Articles https://doi.org/10.1038/s41559-021-01616-8 Climatic and soil factors explain the two-dimensional spectrum of global plant trait variation Julia S. Joswig 1,2 ✉, Christian Wirth 1,3,4, Meredith C. Schuman 2,5, Jens Kattge 1,3, Björn Reu6, Ian J. Wright 7, Sebastian D. Sippel 8,9, Nadja Rüger 3,10,11, Ronny Richter 3,4,12, Michael E. Schaepman 2, Peter M. van Bodegom 13, J. H. C. Cornelissen14, Sandra Díaz 15, Wesley N. Hattingh 16, Koen Kramer 17,18, Frederic Lens 19,20, Ülo Niinemets 21, Peter B. Reich 22,23,24, Markus Reichstein 1,3, Christine Römermann 3,25, Franziska Schrodt 26, Madhur Anand27, Michael Bahn 28, Chaeho Byun 29, Giandiego Campetella 30, Bruno E. L. Cerabolini 31, Joseph M. Craine 32, Andres Gonzalez-Melo33, Alvaro G. Gutiérrez 34, Tianhua He35,36, Pedro Higuchi 37, Hervé Jactel38, Nathan J. B. Kraft39, Vanessa Minden40,41, Vladimir Onipchenko 42, Josep Peñuelas 43,44, Valério D. Pillar 45, Ênio Sosinski 46, Nadejda A. Soudzilovskaia47,48, Evan Weiher 49 and Miguel D. Mahecha3,50,51 Plant functional traits can predict community assembly and ecosystem functioning and are thus widely used in global models of vegetation dynamics and land–climate feedbacks. Still, we lack a global understanding of how land and climate affect plant traits. A previous global analysis of six traits observed two main axes of variation: (1) size variation at the organ and plant level and (2) leaf economics balancing leaf persistence against plant growth potential. The orthogonality of these two axes suggests they are differently influenced by environmental drivers. We find that these axes persist in a global dataset of 17 traits across more than 20,000 species. We find a dominant joint effect of climate and soil on trait variation. Additional independent climate effects are also observed across most traits, whereas independent soil effects are almost exclusively observed for economics traits. Variation in size traits correlates well with a latitudinal gradient related to water or energy limitation. In contrast, varia- tion in economics traits is better explained by interactions of climate with soil fertility. These findings have the potential to improve our understanding of biodiversity patterns and our predictions of climate change impacts on biogeochemical cycles. P lant functional traits have proved useful in identifying life his- dominant for extended sets of traits or when differentiating among tory strategies1,2 for predicting plant community assembly3,4 growth forms. A particular knowledge gap is what environmental and for assessing the impact of vegetation composition and controls determine these two axes of plant form and function. diversity on ecosystem functioning5,6. Consequently, vegetation There is ample evidence that large-scale variation of individual models including coupled climate–vegetation models benefit from plant traits is related to environmental gradients. Early plant bio- a better representation of plant trait variation to adequately anal- geographers suggested that climate and soils together shape plant yse terrestrial biosphere dynamics under global change6,7. Today, form and function15–17 but could not propose a more precise the- in combination with advanced gap-filling techniques8, databases of oretical framework describing these fundamental relationships. plant traits have sufficient coverage to allow quantitative analyses of Over the last decades, examples have thus accumulated without plant form and function at the global scale9. Analysing six funda- an overall framework in which to place them13,18,19. For instance, mental traits, Díaz and colleagues10 revealed that essential patterns tree height depends on water availability20,21 while leaf economics of form and function across the plant kingdom can be captured by traits depend on soil properties, especially soil nutrient supply, as two main axes. The first reflects the size spectrum of whole plants well as on climatic conditions reflected in precipitation18,22,23. Leaf and plant organs. The second axis corresponds to the ‘leaf econom- size, leaf dark respiration rate, specific leaf area (SLA), leaf N and ics spectrum’11 emerging from the necessity for plants to balance leaf P concentration, seed size and wood density, all show broad-scale persistence against plant growth potential. The concept of a global correlations with climate or soil22,24–27. It has also been reported spectrum of plant form and function has since been investigated that many of these traits show latitudinal patterns24–27. Generalizing from various perspectives12–14. It has been shown, for instance, that such insights is, however, not trivial, as soil properties partly mir- orthogonal axes of variation in size and economics traits emerge ror climate gradients, as a consequence of long-term soil formation even in the extreme tundra biome13 or at the scale of plant commu- through weathering, leaching and accumulation of organic mat- nities12. However, it remains unclear whether the two axes remain ter—processes related to temperature and precipitation28; however, A full list of affiliations appears at the end of the paper. 36 Nature Ecology & Evolution | VOL 6 | January 2022 | 36–50 | www.nature.com/natecolevol NATurE Ecology & EVoluTIon Articles a Colour key b 10 SLA 0 2 4 6 8 0 Leaf N 0. 0. 0. 0. 1. Value Leaf P 5 Leaf area 0.32 0.02 0.07 0.04 0.24 0.22 0.22 0.32 0.34 0.34 0.31 0.43 0.47 0.55 0.89 0.92 Seed len 0.99 Leaf f mass 0.37 0 0.12 0.01 0.2 0.21 0.17 0.3 0.34 0.31 0.26 0.39 0.43 0.53 0.84 0.92 Dispersal len 0.95 0.41 0.1 0.03 0 0.26 0.3 0.3 0.43 0.36 0.43 0.46 0.49 0.56 0.66 0.84 0.89 Seed mass Dispersal len 0.14 0.1 0.07 0.12 0.27 0.21 0.34 0.4 0.38 0.4 0.39 0.56 0.61 0.66 0.53 0.55 Height Seed len PC2 (15%) 0.15 0.09 0.06 0.06 0.07 0.15 0.16 0.29 0.17 0.18 0.51 0.92 0.61 0.56 0.43 0.47 Leaf f mass 0.2 Leaf d15N Seed mass Leaf area Height 0.14 0.08 0.04 0.11 0.09 0.02 0.07 0.21 0.14 0.12 0.47 0.92 0.56 0.49 0.39 0.43 0 0.2 0.16 0.1 0.1 0.22 0.07 0.12 0.17 0.37 0.09 0.17 0.47 0.51 0.39 0.46 0.26 0.31 Conduit dens Seeds U 0.5 0.17 0.2 0.08 0.18 0.33 0.28 0.41 0.41 0.29 0.17 0.12 0.18 0.4 0.43 0.31 0.34 Stem dens Vessel len 0.07 0.03 0.06 0.16 0.28 0.23 0.32 0.24 0.29 0.09 0.14 0.17 0.38 0.36 0.34 0.34 Leaf C Leaf C Leaf N:P 0.18 0.09 0.16 0.06 0.27 0.36 0.7 0.24 0.41 0.37 0.21 0.29 0.4 0.43 0.3 0.32 Leaf N:P Stem dens 0.13 0.02 0.26 0.55 0.56 0.23 0.7 0.32 0.41 0.17 0.07 0.16 0.34 0.3 0.17 0.22 Leaf P Conduit dens Leaf N area 0.04 0.08 0.26 0.12 0.69 0.23 0.36 0.23 0.28 0.12 0.02 0.15 0.21 0.3 0.21 0.22 Leaf N area 0.06 0.02 0.05 0.58 0.69 0.56 0.27 0.28 0.33 0.07 0.09 0.07 0.27 0.26 0.2 0.24 SLA −5 0.03 0.08 0.22 0.58 0.12 0.55 0.06 0.16 0.18 0.22 0.11 0.06 0.12 0 0.01 0.04 Leaf N 0.15 0.03 0.22 0.05 0.26 0.26 0.16 0.06 0.08 0.1 0.04 0.06 0.07 0.03 0.12 0.07 Vessel len 0.01 0.03 0.08 0.02 0.08 0.02 0.09 0.03 0.2 0.1 0.08 0.09 0.1 0.1 0 0.02 Leaf d15N 0.01 0.15 0.03 0.06 0.04 0.13 0.18 0.07 0.17 0.16 0.14 0.15 0.14 0.41 0.37 0.32 Seeds U Seeds U Leaf d15N Vessel len Leaf N SLA Leaf N area Leaf P Leaf N:P Leaf C Stem dens Conduit dens Leaf area Leaf f mass Height Seed mass Dispersal len Seed len −10 −10 −5 0 5 10 PC1 (33%) Fig. 1 | Previously identified global axes of variation in size and economics traits hold for an extended trait set (n = 36,197 species per ecoregion median). The set of 17 investigated traits (Supplementary Table 5) can be primarily divided into size and economics traits, which load differently onto the two PC axes describing their global distribution. a, Heatmap of covariation. Trait correlations are indicated using absolute Pearson correlation coefficients, with green shades indicating high absolute correlation and yellow shades indicating low absolute correlation. On the left, the distance tree of traits derived from hierarchical clustering is illustrated. Three resulting groups are: (1) size-related traits (blue) consisting of conduit density (conduit dens), leaf area, leaf fresh mass (leaf f mass), plant height (height), seed mass, dispersal unit length (dispersal len) and seed length (seed len); (2) economics traits (red) comprising SLA, leaf N content per area (leaf N area), leaf N, P and C concentrations, leaf N/P ratio (leaf N:P) and stem specific density (stem dens); and (3) a third (yellow) consisting of the number of seeds per reproduction unit (seeds U), leaf δ15N (leaf d15N) and vessel element length (vessel len). b, The first two PCs of the PCA. Arrow tips refer to the loading of the traits (Supplementary Fig. 6). Contour lines delineate the colour scale that corresponds to the kernel density of species (dense, red to sparse, light yellow; 20%, 50%, 95% and 99% of all species). PC1 explains 33% of trait variation and PC2 15% (Supplementary Fig. 5). climate-independent features reflecting geology and surface mor- aggregation is a suitable means to increase the detectability of global phology also contribute to soil fertility28. Soil may furthermore trait patterns (Supplementary Fig. 3), as described in earlier studies, buffer climate stresses; for example, by alleviating water deficit in where traits have been binned by temperature classes33 or for differ- periods of low precipitation29. ent altitudinal ranges22. Extreme outliers, for instance towering trees Combining the insights suggests that the global spectrum of such as the Californian Sequoia (Sequoiadendron sempervirens), plant traits reveals two internally correlated orthogonal groups and may still exist far away from the equator, where precipitation is suf- that many plant traits are individually linked to environmental gra- ficiently high20 but their influence is outweighed in our approach dients, we expect that both trait groups should closely follow gra- by an increasing fraction of small-statured herbaceous species from dients of climate and soil properties. Here, we investigate to what tropical to temperate and boreal regions. extent the major dimensions underpinning the global spectrum of plant form and function can be attributed to global gradients of cli- Orthogonal axes and trait clusters. To understand whether the mate and soil conditions; and to what extent these factors can jointly axes of variation identified for the grouping of six traits10 also hold or independently explain the global spectrum of form and function. for the extended set of 17 traits, we cluster their trait–trait corre- We compiled and analysed a dataset of 17 functional traits with lations (Fig. 1a and Supplementary Fig. 4) and further represent a sufficient number of records in the TRY database9 to character- these relations on the basis of their principal components (PCA; ize the main ecoregions of the world30, that is, environmentally Methods). This analysis supports the clear distinction of size versus homogeneous areas with distinct biota (Extended Data Fig. 1). The economics traits identified by Díaz and colleagues10. The group of dataset is based on 225,206 georeferenced observations comprising size traits contains two subclusters. The first includes height and records of 20,655 species. The trait data were complemented with 21 seed size traits: plant height (height), seed mass, seed length and climate variables and 107 soil variables (Methods; Supplementary dispersal unit length (dispersal length). The second subset con- Tables 1 and 2). Trait–environment relationships were analysed for tains traits that are linked through plant hydraulic scaling rela- species medians aggregated to ecoregions using ridge regression31, tionships34 and contrasts high conduit density (that is, number of a robust method (Supplementary Figs. 1–3) suitable to deal with conduits per sapwood cross-sectional area) with high leaf area and high-dimensional, unbalanced and collinear predictors in combina- leaf fresh mass (leaf f mass). Economics traits represent dry mass tion with hierarchical partitioning32 (Methods). and nutrient investments in plant tissues, and the rate and dura- tion of returns on those investments11. They are represented by leaf Results nitrogen content per leaf area (leaf N area), leaf nitrogen (leaf N), Our main analysis is based on median trait values of plant spe- phosphorus (leaf P) and carbon (leaf C) content per dry mass, leaf cies per ecoregion. The rationale is that species presence indicates N to P ratio (leaf N:P) and SLA. Stem specific density (stem density) how the trait space can be realized in a given environment. Spatial takes an intermediate position (Fig. 1b) but more closely clusters Nature Ecology & Evolution | VOL 6 | January 2022 | 36–50 | www.nature.com/natecolevol 37 Articles NATurE Ecology & EVoluTIon a 6 b 6 r 2 = 0.39 r 2 = 0.01 0.5 0.95 4 r 2 = 0.8 4 r 2 = 0.08 0.8 0.5 ● ● ● 2 ●● ● ● ● 2 ● ● ● ● ● ● ●●● ● ● 0.5 ● ● ● 0.5 ●● ● ● PCA axis 1 PCA axis 2 ●●●● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●●● ● ●● ●●●●● 0 ● ● ● ● ●●● ● 0 ●●●● ● ●● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●● ● 0.5 ● ● ● ● 0.5 ● ● ●● ● ● ● 0.8 ● ● ● −2 ● ● ●●●● ● ●● ● ● ● ● ●● ● −2 ● ● ● ● ● ● −4 0.5 −4 0.8 0.95 ● −6 −6 −8 −8 0 20 40 60 0 20 40 60 Absolute latitude Absolute latitude Fig. 2 | Size traits, not economics traits vary with latitude: the PC1 of the PCA on 17 plant traits shows a clear latitudinal gradient while PC2 does not (n = 36,197 species per ecoregion median). Contour lines delineate the colour scale that corresponds to the kernel density of species (dense, red to sparse, light yellow; 5%, 95%, 99% quantiles). Mean estimates aggregated at 1° absolute latitude are indicated as black dots. The line refers to a linear model (ordinary least squares). a, PC1 representing mainly size traits (conduit density, leaf area, leaf fresh mass, plant height, seed mass, dispersal unit length, seed length) regressed against absolute latitude. Linear model: r2 = 0.38 without bins; r2 = 0.84 aggregated at 1° absolute latitude. b, PC2 representing mainly economics traits (leaf N, leaf N per area, leaf P, leaf N:P ratio, SLA, leaf C, stem density) regressed against absolute latitude. Linear model: r2 = 0.01 without bins; r2 = 0.08 aggregated at 1° absolute latitude. with this set of economics traits (Fig. 1a), suggesting a syndrome variation within the separate groups of size versus economics traits. of traits promoting slow to fast nutrient and carbon processing at We assess the joint and independent effects of climate and soil on the whole-plant level35–37. Furthermore, we identify a third group trait variability (ridge regression, RR; Table 1 and Fig. 3). Overall, size of traits that appear to be only weakly correlated with any other traits are better explained (RR; r2 = 0.55; maximum r2 = 0.78 for con- trait. This third group contains seed number per reproduction unit duit density; Table 1) than are economics traits (RR; r2 = 0.40; maxi- (seeds U), leaf δ15N (leaf d15N) and vessel element length (vessel mum r2 = 0.55 for leaf N:P ratio; Table 1). We find a substantial joint length). The first two principal components (PC) of the PCA on effect of climate and soil variables—in every case larger than either the trait data represent 48% of the overall variation (Supplementary unique effect—which reflects strong interactions between specific Fig. 5). PC1 is determined by size traits and accounts for 33% of the climate and soil predictors (RR with hierarchical partitioning (HP); variance; PC2 is determined by economics traits and accounts for Fig. 3b and Supplementary Fig. 12). However, we also observe inde- 15% of the variance (Fig. 1b and Supplementary Fig. 6). These two pendent effects of climate and soil (RR with HP; Fig. 3 and Table 1). main axes remain clearly identifiable when the analysis is conducted The independent climate effects are observed across traits but size separately for woody and non-woody species (Supplementary traits tend to be better explained by the independent climate effects Figs. 7 and 8). The remaining PCs each account for less than 10% than are economics traits. In contrast, independent soil predictors of variance (PC3 = 9.36%). In the following, we focus on the two are relevant for all economics traits but not size traits (apart from a groups of size and economics traits (Supplementary Fig. 5). small contribution to leaf area). We interpret these results as evidence for the importance of both joint and independent effects of climate Latitudinal trait variation. As an investigation of broad-scale gra- and soil variables for whole-plant strategies2,37,38 which we show here dients among size and economics traits, we analyse latitudinal gra- at the global scale along with a dichotomous tendency of a stronger dients of the first (PC1) and