Figure 4.

Relationships between taxon coverage of the reference database and identification success. Databases including 100, 500, 1,000 and 1,500 insect species were obtained by randomly sampling dataset A (three replicates for each subset of species). The identification success of Best Match (BM) and Best Close Match (BCM) was averaged across replicates (SD as error bars) and linear regression fitted after including the proportion of correct matches resulting from the analysis of the whole dataset (1,995 species).