This is the accepted version of the journal article: Laughlin, Daniel C.; Siefert, Andrew; Fleri, Jesse; [et al.]. «Rooting depth and xylem vulnerability are independent woody plant traits jointly selected by aridity, seasonality, and water table depth». New Phytologist, Vol. 240, issue 5 (Dec. 2023), p. 1774-1787. DOI 10.1111/nph.19276 This version is available at https://ddd.uab.cat/record/285094 under the terms of the license 1 Rooting depth and xylem vulnerability are independent woody plant traits jointly selected 2 by aridity, seasonality, and water table depth 3 4 Daniel C. Laughlin1*, Andrew Siefert1, Jesse Fleri1, Shersingh Joseph Tumber-Dávila2, William 5 M. Hammond3, Francesco Maria Sabatini4,5, Gabriella Damasceno6,7, Isabelle Aubin8, Richard 6 Field9, Mohamed Z. Hatim10,11, Steven Jansen12, Jonathan Lenoir13, Frederic Lens14,15, James K. 7 McCarthy16, Ülo Niinemets17, Oliver L. Phillips18, Fabio Attorre19, Yves Bergeron20, Hans 8 Henrik Bruun21, Chaeho Byun22, Renata Ćušterevska23, Jürgen Dengler24,25, Michele De 9 Sanctis26, Jiri Dolezal27,28, Borja Jiménez-Alfaro29, Bruno Herault30,31, Jürgen Homeier32,33, Jens 10 Kattge34,35, Patrick Meir36,37, Maurizio Mencuccini38,39, Jalil Noroozi40, Arkadiusz Nowak41,42, 11 Josep Penuelas43,44, Marco Schmidt45, Željko Škvorc46, Fahmida Sultana47, Rosina Magaña 12 Ugarte48, Helge Bruelheide7,6 13 14 1 Department of Botany, University of Wyoming, Laramie, WY USA 15 2 Harvard Forest, Harvard University, Petersham, MA USA 16 3 Agronomy Department, University of Florida, Gainesville, FL, 32611, USA 4 BIOME Lab, Department of Biological, Geological and Environmental Sciences (BiGeA), Alma Mater Studiorum University of Bologna, Via Irnerio 42, 40126, Bologna, Italy. 17 5 Czech University of Life Sciences Prague, Faculty of Forestry and Wood Sciences, Kamýcká 18 129, 165 21 Praha 6 – Suchdol, Czech Republic 19 6 German Centre for Integrative Biodiversity Research (iDiv), Halle-Jena-Leipzig, Leipzig, 20 Germany 21 7 Martin-Luther University, Institute of Biology and Geobotany and Botanical Garden, Halle- 22 Wittenberg, Halle, Germany 23 8 Great Lakes Forestry Centre, Canadian Forest Service, Natural Resources Canada 24 9 School of Geography, University of Nottingham, University Park, Nottingham, NG7 2RD, UK 25 10 Plant Ecology and Nature Conservation Group, Environmental Sciences Department, 26 Wageningen University and Research, Wageningen, The Netherlands 27 11 Botany and Microbiology Department, Tanta University, Egypt 28 12 Ulm University, Institute of Botany, Albert-Einstein-Allee 11, Ulm, 89081, Germany 29 13 UMR CNRS 7058, Ecologie et Dynamique des Systèmes Anthropisés (EDYSAN), Université 30 de Picardie Jules Verne, Amiens, France 31 14 Naturalis Biodiversity Center, Darwinweg 2, 2333 CR Leiden, The Netherlands 1 32 15 Leiden University, Institute of Biology Leiden, Plant Sciences, Sylviusweg 72, 2333 BE 33 Leiden, The Netherlands 34 16 Manaaki Whenua – Landcare Research, Lincoln 7640, New Zealand 35 17 Estonian University of Life Sciences, Crop Science and Plant Biology, Tartu, Estonia 36 18 School of Geography, University of Leeds, Leeds LS2 9JT, U.K. 37 19 Sapienza University of Rome, Department of Environmental Biology 38 20 Institut de recherche sur les forêts Université du Québec en Abitibi-Témiscamingue,445 boul. 39 de l'université, Rouyn-Noranda québec, Canada J9X5E4 40 21 Department of Biology, University of Copenhagen, Denmark 41 22 Andong National University, Department of Biological Science 42 23 Institute of Biology, Faculty of Natural Sciences and Mathematics Ss. Cyril and Methodius 43 University 44 24 Vegetation Ecology Research Group, Institute of Natural Resource Sciences (IUNR), Zurich 45 University of Applied Sciences (ZHAW), Wädenswil, Switzerland 46 25 Plant Ecology, Bayreuth Center of Ecology and Environmental Research (BayCEER), 47 University of Bayreuth, Bayreuth, Germany 48 26 Sapienza University of Rome, Department of Environmental Biology 49 27 Department of Functional Ecology, Institute of Botany, Czech Academy of Sciences, Trebon, 50 Czech Republic 51 28 Department of Botany, Faculty of Science, University of South Bohemia, Ceske Budejovice, 52 Czech Republic 53 29 Biodiversity Research Institute (Univ.Oviedo-CSIC-Princ.Asturias), Mieres, Asturias, Spain 54 30 CIRAD, UPR Forêts et Sociétés, F-34398 Montpellier, France 55 31 Forêts et Sociétés, Univ Montpellier, CIRAD, Montpellier, France 56 32 University of Goettingen, Plant Ecology and Ecosystems Research, Goettingen, Germany 57 33 HAWK University of Applied Sciences and Arts, Resource Management, Goettingen, 58 Germany 59 34 Max Planck Institute for Biogeochemistry, Jena, Germany 60 35 German Centre for Integrative Biodiversity Research (iDiv), Halle-Jena-Leipzig, Leipzig, 61 Germany 62 36 School of Geosciences, University of Edinburgh, UK 63 37 Research School of Biology, Australian National University 64 38 CREAF, Barcelona (Spain) 65 39 ICREA, Barcelona (Spain) 66 40 Department of Botany and Biodiversity Research, Universitiy of Vienna 67 41 Polish Academy of Sciences, Botanical Garden 68 42 University of Warmia and Mazury, Department of Botany and Nature Protection 69 43 CSIC, Global Ecology Unit CREAF-CSIC-UAB, Bellaterra, Barcelona, Catalonia, Spain 70 44 CREAF, Cerdanyola del Vallès, Barcelona, Catalonia, Spain 71 45 Palmengarten der Stadt Frankfurt am Main 72 46 University of Zagreb, Faculty of Forestry and Wood Technology 73 47 Department of Forestry and Environmental Science, Shahjalal University of Science and 74 Technology, Sylhet, Bangladesh 75 48 Department of Pharmacology, Pharmacognosy and Botany, Botany Unit, Faculty of Pharmacy, 76 Universidad Complutense de Madrid 77 2 78 Running title: Rooting depth and xylem vulnerability 79 80 Article Type: Full Paper 81 82 Correspondence: 83 Daniel Laughlin. Email: daniel.laughlin@uwyo.edu 84 Department of Botany, University of Wyoming, Laramie, WY USA 85 86 3 87 Summary 88 • Evolutionary radiations of woody taxa within arid environments were made possible by 89 multiple trait innovations including deep roots and embolism-resistant xylem, but little is 90 known about how these traits have coevolved across the phylogeny of woody plants or 91 how they jointly influence the distribution of species. 92 • We synthesized global trait and vegetation plot datasets to examine how rooting depth 93 and xylem vulnerability across 188 woody plant species interact with aridity, 94 precipitation seasonality, and water table depth to influence species occurrence 95 probabilities across all biomes. 96 • Here we show that xylem resistance to embolism and rooting depth are independent 97 woody plant traits that do not exhibit a trade-off. Resistant xylem and deep roots increase 98 occurrence probabilties in arid, seasonal climates over deep water tables. Resistant xylem 99 and shallow roots increase occurrence probabilities in arid, non-seasonal climates over 100 deep water tables. Vulnerable xylem and deep roots increase occurrence probabilities in 101 arid, non-seasonal climates over shallow water tables. Vulnerable xylem and shallow 102 roots increase occurrence probabilities in humid climates. 