RESULTS .1 Repeatability

Results of the Procrustes distance analyses on the repeated measurements, Figure 3.2 A, show that they did not fall within the range of the paired comparisons which

31 indicates that the degree of measurement error is well below the 5% requirement. Figure 3.2 B shows the effect of the mitigation of allometry on the results. A minimal reduction of overall variability is seen, but the degree of measurement error remains unchanged.

Hence, measurement error was deemed within acceptable limits. These results are visually depicted in Figure 3.3 in which smaller ellipses are indicative of lower degrees of error as compared with larger ellipses. Furthermore, circular ellipses illustrate less directional error than do the more elliptically shaped ones.

Despite being within the 5% acceptable limit, Figure 3.3 shows that the largest amount of measurement error existed along the orbital margins and the outer aspect of the maxillary alveolar margin. Despite the fact that Procrustes superimposition averages out disparities between landmarks, this method highlights potentially problematic landmarks as there are many well-defined highly repeatable landmarks to “anchor” the configurations.

3.3.2 Global Analyses

A significant covariation was observed between tooth loss and cranial shape (RV

= 0.373, p < 0.001), and a two-block PLS analysis was used to elucidate the association between variables (Figure 3.4).

The first PLS vector accounted for 72.01% of the total covariation seen between dental loss and cranial morphology. Similarly factors two, three and four accounted for 4.66%, 3.59% and 2.76% of the covariation in the sample respectively. In Figure 3.4, all edentulous individuals cluster at the bottom of the graph for factor one, independent of skull shape, whereas the remaining vectors only account for a tiny fraction of the covariation and do not align with the order of tooth loss. The statistical program R was used to construct vector diagrams (Figure 3.5 a3, i3 and l3) and polygon representations with a colour spectrum illustrating relative changes of polygon area (Figure 3.5 a-l 1 and 2). The use of both image types facilitates an understanding of the changes taking place in each structure as the vector diagrams show how, and in which direction, each structure is changing. The polygon images also give an indication as to the localised relative shape changes using a colour gradient.

32 In the polygon representations of morphological change (Figure 3.5 a-l 1 and 2), two images are presented. One represents completely edentulous individuals and the other representing dentate individuals. Polygons that experience insignificant change in an area are green in colour, whereas spectra tending toward red and blue are indicative of increasing and decreasing relative polygon area, respectively. The colour red itself indicates a two-fold increase in area, whilst blue represents a halving in area. It is unsurprising that these images indicate that the greatest change associated with tooth loss is the marked recession of the alveolar bone of the maxilla, which seems somewhat localised in the anterior third of this structure. This severe resorption of bone leads to a marked decrease in facial height and prognathism. There is also a significant decrease in the length of the bony bar that extends from the approximate position of the canine jugum to the infraorbital margin near zygomaxillare orbitale (zmo). This is likely to be a result of the decrease in prognathism due to alveolar resorption. As the resorption of bone is not limited to the labial and buccal surfaces of the maxilla, but also occurs on the lingual surface, there is a noticeable increase in the width of the palate, making it appear more hyperbolic.

In the vector diagrams (Figure 3.5 a3 i3 and l3) the two extreme morphologies represented in the data set are superimposed as reference and target images and vectors/arrows are used to show the magnitude and direction of the change that occurs between them. In this case the red image is the reference image that represents completely edentulous samples, whilst the blue image is the target image representing the dentate samples. These images also show that the greatest variation is present in the maxilla, but the vectors further highlight the fact that a larger degree of resorption occurs in a posteromedial direction around the labial/buccal surface. A relatively smaller degree of resorption occurs in an anterolateral direction about the lingual/palatal surface.

Additionally, it is apparent that there is an inferior and anterior flexion of the basicranium relative to the viscerocranium such that the external occipital protuberance comes to lie more inferiorly following tooth loss. The mastoid processes appear to move both inferiorly and anteriorly. The relative size of the basicranium also appears marginally larger in the edentulous individuals. From the vector diagrams it also become apparent

33 that the zygomatic bones shift inferiorly and bizygomatic breadth becomes relatively larger with tooth loss. Finally, both the protrusion of the nasal bones and the size of the orbits appear to undergo a relative increase in cases of extensive tooth loss. It should be noted however, that these observations are relative and it is likely that a decrease in the size of the facial skeleton is responsible for these outcomes rather than an absolute increase in the size of the aforementioned structures.

3.3.3 Individual Curve Analyses

When each of the subsets (basicranium, maxillary alveolar ridge, zygomatic arch, nasal aperture and orbital rim) was analysed individually, the alveolus was the only structure significantly affected by tooth loss. This effect proved to be highly significant (p

< 0.001), as expected. The first vector of the two-block PLS accounts for 82.72% of the covariation between dental loss and maxillary shape, followed by only 5.56%, 2.50% and 2.28% being accounted for by the next three vectors. The dramatic reduction in the thickness of the alveolar bone is striking (Figure 3.6, a and b). The anterior third of the maxilla experiences particularly extreme reduction of bone and it appears as though virtually no bone remains. This severe reduction in bone causes an overall relative widening of the bony palate that results in a more hyperbolic shape. The vector diagram in Figure 3.6 (b) clearly illustrates that the loss of bone associated with edentulism is not limited to the labial/buccal surface, but occurs along the lingual surface as well. The depth of the incisive canal, represented by the triangular eminence of bone that appears to extend below the alveolar margin in Figure 3.6 (a to c), is seen to undergo a relative migration anteriorly and reduce in depth in the images representing the edentulous configuration.

3.4 DISCUSSION