ARTICLE https://doi.org/10.1038/s41467-022-30037-9 OPEN Limits to reproduction and seed size-number trade- offs that shape forest dominance and future recovery The relationships that control seed production in trees are fundamental to understanding the evolution of forest species and their capacity to recover from increasing losses to drought, 1234567890():,; fire, and harvest. A synthesis of fecundity data from 714 species worldwide allowed us to examine hypotheses that are central to quantifying reproduction, a foundation for assessing fitness in forest trees. Four major findings emerged. First, seed production is not constrained by a strict trade-off between seed size and numbers. Instead, seed numbers vary over ten orders of magnitude, with species that invest in large seeds producing more seeds than expected from the 1:1 trade-off. Second, gymnosperms have lower seed production than angiosperms, potentially due to their extra investments in protective woody cones. Third, nutrient-demanding species, indicated by high foliar phosphorus concentrations, have low seed production. Finally, sensitivity of individual species to soil fertility varies widely, limiting the response of community seed production to fertility gradients. In combination, these findings can inform models of forest response that need to incorporate reproductive potential. A full list of authors and their affiliations appears at the end of the paper. NATURE COMMUNICATIONS | (2022)13:2381 | https://doi.org/10.1038/s41467-022-30037-9 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30037-9 T he emergence of extreme disturbance as a global change 714 species. Results show that species differences are not bound phenomenon1,2 highlights the need to understand how tree by a strict size-numbers trade-off, instead ranging over ten orders fecundity influences forest regeneration. Drought-induced of magnitude and mediated by foliar nutrients and soil fertility, forest diebacks are now observed in regions where they were rare with phylogeny accounting for some of the broad variation in in the past3. Forest stands shaped by non-destructive surface fires fecundity. are experiencing catastrophic crown fires, with post-burn seed A size-numbers trade-off could limit variation between species production limited to survivors in unburned fragments (Fig. 1a). through allocation constraints associated with producing big Species capable of vegetative regrowth will contribute to seeds versus many of them5,19,20. Because seed production reforestation4, but colonization and seedling success in many depends on tree size21, we standardize seed production per m2 of landscapes will depend on numbers and sizes of seeds from tree basal area. We define individual standardized production surviving trees. Large seeds that are well-provisioned for seedling (ISP) as (mass per seed) × (mean seeds per tree basal area) or survival are expensive to produce, apparent to seed predators, and ISPis ¼ ms ´ F is ð1Þ dependent on animals for dispersal. If the costs of producing larger seeds are offset by producing fewer of them5,6, then for tree i of species s with standardized seed production Fis = fis/bi reproduction could be capped by a size-numbers trade-off. If this for a tree with basal area bi. The estimate of seed number fis cap depends on resources, then landscape fertility gradients and incorporates year-to-year variation in seed production. Species differing nutrient requirements of species play important roles. seed production (SSP) is the expectation of ISP over trees of a While species differences in seed size are readily observed and the given species (SSPs = Ei(ISPis) = ms × Ei(Fis), Methods). If the subject of a large literature7–9, seed numbers are not. At least benefits of large seeds for attracting mutualist dispersers and since Darwin pondered the rapid ascendance of angiosperms in increased seedling survival22 is offset by the cost of producing the Cretaceous10,11, seed production has been recognized as a fewer of them, then SSP does not vary with seed size. But it does fundamental component of fitness12–15 that lacks systematic vary if producing larger seeds increases or decreases the net quantification for trees, the planet’s dominant life form16–18. reproductive output summarized as ISP. This departure from the Here, we provide a global quantification of tree reproduction 1:1 trade-off is expressed on the proportionate (log10 ) scale as from more than 12.1 million tree-years of observations on SSPs / mβs log10 SSPs ¼ α þ βlog10 ms ð2Þ log10 F^ s ¼ α þ ðβ  1Þlog10 ms where F^ s ¼ Ei ðF is Þ is the expected seeds per basal area for species s, α is the log10 F for a species having seed size equal to 1 g, and β is the proportionate change in F^ s for a proportionate increase in seed size. [Subscripts are hereafter omitted to reduce clutter.] The size-numbers trade-off predicts that β = 0 or, equivalently, a slope of − 1 for log10ms in Eq. (2). A value of β > 0 means that numbers do not decline with seed size as steeply as predicted by the trade- off, i.e., a net (proportionate) gain in reproductive effort of β − 1. Values of (α, β) are unknown, as attempts to quantify species differences in SSP have been limited geographically and to few species20,23 due to missing information on numbers of seeds produced. Because large seeds may confer a competition benefit to seedlings within a crowded understory, while small seeds that are wind-dispersed can promote colonization of distant sites14,24, we further evaluated the potential for a competition-colonization trade-off, examining the hypothesized link between competitive ability and wood density25,26. Species differences in seed size versus numbers could be affected by the costs of auxiliary reproductive structures27. Total reproductive effort includes seeds and flowers supplied with nectar and nutrient-rich fleshy fruits that attract dispersers (e.g., Rosales, Ericales, Cornales, Aquifoliales, Magnoliids, and Lami- Fig. 1 Seed production quantifies forest regeneration potential. nales). It includes cones defended with wood, armaments, and Regeneration of forests devastated by multi-year drought and fire depend on prodigious resin flows (e.g., Pinales) (Fig. 1b). Comprehensive a vastly diminished seed supply. a Seed production is limited to unburned mass and nutrient data for cones, fleshy fruits, nuts, capsules, and landscape fragments in the Sierra Nevada mixed conifer zone following samaras are absent from trait data bases like TRY28, nor is there a 2020 burns at a Masting Inference and Forecasting network (MASTIF) and transparent means for comparing reproductive costs across National Ecological Observatory Network (NEON) site (Shaver Lake, CA). diverse structures and provisioning. For these reasons, species b Total reproduction includes not only seeds, but also defenses, including differences in auxiliary structures like fruits and cones are com- wood, spines, and resin flow in conifer cones; examples from the heavily pared here on a qualitative basis. burned Sierra Nevada and Coast ranges include Calocedrus decurrens, Pinus Limits to SSP could depend not only on the cost of auxiliary albicaulis, P. contorta, P. coulteri, P. flexilis, P. lambertiana, P. monophylla, structures associated with seed production, but also on P. monticola, P. ponderosa, P. radiata, P. sabiniana, Pseudotsuga menzesii, resources. Studies of nutrient effects are limited geographically Sequoiadendron giganteum, and Tsuga mertensiana. Mass fractions for seeds and taxonomically, and results are equivocal29–33. Through the to seeds plus cones ranges from 3% for P. radiata, P. contorta, P. coulteri, and assumption that reproductive allocation scales with net primary P. sabiniana to 61% for C. decurrens. The largest cone in b (Pinus lambertiana) production (NPP)34–37, current models assume that resources is 46 cm. Photo credits: James S. Clark and Jordan Luongo. stimulate fecundity, without consideration of responses that 2 NATURE COMMUNICATIONS | (2022)13:2381 | https://doi.org/10.1038/s41467-022-30037-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30037-9 ARTICLE might come from variation within species (here, ISP) versus nutrient-poor peatlands that dominate many important between species (here, SSP). Nutrient additions in orchard ecosystems52. The Masting Inference and Forecasting (MASTIF) practice use ratios intended to balance demands for nitrogen network provides a global synthesis of individual tree-year (N), phosphorus (P), and potassium (K)29,38, the latter being observations and modeling framework that estimates effects on especially important for fruit yield in agriculture, but less fre- fecundity that accounts for dependence in data. Low signal-to- quently reported in ecological studies39. Agricultural experience noise in seed production data requires massive sample sizes to makes it clear that high-N fertilization can decrease reproduc- estimate effects21, represented here by 12.1 M observations on tive yield40,41 due to allocation to vegetative growth. Variation 714 species from all vegetated continents. Valid estimates of between species could come from their differing nutrient fecundity responses to foliar nutrients and CEC accommodate demands, wherein species with high average foliar nutrients dependence in seed observations between trees and seed traps and tend to occupy fertile sites42,43 and, thus, might be expected to within trees and seed traps over time, the quasi-synchronous, produce more seed. Alternatively, species with high nutrient quasi-periodic