Critical review of drought stress vs temperature in driving bark beetle outbreaks
Disentangling the effects of temperature from drought stress on bark beetles has been a difficult task because many investigations into climatic influences on outbreaks have been retrospective and are lacking in direct measurements of tree drought stress (Kolb et al. 2016). Of the many species of bark beetles, only Ips confusus andDendroctonus brevicomis unequivocally influence tree susceptibility to attack when drought stress is experimentally induced (Gaylord et al. 2013; Kolb et al. 2019). In contrast to these multivoltine-capable bark beetle species that occupy and kill their hosts within forests that occupy consistently dry growing seasons (Breshears et al. 2005; Raffa et al. 2008; Fettig et al. 2019), D. rufipennis hosts occupy cold and wet, montane, boreal and subarctic forests where evidence has clearly demonstrated how temperatures exert strong and direct effects on beetle life cycles (shift from semivoltine to univoltine; Hansen et al. 2001; Hansen and Bentz 2003; DeRose et al. 2013).
During the D. rufipennis outbreaks that peaked in the early 1990s across south-central Alaska, Berg et al. (2006) concluded that decades of warmer than average temperatures were directly affecting bark beetle populations, but that the potential for drought to have contributed to the outbreaks could not be ruled out. The outbreak we investigated was characterized in a similar manner to Berg et al. (2006), and it was concluded that periods of regional drought were associated with pastD. rufipennis outbreaks (DeRose and Long 2012a). However, like most other retrospective studies of bark beetle outbreaks, prior to the instrumental record, DeRose and Long (2012a), had no means to differentiate among how precipitation deficits versus warmer temperatures may have contributed to reconstructions of meteorological drought severity (i.e., PDSI). Additionally, regional tree-ring reconstructions of PDSI across this region can be biased toward sites with greater climate sensitivity (Klesse et al. 2018), which may better reflect mid- to low-elevation forests where cloud cover and precipitation from summer convective storms and/or monsoon rains are lower and less consistent compared to the high elevations that P. engelmannii occupies.
D. rufipennis outbreaks identified from documentary records across the southern Rockies of Utah and Colorado have been found to be driven by three sets of variables, listed in order of importance: winter temperatures, late summer and fall temperatures, and annual PDSI (Hebertson and Jenkins 2008). Since PDSI is strongly influenced by temperature (Sheffield et al. 2012), it can be concluded from the results of Hebertson and Jenkins (2008) that the primary cause ofD. rufipennis outbreaks has been temperature and that there may be indirect evidence for drought stress having had a smaller but detectable influence. It has also been shown that meteorological drought, often in association with the Atlantic Multi-decadal Oscillation (AMO) have induced D. rufipennis outbreaks in Colorado (Hart et al. 2014a). However, the evidence is equivocal whether drought stress can be differentiated from temperature effects on outbreaks for that region because precipitation did not show significant effects for any season (i.e., Hart et al. 2014a, Appendix B). In another retrospective study, D. rufipennis -killed trees in Alaska had higher correlations between interannual tree-ring δ13C and spring and summer temperatures when compared to surviving trees from the same areas, but no differences in precipitation responses were detected (Csank et al. 2016). Similar to the conclusions of Hart et al. (2014a), the results of Csank et al. (2016) fit within the narrative that bark beetles responded to drought, but only to the extent that meteorological drought is a function of water balance driven by precipitation and temperature. However, these results can be viewed through a lens that distinguishes between meteorological drought (see Fig. 1). Indeed, in our study, there was a lack of significant relationships between ring-width growth parameters and various meteorological drought metrics across all of our sites (Appendix S1: Table S2). These results underscore that tree- or stand-level drought stress may be independent or only weakly related to regional meteorological drought conditions for these high elevation spruce forests. Overall, these previous studies indicate that beetle-killed trees were more responsive to temperature than surviving trees, which agrees with our primary conclusion that drought stress played little role in determining patterns of tree mortality.
There may be a distinct role for drought stress in the coldest boreal or montane forests whereby water, still frozen in tree stems or in soils, can limit water transport and photosynthesis and thereby promote susceptibility to D. rufipennis attack (Hard 1987; Bowling et al. 2018). However, drought stress caused via this mechanism occurs at local scales related to topographic variation, and therefore cannot explain the large D. rufipennis outbreaks that have swept across manyP. engelmannii landscapes of the Rocky Mountains in recent decades. Overall, the body of literature on climatic drivers of D. rufipennis outbreaks has found ample evidence that outbreaks were associated with periods of prolonged warmer air temperatures that may have been associated with meteorological drought, whereas evidence is absent or of minor importance that host drought stress increased tree or forest susceptibility leading up to or during outbreaks.