The influence of life history and climate driven diversification on the mtDNA phylogeographic structures of two southern African Mastomys species (Rodentia: Muridae: Murinae)

Introduction

Geographically widespread mammal species often have disjunct phylogeographic patterns that are closely linked with paleoecology (Randi, 2007). This scenario is also evident in glaciation-free regions such as Africa (Arctander, Johansen & Coutellec-Vreto, 1999), and was recently corroborated in a study on a broad niche murid rodent, Mastomys natalensis (Colangelo et al., 2013). Six unique geographically defined M. natalensis haplogroups were identified and it was proposed that the structure obtained is the result of habitat vicariance and climate driven diversification. If this holds, we propose that the evolution of Mastomys coucha, an ecologically and morphologically closely related species to M. natalensis (see below), may similarly be affected by these factors. It is, however, well established that life history of species (amongst others, the ability to disperse, the reproduction potential, and habitat specificity) also affects the genetic structure of taxa (Hodges, Rowell & Keogh, 2007; Pelc, Warner & Gaines, 2009; Puritz et al., 2012; Cangi et al., 2013; du Toit, Matthee & Matthee, 2013). Accurate predictions on the phylogeographic patterns of many un-studied species are thus purely speculative.

In southern Africa, the distribution of M. coucha is characterized by three disjunct centres of endemism (Fig. 1). The reason for the disjunct distribution in M. coucha is not apparent, but several contemporary vicariant barriers occur in the southern African landscape. It is likely that these could have influenced connectivity among multimammate mice populations. In particular, the Orange river (Matthee & Flemming, 2002; Bauer & Lamb, 2005; du Toit et al., 2012, 2013), the Zambezi river (Cotterill, 2004; Faulkes et al., 2004; Van Daele et al., 2004, 2007), altitudinal differences associated with the Great Escarpment (Matthee & Robinson, 1996; Edwards et al., 2011; Schwab et al., 2011; du Toit et al., 2012), and high rainfall variability between the wet eastern and dryer western regions (Rambau, Robinson & Stanyon, 2003; Linder et al., 2010; du Toit et al., 2012; Montgelard & Matthee, 2012; Willows-Munro & Matthee, 2011) could have all influenced the phylogeographic structure of M. coucha (Fig. 1). In addition to these influences, climatic oscillations are also regularly put forward as contributing factor for the diversification of mammals in southern Africa (Matthee & Robinson, 1996; Russo, Chimimba & Bloomer, 2010; Engelbrecht et al., 2011; du Toit et al., 2012; Montgelard & Matthee, 2012).

To test the prediction that vicariance and climatic changes are important mechanisms in shaping the evolution of M. coucha, we embarked on a regional study in which samples were obtained from various localities spanning the known listed biogeographic barriers in southern Africa (see Fig. 1). In addition, to test the effect of life history on evolutionary patterns, we also included representatives of M. natalensis sampled from the same region. The two Mastomys species differ in chromosome number (Granjon et al., 1997 and references therein) and in isozymes and allozymes (Smit & van der Bank, 2001). They are, however, virtually identical in external appearances and they also overlap partly in their distribution (Fig. 1). The distribution of M. natalensis, however, may be influenced by the presence of M. coucha through interspecific competition (Taylor, 1998; Jackson & van Aarde, 2004) and M. natalensis rodents tend to be more abundant in areas of greater rainfall (Taylor, 1998). Mastomys natalensis rodents also have lower population densities at high altitudes (and appear to be totally absent above 2100 m: Kingdon, 1974; Venturi et al., 2004; Makundi, Massawe & Mulungu, 2007) and circumstantial evidence suggest that the two species differ in their reproductive abilities. Population numbers of M. natalensis are highly fluctuating (Leirs, 1994; Venturi et al., 2004; Skinner & Chimimba, 2005; Makundi et al., 2007), while they tend to be more stable for M. coucha (Jackson & van Aarde, 2004).

By studying the comparative phylogeography of two closely related Mastomys species we aimed to provide additional insights into the role of life history, vicariance and climate-driven diversification in the region. Considering the conclusions reached by Colangelo et al. (2013) we hypothesized that the geographic genetic pattern of both Mastomys species might similarly be affected by vicariance and climatic instability in southern Africa.

