Exploring the impacts of unprecedented climate extremes on forest ecosystems: hypotheses to guide modeling and experimental studies

. Climatic extreme events are expected to occur more frequently in the future, increasing the likelihood of unprecedented climate extremes (UCEs) or record-breaking events. UCEs, such as extreme heatwaves and droughts, substantially affect ecosystem stability and carbon cycling by increasing plant mortality and delaying ecosystem recovery. Quantitative knowledge of such effects is limited due to the paucity of experiments focusing on extreme climatic events beyond the range of historical experience. Here, we present a road map of how dynamic vegetation demographic models (VDMs) can be used to investigate hypotheses surrounding ecosystem responses to one type of UCE: unprecedented droughts. As a result of nonlinear ecosystem responses to UCEs that are qualitatively different from responses to milder extremes, we consider both biomass loss and recovery rates over time by reporting a time-integrated carbon loss as a result of UCE, relative to the absence of drought. Ad-ditionally, we explore how unprecedented droughts in combination with increasing atmospheric CO 2 and/or temperature may affect ecosystem stability and carbon cycling. We explored these questions using simulations of pre-drought and post-drought conditions at well-studied forest sites us-ing well-tested models (ED2 and LPJ-GUESS). The severity and patterns of biomass losses differed substantially be-tween models. For example, biomass loss could be sensitive to either drought duration or drought intensity depending on the model approach. This is due to the models having different, but also plausible, representations of processes and interactions, highlighting the complicated variability of UCE impacts that still need to be narrowed down in models. Elevated atmospheric CO 2 concentrations (eCO 2 ) alone did not completely buffer the ecosystems from carbon losses during

. Description of simulation treatments of hypothetical droughts from a 'baseline' case (i.e., no drought treatment) to unprecedented climate extremes (UCEs). Varying drought intensity (precipitation removal) from 5% to 100% removal, in increments of 5%, over drought durations of either 1, 2, or 4 years in length. To explore climate change response, we repeated the drought treatments and increased temperature only (+2K over ambient), eCO2 concentration to 600 ppm and 800 ppm, and increased temperature and eCO2 (+2K 600 ppm; +2K 800 ppm) and compared to the reference simulation.

Meteorological data and initial conditions used to drive ED2 and LPJ-GUESS:
Necessary meteorological drivers for ED2 and LPJ-GUESS include incoming radiation (short-wave and long-wave), air temperature, humidity, and pressure, precipitation and wind speed at sub-daily scale. In-situ meteorological data for Palo Verde is only available since 2008. Using the short-term data as the control climate can lead to biases in ecosystem states and high-frequency cyclic ecosystem dynamics before applying UCEs. Therefore, we use re-analysis data (1970 to 2012) at 0.5 degree resolution from Princeton Global Forcing dataset (Sheffield et al., 2006), and was recycled repeatedly for the Palo Verde simulations.
In-situ meteorological data for EucFACE were obtained from a dataset previously compiled for a simulation study of the EucFACE experimental site (Medlyn et al., 2016).
Daily time series of air temperature, precipitation, downward shortwave radiation and photosynthetically-active radiation for 1992-2011 were extracted from the 1 × 1° grid cell encapsulating the site from the Princeton Global Forcing data set (Sheffield et al., 2006). This 20-year time series was recycled repeatedly to force the simulations. For both sites, the baseline simulations were initialized as a near-bare-ground situation, with small amount of tree seedlings equally from each PFT. The baseline spin-up lasted for 100 years (ED2) or 780 years (LPJ-GUESS) using recycling natural climate variability as described above.

Additional knowledge gaps
With so many compounding interactions contributing to ecosystem resistance, impact, and recovery from droughts, there are still knowledge gaps in compounding processes like response to concurrent or repeated extremes, lag affects, or cascades.
However, it is difficult for planned experiments to include multiple stressors and very extreme environmental conditions, thus making it challenging to assess all impacts and whether biological ecosystem components (e.g. plant-soil, plant-atmosphere, C:N, respiration-photosynthesis) will remain coupled under extreme conditions. Unfortunately, there is a lack of data on key characteristics and responses to UCEs, which greatly impacts our understanding and ability to predict ecosystem responses to such events. In addition to the general understanding of ecosystem responses to UCEs, we describe some issues which can lead to compounded and notable responses to UCEs. However, the sensitivity of ecosystems to repeated or combined extremes as well as their ability to acclimate remains generally unclear.

Concurrent or repeated extremes:
Lag effects: Ecosystems must re-establish resilience following an extreme event, but the time needed for a system to do so is difficult to predict due to unanticipated lag effects of extreme events on ecosystem functioning. Previous drought exposure have been linked to long-term mortality of forest trees in the eastern US (Berdanier and Clark, 2016) and to decreased short-term leaf survival in response to additional extreme events (Dreesen et al., 2014) suggesting a time period following disturbance where forests are particularly susceptible to additional stressors.. Also, transgenerational effects of drought on leaf stoichiometry (C:N) with direct consequences for ecosystem-level C storage has been detected in perennial plant seedlings (Walter et al., 2016). However, such lag effects are generally difficult to study and are therefore generally poorly understood.
Cascades: Despite our understanding that feedbacks among ecosystem components are likely to impact environmental functioning along multiple pathways and ultimately the terrestrial carbon cycle (Reichstein et al., 2013), empirical studies of cascades are rare (but see Jentsch et al., 2011 for plant-soil measurements). In particular, our ability to predict response thresholds is poor, and additional uncertainty in predicting ecosystem responses occurs because thresholds can be passed at any organizational level within an organism (e.g. leaf, individual, plant community levels; Frank et al., 2015;Gutschick and BassiriRad, 2003) and among organisms (e.g. different sensitivities of soil fungi vs. bacteria to different disturbances; Muhr et al., 2009).

Secondary disturbance:
The combination of extreme events and secondary disturbances may increase the susceptibility of carbon loss from ecosystems (e.g., Hicke et al., 2016).
For example, extreme droughts and heatwaves promote forest fires by increasing both fuel flammability and lightning strike frequency (Wendler et al., 2011). Substantial forest damage can also occur through phenological changes of forest vegetation or biotic pests or pathogens. Warm winters can weaken wintertime pest mortality and increase pest growth rates (Bale et al., 2002;Cornelissen, 2011), shifting insect phenologies and triggering outbreaks. Water-stressed trees are susceptible to foliar and woody damage from forest insect and pathogens (Jactel et al., 2012, Flowers andGonzalez-Meler, 2015;Kolb et al., 2016), and combined drought-stress and insect outbreaks can cause massive forest die-off (Allen et al., 2010; Anderegg et al., 2015b) leading to unprecedented levels of tree mortality such as those recorded in western North America (Breshears et al., 2005;Raffa, 2008). Warm winters may advance the leaf-out of deciduous species (Parmesan and Yohe, 2003), increasing their susceptibility to secondary disturbances, such as frostdamage (Gu et al., 2011;Polgar and Primack, 2011). Studies have directly linked such coupled disturbances to a decrease in seasonal C accumulation and to shifts in the development of reproductive structures (Augspurger, 2009), but the global consequences of such phenological shifts and coupled-disturbances has not been quantified (?).
Thresholds: Large-scale ecosystem studies are costly and so rarely include gradients or multiple treatment levels (but see Kreyling et al., 2014). Therefore our ability to detect and understand tipping points is still very limited. Models could play a significant role in identifying 'zones of sensitivity' that can be targeted in field experiments.