High exposure of global tree diversity to human pressure

Abstract

Safeguarding Earth’s tree diversity is a conservation priority due to the importance of trees for biodiversity and ecosystem functions and services such as carbon sequestration. Here, we improve the foundation for effective conservation of global tree diversity by analyzing a recently developed database of tree species covering 46,752 species. We quantify range protection and anthropogenic pressures for each species and develop conservation priorities across taxonomic, phylogenetic, and functional diversity dimensions. We also assess the effectiveness of several influential proposed conservation prioritization frameworks to protect the top 17% and top 50% of tree priority areas. We find that an average of 50.2% of a tree species’ range occurs in 110-km grid cells without any protected areas (PAs), with 6,377 small-range tree species fully unprotected, and that 83% of tree species experience nonnegligible human pressure across their range on average. Protecting high-priority areas for the top 17% and 50% priority thresholds would increase the average protected proportion of each tree species’ range to 65.5% and 82.6%, respectively, leaving many fewer species (2,151 and 2,010) completely unprotected. The priority areas identified for trees match well to the Global 200 Ecoregions framework, revealing that priority areas for trees would in large part also optimize protection for terrestrial biodiversity overall. Based on range estimates for >46,000 tree species, our findings show that a large proportion of tree species receive limited protection by current PAs and are under substantial human pressure. Improved protection of biodiversity overall would also strongly benefit global tree diversity.

Full text

High exposure of global tree diversity to human pressure
Wen-Yong Guo (郭文永)
a,b,c,d,1
, Josep M. Serra-Diaz
e
, Franziska Schrodt
f
, Wolf L. Eiserhardt
b
, Brian S. Maitner
g
, Cory Merow
h,i
, Cyrille Violle
j
,
Madhur Anand
k
,Micha
€
el Belluau
l
, Hans Henrik Bruun
m
, Chaeho Byun
n
, Jane A. Catford
o
, Bruno E. L. Cerabolini
p
,
Eduardo Chac
on-Madrigal
q
, Daniela Ciccarelli
r
, J. Hans C. Cornelissen
s
, Anh Tuan Dang-Le
t,u
, Angel de Frutos
v
, Arildo S. Dias
w
,
Aelton B. Giroldo
x
, Kun Guo
c,d
, Alvaro G. Guti
errez
y,z
, Wesley Hattingh
aa
, Tianhua He (何田华)
bb,cc
, Peter Hietz
dd
, Nate Hough-Snee
ee
,
Steven Jansen
ff
, Jens Kattge
v,gg
, Tamir Klein
hh
, Benjamin Komac
ii
, Nathan J. B. Kraft
jj
, Koen Kramer
kk,ll
, Sandra Lavorel
mm
,
Christopher H. Lusk
nn
, Adam R. Martin
oo
, Maurizio Mencuccini
pp,qq
, Sean T. Michaletz
rr,ss
, Vanessa Minden
tt,uu
, Akira S. Mori
vv
,
€
Ulo Niinemets
ww
, Yusuke Onoda
xx
, Josep Pe~
nuelas
yy,zz
,Val

erio D. Pillar
aaa
, Jan Pisek
bbb
, Bjorn J. M. Robroek
ccc
, Brandon Schamp
ddd
,
Martijn Slot
eee
,
^
Enio Egon Sosinski Jr.
fff
, Nadejda A. Soudzilovskaia
ggg
, Nelson Thiffault
hhh
, Peter van Bodegom
iii
, Fons van der Plas
jjj
,
Ian J. Wright
kkk,lll
, Wu-Bing Xu
a,b,v
, Jingming Zheng
mmm
, Brian J. Enquist
g,nnn
, and Jens-Christian Svenning
a,b,1
Edited by Hugh Possingham, The Nature Conservancy, Sherwood, QLD, Australia; received December 29, 2020; accepted April 13, 2022
Safeguarding Earth’s tree diversity is a conservation priority due to the importance of trees
for biodiversity and ecosystem functions and services such as carbon sequestration. Here,
we improve the foundation for effective conservation of global tree diversity by analyzing
a recently developed database of tree species covering 46,752 species. We quantify range
protection and anthropogenic pressures for each species and develop conservation priori-
ties across taxonomic, phylogenetic, and functional diversity dimensions. We also assess
the effectiveness of several influential proposed conservation prioritization frameworks to
protect the top 17% and top 50% of tree priority areas. We find that an average of
50.2% of a tree species’range occurs in 110-km grid cells without any protected areas
(PAs), with 6,377 small-range tree species fully unprotected, and that 83% of tree species
experience nonnegligible human pressure across their range on average. Protecting high-
priority areas for the top 17% and 50% priority thresholds would increase the average
protected proportion of each tree species’range to 65.5% and 82.6%, respectively, leav-
ing many fewer species (2,151 and 2,010) completely unprotected. The priority areas
identified for trees match well to the Global 200 Ecoregions framework, revealing that
priority areas for trees would in large part also optimize protection for terrestrial biodiver-
sity overall. Based on range estimates for >46,000 tree species, our findings show that a
large proportion of tree species receive limited protection by current PAs and are under
substantial human pressure. Improved protection of biodiversity overall would also
strongly benefitglobaltreediversity.
