The apriori probability is to give a higher weight to certain classes over others. This apriori information is derived from your knowledge of the landscape that is being classified. The signature file contains the statistics of all the classes that will be used in your classification. With the inputs given, the apriori probability file that defines the apriori information specifies a class that is not available in the signature file.
For example, if you are performing a classification in a remote part of Alaska, you might want to give a higher probability to conifers. During the classification, if a particular cell could be one of multiple classes, a higher weight is given to the conifer class during the classification, making it more likely that a cell does in fact contain conifers and is classified as such.
Make sure that all classes in your apriori file exist in the signature file that you will be using in the classification. Remove any classes from the apriori file that do not exist in the signature file or add the class and create a new signature file. Check the Class Probability or Maximum Likelihood Classification documentation for further details on the formatting of the apriori file.