variation known as masting. Individual tree-year demands might preferentially allocate to growth and defense. observations are modeled jointly in a hierarchical Bayesian state- Variation within species could come from phenotypic plasticity space model detailed in Clark et al.53. as genotype × environment (G × E) interactions. Species bearing large seeds and nutrient-rich fleshy fruits might be especially responsive to fertility variation, but data are lacking. Observed Results positive correlations between foliar nutrients and seed production Rather than being constrained to constant values by a size-numbers in the same individuals over time44,45 does not mean that there is trade-off (hypothesis A), species seed production per basal area a positive association in nutrients and seed production taken over (SSP) spans ten orders of magnitude and increases with seed size trees growing at different fertility levels. Our use of ISP (stan- (β = 0.546 ± 0.042, Fig. 2a). Across all species in the synthesis, the dardized for tree size) controls for responses that come from the expected seed number per m2 of basal area for a seed size of 1 g is fact that fertilized trees grow faster and thus are larger21,46, thus E(10α) = 19,700 ± 2920 (Eq. (2) and Methods section). For a small- helping to isolate fertilization responses47 that are caused by statured species this amounts to 0.62 kg at a diameter of 20 cm. For increased seed production for a given tree size. Taken together, a large species, this is 15 kg for a diameter of 100 cm. For a small- reproductive responses to fertility might be positive or negative, seeded species of, say, 0.001 g (e.g., Betula), this amounts to annual depending on allocation to growth and reproduction. averages of > 50, 000 seeds at 40 cm and > 300,000 seeds at 100 cm. As mentioned above in the context of current model If SSP is regulated by a proportionate sacrifice in seed size as the assumptions, the implications of fecundity for landscape cost of producing more of them, then SSP would not vary with seed recruitment has to combine ISP with the turnover of species size (dashed line in Fig. 2a). Equivalently, there would be a pro- across environmental gradients, here termed community seed portionate decline in seed numbers with seed size (slope = − 1, production (CSP) for the amount of seed produced per area of dashed line in Supplementary Fig. 2). Instead, seed numbers forest or woodland. Trees of the same species might produce decline with seed size at less than half the proportionate rate for a more or less seeds on the richest sites, depending on partitioning slope of β − 1 = − 0.454 ± 0.042 (Supplementary Fig. 2). This between reproduction and growth15. Species turnover influences departure from expectation comes from variation in SSP that is CSP if the resource-conservative species that dominate nutrient- related to species traits and partly captured by phylogeny. poor sites42,43,48 differ in their ISP from those on fertile sites. To Species differences in SSP include a phylogenetic component be clear, we do not obtain CSP as an extrapolation of counts in (Fig. 2b, Pagel’sλ = 0.60, p < 10−9, n = 482). Group seed pro- seed traps. That practice would generate huge error due to the duction (GSP, methods) was used to compare fecundity between precise locations of seed traps, especially their siting directly different taxonomic groups, with higher average values for below an individual that produces large seeds that almost never angiosperms than gymnosperms (Supplementary Fig. 3a). The enter seed traps were they not directly below the tree. Reliable possibility that auxiliary costs weigh against SSP (hypothesis B) is estimates of seed production per ha requires summing the pro- consistent with low SSP for cone-bearing Pinales (brown labels in duction over all trees in the plot divided by the area of the plot Fig. 2b). In angiosperms, the Fagales have uniformly high SSP, (see Methods); this is the method used here for CSP. despite seeds ranging from small Betula (1–6 × 10−4 g) to large This synthesis addresses four basic questions, framed as Quercus (0.7–7 g). Cecropia, Miconia and the Fabales have hypotheses: (A) a size-numbers trade-off constrains species dif- notably low SSP. Angiosperms with fleshy fruits (green labels in ferences in SSP; (B) high auxiliary costs related to fleshy fruits and Fig. 2b) do not exhibit lower SSP than other angiosperms on the cones (Fig. 1b) and high nutrient requirements result in high or whole (Supplementary Figure 3b). Some groups are pre- low SSP, depending on allocation trade-offs involving vegetative dominantly high (e.g., Ericales/Cornales), low (e.g., Vismia, growth and defense; (C) ISP increases with soil fertility or not, Cordia, and Campanulids) or mixed (e.g., Rosales and again depending on species-specific resource allocation trade-offs; Magnoliids). and (D) CSP increases or decreases with soil fertility due to the At the species level, SSP declines with increasing mean foliar P combined effects of within-species sensitivity (hypothesis C) and after controlling for phylogeny and leaf habits (hypothesis B). The species turnover across gradients. Confirmation or rejection of surface in Fig. 3 decreases four orders of magnitude from > 10, these hypotheses is central to understanding the evolutionary 000 at the lowest foliar P levels to < 10 g per m2 basal area at the selection pressures that have shaped the dominant forest trees of highest foliar P. By contrast, foliar N has no discernible effects on the world and is likely key to mitigating their future losses and SSP. Broadleaf species span the widest range of concentrations recovery with global change. and contribute most to the declines in SSP with increasing P. To test these hypotheses, we synthesize tree fecundity, foliar Wood density did not explain SSP variation (Supplementary nutrients, and cation exchange capacity (CEC) (Supplementary Table 5). Fig. 1)49, a widely used fertility indicator related to soil capacity to Within species (hypothesis C), phenotypic plasticity in ISP to bind exchangeable ions (e.g., calcium, magnesium, sodium, soil fertility, hereafter βcec, varies widely between species, but with potassium, ammonium, and iron) that are essential to plant phylogenetic coherence (Fig. 4). In the well-represented Fagales, function50,51. CEC has an advantage over soil carbon or nitrogen fertility responses in Castanea and the white and cerris oaks that are high not only on fertile sites, but also in saturated, (sections Quercus and Cerris) are negative, while red oaks (section NATURE COMMUNICATIONS | (2022)13:2381 | https://doi.org/10.1038/s41467-022-30037-9 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30037-9 Fig. 2 seed size-number trade-offs and species seed production. a Species seed production (SSP, g seed per m2 tree basal area) is not constrained by the strict size-number trade-off (dashed line with a slope of zero). Instead, it varies over ten orders of magnitude and has a positive correlation with seed mass across 714 tree species (log10 SSP ¼ 4:29 þ 0:546  log10 m, R2 = 0.189, p < 10−15, n = 714). b SSP exhibits phylogenetic coherence for 482 species having phylogeny data (68% of species). Brown and green text highlight species that produce coniferous cones and fleshy fruits, respectively. The phylogenetic signal is estimated using Pagel's λ = 0.60 (p < 10−9, n = 482). Lobatae) are positive. This divergence coincides with the ten- Community seed productivity (CSP) does not respond to soil dency for many red oaks (Lobatae) to occupy more fertile soils54. fertility (hypothesis D), due to the neutralizing effects of diverging Increasing CEC reduces ISP in the angiosperm groups Fabales, responses within species that occupy a range of CEC levels Magnoliids, Acer, and Fraxinus, while stimulating ISP in Ericales/ (Supplementary Fig. 4) and to the turnover of species across Cornales, and having mixed effects in Rosales and Betulaceae. gradients. The differing sensitivity to CEC within species com- Gymnosperm responses are primarily decreasing for Tsuga, Picea, bined with the species turnover results in seed production that is Cedrus, and soft pines (section Haploxylon), and mixed for hard as high on low-fertility as on high-fertility sites. pines (section Diploxylon) and Abies. Despite species variation, the reconstructed ancestral lineage for gymnosperms is negative Discussion (purple ancestral branches marked by red dashed line in Fig. 4). Species seed production (SSP), measured as seed size × seed By contrast, there is no angiosperm-wide tendency to produce number per basal area, varies over ten orders of magnitude fewer seeds on high-fertility sites. Foliar nutrients do not explain (Fig. 2), far from the equity expected from a strict size-numbers these species differences in response to fertility gradients. trade-off. Ecologists have long recognized a potential trade-off 4 NATURE COMMUNICATIONS | (2022)13:2381 | https://doi.org/10.1038/s41467-022-30037-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30037-9 ARTICLE investment in cones that can exceed seed mass by 30-fold in Fig. 1b. But the generally low SSP in gymnosperms does not imply a fitness disadvantage for large cones. Well-defended seeds certainly arose under selection pressure for the advantages they confer against predation58. The divergent evolutionary options available to angiosperms and gymnosperms may have con- strained reproductive potential for these two groups in different ways; there is no evidence that angiosperms that produce