These are warning messages that occur when you apply the LOESS smoother to two classes. Loess is in the stats package that come with R. ggplot2 is delivering these warnings from loess to you. In fact, these warning messages come from the fortran code that implement the loess method.
I'm not an expert on LOESS, but here's an extract from Statistical Models for S, "Error Messages from the Bowels of Loess", suggesting these warnings come when there are few observations in this areas of plot.