glacial and by Otto-Bliesner (2006) for the last inter- Divergence dating in BEAST estimated the TMRCA of the glacial. The results were visualized using the R packages genus Gerbilliscus in the Late Miocene (7.03 Ma; 95% HPD: ‘maptools’ (Bivand & Lewin-Koh, 2016) and ‘raster’ (Hij- 6.17–10.13; Fig. 4). TMRCA of the eastern clade, i.e. the split mans, 2015). The importance of each predictor was quanti- between the nigricaudus and robustus groups, was estimated fied by its spatial randomization, calculating the Spearman as 4.17 Ma (95% HPD: 3.38–6.78) and the basal split in the correlation (rs) of predictions before and after randomization robustus group as 2.42 Ma (95% HPD: 1.50–4.30). Three and subtracting it from unity. pairs of sister species diverged almost simultaneously about 1.70 Ma: G. robustus and G. sp. n. (Babile) at 1.62 Ma (95% HPD: 0.73–2.95), G. vicinus and G. phillipsi at 1.75 Ma (95% RESULTS HPD: 0.73–3.31) and G. nigricaudus and G. cf. bayeri at 1.80 Ma (95% HPD: 0.86–3.53). Phylogeny, distribution and species delimitation of East African Gerbilliscus Species distribution modelling Both BI (Fig. 2) and ML (see Fig. S2, Fig. S5 in Appendix S3) phylogenetic analyses recovered six main lin- The suitability of climatic conditions for the eastern Gerbillis- eages whose relationships were only partly resolved. The first cus clade appears to have deteriorated from the last inter- split separated the eastern Gerbilliscus into the nigricaudus glacial, through the last glacial to the present (Fig. 5). group (consisting of two lineages G. nigricaudus and G. cf. According to our model the Horn of Africa was more suitable bayeri; see Discussion for more details on the use of names in the last interglacial, and less so from the last glacial maxi- for particular lineages) and the robustus group (consisting of mum to the present. In Tanzania, on the southern border of four lineages G. vicinus, G. phillipsi, G. robustus, G. sp. n. the Somali-Masai area (see Fig. 1) the suitability of condi- (Babile)). The distribution of the eastern clade is largely con- tions also fluctuated. Nevertheless, in the core of the present cordant with the borders of the Somali-Masai region and day distribution (Somali-Masai savanna on both sides of the parapatric with respect to western and southern clades of Rift Valley) climate was apparently favourable during the last Gerbilliscus. The only exceptions are records of G. robustus in glacial cycle. The most important variable affecting the mod- Sudan and Chad, where they may be sympatric with species elled distribution was precipitation during the wettest quarter of the western clade (Fig. 1). (importance = 0.76), but the effect of mean temperature dur- The BP&P analysis supported all six main lineages as phylo- ing the coldest quarter, precipitation during the wettest genetically distinct entities (putative species) regardless of the month and annual precipitation were also significant (impor- combination of priors (posterior probability > 0.98 in all tance ≥ 0.20; see Fig. S4 in Appendix S3 for details). cases). Posterior samples of both s0 and mean h were substan- tially narrower than the corresponding priors, largely coinci- DISCUSSION dent across analyses, and their median corresponded to 0.23– 0.40 quantile of prior density (see Fig. S3 in Appendix S3). Cryptic diversity and distribution of the eastern Thus, the support for distinctiveness of our candidate species clade of Gerbilliscus was neither due to some specific priors nor due to the sam- pling getting stuck at an extreme combination of large diver- In this study, we used multi-locus genetic data to reconstruct gence times and low effective population sizes. The *BEAST the evolutionary history of the eastern African clade of ger- species tree of these six lineages recovered a fully resolved bils, genus Gerbilliscus, distributed mostly in the Somali- topology showing that G. sp. n. (Babile) is the sister species of Masai region. Because of a previous lack of sampling and G. robustus and G. phillipsi is sister to G. vicinus (Fig. 3). genetic data, precise distributions and phylogenetic relation- Phylogenetic analysis (Fig. 2) as well as haplotype net- ships of four currently recognized species in this clade (i.e. works (see Fig. S1 in Appendix S3) revealed significant G. nigricaudus, G. robustus, G. vicinus, G. phillipsi; sensu genetic variation within the three species. We defined two Colangelo et al., 2005) remained obscure. Using phylogenetic haplogroups for G. robustus, and four for both G. vicinus and species delimitation approaches we evaluated the status and G. nigricaudus; genetic K2P distances among intraspeci- of these four species, analysed their intraspecific variability fic haplogroups ranged from 0.036 to 0.100 (Table 1). In G. and provided details on their distributions. robustus, one haplogroup occurred in the Rift Valley in In accordance with previous work (Colangelo et al., 2005, Ethiopia (R1), while the second was found in Chad and 2007; Granjon et al., 2012), we identified two lineages that Sudan (R2). In the two remaining species, the distribution may represent separate but unrecognized species. The first, of intraspecific lineages was mostly parapatric, either sepa- Gerbilliscus sp. n. (Babile), was first reported as a genetically rated by the Rift Valley (e.g. V2 versus V1, V3,V4) or along distinct lineage (12.5–17.6% CYTB genetic distance from the north-south axis (e.g. N4-N2-N1-N3) (Fig. 2, see other species in the eastern clade; see Table 1) from the Fig. S1 in Appendix S3). Babile Elephant Sanctuary in eastern Ethiopia (Lavrenchenko
Journal of Biogeography 5 ª 2017 John Wiley & Sons Ltd T. Aghova et al.
