Can long‐lived species keep pace with climate change? Evidence of local persistence potential in a widespread conifer

Climate change poses significant challenges for tree species, which are slow to adapt and migrate. Insight into genetic and phenotypic variation under current landscape conditions can be used to gauge persistence potential to future conditions and determine conservation priorities, but landscape effects have been minimally tested in trees. Here, we use Pinus contorta, one of the most widely distributed conifers in North America, to evaluate the influence of landscape heterogeneity on genetic structure as well as the magnitude of local adaptation versus phenotypic plasticity in a widespread tree species.


| INTRODUC TI ON
The velocity of anthropogenically driven climate change poses significant challenges for long-lived species, and the longevity of tree species makes them particularly vulnerable to local extirpation. Tree species persistence will depend on rapid adaptation to novel conditions, long-distance dispersal to track ecological niche requirements, or acclimation via plasticity (Aitken et al., 2008). However, rates of both evolutionary change (Etterson & Shaw, 2001) and migration (Davis & Shaw, 2001;de Lafontaine et al., 2018) for many tree species are expected to lag behind the pace of climate change, leaving individuals and ultimately populations, reliant upon plastic responses as they become mismatched with local conditions (Aitken et al., 2008;St Clair & Howe, 2007). Local persistence is most likely in phenotypically plastic and genetically variable populations, as these attributes provide the basis for response to change over both short (i.e., fate of individuals in one generation) and long (i.e., population persistence across generations) timeframes (Benito Garzón et al., 2011;Bontrager & Angert, 2019;Crispo, 2008). Thus, understanding the potential for long-lived tree species to persist through rapid climatic and local environmental change requires insight into the spatial distribution of phenotypic plasticity and genetic variation. This information will be critical to predicting how species distributions may shift with climate change and in determining conservation priorities for at-risk populations.
A species' ability to respond to environmental change is strongly influenced by the landscape conditions where a given population occurs (Manel et al., 2003), and tree species distributed over heterogeneous landscapes likely consist of populations with highly differentiated responses. Heterogeneity in the distribution of abiotic and biotic factors has the potential to impede gene flow by creating geographic barriers or a matrix of unsuitable habitat across which dispersal cannot occur (Sork et al., 1999;Wang & Bradburd, 2014).
On one hand, isolated populations may be able to adapt to local conditions more rapidly because non-adapted alleles no longer flow into the population (Lind et al., 2018). Such populations may also lack the genetic variation to respond to novel environmental selection, thus increasing the risks of local extirpation (Aitken et al., 2008). When landscape heterogeneity does not impede gene flow, selection might instead favour increased plasticity or a generalist phenotype.
Species distributed across complex, climatically variable landscapes are, consequently, subjected to one of two divergent evolutionary pathways, resulting in either specialist populations that are highly locally adapted or the evolution of highly plastic generalists that tolerate a wide range of conditions (Frank et al., 2017).
Widespread tree species provide a natural experiment for evaluating population response to landscape heterogeneity and assessing persistence potential to rapidly changing conditions, but the effect of the landscape has been minimally tested in tree species (Bothwell et al., 2017). Evidence of pollen-mediated gene flow over broad spatial scales (e.g., ~100 km in Pinus sylvestris, Robledo-Arnuncio, 2011; up to 3,000 km in Pinus banksiana, Campbell et al., 1999) suggests that many widely dispersed, wind-pollinated tree species have the potential to overcome the landscape conditions known to limit gene flow in geographically-restricted plant species (Grossenbacher et al., 2014). Despite the potential for high connectivity, geographically based phenotypic variation is common in tree species (Savolainen et al., 2007). Spatially structured phenotypic variation in the face of high gene flow may reflect local adaptation in response to selection imposed by the environment (i.e., role of genetics, environment, and their interaction) or may be a product of high phenotypic plasticity with no heritable genetic basis (i.e., based on environment alone) (Benito Garzón et al., 2011;Kawecki & Ebert, 2004). While plasticity is likely to decrease extirpation risk under rapid climate change if it provides a mechanism by which individuals can phenotypically shift towards a new local optimum, it may also create vulnerability if the plastic response is suboptimal or lags behind local environmental change and reduces the probability for directional selection to support local persistence (Chevin et al., 2013;Chevin & Hoffmann, 2017;Ghalambor et al., 2007). Without insight into the influence of landscape complexity on genetic and phenotypic variability within and among populations, we lack the ability to determine the potential of tree species to evolve in response to ongoing, rapid climate change and thus the ability to identify conservation priorities for forest ecosystems.
Pinus contorta (Douglas Ex. Louden) is one of the most widely distributed tree species in North America, and its occurrence across a topographically and climatically heterogeneous landscape makes it a consummate species for quantifying the influence of landscape complexity on genetic and phenotypic variation. For this reason, P. contorta is one of the most well-studied conifers in biogeographical (e.g., Strong, 2010;Wheeler & Critchfield, 1985;Wheeler & Guries, 1982a, 1982b, forest productivity (e.g., Chuine et al., 2006;McLane et al., 2011;Rehfeldt et al., 1999;Wang et al., 2010) and evolutionary (e.g., Fazekas & Yeh, 2006;Godbout et al., 2008;Liepe et al., 2016;Mahony et al., 2020;Yang & Yeh, 1995) research on forest tree species. However, little is known about the influence of climatically and topographically heterogeneous landscapes on range-wide population genetic structure and population response to novel landscape conditions in this and other widely distributed tree species-information critical to evaluating local population persistence under future conditions.
Here, we use P. contorta to evaluate-(a) the degree to which landscape heterogeneity influences range-wide genetic connectivity K E Y W O R D S adaptational lag, common gardens, landscape genetics, local adaptation, lodgepole pine, microsatellites, persistence potential, phenotypic plasticity, Pinus contorta and variability and (b) the magnitude of local adaptation versus phenotypic plasticity in climatically differentiated environments. We paired landscape genetics with a fully reciprocal in situ common garden study to ask: (a) what are the patterns of genetic differentiation across the topographically and climatically heterogeneous range of this widespread tree species? (b) is there evidence of local adaptation in fitness components such as survival and growth? and (c) what is the degree of phenotypic plasticity in fitness components? Our study provides a unique perspective by pairing a range-wide landscape genetics assessment with a fully reciprocal common garden trial to quantify the influence of landscape complexity on genetic and phenotypic variation in a widespread tree species. Additionally, rather than focusing on well-researched subspecies latifolia, the most widespread and economically important subspecies, we sampled and tested variation across the range of the species and including all subspecies. Our research provides insight into the role of the landscape in shaping population genetic structure in a widespread tree species as well as the potential response of local populations to novel environmental conditions, knowledge critical to understanding how widely distributed species may respond to rapid climate change.

