At least it is a dry cold: the global distribution of freeze–thaw and drought stress and the traits that may impart poly-tolerance in conifers

Abstract Conifers inhabit some of the most challenging landscapes where multiple abiotic stressors (e.g., aridity, freezing temperatures) often co-occur. Physiological tolerance to multiple stressors (‘poly-tolerance’) is thought to be rare because exposure to one stress generally limits responses to another through functional trade-offs. However, the capacity to exhibit poly-tolerance may be greater when combined abiotic stressors have similar physiological impacts, such as the disruption of hydraulic function imposed by drought or freezing. Here, we reviewed empirical data in light of theoretical expectations for conifer adaptations to drought and freeze–thaw cycles with particular attention to hydraulic traits of the stem and leaf. Additionally, we examined the commonality and spatial distribution of poly-stress along indices of these combined stressors. We found that locations with the highest values of our poly-stress index (PSi) are characterized by moderate drought and moderate freeze–thaw, and most of the global conifer distribution occupies areas of moderate poly-stress. Among traits examined, we found diverse responses to the stressors. Turgor loss point did not correlate with freeze–thaw or drought stress individually, but did with the PSi, albeit inverse to what was hypothesized. Leaf mass per area was more strongly linked with drought stress than the poly-stress and not at all with freeze–thaw stress. In stems, the water potential causing 50% loss of hydraulic conductivity became more negative with increasing drought stress and poly-stress but did not correlate with freeze–thaw stress. For these traits, we identified a striking lack of coverage for substantial portions of species ranges, particularly at the upper boundaries of their respective PSis, demonstrating a critical gap in our understanding of trait prevalence and plasticity along these stress gradients. Future research should investigate traits that confer tolerance to both freeze–thaw and drought stress in a wide range of species across broad geographic scales.

Conifers inhabit some of the most challenging landscapes where multiple abiotic stressors (e.g., aridity, freezing temperatures) often co-occur. Physiological tolerance to multiple stressors ("poly-tolerance") is thought to be rare because exposure to one stress generally limits responses to another through functional trade-offs. However, the capacity to exhibit poly-tolerance may be greater when combined abiotic stressors have similar physiological impacts, such as the disruption of hydraulic function imposed by drought or freezing. Here, we reviewed empirical data in light of theoretical expectations for conifer adaptations to drought and freeze-thaw cycles with particular attention to hydraulic traits of the stem and leaf. Additionally, we examined the commonality and spatial distribution of poly-stress along indices of these combined stressors.
We found that the locations with the highest values of our poly-stress index are characterized by moderate drought and moderate freeze-thaw, and most of the global conifer distribution occupies areas of moderate poly-stress. Among traits examined, we found diverse responses to the stressors. Turgor loss point (TLP) did not correlate with freeze-thaw or drought stress individually, but did with the poly-stress index, albeit inverse to what was hypothesized. Leaf mass per area was more strongly linked with drought stress than the poly-stress, and not at all with freeze-thaw stress. In stems, the water potential causing 50% loss of hydraulic conductivity became more negative with increasing drought-and poly-stress, but did not correlate with freezethaw stress. For these traits, we identified a striking lack of coverage for substantial portions of species ranges, particularly at the upper boundaries of their respective poly-stress indices, demonstrating a critical gap in our understanding of trait prevalence and plasticity along these stress gradients. Future research should investigate traits that confer tolerance to both freezethaw and drought stress in a wide range of species across broad geographic scales.

Highlights:
• Majority of conifer extent subjected to moderate poly-stress • Traits that correlated with drought stress also correlated with poly-stress

Introduction
Globally, coniferous forests experience harsh and stressful environmental conditions throughout most of the year. Plants in these ecosystems may be exposed to a range of weather extremes including freezing temperatures in the winter, high air temperatures in the summer, and low water availability seasonally. Transitions between frozen and thawed states (i.e., "freezethaw" cycles) and/or periods of drought can impair physiological functioning in a variety of ways, including photosynthetic decline, hydraulic dysfunction, and mortality (Allen and Ort 2001;Blackman et al. 2019;Choat et al. 2018;Huner et al. 1998Huner et al. , 2003Ivanov et al. 2001, Mayr et al. 2006Sperry 2003, 2006). Despite recent progress in characterizing stress mitigation and tolerance mechanisms to single stressors (e.g., drought, Choat et al. 2018;freezethaw, Guy 2003;Zanne et al. 2014), we currently lack a comprehensive understanding of the mechanisms by which evergreen conifers mitigate multiple (but not necessarily concurring) physiological stresses (i.e., "poly-stress"). In the context of a changing climate, it is particularly important that we understand how plants tolerate stressors that currently exhibit strong selective pressure, such as drought and freeze-thaw cycles, in order to improve our predictions of how our forest compositions may shift in the future.
Both drought and freeze-thaw cycles are expected to become more frequent in many coniferous forest regions due to rising air temperatures and shifting precipitation regimes associated with global climate change (Henry 2008; IPCC 2019). Global warming, which is greatest at higher latitudes, can lead to mid-winter warming events (Williams et al. 2015), increase the occurrence of rain relative to snow (Knowles et al. 2006), decrease total snow accumulation (Mote et al. 2005), and result in earlier spring snowmelt (Clow 2010). Increased variability in winter temperatures and reduced insulation from temperature fluctuations associated with lower snow cover can increase the frequency of freezing and thawing experienced by plants (Henry 2008). These conditions can also have cascading consequences throughout the growing season as reduced snow cover and, consequently, diminished and less prolonged snowmelt, limit available water for plants throughout the growing season (Hu et al. 2010;Williams et al. 2015). When combined with the elevated summer air temperatures and reduced summer precipitation predicted for some of these regions, many trees will likely experience population declines associated with these combined stresses (Mayr et al. 2006;Earles et al. 2018).
Freeze-thaw cycles and drought impose similar impacts on leaves and stems. In leaves, both freezing temperatures and drought can hinder photosynthetic capacity and lead to excess absorbed light energy that causes photoinhibitory damage and leaf death (Demmig-Adams and Adams 2006). In stems, both freeze-thaw cycles and drought disrupt the hydraulic network by inducing or propagating embolisms throughout the xylem (Tyree and Sperry 1989). Although the mechanism of formation differs, embolisms caused by both freeze-thaw events and drought reduce whole plant hydraulic conductance, which can limit plant functioning and growth (Anderegg et al. 2016;Kreyling 2010). Given that these stresses create similar dysfunction in plants, as well as the observations that evergreen conifers often experience freeze-thaw cycles and drought simultaneously in some habitats (Mayr et al. 2006) and that these stresses are likely to have compounding effects (Charrier et al. 2021), the survival and distribution of these taxa likely depends on a tolerance to both stressors (i.e., "poly-tolerance"). However, some studies have noted a distinct trade-off between drought and cold/freezing tolerance in woody taxa mediated by reduced wood density in frost-tolerant species (e.g., Rueda et al. 2017;Laanisto & Niinemets 2015), suggesting divergent trait coordination that promotes either freeze-thaw or drought tolerance across large geographic scales. Though some studies have systematically examined organ-level traits that could promote tolerance to multiple abiotic stressors across a wide range of species (e.g., Hallik et al. 2009;Rueda et al. 2017;Stahl et al. 2013), no study has quantified the degree of geographic overlap between freeze-thaw and drought stress to identify key areas where poly-tolerance to these stressors would be most adaptive. Additionally, we do not yet have a comprehensive understanding of what traits confer both freeze-thaw and drought tolerance, how widespread the coordination of these traits is across species, or how these traits might influence the survival and distribution of conifers worldwide.
In this review we aim to 1) examine indices of the worldwide spatial distribution of drought, freeze-thaw cycles (hereafter, "FT cycles"), and their combination, and 2) identify traits that may impart tolerance to drought and FT cycles and explore to what extent these traits overlap in evergreen conifers. In particular, we focused this review on the physiological, morphological, and anatomical traits of conifer leaves and stems. Although roots are also susceptible to freezing and drought stresses, comparatively little research has investigated the topic (but see Ambroise et al. 2020 for a review on frost resistance in crop roots). This review is not intended to comprehensively examine all aspects of winter or drought ecophysiology (but see Chang et al. 2021;Choat et al. 2018;McDowell et al. 2008;Sakai & Larcher, 2012). Other recent reviews have covered topics such as extreme winter weather events (Casson et al. 2019), winter climate change impacts on plant species composition, ranges, and phenology (Kreyling 2010), and the impacts of severe drought on global forest mortality (Allen et al. 2010;Anderegg et al. 2013;Clark et al. 2016). By identifying the global occurrence of FT and drought, as well as relevant traits that may enhance tolerance to these combined stressors, we aim to identify promising research avenues that will enhance predictions of conifer function, abundance, and distributions in a changing climate.
2. Assessing the worldwide prevalence of drought stress, freeze-thaw stress, and their combination Drought and freezing temperatures profoundly influence the global distribution of plant groups (Engelbrecht et al. 2007;Normand et al. 2009;Stuart et al. 2006;Stahl et al. 2014). The distribution and trait evolution of evergreen woody angiosperms can be sharply defined by exposure to freezing temperatures (Zanne et al. 2018), while maximum temperature and precipitation are more strongly associated with tolerance traits and distributions in conifers (Rueda et al. 2017). This suggests a role for overlapping stressors as a primary driver of conifer occurrence as opposed to temperature or precipitation singly. Furthermore, the impacts of absolute minimum annual temperatures on conifer traits and distributions are likely minimal small compared to those the impacts of the FT cycles that occur in cooler climates. Freezing temperatures alone can reduce photosynthetic capacity (Huner et al. 1998(Huner et al. , 2003Ivanov et al. 2001) and increase the hydraulic resistance in soil (Tranquillini 1976), while FT cycles can cause severe hydraulic dysfunction in xylem via embolisms (Hammel 1967;Sucoff 1969). Cold hardiness, studied extensively in conifers, often varies by genotype and is correlated with local minimum temperatures (e.g., Sebastian-Azcona et al. 2018), although foliage is commonly hardened to temperatures far beyond those minimums (Sakai 1960;Strimbeck et al. 2007;Wisniewski et al. 2018). However, there is no broad-scale relationship between minimum temperature and FT events (Fig. S1). As such, common ranking systems for cold tolerance or cold hardiness do not show any clear relationship with FT events (Fig. S1), thus providing an opportunity to investigate the spatial distribution and impact of FT cycles on plant function and ecology.
To visualize the global patterns of drought, FT cycles, and their combination, we used available global climate data to develop spatial indices of drought stress (Di, based on soil water content and vapor pressure deficit; comparison with other water availability metrics shown in  (Fig. 1C), which are instead a characterized by a high number of freeze-thaw days but low drought stress (Fig. 1A, B). It is worth noting, though, that some regions of the boreal forest do experience some drought stress, particularly in British Columbia, Canada, which has recently experienced severe heatwaves and droughts that have exacerbated widespread wildfires. For the majority of the boreal forests that experience little water stress, the main climatic driver of this biome type is likely associated with the stresses of repeated freezing and thawing, along with the extreme cold. At extreme degrees of drought, conifers are generally replaced by plants (often angiosperms) that exhibit succulence and CAM photosynthesis, or the rapid growth of desert annuals (however, see Larter et al. 2015Larter et al. , 2017, while at extreme FT stress, conifers are replaced by deciduous angiosperms. In locations with relatively extreme poly-stress, the vegetation type tends to be desert and xeric shrubland without conifers (Fig. 2L).
Importantly, the data used to quantify the distributions of drought, FT and polystress (e.g., Fig. 1) are from the recent past (See Supplemental Materials for details). As the global climate changes, shifting precipitation patterns will reduce soil moisture availability in some areas (Dai et al. 2018) and higher temperatures will increase the vapor pressure deficit broadly (Yuan et al. 2019). Furthermore, variations in the extent and location of polar vortices, along with milder winter temperatures in the arctic, could increase the number of FT cycles in regions of the northern hemisphere (Zhang et al. 2016B;Matsumura et al. 2021). Thus, the spatial area experiencing multiple stresses should increase in the future, as should the intensities of polystress.

Traits and stress tolerance
The apparent biome shifts away from conifers at the extreme abiotic stress levels in Figure 2 likely represent the physiological limits of conifers. However, which specific traits limit individual species within this broader distribution remains unclear. Figure 3 shows the occurrence records of multiple species from five genera along the axes of the FT and drought stress indices, illustrating that these taxa vary considerably in their distribution within our polystress space. Some species are absent in areas of drought stress but are present in areas of high FT stress (e.g., Podocarpus nubigenus, Fig. 2W) while others show the reciprocal pattern (e.g.
Pinus herrerae, Fig. 2R), suggesting a tradeoff. Indeed, we found a significant negative relationship between species' median positions on the drought and FT stress index (R 2 = 0.19, p < 0.001; data not shown). However, some species occur in areas subject to both stressors (e.g., Pinus halepensis, Fig. 2Q).
In the next sections we discuss specific hydraulic traits that could impart tolerance to Rosas et al. 2021), which may provide insight to mechanisms facilitating the tolerance of particular conifer species to both stresses. We test how these traits relate to drought and FT stresses using data compiled from databases and our own literature search (see Supplemental Materials). Relationships were analyzed using linear regressions. We checked the appropriateness of a linear model with a Shapiro normality test. All models that included LMA failed to meet the proper assumptions and data were log-transformed.

Traits that impart tolerance in leaves
Among the most well-established characteristics associated with tolerating drought stress are those that maintain turgor pressure in living cells. Positive turgor is crucial for proper metabolic functioning within cells. Turgor can be maintained under water deficits by lowering the turgor loss point (TLP), which can occur in multiple ways. First, the TLP can be lowered through a reduction in the cell's osmotic potential, which limits water loss. Lower TLPs associated with reductions in osmotic potentials have been studied extensively and there is considerable evidence that across biomes, more arid-adapted plants exhibit lower TLP values due to higher concentrations of osmotically active solutes (Bartlett et al. 2012 Conversely, no adjustment was observed in the elastic modulus of Pinus edulis leaves in the same study, nor did Bartlett et al. (2012) find evidence of elastic modulus adjustments in response to drought across a broad range of mostly angiosperm species. This discrepancy suggests not all species adjust TLP using the elastic modulus. Third, plants can reduce TLP by increasing the water content within the apoplast relative to that in the symplast (called the "apoplastic water fraction," Bartlett et al. 2012). This concentrates the solutes within the cell, reduces the osmotic potential, and consequently lowers the TLP (assuming the other components of cellular water relations remain constant). Regardless of how TLP is reduced or maintained, lower TLP typically results in a greater tolerance of water stress. The TLP has also been shown to be tightly coordinated with a large number of other traits that impart drought tolerance, such as the ability to maintain leaf hydraulic conductance (Brodribb and Holbrook 2006;Nardini et al. 2012;Nardini and Luglio 2014;Johnson et al. 2018;Yao et al. 2021), which is known to be an indicator of drought tolerance in conifers (Brodribb and Cochard 2009), and stomatal conductance (Brodribb and Holbrook 2003;Li et al. 2018) at low water potentials.
Extracellular ice formation is one of the most damaging aspects of FT events. The freezing point of water within the apoplast is higher than that of the cytoplasm, and consequently, intercellular water typically freezes first. Freezing lowers the water potential of the apoplast and draws water out from the living cells, which may lead to cellular dehydration and damage to the structure and function of the plasma membrane (Guy 1990(Guy , 2003. The amount of cellular dehydration that occurs with freezing is temperature-dependent because water potential declines with the temperature of the ice, which in turn reduces the water potential of the cytoplasm (Xin and Browse 2000). Thus, plants that are able to reduce the osmotic potential of living cells might experience less damage due to extracellular ice formation (e.g., Tsuga canadensis; Tyree et al. 1978). Indeed, substantial evidence shows that plants acclimated to cold temperatures have lower leaf osmotic potentials and also experience less freezing damage than plants not acclimated to cold temperatures prior to freezing (Kasuga et al. 2007;Charrier et al. 2013;Ting et al. 2014;Arias et al. 2015Arias et al. , 2017. Although a lower osmotic potential is generally associated with lower TLP, we found no significant correlation between FTi and TLP in our dataset (p = 0.74; Fig. 4B). One explanation for the lack of a relationship is that a lower TLP does not impart greater tolerance to higher FT stress. This may be true if a threshold exists below which reducing the TLP does not increase tolerance to FT cycles. Another explanation is that the TLP values we compiled report values that were only collected in the summer. Conifer TLP values have been shown to adjust throughout the growing season in response to drought Adjustment of the elastic modulus may also reduce damage associated with extracellular ice formation, as more rigid cell walls can prevent cell collapse (Scholz et al. 2012;Le Gall et al. 2015;Zhang et al. 2016). Finally, the ability to adjust the apoplastic water fraction while tissues are experiencing freezing temperatures may help plants avoid ice formation and tolerate lower temperatures (Goldstein et al. 1985;Arias et al. 2015Arias et al. , 2017. The extent to which this pattern holds in conifers or imparts tolerance to FT cycles (and not simply low temperatures) remains unclear (Grossnickle 1992). However, the most poly-tolerant conifers would likely exhibit the greatest ability to adjust osmotic, elastic, and apoplastic properties because doing so is associated with tolerance and/or avoidance to both drought and freezing. For example, the ability to seasonally adjust the apoplastic water fraction would allow summer increases during drought stress and non-growing season decreases to tolerate frequent freeze-thaw cycles. However, there are currently no data available to test if conifers adjust in this way, as the available data are heavily biased towards summertime measurements.
Given the theoretical importance of TLP in conifer leaf functioning under drought and FT, as well as empirical support showing that TLP is a key drought tolerance trait in angiosperms (e.g., Bartlett et al. 2012), we predicted conifers would also exhibit strong correlations between TLP and PSi. As discussed above for the individual stresses, the data we compiled indicated no relationships between TLP and Di or FTi (Fig. 4) expressed on a per mass basis (Nardini et al. 2012;Simonin et al. 2012). In habitats with lower water availability, leaves tend to become thicker and/or heavier for a given area (i.e., LMA increases). Mechanistically, this change is caused by more proximal responses to drought stress, such as increases in cell wall thicknesses to withstand more negative water potentials (Onada et al. 2018) and the need for thicker, longer-lived leaves in resource-limiting environments (Givnish 2002;Hodgson et al. 2011;Simonin et al. 2012). Although LMA generally increases across species in more arid environments, the relationship is fairly weak (Reich 2014), likely because considerable variation exists among species and clades that have evolved to fill different niches within a community in any given habitat (Bruelheide et al. 2018;Treurnicht et al. 2020).
As with plants that experience drought stress, plants adapted to colder annual temperatures also tend to produce leaves with higher LMA (Niinemets 2016;Gonzales-Zurdo et al. 2016;Jankowski et al. 2017). However, whether higher LMA is also associated with a greater number of FT cycles or how this trait may vary intraspecifically across the FT and drought poly-stress gradient is unknown. When we compared species' mean LMA to drought, FT, and poly-stress indices, we found that LMA increased with increasing drought (R 2 = 0.27; p < 0.001; Fig. 4D), exhibited no relationship with FTi (R 2 = -0.005; p = 0.63; Fig. 4E), and increased with increasing poly-stress (R 2 = 0.12, p < 0.001; Fig. 4F). This indicates that the response to poly-stress is likely being driven by the strong relationship between LMA and the drought index. However, LMA may have stronger relationships with other aspects of FT cycles that our index does not incorporate. For example, our FT index is based on the typical number of freeze-thaw cycles per year, but it does not account for temperatures prior to or after FT events. Evidence indicates that freezing events can be more damaging (and thus a stronger driver of trait evolution) if they exceed the rate of cold acclimation and/or occur during the deacclimation phase (Sakai 1960;Strimbeck et al. 2007Strimbeck et al. , 2015. It also remains unclear whether relatively minor morphological adjustments result in greater protection to more frequent FT cycles compared to increases in drought stress.

Traits that impart tolerance in stems
Research on traits that impart tolerance to drought and FT stress in stems has largely focused on maintaining xylem function. Loss of xylem function due to drought stress is caused by the propagation of air bubbles (embolisms) throughout the conduit network. The propagation of an embolism from one tracheid to a neighboring, functional tracheid occurs through the bordered pits on the walls between two tracheids. Theory and empirical evidence support that more drought resistant wood in conifers has a greater overlap between the impermeable portion of the pit membrane, the torus, and the pit aperture (Domec et al. 2006;Hacke and Jansen 2009;Delzon et al. 2010;Pittermann et al. 2010;Bouche et al. 2014;Song et al. 2022). As this overlap increases, the water in the functional tracheid must experience increasingly negative pressure to break the pit membrane seal and release a bubble (Cochard et al. 2009 The loss of hydraulic function due to FT-induced embolisms is well-documented in conifers (Sperry and Sullivan 1992;Sparks and Black 2000;Feild and Brodribb 2001;Sparks et al. 2001;Mayr et al. 2002Mayr et al. , 2003aMayr et al. ,b, 2007Sperry 2003, 2006;Mayr andZublasing 2009, McCulloh et al. 2011). In contrast with drought stress, inter-specific tolerances to FT events are much more directly dependent on tracheid dimensions. During FT cycles, gases that are dissolved in the xylem water will form bubbles due to their insolubility in ice (Hammel 1967;Sucoff 1969). Upon thawing, a bubble has two potential fates: it can dissolve back into the water or it can expand and form an embolism within the xylem (Pittermann and Sperry 2006).
The probability of the gas dissolving into the water depends on the size of the bubble, which in turn depends on the diameter of the conduit (and the xylem pressure potential; Pittermann and Sperry 2006). Wider diameter conduits will produce wider diameter bubbles, which will expand at less negative xylem pressures (as described by La Place's Law). Thus, narrower conduits are more resistant to embolism caused by freeze-thaw cycles, a trend that has been observed across conifers Sperry 2003, 2006;Willson and Jackson 2006). FT tolerance may also be related to pit characteristics. Theory suggests that pits with greater porosity facilitate the movement of larger particles that allow ice to nucleate at warmer temperatures (Lintunen et al.

2013). Thus, conifers characterized by pits with lower porosity would likely exhibit greater
tolerance to FT cycles. Despite the mechanistic link between stem P 50 and pit dimensions, we found no relationship between P 50 values and FTi values (R 2 = 0.01, p = 0.1; Fig. 5B).
The impact of wood density and capacitance on tolerating freeze-thaw cycles has not been explored. Theoretically, the high specific heat of water suggests that high water storage in stems may prevent the stem from rapidly freezing or thawing during short periods of low or high temperatures. However, capacitive water that is stored apoplastically would freeze at higher temperatures (compared to symplastically stored water) and could dehydrate the stem symplast, just as the apoplastic water fraction does in leaf tissue (Ball et al. 2002). If so, high density wood with a lower total volume of apoplastically stored water and smaller individual storage compartments may experience less FT damage than low density wood during freeze-thaw cycles.
