I think virtually everyone is used to regard his pet dog as a domesticated wolf. And the majority of scientists does so as well, as the “wolf hypothesis” seems to be well-supported by the evidence - seems. But there are in fact several arguments that suggest the wolf hypothesis is not as water proof as widely assumed. Some authors propose the domestic dog is in fact a descendant of a now extinct wild canid species, a wild-type Canis familiaris. Perhaps we might have to change our view on the origin of our “best friends”, or at least get more open-minded for alternative hypotheses. I came across this interesting re-consideration of dog domestication thanks to the user Joao Ferro.
English Wikipedia on the origin of the domestic dog states: “Some authors wrote as if domestic dogs were descended from a species of now-extinct wild dog distinct from wolves [...] but that has been disproved.“ This is the only sentence treating this alternate hypothesis, and it is anything but disproved. Koler-Matznick  suggests that the actual ancestor of the domestic dog might have been a now-extinct medium-sized generalist canid, comparable to the Golden jackal or coyote. In this article I am going to give an overview over the arguments and evidence pro and contra both the wolf hypothesis and this alternative hypothesis, some of these come very recent studies. Let’s go over it step by step.
Characterising differences between the dog and the wolf are cranial features such as the shortened rostrum, elongated frontals, broader skulls and smaller teeth, smaller body size and floppy ears . Assuming that the dog descended from the wolf, these features can be explained with a pleiotropic effect because of selection for tame behaviour and relaxed selection, the way we see it in other domesticated species . Apart from that, there is considerable overlap between dogs and wolves in respect to these features. Interestingly, a 1986 study comparing the skull morphology of dog skulls which do not show a considerable relative shortening with wild canid species showed that the cranium of the dog is in fact more basal compared to the specialized wolf and similar to those of jackals, coyotes or the dole. This is not only not in accordance with the explanation with neoteny but also not with the wolf hypothesis. A study from 1942 concludes that the shape of the dog’s brain case resembles the coyote closer than the wolf, and a work from 1983 found that cranial and dental data from dogs resembles smaller canids closer than any wolf subspecies . Some dog characters that are diagnostic are not related to paedomorphism, two of them are mandibular . These characters are shared with C. l. chanco and C. l. pallipes , what can either be explained by descent or introgression. Gonzalez 2012 find morphologic similarities between C. l. pallipes and the dog which is not shared by C. l. lupus based on bivariate analyses, and which is probably not the result of hybridization and further recognized cranial similarities between the dingo and Canis lupaster .
Interestingly, the superficial brain morphology of dogs is closest to the coyote, which itself is close to the jackal, and that of the wolf is distinct from that of the those.
So it seems that wolves are specialized canids related to their ecological niche and that dogs share morphologic characters not necessarily connected to paedomorphism that they share with more basal canids . If you wonder why the postcranial skeleton plays little role in these examinations, I think the reason for it is that the postcranial skeleton shows way less variation within Canis and that most of the differences are a product to allometry, but that’s my personal guess.
You might wonder why I didn’t cover canid fossils yet. The problem with fossils is that we are talking about only very scanty and dubious data here with taxonomic uncertainties . Early domestic dogs might have resembled their ancestors to a large extent so that they are not distinguishable, or putative domestic dog skulls might actually be single exceptional wolf skulls . And if canid material definitely shows signs of domestication, it is still uncertain if they belong to the modern lineage of dogs or represent a now extinct, independently domesticated lineage . Distinguishing closely related species from limited osteologic samples is highly problematic and questionable, see the controversy around C. variabilis (or C. l. variabilis) which also was purported as possible dog ancestor [1,3].
But what can be said is that 4000 years old pariah dog skulls are virtually identical to modern counterparts, what indicates phenotypic consistency over millennia [1,2,3] and moreover, the oldest European domestic dogs known by the 1940s and 1961 were claimed to resemble the Dingo , although I don’t know what modern literature says in this case. The oldest genetically confirmed domestic dog is from the Altai 33,000 years ago , so we should use this date as a working hypothesis for the first domestic dogs.
Ecologic and behavioural evidence
One of the most interesting aspects of the controversy around the dog’s ancestry is their ecologic niche. A recent paper from 2013 showed that dogs have genes that enable them to digest higher amounts of starch in their diet than wolves do . This can be either an effect of domestication, as the authors assume, or an ancestral trait inherited from an ancestor with an ecologic niche different from wolves. We can’t say more yet.
