Rarefied data matrices were created using a batch file executed in NONA
(Goloboff, 1993). Initially, the following options were set: AMBIGUOUS= (support
from ambiguous optimizations considered), RSEED 0 (random number generator seed
set based on the computer’s clock). Characters were randomly removed using the
RESAMPLE N command (where N is the percent probability of removing any
individual character from the matrix), HOLD 10000 (tree buffer set to 10,000),
HOLD/10 (to limit the number of trees kept in each replication of “MULT” to ten),
MULT* 50 (performs TBR branch swapping on a Wagner tree generated using random
taxon addition over 50 replicates). The results were saved using KSV* (save all trees in
memory to a file), KSV/ (closes tree file), KEEP 0 (purges tree file), CCODE [.
(restores all characters to the matrix). This sequence was repeated 100 times for each
value of N. The value of N varied between 2 and 98 in intervals of 2. In total, 4900
rarefied matrices were analyzed for each complete data set.