Globally, tree fecundity exceeds productivity gradients

1 2 Global tree-fecundity is linked to the intensity of species interactions — 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Valentin Journe,1 Robert Andrus,2 Marie-Claire Aravena,3 Davide Ascoli,4 Yves Bergeron,5 Roberta Berretti,4 Daniel Berveiller,6 Michal Bogdziewicz,7 Thomas Boivin,8 Raul Bonal,9 Dale Brockway,10 Thomas Caignard,11 Rafael Calama,12 J. Julio Camarero,13 Chia-Hao Chang-Yang,14 Natalie L. Cleavitt,15 Benoit Courbaud,1 Francois Courbet,8 Thomas Curt,16 Adrian J. Das,17 Evangelia Daskalakou,18 Hendrik Davi,8 Nicolas Delpierre ,6 Sylvain Delzon,11 Michael Dietze,19 Sergio Donoso Calderon,3 Laurent Dormont,20 Josep Espelta,21 Timothy J. Fahey,15 William Farfan-Rios,22 Catherine A. Gehring,23 Gregory S. Gilbert,24 Georg Gratzer,25 Cathryn H. Greenberg,26 Qinfeng Guo,27 Andrew Hacket-Pain,28 Arndt Hampe ,11 Qingmin Han,29 Janneke Hille Ris Lambers,30 Kazuhiko Hoshizaki,31 Ines Ibanez,32 Jill F. Johnstone,33 Daisuke Kabeya,29 Christopher L. Kilner,34 Thomas Kitzberger,35 Johannes M.H. Knops,36 Richard K. Kobe,37 Georges Kunstler,1 Jonathan G.A. Lageard,38 Jalene M. LaMontagne,39 Mateusz Ledwon,40 Francois Lefevre ,8 Theodor Leininger,41 Jean-Marc Limousin,42 James A. Lutz,43 Diana Macias,44 Eliot J.B. McIntire,45 Christopher M. Moore,46 Emily Moran,47 Renzo Motta,4 Jonathan A. Myers,48 Thomas A. Nagel,49 Kyotaro Noguchi,50 Jean-Marc Ourcival ,42 Robert Parmenter,51 Ian S. Pearse,52 Ignacio M. Perez-Ramos,53 Lukasz Piechnik,54 John Poulsen,34 Renata Poulton-Kamakura,34 Tong Qiu,34 Miranda D. Redmond,55 Chantal D. Reid,34 Kyle C. Rodman,56 Francisco RodriguezSanchez,57 Javier D. Sanguinetti ,58 C. Lane Scher,34 Harald Schmidt Van Marle,3 Barbara Seget,54 Shubhi Sharma,34 Miles Silman,59 Michael A. Steele,60 Nathan L. Stephenson,17 Jacob N. Straub,61 I-Fang Sun,62 Samantha Sutton,34 Jennifer J. Swenson,34 Margaret Swift,34 Peter A. Thomas,63 Maria Uriarte,64 Giorgio Vacchiano,65 Thomas T. Veblen,2 Amy V. Whipple,23 Thomas G. Whitham,23 Andreas P. Wion,66 Boyd Wright,67 S. Joseph Wright,68 Kai Zhu,24 Jess K. Zimmerman,69 Roman Zlotin,70 Magdalena Zywiec,54 James S. Clark,34 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 1 Universite 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 2 Department of Geography, University of Colorado Boulder, Boulder, CO 80309 USA 3 Universidad de Chile, Facultad de Ciencias Forestales y de la Conservacion de la Naturaleza (FCFCN), La Pintana, 8820808 Santiago, Chile 4 Department of Agriculture, Forest and Food Sciences, University of Torino, 10095 Grugliasco, TO, Italy 5 Forest Research Institute, University of Quebec in Abitibi-Temiscamingue, Rouyn-Noranda, QC J9X 5E4, Canada 6 Universite Paris-Saclay, Centre national de la recherche scientifique, AgroParisTech, Ecologie Systematique et Evolution, 91405 Orsay, France 7 Department of Systematic Zoology, Faculty of Biology, Adam Mickiewicz University, Umultowska 89, 61-614 Poznan, Poland 8 Institut National de Recherche pour Agriculture, Alimentation et Environnement (INRAE), Ecologie des Forets Mediterranennes, 84000 Avignon, France 9 Department of Biodiversity, Ecology and Evolution, Complutense University of Madrid, 28040 Madrid, Spain 10 Southern Research Station, USDA Forest Service, Auburn, AL 36849 USA 11 Universite Bordeaux, Institut National de Recherche pour Agriculture, Alimentation et Environnement (INRAE), 1 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 Biodiversity, Genes, and Communities (BIOGECO), 33615 Pessac, France 12 Centro de Investigacion Forestal - Instituto Nacional de Investigacion y Tecnologia Agraria y Alimentaria (INIACIFOR), 28040 Madrid, Spain 13 Instituto Pirenaico de Ecologla, Consejo Superior de Investigaciones Cientificas (IPE-CSIC), 50059 Zaragoza, Spain 14 Department of Biological Sciences, National Sun Yat-sen University, Kaohsiung 80424, Taiwan 15 Natural Resources, Cornell University, Ithaca, NY, 14853 USA 16 Aix Marseille universite, Institut National de Recherche pour Agriculture, Alimentation et Environnement (INRAE), 13182 Aix-en-Provence, France 17 USGS Western Ecological Research Center, Three Rivers, CA, 93271 USA 18 Institute of Mediterranean and Forest Ecosystems, Greece 19 Earth and Environment, Boston University, Boston, MA, 02215 USA 20 Centre d’Ecologie Fonctionnelle et Evolutive (CEFE), Centre National de la Recherche Scientifique (CNRS), 34293 Montpellier, France. 21 Centre de Recerca Ecologica i Aplicacions Forestals (CREAF), Bellaterra, Catalunya 08193, Spain 22 Washington University in Saint Louis, Center for Conservation and Sustainable Development, Missouri Botanical Garden, St. Louis, MO 63110 USA 23 Department of Biological Sciences, Northern Arizona University, Flagstaff, AZ 86011 USA 24 Department of Environmental Studies, University of California, Santa Cruz, CA 95064 USA 25 Institute of Forest Ecology, Peter-Jordan-Strasse 82, 1190 Wien, Austria 26 Bent Creek Experimental Forest, USDA Forest Service, Asheville, NC 28801 USA 27 Eastern Forest Environmental Threat Assessment Center, USDA Forest Service, Southern Research Station, Research Triangle Park, NC 27709 USA 28 Department of Geography and Planning, School of Environmental Sciences, University of Liverpool, Liverpool, United Kingdom 29 Department of Plant Ecology Forestry and Forest Products Research Institute (FFPRI), Tsukuba, Ibaraki, 3058687 Japan 30 Department of Environmental Systems Science, ETH Zurich, Switzerland 8092 31 Department of Biological Environment, Akita Prefectural University, Akita 010-0195, Japan 32 School for Environment and Sustainability, University of Michigan, Ann Arbor, MI 48109 33 Institute of Arctic Biology, University of Alaska, Fairbanks, AK 99700, USA 34 Nicholas School of the Environment, Duke University, Durham, NC 27708 USA 35 Department of Ecology, Instituto de Investigaciones en Biodiversidad y Medioambiente (Consejo Nacional de Investigaciones Cientificas y Tecnicas - Universidad Nacional del Comahue), Quintral 1250, 8400 Bariloche, Argentina 36 Health and Environmental Sciences Department, Xian Jiaotong-Liverpool University, Suzhou, China, 215123 37 Department of Plant