The Cetartiodactyla (even-toed ungulates, whales and dolphins) radiated approximately 70–80 Myrs ago. The relative positions of the camelid, suiform (pigs), hippopotamus, ruminant and cetacean (whales and dolphins) groups remain unclear, whether morphological or molecular characters are used for attribution 56789. Moreover, polytomies (unresolved tree nodes) within some Cetartiodactyla taxa 8 highlight areas for further data collection (both species and markers) and phylogenetic research. This is a particularly delicate problem within cetaceans, probably due to adaptative radiations within a short period of time 1011.

Males are the heterogametic sex in mammals, and usually, unequal numbers of males and females transmit genes from one generation to the next. Y-specific polymorphisms should allow the inference of sex-specific population parameters and decryption of breeding system patterns and dispersal strategies. Overall, the use of Y-specific markers has been restricted to evolutionary studies of human history and some scarce studies in population genetics, perhaps because of the low diversity of these markers 12. Within Cetartiodactyla, genetic structure or admixture, e.g. in sheep 13 or cattle 1415, has made use of a few Y-specific markers including microsatellites, SNPs and indels.

The matrilineally transmitted mitochondrial control region is commonly used as an informative sequence for population genetics. An equivalent had not been found to date on the Y chromosome. We considered the well-known amelogenin gene to be of particular interest because parts of it do not recombine between X and Y chromosomes.

Sex-chromosome recombination discrepancies have been exploited to develop many molecular sexing techniques. Although it can be ambiguous in some small populations, the amelogenin locus is the most commonly used for gender determination in humans 16. Accurate gender determination in mammals is crucial to various applications in reproductive technologies, forensic investigations and population management. Some techniques rely on specific amplification of loci localized on the Y chromosome (such as Sry 17) while others are based on amplification of homologous fragments from both X and Y chromosomes (e.g. ZFX/ZFY 18) or use both markers 19. Each of these has limitations, such as the need for multiplexing with other markers or additional steps such as digestion, labelling or sequencing. Several amelogenin-based techniques have expanded the taxonomic coverage of molecular sexing for Artiodactyla 202122 but they have not yet been extended to Cetacea. Therefore, there is a need both for new methods that apply to a greater number of species and to increase the number of cross-checking sexing methods, especially in conservation biology 23.

In this study, we found out that sequence variations in the amelogenin locus can be applied in evolutionary and population genetics as well as for molecular sexing in the highly diversified Cetartiodactyla group. We therefore carried out an evolutionary study of orthologous Amel-Y and Amel-X sequences (exons 4 to 5) in Cetartiodactyla. We studied four Cetacea (the striped dolphin, the bottlenosed dolphin, Risso's dolphin and the fin whale) and three Artiodactyla (cow, pig and sheep) species.

Amplification of the studied segment of the amelogenin locus using the species-specific SC1-SC2 primers resulted in an obvious sex-related size polymorphism in all Cetacea (Fig. 1) with a unique 521 bp band for females (two Amel-X copies) and an additional 980 bp band for the Amel-Y in males. This pattern was obvious in male Baleen whales (Mysticetes) but there was no corresponding Amel-X amplification in male dolphins unless by using the primers X5-X6 derived from the human amelogenin sequence. Previous studies showed that amelogenin amplification was prone to allelic drop-out or at least to preferential amplification 24. These phenomena may be explained by several factors. Usually, amplification of the lesser sized allele is favoured when the amount of polymerase is a limiting factor or in case of template DNA degradation 25. Small amounts o DNA may also increase stochasticity of the annealing 26. However, our results are not consistent with these situations since the allele favoured (Amel-Y) is always the greatest one. On the other hand, differences in GC content and mismatches in the annealing sequences may account for differential amplification. The amelogenin fragments that we studied are characterized by a higher GC content when amplified from the X chromosome (56%) than from Y-chromosome (47%). This difference may result from a non-insertion in the Amel-X fragment. This feature as well as a 2 bp-long mismatch between dolphin's Amel-X and the 5' end of the reverse primer SC2 (Fig. 2) may favour preferential amplification of the Y copy in dolphins (Fig. 1b). Indeed, amplifying male dolphin samples SC3 (primer without mismatch, see Fig. 2) instead of the SC2, restores the two bands, seen in baleen whales. The presence of this large insertion in the Amel-Y copy can be used for sex determination in probably all cetacean species.

