Stand structure may mediate drought effects
Although most evidence suggests that surface air temperatures are the predominate influence on D. rufipennis dynamics, drought stress should not be dismissed entirely as a determinant in patterns of outbreak initiation and progression at certain scales, particularly when interacting with stand structure. Indeed, a number of other studies onD. rufipennis mortality have investigated how certain aspects of stand structure have affected percentage mortality across landscapes (e.g., Doak 2004; Hart et al. 2014b; Bakaj et al. 2016; Temperli et al. 2014). These studies have largely found tree size to be the best predictors of tree mortality, whereas stand structure has been of lesser importance (Bakaj et al. 2016). Our study confirms that large trees tended to die earlier in the outbreak, signaling beetles preferred these large trees (Appendix S1: Fig. S3). From a landscape-level perspective,D. rufipennis will encounter stands with a varying amount of large host trees. Once a bark beetle population encounters and occupies a stand, the subsequent timing of death of P. engelmannii may be based on tree size or drought-sensitivity that impacts host defenses, but the importance of each will likely depend on the stand characteristics (Goodsman et al. 2018).
Across the six stands we investigated, drought sensitivity was similar among early-dying trees but different in late-dying trees in some of the stands (Fig. 5; Appendix S1). We hypothesize that, in mature stands where there was an abundance of large trees, the difference in timing of death tended to have been determined in part by the sensitivity of ∆13C to drought or in this case CMD (Fig. 5; Appendix S1: Table S5; Appendix S1: Fig. S6). However, in stands comprised of a mix of large and small P. engelmannii (Appendix S1: Fig. S6), the large trees preferred by beetles were less abundant, thereby concentrating the intensity of beetle attack, and subsequently overwhelming hosts regardless of differences in drought sensitivity or drought stress among trees. This interpretation would be consistent with previous findings of small trees being attacked later during an outbreak after the large trees have been exhausted (e.g., Massey and Wygant 1954; Holsten 1984; Dymerski et al. 2001). The difference in individual timing of mortality on a site-by-site basis highlights the unique stand-specific progression of D. rufipennis outbreaks that cannot be expected to be consistent across a heterogeneous forest landscape. In stands with abundant large trees, drought sensitivity, as it influences tree susceptibility, affects timing of death. However, in stands with fewer large host trees, a limited density of large trees for beetles to choose from causes the best predictor of timing of death to be tree size, consistent with the findings of Temperli et al. (2014) and Bakaj et al. (2016). More extensive research using our iso-demographic approach will be necessary to address whether variability in drought stress across a landscape may lead to initial hot spots of beetle activity. Ideally, this would be paired with a remote sensing approach, such as shown by Hart et al. (2017), as a guide for directing where to establish tree-ring isotope sampling across a landscape.