biodiversity jconservation frameworks jland use jprotected areas jtree species
Trees play a vital role in the biosphere. As key agents in the flow of energy and matter,
they protect catchments and stabilize drainage areas, sequester carbon, and regulate climate
on a local-to-global scale (1–3). Trees also provide habitat for a large proportion of the
diversity of the world’s vertebrates, invertebrates, and fungi (4–9). The magnitude of many
of these functions and services increases as tree diversity increases, and greater functional
diversity of tree assemblages enhances ecosystem productivity and stability (10–12). How-
ever, continued global forest loss and degradation (13–20) have decimated biodiversity
among tree and tree-dependent organisms (8, 21–23). Tree diversity loss diminishes eco-
system resilience and contributions to coregulating the changing climate system (24–26).
While policy makers and land managers are increasingly aware of the importance of trees,
an in-depth global assessment of the adequacy and effectiveness of existing protections for
tree diversity is lacking. A comprehensive assessment of protection and pressures on tree
diversity would provide important input for establishing conservation and restoration prior-
ities to bend the curve of biodiversity loss (27).
Protected areas (PAs) represent a first-order conservation strategy for preventing bio-
diversity loss, aimed to preserve natural ecosystems (28–30) and their inherent services,
such as carbon sequestration (31). At present, PAs cover 15.8% of the Earth’s land
(World Database on Protected Areas; WDPA*). This value remains below the 17% of
Earth’s land area recommended by the Convention on Biological Diversity (CBD)
2020 target (see SI Appendix, Box S1 for a detailed explanation). However, it is not
well-understood how well existing PAs cover tree species diversity.
Significance
Earth’s tree diversity is crucial for
biodiversity and ecosystem
functions and services. Using
species range estimates for 46,752
tree species, we find that an
average of 50.2% of a tree species’
range occurs in 110-km grid cells
without any protected areas, with
a total of 6,377 small-range tree
species entirely unprotected, and
that 83% of tree species
experience nonnegligible human
pressure across their range on
average. Protecting additional
areas selected to optimally cover
multiple dimensions of tree
diversity would strongly improve
this situation. Our results highlight
the need for strengthening efforts
to protect tree diversity via
increased coverage of protected
areas through well-targeted
conservation actions as well as
integration of tree diversity into
restoration efforts in human-
dominated landscapes.
The authors declare no competing interest.
This article is a PNAS Direct Submission.
Copyright © 2022 the Author(s). Published by PNAS.
This article is distributed under Creative Commons
Attribution-NonCommercial-NoDerivatives License 4.0
(CC BY-NC-ND).
1
To whom correspondence may be addressed. Email:
guowyhgy@gmail.com or svenning@bios.au.dk.
This article contains supporting information online at
http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.
2026733119/-/DCSupplemental.
Published June 16, 2022.
*https://www.protectedplanet.net/en, accessed 30 May 2022.
PNAS 2022 Vol. 119 No. 25 e2026733119 https://doi.org/10.1073/pnas.2026733119 1of11
RESEARCH ARTICLE
|
ENVIRONMENTAL SCIENCES
SUSTAINABILITY SCIENCE
Downloaded from https://www.pnas.org by Wageningen University & Research - Library on July 20, 2022 from IP address 137.224.252.11.

Given the current pace of biodiversity loss, debate has arisen
regarding the post-2020 PA targets (32–34). One initiative that
has gained significant momentum is E. O. Wilson’s“Half-
Earth”proposal (35), which argues that half of the Earth’s sur-
face needs to be protected in order to avoid major biodiversity
loss and safeguard major ecosystem processes and services
(36–39). This proposal aligns with the 2050 Vision for Biodi-
versity, proposed by CBD parties for the Post-2020 Global
Biodiversity Framework (40). Different frameworks may sug-
gest different priority areas (SI Appendix, Box S1) for future
PAs, but effective, coordinated global conservation measures
require a strong consensus. While various organizations may
establish global conservation targets and areas of priority based
on different considerations (29, 41), the realized network of
PAs should ideally effectively represent overall biodiversity as
identified through systematic analyses (28).
