For this analysis we pooled the number of individuals for each combination of harvest treatment× sampling year and harvest treatment× position within the stand. Rarefaction curves for each of these vectors was then derived using the rarefy function in the vegan package in R 2.12 (R Development Core Team, 2011). We evaluated overall changes MEK pathway in beetle composition using multivariate regression tree analysis (De’ath 2002) using the mvpart package in R 2.12 (R Development Core Team, 2011). We square-root transformed beetle catch rates an aggregated data matrix (120 samples× 42

species) of catch rates (beetles/day) for a sum of squares multivariate regression tree analysis (ssMRT), where harvesting treatment,

year, and location within machine corridor, partial cut retention strip or uncut vegetation strip were predictor variables. We selected selleck products a final regression tree using cross-validation (based on 1000 iterations). We collected 6692 beetles representing 42 ground beetle species over both years. Overall catch rates were lower in all harvested treatments as compared to uncut stands (Table 1 and Table 2). Mean catch rates in clear cuts during 2009 and 2010 were 19% and 23% of those from uncut stands respectively. Mean catch rates in 2009 and 2010 within shelterwoods were 42% and 36% and in multicohort stands 29% and 33% as compared to uncut stands (Table 1 and Fig. 2a). Overall catch Microbiology inhibitor rates increased in 2010 as compared to 2009 across all cutting treatments as indicated by Wald t-tests ( Table 2 and Fig. 2a). Within shelterwoods in 2009, catch rates in machine corridors were higher than in uncut vegetation strips ( Fig. 2b and Table 2). We did not observe a similar trend in for multicohort treatments. Differences in species richness were greater among harvesting treatments than they were among individual sampling years

(Fig. 3a). Clear cuts had the highest species richness while uncut stands had the lowest species richness in both sampling years. Shelterwood and multicohort stands had similar species richness and fell between clear cuts and uncut sites. However, differences in sampling position within a harvest treatment were larger than differences between harvest treatments, particularly for shelterwood and multicohort stands, where within stand-heterogeneity was higher than either clear cut or uncut stands (Fig. 3b). In both shelterwood and multicohort treatments, the machine corridor treatments had lower species richness than partial cut strips or uncut vegetation strips and were similar to uncut stands in terms of the estimated number of species present. Changes in ground beetle assemblages were best characterized using a ssMRT with 7 terminal nodes. This model explained 36.3% of the total variance within the ground beetle assemblage.