Biology, Program in Ecology, Evolutionary Biology, and Behavior, Michigan State University, East Lansing, MI 48824 38 Department of Natural Sciences, Manchester Metropolitan University, Manchester M1 5GD, UK 39 Department of Biological Sciences, DePaul University, Chicago, IL 60614 USA 40 Institute of Systematics and Evolution of Animals, Polish Academy of Sciences, Slawkowska 17, 31-016 Krakow, Poland 41 USDA, Forest Service, Southern Research Station, PO Box 227, Stoneville, MS 38776 42 CEFE, Univ Montpellier, CNRS, EPHE, IRD, 1919 route de Mende, 34293 Montpellier Cedex 5, France 43 Department of Wildland Resources, and the Ecology Center, Utah State University, Logan, UT 84322 USA 44 Department of Biology, University of New Mexico, Albuquerque, NM 87131 USA 45 Pacific Forestry Centre, Victoria, British Columbia, V8Z 1M5 Canada 46 Department of Biology, Colby College, Waterville, ME 04901 USA 2 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 47 School of Natural Sciences, UC Merced, Merced, CA 95343 USA of Biology, Washington University in St. Louis, St. Louis, MO 49 Department of forestry and renewable forest resources, Biotechnical Faculty, University of Ljubljana, Ljubljana, Slovenia 50 Tohoku Research Center, Forestry and Forest Products Research Institute, Morioka, Iwate, 020-0123, Japan 51 Valles Caldera National Preserve, National Park Service, Jemez Springs, NM 87025 USA 52 Fort Collins Science Center, 2150 Centre Avenue, Bldg C, Fort Collins, CO 80526 USA 53 Inst. de Recursos Naturales y Agrobiologia de Sevilla, Consejo Superior de Investigaciones Cientificas (IRNASCSIC), Seville, Andalucia, Spain 54 W. Szafer Institute of Botany, Polish Academy of Sciences, Lubicz 46, 31-512 Krakow, Poland 55 Department 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 57 Department of Biologia Vegetal y Ecologia, Universidad de Sevilla, 41012 Sevilla, Spain 58 Bilogo Dpto. Conservacin y Manejo Parque Nacional Lanin Elordi y Perito Moreno 8370, San Marten de los Andes Neuqun Argentina 59 Department of Biology, Wake Forest University, 1834 Wake Forest Rd, Winston-Salem, NC 27106 USA 60 Department of Biology, Wilkes University, 84 West South Street, Wilkes-Barre, PA 18766 USA 61 Department of Environmental Science and Ecology, State University of New York-Brockport, Brockport, NY 14420 USA 62 Center for Interdisciplinary Research on Ecology and Sustainability, College of Environmental Studies, National Dong Hwa University, Hualien, Taiwan 63 School of Life Sciences, Keele University, Staffordshire ST5 5BG, UK 64 Department of Ecology, Evolution and Environmental Biology, Columbia University, 1113 Schermerhorn Ext., 1200 Amsterdam Ave., New York, NY 10027 65 Department of Agricultural and Environmental Sciences - Production, Territory, Agroenergy (DISAA), University of Milan, 20133 Milano, Italy 66 Department of Forest and Rangeland Stewardship, Colorado State University, Fort Collins, CO 80523, USA 67 Botany, School of Environmental and Rural Science, University of New England, Armidale, NSW, 2350, Australia 68 Smithsonian Tropical Research Institute, Apartado 0843n03092, Balboa, Republic of Panama 69 Department of Environmental Sciences, University of Puerto Rico, Rio Piedras, PR 00936 USA 70 Geography Department and Russian and East European Institute, Bloomington, IN 47405 USA 48 Department 122 123 124 125 —keywords: climate | competition | forest regeneration | seed consumption | species interactions | tree fecundity 3 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 Abstract Increasing evidence points to intense species competition in wet tropical forests that that may be explained by interactions involving seeds, seedlings, and their consumers. Lack of tree fecundity data across temperate to tropical communities precludes analysis of how the seed resource contributes to biotic interactions that can drive biogeographic diversity patterns. A global synthesis of raw seed-production data shows a 2.4 order of magnitude increase in seed abundance from cold, dry to warm, wet climates, driven by a 2.0 order of magnitude increase in seed production for a given tree size. The modest increase in forest productivity across the same climate gradient cannot explain this 100-fold increase in seed production or the 250-fold increase in seed mass per forest area reported here. The increase in seeds per tree can arise from adaptive evolution driven by intense species interactions or from the direct effects of a warm, moist climate on tree fecundity. Either way, the massive differences in seed supply to temperate versus tropical communities ramifies through food webs, affecting community and ecosystem dynamics, including seedling competition, populations of seed consumers and frugivore-seed dispersers, all of which appear to be especially important in the wet tropics. Introduction Understanding seed production can help resolve the paradox of extreme tree diversity in the warm latitudes where intense competition is expected to limit coexistence [1, 2, 3, 4]. High net primary productivity (NPP) that comes with long growing seasons accelerates growth, increases plant competition, and elevates mortality rates [5, 6, 7]. This coincidence of high diversity with intense competition is increasingly explained by coexistence mechanisms involving interactions between seeds and seedlings through their natural enemies [8, 9, 10]. Tree fecundity determines the density of competing offspring and the diets of consumers and seed dispersers that depend on them [11, 12, 13], and it is clearly subject to adaptive evolution [14, 15]. If there is a latitudinal gradient in seed production, is it a product of greater seed production for a given tree size, or is it the case that tropical trees are simply larger and/or embedded in more productive communities, as assumed in the Dynamic Global Vegetation Models (DGVMs) used to understand effects of climate change [16, 17]? Temperate-tropical fecundity gradients that go beyond what could be explained by differences in tree size or NPP would provide evidence that biogeographic diversity trends depend on this critical demographic variable that is a foundation for many forest