*: outlier in year 1960 removed
Fig. 1 Conceptual diagram of the relationships that possibly influence D. rufipennis outbreaks with particular attention to factors that could influence whether drought stress has played a role in weakening tree resistance to beetle attacks. Box colors; purple represents bark beetle population dynamics across a landscape, blue represents factors affecting water supply, pink represents factors affecting water demand, orange represents regional to local factors driving drought stress, green represents ecophysiological characteristics that vary among trees, gray represents stand- site- or microsite-factors that can modify drought stress and/or likelihood of bark beetle host selection, white represents the likelihood of survival or death for a given timestep during a bark beetle outbreak. Relationship directionality is indicated by symbols in arrows. The shading of arrows reflects the degree to which the relationship has been previously demonstrated, with an emphasis on D. rufipennis and its hosts across North America; dark arrows are more strongly established and lighter arrows are more weakly established or in some cases simply hypothesized. Specific aspects of this system that our study addresses are given in boldface type.
Footnotes: 1Macroclimate evaporative demand is influenced primarily by warm season temperatures and secondarily by cool season precipitation and temperature; by their influence on snowpack persistence into spring or summer months.2Meteorological drought severity is quantified in this study using mean monthly climatic moisture deficit (CMD) across June to August. 3In montane forest of the Rocky Mountains, forest structure can have a strong influence on soil water via snow interception and ablation; or conversely, how the presence of gaps in forests can locally increase snowpack (Hart and Lomas 1979; Hubbart et al. 2015). 4Drought stress is quantified in this study by the sensitivity of latewood carbon isotope discrimination to CMD.5Bark beetle populations, including D. rufipennis , are influenced by many factors not represented here, but likely the single greatest effect is how growing season temperatures modify the proportion of beetles that can reproduce in 1- vs 2-yr lifecycles; for more details see (Hansen et al. 2001; Hansen and Bentz 2003; Berg et al. 2006; Raffa et al. 2008; Bentz et al. 2010).6Stand structure, in this case, reflects the relative abundance of large trees within a stand that D. rufipennis prefer if tree vigor is low enough and/or if beetle populations are high enough to overwhelm tree defenses (Massey and Wygant 1954; Hard 1983, 1985; Doak 2004). 7Drought can affect tree resistance to bark beetles through various mechanisms; for more details see Anderegg et al. (2015). 8Many traits factor into tree defense; for P. engelmannii , the frequency of traumatic resin duct formation appears to be critical for survival (DeRose et al. 2017).9For this study we use the timing of death within a stand as a surrogate for tree/host resistance to bark beetles.
Fig. 2 Conceptual diagram displaying a demographic sampling approach designed to contrast trees that died early during an epidemic (red bars), to those that died late during an outbreak (blue bars). This approach specifically allows for comparisons of early-dying trees that had lower resistance to incipient bark beetles populations (i.e., building from endemic to outbreak), versus late-dying trees from the same stands that resisted pressure by bark beetle attacks for a longer time. If drought stress contributed to the initiation of a beetle outbreak by constraining carbon gain available for defenses, sensitivity to drought stress should be greater in early-dying compared to late-dying trees.
Fig. 3 The six sites where tree cores were collected on the Markagunt Plateau in southern Utah are shown as yellow circles overlaid on a regional map showing the extent of canopy P. engelmanniimortality (red), P. tremuloides or mixed P. tremuloides -conifer (green) meadows (pink), lava fields and other non-vegetated lands (dark purple and white, respectively) are after DeRose et al. (2011). A broader context for the sampling is shown by maps of climatic moisture deficit (CMD; after Wang et al. 2016) for the period of 1961-1990 (left) and projected for 2050 (right) across the distribution of P. engelmannii .
Fig. 4 Predicted relationship between ∆13C values and climatic moisture deficit averaged across June, July, and August (CMD JJA) colored by early/late timing of mortality from repeated measures mixed effects model showing no significant interaction between timing of death (early vs. late) and CMD JJA (model estimate = 0.000268,p = 0.76; Appendix S1: Table S4). Dots represent partial residuals with colors indicating the timing of mortality: red dots/line indicate early-dying trees; blue dots/line indicate late-dying trees.
Fig. 5 Visualization of relationship between raw ∆13C values and climatic moisture deficit averaged across June, July, and August (CMD JJA) for the six sampling sites individually (noted by the three letter code at the top of each panel; Table 1). Linear regression lines have been added for ease of interpretation. Dots represent actual data points with colors indicating timing of mortality: red and blue points/linear regression lines represent early- or late-dying trees, respectively. Gray bands represent confidence intervals.