A new drought tipping point for conifer mortality

Predicting future changes of forest vegetation in response to warming over the next century is important for understanding future ecosystem and economic services and for planning approaches to mitigate negative impacts of climate change. A first step in such prediction is identifying meteorological drivers of tree mortality in current forests. Semi-arid forests of the Southwestern US are a model system for research about tree mortality because they are highly vulnerable to climate-warming-induced drought. Unusually high amounts of tree mortality have already occurred in these forests during recent droughts (van Mantgem et al 2009, Allen et al 2010). In the Southwestern US, the extreme drought of 2002 is an example of the warm 'global climate-change drought' that is expected to increase in frequency in the next century in many global forests (Breshears et al 2005). Consequently, many investigations of approaches for predicting drought-induced tree mortality focus on forests of the Southwestern US (McDowell et al 2013, Williams et al 2013). Huang et al advance these investigations by developing a new approach to predicting tree mortality during drought.

The central premise of Huang et al's approach is that tree death is likely when drought is so severe that no annual radial growth occurs. It is well known in dendrochronology that water availability is a strong control on annual ring width of trees in semi-arid forests (Fritts 1976). Trees experiencing extreme drought stress often do not form rings near the base of the bole where rings are typically sampled ('missing rings'). Ring width is the end product of numerous physiological processes which are sensitive to drought, and is an index of tree carbon balance because carbon allocation to radial growth is a lower priority within the tree than allocation to other carbon sinks, such as growth of leaves and fine roots, stem elongation, and tissue respiration (Waring 1987). Conifer trees with frequent narrow or missing rings typically have lower quantitative resin defenses against bark beetles (McDowell et al 2007, Kane and Kolb 2010), which are strong agents of tree mortality during severe drought. Trees that eventually die during drought typically have more frequent narrow or missing rings before their death than trees of the same species that survive drought (Kane and Kolb 2014, Macalady and Bugmann 2014). Thus, quantification of drought characteristics associated with zero radial growth of trees has potential for predicting tree death.

Huang et al take advantage of the large amount of tree-ring data available from the International Tree-Ring Data Bank for two dominant tree species of the Southwestern US, ponderosa pine and pinyon pine, and data for the standardized precipitation–evapotranspiration index (SPEI), which is precipitation minus potential evapotranspiration (Vicente-Serrano et al 2010). They show that average annual radial growth of ponderosa and pinyon pines is zero when SPEI is −1.64 between the previous September and July of the subject year. They hypothesize that this threshold value of −1.64 for SPEI describes the tipping point of meteorological drought that leads to tree mortality. They test this hypothesis by using the SPEI threshold to predict locations of high mortality of ponderosa and pinyon pines in the Southwestern US during the 2002 global-climate-change drought. Locations of tree death during 2002 were based on satellite-measured light spectral data (normalized difference vegetation index), a well-established approach for detecting large amounts of tree mortality over landscapes. Locations of tree mortality predicted with the SPEI threshold were generally consistent with locations shown by the satellite-measured light spectral data. This result indicates potential for using the SPEI tipping point approach to predict future drought-induced tree mortality in the Southwestern US and other regions. The approach's simplicity is a strength, as calculation of the SPEI tipping point requires only tree-ring chronology and meteorological data for a region.

The approach developed by Huang et al is a step forward in predicting tree mortality during drought, yet the approach has potential limitations that should be addressed in future investigations. First, the value of SPEI for a specific location can be uncertain because of limited availability of climatic data, and because values of SPEI in semi-arid regions often vary over different approaches for calculating potential evapotranspiration (Begueria et al 2014). Second, regionally gridded SPEI does not account for dissimilarities in tree response to water stress caused by local variations in soil properties (Looney et al 2012, Peterman et al 2012), rooting depth, access to ground water, and microclimate. Third, the approach does not fully encompass all drought-related causes of tree mortality. For example, intense forest fires kill many trees during drought when dry fuels, high winds, and ignition coincide. Tree death during wildfire is determined by numerous factors beyond annual radial growth, such as fuel loads, fire intensity, bark thickness, and tree size (McHugh and Kolb 2003). Fourth, tree mortality processes and the importance of low radial growth as an early warning signal of drought-induced mortality can be species specific in forest types that include multiple tree species (Kane and Kolb 2014, Camarero et al 2015). Because no single prediction model of the process of tree mortality during drought is perfect, efforts at predicting future tree mortality should continue to move towards multi-model simulations (McDowell et al 2013), following the lead of climate modelers who composite results from many models (IPCC 2013). The tipping point approach described by Huang et al has much potential to contribute to these efforts.