Recent studies have highlighted the need to consider multi-
ple biodiversity dimensions, including phylogenetic and func-
tional diversity, and their roles in providing ecosystem services
in spatial conservation planning (42–44). Unlike taxonomic
diversity, which is still most often used in biodiversity assess-
ments, functional and phylogenetic diversity represent critical
ecological and evolutionary aspects of biodiversity not fully cap-
tured by species composition alone (45–50) (but see ref. 51).
Given the pivotal role of trees in global terrestrial ecosystems,
questions surrounding how well their multiple diversity dimen-
sions are and could be protected by major biodiversity policy
targets are critical to the domains of both conservation and the
broader sustainability agenda, such as the rising global interest
in tree restoration (3) and integration of trees into agricultural
production systems (52, 53).
While foundational to conservation efforts, PAs are seldom
free from anthropogenic pressures. For example, a recent
study found that approximately one-third of global PAs expe-
rience intensive human pressure (54). Anthropogenic pres-
sures on PAs may increase as human activity near existing PAs
intensifies, and as new PAs are increasingly established in
proximity to global population centers (30, 55). Many regions
that host high biodiversity overlap with human settlements.
As a result, future PAs will be confronted with other land-use
demands (56, 57), given the rising global human population
and natural resource requirements. Examining human pres-
sure within existing PAs and priority areas for tree diversity is
important for assessing both the effectiveness of current PAs
in protecting the tree species they harbor, and the need for
increasing protection of currently unprotected priority areas
for tree diversity.
Here, we analyze a recently developed global database of
46,752 tree species’ranges to 1) assess range protection and
anthropogenic pressures for tree species, 2) identify priority
areas for conservation of tree diversity considering multiple
diversity dimensions, and 3) assess the geographic distribution
of current PAs and different potential conservation prioritiza-
tion scenarios and their respective coverage of global tree spe-
cies diversity. We used complementarity analysis [Zonation
(58, 59)] and integration of taxonomy, phylogenetic related-
ness, and functional traits to assess whether priority areas for
tree conservation overlapped or diverged spatially according to
different diversity dimensions (42, 44). Taxonomic diversity
was represented by species identities. Phylogenetic diversity was
represented by phylogenetic eigenvectors computed from a
genus-level phylogeny (60). Functional diversity was repre-
sented using similar eigenvectors, based on eight commonly
measured, ecologically important traits including maximum
height, wood density, specific leaf area, leaf area, leaf nitrogen
concentration, leaf phosphorus concentration, leaf dry matter
content, and seed dry mass (SI Appendix, Table S1).
We used these diversity data to estimate overlaps in the pri-
ority areas for three diversity dimensions for the top 17 and
50% area targets, representing the CBD 2020 target and the
Half-Earth proposal, respectively. By comparing the tree spe-
cies’range parts covered by existing PAs (as documented by the
WDPA) with those covered by the top 17 and top 50% prior-
ity areas identified through the Zonation algorithm, we quanti-
fied the coverage of existing and potential PAs with respect to
global tree diversity. We further quantified anthropogenic pres-
sures on species by estimating the Human Modification Index
(HMI) (61) for each species’range inside and outside existing
PAs, as well as for areas covered by potential future PAs for the
different conservation targets. As a cumulative measure of
human alteration of terrestrial areas based on 13 anthropogenic
layers, HMI is a unitless index that ranges from 0 to 1, with
0≤HMI ≤0.1 as low (e.g., a value of 0.007 for wildlands),
0.1 <HMI ≤0.4 as moderate (e.g., 0.12 for seminatural lands
and 0.37 for croplands), and 0.4 <HMI ≤1.0 as high to very
high (e.g., 0.58 for dense settlements and 0.65 for villages), fol-
lowing ref. 61.
We also tested coverage of existing PAs and top-priority areas
for tree diversity in relation to the different existing frameworks
for biodiversity conservation (SI Appendix, Box S1), covering
three global biodiversity conservation priority frameworks pro-
posed by leading conservation nongovernmental organizations
(NGOs). These included the Global 200 Ecoregions (G200),
Biodiversity Hotspots (BH), and Last of the Wild (LW) (29)
(SI Appendix, Box S1; hereafter “NGO frameworks”). These
NGO frameworks prioritize either areas of high vulnerability
(BH), the most intact areas (LW), or areas of highest irreplace-
ability (G200) (29). We identified gaps and overlaps between
current conservation efforts (PAs), our estimated priority areas
for tree diversity, and these existing NGO frameworks to assess
their utility for protecting Earth’s tree diversity and guiding
future funding and conservation efforts.
