S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E E N V I RO N M E N TA L S T U D I E S Copyright © 2023 The Authors, some Global beta-diversity of angiosperm trees is shaped by rights reserved; exclusive licensee Quaternary climate change American Association for the Advancement of Science. No claim to Wu-Bing Xu1,2,3*, Wen-Yong Guo1,2,4, Josep M. Serra-Diaz5, Franziska Schrodt6, original U.S. Government Wolf L. Eiserhardt2,7, Brian J. Enquist8,9, Brian S. Maitner8, Cory Merow10, Cyrille Violle11, Works. Distributed Madhur Anand12, Michaël Belluau13, Hans Henrik Bruun14, Chaeho Byun15, Jane A. Catford16, under a Creative Bruno E. L. Cerabolini17, Eduardo Chacón-Madrigal18, Daniela Ciccarelli19, J. Hans C. Cornelissen20, Commons Attribution Anh Tuan Dang-Le21, Angel de Frutos3, Arildo S. Dias22, Aelton B. Giroldo23, License 4.0 (CC BY). Alvaro G. Gutiérrez24,25, Wesley Hattingh26, Tianhua He27,28, Peter Hietz29, Nate Hough-Snee30, Steven Jansen31, Jens Kattge3,32, Benjamin Komac33, Nathan J. B. Kraft34, Koen Kramer35,36, Sandra Lavorel37, Christopher H. Lusk38, Adam R. Martin39, Ke-Ping Ma40, Maurizio Mencuccini41,42, Sean T. Michaletz43, Vanessa Minden44,45, Akira S. Mori46, Ülo Niinemets47, Yusuke Onoda48, Renske E. Onstein3,49, Josep Peñuelas42,50, Valério D. Pillar51, Jan Pisek52, Matthew J. Pound53, Bjorn J. M. Robroek54, Brandon Schamp55, Martijn Slot56, Miao Sun2,57, Ênio E. Sosinski Jr.58, Nadejda A. Soudzilovskaia59, Nelson Thiffault60, Peter M. van Bodegom61, Fons van der Plas62, Jingming Zheng63, Jens-Christian Svenning1,2,64, Alejandro Ordonez1,2,64* As Earth’s climate has varied strongly through geological time, studying the impacts of past climate change on biodiversity helps to understand the risks from future climate change. However, it remains unclear how paleo- climate shapes spatial variation in biodiversity. Here, we assessed the influence of Quaternary climate change on spatial dissimilarity in taxonomic, phylogenetic, and functional composition among neighboring 200-kilometer cells (beta-diversity) for angiosperm trees worldwide. We found that larger glacial-interglacial temperature change was strongly associated with lower spatial turnover (species replacements) and higher nestedness (rich- ness changes) components of beta-diversity across all three biodiversity facets. Moreover, phylogenetic and functional turnover was lower and nestedness higher than random expectations based on taxonomic beta-di- versity in regions that experienced large temperature change, reflecting phylogenetically and functionally se- lective processes in species replacement, extinction, and colonization during glacial-interglacial oscillations. Our results suggest that future human-driven climate change could cause local homogenization and reduction in taxonomic, phylogenetic, and functional diversity of angiosperm trees worldwide. INTRODUCTION knowledge about the impacts of past rapid climate change on One of the main challenges for ecology in the Anthropocene is to spatial patterns of these multiple facets of beta-diversity at large understand how ongoing and near-future climate change reshapes scales (5, 13–15). Examining and comparing paleoclimatic legacies the distribution of biodiversity and ecosystem functioning (1, 2). As on taxonomic, phylogenetic, and functional beta-diversity are Earth’s climate has varied strongly through geological time, study- crucial for understanding the effects of ongoing climate change ing the impacts of past climate change on current biodiversity pat- on future biodiversity patterns. terns provides an opportunity to understand the risks emerging In previous studies, taxonomic beta-diversity was usually de- from ongoing anthropogenic climate change (3–5). Notably, composed into two additive components: spatial species turnover Earth has experienced major glaciations over the past 2.6 million and nestedness of assemblages (16). Variation in these two compo- years (the Quaternary), which have left lasting imprints on many nents is considered to be the consequence of different ecological organismal groups, including trees, through effects on speciation, and evolutionary processes (16–20). Spatial species turnover reflects extinction, and range shifts [as reviewed in (6)]. Although an in- species replacement between sites, while nestedness shows how de- creasing number of studies have assessed the impacts of paleocli- pauperate assemblages are subsets of richer ones, reflecting species mate change on biodiversity, most of these studies focused on losses or gains across sites (16). Compared to regions that experi- patterns of taxonomic diversity of local assemblages (alpha-diver- enced unstable climates over glacial-interglacial cycles, regions sity) [e.g., (7–10)]. Biodiversity, however, includes other comple- with a relatively stable climate (usually low-latitude areas) have mentary dimensions, namely, phylogenetic and functional many more species persisting in situ and experience lower rates of diversity, and also encompasses variation in assemblage composi- species extinction and higher rates of speciation (21), which has led tion across sites (beta-diversity) (11–14). Taxonomic, phylogenetic, to more species with small ranges (22–25). As a result, spatial and functional beta-diversity provide information on how compo- species turnover is expected to be the dominant component of sitions of species identities, evolutionary history, and functional beta-diversity in regions with stable past climates (Fig. 1B). By com- traits vary across spatial gradients (Fig. 1). There is limited parison, regions with unstable past climates experience higher local Xu et al., Sci. Adv. 9, eadd8553 (2023) 5 April 2023 1 of 18 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E extinctions during climatically harsh periods such as glaciations, Fig. 1B) (33, 34). Furthermore, human pressures can decrease and many extant species are colonizers from glacial refugia (5, species turnover by reducing the distributions of small-ranged 21). Because species’ postglacial recolonization may lag the emer- species and expanding the distributions of large-ranged species gence of suitable climatic conditions (26–28), some sites, particular- (36). They may also increase the nestedness of assemblages when ly those far from the glacial refugia, are likely to lose more species a high proportion of species is lost due to habitat conversion in a than other sites during range expansion and contraction through subset of sites within a region. glacial-interglacial cycles, leading to patterns of nestedness in as- There is some evidence for the paleoclimatic legacies on conti- semblage compositions (Fig. 1B) (17, 19, 29). nental and global patterns of both turnover and nestedness compo- Alternatively, spatial patterns of both turnover and nestedness nents of taxonomic beta-diversity in multiple organismal groups, components of beta-diversity may be shaped by contemporary en- e.g., amphibians, freshwater fish, mammals, birds, and plants (14, vironmental conditions (Fig. 1B) (17, 18, 29, 30). In tropical areas 15, 17–19, 29, 37). However, only a few studies have simultaneously with warm and wet conditions, a large number of species have ac- assessed paleoclimatic legacies on large-scale patterns of taxonomic, cumulated, and these species are generally specialized with small phylogenetic, and functional beta-diversity (14, 15, 37–39), partic- range size favored by low climatic seasonality (31), leading to a ularly for megadiverse organismal groups, such as trees. Specifically, high spatial species turnover (13, 32). Topographic heterogeneity it remains unclear whether phylogenetic and functional beta-diver- may also drive high species turnover through geographical isolation sity have been influenced by paleoclimate change through processes limiting dispersal (12). In addition, environmental heterogeneity, beyond those affecting taxonomic beta-diversity. Like taxonomic including topography-driven heterogeneity, can drive species turn- beta-diversity, phylogenetic and functional beta-diversity can be de- over by selecting species with different environmental requirements composed into turnover and nestedness components. Phylogenetic at each local site (termed “habitat filtering”) (18, 33, 34). This filter- and functional turnover reflects the replacement of evolutionary ing would be more apparent when species have narrow niche lineages and traits, and the nestedness captures losses/gains in lin- breadths [usually in tropical regions (35)]. Alternatively, environ- eages and functions across sites (14, 15, 37–41). Because species re- mental gradients can drive the nestedness structure of assemblages placements or losses/gains among assemblages can occur among by gradually filtering out species by tolerance limits along gradients species that are either closely related and have similar traits or are of increasingly stressful conditions (termed “species sorting”; distantly related and dissimilar in traits (40, 41), regions with the 1 Center for Biodiversity Dynamics in a Changing World (BIOCHANGE), Department of Biology, Aarhus University, DK-8000 Aarhus C, Denmark. 2Section for Ecoinformatics and Biodiversity, Department of Biology, Aarhus University, DK-8000 Aarhus C, Denmark. 3German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig, 04103 Leipzig, Germany. 4Zhejiang Tiantong Forest Ecosystem National Observation and Research Station and Research Center for Global Change and Complex Ecosys- tems, School of Ecological and Environmental Sciences, East China Normal University, 200241 Shanghai, P.R. China. 5Université de Lorraine, AgroParisTech, INRAE, Silva, Nancy, France. 6School of Geography, University of Nottingham, Nottingham, NG7 2RD, UK. 7Royal Botanic Gardens, Kew, Surrey TW9 3AE, UK. 8Department of Ecology and Evolutionary Biology, University of Arizona, Tucson, AZ 85721, USA. 9The Santa Fe Institute, 1399 Hyde Park Rd., Santa Fe, NM 87501, USA. 10Department of Ecology and Evolutionary Biology, University of Connecticut, Storrs, CT 06269, USA. 11CEFE, Univ Montpellier, CNRS, EPHE, IRD, Montpellier, France. 12School of Environmental Sciences, University of Guelph, Guelph, Ontario, Canada. 13Centre for Forest Research, Département des sciences biologiques, Université du Québec à Montréal, P.O. Box 8888, Centre-ville station, Montréal, QC H3C 3P8, Canada. 14Department of Biology, University of Copenhagen, 2100 Copenhagen Ø, Denmark. 15Department of Biological Sciences and Biotechnology, Andong National University, Andong, 36729, Korea. 16Department of Geography, King’s College London, London WC2B 4BG, UK. 17 Department of Biotechnologies and Life Sciences, University of Insubria, Via Dunant 3, 21100 Varese, Italy. 18Herbario Luis Fournier Origgi, Centro de Investigación en Biodiversidad y Ecología Tropical (CIBET), Universidad de Costa Rica, San Pedro de Montes de Oca, 11501-2060 San José, Costa Rica. 19Department of Biology, Uni- versity of Pisa, Via Luca Ghini 13, 56126 Pisa, Italy. 20Systems Ecology, A-LIFE, Faculty of Science, Vrije Universiteit, 1081 HV Amsterdam, Netherlands. 21Faculty of Biology - Biotechnology, University of Science - VNUHCM, 227 Nguyen Van Cu, District 5, 700000 Ho Chi Minh City, Vietnam. 22Goethe University, Institute for Physical Geography, Altenhöferallee 1, 60438 Frankfurt am Main, Germany. 23Departamento de Ensino, Instituto Federal de Educação, Ciências e Tecnologia do Ceará - IFCE campus Crateús, Avenida Geraldo Barbosa Marques, 567, 63708-260 Crateús, Brazil. 24Departamento de Ciencias Ambientales y Recursos Naturales Renovables, Facultad de Ciencias Agro- nómicas, Universidad de Chile, Santa Rosa 11315, La Pintana, Santiago, Chile. 25Institute of Ecology and Biodiversity (IEB), Santiago, Chile. 26Global Systems and Analytics, Nova Pioneer, Paulshof, Gauteng, South Africa. 27School of Molecular and Life Sciences, Curtin University, P.O. Box U1987, Perth, WA 6845, Australia. 28College of Science, Health, Engineering and Education, Murdoch University, Murdoch, WA, Australia. 29Institute of Botany, University of Natural Resources and Life Sciences Vienna, 1180 Vienna, Austria. 30Meadow Run Environmental, Leavenworth, WA 98826, USA. 31Institute of Systematic Botany and Ecology, Ulm University, 89081 Ulm, Germany. 32Max Planck Institute for Biogeochemistry, Hans Knöll Str. 10, 07745 Jena, Germany. 33Andorra Recerca + Innovació, AD600 Sant Julià de Lòria (Principat d'), Andorra. 34 Department of Ecology and Evolutionary Biology, University of California, Los Angeles, Los Angeles, CA 90095, USA. 35Wageningen University, Forest Ecology and Management Group, Droevendaalsesteeg 4, 6700AA Wageningen, Netherlands. 36Land Life Company, Mauritskade 63, 1092AD, Amsterdam, Netherlands. 37 Laboratoire d’Ecologie Alpine, LECA, UMR UGA-USMB-CNRS 5553, Université Grenoble Alpes, CS 40700, 38058 Grenoble Cedex 9, France. 38Environmental Research Institute, University of Waikato, Hamilton, New Zealand. 39Department of Physical and Environmental Sciences, University of Toronto Scarborough, 1265 Military Trail, M1C 1A4 Toronto, ON, Canada. 40State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, Chinese Academy of Sciences, Beijing 100093, China. 41 ICREA, Barcelona, 08010, Spain. 42CREAF, Cerdanyola del Vallès, Barcelona 08193, Catalonia, Spain. 43Department of Botany and Biodiversity Research Centre, University of British Columbia, Vancouver, BC V6T 1Z4, Canada. 44Department of Biology, Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussels, Belgium. 45Institute for Biology and Environmental Sciences, University of Oldenburg, 26129 Oldenburg, Germany. 46Research Center for Advanced Science and Technology, the University of Tokyo, 4-6-1 Komaba, Meguro, Tokyo 153-8904, Japan. 47Estonian University of Life Sciences, Kreutzwaldi 1, 51006 Tartu, Estonia. 48Division of Forest and Biomaterials Science, Grad- uate School of Agriculture, Kyoto University, Oiwake, Kitashirakawa, Kyoto, 606-8502 Japan. 49Naturalis Biodiversity Center, Darwinweg 2, 2333CR Leiden, Netherlands. 50 CSIC, Global Ecology Unit CREAF- CSIC-UAB, Bellaterra, Barcelona 08193, Catalonia, Spain. 51Department of Ecology, Universidade Federal do Rio Grande do Sul, Porto Alegre, 91501-970, Brazil. 52Tartu Observatory, University of Tartu, Observatooriumi 1, Tõravere, 61602 Tartumaa, Estonia. 53Department of Geography and Environmental Sciences, Northumbria University, Newcastle upon Tyne NE1 8ST, UK. 54Aquatic Ecology and Environmental Biology, Faculty of Science, Radboud Institute for Biological and Environmental Sciences, Radboud University Nijmegen, 6525 AJ Nijmegen, Netherlands. 55Department of Biology, Algoma University, Sault Ste. Marie, Ontario, P6A 2G4, Canada. 56Smithsonian Tropical Research Institute, Apartado 0843-03092, Balboa, Ancón, Republic of Panama. 57National Key Laboratory for Germplasm Innovation & Utilization of Horticultural Crops, Huazhong Agricultural University, Wuhan 430070, China. 58Embrapa Clima Temperado, 96010-971 Pelotas, RS, Brazil. 59Centre for Environmental Sciences, Hasselt University, Martelarenlaan 42, 3500 Hasselt, Belgium. 60Natural Resources Canada, Canadian Wood Fibre Centre, 1055 du P.E.P.S., P.O. Box 10380, Stn. Sainte-Foy, Quebec, QC G1V 4C7, Canada. 61Institute of Environmental Sciences, Leiden University, 2333 CC Leiden, Netherlands. 62Plant Ecology and Nature Conservation Group, Wageningen University, Netherlands. 63Beijing Key Laboratory for Forest Resources and Ecosystem Processes, Beijing Forestry University, Beijing, 100083, China. 64Center for Ecological Dynamics in a Novel Biosphere (ECONOVO), Department of Biology, Aarhus University, DK-8000 Aarhus C, Denmark. *Corresponding author. Email: wbingxu@gmail.com (W.