The Wolf Within: The Astonishing Evolution of the Wolf into Man’s Best Friend

Although the scope of mDNA and the Y-chromosome is limited to just two genetic systems, we must not make the mistake of underestimating the importance of mitochondrial DNA and the Y-chromosome in crashing through the barriers of uncertainty surrounding the origin of the dog. Scientists from Darwin onwards have pondered this question with no means of coming to a definite conclusion. Were jackals or bush dogs or wolves or coyotes or foxes or hyenas or some other animal, possibly long extinct, the true ancestors of the modern dog? The research, first with mitochondrial DNA and then with the Y-chromosome, has made the answer crystal-clear. Wolves, and only wolves, are without question the ancestors of all living dogs.

4

On the Origin of Wolves

The genetic trees drawn with the help of mitochondrial DNA and the Y-chromosome make it very clear that the only ancestor of all dogs is the wolf. There is more we can decipher from the DNA results, but before we come to that, what do we know about the ancestry of wolves?

The Age of the Mammals began in the Cretaceous period following the sudden extinction of the dinosaurs some 65 million years ago. This extinction left a big gap in the fauna which was gradually filled by mammals, which until then had been an inconspicuous group of small furry animals cowering in the undergrowth. Their numbers increased, and by 40 million years ago, during the Eocene epoch, the emerging mammals began to evolve into today’s familiar groups: horses, deer, elephants, apes, dogs and cats, early forms of modern Orders.

Wolves, and therefore dogs, belong to the last of these Orders, the Carnivora. It has become the most diverse of any Order, embracing over 280 species, and it takes its name from the Latin that describes the main characteristic of its members: ‘flesh-eating’. They are the carnivorans, as distinct from the general term ‘carnivores’, which takes in all meat-eating species, be they fish, reptiles or even plants. A major division of the Order Carnivora is the Family Canidae, which includes wolves, coyotes, jackals and foxes. Cats large and small belong to the Family Felidae, bears and pandas to the Ursidae, while badgers, hyenas and seals belong to other, separate, families. Some carnivorans, like the Giant Panda, are strictly vegetarian but are still included within the same Order. It is their teeth that set the Carnivora apart from other mammalian Orders. All have well-developed third incisors, which serve to pierce the flesh of prey animals to prevent escape and to kill, while unique to carnivorans are their fearsome carnassial teeth, taking the place of our molars. Carnassial teeth are razor sharp and self-sharpening and are designed to slice through flesh like a pair of shears rather than merely tearing at it.

The dog-like carnivorans, the Canidae, and the cat-like Felidae began to diverge from each other and become gradually more specialised. As far as we can tell from the fossil record, the earliest canids evolved in North America where the oldest fossil dog, Cynodesmus, was discovered in Nebraska, USA, and lived between 33 and 26 million years ago. At one metre in length, it resembled a modern coyote and, by its dentition, was clearly carnivorous with large canine teeth for grasping and tearing the flesh of its prey. Soon after, on an evolutionary timescale, some truly fearsome carnivores began to evolve, including the bone-crushing Cynarctus. As these monsters became extinct about 11 million years ago they were replaced in turn by other canids, notably Tomarctus, found all over North America from Florida, north to Montana, west to California and south to Panama. From the size of the jaw muscle insertions in their skulls it is clear that Tomarctus had a bite strength far greater than required to kill its prey. This led to the conclusion that, like modern-day hyenas, Tomarctus was able to crush bones to reach the nutritious marrow of scavenged carcasses.

By the middle of the Miocene epoch, some 10 million years ago, the canids had spread from America, first to Asia, then to Europe and finally to Africa. As they did so, the ancestors of today’s wolves gradually evolved towards a lighter, faster frame in order to hunt swift herding prey like elk and wild horse. They hunted not as individuals but as members of a pack. Thus began the key development in the evolution of the modern wolf, and ultimately of the domestic dog. To be effective hunters they developed the ability to communicate with each other and to work as a team. Wolf packs travelled with the herds, following them throughout the year, a habit that our own human ancestors adopted.

