Drought and freezing vulnerability of the isolated hybrid aspen Populus x smithii relative to its parental species, P. tremuloides and P. grandidentata

Abstract Aim We assessed the vulnerability of an isolated, relictual Pleistocene hybrid aspen population of conservation interest (Populus x. smithii) and the nearest populations of its parent species (Populus grandidentata and Populus tremuloides) to springtime post‐bud break freezing and growing season drought stress. Response to these stressors in the three taxa was compared in terms of avoidance and tolerance. Location North American Midwest; USA. Methods Unique genets from the hybrid Niobrara River population and from the two parental populations were propagated in a common garden from rhizome cuttings. We tracked their phenology before and after bud break and measured their vulnerability to freezing (stem electrolyte leakage and leaf chlorophyll fluorescence) and to drought (stem hydraulic conductance, leaf osmotic potential, stomatal pore index, and gas exchange). Results Populus grandidentata was slower to leaf out, showed lower vulnerability to stem freezing and drought‐induced cavitation, but exhibited a lower capacity to tolerate drought stress through leaf resistance traits compared to P. tremuloides. Hybrids were similar to P. grandidentata in their overwintering strategy, exhibiting later bud break, and in their higher resistance to stem freezing damage, but they were more similar to P. tremuloides in their higher vulnerability to drought‐induced cavitation. The hybrids shared various leaf‐level gas exchange traits with both parents. All aspens showed limited loss of leaf photosynthetic function following moderate freezing. Main Conclusions The Niobrara River hybrid population is vulnerable to drought due to its combination of inherited drought avoidance and tolerance traits. As climate changes, P. x smithii will likely suffer from increased drought stress, while being unaffected by frost during warmer springs. The two parental species contrast in their survival mechanisms in response to climatic stress, with P. tremuloides tending toward freezing tolerance but drought avoidance and P. grandidentata tending toward freezing avoidance and drought tolerance.


