Recent genetic studies, using maternally inherited mitochondrial DNA, indicate a complex evolutionary history for baboons Papio spp. in general, and for eastern African baboons in particular. To further address this topic and to improve our understanding of phylogeographic patterns of baboons in eastern Africa, mitochondrial cytochrome b sequence data were analysed from 148 baboon samples from 103 locations in eastern Africa. The resultant phylogenetic reconstructions suggest an initial split of baboons into four main clades: southern chacma baboons, baboons from Mahale Mountains in Tanzania, main southern, and main northern. We confirm that the boundary between southern and northern clades lies along the Ugalla-Malagarasi River and Ruaha-Rufiji River of central Tanzania. We detected new mitochondrial haplogroups, most notably the Mahale Mountains clade, and refined haplogroup distributions. The evolutionary divergence of baboons in eastern Africa was most likely triggered and maintained by numerous episodes of population division and reconnection, probably related mainly to climate change. To better understand these processes, nuclear DNA information is required, especially to assess gene flow among populations.
Baboons (genus Papio Erxleben, 1777) are widely distributed over most of sub-Saharan Africa, occupying the greatest diversity of habitats of any genus of non-human primates. Baboons are found in all terrestrial habitats from moist forest to the edge of deserts, and from sea level to >3300 m (Swedell, 2011; Butynski et al., 2013; Jolly, 2013). In the savannas and woodlands of eastern Africa, from northeast Sudan southward to Malawi and Zambia, baboons are, second to humans, the most abundant catarrhine species. Several taxa of Papio have been described for eastern Africa (Elliot, 1913; Napier & Napier, 1967; Hill, 1970). Their validity and taxonomic ranks are, however, still disputed (Jolly, 1993, 2013; Sarmiento, 1998; Groves, 2001; Grubb et al., 2003; Butynski et al., 2013). The main eastern African forms are olive baboon Papio anubis (Lesson, 1827) and yellow baboon Papio cynocephalus (Linnaeus, 1766) (figure 1; appendix 1).
Across the distribution of baboons in Africa, morphological clines (Frost et al., 2003; De Jong & Butynski, 2009) and evidence of interspecific hybridization in contact zones (P. kindae x P. ursinus griseipes, e.g. Jolly et al., 2011; P. anubis x P. cynocephalus, e.g. Charpentier et al., 2012; P. hamadryas x P. anubis, e.g. Bergey, 2015) complicate this problem. Taxa are identified primarily by the colour, length, and texture of the pelage (including their mane), body size, body shape, and skull morphology (Hill, 1970; Kingdon, 1971, 2015; Jolly, 1993; Rowe, 1996; Alberts & Altmann, 2001; Groves, 2001; De Jong & Butynski, 2009, 2012; Butynski et al., 2013) (see appendix 1 and visit http://wildsolutions.nl/photomaps/Papio/ to view a large selection of photographs of Papio spp. with localities depicted on an interactive digital map). In eastern African baboon populations, comparisons of cranial and dental morphologies indicate that the morphotypes are distinct, but that intermediate forms do exist (Hayes et al., 1990; Frost et al., 2003; Jolly, 2003). Molecular studies, mainly applying mitochondrial (mt) markers, have not solved this “taxonomic tangle” (Groves, 2001). Phylogenetic reconstructions, based on parts or even complete mt-genomes, reveal several mt-haplogroups or clades. These, however, are only marginally concordant with the morphological variation or taxa (Newman et al., 2004; Zinner et al., 2009a, 2011, 2013).