5

The Living Fossil

All our efforts to reconstruct the past can only ever give an approximation of what really happened. Well-preserved fossils are spectacular but rare and their discovery can only ever convey a patchy record. History is notoriously inaccurate, depending on the inclinations of the author. Mythologies require sophisticated interpretation. Genetics is no different. It is just another foggy lens through which we try to make sense of times gone by. Bearing that in mind, let us clean the eyepiece and take another look.

In Chapter 3 we saw how DNA from living dogs and wolves was able to reconstruct a plausible genetic relationship between the two. These were inferences from modern DNA but, astonishingly, DNA can survive for thousands of years in fossil bone and teeth. As we will see later, it is often in a pretty bad state. Nonetheless it does give us the chance to examine ancient sequences directly rather by inference. Later on, we will have a closer look at how ancient DNA has helped us follow the evolution of the dog. But before that we need to know a little more about DNA itself.

All genetics depends on mutation, the ultimate source of all variation. DNA changes over time. When a cell divides, its DNA is copied so that each of the two daughter cells contains the full set of genetic instructions. The copying process is astonishingly precise and accurate, and the error rate is minuscule. Editing mechanisms within each cell scan the copies for errors and correct them. But the error rate is not zero. After each cell division, roughly 1 in 1,000 million mutations gets through uncorrected. If the emerging mutation changes a vital component of a gene, then the daughter cell will either malfunction or die. Only extremely rarely will a mutation be beneficial. The most dangerous malfunctions are those that turn normal cells into malignant ones which lose the capacity to restrain their own cell divisions, and they develop into tumours. That is why in some rare diseases, where the DNA editing and correction capacity of cells is faulty, it leads to much higher rates of malignancy.

Fortunately, the majority of DNA copying errors has no consequences whatsoever; first, because the errors don’t occur in important genes, or, second, because they are not passed to the next generation. Only mutations in the germ line, being the cells that go on to form eggs and sperm, are capable of travelling on through time. Even then, the vast majority of sperm never get to fertilise an egg, and, in mammals, most eggs are not fertilised anyway. For these reasons alone, the overwhelming majority of germ-line mutations which occur through faulty copying, even the most potentially damaging of them, are not passed on.

Some mutations, however, do get through to the next generation. Most will not be noticed and have no significant effect on the body, either because they occur in unimportant genes or in the gaps between genes in the long stretches of DNA whose function, if any, is still largely unknown. Here it is worth distinguishing between genes and the rest of our DNA. Genes do something, usually instructing cells how to make proteins. There will be more on this when we take a look at gene mutations that have been found in dogs, but for now we will concentrate on the inconsequential mutations that have no effect, neither good nor bad. Precisely because they are so inconsequential, these unassuming ‘neutral’ mutations are the lifeblood of the sort of genetic reconstructions of past events that we have covered so far. A damaging mutation in a vital gene will disadvantage the individual who carries it. Not necessarily fatally, of course, but enough to put him or her at a slight reproductive disadvantage and thus reduce the prospects of the mutation being passed on to the next generation. Generally, over time, the mutant gene will be eliminated by selection, though not always, as we shall see in pedigree dogs. However, the humble and meaningless mutations that have no effect on anything of great importance will escape the scrutiny of selection and will sail on unmolested through future generations. It is these humble mutations that are the guiding lights that illuminate the history written in the language of the genes.

To explain how mutations are used, in dogs and humans, to date past events like the timing of the transformation from wolf to dog, let us imagine a desert island in the middle of a vast ocean. A young couple arrives in a canoe. For our purposes, it could equally well be a couple of shipwrecked dogs. The island is a paradise, with plentiful fresh water in bubbling streams flowing down from high mountains in the interior. There are coconut palms, shellfish and crabs in the sea and no predators or dangerous animals to disturb the idyll. Everything needed for life is on hand, and the couple start a family. Their children grow up in this cradle of abundance and, ignoring incest taboos for the sake of this exercise, have children with their siblings.

