Potential biases in screening for plant resistance to insect pests: an illustration with oilseed rape

Breeding to increase crop resistance is a common strategy to decrease damage caused by insect pests, especially in the current context where insecticides are becoming at the same time less accepted by society and less efficient because of widespread pest resistance. The main bottleneck of this strategy is phenotyping. Although simple, high-throughput methods have been proposed which could be highly useful, they may raise conceptual issues. Using field and laboratory experiments on oilseed rape (Brassica napus) and the pollen beetle (Brassicogethes aeneus syn. Meligethes aeneus), we illustrated possible difficulties with this approach: (i) field screenings might not represent the real attractiveness of the tested genotypes; (ii) plant phenology or spatial organization of the genotypes might bias field screening results; (iii) experiments based on detached plant parts (here, single flower buds or anthers) might not allow to infer the plant-insect relationship of the whole plant. We propose ways to better take these risks into account.