measurements. The length of the fused trochanter/femur segment, the tibia, and the
tarsus was recorded for one the pairs of each leg (prothorcacic, mesothoracic, and
metathoracic). In addition, the width of the fused trochanter/femur and tibia was recorded
at their widest points on each leg. This procedure resulted in a total of 15 measurements
for each specimen. The slide mounted specimens were digitally photographed with a
JVC model KY-F70B 3-CCD digital camera (JVC Americas Corp.) mated to a Leica
DMLB compound microscope (Leica Microsystems AG) with a Diagnostic Instruments
T-49C 0.45x c-mount coupler (Diagnostic Instruments, Sterling Heights, MI). The
Syncroscopy Auto-Montage software (Synoptics Ltd., Frederick, MD) was used to
measure the images at a resolution of 1360 x 1024 pixels. This system also was used to
measure the eggs of T. ovatus. Measurements were recorded digitally in order to reduce
error (the microscope only had to be calibrated once); this method also is less time
consuming than using an ocular micrometer.
Data were analyzed using the SAS statistical software (SAS Institute Inc., Cary,
NC). Due to the number of dimensions measured (15 observations per insect; n = 62), a
principal components analysis was performed using PROC PRINCOMP. This procedure
reduced the dimensions of the data by deriving a small number of linear combinations (in
this case 2 principal components) from the data. Next, a cluster analysis was performed
on the results of the principal components analysis using PROC FASTCLUS in order to
delineate distinct clusters of observations. Due to the discontinuous nature of insect
growth, these data clusters were used to differentiate the number of instars and separate
them from the adult stage. PROC FASTCLUS also calculated the mean and standard
deviation for all 15 morphometric parameters in each cluster. Finally, to obtain a better