IdentiFly

IdentiFly
About IdentiFly
IdentiFly is a software for semiautomated classification based on geometric morphometrics. Primarily it was developed for identification of insect species but it can be used for classification of any objects which differ in shape. Moreover, IdentiFly can be used as an alternative to tpsDig for digitizing landmarks in two dimensions.

Installation

  • Download the ZIP file: identifly10.zip
  • Unpack the ZIP file and copy identfly10 directory to your hard drive.
  • Run the identifly.exe

Use of IdentiFly for identification of honey bee lineages and subspecies
In honey bee there is more than 20 subspecies (or geographic races) which were grouped into four evolutionary lineages.

Before a honey bee colony can be classified as one of the evolutionary lineages you need to obtain forewing images of at least ten workers from the colony. The wing images from different colonies should be saved in separate directories. Each of the wings has to be measured. In case of honey bees 19 landmarks should be located in appropriate part of the wing. The landmarks should be placed in such a way that blue circles around them are tangent to the venation outline in at least three points. Position of the landmarks is essential for correct classification.
Apis wing landmarks
In order to locate landmarks please use appropriate mode: Mode -> Add landmarks. In this mode clicking the image will result in adding a landmark. It is recommended to zoom in the image and verify if the landmark is correctly placed. If needed the position of the landmark can be corrected. In order to correct the position of landmarks you need switch the mode: Mode -> Edit landmarks. In this mode a landmark can be dragged and dropped to a new position. After measurements the image file should be saved: File -> Save. Position of landmarks is saved within the file and can be later verified, corrected and exported.

In case of honey bees there are several possible classification. In order to choose appropriate classification option Classification -> Set classification should be used. For identification of honey bee lineages choose file "classification_apis_mellifera_lineages.dw.xml". With each classification file there is associated a prototype file. It is an image containing information about number and position of the landmarks. In order to see the prototype option Classification -> View prototype can be used. Then the prototype is displayed and can be used as a reference for the correct position of the landmarks.

When all wing images from one colony were measured the colony can be classified: Classification -> Classify directory, next you choose directory where the wing images were saved. It is also possible to classify a single wing (using option Classification -> Classify), however, this is not recommended because identification of honeybee lineages and subspecies based on single wing is very inaccurate. The classification information will be displayed in a new window. The most important is group to which the colony was assigned. For proper interpretation of the classification user should inspect also classification probability. The probability values can range from zero to one. The higher probability of assignment to particular group indicates high similarity of the classified colony to reference samples.
In the best scenario the probability for one subspecies will be high and in all remaining subspecies very low. In this case the focal colony should be classified as the subspecies or lineage with the high classification probability. It is difficult to recommend a threshold value above which a classification probability should be considered as high. In another scenario classification probability is relatively high in two or more subspecies. In this case the focal colony should be considered as hybrid between the subspecies with high classification probabilities. Finally the classification probability can be low in all subspecies. In this case the classification of the focal colony as one of the subspecies should be concluded as doubtful. This can happen in two cases: when the focal colony is very different from all reference samples or when there is an error in measurements. The first case can occur when the colony belongs to species other than A. mellifera, for example A. cerana, or the colony belongs to subspecies not included in the classification model, for example, A. m. pomonella. This should not happen very often because the presented here classification covers most of the honey bee subspecies. Low classification probabilities in all subspecies most often occurs because there is an error in measurements. One of the most common errors is swapping position of two landmarks. In order to detect possible errors option Landmarks -> Outliers can be used. Then files from one folder are compared and the wings most different from the average are displayed. The differences of each landmark from the average are marked with yellow lines.
There is also graphical representation of the identification. The colony is represented as a black point in CVA graph with subspecies or lineages marked as ellipses. The closer the point to the centre of the ellipse the higher similarity of the colony to the particular subspecies or lineage. The graph should be interpreted with care, particularly by users not familiar with multivariate statistical analysis. It is important to notice that in many cases there is more dimensions than the two presented in the graph and the proximity of a point to an ellipse can be misleading. Information about other canonical variates are presented in the table below graph. If the colony differs markedly from all subspecies the black marker can be invisible in the graph.

As an example we present classification results of one colony collected in Cyprus. First we have used "classification_apis_mellifera_lineages.dw.xml" classification file.
Apis CVA classification
The colony was classified as O lineage. The probability of classification to this group was 0.16 which is much higher than values for lineages A, C and M: 2.0x10-10, 1.1x10-8 and 1.5x10-17, respectively. In this case it is safe to conclude that the identified colony was from lineage O.
In the graph showing CV1 and CV2 the black square representing the colony is within O ellipse.

Next we have used "classification_apis_mellifera_subspecies.dw.xml" classification file.
Apis CVA classification
The same colony was classified as as A. m. armeniaca. The classification probability (8.5x10-4) is much lower than previously. The second highest probability (3.9x10-9) is for A. m. syriaca. In this case we should conclude that the identified colony is similar to A. m. armeniaca but markedly different from the reference sample of this subspecies and either it is hybrid with A. m. syriaca or it belongs to subspecies not covered by the identification model. We know that the later explanation is true because using other methods the colony was identified as A. m. cypria. The colony was correctly classified as belonging to lineage O. However, it could not be classified as A. m. cypria because this subspecies was not included in the classification model due to small sample size.