| Study species
Pinus contorta occurs over 33 degrees of latitude from Baja California, Mexico, to the Yukon Territory, Canada, and from sea level along the Pacific to over 3,500 m in the Sierra Nevada of California, USA (Critchfield & Little, 1927;Wheeler & Critchfield, 1985;Wheeler & Guries, 1982; Figure 1). The species is divided into four subspecies (ssp. bolanderi, contorta, latifolia, murrayana;Critchfield, 1957, Figure 1 & Siepielski, 2004) and are hypothesized to be adapted to local climate and environmental conditions (Rehfeldt et al., 1999;Ying & Liang, 1994 Twenty sampling locations were randomly selected from occurrences in regions one through four. In region four, we included two sampling locations representing proposed variety yukonensis

| DNA extraction and microsatellite amplification
Total genomic DNA was extracted using DNeasy plant kits (Qiagen) at the U.S. Department of Agriculture National Forest Genetics Laboratory. Of 15 highly polymorphic SSR markers initially tested (Lesser et al. 2012), nine amplified across all samples (Appendix Table S1). Loci were amplified in multiplex under identical conditions, with locus-specific primers 5'-tailed with universal primer sequences (as described by Missiaggia & Grattapaglia, 2006, see Appendix S1 for details). PCR products were separated on a 3730xl Genetic Analyzer (Life Technologies), and peak sizes were determined using GeneMarker v2.2 (SoftGenetics LLC). Samples were scored three times to verify peaks and resolve conflicts.

