Tocantins river as an effective barrier to gene flow in Saguinus niger populations

The Saguinus represent the basal genus of the Callitrichinae subfamily. Traditionally this genus is divided into three groups: Hairy, Mottled and Bare-face, however, molecular data failed to validate these groups as monophyletic units, as well as raised some subspecies to the species status. This is the case of the former subspecies Saguinus midas midas and S. midas niger, which are now considered as different species. In the present study, we sequenced a portion of the D-loop mtDNA region in populations from the East bank of the Xingu and from both banks of the Tocantins river, in order to test the effectiveness of large rivers as barriers to the gene flow in Saguinus. According to our results, the populations from the East and West banks of the Tocantins river are more divergent than true species like S. mystax and S. imperator. The Tocantins river may be acting as a barrier to gene flow, and consequently these very divergent populations may represent distinct taxonomic entities (species?).

The tamarins, Saguinus spp., are small-bodied primates distributed from the Amazon to the northern coast of South America and Panama.They are the basal members of the Callitrichinae subfamily, which also includes the genera Callithrix, Cebuella, Mico, Leontopithecus and Callimico (Primates -Platyrrhini) (Schneider et al., 1993, 1996: Schneider 2000: Rylands et al., 2000).
Based on DNA sequences of the first three exons and introns of the β2-microglobulin nuclear gene, Canavez et al., (1999) considered S. fuscicollis as the first offshoot within Saguinus, followed by an unresolved trichotomy for S. mystax/S.imperator, S. niger, S. midas/S.bicolor, and S. oedipus.Additionally, these authors suggested that S. midas was more closely related to S. bicolor than to S. niger, contradicting previous studies based on biochemical (Mei-reles et al., 1997) and mitochondrial DNA data (Cropp et al., 1999), which clearly showed S. midas and S. niger to be closely related taxa, separated well apart from the S. bicolor -S.martinsi clade.Rylands et al. (2000) presented a synthesis for Saguinus classification and suggested the occurrence of 15 species without any subdivision.These species are: S. nigricollis, S. fuscicollis, S. graellsi, S. tripartitus, S. labiatus, S. inustus, S. geoffroyi, S. leucopus, S. oedipus, S. mystax, S. imperator, S. midas, S. niger, S. bicolor and S. martinsi.This is the classification we adopted in the present work.
The purpose of this paper was to examine the phylogenetic relationships among taxa of the genus Saguinus based on the fast evolving mitochondrial D-loop nucleotide sequences, aiming to answer two questions: 1) is S. midas from the North of the Amazon river more closely related to S. bicolor than to S. niger from the southern Amazon river? and 2) is the Tocantins river an effective barrier to S. niger populations?
The sequences were aligned using the BIOEDIT sequence editor (Hall, 1999) and ClustalW (Thompson et al., 1994) by constructing a common alignment of 489 bp.The sequences were deposited in GenBank under accession numbers DQ241218-DQ241250 A divergence matrix was constructed using the parameters and evolutionary model selected by the MODELTEST program version 3.07 (Posada and Crandall, 1998).The phylogenetic reconstructions were made using PAUP* version 4b10 (Swofford, 2003) for maximum parsimony and neighbor joining, and Treefinder (Jobb et al., 2004) for maximum likelihood.Saturation test was performed using the DAMBE program (Xia and Xie, 2001).The confidence of tree topologies was assessed by bootstrapping of 1000 pseudo-replicates.Genetic differentiation within, between and for the entire popula-tion was evaluated by computing genetic diversities in Mega version 3.1 (Kumar et al., 2004).
The alignment obtained showed 120 variable sites (87 parsimony-informative), defining 33 distinct haplotypes among the tamarins (Saguinus).Considering only S. niger populations, 64 variable sites were observed generating 21 distinct haplotypes, five from the East bank of the Tocantins, nine from the West bank of the Tocantins, and seven from the East bank of the Xingu river.The haplotype diversity and standard error for each population were: East Tocantins = 1.0 ± 0.126; West Tocantins = 1.0 ± 0.052 and East Xingu = 0.89 ± 0.11.In addition, the estimates of nucleotide diversities showed that East Xingu (Pi = 0.0213 ± 0.006) displayed the highest diversity, followed by East Tocantins (Pi = 0.017 ± 0.004) and West Tocantins (Pi = 0.012 ± 0.003) with similar nucleotide diversities.