expensive, fleshy-fruits realize lower SSP than other angiosperms (Fig. 2b and Supplementary Fig. 3b). Phylogenetic differences (Fig. 2b and Supplementary Fig. 3) bear on Darwin’s ‘abominable mystery’10, the advantages that might explain the explosion of angiosperms in the Cretaceous. The rise of angiosperms has been attributed to rates of growth59, Fig. 3 Effects of foliar nutrients on seed production. Effects of foliar photosynthesis60, and diversification61, all of which tend to be nitrogen (N) and phosphorus (P) on SSP (g seed per m2 tree basal area) higher than in gymnosperms. Capacities to harbor nutrients and from the model in Supplementary Table 1 plotted for the broadleaf generate positive feed-backs with soil fertility through litter deciduous leaf habit (other leaf types exhibit same patterns). The convex production62 and to reproduce vegetatively63 could contribute to hull for the surface is restricted to the data coverage. Symbols indicate leaf angiosperm success. Aridification at the end of the Eocene may habit, including broadleaf deciduous (BD), broadleaf evergreen (BE), have contributed to gymnosperm extinctions11. While all of these needleleaf evergreen (NE), and scalelike evergreen (SE). differences are plausible contributions to angiosperm success64, the fact that angiosperms dominate gymnosperms in SSP represents a between producing few large seeds, each provisioned to attract more direct connection to fitness than most of these mechanisms. dispersers and/or promote high seedling survival, versus many The negative effect of fertility on seed production emerges in the small seeds that might offset low survival with expanded oppor- low SSP for P-demanding species (Fig. 3). Global change studies tunities to reach suitable sites5,6,20. Evaluating whether or not this have steadily improved our understanding of fertility effects on tree compromise between size and numbers limits species variation in growth, foliar nutrients, and primary productivity65–67, while tree SSP is made possible by a large representation of 12.1 M tree- fecundity has remained inconclusive32,33. This synthesis clarifies years from 714 species and an analytical framework that admits three elements of the fertility-fecundity relationship. The species- tree attributes (size and competition), environmental variation, level association between high foliar P and low SSP (Fig. 3) is and dependence between trees and within a tree across years53. consistent with a species-level trade-off between growth and Results show that the increased seed-mass production realized by reproduction15,16. P-rich ribosomal RNA required for rapid cell species that invest in large seeds is mediated by resource access division68,69 can stimulate growth at the expense of reproduction. and the auxiliary costs of woody cones. Moreover, seed production is expensive in resource-acquisitive The expected annual productivity of 19,700 ± 2920 seeds per m2 species23 if nutrient concentrations in seeds increase with foliar basal area at a seed size of 1 g provides a benchmark for basic nutrients70. The decline in SSP with foliar P is consistent with demographic and life-history understanding that has thus far been horticultural evidence that fertilization can increase foliar nutrients absent for trees. For a moderate size tree (e.g., 30 cm diameter) and stimulate vegetative growth at the expense of both quantity and that produces seeds of a moderate size (e.g., 10−2 g) this amounts quality of crop yield41,71,72. However, a plant’s response to nutrient to annual mean production of > 10,000 seeds. The magnitude of addition45,73 need not agree with species differences in nutrient this estimate puts this life form far outside the range of most demand. Potassium (K) promotes reproductive yield in crops74,75, species, including most herbaceous plants8. This massive output but data for trees are limited. Second, within-species responses to certainly contributes to a foundational role of masting in many CEC (i.e., βcec) are phylogenetically conserved despite variation in forest food webs55. It is expected to drive inordinate selection magnitude and sign across species. The negative effect of CEC on pressures required to offset seed input with mortality rates that seed production in gymnosperms (purple shading that dominates would be needed for quasi-stable population densities over time. the ancestry of this group in Fig. 4) is consistent with conifer There is a size-numbers trade-off, but it is not symmetric: dominance in infertile environments62,76 and their resistance to species that invest in large seeds produce more seeds than angiosperm expansion on poor soils77,78. Finally, the fertility gra- expected from a strict 1:1 trade-off. The shallow slope between dients that contribute to trends in net primary productivity (NPP) seed number and size (Supplementary Fig. 2), roughly half that do not translate to trends in fecundity due to the combined effects expected for a strict trade-off56,57, suggests a cost savings as of within-species responses (Fig. 4) and the turnover of species reproductive investment is packaged into a smaller number of across these gradients (Supplementary Fig. 4). larger seeds. If the cost of reproduction increases as β = 0.55, or Seed size versus numbers is among the most fundamental roughly the square root of seed mass (Eq. (2)), then the benefit for constraints on population persistence in competitive stands79–81, large seeds increases according to the slope in Fig. 2a. A fixed cost on landscapes subject to large-scale disturbance82,83, and where associated with the production of each seed would be consistent colonization requires long-distance dispersal84,85. Just as repro- with this shallow slope; the diversity of structures used to support ductive constraints may have contributed to global transforma- reproduction suggests multiple contributing factors. tions in earth history, they will undoubtedly shape the forests that Of course, the slope of the size-numbers relationship between emerge from expanding diebacks, disturbance, and human species does not represent strength of selection for size and exploitation (Fig. 1a). Species and forest types are threatened by numbers within species, which depends on a broad range of traits regeneration failure following climate change, timber harvesting, summarized, in part, by phylogeny. The advantages of producing and other disturbances. Globally, less than 7% of forests are larger/fewer acorns in Quercus cannot be equated with the planted by humans, but, over the last decade, naturally regener- advantages of producing larger/fewer fleshy fruits or cones, due to ated forests have decreased by 8 million ha per year86. The the differing costs and benefits associated with each of these expanding scale of forest losses might increase reliance on arti- structures. Low SSP in gymnosperms is consistent with the ficial regeneration86,87, with attendant costs to forest diversity. NATURE COMMUNICATIONS | (2022)13:2381 | https://doi.org/10.1038/s41467-022-30037-9 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30037-9 Fig. 4 Effects of soil fertility on seed production. Sensitivity of individual standardized production (ISP) to soil fertility based on within-species response to cation exchange capacity (βcec). Text color follows Fig. 2. Red dashed line indicates the ancestry of gymnosperms. Percentages of species that respond negatively to CEC are labeled for species groups. The analysis includes 141 species that span a sufficient CEC range in the Masting Inference and Forecasting (MASTIF) network to estimate a robust effect. The phylogenetic signal, estimated for 129 species (91% of species) having phylogeny data, is highly significant (Pagel's λ = 0.87, p < 0.001, n = 129). Understanding future forests requires integration of seed pro- production (Fig. 4). The global relationships quantified from this duction estimates here with drought- and fire-tolerance88,89, synthesis bring not only a previously unmeasured dimension of ability to resprout following dieback90,91, and changes in forest forest response; they also will allow us to leverage existing structure and composition92. While tree growth and survival have knowledge of growth and survival with the missing link to long been quantifiable using traditional methods, they cannot be regeneration, that of tree fecundity. used to anticipate reforestation potential without knowledge of fecundity. Seed limitation takes on new urgency, as highlighted by fires of uncommon severity in fuels cured by multi-year droughts Methods Fecundity data. Fecundity data from the Masting Inference and Forecasting that leave only remnants of reproductive trees. The relative (MASTIF) network53,96 include two types of raw data, seed traps (ST) and crop benefits of size versus numbers of seeds for seedling success in counts (CC). Each observation references a tree-year (Fig. 5). Data include natural regeneration will shift under novel combinations of longitudinal (repeated) observations on individual trees (99%) and opportunistic temperature and moisture93,94. The positive effects of nutrient observations that come through the iNaturalist project53. ST data consist of numbers of seeds that accumulate annually in mapped seed traps on forest availability on reproduction assumed in current models36 are inventory plots. A fitted dispersal kernel53 relates seed counts to mapped trees, sensible for tree growth and NPP65,95, but not for seed accounting for uncertainty in seed transport and Poisson seed counts. Fecundity 6 NATURE