Hybridization in Large-Bodied New World Primates

Well-documented cases of natural hybridization among primates are not common. In New World primates, natural hybridization has been reported only for small-bodied species, but no genotypic data have ever been gathered that confirm these reports. Here we present genetic evidence of hybridization of two large-bodied species of neotropical primates that diverged ∼3 MYA. We used species-diagnostic mitochondrial and microsatellite loci and the Y chromosome Sry gene to determine the hybrid status of 36 individuals collected from an area of sympatry in Tabasco, Mexico. Thirteen individuals were hybrids. We show that hybridization and subsequent backcrosses are directionally biased and that the only likely cross between parental species produces fertile hybrid females, but fails to produce viable or fertile males. This system can be used as a model to study gene interchange between primate species that have not achieved complete reproductive isolation.

H YBRIDIZATION can be viewed as either a breakdown of species boundaries that could eventually result in the loss of pure parental species or a creative force that can lead to the formation of new recombinant lineages (Arnold 1997; Dowling and Secor 1997;Barton 2001;Mallet 2005;Arnold and Meyer 2006). Regardless of which view is taken, studies of hybridization are crucial for understanding the basis of reproductive isolation and the origins of biodiversity (Coyne and Orr 2004). Hybridization among metazoans has traditionally been viewed as an unusual event, but a variety of genetic studies in the past few decades have shown that this phenomenon is rather common, especially between closely related taxa (Mallet 2005). Among primates, natural hybridization occurs in at least 26 of $233 Old World species (e.g., baboons, guenons, macaques, lemurs) in which hybridization occurs at intraspecific (Groves 1978;Lernould 1988), interspecific (Phillipsconroy and Jolly 1986;Samuels and Altmann 1986;Struhsaker et al. 1988;Watanabe and Matsumura 1991;Bynum et al. 1997;Evans et al. 2001;Wyner et al. 2002), and even intergeneric levels (Dunbar and Dunbar 1974;Jolly et al. 1997). Among neotropical primates, only 8 of $132 New World species have been suggested to form hybrids in the wild (Coimbra-Filho et al. 1993;Peres et al. 1996;Mendes 1997) and these include only small-bodied and very recently separated taxa. Furthermore, of the few reported cases of interspecific hybridization in the wild (Silva et al. 1993;Mendes 1997), the taxonomic status of the species is questionable.
Here we present evidence of hybridization of two large-bodied neotropical primates, Alouatta palliata and A. pigra. These are morphologically (Lawrence 1933;Smith 1970), socially (Crockett and Eisenberg 1987;Treves 2001;Van Belle and Estrada 2006), behaviorally (Crockett and Eisenberg 1987;Neville et al. 1988;Whitehead 1995), and genetically (Cortés-Ortiz et al. 2003) distinct howler monkey species that diverged $3 MYA (Cortés-Ortiz et al. 2003). A. palliata is currently found from southern Veracruz in Mexico through the central part of Guatemala and the southern part of Belize, continuing to the south through Honduras, Nicaragua, Costa Rica, Panama, and the Pacific coast of Colombia and Ecuador. On the other hand, A. pigra is confined to the Yucatan peninsula in Mexico, Belize, and the central and eastern part of Guatemala ( Figure 1). Although A. palliata and A. pigra are allopatric in most of their range, Smith (1970) reported an area of sympatry in the state of Tabasco, Mexico. Currently this area is highly deforested and only small patches of vegetation with various degrees of disturbance can be found. During a series of expeditions, we surveyed the area where sympatry had been reported and found troops with individuals of both A. palliata and A. pigra (based initially on morphological Sequence data for this article have been deposited with the EMBL/ GenBank Data Libraries under accession nos. DQ875611-DQ875741. 1 characters), as well as individuals that possessed morphological features of both species. Using a multilocus approach, we present genetic data that show that these howler monkeys are hybridizing in Mexico.