-B.X.); alejandro.ordonez@bio.au.dk (A.O.) Xu et al., Sci. Adv. 9, eadd8553 (2023) 5 April 2023 2 of 18 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E Fig. 1. Hypothetical effects of past climatic stability and current environmental conditions on spatial turnover and nestedness components of taxonomic, phylogenetic, and functional beta-diversity. (A) Beta-diversity was measured as intraregional multiple-site dissimilarity within each moving window of 25 (5 × 5) grid cells of 200 km by 200 km worldwide. Total taxonomic, phylogenetic, and functional beta-diversity were decomposed into spatial turnover and nestedness com- ponents, respectively. Phylogenetic and functional turnover and nestedness were further compared with random expectations based on observed taxonomic beta-di- versity and site-specific regional species pools and measured as deviations. The map shows the temperature anomaly since the LGM. (B) The turnover component is expected to be high in regions with stable past climate and/or benign current environmental conditions. In contrast, the nestedness component is expected to be high in regions with unstable past climate and/or stressful conditions, owing to both evolutionary and ecological processes. (C and D) Even if two regions have the same mag- nitudes of species turnover or nestedness, their phylogenetic and functional turnover or nestedness can be different because species replacements or losses/gains can occur among species that are closely related and similar in traits or distantly related and dissimilar in traits. In regions with unstable past climate and/or stressful con- ditions, deviations of phylogenetic and functional turnover from random expectations based on taxonomic beta-diversity are expected to be low if the replaced species come from closely related young lineages (shallow branches) and have similar trait values (colors) (C), and deviations of phylogenetic and functional nestedness are expected to be high if species losses/gains are phylogenetically and functionally selective, targeting closely related species and affiliated specific trait values (D). For clarity, only two sites are illustrated in each example region. Geometric shapes represent species and colors of dots functional trait values. same magnitude of taxonomic turnover or nestedness can have dif- selective processes are involved in species replacements, losses, ferent phylogenetic and functional turnover or nestedness (Fig. 1, C and gains, which cumulatively shape phylogenetic and functional and D). If species replacements or losses/gains are phylogenetically turnover and nestedness among assemblages (Fig. 1). and functionally selective, then species that are replaced or lost/ Past climatic stability can drive phylogenetic and functional gained would be more closely (or distantly) related and similar turnover and nestedness to deviate from those randomly expected (or dissimilar) in functional traits compared to those expected given taxonomic beta-diversity (Fig. 1, C and D). Regions with based on random processes. In this study, we propose a null stable climates over glacial-interglacial cycles are usually the model to calculate random expectations of phylogenetic and func- centers of paleoendemic species (42, 43), which would lead to tional turnover and nestedness (i.e., expectations when a random higher phylogenetic turnover than randomly expected from taxo- group of species are replaced or lost/gained) for each given region nomic beta-diversity. In contrast, few and relatively young based on its observed taxonomic beta-diversity and species pool endemic species are expected to be distributed in regions with un- (see the “Null model” section in Materials and Methods for stable past climates (43, 44), which would result in lower phyloge- details). By comparing observed and randomly expected beta-diver- netic turnover than random expectations (Fig. 1C). These sity, we can reveal whether phylogenetically and functionally neoendemic species are expected to have similar functional trait Xu et al., Sci. Adv. 9, eadd8553 (2023) 5 April 2023 3 of 18 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E values if trait divergence is slow (45). Therefore, regions with unsta- assessed their relative roles in shaping present patterns of beta-di- ble past climates may have lower functional turnover than random versity in angiosperm trees. expectations, particularly when traits are phylogenetically con- We hypothesize that both past climate stability and contempo- served. However, because certain traits may not necessarily be phy- rary environmental conditions affect spatial patterns of both turn- logenetically conserved, functional turnover may show a different over and nestedness components of beta-diversity across three pattern than phylogenetic turnover. At the same time, because biodiversity facets, as illustrated in Fig. 1. We expect a strong influ- certain functional traits, e.g., specific leaf area in plants, are ence of temperature anomaly since the LGM because of the impor- closely associated with species adaptation to environmental condi- tance of temperature for species distributions and the large tions (46), deviations of functional turnover from random expecta- temperature change during glacial-interglacial oscillations. We tions may also be strongly influenced by contemporary also hypothesize that phylogenetic and functional turnover would environmental conditions. For example, species from different as- be lower, and the nestedness would be higher than randomly ex- semblages in stressful conditions tend to share similar functional pected from taxonomic beta-diversity in regions with unstable trait values that can improve plant performance under local envi- past climates as a result of phylogenetically and functionally selec- ronmental conditions (47). By comparison, regions with unstable tive processes in species replacement, extinction, and colonization past climates may have higher phylogenetic and functional nested- during glacial-interglacial oscillations (Fig. 1, C and D). ness than random expectations (Fig. 1D). This pattern can emerge when glaciation-driven extinction and postglacial colonization are phylogenetically and functionally selective, targeting species that RESULTS are closely related and affiliated with specific functional trait Global patterns of taxonomic, phylogenetic, and functional values (48, 49). For instance, extinctions of temperate trees during beta-diversity in angiosperm trees glaciation periods are disproportionately common among cold-in- Globally, the total beta-diversity and its turnover and nestedness tolerant species (48, 50), while species’ dispersal capacity has been components exhibited strikingly different spatial patterns across shown to determine the extent to which they recolonize their cli- all three biodiversity facets (Fig. 2). There was relatively low matically suitable areas after glaciations (27). These and other phy- spatial variation in total beta-diversity, with low values mainly con- logenetically and functionally selective processes should drive a centrated in the eastern United States, southwestern Europe, and larger difference in phylogenetic and functional diversity (i.e., eastern South America (Fig. 2, A to C). By contrast, turnover and stronger nestedness structure) between species-rich and species- nestedness components showed clear but opposing latitudinal pat- poor assemblages than those expected based on random species terns (Fig. 2, D to I, and fig. S1). The turnover component decreased gain and loss processes (Fig. 1D). toward the poles, whereas the nestedness component increased While both past climate stability and contemporary environ- toward the poles. On average, the turnover component contributed mental conditions may shape taxonomic, phylogenetic, and func- more to total beta-diversity (Fig. 2, J to L). However, the nestedness tional turnover and nestedness, whether and how these processes component contributed more than half in high latitudinal areas in operate and their relative importance are poorly understood. No North America and Eurasia, as well as in smaller areas elsewhere, studies have so far simultaneously explored global patterns of mostly in the vicinity of deserts. This pattern was especially pro- these three facets of beta-diversity and assessed if phylogenetic nounced for phylogenetic and functional beta-diversity (Fig. 2, J and functional beta-diversity have been influenced by paleoclimate to L). change through processes beyond those affecting taxonomic beta- Across three biodiversity facets, spatial patterns of total beta-di- diversity. Because trees and tree diversity play crucial roles in terres- versity and its components of turnover and nestedness are generally trial ecosystems, global biodiversity, and people (51, 52), it is partic- consistent (Fig. 2), as indicated by strong positive correlations ularly important to clarify global patterns of tree beta-diversity and between taxonomic, phylogenetic, and functional beta-diversity their drivers. Here, we combined the most extensive global database (fig. S2). On average, the total taxonomic beta-diversity was of angiosperm tree species’ distributions (53) with information greater than the total phylogenetic beta-diversity, which, in turn, about their phylogenetic relationships and functional traits to quan- exceeded the total functional beta-diversity; the same pattern was tify intraregional compositional dissimilarity patterns within observed for the turnover component (fig. S2, A to F). However, moving windows of 25 (5 × 5) grid cells of 200 km by 200 km. phylogenetic and functional nestedness were greater than taxonom- We used Sørensen-based multiple-site dissimilarity to calculate ic nestedness, resulting in a higher relative contribution of the nest- the total beta-diversity and partition it into components of turnover edness component to total phylogenetic and functional beta- (Simpson-based dissimilarity) and nestedness (difference between diversity as compared to taxonomic beta-diversity (fig. S2, J to L). Sørensen and Simpson dissimilarities) for all three biodiversity facets (16, 54). We then used a null model to calculate the deviations Effects of past climate stability and current environmental of observed phylogenetic and functional turnover and nestedness conditions from random expectations given the observed taxonomic beta-di- The environmental variables affecting total beta-diversity differ versity and site-specific regional species pools. Last, we used differ- from those affecting the turnover and nestedness components ences in annual temperature and precipitation between the present (Fig. 3, fig. S3, and table S1). Across all three biodiversity facets, and the Last Glacial Maximum (LGM; ~21,000 years ago) (temper- total beta-diversity was positively associated with precipitation sea- ature and precipitation anomalies) to represent past climate stability sonality and negatively associated with mean annual precipitation and current climatic, topographic, and human pressure variables to (Fig. 3). Precipitation seasonality showed the strongest correlation represent contemporary environmental conditions and ultimately with total beta-diversity (table S1). The turnover and nestedness components of all three biodiversity facets were most strongly Xu et al., Sci. Adv. 9, eadd8553 (2023) 5 April 2023 4 of 18 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E Fig. 2. Global patterns of taxonomic, phylogenetic, and functional beta-diversity of angiosperm trees. Total beta-diversity (A to C) and its components of turnover (D to F) and nestedness (G to I) and the proportion of total beta-diversity contributed by nestedness (J to L) are shown for three biodiversity facets, respectively. In (J) to (L), the grid cells with more than 50% of total beta-diversity contributed by nestedness are shown in red. influenced by temperature anomaly since the LGM and mean stability on both turnover and nestedness components of beta-di- annual temperature (Fig. 3, fig. S3, and table S1). However, turnover versity were largely consistent with our predictions (fig. S5). For and nestedness components presented contrasting associations the turnover component of beta-diversity, the LGM temperature (Fig. 3, fig. S3, and table S1). Consistent with our predictions, the anomaly had negative effects in Europe, North America, and turnover component was negatively associated with the LGM tem- Asia, and the LGM precipitation anomaly had negative effects in perature anomaly and positively with mean annual temperature, Australia and North America. For the nestedness component, the whereas the nestedness component was positively associated with LGM temperature anomaly had positive effects in Europe, North the LGM temperature anomaly and negatively with mean annual America, Asia, and Australia, and the LGM precipitation anomaly temperature (Fig. 3 and fig. S4). Both turnover and nestedness com- had positive effects in Australia and North America. Although we ponents of all three biodiversity facets had a stronger correlation did not detect significant associations between beta-diversity com- with the LGM temperature anomaly than with mean annual tem- ponents and the LGM temperature or precipitation anomaly in two perature (table S1). In addition, human modification, as a continents (South America and Africa), no associations were oppo- measure of human pressures, had negative effects on the turnover site to our predictions in any continents (fig. S5). component and positive effects on the nestedness component across all three biodiversity facets (Fig. 3). The nestedness compo- Deviations of phylogenetic and functional beta-diversity nent also had a weak positive relationship with precipitation and associated factors anomaly since the LGM and a weak negative relationship with Null model analyses showed that deviations of the observed phylo- mean annual precipitation across all three biodiversity facets genetic and functional turnover and nestedness from random ex- (Fig. 3). Temperature and precipitation seasonality also had weak pectations based on observed taxonomic beta-diversity and site- effects on the nestedness component, but these effects were incon- specific regional species pools were not evenly distributed across sistent across three biodiversity facets (Fig. 3). For the proportion of space (Fig. 4). Positive deviations of phylogenetic and functional total beta-diversity contributed by the nestedness, the important as- turnover were primarily concentrated in low-latitudinal and moun- sociated factors were similar to those for the nestedness compo- tainous areas, particularly in southwest China and the Andes nent (Fig. 3). region, indicating that species replacement among assemblages Across six continents, beta-diversity was primarily influenced by largely occurred at deep phylogenetic branches and in species different environmental variables, which could be attributed to dif- with dissimilar functional trait values. Meanwhile, most areas in ferences in environmental gradients among these continents (fig. North America had negative deviations in turnover, indicating S5). Despite this geographic variation, the effects of past climatic the replacement of relatively recently evolved lineages and species Xu et al., Sci. Adv. 9, eadd8553 (2023) 5 April 2023 5 of 18 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E Fig. 3. Effects of past climate stability and current environmental conditions on taxonomic, phylogenetic, and functional beta-diversity of angiosperm trees. Total beta-diversity and its components of turnover and nestedness and the nestedness proportion were indicated with different shapes and colors. The averaged es- timates of standardized coefficients (points) and the 95% confidence intervals (bars) were obtained from spatial simultaneous autoregressive models. Nonsignificant variables are shown in gray. Temperate and precipitation anomaly: The differences in annual temperature and precipitation between the present and