The deepest split in the phylogenetic tree of Papio divides the genus into two main mtDNA clades, a northern clade and a southern clade (Burrell, 2008; Zinner et al., 2009a, 2013). The northern clade includes Guinea baboon P. papio (Desmarest, 1820) from West Africa, hamadryas baboon P. hamadryas (Linnaeus, 1758) from the Horn of Africa and southwest Arabia, P. anubis from central and northeast Africa, and P. cynocephalus from Somalia and southeast Kenya, as well as east and central Tanzania. The southern clade includes P. cynocephalus from south Tanzania to north Mozambique, as well as Kinda baboon P. kindae (Lönnberg, 1919) from Zambia, and chacma baboon P. ursinus (Kerr, 1792) from southern Africa. The two main mtDNA clades come into contact within the distribution of P. cynocephalus in central Tanzania, along a line following the Ruaha-Rufiji River from the coast inland westward to the eastern shore of Lake Tanganyika at the mouth of the Malagarasi River (5.25°S, 29.81°E; Zinner et al., 2009a). No morphological differences have been reported among P. cynocephalus within this contact area. Hill (1970) reports only P. cynocephalus cynocephalus for Tanzania (figure 2). Others report P. anubis in northwest Tanzania (Groves, 2001; Anandam et al., 2013; Butynski et al., 2013) and P. kindae in central west Tanzania (Butynski & De Jong, 2009; De Jong & Butynski, 2012). Based on present knowledge, the Ibean baboon P. cynocephalus ibeanus Thomas, 1893 of Somalia and Kenya, and the Nyasa baboon P. cynocephalus strepitus Elliot, 1907 from Malawi (Hill, 1970), are distributed at least 500 km north and south, respectively, from where the two main mtDNA clades meet in central Tanzania.
Although the Papio mtDNA phylogeny has been recently intensively studied, only a few samples from eastern Africa have been analysed. In particular, central Tanzania, where the southern and northern clades meet, has not been sampled in detail. To fill this geographic sampling gap, this study focused on P. cynocephalus in Tanzania, Malawi and Zambia, but also includes samples from P. anubis and P. hamadryas in Eritrea, Ethiopia, Kenya, Uganda, Democratic Republic of Congo (DRC), and northwest Tanzania. The aim of this study is to clarify the geographic distribution of Papio mtDNA clades in eastern Africa, and to relate and combine these clades with previous maps of Papio distributions (e.g. Hill, 1970) and photographs of Papio from the region (e.g. appendix 1).
MATERIAL AND METHODS
We obtained 147 non-invasively collected baboon faecal samples from sites in eastern Africa (figure 2, appendix 2). To these we added one P. hamadryas tissue sample from southwest Saudi Arabia and one museum tissue sample from Somalia [see Zinner et al. (2008) and Kopp et al. (2014) for detailed information on these samples]. The samples originate from 104 locations (103 in eastern Africa and one from southwest Saudi Arabia). Of the 149 samples, 45 were used in previous studies, including the samples from Saudi Arabia and Somalia (Zinner et al., 2008, 2009a; Kopp et al., 2014). For those 45 samples we retrieved respective mtDNA sequence information from GenBank. We also included mtDNA sequence information for two southern chacma P. ursinus ursinus, two grey-footed chacma baboons P. ursinus griseipes Pocock, 1911, two P. papio, and, as outgroup, gelada Theropithecus gelada (Rüppell, 1835), from GenBank (accession numbers in appendix 2).
Faecal samples were collected and stored following the two-step protocol of Roeder et al. (2004) and Nsubuga et al. (2004). Samples were stored at ambient temperature for up to 6 months in the field and at -20°C upon arrival in the laboratory. For each sample, consecutive number, date, location, and GPS coordinates were recorded. Sample collection complied with the laws of the respective countries of origin and Germany, and with the guidelines of the International Primatological Society.
We extracted DNA from 104 faecal samples using the QIAamp DNA Stool Mini Kit (Qiagen, Hilden, Germany) and First DNA All Tissue Kit (Gen-Ial, Troisdorf, Germany) according to the manufacturers' protocols. DNA was eluted in 20μl water (HPLC grade) and stored at -20°C until further processing. To avoid contamination, all working steps (i.e. DNA extraction, PCR set-up, PCR amplification, gel electrophoresis, gel extraction and sequencing) were performed in separate laboratories. All PCR reactions were performed with negative (HPLC-grade water) controls.
