Temperate tree seedlings have similar drought vulnerability despite having different hydraulic drought responses in adults

Climate change is projected result in higher frequencies of drought events across the world and lead to reduced performance in many temperate tree species. However, many studies in this area focus specifically on adult tree drought responses and overlook how trees in other age classes might differ in their vulnerability. Evidence shows that seedling drought response can differ from that of adults and furthermore that demographic performance in the seedling age class will have disproportionately strong effects on the assembly dynamics of future forests, together suggesting that understanding seedling drought responses will be critical to our ability to predict how forests will respond to climate change. In this study, we measured four indices of hydraulic response to drought (leaf water potential, photosynthetic capacity, non-structural carbohydrate concentration, and hydraulic conductivity), as well as interaction effects with shade treatments, for seedlings of two temperate tree species that differ in their adult drought response: isohydric Acer saccharum and anisohydric Quercus rubra. We found a strong isohydric response in A. saccharum seedlings that included conservation of leaf water potentials (>-1.8 MPa) and reductions in non-structural carbohydrate concentrations consistent with reduction of stomatal conductance. Quercus rubra seedlings were able to survive to more negative water potentials, but only rarely, and they showed a similar reduction in photosynthetic capacity as was found for A. saccharum. Our results suggest that, although Q. rubra seedlings display some anisohydric responses to drought, they are more isohydric than adults. Both species seem to be relatively similar in their vulnerability to drought despite the differences predicted from adult drought response, and our results suggest that seedlings of both species will be similarly vulnerable to future drought events.

class will have disproportionately strong effects on the assembly dynamics of future forests, 23 together suggesting that understanding seedling drought responses will be critical to our ability 24 to predict how forests will respond to climate change. In this study, we measured four indices of 25 hydraulic response to drought (leaf water potential, photosynthetic capacity, non-structural 26 carbohydrate concentration, and hydraulic conductivity), as well as interaction effects with shade 27 treatments, for seedlings of two temperate tree species that differ in their adult drought response: 28 isohydric Acer saccharum and anisohydric Quercus rubra. We found a strong isohydric response 29 in A. saccharum seedlings that included conservation of leaf water potentials (> -1.8 MPa) and 30 reductions in non-structural carbohydrate concentrations consistent with reduction of stomatal 31 conductance. Quercus rubra seedlings were able to survive to more negative water potentials, 32 but only rarely, and they showed a similar reduction in photosynthetic capacity as was found for 33 A. saccharum. Our results suggest that, although Q. rubra seedlings display some anisohydric 34 responses to drought, they are more isohydric than adults. Both species seem to be relatively 35 similar in their vulnerability to drought despite the differences predicted from adult drought 36 response, and our results suggest that seedlings of both species will be similarly vulnerable to 37 future drought events. 38 39 Introduction 43 Climate change is projected to increase temperatures and affect global patterns of 44 precipitation, with many areas expected to become drier and hotter [1], and these environmental 45 changes have the potential to strongly affect forest ecosystems [2]. Many studies have thus 46 addressed the effects that water availability has on the performance of tree species, but primarily 47 concentrating on adults [3][4][5]. However, in areas where projected climate may not significantly 48 affect adult trees, relatively few studies address the effect that drought has on saplings and 49 seedlings (but see Maguire and Kobe 2015, Kannenberg and Phillips 2020). This represents an 50 important knowledge gap, as past research has demonstrated that drought response can 51 significantly differ across ontogeny [8], suggesting that tree seedlings are likely to be affected by 52 drought differently than their adult counterparts. Relatively small changes in water availability 53 are likely to have profound effects on survival of younger life stages and could consequentially 54 affect forest community assembly dynamics [2,9,10]. Scientists must therefore reconcile these 55 differences to better predict the effects of climate change on forest demography. 