PHYLOGENETIC RELATIONSHIPS AMONG BRAZILIAN HOWLER MONKEYS , GENUS Alouatta ( PLATYRRHINI , ATELIDAE ) , BASED ON γ 1-GLOBIN PSEUDOGENE SEQUENCES

Os guaribas, do genero Alouatta, que sao os primatas do Novo Mundo com maior distribuicao geografica, tem sido colocados em tres grupos de especies: o grupo Alouatta palliata da America central, e os grupos sulamericanos Alouatta seniculus e Alouatta caraya. Este ultimo e monotipico, mas o grupo A. seniculus inclui pelo menos tres especies (A. seniculus, A. belzebul e A. fusca). Neste estudo, foram sequenciados aproximadamente 600 pares de base do pseudogene globina g1 nas quatro especies brasileiras (A. seniculus, A. belzebul, A. fusca e A. caraya). Os metodos de maxima parcimonia e maxima verossimilhanca produziram arvores filogeneticas com o mesmo arranjo: {A. caraya [A. seniculus (A. fusca, A. belzebul)]}. A arvore mais parcimoniosa apresentou valores de bootstrap maiores de 82% para todos os agrupamentos, e valores de forca de ligacao de pelo menos 2, apoiando o agrupamento irmao de A. fusca e A. belzebul. O estudo tambem confirmou a presenca em A. fusca do elemento de insercao Alu, com 150 pares de base, e uma delecao de 1,8 kb no pseudogene globina g1 ja conhecidos nas demais especies de guaribas. A classificacao cladistica baseada em dados moleculares e congruente com as de estudos morfologicos, com um isolamento claro do grupo monoespecifico A. caraya em relacao ao grupo A. seniculus.

Alouatta is the most widely distributed Neotropical primate genus (Neville et al., 1988), ranging from southern Mexico to northern Argentina, and is found in tropical and subtropical forest ecosystems throughout Brazil (Hirsch et al., 1991).Hershkovitz (1949) recognized three species groups based on the structure of the hyoid apparatus: the Central American Alouatta palliata group, and the South American Alouatta seniculus and Alouatta caraya groups.While the latter group is monotypic, the other two encompass a variety of different forms, whose arrangement has been subject to conflicting interpretations (e.g.Hershkovitz, 1949;Hill, 1962;Mittermeier et al., 1988;Groves, 1993;Bonvicino et al., 1995;Stanyon et al., 1995;Rylands and Brandon-Jones, 1998).
Despite the need for further work, especially on the integration of genetic and morphological data, the most recent major revisions of the genus (Hershkovitz, 1949;Hill, 1962) all divide the A. seniculus group into three species: A. seniculus, A. belzebul (endemic to Brazil) and A. (guariba) fusca.Groves (1993) does recognize a fourth species, Alouatta sara, but it is endemic to Bolivia, and thus not relevant to the present study.More recent studies (Bonvicino et al., 1995) also allocate species status to a number of subspecific forms of both A. seniculus and A. belzebul, but in the absence of a consensus, the arrangement of Groves (1993) will be followed here.
Several recent studies (Schneider et al., 1993(Schneider et al., , 1996;;Harada et al., 1995;Porter et al., 1995Porter et al., , 1997a,b;,b;Meireles et al., 1999) have used DNA sequences of the globin gene to resolve phylogenetic relationships in the platyrrhines, producing results generally consistent with those of more traditional morphological studies in most (see Schneider and Rosenberger, 1996), but not all (Shoshani et al., 1996) cases.In the present study, sequences of the γ 1 -globin pseudogene were used in a molecular analysis of the phylogenetic relationships among Brazilian howler monkeys.The aligned sequences were also used to estimate divergence times based on a local molecular clock, and to investigate the presence of the Alu insertion -found in other Alouatta species (Meireles et al., 1995(Meireles et al., , 1998) ) -in A. fusca.