Results
Global Protection Coverage and Pressures on Tree Species.
Across all 46,752 tree species in our dataset, an average of
50.2% of a species’range occurred in 110-km grid cells with-
out existing PAs. A total of 6,377 species (13.6% of the tree
species evaluated) experience no protection anywhere in their
range (Fig. 1Aand SI Appendix, Table S2), and 10,987 (23.5%
of tested) tree species have less than 25% of their ranges in
such PA cells (Dataset S1). Most tree species experience at least
moderate human pressure within their ranges, with an average
HMI of 0.25 across all species (a value comparable to the HMI
across most of Wales; Fig. 1Aand SI Appendix, Table S2). A
total of 14.8% of tree species experience high to very high aver-
age human pressure within their range (0.4 <HMI ≤1.0),
and an additional 68.5% experience moderate average human
pressure (0.1 <HMI ≤0.4) (following ref. 61). For the por-
tions of species’ranges inside protected cells (110-km cells with
PAs), human pressure is relatively low, with an average HMI of
0.11. In contrast, the unprotected portion of tree species’
ranges (110-km cells without PAs) are exposed to considerably
higher human pressure, with a mean value of 0.25, and 7,776
species experiencing high to very high HMI in these areas (Fig.
1A). The global patterns described above largely reflect rela-
tively wide-ranged tree species (SI Appendix, Fig. S1 and Table
2of11 https://doi.org/10.1073/pnas.2026733119 pnas.org
Downloaded from https://www.pnas.org by Wageningen University & Research - Library on July 20, 2022 from IP address 137.224.252.11.

S2). By contrast, small-range tree species (the first quintile of
tree species’range sizes) have overall greater unprotected range
proportions (mean value of 74.1%), and all of the 6,377 species
that are completely unprotected are such small-range species.
Compared with species with larger ranges, small-range species
have lower HMI (a mean of 0.06) within PA cells but slightly
higher human pressures outside PAs (Fig. 1Band SI Appendix,
Fig. S1 and Table S2).
Global Priority Areas for Tree Conservation across Three
Dimensions of Diversity. The top-priority areas for the 17 and
50% targets show considerable spatial divergence when selected
separately considering taxonomic, phylogenetic, and functional
diversity dimensions (Fig. 2 and SI Appendix,Fig.S2). The
48.2% of the 17% top-priority areas for the different diversity
dimensions are selected according to all three dimensions. Another
43.5% of the 17% target priority areas selected based on taxo-
nomic diversity differ from those based on phylogenetic and func-
tional diversity (Fig. 2B). By contrast, the 17% top-priority areas
determined according to phylogenetic or functional diversity
largely overlapped, reflecting the strong correlation between the
two dimensions (Fig. 2 Aand Band SI Appendix,Fig.S2and
Table S3). Overlaps were stronger for the top 50% priority areas
but otherwise similar (Fig. 2 Cand Dand SI Appendix,Fig.S2).
Areas prioritized according to all three diversity dimensions
occur primarily in the tropical rainforest regions of the Ameri-
cas, Africa, Indo-Malaya, and Australasia, as well as in subtropi-
cal Asia for the 17% target, but also cover subtropical and
warm-temperate regions more broadly under the 50% target.
Generally, areas prioritized by just two diversity dimensions
mainly occur in tropical and subtropical savanna areas (e.g., the
Cerrado in Brazil) for the 17% target, but also include temper-
ate areas in North America and Europe as well as in arid areas,
notably in Australia, for the 50% target (Fig. 2 Aand C). High-
priority sites only selected according to a single diversity dimen-
sion show a scattered global distribution (Fig. 2 Aand C).
Protection Coverage and Pressures Associated with Global
Priority Areas. Protecting the 17 and 50% area targets based
on priority areas according to taxonomic diversity would
increase the average coverage of a species’range to 65.5% (top
17%) and 82.6% (top 50%), strongly exceeding the average
range proportion covered by existing PA cells (49.8%) (red
dashed lines in SI Appendix, Fig. S3Aand Table S4). Expand-
ing current PAs to either the 17 or 50% top-priority areas
would furthermore increase tree species’protection status (SI
Appendix, Figs. S3Aand S4). For example, of the 13.6% of tree
species in our dataset that are small-range and currently lack
any protection, a majority (66.2%) would become partly or
fully covered by 110-km grid cells with potential PAs corre-
sponding to the 17% top-priority areas (yellow flow ribbons in
SI Appendix, Fig. S3A). The proportion of species whose entire
range falls within 110-km grid cells with PAs would increase to
24.8 and 42.5% if the 17 and 50% target top-priority areas
were protected, respectively. Mean HMI values for the priority
areas under the 17 and 50% targets exceed those estimated for
existing PAs (red dashed lines in SI Appendix, Fig. S3Band
Table S4), reflecting that ca. 30% more species experience
moderate to high HMI values within these top-priority areas
relative to the proportion within current PAs.
