Increment core preparation for stable isotope analyses
Six sites were chosen for further analyses based on the abundance and quality of the increment cores. The date of last year of growth was verified for every core using visual crossdating and COFECHA software (Holmes 1983; Yamaguchi 1991). Cross-dated cores were used to identify the last year of growth which indicated when the tree died. The resulting distribution of death years identified trees that died early, when beetle population levels were building, or later during outbreak population levels (Fig. 2; Appendix S1: Fig. S1). Specifically, early- and late-dying trees were identified as those that died during years occupying the respective tails of the death-year distribution within each stand, after excluding outliers that may have died previous to significant beetle activity (Pettit 2018; Appendix S1: Fig. S1). Cores earmarked for stable isotope analyses were re-measured to include earlywood- and latewood-specific growth back to 1930 using a Velmex measuring system. Annual basal area increment (BAI) was calculated for each tree from the outside toward the pith assuming estimates of the diameter inside the bark from the diameter outside the bark (Myers and Alexander 1972), and by estimating areal growth from annual rings could be generalized as concentric circles. Trees identified for stable isotope analyses were determined on a site-by-site basis to obtain the greatest separation of early- and late-dying trees that included at least five dead trees within each group, and to ensure a large enough sample mass for stable isotope analyses (Table 1; Appendix S1: Fig. S1).
We composited the latewood of tree-rings across a site and within timing categories (i.e., each site had two sets of samples for each year that were distinguished by trees that died early versus late). Compositing rings by year for stable isotope analyses assures sufficient cellulose for stable isotope analyses when rings of trees are very small (e.g., Tardif et al. 2008; Voelker et al. 2014; Csank et al. 2016; Ratcliff et al. 2018) and typically yields an expressed population signal greater than 0.85 (Leavitt 2010). Latewood was excised from each ring to better isolate moisture stress during the summer months (Parker and Henoch 1971; Faulstich et al. 2013). Wood was ground to a coarse powder using a Wiley mill before extracting to holocellulose (Leavitt and Danzer 1993), homogenized using a 500-W ultrasonic probe (Laumer et al. 2009), and weighed into tin capsules on a microbalance. Samples were analyzed on a Thermo Delta V Advantage isotope ratio mass spectrometer at the Newell Stable Isotope Laboratory located at Utah State University (http://www.usu.edu/geo/newell/Newell_Website/Stable_Isotope_Lab.html). Carbon isotope composition, is expressed using “delta” notation as δ13C = (Rsample / Rstandard - 1) × 1000, where δ13C is the molar ratio of heavy to light isotopes and Rstandardis Vienna Pee Dee Belemnite (VPDB). Precision across replicate13C analyses of cellulose was 1σ = 0.12 ‰. Stable isotope analyses extended back to the tree-ring formed in 1960 for all sites to include a common period of at least three decades of inter-annual climatic variability leading up to the D. rufipennisoutbreak. Following Farquhar et al. (1989), cellulose δ13C data were converted to Δ13C as: Δ13C = (δ13Cair - δ13Cplant )/(1 + δ13Cplant / 1000), where δ13Cair values for each year followed those given in McCarroll and Loader (2004).