second (PC2) principal components. imprint of climate factors on size traits and of soil conditions on eco- PC1—representing primarily size-related traits—shows a strong nomics traits (Fig. 3, Supplementary Figs. 13–38 and Supplementary linear latitudinal signal (on the basis of species: r2(PC1) = 0.37, Table 3). We propose that the dominance of joint effects implies at the ecoregion level r2(PC1aggregated) = 0.84; Fig. 2a). By contrast, that interactions between soil and climate properties are of primary the axis representing primarily economic traits, PC2, shows little importance in plant trait ecology; as opposed to trait syndromes response to latitude (on the basis of species: r2(PC2) = 0.01, at the being defined by single environmental variables in isolation. ecoregion level r2(PC2aggregated) = 0.08; Fig. 2b, for woody non-woody We next ask how the climate and soil datasets are interdependent species Supplementary Fig. 9), except for a dip at 35° and declin- and which predictors add the most relevant information. For this ing sharply at 60° where the species density also drops (but see purpose, we related all traits to environmental variables in a redun- Supplementary Fig. 10 for comparison to an independent dataset dancy analysis (RDA; Methods; Fig. 4). The RDA again identifies from arctic latitudes which shows the same pattern). Latitudinal two main axes of size and economics traits (Fig. 4a), which are now gradients are known to be strongly related to climate, due to the shown together with the environmental variables that co-vary lin- distribution of solar energy and general atmospheric circula- early with those traits (Fig. 4b and Supplementary Fig. 39). The first tion patterns. Therefore, we propose that those climate (and soil) RDA axis corresponds to size traits (Fig. 4a) and represents an axis aspects that co-vary with latitude consistently determine size traits, of water and energy (for example, precipitation, vapour pressure while they have little effect on economics traits, which are more and temperature; Fig. 4b). Two attributes of soil texture important strongly affected by latitude-independent soil (and climate) effects for water retention—the fraction of gravel and clay—also vary along (Supplementary Fig. 11). this axis. The second RDA axis corresponds to economics traits (Fig. 4a) co-varying with an axis of soil variables generally associ- Climate and soil: joint and independent effects. The differ- ated with soil fertility (that is, soil texture (silt versus sand), water ences in latitudinal relationships between the two PC axes support holding capacity, carbon concentration and stocks), as well as the the hypothesis that different environmental factors should drive climate variable mean solar radiation (Fig. 4b). 38 Nature Ecology & Evolution | VOL 6 | January 2022 | 36–50 | www.nature.com/natecolevol NATurE Ecology & EVoluTIon Articles Table 1 | Showing for each trait the variance explained (r2) by ridge regression models for 220 ecoregions and the independent effects for climate and soil listed from hierarchical partitioning that, respectively, add up with the joint effect to the variance explained by climate or soil Traita Group Explained varianceb by soil Soil (independent effectb) Climate (independent Joint effectb (r2) and climate (r2) (r2) effectb) (r2) Ridge regression model Hierarchical partitioning Hierarchical partitioning Hierarchical partitioning Seed length Size 0.4 –0.01 0.08 0.33 Dispersal length Size 0.26 –0.01 0.03 0.24 Seed mass Size 0.57 –0.01 0.09 0.49 Height Size 0.52 0.01 0.1 0.41 Leaf f mass Size 0.72 0 0.15 0.56 Leaf area Size 0.63 0.03 0.13 0.47 Conduit density Size 0.77 –0.01 0.22 0.56 Stem density Economics 0.41 0.03 0.02 0.36 Leaf C Economics 0.29 0.08 0.05 0.17 Leaf N:P Economics 0.55 0.06 0.12 0.38 Leaf P Economics 0.45 0.15 0.05 0.25 Leaf N area Economics 0.39 0.03 0.02 0.33 SLA Economics 0.41 0.09 0.13 0.19 Leaf N Economics 0.26 0.06 0.16 0.05 Vessel length Other 0.4 0 –0.03 0.42 Leaf d15N Other 0.51 0.05 0.1 0.35 Seeds U Other 0.1 –0.02 0.02 0.1 a For full versions of traits, see main text. Mean values; minimum and maximum values from different cross-validation runs in Supplementary Table 8. Negative values indicate a reduction of explained variance when respective variables are added to the ridge regression model. Discussion Fig. 40) or boreal and arctic areas characterized by short growing This study shows that the proposed global spectrum of plant form periods slowing down mineralization. The dip at ~35° indeed can and function fits well to a substantially extended trait space com- be related to low water availability (Supplementary Fig. 41). At high pared to the original study10, with seven traits that capture the latitudes, cold winters and short growing seasons constrain plant whole-plant size spectrum and seven traits that capture the leaf height13 and require on average more conservative nutrient-use economic spectrum and only three traits that do not fall along strategies (like evergreen leaves) and protection against frost dam- these dimensions (Fig. 1b). One explanation could be that the vary- age than the global mean, despite the high functional diversity in ing fraction of woody and non-woody species would drive these economics traits observed at these latitudes13. Additional datas- patterns. However, we showed that these two main trait groups ets may shed more light on specific conditions, for example see remain clearly identifiable when the analysis is conducted sepa- Bjorkman et al.19. Future studies should quantify how individual rately, yet with fewer samples, for woody and non-woody species stressors, for example radiative stress or water stress, relate to global (Supplementary Fig. 8). patterns of trait variation. However, we cannot discard the possibility that additional traits The climate and soil factors used in this analysis explain up to may add relevant axes of trait variation. For example, our study 77% of observed trait variation—a high fraction given that trait does not include carbon fixation rates39 or fire adaptation traits40, variation is widely known to be determined also by other factors nor does it include any root traits—representing an essential gap to such as biotic interactions (for example, soil biota) and anthro- be filled at the global scale41. The respective data are too scarce to yet pogenic effects or disturbances and local effects such as those of be integrated with global datasets. If such data were available they microclimate12,44–46. Recent findings on how different trait groups would have the potential to fundamentally change our perception vary with the environment indicate that size and economics traits of global plant form and function, and their relation to ecosystem vary differently13 and in particular respond differently to climate functioning. and soil19. Variation in size traits, represented by PC1 in Fig. 1b, shows a Our analyses reveal a dominant joint effect of climate and soil clear latitudinal gradient (Fig. 2b). In contrast, variation in econom- drivers on trait variation—as already suggested by a number of ear- ics traits (represented by PC2) does not show a latitudinal trend. lier studies18,19,22 but not yet quantified globally. Only a dip is apparent at around 35° (absolute), in addition to a The orthogonality of the two main dimensions of plant trait decrease at high latitudes above 60° (absolute) where available data variation suggests that different aspects of climate and soil vari- become increasingly limited. However, comparison to a recent arc- ables are relevant to explain plant trait patterns at the global scale tic dataset indicates that this decrease in variation at high latitudes (Supplementary Figs. 11–39). While latitude-related variables reflects available observations (see Supplementary Fig. 10 for a (mainly climate) explain size traits, variables that share less explana- comparison to independent data). These patterns might represent a tory power with latitude (mainly soil) explain economics traits response to nutrient limitation and drought42,43 in water-scarce and (Supplementary Table 4 and Supplementary Fig. 11). The RDA nutrient-scarce deserts and Mediterranean regions (Supplementary presented in Fig. 4 (Supplementary Fig. 39) provides some insight Nature Ecology & Evolution | VOL 6 | January 2022 | 36–50 | www.nature.com/natecolevol 39 Articles NATurE Ecology & EVoluTIon a b Seed len 8% 33% 0% Dispersal len 3% 24% 0% 0 1. Size trait 1. 0 Economics trait Seed mass 9% 49% 0% Conduit dens Height 10% 41% 1% Other trait Leaf f mass Leaf f mass 15% 56% 0% 8 0. 0. Leaf area 8 13% 47% 3% Leaf area Conduit dens 22% 56% 0% C lim Leaf N:P ) Seed mass Stem dens /2 2% 36% 3% at t Leaf d15N in Height e 6 0. Leaf C jo 5% 17% 8% 0. ef Leaf P 6 nd fe Stem dens Leaf N:P 12% 38% 6% ct ta Seed len Leaf N area ( en in Leaf P 5% 25% 15% de nd SLA Vessel len Leaf N area pe 2% 33% 3% pe 4 Dispersal len 0. nd 0. de 4 SLA 13% 19% 9% en Leaf C (in Leaf N ta Leaf N 16% 5% 6% ct nd fe Vessel len 0% 42% 0% ef Seeds U jo il in 2 0. Leaf d15N So 0. 10% 35% 5% t/2 2 ) Seeds U 2% 10% 0% 100 50 0 50 100 0 0 Percentage of trait variance explained by climate and/or soil Variable Independent_climate Independent_soil Joint Fig. 3 | Climate and soil variables explain up to 77% of variance in size and economics traits. Hierarchical partitioning32 identifies the contribution of climate and soil variables to explain each trait (n = 220, ecoregional median trait: blue, size; red, economics; yellow, other). The joint effect is the fraction explained by both climate and soil together, and is split equally among them. The independent effect is the fraction of r2 explained exclusively by either soil or climate variables. a, Tilted x–y plot of the soil versus climate variables to explain a trait. The axes show the sum of the respective joint and independent effect (hierarchical partitioning). The colours reflect the strength of: the independent effect of climate plus its share of the joint effect (r2; purple); and soils' independent effect plus its share of the joint effect (peach). The sum of both axes equals the total r2 explained by climate and soil; in cases where soil showed a negative independent effect only the climate-independent effect is shown (and vice versa but see Table 1). b, Percentage variation explained by climate (purple, percentages on the left), soil (peach, percentages on the right) and jointly (grey, percentages in the middle) for trait groups— size, economics and other. Total bar length = total r2 explained by climate and soil; in cases where soil showed a negative independent effect only the climate-independent effect is shown (and vice versa but see Table 1). For leaf area, climate and soil jointly explain 47%, the independent climate effect explains an additional 13% of the variance, while soil explains 3%, totalling 63% of variance explained. For trait abbreviations see Fig. 1. a b 1.0 1.0 WaterWilt SLA C_concentration C_stock WaterSat Silt C_density Leaf P Leaf area WaterPot32 CEC C_stocki_0.3m 0.5 Leaf N Vessel len 0.5 WaterPot20 C_stocki_1m Precipitation Leaf f mass C_stocki_2m Precipitation_max Height WaterPot10 Vapour.pressure_sd ClayVapour.pressure_ma Leaf C Precipitation_min Vapour.pressure Seed len Temp_sd Vapour.pressure_min RDA2 RDA2 Seeds U 0 Seed mass 0 Dispersal len Wind_min Temp_min Conduit dens Gravel Temp Solar.rad_sd Wind Precipitation_sd Leaf N:P Wind_max Temp_max Stem dens Wind_sd Solar.rad_min Leaf d15N −0.5 −0.5 Temp_d pH Density Leaf N area Sand Solar.rad Solar.rad_max −1.0 −1.0 −1.0 −0.5 0 0.5 1.0 −1.0 −0.5 0 0.5 1.0 RDA1 RDA1 Fig. 4 | RDA of traits reveals the relationships of climate and soil factors associated with trait distributions (n = 220, ecoregion median, only topsoil layer variables included; variance explained: RDA1 = 63%, RDA2 = 18%). a,b, The output of the RDA is split into two plots: traits (a), where arrows are coloured according to trait groups (blue, size traits; red, economics traits; yellow, other traits; arrow length and point positions scaled to fit the plot); and environmental factors (b), where arrows are coloured according to predictor group (climate, blue and soil, red variables; arrow length and point positions scaled to fit the plot). In a and b, points represent ecoregions and are coloured according to biome (red, tropics; green, temperate; yellow, desert; orange, Mediterranean; dark blue, tundra). Climate variable abbreviations are composed of the variable (average if not stated differently) and a suffix. Variables are: Solar.rad, solar radiation; Vapour.pressure, vapour pressure; Wind, average wind speed; Temp, temperature; Precipitation, precipitation. Suffixes are: no suffix, mean of respective variable; d, diurnal range; min, annual minimum of the respective variable; sd, seasonality of respective variable. Soil variable abbreviations (all topsoil) are: Density, soil density (kg/m3); pH, pH value; Sand, sand fraction (vol%); Silt, silt fraction (vol%); Clay, clay fraction (vol%); C_concentration, organic carbon concentration; C_density, organic carbon density; CEC, cation exchange capacity; C_stock, soil organic carbon stock at depth 0.00 m; C_stocki, organic carbon stock for depth intervals—0.3m (0–0.30 m), 1m (0–1 m), 2m (0–2 m); WaterWilt, available soil water capacity (volumetric fraction) until wilting point; WaterSat, saturated water content; WaterPot, available soil water capacity for moisture potentials—10 (–10 kPa;pF 2.0), 20 (–20 kPa;pF 2.3), 32 (–31.6 kPa;pF 2.5). For trait abbreviations see Fig. 1. 40 Nature Ecology & Evolution | VOL 6 | January 2022 | 36–50 | www.nature.com/natecolevol NATurE Ecology & EVoluTIon Articles on the nature of these climate–soil interactions. The first RDA axis, that interact with climate, pH, soil properties and also plant traits50. which describes variation in size traits, resembles a latitudinal gra- Trait–environment relationships due to smaller scale variation dient. On one extreme end, ample water supply from high and fre- require well-resolved soil data. However, we note that soil physics quent precipitation, abundant water vapour and constant rates of and chemistry explain a large portion of variance along the trait PC high solar radiation meet the fundamental requirements of plant axis three (which itself explains slightly less than 10% of variance physiology—water, sunlight and warm temperatures. Additionally, in the PCA (9.36%); Supplementary Figs. 5, 6 and 38). We expect these conditions promote weathering of soil minerals but also that with improved soil datasets and a higher resolution, the joint microbial activity, contributing to fast turnover rates of organic control of climate and soil on trait variation will probably appear matter supporting nutrient provisioning28,47; in brief, they represent even stronger and more evenly distributed between the two groups conditions that allow plants to grow fast and tall in the race for light. of driver variables. Large vessels supporting large leaves promote high rates of water Our analysis can serve as reference for model developments that transport and thus growth, which is only possible because of the increasingly consider plant functional traits as part of vegetation small risk of embolism under these benign water conditions43. The dynamics under climate change44. high carbon gains can be invested in large fruits and seeds (seed Individual plants and their trait syndromes are considered to be mass, seed length and dispersal unit length). Further along this viable only within specific environmental conditions2. Therefore gradient, the above-mentioned plant requirements become limited: trait–environment relationships should be scale-independent. water supply and temperatures are reduced and slow metabolic However, different plant strategies can be successful under given rates aboveground and belowground. In ecoregions of the boreal environmental conditions, which in addition are often confounded and desert biomes, conduit diameter is constrained by the risk of by small-scale variation. In analyses to date, trait–environment rela- cavitation during freeze–thaw cycles43 and water scarcity, amplified tionships become more apparent for aggregations higher than the by little water holding capacity of gravel-rich soils. Our analysis thus community scale12, where most of the small-scale variation is aver- indicates that size traits appear to be related to a latitudinal gradient aged out. In addition the difference between potential and actual of climatic favorability for plant growth determined by water and vegetation is suggested to