Material and Methods

Results

Within southern Africa, we detected 13 new unique haplotypes for M. natalensis and 24 for M. coucha (a total of 545 bp were analysed; GenBank accession numbers: M. coucha: KJ466152 – KJ466182 and M. natalensis: KJ466183 – KJ466195). Haplotypic diversity (h) values were generally high, but when the two species were compared, it was lower for M. natalensis (h = 0.715) when compared with M. coucha (h = 0.849). Nucleotide diversity (π) show marked differences between the two species, with M. natalensis much lower (π = 0.003) than M. coucha (π = 0.468). Although the AMOVA analyses indicated lower levels of population differentiation for M. natalensis when compared with M. coucha, the majority of pairwise comparisons for both species showed significant differences between sampled populations (Table 2).

SplitsTree and TCS analysis based on the COI haplotypes within M. coucha showed a large degree of haplotype sharing throughout the range (indicated by C(III); Fig. 2). Some geographic structuring of private haplotypes was, however, evident when looking within the region. Individuals sampled along the eastern and western regions of the subcontinent are respectively recognized by clusters containing unique haplotypes (indicated by C(I) and C(II) on the western side of the continent vs. C(IV) on the eastern side; Fig. 2). In southern Africa, M. natalensis also showed a large amount of haplotype sharing throughout the range but, again, some geographically defined sub-clustering was evident (Fig. 3). For example, N(II) individuals were sampled along the southern African eastern lowlands, and N(IV) represented individuals sampled exclusively in Namibia (Fig. 3). The only sampled locality in southern Africa that showed no haplotype sharing with any other locality comprised individuals from East London N(III). At the continental scale, our COI sequence data confirmed the cytochrome b data presented by Colangelo et al. (2013) in that N(VII) represented individuals from north-west Africa; N(VI) consisted of individuals sampled throughout central Africa; and N(V) represented individuals from east Africa (Fig. 3).

There was significant but weak evidence for isolation by distance in both M. natalensis (R² = 0.089, P ≤ 0.05) and M. coucha (R² = 0.028, P ≤ 0.05; Figs S1, S2) and this result supports the intra-clade pattern previously observed for M. natalensis only (Colangelo et al., 2013). The wide occurrence of a common haplotype throughout the ranges of both species would suggest recent expansion events. This is supported by the significant and negative Fu's Fs values for both species (M. coucha Fu's Fs = −13.20, P ≤ 0.05; M. natalensis Fu's Fs = −22.76, P ≤ 0.05) coupled to the results of the mismatch distribution (M. coucha SSD = 0.00, P > 0.05; M. natalensis SSD = 0.01, P > 0.05; Figs S3, S4). Using the optimal model (HKY + G) as part of the Bayesian Skyline analyses, there was little indication that the population expansion could be correlated with the last glacial maximum approximately 18 thousand years ago (Clark et al., 2009). The two species, however, differ in the slope of the line indicative of a gradual expansion in M. natalensis and a more constant effective population size within M. coucha at least during the last 60 000 years (Fig. 4).

Discussion

At the larger continental scale, the results obtained from the COI gene support the distinct M. natalensis phylogroups detected by Colangelo et al. (2013). Within southern Africa, however, we found no conclusive evidence to support geographic vicariance as a contributing factor towards Mastomys diversification. The results from our comparative phylogeographic study rather favours the hypothesis that over smaller geographic scales, the evolution of M. natalensis and M. coucha are more likely influenced by climate driven diversification coupled to life history attributes.

Both Mastomys species included in this study are characterized by extensive haplotype sharing throughout the southern African range. The absence of any clear phylogeographic structure in the regional data is indicative of the fact that none of the previously defined geographic barriers have a strong effect on female geneflow in M. natalensis or M. coucha. Interestingly, this is in sharp contrast to the patterns previously obtained for southern African terrestrial vertebrate taxa where vegetation (associated with rainfall), rivers, and mountainous elevations were regarded as contemporary barriers to dispersal (Matthee & Flemming, 2002; Edwards et al., 2011; Schwab et al., 2011; du Toit et al., 2012, 2013; Fig. 1). It is important to realize, nevertheless, that in the majority of the published cases, the effectiveness of the documented barriers are closely linked to other ecological and life history factors.