| INTRODUC TI ON
Deciduous tree species face environmental challenges in continental environments, with freezing and drought representing two main stressors (Medeiros & Pockman, 2011;Menon, Barnes, & Olson, 2015). Species in temperate North America must be able to survive both freezing winter temperatures and seasonal drought stress.
Adaptation to these abiotic forces has been critical in shaping the diversification of temperate tree lineages and the distribution of these species (Adams et al., 2009;Graham, 2011;Morin, Viner, & Chuine, 2008;Zanne et al., 2014).
Plants survive abiotic stress through some combination of tolerance and avoidance (Levitt, 1980;Savage & Cavender-Bares, 2011). Tolerant plants can survive when their tissues are exposed to freezing temperatures or low water availability, and continue to photosynthesize through adaptations that prevent damage to cells and tissues. These include narrow vessels, antifreeze compounds, and morphological traits (e.g., torus-margo) (Baraloto, Morneau, Bonal, Blanc, & Ferry, 2007;Pittermann, Sperry, Hacke, Wheeler, & Sikkema, 2005). Avoidant plants also survive stress but do so by investing in mechanisms that shield them from stress or allow for temporary reduction in physiological activity, including closing stomata, shedding leaves, or delaying leaf out (Cavender-Bares & Holbrook, 2001;Cavender-Bares et al., 2005;J. Cavender-Bares, 2007). Plants may have different strategies for different stressors; commitment to one strategy may limit a species' access to another strategy and may also result in less growth or productivity (Koehler, Center, & Cavender-Bares, 2012;Savage & Cavender-Bares, 2013). Variation among populations in these strategies can thus result in variation in fitness in response to environmental stress, potentially leading to speciation (Heschel & Riginos, 2005).
The North American aspens are a widely distributed clade of temperate and boreal species that inhabit a range of environments across the continent, including eastern deciduous woodlands, Midwestern prairies, boreal parklands, alpine environments, and arctic environments. The sister species Populus tremuloides and Populus grandidentata are the only two North American representatives of the section Populus (Leuce) Duby. Populus tremuloides is the most widespread tree species in North America and inhabits ecologically diverse environments; P. grandidentata is confined to the mesic eastern United States and Canada. The two co-occur across the range of P. grandidentata and have been known to hybridize (Populus x. smithii) occasionally in this region (Figure 1a). One hybrid population in particular, formed in northern Nebraska, USA during the last glacial maximum, is the focus of this study. It has persisted in isolated north-facing canyons along the Niobrara River (Kantak, 1995;Kaul, Kantak, & Churchill, 1988) and lies well outside of P. grandidentata's current range. Because it is part of a unique community of boreal species along a U.S. National Park System designated National Scenic River, land managers are committed to preserving this culturally and ecologically valuable stand of relictual aspens, comprising a dozen groves consisting of just three unique genotypes (Deacon, Grossman, Schweiger, Armour, & Cavender-Bares, 2017). The population is thought to be in decline due to a variety of stressors (Robertson, Cahlander-Mooers, & Dixon, 2018), potentially including the risk of post-bud break freezing and drought concomitant with climate change.
In order to draw insights relevant to the Niobrara River P. x smithii stand's response to future climate change, we compared genotypes of the hybrid trees-in terms of their vulnerability to freezing and drought stress-to the nearest populations of their parental species from across the Upper Midwest. All three taxa were grown in a common garden in central Minnesota and in a controlled greenhouse environment. We observed the phenology before and after bud break F I G U R E 1 (a) Collections of Populus grandidentata (red triangles), Populus tremuloides (blue squares), and Niobrara Populus x smithii (purple circle) took place within and across the western boundary of the upper Midwestern ranges of Populus grandidentata (hatched) and P. tremuloides (light gray), and from isolated stands of P. tremuloides. Dark gray shading indicates areas where neither species is present. (b) Climatic (annual mean temperature and precipitation) envelopes of the ranges of P. grandidentata (red) and P. tremuloides (blue). A purple dot indicates the climatic conditions experienced by the hybrid P. x smithii population at Smith Falls. Details of climate envelopes in Appendix S3 and measured vulnerability to stem and leaf freezing of the hybrid and both parents in the common garden. This allowed us to compare their responses under the same environmental conditions to milder winters and experimentally simulated post-bud break freezes, which may be expected to occur with increasing frequency as a consequence of climate change. We also assessed stem and leaf responses to drought stress, which is also predicted to occur in some climate change scenarios for the region (Mathys, Coops, & Waring, 2017).
Based on their distributions (Figure 1a), P. tremuloides experiences more variable environments (Figure 1b). If exposure to climatic variation results in greater plasticity (Miner, Sultan, Morgan, Padilla, & Relyea, 2005), we would expect P. tremuloides to exhibit greater plasticity and trait variability than P. grandidentata. Given the low temperatures P. tremuloides and P. grandidentata are both exposed to within their ranges, we would also expect stems of both species to survive severe freezing, either through freezing tolerance mechanisms or by freezing avoidance mediated by the timing of spring bud break and leaf development phenology, which determines the exposure of leaves to freezing temperatures. The range of P. tremuloides also extends into drier environments than P. grandidentata so we would expect differences between them in the mechanisms for dealing with water stress in their stems and leaves and their response to simulated drought. Hybridization in plants offers natural experiments that have long interested evolutionary ecologists (Abbott, 1992;Anderson, 1948;Barton & Hewitt, 1985;Ellstrand, Whitkus, & Rieseberg, 1996;Rieseberg & Carney, 1998), and our research adds to the growing body of evolutionary ecological research in this genus. The genetic composition of Populus hybrids, for example, has been studied in the context of speciation, reproductive isolation, and adaptation (Christe et al., 2016;Gramlich & Hörandl, 2016;Jiang et al., 2016;Lexer, Fay, Joseph, Nica, & Heinze, 2005). In a previous study (Deacon et al., 2017), we found that the Niobrara hybrids shared nuclear genetic material from both parents and inherited the cytoplasmic genome from P. grandidentata. Given this background, we expected that the hybrids would share physiological freezing and drought vulnerability with both parents, and potentially be more similar to P. grandidentata.