Table 1 shows the pairwise genetic distances estimated between all the taxa investigated, based on the HKY model (Hasegawa et al., 1985), assuming among-site variation to follow a gamma distribution with parameter alpha equal to 0.7539, a transition transversion ratio of 3.9555, and the proportion of invariant sites being equal to 0.5317.The divergence between S. niger populations separated by the Tocantins river was 0.086 ± 0.007.This value is in the range of the differences found between S. niger and S. midas (0.079 to 0.105) populations; it is similar to those differences found between S. bicolor and S. martinsi (0.075 ± 0.003) or even S. imperator x S. mystax (0.094 ± 0.001), which are all considered as valid species.The genetic distances between S. midas from the Uatumã and Trombetas (0.049 ± 0.010) were, to some extent, higher than those between S. niger populations from the East Xingu and West Tocantins.Interestingly, the Trombetas population seems to be as related to S. niger from East Tocantins (0.048 ± 0.009) as it is to S. midas from Uatumã (0.049 ± 0.010).However, since only two specimens from each site were studied, these results should be interpreted with caution, but they certainly deserve special attention in future studies.The analysis of mean diversities comparing the three 216 Vallinoto et al. populations of S. niger showed that 78% of the total diversity is ascribable to differences between the East and West banks of the Tocantins river, while only 36% is due to differences between the Xingu and Tocantins West banks (data not showed).Maximum parsimony, maximum likelihood and neighbor joining analyses showed that populations of S. niger from the East bank of the Xingu and the West bank of the Tocantins rivers are more closely related to each other (bootstrap results equal to or greater than 97%) than populations from distinct banks of the Tocantins river.Intriguingly, in spite of the non-significant bootstrap values (73 to 88%), the population of the East Tocantins seems to be more related to S. midas from the North of the Amazon river than to S. niger from the West Tocantins.Finally, as expected, S. bicolor and S. martinsi were found to be monophyletic (bootstrap equal to 100% -Figure 1) and very closely related to each other (bootstrap support equal or greater than 97%).
To test the real effectiveness of the Tocantins river as a barrier for S. niger populations, we estimated the maximum-likelihood under two different assumptions, one constrained, i.e., forcing S. niger from both banks of the Tocantins to be part of the same monophyletic group, and the other unconstrained, as the alternative hypothesis.The likelihood of the unconstrained tree was -1967.08126,but when Tocantins populations from both banks were constrained to be a monophyletic assemblage, a highly significant drop of 44.99627 (p < 0.01) in the likelihood was observed, as revealed by the SH-test (Shimodaira and Hasegawa, 1999).These results strongly suggest that S. niger populations from different banks of the Tocantins river have been isolated for a long time, and are consequently undergoing a process of differentiation driven by geographic isolation.
A second very significant conclusion of this D-loop analysis was that S. midas and S. niger indeed belong to a monophyletic group, strongly supported by bootstrap values of 99-100%.This finding contradicts the results of Canavez et al. (1999), based on nuclear DNA sequences, that indicated a closer relationship between S. midas and S. bicolor.Again, constraining S. midas to be sister group of S. bicolor, as suggested by Canavez et al. (1999), produced a highly significant drop in the likelihood of 77.18821 (p < 0.01), rejecting the proposal of these authors.Previously, Tagliaro et al. (2005) using ND1 mitochondrial DNA data, reached the same conclusions as those of our D-loop analysis, and attributed that incongruity to the conservative nature of the nuclear DNA sequence (900 base pairs from exons 1, 2 and 3 of the β2-microglobulin gene) used by Canavez et al. (1999).
