A reappraisal of the terverticillate penicillia using biochemical, physiological and morphological features. II. Identification

The data from an integrated numerical classification was used to construct identification schemes for some fasciculate penicillia. The identification schemes were presented as a synoptic key and a frequency matrix for computer-assisted identification. Statistical testing of the frequency matrix showed that although character separation values were generally low, only four pairs of taxa showed overlap greater than that expected for a rectangular distribution. The identification schemes were tested practically with 52 previously studied strains and 51 further cultures. A synoptic key based on 10 and 90% cutoff limits was used to correctly identify 44 of the 51 additional strains, although this proved very sensitive to single test discrepancies. The frequency matrix was used to correctly identify 45 of the additional strains with a Willcox probability score and this was compared to identifications based on the modal likelihood fraction.