the LGM. MAT, mean annual temperature; MAP, mean annual precipitation. with similar functional trait values. By contrast, positive deviations nestedness deviations in regions with large Quaternary temperature of phylogenetic and functional nestedness were mainly concentrat- change. In contrast, no significant associations were detected ed in high latitudinal areas, particularly in North America and some between current climatic variables and deviations in phylogenetic areas in Northern Europe. This pattern suggests phylogenetic and turnover and nestedness. Deviations of functional turnover and functional selection for closely related species and those with similar nestedness were significantly associated with mean annual temper- functional trait values in species losses (e.g., glaciation-driven ex- ature and precipitation. Deviation of functional turnover was neg- tinction) or gains (e.g., postglacial recolonization). Nestedness de- atively associated with mean annual temperature and positively viations were usually close to zero for most other regions, suggesting associated with mean annual precipitation, whereas deviation of phylogenetically and functionally random species losses or gains. functional nestedness was positively associated with mean annual Geographic variation in phylogenetic and functional turnover temperature and negatively associated with mean annual precipita- and nestedness deviation was most strongly associated with the tion (Fig. 5). Elevational range also had a weak positive relationship LGM temperature anomaly (Fig. 5, fig. S6, and table S1). Consistent with both phylogenetic and functional nestedness (Fig. 5). Across with our predictions, deviations of phylogenetic and functional continents, we found large variation in the effects of environmental turnover decreased with increasing LGM temperature anomaly, variables on deviations of phylogenetic and functional turnover and whereas deviations of phylogenetic and functional nestedness in- nestedness (fig. S7). Consistent with global-scale analyses, the LGM creased with increasing LGM temperature anomaly (Fig. 6). Specif- temperature or precipitation anomaly was negatively associated ically, there were negative turnover deviations and positive with deviations of phylogenetic or functional turnover and Xu et al., Sci. Adv. 9, eadd8553 (2023) 5 April 2023 6 of 18 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E Fig. 4. Global patterns of the deviation of phylogenetic and functional turnover and nestedness of angiosperm trees. The deviations of phylogenetic and func- tional turnover (A and B) and nestedness (C and D) were calculated as the differences between the observed and random expectations based on taxonomic beta-diversity and site-specific regional species pools. Fig. 5. Effects of past climate stability and current environmental conditions on deviations of phylogenetic and functional turnover and nestedness of angio- sperm trees. The deviations were calculated as the differences between the observed and random expectations based on taxonomic beta-diversity and site-specific regional species pools. The averaged estimates of standardized coefficients (points) and the 95% confidence intervals (bars) were obtained from spatial simultaneous autoregressive models. Nonsignificant variables are shown in gray. Temperate and precipitation anomaly: The differences in annual temperature and precipitation between the present and the LGM. Xu et al., Sci. Adv. 9, eadd8553 (2023) 5 April 2023 7 of 18 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E had a higher trait data availability (figs. S11 and S13). This suggested that a strong correlation between functional and phylogenetic beta- diversity was not merely the consequence of trait imputation con- sidering phylogenetic information. Our results were also robust to the approach used to model tree ranges from distribution occurrences and to the size of moving windows used to calculate intraregional beta-diversity (figs. S14 to S21). Geographic patterns of taxonomic, phylogenetic, and func- tional beta-diversity shown in the main text were based on alpha- hull ranges estimated with an alpha value of 6 (see the “Tree distri- butions” section in Materials and Methods for details). These were very similar to those based on alpha-hull ranges estimated with dif- ferent alpha values (i.e., 2, 4, and 10; figs. S14 to S16), as indicated by strong correlations (Pearson’s r > 0.95; fig. S17). Regarding the in- fluence of sizes of moving windows, intraregional beta-diversity within moving windows of nine (3 × 3) grid cells (fig. S18) showed similar spatial patterns to those within moving windows of 25 (5 × 5) grid cells shown in the main text. There were also similar spatial patterns for phylogenetic and functional turnover de- viations and nestedness (fig. S19). The effects of environmental var- iables on both observed beta-diversity and deviations of phylogenetic and functional turnover and nestedness based on Fig. 6. Relationships between temperature anomaly since the LGM and devi- moving windows of nine grid cells (figs. S20 and S21) were qualita- ations of phylogenetic and functional turnover and nestedness of angio- sperm trees. The deviations of phylogenetic and functional turnover (A and B) tively consistent with those reported in the main text. Specifically, and nestedness (C and D) were calculated as the differences between the observed the LGM temperature anomaly was consistently associated with and random expectations based on taxonomic beta-diversity and site-specific re- both turnover and nestedness components of all three biodiversity gional species pools. The blue lines were fitted with linear regressions. Significance facets, i.e., a negative relationship with turnover and a positive rela- was tested using a modified t test to control for spatial autocorrelation. R 2, coeffi- tionship with nestedness (fig. S20). Although we did not detect a cient of determination. *P < 0.05 and **P < 0.01. significant relationship between the LGM temperature anomaly and deviations of phylogenetic and functional turnover, a positive positively associated with deviations of phylogenetic or functional relationship remained between temperature anomaly and devia- nestedness in North America, Europe, and Asia (fig. S7). tions of phylogenetic and functional nestedness (fig. S21). Sensitivity analyses Patterns of functional beta-diversity shown in this study were robust DISCUSSION to trait imputation despite a large proportion of species requiring Our results demonstrate that Quaternary climate change plays a missing traits to be imputed based on correlation structure major role in shaping present-day patterns of the turnover and nest- among traits and phylogenetic information. Functional beta-diver- edness components of taxonomic, phylogenetic, and functional sity and its components of turnover and nestedness that were qual- beta-diversity in angiosperm trees worldwide. Among the environ- ified based on all species had similar spatial patterns as those based mental variables examined here, temperature anomaly since the on only species that have measured data for at least one or five traits LGM had the strongest relationship with beta-diversity patterns, (figs. S8 and S9), with Pearson’s correlations higher than 0.98 and particularly for phylogenetic and functional biodiversity facets 0.75, respectively (fig. S10). A reduced correlation between func- (see Figs. 3 and 5 and table S1). Consistent with our hypotheses tional beta-diversity based on all species and those based on only (Fig. 1), the temperature anomaly showed a negative relationship species with at least five measured traits was likely the consequence with the turnover and a positive relationship with the nestedness of only a small subset of species within some grid cells being used in components of beta-diversity, reflecting different ecological and calculating beta-diversity (fig. S11). In addition, functional and evolutionary processes driven by Quaternary climate change. The phylogenetic beta-diversities were strongly correlated even when effects of temperature anomaly were also clear at the continental only species with at least one or five measured traits were used in scale, particularly in the Northern Hemisphere, which experienced calculating beta-diversity (fig. S12). Specifically, functional turnover a large temperature decrease during the LGM. These results are gen- was generally lower than phylogenetic turnover, and functional erally consistent with previous regional and global studies that nestedness was similar to phylogenetic nestedness in magnitude; mostly focused only on turnover and nestedness components of tax- these patterns were consistent with the patterns based on all onomic beta-diversity in other organisms (e.g., amphibians, fresh- species (fig. S12). Furthermore, the functional and phylogenetic water fish, mammals, and birds) (17–19, 29), expanding the idea of beta-diversity relationship based on all species was largely consis- the fundamental role played by past climate change in shaping bio- tent across six continents (fig. S13). The correlation between func- diversity patterns to angiosperm trees worldwide. By extending this tional and phylogenetic beta-diversity was not consistently weaker recognized relation to angiosperm trees, one of the ecologically across beta-diversity metrics in Europe and North America, which most important organismal groups worldwide, our results highlight that strong climate change has the potential to notably influence Xu et al., Sci. Adv. 9, eadd8553 (2023) 5 April 2023 8 of 18 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E global biodiversity and ecosystem properties not only via direct area, plant height, seed mass, stem specific density, and leaf nitrogen effects but also via its effects on this ecosystem-defining group. Al- concentration) is still influenced by the Quaternary climate in though few large-scale studies have analyzed taxonomic beta-diver- North and South America (66). This functionally selective process sity together with phylogenetic and/or functional beta-diversity, has also been shown in patterns of functional diversity of plants in they usually directly compared the observed values of different Europe, with regions that experienced large Quaternary climate facets of beta-diversity (14, 15, 37–39). Our study differs from pre- change usually associated with low fractions of maximum potential vious analyses by comparing phylogenetic and functional turnover functional diversity based on current environmental conditions and nestedness with those randomly expected given taxonomic (49). Together, the phylogenetically and functionally selective pro- beta-diversity. Using this novel approach, we found lower phyloge- cesses will cause a lower phylogenetic and functional diversity than netic and functional turnover and higher nestedness than random randomly expected from species richness for species-poor assem- expectations in regions that experienced large Quaternary temper- blages and drive a larger difference in phylogenetic and functional ature changes (Figs. 4 to 6), patterns that have not been reported diversity (i.e., stronger nestedness structure) between species-rich before. This finding suggests that processes such as species replace- and species-poor assemblages. ment, extinction, and colonization during glacial-interglacial oscil- In contrast with the nestedness component, the turnover com- lations are phylogenetically and functionally selective. For instance, ponent of beta-diversity in angiosperm trees is high in regions with species from specific lineages or with specific traits are more likely large Quaternary climate stability, reflecting the prevalence of to experience glaciation-driven extinction or lags on their postgla- species with small ranges (22–25). Long-term stable climatic condi- cial recolonization. tions can benefit species persistence in situ and could increase spe- There is a high nestedness component of beta-diversity in angio- ciation rates due to increased opportunities for regional niche sperm trees in regions with large Quaternary climate change. Pre- differentiation (21, 32). The accumulated species are expected to vious studies at smaller scales or for other organisms often be generally specialized and have small ranges (21, 32), leading to explained this by strong local species extinction during glaciations a high turnover of species across sites. By comparison, small- and incomplete postglacial recolonization from ice-age refugia (17, ranged species in regions with large Quaternary climate change ex- 19, 29). These factors also provide a likely explanation for angio- perience disproportionate extinctions because they usually have low sperm trees. For instance, several studies have shown that geograph- vagility and poor adaptation to changed conditions (21). Therefore, ic accessibility from glacial refugia accounts for much of the regions with large Quaternary climate change have a high propor- variation in tree diversity across Europe (55, 56). In addition, tion of large-ranged species (22–25), decreasing species replace- time since deglaciation is strongly positively associated with region- ment among neighboring sites. Our global analyses of al species richness of vascular plants in the Arctic (57). These pat- angiosperm trees support the above theoretical expectations and terns are consistent with the evidence that the current distributions are generally consistent with previous studies of other organisms of many tree species in Europe and North America are not in equi- or smaller scales (17–19, 29), suggesting a prevalent paleoclimatic librium with the current climate, often due to postglacial dispersal legacy on spatial turnover in biodiversity. limitations (26, 58). Plant species with lower dispersal capacity tend This worldwide study using angiosperm trees shows lower phy- to occupy lower proportions of their potential ranges, and their logenetic and functional turnover than randomly expected from current distributions have stronger associations with accessibility taxonomic beta-diversity in regions with large Quaternary climate to glacial refugia (27, 59). Differential and limited dispersal capacity change. This suggests that species replacements between sites in would result in steep rates of species loss from glacial refugia toward regions with large Quaternary climate change usually stem from higher latitudes, leading to strong nestedness of assemblagecompo- closely related young lineages with similar functional trait values sitions in regions strongly affected by Quaternary climate (Fig. 1C). This finding is consistent with evidence that floras in en- change (60). vironmentally unstable regions usually have more recent evolution- Because species differ in functional traits and evolutionary ary divergence times (44). These floras also tend to be history, they may have differential capacities to survive during gla- phylogenetically clustered, meaning that species are closely related ciations and to recolonize postglacial suitable areas (27, 50, 59, 61). to each other (44, 67). Similarly, assemblages in regions with large Specifically, cold tolerance and dispersal capacity are regarded as Quaternary climate change usually have low functional dispersion, important species attributes affecting species survival during glaci- reflecting species compositions with similar functional trait values ations and recolonization to suitable postglacial areas, respectively (49). However, the associations of Quaternary temperature change (27, 50, 59, 61). These attributes have also been shown to be phylo- with deviations of phylogenetic and functional turnover were genetically conserved (48, 62, 63). Therefore, species from specific weaker than with deviations of phylogenetic and functional nested- lineages with strong