In order to define the breakpoints of the Y insertion location and investigate its evolutionary history, we sequenced various cetaceans (listed in Methods; sequences deposited under the following accessions: EMBL:AM744958 to AM744971). After alignment with available sequences from Artiodactyla (see list in Methods), we detected the same polymorphism in all other Cetartiodactyla except the Pig (Fig. 3): a 460–465 bp insertion (size is a function of indels within different individuals or species) located between the 4^th and 5^th exons (188^th to 651^st position of Y sequences e.g. EMBL:AM744958). Haplotype names and their corresponding accessions are given in Table 1. Sequence similarity was checked by running BLAST (Basic Local Alignment Search Tool) over GenBank nr/nt nucleotide collection sequences with megablast algorithm (intended for high similarity sequences). In addition to the bovine and ovine Amel-Y, the only two relevant (78 and 83% homology, E-values 4.10^-68 and 3.10^-53) hits matched a fragment on the seventh chromosome in pigs (ca. 250 bp), suggesting the insertion might be a transposable element.

We interpret the presence of this insertion as a synapomorphy (shared character) of the Cetartiodactyla excluding pigs and probably other early derived groups (camels, hippopotamuses; 27, see Fig. 3). In addition to this long insertion, 46 other indels were detected by sequence alignment (positions and sizes detailed in Figure 5). Indels are particularly useful for testing phylogenic hypotheses, as they can provide information about ancient divergences rather than population information. We therefore assessed whether phylogenetic topologies differed if we took into account or not the information contained in these indels. Thus, the cetacean sequences summarized in Table 1 as well as Artiodactyla sequences were analyzed first classically, with gaps coded as missing characters, and secondly, with gaps coded as supplementary binary characters (see Fig. 5). For each analysis, two independent Bayesian searches were performed. The phylogenetic trees presented in Figure 4 result from a consensus of 20,000 trees sampled after standard deviation between the two runs dropped below 0.01. They show highly supported nodes. The phylogenetic analysis performed on the complete segment (Fig. 4a) confirmed the clustering by sex-chromosome copy in Cetartiodactyla (Stenella cœruleoalba, Balænoptera physalus, Grampus griseus, Tursiops truncatus, Bos taurus and Ovis aries) whereas Amel-X and Amel-Y clustered together in other mammals (Homo sapiens, Sus scrofa) together with Amel-X from Cetartiodactyla. On the other hand, phylogeny inferred without taking into account the insertion gave a different result (Fig. 4b): whereas haplotypes also clustered by chromosome in cetaceans, no signal related either to species history or to chromosome bearing could be seen in the other Cetartiodactyla. Hence, the phylogenetic signal related to species history seems to strengthen as we follow the tree from Cetartiodactyla towards primates. This partial, homoplasic, persistence of the phylogenetic signal may be explained by the influence of the region surrounding the insertion. This could be the result of the old age of the insertion (74–87 myrs 27). The subsequent loss of homology may have given rise to a more divergent evolution between chromosomes in some taxa (Cetacea) than in others (Artiodactyla).

It would be interesting to study this region at the whole clade level by combining sequence and indel characters in the same analysis. This could give clues to test the many hypotheses about basal radiation of Cetartiodactyla (e.g. 568). Given the presumably basal position of the Suioidea and Tylopoda in the Cetartiodactyla phylogeny (7 and Fig. 3), we hypothesize that the major evolutionary event represented by the insertion (illustrated by an arrow Figure 4a) occurred once in the Cetacea-Ruminantia clade and not in the remaining Cetartiodactyla.

The presence of this large insertion in the Amel-Y copy can be useful for sex determination.