Darren Naish, author of the marvellous blog Tetrapod Zoology, writes in a post on dog origins from 2006 that domestic mammals usually “revert back to wildtype after being feral for a few generations” . While this statement is surely very simplified and in my opinion not true in all respects, but it is certainly correct that most feral domestic animals occupy the same ecologic niche and show a behaviour very similar to that of their wild type. These behavioural aspects include social structure, predation or defence against predators or interaction with man. Feral dogs are opportunistic, voluntary human commensals, just like other generalist canids such as foxes, coyotes and jackals, which readily reproduce in human vicinity . The wolf, however, never is a voluntary human commensal and avoids contact with humans . Furthermore, wolves have a very broad prey spectrum, varying from animals the size of a hare to that of a bison. The fact that they are the largest living canids is certainly helpful for this purpose. Feral dogs on the other hand are rather small compared to most wolves (10-20 kg) and comparable, once again, with other generalist canids . Despite domestic dogs show a considerable variation regarding body size, there is no feral dog population I know which reaches the average size of wolves. As far as I know, feral dogs are hypo- to mesocarnivores that usually prey on animals smaller or as large as themselves and do not take down such large animals as wolves do. (The dingo might be an exception, perhaps as an adaption to being the only apex predator on the Australian mainland since their arrival, but this is my personal speculation.)
The social structure of feral dogs is flexible, less complex and less hierarchic than that of wolves, and they are not as gregarious . One could argue that this is the result of altered brain structures as a result of domestication, but once again, their social behaviour is congruent with that of smaller, less-specialized canids than the wolf .
The most remarkable dogs regarding their vocalization are the dingo-type New guinea singing dogs. They howl at high frequency in bird-like manner. This “singing” is not known for any other living canid (different from the dingo and very different from wolves), although similar sounds have been described for a dhole at the Moscow zoo .
The contribution of female coyotes or jackals to dog domestication can be ruled out based on all genetic studies carried out so far. As far as I know, the dhole and C. simensis were not tested yet, and the contribution of male members of other species cannot be ruled out by mtDNA . The wolf proved to be the genetically closest living canid to the dog, but this alone does not confirm it as its ancestor. A direct proof would be if the domestic dog clustered within modern wolves. Molecular clock analyses based on mtDNA implicate that wolf and dog separated about 76,000 to 136,000 years before present . A study from 1997 showed that some of the domestic dog’s mitochondrial haplotypes intergraded with wolves, indicating an ancestral relationship . Another study from the same year found wolves and dogs to be sister clades, sharing only one of 52 haplotypes, but this does not rule out that both descended from ancient members of C. lupus now extinct.
A paper from 2007  states that genetic studies found C. l. chanco and C. l. pallipes to be actually outside the dog + Holarctic wolf clade and that they diverged much earlier, 800.000 and 400.000 years BP. This indicates that the domestic dog is indeed nested within the wolf. A study published this year  confirmed the sister clade relationship between wolves and dogs again, but this study did not include pallipes and chanco in its phylogenetic analysis. The assumption that dogs were domesticated from wolves thus seems confirmed, but read on.
Another interesting fact: if dogs are domestic wolves, it should be assumed that feral dogs and wolves readily interbreed in the wild and create a panmictic population , like it is the case in other domestic species and, noticeable, in the Australian dingo along with abandoned domestic dogs. Although wolf-dog hybridizations are known to occur, the shared number of mtDNA haplotypes is very small according o Vila et al. 1997 , and Indian wolves and pariah apparently do not often hybridize in the wild . Vila et al further write: “The behavioral and physiological differences between domestic dogs and gray wolves may be sufficiently great such that mating is unlikely and hybrid offspring rarely survive to reproduce in the wild.“ Furthermore, Gonzales 2012 states that there are examples of hybrids in the wild but these are not common and populations are not affected by it . However, he also writes: “Some dog and wolf populations from South Asia, the Middle East and North America appear to be inextricably linked, by a long chain of hybridization events, which may have contributed greatly to the current biological identity of both groups“. Introgression into wild wolves by domestic dogs is likely also for North American populations , and abandoned dogs and wolves seemingly also intermix in Eastern Europe [6 (yeah, not the best reference, but I also heard on other sources that hybrids are common in these areas)]. Also Freedman et al. 2014 suggest post-divergence gene flow between wolves and dogs, and that both underwent severe bottlenecks after their divergence . The ability to produce fertile hybrid offspring is the main character of the “biologic”* species concept , but I think we should remember that wolves and coyotes also hybridize in the wild and yet they are still considered separate species based on other biologic criteria.