Protection coverage and current pressure for prioritizations
based on functional and phylogenetic diversity resemble those
based on taxonomic diversity (SI Appendix,TableS4), with only
minor differences (SI Appendix,Fig.S3Avs. Cand E). However,
we found important differences in the degree of protection in
wide- vs. small-range species, particularly for the top 17% priority
area scenario. Specifically, the priority areas obtained considering
taxonomic diversity would greatly increase the protection coverage
of small-range species to 75% of the mean proportion of ranges,
while the protection coverage of small-range species would only
reach ca. 53% if using the priority areas obtained from the other
two dimensions (SI Appendix,Fig.S4).
Global priority areas identified using all three diversity dimen-
sions simultaneously largely match results from the single dimen-
sions, especially those from taxonomic diversity (Figs. 3 and 4).
Many of the areas designated as top 17% priority areas experience
moderate human pressure. These areas include southern and east-
ern Asia, South America outside the Amazon Basin, and Madagas-
car (Fig. 4). Many of the top 50% priority areas are subject to
high human pressure (Fig. 4). They include many European
countries, India, eastern China, Indonesia, Nigeria, Ethiopia, cen-
tral North America, and eastern Argentina.
Congruence among Top Tree Conservation Priority Areas,
Existing PAs, and NGO Frameworks. Grid cells with existing
PAs cover only about half of the top 17 or 50% priority areas
for tree conservation as jointly defined by taxonomic, phylo-
genetic, and functional diversity (51.2 and 44.6%, respec-
tively, the proportion of “current PAs only”+“shared”in
Fig. 5). In terms of overlap between PAs and the three NGO
frameworks (SI Appendix,Fig.S5), G200 showed the greatest
degree of overlap with the top 17% priority areas (45.4%).
The BH framework showed 34.8% overlap, and the LW
framework showed only 7.3% overlap with the top 17% pri-
ority areas (Fig. 5A). Expansion of PAs based on the G200
Fig. 1. Current protection status and pressures on tree species’ranges for
all (A) and small-range tree species (B) (the first range size quantile). Pro-
tected proportions show the proportion of each tree species‘range within
existing PAs; HMI indicates the mean Human Modification Index within a
tree species’range, overall, or just the range part within or outside PAs.
Mean and median values are indicated by white and black solid lines in the
violin density plots, respectively. Colored panes in yellow, pink, and purple,
respectively, indicate low (0 ≤HMI ≤0.1), moderate (0.1 <HMI ≤0.4), and
high to very high (0.4 <HMI ≤1.0) levels of human modification (61). SI
Appendix, Table S2 lists the mean, median, and first and third quantiles for
each variable.
PNAS 2022 Vol. 119 No. 25 e2026733119 https://doi.org/10.1073/pnas.2026733119 3of11
Downloaded from https://www.pnas.org by Wageningen University & Research - Library on July 20, 2022 from IP address 137.224.252.11.

framework would protect ∼91 or 83% of the top 17 and
50% priority areas for trees, respectively. The LW framework
would only protect about 50% of the top-priority areas
(57 and 53% for the top 17 and 50% priority areas, respec-
tively), thus representing only minor improvements to cur-
rent tree diversity and range protection. The BH framework
offers an intermediate case and would protect 77 and 67% of
the top 17 and 50% priority areas for the three diversity
dimensions for trees.
Discussion
Our results demonstrate that, conservatively estimated, on aver-
age approximately half of a tree species’range lacks protection
Fig. 2. Top 17% and top 50% priority areas (Aand C) according to species taxonomic, phylogenetic, and functional diversity dimensions defined by the
Zonation prioritization. The Venn diagrams show overlapping and unique areas for prioritizations for the 17% target (B) and 50% target (D) based on either
taxonomic, phylogenetic, or functional diversity dimensions. Colors indicate overlap between combinations of two of the three dimensions (green), between
all three dimensions (yellow), or no overlap (purple).
Fig. 3. Proportional changes in the number of tree species with (A) a certain proportion of the species range protected and (B) a certain level of human
influence within the protected species range, computed for existing PAs, the top 17% priority areas, and between the top 17% and top 50% priority areas.
