Biology Reference
In-Depth Information
and Cheyenne River form a loose cluster that reflects geographic proximity, common
temporal period, and similar facial shape. Interestingly, Leavenworth is positioned between
Larson and Cheyenne River, which corresponds to hypotheses about Leavenworth being an
aggregate village for both Bad River peoples similar to those who occupied Cheyenne River,
and Le Beau peoples similar to those who occupied Larson after significant depopulation
primarily due to infectious diseases. Morphologically Leavenworth is also intermediate
between the patterns seen in Larson and Cheyenne River.
The Mahalanobis distances between site components indicate that based on facial
morphology almost all the components are significantly different from one another. The
few exceptions are the Cheyenne River, Mobridge Feature 2, and Nordvold, which formed
a loose cluster on the CV plot. One can visually appreciate the similarities in average facial
shape. The other exception is the lack of significant difference between Rygh and Mobridge
Feature 2, which cluster together well away from all other sites in the CV plot and share
a similar average facial shape.
Both
McKeown (2000)
and
McKeown and Jantz (2005)
present results for these sites using
landmark coordinates for the face and vault. These studies used Mantel tests (
Mantel, 1967;
Manly, 1986; Smouse et al., 1986
) to compute the correlation between the biological distance
matrix (based on theMahalanobis distances), the temporal distancematrix (calculated as differ-
ences in mean site dates), and the geographic distance matrix (calculated as linear distance
between site locations). In both studies, the correlation between biological and geographic
distancematriceswas positive and statistically significant (
p
0.01). Conversely, the correlation
between biological and temporal distance matrices was not statistically significant. This is
consistent with a model of isolation by geographic distance where sites that are biologically
more similar to each other are also geographically closer and vice versa (
Konigsberg, 1990
).
This short case study demonstrates that geometric morphometrics methods are well suited
to biological distance studies as the coordinate data provide greater information about the
range and pattern of morphological variation in an anatomical unit such as the face. By
capturing three-dimensional variation in landmark location (i.e., the geometry of the
form), it increases the level of morphological variation available for statistical analysis.
Output from geometric morphometric techniques such as GPA can be analyzed by standard
statistical procedures and employed in tests of models regarding reasons for certain patterns
of morphological variation. The graphical depictions of morphological variation add consid-
erably to our understanding of shape variation among groups as we can visually interpret the
differences based on variation in landmark location. This takes us beyond interpreting vari-
ation simply as “faces becoming taller through time” and provides the tools for describing
the exact change in morphology (e.g., nasion is more superiorly located or prosthion is
more inferiorly located or both) that produces that overall difference in shape.
<
Thin-Plate Spline Methods
The thin-plate spline (TPS) (
Bookstein, 1991
) provides a sophisticated approach to visu-
alizing differences between coordinate-based configurations. In particular, it can be used
to depict the shape variation between two group mean configurations in terms of the shape
deformation necessary for one mean configuration to be mapped directly onto the other
mean configuration. This is accomplished with an algorithm to exactly map the homologous
