The spatial distribution of nests of the harvester ant Messor barbarus in dryland cereals

The harvester ant Messor barbarus can be responsible for substantial losses of weed seeds in arable fields in NE Spain. The spatial distribution of nests can have consequences for biological weed control, because foraging intensities decline with distance from the nest. The probability that seeds will escape harvesting will be lower if nests occur regularly distributed. We here investigated 'large'-scale variability (up to 150 m), caused by habitat heterogeneity, and 'small'-scale spatial variability (up to 12 m), caused by interactions between colonies, in nest distribution in a 50 x 150 m area in a cereal field in NE Spain, in 2009 and 2010. Large-scale variability was present in the data, but could not be explained by elevation, distance to the nearest field edge, or interpreted as simple trends across the area. Small-scale interactions could successfully be described by a multi-type/hard core Strauss process model, indicating territoriality among nests. Exclusion and interaction zones were identified, with radii that were smaller for small than for large colonies, and smaller for 2009 than for 2010. There was close resemblance between the observed and fitted spatial structure up to a radius of 3-4 m. Large-scale spatial variability, but not small-scale interactions, may be responsible for the existence of areas with few or no nests, where weed seeds have a higher probability of escaping the ants and entering the seed bank. Identifying and understanding the factors that influence the large-scale trends is, therefore, essential for optimizing weed control.