COMMUNICATIONS | (2022)13:2381 | https://doi.org/10.1038/s41467-022-30037-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30037-9 ARTICLE Species traits. Seed dry mass data were collected by our laboratories and sup- plemented with the TRY Plant Trait Database28. We could not incorporate within- species variation in seed size due to the lack of coverage for this variable across the network. There is some speculation that large seeds are associated with a capacity to generate large crowns in Pinus pinea104 and Fagus sylvatica105. Mean values per species came from merging our measurements with values in the TRY database. Genus- or family-level means were used where seed mass was missing at the species level. To insure that results are not affected by using genus- and family-level mean, we show the same patterns using only species with species-level seed mass (Sup- plementary Figs. 5 and 6). Foliar nitrogen and phosphorus concentration, leaf habit, fruit type, and wood density were obtained from the primary literature, TRY, and the BIEN dataset106. Fecundity modeling and inference. We estimate tree fecundity with a hierarchical Bayes state-space model53 that incorporates effects of tree attributes with envir- onment to infer effects on maturation and conditional fecundity (Fig. 5). The methodology detailed in53 builds on models that have provided reliable estimates in many forest types96,107–115. The analysis in53 goes further in showing that the MASTIF model recovers the parameters controlling maturation and fecundity in simulated data, and it accurately predicts the seed-trap data that are used to esti- Fig. 5 The MASTIF model, summarized from Clark et al.53, includes three mate parameters. To allow for uncertain identification of seeds from trees of the levels, observed responses (above), process, and parameters (part of the same genus and for dependence within trees over time and between trees, all tree- posterior distribution, middle), and predictors (below). The data model years of a genus are modeled jointly53,96. For each tree i of species s at stand j in for observed responses includes uncertainty that comes from seed year t, the expected fecundity is the product of maturation probability and con- ditional fecundity, dispersal (seed traps), the fraction of the crop that can be observed (crop counts), and detection of mature status. The process model describes Eð f ijs;t Þ ¼ ^f ijs;t ¼ ^ρijs;t ψ ^ ijs;t ð4Þ change in maturation status ρi,t and, once mature, conditional fecundity ψi,t. Fitted coefficients for conditional fecundity βψ and maturation probability βρ Once a tree has been observed to produce seed in a given year t, then it is known to describe how predictor variables in red affect maturation and fecundity. be mature in subsequent years t 0 > t. A tree never having been observed to produce seed in the past could be immature or not. Maturation probability ^ρijs;t is the Error in the process model is absorbed by process error variance σ2. probability that an individual is in the mature state, Predictors are held in a design matrix xi,t for conditional fecundity. Diameter   di,t is the predictor for maturation status. Additional subscripts for location j ðρÞ ðρÞ ρijs;t  Bernoulli ρijs;t1 þ ð1  ρijs;t1 Þρijs;tþ1 Φðβs0 þ βs1 dijs;t Þ ð5Þ and species s in the main texts are suppressed to reduce clutter. where Φ( ⋅ ) represents the standard normal cumulative distribution function modeling (see Fecundity modeling and inference) further accommodates (CDF) of the probit model, which is determined by tree diameter dijs,t. Note that ðρÞ ðρÞ 0 uncertain assignment of seeds to species where seeds are recorded at the genus coefficients for intercept and diameter βðρÞ s ¼ ½βs0 ; βs1  influence ρijs,t only at the level53. CC data include counts of reproductive structures with estimates of the transition to the mature state, i.e., [ρijs,t∣ρijs,t−1 = 0, ρijs,t+1 = 1]. Conditional fraction of the crop observed53,96. A beta-binomial distribution accounts for fecundity depends on predictors, individual effects, year effects, and error, uncertainty in the count and crop-fraction estimates53,96. The study includes 12,063,723 tree-years from North America, South and Central America, Europe, ðψÞ ^ ijs;t Þ ¼ x0ijs;t βsðψÞ þ βijs þ γg½ijs;t þ ϵijs;t logðψ ð6Þ Africa, Asia, and Oceania (Supplementary Figure 1) from 751 species. Mature trees for 714 species (95% of 751 species) were used in this analysis (see where xijs,t is a design matrix holding individual attributes and environmental Fecundity modeling and inference). Data coverage by sites is included in conditions (see Individual, habitat and climate covariates), and β(ψ, x) are fixed- table S2, with sample sizes (tree-years) summarized by species provided in the ðψÞ file supplementSampleSize.csv. effects coefficients. βijs is the random effect for tree i of species s at stand j. γg[ij]s,t are year effects that are random across groups g and fixed for year t to account for interannual variation that is not fully captured by climate anomalies. Group Individual, habitat, and climate covariates. Covariates were selected based on membership for year effects (g[ij]s) is defined by the species-ecoRegion their ability to explain important variation in the data. In order to test hypotheses combination53; quasi-synchronicity can vary between species and ecoRegions. related to soil fertility we need covariates that can account for the other sources of There is a noise term ϵijs,t. The primary components are summarized in Fig. 5. variation in data that might obscur hypothesized relationships. Covariates included Variable selection is based on model fit using the Deviance information criterion individual attributes for each tree (diameter and shade classes), climate (temperature (DIC) and data coverage, which depends on the range of variation in each pre- and moisture deficit), and habitat conditions (soils and terrain attributes) in Sup- dictor and species. Model implementation is open-access with R package MASTIF, plementary Table 3. The quadratic term in diameter allows for changes in fecundity with full algorithm details provided in53. response with tree size21. Shade classes follow the FIA/NEON classes from 1 (fully- exposed in open space) to 5 (fully-shaded in the understory). Shading quantifies the effects of competition on fecundity14. Cation exchange capacity, a widely used Individual, species, group, and community seed production. Individual stan- indicator for soil fertility97 at 250-m spatial resolution, was evaluated as a weighted dardized production (ISP) is expressed as relative to basal area standardized for tree mean based on uncertainty layers from three soil depths (0–5 cm, 5–15 cm, and size21; it is the product of mass per seed ms times seeds per tree divided by tree 15–30 cm) in a global dataset49 (Supplementary Fig. 1). Slope and aspect quantify basal area, having units g m−2yr−1. Each tree-year has an associated posterior effects on solar radiation and drainage on moisture condition. Slope and aspect were mean estimate ^f ijs;t . Calculation of ISP combines posterior mean estimates with obtained from the NASA Shuttle Radar Topography Mission’s (SRTM)98 digital their uncertainties, as an expectation over years: elevation model (DEM) with a spatial resolution of 30 m. The DEM grid was supplemented with the USGS National Elevation Dataset for MASTIF sites that ms ∑t wijs;t ^f ijs;t were outside SRTM coverage. Slope and aspect were derived following99 ISPijs ¼ ´ ð7Þ 2 3 2 3 bij ∑t wijs;t u1 sinðslopeÞ 6 7 6 7 where ms is seed mass (g), bij is basal area (m2), and weight wijs,t is the inverse of the u ¼ 4 u2 5 ¼ 4 sinðslopeÞ sinðaspectÞ 5 ð3Þ coefficient of variation (CV) for the estimate, u3 sinðslopeÞ cosðaspectÞ wijs;t ¼ CV 1 ^ ijs;t ¼ f ijs;t =sijs;t ð8Þ Temperature T and moisture deficit M are included in the model as site means and anomalies to account for the effects of both site variation and inter-annual variability100. Moisture deficit M summarizes water availability and is obtained and sijs,t is the standard error of the estimate. The CV−1 is used instead of the from monthly M = PET − P from January to December101. T and M were inverse of variance, because the mean tends to scale with variance. Low values for extracted from Terraclimate102, which provides monthly but spatially coarse data, ^f and CHELSA103, which is 1 km spatial resolution but is not available after the year ijs;t are noisy and less important than high values, which are emphasized by the 2016. We integrated the two gridded products with local climate monitoring by CV. Estimates of seed number ^f , its standard error sijs,t, and CV are all ijs;t first using regression to project CHELSA forward based on Terraclimate, with final dimensionless. All estimates are time averages across annual estimates, so we calibration to sites having local weather data. hereafter omit yr−1 from dimensions. Therefore, ISP has the units of g/m2. NATURE COMMUNICATIONS | (2022)13:2381 | https://doi.org/10.1038/s41467-022-30037-9 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30037-9 Species seed production (SSP) is the expectation taken over individual tree ISP Reporting summary. Further information on research design is available in the Nature values for a species s, Research Reporting Summary linked to this article. ∑ij wijs ISPijs SSPs ¼ ð9Þ Data availability ∑ij wijs Seed production data are available at the Duke Data Repository https://doi.org/10.7924/ where wijs is defined as the inverse CVijs for individual i. Analyses of SSP