Towards stochastic simulation of crop yield: a case study of fruit set in sweet pepper

A.M. Wubs

Research output: Thesisinternal PhD, WU


Crop growth simulation models are widely used in research and education, and their use in commercial practice is increasing. Usually these models are deterministic: one set of input values always gives the same output of the model. In reality, however, variation exists between plants of the same crop. A simulation model taking this variation into account is therefore more realistic. The aim of this thesis is to introduce a stochastic component into a dynamic crop simulation model. As case study, fruit set in sweet pepper was used, because large variation in fruit set between the plants exists. Competition with fast growing fruits causes abortion of flowers and young fruits, which results in periods with high and low fruit set, and consequently periods of high and low fruit yield. A literature review showed that most factors influencing fruit abortion can be expressed in the terms source and sink strength. Source strength is the supply of assimilates; a higher source strength increases fruit set. Source strength takes into account leaf area, radiation, and CO2 level and temperature. Sink strength is the demand for assimilates of the fruits and vegetative parts. It is quantified by the potential growth rate, i.e. the growth rate under non-limiting assimilate supply. Assimilate demand of the fruits depends on their number, age, and cultivar. If the total fruit sink strength of a plant is low, fruit set is high. Vulnerable for abortion were very small buds, buds close to anthesis and flowers and young fruits up to 14 days after anthesis. An experiment with six Capsicum cultivars with fruit sizes ranging between 20 and 205g fresh weight showed that variation in weekly fruit yield is highly correlated with variation in weekly fruit set. Fruit yield patterns resembled fruit set patterns, with a lag time being equal to the average fruit growth duration. Further investigation showed that the cultivars not only differed in sink strength of the individual fruits, but also that the source-sink ratio above which fruit set occurred was higher in cultivars with larger fruits. In the second half of the thesis, flower and fruit abortion was modelled. Survival analysis was used as the method to derive the abortion function. Source and sink strength were used as the factors influencing abortion. Their effect on the probability of abortion per day was non-linear: at high values of source and sink strength an increase did not further decrease or increase the probability of abortion, respectively. Flowers on the side shoots turned out to have a higher probability of abortion than flowers on the main shoot. Most flowers and young fruits aborted around 100°Cd after anthesis. The obtained function was used in a crop simulation model for sweet pepper. After calibration the model was able to simulate the observed fruit set pattern, although fruit abortion was not properly simulated when low source strength was combined with high sink strength. Validation with three independent data sets gave reasonable to good results. Survival analysis proved to be a good method for introducing stochasticity in crop simulation models. A case study with constant source strength showed asynchronisation of fruit set between the plants, indicating that fluctuations in source strength are an important factor causing synchronisation between individual plants.

Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • Wageningen University
  • Marcelis, Leo, Promotor
  • Heuvelink, Ep, Co-promotor
  • Hemerik, Lia, Co-promotor
Award date6 Oct 2010
Place of Publication[S.l.]
Print ISBNs9789085856993
Publication statusPublished - 2010


  • capsicum annuum
  • paprika
  • simulation models
  • stochastic models
  • fruit set
  • abortion (plants)
  • survival
  • source sink relations


Dive into the research topics of 'Towards stochastic simulation of crop yield: a case study of fruit set in sweet pepper'. Together they form a unique fingerprint.

Cite this