| Genetic diversity and differentiation
After screening and adjusting for null alleles, genotyping errors, and deviations from Hardy-Weinberg Equilibrium (see Appendix S2 for details), we calculated pairwise F ST (i.e., the inbreeding coefficient or proportion of genetic variance contained within a subpopulation relative to total genetic variance) and the following parameters, averaged across loci, for each sampling location using GenAlex (Peakall & Smouse, 2012): percent polymorphic loci (PPL), allelic richness unbiased expected heterozygosity (uH E ) and inbreeding levels (F IS ).
We used the "pegas" package in R (Paradis, 2010;R Core Team, 2019) to quantify population differentiation within and among sampling locations and subspecies using a hierarchical analysis of molecular variance (AMOVA).

| Population clustering
We estimated the number of population genetic clusters (K) across the range of P. contorta using two approaches: (a) clustering based on genetic information alone and (b) integrating genetic, geographic and phenotypic information to incorporate characteristics typically used in subspecies delineations. We first used Structure 2.3.2 (Falush et al., 2007;Pritchard et al., 2000) to assign individuals to genetic clusters without grouping them a priori based on geographic location or phenotype; model parameters were set according to updated model run and publishing guidelines (Gilbert et al., 2012;Janes et al., 2017; See Appendix S3 for details). Then, we assessed the role of geographic location (i.e., latitude, longitude) and phenotypic variation (i.e., field-collected data, detailed above) in determining population structure using both uncorrelated and correlated models in "Geneland" 4.0.6 (Guillot et al., 2005(Guillot et al., , 2012. Uncorrelated models assume allele frequencies vary among populations. Correlated models, conversely, assume allele frequencies are similar among populations (e.g., rare alleles in certain populations are also rare in others), which can be more powerful in identifying subtle genetic divisions.
See Appendix S3 for methodologies on population assignment and selection of K.

| Landscape genetics
Pairwise genetic distances among sampling locations were calculated using conditional genetic distance (cGD), where genetic distances are based on genetic covariance and estimated from graph distances as the shortest path connecting pairs via population graph topology (Dyer et al., 2010). Pairwise cGD is more sensitive than traditional metrics (e.g., F ST ), accounting for both direct and indirect connectivity (Dyer et al., 2010). We estimated cGD using the "GStudio" package in R (Dyer, 2016).
We tested for range-wide genetic connectivity by comparing pairwise cGD to pairwise spatial and environmental distances, testing hypotheses of isolation by distance (IBD), barrier (IBB), resistance (IBR) and environment (IBE). For tests of IBD, we calculated pairwise Euclidean geographic distance (km) using Vincenty ellipsoid distance in the "geosphere" package in R (Hijmans et al., 2017).  Figure S1). IBE was evaluated using among-population climate dissimilarities irrespective of spatial connectivity, calculated as pairwise Euclidean distances based on the first three principal components from an analysis of seven bioclimatic variables ("prcomp" function in R). See Appendix S4 for detailed methodologies.
We used multiple approaches to evaluate which hypotheses (IBD, IBB, IBR, IBE) best describe observed patterns of genetic distance. First, we used Mantel and partial Mantel tests in the R package "vegan" (Oksanen et al., 2018) under a reciprocal causal modelling framework (Cushman et al., 2013) to evaluate relative support as the difference between reciprocal partial Mantel tests for each hypothesis. Because Mantel and partial Mantel tests are criticized for their tendencies towards inflated type I error rates (Guillot & Rousset, 2013), we also implemented multiple matrix regression with randomization (MMRR, Wang, 2013) in the R package "ecodist" (Goslee & Urban, 2007) to test for consistency of results, comparing all possible combinations of hypotheses to identify the models with the greatest support.

