As expected, the current model did not predict all
possible occurrence areas of the Fairy Pitta in Taiwan.
The number of pitta detected per cell in the active
selection cells outside predictive areas clearly
indicated that the potential omission error in the
model. In particular, the Ilan County was completely
missing in the prediction. Part of the reasons for the
error was the few sighting records available to frame
our selection rules. In contrast, some areas predicted
by the current model, though not verified in the 2001
survey, were found to have pitta presences in either
2002 or 2005 surveys or in some scattered news
reports (L. T. Chao, personal communication).
Although there are some advocacy in promoting