Study species
Howler monkeys are large-bodied, prehensile-tailed platyrrhines more specialized for folivory than frugivory, as is typical of other atelids.They are also highly sexually dimorphic in body size (adult males are 70-80% heavier than females), and two of the Brazilian species (A.caraya and A. fusca) are among only three platyrrhines (the other is Pithecia pithecia) to exhibit sexual dimorphism in pelage coloration (Mittermeier et al., 1988).A. belzebul also exhibits the most marked individual variation in pelage coloration of any platyrrhine (Bonvicino et al., 1989).Only one species (A.belzebul) is endemic to Brazil, but it is also the only one found in both Amazonian and Atlantic forest biomes.The remaining three species are found in Amazonia (A. seniculus), the Brazilian Atlantic forest and neighboring areas (A.fusca), and the savanna and swampy habitats of central and western Brazil (A. caraya).

Preparation and amplification of DNA
Genomic DNA was extracted from the peripheral blood cells of captive specimens of the four Brazilian howler monkey species (Table I), and the γ 1 -globin sequences for A. caraya, A. belzebul and A. seniculus were determined by C.M.M. (Meireles, 1997).DNA extraction for A. caraya and A. seniculus was based on the protocol of Bell et al. (1981), whereas for A. belzebul and A. fusca, that of Sambrook et al. (1989) was used.
The γ 1 fragment was obtained using the PCR protocol described in Meireles et al. (1995), and consisted of an initial denaturation of 3 min at 94 o C followed by 30 cycles of denaturation at 94 o C (30 s), annealing at 55 o C (45 s), extension at 72 o C (45 s) and a final extension of 10 min at 72 o C. The amplification products were separated by agarose gel electrophoresis.The DNA primers (synthesized at the Center for Molecular Medicine and Genetics, Wayne State University, Detroit, USA) were: R1 5' -AAT GTG GAA GAT GCT GGG-3' and R2 5' -GTC ATG TCT GAG CAA CAA AC-3' Despite the presence of a 1.8-kb γ 1 deletion in all four species (Figure 1), these primers amplified a ~600-bp fragment in all cases.
Figure 1 -Model of the β-cluster showing the five globin genes.The γ 1 -globin pseudogene is amplified exhibiting the primers R1 and R2 used in this study and also the 1.8-kb deletion shared by all four Alouatta species as well as by the ateline group (Meireles et al., 1995(Meireles et al., , 1998)).

Cloning of PCR products
The PCR products were separated in 1.0% agarose gels in TBE buffer (89 mM Tris-base, 89 mM boric acid and 2 mM EDTA, pH 8.0) containing 0.5 µg ethidium bromide/ml.The amplified fragments were excised and purified using a Qiaex II gel extraction kit (Qiagen).The purified DNA fragments were cloned into the pGEM-T vector system I (Promega) and transformed into Escherichia coli host JM109 (Promega) according to the protocol supplied by the vendors.Single-stranded DNA was prepared from selected clones using helper phage M13K07 (Promega) and purified by PEG-NaCl precipitation followed by phenolchloroform extraction and ethanol precipitation.

Sequencing
Nucleotide sequences were determined for at least three clones of each species by the dideoxy chain-termination method (Sanger et al., 1977) using a Sequenase version 1.0 kit (United States Biochemical).The sequences were aligned by eye using version 3.0 of the ESEE sequence editor (Cabot and Beckenbach, 1989).Gaps were inserted to minimize the number of nucleotide substitutions and indels (insertions/deletions) needed to account for the descent of the aligned sequences.As PCR reactions commonly result in slightly different sequences for the same cloned fragments, a consensus sequence of three or more clones was defined for each species.These consensus sequences were then aligned against the known orthologous sequences which represented the outgroups for the phylogenetic analysis: Cebus albifrons (Cal), Macaca mulatta (Mmu) and Homo sapiens (Hsa).