| Common gardens
Cones from 10 mature (>30 cm diameter-at-breast-height) individuals at each of nine sampling locations were opportunistically collected in Fall 2010, but seed viability limited testable populations to only three sources (bolded locations in Table 1). Fortunately, one collection was viable from each of the three main contrasting climates across which P. contorta is distributed: warm and wet (contorta 11, GENELAND assignments of sampling locations to each identified genetic cluster for uncorrelated runs (K = 2), using genetic, geographic, and phenotypic data and assuming allele frequencies are similar across populations (Guillot et al., 2012). (c) GENELAND assignments of sampling locations to each identified genetic cluster for correlated runs (K = 9), using genetic, geographic, and phenotypic data and accounting for the abundance of rare alleles in genetic data (Guillot et al., 2012). Map projection is a USA Contiguous Albers Equal Area Conic Here, model support for local adaptation is indicated by a GxE interaction such that there is greater performance of a local genotype compared to a foreign genotype within a single site or greater performance of a genotype at home compared to its performance when planted away. Plasticity is indicated by performance response due to a garden effect, in our case the influence of E alone as well as the GxE interaction. We specifically included starting height in our analysis to account for any influence of greenhouse conditions and initial growth on in situ growth and survival. We additionally fit alternative models that included climate transfer distances, calculated as the difference values were estimated using the variance-function method in R package "rsq" (Zhang, 2018).

| Population genetic variation and variance partitioning
Nine markers successfully amplified across samples from 50 sampling locations and were highly polymorphic (mean = 95% ± 1 standard error (SE), Table 2). The mean inbreeding estimate (F IS ) was 0.08 (±0.02 SE, Table 2), which is in line with estimates of high withinpopulation genetic diversity estimates in other conifer species (see Hamrick, 2004 Table S2).
AMOVAs revealed that 88% of genetic diversity is attributable to variation within sampling locations (Appendix Table S3). A moderate, but significant (p < .001), portion of population structure resided among sampling locations (12%), and hierarchical analyses indicated that more variation resided among sampling locations (10%) than among subspecies (2%) or regions (2%). All test strata were significant at p < .001. Regional analyses were consistently similar to subspecies-level analyses so were not considered further.

| Population clustering
Through Structure analyses, we identified four genetic clusters

| Landscape genetics
All analyses using pairwise cGD (Appendix Table S4) identified isolation by barrier (IBB) as the strongest predictor of genetic differentiation (Appendix Table S5). The relative support matrix ( Results from complementary MMRR analyses (Appendix Table S6) also identified IBB as the strongest predictor of genetic variation (R 2 = 0.07, p < .001), followed by IBE (R 2 = 0.03, p < .001) and IBR (R 2 = 0.03, p = .01), while IBD was non-significant (R 2 = 0.002, p > .34). Adding additional predictors to the IBB model resulted in slight increases in explanatory power (from R 2 = 0.07 to R 2 = 0.08), but IBB was the only significant predictor in all models in which it was included.

| Common gardens
We found significant variation in fitness components ( No<del author="Sarah M Bisbing" command="Delete" timestamp="1604959680386" title="Deleted by Sarah M Bisbing on 11/9/2020, 2:08:00 PM" class="reU3">te</del>: In reciprocal causal modelling, relative support represents the difference between reciprocal partial Mantel coefficients for all pairs of hypothesized landscape influences. Specifically, each cell is calculated as: (genetic distance ~ row model | column model) -(genetic distance ~ column model | row model). IBB = isolation by barrier, IBE = isolation by environment, IBD = isolation by distance, IBR = isolation by resistance. In our analysis, IBB was fully supported (bolded values) independent of all other hypotheses, which is indicated by the IBB row containing all positive values and the IBB column containing all negative values. IBD exhibited no independent support after partialling out the effects of landscape heterogeneity. See Appendix Table S5 for detailed Mantel and Partial Mantel results non-significant effects of G (Df = 2, F = 0.8, p = .45) and starting height (Df = 1, F = 2.0, p = .16). The GxE interaction suggests widespread plasticity, as all subspecies reached the largest basal diameters and heights in the contorta garden, and also lends some support for local adaptation of the latifolia source, which had the highest relative diameter and height growth at home compared to foreign sources (Figure 3d,e). Both latifolia and murrayana grew taller than contorta in its home site, suggesting that the warm, wet environment provided a release from the moisture stress common to their respective home environments.