food webs [18, 19]. While forest inventory data continue to improve estimates of growth and mortality across climate gradients [7, 20], fecundity evidence has remained unavailable. This synthesis allows us to quantify the fecundity gradient on a global scale and determine that it is amplified in warm/moist climates beyond what can be explained by tree size or NPP. Seed and seedling densities are the starting point not only for competition, but also for consumer-based explanations of coexistence that were first recognized in the tropics [8, 9]. Intense plant competition is an inevitable consequence of long growing seasons and high annual growth [21, 22, 23]. Accumulating evidence indicates that consumer pressure is likewise intense [24, 25, 4]. Selection for high seed production might offset high losses to biotic interactions, while at the same time intensifying them by increasing density- and frequencydependent interactions. By the widely invoked Janzen-Connell mechanism [8, 9], a host tree can escape its specialist consumers where that host is rare, i.e., a density-dependent process. A generalist consumer imposes indirect competition between its multiple hosts, as an increase in one attracts the natural enemies it shares with others, a density- and frequency-dependent process. The seed diversity available to consumers could differ from that of trees, because the 4 171 172 173 174 175 abundant species may not produce large seed crops, and vice versa. There is potential for an arms race between species [26, 27] as selective pressures balance the benefits of producing more seed against the costs of diverting resources from growth and defense [28, 29]. Taken together, tree fecundity is a foundation for community interactions that increasingly appear to be most intense in tropical forests. Figure 1: a) Climate effects on fecundity could be I) negligible, in which case there is no latitudinal fecundity gradient (right), or there could be direct (II) or indirect effects through adaptive responses to intense species interactions in tropical climates (III). Both II and III could be amplified beyond what could be explained by gradients in tree size or NPP. Either way, there is potential for positive feedback involving arrows in green. b) Orders of magnitude increases from cold/dry to warm/moist for individual (ISP) and community (CSP) seed production compared with NPP. Curves are sections through surfaces (dashed lines) in Fig. 2, with scales for moisture deficit (above) and temperature (below). Curves are in proportion to minimum values in cold, dry conditions. Confidence intervals (95%) are not visible for ISP and NPP due to the large number of trees. They are wider for CSP due to fewer inventory plots at high temperatures (Fig. 2b). 176 177 178 179 Fecundity could vary due to climate directly or indirectly, the latter through adaptation to biotic interactions that, in turn, respond to climate (Fig. 1a). Because reproductive effort depends on both seed sizes and numbers [30], and it varies with tree size [31], individual seed productivity (ISP) is standardized for tree basal area, 𝑓ˆ𝑖 𝑗 𝑠 × 𝑔𝑠 basal area𝑖 = g m−2 yr−1 𝐼𝑆𝑃𝑖 𝑗 = 180 181 182 183 184 185 186 187 (1) depending on mass of a seed 𝑔𝑠 produced by species 𝑠, where the estimate of mean seed production 𝑓ˆ𝑖 𝑗 𝑠 for tree 𝑖 at location 𝑗 accounts for the uncertainty in its seed production each year, 𝑓𝑖 𝑗 𝑠,𝑡 (see Methods). If seed production is determined solely by tree size, as assumed in ecological models (reviewed in [31]), then climate effects could come through its effects on past growth that results in larger trees (Fig. 1a, II), and ISP (standardized for size) will be constant across climate gradients. While ISP𝑖 𝑗 quantifies production by individuals, community seed production, CSP 𝑗 , quantifies seed density on the forest floor, the starting point for interactions between seeds, seedlings, 5 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 consumers, and dispersers. [We hereafter omit subscripts to reduce clutter.] Like NPP, CSP is a community property, defined as the seed production summed over all trees on a plot (g ha−1 yr−1 , eqn 4). CSP might scale as a fraction of NPP, as suggested by some empirical evidence [32] and assumed in DGVMs [16, 17], predicting high CSP where NPP is highest in warm/moist climates [33]. If ISP is determined as a simple fraction of tree size, or CSP as a fraction of NPP, then it is hard to invoke biotic interactions as an explanation for variation along climate gradients. Alternatively, if the responses are amplified beyond what could be explained by the effects of climate on size or NPP, then climate gradients could be a driver of intense biotic interactions in the tropics. There are at least two potential causes for fecundity amplification (Fig. 1a). First, ISP might have flexibility to respond to a longer growing season [34, 35] well in excess of tree growth, which is limited by mechanical and hydraulic constraints on tree size [36, 37]. At the community scale, NPP is further constrained by the compensatory losses in stand biomass as mortality increases to offset increases in growth [22]. Thus, while NPP increases with warm, wet conditions, the lack of structural constraints on producing more seeds might allow for a disproportionate fecundity response, the amplification of figure 1a, II. Amplification could also be driven by intense species interactions in the wet tropics [4, 38] that increase selection for seed production, mediated by allocation trade-offs between seeds versus growth and defense [39, 40]. Whether amplification occurs as a direct response to climate or as an adaptive response to intense biotic interactions (Fig. 1a, II and III), the densityand frequency-dependent processes involving competition, consumers, and seed dispersers have community-wide implications. A potential arms race follows from the feedback between high seed production and the selection pressures to offset mortality losses (green arrows in Fig. 1a). Biogeographic variation in fecundity remains largely unknown. Fecundity studies typically report on one to a few species from one to a few sites. Recent compilations of seed