Koler-Matznik 2002 writes: “The question of how humans created a domesticated wolf between 15,000 YBP, the oldest estimate from the fossil record for DD (Clutton-Brock 1995), and 135,000 YBP, the highest estimate for separation of DD/wolf from mtDNA (Vilà et al. 1997), is rarely addressed“ . I think this is not an inconsistency. The moment of separation of the two clades (domestic dog + modern wolves) does not implicate that domesticated dogs themselves are of that age, just that the clade they belong to separated from the modern wolf clade about 100.000 years ago.
* I have always been thinking the term “biologic” species concept is rather awkward. All species concepts are biologic. “Genetic species concept” might be a better and more definite term.
Behavioural suitability for domestication and purpose of domestication
Gregariousness often is considered a prerequisite for why the domestication of the dog was so successful. However, Charles Hamilton Smith (this man strangely keeps showing up on my blog) reported the case of an Andean fox, which belongs to a solitary species, tamed by South American people and used as hunting aid. Therefore the ancestor of the domestic dog was not necessarily gregarious.
There are several hypotheses that try to explain why and how humans domesticated the wolf, assuming it is the ancestor of the dog. One that is quite popular is that they were used as a hunting aid. There are several problems with this hypothesis. Wolves are chase hunters of medium-sized to large prey while early human hunters were “ambush predators”, so the wolves would have chased their prey away . Furthermore, wolves are strongly food-possessive, so trying to secure the kill would be a dangerous game .
Pariah dogs and smaller canids may be a more conclusive model for canids as hunting aid. Aboriginal people prevent their tamed dingoes from following when hunting kangoroos because they would chase them off, but dingoes are very useful to locate smaller prey either on the ground or in trees . Also the Basenji hunts alongside humans. Furthermore, early humans could have been kleptoparasitic by following inoffensive small to medium-sized canids and stealing off their prey, as still some people do with pariah dogs and also the dhole today . This could be a reasonable reason for early humans to raise and keep these canids and use them for hunting purposes.
Another hypothesis claims that wolves might have served as guards; this is unlikely considering that all canids make poor guards and even abandon their offspring if their personal life is threatened . Furthermore, it is not unlikely that wolves considered Pleistocene humans as prey. Indian wolves are known to prey upon children, and before long-range weapons and persecution they would have been bolder . It is known that wolves do attack people, even adult ones and especially in the past, and a predatory motivation is one among many .
Many authors propose that wolves domesticated themselves by voluntary becoming a human commensal, feeding of camp refuse and leftovers . According to them, this different ecological niche separated these commensal wolves from their conspecific, so that they went different evolutional ways, one leading to the domestic dog . Those wolves would have become increasingly tame and humans started to breed them for whatever purpose. This hypothesis is problematic because early nomadic hunter-gatherers probably used as much nutritious portions of a carcass as possible, like modern hunter-gatherers do, so that it is unlikely that there was enough nutrition to support a reproducing population of wolves , and it is very dubious why a large, specialized big game hunting predator would niche shift to scavenger of scarce refuse . A generalized, smaller canid in the pariah niche is more likely to become a human commensal .
Summary and conclusion
Let’s put it all together: the dog is not phenotypically closer to the wolf than to other canids, actually it has features that cannot be put aside as results of neoteny but likely are plesiomorphic and similar to other, less derived canid species, suggesting that the wolf is a more specialized canid than the other species, including the dog. This is also the case in the brain structure of the dog and the social structure of feral dogs, which is more flexible, less gregarious and less hierarchical than that of wolves, like in other smaller and less-specialized canids. Dogs are adapted to a more starch-rich diet than wolves, this is either an adaption to their domestic niche or a trait inherited from their wild ancestors. Generalist canids and feral dogs are opportunistic, voluntary human commensals that reproduce in humans neighbourhood, while wolves never do so voluntarily. The former usually are meso- to hypocarnivores, while wolves, being the largest extant canid, regularly take down big game. The vocalization of the New guinea singing dog is exceptional among living canids, it might or might not be an ancestral trait (far-fetched because no other pariah dog displays it).