103 • Traits related to water uptake and transport explain biogeographic-scale species 104 occurrences, suggesting that responses of deeply rooted vegetation may be buffered if 105 evaporative demand changes faster than water table depth under climate change. 106 107 Keywords: cavitation, drought avoider, drought resistant, embolism, species distribution 108 modeling, water availability 109 4 110 ‘Species that show deviations from statistical correlations (and there are always many such 111 species) may tell us more about the adaptive value of particular characteristics than species that 112 fit the correlation.’ (Sherwin Carlquist, 1975, p. 1) 113 114 Introduction 115 Hotter droughts are expected to become more frequent and severe under most climate change 116 scenarios (Spinoni et al., 2018; Hammond et al., 2022) and they have already resulted in 117 widespread tree mortality (Bauman et al., 2022; Hartmann et al., 2022; McDowell et al., 2022). 118 Understanding which plant species are winners and losers under scenarios of decreasing water 119 availability is critical for improving predictions of vegetation dynamics (Hammond et al., 2019) 120 and ecosystem restoration success (Laughlin et al., 2017). Rooting depth and xylem vulnerability 121 to drought-induced embolism are strongly linked to how woody plant species respond to 122 moisture availability (Aubin et al., 2016; Volaire, 2018) because the physiological link between 123 water uptake and xylem conductance is vital to maintaining photosynthesis in arid climates and 124 during dry seasons. However, little is known about how these traits have coevolved within 125 woody taxa, nor do we know the joint effects of these traits on species distributions and potential 126 responses to global change. 127 Survival in water-limited landscapes can be achieved by resisting, avoiding, or escaping 128 dehydration caused by low soil water potentials (May & Milthorpe, 1962; Fischer & Maurer, 129 1978; Levitt, 1980; Volaire, 2018). First, drought resistance (also called drought tolerance) is 130 achieved by constructing xylem that can resist embolism formation and maintain water 131 conductance to enable CO2 uptake even under extreme negative xylem pressure. Woody plants 132 vary widely in their vulnerability to embolism (Hacke et al., 2001; Maherali et al., 2004; Choat 5 133 et al., 2012), which can be quantified by P50, the xylem pressure (MPa) at which xylem 134 conductance is reduced to 50% of its maximum. Species with lower P50 values (i.e., more 135 negative values) are more resistant to embolism. Second, drought avoidance can be achieved by 136 constructing root systems that access deeper soil water in the unsaturated vadose zone or directly 137 from the water table in the saturated zone (Loheide et al., 2005; Ryel et al., 2008). Maximum 138 rooting depth varies widely among woody plants (Canadell et al., 1996; Jackson et al., 1996). 139 Deep root systems evolved for a variety of reasons, including anchorage, support, and increasing 140 the available soil volume, but here we explore its vital role in water uptake. Shallow-rooted 141 species rely on rainfall directly, whereas deep-rooted species can access water from deeper and 142 older sources of water (Miguez-Macho & Fan, 2021). Both phreatophytic species that tap into 143 water tables and deeply-rooted xerophytic plants that access vadose zone soil water can avoid the 144 drought-induced reductions in soil water potential near the soil surface (Ryel et al., 2008; 145 Miguez-Macho & Fan, 2021). In arid climates, species that develop only shallow roots confront 146 drought directly and must be able to resist embolism formation. If "roots grow only as deeply as 147 needed" (Schenk & Jackson, 2002, p. 481), then in humid climates plants should grow shallow 148 roots, and may even need to restrict their roots to the shallow oxygenated zone to prevent anoxia 149 (Jackson et al., 1996; Fan et al., 2017; Tumber-Dávila et al., 2022). Third, drought escape can be 150 achieved by completing a short life cycle during the wet growing season or by deciduous leaf 151 senescence in the dry season. 152 The large carbon investment into growth and maintenance of deep roots must be offset by 153 carbon gains made possible by having such deep roots, otherwise the strategy would be unlikely 154 to have evolved (Seyfried et al., 2005). The construction costs of deep roots could possibly be 155 offset by lower stem xylem construction costs, and evidence for a trade-off between rooting 6 156 depth and xylem resistance to embolism was demonstrated across 12 species in a humid tropical 157 forest, where shallow-rooted species were more tolerant of low water potentials (Brum et al., 158 2019). However, the relationship between rooting depth and xylem vulnerability across global- 159 scale climatic gradients has never been rigorously quantified. There are well-known examples of 160 deep-rooted species in arid climates that also exhibit remarkably resistant stem xylem (Jackson et 161 al., 2000; Seyfried et al., 2005), suggesting that the carbon costs of constructing both deep roots 162 and resistant xylem are offset by the carbon gains that are achieved by simultaneously avoiding 163 and resisting dehydration. 164 Rooting depth and xylem vulnerability may jointly influence species distributions along 165 gradients of water availability in complex ways (Fig. 1A). Water limitation in plants is affected 166 by the regional macroclimate, precipitation seasonality, and local topography and drainage 167 gradients (Schwinning & Ehleringer, 2001; Fan et al., 2017; Grossiord et al., 2020). Climatic 168 aridity is defined by both the total precipitation relative to potential evapotranspiration, as well as 169 the occurrence of dry seasons during the year. From a hydrological perspective, water from 170 ‘uplands’ (defined as well-drained topographic positions independent of elevation above sea 171 level) flows into local topographic depressions creating shallow water tables in ‘lowlands’ 172 (defined as poorly drained topographic depressions independent of elevation above sea level) 173 (Fig. 1A) (Nobre et al., 2011; Fan et al., 2017). Water table depth is largely independent of 174 regional climate because it is primarily under hydrologic control (Sousa et al., 2022), and plant 175 species responses to regional drought may depend on water table depth (Sousa et al., 2020). For 176 example, phreatophytic plants growing in arid riparian areas experience severe atmospheric 177 vapor pressure deficits yet have ample access to soil water. Conversely, plants growing in coarse, 178 shallow, or excessively drained soils within humid regions will experience local soil water 7 179 deficits despite receiving high amounts of precipitation. In previous studies, species with low P50 180 were associated with arid regional climates (Larter et al., 2017; Laughlin et al., 2020a) and 181 upland topographical positions (Oliveira et al., 2019). Plants with shallow roots tend to occur in 182 humid climates and waterlogged soil, whereas both shallow and deep-rooted species occur in 183 well-drained soil in arid climates (Schenk & Jackson, 2005; Fan et al., 2017). Determining the 184 joint response of both traits to gradients of regional aridity, seasonality, and local water table 185 depth will improve our understanding of vegetation response to drought. 