| Sampling locations
From May 2013 to June 2015, we collected rhizome cuttings from aspen populations throughout the Midwestern U.S. range of P. tremuloides and P. grandidentata (Figure 1a; Appendix S1). Collections of P. grandidentata represent the species' western range edge in Minnesota and Iowa and interior populations in Minnesota and Wisconsin. Cuttings of P. tremuloides were collected from these sites as well as from small, isolated stands in Nebraska and from the South Dakota Black Hills.
Finally, we collected cuttings from the hybrid P. x smithii aspens in the Niobrara River Valley (NRV) of north-central Nebraska (Deacon et al., 2017). Though this hybrid can and does form across its parents' overlapping ranges (Barnes, 1961;Barnes & Pregitzer, 1985), we were, in the work reported here, interested in the physiology of the NRV hybrid aspens. We therefore confined our sampling of hybrid trees to this stand. Due to the few remaining stands of the NRV hybrids, and the attempt to collect from the nearest parental species populations in all directions surround the NRV hybrids, many more collections were made of P. tremuloides, which occurs in closer proximity to the NRV than of P. grandidentata and P. x smithii.

| Common garden design and establishment
We established a common garden with all three aspen taxa grown out from collected rhizomes in the greenhouse. When collected, rhizomes were trimmed to 10-50 cm in length, transported to the Trees produced in the greenhouse were randomly planted with respect to genotype at 1 m 2 intervals in a cleared field that had been fenced to exclude deer. We watered all seedlings and mowed the space between them following establishment. Propagation success was variable among taxa and more of the P. grandidentata and P. x smithii stems failed than P. tremuloides. We transplanted 362 P. tremuloides, 42 P. grandidentata, and 31 P. x smithii stems for a total of 435 stems into the common garden. The number of physiologically and genetically distinct individuals of each species included in measurements and experiments below is given in Appendix S2.

| Leaf phenology
We recorded the status of overwintering leaf buds to monitor ex-

| Greenhouse bud break
We collected branch segments containing at least 5 buds each from 10 naturally occurring mature P. tremuloides and 10 naturally occurring mature P. grandidentata trees near our common garden  to the boiled electrolyte leakage (Flint, Boyce, & Beattie, 1967).

| Stem freezing-electrolyte leakage method
Freezing tolerance was calculated as 100 less the index of injury such that positive values indicate higher tolerance.

| Leaf freezing-chlorophyll fluorescence method
Branches with at least 2 attached leaves were collected from com-  1999, p.;Feild & Brodribb, 2001;Cavender-Bares et al., 2005, p.). potential (Ψ PD ) was measured on leaves of the same trees prior to stem collection using a pressure chamber (PMS Instruments). Leaves on the collected stems were removed, and leaf area of all leaves distal to the segment was quantified. Axillary leaf scars were covered with superglue (LocTite Gel), and stem diameter was recorded prior to placing stems in a tubing apparatus that allowed the solution to move gravimetrically under a known pressure gradient (Sperry, Donnelly, & Tyree, 1988

| Leaf osmotic potential
Leaf osmotic potential was determined from leaf disks from fully saturated plants following a modified protocol from (Bartlett et al., 2012). 6.35-mm-diameter leaf disks were cut using a paper hole

| Leaf stomatal pore index
Three to five leaves per plant were collected from the same plants as gas exchange measurements. Stomatal densities and guard cell lengths were determined from clear nail varnish impressions on the abaxial leaf surface between the mid-vein and the leaf margin and measured on images taken from a microscope camera using ImageJ software (Rasband, 1997). Measurements were averaged over three regions of the impression at 20× and 40× magnification, respectively. Total stomatal pore area index (SPI; a dimensionless index of stomatal pore area per lamina area) was calculated as stomatal density × guard cell length 2 (Sack, Cowan, Jaikumar, & Holbrook, 2003).

