EVALUATION OF COUNTING ERROR DUE TO COLONY MASKING IN BIOAEROSOL SAMPLING

Colony counting error due to indistinguishable colony overlap (i.e., m asking) was evaluated theoretically and experimentally. A theoretical model to predict colony masking was used to determine colony counting efficiency by Monte Carlo computer simulation of microorganism collect ion and development into CFU. The computer simulation was verified exp erimentally by collecting aerosolized Bacillus subtilis spores and exa mining micro- and macroscopic colonies. Colony counting efficiency dec reased (i) with increasing density of collected culturable microorgani sms, (ii) with increasing colony size, and (iii) with decreasing abili ty of an observation system to distinguish adjacent colonies as separa te units. Counting efficiency for 2-mm colonies, at optimal resolution , decreased from 98 to 85% when colony density increased from 1 to 10 microorganisms cm(-2), in contrast to an efficiency decrease from 90 t o 45% for 5-mm colonies. No statistically significant difference (alph a = 0.05) between experimental and theoretical results was found when colony shape was used to estimate the number of individual colonies in a CFU. Experimental colony counts were 1.2 times simulation estimates when colony shape was not considered, because of nonuniformity of act ual colony size and the better discrimination ability of the human eye relative to the model. Colony surface densities associated with high counting accuracy were compared with recommended upper plate count lim its and found to depend on colony size and an observation system's abi lity to identify overlapped colonies. Correction factors were develope d to estimate the actual number of collected microorganisms from obser ved colony counts. This study determined that computer simulation of c olony surface density and resulting masking can identify suitable air sample volumes (i.e., flow rates and collection times) for measuring c oncentrations of airborne microorganisms and that errors due to colony masking can be reduced by applying correction factors to observed col ony counts.