The domestic dog, Canis familiaris, presents a unique opportunity to identify and study the relationship between genotype and phenotype. Over the approximately 15,000 years since its domestication from the gray wolf, Canis lupus, the dog has undergone intense artificial selection for a variety of functional and aesthetic forms, resulting in hundreds of modern breeds that exhibit a wide range of behavior and morphology. Restrictive breeding histories have rendered each breed as a distinct genetic unit, facilitating the genetic mapping of breed-specific phenotypes.

Brachycephaly, or a short, wide head, is a phenotype that is observed across a variety of breeds. However, brachycephaly has not been consistently defined or quantified and it remains unclear whether different forms of brachycephaly exist across breeds. The inclusion of brachycephalic breeds within two genetically distant breed groups, the toy dogs and the Mastiff-like dogs, supports the possibility that distinctions within the brachycephalic category exist. In addition, mouse developmental gene expression studies demonstrate that multiple genetic pathways can be manipulated to produce a brachycephalic mouse, suggesting that distinctions between brachycephalic dog breeds may correspond to different genetic mechanisms.

The objectives of this dissertation are to identify patterns of morphological distinction between brachycephalic dog breeds and to draw connections between these patterns and genetic relationships among breeds and craniofacial developmental genetics in mice. The following three research questions are addressed:

(1) Are there significant distinctions in the patterns of shape differences between the crania of various brachycephalic breeds and the ancestral gray wolf cranium?

(2) Do any distinctions in patterns of shape differences correspond to genetic relationships between breeds?

(3) Are these patterns of shape differences comparable to genetic pathways identified in mouse developmental genetic studies?

Three-dimensional landmark coordinates representing craniofacial shape variation were collected from 527 adult dog crania, representing sixty-nine breeds. Dog breed crania were compared to a sample of 120 adult gray wolf crania from Alaska. Comparison of average cranial shape between groups was performed using a geometric morphometric approach, wherein landmark coordinate configurations were superimposed to remove the effects of size and orientation. Principal component and discriminant function analyses were implemented to describe the major axes of variation within each group and to identify the shape differences from the gray wolf characterizing each breed.

Three distinct patterns of canine brachycephaly were observed across breeds. First, the Boxer and Bulldog breeds share a pattern of facial shortening, which, relative to the gray wolf, includes a more rostro-dorsal position of the frontal bones and a reduction in length and a dorsal tilt of the rostral nasal bones. The short snout of these breeds is also proportionally wider. Second, the Pug, Pekingese, and French Bulldog breeds exhibit a pattern that is distinct from the Boxer and Bulldog. Relative to the gray wolf, these breeds display an even greater degree of reduction in the length of the rostral-most snout elements than observed in the Boxer and Bulldog. This reduction is combined with extremely dished nasal bones at the midface. Finally, the Chihuahua presents a third form of brachycephaly in which, relative to the gray wolf, the rostral snout is neither tilted dorsally nor proportionally increased in width.

These patterns do not correspond to recently determined genetic relationships between breeds. The presence of different forms of brachycephaly within the toy dog and Mastiff-like dog groups suggests that the overall genetic similarity between breeds may not reflect shared genetic mechanisms for individual traits that are part of complex breeding histories.

Three candidate genes for canine brachycephaly were identified from the mouse developmental genetic literature. Mouse mutant phenotypes for each gene were compared with the patterns of brachycephaly observed in dog breeds. Mouse haploinsufficient for the gene Tcof1 exhibit brachycephalic features resembling the Pug/Pekingese/French Bulldog pattern, whereas mice lacking Msx1 and Msx2 alleles present a phenotype most similar to the Boxer/Bulldog pattern. Finally, loss- and gain-of-function mutations in the gene Fgfr2c in mice produce a form of brachycephaly that parallels that seen in the Chihuahua.

By quantifying variation in the brachycephalic phenotype and identifying candidate genes that may underlie this variation, these findings provide direction for future studies of genetic association. This insight into the relationship between genetic variation and craniofacial phenotypic variation in the dog can also be applied to our understanding of the processes of natural selection that have produced brachycephalic forms in the canid fossil record, as well as throughout vertebrate evolution.