We amplified and sequenced the complete mitochondrial cytochrome b gene (cyt b, 1140 bp). This allowed us to include published data in the statistical analyses. In addition, this marker can be reliably amplified from low quality samples, such as faecal material and tissue samples from museum specimens. We used established protocols (Zinner et al., 2009a) and amplified cyt b via two over-lapping fragments to ensure that sequences were obtained even if DNA was degraded. Zinner et al. (2009a) showed that the primers and PCR conditions applied here solely amplify mtDNA and not nuclear mitochondrial pseudogenes (numts). PCR reactions with a total volume of 30 μl included 1 U BiothermTaq 5000 DNA polymerase (Genecraft, Cologne, Germany), 1x reaction buffer, 0.16 mM dNTPs, and 0.33μM of each primer. PCR conditions for amplification comprised a pre-denaturation step at 94°C for 2 min, followed by 40 cycles at 94°C for 1 min, 60°C for 1 min, 72°C for 1 min, and a final extension step at 72°C for 5 min. Results of the PCR amplifications were checked on 1% agarose gels. PCR products were cleaned with the Qiagen PCR Purification Kit and subsequently sequenced on an ABI 3130xL sequencer using the BigDye Terminator Cycle Sequencing Kit (Applied Biosystems, Foster City, USA) and both amplification primers.
Sequences were checked, edited, and aligned manually using BIOEDIT 22.214.171.124 (Hall, 1999). The final alignment comprised 156 sequences of which 149 derived from eastern African and Saudi Arabian baboons, two sequences each from P. papio, P. u. ursinus, and P. u. griseipes, and as outgroup, one sequence from T. gelada. All sequences were deposited in GenBank (details of samples and accession numbers are given in Table S1).
Phylogenetic trees were reconstructed with maximum-likelihood (ML) and Bayesian methods using RAxML 8 (Stamatakis, 2014) as implemented in raxmlGUI 1.5b.1 (Silvestro & Michalak, 2012) and MrBayes 3.2.6 (Ronquist et al., 2012), respectively. ML calculations were run with the GTR-CAT-I model and 1,000 rapid bootstrapping replications. For Bayesian tree reconstructions, we conducted four Markov Chain Monte Carlo (MCMC) runs with a default temperature of 0.2 and the HKY+I+G model as selected as best-fit model in jModeltest 2.1.7 (Darriba et al., 2012) under the Bayesian information criterion (BIC) and the Decision Theory Performance-based Selection (DT). All repetitions were run for 1 million generations with tree and parameter sampling setting in every 100 generations. The first 25% of samples were discarded as burn-in. The adequacy of the burn-in and convergence of all parameters was assessed via the uncorrected Potential Scale Reduction Factor (PSRF) (Gelman & Rubin, 1992) as calculated by MrBayes and by visual inspection of the trace of the parameters across generations using TRACER 1.6 (Rambaut et al., 2014). To check whether posterior clade probabilities were also converging, AWTY (Nylander et al., 2008) was applied. Posterior probabilities for each split and a phylogram with mean branch lengths were calculated from the posterior density of trees. Trees were visualized and edited in FigTree 1.4.2. Additionally, a haplotype network was built in POPART 1.7 (Leigh & Bryant, 2015) using the median-joining network algorithm (Bandelt et al., 1999).
Among the 149 eastern African and Arabian baboon sequences, we detected 69 unique cyt b haplotypes. Phylogenetic analyses were conducted with these unique sequences, plus the additional six non-eastern African sequences and the outgroup sequence. The final alignment comprised 76 sequences. Phylogenetic trees derived from Bayesian inference and the ML algorithm yielded identical tree topologies and mostly well-supported terminal clades (ML bootstrap values: >75%, Bayesian posterior probabilities: >0.95) (figure 3). The analysis suggests that baboons initially divided into four clades [southern chacma, Mahale Mts (central west Tanania), main southern, main northern], although the branching pattern among them remains largely unknown. The main southern clade contains Kinda, northern chacma, and southern yellow baboons, while the main northern clade comprises Guinea, hamadryas, eastern olive, and northern yellow baboons.