While wood density has been shown to be negatively correlated with frost tolerance in the woody taxa of North America (Rueda et al. 2017), data is not currently available to determine if these patterns extend to tolerance of freeze-thaw cycling.  Mayr et al. (2003) found that Picea abies lost more hydraulic function than that of neighboring Pinus cembra in the winter but was more resistant to loss of function caused by drought alone. The key difference among these alpine species seems to be the cuticular conductance, which was higher in Picea abies and caused lower water potential values over the winter than Pinus cembra experienced (Mayr et al. 2003). If xylem pressure decreases during FT cycles, the probability increases that a gas bubble will expand instead of redissolving, even in narrower conduits (Pittermann and Sperry 2006). Low cuticular and/or minimum stomatal conductance has long been identified as a drought tolerance trait in crops (Sinclair 2000) and has been linked to desiccation tolerance in angiosperms more generally (Gleason et al. 2014;Blackman et al. 2016). A recent review found that 15 Pinales species had relatively low values of minimum stomatal conductance compared to angiosperms (Duursma et al. 2019), but it remains unclear the extent to which this trait changes across species ranges in response to greater stress from drought, FT, and/or their combination.
Our results also suggest that a more negative P 50 value imparts some advantage in habitats with higher PSi values (R 2 = 0.11, p < 0.001; Fig 5C). Although this correlation was weak, it was highly significant, and, like the relationship between P 50 and Di, was driven by trends in the Cupressaceae and Taxaceae. The remaining families did not have significant relationships, although the correlation for the Podocarpaceae suggested a trend (p = 0.09). The similarity in the relationships between P 50 vs. Di and PSi suggests that drought tolerance is driving the observed relationship between P 50 and PSi.

Discussion and future directions
Our compiled dataset indicates that conifers occupy a large percentage of the worldwide FT and drought poly-stress space (Fig. 2), and that individual species differ substantially within this space (Fig. 3). Though some species occupy regions with moderate drought and FT stress, no species occupy space with both severe drought and FT stress (Fig. 3). Along with the negative relationship of all species' occurrence along the drought and FT stress indices (R2 = 0.17; p < 0.001; data not shown), this distribution suggests a potential functional trade-off in tolerance to these stresses (Fig. 3). Although the observed species' ranges may be limited by other stressors and/or geographic barriers, the lack of data on traits across the FT, drought, and poly-stress indices prevents us from fully quantifying the impact of these stresses on distribution patterns.
Further analysis of traits across species' poly-stress ranges may answer whether trait combinations conferring poly-stress tolerance are equivalent for all conifers or if different trait combinations lead to the same tolerance across species.
Our analyses indicate a tendency for individual conifer species to occupy a larger range of FTi than of Di (except when FT tolerance is very low; Fig. 3). This tendency remains across all 603 species studied. As a whole, conifers generally occur in habitats with more FT stress than drought stress (Fig 2). These findings may suggest that conifers are better at tolerating FT stress and/or that one unit of the FT index is inherently less stressful than one unit of the drought index.
This observation drives several questions as follows. In species that occupy a broad range of FT space (e.g., Callitris rhomboidea in Fig. 3 coverage is 10%. Without a more complete picture of how and if traits vary across species' ranges, our ability to determine which specific traits are limiting is hindered.
Conifers have a worldwide presence, collectively experiencing a wide range of FT and drought stresses, yet the trait data do not yet exist to fully determine how exactly conifers cope with this poly-stress. For example, the numerous studies that have measured hydraulic vulnerability to drought in a wide range of species contrasts with the comparatively fewer studies that have characterized hydraulic vulnerability to FT cycles. Studies that examine the interaction between low xylem water potential and FT induced embolism (i.e., the combination of drought and FT cycles) in conifers are even rarer still (Pittermann and Sperry 2006;Mayr et al. 2006;Mayr and Sperry 2010). To help make progress in the study of poly-tolerance in conifers, contrasting high versus low poly-stress ecoregions with abundant conifer species (e.g., Western Turkey sclerophyllous and mixed forests vs. Sinaloan dry forests of Mexico) may provide the most facile route to understanding traits conferring FT-drought poly-tolerance.
