Use of the most probable number technique to quantify soil-borne plant pathogens

Bioassays using serial soil dilutions and most probable number (MPN) estimations have been used by various authors to quantify inoculum of soil-borne plant pathogens. The requirements of a reliable bioassay are discussed; they include a good choice of dilution series and reproducible growing conditions. Sources of computer programs for analysis of the data are listed. The importance of testing the fit to the mathematical model used is illustrated and emphasised. Factors affecting the size and stability of the standard errors and of the inherent bias of the most probable number estimate are discussed. Equations are presented for calculating the expected standard errors, approximate confidence limits and least significant differences for different dilution factors and numbers of replicates. The benefits of using uneven replication are illustrated. Mathematical considerations show that the technique should enable differences of an order of magnitude to be detected and MPNs should be quoted with a maximum of two significant figures. Dilution ratios as large as 10 should be avoided. Statistical and biological difficulties, especially in standardising growing conditions when soil moisture is critical, indicate that results should normally be regarded as relative, rather than absolute, measurements of inoculum.