For eastern Africa, our reconstruction shows nine well-supported mt-clades or haplogroups. These indicate paraphyletic or polyphyletic relationships within at least the two main East African baboon taxa (P. cynocephalus and P. anubis). The nine mt-clades (figures 3 & 4) are: (1) Mahale Mts baboons; (2) Kinda baboons; (3) Luangwa Valley (Zambia) and west Malawi baboons (with northern chacma baboons as a sister clade); (4) southern yellow baboons 1 from southwest Tanzania, north Zambia, and south Malawi; (5) southern yellow baboons 2 from southeast Tanzania and south Malawi east of the Shire River; (6) yellow baboons from northeast Tanzania and olive baboons from central Ethiopia; (7) olive baboons from central Ethiopia; (8) olive baboons from Uganda, east DRC and west Tanzania, and one yellow baboon from the Ugalla region; and (9) hamadryas baboons from Saudi Arabia, Eritrea, and Ethiopia, and eastern and northern clades of olive baboons and yellow baboons from Eritrea, Ethiopia, Somalia, Kenya, and north Tanzania. Clades 2–5 represent the main southern East African clade, while clades 6–9 represent the main northern East African clade. Within clade 9, distinctions can be made among possible clades 9a to 9e. Hamadryas baboons cluster with olive baboons from Eritrea (9a) and Ethiopia (9b and 9c). Olive baboons from east of the Ethiopian Rift Valley form another clade (9d). Northern yellow baboons from Somalia, Kenya and north Tanzania cluster with olive baboons from Kenya and north Tanzania (9e).
The nine mt-clades are also visible within the haplotype network (figure 4), and the subdivision within clade 9 becomes even more obvious than in the tree reconstruction. Among the northern East African clades, olive baboon haplotypes appear in every clade—usually in combination with yellow baboon or hamadryas baboon haplotypes. Interestingly, hamadryas and yellow baboons do not share the same clade or subclade.
Identical mt-haplotypes were mainly found in samples collected at the same locality (same social group or neighbouring groups), but identical sequences were found at different sites and in different taxa. Median distance between sites with identical sequences was 63 km (quartiles 39–121; n=49 sites with distance >10 km). Largest distances among sites (figures 2 & 5) with identical haplotypes were between Webi Shebelli River, south Somalia (Web, yellow baboon) and Lolldaiga Hills Ranch, central Kenya (Lo3, olive baboon; ca. 950 km), and between Serengeti National Park, north Tanzania (Ni2, olive baboon) and Dodoma-Iringa, central Tanzania (nIg, yellow baboon; ca. 550 km).
There are unresolved relationships between the two main East African clades, and with the southern chacma and Mahale Mts clades (figures 3 & 4). The major split separates southern and northern clades. This is the first study to reveal the Mahale Mts clade (as no samples from Mahale Mts were included in previous studies).
We confirm the previously assumed boundary between the distributions of southern and northern clades in Tanzania (figure 5; Zinner et al., 2011). This boundary runs from the lower Rufiji River in the east across central Tanzania to the mouth of the Malagarasi River at Lake Tanganyika in the west. These two large rivers are biogeographic barriers, or at least boundaries, for a number of genera, species, and subspecies of terrestrial mammals (Kingdon, 1971–1982, 2015; Kingdon et al., 2013; Butynski & De Jong, in press). As concerns primates:
- The Ugalla-Malagarasi River appears to be a barrier between P. anubis (north) and P. cynocephalus (south) (Kano, 1971).
- The Ruaha-Rufiji River seems to be the south limit in coastal Tanzania for the Zanzibar dwarf galago Galagoides zanzibaricus (Matschie, 1893) (Butynski et al., 2006, 2013; De Jong, 2012; Butynski & De Jong, in press).
- The Ruaha-Rufiji River appears to be the north limit for the Mozambique dwarf galago Galagoides granti (Thomas & Wroughton, 1907) (Butynski et al., 2006, 2013; De Jong, 2012; Butynski & De Jong, in press).
Our phylogenetic reconstruction and the haplotype network reveal nine reasonably wellsupported mt-haplogroups or clades within eastern African baboons. Tree topology and statistical support values suggest five clades for southern East African baboons: Mahale Mts, Kinda, Luangwa Valley, and two southern yellow clades. Luangwa Valley baboons form the sister group of northern chacmas, whereas southern chacmas are either basal among baboons or represent the first split in the southern baboon clade (Zinner et al., 2009a, 2013). For the main northern East African clade, the situation looks different. Here we found only three well-supported haplogroups, but a diverse clade consisting of hamadryas, eastern olive, and northern yellow baboons stretching from Eritrea and southwest Saudi Arabia south to the Ruaha-Rufiji River in central Tanzania (clade 9). Within the distribution of clade 9, identical haplotypes occur at sites that are almost 1,000 km apart and in different baboon species (e.g. yellow and olive baboons [this study], and olive and hamadryas baboons [Hapke et al., 2001]). In general, the distribution pattern of baboon haplotypes appears to be in a geographic cline—irrespective of their taxonomic relationships.