cold tolerance are more likely to survive during ness (Fig. 6), and these associations became insignificant in sensi- glaciations, and those from lineages with strong dispersal capacity tivity analyses using the moving window size of nine grid cells (fig. are more likely to recolonize their postglacial climatically suitable S21). These differences suggest that the phylogenetically and func- areas (27, 48). Because many traits are correlated with each other tionally selective processes leading to higher phylogenetic and func- because of biophysical constraints and trade-offs among traits (64, tional nestedness than random expectations (see above) are the 65), species with strong cold tolerance and dispersal capacity are main glaciation-driven processes that result in a departure of phy- probably also affiliated with specific trait values for some other logenetic and functional beta-diversity from the taxonomic beta- functional traits such as those used in this study (e.g., maximum diversity. height and seed dry mass), which would be selected during In addition to paleoclimate change, contemporary environmen- glacial-interglacial oscillations. The contemporary functional com- tal conditions also play important roles in explaining global patterns position of plant assemblages measured with five traits (specific leaf of beta-diversity in angiosperm trees. Consistent with the major role Xu et al., Sci. Adv. 9, eadd8553 (2023) 5 April 2023 9 of 18 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E of temperature in shaping plant beta-diversity across the Americas keystone organismal group in terrestrial ecosystems. Notably, high (13), annual temperature is the strongest current climatic variable to levels of Quaternary temperature instability are associated with explain the turnover component of beta-diversity, with high turn- marked gradients in species, lineage, and trait diversity across over in regions with warm conditions. By comparison, the nested- sites, reflecting severe past losses, incomplete recolonization, and ness component is high in regions with cold conditions, likely low levels of replacement in all three diversity facets. Furthermore, reflecting strong filtering of species under stressful conditions, strong Quaternary climate change is also associated with lower phy- because most lineages probably originate in tropical climates and logenetic and functional turnover and higher nestedness than ran- are difficult to adapt in freezing conditions [“tropical niche conser- domly expected from taxonomic beta-diversity, reflecting vatism” hypothesis (68)]. The nestedness component also tends to phylogenetically and functionally selective processes during be higher when annual precipitation is lower, likely because water glacial-interglacial oscillations. These findings suggest that large availability is critical for trees, and species would be gradually fil- Quaternary climate changes resulted in the reduction and homog- tered out in areas with limited water availability. In addition, enization of the diversity of angiosperm tree across large landscapes while controlling for other variables, human pressures significantly worldwide, with particularly severe effects on phylogenetic and impact tree beta-diversity patterns, decreasing the turnover compo- functional composition. These diversity reductions are likely to nent but increasing the nestedness component. This decreased affect ecosystem functions, such as plant productivity in forests turnover reflects the pattern of biotic homogenization driven by and woodlands, because of the critical ecological importance of human activities as a consequence of the replacement of narrow- tree diversity (51, 73). Our results highlight that future human- ranged species by widespread species (36, 69). The increased nested- driven climate change is likely to have strong and long-lasting ness probably suggests greater species losses in some sites than effects on tree assemblages globally via strong reduction and ho- others, which reflect landscapes being differentially influenced by mogenization of taxonomic, phylogenetic, and functional diversity, human activities such as deforestation (70). potentially leading to impaired forest ecosystem functioning. Although generally similar, there were clear differences in spatial patterns of phylogenetic and functional beta-diversity and associa- tions with environmental variables. First, functional turnover was MATERIALS AND METHODS generally lower than phylogenetic turnover, particularly in Tree distributions regions with high phylogenetic turnover (usually tropical regions; In this study, we used the global tree species list and distributions Fig. 2, E and F, and fig. S2F), suggesting that tree species among compiled in the TREECHANGE database (53). The tree species neighboring sites are more similar in functional trait values than checklist came from the GlobalTreeSearch v.1.2 (74), which was as- in phylogenetic relationship. This pattern reflects functional redun- sembled from a range of botanical publications and extended by dancy among regional co-occurring species, which may result from many botanical experts. Taxonomic names were standardized environmental filtering selecting functionally similar species (71) using the Taxonomic Name Resolution Service (75), resulting in and/or evolutionary convergence leading to similar traits among in- 54,020 tree species. dependent lineages (72). Second, we found significant effects of Tree species occurrences were compiled from five major com- mean annual temperature and precipitation on deviations of func- prehensive biodiversity infrastructures, including the Global Biodi- tional turnover and nestedness but no effects of current climatic versity Information Facility (GBIF; www.gbif.org) (76), the conditions on deviations of phylogenetic turnover and nestedness Botanical Information and Ecological Network v.3 (BIEN; http:// (Fig. 5). These differences suggest a stronger functional beta-diver- bien.nceas.ucsb.edu/bien) (77), the Latin American Seasonally sity response to contemporary environmental conditions compared Dry Tropical Forest Floristic Network (DRYFLOR; www.dryflor. to phylogenetic beta-diversity. Although we observed differences in info) (78), the RAINBIO database (https://gdauby.github.io/ the phylogenetic and functional beta-diversity patterns, we note rainbio/index.html) (79), and the Atlas of Living Australia (www. that these two measures are not independent in this study, ala.org.au). These records were assessed using a quality control because phylogenetic information was used in the imputation of workflow considering geographic coordinates, duplications, native missing functional traits. Increasing the amount of measured trait ranges, and geographical and environmental outliers and labeled data is expected to result in greater differences between these two from AAA (high geographic precision and low environmental un- diversity dimensions. Because trait data coverage was not evenly dis- certainty) to E (missing coordinates) (53). In this study, we used tributed across regions, with higher coverage in Europe and North high-quality occurrences and those records without geographic America (fig. 11), the functional beta-diversity could be more re- bias distributed in native ranges, which were labeled as AAA, AA, dundant with phylogenetic beta-diversity due to trait imputation A, and C [see (53) for further information]. The final dataset had in regions with poor trait coverage, leading to spatial uncertainties 46,752 species with 7,066,785 occurrences at a resolution of 30 arc in spatial patterns of functional beta-diversity. This concern, sec (fig. S22). however, is unlikely to be a major issue, because our sensitivity anal- We then estimated species ranges using alpha hulls for species yses using only species with at least one and five measured traits with 20 or more occurrences using the R package alphahull (80). showed similar spatial patterns of functional beta-diversity as The alpha hull is a generalization of the convex hull and allows those using all species (figs. S8 to S10) and because a strong corre- the constructed geometric shape to consist of several discrete lation between functional and phylogenetic beta-diversity was prev- hulls dependent on the value of the parameter alpha (81). For alent across continents regardless of trait coverage (fig. S13). species with fewer than 20 occurrences or disjunct records, a 10- In summary, our results show that Quaternary climate change km buffer was built around each point and then combined with has left a strong legacy in current global patterns of taxonomic, phy- alpha hulls. Following recommendations in previous studies (82), logenetic, and functional beta-diversity for angiosperm trees, a we used the alpha value of 6° to construct alpha hulls. To test the Xu et al., Sci. Adv. 9, eadd8553 (2023) 5 April 2023 10 of 18 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E sensitivity of our results to alpha values, we also constructed alpha Phylogenetic tree hulls using other alpha values and repeated analyses based on dif- The phylogenetic information for the tree species was extracted ferent range estimates (see the “Sensitivity analyses” section). from the largest seed plant phylogeny currently available [the Species ranges were then rasterized to grid cells in a resolution of “ALLBM” tree in (90)]. This synthesis phylogeny was constructed 200 km with an equal-area Behrmann projection using the R by combining sequence data from GenBank with a backbone tree package letsR (83). Species assemblages in each grid cell were reflecting deep relationships (91) and adding species not found in defined as all species with ranges falling within the grid cell. We GenBank based on their placement in the Open Tree of Life, which chose the resolution of 200 km because biodiversity assessment at reflects previous knowledge about phylogenetic relationships and a coarse resolution can reduce the undersampling of species distri- taxonomy (90). Because there were 5791 tree species (10.7% of a butions. Although biodiversity patterns based on coarse species dis- total of 54,020 species) missing in the phylogeny due to different tributions probably underestimate the variation in species taxonomic backbone lists, these missing species were manually compositions at small spatial scales, this study focuses on global var- added into the phylogeny based on the most recent common ances- iation in beta-diversity at broad spatial scales and how past and con- tor using “add.species.to.genus” function in the R package phytools temporary macroclimate gradients determine these changes. We (92), following the same approach used in (90) to add missing excluded gymnosperms from our analyses because including species. We then reduced this phylogeny by removing any species them can strongly influence plant phylogenetic structure due to absent in our list of tree species with distribution data (46,752 deep split time between gymnosperms and angiosperms, which species). The generated phylogeny was further pruned to contain means whether there is a gymnosperm in a grid cell can strongly only the angiosperm species used in this study. Although the gen- influence its phylogenetic diversity, obscuring patterns within an- erated phylogeny contains some polytomies, this is unlikely to bias giosperms (84). We did not perform analyses for gymnosperms sep- the global analyses of phylogenetic beta-diversity. Generating a phy- arately because there were only 579 gymnosperms in our dataset logeny for a group of species by pruning from a synthesis tree, such and 792 grid cells worldwide that had at least five gymnosperms as “ALLBM” used in this study, has been widely used in ecological (fig. S23). We removed grid cells with fewer than five species to analyses, and it has been shown that the common community phy- avoid potentially unreliable beta-diversity estimates (see fig. S23 logenetic analyses based on a synthesis tree and a purpose-built re- for spatial pattern of species richness). A total of 2319 assemblages solved tree can produce consistent results (93). and 43,635 species were kept and used in subsequent analyses. Functional traits External validation of tree distributions We used the trait product for trees from (85). Specifically, we used We performed three types of external validations to validate the eight functional traits related to plant growth, survival and repro- alpha-hull ranges constructed in this study. The details of these ex- duction (64), including specific leaf area, leaf area, leaf dry matter ternal validations have been published in (85), which used the same content, leaf nitrogen concentration, leaf phosphorus concentra- range maps as those used in this study. First, we compared global tion, wood density, seed dry mass, and plant maximum height. patterns of species richness based on our range maps with predicted These traits are often used in functional diversity analyses of richness from an independent study that integrated 1336 forest plots plants and have relatively good data coverage (49, 64, 66, 94). and 282 regional checklists (86). Our estimated richness shows a These traits were selected from a total of 20 traits that were compiled similar spatial pattern to those predicted based on plots and region- for the TREECHANGE project (85) (see table S2 for all trait names). al checklists, although the predicted richness was generally higher Trait observations were compiled from three databases, including than our estimates. Second, we compared our range maps with TRY Plant Trait Database (www.try-db.org) (94), BIEN (77), and ranges from two external datasets: EU-Forest (87) and Little’s TOPIC (95). However, only 11,659 species (21.6% of 54,020 tree “Atlas of the United States trees” (88). EU-Forest includes range species) had data for at least one functional trait. The trait gaps maps of 242 tree species in Europe based on 1,000,525 occurrence were filled by Bayesian hierarchical probabilistic matrix factoriza- records, and Little’s Atlas includes ranges of 680 tree species in tion (BHPMF), which is a robust machine learning approach im- North America based on botanical lists, forest surveys, field notes, puting trait values based on the taxonomic hierarchy and and herbarium specimens. There were 120 and 536 species in our correlation structure among traits (96). Following the suggestion dataset matched to EU-Forest and Little’s Atlas. We calculated a co- of the BHPMF approach, trait values were log-transformed before occurrence index to measure the extent to which range maps from imputation, and imputed trait values were back-transformed. All of our dataset and external datasets overlapped. The co-occurrence the 20 functional traits were used in the imputation to maximize index varies from 0 (no overlap) to 1 (complete overlap). For the benefits from the correlation structure among traits. To improve matched tree species between our dataset and EU-Forest and the estimation of missing trait values, phylogenetic information in Little’s Atlas, 73 and 85% of species had co-occurrence values the form of phylogenetic eigenvectors was also used as predictors higher than 0.8, and the mean co-occurrence values were 0.84 (97). These phylogenetic eigenvectors were extracted from a princi- and 0.92, with the SD as 0.16 and 0.11, respectively. This suggests pal coordinate analysis on the genus-level phylogenetic distance that the range maps are well matched between our dataset and two matrix (98). To determine the number of phylogenetic eigenvectors external datasets. Third, we compared the species richness of 459 included in the trait imputation, an increasing number of phyloge- geographic regions worldwide based on our range maps with netic eigenvectors were added as predictors. The predicted trait those based on subnational species