After selecting the CharSets it is necessary to Set Character Types and Weights, under the Data menu. The character type should be "myordering," which assigns an ordered status to subcharacters derived from cumulative frequencies and an unordered status to those from noncumulative frequencies. The weights should be either USW or ESW. X
|Preliminary Pokemon Phylogenetic Hypothesis|
For the particular Pokemon example I rooted the "animal" Pokemon with Electrode and obtained a single mp-tree when using USW.
Continuous characters
Continuous characters should be ranked, this can be easily accomplished by dividing the maximum range between any two measurements by 24 (arbitrary convention, or other number of steps we would like for the analysis, we could even just round the measurement), and then using this product to divide all the measurements. The result can be rounded to the nearest integer and then is ready to be used as raw data for GFC analyses. In the case of morphological measurements that are correlated to size we recommend normalizing by obtaining ratios against the largest measurement (e.g., tibia size/Total Length).
COMPARISON BETWEEN NORMALIZING AND RESIDUALIZING MEASUREMENTS FROM MORPHOLOGY FOR GFC ANALYSIS
Comparison of normalizing vs. residualizing measurements for GFC analysis. Although the graphs look similar at first glance there is a difference that could be significant for GFC analyses. Compare sets of arrows on both graphs. Residualizing is affected by the included taxa and their relationships. Ratios do not have this problem. Residualizing over log-transformed variables gives plot results similar to those obtained by calculating ratios, almost undistinguishable, nevertheless residuals will still be affected by the included taxa.
Plot of Residuals
Plot of Ratios
