Project: 2001-08. Report - Dothistroma prediction model (2001–08)

Date: 2003 Author: Forest Research Publication: Report Project reference: 2001-08 Report is available online Download report

Executive summary:

A model to predict the Dothistroma spray programme was developed by testing the effect of rainfall, raindays, and inoculum in previous season on the percentage of susceptible area sprayed in individual years from 1966 to 2002 at Kaingaroa and Kinleith. For both data sets a multiple regression using average monthly rainfall from November–February (including October or March rainfall if 200 mm or more), total raindays of at least 0.1 mm from November–February, and the percentage of susceptible area sprayed in the previous season to predict area sprayed gave reasonable results. The Kaingaroa data set gave an R2 of 0.53 and for Kinleith R2 = 0.70. Some of the unexplained variation was attributable to incomplete or inaccurate input data, management decisions influencing area sprayed, and, possibly, exceptional climatic events such as a wetter and warmer than average September or April.

Further work to enhance the model should be carried out. This includes refining the input data, adding more observations, testing other combinations of predictive variables, and establishing disease progress field trials. Such enhancements should result in a model sufficiently robust to predict accurately the size of spray programmes and in turn reduce costs associated with over– or under–estimating supplies needed to service the programme.

Report is available online