We identified 10 Alu insertions shared by C. lhoesti, Chlorocebus aethiops and E. patas. This result strongly supported (p < 0.001) 32 the monophyly of the terrestrial group and the paraphyly of the previously defined genus Cercopithecus. This result is congruent with the single transition hypothesis, which states that there was only one transition from an arboreal to terrestrial lifestyle within the tribe 71433. Unfortunately, we did not identify enough informative markers to gain significant support for the relationships between taxa within the terrestrial group.

Given the unambiguous clustering of a single terrestrial clade, our results support the division of the tribe Cercopithecini into four major clades: the arboreal Cercopithecus species, a terrestrial clade, Miopithecus, and Allenopithecus. Six Alu insertions were shared by all taxa except A. nigroviridis. This unambiguously placed Allenopithecus as the basal lineage within the tribe (p < 0.01). Another seven insertions joined M. talapoin to all arboreal Cercopithecus species (i.e. C. ascanius, C. cephus, C. petaurista, C. wolfi, C. diana and C. nictitans). These results suggest that the terrestrial group diverged after Allenopithecus but before the separation of Miopithecus from the arboreal Cercopithecus species (p < 0.01). The divergence of Miopithecus and the monophyly of the six arboreal Cercopithecus species were supported by nine Alu insertion loci present in all arboreal taxa but not in Miopithecus (p < 0.01).

Of the six arboreal Cercopithecus species we tested, two clades were well supported. C. ascanius and C. cephus were clustered together to the exclusion of C. petaurista, based on four Alu insertions (p < 0.05). The C. ascanius and C. cephus clade was linked to C. petaurista and C. nictitans by seven insertions to form a polytomy (p < 0.01). Despite the relatively strong support for these two nodes, we should note that there were loci that supported alternative topologies among the arboreal Cercopithecus species. For example, when the affiliation of C. diana is considered, Alu insertions present in C. diana suggested at least four alternative phylogenetic hypothesis (Figure 3). Two Alu insertions (TA26 and CD_Yd_40) grouped C. diana with C. ascanius, C. cephus and C. petaurista, but not C. nictitans. In contrast, two other Alu elements (CD_Yd_27 and CN_PY2_53A) were found only in C. diana and C. nictitans. In addition, one marker (CD_PY2_36) supported the grouping of C. diana, C. nictitans and C. cephus, and one marker (CD_PY2_41) grouped C. diana, C. wolfi and C. petaurista together. Owing to this incongruence and the small number of makers supporting each hypothesis, none of these hypotheses gained statistically significant support using the likelihood test.

Another noteworthy result is that a number of Alu insertion loci appeared to be heterozygous in certain species (e.g. both a filled band and an empty band are present) (Figures 1B and 1C). Several mechanisms, including incomplete lineage sorting, concurrent polymorphism and introgression may have contributed to this finding.

Our results indicate that Allenopithecus is the most basal lineage of the tribe. A monophyletic terrestrial clade comprising C. lhoesti, E. patas, and Chlorocebus aethiops was the next to diverge from other guenons after the split of Allenopithecus, and Miopithecus subsequently diverged from the rest of arboreal Cercopithecus. This placement of Miopithecus represents a major difference compared with other phylogenetic analyses of the Cercopithecini. Despite the general view that Miopithecus may share some primitive features of the group (e.g. sexual swelling), no studies to date have provided robust support for a definitive placement for Miopithecus. For example, in a study using Y-chromosome DNA data 14, a four-way polytomy for Allenopithecus; Miopithecus, a terrestrial clade and an arboreal Cercopithecus clade was obtained. Studies using X-chromosome sequences identified Allenopithecus as the most basal lineage of the tribe, although a three-way polytomy consisting of Miopithecus, the terrestrial clade and the arboreal Cercopithecus clade persisted 714. Our results represent the first statistically significant support for the clustering of Miopithecus with the arboreal Cercopithecus clade.

Owing to rapid speciation and the resulting close relationships within the genus Cercopithecus, a detailed lower-level hypothesis of relationships has been lacking. To date, the only robustly supported relationships are the separation of a cephus group + mitis group clade, and a neglectus + diana + mona group clade 714. In the current study, several relationships within the genus are supported by Alu insertions that have statistical support. For example, the clustering of C. ascanius and C. cephus within the cephus group (represented in this study by C. ascanius; C. cephus and C. petaurista) is well supported. The close relationship between the mitis group (represented here by C. nictitans) and the cephus group is also strongly supported. The diana group (represented here by C. diana) appears to be basal within the arboreal Cercopithecus clade. However, this hypothesis deserves further investigation, owing to the relatively low level of support for the placement of C. wolfi. Overall, these results are in good agreement with previous studies using X and Y chromosome DNA data, and provide robust independent evidence for the suggested relationships 714.

Our results provide strong support for a single evolutionary transition from arboreality to terrestriality within the tribe Cercopithecini. The inference of a single transition to a ground-dwelling lifestyle suggests that such changes in substrate preference are probably rare in the history of the primates and therefore may lend greater significance to a similar change at the beginning of the hominin radiation. In addition, the guenon transition to a terrestrial lifestyle is estimated to have occurred at the Miocene/Pliocene boundary 7. This is significant because the same time window brackets the origins of hominin terrestrial bipedalism. Thus, similar factors may have driven early human ancestors and the terrestrial guenon progenitor to a largely ground-dwelling way of life.

