WILKINS ET AL: FLORIDA PANTHER MORPHOLOGY


 CRANIAL PROPORTIONS
 Methods

 Cranial measures provide a suitable method for the study of geographic
variation. According to Goldman (1946), the skull of the Florida panther differs
from that of western subspecies in a number of proportions. Earlier work showed
P. c. coryi could be correctly classified with a high level of confidence when
compared with other North American subspecies using cranial measurements and
statistical methods (Abercrombie 1984, Belden 1986b). Eighteen cranial
measurements were taken on adult specimens of P. c. coryi (n=55), including
historic specimens from Louisiana (n=3) and other North American subspecies
(n=183). Lack of sufficient samples of all South American forms and some North
American subspecies limited the cranial analysis to the use of six subspecies: P. c.
azteca, P. c. californica, P. c. coryi, P. c. hippolestes, P. c. kaibabensis, P. c.
oregonensis. There were too few specimens available of the now-extinct P. c.
cougar, from eastern U.S. to include in this analysis.
 Unfortunately, many P. c. coryi skulls were damaged, the result of having
been shot in the head (older specimens) or hit by cars (recent specimens), resulting
in numerous missing values. In all multivariate SAS statistical procedures,
observations with missing variables will be eliminated from the analysis.
Therefore, even with a sufficient sample of specimens, elimination of damaged
individuals and separate analyses by sex would preclude statistical methodology.
Missing values were replaced with mean values for a particular variable which was
calculated using existing values within the appropriate class (i.e. historic males,
historic females, recent males, etc.). Specimens of unknown sex were classified
with discriminant-function analysis also using specimens of known sex from the
appropriate class (as above) to create a calibration data set. These procedures were
necessary only for the Florida data set, since complete skulls of known-sex
individuals from other subspecies groups were selected for measurement. Skulls of
adult males are not only larger, but more angular and massive. Females are
smaller and have a more smoothly rounded brain case and lesser development of
sagittal and lambdoid crests (Goldman 1946). The two sexes might, therefore, be
described by a different set of variables. Stepwise discriminant analysis selected 11
variables as important for discriminating females and 16 for males.
 Geographic variation was explored using PCA. Because no discrete clusters
were formed, subsequent CDA was conducted to maximize intergroup differences.
Presented here is only one of several analyses conducted, including only three
subspecies: Florida and Louisiana specimens (P. c. coryi), southwestern U.S. (P. c.
azteca), and Texas (P. c. stanleyana), those groups inhabiting the southern part of
the U.S. This was thought to be an appropriate strategy considering the disjunct
distribution of P. c. coryi, the considerable variation exhibited by the subspecies in
North America, and the limited scope of this study. Also, PCA suggested that
clinal variation might be a confounding factor that could not be resolved with the