Evolutionary history also indicates that our sexing technique is applicable, in addition to cetaceans, to over a wide range of Cetartiodactyla species including domestic and wild species, in particular the widespread Ruminantia (Bovidae, Capridae and most likely Cervidae). It is however not suitable to Suiformes and further studies are required to confirm that the technique is also not applicable to Camelidae, given their even more basal position in the Cetartiodactyla phylogeny.

In dolphins, the Amel-Y fragments amplified with the SC1-SC2 primer pair were easily sequenced without the need of cloning since amplification was Y chromosome-specific. From the ten Striped dolphin samples sequenced, nine were males, and we could deduce seven distinct Y-haplotypes (one haplotype represented by three individuals and four individual haplotypes) bearing 64 polymorphic sites (nucleotide diversity π = 0.004 ± 0.0007). Half of these were in the ~460 bp insertion. An alignment of polymorphic sites is presented in Figure 5 (a). Strikingly, these sequences showed two highly divergent haplogroups, diverging by a mean of 49 substitutions. This concords with our results that support the probable existence of two subspecies within the Mediterranean sea (unpublished data). Moreover, one of these haplogroups displayed a high degree of polymorphism, with 24 informative sites, whereas the others showed only eight. These values are sufficient for use in pedigree analysis and population genetics, as the Y chromosome counterpart of the mitochondrial d-loop in this species. Indeed, in striped dolphin the intra-specific (inter-group) divergence is greater than inter-specific divergence with a mean of 45 nucleotide substitutions between the striped dolphin and fin whale sequences. There is an average of 0.048 ± 0.01 substitutions per site when comparing the two striped dolphin populations. This is comparable to the divergence observed between each population and the Common dolphin (0.058 ± 0.03) and confirms that nucleotide diversity is one order of magnitude higher than the range observed (10^-4) for Y chromosome markers in mammals 12. As for the mitochondrial d-loop, the size of the amplified fragment slightly limits the use of the technique. Some degraded samples do not amplify; even so, a particularly degraded sperm whale sample was still amplifiable (data not shown).

Since the Y chromosome population is expected to have a small effective size, it is more likely to be affected by genetic drift. Thus, it reflects more recent demographic events such as bottlenecks, expansions or founder effects 28. To study this sort of event, one needs a marker whose diversity is high enough to allow the reconstruction of gene genealogies with the least ambiguities and in regions where recombination does not interfere with the uniqueness of the trees. For this purpose, highly variable microsatellites represent valuable markers but they require intensive computing methods to take into account uncertainties in the trees arising from alleles that are identical by state and not by descent (homoplasies) 2930. Adding a new sequence marker is therefore of interest for Y-chromosome population genetics in Cetartiodactyla. Moreover, the Bayesian estimate of mutation rate on each edge of both trees in Fig. 4, jointly computed with phylogenetic inference, shows high values for a marker of nuclear DNA: between 10^-8 and 10^-10 substitutions per site and per year in all Cetartiodactyla branches. This value is intermediate between those of mitochondrial d-loop and nuclear DNA in mammals 3132.

We found two stop codons at amino acid positions 98 and 99 of exon 5 in all Y chromosome copies of amelogenin in the four studied cetacean species (positions 988–993 of sequence EMBL:AM744959). The Amel-Y gene product may therefore be truncated in these species or represent a pseudogene as already observed in species from most of the other eutherian clades 33

Biological material was isolated from soft tissues sampled from dead stranded cetaceans and extracted using a classical phenol-chloroform protocol 36. We used heterologous primers, X5 (5'-GTGCTTACCCCTTTGAAGTG-3') and X6 (5'-CTTCCTCCCGCTTGGTCTTG-3'), designed from the amelogenin intron 4 and exon 5 of Homo sapiens X chromosome (GenBank:NC 000023, chrX:11221454–11228802) reference assembly Build 36.3, to amplify the homologous region in cetacean amelogenin. This region of Amel-X is 92% identical to Amel-Y.