There are three major scenarios in SINE-based phylogenetic analyses that can lead to confounding results: adjacent independent insertions, incomplete lineage sorting, and hybridization between species. Adjacent independent insertions refer to the scenario that in different species, different Alu elements are inserted in close genomic proximity and are therefore amplified within the same PCR amplicon during genotyping. In this study, we sequenced all informative loci to confirm that the appropriately size amplicon contained the same Alu insertions, and all loci containing parallel independent insertions have been excluded or treated as independent markers (see Results section). Therefore, the incongruence observed in our current dataset cannot be explained by invoking this explanation.

Incomplete lineage sorting is caused by alternative fixation/elimination of ancestral polymorphisms in populations of daughter lineages. Observations of incomplete lineage sorting can be particularly problematic when examining taxa that have undergone rapid bursts of speciation 2234. It is believed that the tribe Cercopithecini diverged from tribe Papionini about 11 million years ago 735. However, owing to the lack of fossil evidence, the divergence times of subsequent groups are less certain. For the arboreal Cercopithecus clade, multiple speciation events may have occurred during a very short period to create the six species groups, containing the >20 species we see today 1. If that is the case, many polymorphic Alu elements present in the guenon ancestral population have the potential to remain polymorphic before, during, and after these speciation events. These insertions were eventually fixed in or lost from the genomes of daughter species, and may have resulted in the incongruent insertion presence/absence patterns observed.

Finally, occasional introgressions via hybridization present yet another hypothesis to explain the incongruent Alu patterns. The ranges of closely related species within many guenon species groups often overlap geographically, and many of these species are known to hybridize with each other. For example, it is known that hybridizations occur between the red-tailed monkey group (cephus species group) and the blue monkey group (mitis species group, represented here by C. nictitans) in their contact zones 36. These hybridizations may have contributed to the polytomy of C. ascanius, C. cephus, C. petaurista (the cephus species group) and C. nictitans (the mitis species group) in our study. In fact, introgression may even have introduced certain Alu insertion loci into species other than the arboreal Cercopithecus clade and provide a plausible explanation for some of the incongruent results in our study.

Genomic sequences from Chlorocebus aethiops(African green monkey) were obtained from the NIH Intramural Sequencing Center 37, as part of the Comparative Vertebrate Sequencing Initiative. The sequences were broken into 10000-bp fragments and compared with the human genome (hg18) and the rhesus monkey genome (rheMac2) using the BLAST-Like Alignment Tool (BLAT) 38. Fragments containing insertion/deletions were extracted and annotated to identify putative lineage-specific Alu insertions using previously reported methods 39. In order to allow a focus on the cercopithecine lineage, only Alu elements present in the Chlorocebus aethiops sequence but absent in the orthologous regions of the rhesus monkey genome were excised along with 1000 bp of flanking sequence in both directions. Flanking oligonucleotide primers for PCR amplification of each Alu element were then designed using Primer3 software 40. The primers were subsequently screened against the GenBank NR database using the Basic Local Alignment Search Tool (BLAST) program 41 to determine if they resided in unique DNA sequences.

The Alu element PCR-display methodology and Alu-specific primers described by Xing et al 28 were used for identifying species-specific Alu insertions. To prevent ascertainment bias, every cercopithecine species in this study was subject to this methodology and Alu insertion loci were identified from all species. Oligonucleotide primer pairs were initially tested using Chlorocebus aethiops DNA templates with a temperature gradient PCR (48–60°C) to determine the most appropriate annealing temperature for further analysis. All loci were screened against a primate panel that was composed of DNA samples from 12 Old World primate species (Table 1). Because the quantity of genomic DNA samples for most species is limited, all samples except Chlorocebus aethiops were subjected to whole genome amplification using the GenomiPhi genome amplification kit (Amersham, Sunnyvale, CA, USA) following the manufacturer's instructions. The amplified samples were then purified and divided into aliquots for locus-specific PCR analysis.

PCR amplification of each locus was performed as described by Xing et al 28. The resulting PCR products were run on 2% agarose gels with 0.25 μg of ethidium bromide and visualized using ultraviolet fluorescence. Detailed information for each locus including primer sequences, annealing temperature, PCR product sizes, chromosomal locations and amplification results are available on our website (supplemental table. located under "Publications") 42.

For the potentially informative loci, all PCR amplicons that appeared to have Alu insertions were subjected to sequence analysis to verify the presence of the Alu elements. Selected taxa from several uninformative loci were also sequenced to confirm the presence of the Alu elements. Individual PCR products were either directly sequenced or were cloned and sequenced as described previously 29. Sequences for each locus were aligned against rhesus monkey (when available) or human orthologous sequence obtained via the BLAT search. Sequence alignments of these loci are available from our website (supplemental table. located under "Publications") 42. The DNA sequences generated for this project have been deposited in GenBank (accession numbers DQ977747-DQ978210).

Alu insertion loci were included in phylogenetic analysis if amplicons were generated in at least seven out of the 11 cercopithecine taxa in our panel, and only 2 distinct classes of amplicons were obtained (i.e. Alu filled size and pre-integration or Alu empty size). Any primer pair that generated multiple paralogous fragments across the panel was excluded from the analysis.

We implemented an exhaustive search in PAUP* software (version 4.0b10) 43 using Dollo parsimony and designating the Pygathrix nemaeus as an outgroup taxon. Presence of the insert was coded as "1" and absence of the insert as "0". If no amplification was observed for a given locus in any taxon, the character state was coded as unknown ("?"). For loci containing adjacent independent insertion events, the independent insertions were treated as independent markers. In total, 10000 bootstrap replicates were performed on the data. A statistical test for evaluating SINE insertions based on a likelihood model 32 was also performed to assess the statistical significance of each branch of the resulting tree.