Genetically, modern wolves and dogs are close relatives, but they share only few haplotypes and seem to be sister clades and diverged about 100.000 years ago. C. l. chanco and C. l. pallipes seemingly are considerably more basal, being outside the northern wolves + dogs clade and diverged about 400.000 years before present. Therefore, dogs cluster within C. lupus genetically. Hybridizations occur in the wild, sometimes also leaving tracks in each other’s populations, but this occurs also between other canid species.
It is unlikely that wolves are suitable as a hunting aid. Paria dogs and other less-specialized canids are more suitable for this purpose. It is also unlikely that large predators like wolves voluntarily became human commensals feeding on scarce refuse (if there was any) of hunter-gatherer camps, while many less-specialized canids already are. Any canids make poor guards and it is possible that wolves considered early humans prey.
So, now my personal interpretation of all this. Behavioural, morphological and ecological data seems to contradict the wolf hypothesis, or at least makes it unlikely. Genetic data, on the other hand, seems to strongly confirm it (unless you consider chanco and pallipes separate species). How does that fit together? I discovered something that seems to be overlooked in this discussion. Chanco and pallipes are not only more basal regarding their phylogenetics, but also regarding their ecology and behaviour. They are less gregarious and their social structure is less hierarchical and very similar to dingoes, they are hypo- to mesopredators and take down larger prey on occasion, just like pariah dogs. They are, simply spoken, less specialized and more plesiomorphic, like pariah dogs and unlike the northern wolf clade. Parsimony thus dictates that the common ancestor of the northern wolf + dog clade was so as well. It is also interesting to note that the coat of both chanco and pallipes is lighter in colour on average than that of northern wolves, having a sand-coloured to reddish tint, slightly resembling dingo-type pariah dogs. Therefore I consider it likely that domestic dogs descend from basal, plesiomorphic wolves less specialized than modern northern wolves, what also was also concluded by the authors on recent papers on wolf/dog genetics. However, I take one speculative step and state it might be possible that within the approximately 100.000 years between the split from primitive wolves, the familiaris-lineage might have undergone evolution on species level (therefore, wild-type Canis familiaris) to a niche and phenotype similar to dingo-type pariah dogs, since the colour of their coat is an effective camouflage in their environment and it would explain the almost uniform appearance of dingo-type pariah dogs, ranging from Israel (Canaan Dog) over South East Asia, Australia to North America (Carolina dog). However, that’s just a guess of mine. But regarding the species status of domestic dogs, I think it is premature to label them under Canis lupus, especially since there is no general opinion regarding the species status of chanco and pallipes either.
By the way, that being said does not mean that northern wolves played absolutely no role in the domestication of the dog. The use of wolf-dog hybrids in the creation of certain breeds, such as the German Shepherd, is documented. Genetically, such dog breeds show a stronger relationship to wolves than the main dog group consisting of a large part of the domestic dog, including dingo-type pariah dogs as well as popular, highly derived breeds like the collie and retriever .
 Koler-Matznick, 2002: The Origin of the dog revisited.
 Primitive dogs
 Erik Axelsson et al.: The genomic signature of dog domestication reveals adaption to a starch-rich diet. 2013.
 Vila, C et al.: "Multiple and ancient origins of the domestic dog". Science 276 (5319): 1687–9. doi:10.1126/science.276.5319.1687. PMID 9180076. Archived from the original on 2007-09-26.
 Shaun Ellis: The man who lives with wolves. 2010.
 Kohler-Matznick et al.: An updated description of the New Guinea singing dog (Canis hallstromi, Troughton 1957). 2003
 Freedman et al. 2014: Genome Sequencing highlights the dynamic early history of domestic dogs.
 Druzhkova et al. 2013: Ancient DNA analysis affirms the Canid from Altai as a primitive dog
 Aggarwal et al. Mitochondrial DNA coding region sequences support the phylogenetic distinction of two indian wolf species. 2007.