370 University of California Publications in Geological Sciences 



Paleontology can with increasing fidelity trace the intricate ramifications of the 

 evolutionary course of many organic groups; many existing hiatuses in the fossil 

 record can be spanned confidently because of their smallness in comparison with 

 the period of time involved. These hiatuses, however, impede the observation of the 

 results of evolutionary mechanisms as shown in animal remains. Simpson (1953) 

 emphasizes the minuteness of the change undergone by the critical unit, the 

 breeding population; the small selection pressure exerted through minor differ- 

 ences in relative fertility; and the very moderate degree to which mutations 

 augment genetic variability, as characteristic features utilized in the synthetic 

 approach. But the chance factors affecting fossil preservation generally lead to 

 small, eclectic, discontinuous samples that demonstrate only gross changes. 



There are, however, some unusual fossil accumulations that permit analyses to 

 take into account at least some of the variability characteristic of natural popula- 

 tions. The Rancho La Brea vertebrate fauna provides one of the best available 

 series of mammalian fossils. Among the different groups of mammals found in 

 these tar-impregnated deposits are certain canids whose taxonomic assignment 

 has been somewhat difficult. Coyotelike dogs are given special attention in this 

 paper. The abundant wolflike forms, which have previously been subjected to 

 statistical analysis (Nigra and Lance, 1947), and the gray fox will not be consid- 

 ered. 



METHOD 



Influences from genetic and zoological research as well as evaluations of their own 

 material have prompted paleontologists to greater concern for the nature and 

 characteristics of the populations of which their fossil specimens are generally very 

 small samples. This interest can lead to a consideration of possible ways of deter- 

 mining the extent and expression of inter- and intrapopulation variation and the 

 means whereby it may be taken into account. Methods of elucidating the relation- 

 ships among populations, the units of evolution, are of prime importance. Multi- 

 variate analysis provides one mode of attack on these problems. 



The biological significance of linear measurements from a specimen with irreg- 

 ular morphology may be open to some question, but it is standard practice to use 

 them in taxonomic studies — either actual dimensions or merely their relative pro- 

 portions. The measurements taken of the coyotes are of the sort used in such 

 studies. The proposed taxonomic distinctions among the canids considered here 

 have been almost uniformly based on quantitative differences, as is indicated 

 throughout. 



P. C. Mahalanobis (e.g., Mahalanobis, Majumdar, and Rao, 1949) has developed 

 a technique by which a number of variables may all be taken into account at the 

 same time when several groups are being analyzed together. Consider a problem in- 

 volving four closely related groups or populations that are all suspected of differing 

 from one another, yet may not differ from each other to the same degree. If evo- 

 lutionary significance is placed on differences in over-all morphology, a measure of 

 the degree of difference, or distance, between each population and each of the 

 others would be desirable. Thus, with populations P 1? P 2 , P 3 , and P 4 , some way of 

 reckoning the distances D 12 , D 13 , D 14 , D 23 , D 24 , and D 34 should be sought. These 

 distances, in numbers, might make possible a statement about intergroup relations, 