MATERIALS AND METHODS
Blood and/or hair samples were collected from A. palliata and A. pigra individuals from sites in Tabasco, Mexico and other areas throughout Mexico (Figure 1). Genomic DNA was extracted using the DNeasy tissue kit (QIAGEN, Valencia, CA). Primers for eight microsatellite loci ½Ap68 (Ellsworth and Hoelzer 1998), Ap74 (Ellsworth andHoelzer 1998), PEPC8 (Escobar-Páramo 2000), and MapPairs (Invitrogen, Carlsbad, CA) loci D5S111, D6S260, D8S165, D14S51, and D17S804 were used to identify diagnostic alleles in each species and to identify hybrid individuals on the basis of the presence of these alleles. We used primers CB1-59 and CB2-39 (Palumbi 1996) to amplify a region of the mitochondrial cytochrome b (cytb) gene and/or primers LCO-CO2-L and LCO-CO3-H (Cortés-Ortiz et al. 2003) to amplify a fragment of the ATP-synthase 6 and 8 genes (ATPase). A fragment of the Y chromosome Sry gene was amplified using primers SW2 (Whitfield et al. 1993) and SRY (Moreira 2002). To determine whether hybridization and subsequent crosses are directionally biased, we used a chi-square goodness-of-fit test to compare the observed frequencies of genotypes of hybrid individuals to those expected if all possible crosses among hybrids and backcrosses with parental species occur. We also estimated the probabilities of observing the detected genotypes on the basis of equal proportions of alleles/haplotypes in the parental species.

RESULTS AND DISCUSSION
We genotyped 104 individuals of A. palliata and A. pigra and putative hybrid individuals for the eight microsatellite loci. These individuals include 40 A. palliata and 28 A. pigra individuals from outside of the putative hybrid zone and 36 individuals from within this zone ( Figure 1). On the basis of the genotypes of A. palliata and A. pigra outside of the zone, three loci contained alleles that were distinct for each species (Ap68, D5S111, and D8S185) ( Table 1). Sequences of these alleles confirmed that size differences are due to differences in the number of repeat units. The two species shared alleles at the other loci examined or potentially diagnostic alleles occurred at low frequencies in one or the other species. Several alleles showed clines in allele frequencies through the hybrid zone. We also sequenced a 307-bp region of the mitochondrial cytb gene and/or an 817-bp fragment of the ATPase locus from the same 104 individuals listed above (GenBank accession nos. DQ875685-DQ875741 and DQ875611-DQ875672, respectively). Sequences from the two parental species have fixed differences at 14 sites for the cytb fragment and 46 sites for the ATPase fragment, and each locus showed $5% sequence divergence among species. On the basis of these levels of sequence divergence, A. palliata and A. pigra likely separated $3 MYA (Cortés-Ortiz et al. 2003).
In total, 23 individuals from the putative hybrid zone wholly possessed alleles of either A. palliata (n ¼ 11) or A. pigra (n ¼ 12) and contained the respective species' mitochondrial haplotype; this suggests that individuals of both parental species are nearly equally abundant within the hybrid zone. Thirteen other individuals were identified as hybrids on the basis of mitochondrial and microsatellite data ( Table 2). The hybrid individuals included seven adult females, one infant female, and five adult males. Twelve individuals possessed microsatellite alleles, diagnostic of both parental species, although no individuals were F 1 hybrids ( Table 2). The lack of F 1 's may be because these individuals are ephemeral, or the hybrid zone is old, or it could reflect a low incidence of hybridization of pure parental forms (see Goodman et al. 1999). All adult hybrids contained the mitochondrial haplotype of A. pigra. The infant was the only hybrid that possessed A. palliata's haplotype. The presumed mother of this infant (based on genotypic data and the fact that the female was carrying the infant) was pure A. palliata based on the genetic markers used here and her appearance and occurred with a hybrid male that was likely the father of this infant based on the genotypic evidence. Hybrid individuals occurred in fragmented habitats where the two species' distributions overlap and were members of ''mixed troops'' that contained individuals of both parental species and in some cases individuals with unique or intermediate morphologies (Figure 2).
We attempted amplifications of a region of the Sry gene with genomic DNA of 4 A. palliata males, 2 A. palliata females, 3 A. pigra males, and 2 A. pigra females from outside of the putative hybrid zone and all 13 individuals from within the hybrid zone that were characterized as hybrids. Amplifications were successful only with genomic extractions of males; this and the fact that direct sequencing of amplification products yielded chromatograms without double peaks or other ambiguities strongly imply that the gene amplified occurs on the Y chromosome in these individuals. The sequences obtained from individuals outside of the hybrid zone were $821 bp in length and showed fixed differences at three sites among species (GenBank accession nos. DQ875673-DQ875684). All male hybrid individuals (n ¼ 5) possessed the Sry gene of A. pigra (Table 2).