| D ISCUSS I ON
The fate of tree species under rapid climate change will hinge on a match between genotypes and environments (Aitken & Bemmels, 2016;Aitken et al., 2008), and insight into genetic and phenotypic variation under current landscape conditions can be used to gauge persistence potential to future conditions and determine con- Collectively, our findings indicate that, despite generally high connectivity, reduced survival under water-limited conditions may make some populations of P. contorta more vulnerable to local maladaptation and extirpation, and these populations should be prioritized in conservation efforts. However, our findings also suggest that widespread tree species possess genetically diverse and phenotypically plastic populations likely to have high persistence potential under rapid climate change.

| How do heterogeneous landscapes influence genetic connectivity?
Genetic connectivity is a well-documented phenomenon in widely distributed, wind-pollinated tree species (Hamrick, 2004;Kremer et al., 2012), and the limited population genetic structure identified here provides another data point supporting genetic connectivity across the nearly continuous distribution of P. contorta (Fazekas & Yeh, 2006;Wheeler & Guries, 1982a, 1982bYang & Yeh, 1995).
Greater structure or landscape influence may have been more apparent had we utilized a greater number of neutral markers or identified areas of the genome undergoing selection. However, despite this limitation, our sampling across the species' range allowed us to identify subtle landscape constraints to gene flow, which limited connectivity to isolated or narrowly distributed populations and created geographic substructure (Figure 2a). The presence of persistent geographical barriers drove substructure and was the only measurable landscape effect on gene flow (Table 3), in contrast to the strong influence of distance and the environment in many other plant species . Notably, the geographic-genetic structure quantified using the markers tested here does not overlay subspecies delineations but does match expectations of high gene flow over large distances for widespread conifers (Hamrick, 2004;Kremer et al., 2012) while also mapping geographic substructure for isolated regions of the species' range.
Genetic structure in widespread tree species, such as P. contorta, may be further influenced by the now-obscured landscape and climate conditions present when seedlings of long-lived species established (Yeaman & Jarvis, 2006) or even much older historic processes that influenced colonization and migration (e.g., Pleistocene glaciations, Ortego et al., 2015), but historic climate datasets of sufficient resolution do not exist to test these hypotheses. Moreover, the fact that environmental conditions did not structure genetic variation in P. contorta suggests that population genetic structure of long-lived tree species may not yet reflect contemporary patterns of gene flow as mediated by current landscape conditions, revealing a potential lag in the response of widespread tree species to climate change (Gugger et al., 2013;Ortego et al., 2015).