lists (89). The two richness es- values were then evaluated using the root mean squared error timates were highly correlated (Pearson’s r = 0.89), and the map of (RMSE) (96). When the first six phylogenetic eigenvectors were in- residuals from their relationship showed that the two richness esti- cluded in the imputation, the RMSE was minimized to 0.087, which mates were similar in most regions. means a high overall prediction accuracy [see (85) for more details]. Xu et al., Sci. Adv. 9, eadd8553 (2023) 5 April 2023 11 of 18 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E The maximum and minimum observed trait values were used as the margins of continents have few neighboring cells, we only in- thresholds to constrain imputed data to avoid outliers in the impu- cluded grid cells with at least 13 neighboring cells as focal cells tation of missing values. For plant maximum height, a height of 2 m (leaving a total of 1838 grid cells) to calculate beta-diversity. For was used to replace imputed values lower than that, considering the each focal cell, we further randomly sampled 13 cells from its 24 definition of trees in the GlobalTreeSearch database (74). neighboring cells and used them together with the focal cell to cal- We noted that although a large proportion of tree species is culate multiple-site dissimilarity. We performed this resampling missing in measured trait values, 55.8% of 4157 genera and 88.9% 200 times and calculated the average beta-diversity. of 253 families have data for at least one functional trait (table S2). In We calculated taxonomic, phylogenetic, and functional beta-di- addition, a lack of trait data for a large proportion of species did not versity using indices from the family of Sørensen-based multiple- mean low coverage of trait data across regions. We thus calculated site dissimilarity measures (16, 40). As one of the most commonly the proportion of species in each grid cell that had measured data used taxonomic dissimilarity indices, Sørensen dissimilarity index for at least one, five, or all eight traits. Although all regions, except was used to measure the proportion of exclusive species among as- Europe, had a low proportion of species with all eight traits, a con- semblages. The analogs of Sørensen dissimilarity index for phyloge- siderable proportion of species had at least one or five traits in most netic and functional beta-diversity were used to measure the regions, particularly Europe and North America (fig. S11). For 78% proportion of exclusive branch lengths in a phylogenetic tree or of the grid cells (1818 of a total of 2319 grid cells), more than half functional dendrogram for the species among assemblages (101). (>50%) of their species had at least one measured trait data. There- We then partitioned the three facets of Sørensen dissimilarity into fore, our functional trait dataset provides substantial information two additive components due to spatial turnover and nestedness about trait variation across species in most regions. (16, 40). The turnover component, which is the Simpson dissimilar- We further tested phylogenetic signals of eight traits for species ity, reflects the effect of replacements of species or branches in a with measured trait data. We calculated phylogenetic signals as phylogeny or functional dendrogram without the impact from dif- Pagel’s λ using the “phylosig” function in the R package phytools ferences in species richness, phylogenetic, and functional diversity (92). Pagel’s λ can range from zero to one, with zero indicating among sites (16, 40). The nestedness component, which is the dif- no phylogenetic signal and one indicating the strongest phylogenet- ference between Sørensen and Simpson dissimilarities, reflects the ic signal (99). We found that all eight traits had significant phylo- contribution due to differences in species richness, phylogenetic, genetic signals (P < 0.001), with lambda values ranging from 0.602 and functional diversity when species-poor assemblages are to 0.954 (table S3). This supported our approach using phylogenetic nested in species-rich assemblages (16, 40). All calculations of relatedness to predict traits. We also tested phylogenetic signals of beta-diversity were performed using the R package betapart (54). traits using imputed trait data. Because all species had imputed We noted that although some other indices can be used to traits and the number of species may influence the phylogenetic measure different patterns of dissimilarity between assemblages, signal estimates, we calculated the number of species that had mea- these indices were usually related to Sørensen and Simpson dissim- sured data for each trait and randomly sampled this number of ilarity (102). species from the imputed traits. We repeated this random sampling Although functional beta-diversity can be calculated based on 200 times and calculated the mean and SD of phylogenetic signals the volume of intersections between convex hulls in a multidimen- across iterations. All eight imputed traits had stronger phylogenetic sional functional space (41), our calculation was based on a func- signals than the respective measured traits (table S3). This result was tional dendrogram because it has a similar data structure to the not unexpected because phylogenetic information was used to phylogenetic tree, making functional and phylogenetic beta-diver- impute missing functional traits. This suggests that the imputed sity comparable (103), and because the convex hull–based method traits contain some redundant information with the phylogenetic is sensitive to extreme trait values or outliers and requires very long tree, and thus the results of phylogenetic and functional analyses computation time, particularly for assemblages with many species would be dependent to some degree due to trait imputation. To similar to ours (41, 104). We performed a principal components assess the influence of trait imputation, we performed sensitivity (PCs) analysis on all eight log- and z-transformed trait values to analyses using only species that had measured trait data for calcu- remove the correlation structure among traits. The first four PCs lating phylogenetic and functional beta-diversity and compared re- explained a high amount (93.5%) of the total variation. We thus lationships between phylogenetic and functional beta-diversity used these four PCs to calculate the Euclidean distances among across six continents (see the “Sensitivity analyses” section). species and performed hierarchical clustering to generate a func- tional dendrogram based on the method of “complete” using the Beta-diversity R package fastcluster (105). We used a moving window, containing 25 (5 × 5) grid cells of 200 To reflect the relative importance of turnover and nestedness km by 200 km, to measure intraregional compositional heterogene- components, we also calculated the proportion of total beta-diver- ity as beta-diversity for each focal cell (13, 29). As a sensitivity anal- sity contributed by the nestedness component for all three evaluated ysis, we further used a moving window of nine (3 × 3) grid cells to biodiversity facets (15, 19, 39). Proportions lower than 0.5 indicate measure intraregional beta-diversity (see the “Sensitivity analyses” that total beta-diversity is mainly determined by turnover, whereas section). In this study, we used the multiple-site dissimilarity, proportions higher than 0.5 suggest that the nestedness-resultant rather than the averaged pairwise dissimilarity between a focal cell dissimilarity is the main component. and its neighboring cells, because the averaged pairwise dissimilar- ity cannot reflect co-occurrence patterns among more than two sites Null model (100). Because the number of grid cells within a region can affect the Because the three facets of beta-diversity are not independent and multiple-site dissimilarity (16) and some cells on islands and along usually strongly correlated, the processes driving taxonomic beta- Xu et al., Sci. Adv. 9, eadd8553 (2023) 5 April 2023 12 of 18 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E diversity also affect phylogenetic and functional beta-diversity (38– further measured as the average of the focal cell and its 24 neighbor- 41). To investigate whether phylogenetic and functional beta-diver- ing cells (fig. S24). sity are affected by processes beyond those shaping taxonomic beta- diversity, we constructed a null model to calculate the random ex- Statistical analysis pectations of phylogenetic and functional beta-diversity based on We used the Pearson correlation to check pairwise relationships the observed taxonomic beta-diversity and site-specific regional among environmental variables and also among multiple dimen- species pools. For each focal cell, we defined the site-specific region- sional beta-diversity. To account for the spatial autocorrelation of al species pool as all species in the 25-cell moving window that was these variables, a spatially corrected t test was used to assess statis- used to calculate the beta-diversity for the given focal cell. In the tical significance (110). Following suggestions in (29), we per- null model, the identities of species were randomly shuffled formed piecewise regressions to examine latitudinal patterns of among the species pool in the phylogenetic tree and functional den- beta-diversity and its turnover and nestedness components. drogram. Therefore, species richness of each grid cell, intraregional We then tested the effects of environmental variables on the taxonomic beta-diversity and its components of turnover and nest- global-scale variation in each of the beta-diversity variables (three edness, as well as the pool of species branches in the phylogenetic facets of total beta-diversity and turnover and nestedness compo- tree and functional dendrogram were kept constant with the ob- nents, and deviations of phylogenetic and functional turnover and served in the null model. nestedness). First, we evaluated relationships between beta-diversity Because this study focused on whether there are some processes and eight environmental variables using bivariate linear regressions driving stronger or weaker intraregional turnover and nestedness and the Pearson correlation. We log-transformed elevational range patterns in phylogenetic and functional compositions than expected and square root–transformed human modification to improve these by chance, we calculated the random expectations of phylogenetic variables' normality and the regressions' linearity. Significances and functional turnover and nestedness. We repeated our null were assessed using a spatially corrected t test (110). Then, multiple model 200 times and calculated the average random expectations. ordinary least squares (OLS) linear regressions were used to calcu- Then, we calculated the deviations of phylogenetic and functional late standardized regression coefficients and determine the relative turnover and nestedness as the differences between the observed importance of environmental variables using both the scaled beta- and the average expected values. diversity and environmental variables. Residuals of OLS models, however, showed strong spatial autocorrelation, which could Environmental data affect significance test and bias parameter estimates (111). To To measure past climate stability, we calculated the change in mean account for spatial autocorrelation, we used spatial simultaneous annual temperature and precipitation between the present and the autoregressive (SAR) models that include a spatial weight matrix LGM, defined as the temperature anomaly and precipitation as an additional error term (112). We run SAR models with a anomaly. Contemporary environmental conditions were represent- range of neighbor distances from 200 to 1000 km in a step of 100 ed by current climatic conditions, topography, and human pressure. km and row-standardized coding style to define the spatial weight We note that topography may also capture the effects of past climate matrix and used the minimum residual spatial autocorrelation stability because topographic heterogeneity favors species to track (minRSA) to select the most appropriate SAR model for each re- climate change by shifting short altitudinal ranges (23). We includ- sponse variable (112). The criteria of minRSA is to minimize the ed mean annual temperature and precipitation because of their roles summed absolute values of Moran’s I of the first 20 distance in determining species distributions (106). Temperature and pre- classes in the model residuals. For all beta-diversity variables, the cipitation seasonality were also included because they were impor- SAR model with a neighbor distance of 300 km produced a tant drivers of species range size (24, 25), which was directly related minimal residual spatial autocorrelation and a lowest Akaike infor- to beta-diversity. Temperature seasonality was measured as the SD mation criterion value. The final SAR models substantially reduced of monthly temperature (bio4), and precipitation seasonality was spatial autocorrelation in residuals relative to the OLS models (table measured as the coefficient of variation of monthly precipitation S4). In the multiple regression models, we included all eight envi- (bio15). Topography was measured as the elevational range ronmental variables to estimate their relative importance. Because within a grid cell using elevation data in a spatial resolution of 1 of a strong correlation between mean annual temperature and tem- km. Human pressure was represented with the human modification perature seasonality (fig. S25), we also fit models without tempera- index, which was modeled based on 13 anthropogenic stressors, ture seasonality in a preliminary analysis. These models estimated a considering different types of human activities (e.g., human settle- very similar coefficient for the mean annual temperature with that ment, agriculture, and energy production) (107). Human modifica- from models with all environmental variables, suggesting that our tion values range from 0 to 1, with higher values indicating stronger models were not sensitive to the strong correlation between envi- human pressures. ronmental variables. The temperature and precipitation data at the LGM were extract- We also used Akaike-based model selection and multimodel in- ed from the PaleoClim database with a resolution of 5 arc min (108). ference to qualify the relative importance of each environmental The contemporary climate and elevation data were downloaded variable by assessing all subsets of full SAR models (113). The from the WorldClim v1.4 database, with a resolution of 5 arc min Akaike weight of each model was first calculated. Then, we averaged for climate (109). The human modification map was accessed from the standardized regression coefficients for each variable across all (107), with a resolution of 1 km. We first calculated the mean values evaluated models by weighting each value with the Akaike weight of of all climatic variables and human modification for each grid cell of the model that contained it. We also measured the importance of 200 km. To match the scale of beta-diversity and environmental each environmental variable by summing the weights of all variables (13), the environmental conditions for each cell were models including that variable. Xu et al., Sci. Adv. 9, eadd8553 (2023) 5 April 2023 13 of 18 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E To assess the variation in the effects of environmental variables species in the nine-cell moving window for each focal cell. Effects of on beta-diversity across continents, we further fit SAR models re- environmental variables on spatial variation in beta-diversity were gressing each of beta-diversity variables against eight environmental then assessed using the same methods used in our main analyses. variables within each of six continents (Africa, Asia, Australia, Europe, North America, and South America). We calculated the av- eraged standardized regression coefficients based on multimodel Supplementary Materials inference to estimate the relative effects of environmental variables. This PDF file includes: Figs. S1 to S25 All analyses were performed in R using the packages spatialreg and Tables S1 to S4 spdep (114) for SAR models, MuMIn (115) for multimodel infer- ence, and SpatialPack (116) for the spatially corrected t test. REFERENCES AND NOTES Sensitivity analyses 1. G. T. Pecl, M. B. Araujo, J. D. Bell, J. Blanchard, T. C. Bonebrake, I. C. Chen, T. D. Clark, To test the influence of trait imputation on functional beta-diversity R. K. Colwell, F. Danielsen, B. Evengard, L. Falconi, S. Ferrier, S. Frusher, R. A. Garcia, and its relationship with phylogenetic beta-diversity, we repeated R. B. Griffis, A. J. Hobday, C. Janion-Scheepers, M. A. Jarzyna, S. Jennings, J. Lenoir, the calculation of functional and phylogenetic beta-diversity using H. I. Linnetved, V. Y. Martin, P. C. McCormack, J. McDonald, N. J. Mitchell, T. Mustonen, J. M. Pandolfi, N. Pettorelli, E. Popova, S. A. Robinson, B. R. Scheffers, J. D. Shaw, C. J. Sorte, only species that have measured data for at least one or five traits. J. M. Strugnell, J. M. Sunday, M. N. Tuanmu, A. Verges, C. Villanueva, T. Wernberg, Because only 114 species had measured data for all eight traits, we E. Wapstra, S. E. Williams, Biodiversity redistribution under climate change: Impacts on did not perform a sensitivity analysis using only these species. We ecosystems and human well-being. Science 355, eaai9214 (2017). calculated correlations of functional and phylogenetic beta-diver- 2. C. H. Trisos, C. Merow, A. L. Pigot, The projected timing of abrupt ecological disruption sity based on all species with those based on species with at least from climate change. Nature 580, 496–501 (2020). 