Phil Busey Agronomy
Consulting Inc.


Publications list

Busey, P. 2009. Proliferation and detection of contamination in turfgrass vegetative propagation. Internat. Turfgrass Soc. Res. J. 11:1177-1182.


Turfgrass cultivars are often vegetatively propagated clones which can be contaminated by genetic off-types of the same or closely related species. If a contaminant clone is more competitive than the intended cultivar there is potential for proliferation (increased contamination) during propagation. The objectives of this study were to estimate the rate of proliferation of a competitive genetic contaminant into a less competitive cultivar, and to estimate the likelihood of detecting a contaminant, once it occurs. The logistic equation was used to model the sigmoidal growth in cover, as a function of days since planting, for a weedy bermudagrass, Cynodon dactylon (L.) Pers 'PI-291586', grown without competition. The observed growth rate was depreciated in steps to simulate the growth of a hypothetically less competitive cultivar growing together with a small admixture of the more competitive clone. This was repeated in steps across a range of admixture rates. It was shown that a small rate of admixture, 0.001, of a contaminant with a 50% faster growth rate, could in one planting cycle proliferate 140 times in the planting. The detection of contamination through random sampling depends on the power required (the likelihood of finding a contaminant), the frequency of occurrence of the off-type, and the accuracy of diagnosis. While the accuracy of diagnosing contaminant grasses may vary, the binomial expectations of random sampling can be precisely estimated. The required number of samples is ln(1-power)/ln(purity), where the power is the likelihood of finding a contaminant and the purity is the predominance of the intended cultivar. The likelihood of detecting a genetic variation through random sampling is rare, unless many samples are analyzed. To be 95% sure of detecting a 0.8% contaminant, one would need to collect and analyze 373 samples, which is probably impractical. Therefore, visual sampling or other survey approaches are necessary in any quality assurance program.