are done r4348ph5t122. Species traits are downloaded from TRY Plant Trait database at https:// on the proportionate (log) scale to avoid dominance of results by species having www.try-db.org/TryWeb/Home.php28. Cation exchange capacity data are obtained at highest seed production. Visualizations are based on log10 to facilitate https://soilgrids.org/49. Climate data are extracted from Terraclimate at http://www. interpretation of scale. climatologylab.org/102 and CHELSA at https://chelsa-climate.org/103. Elevation data are Group seed production (GSP) was evaluated to determine differences between obtained from SRTM at https://srtm.csi.cgiar.org/ and USGS National Elevation Dataset angiosperms versus gymnosperms and between the cone-producing Pinales, large- at https://ned.usgs.gov/. seeded Fagales, fleshy-fruited members of all groups, and remaining species. For comparisons of species groups k, we evaluated means and uncertainties: Code availability ∑i;j;s2k wijs ISPijs R statistical software v4.0.2 was used in this work. All analyses used published R GSPk ¼ ð10Þ packages, with details stated in the section Methods. MASTIF includes code in R and C++, ∑i;j;s2k wijs which is published on CRAN at https://cran.r-project.org/web/packages/mastif/index.html. with standard error, 1 rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi Received: 1 October 2021; Accepted: 13 April 2022; SEðGSPk Þ ¼ ∑ s2 ð11Þ nk i;j;s2k;t ijs;t where nk is the number of observations included in the estimate. The comparisons between the groups is shown in Supplementary Fig. 3. To understand the collective impacts of soil fertility on seed production, we calculated community seed production (CSP) by extending tree reproduction to the References stand level by summing fecundity over all trees and species and dividing by plot 1. Curtis, P. G., Slay, C. M., Harris, N. L., Tyukavina, A. & Hansen, M. C. area: Classifying drivers of global forest loss. Science 361, 1108–1111 (2018). 1 2. Duane, A., Castellnou, M. & Brotons, L. Towards a comprehensive look at CSPj ¼ ∑ ISPijs ð12Þ global drivers of novel extreme wildfire events. Clim. Change 165, 43 (2021). Aj is 3. Allen, C. D. et al. A global overview of drought and heat-induced tree where Aj is plot area (ha) for each plot. Note that we did not standardize ISP by mortality reveals emerging climate change risks for forests. For. Ecol. Manag. basal area to yield instead seeds per area of forest floor. The CSP thus represents 259, 660–684 (2010). Adaptation of Forests and Forest Management to stand production, much like NPP represents stand vegetative production. Because Changing Climate. calculation of CSP requires plot area, analysis on CSP is based on inventory plots. 4. Franklin, J. F., Mitchell, R. J. & Palik, B. J. Natural disturbance and stand development principles for ecological forestry. General Technical Report. NRS-19. Newtown Square, PA: US Department of Agriculture, Forest Service, Phylogeny, foliar nutrients, and soil fertility. Phylogeny was obtained for Northern Research Station. 44. p. 19 (2007). 482 species (67% of total) from116. To determine if SSP can be explained by shared 5. Westoby, M., Jurado, E. & Leishman, M. Comparative evolutionary ecology of ancestry, we estimated the phylogenetic signal using Pagel’sλ117, which ranges from seed size. Trends Ecol. Evol. 7, 368–372 (1992). 0 (independence over species) to 1 (related species completely resemble each 6. Smith, C. C. & Fretwell, S. D. The optimal balance between size and number of other). Continuous character mapping was used to visualize the phylogenetic signal offspring. Am. Nat. 108, 499–506 (1974). in Fig. 2 using R package phytools118. We explored the linkage between foliar 7. Lord, J., Westoby, M. & Leishman, M. Seed size and phylogeny in six nutrients and species difference in seed production using the phylogenetic temperate floras: Constraints, niche conservatism, and adaptation. Am. Nat. regression to account for species relatedness, where the correlation between species 146, 349–364 (1995). increases proportionally with the branch length (i.e., from the root to the most recent common ancestor) in the phylogeny119,120. Phylogenetic regression was 8. Moles, A. T. et al. Global patterns in seed size. Glob. Ecol. Biogeogr. 16, implemented using R package phylolm121 for 427 species that has foliar nutrients 109–116 (2007). data (89% of the 482 species with phylogeny). 9. Tautenhahn, S. et al. On the biogeography of seed mass in germany - distribution We estimated within-species response to CEC, that is, βcec in Eq. (6), for patterns and environmental correlates. Ecography 31, 457–468 (2008). 141 species that has sufficient range of variation across the soil fertility 10. Lidgard, S. & Crane, P. R. Quantitative analyses of the early angiosperm (CEC) gradient; βcec coefficients cannot be estimated for species that are limited to radiation. Nature 331, 344–346 (1988). one or a few sites. We quantified phylogenetic coherence in βcec (Fig. 4) between 11. Crisp, M. D. & Cook, L. G. Cenozoic extinctions account for the low diversity species following the same procedure used for SSP. To determine how the ability to of extant gymnosperms compared with angiosperms. New Phytol. 192, estimate βcec depended on the range in CEC over which it was observed, we plotted 997–1009 (2011). predicted seed production against CEC while holding all other covariates constant 12. Stearns, S. C. Life-history tactics: a review of the ideas. Quart. Rev. Biol. 51, (i.e., half of maximum diameter, medium shade, and mean temperature and 3–47 (1976). moisture in Supplementary Fig. 4). This figure demonstrates the large differences in 13. Grubb, P. J. The maintenance of species-richness in plant communities: the both directions and magnitudes of βcec along the soil fertility gradients where the importance of the regeneration niche. Biol. Rev. 52, 107–145 (1977). species have been observed. 14. Clark, J. S., LaDeau, S. & Ibanez, I. Fecundity of trees and the colonization- Finally, we evaluated the effects of soil fertility on stand productivity (i.e., CSP) competition hypothesis. Ecol. Monogr. 74, 415–442 (2004). using regression. The analysis is restricted to sites having all trees sampled within a 15. Salguero-Gómez, R. et al. Fast-slow continuum and reproductive strategies known area, which is true for all inventory studies with seed-traps. structure plant life-history variation worldwide. Proc. Natl Acad. Sci. USA 113, 230–235 (2016). Calculation of seed number per 1 g. Provided the estimate of α does not have 16. Thomas, S. C. Age-Related Changes in Tree Growth and Functional Biology: large error, the expected seeds per basal area at seed mass = 1 g for the log10-log10 The Role of Reproduction, p. 33-64 (Springer Netherlands, 2011). regression is approximated with two terms from the Taylor expansion 17. Wenk, E. H. & Falster, D. S. Quantifying and understanding reproductive   allocation schedules in plants. Ecol. Evol. 5, 5521–5538 (2015). s2 l 2 18. Bar-On, Y. M., Phillips, R. & Milo, R. The biomass distribution on earth. Proc. Eð10α Þ ¼ 10α^ 1 þ α ± 10α^ sα l ð13Þ Natl Acad. Sci. USA 115, 6506–6511 (2018). 2 19. Turnbull, L. A., Rees, M. & Crawley, M. J. Seed mass and the competition/ where α^ ¼ 4:29 and sα = 0.065 are the estimate and standard error for parameter α, colonization trade-off: a sowing experiment. J. Ecol. 87, 899–912 (1999). and the constant l ¼ logð10Þ. For this model α ^  sα , this approximation is con- 20. Moles, A., Falster, D., Leishman, M. & Westoby, M. Small-seeded species fident at 19, 720 ± 2, 920. For a tree of given size, its seed production is 19, 720 × its produce more seeds per square metre of canopy per year, but not per basal area. individual per lifetime. J. Ecol. 92, 384–396 (2004). 21. Qiu, T. et al. Is there tree senescence? the fecundity evidence. Proc. Natl Acad. Seed size-number trade-offs within orders. We have included taxonomic rank Sci. USA 118, e2106130118 (2021). (i.e., order) as a factor level, including an interaction with seed size, in the 22. Westoby, M., Falster, D. S., Moles, A. T., Vesk, P. A. & Wright, I. J. Plant regression. We found that SSP was not constrained by the strict 1:1 trade-off within ecological strategies: Some leading dimensions of variation between species. each order (Supplementary Table 4). Annu. Rev. Ecol. Syst. 33, 125–159 (2002). 8 NATURE COMMUNICATIONS | (2022)13:2381 | https://doi.org/10.1038/s41467-022-30037-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30037-9 ARTICLE 23. Henery, M. L. & Westoby, M. Seed mass and seed nutrient content as 53. Clark, J. S., Nuñez, C. L. & Tomasek, B. Foodwebs based on unreliable predictors of seed output variation between species. Oikos 92, 479–490 (2001). foundations: spatiotemporal masting merged with consumer movement, 24. Turnbull, L. A., Coomes, D., Hector, A. & Rees, M. Seed mass and the storage, and diet. Ecol. Monogr. 89, e01381 (2019). competition/colonization trade-off: competitive interactions and spatial 54. Burns, R. M. Silvics Of North America (US Department of Agriculture, Forest patterns in a guild of annual plants. J. Ecol. 92, 97–109 (2004). Service, 1990). 25. Chave, J. et al. Towards a worldwide wood economics spectrum. Ecol. Lett. 12, 55. Koenig, W. D. & Knops, J. M. H. Seed-crop size and eruptions of north 351–366 (2009). american boreal seed-eating birds. J. Anim. Ecol. 70, 609–620 (2001). 26. Poorter, L. et al. The importance of wood traits and hydraulic conductance for 56. Greene, D. F. & Johnson, E. A. Estimating the mean annual seed production of the performance and life history strategies of 42 rainforest tree species. New trees. Ecology 75, 642–647 (1994). Phytol. 