A third and unexpected finding was the separation in the phylogenetic tree of S. midas from the Uatumã (State of Amazonas, Brazil) of those from Trombetas (State of Pará, Brazil) banks, which are about 200 km away from each other, almost the same distance that separates S. niger populations of the Tocantins and Xingu rivers.The genetic distances between populations of S. midas from Uatumã and Trombetas (0.049) are of the same magnitude as those between Trombetas and Tocantins East bank (0.048).Conversely, the genetic distance between S. midas from Uatumã and S. niger from Tocantins East bank is 0.079.
It is obvious that description of biodiversity has a great importance to wildlife conservation.However, as it implies the definition of species, it has been a matter of great and unsolved debate (see Groves, 2004, for a good review on Primates).Distance methods have been proposed as a mean of objectively identifying species.The idea is that genetic similarity is greater within than between species (Nei, 1972).Some authors caution against this approach (see Vogle and DeSalle, 1994 for details).Alternatively, a number of authors have advocated the use of the concept of evolutionarily significant units (ESUs) to characterize useful units for conservation purposes (Ryder, 1986;Cracraft, 2002).According to Moritz (1994), in order to be identified as ESUs, the populations should be reciprocally monophyletic for mtDNA alleles and also show divergence of allele frequencies at nuclear loci.Yet, Vogle and DeSalle (1994) favor a definition of ESUs based on patterns of variation.According to them, "in the theoretical framework of the phylogenetic species concept, conservation units are delimited by characters that diagnose clusters of individuals or populations to the exclusion of other such clusters".Another important concept is the management unit (MU), also proposed by Moritz (1994), which represents populations that do not show reciprocal monophyly, but have diverged in allele frequency.However, Hudson and Coyne (2002) observed that we must be cautious about recognizing species using only mitochondrial DNA, since this DNA becomes monophyletic more rapidly than does a single nuclear gene, and far more rapidly than does a sample of several nuclear genes.Therefore, the use of mtDNA alone is not a good strategy for assessing reciprocal monophyly, unless the population divergence is very ancient, which in the present study does not seem to be the case.According to the phylogenetic species concept (Hennig, 1966), species can be identified as the smallest aggregation of populations or lineages diagnosable by a unique combination of character states.It is important to note that our phylogenetic tree shows S. niger from the West and East banks of the Xingu and Tocantins rivers, respectively, as belonging to a monophyletic clade (bootstrap supports equal to or greater than 97%), excluding Saguinus populations from the East bank.This eastern population tends to form a moderately well supported group (bootstrap support 73% to 88%) with S. midas, populations which are separated not only by a large distance but also by a huge barrier, represented by the Amazon river.So, following the interpretation of Vogle and DeSalle (1994) "in the theoretical framework of the phylogenetic species concept", Saguinus from different banks of the Tocantins river should be considered as distinct ESUs, if these results were confirmed by other molecular markers.
On the other hand, using the criterion of genetic distances, they also can be considered as distinct entities, such as S. bicolor and S. martinsi or S. midas and S. niger, for example.Regarding the XE and TW populations, they fit the criteria of management units (MU), as they are not reciprocally monophyletic and present a discrete percentage of nucleotide divergence (0.039 ± 0.007).
The present study reinforces the existence of two divergent populations of Saguinus niger separated by the Tocantins river, it corroborates the monophyletic status of the S. midas-S.niger group and it calls attention to the divergence of S. midas from the northern Amazon, indicating that much has still to be done in order to clarify all aspects of the biological diversity of New World primates.

Figure 1 -
Figure1-Phylogenetic consensus tree depicted by bootstrapping 1000 pseudoreplicates.Numbers at nodes mean bootstrap support levels for maximum parsimony (MP), minimal evolution (ME) and maximum likelihood (ML), respectively.When bootstrap values of MP, ME and ML were similar the signal greater than (>) was used.

Table 1 -
Estimates of genetic distances among Saguinus lineages based on HKY85 model assuming heterogeneous substitution rates among lineages and rates among sites following a gamma distribution.