Survival analysis of flower and fruit abortion in sweet pepper

In order to obtain a crop growth model that can simulate inter- and intra-plant variation in fruit set, fruit abortion times in sweet pepper were analysed by means of survival analysis. Survival analysis is a statistical technique dealing with the timing of events. The Cox proportional hazards model estimates the instantaneous baseline probability of abortion per time (hazard rate), given that the fruit has not aborted yet. It also estimates the effect of explanatory factors (covariates) on the abortion rate. Two important factors known to influence abortion in sweet pepper are the supply and demand for assimilates (source and sink strength, respectively). A plant density experiment was analysed, as density influences plant source and indirectly also sink strength. Flowering as well as abortion or harvest dates were recorded for all individual flowers and fruits. LAI was measured to calculate the source strength with a photosynthesis-based simulation model. Empirical curves showed that survival of flowers and fruits (per plant) was higher at 1.6 plants/m2 than at 3.1 or 4.6 plants/m2. Plant density, source strength and source/sink ratio all correlated with the treatment plant density, but source/sink ratio explained most of the variation in the data in a Cox proportional hazards model. Averaging the source/sink ratio over seven days gave the best fit. Adding the position within the node (main or side branch) improved the fit; flowers from the side branch had a higher probability of abortion per unit time than flowers from the main branch. The susceptibility for fruit abortion differed among plants. The baseline hazard rate indicated that between 0 and 13 days after anthesis, flowers were most susceptible for abortion