185, 481–492 (2010). 57. Lord, J. M. & Westoby, M. Accessory costs of seed production and the 27. Hanley, M. E., Cook, B. I. & Fenner, M. Climate variation, reproductive evolution of angiosperms. Evol. Int. J. Org. Evol. 66, 200–210 (2012). frequency and acorn yield in english oaks. J. Plant Ecol. 12, 542–549 (2018). 58. Hulme, P. & Benkman, C. Granivory. vol. 23, 132-154 (Oxford: Blackwell, 2002). 28. Kattge, J. et al. Try plant trait database - enhanced coverage and open access. 59. Bond, W. J. The tortoise and the hare: ecology of angiosperm dominance and Glob. Change Biol. 26, 119–188 (2020). gymnosperm persistence. Biol. J. Linn. Soc. 36, 227–249 (1989). 29. Ran, E., Arnon, D., Alon, B.-G., Amnon, S. & Uri, Y. Flowering and fruit set of 60. Brodribb, T. J. & Feild, T. S. Leaf hydraulic evolution led a surge in leaf olive trees in response to nitrogen, phosphorus, and potassium. J. Am. Soc. photosynthetic capacity during early angiosperm diversification. Ecol. Lett. 13, Hortic. Sci. Am. Soc. Hortic. Sci. 133, 639–647 (2008). 175–183 (2010). 30. Fernández-Martínez, M., Vicca, S., Janssens, I. A., Espelta, J. M. & Peñuelas, J. 61. Davies, T. J. et al. Darwin’s abominable mystery: Insights from a supertree of The role of nutrients, productivity and climate in determining tree fruit the angiosperms. Proc. Natl Acad. Sci. USA 101, 1904–1909 (2004). production in european forests. New Phytol. 213, 669–679 (2017). 62. Berendse, F. & Scheffer, M. The angiosperm radiation revisited, an ecological 31. Fortier, R. & Wright, S. J. Nutrient limitation of plant reproduction in a explanation for darwin’s ‘abominable mystery’. Ecol. Lett. 12, 865–872 (2009). tropical moist forest. Ecology 102, e03469 (2021). 63. Barrett, S. C. H. Influences of clonality on plant sexual reproduction. Proc. 32. Canham, C. D., Ruscoe, W. A., Wright, E. F. & Wilson, D. J. Spatial and Natl Acad. Sci. USA 112, 8859–8866 (2015). temporal variation in tree seed production and dispersal in a new zealand 64. Condamine, F. L., Silvestro, D., Koppelhus, E. B. & Antonelli, A. The rise of temperate rainforest. Ecosphere 5, art49 (2014). angiosperms pushed conifers to decline during global cooling. Proc. Natl Acad. 33. Pérez-Ramos, I. M., Aponte, C., García, L. V., Padilla-Díaz, C. M. & Marañón, Sci. USA 117, 28867–28875 (2020). T. Why is seed production so variable among individuals? a ten-year study 65. Oren, R. et al. Soil fertility limits carbon sequestration by forest ecosystems in with oaks reveals the importance of soil environment. PLoS ONE 9, e115371 a co2-enriched atmosphere. Nature 411, 469–472 (2001). (2014). 66. Reich, P. B. et al. Nitrogen limitation constrains sustainability of ecosystem 34. Sitch, S. et al. Evaluation of ecosystem dynamics, plant geography and response to co2. Nature 440, 922–925 (2006). terrestrial carbon cycling in the LPJ dynamic global vegetation model. Glob. 67. Firn, J. et al. Leaf nutrients, not specific leaf area, are consistent indicators of Change Biol. 9, 161–185 (2003). elevated nutrient inputs. Nat. Ecol. Evol. 3, 400–406 (2019). 35. Krinner, G. et al. A dynamic global vegetation model for studies of the coupled 68. Elser, J. et al. Biological stoichiometry from genes to ecosystems. Ecol. Lett. 3, atmosphere-biosphere system. Glob. Biogeochem. Cycles 19, 1–33 (2005). 540–550 (2000). 36. Fisher, R. A. et al. Vegetation demographics in earth system models: a review 69. Niklas, K. J., Owens, T., Reich, P. B. & Cobb, E. D. Nitrogen/phosphorus leaf of progress and priorities. Glob. Change Biol. 24, 35–54 (2018). stoichiometry and the scaling of plant growth. Ecol. Lett. 8, 636–642 (2005). 37. Hanbury-Brown, A., Ward, R. & Kueppers, L. M. Future forests within earth 70. Kerkhoff, A. J., Fagan, W. F., Elser, J. J. & Enquist, B. J. Phylogenetic and system models: regeneration processes critical to prediction. New Phytol. in growth form variation in the scaling of nitrogen and phosphorus in the seed press https://doi.org/10.1111/nph.18131 (2022). plants. Am. Nat. 168, E103–E122 (2006). 38. Stiles, W. C. & Reid, W. S. Orchard nutrition management. Inf. Bull. (1991). 71. Weinbaum, S. A., Johnson, R. S. & DeJong, T. M. Causes and consequences of https://ecommons.cornell.edu/bitstream/handle/1813/3305/Orchard% overfertilization in orchards. HortTechnology 2, 112b (1992). 20Nutrition%20Management.pdf?sequence=2&isAllowed=y. 72. Fernandez-Escobar, R. et al. Olive oil quality decreases with nitrogen over- 39. Schlesinger, W. H. Some thoughts on the biogeochemical cycling of potassium fertilization. HortScience 41, 215 (2006). in terrestrial ecosystems. Biogeochemistry 154, 427–432 (2021). 73. Han, Q., Kabeya, D., Iio, A. & Kakubari, Y. Masting in fagus crenata and its 40. Neilsen, D. & Neilsen, G. Efficient use of nitrogen and water in high-density influence on the nitrogen content and dry mass of winter buds. Tree Physiol. apple orchards. HortTechnology 12, 19 (2002). 28, 1269–1276 (2008). 41. Rubio Ames, Z., Brecht, J. K. & Olmstead, M. A. Nitrogen fertilization rates in 74. Pettigrew, W. T. Potassium influences on yield and quality production for a subtropical peach orchard: effects on tree vigor and fruit quality. J. Sci. Food maize, wheat, soybean and cotton. Physiol. Plant. 133, 670–681 (2008). Agric. 100, 527–539 (2020). 75. Leeper, A. C., Lawrence, B. A. & LaMontagne, J. M. Plant-available soil 42. Elser, J. J. et al. Growth rate-stoichiometry couplings in diverse biota. Ecol. nutrients have a limited influence on cone production patterns of individual Lett. 6, 936–943 (2003). white spruce trees. Oecologia 194, 101–111 (2020). 43. Seyednasrollah, B. & Clark, J. S. Where resource-acquisitive species are 76. Chapin, F. S., Autumn, K. & Pugnaire, F. Evolution of suites of traits in located: the role of habitat heterogeneity. Geophys. Res. Lett. 47, response to environmental stress. Am. Nat. 142, S78–S92 (1993). e2020GL087626 (2020). 77. Westoby, M. & Wright, I. J. Land-plant ecology on the basis of functional 44. Rosecrance, R. C., Weinbaum, S. A. & Brown, P. H. Alternate bearing affects traits. Trends Ecol. Evol. 21, 261–268 (2006). nitrogen, phosphorus, potassium and starch storage pools in mature pistachio 78. Brodribb, T. J., Pittermann, J. & Coomes, D. A. Elegance versus speed: trees. Ann. Bot. 82, 463–470 (1998). Examining the competition between conifer and angiosperm trees. Int. J. Plant 45. Sala, A., Hopping, K., McIntire, E. J. B., Delzon, S. & Crone, E. E. Masting in Sci. 173, 673–694 (2012). whitebark pine (pinus albicaulis) depletes stored nutrients. New Phytol. 196, 79. Clark, J. S., Macklin, E. & Wood, L. Stages and spatial scales of recruitment 189–199 (2012). limitation in southern appalachian forests. Ecol. Monogr. 68, 213–235 (1998). 46. LaDeau, S. L. & Clark, J. S. Rising co2 levels and the fecundity of forest trees. 80. McEuen, A. B. & Curran, L. M. Seed dispersal and recruitment limitation Science 292, 95–8 (2001). across spatial scales in temperate forest fragments. Ecology 85, 507–518 47. Callahan, H. S., Del Fierro, K., Patterson, A. E. & Zafar, H. Impacts of elevated (2004). nitrogen inputs on oak reproductive and seed ecology. Glob. Change Biol. 14, 81. Emsweller, L. N., Gorchov, D. L., Zhang, Q., Driscoll, A. G. & Hughes, M. R. 285–293 (2008). Seed rain and disturbance impact recruitment of invasive plants in upland 48. Lambers, H. & Poorter, H. Inherent Variation in Growth Rate Between Higher forest. Invasive Plant Sci. Manag. 11, 69–81 (2018). Plants: A Search for Physiological Causes and Ecological Consequences, vol. 23, 82. Lindgren, s, Eriksson, O. & Moen, J. The impact of disturbance and seed 187-261 (Academic Press, 1992). availability on germination of alpine vegetation in the scandinavian 49. Hengl, T. et al. Soilgrids250m: global gridded soil information based on mountains. Arct. Antarct. Alp. Res. 39, 449–454 (2007). machine learning. PLoS ONE 12, 1–40 (2017). 83. Cai, W. H., Liu, Z., Yang, Y. Z. & Yang, J. Does environment filtering or seed 50. Sharma, A., Weindorf, D. C., Wang, D. D. & Chakraborty, S. Characterizing limitation determine post-fire forest recovery patterns in boreal larch forests? soils via portable x-ray fluorescence spectrometer: 4. cation exchange capacity Front. Plant Sci. 9, 1318 (2018). (cec). Geoderma 239, 130–134 (2015). 84. Darwin, C. On the Origin of Species (John Murray, 1859). 51. Hazelton, P. & Murphy, B. Interpreting Soil Test Results: What Do All The 85. Black, M. Darwin and seeds. Seed Sci. Res. 19, 193–199 (2009). Numbers Mean? (CSIRO publishing, 2016). 86. FAO. Global forest resources assessment 2020-key findings. un food and 52. Chowdhury, S. et al. Chapter Two - Role Of Cultural And Nutrient agriculture organization. Report (2020). Management Practices In Carbon Sequestration In Agricultural Soil, vol. 166, 87. Payn, T. et al. Changes in planted forests and future global implications. For. 131-196 (Academic Press, 2021). Ecol. Manag. 352, 57–67 (2015). NATURE COMMUNICATIONS | (2022)13:2381 | https://doi.org/10.1038/s41467-022-30037-9 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30037-9 88. Clark, J. S. et al. The impacts of increasing drought on forest dynamics, 119. Felsenstein, J. Phylogenies and the comparative method. Am. Nat. 125, 1–15 structure, and biodiversity in the united states. Glob. Change Biol. 22, (1985). 2329–2352 (2016). 