Figure 2 Bayesian inference phylogeny for East African Gerbilliscus based on concatenated dataset, i.e. supermatrix of one mitochondrial (CYTB) and three nuclear loci (FGB, BRCA1, IRBP). Black points indicate nodes with posterior probability support > 0.95.
6 Journal of Biogeography ª 2017 John Wiley & Sons Ltd Phylogeny of Somali-Masai gerbils
nigricaudus bayeri (RMCA5183-M) from Maroon River in western Kenya, and it clustered within a clade distributed west of the Rift Valley, as sister to G. nigricaudus. Based on the results of a coalescent species delimitation and high CYTB genetic distances (13.4–15.1% from G. nigricaudus), this clade may deserve the status of a distinct species status (see also Bates, 1988). Taxonomic revision of the group, including morphological analyses and formal descriptions of these taxa, is currently in preparation and will be published elsewhere (M.M. McDonough et al., unpublished data). Three taxa, G. sp. n. (Babile), G. phillipsi and G. cf. bayeri, exhibited relatively small geographical distributions (Fig. 1), although this could be a result of limited sampling in some Figure 3 Coalescent based species tree for Gerbilliscus in East regions. For example, G. sp. n. (Babile) may be more wide- Africa generated using *BEAST, visualized by DensiTree. Posterior probability of nodes is indicated in boxes. spread in southeastern Ethiopia and part of Somalia, and museum records of G. robustus or G. phillipsi from Somalia et al., 2010). We captured additional individuals at the same (Monadjem et al., 2015) may in fact represent this species site, and increased our sample by adding museum specimens (M.M. McDonough et al., unpublished data). On the other from additional localities in Ethiopia and Somalia. A coales- hand, present records of G. sp. n. (Babile) suggest its occur- cence-based species tree demonstrated that G. sp. n. (Babile) rence is associated with transitional semi-evergreen bushland is a sister lineage to G. robustus, from which it seems to be (van Breugel et al., 2016), which is localized only in the nar- separated by a mountain ridge in the eastern Ethiopian high- row belt along mountain chains in Ethiopia. Gerbilliscus cf. lands (Fig. 1). The second currently unrecognized species we bayeri has a limited distribution in western Kenya and South referred to as G. cf. bayeri. We sequenced a syntype of G. Sudan (and potentially also in northern Uganda). On the
Figure 4 Ultrametric Bayesian tree generated using BEAST with dated divergences for East African Gerbilliscus (i.e. TMRCAs) and 95% highest posterior densities (HPDs). Stars indicate the position of palaeontological records used for calibration of the tree.
Journal of Biogeography 7 ª 2017 John Wiley & Sons Ltd T. Aghova et al.
Figure 5 MaxEnt results of (a) the present distribution of suitable habitats for East African Gerbilliscus clade as typical representatives of Somali-Masai fauna (red dots indicate confirmed presence data); (b) MaxEnt prediction for last glacial maximum (21 ka). (c) MaxEnt prediction for last interglacial period (120–140 ka). Green colour represent habitats that are suitable for the eastern Gerbilliscus, while yellow-orange are not. other hand, the three remaining species (G. robustus, G. vici- wet-dry climatic oscillations (Potts, 2013) generated direc- nus, G. nigricaudus) have large distributions. Conspicuous tional selection pressures driven by the expansion and con- intraspecific structure of these species (e.g. Fig. S1 in traction of grasslands. The oldest divergence within the Appendix S3) can be used to propose and test phylogeo- robustus group occurred around 2.5 Ma. Two Pleistocene graphical hypotheses related to the Plio-Pleistocene history climatic transitions, the development of the Walker circula- of gerbils and, to some extent, also the biota of the Somali- tion (2.0–1.7 Ma; Ravelo et al., 2004) and the early-middle Masai bioregion in general. Pleistocene transition (1.2–0.8 Ma; Berger & Jansen, 1994), also significantly affected the ecosystems in eastern Africa. The Walker circulation increased the interannual variation in Evolutionary history of the eastern clade of rainfall and the early-mid Pleistocene transition prolonged Gebilliscus and a scenario of diversification in the and intensified glacial-interglacial climatic cycles. Both