In this study we used two regional non-host controls and found that, for the common period, the reconstructed outbreaks had high fidelity in terms of timing, duration and frequency (Fig. 2). Sources of inconsistency between the two reconstructions were associated with the start and end years of outbreaks, a broad problem with the outbreak detection method employed here due to lag effects between budworm defoliation and subsequent growth suppression (Thomson and van Sickle, 1980, Alfaro et al., 1982 and Swetnam and Lynch, 1989) and in the intensity of individual outbreaks.

For example, outbreak reconstructions using the regional lodgepole pine non-host show a higher intensity outbreak in the 1800s than the ponderosa pine non-host, while the reverse was true for the 1900s outbreak (Fig. 2b). We attribute GSK2118436 chemical structure these differences to the degree and type of climatic variability captured by each non-host (limitation 2 identified above), as well as the potential for local endogenous processes to be reflected in the year-to-year variation in the tree-ring series.

Using the longer regional ponderosa pine non-host chronology (Table 1), we identified 12 low-intensity WSB outbreaks over a 435-year period, or one outbreak approximately every 36 years. This finding is similar to those of Campbell et al. (2006) who identified 8 WSB outbreaks over a 300-year period or Gemcitabine mouse one Linifanib (ABT-869) outbreak approximately every 37 years in the southern interior of BC. While we identified low-intensity events when ⩾15% of trees recorded

an outbreak, Campbell et al. (2006) identified an outbreak when ⩾35% of trees recorded an outbreak (Table 5). The application of a minimum threshold can be effective at differentiating between low and moderate intensity outbreaks. However, the threshold itself is somewhat subjective, as it is not based on theoretical or experimental values. It is possible that the threshold minimum of 35% may be too conservative and exclude small and/or low intensity events (Ann M. Lynch, personal communication). Defoliation impacts are often highest among trees in the suppressed and intermediate height classes ( Alfaro and Maclauchlan, 1992), yet in our study (and others) canopy dominants were selected for reconstruction purposes to obtain the longest possible records. These individuals, however, may not be capturing the full impact of budworm feeding. When we increased the minimum threshold to 50% (moderate) the number of reconstructed outbreaks dropped to 5 (from 12) that on average lasted 11 years with a return interval of 64 years ( Table 5). While the duration of low and moderate intensity events were similar (15 versus 11 years), the return interval increased two-fold ( Table 5), which is consistent with return intervals reconstructed for WSB outbreaks in the Colorado Front Range ( Ryerson et al., 2003).