Language selection


Development of bioclimatic models to forecast the dynamics of two insect pests: carrot weevil and carrot fly

Project Code: PRR07-090

Project Lead

Gaétan Bourgeois - Agriculture and Agri-Food Canada


To develop, validate, demonstrate and assist carrot growers to adopt anintegrated system for the management of carrot weevil and rust fly

Summary of Results


In Canada, carrot weevil (Listronotus oregonensis) and rust fly (Psila rosae) are major pests causing serious economic damage across all carrot production areas. In mid-1980’s, forecasting models had been developed for both pests using limited documented observations. However, new information on changes in population dynamics indicates that crop damage could also occur from a 2nd generation of carrot weevil and that life cycles of carrot rust fly may have evolved with time.

This 3-year (2007-2010) project aimed at updating the existing models using extensive sets of new data, so that the models account for the changing population dynamics of these pests, and facilitating the adoption of the tools by carrot growers across Canada.


The data used for updating the prediction models for these two pests were collected at the experimental farm of Agriculture and Agri-Food Canada (AAFC) in Sainte-Clotilde (Quebec) from 1983 to 2009. Meteorological data collected from the weather stations at the same sites for the same period of time allowed the elaboration of bioclimatic models based on cumulative thermal units (degree-days).

Carrot weevil: Data were collected from two plots seeded with carrots at beginning of May of each year of the project. Standard crop production practices were implemented in these plots, except that no insecticide applications were used. In late April, 3 wooden-plate weevil traps were placed 2 m from the edge of each plot, and 3 other traps were placed in the middle of each plot. The traps were monitored twice a week till mid-September by counting the number of adult weevils found in each trap.

Carrot rust fly: Data were collected from the same plots as for the weevils. Yellow sticky traps were installed along the borders of the plots at every 5 m and at 1 m high from the ground between 10 and 15 May of each year of the project. The traps were monitored twice a week till the end of season by counting the number of adult carrot rust flies found in each trap.

The combined data collected during this project and over the previous 25 years were used to elaborate the bioclimatic models for each pest, as well as to evaluate their predictive capacity. The updated models were validated in a select number of commercial carrot crops by comparing the timings of observed insects captured in the traps and those predicted by the models. The meteorological data obtained from respective weather stations served to complete the validation. For the weevil, the model validation was conducted in Lanaudière (Quebec) and Bradford (Ontario) regions over the periods between 1995 - 2010 and 2005 - 2008, respectively. For the carrot rust fly, the model validation was conducted in the Bradford region from 2005 to 2008.


Overall, the analyses of comparisons between observations and predictions indicated an improvement of the predictive capacity of the updated models, given that observations of over 20 years had been taken into account.

Carrot weevil: Comparison statistics obtained for thresholds of 5, 50 and 95% of weevil populations indicate that the variations between observed and predicted timings averaged 3.7 days. Moreover, statistics of model efficiency indicate that the model has a better predictive value then the use of the mean of observations. In general, the weevil completes one generation per season. Occasionally, a 2nd generation of the weevil can be reached and cause damage to the crop. Unfortunately, the trap used in these experiments did not allow obtaining enough data to characterize this 2nd generation. Indeed, the type of trap used works well for the 1st generation since it provides a feeding source and an egg laying site for weevils at a time when the carrots are at the early stages of their development. When the 2nd generation of adults emerges, the carrots are almost mature and weevils are much less attracted to traps. This study was therefore limited to modeling the emergence of overwintering adults (1st generation) of the carrot weevil.

Carrot rust fly: In general, this pest completes two generations per season and the traps used are quite efficient for an adequate monitoring of the adult populations. The updated model offers a good prediction for the 1st generation of the insect. The variations between the observed and predicted timings averaged 6 days for the 1st generation, while the difference was 8.6 days on average for the 2nd generation, therefore less accurate by a few days. Moreover, the model indicates that to complete its life cycle, this 2nd generation requires about 200 degree-days more than it was observed in studies conducted in early 1980’s, which implies that the carrot rust flies remain active longer in the fall.

Validation results obtained from Quebec and Ontario demonstrated the excellent predictive capacity of these two models in different carrot growing regions, provided that the time of trap installation and seeding dates are taken into consideration. Early installation of traps (soon after carrot seeding) is especially important for best results with the carrot weevil model.

These prediction tools are part of an integrated system that includes also economic damage thresholds, appropriate scouting techniques and control recommendations, allowing a system approach for weevil and rust fly management in carrots. These two models updated through this project are integrated into CIPRA (Computer Centre for Agricultural Pest Forecasting), a multi-crop/pest decisions support system software developed by the AAFC’s team of researchers in bioclimatology and modeling at the Saint-Jean-sur-Richelieu Centre (Quebec). The information and service are now available to carrot growers in Canada through the provincial extension personnel. It is anticipated that these models will allow growers to better target their pest control practices and avoid unnecessary pesticide applications, thus contributing to judicious pest management decisions and pesticide risk reduction.

Numerous extension activities were conducted in the context of technology transfer (e.g. publications, presentations at conferences and posters) to communicate the results of this studies to stakeholders.

For more details, please contact Dr. Gaétan Bourgeois.

Report a problem on this page
Please select all that apply:
Date modified: