Feeding pollinators from weeds could promote pollen allergy. A simulation study

Weed pollen both feeds many insects and causes allergies in humans. The objective of the study was to in-vestigate the determinants in terms of management practices and weed species traits on those pollen-borne services and dissservices (hence (dis)services). We developed a set of indicators to assess the magnitude of these (dis)services combining species pollinating values (or allergizing potential) with species densities by means of the CONTRA method for designing fuzzy decision trees. Densities were calculated by the process-based weed dynamics model FLORSYS, which simulates the dynamics of a multispecies weed flora at a daily time step and over the years depending on the field's cropping system and pedoclimate. We then simulated a virtual farm-field network consisting of 272 cropping systems from 7 regions in France and Spain, with FLORSYS over 27 years and with 10 weather repetitions. The simulated weed flora and (dis)service indicators were analysed with analyses of variance, regression trees, RLQ and fourth-corner analyses. The simulations showed that the management practices that increased pollen-borne allergy risk were also often those providing large trophic resources for butterflies and, to a lesser degree, to other pollinating insects. Practices reducing pollen-borne (dis)services included frequent and intensive tillage, frequent spring crops, herbicides sprayed on cash crops, preferably broad-spectrum products targeting emerged weeds but the effect of all these techniques greatly varied with the date of their implementation. The weed species that most contributed to pollinator food were those that were present and growing best during the critical feeding period of the pollinators rather than those with the largest pollinating value or those with the largest densities during the cropping campaign. Species characteristics fa-vouring pollinator food included low dormancy, tall ``stemmy'' plants and high temperature requirements in terms of photosynthesis and frost sensitivity. Then, we identified the management strategies that reconciled pollinator food with low allergy risk. For instance, direct sowing or several weeks without tillage after harvest or prior to crop sowing increase food offer. Finally, we proposed rules for ex post monitoring following the in-troduction of novel crops or management techniques. For instance, ex post biodiversity monitoring following the introduction of glyphosate-tolerant maize should focus on situations with frequent tillage and on the presence of butterflies.