Evaluation of the reliability of determining soft rot resistance in potatoes by the tuber slice method

Abstract

Soft rot resistance in potato (Solanum tuberosum L.) tubers can be determined by inoculating tuber slices with soft rotErwinia species. Questions have been raised in the literature concerning the reliability of the tuber slice method. The objectives of this study were to 1) examine the statistical assumptions underlying the analysis of variance for different response variables as measures of soft rot resistance using the tuber slice method of evaluation; 2) estimate the sample sizes necessary to detect specified differences with power 0.83; and, 3) choose the “best” variable for measuring resistance to soft rot based on valid statistical analysis and minimal sample size. Slices from fifteen tubers from each of three cultivars (Atlantic, Norchip, Superior) were inoculated withErwinia carotovora subsp.atroseptica, E. carotovora subsp.carotovora orE. chrysanthemi and incubated at two different temperatures (20 C and 30 C) for 48 hrs. The test was conducted on two dates. Tuber samples were sliced and weighed prior to inoculation and after the macerated tissue was removed following a 48 hr incubation period. The maximum diameter of macerated tissue, actual weight loss, four measures of proportional weight loss, and various transformations of these variables were analyzed. The “best” response variables for measuring resistance to soft rot by the tuber slice method were the diameter of the macerated tissue and the square root transformation of one of the measures of proportional weight loss. No differences were found among the cultivars nor theErwinia subsp. for either of these response variables, and the cultivar ×Erwinia subsp., cultivar × temperature and cultivar ×Erwinia subsp. × temperature interactions were not significant. However, there were significant differences between the incubation temperatures, and theErwinia subsp. × temperature interaction was significant for both response variables. Estimates of sample sizes necessary to find a 20% difference in main effect and interaction effect means were calculated.