1. A time thermal model for predicting the parasitism of Orobanche cumana in sunflower - five years of field validation HANAN EIZENBERG J. HERSHENHORN, G. ACHDARI. AND J. E. EPHRATH Department of Plant Pathology and Weed Research, Newe Ya’ar Research Center, ARO, Ramat Yishay, Israel.

4. Background During its initial stages of parasitism, the broomrapes grow underground Predicting their developmental stages at this phase is a necessity in order to properly apply control measures This challenge can be met by using the modeling approach, as reported for P. aegyptiaca, O. minor and O. cumana, in tomato, red clover and sunflower, respectively In those studies, the relations between parasitism dynamics and thermal time has been described by mathematical functions, e.g. sigmoid, logistic, Weibull, and polynomial functions

5. Weed Research 50, 140–152

7. Y - broomrape number b - the slope at x 0 a - the upper asymptote (maximum) Y 0 - the lower asymptote (minimum) X 0 - the GDD when Y is 50% of maximum (median) Four parameters logistic equation

8. Weed Research 50, 140–152

9. The objective of this study is to: Calibrate under field conditions an equation that describes the parasitism dynamics of Orobanche cumana in sunflower To estimate the contribution of an additional estimated parameters (lag phase) to reduce the RMSE of the model

10. Fit model for individual field Model calibration field Trails 4 locations 4 years Model validation field Trails 5 locations 2years Input Test combined model 4 locations Does the model consist? Yes No Model test Model adjustment base on multi years data Flow chart for model development

11. Fit model for individual field Model calibration field Trails 4 locations 4 years Model validation field Trails 5 locations 2years Input Test combined model 4 locations Does the model consist? Yes No Model test Model adjustment base on multi years data Flow chart for model development

13. Minirhizotron system

14. 50 cm 45°

18. 9 Field experiments through 2005-2009 Model calibration Model validation

19. To estimate the number of attachments related to thermal time, the following equations were tested: Sigmoid, Gompertz (both three parameters) and Weibull (four parameters) These equations are characterized with the pattern lag, and with the log and maximal asymptote for the number of parasite tubercles as a function of thermal time. Fit of equations was evaluated by RMSE, and by the corrected Akaike Information Criterion (AIC)

20. Model calibration field trails Logistic (RMSE=0.9)

21. Model calibration field trails Weibull (RMSE=0.06) Logistic (RMSE=0.09)

22. In the calibration studies, the number of attachments was best fitted to thermal time using the Weibull equation, which resulted in a great fit in the validation studies (RMSE = 0.066; R 2 = 0.99; slope a ~ 1). a = 2.1 P <0.0001 σ = 331.7 P <0.0001  = 1.9 P <0.0001 µ (lag) = 420 P <0.0001 µ = lag (location) σ = scale (63% of maximum) λ = shape a = maximum asymptote

23. Validation test of the model

24. A four parameters modified Weibull equation (estimated the lag phase) based on the parameters obtained from model calibration This is not a fit! This curve is based on the parameters estimated from the calibration model

25. Validation test of the model A four parameters modified Weibull equation (estimated in the lag phase) based on the parameters estimated from model calibration Blue circle obtained from field validation studies Curve obtained from Calibration study

26. Model test R 2 = 0.99; P < 0.001

27. Where such a model could be applied? Smart control of O. cumana in sunflower 200 GDD600 GDD1000 GDD Weibull equation estimated the parameters: µ (lag) = 420 and σ (63% of maximal asymptote) = 331.7

28. Lag=420  =331 Imazapic (as other imidazolinone herbicides) effectively controls broomrape when it is attached to the roots Imazapic (4.8 g a.i. ha -1 ) applied at 720 GDD

29. Where such a model could be applied? Smart control of O. cumana in sunflower 200 GDD600 GDD1000 GDD Non herbicide treated control Imazapic (4.8 g a.i. ha -1 ) applied at 720 GDD

30. SED=5.65 Thermal time (degree days) 600 GDD1000 GDD Control efficacy based on the model

31. Conculsions (chemical control) The example that has been given demonstrates control efficacy of one foliar treatment with imazapic Further chemical treatments should be applied according to the model but not as foliar applications as imazapic may injure the sunflower reproductive tissues after initiation Herbigation may be considered for further treatments but a protocol should be developed

32. Conclusions Thermal time can robustly predict O. cumana parasitism in sunflower using the Weibull equation The Weibull equation adds a biological dimension to the model, compared to the other equations, as the lag phase allows to estimate the precise timing of parasite attachment to host roots This information is crucial in any attempt to develop control strategies for these parasitic weeds

33. Taking home message The modeling approach is essential for the development of control strategy and decision support systems for Orobanche managment However, It could be applied in other field of studies related to parasitic plants such as resistance, biological aspects and strigolactones