The prioritization jointly considers taxonomic, phylogenetic, and functional diversity; results for prioritizations for taxonomic, phylogenetic, and functional
diversity separately are shown in SI Appendix, Fig. S3. Ribbons represent proportional flows of species in terms of changing scores (i.e., either protection cov-
erage [A] or human influence level [B]) between two consecutive grouping bars. (A) Protection percentage categories indicate the proportion of a species’
range inside 110-km grid cells overlapping current PAs or the top 17% or top 50% priority areas, respectively. Red dashed lines indicate mean protection
percentages for all tree species, with exact values given in SI Appendix, Table S4.(B) HMI categories based on the mean HMI value for the proportion of each
species’range overlapping with existing PAs or the top 17% or top 50% priority areas. HMI values were divided into three categories following ref. 61 repre-
senting low (0 ≤HMI ≤0.1), moderate (0.1 <HMI ≤0.4), and high to very high (0.4 <HMI ≤1.0) degrees of human modification. The yaxis and dashed red
lines (Right) show the average HMI values across all tree species’range proportions overlapping with either existing PAs or the top 17% or top 50% priority
areas, respectively. SI Appendix, Table S4 provides exact values.
4of11 https://doi.org/10.1073/pnas.2026733119 pnas.org
Downloaded from https://www.pnas.org by Wageningen University & Research - Library on July 20, 2022 from IP address 137.224.252.11.

under the current PA network. A majority of species’ranges
experience moderate (32,003 species) or high (6,928 species)
human pressure, even for species with ranges that are fully cov-
ered with existing PA grid cells. About 13.6% of tree species,
all of which are small-range species, occur completely outside
grid cells with existing PAs. Compared with average PA cover-
age (49.8% of species range in grid cells with PAs), the ranges
of small-range tree species are only about half as well-covered
(25.9%) (SI Appendix, Table S2). In addition, nearly one-quarter
of the 46,752 tree species have less than 25% of their ranges over-
lapping protected cells (i.e., grid cells with existing PAs). Overall,
our results indicate that, even when optimistically equating PA
presence in a 110-km grid cell as PA coverage (Methods), the cur-
rent PA network is insufficient to protect Earth’streediversity,
particularly for small-range trees, given the fact that a large pro-
portion of Earth’s total tree species are estimated to be small-
range, as found in our analysis. Further, a recent analysis estimates
ca. 9,000 further undiscovered tree species, which will be mostly
small-range species (62). However, our results also show that pro-
tecting the top 17 and 50% priority areas would strongly improve
protection coverage and would include large numbers of areas of
global importance for tree diversity that are currently exposed to
moderate to high human pressure (Fig. 3 and SI Appendix,Fig.
S4). Importantly, species with limited ranges would also be better
protected (Fig. 3 and SI Appendix,Fig.S4), which is crucial as
they experience greater anthropogenic pressures and extinction
risk (63, 64).
Critically, 49.2% of tree species experience moderate to very
high human pressure even within protected cells (Fig. 1A),
highlighting the need to enhance protection effectiveness
within and around PAs. Furthermore, half of the high-priority
areas for tree diversity conservation currently have moderate to
high human pressures. Here, pressures such as habitat conver-
sion, overharvesting, or overgrazing by livestock threaten tree
populations and may also negatively influence conservation
efforts in nearby PAs (17, 18, 65–67). Our analysis thus identi-
fies vulnerable areas in which protection efforts, mitigation of
human pressure, and restoration efforts (sensu ref. 29), includ-
ing cost-effective approaches such as natural regeneration of
degraded habitat (68, 69), would yield high returns in terms of
biodiversity protection goals (57). Mechanisms such as pay-
ments for ecosystem services programs could help achieve the
reduction in human pressures in these critical areas identified
in our study.
Studies of conservation priorities have often used taxonomic
diversity or other singular dimensions (70–72) as a surrogate
for other aspects of biodiversity or ecosystem function (50, 51,
73) (but see ref. 45). In the present study, we found that
Fig. 4. Overlap between current PAs, top-priority areas for 17% and 50% targets, and the HMI. The priority areas for tree conservation are jointly defined
according to taxonomic, phylogenetic, and functional diversity. HMI is categorized into low (0 ≤HMI ≤0.1), moderate (0.1 <HMI ≤0.4), and high to very
high (0.4 <HMI ≤1.0) levels(61). The HMI layer is shown at a resolution of 1 km
2
.
Fig. 5. Percentages of the (A) top 17% and (B) top 50% priority areas for
tree diversity covered by existing PAs or by each NGO framework (G200,
BH, and LW) for global biodiversity conservation. Colors indicate overlaps
between combinations. Unprotected: areas not overlapping with either PAs
or a conservation priority framework; NGO framework only: areas overlap-
ping only with the considered NGO framework; shared: areas overlapping
with both PAs and a given NGO framework; current PAs only: areas only
overlapping with PAs. Priority areas for tree conservation are jointly
defined according to taxonomic, phylogenetic, and functional diversity.
