Fig. S1 Results from BARRIER after removing the effect of isolation by distance (IBD) across Kenya (above) and within Mpala Ranch (below), using the sibship-controlled dataset. Barriers are shown in red, and are labelled with capital letters, with numbers at the end of each barrier indicating the number of loci supporting that section of the barrier. The thickness of the barrier is proportional to the number of loci supporting the observed split. The figure shows that the genetic discontinuities observed at both geographic scales are not the result of sampling artefacts, with all main significant barriers still present even after accounting for IBD patterns.

Fig. S2 The Kenyan Rift Valley (RV) is a complex orographic system spanning throughout most of middle and western Kenya (green boundary). Three sub-regions of genetic diversity identified using Bayesian population structure and barrier analyses were detected within the RV. The Western RV (blue), Central RV (red), and Eastern RV (yellow) represent areas within which gene flow is mainly localised. Sampled localities across Kenya are labelled with numbers.

Table S1 All sampled localities with their full names, and ID as shown in Figures 4 and 2 (locations 1-11; and 12-28 respectively). Populations within MR are highlighted in bold. Universal Transverse Mercator geographic coordinates (UTM) taken at the centre of each sampled locality using a handheld Global Positioning System (GPS; Garmin ETREX). D represents the pairwise geographic distance in km between sites taking Marich Pass (12) as the reference point. N is the sample size collected at each locality, and A is the altitude in meters above the sea level.

Table S2 Primer names, fluorescent dyes, annealing temperatures and mean number of alleles corrected for sample size for the seven microsatellite loci used. GenBank accession numbers are provided below each locus name. Forward (5′–3′) primers were directly labelled with FAM (blue), VIC (green), NED (black) or PET (red) fluorescent dye. Loci with the same number in the last column were multiplexed together during PCR reactions.

Table S3 Population genetic structure results for the full data using STRUCTURE v2.2.3. The results shown in here correspond to K = 6 the most likely number of genetic partitions supported by the data according to the DeltaK criteria proposed by (Evanno et al. 2005 c.f at bottom of Table). The analysis was conducted with the following basic parameters: 791 individuals, 7 loci, 6 populations assumed, 250000 burn-in period, 1000000 repetitions, and estimating admixture coefficients.

Table S4 Allele frequencies for all populations of Senegalia mellifera grouped by regions as defined by Barrier analyses. Data shown below represent allele frequencies of the full data set. Population identities along with sample sizes are shown in Tables 1 and S1 (> 9 < 36). Numbers in bold indicate regional unique alleles, numbers both in bold and underlined denote unique alleles for that population, FD indicates significant deviations from HW for that particular locus in analysis with the full data set, and SC after sibship-control with COLONY. Populations outside HW equilibrium before and after sibship-control are marked as FD/SC.

MEC_4833_sm_FigS1-2andTableS1-4.doc4025KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.