We accomplish our goals by applying discriminant function analysis and thin plate spline analysis to wing landmarks. We have established a database of expected wing landmark variability for several species and populations of species: (1) Drosophila species, Drosophila melanogaster wild type and selected mutant populations. (2) gypsy moth, Lymantria dispar, populations. (3) Catocaline moth species. (4) Dragonfly (Odonata) species.
For example we have samples of North American, European and Asian gypsy moth populations. A male Gypsy moth forewing of unknown origin is compared to this database and the wing obtains a discriminant score assessing its likelihood of belonging to one of the three origins we can recognise: North America, Europe or Asia. A higher score for membership in one of the foreign databases marks a wing as being of potential foreign origin.
The above approach uses established statistical concepts on a relatively
new type of morphological data, landmark coordinates. While landmark
coordinates have become a popular subject in the community of biologists
interested in biological
shape, our application of this approach is one of the first cases in
which it is being used to solve a practical problem, the
Asian gypsy moth incursion into North America. We need to explain the
general approach starting with its theoretical basis in order to make it
acceptable and, moreover, the method of choice in this arena of applied
technology. We are using the Drosophila
wing project to establish the credibility of our method. We also need
to show examples of how this approach successfully discriminates species,
sibling species and populations. Our chosen species fit these several needs.