transi- Somali-Masai savanna tions also helped the spread of C4 plants (Cerling et al., This study is the first, to our knowledge, to focus on the 1988). Both hominins (Homo ergaster and Homo erectus) and detailed phylogeography of any group living predominantly a number of ungulate species associated with grassland habi- in the Somali-Masai savanna. We estimated the origin of the tats first appeared between 2.0 and 1.9 Ma (Bobe, 2004). In genus Gerbilliscus as dating back to the late Miocene. Previ- eastern Gerbilliscus, this period seems to be associated with ous estimates of TMRCA varied, according the calibration synchronous diversification of extant sister species (Fig. 4), points and genetic markers used, between 6.3 and 8.5 Ma while the subsequent intensification of climatic cycles appar- (Chevret & Dobigny, 2005; Colangelo et al., 2005, 2007), ently affected allopatric divergences within species with large which is in agreement with our results. Based on our esti- distribution areas (see e.g. Fig. S1 in Appendix S3). mated divergence dates, the Plio-Pleistocene evolution of the Additionally, the local geomorphology played an impor- eastern Gerbilliscus clade seems to have been influenced by tant role in evolution of the East African biota. For example, several major climatic events. Very little is known about the Gerbilliscus robustus is separated from G. sp. n. (Babile) by a effect of the Messinian salinity crisis on eastern African cli- mountain chain that apparently prevents the dispersal of mate (Maslin & Christensen, 2007), but it is generally held both taxa across high altitude areas with very different envi- that overall aridification at the Miocene/Pliocene boundary ronmental conditions (Fig. 1). One of most influential land- promoted the expansion of very dry habitats (Hodell et al., scape features in the evolution of small mammals in the 2001). Due to the fact that arid parts of the extant Somali- Somali-Masai region is the Rift Valley (Trauth et al., 2010). Masai savanna (e.g. Masai-xeric shrubland) are likely to by Rifting generated numerous depressions which were repeat- uninhabitable by Gerbilliscus (as is northern Kenya today; edly filled with water, and which were highly sensitive to Figs 1 & 5), we hypothesize that such aridification led to the changes in the local precipitation-evaporation regime (Maslin early-Pliocene split between the robustus group in the north et al., 2014). Rift basins, either filled with water or extremely and the nigricaudus group in the south. dry with halophytic vegetation (White, 1983) could prevent The intensification of Northern Hemisphere glaciation is the dispersal of small terrestrial animals (Trauth et al., 2010). likely to have had similar effects, reflected in the African as Gerbilliscus biogeography appears to reflect the influence of an increase in aridity c. 2.7 Ma (deMenocal, 2004). Our esti- the Rift Valley both from east to west (G. cf. bayeri versus G. mates of divergence dates suggest that prolonged intervals of nigricaudus, G. nigricaudus N2 versus N4 and N1 versus N3,
8 Journal of Biogeography ª 2017 John Wiley & Sons Ltd Phylogeny of Somali-Masai gerbils
G. vicinus V2 versus V3) and from north to south (G. vicinus cf.
G. bayeri versus G. phillipsi versus G. robustus; Fig. 1). An interesting observation is the migration of G. robustus R2 into Central Africa, because very few taxa originating in the Somali-Masai region have spread westwards or south- wards. Conversely, numerous species typically inhabiting the
G. nigricaudus N4 Sudanian savanna have their origins in the east, including rodents (e.g. Brouat et al., 2009; Dobigny et al., 2013) and ungulates (Lorenzen et al., 2012). Our ecological niche mod- elling (Fig. 5) suggests that conditions have been suitable, especially during interglacials, for the spread of savanna spe- G. nigricaudus N3
. Values above diagonal represent cies into the belt of Sudanian region, although it is not clear why other Somali-Masai taxa (e.g. Gr evy’s zebra or pouched
mega mice) remained restricted to the easternmost part of Africa (IUCN 2016, Mikula et al., 2016). gene in G. nigricaudus N2 Gerbilliscus as a model for understanding the CYTB evolution of the Somali-Masai savanna
According to White (1983), the Somali-Masai region com-
