where x is the absolute value of the difference between ΨPD corresponding to a particular photosynthetic measurement and the least negative ΨPD observed during the study.  The difference was used to avoid potential overestimation of Amax (or gmax). Multi-parameter Weibull-type drought response curves are typically used to describe how tree trait respond to changes in drought stress, as they account for the often-observed relative unresponsiveness of trees when experiencing mild stress (Bateman et al. 2018; Vico and Porporato 2008; Wolfe et al. 2016). We calculated for each genotype the curve parameter values Amax , b and c based on minimization of the sum of squared differences between the observed and predicted values for Aarea at different stages of drought stress. We then generated LMMs with curve parameter values as response variables and ploidy, defoliation and their interaction as explanatory variables and genotype as random intercepts. If necessary, response variables were transformed to meet model assumptions as described above.
In Experiment 3, the effects of ploidy, genotype, and defoliation on tree growth, morphological and allocational traits were evaluated using LMMs as described above. Individual values of final above ground tree dry weights and dry weight of newly produced tissue were normalized for initial d2h. Total leaf weight was normalized for total final tree dry weight.
Analyses were conducted with R 4.0.4 (R Core Team 2021). Linear mixed effects models (LMMs) were fitted using the “lme4” package (version 1.1-20, Bates et al. (2014)). LMMs were subjected to type III ANOVAs with Satterthwaite’s method to produce a summary of the F and p statistics. We calculated the relative explanatory power of individual fixed effects by quantifying the semi-partial R2 to assess the relative importance of each fixed effect while accounting for all other fixed and random effect terms. The r2beta function in the package “r2glmm” was used for calculating semi-partial R2 statistics (v. 0.1.2 ,(Jaeger et al. 2017)). We then calculated the proportion of total variance explained by all fixed effects (R2marginal) and all random effects (R2conditional - R2marginal). Lastly individual random effects were calculated by dividing the variance for that effect by the total variance. The getvariance function of the “insight” package (v. 0.13.1(Lüdecke et al. 2019)) was used to calculate proportions of model variances explained by the total fixed effects,  the total random effects, and the model residuals as well as for quantifying individual variance contributions of each random intercept of an LMM. The significance of random effects was tested with log-likelihood-ratio tests using the ranova function of the “lmerTest” package (v.3.1-3 ) (Kuznetsova et al. 2017). Weibull curve parameters were generated in Excel solver.
RESULTS
Experiment 1: Ploidy effects on traits of nonstressed trees across populations
After 93 days of growth, genotypes from the Great Lakes, Wisconsin (WI) population, had on average, more than a 3-fold higher growth rate than those from the Intermountain West, Utah (UT) (Fig. 1, Supplementary data Table S1). Differences between populations explained over 45% of the total observed variation in relative growth (effect size plot, Fig. 1). RG did not differ significantly between ploidy levels (Fig. 1, Supplementary data Table S1). Population differences in LWR indicated that, across ploidy levels, WI genotypes invested on average 72% more in leaves than did UT genotypes (Fig. 1, Supplementary data Table S1). Specific leaf area (SLA) averaged 19% higher across WI genotypes than in UT genotypes (not shown, Supplementary data Table S1). LWR and SLA differences between cytotypes were slight and nonsignificant.
All leaf phytochemical traits differed significantly between populations. Levels of total salicinoid phenolic glycosides were on average, 17% higher in WI compared with UT genotypes (Fig. 1, Supplementary data Table S1). WI genotypes also exceeded UT genotypes with respect to foliar concentrations of individual phenolic glycosides tremulacin (36%) and salicin (124%). Salicortin levels, however, did not differ between WI and UT genotypes. Condensed tannin levels were 33% higher (not shown, Supplementary data Table S1) and nitrogen levels were 15% higher in WI than in UT genotypes (Fig. 1, Supplementary data Table S1). In no case did measures of phytochemistry differ significantly between cytotypes.
A population contrast was also observed for light-saturated photosynthesis (29% higher in WI than in UT genotypes) and stomatal conductance (21% higher in WI than in UT genotypes). Again, however, none of the leaf physiological parameters differed significantly between ploidy levels (Fig. 1, Supplementary data Table S1).
Genotypic differences significantly influenced all growth, allocation, phytochemical and leaf physiological traits. Genotypic differences had a higher explanatory importance for phytochemical and leaf physiological trait variations than population differences (effect size plots, Fig. 1). In contrast, growth traits variations were driven more by differences between populations than among genotypes.