56 Drought tolerance is a broad term that encompasses many plant traits including stomatal 57 regulation behavior (e.g., iso/anisohydry), root morphology, xylem anatomy, and leaf abscission 58 behavior [11][12][13][14][15]. Iso/ansiohydry, referring to whether plants close their stomata during drought 59 to limit water loss [11], is an overarching concept that has been used to categorize a broad range 60 of other drought-related traits. Plants are typically sorted along a gradient ranging from isohydric 61 species which exhibit strong stomatal control on one end and anisohydric species which show 62 drought and took longer to recover after the drought ended [41], suggesting that small size 132 classes will have different vulnerability to drought even if they follow the hydraulic strategies of 133 adult trees. Together, results from these studies support the recent call for a full quantification of 134 hydraulic responses to drought [21,22] that is more robust in addressing suites of traits and 135 within the context of ontogenetic differences. 136 Seedlings are the size class most likely to experience directional mortality effects 137 [10,42,43], meaning that differences in species' response to environmental drivers should be 138 strongest in this phase of recruitment. Therefore, improving our understanding of ontogenetic 139 differences in drought response will be critical to forecast changes to forest systems. Predicting 140 future seedling performance will require an accurate estimation of hydraulic traits and behavior; 141 if seedlings tend toward the same hydraulic strategies as their adult counterparts, climate change-142 related drought could accentuate differences in species performance based on these traits. 143 However, if seedlings tend to be more isohydric than adult trees in general, as suggested by 144 Following transplantation to individual pots, seedlings were immediately moved under 207 moderate shade cloth (~40% ambient PAR) and allowed to grow for an entire growing season 208 (summer of 2017) under well-watered conditions. This was done to minimize the effects of first-209 year transplant stress and to allow seedlings the ability to assimilate enough carbon to allocate 210 photosynthate to storage tissue and other labile carbon pools. Seedlings were moved to an 211 outdoor pit following the onset of leaf color change in fall to allow seedling foliar phenology to 212 respond to natural climate conditions. The pit was used to help insulate seedling roots from frost 213 conditions they would otherwise not experience, and seedlings were further insulated by 214 surrounding the pots with potting soil. 215 Seedlings were removed from the pit and moved back into the greenhouse in early spring 216 2018 corresponding to when leaf bud expansion was noted for both species. All pots were moved 217 under one layer of shade cloth and were regularly watered until treatments were implemented. 218 Environmental sensors were added simultaneously with when the seedlings were moved back 219 inside, measuring air temperature and relative humidity (HOBO U23 Pro v2 data loggers) and 220 light (HOBO Pendant data loggers; Onset Computer Corporation; Bourne, MA, USA). 221 Temperature, relative humidity, and light were all measured at 30-minute intervals. Soil moisture 222 was also measured at the individual (pot) level coinciding with harvesting or gas exchange 223 measurement using a FieldScout TDR300 soil moisture meter (Spectrum Technologies; Aurora, 224 IL, USA). 225 We took pre-treatment measurements of photosynthetic capacity and non-structural 226 carbohydrate concentrations approximately four weeks following initial seedling leaf-out (June 227 14 th ), after which treatment conditions were initiated. Our four treatments were drought (D; no 228 additional shade cloth added, seedlings are no longer watered), shade (S; extra shade cloth used 229 to reduce PAR to 10% of ambient, seedlings remained well-watered), shade and drought (DS; 230 extra shade cloth added, seedlings are no longer watered), and a control treatment (C; no 231 additional shade cloth, seedlings remain well-watered). All seedlings acclimated to the study 232 treatments for two weeks before three harvests were made at weekly intervals. Seedlings in each 233 treatment combination were randomly assigned to be harvested for measurement of either xylem 234 conductivity or [NSC]. Each harvest included six seedlings from each group: two for 235 measurement of xylem conductivity and four for measurement of [NSC]. Predawn leaf water 236 potential (ΨPD) was measured on the morning of each harvest before sunrise as an approximation 237 for soil water potential. Water potential was measured using excised leaves and a Scholander 238 pressure chamber (PMS Instrument Company, Albany, OR, USA). 239 240

Gas exchange measurements (Amax) 241
