Using matrix population models to inform biological control management of the wheat stem sawfly, Cephus cinctus

Demographic models are a powerful means of identifying vulnerable life stages of pest species and assessing the potential effectiveness of various management approaches in reducing pest population growth and spread. In a biological control context, such models can be used to focus foreign exploration or conservation efforts on enemies that attack life stages identified to have the greatest impact, and determine target levels of predation or parasitism that would be necessary to suppress population growth. In this study, we constructed a matrix population model to assess the potential effectiveness of biological controls against the wheat stem sawfly, Cephus cinctus, a major pest of wheat in North America. We calculated the sensitivity of C. cinctus population growth to changes in stage-specific survivorship, to identify the stage at which parasitoid attack would have the largest impact on pest population growth. We also calculated the stage-specific rate of mortality needed to reduce C. cinctus population growth rate to zero, to set targets for conservation biological control approaches. Our model indicates that C. cinctus populations are growing (λ = 1.022), and are predicted to triple in a year in the absence of added control measures. The winter larval stage had the highest elasticities, suggesting this stage is the weakest link in the pest life-cycle, in part reflecting the much longer average duration of the winter compared with the summer larval stage (45 versus 5 weeks). Parasitism levels by native Bracon spp. parasitoids necessary to suppress C. cinctus population growth were the same for summer and overwintering stages (68%). These target parasitism levels far exceeded those typically observed in the field, and conservation measures employed to date suggest that single actions do not bolster parasitism to target thresholds. Thus multiple conservation measures (e.g. reduced tillage, increased cutting height and the provisioning of floral resources) will likely need to be complemented by other management approaches (e.g., host plant resistance), to suppress C. cinctus populations. Our results reinforce previous work demonstrating the utility of matrix models for evaluating the potential efficacy of biological control agents, and further illustrates how they can be used to evaluate, and set targets for, conservation management approaches using specific natural enemies