Experiment 2: Ploidy effects on tree traits and their response to stress
Among Wisconsin genotypes, we found no difference between diploid and triploid trees for any aboveground growth traits when averaged across all stress treatments (Fig. 2, Supplementary data Table S2). In contrast, genotypic differences significantly influenced most growth traits. Tree genotype was particularly relevant for determining final height and SLA and explained 39% and 30% of the total observed variance in these two traits (effect size plots, Fig. 2).
Tree growth and biomass allocation traits were affected by drought stress, defoliation or their interaction (Fig. 2). At a given tree weight, drought-stressed trees had on average 8% lower leaf weights than did watered trees. Similarly, defoliation reduced total leaf weight and SLA. For total tree weight we observed significant drought × defoliation interactions. Nondefoliated, watered trees had 25% greater total weight when compared with nondefoliated, drought-stressed trees. However, when trees were defoliated, drought-treatment differences disappeared. Similar, albeit less pronounced interaction effects between drought and defoliation treatments were also observed for total leaf weight. Cytotypes showed similar responses to drought stress and defoliation (no significant ploidy × treatment interactions were found). Furthermore, we did not observe any genotype × treatment interactions.
Concentrations of most chemical compounds did not differ between the cytotypes when averaged across all treatments (Fig. 3, Supplementary data Table S2, Fig. S2). The only exception was tremulacin, which was 17% higher in triploid trees than in diploids (Supplementary data Fig. S2). However, differences in ploidy levels could mainly be attributed to a single triploid genotype (Supplementary data Fig. S2). Genotypic differences significantly affected most chemical compounds (Fig. 3, Supplementary data Table S2, Fig. S2). Genotypic effects accounted for 30% of the variation in total salicinoid phenolic glycosides and almost 50% of the variation in condensed tannins (effect size plot, Fig. 3).
Drought stress and defoliation did not affect concentrations of total phenolic glycoside levels. Drought, however, led to increased levels of the phenolic glycosides salicortin (7% increase) and salicin (44% increase) (Supplementary data Table S2). Condensed tannin concentrations increased by 25% under drought stress and by 56% in response to defoliation. Nitrogen levels increased by 11% under drought stress, and decreased minimally (<5%) due to defoliation. Ploidy levels as well as the different genotypes responded similarly to drought stress and defoliation (no significant ploidy × treatment or genotype × treatment interactions).
Differences in ploidy affected photosynthetic rates (Fig. 4a) and Vcmax (Supplementary data . Fig,S3). Well-watered triploids exhibited 11% higher photosynthetic rates and 23% higher Vcmax. Cytotype explained almost 20% of the total variation in Vcmax (Fig. 4a, effect size plot , Supplementary data Fig. S3,). No other leaf physiological traits were affected by cytotype levels. Genotype significantly affected variation of photosynthetic rate, stomatal conductance and the quantum  efficiency of PSII, explaining 10-20% of the variation in all traits (Fig. 4a, Supplementary data Fig. S3).
 In well-watered trees, defoliation affected only stomatal conductance, which was 3% higher in defoliated than in nondefoliated trees (Supplementary data Fig. S3). No significant ploidy × treatment or genotype × treatment interactions were found. For trees that were exposed to drought stress, we observed similar drought response curve progressions for diploid and triploid trees of the same defoliation treatment. No significant impact of ploidy or ploidy × defoliation interactions was detected for any curve parameter (Fig. 4b, Supplementary data Fig. S4, Fig.S5,). However, defoliation treatments per se significantly affected net photosynthesis and stomatal conductance responses to drought. Photosynthetic rates of nondefoliated trees plummeted when experiencing ΨPD of -7 bars or lower. In contrast, photosynthesis rates of defoliated trees decreased more gradually. Defoliated trees were still photosynthetically active, even when experiencing severe drought stress. Correspondingly, we found significant differences between curve parameters describing the defoliated and nondefoliated tree curves (Fig. 4b, Supplementary data Fig. S4, S5).
Experiment 3: Ploidy effects under drought recovery
Diploid and triploid trees differed in growth responses during recovery from extensive drought stress. Across defoliation treatments, triploids produced 35% more new tissue than did diploid trees (Fig. 5). None of the other growth traits was affected by cytotype differences (Fig. 5, Supplementary data Fig. S6, Table S3). In contrast, genotype significantly affected all growth traits except new tissue production and explained 10-20% of the variation in the affected traits (Fig.5, Supplementary data Fig. S6, Table S3).
