The relationship between humans and animals is as old as humanity itself. Dogs are believed to be the first animals domesticated by humans, and much of the wide physical variation that exists in dogs today is thought to result from this long-term relationship. The prevailing narrative is that early dogs were not as diverse as modern dogs and that the diversity of modern dogs resulted from intensive breeding programs in the Victorian era (mid- to late 19th century) that led to the selection of traits such as large round eyes, shortened snouts, and floppy ears. On page 741 of this issue, Evin et al. (1) report that dog populations were already diverse millennia before modern breeding programs.

The process of dog domestication is still under debate. It is currently thought to have occurred multiple times and in multiple places (2), likely from an extinct common ancestor of wolves and dogs (3). Whether an increased proximity between wolves and humans created the conditions for domestication is unclear. Analyses of ancient DNA from around the globe suggest that the first domestic dogs appeared around 11,000 years ago (2, 4), whereas archaeological evidence from Altai, Siberia, indicates that they might have existed as early as 33,000 years ago (5). Carbon-14 is the most common direct method of dating ancient bone, whereas ancient DNA, which is less likely to be preserved, can be used to estimate the age of a specimen on the basis of genetic changes through time. Both methods rely on the preservation of organic material, such as collagen. However, archaeological remains of ancient canids (mammals in the family Canidae that also includes foxes, coyotes, and jackals) are often poorly preserved, fragmentary bones, and direct dating is not possible. In these cases, dating is done indirectly on archaeological finds associated with the specimen. This approach introduces a wide margin of error, which partially explains the discrepancy in estimates of dog domestication timings based on genetic and archaeological evidence. The physical and genetic identification of ancient domestic dogs is also not always straightforward, as they are difficult to differentiate from wolves. Making this distinction has relied on modern datasets of wolves and dogs for morphological comparison and on whether the remains are associated with archaeological evidence of humans.

Evin et al. used three-dimensional (3D) morphometric analysis to compare the physical features of a global sample of 643 canid skulls from the past 50,000 years. This quantitative analytical technique measures fine differences in shape across specimens. A detailed 3D model is created from a specimen using either a laser scanner or photogrammetry, and physical features of interest are identified, measured, and compared across specimens using multivariate statistics. Physical differences between specimens may be markers of domesticated traits, or they may result from other factors, such as changes in diet or environment. Evin et al.'s results indicate that a distinctive dog skull morphology emerged around 11,000 years ago, which aligns with the time frame suggested by genetic analyses of ancient dogs (4). The results also reveal a high amount of diversity among specimens identified as domesticated, challenging the idea that Victorian-era breeding was the sole driver of the immense diversity of modern dogs.

Evin et al.'s analysis has one main limitation. The older a specimen is, the more difficult it is to date it directly and obtain a narrow time frame. The authors used a large, global sample of canids that had been dated by a variety of methods across different laboratories. The uncertainty about the accuracy and comparability of these dates calls into question the apparent alignment of the dog domestication timings obtained from physical and genetic analyses.

Evin et al.'s study raises additional questions for further investigation. Their findings indicate that Pleistocene wolves were morphologically more diverse than modern wolves, although the sample size of Pleistocene wolves was small (17 skulls). If ancient wolves displayed great physical variability, this diversity was probably also present in the first domesticated canids, which are their descendants. If early dogs were already diverse, humans might have had less influence on their evolution than previously thought. Could climate, geography, or resource availability have contributed more than people to the diversification of early domesticated canids?

More research is also needed to understand the evolution of morphological and functional features in specific canid groups. For example, how have wolflike morphological features been retained in some modern dog breeds, such as the German shepherd? Are all dogs the product of human interference? Dingoes, for instance, have a narrower range of physical variability than domesticated dogs and, unlike dogs, cannot digest starch. Ethnographic and archaeological evidence reveals diversity in their relationships with people (6). Studying the evolution of canids such as dingoes may challenge assumptions that dog domestication was wholly human driven.

The domestication of dogs has captivated attention because of the close bonds that many humans share with dogs. Evin et al. prompt a rethink about the timing and development of this relationship. Their research contributes to the wider understanding of domestication as a complex, multifaceted biological and cultural process in which thousands of years of human and animal history are intertwined.