Time passes. The population settles down to a stable total of 1,000. Nobody leaves the island, there is plenty for everybody, and no one else arrives. Until one day a scientist and a research assistant turn up and begin taking a DNA sample from each of the inhabitants. The samples go off to the lab and the sequences are read. A few weeks later the scientist and his assistant, now back home, get the results. What can they deduce about the people on the island from the results? It doesn’t matter all that much which genetic markers we are talking about for this example to work, so let’s keep it simple and imagine that we are working with mitochondrial DNA. The first things the researchers notice is that everyone’s DNA sequence is very similar. Some sequences are identical, and we will call that the ‘core’ sequence of the island. However, about half the people have a sequence that differs at just one DNA base from the core.

DNA sequences are written using a childishly simple alphabet with only four letters. These letters represent simple organic chemicals, or bases, joined together in a linear sequence. Their abbreviations are even simpler : A, G, C and T. Any DNA sequence is a long string of these bases: … CCGGTAA … and so on. A mutation might change a T to an A, making the new sequence read as … CCGGAAA … The language may be child’s play, but the meaning is far from simple, as we will explore further in a later chapter, but not now. Instead we travel back to our island.

Of the 1,000 people who were tested, 500 have the core sequence and 500 have a one-base difference from it, but not all at the same one.

The researchers draw the reasonable conclusion that everybody on the island is ultimately descended from one couple, or rather from one woman, as we are dealing with mitochondrial DNA. Can they tell from the results how long ago the island was settled? To get an answer, we need to agree a very important factor. The mutation rate. That is the rate at which mDNA mutations occur and get passed on. It is going to be an estimate, drawn from other results. The factors which contribute to the estimate are sometimes astonishingly crude.

A common approach is to take two species, say human and chimpanzee, compare their DNA sequences and make an assumption about how long it is since they last shared a common ancestor. In this example the usual figure is 6 million years, based on fossil evidence which, for both species, is extremely flimsy. All genetic dating of past events depends crucially on the accuracy of the mutation rate and that it has remained stable over the period.

Fortunately, the estimates of the mitochondrial DNA mutation rate by the various methods come up with a figure that most are happy to accept. For the segment of mitochondrial DNA that Wayne and Vilà used, the rate is estimated to be one base change every 20,000 years. Mutations occur randomly as cells divide, so we must turn to discussing probabilities. A mutation rate of one per 20,000 years doesn’t mean that no mutations occur until that time has passed. It is an average. It could happen in the first generation or the last or, more likely, somewhere in between. Let us say the time between generations on the island is twenty years. If a quarter of the population has a mitochondrial DNA sequence that is one mutation away from the core, the average number of mutations per person across the whole island is then one quarter. The estimated time from first settlement then becomes the average number of mutations (a quarter) from the core, multiplied by the mutation rate (20,000), which comes to 5,000 years.

Returning to our scientists, they go back to the island to inform the council of elders of their results of the project. They also reveal that the original settlers had come from the mainland far away to the east because that is where they have also found the core mDNA sequence among the inhabitants. After listening politely to the presentation, the elders turn to the scientists and, as I have experienced first-hand, they say, most politely, something like ‘Thank you for your trouble. We knew that all along.’

In my deliberately simplified example we were dealing with just one segment of DNA on an island originally settled by only one couple. No one arrived or left for millennia. It doesn’t get any simpler than that.

Let us now suppose that other things happened on the island. Perhaps half the population died in an earthquake, or the central volcano erupted, destroying the crops, and three-quarters of the people starved, or an epidemic killed 90 per cent of the population. These are the sorts of catastrophes which might have happened in real life. Those events can severely distort our calculations. For instance, and let’s make it extreme, a tsunami kills everybody on the island except a couple who were far out to sea fishing at the time. They survived and, over time, their offspring repopulated the island. In this scenario, the genetic calculations would give the time that had elapsed since the tsunami rather than since the original settlement. The island would have undergone a ‘population bottleneck’. There would be no way of telling, by genetics alone, for how long the island had been settled before the tsunami struck. If we introduce further complexity, like a few boatloads of new arrivals, then all hope of being precise about the original settlement date goes up in a puff of smoke.