numbers recognise the challenges posed by differing methods, some yielding estimates of stand averages and others offering various individual-tree estimates that are difficult to compare [41, 42]. Efforts to synthesize this literature globally report that seed size [43] or variation in seed numbers (e.g., [42]) increase with temperature or with variability in precipitation or temperature. Latitudinal trends in seed size [43, 44] may not translate to trends in fecundity, which depends on the number of seeds × seed size. A decline in predicted seed-mass density (per forest floor) with increasing latitude reported from a study that included only forests at low latitudes and mostly heath and grasslands at high latitudes [43] highlights the need to separate variation in tree fecundity from variation in tree abundance. This synthesis extends the Masting Inference and Forecasting (MASTIF) network [45] to determine the climate controls on seed production globally and the extent to which those trends go beyond what can be explained by effects of size and productivity. Climate trends are summarized by mean annual temperature and moisture deficit. We additionally allow for effects of individual condition and local habitat variation by including tree diameter, shade class, and soil cation exchange capacity (Materials and Methods). Results and Discussion Community seed production (CSP) increases 2.4 orders of magnitude to a global maximum in the warm, moist tropics, primarily driven by a two order-of-magnitude increase in seed production for a given tree size (ISP) (Fig. 1b). These increases align with the geographic trend in NPP (panels in Fig. 2), but the amplification for seed production in excess of the NPP gradient provides first evidence that it can play a central role in the species interactions hypothesized to 6 Figure 2: a) Climate responses for (a) ISP (seed mass per tree basal area, log10 g m−2 y−1 ) (b) CSP (seed mass per ha forest floor, log10 g ha−1 y−1 ), and (c) NPP (kg C m−2 y−1 ). Dashed lines indicate the transect from dry taiga to wet tropics in Fig. 1b. Note the linear scale for (c) and log10 scales for (a) and (b). Convex hulls are defined by observations (red), including individual trees (a, c) and inventory plots (b). Surface transparency increases as the inverse of the predictive standard error–faded edges reflect increased uncertainty at data extremes. Coefficients are reported in Table S4 and Table S2 for NPP. 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 be most intense in warm, moist climates. If individual fecundity scales with tree basal area, then ISP (seed mass per tree basal area) would be flat in Fig. 1b. If community seed production scales with NPP, then CSP would increase no faster than NPP on the proportionate scale in Fig. 1b. The amplification over size and NPP observed here has implications for trophic interactions, and it provides insights into cause. A first important benchmark of this study is the exposure of fecundity trends with global climate. The average seed production for 95% of trees of a given size varies over five orders of magnitude, with ISP ranging from 0.000025 to 50 g per cm2 of basal area (Figure S5a). The increase in ISP to its highest values in warm, moist climates (Fig. 2b) is driven more by temperature than by moisture (Table S4), amplified by moisture where temperatures are high (Figure S2c). The five order-of-magnitude range for individual ISP is matched by that for community seed production, with 95% of CSP values ranging from 50 g to 2500 kg ha−1 (Figure S5b). The 100-fold increase in ISP across the climate gradient is more than matched by the 250-fold increase in CSP. Forest productivity cannot explain the global fecundity gradient evident at the individual or community levels. Like fecundity trends, NPP shows high values in warm, moist climates (Fig. 2c). However, the three-fold range of NPP across this climate space is swamped by the 100and 250-fold ranges for ISP and CSP (Fig. 1b). The amplification of both ISP and CSP means that not only do individual trees produce more seed for a given size in the wet tropics, but also that seed abundance is amplified at the community level (Figure S2f). [Community-level CSP need not necessarily track ISP responses due to heterogeneous size-species structures associated with local site conditions, past disturbance, and competition.] These results extend the previous discovery of a fecundity hotspot in the warm, moist southeastern North America [45] to a global pattern. Fecundity trends that are amplified well beyond what can be explained by size or productivity alone could be driven by direct climate effects, by selective pressures due to high losses to biotic interactions, or both (Fig. 1a). The two order-of-magnitude climatic and latitudinal trend in seed mass per forest-floor area (CSP) has its most direct implications for density-dependent interactions, which include competition within tree species and frequency-dependent consumers. All else being equal, a 100-fold gradient in seed supply requires corresponding mortality losses 7 Figure 3: Species diversity in seeds (vertical axis) is lower than expected from Í species diversity in trees (horizontal axis). In both cases, diversity is evaluated from the Shannon index, − 𝑠 𝑝 𝑠 log 𝑝 𝑠 , where 𝑝 𝑠 is the fraction of species 𝑠 in basal area (trees) and CSP (seeds). Each point represents an inventory plot. Except at low tree diversity, points lie almost entirely below the 1:1 line (dashed). The legend at top left shows mean annual temperature (symbol color) and mass of the average seed (symbol size). 