Finally, it is worth reflecting on the limitation of our analyses regarding future conifer distributions. Rising air temperatures that increase VPD values, unstable atmospheric circulation that drive polar vortices, and increased precipitation variability will combine to increase drought, FT, and poly-stress globally. The vast boreal forests provide an excellent example of the need for more information on which traits are needed for tolerance to each stress and their combination and what the limits are to those tolerances. Specifically, our FT, drought, and poly-stress indices showed that boreal forests currently experience high levels of FT stress but low drought and, consequently, low levels of poly-stress. However, these regions may become drier in the future as altered precipitation patterns result in long, dry periods between summer rain events (Wang et al. 2014). This potential decrease in soil water content will be exacerbated by increases in VPD, which would intensify the drought stress (Lopez et al. 2021;Yuan et al. 2019). These changes will increase the occurrence and intensity of poly-stress, and thus, a tolerance to both FT and drought will be critical for the survival and growth of coniferous species at these latitudes.
Consequently, we need a more fundamental understanding of the traits that permit tolerance of these stresses for a broader range of species across their distributions to predict how these forests will respond to FT and drought stress in the future.

Conclusion
In this review we present four key points: (1) the global peak co-occurrence of drought and freeze-thaw stresses is the combination of moderate drought and moderate freeze-thaw; (2) the distribution of conifers overlaps considerably with moderate levels of FT and drought stress; (3) the distribution of individual species along the two indices varies, with some species exhibiting broad tolerance to both stresses and others exhibiting tolerance to only one (typically FT); and (4) while several hydraulic traits may confer tolerance to both FT cycles and drought in stems and leaves, the lack of trait data across species ranges limits our understanding of polytolerance in conifers globally. Given that global climate change will likely increase the frequency and severity of both FT cycles and drought, addressing this knowledge gap will be of critical importance for predicting how coniferous forests will function in the future.
The emergent science of poly-tolerance is primarily limited by the scarcity of comprehensive intra-and interspecific datasets across broad geographical scales. Many hydraulic traits that may be involved in poly-tolerance -such as conduit and pit dimensions or TLP-are typically measured in a few species at small spatial scales due to the time-intensive nature of these measurements. Recent efforts to expedite these measurements (e.g., measuring TLP by osmometry; Bartlett et al. 2012) have been successful and will facilitate the broad-scale survey of these traits across ecological gradients, especially with increasing utilization of global field experiment networks such as DroughtNet (Knapp et al 2015) and the Center for Forest Global Earth Observatory "ForestGeo" (Davies et al. 2021). Furthermore, the observed distribution of species across these indices suggests that different species may have varying levels of tolerance to each stress, indicating the response of conifers to increasing poly-stress will likely be speciesspecific. Coniferous forests dominated by species that have low poly-tolerance to these stressors may consequently experience range shifts or extirpation in the future. Drastic changes in the distribution of conifer species will have far-reaching industrial, agricultural, and cultural implications, underscoring the need to strengthen poly-tolerance concepts.
Blackman, C. J., Pfautsch, S., Choat, B., Delzon, S., Gleason, S. M., & Duursma, R. A. (2016). Toward Goldstein, G., Rada, F., & Azocar, A. (1985).   . The availability of occurrence data (grey), trait data (light green) and their overlap (dark green) in FTi-Di space for five representative species in five representative genera. Each data point was rounded to the nearest FT and drought index and then buffered by a radius of two index units. The approximate completeness of trait data that is available relative to the environmental space that each species occupies was calculated as the area of light and dark green pixels divided by the area of any-colored pixels, and expressed as a percentage in the top right of each panel.