Given these unknown and often unknowable factors, I take claims of accurate genetic dating of past events with a large pinch of salt. That does not mean they lack value, but it is a mistake to become a slave to such calculations. We will use the island metaphor again when we come to consider the origins of pedigree dog breeds. Wayne and Vilà also used this kind of calculation to estimate the timing of the wolf–dog transition. The answer was much further back than anyone suspected, between 76,000 and 135,000 years ago.

6

Let the Bones Speak

At some point in the past the lives of wolf and human became intertwined and it is from this partnership that the dog eventually emerged. Until genetics entered the fray, the only way of following this transition through the intervening millennia was through fossils. Good fossils are in short supply and the fossil record is understandably full of gaps.

In terms of time, the oldest skulls that could even remotely be differentiated from wolves were excavated in the Goyet cave in southern Belgium in the 1860s. Like all good fossil sites, Goyet is a limestone cave whose alkaline environment helps to preserve the calcified bones and, importantly, any DNA that might lie within.

From studying the style of stone and bone tools found there, it was clear that the cave had been occupied by humans for a very long time. Neanderthals lived there during the time of the Mousterian culture, which lasted from about 160,000 years to 40,000 years BP (the standard archaeological abbreviation for ‘before present’). It takes its name from the rock shelter at Le Moustier in the Dordogne region of central France. The Mousterian lasted until the arrival of modern humans, our ancestors, about 40,000 years ago. As is not uncommon with early excavations, disturbance of the layers within the cave made precise stratigraphic dating of the different artefacts found there problematic. However, carbon-dating of the fossils gave precise dates for the organic remains at least. The cave fauna was a rich assemblage of cave bear, cave lion, horse, reindeer, lynx, red deer and mammoth. In the deeper recesses of the cave archaeologists found the skull of a ‘large canid’ carbon-dated to 31,700 years BP. Was it a wolf or was it a dog?

Of course, there must have been a period after the first wolf was adopted into a human band when its skull was exactly the same as a wolf’s – because it was a wolf. There was no exact moment of transition from one to the other, and the whole debate has a strong flavour of semantics. The more cautious authors merely refer to these intermediates as ‘canids’ or ‘wolf-dogs’, thereby sidestepping the argument altogether.

A similar conundrum faced archaeologists excavating the nearby site of Trou des Nutons, a cave formed in the limestone hills of the Ardennes by the River Lesse, a tributary of the Meuse. Among the fossils found in the Trou des Nutons were beaver, roe deer, horse, bison and wild sheep, suggesting a later occupation than at Goyet. This was confirmed when another skull of a mystery ‘large canid’ was given a carbon date of 21,800 years BP. This is a surprisingly early date and in the middle of the last Ice Age. But was it the skull of a dog or a wolf?

These skulls from France were subjected to a series of precise measurements of snout-length and width, the length of the tooth row and the size of the flesh-shearing, self-sharpening carnassial teeth that wolves and dogs have where we have molars.

Fossil canid skulls from two archaeological sites in Russia and Ukraine, one at Mezin (Ukraine) and the other at Avdeevo just over the Russian border, were given the same treatment. These two sites were inhabited by early humans who constructed huts of mammoth bones and left behind an abundance of beads and other artefacts carved from mammoth ivory. The objective of the osteometric study of candid fossils from these two sites was to discover whether the remains of these ‘large canids’ differed sufficiently from wolves in their skull morphology to be classified as dogs on their way to domestication rather than unmodified wolves.