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 to yield similar densities of adult trees [21, 22]. Elevated densities of seeds, fruits, and nuts and their offsetting mortality losses increase selective pressure for the most competitive phenotypes. The bottom-up enrichment of food webs that cascades to higher trophic levels [18, 19, 25] must inevitably increase consumer and disperser densities that, in turn, impose frequency-dependence selection on seed and seedling survival [8]. The magnitude of amplification leaves no doubt that it intensifies species interactions in the wet tropics. Frequency-dependent consumer pressures depend on diversity of the seed resource, which is poorly predicted by the standard inventory of trees. Species diversity of both seed mass and tree basal area is highest in the warm climates where diversity of the seed resource would be overestimated on the basis of tree diversity (Fig. 3). The lower species diversity for seeds in warm climates results from the fact that species having modest differences in tree basal area vary widely in fecundity; tendency for a subset of species to dominate seed production reduces seed diversity below that for trees. In the cool climates where seeds tend to be small (small, blue symbols in Fig. 3), the low diversity that would be estimated on the basis of trees masks an unexpectedly high seed diversity (Fig. 2). Although many studies do not record fecundity for species having the smallest seeds (e.g., Salicaceae), these are also the seeds that are least apparent to vertebrate consumers. Omission of these smallest seeds means that values are overestimates, but still relevant for many consumers. The net effect of reduced seed diversity in warm climates affects frequency-dependent processes [46], such as host-specific seed predation. The concentration of seed mass in a smaller than expected species diversity reduces the apparency of weak producers, while potentially concentrating consumption on species that are not necessarily abundant, but that can dominate seed production. The biogeographic variation between trees (ISP) and communities (CSP) is distinct from the large masting literature focused on variation in the magnitude of reproduction over time within trees or stands. Temporal variation in climate [47, 48, 49] that interacts with variable 8 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 storage and pollen supply [50, 51] is of great interest for understanding allocation shifts within individuals, but it fundamentally differs from geographic variation in populations subjected to divergent selection histories [47]. The 100-fold trend in expected ISP and CSP (Fig. 2a) is still modest relative to the within-tree (over time) and between-tree variation that motivates local-scale studies (Figure S2). The fact that the massive geographic trend in Fig. 2a can be readily masked by other sources of variation emphasizes the importance of large data sets that span broad coverage in individual condition, habitat, and climate. Whether or not the amplified fecundity response in warm, moist climates represents a legacy of adaptive evolution to intense species interactions, its 100-fold biogeographic gradient adds a new dimension to the understanding of trophic processes that may control latitudinal diversity gradients. The fact that both individual fecundity and community-level CSP overwhelm climate responses in NPP (Fig. 2a) means that fecundity of many species can contribute to the selection pressures on others and on their consumers [52]. If host-specific consumers regulate diversity through density- and frequency-dependent attack, then their strongest impacts are occurring where seed supply can support their highest numbers. The dramatic biogeographic trend sets up the potential for an evolutionary arms race to increase fecundity in the warm, moist tropics. Regardless of whether this arms race has occurred, the trends in stand-level seed rain imply profound implications for food web dynamics. A positive feedback on selection pressure in diverse tropical forests could ensue where species from every major angiosperm clade enrich functional space and niche overlap. Declines in biodiversity that result from climate change, habitat degradation, and human exploitation in the tropical regions where interaction strength is intense is expected to ramify through food webs to a degree that is not expected where interactions are loose and generally weak [53]. The temperate-tropical gradient can motivate research on its contribution to consumer and disperser guilds [4] and the broader implications for diversity. 9 314 References 316 [1] R. H. MacArthur, Geographical Ecology: Patterns in the Distribution of Species (1972), princeton edn. 317 [2] J. Chave, H. C. Muller-Landau, S. A. Levin, The American Naturalist 159, 1 (2002). 318 [3] J. H. Brown, Journal of Biogeography 41, 8 (2014). 319 [4] A. L. Hargreaves, et al., Science Advances 5, 1 (2019). 320 [5] O. L. Phillips, A. H. Gentry, Science 263, 954 (1994). 321 [6] S. L. Lewis, et al., Journal of Ecology 92, 929 (2004). 322 [7] N. L. Stephenson, P. J. Van Mantgem, Ecology Letters 8, 524 (2005). 323 [8] D. Janzen, The American Naturalist 104, 501 (1970). 315 324 325 326 [9] J. Connell, Herbivores and the Number of Tree Species in Tropical Forests (Pudoc, Wageningen, 1970), p. 298–312. [10] K. Zhu, C. W. Woodall, J. V. Monteiro, J. S. Clark, Ecology 96, 2319 (2015). 328 [11] J. Terborgh, Community aspects of frugivory in tropical forests (Springer, Dordrecht, 1986), vol. 15 of Tasks for Vegetation Science. 329 [12] R. T. Corlett, Biological Conservation 163, 13 (2013). 330 [13] K. Mokany, S. Prasad, D. A. Westcott, Nature Communications 5, 3971 (2014). 327 332 [14] R. J. Petit, R. Bialozyt, P. Garnier-Géré, A. Hampe, Forest Ecology and Management 197, 117 (2004). 333 [15] H. X. Wu, R. Ker, Z. Chen, M. Ivkovic, Evolutionary Applications 14, 834 (2021). 334 [16] S. Sitch, et al., Global Change Biology 9, 161 (2003). 335 [17] G. Krinner, et al., Global Biogeochemical Cycles 19, 1 (2005). 336 [18] R. S. Ostfeld, F. Keesing, Trends in Ecology and Evolution 15, 232 (2000). 337 [19] A. E. Rosenblatt, O. J. Schmitz, Trends in Ecology and Evolution 31, 965 (2016). 338 [20] R. J. Brienen, et al., Nature Communications 11, 1 (2020). 331 340 [21] E. Assmann, The principles of forest yield study. Studies in the organic production, structure, increment and yield of forest stands. (1970). 341 [22] J. S. Clark, Journal of Ecology 78, 275 (1990). 342 [23] N. L. Stephenson, et al., Ecological Monographs 81, 527 (2011). 343 [24] R. Bagchi, et al., Nature 506, 85 (2014). 344 [25] T. Levi, et al., Proceedings of the National Academy of Sciences 116, 581 (2019). 