120. Martins, E. P. & Hansen, T. F. Phylogenies and the comparative method: A 89. Gazol, A., Camarero, J. J., Anderegg, W. R. L. & Vicente-Serrano, S. M. general approach to incorporating phylogenetic information into the analysis Impacts of droughts on the growth resilience of northern hemisphere forests. of interspecific data. Am. Nat. 149, 646–667 (1997). Glob. Ecol. Biogeogr. 26, 166–176 (2017). 121. Tung Ho, L. S. & Ané, C. A linear-time algorithm for gaussian and non- 90. Stephens, S. L. et al. Managing forests and fire in changing climates. Science gaussian trait evolution models. Syst. Biol. 63, 397–408 (2014). 342, 41–42 (2013). 122. Clark, J. S. Data from: continent-wide tree fecundity driven by indirect climate 91. North, M. P. et al. Tamm review: reforestation for resilience in dry western u.s. effects https://doi.org/10.7924/r4348ph5t (2020). forests. For. Ecol. Manag. 432, 209–224 (2019). 92. Seidl, R., Rammer, W. & Spies, T. A. Disturbance legacies increase the resilience of forest ecosystem structure, composition, and functioning. Ecol. Acknowledgements Appl. 24, 2063–2077 (2014). For access to sites and logistical support we thank the National Ecological Observatory 93. Serra-Diaz, J. M. et al. Averaged 30 year climate change projections mask Network (NEON). The project has been funded continuously since 1992 by National opportunities for species establishment. Ecography 39, 844–845 (2016). Science Foundation grants to J.S.C, most recently DEB-1754443, and by the Belmont 94. Davis, F. W. et al. Shrinking windows of opportunity for oak seedling Forum (1854976), NASA (AIST16-0052, AIST18-0063), and the Programme d’Inves- establishment in southern california mountains. Ecosphere 7, e01573 (2016). tissement d’Avenir under project FORBIC (18-MPGA-0004)(Make Our Planet Great 95. LeBauer, D. S. & Treseder, K. K. Nitrogen limitation of net primary productivity Again). Puerto Rico data were funded by NSF grants to M.U., most recently, DEB 0963447 in terrestrial ecosystems is globally distributed. Ecology 89, 371–379 (2008). and LTREB 11222325. Data from the Andes Biodiversity and Ecosystem Research Group 96. Clark, J. S. et al. Continent-wide tree fecundity driven by indirect climate were funded by the Gordon and Betty Moore Foundation and NSF LTREB 1754647 to effects. Nat. Commun. 12, 1242 (2021). M.S. Additional funding to M.Z. came from the W.Szafer Institute of Botany of the Polish 97. Brady, N. C., Weil, R. R. & Weil, R. R. The Nature And Properties Of Soils, vol. Academy of Sciences and the Polish National Science Foundation (2019/33/B/NZ8/0134). 13 (Prentice Hall Upper Saddle River, 2008). M.B. was supported by the Polish National Science Centre grant no. 2019/35/D/NZ8/ 98. Farr, T. G. et al. The shuttle radar topography mission. Rev. Geophys. 45, 00050, and Polish National Agency for Academic Exchange Bekker programme PPN/BEK/ RG2004 (2007). https://doi.org/10.1029/2005RG000183. 2020/1/00009/U/00001. J.M.L. was supported by NSF grant DEB 1745496. Jerry Franklin’s 99. Clark, J. S. Landscape interactions among nitrogen mineralization, species data remain accessible through NSF LTER DEB-1440409. USDA Forest Service and USGS composition, and long-term fire frequency. Biogeochemistry 11, 1–22 (1990). research was funded by those agencies. We thanked Walt Koenig for providing data. We 100. Clark, J. S., Bell, D. M., Kwit, M. C. & Zhu, K. Competition-interaction thanked Christopher W. Woodall for comments on the manuscript. Any use of trade, firm, landscapes for the joint response of forests to climate change. Glob. Change or product names does not imply endorsement by the U.S. Government. Biol. 20, 1979–1991 (2014). 101. Begueria, S., Vicente-Serrano, S. M., Reig, F. & Latorre, B. Standardized Author contributions precipitation evapotranspiration index (spei) revisited: parameter fitting, J.S.C and T.Q. designed the study, performed analyses, and wrote the paper, J.S.C. compiled evapotranspiration models, tools, datasets and drought monitoring. Int. J. the MASTIF data and wrote the MASTIF model and software, B.C., V.J, and G.K. co-wrote Climatol. 34, 3001–3023 (2014). the paper. T.Q., R.A., M.A., D.A., Y.B., R.B., D.B., M.B., T.B., R.B., D.C.B., T.C., R.C., J.J.C., 102. Abatzoglou, J. T., Dobrowski, S. Z., Parks, S. A. & Hegewisch, K. C. C.C., N.L.C., B.C., F.C., T.C., A.J.D., E.D., H.D., N.D., S.D., M.D., S.D., L.D., J.E., T.J.F., W.F., Terraclimate, a high-resolution global dataset of monthly climate and climatic C.A.G., G.S.G., G.G., C.H.G., Q.G., A.H., A.H., Q.H., J.H., K.H., I.I., J.F.J., V.J., D.K., C.L.K., water balance from 1958-2015. Sci. Data 5, 170191 (2018). T.K., J.M.K., R.K.K., G.K., J.G.L., J.M.L., M.L., F.L., T.L., J.L., J.A.L., D.M., E.J.M., C.M.M., 103. Karger, D. N. et al. Climatologies at high resolution for the earth’s land surface E.M., R.M., J.A.M., T.A.N., K.N., J.O., R.P., I.S.P., I.M.P., L.P., J.P., R.P., M.D.R., C.D.R., areas. Sci. Data 4, 170122 (2017). K.C.R., F.R., J.D.S., C.L.S., W.H.S., H.S., B.S., S.S., M.S., M.A.S., N.L.S., J.N.S., I.S., S.S., J.J.S., 104. Schneider, R., Calama, R. & Martin-Ducup, O. Understanding tree-to-tree M.S., P.A.T., M.U., G.V., T.T.V., A.V.W., T.G.W., A.P.W., B.W., S.J.W., K.Z., J.K.Z., R.Z., variations in stone pine (pinus pinea l.) cone production using terrestrial laser M.Z., and J.S.C contributed data and revised the paper. scanner. Remote Sens. 12, 173 (2020). 105. Gavranović, A., Bogdan, S., Lanšćak, M., Čehulić, I. & Ivanković, M. Seed yield and morphological variations of beechnuts in four european beech (fagus Competing interests sylvatica l.) populations in croatia. South-East Eur. For. 9, 17–27 (2018). The authors declare no competing interests. 106. Maitner, B. S. et al. The bien r package: a tool to access the botanical information and ecology network (bien) database. Methods Ecol. Evol. 9, 373–379 (2018). Additional information Supplementary information The online version contains supplementary material 107. Clark, J. S., Silman, M., Kern, R., Macklin, E. & HilleRisLambers, J. Seed available at https://doi.org/10.1038/s41467-022-30037-9. dispersal near and far: patterns across temperate and tropical forests. Ecology 80, 1475–1494 (1999). Correspondence and requests for materials should be addressed to James S. Clark. 108. LePage, P. T., Canham, C. D., Coates, K. D. & Bartemucci, P. Seed abundance versus substrate limitation of seedling recruitment in northern temperate Peer review information Nature Communications thanks Francesco Chianucci, Eric forests of british columbia. Can. J. For. Res. 30, 415–427 (2000). Ribbens and the other, anonymous, reviewer(s) for their contribution to the peer review 109. Clark, J. S., LaDeau, S. & Ibanez, I. Fecundity of trees and the colonization- of this work. Peer reviewer reports are available. competition hypothesis. Ecol. Monogr. 74, 415–442 (2004). 110. Muller-Landau, H. C., Wright, S. J., Calderon, O., Condit, R. & Hubbell, S. P. Reprints and permission information is available at http://www.nature.com/reprints Interspecific variation in primary seed dispersal in a tropical forest. J. Ecol. 96, 653–667 (2008). Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in 111. Jones, F. A. & Muller-Landau, H. C. Measuring long-distance seed dispersal in published maps and institutional affiliations. complex natural environments: an evaluation and integration of classical and genetic methods. J. Ecol. 96, 642–652 (2008). 112. Clark, J. S. Individuals and the variation needed for high species diversity in forest trees. Science 327, 1129–1132 (2010). Open Access This article is licensed under a Creative Commons 113. Clark, J. S. et al. High-dimensional coexistence based on individual variation: a Attribution 4.0 International License, which permits use, sharing, synthesis of evidence. Ecol. Monogr. 80, 569–608 (2010). adaptation, distribution and reproduction in any medium or format, as long as you give 114. Clark, J. S., Bell, D. M., Kwit, M. C. & Zhu, K. Competition-interaction appropriate credit to the original author(s) and the source, provide a link to the Creative landscapes for the joint response of forests to climate change. Glob. Change Commons license, and indicate if changes were made. The images or other third party Biol. 20, 1979–91 (2014). material in this article are included in the article’s Creative Commons license, unless 115. Minor, D. M. & Kobe, R. K. Fruit production is influenced by tree size and size- indicated otherwise in a credit line to the material. If material is not included in the asymmetric crowding in a wet tropical forest. Ecol. Evol. 9, 1458–1472 (2019). article’s Creative Commons license and your intended use is not permitted by statutory 116. Zanne, A. E. et al. Three keys to the radiation of angiosperms into freezing regulation or exceeds the permitted use, you will need to obtain permission directly from environments. Nature 506, 89–92 (2014). the copyright holder. To view a copy of this license, visit http://creativecommons.org/ 117. Pagel, M. Inferring the historical patterns of biological evolution. Nature 401, licenses/by/4.0/. 877–884 (1999). 