Phylogenetic analysis
Phylogenetic analyses were performed using the maximum parsimony and maximum likelihood methods.The most parsimonious and the maximum likelihood trees were determined using the phylogenetic analysis using parsimony (PAUP) parsimony program, version 4.0 for DOS (written by David Swofford, Smithsonian Institute, Washington, DC, USA).For both trees, the strength of grouping values or Bremer decay indices were estimated using the PAUP program, and bootstrap analysis was applied to test the support for each grouping.Two thousand bootstrap replications with one shuffle per replication were carried out on the seven data sets.The number of parsimony-informative characters (synapomorphies) was established by eye and confirmed by the PAUP program.Divergence times were estimated using the branch lengths of the maximum likelihood tree constructed from a distance matrix calculated using Kimura's (1980) model.The molecular clock was calibrated using the local clock procedure, based on Goodman's (1996) estimate of 21 MYA for the divergence between Cebus and Alouatta.
The natural logarithm (ln) of the likelihood for the best maximum likelihood tree using the aligned γ-globin sequences was -9329.13944.Bootstrap values (as a percentage of 2,000 replicates) are shown below the lines (Figure 5).Although the topologies of the tree and bootstrap values for the first (88) and last (100) clades were the same, the A. seniculus-A.belzebul/A.fusca clade had a borderline bootstrap value of 72.This analysis confirmed the cladistic groupings shown by the maximum parsimony tree (Figure 4).
Although the A. belzebul-A.fusca clade was relatively well defined, with a high bootstrap value (88%) and four shared synapomorphies, the analysis did not permit identification of the species closest to A. seniculus, despite the relative geographic proximity of A. belzebul in southwestern Amazonia (Hirsch et al., 1991).The data, nevertheless, supported the presence of a well-defined A. seniculus group [A.seniculus (A.belzebul, A. fusca)] as proposed by Hershkovitz (1949) and still widely accepted (e.g.Mittermeier et al., 1988).There is little doubt from these data that A. caraya should be placed in a distinct, monotypic group, and that Alouatta is monophyletic, with 15 synapomorphies.
The divergence times between the different Alouatta lineages (Table II) were calculated by the local molecular clock method, based on Goodman's (1996) estimate of 21 MYA for the split between Alouatta and Cebus, using the maximum likelihood branch lengths for the sequences of the γ 1 -globin pseudogene.This analysis indicated that the different Alouatta lineages diverged relatively recently in relation to the putative + 11 MYA split of the alouattine and ateline groups (Meireles et al., 1999).
Overall, then, the results of the present study corroborate the phylogenetic relationships among Brazilian howler monkey species suggested by morphological characteristics (Mittermeier et al., 1988), principally the existence of distinct A. caraya and A. seniculus groups, which diverged approximately 2.5 MYA (Table II).
The nucleotide sequences bordered by primers R1 and R2 in the seven primate species studied here spanned a total of 2,482 aligned base positions (Figure 3) consisting of the following γ 1 gene regions: the 5' coding region of exon 1 (1-35), intron 1 (36-160), exon 2 (161-211), a ~1.8-kb deletion (212-2092), a region between the deletion and insertion (2093-2149), a ~150-bp insertion (2150-2298), and the 3' flanking region (2299-2482).In the phylogenetic analysis, all sequences except that of A. fusca were also represented by a 1.8-kb region extending upstream from the R1 primer.The sequences in this region are described in Meireles et al. (1999).The DNA sequences were deposited in GenBank (Table I), and are also available from the authors on request, or can be accessed via internet at http://ns.med.wayne.edu/.

Figure 3
Figure3-Aligned DNA sequences of the γ 1 -globin pseudogene corresponding to the region between the R1 and R2 primers in four simian genera, including four Alouatta species (see abbreviations in TableI).Dots indicate the same nucleotide as that found in the first (Abe) sequence, and dashes designate gaps resulting from insertions/deletions in the alignment.Sequences corresponding to positions 2041-3720 in Cebus albifrons (Cal), Macaca mulatta (Mmu) and Homo sapiens (Hsa) were published byBailey et al. (1992) and are thus omitted here.

Figure 4 -
Figure4-Maximum parsimony tree derived from the aligned γ 1 -globin nucleotide sequences in seven simian species.The length, or nucleotide substitution score, is 813.Values above the lines represent the minimum number of additional substitutions required to break up the clade designated by that node and those below the lines, the percentage of bootstrap values obtained in 2,000 replicates.