The shape of speed: 3D geometric morphometrics of the humerus predicts maximum running speed in canids (Carnivora: Canidae)

Abstract Estimates of running speed are useful for many kinds of palaeontological reconstructions, including the co-evolution of predators and prey, and inferring the hunting strategies of extinct species; however, previous studies have failed to find a strong relationship between skeletal morphology and speed. This study capitalizes on the high degree of variation in morphology and functional ability across domestic dog breeds to show that the shape of the forelimb bone provides accurate estimates of running speed in Canidae. To accomplish this, 3D geometric morphometrics of the humerus was combined with the exceptional historical records from competitive dog races and used to quantify the relationship between humerus shape and relative maximum running speed. Selective breeding of dogs has pushed the morphological variation in dog breeds to the extremes, creating breeds with a variety of humeral shapes and a wide range of maximum running speeds from the basset hound at 34 km/h to the greyhound at 65 km/h. For each breed, maximum running speed was determined using records from the American Kennel Club’s timed 100-yard sprint competition. Speeds were normalized by calculating their Froude number and then regressed onto shoulder height in order to obtain an accurate metric of breeds that are fast for their size. Results indicate that there is a strong and significant relationship between the maximum relative speed of the breed and shape of the humerus driven by a combination of the shape of the distal articulation and bone robusticity (R2 = 0.42, P < .001). Tests of this dog-based predictive equation on wild canids have relatively low error (%Prediction Error = 10%, %Standard Error of the Estimate = 13%) and inclusion of wild canids in the model reduces predictive error even further (%PE = 8%, %SEE = 10%), supporting the use of these methods to estimate locomotor performance in fossil canids. This method is applied to estimate maximum running speed in four fossil canids: the dire wolf, Aenocyon dirus; the fossil coyote, Canis latrans; Hesperocyon gregarius, and Phlaocyon multicuspus.