| Data analysis
For each set of measurements or experiments described above, we used a two-step linear regression approach to determine the factors affecting aspen physiological performance. First, we used bivariate models to assess whether there was an effect of taxonomic identity (P. grandidentata, P. x smithii, or P. tremuloides) on physiological performance. Then, because P. grandidentata and P. x smithii represent a restricted range of states and climates (mesic climates in only three and one states, respectively) relative to P. tremuloides, we constructed a second set of multiple regression models for P. tremuloides only. These models predicted physiological performance based on state of origin (IA, MN, NE, SD, or WI; these sometimes served as surrogates for latitude and longitude of collection) or climate effects (relevant climate variables downloaded from Climate WNA) (Wang, Hamann, Spittlehouse, & Carroll, 2016)). In particular, we considered annual heat moisture index (AHM; mean annual temperature divided by mean annual precipitation) and number of frost-free days (NFFD) as potential climatic predictors of physiology. When repeated measures were taken on a plant over time or at different freezing temperatures, this information was also included in regression models, including interaction effects when they improved model fit. In all cases, we compared hierarchical models using likelihood ratio tests and nonhierarchical models using AIC tests and present findings from the best-supported, most parsimonious model for each measurement or experiment. When significant ANOVAs indicated differences among levels of categorical predictors, we used post hoc tukey tests ("HSD.test" in agricolae; (De Mendiburu, 2016)) to differentiate between factor levels at α = 0.05.
Curves representing loss of stem hydraulic conductance with decreasing water potential were fit with a Weibull distribution using the fitplc package (Duursma & Choat, 2017). PLC 50 values, which give the water potential at which 50% of hydraulic conductance is lost, were extracted from these curves and confidence intervals were generated using bootstrapping. We also use ANOVA to assess differences among taxa in predawn water potential, native stem embolism (before spinning), and the PLC reached by each individual at the minimum water potential imposed (−2.2 MPa). Because sample sizes for this set of measurements are low, we did not model the consequences of state or climate of origin for stem hydraulic conductance properties. All data were analyzed in R (R Core Team 2013, v. 3.4.1).

| Phenology
The leafout phenology of aspens in the common garden varied significantly among taxa and, within P. tremuloides, by state of origin ( Figure 2; Table 1). Populus tremuloides showed advanced phenology relative to P. grandidentata, with their hybrid P. x smithii more similar to P. grandidentata (Figure 2a; For instance, P. tremuloides individuals from Wisconsin and Iowa collections were more similar to phenologically slower P. grandidentata or hybrid aspens compared to individuals from Nebraska and South Dakota, which leafed out much earlier. Greenhouse studies of bud break among aspen branches collected on mature trees in central Minnesota showed that P. tremuloides required briefer exposure to freezing temperatures for bud break to occur relative to P. grandidentata, breaking bud on average 4 days earlier in a controlled environment ( Figure 3; Table 1; df = 3, 100; F = 24.9, p < 0.001). Both species reacted similarly to prolonged exposure to winter freezing (no significant interaction), although longer winter duration had a stronger effect on P. tremuloides than on P. grandidentata.

| Stem freezing
Stems of all aspens were relatively tolerant of freezing (at −5 to −15°C) following bud break. P. grandidentata and the hybrid P. x smithii aspens were more resistant than P.  Figure 4).
Timing of freezing events significantly influenced stem damage.
Early spring freezes were least damaging to aspen stems, causing an average of only 26% damage, while late-fall freezes caused slightly more (35%) damage ( Figure 5; Table 1). The greatest freezing damage (50%) was exhibited in May, once the growing season had already begun. Freezing during this period tended to cause more damage across all taxa and temperatures (df = 2, 198; F = 32.0, p < 0.001).
There was no strong influence of state or climate of origin on freezing damage to stems of P. tremuloides; freezing damage was not more or less severe in individuals from warmer climates.