The day before each harvest, for each treatment, we measured gas exchange in two of the 242 four seedlings selected for the NSC analyses. We used an LI-6400 Portable Photosynthesis 243 System equipped with a CO2 mixer assembly, LI-02B LED red/blue light source, and LI-06 PAR 244 sensor (Li-COR Biosciences, Lincoln, NE, USA). We constructed light curves (i.e., A-Q curves) 245 for each plant by recording gas exchange at 1500, 1000, 750, 500, 250, 100, 50, 25, and 0 μmol 246 photons m -2 s -1 at CO2 concentrations of 400 ppm, ambient temperature, and ambient humidity. 247 Maximum photosynthetic capacity (Amax) was calculated using equations published by Marshall 248 and Biscoe (1980) and using the nls command in the stats package in R v3.5.3. This parameter 249 represents the maximum photosynthetic rate that a leaf is capable of under saturating light 250 conditions. Reductions in Amax indicate limitations on the photosynthetic machinery, such as 251 from reduced stomatal conductance [24]. 252 253

Non-structural glucose concentrations [NSCGlu] 254
After ΨPD measurement, tissue from seedlings selected for the NSCGlu analysis were 255 immediately separated into three pools: leaves (including petioles), stem (above root collar), and 256 roots (below root collar). Leaves were microwaved for 180 seconds at 800 watts to stop leaf 257 enzyme activity [25] and then all tissues were transferred to a drying oven and dried for 48 hours 258 at 70 °C. We weighed each sample, ground them using a ball mill, and then stored the samples at 259 20 °C in airtight containers. We measured out 50 mg of each sample into screw-top conical tubes 260 and soluble sugars were extracted according to Quentin et al. (2015) using repeated incubation 261 and centrifuging in 80% ethanol. We measured glucose concentrations using a phenol-sulfuric 262 acid colorimetric assay as described by Dubois et al. (1956). Glucose concentration was 263 measured against a glucose standard curve using absorbance measured at 490 nm. Measurements 264 were then converted to units of mg glucose per gram of dry tissue. 265 266 Xylem conductivity (k) 267 Following ΨPD measurement, seedlings harvested for xylem conductivity measurements 268 were removed from their pots and their root balls were soaked in water for 10 minutes to 269 alleviate stress on the water column. We then cut stem segments from each plant underwater, 270 recording the length of each segment and its average diameter. Conductance was measured using 271 protocol established by Kolb et al. (1996). We slightly modified this protocol by using 20 mM 272 KCl solution filtered to 0.22 μm (used to prevent microbial growth within the system that could Leaf water potential measurements (ΨPD) taken over the span of the experiment (Fig 1) were 294 used to assess the range of water potentials experience by seedlings of each species. 295 Measurements were pooled across all seedlings harvested for both NSC and xylem conductivity 296 experiments. We conducted a one-way ANOVA to assess whether ΨPD differed significantly 297 between the two species. We modeled maximum photosynthetic capacity (Amax) as a function of light treatment and of leaf 307 water potential (our proxy for drought effects); thus we combined seedlings from the two shaded 308 treatments (S and DS) as well as from the two unshaded treatments (C and D) to assess the two 309 light treatments and then used the full range of observed water potentials to account for drought 310 treatments. Photosynthetic capacity for seedling i was estimated from a normal likelihood: 311 And an exponential process model that described well the reductions in Amax observed in the data 313 (Fig 2a-b): 314 Parameter β1 represents the maximum photosynthetic rate at water potential equal to zero (i.e., 316 full water availability). Parameter β2 indicates the decay rate at which Amax changes in response 317 to changes in ΨPD; we used this parameter to assess reductions in photosynthetic capacity. Both 318 parameters were estimated for each shade treatment. The model was also evaluated for the 319 effects of seed source and initial height, but they did not improve model fit and so we did not 320 include them in the final model. Since seedling glucose concentrations did not change significantly over the duration of 343 the harvest periods (Fig S1), data was pooled across harvest dates for which seedlings from all 344 treatments were still alive (harvest 1 for A. saccharum and harvests 1 and 2 for Q. rubra). Non-345 structural glucose concentrations ([NSCGlu]) were analyzed using ANOVA that estimated the 346 effects of treatment on values pooled by tissue type (leaf, stem, and root), or averaged across all 347 pools (using averages weighted by the mass of each pool). Analyses were conducted separately 348 for each pool x species combination. We only included data for seedlings that were recorded as 349 alive at the time of harvest because most of the dead seedlings appeared to have died from 350 hydraulic failure, and therefore could skew the results. 351

Xylem conductivity 353