G. nigricaudus N1 prises several vegetation types. The most widespread vegeta- tion in eastern Africa including the Horn of Africa is Acacia- Commiphora bushland and thicket (shown as orange back- sp. n. ground in Fig. 1; Dixon et al., 2014). Overlap between this G. (Babile) vegetation type and the distribution of eastern Gerbilliscus spp. suggests that gerbils may be informative regarding the Plio-Pleistocene evolution of this bioregion, although the actual distribution of the genus may be more restricted than G. robustus R2 can be predicted from vegetation maps. The absence of Gerbilliscus in large areas of southeastern Ethiopia, Somalia and northeastern Kenya may partly be caused by a lack of sampling or of sampling at appropriate G. robustus R1 times of the year. However, two indicators argue against this
clade (below diagonal) calculated from mitochondrial interpretation. First, our recent sampling in southeastern Ethiopia and northern Kenya (Fig. 1) failed to find any Ger-
G. phillipsi billiscus, and there is also a dearth of museum specimens from this area (Monadjem et al., 2015; ‘Free and Open Gerbilliscus Access to Biodiversity Data|GBIF.Org’). These arid areas belonging to the Masai-xeric shrubland ecoregion (sensu G. vicinus V4 White, 1983) have an annual rainfall of < 150 mm, and are often almost devoid of vegetation, covered only by stones. Small mammal communities are dominated by Acomys, Ger-
G. vicinus V3 billus and Arvicanthis, rodents that can tolerate extreme arid- ity. Second, our species distribution modelling predicts (Fig. 5) that virtually all of the Horn of Africa represents unsuitable habitat for Gerbilliscus, which should be explored G. vicinus V2 more extensively in the field. In the contrast, climatic condi- tions in southern Somalia seem to support the occurrence of Gerbilliscus (Fig. 5), which is in agreement with previous biodiversity surveys (Bates, 1988; Varty, 1988). G. vicinus V1 0.146 0.159 0.158 0.1520.242 0.263 0.254 0.014 0.239 0.019 0.219 0.017 0.221 0.024 0.220 0.024 0.225 0.134 0.025 0.151 0.024 0.139 0.024 0.136 Our study provides the first detailed genetic structure of
N1N2 0.213N3 0.201N4 0.210 0.228 0.217 0.226 0.221 0.223 0.227 0.216 0.221 0.204 0.224any 0.207 0.198 0.225mammal 0.215 0.238 0.228 0.208 0.232from 0.230 0.224 0.236 0.220 the 0.231 0.224 Somali-Masai 0.201 0.235 0.224 0.205 0.214 0.075 0.069 savanna, 0.080 0.011 0.101 including 0.054 0.010 0.014 0.099 0.013 0.010 0.015 0.018 0.020 0.016 0.018 R1R2 0.161 0.169 0.179 0.181 0.159 0.184 0.163 0.171 0.113 0.141 0.100 0.015 0.017 0.022 0.026 0.026 0.025 0.024 0.025 0.025 0.026 0.025 0.027 0.028
V1V2V3V4 0.053 0.043 0.036 0.010 0.044 0.057 0.009 0.008 0.053 0.008 0.011 0.010 0.018 0.019 0.018 0.019 0.020identification 0.022 0.021 0.022 0.020 0.023 0.023 of 0.019 0.023 the 0.020 0.021 0.021main 0.019 0.021 0.023 factors 0.022 0.022 responsible 0.023 0.024 0.023 0.023 0.023 for 0.024 their 0.023 0.023 evolu- 0.021 0.023 0.023 0.030 0.030 0.029 0.030 K2-P genetic distances among lineages of East African tionary history since the Miocene. The general validity of bayeri
sp. n. (Babile) 0.156 0.170 0.153 0.161evolutionary 0.144 0.125 patterns 0.176 observed in Gerbilliscus 0.022should 0.024 now be 0.021 0.022 0.023 . cf. G. vicinus G. vicinus G. vicinus G. vicinus G. phillipsi G. robustus G. robustus G. G. nigricaudus G. nigricaudus G. nigricaudus G. nigricaudus G Table 1 standard errors. tested with comparative data from other taxa that have
Journal of Biogeography 9 ª 2017 John Wiley & Sons Ltd T. Aghova et al. similar ecological requirements distributions (‘community of the genus Homo. Palaeogeography, Palaeoclimatology, phylogeography; e.g. Hickerson & Meyer, 2008). Several Palaeoecology, 207, 399–420. other small mammals associated with Somali-Masai savanna Bouckaert, R.R. (2010) DensiTree: making sense of sets of habitats (e.g. spiny mice, genus Acomys; grass rats, Arvican- phylogenetic trees. Bioinformatics, 26, 1372–1373. this; meadow mice, Myomyscus; arid-adapted shrews, genus Braconnot, P., Otto-Bliesner, B., Harrison, S. et al. (2007) Crocidura) are suitable models for such studies. The results Results of PMIP2 coupled simulations of the Mid-Holo- of a comparative study could provide robust conclusions cene and Last Glacial Maximum - Part 1: experiments and regarding biogeographical processes operating during the large-scale features. Climate of the Past, 3, 261–277. Plio-Pleistocene in these dry and endangered ecosystems in van Breugel, P., Friis, I. & Demissew, S. (2016) The transi- northeastern Africa. tional semi-evergreen bushland in Ethiopia : characteriza- tion and mapping of its distribution using predictive modelling. Applied Vegetation Science, 19, 355–367. ACKNOWLEDGEMENTS Brouat, C., Tatard, C., B^a, K., Cosson, J.