PNAS 2022 Vol. 119 No. 25 e2026733119 https://doi.org/10.1073/pnas.2026733119 5of11
Downloaded from https://www.pnas.org by Wageningen University & Research - Library on July 20, 2022 from IP address 137.224.252.11.

different diversity dimensions resulted in substantially different
spatial prioritizations (Fig. 2 and SI Appendix, Figs. S2 and S3)
and distinct relationships between species’range sizes and pro-
tection percentages (SI Appendix, Fig. S4), demonstrating the
importance of considering multiple aspects of biodiversity in
conservation planning (42, 44). Specifically, in using a compre-
hensive view of trees beyond only species in forests (74), we
found certain regions usually not considered as tree diversity
hotspots (e.g., tropical and subtropical savanna, temperate areas
in North America and Europe, arid areas in Australia) are also
priority areas within one- or two-dimensional priority analyses
(Fig. 2). This finding apparently owes to certain traits or phylo-
genetic lineages evolved in situ, such as Cerrado woody species
in Brazil (75) and the family Gyrostemonaceae, endemic to
Australia and concentrated in the drier parts of the continent.
In addition, phylogenetic and functional diversity dimensions
support more spatially continuous priority areas than those
determined by taxonomic diversity (Fig. 2; similar to ref. 44).
Growing recognition of the importance of ecological integrity
of large PAs supports the usefulness of phylogenetic and func-
tional diversity dimensions in conservation planning (28, 35,
76). Further, priority areas defined by these dimensions include
more temperate areas (SI Appendix, Fig. S2) and areas with less
human pressure (SI Appendix, Fig. S3 and Table S4).
Organizations such as the International Union for Conserva-
tion of Nature (IUCN) have conducted similar conservation
status evaluations for other, mostly smaller taxonomic groups.
The IUCN Red List reports that 18% of vertebrates are threat-
ened. The Global Tree Assessment has used the IUCN
approach to evaluate the global conservation status of
58,497 tree species and found that 30% (17,510 species) are
threatened (77). Based on quantitative range estimates for
>46,000 tree species, we report that 83.3% of tree species expe-
rience moderate to high human pressure, in which 14.8% of
species are exposed to high or very high human pressure. The
differing risk estimates clearly arise from diverging approaches
used by the different evaluations. The IUCN system considers
population loss and decline of range size as indicators of extinc-
tion risk (78), while our analysis evaluates conservation status
according to PA coverage and human pressure within species’
ranges. Despite the differences, the estimations are not incom-
patible and consistently show that much stronger conservation
efforts for trees are needed to reduce the risk of losing large pro-
portions of tree species diversity. Both approaches also highlight
the need to have special focus on small-range tree species, in line
with a global assessment for plants overall (79). Similarly, a
regional study on the Brazilian Amazon found that tree species
with small-range sizes are more likely to become extinct from ris-
ing human pressures (80), supporting the conclusion that better
PA coverage is especially needed for the world’s many small-
range tree species.
As a cornerstone of biodiversity conservation, PAs are estab-
lished to protect biodiversity and ecosystem services. However,
as species, ecosystems, and PAs are experiencing dynamic
changes and pressures (54–56) from land use and climate
change, alien species introductions, and pollution, other conser-
vation actions are increasingly proposed and realized outside
PAs, such as restoration and reforestation through natural
regeneration (81), active promotion of rare species in restora-
tion and reforestation (82), stronger integration of tree diversity
into forest management (11, 81), and integration of tree diver-
sity into agricultural landscapes via agroforestry (e.g., ref. 83).
Those implementations are not only critical for preserving and
enhancing tree diversity but also for mitigating climate change
(3, 83, 84) and enhancing rates of provisioning of other ecosys-
tem services such as biomass productivity (11). Indeed, PA roles
have been transforming, and are increasingly extended to cover
ecosystem services (85, 86). PAs are thus increasingly regarded
as a multipurpose solution to ensure biodiversity while provid-
ing key ecosystem services (87), a view that is supported by
high associations between biodiversity, carbon stocks, and other
key ecosystem services (84, 86, 88). Thus, a comprehensive
evaluation of existing PAs and proposed conservation prioritiza-
tion frameworks using key organismal groups is critical for
understanding their effectiveness and for guiding future PA
expansions (89).
