Conclusiones e implicancias generales

Two different approaches have been used to date the time when the domestica-tion of dogs took place. The first one is to try to identify bone remains whose mor-phology corresponds to dogs and not to wolves (the ancestor of the domestic dog).

The second method consists of estimating the date when the mtDNA sequences present in dogs separated from the sequences in wolves. Although both methods target the moment when the dog and wolf lineages split, they look at different things and the dates obtained with both methods are expected to be different. To understand this, imagine one lineage that at a certain time in the past splits in two, as indicated in Fig. 3.4. This could correspond to the split of dogs from the ances-tral wolf lineage, and the goal of the studies about domestication is to character-ize when this separation occurred. It is at this point that the morphology of the two species starts to diverge. However, it will take some time for their morphol-ogy to be sufficiently divergent as to allow the unambiguous identification of bone remains as belonging to one species or the other. Zeder and Hesse (2000) and Zeder (2003) have shown that for other domestic mammals the morphological differentiation from the ancestor did not occur until some time after the initial domestication, when the domesticates expanded to neighbouring areas where they could not mate with wild animals. Similarly, we could expect that the mor-phological change in dogs may have occurred some time after the domestication

Fig. 3.4. Difference between the species divergence time, and the estimates obtained using genetic and morphological data. Species are indicated as grey bands, and the black lines represent the genetic relationships between sequences from each species. Morphological differentiation between species increases from the time of reproductive isolation.

(although we do not know how much later). Since this is the kind of evidence used by archaeologists, and because the ancient dog bone remains are very scarce, the dates of domestication based on archaeological data will tend to be more recent than the split between dogs and wolves.

Molecular estimates of the time of divergence are based on the comparison of DNA sequences. By applying a molecular clock it is possible to estimate the time necessary to make the two lineages as different as they currently are. In Fig.

3.4, the mtDNA lineages present in wolves and in dogs are represented as the lines inside the grey area. However, the split of the lineages is necessarily older than the split of the populations (the domestication). Consequently, estimates of the date of the domestication based on DNA sequences are likely to be older than the actual event.

Another way to use mtDNA sequences to estimate the time of domestication is to calculate the time of divergence between the most divergent sequences within the domestic clade (the time when dogs started diverging from one another; Fig.

3.4). This estimate is highly dependent on the demographic history of dogs: the origin of these sequences might appear more recent when the popula-tion remained small (Avise, 2000). However, it seems logical to think that dogs started from a limited number of individuals and soon after domestication the population started to expand to produce the current population size of several hundred million dogs worldwide (Coppinger and Coppinger, 2001). In this case, the estimates should be reasonably close to the beginning of the dog population expansion, and probably also close to the time of domestication. This is likely a very much better estimator of the age of domestication than the previous ones and should be expected to be younger than the split between dogs and wolves, and older than the first appearance of morphological divergence (see Fig. 3.4).

In summary, while archaeological approaches may lead to an underestima-tion of the date of domesticaunderestima-tion, molecular methods tend to produce overesti-mates. Estimating the age of the diversity within the domestic lineage as opposed to the split between the domestic and wild lineages may produce a more accurate date. While the search for old dog-remains continues, molecular analysis methods are constantly improving, and integrating both approaches will lead us towards a more accurate view of the domestication process.

The archaeological record suggests that dogs were present in the Middle East and Central Europe about 14,000–15,000 years ago (Nobis, 1979; Olsen, 1985;

Dayan, 1994; Clutton-Brock, 1999; Sablin and Khlopachev, 2002). Early dogs were morphologically distinct from grey wolves. They are often identified by their smaller body size and wider crania, a more prominent stop on the face and a shortened, crowded jaw (Olsen, 1985; Morey, 1992). The identification of these bones as belonging to dogs and not to wolves implies that the animals had been living with humans for a time period sufficiently long to allow some morphologi-cal differentiation from their wild ancestor. The small body size of Asian wolves and the shared presence with early dogs of some traits led to the suggestion that Asian wolves are the direct ancestor of the dog (Olsen and Olsen, 1977).

However, the oldest dog-remains unmistakably identified so far have recently been discovered in western Russia (Sablin and Khlopachev, 2002) and date to 13,000–17,000 radiocarbon years before present (this is equivalent to 16,000–20,000 calendar years before present). These early dogs do not show a reduction in body size when compared to the local populations of grey wolves and have cranial proportions similar to the great dane.