-F., Dobigny, G., This study was supported by two projects of the Czech Science Fichet-Calvet, E., Granjon, L., Lecompte, E., Loiseau, A., Foundation, nos. P506/10/0983 and 15-20229S, the Ministry Mouline, K., Piry, S. & Duplantier, J.-M. (2009) Phylo- of Culture of the Czech Republic (DKRVO 2017/15, National geography of the Guinea multimammate mouse (Masto- Museum, 00023272) and the Russian Foundation for Basic mys erythroleucus): a case study for Sahelian species in Research (project no. 15-04-03801-a). For help during the West Africa. Journal of Biogeography, 36, 2237–2250. field work we acknowledge V. Mazoch, H. Konvi ckov a, J. Bryja, J., Mikula, O., Sumbera, R., Meheretu, Y., Aghov a, T., Vrbov a Kom arkov a, J. Kr asov a, A. H anov a, A. Kone cn y, L. Lavrenchenko, L.A., Mazoch, V., Oguge, N., Mbau, J.S., Cuypers, C. Sabuni, A. Katakweba, A. Massawe, J. Skl ıba, K. Welegerima, K., Amundala, N., Colyn, M., Leirs, H. & Welegerima, A. Ribas, F. Sedl a cek and all local collaborators. Verheyen, E. (2014) Pan-African phylogeny of Mus (sub- We would like to thank also J. Vot ypka and S. Gryseels for genus Nannomys) reveals one of the most successful mam- providing samples. For help with genotyping we acknowledge mal radiations in Africa. BMC Evolutionary Biology, 14, H. Konvi ckov a and A. Bryjov a. A. Dehne-Garcia, G. J. Ker- 256. goat, J. Sm ıd and A. Drummond helped with the design of Burgess, N.D., Butynski, T.M., Cordeiro, N.J., Doggart, N.H., data analysis. We thank P.J.J. Bates, P. van Breugel and I. Friis Fjelds, J., Howell, K.M., Kilahama, F.B., Loader, S.P., for additional information. Most analyses were run on the Lovett, J.C., Mbilinyi, B., Menegon, M., Moyer, D.C., Czech-grid infrastructure METACENTRUM, minor part then Nashanda, E., Perkin, A., Rovero, F., Stanley, W.T. & Stu- on CIPRES Gateway or CBGP cluster. We also thank the cura- art, S.N. (2007) The biological importance of the Eastern tors that allowed us to study the tissue collections in their Arc Mountains of Tanzania and Kenya. Biological Conser- care: J. Phelps and B. Stanley (FMNH), N. Duncan (AMNH), vation, 134, 209–231. D. Moerike (SMNS), and W. Wendelen (RMCA). We also Cerling, T.E., Bowman, J.R. & O’Neil, J.R. (1988) An iso- would like to thank Judith Masters and four anonymous topic study of a fluvial-lacustrine sequence: the Plio-Pleis- reviewers for their useful comments on an earlier version of tocene koobi fora sequence, East Africa. Palaeogeography, the manuscript. Palaeoclimatology, Palaeoecology, 63, 335–356. Cerling, T.E., Wynn, J.G., Andanje, S.A., Bird, M.I., Korir, D.K., Levin, N.E., Mace, W., Macharia, A.N., Quade, J. REFERENCES & Remien, C.H. (2011) Woody cover and hominin Bandelt, H.J., Forster, P. & Rohl,€ A. (1999) Median-joining environments in the past 6 million years. Nature, 476, networks for inferring intraspecific phylogenies. Molecular 51–56. Biology and Evolution, 16,37–48. Chevret, P. & Dobigny, G. (2005) Systematics and evolution Bates, P.J.J. (1988) Systematics and zoogeography of Tatera of the subfamily Gerbillinae (Mammalia, Rodentia, Muri- (Rodentia: Gerbillinae) of north-east Africa and Asia. Bon- dae). Molecular Phylogenetics and Evolution, 35, 674–688. ner zoologische Beitr€age, 39, 265–303. Colangelo, P., Corti, M., Verheyen, E., Annesi, F., Oguge, Berger, W.H. & Jansen, E. (1994) Mid-Pleistocene climate N., Makundi, R.H. & Verheyen, W. (2005) Mitochon- shift: the Nansen connection. The polar oceans and their drial phylogeny reveals differential modes of chromoso- role in shaping the global environment (ed. by O.M. Johan- mal evolution in the genus Tatera (Rodentia: nessen, R.D. Muench and J.E. Overland). AGU Geophysical Gerbillinae) in Africa. Molecular Phylogenetics and Evolu- Monograph Series, 85, 295–311. tion, 35, 556–568. Bivand, R. & Lewin-Koh, N. (2016) maptools: Tools for Colangelo, P., Granjon, L., Taylor, P.J. & Corti, M. (2007) reading and handling spatial objects. R package version Evolutionary systematics in African gerbilline rodents of 0.8-39. Available at https://CRAN.R-project.org/package= the genus Gerbilliscus: inference from mitochondrial genes. maptools Molecular Phylogenetics and Evolution, 42, 797–806. Bobe, R. (2004) The expansion of grassland ecosystems in Colangelo, P., Castiglia, R., Franchini, P. & Solano, E. (2010) Africa in relation to mammalian evolution and the origin Pattern of shape variation in the eastern African gerbils of