186 In this paper, we combined species-level trait data with two macroclimatic gradients and 187 one local gradient in water availability and asked the following two research questions: (1) Are 188 P50 and maximum rooting depth correlated across the phylogeny of woody plants? and (2) Are 189 species occurrences along regional aridity, seasonality, and local water table depth gradients 190 jointly explained by species P50 and maximum rooting depth? We demonstrate that there is no 191 evidence for a trade-off between drought resistance and avoidance and that each combination of 192 xylem vulnerability and rooting depth values differentially optimizes occurrence probabilities 193 along multiple gradients of water availability. 194 195 8 (A) Three drivers of water limitation 1) Regional macroclimate 2) Precipitation seasonality Seasonality = Coefficient of Variation of Monthly Precipitation Aridity Index = Precipitation / Potential Evapotranspiration Hyper Semi Dry Moist Very Low seasonality High seasonality Arid arid subhumid subhumid Humid arid humid 0.05 0.20 0.50 0.65 0.80 1.5 Deep roots Shallow roots Resistant xylem Vulnerable xylem Shallow roots Deep roots Vulnerable xylem Resistant xylem 3) Local topography and drainage Drought deciduous leaves Uplands Deep water tables Elevation Land surface Green text: Lowlands Shallow water tables Predictions of optimal trait values Water table in each environment Shallow Deep Shallow roots roots roots (B) Geographic distribution of plots (C) Whittaker biome distribution of plots 196 197 Figure 1. (A) Gradients of regional aridity (macroclimate), precipitation seasonality, and local 198 water table depth drive variation of water limitation in plants. Green text reports predictions of 199 optimal trait values along each environmental gradient. Note rooting depth is predicted to be 200 highest at intemediate water table depth (Fan et al. 2017) (B) Biogeographic distribution of 201 vegetation plots from the sPlot 3.0 database around the planet illustrated as density of plots per 202 hexagonal grid cell. (C) Bioclimatic distribution of vegetation plots within Whittaker biome 203 defined by mean annual temperature and total annual precipitation. 204 205 206 Material and Methods 207 Dataset compilation 9 208 We combined the Xylem Functional Traits (XFT) database (Choat et al., 2012) with recently 209 published data to quantify average P50 trait data at the species level (Table S1). We discarded 210 vulnerability curves that were non-sigmoidal in shape or were generated using the air-injection 211 method (Cochard et al., 2013). We merged this P50 dataset with the Root Systems of Individual 212 Plants (RSIP) database based on species (Tumber-Dávila et al., 2022). We calculated maximum 213 rooting depth for each species as the absolute maximum rooting depth recorded for that species 214 in the database. This estimates the potential rooting depth of a species and is analogous to 215 maximum height, another size-dependent trait and key indicator of plant ecological strategies 216 (Díaz et al., 2022). Maximum rooting depth is meaningful because it captures the potential for 217 individuals of a species to construct deep roots. This potential may be more informative than 218 individual trait values, which are highly plastic and noisy, for explaining species occurrence 219 across environmental gradients. Maximum rooting depth was strongly correlated with the median 220 of individual plant rooting depths across species (r = 0.85), and models that used median rooting 221 depth were qualitatively indistinguishable from models that used maximum rooting depth. After 222 standardizing species names according to the World Flora Online taxonomic backbone (Miller & 223 Ulate, 2017) and filtering out two outlying measurements of maximum rooting depth that were < 224 0.3 m, the database contained 903 species (2458 observations) with average P50 and 1089 species 225 (2716 observations) with maximum rooting depth. We merged the datasets based on species 226 names and found that both traits have been measured on 207 woody species (Table S1). For 227 these 207 species, there were 871 observations of P50 and 950 observations of rooting depth 228 (Table S1). 229 We extracted vegetation plots from the sPlot 3.0 database (Bruelheide et al., 2019) that 230 contained at least one of the 207 species with P50 and rooting depth data and had a location 10 231 uncertainty of 1 km or less (508,443 plots). Of the 207 species, 188 were present in at least one 232 plot (1,733,795 occurrences). The selected plots were broadly distributed worldwide and 233 throughout Whittaker's bioclimatic space (Fig. 1B-C). Vegetation plots were assigned to 234 ecoregions using Olson’s updated ecoregion map (Olson et al., 2001; Dinerstein et al., 2017). 235 We defined regional species pools as all of our species that were detected within an ecoregion, 236 and we defined species absences as species that were present in the regional species pool but not 237 detected in the plot itself (Laughlin et al., 2021). For each plot, we extracted the aridity index 238 (the ratio of precipitation-to-potential evapotranspiration) from the Global Aridity Index and 239 Potential Evapo-Transpiration Climate Database (Zomer et al., 2008; Trabucco & Zomer, 2018), 240 water table depth (WTD) from the ‘wtd’ database (Döll & Fiedler, 2008; Fan et al., 2013), and 241 precipitation seasonality (the CV of monthly precipitation) from WorldClim (Fick & Hijmans, 242 2017). The aridity index estimates macroclimatic water availability, where ratios <0.65 indicate 243 arid climates and >0.65 indicate humid climates (Fig 1A). Water table depth varies 244 independently of climate (in this dataset, the correlation coefficient between aridity and WTD is 245 r = 0.13) and is thought to be a major driver of rooting depth distributions (Fan et al., 2017). 246 Of the 207 species for which rooting depth and P50 were available, 105 were evergreen 247 and 102 were deciduous (Table S1). We focus on drought resistance and avoidance in this paper 248 because only a few deciduous species in our dataset were drought-deciduous escapers; the vast 249 majority were cold-deciduous species. Nevertheless, we evaluate deciduous and evergreen 250 species to understand their effects on the overall model results. 