Mapping the approximate distributions of the mt-haplogroups reveals some disjunct distributions. Among southern East African baboons, clade 4 was found in southeast Tanzania and northeast Zambia, but also east of the Shire River in south Malawi. Between these two distributions, there are haplotypes belonging to clade 3 (Luangwa Valley baboons). Among northern East African baboons, haplotypes of clade 6 occur in yellow baboons at the south Kenyan and north Tanzanian coast, and in olive baboons in west Ethiopia. A possible explanation for such disjunct distributions is that they are relicts of a former wider distribution; the respective haplotypes are now extinct, having been replaced by haplotypes of other clades or haplogroups in parts of the former distribution. Alternatively, the disjunct pattern is a result of incomplete geographic sampling. Increasing the number of samples from areas in between the two distributions might resolve this question.
As in previous studies, the mt-clades only partly match with recognized baboon taxa, most obvious in the case of southern and northern yellow baboons. Their mt-genomes diverged around 2 million years ago (Zinner et al., 2013), though no taxonomic differentiation was reported among yellow baboons from northeast and southeast Tanzania; both are regarded as P. c. cynocephalus. Several of the clades, however, occur in areas where Hill (1970) identified different morphotypes or subspecies of baboons, e.g. Luangwa Valley baboon P. c. (u.) jubilaeus Schwarz, 1928 and central olive baboon P. a. tesselatum Elliot, 1909 (figure 2, appendix 1). Whether P. c. strepitus Elliot, 1907 is at least partly equivalent with our clade 4 needs further investigation. MtDNA information is often used to delimit species within taxonomic groups (e.g. barcoding; Blaxter, 2004; Galimberti et al., 2015; Raupach et al., 2016). This study, however, indicates that (at least) the two main East African baboon taxa are paraphyletic or polyphyletic groups. As such, mtDNA information alone cannot delimit species within Papio in East Africa.
The geographic pattern of mt-haplogroup distributions suggests a complicated biogeographic history for baboons. Like other savanna-living mammals (Lorenzen et al., 2012), baboons were impacted by multiple cycles of expansion and retreat of savanna biomes during Pleistocene glacial and inter-glacial periods, and other climatic changes (Maslin et al., 2015; Trauth et al., 2015). Recurrent fragmentation and reconnection of populations, extinctions, and distribution shifts of demes likely led to multiple phases of isolation, hybridization, and introgression among populations (Zinner et al., 2009a, 2011).
The geographic distribution pattern of mt-haplogroups can provide insights into the biogeographic history of taxa (Avise, 2000, 2004). We found better supported clades, and geographically more confined clades, among southern East African baboons than among northern East African baboons. This suggests that southern East African baboons were affected more by ecological change, leading to longer periods of isolation (thereby hampering movement among demes and/or dispersal into new areas), than northern East Africa baboons. Northern East African baboons, on-the-other-hand, seem to have experienced more introgression due to more frequent reconnection of populations. Here the possible introgression of olive baboons into hamadryas and yellow baboons might be of particular interest (see Zinner et al., 2011 for a more detailed scenario). There is some evidence that ancient introgression occurred even beyond the species level, as found in kipunji Rungwecebus kipunji (Ehardt, Butynski, Jones & Davenport, 2005) (Jones et al., 2005). Individuals from the Mt Rungwe population carry mitochondrial DNA-sequences that are highly similar to those in south Tanzanian yellow baboons (Zinner et al., 2009b), whereas the Udzungwa Mts population seems not to be affected by baboon introgression (Roberts et al., 2010).
