9 and 49.2 cm in diameter. Within each of the stands sampled, four sampling plots were set up: an “edge plot” (EP) of 20 neighboring trees was established along the stand edge and three “interior plots” (IP1, IP2, IP 3)

of 20 neighboring trees were established within the heart of the stand, 25 m apart, in a cross-shaped design (Fig. 1B). Tree density decreased with increasing stand age. When stand density was very low (e.g. in 7 old stands) only 10 trees per plot were sampled to make sure that sampled plots were small enough to be homogeneous in terms of site conditions. AZD2281 In spring 2005, tree height was measured for a subsample of 29 trees per stand, corresponding to all 20 trees from one of the inner plots plus the three largest trees of the other three plots. Diameter at breast height was recorded for all trees of each plot. PPM population density (number of nests/ha) was calculated from the number of nests per sampled tree, the number of trees sampled, tree density and the area of the sampled plots. In total, 11,353 pine trees were included in this analysis (see Samalens, 2009 for further details). Egg batches were obtained from a laboratory rearing program in spring 2011. Details of the method used have been reported elsewhere (Castagneyrol et

al., 2014). Fifty egg batches were distributed between five pine stands, in which two trees were selected at random at each of five different distances this website from the stand edges (0, 2, 6, 8 and 16 m). A single egg batch was attached to each tree, on a branch at the base of the tree crown. Sentinel egg batches were protected against predators and parasitoids with a fine mesh (0.05 × 0.05 cm),

to ensure that any deaths were due to abiotic factors only. One of the two egg batches exposed at each distance from the edge was associated with a Hobo® data logger (Fig. 2). Temperatures were recorded at 30-min intervals, from the start of the experiment until the end of the egg hatching period Sclareol (i.e. 50 days later). The egg batches were removed at the end of August and egg mortality was determined, as a percentage, in the laboratory (see Castagneyrol et al., 2014). The data for this experiment were recorded as dataset 2. Analyses were carried at both the plot and tree scales. The number of nests per hectare, stem density and aspect were determined at stand scale. These variables were therefore included in models with stands as replicates. Tree height and diameter, and the presence/absence of nests on sampled trees were tree-specific attributes and were analyzed in models with trees as replicates. All statistical analyses were performed with R software (R Core Team, 2012). Generalized linear models were used to assess the effect of stand characteristics on PPM infestation in the 145 independent sampled stands. All stand characteristics (age, stem density, mean tree diameter and height) were strongly correlated (all pairewise correlations with |r| > 0.84 and P < 0.001).