In the absence of any strongly defined vicariant barriers affecting M. coucha and M. natalensis dispersal, we propose that the significant differences detected between sampling sites in southern Africa is largely the result of climate driven diversification coupled to weak but significant isolation by distance detected in the data (also see Colangelo et al., 2013). This is particularly evident when considering the unique haplotypes confined to certain geographic areas. The latter is sufficient to result in a signature of significant population differentiation between the majority of sampling sites, and this pattern is particularly strong in M. coucha (Table 2). When considering the extensive haplotype sharing throughout the region (present in both species), and this is combined with the pattern of isolation by distance, and the fairly large number of private haplotypes restricted to each region (South Africa and Namibia), it is possible to speculate on the mechanisms that could have resulted in such an outcome. Mastomys natalensis and M. coucha are regarded as ecologically tolerant species (Coetzee, 1975; Ntshotsho et al., 2004) and have the ability to proliferate (Jackson & van Aarde, 2004). If this holds, then the dramatic changes in vegetation cover (expansion and contraction of C₃ and C₄ grasses) as a result of climatic oscillations (deMenocal, 2004; Chase & Meadows, 2007), were clearly not severe enough to confine the species to single refugia. It is thus more likely that multiple Mastomys refugia persisted during unfavourable conditions. The mtDNA patterns obtained in the present study reflect periods of isolation among populations (inferred from the unique haplotypes within regions: M. coucha Fig. 2 – C(I), C(II) and C(IV); and M. natalensis Fig. 3 – N(II), N(III) and N(IV)) and periods of secondary contact (inferred from the large number of individuals sharing a single haplotype: M. coucha Fig. 2 – C(III); and M. natalensis Fig. 3 – N(II), N(I)) (also see Willows-Munro & Matthee, 2011; du Toit et al., 2012; Montgelard & Matthee, 2012). During more favourable periods, populations expand (partly supported by Fu's Fs and the mismatch analyses). The Bayesian Skyline Plots suggest that the last expansions predate the last glacial maximum 19 thousand years ago (Fig. 4), and for M. natalensis, this finding is in agreement with the published report for this haplogroup (Colangelo et al., 2013). In totality, this outcome effectively support the hypothesis of Montgelard & Matthee (2012) who suggested that although climatic oscillations are important in generating genetic structure of rodents in the region, it is not possible to implicate a single glaciation event as a vicariance factor.

When the nucleotide diversity is compared between the two species there are marked differences. Regionally, M. natalensis show similarly low levels of intra-phylogroup nucleotide diversity to those reported previously (Colangelo et al., 2013). In contrast, the haplotypes of M. coucha are more divergent from each other showing a higher number of site changes among haplotypes (Figs 2, 3), and this is also reflected in the AMOVA analyses where all pairwise comparisons between sampling sites show significant differentiation (Table 2). As both taxa are separated by the same documented potential barriers to dispersal, and were exposed to similar climatic oscillations, we propose that the differences in genetic diversity between M. natalensis and M. coucha can mainly be attributed to life history differences between the two species (also see Hodges et al., 2007; Puritz et al., 2012). Throughout the landscape, these two species occupy different niches in the environment (Ntshotsho et al., 2004) but both are characterized by exceptionally high propagation rates (Coetzee, 1975). It has, however, been documented that the reproductive success of M. natalensis is influenced by the fact that the species can breed on comparatively lower quality diets when compared to M. coucha (Jackson & van Aarde, 2004). It thus seems reasonable to speculate that M. coucha is more of a habitat specialist when compared to M. natalensis, and this phenomena can result in a comparatively higher degree of structure (lower levels of connectivity) between sampled populations of M. coucha (Table 2).

Genetic diversification among regional populations of a number of southern African rodent species have now been examined and strong genetic differentiation is nearly always detected (Matthee & Robinson, 1997; Rambau et al., 2003; Russo et al., 2010; Engelbrecht et al., 2011; du Toit et al., 2012). The initial predictions of finding similar mtDNA structures in Mastomys are however not supported by this study. It is thus likely that the ability of M. natalensis and M. coucha to survive in multiple refugia during unfavourable periods can possibly be attributed to the lineages' ability to increase population numbers rapidly, and to exploit a variety of habitat types. The outcome of this study, compared with previously published work from the same region, highlights the need for additional case studies to be performed.