Differential weed seed removal in dryland cereals

Weeds persist in rain-fed cereal fields in NE Spain, despite intense herbicide use and high seed removal rates by granivorous harvester ants. Herbicide resistance is involved, but certain weed species also appear to escape seed removal by granivores. To identify the mechanisms involved, we measured seed removal rates (three fields in 2010) and the timing of seed shed (one field in 2009 and three fields in 2010) and used an existing model, which integrates short-term rates of seed shed, burial and removal, to estimate long-term seed removal rates. Averaged over years, fields and weed species, the long-term seed removal rate was estimated at 72 % (range 46–100 %). Fifteen to 25 % of the seeds of Bromus diandrus avoided removal by being less attractive (low removal rates), and another 0–29 % escaped through crop harvest, which made seeds inaccessible to granivores. Similarly, 20–32 % of the Papaver rhoeas seeds escaped through crop harvest, while another 13–17 % escaped by burial into the soil (small seed size). Other species, such as Galium spurium or Diplotaxis erucoides , had no means of avoiding seed removal by harvester ants. In particular, the more troublesome weeds, such as B. diandrus , P. rhoeas and L. rigidum , combined herbicide resistance or tolerance with avoidance mechanisms against granivory.