Simultaneous assessment of germination and infection dose-responses in fungicide-treated seeds with non- and semiparametric statistical methods

摘要

Assessing time-varying competing events presents several challenges to standard statistical analyses. Here, we demonstrate the assessment of competing transitional events with non- and semi-parametric time-to-event analyses using a seed germination versus pathogen infection system. We investigated the incidence of germination or visible infection in Unto la paniculata seeds following treatment with an anti-microbial biocide (PPM (TM)). Kaplan-Meier estimates of survivor functions indicated that temporal patterns of germination and infection were distinctly different. PPM improved germination and reduced infection up to 2- and 60-fold, respectively, compared with controls. Extended Cox regression models indicated that PPM dose and the PPM x day interaction were significant when the event of interest was infection. Neither PPM dose nor the PPM x day interaction was significant when germination was the event of interest. No difference in early seedling dry mass was detected among PPM doses. However, visible root stunting was observed with PPM doses > 2.0 ml l(-1). We conclude that PPM may be beneficial at = 2.0 ml l(-1) for germination and subsequent seedling development. In summary, the application of time-to-event analyses provides appropriate and robust methods for the evaluation of competing transitional events and we encourage more wide-spread use in seed research.