251 252 Data Analysis 11 253 To answer question 1, we extracted a fully resolved phylogeny based on the PhytoPhylo 254 megaphylogeny (Qian & Jin, 2016) using the R package V.PhyloMaker version 0.1.0 (Jin & 255 Qian, 2019). We calculated Pagel’s λ under a Brownian motion model of evolution to quantify 256 the phylogenetic signal for each trait (Pagel, 1999) using the phylosig function in the R package 257 phytools version 1.2.0 (Revell 2012). We quantified the phylogenetic covariance between the 258 two traits by fitting phylogenetic generalized least squares (PGLS) regression and a phylogenetic 259 correlation structure to account for the non-independence of species across the phylogeny 260 (Revell & Harmon, 2022). Ancestral state reconstruction used maximum likelihood estimates 261 under a Brownian motion model of evolution. The phylogeny was plotted using the R package 262 phytools (Revell, 2012). 263 To answer question 2, we compared two generalized linear mixed effects models of 264 species presence–absence data using the binomial family and logit link. Both models included 265 species random intercepts to account for the fact that trait data were species-level averages and 266 not measured at the intraspecific level at each location, and ecoregion random intercepts to 267 account for the different numbers of observations in each ecoregion. Model 1 was an 268 environment-only model that included linear and quadratic fixed effects terms for aridity, 269 seasonality, and WTD, and interactions among their linear terms. Model 2 included both the 270 environment and traits, where P50, rooting depth, aridity, seasonality, and WTD, their squared 271 terms, and interactions among linear terms up to a five-way interaction, were included as fixed 272 effects. Aridity index and the absolute value of P50 were square root transformed, seasonality, 273 WTD and rooting depth were log transformed, and all predictors were mean-centered and scaled 274 to a standard deviation of 1 prior to fitting the models. 275 For example, in Model 1: 12 276 𝑦! ~ Bernoulli(𝑝! ) 277 𝑝! = logit "# 2𝛽$ + 𝛽# arid! + 𝛽% arid! % + 𝛽& WTD! + 𝛽' WTD! % + 𝛽( seas! + 𝛽) seas! % 278 + 𝛽* arid! WTD! + +𝛽+ arid! seas! + 𝛽, WTD! seas! + 𝛽#$ arid! WTD! seas! 279 + 𝜂-[!] + 𝛾0[!] = 280 where yi is the ith presence-absence record, pi is the expected occurrence probability, the β’s are 281 regression coefficients, ηj is a random intercept for species j drawn from a normal distribution, 282 N(0, ση), and γk is a random intercept for ecoregion k drawn from a normal distribution, N(0, σγ). 283 This models a two-dimensional surface of P50 and rooting depth and its interaction with a three- 284 dimensional surface of aridity, seasonality, and WTD. The shape of the trait surface changes in 285 relation to aridity, seasonality, and WTD, accounting for trait-by-environment interactions that 286 cause different trait combinations to be selected in different environments. We compared models 287 1 and 2 using a likelihood ratio test to determine the importance of traits for explaining species 288 occurrences when environmental variables were already in the model. The R syntax of Model 2 289 using the bam function in the R package mgcv 1.8.41 (Wood, 2011) was as follows: pres ~ arid * 290 wtd * ps * p50 * rd + arid2 + wtd2 + ps2 + p502 + rd2 + s(spp, bs = "re") + s(eco, bs = "re"). We 291 plotted 90% confidence intervals using posterior simulations from the fitted model. 292 We plotted model predictions of occurrence probabilities at different combinations of 293 environment and trait values. We plotted conditional effects of each trait by illustrating the 294 effects of one trait, while holding the other trait at its mean value, at different combinations of 295 high (95th percentile) or low (5th percentile) aridity, seasonality, and water table depth. To 296 illustrate trait-by-trait interactions, we plotted landscapes of occurrence probabilities across a 297 two-dimensional surface of P50 and rooting depth at different combinations of aridity, 298 seasonality, and water table depth. To test the four hypotheses directly, we fixed the two traits at 13 299 their 5th and 95th percentiles to represent four plant strategies and plotted landscapes of 300 occurrence probabilities along a two-dimensional surface of aridity and WTD at two levels of 301 seasonality. The average occurrence probability across this large dataset is approximately 5%, so 302 predicted probabilities vary above and below this mean value. All analyses used R version 4.2.2. 303 304 Results 305 Phylogenetic correlation 306 Mean P50 and maximum rooting depth exhibited significant phylogenetic signal (Fig. 2), 307 although the signal was weaker for rooting depth (P50: Pagel's 𝜆 = 0.91, P < 0.001; rooting depth: 308 Pagel's 𝜆 = 0.58, P < 0.01). These two traits were uncorrelated (Fig. 3) across the phylogeny of 309 woody plants (PGLS slope = 0.06, SE = 0.07, t = 0.95, P = 0.35), and inspection of intraspecific 310 trait ranges provided no indication that these traits were correlated (Figure S1). Deciduous and 311 evergreen species were well distributed throughout this two-dimensional trait space, but 312 deciduous species had on average more vulnerable xylem than evergreen species (Fig. 3). 313 14 314 315 Figure 2. Xylem vulnerability to embolism (P50) mapped onto the phylogeny of woody plants, 316 where the color of the internal tree branches show vulnerability based on maximum likelihood- 317 based ancestral state reconstruction. Dark colors indicate more resistant xylem and light colors 318 indicate more vulnerable xylem (see legend in the center). The outer bars represent maximum 319 rooting depth (log-scale) for each species, where longer bars represent deeper roots (see lower 320 left for rooting depth scale legend). Pictures of representative species of major clades are 321 illustrated around the phylogeny. 15 322 323 Figure 3. Distribution of species in the bivariate trait space defined by maximum observed 324 rooting depth and mean P50. Deciduous species are red and evergreen species are blue. Vertical 325 and horizontal lines denote the 50th percentiles of the traits. Species names associated with bold 326 symbols were selected by an algorithm to represent as much as the occupied trait space as 327 possible because not all names could easily fit. The four grey points represent the 5th and 95th 328 percentiles of the trait distributions (-7.0 and -1.5 MPa for P50; 12.9 and 0.6 m for rooting depth) 329 to represent each of the four plant strategies described in Figure 1B (see Methods). The curves 330 depict the distribution of values across each of the two trait axes for each leaf type. 331 16 332 333 Models of occurrence probabilities 334 The generalized linear mixed effects model that included traits (see Model 2 results in Table S2) 335 explained 16.7% of the deviance (marginal R2 = 0.24; conditional R2 = 0.70). A likelihood ratio 336 test (LRT) determined that traits were a significant improvement to the model that only included % 337 environmental data as predictors (LRT 𝜒%# = 188,143; P < 0.0001) (see Model 1 results in Table 338 S3). Traits exhibited no significant main effects (i.e., the trait effects depended on the 339 environment), and the quadratic term for rooting depth was not significant. Nearly all trait-by- 340 environment interactions were significant (Table S2). 