3. S. C. Brown, T. M. L. Wigley, B. L. Otto-Bliesner, C. Rahbek, D. A. Fordham, Persistent one or five measured traits. We also analyzed the relationship Quaternary climate refugia are hospices for biodiversity in the Anthropocene. Nat. Clim. between functional and phylogenetic beta-diversity using values Chang. 10, 244–248 (2020). based on all species and only species with at least one and five mea- 4. D. A. Fordham, S. T. Jackson, S. C. Brown, B. Huntley, B. W. Brook, D. Dahl-Jensen, sured traits, respectively, and assessed whether the relationship M. T. P. Gilbert, B. L. Otto-Bliesner, A. Svensson, S. Theodoridis, J. M. Wilmshurst, based on all species was consistent with that based on only J. C. Buettel, E. Canteri, M. McDowell, L. Orlando, J. Pilowsky, C. Rahbek, D. Nogues-Bravo, Using paleo-archives to safeguard biodiversity under climate change. Science 369, species with some measured trait data. Because Europe and North eabc5654 (2020). America had higher trait coverage than other continents (fig. S11), 5. B. Saladin, L. Pellissier, C. H. Graham, M. P. Nobis, N. Salamin, N. E. Zimmermann, Rapid the functional beta-diversity within these continents was less likely climate change results in long-lasting spatial homogenization of phylogenetic diversity. influenced by trait imputation. Thus, its relationship with phyloge- Nat. Commun. 11, 4663 (2020). netic beta-diversity tends to reflect their intrinsic association rather 6. J. C. Svenning, W. L. Eiserhardt, S. Normand, A. Ordonez, B. Sandel, The influence of than the consequence of trait imputation. We, therefore, performed paleoclimate on present-day patterns in biodiversity and ecosystems. Annu. Rev. Ecol. Evol. Syst. 46, 551–572 (2015). an analysis by comparing relationships between phylogenetic and 7. B. A. Hawkins, R. Field, H. V. Cornell, D. J. Currie, J.-F. Guégan, D. M. Kaufman, J. T. Kerr, functional beta-diversity within each of the six continents (Africa, G. G. Mittelbach, T. Oberdorff, E. M. O'Brien, E. E. Porter, J. R. G. Turner, Energy, water, and Asia, Australia, Europe, North America, and South America) to test broad-scale geographic patterns of species richness. Ecology 84, 3105–3117 (2003). whether they were less strongly correlated in Europe and 8. D. J. Currie, G. G. Mittelbach, H. V. Cornell, R. Field, J.-F. Guegan, B. A. Hawkins, North America. D. M. Kaufman, J. T. Kerr, T. Oberdorff, E. O'Brien, J. R. G. Turner, Predictions and tests of climate-based hypotheses of broad-scale variation in taxonomic richness. Ecol. Lett. 7, We then tested the sensitivity of our results to the values of the 1121–1134 (2004). parameter alpha in constructing alpha-hull ranges. The alpha pa- 9. J. Alahuhta, H. Antikainen, J. Hjort, A. Helm, J. Heino, Current climate overrides historical rameter determines the resolution of constructed alpha hulls, with effects on species richness and range size of freshwater plants in Europe and North a smaller value representing a higher resolution (81, 82). When America. J. Ecol. 108, 1262–1275 (2020). alpha approaches zero, the constructed shape is close to the original 10. R. Testolin, F. Attorre, P. Borchardt, R. F. Brand, H. Bruelheide, M. Chytrý, M. De Sanctis, point set, whereas the constructed alpha hulls will be the typical J. Dolezal, M. Finckh, S. Haider, A. Hemp, U. Jandt, M. Kessler, A. Y. Korolyuk, J. Lenoir, N. Makunina, G. P. Malanson, D. B. Montesinos-Tubée, J. Noroozi, A. Nowak, R. K. Peet, convex hull when alpha takes infinite values. In the main text, we G. Peyre, F. M. Sabatini, J. Šibík, P. Sklenář, S. P. Sylvester, K. Vassilev, R. Virtanen, W. Willner, reported results based on alpha-hull ranges estimated with the S. K. Wiser, E. G. Zibzeev, B. Jiménez-Alfaro, Global patterns and drivers of alpine plant alpha value of 6°. As a sensitivity analysis, we calculated beta-diver- species richness. Glob. Ecol. Biogeogr. 30, 1218–1231 (2021). sity based on alpha-hull ranges estimated with alpha values of 2°, 4°, 11. Z. Wang, J. Fang, Z. Tang, L. Shi, Geographical patterns in the beta diversity of China's and 10° and compared them with beta-diversity values reported in woody plants: The influence of space, environment and range size. Ecography 35, 1092–1102 (2012). the main text. 12. B. G. Weinstein, B. Tinoco, J. Luis Parra, L. M. Brown, J. A. McGuire, F. Gary Stiles, To test the sensitivity of our results to the sizes of moving C. H. Graham, Taxonomic, phylogenetic, and trait beta diversity in South American windows to define regions, we repeated analyses using a moving hummingbirds. Am. Nat. 184, 211–224 (2014). window of nine (3 × 3) grid cells. To control for different 13. I. R. McFadden, B. Sandel, C. Tsirogiannis, N. Morueta-Holme, J. C. Svenning, B. J. Enquist, numbers of cells within regions, we only included grid cells with N. J. B. Kraft, Temperature shapes opposing latitudinal gradients of plant taxonomic and phylogenetic β diversity. Ecol. Lett. 22, 1126–1135 (2019). at least five neighboring cells as focal cells to calculate beta-diversity. 14. M. F. Maxwell, F. Leprieur, J. P. Quimbayo, S. R. Floeter, M. G. Bender, Global patterns and For each focal cell, we resampled five grid cells from its eight neigh- drivers of beta diversity facets of reef fish faunas. J. Biogeogr. 49, 954–967 (2022). boring cells 200 times and calculated the average values for all 15. H. Qian, Y. Jin, F. Leprieur, X. Wang, T. Deng, Geographic patterns and environmental metrics of beta-diversity. We also calculated the deviation of phylo- correlates of taxonomic and phylogenetic beta diversity for large-scale angiosperm as- genetic and functional turnover and nestedness from their random semblages in China. Ecography 43, 1706–1716 (2020). expectations using the null model in our main analyses. Here, the 16. A. Baselga, Partitioning the turnover and nestedness components of beta diversity. Glob. Ecol. Biogeogr. 19, 134–143 (2010). site-specific regional species pool needed for the null model was all Xu et al., Sci. Adv. 9, eadd8553 (2023) 5 April 2023 14 of 18 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E 17. F. Leprieur, P. A. Tedesco, B. Hugueny, O. Beauchard, H. H. Durr, S. Brosse, T. Oberdorff, 41. S. Villéger, G. Grenouillet, S. Brosse, Decomposing functional β-diversity reveals that low Partitioning global patterns of freshwater fish beta diversity reveals contrasting signatures functional β-diversity is driven by low functional turnover in European fish assemblages. of past climate changes. Ecol. Lett. 14, 325–334 (2011). Glob. Ecol. Biogeogr. 22, 671–681 (2013). 18. J. C. Svenning, C. Flojgaard, A. Baselga, Climate, history and neutrality as drivers of 42. J. López-Pujol, F.-M. Zhang, H.-Q. Sun, T.-S. Ying, S. Ge, Centres of plant endemism in mammal beta diversity in Europe: Insights from multiscale deconstruction. J. Anim. Ecol. China: Places for survival or for speciation? J. Biogeogr. 38, 1267–1280 (2011). 80, 393–402 (2011). 43. S. Harrison, R. Noss, Endemism hotspots are linked to stable climatic refugia. Ann. Bot. 19. R. Dobrovolski, A. S. Melo, F. A. S. Cassemiro, J. A. F. Diniz-Filho, Climatic history and 119, 207–214 (2017). dispersal ability explain the relative importance of turnover and nestedness components 44. L. M. Lu, L. F. Mao, T. Yang, J. F. Ye, B. Liu, H. L. Li, M. Sun, J. T. Miller, S. Mathews, H. H. Hu, of beta diversity. Glob. Ecol. Biogeogr. 21, 191–197 (2012). Y. T. Niu, D. X. Peng, Y. H. Chen, S. A. Smith, M. Chen, K. L. Xiang, C. T. Le, V. C. Dang, A. M. Lu, 20. J. Soininen, J. Heino, J. Wang, A meta-analysis of nestedness and turnover components of P. S. Soltis, D. E. Soltis, J. H. Li, Z. D. Chen, Evolutionary history of the angiosperm flora of beta diversity across organisms and ecosystems. Glob. Ecol. Biogeogr. 27, 96–109 (2018). China. Nature 554, 234–238 (2018). 21. M. Dynesius, R. Jansson, Evolutionary consequences of changes in species' geographical 45. G. Hunt, M. J. Hopkins, S. Lidgard, Simple versus complex models of trait evolution and distributions driven by Milankovitch climate oscillations. Proc. Natl. Acad. Sci. U.S.A. 97, stasis as a response to environmental change. Proc. Natl. Acad. Sci. U.S.A. 112, 9115–9120 (2000). 4885–4890 (2015). 22. R. Jansson, Global patterns in endemism explained by past climatic change. Proc. Biol. Sci. 46. Z. Liu, M. Zhao, H. Zhang, T. Ren, C. Liu, N. He, Divergent response and adaptation of 270, 583–590 (2003). specific leaf area to environmental change at different spatio-temporal scales jointly 23. B. Sandel, L. Arge, B. Dalsgaard, R. G. Davies, K. J. Gaston, W. J. Sutherland, J. C. Svenning, improve plant survival. Glob. Chang. Biol. 29, 1144–1159 (2023). The influence of Late Quaternary climate-change velocity on species endemism. Science 47. B. Lhotsky, B. Kovács, G. Ónodi, A. Csecserits, T. Rédei, A. Lengyel, M. Kertész, Z. Botta- 334, 660–664 (2011). Dukát, Changes in assembly rules along a stress gradient from open dry grasslands to 24. N. Morueta-Holme, B. J. Enquist, B. J. McGill, B. Boyle, P. M. Jørgensen, J. E. Ott, R. K. Peet, wetlands. J. Ecol. 104, 507–517 (2016). I. Šímová, L. L. Sloat, B. Thiers, C. Violle, S. K. Wiser, S. Dolins, J. C. Donoghue II, N. J. B. Kraft, 48. W. L. Eiserhardt, F. Borchsenius, C. M. Plum, A. Ordonez, J.-C. Svenning, Climate-driven J. Regetz, M. Schildhauer, N. Spencer, J.-C. Svenning, Habitat area and climate stability extinctions shape the phylogenetic structure of temperate tree floras. Ecol. Lett. 18, determine geographical variation in plant species range sizes. Ecol. Lett. 16, 263–272 (2015). 1446–1454 (2013). 49. A. Ordonez, J. C. Svenning, Geographic patterns in functional diversity deficits are linked 25. W. Xu, J. C. Svenning, G. Chen, B. Chen, J. Huang, K. Ma, Plant geographical range size and to glacial-interglacial climate stability and accessibility. Glob. Ecol. Biogeogr. 24, climate stability in China: Growth form matters. Glob. Ecol. Biogeogr. 27, 506–517 (2018). 826–837 (2015). 26. J. C. Svenning, F. Skov, Limited filling of the potential range in European tree species. Ecol. 50. J. C. Svenning, Deterministic Plio-Pleistocene extinctions in the European cool-temperate Lett. 7, 565–573 (2004). tree flora. Ecol. Lett. 6, 646–653 (2003). 27. S. Normand, R. E. Ricklefs, F. Skov, J. Bladt, O. Tackenberg, J.-C. Svenning, Postglacial 51. E. G. Brockerhoff, L. Barbaro, B. Castagneyrol, D. I. Forrester, B. Gardiner, J. R. González- migration supplements climate in determining plant species ranges in Europe. Proc. Biol. Olabarria, P. O. B. Lyver, N. Meurisse, A. Oxbrough, H. Taki, I. D. Thompson, F. van der Plas, Sci. 278, 3644–3653 (2011). H. Jactel, Forest biodiversity, ecosystem functioning and the provision of ecosystem 28. U. Herzschuh, H. J. Birks, T. Laepple, A. Andreev, M. Melles, J. Brigham-Grette, Glacial services. Biodivers. Conserv. 26, 3005–3035 (2017). legacies on interglacial vegetation at the Pliocene-Pleistocene transition in NE Asia. Nat. 52. B. J. Enquist, A. J. Abraham, M. B. J. Harfoot, Y. Malhi, C. E. Doughty, The megabiota are Commun. 7, 11967 (2016). disproportionately important for biosphere functioning. Nat. Commun. 11, 699 (2020). 29. A. Baselga, C. Gomez-Rodriguez, J. M. Lobo, Historical legacies in world amphibian di- 53. J. M. Serra-Diaz, B. J. Enquist, B. Maitner, C. Merow, J.-C. Svenning, Big data of tree species versity revealed by the turnover and nestedness components of beta diversity. PLOS ONE distributions: How big and how good? For. Ecosyst. 4, 30 (2017). 7, e32341 (2012). 54. A. Baselga, C. D. L. Orme, betapart: An R package for the study of beta diversity. Methods 30. J. Douda, A. Havrdová, J. Doudová, B. Mandák, Legacy of post-glacial colonization affects Ecol. Evol. 3, 808–812 (2012). β-diversity: Insights into local community assembly processes. J. Biogeogr. 45, 55. J. C. Svenning, F. Skov, Could the tree diversity pattern in Europe be generated by 1604–1615 (2018). postglacial dispersal limitation? Ecol. Lett. 10, 453–460 (2007). 31. G. C. Stevens, The latitudinal gradient in geographical range: How so many species 56. B. Jimenez-Alfaro, M. Girardello, M. Chytry, J. C. Svenning, W. Willner, J. C. Gegout, coexist in the tropics. Am. Nat. 133, 240–256 (1989). E. Agrillo, J. A. Campos, U. Jandt, Z. Kacki, U. Silc, M. Slezak, L. Tichy, I. Tsiripidis, 32. J. F. Colville, C. M. Beale, F. Forest, R. Altwegg, B. Huntley, R. M. Cowling, Plant richness, P. D. Turtureanu, M. Ujhazyova, T. Wohlgemuth, History and environment shape species turnover, and evolutionary diversity track gradients of stability and ecological opportu- pools and community diversity in European beech forests. Nat. Ecol. Evol. 2, nity in a megadiversity center. Proc. Natl. Acad. Sci. U.S.A. 117, 20027–20037 (2020). 483–490 (2018). 33. W. Ulrich, W. Kryszewski, P. Sewerniak, R. Puchałka, G. Strona, N. J. Gotelli, A compre- 57. L. Stewart, I. G. Alsos, C. Bay, A. L. Breen, C. Brochmann, N. Boulanger-Lapointe, hensive framework for the study of species co-occurrences, nestedness and turnover. O. Broennimann, H. Bültmann, P. K. Bøcher, C. Damgaard, F. J. A. Daniëls, D. Ehrich, Oikos 126, 1607–1616 (2017). P. B. Eidesen, A. Guisan, I. S. Jónsdóttir, J. Lenoir, P. C. le Roux, E. Lévesque, M. Luoto, 34. A. Menegotto, C. S. Dambros, S. A. Netto, The scale-dependent effect of environmental J. Nabe-Nielsen, P. Schönswetter, A. Tribsch, L. U. Tveraabak, R. Virtanen, D. A. Walker, filters on species turnover and nestedness in an estuarine benthic community. Ecology K. B. Westergaard, N. G. Yoccoz, J.-C. Svenning, M. Wisz, N. M. Schmidt, L. Pellissier, The 100, e02721 (2019). regional species richness and genetic diversity of Arctic vegetation reflect both past 35. E. E. Saupe, C. E. Myers, A. T. Peterson, J. Soberón, J. Singarayer, P. Valdes, H. Qiao, Non- glaciations and current climate. Glob. Ecol. Biogeogr. 25, 430–442 (2016). random latitudinal gradients in range size and niche breadth predicted by spatial patterns 58. B. J. Seliger, B. J. McGill, J. C. Svenning, J. L. Gill, Widespread underfilling of the potential of climate. Glob. Ecol. Biogeogr. 