For each species, we analyzed the differences in xylem conductivity between seedlings 354 harvested before and after the initiation of drought and shade treatments. We limited our analysis 355 to compare differences only using seedlings harvested in the second harvest period (three weeks 356 after the initiation of treatments) because there were no surviving seedlings in either of the two 357 drought treatments past this harvest. We used initially carried out an ANCOVA to estimate the We found some significant differences in height and mass between populations for each species 375 (summarized in Appendix S1), but they did not significantly affect any of our analyses and are 376 not included in the following results. 377 378 Predawn leaf water potential (ΨPD) 379 Leaf water potential decreased under the drought treatments for both species (Fig 1) with Q. 380 rubra seedlings reaching more negative water potentials (-3.28 MPa) than A. saccharum 381 seedlings (-1.76 MPa). However, we did not find any significant difference in ΨPD between the 382 two species (Pr(>F) = 0.291). All A. saccharum seedlings in the D and DS treatments died before 383 the second harvest (~21 days after treatments were initiated) and all Q, rubra seedlings died 384 before the third harvest (~28 days after treatment initiation). 385

386
Photosynthetic capacity 387 Our model fit (r 2 of predicted vs observed) was 0.289 for A. saccharum and 0.305 for Q. 388 rubra. Posterior parameter estimates can be found in Table S1. There were no significant 389 differences in posterior estimates of intercept parameter β1 between light treatments for either 390 species (Fig 2c), although values were higher in the shade treatment for Q. rubra. Decay 391 parameters β2 were statistically significant, different from zero, for both shade treatments for Q. 392 rubra but was only significant for the unshaded treatment in A. saccharum (Fig 2d). These decay 393 parameters did not significantly vary between light treatments for either of the species. Predicted 394 Amax decreased from 3.375 μmol m -2 s -1 (± standard deviation of 0.692) at ΨPD = 0 MPa to 1.309 395 ± 0.413 μmol m -2 s -1 at ΨPD = -1 MPa for unshaded A. saccharum seedlings, a decrease of 61.2% 396 (Fig 2a). This drop was proportionally smaller in the shade treatment, with Amax decreasing from 397 2.936 ± 0.568 to 2.577 ± 0.929 μmol m -2 s -1 (12.3% decrease). This trend was the opposite for Q. 398 rubra seedlings (Fig 2b), where predicted Amax decreased in light conditions from 2.775 ± 0.383 399 to 1.893 ± 0.325 μmol m -2 s -1 (31.8% decrease) and in shade conditions from 4.934 ± 0.746 to 400 1.137 ± 0.411 μmol m -2 s -1 (77% decrease). 401 402

Non-structural carbohydrates 403
Non-structural glucose concentrations ([NSCGlu]) showed a general decrease over time across all 404 treatments and pools for both species (Fig 3), but there were no significant differences between 405 treatments for any of the species x pool combinations (Table S2). Some carbon pools 406 significantly decreased from the pre-treatment values (Fig 3). There were no statistically 407 significant drops in [NSCGlu] between control and treatment seedlings in either of the two 408 species, and in some cases mean [NSCGlu] values were higher in the treatments than the control 409 (particularly for Q. rubra , Fig 3e-h). There were no significant differences in conductivity between any of the treatments for 422 either species (Table 1a; Fig S4). Average A. saccharum conductivity was 0.075 ± 0.016 g s -1 423 MPa -1 mm -1 (mean ± s.d.) in the control treatment and 0.073 ± 0.027 g s -1 MPa -1 mm -1 in the 424 shade treatment (n = 4 in each treatment). There were no A. saccharum seedlings that survived in 425 the drought treatments that could be used in this analysis. Quercus rubra average conductance 426 was 0.076 ± 0.024 g s -1 MPa -1 mm -1 in the control treatment, 0.041 g s -1 MPa -1 mm -1 in the 427 drought treatment, 0.082 ± 0.011 g s -1 MPa -1 mm -1 in the shade treatment, and 0.043 ± 0.021 g s -1 428 MPa -1 mm -1 in the combined drought and shade treatment (n = 3, 1, 4, and 3, respectively). We 429 found a significant effect by drought treatment for Q. rubra seedlings when the different shade 430 treatments were grouped together (Table 1b, Fig S5b), with seedlings of this species exposed to 431 drought having significantly lower hydraulic conductivity (0.043 ± 0.017 g s -1 MPa -1 mm -1 ) 432 compared to those in the well-watered treatments (0.079 ± 0.021 g s -1 MPa -1 mm -1 ). Acer 433 saccharum seedling conductivity in the combined well-watered treatments was 0.074 ± 0.021 g 434 s -1 MPa -1 mm -1 . 435 436 In this experiment we measured the hydraulic strategies used by seedlings of two 455 dominant tree species that commonly co-occur across a wide range of eastern North American 456 forests, and that differ in their response to drought (as measured in adults). We quantified four 457 commonly measured indicators of drought tolerance (leaf water potential, photosynthetic 458 capacity, non-structural carbohydrate concentrations, and hydraulic conductance) over the 459 duration of a greenhouse dry-down experiment. We used two different light treatments to 460 investigate the potential interaction effects between shade and drought on tree seedling 461 performance, since shade can both exacerbate carbon starvation via low light levels and/or 462 ameliorate drought stress via decreasing temperature. 