341 P50 interacted with aridity, seasonality, and water table depth (Fig. 4, Table S2). The 342 effect of P50 on probability of occurrence switched from negative to positive from arid to humid 343 climates, where the probability of occurrence of species resistant to embolism increased in more 344 arid climates (Fig 4A,B). In arid climates with shallow water tables, low P50 values increased 345 occurrence probability in seasonal environments, whereas high P50 values increased occurrence 346 probabilities in non-seasonal environments (Fig 4C). 347 Rooting depth interacted weakly with aridity, seasonality, and water table depth (Fig. 4, 348 Table S2). The effect of rooting depth on probability of occurrence switched from positive to 349 negative in arid to humid climates, where the probability of occurrence of species that avoid 350 drought by constructing deep roots increased in more arid climates (Fig. 4E,F). The interaction 351 was weak because the importance of shallow roots in humid climates increased slightly in sites 352 with shallower water tables (Fig 4F,H). 353 354 17 355 356 Figure 4. (A-D) Effects of P50 on occurrence probability (with rooting depth held at its average 357 value, 2.6 m) at the 5th and 95th percentiles of the aridity index, seasonality, and water table depth 358 (WTD). (E-H) Effects of rooting depth on occurrence probability (with P50 held at its average 359 value, -3.6 MPa) at the 5th and 95th percentiles of the aridity index, seasonality, and water table 360 depth (WTD). Note that the 5th percentile of WTD was close to zero meters. 361 18 362 Trait-by-environment interactions generated highly contingent occurrence landscapes. 363 Landscapes of probabilities illustrate the trait values and environments that optimize species 364 occurrence probabilities and can be viewed through two perspectives: first, as functions of 365 aridity, seasonality, and water table depth for each of the four plant strategies (Figure 5); and 366 second, as functions of P50 and rooting depth in different combinations of aridity, seasonality, 367 and water table depth (Figure 6). 368 Each strategy was most likely to occur in different environments. First, resistant avoiders 369 (low P50 and deep roots) were most likely to occur in arid and seasonal climates in sites with 370 deep water tables (Figures 5A, 6A). Second, vulnerable confronters (high P50 and shallow roots) 371 were most likely to occur in humid climates at any water table depth (Figures 5D, 6B, 6D). 372 Third, resistant confronters (low P50 and shallow roots) were most likely to occur in arid and 373 non-seasonal environments in sites with deep water tables (Figure 5C). Fourth, vulnerable 374 avoiders (high P50 and deep roots) were most likely to occur in arid and non-seasonal climates in 375 sites with shallow water tables (Figure 5B, 6C). 376 Trends in occurrence probabilities for deciduous species were qualitatively similar to full 377 model results (Figure S2). However, resistant confronters and vulnerable avoiders with 378 evergreen leaves exhibited relatively flat occurrence landscapes, suggesting that the full model 379 results for these two strategies were driven by deciduous species. 380 19 Low seasonality (CV = 11.6) A B Arid, Shallow water table C D Arid, Deep water table High seasonality (CV = 49.1) E F Arid, Deep water table G H Humid 381 382 Figure 5. Occurrence probability landscapes illustrating changes in the probability of woody 383 species occurrence for each of the four plant strategies (i.e., set as the 5th and 95th percentiles of 384 P50 and maximum rooting depth) along continuous gradients of the aridity index and water table 385 depth at the 5th and 95th percentiles of precipitation seasonality. Warm colors indicate higher 386 probability of occurrence and cool colors indicate lower probability of occurrence. The 387 environmental space is masked to emphasize the 99th percentile of observed environmental 388 variation. Contour interval = 0.01. 20 Low seasonality (CV = 11.6) A B Vulnerable, Shallow roots C D Vulnerable, Deep roots Vulnerable, Shallow roots High seasonality (CV = 49.1) E Resistant, F Deep roots Vulnerable, Shallow roots G H Vulnerable, Shallow roots 389 390 Figure 6. Occurrence probability landscapes illustrating changes in the probability of woody 391 species occurrence in different environments (i.e., set as the 5th and 95th percentiles of the aridity 392 index and water table depth (WTD) at the 5th and 95th percentiles of precipitation seasonality 393 along continuous gradients of P50 and maximum rooting depth. Warm colors indicate higher 394 probability of occurrence and cool colors indicate lower probability of occurrence. The trait 395 space is masked to illustrate the convex hull of observed trait values. Contour interval = 0.01. 396 21 397 Discussion 398 Xylem vulnerability to drought-induced embolism and maximum rooting depth have each been 399 separately investigated as drivers of drought resistance and avoidance in woody species 400 (Canadell et al., 1996; Choat et al., 2012), yet their phylogenetic correlation and the joint effect 401 of these traits on species distributions were unknown until now. Here we have shown that 1) 402 drought resistance and avoidance are orthogonal woody plant strategies, and 2) P50 and rooting 403 depth jointly improve predictions of species occurrence probabilities along gradients of aridity, 404 seasonality, and water table depth. 405 First, it has been suggested that drought resistance trades-off with drought avoidance 406 because species in a tropical forest were either shallow-rooted trees that built resistant xylem or 407 were deep-rooted trees that built vulnerable xylem (Brum et al., 2019). This proposed trade-off 408 assumed that investment in deep root systems comes at a cost of constructing vulnerable xylem. 409 However, our large sample across the phylogeny of woody plants provides strong evidence that 410 P50 and rooting depth are phylogenetically uncorrelated. The many species that "show deviations 411 from statistical correlations" (Carlquist, 1975, p. 1) provide insight into the adaptive value of 412 these traits. The large carbon investment into growth and maintenance of deep roots and resistant 413 xylem must be offset by the large carbon gains that are made possible by deep roots and resistant 414 xylem in dry environments, otherwise these phenotypes could not persist in these environments. 415 Moreover, many plant species construct shallow root systems and vulnerable xylem to maximize 416 carbon gain in wetter environments. Drought resistance and avoidance do not exhibit a trade-off 417 and species representing most major clades of woody seed plants have been able to explore the 418 full range of combinations of these two traits to occupy a broad range of environments. 