339 10 345 346 [26] R. Dawkins, J. R. Krebs, Proceedings of the Royal Society of London. Series B, Biological Sciences 205, 489 (1979). 348 [27] M. Gruntman, D. Groß, M. Májeková, K. Tielbörger, Nature Communications 8, 2235 (2017). 349 [28] A. B. Berdanier, J. S. Clark, Ecosphere 7, e01313 (2016). 350 [29] J. D. Lauder, E. V. Moran, S. C. Hart, Tree Physiology 39, 1071 (2019). 351 [30] M. Westoby, E. Jurado, M. Leishman, Trends in Ecology and Evolution 7, 368 (1992). 352 [31] T. Qiu, et al., Proceedings of the National Academy of Sciences 118 (2021). 353 [32] G. Vacchiano, et al., Ecological Modelling 376, 40 (2018). 354 [33] S. Del Grosso, et al., Ecology 89, 2117 (2008). 355 [34] S. H. Yeoh, et al., Molecular Ecology 26, 5074 (2017). 356 [35] I. Mendoza, et al., Biotropica 50, 431 (2018). 357 [36] G. W. Koch, S. C. Sillett, G. M. Jennings, S. D. Davis, Nature 428, 851 (2004). 358 [37] D. A. King, S. J. Davies, S. Tan, N. S. Md. Noor, Functional Ecology 23, 284 (2009). 359 [38] L. S. Comita, et al., Journal of Ecology 102, 845 (2014). 360 [39] J. R. Obeso, New Phytologist 155, 321 (2002). 361 [40] J. D. Fridley, Journal of Ecology 105, 95 (2017). 362 [41] J. M. LaMontagne, I. S. Pearse, D. F. Greene, W. D. Koenig, Nature Plants 6, 460 (2020). 347 363 364 [42] I. S. Pearse, J. M. LaMontagne, M. Lordon, A. L. Hipp, W. D. Koenig, New Phytologist 227, 1557 (2020). 366 [43] A. T. Moles, I. J. Wright, A. J. Pitman, B. R. Murray, M. Westoby, Ecography 32, 78 (2009). 367 [44] S. Tautenhahn, H. Heilmeier, L. Go, S. Klotz, C. Wirth, Ecography 31, 457 (2008). 368 [45] J. S. Clark, et al., Nature Communications 12, 1 (2021). 365 370 [46] P. T. Green, K. E. Harms, J. H. Connell, Proceedings of the National Academy of Sciences 111, 18649 (2014). 371 [47] J. S. Clark, D. M. Bell, M. C. Kwit, K. Zhu, Global Change Biology 20, 1979 (2014). 372 [48] T. Caignard, et al., Scientific Reports 7, 1 (2017). 369 374 [49] M. Bogdziewicz, M. Fernández-Martínez, J. M. Espelta, R. Ogaya, J. Penuelas, New Phytologist 227, 1073 (2020). 375 [50] D. Kelly, et al., Ecology Letters 16, 90 (2013). 373 11 376 377 [51] M. Bogdziewicz, M. Pesendorfer, E. E. Crone, C. Pérez-Izquierdo, R. Bonal, Ecology Letters 23, 1820 (2020). 379 [52] T. G. Whitham, G. J. Allan, H. F. Cooper, S. M. Shuster, Annual Review of Ecology, Evolution, and Systematics 51, 587 (2020). 380 [53] D. R. Strong, K. T. Frank, Annual Review of Environment and Resources 35, 1 (2010). 378 12 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 Acknowledgements We thank the National Ecological Observatory Network (NEON) for access to sites and vegetation structure data, W. Koening for useful data on crop production and S. Sitch for access to TRENDY products. The project has been funded since 1992 by grants to JSC from the National Science Foundation, most recently DEB-1754443, and by the Belmont Forum (1854976), NASA (AIST16-0052, AIST18-0063), and the Programme d’Investissement d’Avenir under project FORBIC (18-MPGA-0004) (Make Our Planet Great Again). Jerry Franklin’s data remain accessible through NSF LTER DEB-1440409. Puerto Rico data were funded by NSF grants, most recently, DEB 0963447 and LTREB 11222325. Data from the Andes Biodiversity and Ecosystem Research Group were funded by the Gordon and Betty Moore Foundation and NSF LTREB 1754647. MB was supported by grant no. 2019/35/D/NZ8/00050 from (Polish) National Science Centre, and Polish National Agency for Academic Exchange Bekker programme PPN/BEK/2020/1/00009/U/00001. Research by the USDA Forest Service and the the USGS was funded by these agencies. Any use of trade, firm, or product names does not imply endorsement by the U.S. Government. 396 397 398 399 400 401 Supplementary Materials Materials and Methods Supplementary Text Table S1 – S4 Fig S1 - S5 References (1 – 13) 13 402 Supplementary material 403 Materials and Methods 404 Fecundity data 430 Comprehensive data were needed to estimate climate effects due to the large variation in seed production. Masting, where large crop years exceed intervening years by orders of magnitude, is further complicated by spatio-temporal variation in habitat and climate. The many sources of variation means that biogeographic trends of interest can only be identified from broad coverage and large sample sizes, while accounting for individual tree condition, local habitat, and climate [45, 31]. The study uses crop-count and seed-trap data from the Masting Inference and Forecasting (MASTIF) project, including opportunistic data through the iNaturalist project MASTIF [56]. Observations include 12,053,732 tree-year observation from 748 species and 146,744 mature individuals. For crop-count data, an observation consists of tree species, diameter, crown shade class, the number of seeds/fruits or cones counted, and an estimate of the fraction of the total crop represented by the count. For seed-trap data, an observation consists of a count for a seed trap, trap location from an inventory plot where trees are measured and mapped, and trap area. Data models for the two data types in the MASTIF model include a beta-binomial distribution for crop counts (uncertainty in the count and in the crop-fraction estimate) and a redistribution model for seed counts (uncertainty in seed transport and in the count) [56]. Seed mass is taken as an average for the species, obtained from collections in our labs, supplemented with the TRY Plant Trait Database [57]. All observations provide estimates of ISP, including those on isolated trees. Because it requires seed production from a known area, only inventory plots offer estimates of CSP (Table S3). Together, ISP and CSP allow us to test how size-standardized seed production by individuals (ISP) and stand-level density of seed (CSP) vary with climate. As in all observational studies, geographic coverage is not uniform. The majority of sites are temperate (98%), while most observations (tree-years, 80%) and species (74%) are tropical. Sample sizes are included in Table S3 and their locations are shown in Figure S1. To clarify coverage, the distribution of data is displayed in each figure and detailed in the Supplement (Table S3, Figure S1). 