10 Journal of Biogeography ª 2017 John Wiley & Sons Ltd Phylogeny of Somali-Masai gerbils
the genus Gerbilliscus (Rodentia, Muridae): environmental Granjon, L., Colangelo, P., Tatard, C., Colyn, M., Dobigny, correlations and implication for taxonomy and systemat- G. & Nicolas, V. (2012) Intrageneric relationships within ics. Mammalian Biology – Zeitschrift fur€ S€augetierkunde, Gerbilliscus (Rodentia, Muridae, Gerbillinae), with charac- 75, 302–310. terization of an additional West African species. Zootaxa, Cowling, S.A., Cox, P.M., Jones, C.D., Maslin, M.A., Peros, 25,1–25. M. & Spall, S.A. (2008) Simulated glacial and interglacial Heled, J. & Drummond, A.J. (2010) Bayesian inference of vegetation across Africa: implications for species phyloge- species trees from multilocus data. Molecular Biology and nies and trans-African migration of plants and animals. Evolution, 27, 570–580. Global Change Biology, 14, 827–840. Hickerson, M.J. & Meyer, C.P. (2008) Testing comparative Craigie, I.D., Baillie, J.E.M., Balmford, A., Carbone, C., Col- phylogeographic models of marine vicariance and dispersal len, B., Green, R.E. & Hutton, J.M. (2010) Large mammal using a hierarchical Bayesian approach. BMC Evolutionary population declines in Africa’s protected areas. Biological Biology, 8, 322. Conservation, 143, 2221–2228. Hijmans, R.J. (2015) raster: Geographic data analysis and Degnan, J. H., & Rosenberg, N. A. (2009). Gene tree discor- modeling. R package version 2.5-2. Available at: https:// dance, phylogenetic inference and the multispecies coales- CRAN.R-project.org/package=raster cent. Trends in Ecology & Evolution, 24, 332–340. Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G. & Jarvis, deMenocal, P.B. (2004) African climate change and faunal A. (2005) Very high resolution interpolated climate sur- evolution during the Pliocene-Pleistocene. Earth and Plan- faces for global land areas. International Journal of Clima- etary Science Letters, 220,3–24. tology, 25, 1965–1978. Dixon, A.P., Faber-Langendoen, D., Josse, C., Morrison, J. & Hijmans, R.J., Phillips, S., Leathwick, J. & Elith, J. (2016) dismo: Loucks, C.J. (2014) Distribution mapping of world grass- Species distribution modeling. R package version 1.0-15. land types. Journal of Biogeography, 41, 2003–2019. Available at: https://CRAN.R-project.org/package=dismo Dobigny, G., Tatard, C., Gauthier, P., B^a, K., Duplantier, Hodell, D.A., Curtis, J.H., Sierro, F.J. & Raymo, M.E. (2001) J.M., Granjon, L. & Kergoat, G.J. (2013) Mitochondrial Correlation of Late Miocene to Early Pliocene sequences and nuclear genes-based phylogeography of Arvicanthis between the Mediterranean and North Atlantic. Paleo- niloticus (Murinae) and sub-Saharan open habitats Pleis- ceanography, 16, 164–178. tocene history. PLoS ONE, 8, e77815. IUCN (2016) The IUCN red list of threatened species. Version Drummond, A.J., Ho, S.Y.W., Phillips, M.J., & Rambaut, A. 2016-2. Available at: www.iucnredlist.org (accessed 4 (2006). Relaxed phylogenetics and dating with confidence. November 2016). PLoS Biology, 4, e88. Katoh, K. & Standley, D.M. (2013) MAFFT Multiple Drummond, A.J. & Rambaut, A. (2007) BEAST: Bayesian Sequence Alignment Software Version 7: improvements in evolutionary analysis by sampling trees. BMC Evolutionary performance and usability. Molecular Biology and Evolu- Biology, 7, 214. tion, 30, 772–780. Drummond, A.J., Suchard, M.A., Xie, D. & Rambaut, A. Kebede, F., Rosenbom, S., Khalatbari, L., Moehlman, P.D., (2012) Bayesian phylogenetics with BEAUti and the BEAST Beja-Pereira, A., & Bekele, A. (2016). Genetic diversity of 1.7. Molecular Biology and Evolution, 29, 1969–1973. the Ethiopian Grevy’s zebra (Equus grevyi) populations Dupont, L. (2011) Orbital scale vegetation change in Africa. that includes