22 419 Second, P50 and rooting depth jointly explain species occurrence probabilities, thereby 420 advancing our conceptual understanding of how species may respond to changing water 421 availability under climate change. Our modeling results suggest that forecasting woody species 422 range shifts in response to changing climate must also consider topographically mediated 423 hydrologic regimes. While both xylem vulnerability and rooting depth were more strongly 424 related to aridity, they interacted with water table depth to determine species occurrence 425 probabilities. Given that water table depth is relatively independent of regional climate (Fan et 426 al., 2017), if evaporative demand changes faster than water table depth under climate change, 427 then deeply-rooted vegetation responses may be buffered in the near-term by stable water table 428 depths. 429 To facilitate the interpretation of model results based on a complex five-way interaction 430 of continuous traits and environments, we discuss each of the four strategies that represent low 431 and high values of each trait in turn (Fig. 1B, Fig. 3, Fig. 5). First, resistant avoiders (low P50 and 432 deep roots) exhibit, in theory, the most drought-tolerant strategy because they can withstand 433 substantial negative pressures in their xylem and at the same time are also able to avoid these 434 strongly negative pressures when accessing deep soil water reservoirs (Hammond et al., 2019; 435 Tumber-Dávila et al., 2022). This strategy is most likely to occur in arid, seasonal climates with 436 deep water tables (Figs. 5E and 6E). This result was consistent for deciduous and evergreen 437 species, but most resistant avoiders were evergreen. For example, the evergreen angiosperm 438 Arbutus unedo, a small tree that grows in Mediterranean chaparral on dry slopes and ridges, and 439 the evergreen gymnosperm Juniperus monosperma, a small tree that grows in arid deserts, 440 illustrate how resistant xylem and deep roots optimize growth and survival in arid uplands. In 23 441 order to survive in some of the driest environments on earth, constructing deep roots and 442 resistant xylem is the optimal strategy for maintaining a positive carbon balance. 443 Second, resistant confronters (low P50 and shallow roots) are most likely to occur in arid, 444 non-seasonal climates over deep water tables. This agrees with previous empirical work that arid 445 environments contain both deep and shallow-rooted species (Fan et al., 2017) yet clarifies that 446 shallow roots are more adaptive in non-seasonal arid climates where more consistent 447 precipitation (albeit still modest amounts) wets the top soil layers throughout the year. This result 448 appeared to be driven by deciduous species, such as Amelanchier ovalis, which grows in open 449 sites in dry climates, but evergreen species, such as Encelia farinosa and Juniperus scopulorum, 450 also grow in dry habitats with well-drained soil (Ehleringer, 1993). 451 Third, vulnerable avoiders (high P50 and deep roots) are most likely to occur in arid, non- 452 seasonal climates over shallow water tables. This result was also most evident in deciduous 453 species. For example, the deciduous Populus euphratica grows in central Asian river flood plains 454 in arid regions (Bruelheide et al., 2010). This strategy may be emblematic of deep-rooted, 455 phraetophytic species that occur in desert riparian zones (Loheide et al., 2005). 456 Fourth, vulnerable confronters (high P50 and shallow roots) present, in theory, the least 457 drought-tolerant strategy because root-to-shoot conductance can decline under even weak 458 negative xylem pressure that cannot be overcome by access to deep soil water (Hammond et al., 459 2019; Tumber-Dávila et al., 2022). This strategy is most likely to occur in humid climates, 460 regardless of water table depth. This result was consistent for deciduous and evergreen species. 461 For example, the deciduous angiosperm Salix cinerea is a medium-sized shrub that grows in 462 wetlands in humid climates. Juglans cinerea, on the other hand, is a deciduous angiosperm tree 463 that grows in humid climates ranging from streambanks to well-drained slopes. Other drought- 24 464 related traits that were not measured here, such as leaf turgor loss point, could further explain 465 how species that are vulnerable confronters such as Juglans cinerea can survive in well-drained 466 dry soil, yet turgor loss point was not available for most of the 188 species in this dataset. 467 Our work represents the first systematic analysis of the combined effect of drought 468 resistance and avoidance on the global distribution of woody species. Yet, some limitations of 469 the study deserve attention. First, maximum rooting depth data is underestimated due to the 470 logisitical difficulty of measuring this trait, and rooting depth within species responds plastically 471 to variation in soil water depth (Schenk & Jackson, 2005; Fan et al., 2017). Given the low 472 number of replicated trait observations within species (Table S1, Figure S1), our analysis was 473 unable to evaluate the phenotypic plasticity of each trait and whether individual species could 474 acclimate to changing vapor pressure deficits and soil water availability. We urge researchers to 475 develop methods to simultaneously measure xylem vulnerability and rooting depth on individual 476 plants in their environments. Second, our analysis does not predict the growth, survival, and 477 reproduction of different ontogenetic stages of trees, and these are the fitness components that 478 are important to population dynamics that lead to range shifts (Merow et al., 2017). Our analysis 479 is an important first step to determine the joint effect of these critical traits on species responses 480 to changing aridity, seasonality, and water table depth, but future work should link these trait-by- 481 environment interactions to demographic rates and population growth rates to improve 482 forecasting generality (Laughlin et al., 2020b). Third, the limited data availability for drought- 483 deciduous species prevented us from understanding how the drought escape strategy relates to 484 drought resistance and avoidance. Future work should address these limitations. 485 The earliest plants to colonize the land emerged in humid tropical environments and 486 likely did not stray too far from shorelines and streambanks where evaporative demand was 25 487 moderated by humid atmospheres and shallow water tables (Willis & McElwain, 2014; Bouda et 488 al., 2022). The evolution of increasing resistance to embolism and extension of roots deep into 489 the soil profile permitted species to occupy increasingly drier environments, and these traits have 490 retained their adaptive advantages to this day. Simultaneous understanding of drought resistance 491 and avoidance strategies that have evolved across the phylogeny of woody plants will enhance 492 our predictions of vegetation response to changing water availability and guide species selection 493 in ecological restoration projects (Laughlin et al., 2017; McDowell et al., 2022). 