Results
We placed an average of 392 models per site over six sites for a total of 2,356 throughout all our field sites of which 2,271 were recovered (96.39%) with 158 showing evidence of an attack event (6.96%). In French Guiana, of 1,604 model placements throughout 4 field sites, 1,524 were recovered (95.01 %) and 123 models were attacked (8.07%). In Panama, we placed 752 models throughout two field sites of which 747 were recovered (99.34%) and 35 models were attacked (4.69%). Attacks were recorded as damage caused on the wax bodies or paper wings by either avian, unknown or invertebrate predators.
We found palatable cryptic models to be similarly attacked to warningly colored phenotypes throughout our sites. This suggests that warning signals are at least as advantageous as crypsis even though they are more easily detected and even more so when co-mimics to a given aposematic signal are frequent (see Figure 4). In contrast, aposematism appears less effective than crypsis when co-mimics are absent (see Figure 3).