Leaf Age Category Distribution
Functional Group Distribution
Cell Means - Mite Counts
ANOVA
Multidimensional scaling is a statistical technique that identifies patterns in community composition. Given different taxa or species measured at different sites, MDS can determine which sites show similar patterns of abundance. Computationally, MDS measures community similarity as a distance score -- sites with similar community compositions will be close together on this metric; sites with dissimilar community compositions will be far apart. These distances are plotted on a 2-dimensional surface like an ordinary road map -- similar sites are near each other, dissimilar sites are far apart. In the graph on the left below, each point represents one leaf (one site).
The display of sites is not, in itself, very informative. However, if we augment the graph, we can see the effect of our grouping variable (leaf age in this example).
To see the effect of leaf age on community composition, we colour each point in the plot (each leaf) according to its level on the Leaf Age categorical variable, as shown in the graph on the right. Remember that leaves with similar community composition will be close together on the MDS plot. Thus, if community composition varies according to leaf age, points of the same age (and hence same colour) will cluster together -- we will see distinct blocks of colour in the plot. If community composition does not depend on leaf age, the colours will appear to be mixed (distributed evenly) across the plot.
We can illustrate the variability of each level of the grouping variable by computing its centroid -- the average x,y position of all sites in the group. We can draw an ellipse around the centroid defined by the standard deviations of the x and y locations of all points in the group.
If community composition within a group (for example Young Leaves) is very homogeneous, the ellipse for that group will be small. The more variability in the group's community composition, the larger the associated ellipse.
Finally, we can illustrate the relationships between taxa by adding their centroids (as computed by the MDS analysis) to the plot. Taxa that tended to occur together in the samples will be close together on the plot.
Age category (sd around centroid)