494 495 Acknowledgements 496 This study was funded by a US National Science Foundation grant 2019528 to DL. We thank the 497 German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig for supporting 498 the sPlot working group and the University of Wyoming Advanced Research Computing Center 499 for their technical support. sPlot was initiated by sDiv and funded by the German Research 500 Foundation (FZT 118, 202548816), and is now a strategic project of iDiv. FMS acknowledges 501 the support of the Italian Ministry of University and Research, under the Maria Levi Montalcini 502 programme (2019). CB was supported by the National Research Foundation of Korea (NRF) 503 grant funded by the Korea government (MSIT) (2022R1A2C1003504). 504 505 Author Contributions 506 DL designed the study. AS, JF, and DL conducted the data analyses. DL and AS wrote the first 507 draft, and all authors contributed substantially to revisions. 508 509 Competing Interests 26 510 None declared. 511 512 Data Availability 513 All code and data will be archived on Dryad upon acceptance and the data DOI will be included 514 at the end of the article. Code is available for review at Andrew Siefert’s GitHub repository 515 https://github.com/andrewsiefert/xylem_root3. 516 517 References 518 Aubin I, Munson AD, Cardou F, Burton PJ, Isabel N, Pedlar JH, Paquette A, Taylor AR, 519 Delagrange S, Kebli H, et al. 2016. Traits to stay, traits to move: a review of functional 520 traits to assess sensitivity and adaptive capacity of temperate and boreal trees to climate 521 change. Environmental Reviews 24(2): 164-186. 522 Bauman D, Fortunel C, Delhaye G, Malhi Y, Cernusak LA, Bentley LP, Rifai SW, Aguirre- 523 Gutiérrez J, Menor IO, Phillips OL, et al. 2022. Tropical tree mortality has increased 524 with rising atmospheric water stress. Nature 608(7923): 528-533. 525 Bouda M, Huggett BA, Prats KA, Wason JW, Wilson JP, Brodersen CR. 2022. Hydraulic 526 failure as a primary driver of xylem network evolution in early vascular plants. Science 527 378(6620): 642-646. 528 Bruelheide H, Dengler J, Jiménez-Alfaro B, Purschke O, Hennekens SM, Chytrý M, Pillar 529 VD, Jansen F, Kattge J, Sandel B, et al. 2019. sPlot – A new tool for global vegetation 530 analyses. Journal of Vegetation Science 30(2): 161-186. 531 Bruelheide H, Vonlanthen B, Jandt U, Thomas FM, Foetzki A, Gries D, Wang G, Zhang X, 532 Runge M. 2010. Life on the edge – to which degree does phreatic water sustain 27 533 vegetation in the periphery of the Taklamakan Desert? Applied Vegetation Science 13(1): 534 56-71. 535 Brum M, Vadeboncoeur MA, Ivanov V, Asbjornsen H, Saleska S, Alves LF, Penha D, Dias 536 JD, Aragão LEOC, Barros F, et al. 2019. Hydrological niche segregation defines forest 537 structure and drought tolerance strategies in a seasonal Amazon forest. Journal of 538 Ecology 107(1): 318-333. 539 Canadell J, Jackson RB, Ehleringer JB, Mooney HA, Sala OE, Schulze ED. 1996. 540 Maximum rooting depth of vegetation types at the global scale. Oecologia 108(4): 583- 541 595. 542 Carlquist S. 1975. Ecological strategies of xylem evolution. Berkeley, CA: University of 543 California Press. 544 Choat B, Jansen S, Brodribb TJ, Cochard H, Delzon S, Bhaskar R, Bucci SJ, Feild TS, 545 Gleason SM, Hacke UG, et al. 2012. Global convergence in the vulnerability of forests 546 to drought. Nature 491: 752-755. 547 Cochard H, Badel E, Herbette S, Delzon S, Choat B, Jansen S. 2013. Methods for measuring 548 plant vulnerability to cavitation: a critical review. Journal of Experimental Botany 549 64(15): 4779-4791. 550 Díaz S, Kattge J, Cornelissen JH, Wright IJ, Lavorel S, Dray S, Reu B, Kleyer M, Wirth C, 551 Prentice IC. 2022. The global spectrum of plant form and function: enhanced species- 552 level trait dataset. Scientific Data 9(1): 755. 553 Dinerstein E, Olson D, Joshi A, Vynne C, Burgess ND, Wikramanayake E, Hahn N, 554 Palminteri S, Hedao P, Noss R, et al. 2017. An Ecoregion-Based Approach to 555 Protecting Half the Terrestrial Realm. Bioscience 67(6): 534-545. 28 556 Döll P, Fiedler K. 2008. Global-scale modeling of groundwater recharge. Hydrol. Earth Syst. 557 Sci. 12(3): 863-885. 558 Ehleringer JR. 1993. Variation in leaf carbon isotope discrimination in Encelia farinosa: 559 implications for growth, competition, and drought survival. Oecologia 95(3): 340-346. 560 Fan Y, Li H, Miguez-Macho G. 2013. Global patterns of groundwater table depth. Science 561 339(6122): 940-943. 562 Fan Y, Miguez-Macho G, Jobbágy EG, Jackson RB, Otero-Casal C. 2017. Hydrologic 563 regulation of plant rooting depth. Proceedings of the National Academy of Sciences 564 114(40): 10572-10577. 565 Fick SE, Hijmans RJ. 2017. WorldClim 2: new 1‐km spatial resolution climate surfaces for 566 global land areas. International Journal of Climatology 37(12): 4302-4315. 567 Fischer RA, Maurer R. 1978. Drought resistance in spring wheat cultivars. I. Grain yield 568 responses. Australian Journal of Agricultural Research 29(5): 897-912. 569 Grossiord C, Buckley TN, Cernusak LA, Novick KA, Poulter B, Siegwolf RTW, Sperry JS, 570 McDowell NG. 2020. Plant responses to rising vapor pressure deficit. New Phytologist 571 226(6): 1550-1566. 572 Hacke UG, Sperry JS, Pockman WT, Davis SD, McCulloh KA. 2001. Trends in wood density 573 and structure are linked to prevention of xylem implosion by negative pressure. 574 Oecologia 126(4): 457-461. 575 Hammond WM, Williams AP, Abatzoglou JT, Adams HD, Klein T, López R, Sáenz- 576 Romero C, Hartmann H, Breshears DD, Allen CD. 2022. Global field observations of 577 tree die-off reveal hotter-drought fingerprint for Earth’s forests. Nature communications 578 13(1): 1761. 29 579 Hammond WM, Yu K, Wilson LA, Will RE, Anderegg WRL, Adams HD. 2019. Dead or 580 dying? Quantifying the point of no return from hydraulic failure in drought-induced tree 581 mortality. New Phytologist 223(4): 1834-1843. 582 Hartmann H, Bastos A, Das AJ, Esquivel-Muelbert A, Hammond WM, Martínez-Vilalta J, 583 McDowell NG, Powers JS, Pugh TAM, Ruthrof KX, et al. 2022. Climate change risks 584 to global forest health: Emergence of unexpected events of elevated tree mortality 585 worldwide. Annual Review of Plant Biology 73(1): 673-702. 586 Jackson RB, Canadell J, Ehleringer JR, Mooney HA, Sala OE, Schulze ED. 1996. A global 587 analysis of root distributions for terrestrial biomes. Oecologia 108(3): 389-411. 588 Jackson RB, Sperry JS, Dawson TE. 2000. Root water uptake and transport: using 589 physiological processes in global predictions. Trends in Plant Science 5(11): 482-488. 590 Jin Y, Qian H. 2019. V.PhyloMaker: an R package that can generate very large phylogenies for 591 vascular plants. Ecography 42: 1353-1359. 592 Larter M, Pfautsch S, Domec JC, Trueba S, Nagalingum N, Delzon S. 2017. Aridity drove 593 the evolution of extreme embolism resistance and the radiation of conifer genus Callitris. 