We subsequently cloned and sequenced these PCR fragments and designed oligonucleotide primers specific to the Artiodactyla. They are anchored in exons 4 and 5 of Amel-X and Amel-Y, which allows complete amplification of the 4^th intron (SC1: 5'-CAAGCATGCATTTCAATTCCC-3' and SC2: 5'-CTGCATGGGGAACATCGGAG-3'). Optimal PCR conditions were adjusted by using temperature (49–62°C) and MgCl₂ (1–2.5 mM) gradients. Following this optimization step, the PCR amplifications were conducted in a reaction mix consisting of 477 mM KCl, 1 mM Tris/HCl pH 8.3, 1.5 mM MgC1₂, 250 μM each dNTP, 2 pmol/μl of each primer, 3 units of Taq polymerase in a final volume of 25 μl. Cycling was conducted as follows: 95°C for 2 min, followed by 30 cycles of denaturation at 94°C for 1 min, annealing at 55°C for 45 s, extension at 72°C for 1 min, and a final extension at 70°C for 10 min. PCR products were run on 1.2% agarose gel ethidium bromide stained, alongside a 1 kbp ladder (New England Biolabs, County Road, MA). In order to validate the assay, the gender, if identified during examination of the stranded carcasses, was recorded (22 out of 38 samples). These amplifications were conducted in eight Cetacean species: 20 Striped dolphin (Stenella cœruleoalba), five Fin whales (Balaenoptera physalus), four Bottlenosed dolphins (Tursiops truncatus), three Common dolphins (Delphinus delphis), three Gray whales (Eschrichtius robustus), two Sperm whales (Physeter macrocephalus), one Minke whale (Balænoptera acutorostrata) and one Risso's dolphin (Grampus griseus).

Of these, we sequenced striped dolphins (9 males and one female), fin whales (two males and two females), common dolphin (2 males), gray whale (one female) and Risso's dolphin (one female). Sequencing was performed on an ABI Prism sequencer (Applied Biosystems, Foster City, CA) with the dye terminator protocol directly for fragments amplified using the same (SC1-SC2) primer pair or after cloning in pGEM-T (Promega, Madison, WI) for fragments amplified using the X5-X6 pair. Each sample was sequenced from at least two independent amplifications for greater reliability.

The obtained sequences were aligned using ClustalW, a clustering-based sequence alignment algorithm 37, with bottlenosed dolphin (GenBank:AY787743S2), cow (GenBank:AB091789–AB091790), pig (GenBank:AF328419 and AB091792), sheep (GenBank:AY604731) and human (GenBank:NT_011757 from 9098117 to 9098612 and GenBank:NC_000024 from 6796200 to 6796719) sequences. Striped dolphin haplotypes were named according to membership in one of the two Mediterranean populations inferred by population genetic data on mitochondrial and microsatellite DNA (data not shown). Stenella cœruleoalba YAn thus represents the n^th haplotype from population 1 while Stenella cœruleoalba YBn represents the n^th haplotype from population 2.

Phylogenic inference was performed using a Bayesian Monte Carlo Markov Chain (MCMC) approach, implemented in Mrbayes 38, that allows determining posterior probability of each node of the tree.

Given the high indel density in this region, data were divided into two partitions: a sequence matrix (1120 characters corresponding to nucleotides) and a binary character matrix (presence/absence of the indels). Gap coding was achieved using Modified Complex Indel Coding 39 in which parsimony-informative gap characters were scored from pairwise unambiguously aligned sequences. Hence, a total of 47 gap characters were included (summarized in Figure 5b). We used a General Time Reversible (GTR) model, allowing for a proportion of invariant sites for sequence partition, and a binary F81-like model with a variable ascertainment bias for the insert's absence/presence (state 0/state 1) for the binary partition. In this model, π₀ and π₁ are the stationary probabilities of the two states and π₀/π₁ the rate of transition from state 1 to 0.

Patterns of genetic diversity were computed using DnaSP 40. Specifically, the average number of substitutions was computed using Nei's equations 10.5, 10.6, 10.7 41.