28, 928–942 (2019). ranges of North American trees. J. Biogeogr. 48, 359–371 (2021). 36. W. B. Xu, J. C. Svenning, G. K. Chen, M. G. Zhang, J. H. Huang, B. Chen, A. Ordonez, K. P. Ma, 59. S. Dullinger, W. Willner, C. Plutzar, T. Englisch, L. Schratt-Ehrendorfer, D. Moser, S. Ertl, Human activities have opposing effects on distributions of narrow-ranged and wide- F. Essl, H. Niklfeld, Post-glacial migration lag restricts range filling of plants in the Euro- spread plant species in China. Proc. Natl. Acad. Sci. U.S.A. 116, 26674–26681 (2019). pean Alps. Glob. Ecol. Biogeogr. 21, 829–840 (2012). 37. J. F. M. Rodrigues, J. A. F. Diniz-Filho, Dispersal is more important than climate in struc- 60. J. Hortal, J. A. Diniz-Filho, L. M. Bini, M. A. Rodriguez, A. Baselga, D. Nogues-Bravo, turing turtle communities across different biogeographical realms. J. Biogeogr. 44, T. F. Rangel, B. A. Hawkins, J. M. Lobo, Ice age climate, evolutionary constraints and di- 2109–2120 (2017). versity patterns of European dung beetles. Ecol. Lett. 14, 741–748 (2011). 38. F. P. Peixoto, F. Villalobos, A. S. Melo, J. A. F. Diniz-Filho, R. Loyola, T. F. Rangel, 61. W. L. Eiserhardt, F. Borchsenius, B. Sandel, W. D. Kissling, J.-C. Svenning, Late Cenozoic M. V. Cianciaruso, Geographical patterns of phylogenetic beta-diversity components in climate and the phylogenetic structure of regional conifer floras world-wide. Glob. Ecol. terrestrial mammals. Glob. Ecol. Biogeogr. 26, 573–583 (2017). Biogeogr. 24, 1136–1148 (2015). 39. J. N. Pinto-Ledezma, D. J. Larkin, J. Cavender-Bares, Patterns of beta diversity of vascular 62. N. G. Beckman, J. M. Bullock, R. Salguero-Gómez, High dispersal ability is related to fast plants and their correspondence with biome boundaries across North America. Front. life-history strategies. J. Ecol. 106, 1349–1362 (2018). Ecol. Evol. 6, 194 (2018). 63. L. T. Lancaster, A. M. Humphreys, Global variation in the thermal tolerances of plants. Proc. 40. F. Leprieur, C. Albouy, J. De Bortoli, P. F. Cowman, D. R. Bellwood, D. Mouillot, Quantifying Natl. Acad. Sci. U.S.A. 117, 13580–13587 (2020). phylogenetic beta diversity: Distinguishing between 'true' turnover of lineages and 64. S. Diaz, J. Kattge, J. H. Cornelissen, I. J. Wright, S. Lavorel, S. Dray, B. Reu, M. Kleyer, C. Wirth, phylogenetic diversity gradients. PLOS ONE 7, e42760 (2012). I. C. Prentice, E. Garnier, G. Bonisch, M. Westoby, H. Poorter, P. B. Reich, A. T. Moles, J. Dickie, Xu et al., Sci. Adv. 9, eadd8553 (2023) 5 April 2023 15 of 18 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E A. N. Gillison, A. E. Zanne, J. Chave, S. J. Wright, S. N. Sheremet'ev, H. Jactel, C. Baraloto, 83. B. Vilela, F. Villalobos, letsR: A new R package for data handling and analysis in macro- B. Cerabolini, S. Pierce, B. Shipley, D. Kirkup, F. Casanoves, J. S. Joswig, A. Gunther, ecology. Methods Ecol. Evol. 6, 1229–1234 (2015). V. Falczuk, N. Ruger, M. D. Mahecha, L. D. Gorne, The global spectrum of plant form and 84. G. Feng, J.-C. Svenning, X. Mi, Q. Jia, M. Rao, H. Ren, D. P. Bebber, K. Ma, Anthropogenic function. Nature 529, 167–171 (2016). disturbance shapes phylogenetic and functional tree community structure in a sub- 65. A. Ordonez, J. C. Svenning, The potential role of species and functional composition in tropical forest. For. Ecol. Manage. 313, 188–198 (2014). generating historical constraints on ecosystem processes. Glob. Ecol. Biogeogr. 29, 85. W. Y. Guo, J. M. Serra-Diaz, F. Schrodt, W. L. Eiserhardt, B. S. Maitner, C. Merow, C. Violle, 207–219 (2020). M. Anand, M. Belluau, H. H. Bruun, C. Byun, J. A. Catford, B. E. L. Cerabolini, E. Chacon- 66. B. Blonder, B. J. Enquist, B. J. Graae, J. Kattge, B. S. Maitner, N. Morueta-Holme, A. Ordonez, Madrigal, D. Ciccarelli, J. H. C. Cornelissen, A. T. Dang-Le, A. de Frutos, A. S. Dias, I. Šímová, J. Singarayer, J. C. Svenning, P. J. Valdes, C. Violle, Late Quaternary climate A. B. Giroldo, K. Guo, A. G. Gutierrez, W. Hattingh, T. He, P. Hietz, N. Hough-Snee, S. Jansen, legacies in contemporary plant functional composition. Glob. Chang. Biol. 24, J. Kattge, T. Klein, B. Komac, N. J. B. Kraft, K. Kramer, S. Lavorel, C. H. Lusk, A. R. Martin, 4827–4840 (2018). M. Mencuccini, S. T. Michaletz, V. Minden, A. S. Mori, U. Niinemets, Y. Onoda, J. Penuelas, 67. H. Qian, T. Deng, Y. Jin, L. Mao, D. Zhao, R. E. Ricklefs, Phylogenetic dispersion and di- V. D. Pillar, J. Pisek, B. J. M. Robroek, B. Schamp, M. Slot, E. E. Sosinski Jr., versity in regional assemblages of seed plants in China. Proc. Natl. Acad. Sci. U.S.A. 116, N. A. Soudzilovskaia, N. Thiffault, P. van Bodegom, F. van der Plas, I. J. Wright, W. B. Xu, 23192–23201 (2019). J. Zheng, B. J. Enquist, J. C. Svenning, High exposure of global tree diversity to human 68. J. J. Wiens, M. J. Donoghue, Historical biogeography, ecology and species richness. Trends pressure. Proc. Natl. Acad. Sci. U.S.A. 119, e2026733119 (2022). Ecol. Evol. 19, 639–644 (2004). 86. P. Keil, J. M. Chase, Global patterns and drivers of tree diversity integrated across a 69. M. L. McKinney, J. L. Lockwood, Biotic homogenization: A few winners replacing many continuum of spatial grains. Nat. Ecol. Evol. 3, 390–399 (2019). losers in the next mass extinction. Trends Ecol. Evol. 14, 450–453 (1999). 87. A. Mauri, G. Strona, J. San-Miguel-Ayanz, EU-Forest, a high-resolution tree occurrence 70. G. N. Daskalova, I. H. Myers-Smith, A. D. Bjorkman, S. A. Blowes, S. R. Supp, A. E. Magurran, dataset for Europe. Sci Data 4, 160123 (2017). M. Dornelas, Landscape-scale forest loss as a catalyst of population and biodiversity 88. E. L. Little, Digital Representation of “Atlas of United States Trees”. (US Geological Survey, change. Science 368, 1341–1347 (2020). Denver, 1999). 71. M. Asefa, M. Cao, G. Zhang, X. Ci, J. Li, J. Yang, Environmental filtering structures tree 89. B. Sandel, P. Weigelt, H. Kreft, G. Keppel, M. T. van der Sande, S. Levin, S. Smith, D. Craven, functional traits combination and lineages across space in tropical tree assemblages. Sci. T. M. Knight, Current climate, isolation and history drive global patterns of tree phylo- Rep. 7, 132 (2017). genetic endemism. Glob. Ecol. Biogeogr. 29, 4–15 (2020). 72. C. M. Caruso, H. Maherali, R. A. Martin, A meta-analysis of natural selection on plant 90. S. A. Smith, J. W. Brown, Constructing a broadly inclusive seed plant phylogeny. Am. J. Bot. functional traits. Int. J. Plant Sci. 181, 44–55 (2020). 105, 302–314 (2018). 73. T. Conradi, K. Van Meerbeek, A. Ordonez, J. C. Svenning, Biogeographic historical legacies 91. S. Magallon, S. Gomez-Acevedo, L. L. Sanchez-Reyes, T. Hernandez-Hernandez, A meta- in the net primary productivity of Northern Hemisphere forests. Ecol. Lett. 23, calibrated time-tree documents the early rise of flowering plant phylogenetic diversity. 800–810 (2020). New Phytol. 207, 437–453 (2015). 74. E. Beech, M. Rivers, S. Oldfield, P. P. Smith, GlobalTreeSearch: The first complete global 92. L. J. Revell, phytools: An R package for phylogenetic comparative biology (and other database of tree species and country distributions. J. Sustain. For. 36, 454–489 (2017). things). Methods Ecol. Evol. 3, 217–223 (2012). 75. B. Boyle, N. Hopkins, Z. Lu, J. A. R. Garay, D. Mozzherin, T. Rees, N. Matasci, M. L. Narro, 93. D. Li, L. Trotta, H. E. Marx, J. M. Allen, M. Sun, D. E. Soltis, P. S. Soltis, R. P. Guralnick, B. Baiser, W. H. Piel, S. J. Mckay, S. Lowry, C. Freeland, R. K. Peet, B. J. Enquist, The taxonomic name For common community phylogenetic analyses, go ahead and use synthesis phylogenies. resolution service: An online tool for automated standardization of plant names. BMC Ecology 100, e02788 (2019). Bioinformatics 14, 16 (2013). 94. J. Kattge, G. Bonisch, S. Diaz, S. Lavorel, I. C. Prentice, P. Leadley, S. Tautenhahn, 76. GBIF.org, GBIF Occurrence Download (2022);https://doi.org/10.15468/dd.4jvnmv. G. D. A. Werner, T. Aakala, M. Abedi, A. T. R. Acosta, G. C. Adamidis, K. Adamson, M. Aiba, 77. B. J. Enquist, X. Feng, B. Boyle, B. Maitner, E. A. Newman, P. M. Jørgensen, P. R. Roehrdanz, C. H. Albert, J. M. Alcantara, C. C. Alcazar, I. Aleixo, H. Ali, B. Amiaud, C. Ammer, B. M. Thiers, J. R. Burger, R. T. Corlett, T. L. P. Couvreur, G. Dauby, J. C. Donoghue, W. Foden, M. M. Amoroso, M. Anand, C. Anderson, N. Anten, J. Antos, D. M. G. Apgaua, T. L. Ashman, J. C. Lovett, P. A. Marquet, C. Merow, G. Midgley, N. Morueta-Holme, D. M. Neves, D. H. Asmara, G. P. Asner, M. Aspinwall, O. Atkin, I. Aubin, L. Baastrup-Spohr, K. Bahalkeh, A. T. Oliveira-Filho, N. J. B. Kraft, D. S. Park, R. K. Peet, M. Pillet, J. M. Serra-Diaz, B. Sandel, M. Bahn, T. Baker, W. J. Baker, J. P. Bakker, D. Baldocchi, J. Baltzer, A. Banerjee, A. Baranger, M. Schildhauer, I. Šímová, C. Violle, J. J. Wieringa, S. K. Wiser, L. Hannah, J.-C. Svenning, J. Barlow, D. R. Barneche, Z. Baruch, D. Bastianelli, J. Battles, W. Bauerle, M. Bauters, B. J. McGill, The commonness of rarity: Global and future distribution of rarity across land E. Bazzato, M. Beckmann, H. Beeckman, C. Beierkuhnlein, R. Bekker, G. Belfry, M. Belluau, plants. Sci. Adv. 5, eaaz0414 (2019). M. Beloiu, R. Benavides, L. Benomar, M. L. Berdugo-Lattke, E. Berenguer, R. Bergamin, 78. DryFlor, K. Banda-R, A. Delgado-Salinas, K. G. Dexter, R. Linares-Palomino, A. Oliveira-Filho, J. Bergmann, M. B. Carlucci, L. Berner, M. Bernhardt-Romermann, C. Bigler, A. D. Bjorkman, D. Prado, M. Pullan, C. Quintana, R. Riina, G. M. Rodríguez M., J. Weintritt, P. Acevedo- C. Blackman, C. Blanco, B. Blonder, D. Blumenthal, K. T. Bocanegra-Gonzalez, P. Boeckx, Rodríguez, J. Adarve, E. Álvarez, A. Aranguren B, J. C. Arteaga, G. Aymard, A. Castaño, S. Bohlman, K. Bohning-Gaese, L. Boisvert-Marsh, W. Bond, B. Bond-Lamberty, A. Boom, N. Ceballos-Mago, Á. Cogollo, H. Cuadros, F. Delgado, W. Devia, H. Dueñas, L. Fajardo, C. C. F. Boonman, K. Bordin, E. H. Boughton, V. Boukili, D. Bowman, S. Bravo, M. R. Brendel, Á. Fernández, M. Á. Fernández, J. Franklin, E. H. Freid, L. A. Galetti, R. Gonto, R. González- M. R. Broadley, K. A. Brown, H. Bruelheide, F. Brumnich, H. H. Bruun, D. Bruy, M., R. Graveson, E. H. Helmer, Á. Idárraga, R. López, H. Marcano-Vega, O. G. Martínez, S. W. Buchanan, S. F. Bucher, N. Buchmann, R. Buitenwerf, D. E. Bunker, J. Burger, H. M. Maturo, M. McDonald, K. McLaren, O. Melo, F. Mijares, V. Mogni, D. Molina, N. del Pilar S. Burrascano, D. Burslem, B. J. Butterfield, C. Byun, M. Marques, M. C. Scalon, Moreno, J. M. Nassar, D. M. Neves, L. J. Oakley, M. Oatham, A. R. Olvera-Luna, F. F. Pezzini, M. Caccianiga, M. Cadotte, M. Cailleret, J. Camac, J. J. Camarero, C. Campany, O. J. R. Dominguez, M. E. Ríos, O. Rivera, N. Rodríguez, A. Rojas, T. Särkinen, R. Sánchez, G. Campetella, J. A. Campos, L. Cano-Arboleda, R. Canullo, M. Carbognani, F. Carvalho, M. Smith, C. Vargas, B. Villanueva, R. T. Pennington, Plant diversity patterns in neotropical F. Casanoves, B. Castagneyrol, J. A. Catford, J. Cavender-Bares, B. E. L. Cerabolini, dry forests and their conservation implications. Science 353, 1383–1387 (2016). M. Cervellini, E. Chacon-Madrigal, K. Chapin, F. S. Chapin, S. Chelli, S. C. Chen, A. Chen, P. Cherubini, F. Chianucci, B. Choat, K. S. Chung, M. Chytry, D. Ciccarelli, L. Coll, C. G. Collins, 79. G. Dauby, R. Zaiss, A. Blach-Overgaard, L. Catarino, T. Damen, V. Deblauwe, S. Dessein, L. Conti, D. Coomes, J. H. C. Cornelissen, W. K. Cornwell, P. Corona, M. Coyea, J. Craine, J. Dransfield, V. Droissart, M. C. Duarte, H. Engledow, G. Fadeur, R. Figueira, R. E. Gereau, D. Craven, J. Cromsigt, A. Csecserits, K. Cufar, M. Cuntz, A. C. da Silva, K. M. Dahlin, O. J. Hardy, D. J. Harris, J. de Heij, S. Janssens, Y. Klomberg, A. C. Ley, B. A. Mackinder, M. Dainese, I. Dalke, M. D. Fratte, A. T. Dang-Le, J. Danihelka, M. Dannoura, S. Dawson, P. Meerts, J. L. van de Poel, B. Sonke, M. S. Sosef, T. Stevart, P. Stoffelen, J. C. Svenning, A. J. de Beer, A. De Frutos, J. R. De Long, B. Dechant, S. Delagrange, N. Delpierre, P. Sepulchre, X. van der Burgt, J. J. Wieringa, T. L. Couvreur, RAINBIO: A mega-database of G. Derroire, A. S. Dias, M. H. Diaz-Toribio, P. G. Dimitrakopoulos, M. Dobrowolski, D. Doktor, tropical African vascular plants distributions. PhytoKeys 74, 1–18 (2016). P. Drevojan, N. Dong, J. Dransfield, S. Dressler, L. Duarte, E. Ducouret, S. Dullinger, 80. B. Pateiro-López, A. Rodríguez-Casal, Generalizing the convex hull of a sample: The R W. Durka, R. Duursma, O. Dymova, A. E-Vojtkó, R. L. Eckstein, H. Ejtehadi, J. Elser, T. Emilio, Package alphahull. J. Stat. Softw. 34, 1–28 (2010). K. Engemann, M. B. Erfanian, A. Erfmeier, A. Esquivel-Muelbert, G. Esser, M. Estiarte, 81. M. A. Burgman, J. C. Fox, Bias in species range estimates from minimum convex polygons: T. F. Domingues, W. F. Fagan, J. Fagundez, D. S. Falster, Y. Fan, J. Fang, E. Farris, F. Fazlioglu, Implications for conservation and options for improved planning. Anim. Conserv. 6, Y. Feng, F. Fernandez-Mendez, C. Ferrara, J. Ferreira, A. Fidelis, B. Finegan, J. Firn, 19–28 (2003). T. J. Flowers, D. F. B. Flynn, V. Fontana, E. Forey, C. Forgiarini, L. Francois, M. Frangipani, 82. E. García-Roselló, C. Guisande, A. Manjarrés-Hernández, J. González-Dacosta, J. Heine, D. Frank, C. Frenette-Dussault, G. T. Freschet, E. L. Fry, N. M. Fyllas, G. G. Mazzochini, P. Pelayo-Villamil, L. González-Vilas, R. P. Vari, A. Vaamonde, C. Granado-Lorencio, S. Gachet, R. Gallagher, G. Ganade, F. Ganga, P. Garcia-Palacios, V. Gargaglione, E. Garnier, J. M. Lobo, Can we derive macroecological patterns from primary Global Biodiversity J. L. Garrido, A. L. de Gasper, G. Gea-Izquierdo, D. Gibson, A. N. Gillison, A. Giroldo, Information Facility data? Glob. Ecol. Biogeogr. 24, 335–347 (2015). M. C. Glasenhardt, S. Gleason, M. Gliesch, E. Goldberg, B. Goldel, E. Gonzalez-Akre, Xu et al., Sci. Adv. 9, eadd8553 (2023) 5 April 2023 16 of 18 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E J. L. Gonzalez-Andujar, A. Gonzalez-Melo, A. Gonzalez-Robles, B. J. Graae, E. Granda, 95. I. Aubin, C. Messier, S. Gachet, K. Lawrence, D. McKenney, A. Arseneault, W. Bell, L. De S. Graves, W. A. Green, T. Gregor, N. Gross, G. R. Guerin, A. Gunther, A. G. Gutierrez, Grandpré, B. Shipley, J. Ricard, A. D. Munson, TOPIC–traits of plants in Canada (Natural L. Haddock, A. Haines, J. Hall, A. Hambuckers, W. Han, S. P. Harrison, W. Hattingh, Resources Canada, Canadian Forest Service, 2012); www.nrcan.gc.ca/forests/research- J. E. Hawes, T. He, P. He, J. M. Heberling, A. Helm, S. Hempel, J. Hentschel, B. Herault, centres/glfc/20303 A. M. Heres, K. Herz, M. Heuertz, T. Hickler, P. Hietz, P. Higuchi, A. L. Hipp, A. Hirons, 96. F. Schrodt, J. Kattge, H. Shan, F. Fazayeli, J. Joswig, A. Banerjee, M. Reichstein, G. Bönisch, M. Hock, J. A. Hogan, K. Holl, O. Honnay, D. Hornstein, E. Hou, N. Hough-Snee, S. Díaz, J. Dickie, A. Gillison, A. Karpatne, S. Lavorel, P. Leadley, C. B. Wirth, I. J. Wright, K. A. Hovstad, T. Ichie, B. Igic, E. Illa, M. Isaac, M. Ishihara, L. Ivanov, L. Ivanova, C. M. Iversen, S. J. Wright, P. B. Reich, BHPMF - a hierarchical Bayesian approach to gap-filling and trait J. Izquierdo, R. B. Jackson, B. Jackson, H. Jactel, A. M. Jagodzinski, U. Jandt, S. Jansen, prediction for macroecology and functional biogeography. Glob. Ecol. Biogeogr. 