It also might be that at least part of the difference between southern East African and northern East African populations is the result of incomplete geographic sampling. It is, therefore, important to collect more samples in under-sampled areas: (1) central and west Kenya, easten and central Uganda, south South Sudan into Ethiopia; (2) Somalia, north coast of Kenya; (3) west Uganda, east DRC; (4) southwest Tanzania; (5) north Mozambique, east of Lake Malawi, and (6) Mahale Mts, Ugalla, east DRC west of Lake Tanganyika. The latter is of particular interest since the Mahale Mts baboons share morphological traits with Kinda baboons. Here are two scenarios for this similarity: (1) Kinda baboons once occurred east of Lake Tanganyika as far north as Mahale Mts; (2) during times of extreme lake level lowstands a land bridge connected the east and west shores of Lake Tanganyika (Scholz & Rosendahl, 1988; Lezzar et al., 1996; Cohen et al., 1997; Nevado et al., 2013) that enabled Kinda baboons from the west side to come into contact with yellow baboons from the east side—and the possibility of gene flow.
This study provides further insights into the evolutionary history of eastern African baboons and confirms the Ugalla-Malagarasi River and Ruaha-Rufiji River as the boundary between the main southern East African and main northern East African baboon clades. Subclades are obvious in both of these main clades. Interestingly, the distribution of the subclades seems to be more geographically structured in southern East African baboons, while a more clinal pattern is evident in northern East African baboons. A possible explanation is that southern East African baboons were historically affected more by population division due to changing ecological conditions, while northern East African baboons were affected more by frequent reconnection and gene flow among populations.
Although we obtained new information and insights on the phylogeographic pattern of baboons in eastern Africa, there are several topics that remain to be addressed. For instance, the disjunct distribution of several subclades could be a relict of a former wide distribution, but the possibility that the distribution pattern of subclades is due to incomplete geographic sampling cannot be ruled out. Likewise, it remains unknown (1) how Mahale Mts baboons are related to Kinda baboons on the west shore of Lake Tanganyika; (2) which yellow baboon haplotypes occur around Lake Malawi, particularly in Mozambique, and (3) how baboons in central and west central Ethiopia are related to baboons in Uganda and DRC. To address these topics, additional sampling is required. Further, since gene flow and hybridization among baboon populations most likely occurred repeatedly, nuclear DNA information, ideally nuclear genome data, are required to better reconstruct the evolutionary history and phylogeography of Papio. In combination with baboon genomics, spatial and ecological modelling of historic and current distributions of Papio taxa and clades will provide insights into habitat preferences (niches) and adaptations.
We thank the national administrations of countries in which samples were collected. We are particular grateful to the University of Dodoma, TAWIRI, and COSTECH in Tanzania (Research Permit issued 24th September 2008, 2008-272-NA-2008-64), and ZAWA in Zambia (Research Permit issued 13th July 2007, Export Certificate 212007007240). We are grateful to the many colleagues and field assistants who helped collect the samples, to Lorna Depew and two anonymous reviewers for their comments on the manuscript, and to Hakan Pohlstrand, Chris Roche, Katerina Guschanski, Ian Salisbury, Dean Gaffigan, Celesta von Chamier, Daniel Cara, Colleen Begg, Anna Weyher, Jim Auburn, and Mike Haworth for the use of their photographs in appendix 1. Jonathan Kingdon kindly allowed the reproduction of his drawings of Papio anubis and Papio cynocephalus in figure 1. This paper is dedicated to Jonathan Kingdon on the event of his 80th birthday and in gratitude for all that he has accomplished on behalf of the conservation of East Africa's mammals.
Supporting information for 22 photographs of adult male baboons from 21 sites in eastern Africa. The baboons in these 22 photographs represent five Papio species and 11 Papio subspecies. The taxonomy followed here is that of Hill (1970), except for the Kinda baboon, which is here recognized as species P. kindae. Numbers in the first column correspond to the numbers on the map (appendix 1b) and with the photographs (appendix 1c). More than 650 photographs of baboons in the wild from many sites across Africa can be viewed at: http://wildsolutions.nl/photomaps/Papio/
List of Papio spp. samples. The following information is given: taxon (based on morphotype and location), sample ID, country and site of origin (coordinates in decimal degrees), haplotype ID, and GenBank accession number. Abbreviations for taxa are given at the end of this table.