| Local adaptation despite gene flow?
While gene flow can maintain connectivity between populations distributed across complex landscapes, climatically-or spatially-varying selection can be strong enough to overcome the homogenizing effects of gene flow (Kawecki & Ebert, 2004). In our climaticallydifferentiated gardens, we observed some patterns of survival consistent with local adaptation despite gene flow across the range of P.
contorta. This finding, combined with outcomes from the Illingworth provenance trials in British Columbia (e.g., Rehfeldt et al., 1999Rehfeldt et al., , 2001Ying & Liang, 1994), supports the notion that P. contorta populations are locally adapted to somewhat narrower ranges of climatic conditions than are present across its entire range. Cold-tolerance, for example, may have affected survival in our study, a characteristic observed to strongly affect P. contorta survival and growth (Liepe et al., 2016;Mahony et al., 2020;Rehfeldt et al., 1999Rehfeldt et al., , 2001Wang et al., 2010). In our experiment, murrayana and contorta populations had strikingly low survival in the cooler latifolia garden (Figure 3c), suggesting maladaptation to the extreme winter temperatures of this intermountain climate. Warming winter temperatures predicted across the range of P. contorta (Mahony et al., 2017) may relieve maladapted populations of this limitation. Climate change is, however, simultaneously generating novel springtime freezing events and increasing growing-season minimum temperatures, which are documented to drive declines in P. contorta (Mulvey & Bisbing, 2016;Sullivan et al., 2015) and co-occurring species (Buma et al., 2017) and may lead to regeneration failures in temperature-constrained populations.
Water availability is also documented to drive local adaptation in P. contorta (Mahony et al., 2020), and, in our gardens, reciprocal transfers between wet and dry environments had the most profound impact on survival. Specifically, the exclusive survival of murrayana but complete mortality of other populations in the drought-impacted murrayana garden (2012-2016 California drought, Lund et al., 2018) is consistent with greater drought tolerance of populations with a history of exposure to aridity (Figure 3b, Kolb et al., 2016). At the other extreme, transfer to the wet, maritime climate of the contorta garden led to the highest absolute survival for all populations, and water availability appears to be a significant driver of P. contorta success. Local declines are likely in portions of the species' distribution where, despite predicted increases in precipitation (Mahony et al., 2017), concurrent temperature increases will change the timing and type of precipitation (e.g., from snow-to rain-dominated precipitation; Buma et al., 2019) and thus growing-season water availability. Given that drought is expected to become increasingly common across its range (Coops & Waring, 2011;Mahony et al., 2020), drought adaptation may be key to local P. contorta population persistence.
Our common garden interpretations do, however, need to be made with caution given several limitations. Testing one population per subspecies (due to limited seed viability) did not allow us to determine whether or not there are clear breaks among subspecies or rather continuous variation across the species' range. Future work should include more populations per subspecies as well as test sites covering the range of current and predicted future P. contorta habitat conditions. Moreover, short-term experiments for long-lived tree species may not provide definitive evidence for local adaptation (e.g. Pinus ponderosa, Wright, 2007), and long-term tracking of individuals will be required to validate findings. Finally, trait responses may be controlled by many genes, and populations may harbour a vast reservoir of adaptive variation to facilitate rapid evolutionary responses (Barghi et al., 2019). Despite these limitations, our conclusions remain consistent with findings of local adaptation in P. contorta and other widespread conifers (Rehfeldt et al., 2001(Rehfeldt et al., , 2002(Rehfeldt et al., , 2014(Rehfeldt et al., , 2018Wright, 2007), and we hypothesize that patterns of local adaptation will become more apparent over time (Germino et al., 2019).

| Persistence potential via phenotypic plasticity?
We also observed evidence of high phenotypic plasticity in all populations, and this response, despite local adaptation, is likely to promote local population persistence in P. contorta and other widely distributed tree species (Alberto et al., 2013). With an estimated 12 generations required for Pinus species to adapt to projected future conditions (Rehfeldt et al., 2001(Rehfeldt et al., , 2002, evolutionary change is unlikely to match the pace of climate change, and phenotypic plasticity may allow population persistence under a wide range of future local conditions. Previous work found that plasticity in P. contorta growth potential was highest for populations from warmer environments, whereas cold-hardy populations were limited in growth plasticity but exhibited higher survival in colder environments (Rehfeldt et al., 2018). Population response in our gardens was consistent with these expectations: murrayana from the warm, dry Sierra Nevada had low survival in the cold, dry latifolia garden but exceptional growth rates across all environments, while latifolia had high survival at home and a limited growth response elsewhere. This ability of P.
contorta genotypes to be plastic in their response to environmental heterogeneity may provide the foundation for persistence potential by buffering local populations from negative selection and giving this long-lived, slow-to-migrate tree species more time to adapt to novel local conditions (Alberto et al., 2013;Crispo, 2008).