To complete the comparisons, the analysis was extended to include later, but still prehistoric, unambiguous fossil dogs from France and Germany. Also included were a selection of modern and fossil wolves from Europe and Asia along with modern dogs from several large breeds including Great Dane, Tibetan Mastiff, Siberian Husky, Chow Chow, Irish Wolfhound, Malinois, Dobermann Pinscher and German Shepherd.1

Comparing multiple skull measurements from dogs of different sizes is a complicated business, and I will spare you the details of the multivariate analysis and go straight to the main conclusion. The Palaeolithic skulls from the oldest sites, including Goyet at 31,700 years BP, had a significantly different shape from modern, or indeed fossil, wolves. This suggests that, even by that early date, these animals were dogs already on the way to modification through ‘domestication’. An alternative explanation, though in my opinion rather less likely, is that these were the skulls of one or more wolf species that later became extinct. As we shall see later, there is other enticing evidence to support the former scenario and suggest that the close association between wolf and man began a very long time ago.

The next layer of evidence about the changing appearance of domesticated dogs comes from the late glacial period around 17,000 years BP, when the ice sheets covering northern Europe were fast retreating. The shrinking tundra no longer supported herds of large prey animals. The climate warmed considerably, rainfall increased and forests covered much of the formerly open tundra. The fauna changed with the landscape and many prey animals disappeared. Mammoths, woolly rhinoceros and their predators, the sabre-tooth tiger and cave bear, were forced into extinction. Others, like the wild horse, reindeer and bison, shifted their ranges. Humans began to spread north, first following the shrinking herds and later, as they entered the Mesolithic period, changing their diet to smaller woodland prey, like wild boar, pine marten, red and roe deer. On the coastal settlements, shellfish became a major source of food and the first boats ventured out to sea to catch fish. Supplementing this meagre protein diet were roots and tubers, insects and snails. The heroics of the mammoth hunt became a thing of the past and life became a gruelling fight for survival.

The close cooperation between human and dogs, by now thoroughly assimilated into human society, continued even though the superbly effective working partnership that had developed in the Upper Palaeolithic was at its best when killing large prey, a practice which by now was rapidly disappearing.

Around 12,000 years BP much smaller dogs made their debut in the fossil record. A team of French archaeologists found the remains of thirty-nine dogs at the Pont d’Ambron rock-shelter in the Dordogne. From an osteometric analysis similar to that carried out at the earlier sites of Goyet and Trou des Nutons in the Ardennes, it was clear that the Pont d’Ambron dogs were considerably smaller. The same was true with the remains excavated at the Montespan cave in the northern foothills of the Pyrenees and at the open-air site of Le Closeau in an old channel of the River Seine.

The authors of the exhaustive paper summarising this body of work confidently concluded that they were dealing with the remains of dogs and not wolves. In France at least, and also in Spain, dogs were clearly changing. In Russia, however, at around the same time, wolf-dogs were still very large. Whether this was a result of separate wolf domestications in the two regions or for some other reason, it was impossible to say. One firm but rather grisly conclusion, drawn from cut-marks on the bones of the Pont d’Ambron dogs, was that they had been butchered and, presumably, cooked and eaten.

As well as the issue of timing, the identification of the geographical location of the wolf–dog transition has absorbed many researchers and continues to do so. The first scenario to be proposed, by a group from the University of Konstanz in Switzerland led by Peter Savolainen, was that the major ‘domestication’ event happened only once, in East Asia.2 This was the conclusion of an mDNA study of 654 dogs from different regions of the world where the focus was on the diversity of sequences. The perfectly sensible rationale was that the highest diversity, that is the highest number of different mDNA lineages, would be found in the places where dogs had been around the longest and had the most time to accumulate new mutations, rather like the islanders in our metaphorical example. Savolainen’s team found mDNA sequence diversity was highest in south-east Asia and located the first ‘domestication’ to the region. This was a very controversial conclusion at the time, and it would be another decade before the debate was settled, although it still rumbles on in some quarters.3