If matings of hybrids are random and occur among all possible combinations of hybrids and parental species, we expect to find equal frequencies of the four possible genotypes of males and the two possible genotypes of females at the maternally inherited mitochondrial locus and the paternally inherited nuclear locus located on the Y chromosome of males. Although sample sizes are small, the chi-square goodness-of-fit tests suggest that the observed frequencies of males' and females' genotypes differed significantly from these expectations (Table 3). Moreover, probabilities of detecting 12 adult hybrids with the mitochondrial haplotype of A. pigra (P ¼ 2.4 3 10 À4 ), 12 hybrid individuals with the mitochondrial haplotype of A. pigra plus 1 hybrid individual with the mitochondrial haplotype of A. palliata (P ¼ 1.6 3 10 À3 ), and all 5 hybrid males with the Sry gene of A. pigra (P ¼ 3.1 3 10 À2 ) are low. These patterns imply that the direction of hybridization and subsequent backcrosses is strongly biased. Only crosses between A. pigra females or hybrid females carrying the mitochondrial haplotype of A. pigra and A. palliata males or hybrid males with the Sry gene of A. palliata occur and give rise to female offspring (Figure 3). However, no male  Api All individuals except S157 were adults; S157 was an infant still being carried by its presumed mother.
c Phenotype based on size and pelage coloration and texture. Apa, A. palliata-like; Api, A. pigra-like individuals.
d Mitochondrial haplotype. a, A. palliata; i, A. pigra. e Identity of alleles for parental species at each diagnostic microsatellite locus. a, A. palliata; i, A. pigra.
f Identity of the Y chromosome Sry gene. I, A. pigra; NA, primers did not amplify a product and were not expected to on the basis of the sex of this individual.
hybrids with the Sry gene of A. palliata were observed and so, in accordance with Haldane's rule (Haldane 1922), the data strongly suggest that the aforementioned crosses fail to produce viable males. Furthermore, on the basis of the low probability of detecting only the mtDNA haplotype of A. palliata in 12 adult hybrid individuals, A. palliata females and A. pigra males or hybrid males either mate infrequently or typically fail to produce viable offspring. Nonetheless, the genotypes of the hybrid infant and its suspected parents imply that this infant (S157, Table 2) was produced from a backcross between a male hybrid (S154, Table 2) and a female A. palliata. This demonstrates that such matings occur and that female offspring are produced. However, because no adult females were observed with the mitochondrial haplotype of A. palliata, we suspect that such crosses are uncommon or this infant is either infertile or will not survive to reproductive age.
We are currently investigating the potential role of morphological, behavioral, genetic, and cytogenetic differences as causes of the bias in direction of hybridization of these species. This work should advance our understanding of the speciation process and origins of reproductive isolation among primates, as well as the role of hybridization in primate evolution (Arnold and Meyer 2006). Moreover, study of the presence of hybrids in fragmented and intact forest tracts will reveal whether human-induced forest fragmentation has instigated hybridization by confining members of both species to small areas and limiting access to conspecific mates.
The Secretaría del Medioambiente y Recursos Naturales, Mexico, and the Autoridad Nacional del Ambiente, Panama, are gratefully acknowledged for providing the collecting, export, and import permits that made our research possible. This research was supported by the PROMEP Ph.D. scholarship no. UVER 98-11-019 and PROMEP 103.5/03/1154EXB-9 to L.C.   A. palliata male from within the hybrid zone; (B) hybrid male that occurred with an A. palliata female and a hybrid female (note that males A and B belonged to different neighboring groups); (C) hybrid male with mixed characteristics (size of body and hyoid bone similar to A. pigra but note the light hairs in the flanks); (D) A. pigra male from outside the hybrid zone. a Lowercase letters refer to the mitochondrial haplotype of A. palliata (a) and A. pigra (i); uppercase letters denote the genotype of males at the Y chromosome Sry gene for alleles of A. palliata (A) and A. pigra (I).