Trees that were defoliated prior to drought showed higher growth responses during drought recovery than trees that were not defoliated. Defoliated trees allocated more of their stem mass to height growth (increased by 10%), produced 45% more new tissue during recovery, and exceeded nondefoliated trees by 19% in total weight at the end of the experiment (Supplementary data Fig. S6). Defoliation also caused trees to invest more in leaf production (leaf weight at a given tree weight increased by 30%) and in thicker, smaller leaves (SLA decreased by 9%) (Fig. 5, Supplementary data Fig. S6). Cytotypes as well as genotypes responded similarly to the defoliation treatment (no ploidy × treatment or genotype × treatment interactions).
DISCUSSION
In this research, we disentangled two components of genetic diversity, genotypic variation and ploidy level, and quantified their relevance for growth, biomass allocation, morphology, phytochemistry, and physiological traits. Population and genotypic differences typically were stronger than ploidy effects for both a wide array of traits and responses of those traits to drought and defoliation. One key illustration of this is observed variation in tree growth, along with it physiological, morphological and allocational determinants, among nonstressed trees in Experiment 1. Ploidy level had no clearly discernible influence on growth, which differed substantially across populations as well as genotypes within each population. This variation was consistent with corresponding population/genotype differences in LWR, SLA, and, to a lesser extent, Aarea, the three principal growth determinants in young trees (Kruger and Volin, 2006). Only a few traits, such as drought resilience, photosynthetic performance and levels of certain chemical defenses, showed subtle, albeit significant, differences between diploid and triploid aspen. Our research highlights that, despite the recognized importance of ploidy for tree ecology (Diatta et al., 2022; Blonder et al., 2021; Park et al., 2021), ploidy effects may pale in contrast to genotypic effects in determining functional trait expression and trait responses to stress in highly diverse tree species. Similar findings were recently reported by DeRose et al. (2022) for natural populations of mature aspen in northern Utah.  
Trembling aspen is known for its exceptionally high genetic and trait variation (Callahan et al., 2013; Mitton and Grant, 1996; Jelinski and Cheliak, 1992). Our finding that the vast majority of trait variation was driven by genotypic rather than ploidy differences confirms the well-documented importance of genotypic diversity for determining tree environmental adaptation and evolution (Petit and Hampe, 2006; Westerband et al., 2021).
Regarding ploidy effects on phytochemistry, the observed difference in foliar tremulacin level between triploids and diploids (Supplementary data Fig. S2, Table S2) accords with results from studies on other plant systems finding that polyploidy can lead to altered secondary chemistry (Te Beest et al., 2012; Park et al., 2021; Gaynor et al., 2020). The ploidy effect on tremulacin level, however, was moderate (absolute differences of ~1.5% dw  across defoliation and drought treatments) and mainly driven by a single triploid genotype that showed increased defense levels. In contrast to our results, DeRose et al. (2022) reported for mature aspen in northern Utah that triploids exhibited 21% higher levels of salicinoid phenolic glycosides than diploids. Similar to our findings, though, expression of condensed tannins did not differ between cytotypes (DeRose et al., 2022).
The influence of ploidy level on light-saturated photosynthesis in our Wisconsin genotypes (Fig. 4a) is also generally consistent with previous findings. Greer et al. (2018) reported that mature triploid aspen clones in the Rocky Mountains exhibited higher net carbon assimilation rates than did diploid clones. In contrast to our results, however, the key driver of enhanced photosynthetic performance in mature triploids was increased stomatal conductance rather than heightened rubisco activity (Vcmax, Supplementary data Fig. S3). Regardless, given the central role that leaf photosynthetic performance plays in plant carbon balance, differences in photosynthesis between diploid and triploid trees might ultimately drive other trait-level divergences between the two cytotypes. For example, increased photosynthetic rates in triploids possibly contributed, via greater availability of assimilated carbon, to higher tremulacin levels in triploid foliage. Enhanced photosynthetic activity would also likely increase biomass growth in triploids, but corresponding growth differences were not observed in our Experiment 2. We acknowledge, however, that our ability to thoroughly assess links between growth and its determinants is precluded by the absence of data on belowground biomass allocation in that experiment. Other studies on mature aspen have found that triploid genotypes exceed diploids in stem growth rates and stem diameter (DeRose et al., 2015).