explain some of this gap13. Dynamic light availability. global vegetation models predict individual plant processes well Important correlates of water and nutrient availability are associ- but fail to produce reliable forecasts with a changing environ- ated with the second RDA axis, describing variation in economics ment44. Deciphering at which spatial and temporal scale, or condi- traits. Traits associated with an acquisitive strategy are related to tions, actual vegetation is representative of potential vegetation may indicators of soil fertility, most importantly silt and organic mat- advance our understanding of community assembly and necessary ter concentration as well as pH (refs. 18,28). Soil pH is intermediate model complexity. between the two axes, as might be expected given that pH reflects Trait–environment correlations identified in our study should both broad-scale climate variation (especially aridity47) and a variety not be confounded with causality. Yet, the ubiquitous importance of processes related to nutrient availability and soil microbial com- of climate variables for explaining current differences in trait munities18,48–50. Silt forms the substrate of our most fertile soils as its expression at ecoregion scale, suggests that trait shifts will occur structure is able to retain water against gravitation (unlike sand) but with climate change. Trait shifts are constrained by available trait renders it accessible to plants under drought conditions28,51 (unlike combinations in addition to other constraints such as species dis- clay). The high fertility is associated with a high concentration of persal. For example, our results indicate that plant size increases organic matter, which has a high cation exchange capacity espe- with temperature so long as sufficient water is available (Fig. 4 and cially under high pH (ref. 47). On the opposite end of the gradient, Supplementary Figs. 19, 20 and 21), in line with the finding that spe- sandy soils require adaptations to both water and nutrient limita- cies become larger and large species are more prevalent at warmer tion. The trait configuration at the conservative end of the econom- and wetter sites in the tundra19. Global change is also reflected by ics traits (low SLA, high tissue density and high organ longevity) soil degradation. Changes in soil parameters can be considered to represents an adaptation to both11,37. Various processes exist that also correspond with trait shifts, especially for economics traits. lead to variation in the soil characteristics underlying the second Human-induced soil degradation has many facets: often fertile top- RDA axis independent of latitude18—for example, sandstone as a soil is lost or toxic substances accumulate; rooting is impeded and geological substrate giving rise to sandy soils exists from the tropics altered by artificial fertilizers; while soil formation takes millenia57. to the arctic28,51. However, different climate variables related to solar The trait shifts may thus be similarly complex and depend on the radiation, temperature and precipitation, which influence long- extent and type of soil degradation. For example, in areas of wind and short-term soil development processes directly and indirectly and water erosion, species that tolerate lower nutrient availability via soil biology28,51,52, are related to this axis. Variation in economic may be more successful and this may be reflected in lower leaf nutri- traits is most probably the evolutionary response to exploiting this ent contents (Fig. 4 and Supplementary Fig. 30). The fertilization partly climate-independent edaphic niche axis. of nutrient-poor grasslands, for example resulting from agricultural Size traits are on average explained better than economics traits run-off, may shift these areas from more conservative to more com- by the environmental variables considered in this study. The lower petitive species with higher leaf nutrient contents. fraction of explained variance for economics traits could have sev- Plants as a whole need to balance both size and economics traits. eral causes. Firstly, data on soil factors that are likely to be very To sustain human livelihoods, it may be important to understand important, such as soil nitrogen and phosphorus availability18,23, the local expression of trait shifts and their global consequences for are not yet available at a global scale. Secondly, economics traits biodiversity when viable trait combinations change. show relatively more within-site variation than across-site varia- In conclusion, the insights extracted here advance our under- tion in comparison to size traits (Supplementary Fig. 42), prob- standing of broad-scale plant functional patterns. In particular, we ably because economics traits vary more than size traits within one highlight the combination of independent and particularly joint plant; for example, leaf N per area and SLA vary with age and light effects of climate and soil on trait variation, an interaction that has availability53. Thirdly, soil heterogeneity within ecoregions—both to date been neglected because few studies include both in a single abiotic and biotic—may weaken the relationship between econom- analysis, at the global scale as we have done here. In doing so, we ics traits and environmental variables12,54,55. Reasons for small-scale identify an important gap in knowledge: what is the nature of cli- soil variation are, for example, topography, soil age and thus fertil- mate–soil interactions that drive whole-plant trait variation and ity56 but also abundance of microbial communities and mycorrhiza what distinguishes the majority of climate and soil factors having Nature Ecology & Evolution | VOL 6 | January 2022 | 36–50 | www.nature.com/natecolevol 41 Articles NATurE Ecology & EVoluTIon joint effects on plant traits from those with independent effects? part of the probability density distribution to be sampled by BHPMF. To avoid These are the sorts of questions that require answers to increase our autocorrelation, only every twentieth iteration was used to calculate the resulting trait values. The mean of these predictions result in the final trait values used as the capacity to predict plant functional diversity in a changing envi- output. Compared to the original data, the imputed values are similar in terms of ronment. Such predictive power would contribute to a sound basis trait–trait correlation, according to the Procrustes test provided in ref. 10. for assessing long-term feedbacks between global environmental change and the terrestrial biosphere, helping to constrain param- Trait clustering. To define groups of correlated traits, we clustered species’ traits eters of global coupled climate–vegetation models. Humans are cur- (species median per ecoregion, A1) on the basis of absolute pairwise Pearson correlation coefficients using a hierarchical clustering algorithm (‘complete linkage rently modifying both climatic and edaphic conditions at the global clustering’). Variables were transformed into distances previous to the clustering. scale. Climate envelope models used to predict vegetation shifts Hierarchical clustering then attributes variables (here, traits) to groups of least must be complemented by drivers related to large-scale anthro- distance and highest similarity. Traits were more like each other if they exhibited pogenic alterations of soil conditions resulting, for example, from similar correlation patterns with all other traits. We set a distance between traits of land-use change, atmospheric nitrogen deposition, fertilization, 1 as the threshold for defining trait clusters. We used the R package ‘stats’ function ‘hclust’ included in R (ref. 315). liming and salinization. Our global analysis provides an essential context for finer-scale studies to directly tackle questions of biologi- PCA. Values for all 17 traits (unique species per ecoregion, A1) were natural cal processes and mechanisms at landscape and community scales. log-transformed and then projected onto components (PCA). We used the R package FactoMineR316 that scales data internally. After the PCA (A1), we extracted Methods the variance explained (Fig. 1b and Supplementary Fig. 5) and respective loadings We extracted data on 17 plant functional traits from a gap-filled version of TRY for the first five principle components (Fig. 1b and Supplementary Fig. 6), which database9 (Supplementary Table 5; www.try-db.org, accession date July 2017, are significant according to the number of axes to keep estimated using a sequential request no. 3282) which includes published literature11,58–101,101–310. Quality control Bonferroni procedure (R package ade4 (refs. 317–321), function testdim). For the was conducted according to the published protocol of TRY.9, 311 Traits with analysis (ridge regression package ‘glmnet’322,323) for Fig. 3, all environmental z-score > 4 were excluded and those with z-score > 3 were checked for plausibility. variables (climate and soil) were first reduced with this package to 20 PCs. Before this, missing data were imputed using a Bayesian hierarchical probabilistic matrix factorization (BHPMF) algorithm8,312 for an extended dataset, derived Environmental variables. To represent climate conditions we used 21 variables from TRY (Supplementary Table 6). Imputation was done to be able to include derived from WorldClim at a resolution of 1 km for temperature, precipitation, the maximum number of species in our analyses. Then the 17 traits were selected vapour pressure, solar radiation and wind (Supplementary Table 1). To characterize among the traits with the largest total number of entries. The data were attributed soil conditions we used 107 variables derived from the ISRIC data product to ecoregions30 (Supplementary Table 7 and Extended Data Fig. 1) and aggregated ‘SoilGrids’324–326 (https://soilgrids.org/ through ISRIC—WDC Soils). ‘SoilGrids’ to species median values. The imputed values were calculated using the whole provides global predictions of 17 fundamentally different soil characteristics (some dataset at the individual record level. BHPMF calculates the imputations from for seven depths, that is 0, 5, 15, 30, 60, 100, 200 cm; Supplementary Table 2) at a 1,000 Gibbs sampler (Markov chain Monte Carlo) imputations by taking the resolution of 1 km. SoilGrids are publicly accessible environmental data (Creative mean of every twentieth imputation of these 1,000 ‘versions’, after the first 200 Commons Attribution 4.0 International), with a collection of georeferenced soil are removed. Then the species median was calculated at the ecoregion level. We profile data and are managed in World Soil Information Service324. excluded observations that were not georeferenced because we could not attribute them to ecoregions. According to TRY regulations, data from experimental Aggregation of traits and environmental variables to ecoregions. To determine treatments (for example, fertilization) or from botanic gardens were also excluded. trait–environment relationships, we aggregated trait as well as environmental In total, we included 225,206 observations from 20,655 global unique species data to regions, here ecoregions30 (ecoregion aggregation A2, see also above; (36,197 unique species to ecoregion combinations). Throughout this study we used Supplementary Table 7). Ecoregions are environmentally homogeneous areas, one of two aggregation levels: either species median per ecoregion (ER)30 resulting nested within biogeographic realms (defined by refs. 327,328) and biomes (modified in unique species values per ecoregion (termed A1, n = 36,197 with n = 20,655 after refs. 329,330 but see ref. 30). As a first estimate, ecoregions are distinct biotas328,331 globally unique species) or the aggregation to median ecoregions calculated from defined by the physiognomy of the prevailing climatic climax vegetation30. These median species per ecoregion (termed A2, n = 220). R was used for all analyses and areas of distinct biotas, areas of relatively uniform flora or fauna, are next subset figures313. into provinces with substantial differences of vegetation on the basis of a selection of plants and animals, maps and expert knowledge30,331. At global scale Olson Hierarchical probabilistic matrix factorization. Description. BHPMF decomposes et al.30 defined 867 ecoregions. Ecoregions were chosen as the scale of aggregation or factorizes probabilistically a matrix (probabilistic matrix factorization, PMF314) for their high signal-to-noise ratio and the ability to correct for sampling bias. using information contained within different hierarchical levels (here, taxonomy) While the grid scale has higher spatial resolution, it lacks estimates of species within a Bayesian framework8. The underlying premise of BHPMF is to gap-fill richness (equivalent of Kier species richness174) and is not as well explained by (or more accurately, to predict) traits of an individual plant using trait–trait the climate and soil (Supplementary Fig. 3) and distribution of grids is globally correlations as well as intraspecific and interspecific trait variability.8. Using a uneven (Extended Data Fig. 1) in comparison to ecoregions. The global sampling Gibbs sampler (a Markov Chain Monte Carlo algorithm), BHPMF also provides distribution is recognized to show a bias towards Europe9, which is even more a prediction confidence in the form of standard deviations which is a per-value pronounced in the lower level data (grid scale) than in the more aggregated estimate of uncertainty in trait predictions8. BHPMF can fill gaps if there is at least one (Extended Data Fig. 1). Our method accounts for this oversampling and one value per row (species) and column (trait). reproduces a stable pattern, even when species in oversampled ecoregions are deleted (Supplementary Fig. 2). Implementation. The largest possible dataset was retrieved at the time when study For each of the 867 ecoregions, we calculated the median ecoregion aggregate was conducted, including 172 traits of 652,957 individuals (Supplementary trait value from the median trait values of all species identified in each region. Table 6). For data preparation before BHPMF, all individual-level trait data For further analyses, we only used regions with >20 species and a representation were firstly log-transformed and secondly normalized via zlog transformation of >1% of the estimated species richness of the ecoregion30. Preliminary tests ( z = xs.d. −x̄ ). Log transformation was chosen to achieve a closer-to-normal with different selection criteria (for example, number of species and inclusion or distribution of values per trait311,313. This transformation is considered necessary exclusion of 1% of species richness estimate by Kier et al.174) showed that lower because a given difference for small trait values (absolute value) is likely to be numbers of species per ecoregion result in weaker explained variance, while physiologically more relevant than the same difference (absolute value) for large stricter rules reduced the number of ecoregions. These selection criteria serve as trait values. a quality control because ecoregions with poor representation of species richness BHPMF internally splits the datasets randomly into a training dataset (80%), a are excluded, as we can expect the regression to the mean to be stronger with test dataset (10%) and a validation dataset (10%). more species data. A total of 220 ecoregions met these criteria and were included The training dataset is used during training of latent vectors, while the test data in the analysis. On average, these ecoregion-level trait values were based on 164 are tested against to improve the latent vectors, and finally the validation dataset species-level trait medians (with a maximum of 1,245 species in Tapajós-Xingu serves as the basis for calculation of the root mean square error (RMSE) and moist forests; Supplementary Table 7). In total, we aggregated 36,197 median stopping the optimization of latent vectors within BHPMF8. The validation dataset species trait values to ecoregions. These ecoregions cover the global latitudinal ensures ongoing amelioration of the model performance during the training gradient (Fig. 2) as well as a substantial fraction of the geographic space (Extended process and stops the process after five consecutive iterations with stable RMSE. Data Fig. 1). To aggregate environmental variables to ecoregions, we associated The test dataset is used only on the lowest taxonomic level (individuals × traits). each trait observation with its corresponding values of climate and soil variables. BHPMF was run with a maximum of 1,000 iterations, whereas the first 200 Then, we averaged over all values within one ecoregion. Thus, the selected were discarded during the ‘burn-in’ phase, as predictions of these iterations environmental variables represent averages that are weighted by the number and are likely to be influenced by the initialization of BHPMF rather than being locations of trait observations within ecoregions. 