Of 376 fecal samples were collected for DNA study in the Democratic Republic of Congo between July 2010 and February 2012. DNA extraction was performed using a proprietary procedure combining a sampling lysis buffer [31], removal of the potential PCR inhibitors such as bile salts and bilirubin with starch 
[32], and a commercially available DNA cleanup system with a silica membrane (Wizard SV Gel and PCR Clean-Up System; Promega, Madison, WI, USA) (details are presented in Table S1). The SDS lysate of gut cells was preserved in a tube at ambient temperature until DNA extraction (maximum of 6 months). Qualification of purified DNA samples was made quantitatively [33] or qualitatively by an electrophoretic procedure (Table S1). Multiple sampling from the same individual was inspected as much as possible by genotyping with 10 microsatellite markers (data not shown), but this evaluation was incomplete for some samples due to difficulty with the genotyping. Among the collected samples, we failed to sequence the target mtDNA region for 114 samples due to low DNA quality or recovery and confirmed 126 cases of multiple sampling. Finally, we judged that 136 DNA samples were taken from different individuals and JSI124 site subjected them to comparative analysis in this study.FactorsAll areas (n = 21) (AIC = 223.21) t p 0.When TL2 was removed (n = 15) (AIC = 221.68) t 3.6 (+) p 0.Straight distance5.2 (+)Number of tributaries 23.0 (2) 0.20.5 (2) 0.FST was used as a response variable and Gaussian (identity) was used as a family (link function). Signs in parenthesis mean direction to increase FST. doi:10.1371/journal.pone.0059660.tGenetic Structure of BonobosDNA SequencingA complete sequence of mtDNA noncoding region was determined from each fecal sample. A DNA fragment spanning the target region was amplified and Potassium clavulanate sequenced using the seven primers listed in Table S1. PCR amplification was performed with a high-success rate DNA polymerase KOD FX (Toyobo, Osaka, Japan) and sequences were read manually by direct sequencing. False readings caused by nuclear mitochondrial DNA (Numt) were verified by aligning the obtained sequence reads with published data. Obtained sequence data were deposited in DDBJ/EMBL/ GenBank databases (Accession Numbers AB780372 B780425). The data were subjected to size adjustments by sorting with Gblocks [34] under the default stringent condition for haplotyping and subsequent phylogenetic or population analyses.map. The width of the Luo River 1662274 between the Wamba and Iyondji populations currently prevents bonobos from moving to the opposite bank. The number of tributaries between two populations was greater than the number of tributaries expected from Figure 1 in most cases. The analyses were performed with R and JMP (SAS Institute, Cary, NC, USA). For the dataset of genetic distance (pairwise FST), a normal distribution was not rejected. (Kolmogorov-Smirnov test, D = 0.16, p = 0.57, n = 21). Correlations were tested using Pearson’s correlation test. A generalized linear model (GLM) was used for calculations of AIC to estimate whether tributaries of the Congo River influenced genetic distances.Supporting InformationTable S1 Summary of DNA experiments.Molecular Data AnalysisHaplotypes were defined by multiple alignments with ClustalX ver. 2.1 
[35] for the sorted sequences. The Tamura-Nei model [36] was assumed in the computation of evolutionary distance. Molecular phylogenetic relations were inferred by using neighborjoining (NJ), maximum likelih.Of 376 fecal samples were collected for DNA study in the Democratic Republic of Congo between July 2010 and February 2012. DNA extraction was performed using a proprietary procedure combining a sampling lysis buffer [31], removal of the potential PCR inhibitors such as bile salts and bilirubin with starch [32], and a commercially available DNA cleanup system with a silica membrane (Wizard SV Gel and PCR Clean-Up System; Promega, Madison, WI, USA) (details are presented in Table S1). The SDS lysate of gut cells was preserved in a tube at ambient temperature until DNA extraction (maximum of 6 months). Qualification of purified DNA samples was made quantitatively [33] or qualitatively by an electrophoretic procedure (Table S1). Multiple sampling from the same individual was inspected as much as possible by genotyping with 10 microsatellite markers (data not shown), but this evaluation was incomplete for some samples due to difficulty with the genotyping. Among the collected samples, we failed to sequence the target mtDNA region for 114 samples due to low DNA quality or recovery and confirmed 126 cases of multiple sampling. Finally, we judged that 136 DNA samples were taken from different individuals and subjected them to comparative analysis in this study.FactorsAll areas (n = 21) (AIC = 223.21) t p 0.When TL2 was removed (n = 15) (AIC = 221.68) t 3.6 (+) p 0.Straight distance5.2 (+)Number of tributaries 23.0 (2) 0.20.5 (2) 0.FST was used as a response variable and Gaussian (identity) was used as a family (link function). Signs in parenthesis mean direction to increase FST. doi:10.1371/journal.pone.0059660.tGenetic Structure of BonobosDNA SequencingA complete sequence of mtDNA noncoding region was determined from each fecal sample. A DNA fragment spanning the target region was amplified and sequenced using the seven primers listed in Table S1. PCR amplification was performed with a high-success rate DNA polymerase KOD FX (Toyobo, Osaka, Japan) and sequences were read manually by direct sequencing. False readings caused by nuclear mitochondrial DNA (Numt) were verified by aligning the obtained sequence reads with published data. Obtained sequence data were deposited in DDBJ/EMBL/ GenBank databases (Accession Numbers AB780372 B780425). The data were subjected to size adjustments by sorting with Gblocks [34] under the default stringent condition for haplotyping and subsequent phylogenetic or population analyses.map. The width of the Luo River 1662274 between the Wamba and Iyondji populations currently prevents bonobos from moving to the opposite bank. The number of tributaries between two populations was greater than the number of tributaries expected from Figure 1 in most cases. The analyses were performed with R and JMP (SAS Institute, Cary, NC, USA). For the dataset of genetic distance (pairwise FST), a normal distribution was not rejected. (Kolmogorov-Smirnov test, D = 0.16, p = 0.57, n = 21). Correlations were tested using Pearson’s correlation test. A generalized linear model (GLM) was used for calculations of AIC to estimate whether tributaries of the Congo River influenced genetic distances.Supporting InformationTable S1 Summary of DNA experiments.Molecular Data AnalysisHaplotypes were defined by multiple alignments with ClustalX ver. 2.1 [35] for the sorted sequences. The Tamura-Nei model [36] was assumed in the computation of evolutionary distance. Molecular phylogenetic relations were inferred by using neighborjoining (NJ), maximum likelih.