Mesoscale Modelling of Maize Pollen Dispersal

The growing introduction of genetically modified (GM) crops has generated a host of research efforts aimed at investigating the possibilities for coexistence between GM, conventional and organic farming systems. This is particularly true for maize, a plant that is extensively grown in Europe. Published experimental and modelling studies aimed at characterizing pollen dispersal have shown that most pollen emitted by a source field deposits within a short distance from the latter, but also that the observed dispersal functions have long fat tails, making it possible for pollen to contaminate plants at rather long distances. Such possibility has been confirmed recently from (i) a series of airborne measurements of pollen concentration and viability in the atmospheric boundary layer, (ii) chamber measurements of pollen viability in a range of temperature and humidity conditions and (iii) observations of fecundations in isolated plots of white-kernel maize, at several km from any maize field.

In order to better understand long-range dispersal of maize pollen we have developed an approach aimed at simulating the trajectories and dehydration of pollen grains in the atmosphere at regional scale. To this purpose we have modified the non-hydrostatic mesoscale Meso-NH model so as to introduce source terms for pollen emission, conservation equations for pollen concentration and moisture as well as a deposition velocity. Several simulations are performed over the Aquitaine region in South-West France, on several days during the maize pollination period. MesoNH is run in a two-way nested configuration including three nested computational domains down to a 2-km horizontal resolution. All the maize fields of the region have been previously identified from remotely sensed data.

Considering several days during which airborne measurements were performed at several times, observed and simulated concentration profiles are found to agree well throughout the atmospheric boundary layer. The simulations allow the pollen plume to be characterized through each day and maps of pollen deposition to be produced. The deposition rates at remote distances from the maize fields are in the same range as those measured in situ. Test simulations are also performed using specific landuse patterns. For example, on a typical convective day a single 12-km square maize source plot is shown to generate a plume that extends over about 100 km in the mean wind direction, and the accumulated deposition downwind from the source exhibits a long tail spanning over five orders of magnitude.