42 Nature Ecology & Evolution | VOL 6 | January 2022 | 36–50 | www.nature.com/natecolevol NATurE Ecology & EVoluTIon Articles Model building. Ecoregion trait values (natural log-transformed, A2) were related 8. Schrodt, F. et al. BHPMF—a hierarchical Bayesian approach to gap-filling to all environmental variables using ridge regression31, which is a well-established and trait prediction for macroecology and functional biogeography. linear regression method that is suitable to deal with a large number of collinear Glob. Ecol. Biogeogr. 24, 1510–1521 (2015). predictors and uneven numbers of predictors for climate and soil. We used the 9. Kattge, J. et al. TRY plant trait database—enhanced coverage and open R package ‘glmnet’322,323. From aggregating trait values to ecoregion medians we access. Glob. Change Biol. 26, 119–188 (2020). obtain 220 samples for each trait. The environmental predictors of climate and 10. Díaz, S. et al. The global spectrum of plant form and function. Nature soil were reduced to 20 each by means of a PCA. In addition, the environmental 529,167–171 (2015). predictors show relatively high collinearity, thus duplicated information. Ridge 11. Wright, I. J. et al. The worldwide leaf economics spectrum. Nature 428, regression addresses collinearity among predictors by shrinking (regularizing) 821–827 (2004). regression coefficients according to a penalty on the L2 norm of the vector of 12. Bruelheide, H. et al. Global trait–environment relationships of plant regression coefficients. The regularization parameter lambda was obtained via communities. Nat. Ecol. Evol. 2, 1906–1917 (2018). tenfold cross validation. The variance explained was derived from an iterative 13. Thomas, H. J. et al. Global plant trait relationships extend to the climatic holdout set (tenfold cross validation), that is, prediction of 90% of randomly extremes of the tundra biome. Nat. Commun. 11, 1351 (2020). sampled ecoregions for inclusion in model building and then predicting the 14. Kong, D. et al. Nonlinearity of root trait relationships and the root remaining 10% of the data to evaluate the quality of the models. The final model economics spectrum. Nat. Commun. 10, 2203 (2019). predicts the remaining 10% of unused ecoregions. This prediction-loop was 15. Schimper, A. Plant-Geography Upon A Physiological Basis (Clarendon Press, repeated until all ER trait values are predicted, that is, resulting in different linear 1903). models. Repeated r2: the r2 is the squared correlation of predicted versus original 16. Warming, E. Oecology Of Plants (Oxford, 1909). ER trait values. This procedure was repeated 50 times and the explained variances’ 17. Raunkiær, C. in Life Forms of Plants and Statistical Plant Geography, 4-16 (r2) mean, minimum and maximum were calculated. For the purpose of defining (Clarendon Press, 1934). how much of the explained variance is due to independent and joint information 18. Maire, V. et al. Global effects of soil and climate on leaf photosynthetic in the data streams, we used hierarchical partitioning32. Model outputs (r2) of traits and rates. Glob. Ecol. Biogeogr. 24, 706–717 (2015). all repetitions (n = 50, if not indicated differently) were used as input (ridge 19. Bjorkman, A. D. et al. Plant functional trait change across a warming regression, partial least squares (PLS) with and without PCA, random forest). tundra biome. Nature 562, 57–62 (2018). 20. Olson, M. E. et al. Plant height and hydraulic vulnerability to drought and Redundancy analysis. To relate trait–trait covariation to trait–environment cold. Proc. Natl Acad. Sci. USA 115, 7551–7556 (2018). covariation, we performed a redundancy analysis (R package ‘vegan’). 21. Moles, A. T. et al. Global patterns in plant height. J. Ecol. 97, 923–932 (2009). Ecoregion-aggregated traits (A2) were normalized and natural log-transformed. 22. Ordoñez, J. C. et al. A global study of relationships between leaf traits, Scaled climate and soil variables were used as predictors. To decrease the factor climate and soil measures of nutrient fertility. Glob. Ecol. Biogeogr. 18, that quantifies collinearity (variance inflation factor, vif), only the topsoil layer was 137–149 (2009). selected (Fig. 4). For Supplementary Fig. 39, additional model tuning based on vif, 23. Simpson, A. H., Richardson, S. J. & Laughlin, D. C. Soil–climate with the exclusion of two variables with vif > 20, led to a model with vif < 10, which interactions explain variation in foliar, stem, root and reproductive traits can be considered low cocorrelation. across temperate forests. Glob. Ecol. Biogeogr. 25, 964–978 (2016). 24. Wright, I. J. et al. Global climatic drivers of leaf size. Science 357, 917–921 Reporting Summary. Further information on research design is available in the (2017). Nature Research Reporting Summary linked to this article. 25. Atkin, O. K. et al. Global variability in leaf respiration in relation to climate, plant functional types and leaf traits. New Phytol. 206, 614–636 (2015). Data availability 26. Asner, G. P., Knapp, D. E., Anderson, C. B., Martin, R. E. & Vaughn, N. Plant trait data were accessed from the TRY database (https://try-db.org, request no. Large-scale climatic and geophysical controls on the leaf economics 3282, date accessed July 2017, see also Extended Data Fig. 1). All TRY data required spectrum. Proc. Natl Acad. Sci. USA 113, E4043–E4051 (2016). to reproduce this analysis, and the corresponding R scripts, are provided in an 27. Moles, A. T. et al. Global patterns in seed size. Glob. Ecol. Biogeogr. 16, open TRY File Archive (https://www.try-db.org/TryWeb/Data.php). Climate data 109–116 (2007). WorldClim are publicly available via https://www.worldclim.org/ (accessed May 28. Blume, H.-P. et al. Soil Science 1st edn.(Springer, Berlin-Heidelberg, 2016). 2018). Soil data, namely SoilGrids (https://soilgrids.org/, accessed June 2018) are 29. Seneviratne, S. I. et al. Investigating soil moisture–climate interactions in a publicly available. Ecoregion information30 shapefiles are publicly available (accessed changing climate: a review. Earth-Sci. Rev. 99, 125–161 (2010). January 2014, Sciencebase.gov), The estimate of species richness per ecoregion174 30. Olson, D. M. et al. Terrestrial ecoregions of the world: a new map of life on is publicly available (accessed January 2014, databasin.org. Data for this study Earth. BioScience 51, 933–938 (2001). can be accessed on Github (https://github.com/juliajoswig/ Repo_ClimateSoil_ 31. Hastie, T., Tibshirani, R. & Friedman, J. The Elements of Statistical Learning TraitSpectrum). For Extended Data Fig. 1 and Supplementary Fig. 7, the Geodata (Springer, 2008). product of the Missions Database ‘ArcWorld Supplement’ (GMI) was used. 32. Chevan, A. & Sutherland, M. Hierarchical partitioning. Am. Stat. 45, 90–96 (1991). 33. Reich, P. B. & Oleksyn, J. Global patterns of plant leaf N and P in relation Code availability to temperature and latitude. Proc. Natl Acad. Sci. USA 101, 11001–11006 The code is available on Github (https://github.com/juliajoswig/ (2004). Repo_ClimateSoil_TraitSpectrum). 34. Corner, E. J. H. The Durian theory or the origin of the modern tree. Ann. Bot. XIII, 367–414 (1949). Received: 21 January 2021; Accepted: 10 November 2021; 35. Pietsch, K. A. et al. Global relationship of wood and leaf litter Published online: 23 December 2021 decomposability: the role of functional traits within and across plant organs. Glob. Ecol. Biogeogr. 23, 1046–1057 (2014). 36. FloresâMoreno, H. et al. Robustness of trait connections across environmental References gradients and growth forms. Glob. Ecol. Biogeogr. 28, 1806–1826 (2019). 1. Westoby, M. A leaf–height–seed (LHS) plant ecology strategy scheme. Plant 37. Chapin, F. S. The mineral nutrition of wild plants. Annu. Rev. Ecol. Syst. 11, Soil 199, 213–227 (1998). 233–260 (1980). 2. Kraft, N. J. B. et al. Community assembly, coexistence and the 38. Vitousek, P. Nutrient Cycling and Limitation: Hawai’i as a Model System environmental filtering metaphor. Funct. Ecol. 29, 592–599 (2015). (Princeton Univ. Press, 2004). 3. McGill, B. J., Enquist, B. J., Weiher, E. & Westoby, M. Rebuilding community 39. Shipley, B., Vile, D., Garnier, E., Wright, I. J. & Poorter, H. Functional ecology from functional traits. Trends Ecol. Evol. 21, 178–185 (2006). linkages between leaf traits and net photosynthetic rate: reconciling 4. Lavorel, S. & Garnier, E. Predicting changes in community composition and empirical and mechanistic models. Funct. Ecol. 19, 602–615 (2005). ecosystem functioning from plant traits: revisiting the Holy Grail. Funct. 40. He, T., Belcher, C. M., Lamont, B. B. & Lim, S. L. A 350-million-year legacy Ecol. 16, 545–556 (2002). of fire adaptation among conifers. J. Ecol. 104, 352–363 (2016). 5. Musavi, T. et al. Potential and limitations of inferring ecosystem 41. Bergmann, J., Ryo, M., Prati, D., Hempel, S. & Rillig, M. C. Root traits are photosynthetic capacity from leaf functional traits. Ecol. Evol. 6, more than analogues of leaf traits: the case for diaspore mass. New Phytol. 7352–7366 (2016). 216, 1130–1139 (2017). 6. Scheiter, S., Langan, L. & Higgins, S. I. Next-generation dynamic global 42. Aerts, R. The advantages of being evergreen. Trends Ecol. Evol. 10, 402–407 vegetation models: learning from community ecology. New Phytol. 198, (1995). 957–969 (2013). 43. Zanne, A. E. et al. Functional biogeography of angiosperms: life at the 7. Van Bodegom, P. M., Douma, J. C. & Verheijen, L. M. A fully traits-based extremes. New Phytol. 218, 1697–1709 (2018). approach to modeling global vegetation distribution. Proc. Natl Acad. Sci. 44. Franklin, O. et al. Organizing principles for vegetation dynamics. USA 111, 13733–13738 (2014). Nat. Plants 6, 444–453 (2020). Nature Ecology & Evolution | VOL 6 | January 2022 | 36–50 | www.nature.com/natecolevol 43 Articles NATurE Ecology & EVoluTIon 45. Legay, N. et al. Contribution of above- and below-ground plant traits to the 76. Blonder, B., Violle, C., Bentley, L. P. & Enquist, B. J. Venation networks and structure and function of grassland soil microbial communities. Ann. Bot the origin of the leaf economics spectrum. Ecol. Lett. 14, 91–100 (2011). 114, 1011–1021 (2014). 77. Bocanegra-González K.T., Fernández-Méndez, F. & David Galvis-Jiménez, J. 46. Grime, J. P. Vegetation classification by reference to strategies. Nature 250, Funtional groups of tres in secondary forests of the bajo calima region 26–31 (1974). (Buenaventura, Colombia) Boletín CientífiCo Centro de Museos Museo de 47. Slessarev, E. W. et al. Water balance creates a threshold in soil pH at the Historia natura 19, (2015). global scale. Nature 540, 567–569 (2016). 78. Bodegom, P. M. V., Kanter, M. D. & Aerts, C. B. R. Radial oxygen loss, a 48. Fierer, N. & Jackson, R. B. The diversity and biogeography of soil bacterial plastic property of dune slack plant species. Plant Soil 271, 351–364 (2005). communities. Proc. Natl Acad. Sci. USA 103, 626–631 (2006). 79. Bond-Lamberty, C. W. B. & Gower, S. T. Above- and belowground biomass 49. Sinsabaugh, R. L. & Follstad Shah, J. J. Ecoenzymatic stoichiometry and and sapwood area allometric equations for six boreal tree species of ecological theory. Annu. Rev. Ecol. Evol. Syst. 43, 313–343 (2012). northern Manitoba. Can. J. For. Res. 32, 1441–1450 (2002). 50. de Vries, F. T. et al. Abiotic drivers and plant traits explain landscape-scale 80. Bond-Lamberty, C. W. B. & Gower, S. T. Leaf area dynamics of a boreal patterns in soil microbial communities. Ecol. Lett. 15, 1230–1239 (2012). black spruce fire chronosequence. Tree Physiol. 22, 993–1001 (2002). 51. Zech, W., Schad, P. & Hintermaier-Erhard, G. Böden der Welt—Ein Bildatlas 81. Bond-Lamberty, C. W. B. & Gower, S. T. The use of multiple measurement (Springer Spectrum, 2014). techniques to refine estimates of conifer needle geometry. Can. J. For. Res. 52. Rosenberg, E. et al. (eds) The Prokaryotes: Prokaryotic Communities and 33, 101–105 (2003). Ecophysiology 4th edn. (Springer-Verlag, 2013). 82. Bond-Lamberty, C. W. B. & Gower, S. Net primary production and net 53. Niinemets, Ã. Leaf age dependent changes in within-canopy variation in ecosystem production of a boreal black spruce fire chronosequence. leaf functional traits: a meta-analysis. J. Plant Res. 129, 313–338 (2016). Glob. Change Biol. 10, 473–487 (2004). 54. Butler, E. E. et al. Mapping local and global variability in plant trait 83. Bragazza, L. Conservation priority of Italian alpine habitats: a floristic distributions. Proc. Natl Acad. Sci. USA 114, E10937–E10946 (2017). approach based on potential distribution of vascular plant species. Biodivers. 55. Freschet, G. T. et al. Global to community scale differences in the Conserv. 18, 2823–2835 (2009). prevalence of convergent over divergent leaf trait distributions in plant 84. Choat, B. et al. Global convergence in the vulnerability of forests to assemblages. Global Ecol. Biogeogr. 20, 755–765 (2011). drought. Nature 491, 752–755 (2012). 56. Yemefack, M., Rossiter, D. G. & Njomgang, R. Multi-scale characterization 85. Briemle, G., Nitsche, S. & Nitsche, L. in BIOLFLOR—Eine Datenbank mit of soil variability within an agricultural landscape mosaic system in Biologisch-ökologischen Merkmalen zur Flora von Deutschland (eds Klotz, S. southern Cameroon. Geoderma 125, 117–143 (2005). et al.) 203–225 (Bundesamt für Naturschutz, 2002). 57. Oldeman, L., Hakkeling, R. & Sombroek, W. Global Assessment of Soil 86. Brown, K. et al. Assessing natural resource use by forest-reliant communities Degradation (GLASOD): World Map of the Status of Human-induced Soil in Madagascar using functional diversity and functional redundancy Degradation (United Nations Environment Programme, 1991). metrics. PLoS ONE https://doi.org/10.1371/journal.pone.0024107 (2011). 58. Ackerly, D. D. & Cornwell, W. K. A trait-based approach to community 87. Burrascano, S. et al. Wild boar rooting intensity determines shifts in assembly: partitioning of species trait values into within- and understorey composition and functional traits. Community Ecol. 16, among-community components. Ecol. Lett. 10, 135–145 (2007). 244–253 (2015). 59. Adler, P. B. A Comparison of Livestock Grazing Effects on Sagebrush Steppe, 88. Butterfield, B. J. & Briggs, J. M. Regeneration niche differentiates functional USA, and Patagonian Steppe, Argentina. PhD thesis (Colorado State strategies of desert woody plant species. Oecologia 165, 477–487 (2011). University, 2003). 89. Byun, C., de Blois, S. & Brisson, J. Plant functional group identity and 60. Adler, P. B., Milchunas, D. G., Lauenroth, W. K., Sala, O. E. & Burke, I. C. diversity determine biotic resistance to invasion by an exotic grass. J. Ecol. Functional traits of graminoids in semi-arid steppes: a test of grazing 101, 128–139 (2013). histories. J. Appl. Ecol. 41, 653–663 (2004). 90. Campbell, C. et al. Acclimation of photosynthesis and respiration is 61. Adriaenssens, S. Dry deposition and canopy exchange for temperate tree asynchronous in response to changes in temperature regardless of plant species under high nitrogen deposition. PhD thesis, Ghent Univ. (2012). functional group. New Phytol. 176, 375–389 (2007). 62. Atkin, O. K., Schortemeyer, M., McFarlane, N. & Evans, J. R. The response 91. Campetella, G. et al. Patterns of plant trait–environment relationships along a of fast- and slow-growing Acacia species to elevated atmospheric CO2: an forest succession chronosequence. Agric. Ecosyst. Environ. 145, 38–48 (2011). analysis of the underlying components of relative growth rate. Oecologia 92. Carswell, F. E. et al. Photosynthetic capacity in a central Amazonian rain 120, 544–554 (1999). forest. Tree Physiol. 20, 179–186 (2000). 63. Atkin, O. K., Westbeek, M., Cambridge, M. L., Lambers, H. & Pons, T. L. 93. Castro-Diez, P., Puyravaud, J. P., Cornelissen, J. H. C. & Villar-Salvador., P. Leaf respiration in light and darkness (a comparison of slow- and Stem anatomy and relative growth rate in seedlings of a wide range of fast-growing Poa species). Plant Physiol. 113, 961–965 (1997). woody plant species and types. Oecologia 116, 57–66 (1998). 