431 Environmental and individual covariates 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 432 433 434 435 436 437 438 439 440 441 442 443 444 445 Predictors of fecundity for a given tree include diameter, crown class, climate, and soil and terrain covariates (Table S1). We included both linear and quadratic terms for diameter to allow changes of fecundity with tree size [31]. Crown class ranges from 1 (full sun) to 5 (full shade), following the protocol used in the National Ecological Observation Network (NEON) and the USDA Forest Inventory and Analysis (FIA) program. Climate variables include annual temperature (◦ C) from the previous year, and moisture deficit (summed monthly evapotranspiration minus precipitation, mm) from the previous and current years. Because seasonality varies globally (there is no uniform definition of a ’growing season’), we describe climate with annual norms for temperature and moisture-deficit. Moisture Í deficit is defined as (𝐷 𝑗 = 12 𝑘=1 𝑃𝐸𝑇 𝑗 𝑘 − 𝑃 𝑗 𝑘 for location 𝑗 and month 𝑘), which is the basis for the familiar Standardized Precipitation Evapotranspiration Index (SPEI) [58], but omitting here the standardization, which allows for comparisons between sites. To allow for changes in moisture access with tree size we included the interaction between moisture deficit and tree diameter. Climate variables were derived from CHELSA [59], Terra14 446 447 448 449 450 451 452 453 454 455 456 457 Climate [60], and local climate monitoring data where available. TerraClimate provides monthly but spatially coarse resolution [60] through 2020. CHELSA provides high spatial resolution (1 km) but it is not available after 2016. We used regression to project CHELSA climate forward based on Terraclimate, followed by calibration to local weather data where available. Details are available in [45]. Cation exchange capacity (CEC), used as an indicator of soil fertility, was obtained from soilGrid250 [61] as weighted mean from three soil depths: 0-5, 5-15 and 15-30 cm, weights are reported uncertainty. Slope and aspect were obtained from the global digital elevation model from the NASA shuttle radar topography mission [62] and, for latitudes above 61°, from USGS National Elevation Dataset [63] with a resolution at 30 meters for both products. The covariates for slope and aspect (𝑢 1 , 𝑢 2 , 𝑢 3 ) constitute a length-3 vector,    𝑢 𝑗,1 = sin(𝑠 𝑗 )  u 𝑗 = 𝑢 𝑗,2 = sin(𝑠 𝑗 ) sin(𝑎 𝑗 )   𝑢 𝑗,3 = sin(𝑠 𝑗 ) cos(𝑎 𝑗 )  (1) for slope 𝑠, where aspect 𝑎 is taken in radians [64]. Table S1: Covariates used to fit the MASTIF model and data sources. Subscripts are tree 𝑖, site 𝑗, and year 𝑡. Covariate Diameter (𝐺 𝑖 𝑗,𝑡 , 𝐺 𝑖2𝑗,𝑡 ) Crown class (𝐶𝑖 𝑗,𝑡 ) Moisture deficit (𝐷 𝑗 ) Deficit anomaly (𝐷 𝑗,𝑡 ) Temperature (𝑇 𝑗 ) Temperature anomaly (𝑇 𝑗,𝑡 ) 𝐷 𝑗 × 𝐺 𝑖 𝑗,𝑡 CEC 𝑗 (0 - 30cm) Slope, aspect (𝑢 1 𝑗 , 𝑢 2 𝑗 , 𝑢 3 𝑗 ) 458 459 460 461 462 463 464 465 466 467 468 469 Units cm, cm2 ordinal (class 1-5) mm mm ◦C ◦C mm × cm mmolc/kg radians Data source MASTIF MASTIF terraClimate, CHELSA terraClimate, CHELSA terraClimate, CHELSA terraClimate, CHELSA soilgrid250m DEM, USGS Model inference with MASTIF The MASTIF model used to analyse seed trap/crop count data is detailed in [56]. This (hierarchical) state-space, auto-regressive model accommodates dependence between trees and within trees over years through a joint analysis. For each tree 𝑖 and year 𝑡 there is a mean fecundity estimate 𝑓ˆ𝑖,𝑡 = 𝜌ˆ𝑖,𝑡 𝜓ˆ 𝑖,𝑡 that is the product of conditional fecundity 𝜓ˆ and maturation probability 𝜌ˆ𝑖,𝑡 . The model for conditional fecundity is given by log 𝜓𝑖𝑡 = x′𝑖,𝑡 𝜷 (𝑥) + 𝛽𝑖(𝑤) + 𝛾𝑔[𝑖],𝑡 + 𝜖𝑖,𝑡 , where x𝑖𝑡 is a design vector holding climate, soils, local crowding, and individual attributes (sTable S1), 𝜷 (𝑥) are fixed-effects coefficients, 𝛽𝑖(𝑤) is the random effect for tree 𝑖, 𝛾𝑔[𝑖],𝑡 are year effects that are random across groups 𝑔 and fixed for year 𝑡, and 𝜖𝑖,𝑡 is Gaussian error. The group membership for a tree 𝑖 is 𝑔[𝑖], which is defined by species-ecoregions [56]. Conditional log fecundity 𝜓 is censored at zero to allow for the immature state and for failed seed crops in larger individuals,  0 𝜓≤1 (2) 𝐸[ 𝑓] = 𝜓 𝜓>1 15 478 This censoring means that seed production requires the potential to produce at least one seed and follows the same approach as a Tobit model for the linear scale, which is censored at zero rather than one. Fecundity can be calculated as mass of seeds, and it can be standardized for tree basal area as in eqn 1. The posterior distribution includes parameters and latent variables for maturation state and tree-year seed production. Posterior simulation uses direct sampling and Metropolis and Hamiltonian Markov Chain (HMC) updates within Gibbs. Model structure and methodology was implemented with R (version 4.0, [65]) and the R package Mast Inference and Forecasting (MASTIF), detailed in [56]. 479 Uncertainty in fecundity estimates 470 471 472 473 474 475 476 477 480 481 482 483 484 485 486 487 488 489 We evaluated weighted mean fecundity at the individual and plot scales (CSP), where weights accommodate year-to-year uncertainty for an individual tree and tree-to-tree uncertainty for a stand. For individual and stand-level CSP we included only trees > 7 cm in diameter, i.e., at least as larges as the smallest measured size in inventory data. Individual mean fecundity was obtained as Í 𝑤 𝑓ˆ ˆ𝑓𝑖 𝑗 𝑠 = 𝑡Í 𝑖 𝑗 𝑠,𝑡 𝑖 𝑗 𝑠,𝑡 (3) 𝑡 𝑤 𝑖 𝑗 𝑠,𝑡 where the weight 𝑤 𝑖 𝑗 𝑠,𝑡 is the inverse of the predictive coefficient of variation for the estimate, 𝑤 𝑖 𝑗 𝑠,𝑡 = 𝐶𝑉𝑖−1 𝑗 𝑠,𝑡 . This is used rather than the predictive variance, because the mean tends to scale with the variance such that a variance weight would have the undesirable property of downweighting the important large values while up-weighting the low values, which are dominated by noise. Community seed production (CSP) was evaluated from the individual means Í 1 𝑖𝑠 𝑤 𝑖 𝑗 𝑠 𝑓ˆ𝑖 𝑗 𝑠 Í (4) 𝐶𝑆𝑃 𝑗 = 𝐴𝑗 𝑖𝑠 𝑤 𝑖 𝑗 𝑠 492 where 𝐴 𝑗 is plot area, and 𝑤 𝑖 𝑗 𝑠 is the inverse of the coefficient of variation evaluated as the root mean predictive variance for tree 𝑖 𝑗 𝑠 divided by the the mean prediction for that individual. Because CSP requires a plot area, only trees on inventory plots are included in the CSP analysis. 