463 Our results indicate that, as predicted from adult characteristics, A. saccharum seedlings 464 appear to be slightly more vulnerable to drought due to their inability to survive past relatively 465 moderate reductions in soil water potential (Fig 1a), whereas Q. rubra were recorded to survive 466 to considerably lower levels (Fig 1b). Quercus rubra seedlings were also able to survive longer 467 in the drought treatments, but only by about a week. Still, we did not find statistically significant 468 differences between species in any of the drought indicators, suggesting that seedlings of these 469 two species are still similarly vulnerable to drought. However, due to the relatively small sample 470 sizes used in this study, it is important to be cautious with these results and the conclusions 471 drawn from them. Overall, our results suggest that the increased drought frequency predicted for 472 (e.g., by giving trees access to more resources or by helping them resist cavitation), but there is 485 strong evidence that tree performance in drought conditions is strongly determined by tradeoffs 486 between these two processes [20]. Furthermore, juvenile trees may not be able to make use of the 487 same mitigating strategies used by adults due their size and relative lack of access to resources 488 (i.e., deep water sources; Cavender-Bares and Bazzaz 2000). Hydraulic outcomes at the seedling 489 level may therefore be more similar among species than they are at larger size classes. 490 Our results do not fully support the idea that seedling hydraulic strategies are similar to 491 those of the adults; while leaf water potentials reached lower levels in Q. rubra seedlings, as we 492 expected, the other responses to drought did not differ between the two species. First, while adult 493 Q. rubra (as well as other Quercus species in general) respond to drought stress by maintaining 494 photosynthetic capacity at the cost of reduced leaf water potential [8,24], we found that seedlings 495 of this species exhibited declines in photosynthetic capacity that began at ΨPD < -1 MPa (Fig 2b). 496 This was not significantly different from the trend found in A. saccharum seedlings (Fig 2a), 497 which matched the photosynthetic response demonstrated in adults of this species [24]. This 498 relatively isohydric response of Q. rubra seedlings agrees with previous work done by  Bares and Bazzaz (2000) and provides support for the idea that seedling hydraulic strategies will 500 be more similar between species than they are in conspecific adults. 501 Adults of these two species have also been shown to have different wood densities, with 502 A. saccharum having dense wood and diffuse-porous xylem and Q. rubra having ring-porous 503 xylem and wood that is less dense [14]. Diffuse-porous xylem can help trees avoid embolism 504 formation due to the narrower conduits, but they are also more difficult to refill after embolism 505 occurs [31]. Narrow xylem conduits and strong stomatal control help Acer species maintain 506 hydraulic conductance during drought [72,73]. In contrast, Quercus species are more prone to 507 gradual but significant declines in conductivity [40,73,74]. We hypothesized that A. saccharum 508 seedlings would demonstrate stricter control of internal water potentials compared to Q. rubra, 509 and thus show little variation in the conductivity of living seedlings, whereas Q. rubra seedlings 510 would show a gradual decline in conductivity associated with anisohydric stomatal regulation. 511 We found tentative support for this hypothesis with respect to Q. rubra seedlings, for 512 which conductivity appeared to decrease gradually beginning at ΨPD < -1 (Fig 4), which closely 513 resembles the pattern seen in conspecific adults [74] and agrees with previous work done on 514 seedlings of this species [40]. However, we were not able to fully quantify this trend in a more 515 complex analysis because we did not measure percent loss in conductivity (sensu Kolb et al. 516 1996) and because we had limited survival in our seedlings. There was high mortality for this 517 species past the -1 MPa threshold, suggesting that these seedlings are still vulnerable to drought. 518 Acer saccharum seedlings supported the hypothesis, with conductivity maintained at ΨPD > -1 519 and hydraulic failure past that point (Fig 4), which agrees with the strategy used by adults. We 520 observed seedlings of this species surviving to slightly more negative ΨPD in the NSC harvests 521 (ΨPD =-1.76 MPa , Fig 1), suggesting that reductions in conductivity may follow a trend more 522 similar to that of the Q. rubra seedlings, but conductivity was not quantified for these individuals 523 and thus we lack the evidence needed to better support this conclusion. We found no significant differences in non-structural glucose between treatment 541 conditions for A. saccharum in any of the tissue pools, and limited differences between treatment 542 and pre-treatment [NSCGlu] (Fig 3a-d). This suggests that seedlings were generally able to 543 maintain their labile glucose pools throughout the experiment. This is consistent with a previous 544 study that worked with Acer rubrum seedlings (which are closely related to A. saccharum and 545 typically exhibit a similar hydraulic strategy; Davies and Kozlowski 1977), which found 546 reductions in soluble sugar concentrations over time in similar treatments [6]. We also found no 547 significant differences between treatments in any of the Q. rubra NSCGlu pools (Fig 3e-h suggests that seedlings of this species experienced similar stress across all treatments. 