594 New Phytologist 215(1): 97-112. 595 Laughlin DC, Delzon S, Clearwater MJ, Bellingham PJ, McGlone MS, Richardson SJ. 596 2020a. Climatic limits of temperate rainforest tree species are explained by xylem 597 embolism resistance among angiosperms but not among conifers. New Phytologist 598 226(3): 727-740. 599 Laughlin DC, Gremer JR, Adler PB, Mitchell RM, Moore MM. 2020b. The net effect of 600 functional traits on fitness. Trends in Ecology & Evolution 35(11): 1037-1047. 30 601 Laughlin DC, Mommer L, Sabatini FM, Bruelheide H, Kuyper TW, McCormack ML, 602 Bergmann J, Freschet GT, Guerrero-Ramírez NR, Iversen CM, et al. 2021. Root 603 traits explain plant species distributions along climatic gradients yet challenge the nature 604 of ecological trade-offs. Nature Ecology & Evolution 5(8): 1123-1134. 605 Laughlin DC, Strahan RT, Huffman DW, Sánchez Meador AJ. 2017. Using trait-based 606 ecology to restore resilient ecosystems: historical conditions and the future of montane 607 forests in western North America. Restoration Ecology 25(S2): S135-S146. 608 Levitt J. 1980. Responses of plants to environmental stress, Volume 2: Water, radiation, salt 609 and other stresses: Academic Press. 610 Loheide SP, Butler JJ, Gorelick SM. 2005. Estimation of groundwater consumption by 611 phreatophytes using diurnal water table fluctuations: A saturated-unsaturated flow 612 assessment. Water resources research 41(7). 613 Maherali H, Pockman WT, Jackson RB. 2004. Adaptive variation in the vulnerability of 614 woody plant to xylem cavitation. Ecology 85(8): 2184-2199. 615 May LH, Milthorpe FL. 1962. Drought resistance of crop plants. Field Crop Abstracts 15(3): 616 171-179. 617 McDowell NG, Sapes G, Pivovaroff A, Adams HD, Allen CD, Anderegg WRL, Arend M, 618 Breshears DD, Brodribb T, Choat B, et al. 2022. Mechanisms of woody-plant 619 mortality under rising drought, CO2 and vapour pressure deficit. Nature Reviews Earth & 620 Environment 3: 294-308. 621 Merow C, Bois ST, Allen JM, Xie Y, Silander JA. 2017. Climate change both facilitates and 622 inhibits invasive plant ranges in New England. Proceedings of the National Academy of 623 Sciences 114(16): E3276-E3284. 31 624 Miguez-Macho G, Fan Y. 2021. Spatiotemporal origin of soil water taken up by vegetation. 625 Nature 598(7882): 624-628. 626 Miller C, Ulate W. 2017. World Flora Online Project: An online flora of all known plants. 627 Biodiversity Information Science and Standards 1: e20529. 628 Nobre AD, Cuartas LA, Hodnett M, Rennó CD, Rodrigues G, Silveira A, Waterloo M, 629 Saleska S. 2011. Height Above the Nearest Drainage – a hydrologically relevant new 630 terrain model. Journal of Hydrology 404(1): 13-29. 631 Oliveira RS, Costa FRC, van Baalen E, de Jonge A, Bittencourt PR, Almanza Y, Barros 632 FdV, Cordoba EC, Fagundes MV, Garcia S, et al. 2019. Embolism resistance drives 633 the distribution of Amazonian rainforest tree species along hydro-topographic gradients. 634 New Phytologist 221(3): 1457-1465. 635 Olson DM, Dinerstein E, Wikramanayake ED, Burgess ND, Powell GVN, Underwood EC, 636 D'amico JA, Itoua I, Strand HE, Morrison JC, et al. 2001. Terrestrial Ecoregions of 637 the World: A New Map of Life on Earth: A new global map of terrestrial ecoregions 638 provides an innovative tool for conserving biodiversity. Bioscience 51(11): 933-938. 639 Pagel M. 1999. Inferring the historical patterns of biological evolution. Nature 401(6756): 877- 640 884. 641 Qian H, Jin Y. 2016. An updated megaphylogeny of plants, a tool for generating plant 642 phylogenies and an analysis of phylogenetic community structure. Journal of Plant 643 Ecology 9: 233-239. 644 Revell LJ. 2012. phytools: an R package for phylogenetic comparative biology (and other 645 things). Methods in Ecology and Evolution 3(2): 217-223. 32 646 Revell LJ, Harmon LJ. 2022. Phylogenetic comparative methods in R: Princeton University 647 Press. 648 Ryel RJ, Ivans CY, Peek MS, Leffler AJ 2008. Functional differences in soil water pools: a 649 new perspective on plant water use in water-limited ecosystems. In: Lüttge U, Beyschlag 650 W, Murata J eds. Progress in Botany. Berlin, Heidelberg: Springer Berlin Heidelberg, 651 397-422. 652 Schenk HJ, Jackson RB. 2002. Rooting depths, lateral root spreads and below-ground/above- 653 ground allometries of plants in water-limited ecosystems. Journal of Ecology 90: 480- 654 494. 655 Schenk HJ, Jackson RB. 2005. Mapping the global distribution of deep roots in relation to 656 climate and soil characteristics. Geoderma 126(1): 129-140. 657 Schwinning S, Ehleringer JR. 2001. Water use trade‐offs and optimal adaptations to pulse‐ 658 driven arid ecosystems. Journal of Ecology 89(3): 464-480. 659 Seyfried MS, Schwinning S, Walvoord MA, Pockman WT, Newman BD, Jackson RB, 660 Phillips FM. 2005. Ecohydrological control of deep drainage in arid and semiarid 661 regions. Ecology 86(2): 277-287. 662 Sousa TR, Schietti J, Coelho de Souza F, Esquivel-Muelbert A, Ribeiro IO, Emílio T, 663 Pequeno PACL, Phillips O, Costa FRC. 2020. Palms and trees resist extreme drought 664 in Amazon forests with shallow water tables. Journal of Ecology 108(5): 2070-2082. 665 Sousa TR, Schietti J, Ribeiro IO, Emílio T, Fernández RH, ter Steege H, Castilho CV, 666 Esquivel-Muelbert A, Baker T, Pontes-Lopes A, et al. 2022. Water table depth 667 modulates productivity and biomass across Amazonian forests. Global Ecology and 668 Biogeography 31(8): 1571-1588. 33 669 Spinoni J, Vogt JV, Naumann G, Barbosa P, Dosio A. 2018. Will drought events become 670 more frequent and severe in Europe? International Journal of Climatology 38(4): 1718- 671 1736. 672 Trabucco A, Zomer RJ. 2018. Global aridity index and potential evapotranspiration (ET0) 673 climate database v2. Consortium for Spatial Information (CGIAR-CSI). Published online, 674 available from the CGIAR-CSI GeoPortal at https://cgiarcsi.community. 675 Tumber-Dávila SJ, Schenk HJ, Du E, Jackson RB. 2022. Plant sizes and shapes above and 676 belowground and their interactions with climate. New Phytologist 235(3): 1032-1056. 677 Volaire F. 2018. A unified framework of plant adaptive strategies to drought: crossing scales 678 and disciplines. Global Change Biology 24(7): 2929-2938. 679 Willis KJ, McElwain JC. 2014. The evolution of plants. Oxford: Oxford University Press. 680 Wood SN. 2011. Fast stable restricted maximum likelihood and marginal likelihood estimation 681 of semiparametric generalized linear models. Journal of the Royal Statistical Society: 682 Series B (Statistical Methodology) 73(1): 3-36. 683 Zomer RJ, Trabucco A, Bossio DA, Verchot LV. 2008. Climate change mitigation: A spatial 684 analysis of global land suitability for clean development mechanism afforestation and 685 reforestation. Agriculture, Ecosystems & Environment 126(1-2): 67-80. 686 687 688 Supporting Information 689 Additional supporting information can be found online in the Supporting Information section at 690 the end of this article. 34