| Leaf freezing
Aspen leaves were generally resistant to freezing, rarely losing more than 10% of their maximum quantum yield following a freeze event and recovering after freezing to Fv/Fm values of 0.8, indicating nearfull function (Appendix S4). Taxa did not differ significantly in loss of photosynthetic function following freezing, and P. tremuloides individuals did not vary by state or climate of origin.

| Stem hydraulic conductance and vulnerability to drought embolism
Populus tremuloides was more vulnerable to cavitation with spinning than was P. grandidentata, with the P. x smithii hybrid intermediate between them ( Figure 6; Table 1

| Leaf osmotic potential and stomatal pore index
The parental aspen species differ in two heritable traits controlling water use and drought tolerance. Populus tremuloides had a more greater capacity to absorb water in dry conditions, but a larger stomatal pore surface area, associated with higher leaf hydraulic conductance (Sack et al., 2003), and lower ability to prevent water loss and desiccation (Cavender-Bares, Sack, & Savage, 2007). The hybrid P. x smithii showed less negative osmotic potential, like P. grandidentata, but high SPI, like P. tremuloides (Figure 7a; Table 1), indicating lower water absorption capability and low resistance to water loss.

| Gas exchange and water use efficiency
Intrinsic water use efficiency (WUEi) in the common garden did not vary by taxon or by state, but roughly 10% of the variation in gas exchange (light saturated A and g s ) was associated with climate of origin (AHM, increasing as conditions become more hot and dry)