a unique population of the Alledeghi Plain. Quaternary Science Reviews, 30, 3589–3602. Mitochondrial DNA Part A, 27, 397–400. Faber-Langendoen, D., Keeler-Wolf, T., Meidinger, D., Tart, Kingdon, J. (1990) Island Africa: the evolution of Africa’s rare D., Hoagland, B., Josse, C., Navarro, G., Ponomarenko, S., animals and plants. Collins, London. Saucier, J.P., Weakley, A. & Comer, P. (2014) EcoVeg: a Lanfear, R., Calcott, B., Kainer, D., Mayer, C. & Sta- new approach to vegetation description and classification. matakis, A. (2014) Selecting optimal partitioning Ecological Monographs, 84, 533–561. schemes for phylogenomic datasets. BMC Evolutionary Galan, M., Page, M. & Cosson, J.F. (2012) Next-gseneration Biology, 14,1. sequencing for rodent barcoding: species identification Lavrenchenko, L.A., Kruskop, S.V., Bekele, A., Belay, G., from fresh. Degraded and environmental samples. PloS Morozov, P.N., Ivlev, Y.F. & Warshavsky, A.A. (2010) ONE, 7, e48374. Mammals of the Babille elephant sanctuary (Eastern Ethio- Geist, H.J. & Lambin, E.F. (2004) Dynamic causal patterns of pia). sRussian Journal of Theriology, 9,47–60. desertification. BioScience, 54, 817. Librado, P. & Rozas, J. (2009) DnaSP v5: a software for com- Gernhard, T. (2008) The conditioned reconstructed process. prehensive analysis of DNA polymorphism data. Bioinfor- Journal of Theoretical Biology, 253, 769–778. matics, 25, 1451–1452. Gouy, M., Guindon, S. & Gascuel, O. (2010) SeaView Ver- Linder, H.P., de Klerk, H.M., Born, J., Burgess, N.D., Fjelds a, sion 4: a multiplatform graphical user interface for J. & Rahbek, C. (2012) The partitioning of Africa: statisti- sequence alignment and phylogenetic tree building. Molec- cally defined biogeographical regions in sub-Saharan ular Biology and Evolution, 27, 221–224. Africa. Journal of Biogeography, 39, 1189–1205.
Journal of Biogeography 11 ª 2017 John Wiley & Sons Ltd T. Aghova et al.
Lorenzen, E.D., Heller, R., & Siegismund, H.R. (2012). Com- Ronquist, F., Teslenko, M., van der Mark, P., Ayres, D.L., parative phylogeography of African savannah ungulates. Darling, A., Hohna,€ S., Larget, B., Liu, L., Suchard, M.A. Molecular Ecology, 21, 3656–3670. & Huelsenbeck, J.P. (2012) MrBayes 3.2: efficient Bayesian Maslin, M.A. & Christensen, B. (2007) Tectonics, orbital phylogenetic inference and model choice across a large forcing, global climate change, and human evolution in model space. Systematic Biology, 61, 539–542. Africa: introduction to the African paleoclimate special Sayre, R., Comer, P., Hak, J., et al. (2013). A new map of volume. Journal of Human Evolution, 53, 443–464. standardized terrestrial ecosystems of Africa. Washington, Maslin, M.A., Brierley, C.M., Milner, A.M., Shultz, S., DC: Association of American Geographers. 24 pp. Trauth, M.H. & Wilson, K.E. (2014) East african climate Sepulchre, P., Ramstein, G., Fluteau, F., Schuster, M., Tierce- pulses and early human evolution. Quaternary Science lin, J.-J. & Brunet, M. (2006) Tectonic uplift and Eastern Reviews, 101,1–17. Africa aridification. Science, 313, 1419–1423. Maslin, M.A., Shultz, S. & Trauth, M.H. (2015) A synthesis Sm ıd, J., Carranza, S., Kratochv ıl, L., Gvo zd ık, V., Nasher, of the theories and concepts of early human evolution. A.K. & Moravec, J. (2013) Out of Arabia: a complex bio- Philosophical Transactions of the Royal Society B: Biological geographic history of multiple vicariance and dispersal Sciences, 370, 20140064. events in the gecko genus Hemidactylus (Reptilia: Gekkoni- McDonough, M.M., Sumbera, R., Mazoch, V., Ferguson, dae). PLoS ONE, 8, e64018. A.W., Phillips, C.D. & Bryja, J. (2015) Multilocus phylo- Stamatakis, A. (2006) RAxML-VI-HPC: maximum likeli- geography of a widespread savanna-woodland-adapted hood-based phylogenetic analyses with thousands of taxa rodent reveals the influence of Pleistocene geomorphology and mixed models. Bioinformatics (Oxford, England), 22, and climate change in Africa’s Zambezi region. Molecular 2688–2690. Ecology, 24, 5248–5266. Thulin, M. (ed.) (1993) Flora of Somalia: Vol. 1 – Pterido- Mikula, O., Sumbera, R., Aghov a, T., Mbau, J.S., Katak- phyta; Gymnospermae; Angiospermae (Annonaceae-Faba- weba, A.S., Sabuni, C.A., & Bryja, J. (2016). Evolution- ceae). Royal Botanic Gardens, Kew. ary history and species diversity of African pouched