64. Auger, S. L’Importance de la Variabilité Interspécifique des Traits Fonctionnels 94. Castro-Diez, P., Puyravaud, J. P. & Cornelissen, J. H. C. Leaf structure and par Rapport à la Variabilité Intraspécifique Chez les Jeunes Arbres en Forêt anatomy as related to leaf mass per area variation in seedlings of a wide Mature. MSc thesis (Université de Sherbrooke, 2012). range of woody plant species and types. Oecologia 124, 476–486 (2000). 65. Bahn, M. et al. in Land-Use Changes in European Mountain Ecosystems. 95. Cavender-Bares, A. K. J. & Miles, B. Phylogenetic structure of Floridian ECOMONT—Concept and Results (eds Cernusca, A. et al.) 247–255 plant communities depends on taxonomic and spatial scale. Ecology 87, (Blackwell Wissenschaft, 1999). 109–122 (2006). 66. Baker, T. R. et al. Do species traits determine patterns of wood production 96. Cavender-Bares, L. S. J. & Savage, J. Atmospheric and soil drought reduce in Amazonian forests? Biogeosciences 6, 297–307 (2009). nocturnal conductance in live oaks. Tree Physiol. 27, 522–620 (2007). 67. Bakker, C., Van Bodegom, P. M., Nelissen, H. J. M., Ernst, W. H. O. & 97. Cerabolini, B. E. L. et al. Can CSR classification be generally applied outside Aerts, R. Plant responses to rising water table and nutrient management in Britain? Plant Ecol. 210, 253–261 (2010). calcareous dune slacks. Plant Ecol. 185, 19–28 (2006). 98. Chave, J. et al. Towards a worldwide wood economics spectrum. Ecol. Lett. 68. Bakker, C., Rodenburg, J. & van Bodegom, P. M. Effects of Ca- and Fe-rich 12, 351–366 (2009). seepage on P availability and plant performance in calcareous dune soils. 99. Chen, Y., Han, W., Tang, L., Tang, Z. & Fang, J. Leaf nitrogen and Plant Soil 275, 111–122 (2005). phosphorus concentrations of woody plants differ in responses to climate, 69. Baraloto, C. et al. Decoupled leaf and stem economics in rainforest trees. soil and plant growth form. Ecography 36, 178–184 (2011). Ecol. Lett. 13, 1338–1347 (2010). 100. Choat, B. et al. Global convergence in the vulnerability of forests to 70. Baraloto, C. et al. Functional trait variation and sampling strategies in drought. Nature 491, 752–755 (2012). species-rich plant communities. Funct. Ecol. 24, 208–216 (2010). 101. Choat, B., Sack, L. & Holbrook, N. M. Diversity of hydraulic traits in nine 71. Beckmann, M., Hock, M., Bruelheide, H. & Erfmeier, A. The role of UV-B Cordia species growing in tropical forests with contrasting precipitation. radiation in the invasion of Hieracium pilosella—a comparison of German New Phytol. 175, 686–698 (2007). and New Zealand plants. Environ. Exp. Bot. 75, 173–180 (2012). 102. Coomes, D. A., Heathcote, S., Godfrey, E. R. & Shepherd, J. J. Scaling of xylem 72. Blanco, C. C., Sosinski, E. E., dos Santos, B. R. C., da Silva, M. A. & Pillar, vessels and veins within the leaves of oak species. Biol. Lett. 4, 302–306 (2008). V. D. On the overlap between effect and response plant functional types 103. Cornelissen, J., Aerts, R., Cerabolini, B., Werger, M. & van der Heijden, M. linked to grazing. Community Ecol. 8, 57–65 (2007). Carbon cycling traits of plant species are linked with mycorrhizal strategy. 73. Blonder, B. et al. The shrinkage effect biases estimates of paleoclimate. Oecologia 129, 611–619 (2001). Am. J. Bot. 99, 1756–1763 (2012). 104. Cornelissen, J. H. C. An experimental comparison of leaf decomposition 74. Blonder, B., Violle, C. & Enquist, B. J. Assessing the causes and scales of the rates in a wide range of temperate plant species and types. J. Ecol. 84, leaf economics spectrum using venation networks in Populus tremuloides. 573–582 (1996). J. Ecol. 101, 981–989 (2013). 105. Cornelissen, J. H. C., Diez, P. C. & Hunt., R. Seedling growth, allocation 75. Blonder, B. et al. Testing models for the leaf economics spectrum with leaf and leaf attributes in a wide range of woody plant species and types. J. Ecol. and whole-plant traits in Arabidopsis thaliana. AoB Plants 7, plv049 (2015). 84, 755–765 (1996). 44 Nature Ecology & Evolution | VOL 6 | January 2022 | 36–50 | www.nature.com/natecolevol NATurE Ecology & EVoluTIon Articles 106. Cornelissen, J. H. C., Werger, M. J. A., Castro-Diez, P., van Rheenen, J. W. 133. DunbarâCo, S., Sporck, M. J. & Sack, L. Leaf trait diversification and design A., & Rowland, A. P. Foliar nutrients in relation to growth, allocation and in seven rare taxa of the Hawaiian Plantago radiation. Int. J. Plant Sci. 170, leaf traits in seedlings of a wide range of woody plant species and types. 61–75 (2009). Oecologia 111, 460–469 (1997). 134. Durka, W. In BIOLFLOR—Eine Datenbank mit Biologisch-ökologischen 107. Cornelissen, J. H. C. et al. Leaf structure and defence control litter Merkmalen zur Flora von Deutschland (eds Klotz, S. et al.) 75–91 decomposition rate across species and life forms in regional floras on two (Bundesamt für Naturschutz, 2002). continents. New Phytol. 143, 191–200 (1999). 135. Durka, W. In BIOLFLOR—Eine Datenbank mit Biologisch-ökologischen 108. Cornelissen, J. H. C. A triangular relationship between leaf size and seed Merkmalen zur Flora von Deutschland (eds Klotz, S. et al.) 57–74 size among woody species: allometry, ontogeny, ecology and taxonomy. (Bundesamt für Naturschutz, 2002). Oecologia 118, 248–255 (1999). 136. Durka, W. In BIOLFLOR—Eine Datenbank mit Biologisch-ökologischen 109. Cornelissen, J. H. C., Aerts, R., Cerabolini, B., Werger, M. J. A. & van der Merkmalen zur Flora von Deutschland (eds Klotz, S. et al.) 133–175 Heijden., M. G. A. Carbon cycling traits of plant species are linked with (Bundesamt für Naturschutz, 2002). mycorrhizal strategy. Oecologia 129, 611–619 (2001). 137. Medlyn, B. E. & Jarvis, P. G. Design and use of a database of model 110. Cornelissen, J. H. C. et al. Leaf digestibility and litter decomposability are parameters from elevated [CO2] experiments. Ecol. Model. 124, 69–83 (1999). related in a wide range of subarctic plant species and types. Funct. Ecol. 18, 138. Everwand, G., Fry, E. L., Eggers, T. & Manning, P. Seasonal variation in the 779–786 (2004). capacity for plant trait measures to predict grassland carbon and water 111. Cornelissen, J. H. C. et al. Functional traits of woody plants: correspondence fluxes. Ecosystems 17, 1095–1108 (2014). of species rankings between field adults and laboratory-grown seedlings? 139. Fazayeli, F., Banerjee, A., Kattge, J., Schrodt, F. & Reich, P. B. Uncertainty J. Veg. Sci. 14, 311–322 (2003). quantified matrix completion using Bayesian Hierarchical Matrix 112. Cornelissen, J. H. C., Diez, P. C. & Hunt, R. Seedling growth, allocation and factorization. In Proc. 13th International Conference on Machine Learning leaf attributes in a wide range of woody plant species and types. J. Ecol. 84, and Applications (eds Ferri, C. et al.) 312–317 (International Conference on 755 (1996). Machine Learning and Applications (ICMLA), 2014). 113. Cornelissen, J. H. C. et al. Leaf structure and defence control litter 140. Fagúndez, J. & Izco, J. Seed morphology of the European species of Erica L. decomposition rate across species and life forms in regional floras on two sect. Arsace Salisb. ex Benth. (Ericaceae). Acta Bot. Gall. 157, 45–54 (2010). continents. New Phytol. 143, 191–200 (1999). 141. Fonseca, C. R., Overton, J. M., Collins, B. & Westoby, M. Shifts in 114. Schwilk, D. W., Cornwell, W. K. & Ackerly., D. D. A trait-based test for trait-combinations along rainfall and phosphorus gradients. J. Ecol. 88, habitat filtering: convex hull volume. Ecology 87, 1465–1471 (2006). 964–977 (2000). 115. Cornwell, W. K. & Ackerly, D. D. Community assembly and shifts in plant 142. Fortunel, C. et al. Leaf traits capture the effects of land use changes and trait distributions across an environmental gradient in coastal California. climate on litter decomposability of grasslands across Europe. Ecology 90, Ecol. Monogr. 79, 109–126 (2009). 598–611 (2009). 116. Cornwell, W. K., Bhaskar, R., Sack, L., Cordell, S. & Lunch, C. K. 143. Frainer, A. & McKie, B. G. Shifts in the diversity and composition of Adjustment of structure and function of Hawaiian Metrosideros polymorpha consumer traits constrain the effects of land use on stream ecosystem at high vs. low precipitation. Funct. Ecol. 21, 1063–1071 (2007). functioning. Adv. Ecol. Res. 52, 169–200 (2015). 117. Cornwell, W. K. et al. Plant species traits are the predominant control on 144. Frenette-Dussault, C., Shipley, B., Léger, J.-F., Meziane, D. & Hingrat, Y. litter decomposition rates within biomes worldwide. Ecol. Lett. 11, Functional structure of an arid steppe plant community reveals similarities 1065–1071 (2008). with Grime’s C-S-R theory. J. Veg. Sci. 23, 208–222 (2011). 118. Craine, J. M. et al. Global patterns of foliar nitrogen isotopes and their 145. Freschet, G. T., Cornelissen, J. H. C., van Logtestijn, R. S. P. & Aerts, R. relationships with climate, mycorrhizal fungi, foliar nutrient concentrations, Evidence of the plant economics spectrum in a subarctic flora. J. Ecol. 98, and nitrogen availability. New Phytol. 183, 980–992 (2009). 362–373 (2010). 119. Craine, J. M., Lee, W. G., Bond, W. J., Williams, R. J. & Johnson, L. C. 146. Freschet, G. T., Cornelissen, J. H. C., van Logtestijn, R. S. P. & Aerts, R. Environmental constraints on a global relationship among leaf and root Substantial nutrient resorption from leaves, stems and roots in a subarctic traits of grasses. Ecology 86, 12–19 (2005). flora: what is the link with other resource economics traits? New Phytol. 120. Craine, J. M. et al. Functional consequences of climate change-induced 186, 879–889 (2010). plant species loss in a tallgrass prairie. Oecologia 165, 1109–1117 (2011). 147. Fry, E. L., Power, S. A. & Manning, P. Trait-based classification and 121. Craine, J. M. et al. Global diversity of drought tolerance and grassland manipulation of plant functional groups for biodiversity–ecosystem climate-change resilience. Nat. Clim. Change 3, 63–67 (2012). function experiments. J. Veg. Sci. 25, 248–261 (2013). 122. Craven, D. et al. Between and within-site comparisons of structural and 148. Fyllas, N. M. et al. Basin-wide variations in foliar properties of Amazonian physiological characteristics and foliar nutrient content of 14 tree species at forest: phylogeny, soils and climate. Biogeosciences 6, 2677–2708 (2009). a wet, fertile site and a dry, infertile site in Panama. For. Ecol. Manag. 238, 149. Gachet, S., Véla, E. & Tatoni, T. BASECO: a floristic and ecological database 335–346 (2007). of Mediterranean French flora. Biodivers. Conserv. 14, 1023–1034 (2005). 123. Craven, D. et al. Seasonal variability of photosynthetic characteristics 150. Gallagher, R. V. & Leishman, M. R. A global analysis of trait variation and influences growth of eight tropical tree species at two sites with contrasting evolution in climbing plants. J. Biogeogr. 39, 1757–1771 (2012). precipitation in Panama. For. Ecol. Manag. 261, 1643–1653 (2011). 151. Garnier, E. et al. Assessing the effects of land-use change on plant traits, 124. Dainese, M. & Bragazza, L. Plant traits across different habitats of the communities and ecosystem functioning in grasslands: a standardized Italian alps: a comparative analysis between native and alien species. Alpine methodology and lessons from an application to 11 European sites. Ann. Bot. 122, 11–21 (2012). Bot. 99, 967–985 (2007). 125. de Araujo, A. et al. LBA-ECO CD-02 C and N Isotopes in Leaves and 152. Givnish, T. J., Montgomery, R. A. & Goldstein, G. Adaptive radiation of Atmospheric CO2, Amazonas, Brazil (ORNL DAAC, 2012); http://daac. photosynthetic physiology in the Hawaiian lobeliads: light regimes, static ornl.gov light responses, and whole-plant compensation points. Am. J. Bot. 91, 126. de Vries, F. T. & Bardgett, R. D. Plant community controls on short-term 228–246 (2004). ecosystem nitrogen retention. New Phytol. 210, 861–874 (2016). 153. Guerin, G. R., Wen, H. & Lowe, A. J. Leaf morphology shift linked to 127. Demey, A. et al. Nutrient input from hemiparasitic litter favors plant species climate change. Biol. Lett. 8, 882–886 (2012). with a fast-growth strategy. Plant Soil 371, 53–66 (2013). 154. Gutiérrez, A. G. & Huth, A. Successional stages of primary temperate 128. Diaz, S. et al. The plant traits that drive ecosystems: evidence from three rainforests of Chiloé Island, Chile. Perspect. Plant Ecol. Evol. Syst. 14, continents. J. Veg. Sci. 15, 295–304 (2004). 243–256 (2012). 129. Domingues, T. F., Berry, J. A., Martinelli, L. A., Ometto, J. P. H. B. & 155. Guy, A. L., Mischkolz, J. M. & Lamb, E. G. Limited effects of simulated Ehleringer, J. R. Parameterization of canopy structure and leaf-level acidic deposition on seedling survivorship and root morphology of endemic gas exchange for an eastern Amazonian tropical rain forest (Tapajós plant taxa of the Athabasca sand dunes in well-watered greenhouse trials. National Forest, Pará, Brazil). Earth Interact. https://doi.org/10.1175/EI149.1 Botany 91, 176–181 (2013). (2005). 156. Han, W. et al. Floral, climatic and soil pH controls on leaf ash content in 130. Domingues, T. F., Martinelli, L. A. & Ehleringer, J. R. Ecophysiological traits China’s terrestrial plants. Glob. Ecol. Biogeogr. 21, 376–382 (2011). of plant functional groups in forest and pasture ecosystems from eastern 157. Han, W., Fang, J., Guo, D. & Zhang, Y. Leaf nitrogen and phosphorus Amazônia, Brazil. Plant Ecol. 193, 101–112 (2007). stoichiometry across 753 terrestrial plant species in China. New Phytol. 168, 131. Domingues, T. F. et al. Co-limitation of photosynthetic capacity by 377–385 (2005). nitrogen and phosphorus in West Africa woodlands. Plant Cell Environ. 33, 158. Hao, G.-Y., Sack, L., Wang, A.-Y., Cao, K.-F. & Goldstein, G. Differentiation 959–980 (2010). of leaf water flux and drought tolerance traits in hemiepiphytic and 132. Duarte, Ld. S., Carlucci, M. B., Hartz, S. M. & Pillar, V. D. Plant dispersal non-hemiepiphytic Ficus tree species. Funct. Ecol. 24, 731–740 (2010). strategies and the colonization of Araucaria forest patches in a grassland– 159. He, J.-S. et al. A test of the generality of leaf trait relationships on the forest mosaic. J. Veg. Sci. 18, 847–858 (2007). Tibetan plateau. New Phytol. 170, 835–848 (2006). Nature Ecology & Evolution | VOL 6 | January 2022 | 36–50 | www.nature.com/natecolevol 45 Articles NATurE Ecology & EVoluTIon 160. Hickler, T. Plant Functional Types and Community Characteristics along 189. Laughlin, D. C., Fulé, P. Z., Huffman, D. W., Crouse, J. & Laliberté, E. Climatic Environmental Gradients on Öland’s Great Alvar (Sweden). Masters thesis constraints on trait-based forest assembly. J. Ecol. 99, 1489–1499 (2011). (University of Lund, 1999). 190. Laughlin, D. C., Leppert, J. J., Moore, M. M. & Sieg, C. H. A multi-trait test 161. Hoof, J., Sack, L., Webb, D. T. & Nilsen, E. T. Contrasting structure and of the leaf-height-seed plant strategy scheme with 133 species from a pine function of pubescent and glabrous varieties of Hawaiian Metrosideros forest flora. Funct. Ecol. 24, 493–501 (2009). polymorpha (Myrtaceae) at high elevation. Biotropica 40, 113–118 (2008). 191. Lens, F. Comparative wood anatomy of Epacrids (Styphelioideae, 162. Husson, A. F., Josse, J., Le, S., Mazet, J. & Husson, M. F. Package Ericaceae s.l.). Ann. Bot. 91, 835–856 (2003). ‘FactoMineR’ (CRAN, 2017). 192. Lens, F., Baas, P., Jansen, S. & Smets, E. A search for phylogenetically 163. Jacobs, B. et al. Unraveling the Phylogeny of Heptacodium and informative wood characters within Lecythidaceae s.l. Am. J. Bot. 94, Zabelia (Caprifoliaceae): An Interdisciplinary Approach. Syst. Bot. 36, 483–502 (2007). 231–252 (2011). 193. Lens, F., Dressler, S., Jansen, S., van Evelghem, L. & Smets, E. Relationships 164. Jansen, S., Decraene, L. P. R. & Smets, E. On the wood and stem anatomy within balsaminoid Ericales: a wood anatomical approach. Am. J. Bot. 92, of Monococcus echinophorus (Phytolaccaceae s.l.). Syst. Geogr. Plants 70, 941–953 (2005). 171 (2000). 194. Lens, F., Eeckhout, S., Zwartjes, R., Smets, E. & Janssens, S. B. The multiple 165. Jansen, S. et al. Contributions to the wood anatomy of the Rubioideae fuzzy origins of woodiness within Balsaminaceae using an integrated (Rubiaceae). J. Plant Res. 114, 269–289 (2001). approach: where do we draw the line? Ann. Bot. 109, 783–799 (2011). 