Revised thresholds for cabbage stem flea beetle on oilseed rape


Cereals & Oilseeds
Project code:
01 July 2004 - 30 June 2007
AHDB Cereals & Oilseeds.
AHDB sector cost:
£66,044 from HGCA (project no. 3023).
Project leader:
David B. Green ADAS Woodthorne, Wergs Road Wolverhampton, WV6 8TQ


pr428_final_project_report pr428_final_project_report_abstract_and_summary

About this project


The aim of the study was to investigate a field-based method using water traps to provide information in early autumn on the need for control of cabbage stem flea beetle. At each of 71 sites (27, 25 and 19 sites in autumn 2004, 2005 and 2006 respectively) in central, eastern and northern England, four yellow water traps 25 cm in diameter were placed on the soil surface in winter oilseed rape crops soon after drilling or at early crop emergence. Two traps were sited on the crop headland with two traps within the field; 12 and 24 metres from the crop headland. Weekly and total catches of cabbage stem flea beetles in traps were recorded between crop emergence and late October or early November. Peaks of adult activity were recorded in late September or early October with higher totals of beetles recorded in each year of the study in central and northern England than in eastern England. Plant samples were collected in December to determine the number of larvae per plant and seventeen from the total of 71 sites subsequently developed infestations averaging two or more larvae per plant.

Regression analysis using data from all 71 sites showed that mean numbers of larvae per plant were significantly related to mean number of adults per water trap (P < 0.001). An infestation averaging two larvae per plant was likely to be attained from an average of 36 (SE 3.2) adults per trap with 69.3% of the variance explained. Regressions testing the relationships between adult numbers and larval infestations for each of the three study years were also significant (P=or< 0.001). Regressions were tested for headland or field-sited traps with two larvae per plant likely to be attained from means of 33 and 40 beetles per trap respectively.

The use of water traps enabled successful decisions to be made whether to spray or not at 87% of sites using a mean of 36 beetles per water trap. Overall predictive success was improved to 89% if the lower or upper 95% confidence limit values of 30 and 43 respectively per trap were used. Similar predictive successes were also obtained from headland or field-sited traps with correct treatment decisions made at 86% and 90% of sites respectively.

At sites where infestations averaging two or more larvae per plant were recorded, predictions of the need for control using the lower 95% confidence limit value of 30 cabbage stem flea beetle adults per water trap enabled 82% correct treatment decisions to be made, compared with 65% of correct treatment decisions using the median and upper 95% confidence limit values of 36 and 43 adults per water trap. A threshold value for water trap catches averaging 30-35 per trap was shown to be an action threshold above which an autumn pyrethroid spray treatment would be justified, irrespective of whether an earlier seed treatment had been applied.

In autumn 2004, four yellow sticky traps were compared with water traps as predictive methods at 27 sites. Sticky traps caught fewer cabbage stem flea beetles than water traps with a mean of 1.3 per sticky trap compared with a mean of 8.0 per water trap. A significant regression was obtained (P < 0.001) with 51.0% of variance explained with two larvae per plant likely to be attained from a mean of 5.7 beetles per sticky trap. The use of sticky traps provided a poor predictive method compared with water traps and the method tested did not predict the two sites in 2004 where above threshold numbers of larvae developed.

Regressions between larval numbers and plant, cotyledon and first true leaf damage were also tested at 52 sites in the first two years of the study during harvest years 2005 and 2006. Although larval numbers were significantly correlated with plant and cotyledon damage, only 14.0% and 10.8% of variance was explained and these methods were overall poor predictors of larval damage with only 20% of sites that developed larval infestations greater than two per plant being correctly predicted for treatment.