| D ISCUSS I ON
Populus tremuloides and P. grandidentata show different strategies for surviving drought and freezing, with P. x smithii sharing qualities of both parents. Based on current species ranges, we expected P. tremuloides to vary in its response to these stressors more than the other two taxa. Delayed bud break and leafout phenology play an important role in the avoidance of potential late-spring freeze events in P. grandidentata, and its stems were not as injured in our freezing experiments compared to P. tremuloides. These factors correspond to a freezing avoidance strategy in P. grandidentata and a freezing tolerance strategy in P. tremuloides. In contrast, given its lower stem vulnerability to embolism, P. grandidentata exhibits greater drought tolerance than P. tremuloides. Meanwhile, the more negative osmotic potential of P. tremuloides allows it to access water more readily during drought, indicating a drought avoidance strategy. The two parental species thus contrast in important ways in their survival mechanisms in response to climatic stress, with P. tremuloides tending toward freezing tolerance but drought avoidance and P. grandidentata tending toward freezing avoidance and drought tolerance.   (Wang, Ives, & Lechowicz, 1992). The connection between early spring leaf flushing and traits associated with stress response is common to both drought and freezing (Bohnert, Nelson, & Jensen, 1995) so these sister species may be preadapted to tolerate both of these stresses.
Populus x smithii shares freezing avoidance traits with P. grandidentata, but freezing is not currently, nor is it projected to be, a big challenge for our focal population. The freezing temperatures imposed in our experiments that resulted in a high index of injury are much lower than climate records for occasional spring freeze events for the Niobrara River Valley of Nebraska. In fact, the freezing avoidance traits of P. x smithii may put it at a disadvantage in a warming climate where protection from freezing damage comes with tradeoffs for growth and productivity (Koehler et al., 2012;Savage & Cavender-Bares, 2013).
In contrast to the similarity between the hybrid and one parent for freezing avoidance, P. x smithii shares drought tolerance and avoidance traits with both parents. The ability to regulate leaf osmotic potential allows plants to deal with both the stress of drought and freezing (Sakai & Larcher, 1987)  Freezing tolerance atof P. x smithii is more similar to P. grandidentata and less negative than P. tremuloides. Solutes in the cell can protect membranes and proteins while also helping to maintain turgor and there can be strong selection for low (more negative) osmotic potential in summer drought or winter cold (Callister et al., 2008). Unlike regionally sympatric P. tremuloides, P. x smithii has less negative osmotic potential and does not regulate stomatal conductance like the western P. tremuloides, where much lower water use efficiency (WUE) was observed. Typically high WUE and low stomatal conductance are characteristics of drought tolerant plants, whereas low WUE and high stomatal conductance are seen in drought avoiding species that maximize growth before drought intensifies (Heschel & Riginos, 2005). P. x smithii has a lower stomatal pore index (SPI) than P. tremuloides, similar to P. grandidentata. SPI combines density and size of stomata and is usually correlated with stomatal conductance (Sack et al., 2003). Species or populations with a lower SPI tend to have lower stomatal conductance and higher WUE.
The greater loss of stem hydraulic conductivity with increasing xylem tension (more negative water potential) in P. tremuloides provided evidence that is the most vulnerable of the three taxa to cavitation/embolism. Its lack of drought tolerance in the stem and the higher stomatal pore area in the leaves are both indicative of drought avoidance. This characterization is further supported by its more negative osmotic potential that promotes greater water access during drought. In contrast, P. grandidentata's more embolism-resistant stems and leaves with low stomatal pore area that minimize water loss are indicative of a drought tolerance strategy.  (-0.48, -1.74) F I G U R E 7 Species differed significantly in leaf traits related to water use and drought vulnerability. (a) Populus tremuloides (blue) leaves had a significantly more negative osmotic potential, allowing increased water access during drought (a drought avoidance strategy) than Populus grandidentata (red). (b) Compared to P. tremuloides, P. grandidentata had lower stomatal pore index (SPI), which allows leaves to minimize water loss.
Hybrid Populus x smithii (purple) leaves showed lower negative osmotic potential similar to P. grandidentata but higher SPI similar to P. tremuloides. Letters indicate significant differences at the 0.05 level from a Tukey post hoc test 2-6°C temperature increase from 2025-2034 compared to  and increased precipitation (0%-20%) that will be offset by increasing temperatures, resulting in further drought stress. Warmer and shorter winters will likely not have any negative effects on P. x smithii because spring freeze events are unlikely to be cold enough, but successful management of the NRV aspens may require greater intervention than has been attempted to date.

ACK N OWLED G M ENTS
The authors thank Jeffrey Carstens (USDA-ARS) provided invaluable assistance in collecting root and leaf cuttings, providing natural history knowledge, and suggesting collection locations. Bruce

McIntosh (Western Nebraska Resources Council) and Joseph
Zeleznik (NDSU) generously accompanied us in field collections F I G U R E 8 (a) Intrinsic water use efficiency (WUEi; light saturated A max /g s ) in the common garden did not vary by species or in plants collected in more mesic, eastern sites. However, when only Populus tremuloides individuals from drier, hotter western sites (those with a high summer heat moisture index) were considered, plants from the driest and hottest sites had lower WUE in the common garden. The dashed line indicates the highest summer heat moisture index value recorded for a Populus grandidentata observation in the GBIF database. Populus tremuloides individuals collected from sites with a higher Index value than this are included in the regression. (b) Observed light saturated stomatal conductance (g s ) and (c) photosynthetic rate (A max ), in the common garden did not vary by species (red: P. grandidentata, purple: Populus x smithii, blue: P. tremuloides). There was no relationship between g s or A max and climate (summer heat moisture index) in the range of P. grandidentata and the eastern part of the range of P. tremuloides. Yet in the western part of the range of P. tremuloides, A max declined precipitously (and g s did to a lesser extent) contributing to a significantly lower WUEi of plants from warmer and drier climates

DATA ACCE SS I B I LIT Y
All data are accessible in the Cedar Creek Ecosystem Science