166. Jansen, S., Piesschaert, F. & Smets, E. Wood anatomy of Elaeagnaceae, with 195. Lens, F., Endress, M. E., Baas, P., Jansen, S. & Smets, E. Vessel grouping comments on vestured pits, helical thickenings, and systematic patterns in subfamilies Apocynoideae and Periplocoideae confirm relationships. Am. J. Bot. 87, 20 (2000). phylogenetic value of wood structure within Apocynaceae. Am. J. Bot. 96, 167. Jansen, S., Robbrecht, E., Beeckman, H. & Smets, E. Gaertnera and 2168–2183 (2009). Pagamea: genera within the Psychotrieae or constituting the tribe 196. Lens, F., Groeninckx, I., Smets, E. & Dessein, S. Woodiness within the Gaertnereae? A wood anatomical and palynological approach. Bot. Acta Spermacoceae–Knoxieae alliance (Rubiaceae): retention of the basal 109, 466–476 (1996). woody condition in Rubiaceae or recent innovation? Ann. Bot. 103, 168. S., J., E., R., H., B. & Smets, E. Comparative wood anatomy of African 1049–1064 (2009). Coffeae (Rubiaceae-Rubioideae). Belg. J. Bot. 130, 47–58 (1997). 197. Lens, F., Jansen, S., Caris, P., Serlet, L. & Smets, E. Comparative wood 169. Kattge, J., Knorr, W., Raddatz, T. & Wirth, C. Quantifying photosynthetic anatomy of the primuloid clade (Ericales s.l.). Syst. Bot. 30, 163–183 (2005). capacity and its relationship to leaf nitrogen content for global-scale 198. Lens, F., Jansen, S., Robbrecht, E. & Smets, E. Wood anatomy of the terrestrial biosphere models. Glob. Change Biol. 15, 976–991 (2009). Vangueriaea (Ixoroideae-Rubuaceae), with special emphasis on some 170. Kazakou, E., Vile, D., Shipley, B., Gallet, C. & Garnier, E. Co-variations in geofrutices. IAWA J. 21, 443–455 (2000). litter decomposition, leaf traits and plant growth in species from a 199. Lens, F. et al. The wood anatomy of the polyphyletic Icacinaceae s.l., and Mediterranean old-field succession. Funct. Ecol. 20, 21–30 (2006). their relationships within asterids. Taxon 57, 525–552 (2008). 171. Kerkhoff, A. J., Fagan, W. F., Elser, J. J. & Enquist, B. J. Phylogenetic and 200. Lens, F., Kron, K. A., Luteyn, J. L., Smets, E. & Jansen, S. Comparative growth form variation in the scaling of nitrogen and phosphorus in the wood anatomy of the blueberry tribe (Vaccinieae, Ericaceae s.l). Ann. seed plants. Am. Nat. 168, E103–E122 (2006). Missouri Bot. Gard. 91, 566–592 (2004). 172. Kew, R. B. G. Seed Information Database—SID (Kew, 2008); http://data.kew. 201. Lens, F., Smets, E. & Jansen, S. Comparative wood anatomy of org/sid/ Andromedeae s.s., Gaultherieae, Lyonieae and Oxydendreae (Vaccinioideae, 173. Kichenin, E., Wardle, D. A., Peltzer, D. A., Morse, C. W. & Freschet, G. T. Ericaceae s.l.). Bot. J. Linn. Soc. 144, 161–179 (2004). Contrasting effects of plant inter- and intraspecific variation on 202. Lens, F., Smets, E. & Melzer, S. Stem anatomy supports Arabidopsis thaliana community-level trait measures along an environmental gradient. Funct. as a model for insular woodiness. New Phytol. 193, 12–17 (2011). Ecol. 27, 1254–1261 (2013). 203. Lens, F. et al. Testing hypotheses that link wood anatomy to cavitation 174. Kier, G. et al. Global patterns of plant diversity and floristic knowledge. resistance and hydraulic conductivity in the genus Acer. New Phytol. 190, J. Biogeogr. 32, 1107–1116 (2005). 709–723 (2010). 175. Kirkup, D., Malcolm, P., Christian, G. & Paton, A. Towards a digital African 204. Li, H., Liang, Y., Xu, Q. & Cao, D. Key wavelengths screening using flora. Taxon 54, 457 (2005). competitive adaptive reweighted sampling method for multivariate 176. Kleyer, M. et al. The LEDA traitbase: a database of life-history traits of the calibration. Anal. Chim. Acta 648, 77–84 (2009). northwest European flora. J. Ecol. 96, 1266–1274 (2008). 205. Louault, F., Pillar, V. D., Aufrèère, J., Garnier, E. & Soussana, J. F. Plant 177. Klotz, S. & Kühn, I. in BIOLFLOR—Eine Datenbank mit traits and functional types in response to reduced disturbance in a Biologisch-ökologischen Merkmalen zur Flora von Deutschland (eds Klotz, S. semi-natural grassland. J. Veg. Sci. 16, 151–160 (2005). et al.) 119-126 (Bundesamt für Naturschutz, 2002). 206. Loveys, B. R. et al. Thermal acclimation of leaf and root respiration: an 178. Klotz, S. & Kühn, I. in BIOLFLOR—Eine Datenbank mit investigation comparing inherently fast- and slow-growing plant species. Biologisch-ökologischen Merkmalen zur Flora von Deutschland (eds Klotz, S. Glob. Change Biol. 9, 895–910 (2003). et al.) 241–246 (Bundesamt für Naturschutz,2002). 207. Malhado, A. C. M. et al. Drip-tips are associated with intensity of 179. Klotz, S. & Kühn, I. in BIOLFLOR—Eine Datenbank mit precipitation in the Amazon rain forest. Biotropica 44, 728–737 (2012). Biologisch-ökologischen Merkmalen zur Flora von Deutschland (eds Klotz, S. 208. Malhado, A. C. M. et al. Spatial trends in leaf size of Amazonian rainforest et al.) 273–281 (Bundesamt für Naturschutz, 2002). trees. Biogeosciences 6, 1563–1576 (2009). 180. Klotz, S. & Kühn, I. in BIOLFLOR—Eine Datenbank mit 209. Malhado, A. C. M. et al. Spatial distribution and functional significance of leaf Biologisch-ökologischen Merkmalen zur Flora von Deutschland (eds Klotz, S. lamina shape in Amazonian forest trees. Biogeosciences 6, 1577–1590 (2009). et al.) 197–201 (Bundesamt für Naturschutz, 2002). 210. Malhado, A. C. M. et al. Are compound leaves an adaptation to seasonal 181. Koike, F. Plant traits as predictors of woody species dominance in climax drought or to rapid growth? Evidence from the Amazon rain forest. Glob. forest communities. J. Veg. Sci. 12, 327–336 (2001). Ecol. Biogeogr. 19, 852–862 (2010). 182. Kraft, N. J. B. & Ackerly, D. D. Functional trait and phylogenetic tests of 211. Manning, P., Houston, K. & Evans, T. Shifts in seed size across experimental community assembly across spatial scales in an Amazonian forest. Ecol. nitrogen enrichment and plant density gradients. Basic Appl. Ecol. 10, Monogr. 80, 401–422 (2010). 300–308 (2009). 183. Kraft, N. J. B., Valencia, R. & Ackerly, D. D. Functional traits and 212. Markesteijn, L., Poorter, L., Paz, H., Sack, L. & Bongers, F. Ecological niche-based tree community assembly in an Amazonian forest. Science 322, differentiation in xylem cavitation resistance is associated with stem and 580–582 (2008). leaf structural traits. Plant Cell Environ. 34, 137–148 (2011). 184. Krumbiegel, A. in BIOLFLOR—Eine Datenbank mit Biologisch-ökologischen 213. Martin, R. E., Asner, G. P. & Sack, L. Genetic variation in leaf pigment, Merkmalen zur Flora von Deutschland (eds Klotz, S. et al.) 93–118 optical and photosynthetic function among diverse phenotypes of (Bundesamt für Naturschutz, 2002). Metrosideros polymorpha grown in a common garden. Oecologia 151, 185. Kühn, I. in BIOLFLOR—Eine Datenbank mit Biologisch-ökologischen 387–400 (2007). Merkmalen zur Flora von Deutschland (eds Klotz, S. et al.) 47–56 214. McDonald, P. G., Fonseca, C. R., Overton, J. M. & Westoby, M. Leaf-size (Bundesamt für Naturschutz, 2002). divergence along rainfall and soil-nutrient gradients: is the method of size 186. Kuhn, I., Durka, W. & Klotz, S. Biolflor—a new plant-trait database as a reduction common among clades? Funct. Ecol. 17, 50–57 (2003). tool for plant invasion ecology. Divers. Distrib. 10, 363–365 (2004). 215. McKenna, M. F. & Shipley, B. Interacting determinants of interspecific 187. Kühn, I. & Klotz, S. in BIOLFLOR—Eine Datenbank mit relative growth: empirical patterns and a theoretical explanation. Écoscience Biologisch-ökologischen Merkmalen zur Flora von Deutschland (eds Klotz, S. 6, 286–296 (1999). et al.) 227–239 (Bundesamt für Naturschutz, 2002). 216. Medlyn, B. E. et al. Effects of elevated [CO 2] on photosynthesis in European 188. Kurokawa, H. & Nakashizuka, T. Leaf herbivory and decomposability in a forest species: a meta-analysis of model parameters. Plant Cell Environ. 22, Malaysian tropical rain forest. Ecology 89, 2645–2656 (2008). 1475–1495 (1999). 46 Nature Ecology & Evolution | VOL 6 | January 2022 | 36–50 | www.nature.com/natecolevol NATurE Ecology & EVoluTIon Articles 217. Medlyn, B. E. et al. Stomatal conductance of forest species after long-term 246. Papanastasis, M. et al. Leaf traits capture the effects of land use changes and exposure to elevated CO2 concentration: a synthesis. New Phytol. 149, climate on litter decomposability of grasslands across Europe. Ecology 90, 247–264 (2001). 598–611 (2009). 218. Meir, P. et al. Acclimation of photosynthetic capacity to irradiance in tree 247. Patiño, S. et al. Branch xylem density variations across the Amazon basin. canopies in relation to leaf nitrogen concentration and leaf mass per unit Biogeosciences 6, 545–568 (2009). area. Plant Cell Environ. 25, 343–357 (2002). 248. Paula, S. et al. Fire-related traits for plant species of the Mediterranean 219. Meir, P., Levy, P. E., Grace, J. & Jarvis, P. G. Photosynthetic parameters from basin. Ecology 90, 1420–1420 (2009). two contrasting woody vegetation types in West Africa. Plant Ecol. 192, 249. Paula, S. & Pausas, J. G. Burning seeds: germinative response to heat 277–287 (2007). treatments in relation to resprouting ability. J. Ecol. 96, 543–552 (2008). 220. Mencuccini, M. The ecological significance of long-distance water transport: 250. Peco, B., de Pablos, I., Traba, J. & Levassor, C. The effect of grazing short-term regulation, long-term acclimation and the hydraulic costs of abandonment on species composition and functional traits: the case of stature across plant life forms. Plant Cell Environ. 26, 163–182 (2003). Dehesa grasslands. Basic Appl. Ecol. 6, 175–183 (2005). 221. Meng, T.-T. et al. Responses of leaf traits to climatic gradients: Adaptive 251. Peñuelas, J. et al. Faster returns on ‘leaf economics’ and different variation versus compositional shifts. Biogeosciences 12, 5339–5352 (2015). biogeochemical niche in invasive compared with native plant species. 222. Messier, J., McGill, B. J., Enquist, B. J. & Lechowicz, M. J. Trait variation Glob. Change Biol. 16, 2171–2185 (2009). and integration across scales: is the leaf economic spectrum present at local 252. Peñuelas, J. et al. Higher allocation to low cost chemical defenses in scales? Ecography 40, 685–697 (2016). invasive species of Hawaii. J. Chem. Ecol. 36, 1255–1270 (2010). 223. Messier, J., McGill, B. J. & Lechowicz, M. J. How do traits vary across 253. Petter, G. et al. Functional leaf traits of vascular epiphytes: vertical trends ecological scales? A case for trait-based ecology. Ecol. Lett. 13, within the forest, intra- and interspecific trait variability, and taxonomic 838–848 (2010). signals. Funct. Ecol. 30, 188–198 (2015). 224. Meziane, D. & Shipley, B. Interacting components of interspecific relative 254. Pierce, S., Brusa, G., Sartori, M. & Cerabolini, B. E. L. Combined use of leaf growth rate: constancy and change under differing conditions of light and size and economics traits allows direct comparison of hydrophyte and nutrient supply. Funct. Ecol. 13, 611–622 (1999). terrestrial herbaceous adaptive strategies. Ann. Bot. 109, 1047–1053 (2012). 225. Milla, R. & Reich, P. B. Multi-trait interactions, not phylogeny, 255. Pierce, S., Brusa, G., Vagge, I. & Cerabolini, B. E. L. Allocating CSR plant fine-tune leaf size reduction with increasing altitude. Ann. Bot. 107, functional types: the use of leaf economics and size traits to classify woody 455–465 (2011). and herbaceous vascular plants. Funct. Ecol. 27, 1002–1010 (2013). 226. Minden, V., Andratschke, S., Spalke, J., Timmermann, H. & Kleyer, M. 256. Pierce, S., Ceriani, R. M., De Andreis, R., Luzzaro, A. & Cerabolini, B. The Plant trait–environment relationships in salt marshes: deviations from leaf economics spectrum of Poaceae reflects variation in survival strategies. predictions by ecological concepts. Perspect. Plant Ecol. Evol. Syst. 14, Plant Biosyst. 141, 337–343 (2007). 183–192 (2012). 257. Pierce, S., Luzzaro, A., Caccianiga, M., Ceriani, R. M. & Cerabolini, B. 227. Minden, V. & Kleyer, M. Testing the effect–response framework: key Disturbance is the principal α-scale filter determining niche differentiation, response and effect traits determining above-ground biomass of salt coexistence and biodiversity in an alpine community. J. Ecol. 95, marshes. J. Veg. Sci. 22, 387–401 (2011). 698–706 (2007). 228. Mischkolz, J. M. Selecting and Evaluating Native Forage Mixtures for the 258. Pillar, V. D. & Sosinski, E. E. An improved method for searching plant Mixed Grass Prairie. Msc thesis (University of Saskatchewan, 2013). functional types by numerical analysis. J. Veg. Sci. 14, 323–332 (2003). 229. Moretti, M. & Legg, C. Combining plant and animal traits to assess 259. Powers, J. S. & Tiffin, P. Plant functional type classifications in tropical dry community functional responses to disturbance. Ecography 32, forests in Costa Rica: leaf habit versus taxonomic approaches. Funct. Ecol. 299–309 (2009). 24, 927–936 (2010). 230. Müller, S. C., Overbeck, G. E., Pfadenhauer, J. & Pillar, V. D. Plant 260. Prentice, I. C. et al. Evidence of a universal scaling relationship for leaf CO 2 functional types of woody species related to fire disturbance in forest– drawdown along an aridity gradient. New Phytol. 190, 169–180 (2010). grassland ecotones. Plant Ecol. 189, 1–14 (2006). 261. Preston, K. A., Cornwell, W. K. & DeNoyer, J. L. Wood density and vessel 231. Nakahashi, C. D., Frole, K. & Sack, L. Bacterial leaf nodule symbiosis in traits as distinct correlates of ecological strategy in 51 California coast range Ardisia (Myrsinaceae): does it contribute to seedling growth capacity? angiosperms. New Phytol. 170, 807–818 (2006). Plant Biol. 7, 495–500 (2005). 262. Price, C. A. & Enquist, B. J. Scaling mass and morphology in leaves: an 232. Niinemets, U. Components of leaf dry mass per area—thickness and extention of the WBE model. Ecology 88, 1132–1141 (2007). density—alter leaf photosynthetic capacity in reverse directions in woody 263. Price, C. A., Enquist, B. J. & Savage, V. M. A general model for allometric plants. New Phytol. 144, 35–47 (1999). covariation in botanical form and function. Proc. Natl Acad. Sci. USA 104, 233. Niinemets, U. Global-scale climatic controls of leaf dry mass per area, 13204–13209 (2007). density, and thickness in trees and shrubs. Ecology 82, 453–469 (2001). 264. Pyankov, V. I., Kondratchuk, A. V. & Shipley, B. Leaf structure and specific 234. Ogaya, R. & Peñuelas, J. Comparative field study of Quercus ilex and leaf mass: the alpine desert plants of the Eastern Pamirs, Tadjikistan. New Phillyrea latifolia: photosynthetic response to experimental drought Phytol. 143, 131–142 (1999). conditions. Environ. Exp. Bot. 50, 137–148 (2003). 265. Quero, J. L. et al. Relating leaf photosynthetic rate to whole-plant growth: 235. Ogaya, R. & Penuelas, J. Contrasting foliar responses to drought in Quercus drought and shade effects on seedlings of four Quercus species. Funct. Plant ilex and Phillyrea latifolia. Biol. Plant. 50, 373–382 (2006). Biol. 35, 725 (2008). 236. Ogaya, R. & Peñuelas, J. Tree growth, mortality, and above-ground biomass 266. Quested, H. M. et al. Decomposition of sub-arctic plants with differing accumulation in a holm oak forest under a five-year experimental field nitrogen economies: a functional role for hemiparasites. Ecology 84, drought. Plant Ecol. 189, 291–299 (2006). 3209–3221 (2003). 237. Ogaya, R. & Peñuelas, J. Changes in leaf δ13C and δ15N for three 267. Reich, P. B., Oleksyn, J. & Wright, I. J. Leaf phosphorus influences the Mediterranean tree species in relation to soil water availability. Acta Oecol. photosynthesis–nitrogen relation: a cross-biome analysis of 314 species. 34, 331–338 (2008). Oecologia 160, 207–212 (2009). 238. Onoda, Y. et al. Global patterns of leaf mechanical properties. Ecol. Lett. 14, 268. Reich, P. B. et al. Scaling of respiration to nitrogen in leaves, stems and 301–312 (2011). roots of higher land plants. Ecol. Lett. 11, 793–801 (2008). 239. Ordoñez, J. C. et al. Leaf habit and woodiness regulate different leaf 269. Auger, S. & Shipley, B. Inter-specific and intra-specific trait variation along economy traits at a given nutrient supply. Ecology 91, 3218–3228 (2010). short environmental gradients in an old-growth temperate forest. J. Veg. Sci. 240. Otto, B. in BIOLFLOR—Eine Datenbank mit Biologisch-ökologischen 24, 419–428 (2012). Merkmalen zur Flora von Deutschland (eds Klotz, S. et al.) 177–196 270. Sack, L., Cowan, P. D., Jaikumar, N. & Holbrook, N. M. The ’hydrology’ of (Bundesamt für Naturschutz, 2002). leaves: co-ordination of structure and function in temperate woody species. 241. Overbeck, G. E., Müller, S. C., Pillar, V. D. & Pfadenhauer, J. Fine-scale Plant Cell Environ. 26, 1343–1356 (2003). post-fire dynamics in southern Brazilian subtropical grassland. J. Veg. Sci. 271. Sack, L. & Frole, K. Leaf structural diversity is related to hydraulic capacity 16, 655–664 (2005). in tropical rain forest trees. Ecology 87, 483–491 (2006). 