118. Revell, L. J. phytools: an r package for phylogenetic comparative biology (and other things). Methods Ecol. Evol. 3, 217–223 (2012). © The Author(s) 2022 10 NATURE COMMUNICATIONS | (2022)13:2381 | https://doi.org/10.1038/s41467-022-30037-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30037-9 ARTICLE Tong Qiu 1, Robert Andrus2, Marie-Claire Aravena 3, Davide Ascoli4, Yves Bergeron 5, Roberta Berretti4, Daniel Berveiller6, Michal Bogdziewicz 7, Thomas Boivin8, Raul Bonal9, Don C. Bragg10, Thomas Caignard11, Rafael Calama12, J. Julio Camarero 13, Chia-Hao Chang-Yang 14, Natalie L. Cleavitt15, Benoit Courbaud16, Francois Courbet8, Thomas Curt17, Adrian J. Das 18, Evangelia Daskalakou 19, Hendrik Davi8, Nicolas Delpierre6, Sylvain Delzon11, Michael Dietze 20, Sergio Donoso Calderon3, Laurent Dormont21, Josep Espelta 22, Timothy J. Fahey15, William Farfan-Rios 23, Catherine A. Gehring24, Gregory S. Gilbert 25, Georg Gratzer26, Cathryn H. Greenberg27, Qinfeng Guo 28, Andrew Hacket-Pain 29, Arndt Hampe11, Qingmin Han 30, Janneke Hille Ris Lambers31, Kazuhiko Hoshizaki 32, Ines Ibanez 33, Jill F. Johnstone 34, Valentin Journé 16, Daisuke Kabeya30, Christopher L. Kilner 1, Thomas Kitzberger35, Johannes M. H. Knops36, Richard K. Kobe 37, Georges Kunstler16, Jonathan G. A. Lageard 38, Jalene M. LaMontagne 39, Mateusz Ledwon40, Francois Lefevre 8, Theodor Leininger41, Jean-Marc Limousin42, James A. Lutz 43, Diana Macias44, Eliot J. B. McIntire45, Christopher M. Moore46, Emily Moran 47, Renzo Motta4, Jonathan A. Myers 48, Thomas A. Nagel49, Kyotaro Noguchi50, Jean-Marc Ourcival42, Robert Parmenter 51, Ian S. Pearse52, Ignacio M. Perez-Ramos53, Lukasz Piechnik54, John Poulsen1, Renata Poulton-Kamakura1, Miranda D. Redmond55, Chantal D. Reid1, Kyle C. Rodman 56, Francisco Rodriguez-Sanchez57, Javier D. Sanguinetti58, C. Lane Scher1, William H. Schlesinger1, Harald Schmidt Van Marle3, Barbara Seget54, Shubhi Sharma1, Miles Silman59, Michael A. Steele60, Nathan L. Stephenson 18, Jacob N. Straub61, I-Fang Sun 62, Samantha Sutton1, Jennifer J. Swenson1, Margaret Swift1, Peter A. Thomas 63, Maria Uriarte 64, Giorgio Vacchiano 65, Thomas T. Veblen 2, Amy V. Whipple24, Thomas G. Whitham24, Andreas P. Wion66, Boyd Wright67, S. Joseph Wright 68, Kai Zhu 25, Jess K. Zimmerman69, Roman Zlotin70, Magdalena Zywiec54 & James S. Clark 1,16 ✉ 1 Nicholas School of the Environment, Duke University, Durham, NC 27708, USA. 2Department of Geography, University of Colorado Boulder, Boulder, CO 80309, USA. 3Universidad de Chile, Facultad de Ciencias Forestales y de la Conservacion de la Naturaleza (FCFCN), La Pintana 8820808 Santiago, Chile. 4Department of Agriculture, Forest and Food Sciences, University of Torino, 10095 Grugliasco, TO, Italy. 5Forest Research Institute, University of Quebec in Abitibi-Temiscamingue, Rouyn-Noranda, QC J9X 5E4, Canada. 6Universite Paris-Saclay, Centre national de la recherche scientifique, AgroParisTech, Ecologie Systematique et Evolution, 91405 Orsay, France. 7Department of Systematic Zoology, Faculty of Biology, Adam Mickiewicz University, Umultowska 89, 61-614 Poznan, Poland. 8Institut National de Recherche pour Agriculture, Alimentation et Environnement (INRAE), Ecologie des Forets Mediterranennes, 84000 Avignon, France. 9Department of Biodiversity, Ecology and Evolution, Complutense University of Madrid, 28040 Madrid, Spain. 10USDA Forest Service, Southern Research Station, Monticello, AR 71656, USA. 11 Universite Bordeaux, Institut National de Recherche pour Agriculture, Alimentation et Environnement (INRAE), Biodiversity, Genes, and Communities (BIOGECO), 33615 Pessac, France. 12Centro de Investigacion Forestal - Instituto Nacional de Investigacion y Tecnologia Agraria y Alimentaria (INIA-CIFOR), 28040 Madrid, Spain. 13Instituto Pirenaico de Ecologla, Consejo Superior de Investigaciones Cientificas (IPE-CSIC), 50059 Zaragoza, Spain. 14Department of Biological Sciences, National Sun Yat-sen University, Kaohsiung 80424, Taiwan. 15Natural Resources, Cornell University, Ithaca, NY 14853, USA. 16Universite Grenoble Alpes, Institut National de Recherche pour Agriculture, Alimentation et Environnement (INRAE), Laboratoire EcoSystemes et Societes En Montagne (LESSEM), 38402 St. Martin-d’Heres, France. 17Aix Marseille universite, Institut National de Recherche pour Agriculture, Alimentation et Environnement (INRAE), 13182 Aix-en-Provence, France. 18USGS Western Ecological Research Center, Three Rivers, CA 93271, USA. 19Institute of Mediterranean and Forest Ecosystems, Athens, Greece. 20Earth and Environment, Boston University, Boston, MA 02215, USA. 21Centre d’Ecologie Fonctionnelle et Evolutive (CEFE), Centre National de la Recherche Scientifique (CNRS), 34293 Montpellier, France. 22Centre de Recerca Ecologica i Aplicacions Forestals (CREAF), Bellaterra, Catalunya 08193, Spain. 23Washington University in Saint Louis, Center for Conservation and Sustainable Development, Missouri Botanical Garden, St. Louis, MO 63110, USA. 24Department of Biological Sciences, Northern Arizona University, Flagstaff, AZ 86011, USA. 25Department of Environmental Studies, University of California, Santa Cruz, CA 95064, USA. 26Institute of Forest Ecology, Peter-Jordan-Strasse 82, 1190 Wien, Austria. 27Bent Creek Experimental Forest, USDA Forest Service, Asheville, NC 28801, USA. 28Eastern Forest Environmental Threat Assessment Center, USDA Forest Service, Southern Research Station, Research Triangle Park, NC 27709, USA. 29Department of Geography and Planning, School of Environmental Sciences, University of Liverpool, Liverpool, UK. 30Department of Plant Ecology Forestry and Forest Products Research Institute (FFPRI), Tsukuba, Ibaraki 305-8687, Japan. 31Department of Environmental Systems Science, ETH Zurich, Zurich 8092, Switzerland. 32Department of Biological Environment, Akita Prefectural University, Akita 010-0195, Japan. 33School for Environment and Sustainability, University of Michigan, Ann Arbor, MI 48109, USA. 34Institute of Arctic Biology, University of Alaska, Fairbanks, AK 99700, USA. 35Department of Ecology, Instituto de Investigaciones en Biodiversidad y Medioambiente (Consejo Nacional de Investigaciones Cientificas y Tecnicas - Universidad Nacional del NATURE COMMUNICATIONS | (2022)13:2381 | https://doi.org/10.1038/s41467-022-30037-9 | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30037-9 Comahue), Quintral 1250, 8400 Bariloche, Argentina. 36Health and Environmental Sciences Department, Xian Jiaotong-Liverpool University, Suzhou 215123, China. 37Department of Plant Biology, Program in Ecology, Evolutionary Biology, and Behavior, Michigan State University, East Lansing, MI 48824, USA. 38Department of Natural Sciences, Manchester Metropolitan University, Manchester M1 5GD, UK. 39Department of Biological Sciences, DePaul University, Chicago, IL 60614, USA. 40Institute of Systematics and Evolution of Animals, Polish Academy of Sciences, Slawkowska 17, 31-016 Krakow, Poland. 41USDA, Forest Service, Southern Research Station, PO Box 227, Stoneville, MS 38776, USA. 42CEFE, Univ Montpellier, CNRS, EPHE, IRD, 1919 route de Mende, 34293 Montpellier Cedex 5, France. 43Department of Wildland Resources, and the Ecology Center, Utah State University, Logan, UT 84322, USA. 44Department of Biology, University of New Mexico, Albuquerque, NM 87131, USA. 45Pacific Forestry Centre, Victoria, BC V8Z 1M5, Canada. 46Department of Biology, Colby College, Waterville, ME 04901, USA. 47School of Natural Sciences, UC Merced, Merced, CA 95343, USA. 48Department of Biology, Washington University in St. Louis, St. Louis, MO, USA. 49Department of forestry and renewable forest resources, Biotechnical Faculty, University of Ljubljana, Ljubljana, Slovenia. 50Tohoku Research Center, Forestry and Forest Products Research Institute, Morioka, Iwate 020-0123, Japan. 51Valles Caldera National Preserve, National Park Service, Jemez Springs, NM 87025, USA. 52Fort Collins Science Center, 2150 Centre Avenue Bldg C, Fort Collins, CO 80526, USA. 53Inst. de Recursos Naturales y Agrobiologia de Sevilla, Consejo Superior de Investigaciones Cientificas (IRNAS-CSIC), Seville, Andalucia, Spain. 54W. Szafer Institute of Botany, Polish Academy of Sciences, Lubicz 46, 31-512 Krakow, Poland. 55Department of Forest and Rangeland Stewardship, COlorado State University, Fort COllins, CO, USA. 56 Department of Forest and Wildlife Ecology, University of Wisconsin-Madison, Madison, WI 53706, USA. 57Department of Biologia Vegetal y Ecologia, Universidad de Sevilla, 41012 Sevilla, Spain. 58Bilogo Dpto. Conservacin y Manejo Parque Nacional Lanin Elordi y Perito Moreno, 8370 San Marten de los Andes Neuqun, Argentina. 59Department of Biology, Wake Forest University, 1834 Wake Forest Rd, Winston-Salem, NC 27106, USA . 60Department of Biology, Wilkes University, 84 West South Street, Wilkes-Barre, PA 18766, USA. 61Department of Environmental Science and Ecology, State University of New York-Brockport, Brockport, NY 14420, USA. 62Center for Interdisciplinary Research on Ecology and Sustainability, College of Environmental Studies, National Dong Hwa University, Hualien, Taiwan. 63School of Life Sciences, Keele University, Staffordshire ST5 5BG, UK. 64Department of Ecology, Evolution and Environmental Biology, Columbia University, 1113 Schermerhorn Ext., 1200 Amsterdam Ave., New York, NY 10027, USA. 65Department of Agricultural and Environmental Sciences - Production, Territory, Agroenergy (DISAA), University of Milan, 20133 Milano, Italy. 66Department of Forest and Rangeland Stewardship, Colorado State University, Fort Collins, CO 80523, USA. 67Botany, School of Environmental and Rural Science, University of New England, Armidale, NSW 2350, Australia. 68Smithsonian Tropical Research Institute, Apartado 0843n03092, Balboa, Republic of Panama. 69Department of Environmental Sciences, University of Puerto Rico, Rio Piedras, PR 00936, USA. 70Geography Department and Russian and East European Institute, Bloomington, IN 47405, USA. ✉email: jimclark@duke.edu 12 NATURE COMMUNICATIONS | (2022)13:2381 | https://doi.org/10.1038/s41467-022-30037-9 | www.nature.com/naturecommunications