Multivariate analysis of variance (MANOVA)
and canonical variates analysis (CVA) of the resulting uniform
and non-uniform shape scores confirmed the significant
discrimination between Pliocene and post-Pliocene samples (Wilks'
lambda, P < 0.01; Fig. 6).
Only the first canonical factor discriminated significantly among
the samples.
The morphological bases for the
discrimination can be visualized with thin-plate splines of the
transformations (Fig. 7).
The uniform analyses illustrate C. erosa as a
sheared transformation, compared to C. cancellata.
This is reflective of the observed greater
circularity of the C. cancellata valve outline, and
the relatively longer escutcheon of C. erosa. The
non-uniform splines indicate that the major difference between
these two species are the relative placements of the anterior
cardinal tooth and the dorsal tip of the anterior adductor muscle
scar (landmarks 3 and 5 respectively) that results in C. erosa
having a broader and deeper hinge relative to C. cancellata.
Also apparent is the relatively shorter lunule of C. cancellata
(visualized as the distance between landmarks 1 and 2), as noted
in previous studies (Dall 1903; Roopnarine 1995)
Naticid drill hole location was found to be
selective and non-random within all samples (chi2
goodness-of-fit test; P < 0.05 for all samples), and the mean and
variance of hole location varied significantly among samples. The
mean location of drilling was the same for the Pliocene Pinecrest
Beds and Caloosahatchee Fm., the Lower Pleistocene Bermont Fm.
and the Recent, but differed significantly for the Middle
Pleistocene Ft. Thompson Fm. (likelihood ratio test of sample
means [Manly 1986]; P < 0.001).The mean location of drilling in the
latter sample is on the dorso-posterior region of the valve,
whereas it is situated close to the ventral end of the posterior
adductor muscle in the other samples (Fig. 5). The most obvious
difference however between Pliocene and post-Pliocene valves, is
found in the variance of drill hole location. 
There is a significantly smaller spatial
distribution of drill holes in samples from the Bermont and Ft.
Thompson formations when compared to those from the Pliocene
Pinecrest Beds and the Caloosahatchee Fm. (likelihood ratio test
of deviations from the mean; P < 0.001), but the Recent sample exhibits as
much variation as pre-extinction samples (Fig. 8). This can be explained by noting that in
addition to being drilled in the ``typical'' posterior adductor
region, Recent specimens are also drilled frequently on the
dorso-posterior region, typical of Middle Pleistocene specimens
from the Ft. Thompson Fm.
Model II regression analysis of the
relationship between drill hole size (estimate of predator size)
and prey size indicates a change after the end of the Pliocene.
The relationship between these two parameters is statistically
significant before the Pleistocene, but only marginally
significant during Bermont time (Fig. 9). Naticids of all sizes seem to prey upon Chione
in a medium size range. 
By the Middle Pleistocene (Ft. Thompson Fm.)
however, prey selection is again stereotyped by prey size.
Reduction in stereotypy of prey size selection is predicted by
the Kelley-Hansen Hypothesis.
There is no corresponding change in valve thickness. In fact, individuals of C. cancellata from the Bermont Fm. were of almost identical thickness to the Pliocene C. erosa (Figure 10). During the Pleistocene however, valve thickness of C. cancellata increased significantly, and remains significantly greater today.