242. Overbeck, G. E. & Pfadenhauer, J. Adaptive strategies in burned subtropical 272. Sack, L., Melcher, P. J., Liu, W. H., Middleton, E. & Pardee, T. How strong is grassland in southern Brazil. Flora 202, 27–49 (2007). intracanopy leaf plasticity in temperate deciduous trees? Am. J. Bot. 93, 243. Baas, P., Smets, E. & Jansen, S. Vegetative anatomy and effinities of 829–839 (2006). Dirachma socotrana (Dirachmaceae). Syst. Bot. 26, 231–241 (2001). 273. Sack, L., Tyree, M. T. & Holbrook, N. M. Leaf hydraulic architecture 244. Pakeman, R. J. et al. Impact of abundance weighting on the response of correlates with regeneration irradiance in tropical rainforest trees. New seed traits to climate and land use. J. Ecol. 96, 355–366 (2008). Phytol. 167, 403–413 (2005). 245. Pakeman, R. J., Lep, J., Kleyer, M., Lavorel, S. & Garnie, E. Relative climatic, 274. Sanda V., Bita-Nicolae, C. D. & Barabas, N. The Flora of Spontaneous and edaphic and management controls of plant functional trait signatures. Cultivated Cormophytes from Romania (in Romanian) (Editura Ion J. Veg. Sci. 20, 148–159 (2009). Bacău, 2003). Nature Ecology & Evolution | VOL 6 | January 2022 | 36–50 | www.nature.com/natecolevol 47 Articles NATurE Ecology & EVoluTIon 275. Sandel, B., Corbin, J. D. & Krupa, M. Using plant functional traits to 303. Wohlfahrt, G. et al. Inter-specific variation of the biochemical limitation to guide restoration: a case study in California coastal grassland. Ecosphere 2, photosynthesis and related leaf traits of 30 species from mountain grassland art23 (2011). ecosystems under different land use. Plant Cell Environ. 22, 1281–1296 276. Sardans, J., Penuelas, J. & Ogaya, R. Drought-induced changes in C (1999). and N stoichiometry in a Quercus ilex Mediterranean forest. For. Sci. 54, 304. Wright, I. J. et al. Relationships among ecologically important dimensions 513–522 (2008). of plant trait variation in seven neotropical forests. Ann. Bot. 99, 1003–1015 277. Sardans, J., Peñuelas, J., Prieto, P. & Estiarte, M. Changes in Ca, Fe, Mg, (2007). Mo, Na, and S content in a Mediterranean shrubland under warming and 305. Wright, J. P. & Sutton-Grier, A. Does the leaf economic spectrum hold drought. J. Geophys. Res. https://doi.org/10.1029/2008jg000795 (2008). within local species pools across varying environmental conditions? Funct. 278. Scherer-Lorenzen, M., Schulze, E., Don, A., Schumacher, J. & Weller, E. Ecol. 26, 1390–1398 (2012). Exploring the functional significance of forest diversity: a new long-term 306. Wright, S. J. et al. Functional traits and the growth–mortality trade-off in experiment with temperate tree species (biotree). Perspect. Plant Ecol. Evol. tropical trees. Ecology 91, 3664–3674 (2010). Syst. 9, 53–70 (2007). 307. Xu, L. & Baldocchi, D. D. Seasonal trends in photosynthetic parameters and 279. Schurr, F. M. et al. Colonization and persistence ability explain the extent to stomatal conductance of blue oak (Quercus douglasii) under prolonged which plant species fill their potential range. Global Ecol. Biogeogr. 16, summer drought and high temperature. Tree Physiol. 23, 865–877 (2003). 449–459 (2007). 308. Yguel, B. et al. Phytophagy on phylogenetically isolated trees: why hosts 280. Schwallier, R. et al. Evolution of wood anatomical characters in Nepenthes should escape their relatives. Ecol. Lett. 14, 1117–1124 (2011). and close relatives of Caryophyllales. Ann. Bot. 119, 1179–1193 (2017). 309. Zanne, A. E. et al. Global Wood Density Database (EOL, 2009); https:// 281. Schweingruber, F. H., & Poschlod, P. Growth rings in herbs and shrubs: life opendata.eol.org/dataset/dde86ffb-7741-44a1-acf2-808b3dd6bc97/resource/ span, age determination and stem anatomy. Forest Snow Landsc. Res. 79, d1e2b018-a7ce-444b-ac8a-ac43b2355cc9/download/archive 195–415 (2005). 310. Zanne, A. E. et al. Angiosperm wood structure: global patterns in vessel 282. Scoffoni, C., Pou, A., Aasamaa, K. & Sack, L. The rapid light response of anatomy and their relation to wood density and potential conductivity. Am. leaf hydraulic conductance: new evidence from two experimental methods. J. Bot. 97, 207–215 (2010). Plant Cell Environ. 31, 1803–1812 (2008). 311. Kattge, V. et al. TRY - a global database of plant traits. Global Change Biol 283. Shiodera, S., Rahajoe, J. S. & Kohyama, T. Variation in longevity and traits 9, 2905–2935 (2011). of leaves among co-occurring understorey plants in a tropical montane 312. Shan, H. et al. Gap Filling in the Plant Kingdom—Trait Prediction Using forest. J. Trop. Ecol. 24, 121–133 (2008). Hierarchical Probabilistic Matrix Factorization (ICML, 2012); http://arxiv. 284. Shipley, B. The use of above-ground maximum relative growth rate as an org/abs/1206.6439 accurate predictor of whole-plant maximum relative growth rate. Funct. 313. R Core Team. R: A Language and Environment for Statistical Computing (R Ecol. 3, 771 (1989). Foundation for Statistical Computing, 2021). 285. Shipley, B. Trade-offs between net assimilation rate and specific leaf area in 314. Salakhutdinov, R. & Mnih, A. Probabilistic matrix factorization. In Proc. determining relative growth rate: relationship with daily irradiance. Funct. 20th International Conference on Neural Information Processing Systems (eds Ecol. 16, 682–689 (2002). Platt, J. C. et al.) 1257–1264 (Curran Associates Inc., 2007). 286. Shipley, B. & Lechowicz, M. J. The functional co-ordination of leaf 315. R Core Team. R: A Language and Environment for Statistical Computing (R morphology, nitrogen concentration, and gas exchange in 40 wetland Foundation for Statistical Computing, 2013). species. Écoscience 7, 183–194 (2000). 316. Lê, S., Josse, J. & Husson, F. FactoMineR: a package for multivariate 287. Shipley, B. & Parent, M. Germination responses of 64 wetland species in analysis. J. Stat. Softw. 25, 1–18 (2008). relation to seed size, minimum time to reproduction and seedling relative 317. Dray, S. & Dufour, A.-B. The ade4 package: implementing the duality growth rate. Funct. Ecol. 5, 111 (1991). diagram for ecologists. J. Stat. Softw. 22, 1–20 (2007). 288. Shipley, B. & Vu, T.-T. Dry matter content as a measure of dry matter 318. Bougeard, S. & Dray, S. Supervised multiblock analysis in R with the ade4 concentration in plants and their parts. New Phytol. 153, 359–364 (2002). package. J. Stat. Softw. 86, 1–17 (2018). 289. Spasojevic, M. J. & Suding, K. N. Inferring community assembly 319. Chessel, D., Dufour, A.-B. & Thioulouse, J. The ade4 package—I: one-table mechanisms from functional diversity patterns: the importance of multiple methods. R News 4, 5–10 (2004). assembly processes. J. Ecol. 100, 652–661 (2012). 320. Dray, S., Dufour, A.-B. & Chessel, D. The ade4 package—II: two-table and 290. Swaine, E. K. Ecological and Evolutionary Drivers of Plant Community K-table methods. R News 7, 47–52 (2007). Assembly in a Bornean Rain Forest. PhD Thesis (University of Aberdeen, 321. Thioulouse, J. et al. Multivariate Analysis of Ecological Data with ade4 2007). (Springer, 2018). 291. Trefflich, A., Klotz, S. & Kuhn, I. in BIOLFLOR—Eine Datenbank mit 322. Friedman, J., Hastie, T. & Tibshirani, R. Regularization paths for generalized Biologisch-ökologischen Merkmalen zur Flora von Deutschland (eds Klotz, S. linear models via coordinate descent. J. Stat. Softw. 33, 1–22 (2010). et al.) 127–131 (Bundesamt für Naturschutz, 2002). 323. Simon, N., Friedman, J., Hastie, T. & Tibshirani, R. Regularization paths for 292. Tucker, S. S., Craine, J. M. & Nippert, J. B. Physiological drought tolerance Cox’s proportional hazards model via coordinate descent. J. Stat. Softw. 39, and the structuring of tallgrass prairie assemblages. Ecosphere 2, art48 (2011). 1–13 (2011). 293. Ciocarlan, V. The Illustrated Flora of Romania. Pteridophyta et Spermatopyta 324. Batjes, N. H., Ribeiro, E. & van Oostrum, A. Standardised soil profile data (in Romanian) (Editura Ceres, 2009). to support global mapping and modelling (WoSIS snapshot 2019). Earth 294. van Bodegom, P. M., Sorrell, B. K., Oosthoek, A., Bakker, C. & Aerts, R. Syst. Sci. Data 12, 299–320 (2020). Separating the effects of partial submergence and soil oxygen demand on 325. Hengl, T. et al. SoilGrids1km—global soil information based on automated plant physiology. Ecology 89, 193–204 (2008). mapping. PLoS ONE 9, e105992 (2014). 295. Vergutz, L. et al. A Global Database of Carbon and Nutrient Concentrations 326. Arrouays, D. et al. Soil legacy data rescue via GlobalSoilMap and other of Green and Senesced Leaves (ORNL DAAC, 2012); https://doi.org/10.3334/ international and national initiatives. GeoResJ 14, 1–19 (2017). ORNLDAAC/1106 327. Richard, P. & Pielou, E. C. Biogeography (John Wiley & Sons, 1979). 296. Vergutz, L., Manzoni, S., Porporato, A., Novais, R. F. & Jackson, R. B. 328. Udvardy, M. D. F. A Classification of the Biogeographical Provinces of the Global resorption efficiencies and concentrations of carbon and nutrients in World (International Union for Conservation of Nature and Natural leaves of terrestrial plants. Ecol. Monogr. 82, 205–220 (2012). Resources, 1975). 297. Vile, D. Significations Fonctionnelle et Ecologique des Traits des Especes 329. Dinerstein, E. et al. A Conservation Assessment of the Terrestrial Ecoregions Vegetales: Exemple dans une Succession Post-cultural Méditerranéenne et of Latin America and the Caribbean (The World Bank, 1995). Generalisations. PhD thesis (University of Montpellier II, 2005). 330. Ricketts, T. H. et al. Terrestrial Ecoregions of North America: A Conservation 298. Von Holle, B. & Simberloff, D. Testing Fox’s assembly rule: does plant invasion Assessment (Island Press, 1999). depend on recipient community structure? Oikos 105, 551–563 (2004). 331. Dasmann, R. F. A System for Defining and Classifying Natural Regions for 299. Williams, M., Shimabukuro, Y. E. & Rastetter, E.B. LBA-ECO CD-09 Soil Purposes of Conservation: A Progress Report (IUCN, 1973). and Vegetation Characteristics, Tapajos National Forest, Brazil (ORNL DAAC, 2012); https://doi.org/10.3334/ORNLDAAC/1104 300. Willis, C. G. et al. Phylogenetic community structure in Minnesota oak Acknowledgements savanna is influenced by spatial extent and environmental variation. The study was supported by the TRY initiative on plant traits (http://www.try-db. Ecography 33, 565–577 (2010). org). The TRY database is hosted at the Max Planck Institute for Biogeochemistry 301. Wilson, K. B., Baldocchi, D. D. & Hanson, P. J. Spatial and seasonal (MPI BGC, Germany) and supported by Future Earth and the German Centre for variability of photosynthetic parameters and their relationship to leaf Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig. We would like to thank all nitrogen in a deciduous forest. Tree Physiol. 20, 565–578 (2000). PIs contributing to the TRY database, whose efforts allowed this analysis. In detail, we 302. Wirth, C. & Lichstein, J. W. in Old-Growth Forests: Function, Fate and Value thank: J.H.C. Cornelissen, R. Milla, W. Cornwell, K. Kramer, S. Gachet, Ingolf Kühn, P. (eds Wirth, C. et al.) 81–113 (Springer, 2009). Poschlod, M. Scherer, J. Pausas, B. Sandal, K. Verheyen, J. Penuelas, N. Soudzilovskaia, 48 Nature Ecology & Evolution | VOL 6 | January 2022 | 36–50 | www.nature.com/natecolevol NATurE Ecology & EVoluTIon Articles P. Reich, J. Fang, S. Harrison, R. Gallagher, B. Hawkins, B. Finegan, J. Powers, F. Lenti, funded by FONDECYT grant nos 11150835 and 1200468. V.O. thanks Russian science S. Higgins, B. Medlyn, H. Ford, V. Pillar, M. Bahn, E. Sosinski, T. He, B. Cerabolini, foundation (RSF, 19-14-00038) for financial support. J. Cavender-Bares, I. J. Wright, F. Louault, B. Amiaud, G. Gonzalez-Melo, P. Adler, F. Schurr, J. Craine, Y. Niinemets, A. Zanne, H. Jactel, M. Harze, R. Montgomery, C. Author contributions Römermann, T. Hickler, A. Pahl, M. Dainese, D. Kirkup, J. Dickie, W. Hattingh, P. B.R., C.W., J.K., J.S.J. and M.D.M. were responsible for conceptualization. B.R., C.W., J.S.J. Higuchi, T. Domingues, A. Araujo, M. Williams, C. Price, B. Shipley, L. Sack, B. Schamp, and M.D.M. developed the methodology. J.S.J. undertook the formal analysis, validation, W. Han, Y. Onoda, K. Fleischer, J.P. Wright, G. Guerin, F. de Vries, D.D. Baldocchi, J. visualization, investigation and project administration. B.R., C.R., C.W., F.L., F.S., I.W., Kattge, B. Blonder, K. Brown, D. Campetella, G. Frechet, Q. Read, N. G. Swenson, V. J.H.C.C., J.K., J.S.J., K.K., M.D.M., M.R., N.R., P.B.R., P.v.B., R.R., S.D., S.D.S., Ü.N. and Lanta, E. Weiher, M. Leishman, A. Siefert, M. Spasojevic, R. Jackson, J. Messier, S. J. W.N.H. wrote the original draft. J.K. and J.S.J. carried out data curation. A.G.M., A.G.G., Wright, D. Craven, J. Molofsky, P. Meir, E. Forey, A. Totte, C. Frenette Dussault, O. Atkin, B.E.L.C., C.B., C.R., C.W., E.S., E.R.W., F.L., G.C., H.J., I.W., J.M.C., J.H.C.C., J.P., K.K., F. Koike, D. Laughlin, S. Burrascano, K. Ollerer, N. Gross, A. Madhur, P. Begonna, B. M.A., M.B., N.K., N.A.S., P.H., P.B.R., P.v.B., S.D., T.H., Ü.N., V.M., V.O., V.D.P. and Bond-Lamberty, B. von Holle, W. Green, B. Yguel, A. C. Malhado, P. Manning, G. Zotz, W.N.H. obtained resources. A.G.M., A.G.G., B.E.L.C., B.R., C.B., C.R., C.W., E.S., E.R.W., E. Lamb, J. Fagundez, Z. Wang, S. Diaz, C. Byun, W. Bond, B. Enquist, C. Baraloto, F.L., F.S., G.C., H.J., I.W., J.M.C., J.S.J., J.H.C.C., J.P., K.K., M.A., M.B., M.C.S., M.D.M., P. Manning, M. Kleyer, W. Ozinga, J. Ordonez, J. Lloyd, H. Poorter, E. Garnier, F. M.R., M.E.S., N.K., N.R., N.A.S., P.H., P.B.R., P.v.B., S.D., S.D.S., T.H., Ü.N., V.M., V.O., Valladares, C. Pladevall, G. Freschet, M. Moretti, H. Kurokawa, V. Minden, A. Demey, V.D.P. and W.N.H. wrote, reviewed and edited the manuscript. B.R., C.W., J.K., M.C.S., F. Férnandez-Méndez, J. Butterfield, T. Domingu, E. Swaine, L. Poorter, S. Shiodera, M.D.M., M.E.S. and N.R. undertook supervision. C.W., J.K., M.R. and M.E.S. were T. Chapin, M. Beckmann, J.A. Gutierrez, M. Mencuccini, S. Jansen, and N. J. B. Kraft. involved in project administration and funding. B.R., J.S.J., S.D.S. and R.R. ran the We appreciate the discussions at the MPI BGC. We thank F. Fazayeli for preparing the software. gap-filled trait data. We thank F. Gans and U. Weber for preparing ancillary data and B. Ahrens for pointing out some soil data availability. We acknowledge Environmental Systems Research Institute (ESRI) and its licensor(s) for the Geodata product of the Competing interests Missions Database ‘ArcWorld Supplement’ (GMI), published by Global Mapping The authors declare no competing interests. International and originated from Global Mapping International for producing Extended Data Fig. 1 and Supplementary Fig. 7 and available in ArcGIS software by ESRI. ArcGIS and ArcMap are the intellectual property of ESRI and are used herein under license. For Additional information Extended data is available for this paper at https://doi.org/10.1038/s41559-021-01616-8. more information about ESRI software, please visit www.esri.com. The authors affiliated with the MPI BGC acknowledge funding by the European Union’s Horizon 2020 Supplementary information The online version contains supplementary material project BACI under grant agreement no. 640176. We are thankful to the data providers available at https://doi.org/10.1038/s41559-021-01616-8. for the SoilGrids, hosted by ISRIC. J.S.J. acknowledges the International Max Planck Correspondence and requests for materials should be addressed to Julia S. Joswig. Research School for global biogeochemical cycles. J.S.J., M.E.S. and M.C.S. acknowledge Peer review information Nature Ecology & Evolution thanks Haydn J.D. Thomas, support from the University of Zurich University Research Priority Program on Global Vanessa Buzzard and the other, anonymous, reviewer(s) for their contribution to the peer Change and Biodiversity. P.B.R., M.E.S. and M.C.S. acknowledge membership in the review of this work. Peer reviewer reports are available. US NSF 20-508 BII-Implementation project, ‘The causes and consequences of plant biodiversity across scales in a rapidly changing world’. M.E.S. acknowledges the NOMIS Reprints and permissions information is available at www.nature.com/reprints. grant of Remotely Sensing Ecological Genomics that funds J.S.J. and M.C.S. C.W. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in acknowledges the support of the Max Planck Society via its fellowship programme. published maps and institutional affiliations. N.R. was funded by a research grant from Deutsche Forschungsgemeinschaft DFG (RU 1536/3-1). K.K. was supported by the project Resilient Forests (KB-29-009-003) of the Open Access This article is licensed under a Creative Commons Dutch Ministry of Economic Affairs. The trait data supplied were co-funded by the Attribution 4.0 International License, which permits use, sharing, adap- EU-FP7-KBBE project: BACCARA—Biodiversity and climate change, a risk analysis tation, distribution and reproduction in any medium or format, as long (project ID 226299). I.W. acknowledges support from the Australian Research Council as you give appropriate credit to the original author(s) and the source, provide a link to (DP170103410). J.P. acknowledges financial support from the European Research the Creative Commons license, and indicate if changes were made. The images or other Council Synergy grant ERC-SyG-2013-610028 IMBALANCE-P. N.A.S. is financed third party material in this article are included in the article’s Creative Commons license, by a VIDI grant (016.161.318) issued by the Netherlands Organization of Scientific unless indicated otherwise in a credit line to the material. If material is not included in Research. The data V.M. provided were funded by II. Oldenburgischer Deichband and the article’s Creative Commons license and your intended use is not permitted by statu- the Wasserverbandstag e.V. (NWS 10/05). We thank M. Kleyer for his critical input. P.H. tory regulation or exceeds the permitted use, you will need to obtain permission directly and V.D.P. have been supported by CNPq (grant nos 369617/2017-2 and 307689/2014-0, from the copyright holder. To view a copy of this license, visit http://creativecommons. respectively). C.B. was supported by the National Research Foundation of Korea (NRF) org/licenses/by/4.0/. grant funded by the Korea government (MSIT) (2018R1C1B6005351). A.G.G. was © The Author(s) 2021 1 Max-Planck-Institute for Biogeochemistry, Jena, Germany. 2Remote Sensing Laboratories, Department of Geography, University of Zurich, Zurich, Switzerland. 3German Centre for Integrative Biodiversity Research (iDiv), Leipzig, Germany. 4Institute of Systematic Botany and Functional Biodiversity, University of Leipzig, Leipzig, Germany. 5Department of Chemistry, University of Zurich, Zurich, Switzerland. 6Escuela de Biología, Universidad Industrial de Santander, Bucaramanga, Colombia. 7Department of Biological Sciences, Macquarie University, Sydney, New South Wales, Australia. 8Institute for Atmospheric and Climate Science, ETH Zurich, Zurich, Switzerland. 9Norwegian Institute of Bioeconomy Research, Oslo, Norway. 10Department of Economics, University of Leipzig, Leipzig, Germany. 11Smithsonian Tropical Research Institute, Ancón, Panama. 12Geoinformatics and Remote Sensing, Institute for Geography, University of Leipzig, Leipzig, Germany. 13Environmental Biology Department, Institute of Environmental Sciences, CML, Leiden University, Leiden, the Netherlands. 14Systems Ecology, Department of Ecological Science, Faculty of Science, Vrije Universiteit Amsterdam, Amsterdam, the Netherlands. 15Instituto Multidisciplinario de Biología Vegetal (IMBIV), CONICET and FCEFyN, Universidad Nacional de Córdoba, Córdoba, Argentina. 16 Global Systems and Analytics, Nova Pioneer, Johannesburg, South Africa. 17Chairgroup Forest Ecology and Forest Management, Wageningen University, Wageningen, the Netherlands. 18Land Life Company, Amsterdam, the Netherlands. 19Research Group Functional Traits, Naturalis Biodiversity Center, Leiden, the Netherlands. 20Plant Sciences, Institute of Biology Leiden, Leiden University, Leiden, the Netherlands. 21Estonian University of Life Sciences, Tartu, Estonia. 22Department of Forest Resources, University of Minnesota, St Paul, MN, USA. 23Hawkesbury Institute for the Environment, Western Sydney University, Penrith, New South Wales, Australia. 24Institute for Global Change Biology and School for Environment and Sustainability, University of Michigan, Ann Arbor, MI, USA. 25Department of Plant Biodiversity, Institute of Ecology and Evolution, Friedrich-Schiller University, Jena, Germany. 26School of Geography, University of Nottingham, Nottingham, UK. 27School of Environmental Sciences, University of Guelph, Guelph, Canada. 28Department of Ecology, University of Innsbruck, Innsbruck, Austria. 29Department of Biological Sciences and Biotechnology, Andong National University, Andong, Korea. 30 Plant Diversity and Ecosystems Management Unit, School of Biosciences and Veterinary Medicine, University of Camerino, Camerino, Italy. 31Department of Biotechnologies and Life Sciences (DBSV), University of Insubria, Varese, Italy. 32Jonah Ventures LLC, Boulder, CO, USA. 33Facultad de Ciencias Naturales y Matemáticas, Universidad del Rosario, Bogotá, Colombia. 34Departamento de Ciencias Ambientales y Recursos Naturales Renovables, Facultad de Ciencias Agronómicas, Universidad de Chile, Santiago, Chile. 35School of Molecular and Life Sciences, Curtin University, Perth, Western Australia, Australia. 36College of Science, Health, Engineering and Education, Murdoch University, Murdoch, Western Australia, Australia. 37Department of Forestry, Universidade do Estado de Santa, Catarina, Lages, Brazil. 38INRAE University Bordeaux, BIOGECO, Cestas, France. 39Department of Ecology and Evolutionary Biology, University of California, Los Angeles, CA, USA. 40Department of Biology, Vrije Universiteit Brussel, Brussels, Belgium. 41Landscape Nature Ecology & Evolution | VOL 6 | January 2022 | 36–50 | www.nature.com/natecolevol 49 Articles NATurE Ecology & EVoluTIon Ecology Group, Institute of Biology and Environmental Sciences, University of Oldenburg, Oldenburg, Germany. 42Department of Ecology and Plant Geography, Moscow State Lomonosov University, Moscow, Russia. 43CSIC, Global Ecology Unit CREAF-CSIC-UAB, Bellaterra, Spain. 44CREAF, Cerdanyola del Vallés, Spain. 45Department of Ecology, Universidade Federal do Rio Grande do Sul, Porto Alegre, Brazil. 46Embrapa Recursos Genéticos e Biotecnologia, Brasília, Brazil. 47Centre for Environmental Sciences, Hasselt University, Diepenbeek, Belgium. 48Institute of Environmental Sciences, Leiden University, Leiden, the Netherlands. 49Department of Biology, University of Wisconsin, Eau Claire, WI, USA. 50Remote Sensing Centre for Earth System Research, University of Leipzig, Leipzig, Germany. 51Helmholtz Centre for Environmental Research, Leipzig, Germany. ✉e-mail: juliajoswigjj@gmail.com 50 Nature Ecology & Evolution | VOL 6 | January 2022 | 36–50 | www.nature.com/natecolevol NATurE Ecology & EVoluTIon Articles Extended Data Fig. 1 | Map of Ecoregion data. Map of ecoregions (30) included in this study (n=220). Nature Ecology & Evolution | www.nature.com/natecolevol nature portfolio | reporting summary Corresponding author(s): Julia Joswig Last updated by author(s): Nov 1, 2021 Reporting Summary Nature Portfolio wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Portfolio policies, see our Editorial Policies and the Editorial Policy Checklist. Statistics For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section. n/a Confirmed The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one- or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section. A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted Give P values as exact values whenever suitable. For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated Our web collection on statistics for biologists contains articles on many of the points above. Software and code Policy information about availability of computer code Data collection Trait data: TRY (www.try-db.org, accession date July 2017, request nb.3282) Climate data: WorldClim is publicly available via https://www.worldclim.org/ (accession date: May 2018), Soil data: SoilGrids (soilgrids.org, accession date: June 2018) is publicly available Ecoregions: of Olson et al. 2001 Shapefiles are publicly available (accession date January 2014, https://www.sciencebase.gov/catalog/ item/508fece8e4b0a1b43c29ca22) Estimate of species richness per ecoregion: Kier species richness (accession date January 2014, https://databasin.org/ datasets/43478f840ac84173979b22631c2ed672/) Spatial data (licenced, see third party rights form) Data analysis R, R-Studio, Microsoft Excel For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Portfolio guidelines for submitting code & software for further information. March 2021 1 Data nature portfolio | reporting summary Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: - Accession codes, unique identifiers, or web links for publicly available datasets - A description of any restrictions on data availability - For clinical datasets or third party data, please ensure that the statement adheres to our policy Plant trait data were accessed from the TRY data base (try-db.org, request number: 3282, accession date: July 2017, see also Figure \ref{fig_extDat}). All TRY data required to reproduce this analysis, and the corresponding R-scripts, are provided in an open TRY File Archive (https://www.try-db.org/TryWeb/Data.php). Climate data WorldClim is publicly available via https://www.worldclim.org/ (accession date: May 2018), Soil data, namely SoilGrids (soilgrids.org, accession date: June 2018) is publicly available. Ecoregion information \cite{Olson2001} shapefiles are publicly available (accession date January 2014, https://www.sciencebase.gov/catalog/ item/508fece8e4b0a1b43c29ca22), The estimate of species richness per ecoregion (Kier et al. 2005) is publicly available (accession date January 2014, https:// databasin.org/datasets/43478f840ac84173979b22631c2ed672/. Data for this study can be accessed on Github (https://github.com/juliajoswig/ Repo\ClimateSoil \TraitSpectrum). For the extended data figure 1, Supplementary Fig.7 , the Geodata product of the Missions Database “ArcWorld Supplement” (GMI) was used, published by Global Mapping International and originated from Global Mapping International for producing and available in ArcGIS software by Esri. ArcGIS and ArcMapTM are the intellectual property of Esri and are used herein under license. Copyright Esri. All rights reserved. For more information about Esri software, please visit www.esri.com. Field-specific reporting Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection. Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf Ecological, evolutionary & environmental sciences study design All studies must disclose on these points even when the disclosure is negative. Study description The overarching objective of this study is to test whether the major dimensions underpinning the global spectrum of plant variation, such as the size and leaf economics spectra, are the result of joint and/or independent variation of climate and soil. For this study, we compiled and analyze a dataset of 17 functional traits for 225,206 geo-referenced observations comprising records of 20,655 global unique species (36,197 unique species to ecoregion combinations) covering the main worlds’ ecoregions – environmentally homogeneous areas characterized by distinct plant assemblages (Methods, Extended Data Figure 1). The trait data were complemented with 21 climate variables (Supplementary Tab. 1) as well as 107 soil variables (Supplementary Tab. 2). Trait- environment relationships were analyzed on the basis of ecoregions using a combination of a regression technique and hierarchical partitioning (for details see Methods). Research sample We extracted data of 17 plant functional traits from the TRY data base1 (Supplementary Tab. 5, Supplementary Tab. 6, www.try- db.org, accession date July 2017, request nb.3282) including published literature (see manuscript). To be able to include the maximum number of species in our analyses, we used the gap-filled version of TRY, with missing data were imputed using a Bayesian Hierarchical Probabilistic Matrix Factorization (BHPMF) algorithm, but observed values were kept. We excluded observations that were not geo-referenced. In total, we included 225,206 observations from 20,655 global unique species (36,197 unique species to ecoregion combinations). The data were stratified by ecoregions (see below), aggregated to species median values per ecoregion and log-transformed. To represent climate conditions we used 21 variables derived from the WorldClim reanalysis product at a resolution of 1 km (accession date May 2018). We extracted values for temperature (annual average, diurnal range, max of warmest month, min of coldest month, sd), precipitation (annual average, min, max, sd), vapour pressure (annual average, min, max, sd), solar radiation (annual average, min, max, sd) and wind (annual average, min, max, sd; see Supplementary Tab. 1). To characterise soil conditions we used 107 variables derived from the ISRIC data product ‘SoilGrids’ (soilgrids.org through ISRIC - WDC Soils, Supplementary Tab. 2). ‘SoilGrids’ provides global predictions of soil characteristics for seven depths, i.e. 0, 5, 15, 30, 60, 100, 200 cm at a resolution of 1 km. In total, we included 128 environmental variables. Before performing the ridge regression, climate and soil variables were reduced to 20 variables each by means of a principle component analysis (PCA). March 2021 Sampling strategy The data were stratified by ecoregions, aggregated to species median values per ecoregion and log-transformed. To determine trait- environment relationships, we aggregated trait as well as environmental data to ecoregions (Supplementary Tab.7) Ecoregions are environmentally homogeneous areas characterized by distinct plant assemblages. At global scale Olson et al. defined 867 ecoregions. For each of the 867 ecoregions, we calculated the median of all species median trait values. For further analyses we only included regions with >20 species and a representation of >1% of the estimated species richness of the ecoregion. To aggregate environmental variables to ecoregion level, we associated each geo-referenced trait observation with its corresponding values of climate and soil variables. Then, we averaged over all values within one ecoregion. Thus, the selected environmental 2 variables represent averages that are weighted by the number and location of trait observations within ecoregions. nature portfolio | reporting summary Data collection Trait data was collected in the framework of TRY (Kattge et al. 2011). Timing and spatial scale Trait-data: geo-referenced global point data Climate- data: gridded data (1km resolution) Soil: gridded data of up to 7 different depths (1km resolution) Data exclusions For trait observations, we only included geo-referenced data. For environmental data, we only included the grid-level information if a trait data point was located there. For ecoregions we only included regions with >20 species and a representation of >1% of the estimated species richness of the ecoregion. Reproducibility All attempts to repeat the results were successful. Per default, the ridge regression was repeated 50 times for each trait. The mean, minimum and maximum of r2 and independent effect give an estimate of model spreads (Table 1, Supplementary Tab. 9). All TRY data required to reproduce this analysis, and the corresponding R-scripts, are provided in an open TRY File Archive (https:// www.try-db.org/TryWeb/Data.php). Randomization To estimate how much of the ecoregion trait variation (r2) is explained by random environmental variables, we performed the ridge regression followed by hierarchical partitioning (see methods „ridge regression“ and “hierarchical partitioning”) and paired climate or soil with noise. The noise data set comprised randomly sampled values for a variable set, as large as the soil variable set (n=107). We performed ridge regression analysis with noise data, together with climate or soil. Then we calculated the independent effect of noise from climate or soil data. The independent effect of randomized data (noise) is never above 0 (negative values due to large difference between r2_total and r2_noise and model variabiliy; Supplement Fig 2). Moreover, we sampled for each ecoregion the minimum requirement for selection (Kier species richness of 1% and 20 species) and erased randomly (n=3) the rest of the species observations. See Supplement Fig. 2. Blinding Describe the extent of blinding used during data acquisition and analysis. If blinding was not possible, describe why OR explain why blinding was not relevant to your study. Did the study involve field work? Yes No Reporting for specific materials, systems and methods We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material, system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response. Materials & experimental systems Methods n/a Involved in the study n/a Involved in the study Antibodies ChIP-seq Eukaryotic cell lines Flow cytometry Palaeontology and archaeology MRI-based neuroimaging Animals and other organisms Human research participants Clinical data Dual use research of concern March 2021 3