24, T. Jenkins, A. Jentsch, J. R. P. Jespersen, G. F. Jiang, J. L. Johansen, D. Johnson, E. J. Jokela, 1510–1521 (2015). C. A. Joly, G. J. Jordan, G. S. Joseph, D. Junaedi, R. R. Junker, E. Justes, R. Kabzems, J. Kane, 97. C. Penone, A. D. Davidson, K. T. Shoemaker, M. Di Marco, C. Rondinini, T. M. Brooks, Z. Kaplan, T. Kattenborn, L. Kavelenova, E. Kearsley, A. Kempel, T. Kenzo, A. Kerkhoff, B. E. Young, C. H. Graham, G. C. Costa, Imputation of missing data in life-history trait M. I. Khalil, N. L. Kinlock, W. D. Kissling, K. Kitajima, T. Kitzberger, R. Kjoller, T. Klein, datasets: Which approach performs the best? Methods Ecol. Evol. 5, 961–970 (2014). M. Kleyer, J. Klimesova, J. Klipel, B. Kloeppel, S. Klotz, J. M. H. Knops, T. Kohyama, F. Koike, 98. J. A. F. Diniz-Filho, C. E. R. de Sant'Ana, L. M. Bini, An eigenvector method for estimating J. Kollmann, B. Komac, K. Komatsu, C. Konig, N. J. B. Kraft, K. Kramer, H. Kreft, I. Kuhn, phylogenetic inertia. Evolution 52, 1247–1262 (1998). D. Kumarathunge, J. Kuppler, H. Kurokawa, Y. Kurosawa, S. Kuyah, J. P. Laclau, B. Lafleur, 99. M. Pagel, Inferring the historical patterns of biological evolution. Nature 401, E. Lallai, E. Lamb, A. Lamprecht, D. J. Larkin, D. Laughlin, Y. Le Bagousse-Pinguet, G. le 877–884 (1999). Maire, P. C. le Roux, E. le Roux, T. Lee, F. Lens, S. L. Lewis, B. Lhotsky, Y. Li, X. Li, 100. A. Baselga, Multiple site dissimilarity quantifies compositional heterogeneity among J. W. Lichstein, M. Liebergesell, J. Y. Lim, Y. S. Lin, J. C. Linares, C. Liu, D. Liu, U. Liu, several sites, while average pairwise dissimilarity may be misleading. Ecography 36, S. Livingstone, J. Llusia, M. Lohbeck, A. Lopez-Garcia, G. Lopez-Gonzalez, Z. Lososova, 124–128 (2013). F. Louault, B. A. Lukacs, P. Lukes, Y. Luo, M. Lussu, S. Ma, C. M. R. Pereira, M. Mack, V. Maire, A. Makela, H. Makinen, A. C. M. Malhado, A. Mallik, P. Manning, S. Manzoni, Z. Marchetti, 101. J. A. Bryant, C. Lamanna, H. Morlon, A. J. Kerkhoff, B. J. Enquist, J. L. Green, Microbes on L. Marchino, V. Marcilio-Silva, E. Marcon, M. Marignani, L. Markesteijn, A. Martin, mountainsides: Contrasting elevational patterns of bacterial and plant diversity. Proc. C. Martinez-Garza, J. Martinez-Vilalta, T. Maskova, K. Mason, N. Mason, T. J. Massad, Natl. Acad. Sci. U.S.A. 105, 11505–11511 (2008). J. Masse, I. Mayrose, J. McCarthy, M. L. McCormack, K. McCulloh, I. R. McFadden, B. J. McGill, 102. A. Šizling, P. Keil, E. Tjørve, K. Tjørve, J. Žárský, D. Storch, Mathematically and biologically M. Y. McPartland, J. S. Medeiros, B. Medlyn, P. Meerts, Z. Mehrabi, P. Meir, F. P. L. Melo, consistent framework for presence-absence pairwise indices. bioRxiv 452244 M. Mencuccini, C. Meredieu, J. Messier, I. Meszaros, J. Metsaranta, S. T. Michaletz, (2022).https://doi.org/10.1101/2021.07.14.452244. C. Michelaki, S. Migalina, R. Milla, J. E. D. Miller, V. Minden, R. Ming, K. Mokany, A. T. Moles, 103. N. G. Swenson, Phylogenetic beta diversity metrics, trait evolution and inferring the A. t. Molnar, J. Molofsky, M. Molz, R. A. Montgomery, A. Monty, L. Moravcova, A. Moreno- functional beta diversity of communities. PLOS ONE 6, e21264 (2011). Martinez, M. Moretti, A. S. Mori, S. Mori, D. Morris, J. Morrison, L. Mucina, S. Mueller, 104. S. Mammola, C. P. Carmona, T. Guillerme, P. Cardoso, Concepts and applications in C. D. Muir, S. C. Muller, F. Munoz, I. H. Myers-Smith, R. W. Myster, M. Nagano, S. Naidu, functional diversity. Funct. Ecol. 35, 1869–1885 (2021). A. Narayanan, B. Natesan, L. Negoita, A. S. Nelson, E. L. Neuschulz, J. Ni, G. Niedrist, J. Nieto, 105. D. Müllner, fastcluster: Fast hierarchical, agglomerative clustering routines for R and U. Niinemets, R. Nolan, H. Nottebrock, Y. Nouvellon, A. Novakovskiy, N. Nutrient, Python. J. Stat. Softw. 53, 1–18 (2013). K. O. Nystuen, A. O'Grady, K. O'Hara, A. O'Reilly-Nugent, S. Oakley, W. Oberhuber, 106. H. Kreft, W. Jetz, Global patterns and determinants of vascular plant diversity. Proc. Natl. T. Ohtsuka, R. Oliveira, K. Ollerer, M. E. Olson, V. Onipchenko, Y. Onoda, R. E. Onstein, Acad. Sci. U.S.A. 104, 5925–5930 (2007). J. C. Ordonez, N. Osada, I. Ostonen, G. Ottaviani, S. Otto, G. E. Overbeck, W. A. Ozinga, 107. C. M. Kennedy, J. R. Oakleaf, D. M. Theobald, S. Baruch-Mordo, J. Kiesecker, Managing the A. T. Pahl, C. E. T. Paine, R. J. Pakeman, A. C. Papageorgiou, E. Parfionova, M. Partel, middle: A shift in conservation priorities based on the global human modification gra- M. Patacca, S. Paula, J. Paule, H. Pauli, J. G. Pausas, B. Peco, J. Penuelas, A. Perea, P. L. Peri, dient. Glob. Chang. Biol. 25, 811–826 (2019). A. C. Petisco-Souza, A. Petraglia, A. M. Petritan, O. L. Phillips, S. Pierce, V. D. Pillar, J. Pisek, 108. J. L. Brown, D. J. Hill, A. M. Dolan, A. C. Carnaval, A. M. Haywood, PaleoClim, high spatial A. Pomogaybin, H. Poorter, A. Portsmuth, P. Poschlod, C. Potvin, D. Pounds, A. S. Powell, resolution paleoclimate surfaces for global land areas. Sci. Data 5, 180254 (2018). S. A. Power, A. Prinzing, G. Puglielli, P. Pysek, V. Raevel, A. Rammig, J. Ransijn, C. A. Ray, 109. R. J. Hijmans, S. E. Cameron, J. L. Parra, P. G. Jones, A. Jarvis, Very high resolution inter- P. B. Reich, M. Reichstein, D. E. B. Reid, M. Rejou-Mechain, V. R. de Dios, S. Ribeiro, polated climate surfaces for global land areas. Int. J. Climatol. 25, 1965–1978 (2005). S. Richardson, K. Riibak, M. C. Rillig, F. Riviera, E. M. R. Robert, S. Roberts, B. Robroek, A. Roddy, A. V. Rodrigues, A. Rogers, E. Rollinson, V. Rolo, C. Romermann, D. Ronzhina, 110. P. Dutilleul, P. Clifford, S. Richardson, D. Hemon, Modifying the t test for assessing the C. Roscher, J. A. Rosell, M. F. Rosenfield, C. Rossi, D. B. Roy, S. Royer-Tardif, N. Ruger, R. Ruiz- correlation between two spatial processes. Biometrics 49, 305–314 (1993). Peinado, S. B. Rumpf, G. M. Rusch, M. Ryo, L. Sack, A. Saldana, B. Salgado-Negret, 111. C. F. Dormann, J. M. McPherson, M. B. Araújo, R. Bivand, J. Bolliger, G. Carl, R. G. Davies, R. Salguero-Gomez, I. Santa-Regina, A. C. Santacruz-Garcia, J. Santos, J. Sardans, A. Hirzel, W. Jetz, W. Daniel Kissling, I. Kühn, R. Ohlemüller, P. R. Peres-Neto, B. Reineking, B. Schamp, M. Scherer-Lorenzen, M. Schleuning, B. Schmid, M. Schmidt, S. Schmitt, B. Schröder, F. M. Schurr, R. Wilson, Methods to account for spatial autocorrelation in the J. V. Schneider, S. D. Schowanek, J. Schrader, F. Schrodt, B. Schuldt, F. Schurr, G. S. Garvizu, analysis of species distributional data: A review. Ecography 30, 609–628 (2007). M. Semchenko, C. Seymour, J. C. Sfair, J. M. Sharpe, C. S. Sheppard, S. Sheremetiev, 112. W. D. Kissling, G. Carl, Spatial autocorrelation and the selection of simultaneous autor- S. Shiodera, B. Shipley, T. A. Shovon, A. Siebenkas, C. Sierra, V. Silva, M. Silva, T. Sitzia, egressive models. Glob. Ecol. Biogeogr. 17, 59–71 (2008). H. Sjoman, M. Slot, N. G. Smith, D. Sodhi, P. Soltis, D. Soltis, B. Somers, G. Sonnier, 113. K. P. Burnham, D. R. Anderson, Model selection and multi-model inference: A practical in- M. V. Sorensen, E. E. Sosinski Jr., N. A. Soudzilovskaia, A. F. Souza, M. Spasojevic, formation-theoretic approach (Springer, 2002). M. G. Sperandii, A. B. Stan, J. Stegen, K. Steinbauer, J. G. Stephan, F. Sterck, D. B. Stojanovic, 114. R. Bivand, A. M. Altman., L. Anselin, R. Assunção, O. Berke, A. Bernat, G. Blanchet, T. Strydom, M. L. Suarez, J. C. Svenning, I. Svitkova, M. Svitok, M. Svoboda, E. Swaine, E. Blankmeyer, M. Carvalho, B. Christensen, Y. Chun, C. Dormann, S. Dray, V. Gómez-Rubio, N. Swenson, M. Tabarelli, K. Takagi, U. Tappeiner, R. Tarifa, S. Tauugourdeau, M. Gubri, R. Halbersma, E. Krainski, P. Legendre, N. Lewin-Koh, A. Li, H. Li, J. Ma, A. Mallik, C. Tavsanoglu, M. Te Beest, L. Tedersoo, N. Thiffault, D. Thom, E. Thomas, K. Thompson, G. Millo, W. Mueller, H. Ono, P. Peres-Neto, G. Piras, M. Reder, M. Tiefelsdorf, R. Westerholt, P. E. Thornton, W. Thuiller, L. Tichy, D. Tissue, M. G. Tjoelker, D. Y. P. Tng, J. Tobias, P. Torok, D. Yu, Package ‘spdep’. Spatial dependence: Weighting schemes, statistics, R package T. Tarin, J. M. Torres-Ruiz, B. Tothmeresz, M. Treurnicht, V. Trivellone, F. Trolliet, V. Trotsiuk, version 1.1-5. (R Foundation for Statistical Computing, 2020). J. L. Tsakalos, I. Tsiripidis, N. Tysklind, T. Umehara, V. Usoltsev, M. Vadeboncoeur, J. Vaezi, 115. K. Barton, Package ‘MuMIn'. Model selection and model averaging base on information F. Valladares, J. Vamosi, P. M. van Bodegom, M. van Breugel, E. Van Cleemput, M. van de criteria, R package version 1.43.17 (R Foundation for Statistical Computing, 2020). Weg, S. van der Merwe, F. van der Plas, M. T. van der Sande, M. van Kleunen, K. Van 116. F. Osorio, R. Vallejos, F. Cuevas, D. Mancilla, SpatialPack: Package for analysis of spatial Meerbeek, M. Vanderwel, K. A. Vanselow, A. Varhammar, L. Varone, M. Y. V. Valderrama, data, R package version 0.3-8196 (R Foundation for Statistical Computing, 2020). K. Vassilev, M. Vellend, E. J. Veneklaas, H. Verbeeck, K. Verheyen, A. Vibrans, I. Vieira, J. Villacis, C. Violle, P. Vivek, K. Wagner, M. Waldram, A. Waldron, A. P. Walker, M. Waller, Acknowledgments: We thank all of the authors who contributed their data or gave us G. Walther, H. Wang, F. Wang, W. Wang, H. Watkins, J. Watkins, U. Weber, J. T. Weedon, permission to include it in this study. We thank TRY contributors for sharing their data. This L. Wei, P. Weigelt, E. Weiher, A. W. Wells, C. Wellstein, E. Wenk, M. Westoby, A. Westwood, work was also conducted as a part of the BIEN Working Group. We thank all the data P. J. White, M. Whitten, M. Williams, D. E. Winkler, K. Winter, C. Womack, I. J. Wright, contributors and numerous herbaria who have contributed their data to various data compiling S. J. Wright, J. Wright, B. X. Pinho, F. Ximenes, T. Yamada, K. Yamaji, R. Yanai, N. Yankov, organizations (see the herbarium list below) for the invaluable data and support provided to B. Yguel, K. J. Zanini, A. E. Zanne, D. Zeleny, Y. P. Zhao, J. Zheng, J. Zheng, K. Zieminska, BIEN. We thank the New York Botanical Garden; Missouri Botanical Garden; Utrecht Herbarium; C. R. Zirbel, G. Zizka, I. C. Zo-Bi, G. Zotz, C. Wirth, TRY plant trait database – enhanced the UNC Herbarium; and GBIF, REMIB, and SpeciesLink. The staff at CyVerse provided critical coverage and open access. Glob. Chang. Biol. 26, 119–188 (2020). computational assistance. We acknowledge the herbaria that contributed data to this work: A, Xu et al., Sci. Adv. 9, eadd8553 (2023) 5 April 2023 17 of 18 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E AAH, AAS, AAU, ABH, ACAD, ACOR, AD, AFS, AK, AKPM, ALCB, ALTA, ALU, AMD, AMES, AMNH, Foundation (AUFF) Starting Grant (AUFF-F-2018-7-8). J.-C.S. acknowledges support from the AMO, ANGU, ANSM, ANSP, AQP, ARAN, ARIZ, AS, ASDM, ASU, AUT, AV, AWH, B, BA, BAA, BAB, Danish Council for Independent Research|Natural Sciences (grant 6108-00078B) to the BABY, BACP, BAF, BAFC, BAI, BAJ, BAL, BARC, BAS, BBB, BBS, BC, BCMEX, BCN, BCRU, BEREA, TREECHANGE Project. J.-C.S. also considers this work a contribution to the Center for Ecological BESA, BG, BH, BHCB, BIO, BISH, BLA, BM, BOCH, BOL, BOLV, BONN, BOON, BOTU, BOUM, BPI, BR, Dynamics in a Novel Biosphere (ECONOVO) funded by Danish National Research Foundation BREM, BRI, BRIT, BRLU, BRM, BSB, BUT, C, CALI, CAN, CANB, CANU, CAS, CATA, CATIE, CAY, CBM, (grant DNRF173) and the VILLUM Investigator project “Biodiversity Dynamics in a Changing CDA, CDBI, CEN, CEPEC, CESJ, CGE, CGMS, CHAM, CHAPA, CHAS, CHR, CHSC, CIB, CICY, CIIDIR, World” funded by VILLUM FONDEN (grant 16549). W.L.E. was supported by the VILLUM CIMI, CINC, CLEMS, CLF, CMM, CMMEX, CNPO, CNS, COA, COAH, COCA, CODAGEM, COFC, COL, FONDEN (grant 00025354). M.A. was supported by the Natural Sciences and Engineering COLO, CONC, CORD, CP, CPAP, CPUN, CR, CRAI, CRP, CS, CSU, CSUSB, CTES, CTESN, CU, CUVC, Research Council of Canada (NSERC) Discovery grants program. C.B. was supported by the CUZ, CVRD, DAO, DAV, DBG, DBN, DES, DLF, DNA, DPU, DR, DS, DSM, DUKE, DUSS, E, EA, EAC, National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) EAN, EBUM, ECON, EIF, EIU, EMMA, ENCB, ER, ERA, ESA, ETH, F, FAA, FAU, FAUC, FB, FCME, FCO, (2022R1A2C1003504). A.G.G. was supported by the ANID PIA/BASAL (grant FB210006) and FCQ, FEN, FHO, FI, FLAS, FLOR, FM, FR, FRU, FSU, FTG, FUEL, FULD, FURB, G, GAT, GB, GDA, GENT, FONDECYT (grant 1200468). R.E.O. was supported by the German Research Foundation to GES, GH, GI, GLM, GMDRC, GMNHJ, GOET, GRA, GUA, GZU, H, HA, HAC, HAL, HAM, HAMAB, HAO, German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig (DFG-FZT 118, HAS, HASU, HB, HBG, HBR, HCIB, HEID, HGM, HIB, HIP, HNT, HO, HPL, HRCB, HRP, HSC, HSS, HU, 202548816). J.Pe. was supported by the Fundación Ramón Areces project (CIVP20A6621) and HUA, HUAA, HUAL, HUAZ, HUCP, HUEFS, HUEM, HUFU, HUJ, HUSA, HUT, HXBH, HYO, IAA, IAC, Spanish Government project (PID2019-110521GB-I00). J.Pi. was supported by the Estonian IAN, IB, IBGE, IBK, IBSC, IBUG, ICEL, ICESI, ICN, IEA, IEB, ILL, ILLS, IMSSM, INB, INEGI, INIF, INM, Research Council Grants (PUT1355, PRG 1405) and Mobilitas Pluss (MOBERC11). V.D.P. was INPA, IPA, IPRN, IRVC, ISC, ISKW, ISL, ISTC, ISU, IZAC, IZTA, JACA, JBAG, JBGP, JCT, JE, JEPS, JOTR, supported by the National Council of Scientific and Technological Development (CNPq) of JROH, JUA, JYV, K, KIEL, KMN, KMNH, KOELN, KOR, KPM, KSC, KSTC, KSU, KTU, KU, KUN, KYO, L, LA, Brazil (grant 307689/2014-0). Author contributions: W.-B.X, A.O., and J.-C.S. conceived the LAGU, LBG, LD, LE, LEB, LIL, LINC, LINN, LISE, LISI, LISU, LL, LMS, LOJA, LOMA, LP, LPAG, LPB, LPD, study. W.-Y.G., J.M.S.-D., W.-B.X., F.S., W.L.E., B.J.E., B.S.M., C.M., C.V., M.A., M.B., H.H.B., C.B., J.A.C., LPS, LSU, LSUM, LTB, LTR, LW, LYJB, LZ, M, MA, MACF, MAF, MAK, MARS, MARY, MASS, MB, MBK, B.E.L.C., E.C.-M., D.C., J.H.C.C., A.T.D.-L., A.d.F., A.S.D., A.B.G., A.G.G., W.H., T.H., P.H., N.H.-S., S.J., J.K., MBM, MBML, MCNS, MEL, MELU, MEN, MERL, MEXU, MFA, MFU, 816 MG, MGC, MICH, MIL, MIN, B.K., N.J.B.K., K.K., S.L., C.H.L, A.R.M., K.-P.M., M.M., S.T.M, V.M., A.S.M., U.N., Y.O., R.E.O., J.Pe., V.D.P., MISSA, MJG, MMMN, MNHM, MNHN, MO, MOL, MOR, MPN, MPU, MPUC, MSB, MSC, MSUN, MT, J.Pi., M.J.P., B.J.M.R., B.S., M.Sl., M.Su., Ê.E.S., N.A.S., N.T., P.M.v.B., F.v.d.P., J.Z., J.-C.S., and A.O. MTMG, MU, MUB, MUR, MVFA, MVFQ, MVJB, MVM, MW, MY, N, NA, NAC, NAS, NCU, NE, NH, NHM, collated the data. W.-B.X. performed the analyses with help from W.-Y.G. and J.M.S.-D. W.-B.X. NHMC, NHT, NLH, NM, NMB, NMNL, NMR, NMSU, NSPM, NSW, NT, NU, NUM, NY, NZFRI, O, OBI, and A.O. wrote the first draft of the manuscript. All authors contributed to interpreting the ODU, OS, OSA, OSC, OSH, OULU, OWU, OXF, P, PACA, PAMP, PAR, PASA, PDD, PE, PEL, PERTH, results, the writing, and approved the final manuscript. Competing interests: The authors PEUFR, PFC, PGM, PH, PKDC, PLAT, PMA, POM, PORT, PR, PRC, PRE, PSU, PY, QCA, QCNE, QFA, QM, declare that they have no competing interests. Data and materials availability: All species QRS, QUE, R, RAS, RB, RBR, REG, RELC, RFA, RIOC, RM, RNG, RSA, RYU, S, SACT, SALA, SAM, SAN, occurrences used in this article have been deposited in the BIEN database (https://bien.nceas. SANT, SAPS, SASK, SAV, SBBG, SBT, SCFS, SD, SDSU, SEL, SEV, SF, SFV, SGO, SI, SIU, SJRP, SJSU, ucsb.edu/bien/). All data (species distributions, the phylogeny, and imputed trait data) used in SLPM, SMDB, SMF, SNM, SOM, SP, SPF, SPSF, SQF, SRFA, STL, STU, SUU, SVG, TAES, TAI, TAIF, TALL, this study and all code for this analysis are available on Github (https://github.com/Wubing-Xu/ TAM, TAMU, TAN, TASH, TEF, TENN, TEPB, TEX, TFC, TI, TKPM, TNS, TO, TOYA, TRA, TRH, TROM, Global_tree_beta-diversity) and are mirrored on Zenodo (https://doi.org/10.5281/ TRT, TRTE, TU, TUB, U, UADY, UAM, UAMIZ, UB, UBC, UC, UCMM, UCR, UCS, UCSB, UCSC, UEC, zenodo.7699130). UESC, UFG, UFMA, UFMT, UFP, UFRJ, UFRN, UFS, UGDA, UH, UI, UJAT, ULM, ULS, UME, UMO, UNA, UNB, UNCC, UNEX, UNITEC, UNL, UNM, UNR, UNSL, UPCB, UPEI, UPNA, UPS, US, USAS, USF, USJ, Submitted 8 July 2022 USM, USNC, USP, USZ, UT, UTC, UTEP, UU, UVIC, UWO, V, VAL, VALD, VDB, VEN, VIT, VMSL, VT, W, Accepted 8 March 2023 WAG, WAT, WELT, WFU, WII, WIN, WIS, WMNH, WOLL, WS, WTU, WU, XAL, YAMA, Z, ZMT, ZSS, and Published 5 April 2023 ZT. Funding: W.-B.X. and A.O. acknowledge support from the Aarhus University Research 10.1126/sciadv.add8553 Xu et al., Sci. Adv. 9, eadd8553 (2023) 5 April 2023 18 of 18