Growth plasticity of conifers did not avoid declining resilience to soil and atmospheric droughts during the 20th century

Background: Plasticity in response to environmental drivers can help trees cope with droughts. However, our understanding of the importance of plasticity and physiological adjustments in trees under global change is limited. Methods: We used the International Tree-Ring Data Bank (ITRDB) to examine 20th century growth responses in conifer trees during (resistance) and following (resilience) years of severe soil and atmospheric droughts occurring in isolation or as compound events. Growth resilience indices were calculated using observed growth divided by expected growth to avoid spurious correlations, in which the expected values were obtained by the autoregressive moving average (ARIMA) model. We used high atmospheric vapour pressure deficit (VPD) to select years of atmospheric drought and low annual values of the Standardized Precipitation-Evapotranspiration Index (SPEI) to select years with soil drought. We acquired the sensitivities (i.e., the slopes of the relationships) by fitting the resilience indices as a function of environmental drivers, and assessed how these sensitivities changed over time for different types of drought events using linear mixed models. We also checked whether plasticity in growth responses was sufficient to prevent long-term trends of growth reductions during or after severe droughts. We acknowledge that by focusing on the response of surviving trees from the ITRDB we are potentially biasing our results towards higher resilience, as stand level responses (e.g., mortality) may result in lowered competition after the disturbance event. Results: Sensitivities of resilience to VPD and SPEI changed throughout the 20th century, with the directions of these changes often reversing in the second half of the century. For the 1961–2010 period, changing sensitivities had positive effects on resilience, especially following years of high-VPD and compound events, avoiding growth losses that would have occurred if sensitivities had remained constant. Despite sensitivity changes, resilience was still lower at the end of the 20th century compared to the beginning of the century. Conclusions: Future adjustments to low-SPEI and high-VPD events are likely to continue to compensate for the trends in climate only partially, leading to further generalized reductions in tree growth of conifers. An improved understanding of these plastic adjustments and their limits, as well as potential compensatory processes at the stand level, is needed to project forest responses to climate change.