552 The lack of a significant difference between our control treatments and the three stress 553 treatments prevents us from being able to make strong conclusions about how drought will affect 554 [NSCGlu] for seedlings of either species. The difference between our results and the results from 555 other studies that have found significant effects in similar experiments (e.g., Maguire and Kobe 556 2015) could be due to the extremely high temperatures experienced by seedlings in the 557 greenhouse environment (Fig S3) Contrary to other tree seedling studies (e.g., Piper and Fajardo 2016), we did not find any 574 significant effects associated with light availability in any of our drought response indicators. 575 Although including light treatments improved our photosynthetic capacity model's performance, 576 there were no significant differences between light condition effects for either β1 (photosynthetic 577 capacity at saturating water availability) or β2 (the decay rate parameter) (Fig 2c-d). There were 578 no significant shade effects in any of the glucose pools (Fig 3) and hydraulic conductance was 579 better predicted in an analysis that explicitly omitted light treatment and focused solely on 580 differences in drought treatment (Fig S5). The lack of a shade effect on glucose concentrations is 581 consistent with previous research that found no significant change in soluble sugar 582 concentrations over time in seedlings [6] and saplings [7] of temperate tree species common in 583 our study region. However, both studies found significant reductions in starch concentrations that 584 suggest that plants prioritize the mobilization of starch for use in metabolism over the 585 consumption of soluble sugars, potentially due to the importance of soluble sugars in 586 osmoregulation [12,20]. We did not measure starch concentrations, so we can only speculate that 587 this mechanism may have affected our results. The extreme temperatures experienced in the 588 greenhouse could have also created respiration demands across all treatments that overwhelmed 589 any signal that we might have otherwise observed with differences in light availability [76]. Still, 590 our results do not support the existence of either mitigating or exacerbating effects of shade on 591 hydraulic performance of temperate tree seedlings. 592 593 3) What are the implications for seedling demographic performance under climate change? 594 Altogether, the results from our experiment suggest a convergence in the vulnerability of 595 temperate tree seedlings to drought that is inconsistent with drought responses expected from 596 evidence collected in adult trees of the same species. There were no drought treatment A. 597 saccharum seedlings that survived longer than two weeks past the initiation of experimental 598 treatments and no water-limited Q. rubra seedlings that survived past three weeks. Although our 599 stress treatments were harsh (and also likely exacerbated by high temperatures) the lack of a 600 significant difference between seedling responses is consistent with results from previous 601 research [6]. Still, the relatively small sample size of seedlings surviving in this experiment could 602 have affected our ability to pick up on differences between the two species at finer scales of time, 603 temperature, and soil moisture, and it therefore prevents us from making stronger conclusions 604 about seedling drought dynamics. 605 Even so, with no clear distinction in seedling vulnerability to drought, potential 606 differences in seedling demography and recruitment between species within the context of 607 climate change are more likely to be driven by other factors. For example, we have previously 608 shown that access to spring light [48] and the capacity to track climate change with spring leaf 609 out phenology [47] differs between species and that this mechanism can help account for carbon 610 starvation dynamics expected under hotter and drier summers. The lack of a difference between 611 species further indicates that seedling drought vulnerability is decoupled from adult vulnerability 612 and suggests that future community assembly in temperate forests will not be strongly limited by 613 the vulnerability of tree seedlings to drought. This places a stronger emphasis on traits and 614 behaviors that do differ between species and that could lead to differential outcomes during 615 climate change.