| Is persistence potential enough?
Although plasticity may provide populations time to adapt, it is concerning that many populations of P. contorta and other conifers of western North America already lag behind their climatic optimum (Gray & Hamann, 2013;Johnstone & Chapin, 2003). Climate change projections indicate a decline in P. contorta suitable habitat across much of the species range by 2080 (Coops & Waring, 2011;Oney et al., 2013), and productivity and growth are expected to decline at lower latitudes and elevations in the near future (Rehfeldt et al., 2001;Wang et al., 2006). Populations occurring at lower elevations, particularly at southern latitudes, are at particular risk of local extirpation due to compounding warming and drying (Coops & Waring, 2011;Mahony et al., 2017;Rehfeldt et al., 2001). The pace of evolutionary change for long-lived tree species is expected to be slow, and habitat suitability (Gray & Hamann, 2013), provenance testing (Rehfeldt et al., 2001), and growth chamber (Liepe et al., 2016) studies of P. contorta corroborate our findings that some populations already lag substantially behind their climatic optima.
Similar mismatches to contemporary climate were recently identified in Pinus ponderosa (Martínez-Berdeja et al., 2019) and Quercus lobata (Browne et al., 2019) and interpreted as evidence of environmental change that exceeds the pace of evolutionary change (i.e., adaptational lag; Mátyás, 1994). In these species, populations from warmer, drier climates had the highest growth potential when grown in cooler or wetter conditions, suggesting a mismatch to current climate and high vulnerability to ongoing warming and drying.
Consistent with these findings, maximum survival and growth of latifolia and murrayana populations tested here occurred under the mild climate of our contorta garden, providing additional evidence for a lag between P. contorta occurrence and its climatic optimum. Populations growing under extreme local conditions may still possess adaptations (e.g., drought tolerance) making them optimally suited for the home environment but be diminished in growth and survival due to the climatic lag between local and optimal conditions, which may be best explained by adaptation to historic colder, wetter climates.
Prior research on P. contorta historical migrations and contemporary invasions into meadows (latifolia: Jakubos & Rommer, 1993;murrayana: Helms, 1987;Anderson, 1996;Lubetkin et al., 2017) provides support for an adaptational lag across much of its current distribution. Slow, progressive warming and drying during the early Holocene are the likely origin of P. contorta's adaptational lag, which led to extirpation of murrayana from lower elevations and forced populations to track cooler, wetter climates by migrating to higher elevation (Anderson, 1996). Warmer growing seasons since the end of the Little Ice Age (ca. 1,870) have led to further moisture stress for latifolia and murrayana, and montane meadows provide a local source of relief and opportunity for establishment (Helms, 1987;Jakubos & Rommer, 1993;Lubetkin et al., 2017). For long-lived tree species, such as P. contorta, persistence through climate fluctuations over geologic time may mean that maladaptation to contemporary climate is common (Gray & Hamann, 2013), populations are instead adapted to historic climates (Browne et al., 2019), and projected climate conditions will only exacerbate adaptational lags and perpetuate growth under suboptimal conditions.

| CON CLUS IONS
Our findings suggest that P. contorta populations likely have high persistence potential via phenotypic plasticity and high genetic variability. However, geographically-based genetic substructure in some portions of the species' range as well as complete mortality of non-local populations in our most water-limited garden also indicate that some populations may be vulnerable to local maladaptation and extirpation with rapid climate change. Management of conifers is already incorporating assisted migration as part of a conservation strategy for maintaining viable populations of these long-lived species (e.g., O'Neill et al., 2008;Young et al., 2020), and our results suggest that such efforts may be warranted for vulnerable populations, complementing the natural processes of high gene flow and local adaptation within widespread conifers.

PEER R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/ddi.13191.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data generated for this study are available through the Dryad Digital Repository: https://doi.org/10.5061/dryad.nvx0k 6dqv (Bisbing, 2021)