493 Net Primary Production 490 491 494 495 496 497 498 499 500 501 502 503 504 505 506 507 We extracted Net Primary Production (NPP) from the Moderate Resolution Imaging Spectroradiometer (MODIS) product MOD17 at 500 m resolution (MOD17A3HGFv006, [66]). For 2000 to 2020, we merged yearly CSP estimates with NPP from matching site years, which are available from 2000 to 2020. Because seed production data span the interval 1959 to 2020, we used the location-specific mean NPP values for the limited number of earlier years. Because MODIS NPP can depend on uneven cloud coverage, we compared MODIS with NPP predictions from DGVMs in the TRENDY project [67], using the S3 experiment. For each site we averaged NPP from 11 models (CABLE-POP, CLASSIC, CLM5.0, ISAM, JSBACH, JULES, LPJ-GUESS, LPX, OCN, ORCHIDEE, ORCHIDEE-CNP) for all sites and fitted them to the same climate variables (temperature, moisture deficit) used for ISP and CSP (Table S2). The two NPP products show similar main effects, but differ in temperature × moisture interaction, which is negative for MODIS and positive for the aggregated DGVM. Despite this difference in the interaction term, the main effects dominated such that surfaces show the same trends (Figure S3). Thus, we included only MODIS results in S4. 16 Table S2: Coefficients for climate on NPP MODIS and NPP DGVM products. 𝑟 2 for NPP MODIS = 0.48, NPP DGVM = 0.52. Variable Climate effects on NPP (MODIS) 𝐼𝑛𝑡𝑒𝑟𝑐𝑒 𝑝𝑡 𝑇 𝐷 𝑇×𝐷 Climate effects on NPP (DGVMs TRENDY) 𝐼𝑛𝑡𝑒𝑟𝑐𝑒 𝑝𝑡 𝑇 𝐷 𝑇×𝐷 NPP effect (MODIS) on log𝑒 ISP 𝐼𝑛𝑡𝑒𝑟𝑐𝑒 𝑝𝑡 𝑁 𝑃𝑃 NPP effect (DGVMs TRENDY) on log𝑒 ISP 𝐼𝑛𝑡𝑒𝑟𝑐𝑒 𝑝𝑡 𝑁 𝑃𝑃 NPP effect (MODIS) on log𝑒 CSP 𝐼𝑛𝑡𝑒𝑟𝑐𝑒 𝑝𝑡 𝑁 𝑃𝑃 NPP effect (DGVMs TRENDY) on log𝑒 CSP 𝐼𝑛𝑡𝑒𝑟𝑐𝑒 𝑝𝑡 𝑁 𝑃𝑃 508 Parameter Estimate SE P-value 𝛽 𝑁,𝑇 𝛽 𝑁,𝐷 𝛽 𝑁,𝐷𝑇 3.52e-01 2.46e-02 < 2e-16 1.54e-02 1.92e-03 4.99e-15 -1.81e-04 3.35e-05 8.41e-08 -1.11e-05 2.65e-06 2.99e-05 - 1.455e-01 2.2e-02 7.71e-11 3.19e-02 1.72e-03 < 2e-16 -3.25e-04 3.01e-05 < 2e-16 7.36e-06 2.38e-06 0.00205 𝛽 𝑁,𝑇 𝛽 𝑁,𝐷 𝛽 𝑁,𝐷𝑇 𝛽𝑓𝑁 3.98 2.06 0.027 0.024 <2e-16 <2e-16 𝛽𝑓𝑁 4.88 1.64 0.037 0.047 <2e-16 <2e-16 𝛽𝑓𝑁 8.70 2.70 0.38 0.44 <2e-16 3.31e-09 𝛽𝑓𝑁 9.26 2.21 0.44 0.58 <2e-16 1.77e-4 Supplementary Tables Table S3: Numbers of species, stands, trees, and tree-years for ISP analysis and complete inventories for CSP analysis by tropical and temperate regions. Complete inventories include all trees within a mapped plot and are needed to determine seeds per area in CSP. Because not all inventory plots use the same minimum diameter, the latter is based on trees > 7 cm. Floristic Region Tropical Temperate Complete Species Sites Tree-years Trees inventories 559 64 9,723,438 85,261 47 194 3506 2,330,294 61,461 204 17 Table S4: Coefficients for climate effect on individual (ISP), community fecundity (CSP). ISP and CSP fecundity are fitted on a natural log scale. 𝑟 2 for ISP = 0.2, CSP = 0.15. Variable Parameter Climate effects on log𝑒 ISP 𝐼𝑛𝑡𝑒𝑟𝑐𝑒 𝑝𝑡 𝑇 𝛽 𝑓 ,𝑇 2 𝑇 𝛽 𝑓 ,𝑇 2 𝐷 𝛽 𝑓 ,𝐷 2 𝐷 𝛽 𝑓 ,𝐷 2 𝑇×𝐷 𝛽 𝑓 ,𝐷𝑇 Climate effects on log𝑒 CSP 𝐼𝑛𝑡𝑒𝑟𝑐𝑒 𝑝𝑡 𝑇 𝛽 𝑓 ,𝑇 𝑇2 𝛽 𝑓 ,𝑇 2 𝐷 𝛽 𝑓 ,𝐷 𝐷2 𝛽 𝑓 ,𝐷 2 𝑇×𝐷 𝛽 𝑓 ,𝐷𝑇 18 Estimate SE P-value 4.64e+00 4.93e-02 1.78e-01 6.01e-03 -5.60e-03 1.770e-04 2.72e-03 4.80e-05 -1.12e-07 1.14e-08 -1.84e-04 1.73e-06 <2e-16 <2e-16 <2e-16 <2e-16 <2e-16 <2e-16 9.88e+00 9.96e-02 -2.38e-03 9.21e-04 2.87e-08 -1.19e-04 5.61e-01 <2e-16 7.88e-02 0.21 2.82e-03 0.40 7.16e-04 0.20 2.20e-07 0.90 4.05e-05 3.60e-3 Supplementary Figures (b) (c) (e) (f) Latitude (a) (d) Latitude 509 Longitude Longitude Longitude sample size 1e+00 1e+04 1e+06 Figure S1: MASTIF network data, including longitudinal (in green) and opportunistic (in orange) observations in North America (a), Europe (b), Asia (c), South and Central America (d), Africa (e) and Oceania (f). Number of observations are reported in Table S3. 19 Figure S2: Climate responses for ISP (seed mass per basal area) (a, b, c) and stand-level CSP, as g ha−1 (d, e, f) showing marginal responses to temperature (a and d) and moisture deficit (d and e) with observations (dots) and the fitted model, and interactions between temperature and moisture deficit (c and f). Coefficient are reported in Table S4. Low and high values used for conditional plots in (c and f), labelled as Moist (𝐷 = −1500 mm) and Dry (𝐷 = −50 mm). Due to large sample size, confidence intervals around lines in (a, b, c) are not distinct from the predictive mean. Temperature and moisture deficit correspond here to a mean annual value for each sites. Figure S3: Climate response for NPP from MODIS product (a) and DGVM product from TRENDY DGVM products 20 Figure S4: Relationships between NPP from MODIS and individual (standardized) fecundity ISP (a) and stand CSP (b), both positive (𝑝 < 0.00001) and both accounting for little of the variability (𝑟 2 = 0.05 and 0.13, respectively). Coefficient are reported in Table S4 Figure S5: Distribution of (a) ISP (g seed per m2 basal area) and (b) CSP (g seed per ha basal area) fecundities. Black dotted lines represent the quantile at 2.5 and 97.5%. 21 510 References in Supplementary materials 511 [54] J. S. Clark, et al., Nature Communications 12, 1 (2021). 512 [55] T. Qiu, et al., Proceedings of the National Academy of Sciences 118 (2021). 513 [56] J. S. Clark, C. L. Nuñez, B. Tomasek, Ecological Monographs 89, 1 (2019). 514 [57] J. Kattge, et al., Global Change Biology 26, 119 (2020). 516 [58] S. M. Vicente-Serrano, S. Beguería, J. I. López-Moreno, Journal of Climate 23, 1696 (2010). 517 [59] D. N. Karger, et al., Scientific Data 4, 1 (2017). 515 519 [60] J. T. Abatzoglou, S. Z. Dobrowski, S. A. Parks, K. C. Hegewisch, Scientific Data 5, 170191 (2018). 520 [61] T. Hengl, et al., PLoS ONE 12 (2017). 521 [62] T. G. Farr, et al., Reviews of Geophysics 45 (2007). 518 523 [63] D. Gesch, et al., Photogrammetric Engineering and Remote Sensing (American Society for Photogrammetry and Remote Sensing, 2002), vol. 68, pp. 5–11. 524 [64] J. S. Clark, Ecological Monographs 60, 135 (1990). 522 526 [65] R Core Team, R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, Vienna, Austria, 2020). 527 [66] S. W. Running, M. Zhao . 528 [67] S. Sitch, et al., Biogeosciences 12, 653 (2015). 525 22