Thulin, M., Lavin, M., Pasquet, R. & Delgado-Salinas, A. mice (Rodentia: Nesomyidae: Saccostomus). Zoologica (2004) Phylogeny and biogeography of Wajira (Legumi- Scripta, 45, 595–617. nosae): a monophyletic segregate of Vigna centred in the Monadjem, A., Taylor, P.J., Denys, C. & Cotterill, F.P.D. Horn of Africa region. Systematic Botany, 29, 903–920. (2015) Rodents of Sub-Saharan Africa. A biographic and Trauth, M.H., Larrasoana,~ J.C. & Mudelsee, M. (2009) taxonomic synthesis. Walter de Gruyter GmbH, Berlin/ Trends, rhythms and events in Plio-Pleistocene African cli- Munich/Boston. mate. Quaternary Science Reviews, 28, 399–411. Muscarella, R., Galante, P.J., Soley-Guardia, M., Boria, R.A., Trauth, M.H., Maslin, M.A., Deino, A.L., Junginger, A., Kass, J.M., Uriarte, M. & Anderson, R.P. (2014) ENMeval: Lesoloyia, M., Odada, E.O., Olago, D.O., Olaka, L.A., an R package for conducting spatially independent evalua- Strecker, M.R. & Tiedemann, R. (2010) Human evolution tions and estimating optimal model complexity for Max- in a variable environment: the amplifier lakes of Eastern ent ecological niche models. Methods in Ecology and Africa. Quaternary Science Reviews, 29, 2981–2988. Evolution, 5, 1198–1205. Uno, K.T., Pratigya, J.P. & Jackson, K.E. (2016) Neogene Otto-Bliesner, B.L. (2006) Simulating Arctic climate warmth biomarker record of vegetation change in eastern Africa. and icefield retreat in the last interglaciation. Science, 311, Proceedings of the National Academy of Sciences USA, 113, 1751–1753. 6355–6363. Phillips, S.J., Anderson, R.P. & Schapire, R.E. (2006) Maxi- Varshavsky, A.A., Karimova, T.Y. & Neronov, V.M. (2007) mum entropy modeling of species geographic distribu- Specific features of the rodent fauna formation in Somali tions. Ecological Modelling, 190, 231–259. Peninsula. Zoologichesky Zhurnal, 86, 1498–1506. Potts, R. (2013) Hominin evolution in setings of strong envi- Varty, N. (1988) Mammals and problems of their conserva- ronmental variability. Quaternary Science Reviews, 73,1–13. tion in the riverine forests of the Jubba Valley, southern Rambaut, A., Suchard, M.A., Xie, D. & Drummond, A.J. Somalia. Tropical Zoology, 6975, 179–192. (2014) Tracer v1.6. Available at: http://beast.bio.ed.ac.uk/ Warren, D.L. & Seifert, S.N. (2011) Ecological niche model- Tracer. ing in Maxent : the importance of model complexity and Randi, E., D’Huart, J.P., Lucchini, V. & Aman, R. (2002) the performance of model selection criteria. Ecological Evidence of two genetically deeply divergent species of Applications, 21, 335–342. warhthog, Phacochoerus africanus and P. aethiopicus White, F. (1983) Vegetation of Africa – A descriptive memoir (Artiodactyla: Suiformes) in Eastern Africa. Mammalian to accompany the UNESCO/AETFAT/UNSO vegetation map Biology, 67,91–96. of Africa. Natural Resources Research Report XX. Ravelo, A.C., Andreasen, D.H., Lyle, M., Olivarez Lyle, A. & UNESCO, Paris. Wara, M.W. (2004) Regional climate shifts caused by Yang, Z. & Rannala, B. (2014) Unguided species delimitation gradual global cooling in the Pliocene epoch. Nature, 429, using DNA sequence data from multiple loci. Molecular 263–267. Biology and Evolution, 31, 3125–3135.
12 Journal of Biogeography ª 2017 John Wiley & Sons Ltd Phylogeny of Somali-Masai gerbils
SUPPORTING INFORMATION BIOSKETCH
Additional Supporting Information may be found in the Tatiana Aghova is PhD student working on phylogeogra- online version of this article: phy of African savanna ecosystems using rodents as a model organism, under supervision of Josef Bryja and Radim Appendix S1 Details on used individuals. Appendix S2 Additions to materials and methods. Sumbera. Appendix S3 Additions to results. Author contributions: J.B. and R.S. conceived the idea; T.A., R.S., O.M., M.M., L.L., Y.M., J.M. and J.B. participated to DATA ACCESSIBILITY sampling in the field; T.A., L.P. and M.M. genotyped the material. T.A. and O.M. analysed the data, T.A., R.S. and New sequences used in this study are available in GenBank J.B. wrote the first version of the manuscript that was – under accession numbers KU965898 KU966133 and approved by all authors. KY352044–KY352064. Further details on used specimens and museum vouchers are specified in Table S1 in Appendix S1. Editor: Judith Masters
Journal of Biogeography 13 ª 2017 John Wiley & Sons Ltd