Morphometric analysis of humerus and femur shape in Morrison sauropods: implications for functional morphology and paleobiology

Paleobiology, Summer 2004 by Bonnan, Matthew F

Humerus Results.-Nine incomplete specimens were excluded from the PCA. Variable loadings suggest PC 2 is a component of olecranon fossa breadth (H9), PC 3 is anteroposterior midshaft width (H4), and PC 4 is a contrast of lateral condyle breadth (H7) with olecranon fossa height (H10) (Table 1). Together, these three PCs account for less than 6% of the total sample variation. All PCA plots show little else but size-related distributions: specimens plot in order of size along PC 1, and the four Apatosaurus specimens clustered at the negative end of the PC 1 axis are juveniles. No significant differences were reported among the three genera on any of the variables using the KW test. Few distinct allometric trends were present in most bivariate plots, and regression of humerus variables against maximum length consistently produced nearly isometric slopes not significantly different from one for most variables (Table 3). Certain humeral dimensions have moderately negative allometric slopes (b^sub 1^ [congruent with] 0.6354-0.8420) in Camarasaurus (proximal and distal breadth [H2, H6], mediolateral midshaft width [H5], medial "condyle" breadth [H8], and olecranon fossa dimensions [H9, H10]). These results suggest that mediolateral and olecranon fossa expansion lag slightly behind an increase in humerus length without affecting its overall isometric trend. Linear regression results show a great deal of individual variation for Diplodocus humeri. Several variables (anteroposterior midshaft width, condyle breadths, olecranon fossa dimensions) have low r^sup 2^ values and slopes that are not significantly different from zero (see Table 3). These variables show no discernible allometric or isometric trends for Diplodocus, but the very small sample size of Diplodocus humeri may have a substantial effect on these results. In the plot of mediolateral versus anteroposterior midshaft widths (H4, H5), Diplodocus humeri appear to expand more mediolaterally than anteroposteriorly as size increases, but it is unclear if this trend is a real biological signal or the re-sult of a weak allometric trend for anteroposterior midshaft width. As with the PCA plots, the four Apatosaurus specimens that cluster at the left of each bivariate plot are juveniles.

Femur Results.-Data were generally absent for three variables in all genera, but especially in the diplodocids (Apatosaurus, Diplodocus): the crural extensor fossa breadth (F8), the crural extensor fossa height (F9), and the anteroposterior midshaft width (F12). Many of the diplodocid femora examined were stored, mounted, or damaged in such a way that measurement of their anterior surfaces was difficult or impossible. The missing data for the crural extensor fossa variables (F8, F9) alone would exclude 18 specimens from the PCA, severely truncating the large sample size and reducing the number of Apatosaurus (n = 6) and Diplodocus (n = 6) femora available for analysis. The truncation of diplodocid specimens over the three variables is worse, resulting in the retention of a single Apatosaurus femur for the PCA. Therefore, two PCAs were run. In the first PCA, the three problematic variables were deleted. In the second PCA, the missing values were estimated by substitution of means in the SPSS program (see Strauss et al. 2003: p. 289 for a recent review of this technique). Although this method is far from ideal because it tends to decrease variance and covariance estimates (Strauss et al. 2003), comparison of these results with those in the first PCA revealed no discernible differences, and both the PCA plots and their loadings were very similar. Therefore, to analyze and plot as many specimens as possible, the results of the second PCA were used and reported here. In subsequent KW tests and bivariate plots of the femora, the patterns displayed in the PCA plots closely adhere to the patterns observed in the unmodified data.