The absence of marked cytotype differences in trait response to defoliation and/or recurring drought in our study (no ploidy × treatment interactions) agrees with results from research on other plant species such as strawberry, Fragaria vesca (Wei et al., 2020) the herbaceous perennial Rhodohypoxis baurii (Mtileni et al., 2021), and the annual herb Mercurialis annua (Vilas and Pannell, 2017). Yet a growing body of research documents that polyploidy in plants confers a benefit under environmental stress (Van de Peer et al., 2021). For example, allotetraploid grass (Brachypodium distachyon) (Manzaneda et al., 2012), autotetraploid Rangpur lime hybids (Citrus limonia × Citrus sinensis) (Allario et al., 2013) and polyploid gum Arabic tree (Acacia senegal) (Diallo et al., 2016) were more drought tolerant than their diploid counterparts. Several reasons may explain the discrepancy between our results and the aforementioned examples. Firstly, the extent to which polyploidy can confer increased stress tolerance can differ greatly among plant species (Van de Peer et al., 2021; Ramsey and Ramsey, 2014). Moreover, the importance of cytotype for stress adaptation may depend on the source of cytotype variation (e.g. polyploids that arise within a species (autopolyploidy) or from hybridization of two distinct species (allopolyploidy)), the number of multiplied chromosome sets (triploidy, tetraploidy, etc.) or gene dosage effects (Bastiaanse et al., 2019; Van de Peer et al., 2017; Van de Peer et al., 2021).Triploid and diploid genotypes from Wisconsin did not differ in their response to immediate drought stress (Experiment 2). However, the importance of polyploidy for stress responses may depend on the ecological context in which a plant population evolved (Van de Peer et al., 2021). Ploidy differences in drought-stress response may therefore be more pronounced in genotypes that evolved in more arid regions such as the Intermountain West (Utah population). Finally, ploidy effects have rarely been explored in the context of intraspecific genotypic variation. Our findings underscore the importance of considering conventional genotypic variation when exploring the ecological relevance of ploidy effects in future studies.
Cytotype differences affected trait response to environmental stress only in one case: triploid genotypes displayed a slightly greater drought resilience than did diploid genotypes, as measured via increased production of new lateral shoot tissue during post-drought recovery (Fig. 5). As a consequence, when growing in regions frequented by drought, young triploid trees might outcompete young diploid trees. A cytotype difference in drought resilience could help explain the comparative abundance of triploid clones in the arid regions of North America (Mock et al. 2012; but see Latutrie et al. (2019)). However, recent studies by Blonder et al. (2021), Blonder et al. (2022) and (Greer et al., 2018) found that triploid mature aspen appear to be more susceptible to drought stress than diploids. These findings suggest that cytotype effects on drought-tolerance in aspen may vary among developmental stages.
Defoliation rivaled or exceeded polyploidy with respect to the magnitude of its positive impact on drought resilience (Fig. 5, Supplementary data Fig. S6). This outcome can be explained in part by the fact that leaves of defoliated trees sustained stomatal opening and photosynthesis at soil moisture deficits well beyond the threshold causing pronounced stomatal closure (and decreased photosynthesis) in foliage of nondefoliated trees (Fig. 4b, Supplementary data Fig. S5). A similar influence of defoliation on leaf stomatal and photosynthetic responses to water stress has been observed in other studies (McGraw et al., 1990; Pinkard et al., 2011). It may stem from the effects of defoliation on the fundamental balance between a tree’s transpirational demand and its intrinsic capacity to acquire and transport water. Namely, defoliation would immediately decrease the ratios of leaf area to root surface area and sapwood cross-sectional area, both of which are thought to exert considerable control over stomatal behavior (Sperry et al., 1998). Moreover, the latter appears to have governed the photosynthetic drought response in our study, as neither the slope nor intercept of the photosynthesis-conductance relationship differed significantly between control and defoliated trees, regardless of water status (P > 0.65, data not shown). Consistent with the observed growth responses of droughted trees in this study (Fig. 2) and others (Pinkard et al., 2011; Jacquet et al., 2014), a comparative increase in leaf carbon gain would help mitigate the negative effects of defoliation on biomass growth. Correspondingly, the increase would allow defoliated trees to accumulate more carbohydrate reserves and, during subsequent drought recovery, invest more resources into biomass production relative to nondefoliated trees. These phenomena and their underlying mechanisms warrant further investigation, as we have not found published corroboration of this impact of defoliation impact on drought recovery.
In conclusion, our study shows that in young trembling aspen, genotypic differences were key determinants of functional trait variation, and the magnitude of that variation obscured potential ploidy effects. Upon accounting for genotypic effects, only a few traits differed between diploid and triploid trees. Triploid trees exhibited slightly higher levels of defense and photosynthesis, as well as higher resilience to extended drought events. Hence, while genotypic differences are key determinants of aspen’s adaptation to environmental stress, young triploid trees might possess additional subtle advantages when confronted with drought or herbivory. With the advance of climate change, forests are increasingly exposed to both environmental stressors (Aber et al., 2001; Seidl et al., 2017). Accordingly, the possibility that herbivore-mediated defoliation, in addition to genetic determinants, can buffer negative drought effects is an important consideration when assessing the consequences of future stressors on forest ecosystems.