The first attempts to use molecular genetics to study the domestication of dogs took place in the mid-1990s. Okumura et al. (1996) were the first to sequence the control region of a large sample of dogs, but focused on Asian dogs. These authors observed that the mitochondrial DNA sequences grouped into a discrete number of clades. Tsuda et al. (1997) similarly studied a large sample of dogs from Asia, but also included some samples from wolves. Consequently, they were able to convincingly demonstrate a close relationship between wolves and dogs. The close relationship was confirmed in a study published at the same time (Vilà et al., 1997), but in this study an attempt to use the molecular information to estimate the date of domestication was made. The study by Vilà et al. included 162 wolves from 27 populations from throughout Europe, Asia and North America and 140 dogs representing 67 breeds. As Okumura et al. did 1 year earlier, Vilà et al.

observed that when constructing an evolutionary tree with all dog sequences, they clustered in just four groups (clades I–IV, Fig. 3.3). One of these groups, clade I, contained most of the studied dogs, and a more recent study including more than 600 modern dogs (Savolainen et al., 2002) concluded that about 71% of them had sequences belonging to this clade. This large diversity could indicate that this group of sequences was the oldest one and thus its origin could correspond to the original domestication event.

To estimate the date of the domestication, Vilà et al. (1997) calculated how much time would be necessary to reach that level of diversity starting from a single mtDNA sequence, using the molecular clock. Vilà et al. first estimated the average sequence divergence between wolves and coyotes for the studied region of the mtDNA and found that it was 7.5%. The fossil record indicates that mor-phologically differentiated wolves and coyotes existed at least 1 million years ago (Kurtén and Anderson, 1981). Next, the maximum divergence for sequences belonging to clade I was estimated to be 1%. This was used to calculate when the sequences from that clade started to diverge. To confirm that the rate of evolu-tion was not significantly different for the different lineages, the lengths of the dif-ferent branches in the tree were compared. As no difference was found, the same molecular clock could be used for all the sequences. Finally, since the rate of divergence was around 7.5% per million years, the estimated time needed to gen-erate divergences up to 1% was 135,000 years! The result of the study of Vilà et al. (1997) has been taken as a firm date. However, that estimation was not pre-sented with confidence intervals and these are usually large in this sort of estimate.

Also, the region of the mtDNA studied may not be the best to apply a strict molecular clock. On the other hand, the results very clearly suggested that the domestication of dogs had to be much older than indicated by the archaeological record.

More recently, another attempt has been made to use dog mtDNA informa-tion to estimate when dogs were domesticated (Savolainen et al., 2002). These authors also focused on the diversity of clade I and concluded that the domesti-cation could have taken place more recently. These authors used a different model of sequence evolution and only considered the sequences in their sample that were currently present on each continent. Since they found the largest diver-sity in East Asia they concluded that this could have been the place of domesti-cation. Looking at the average sequence divergence among clade I sequences in East Asia, they estimated that the age of the clade was 41,000 years (SD±4000), supporting an origin much older than suggested by the bone remains as the pre-vious study. However, they considered that it was possible that clade I did not derive from one unique domestication event and that several wolf lineages were involved. Clade I was arbitrarily subdivided into various smaller groups and the ages of these sub-clades were estimated to be between 11,000 and 26,000 years old. Despite this wide range of dates, they suggested an East Asian origin for the domestic dog 15,000 years ago (Savolainen et al., 2002). There is a series of prob-lems with this approach. First, East Asia was suggested as the centre of domesti-cation because of its higher diversity. However, whereas most of the samples from the rest of the world corresponded to purebred dogs, this region was character-ized by mongrel dogs and dogs of non-recogncharacter-ized breeds. As these animals do not belong to inbred lines it seems likely that they may have a much higher genetic diversity for this reason alone, and this can bias the results. Second, the patterns of current diversity may be very different from ancestral patterns (see the study by Leonard et al., 2002), so looking at modern dogs may not be appropriate to recon-struct patterns of diversity that existed in one region in the past. Third, no objec-tive evidence indicating that clade I derived from one or more founding wolf lineages was provided. Subjectively dividing the clade in smaller groups can produce whatever result the researcher wishes to obtain. Lastly, despite estimat-ing a very old date of origin for one of their sub-clades, 26,000 years (confidence interval: 18,000–34,000 years), the authors conclude that the domestication took place about 15,000 years ago.

Although the results of this study apparently fit with the dates derived from archaeological remains, the conclusion that dogs could have been domesticated in East Asia 15,000 years ago faces one main problem: dog remains (already mor-phologically differentiated from wolves) of about that time have already been dis-covered in several places in Europe (Nobis, 1979; Sablin and Khlopachev, 2002) and Asia (Dayan, 1994; Clutton-Brock, 1999), and probably also existed in America (Leonard et al., 2002). The analysis of mtDNA sequences in pre-Columbian dogs of Leonard et al. (2002) has shown that American dogs clearly had Eurasian origin, implying that they arrived in the Americas with the first humans. Since all these dogs in Europe, Asia and America shared a common origin, we can presume that the domestication of the dog had to take place sig-nificantly before 15,000 years ago to allow their expansion over three continents.