Addressing this need, we here focus on a key organismal
group, namely trees, to test the global effectiveness of multiple
influential biodiversity conservation frameworks. The high
degree of overlap between the NGO frameworks and the top-
priority areas for tree diversity (particularly the G200 and BH
frameworks) demonstrates that tree diversity is fairly congruent
with that of broader sets of organisms, because these frameworks
are assumed to select ecoregions that are most crucial (either
highly irreplaceable or vulnerable) to the global biodiversity and
include socioecological factors (90, 91). Importantly, this finding
shows that enhanced protection of biodiversity overall would
also strongly benefit tree diversity. The LW framework showed
the smallest degree of overlap with tree conservation priority
areas, because this NGO framework primarily captures remote
wilderness areas, which are often located in arid or cold high-
altitude or -latitude regions, which have limited suitability for
trees and inherently harbor low tree species diversity (92). Such
areas have also historically experienced lower levels of human
pressure, allowing large natural areas to persist and reducing
opportunity costs of conservation (29, 32, 54, 93).
Despite the coverage and quality of the tree species dataset
analyzed here, limitations such as a necessary coarse spatial reso-
lution (110-km grid cells), limited geographical coverage in
parts of Russia and southern Asia (94), and limited data for
many functional traits introduce uncertainties. Phylogenetic
and functional diversity often correlate strongly (44, 95), and
we also found significant phylogenetic signals for four out of
the eight functional traits analyzed in this study (SI Appendix,
Table S1). However, this correlation was certainly enhanced by
imputing functional traits using phylogenetic eigenvectors.
Nevertheless, functional and phylogenetic diversity dimensions
gave somewhat distinct priority areas (Fig. 2 and SI Appendix,
Fig. S3), indicating that these diversity dimensions still pro-
vided unique information. Even though range size is generally a
good proxy to represent species’vulnerability (96, 97) and is
commonly used in conservation status assessments (78), further
research is clearly needed to expand our understanding of geo-
graphical distributions. Better data coverage for tree functional
traits is also important for conserving or restoring tree diversity,
as tree species with different sets of traits have distinct ecologi-
cal demands and functions; for example, large seeds and animal
pollination are positively related to tree species’extinction risk
(96). Further, our prioritization analysis did not consider socio-
economic costs and local or regional social or political contexts
(98), factors which are important for systematic conservation
planning (99). Moreover, as the goal of our study is to under-
stand pressures and protections of tree species, we did not spe-
cifically consider primary vs. secondary forest. Notably, the
top-priority areas under high pressure would inherently tend to
be or develop as secondary forests if such protection was real-
ized, with expected time lags in developing their full biodiver-
sity and ecosystem benefits.
6of11 https://doi.org/10.1073/pnas.2026733119 pnas.org
Downloaded from https://www.pnas.org by Wageningen University & Research - Library on July 20, 2022 from IP address 137.224.252.11.

Conclusions
Globally, 83.8% of the 46,752 tree species evaluated in our analy-
sis are subject to moderate to very high human pressure, with PA
grid cells covering only ≤25% of the ranges for 23.5% of tree spe-
cies. Further, a total of 6,377 small-range tree species remain
completely unprotected. At the same time, a total of 14.8% tree
species experience high to very high human pressure even within
existing PAs. Our analysis further found existing PA grid cells are
estimated to cover only about half of the critical areas for tree
diversity, as quantified by taxonomic, phylogenetic, and functional
diversity dimensions. These results highlight the pressing need for
stronger protection of Earth’s tree diversity. Our results also show
that expanding PAs according to the top 17% and especially the
50% priority areas would yield strong improvements in PA cover-
age of trees, as would implementing some of the major proposals
for increased general biodiversity protection, notably the G200
framework. Such efforts are critically needed to counter expanding
human pressure on natural and seminatural areas in many parts of
the world (100), and also require effective protection of current
and future PAs as pressures may penetrate into the PAs despite
formal protection (3, 41, 80). We further suggest that promoting
protection and inclusion of tree diversity, notably rare and threat-
ened species, into land-sharing approaches such as small-scale
nature areas, sustainable forestry, multiuse reforestation, and agro-
forestry (e.g., ref. 82) would help to achieve adequate protection
of tree diversity and facilitate contributions to ecosystem services
in rural and urban landscapes broadly (101). Given the enormous
importance of tree diversity for biodiversity overall, for people,
and for the functioning of the biosphere, it should be a major pri-
ority for environmental policy to bend the curve from biodiversity
loss to recovery (27) for Earth’srichflora of trees.
Methods
Tree Species, Their Occurrence Records, and Range Estimates. We used
the world tree species list and species occurrence data compiled and cleaned in
ref. 94. Briefly, they extracted the records from the world tree species checklist
[GlobalTreeSearch; GTS (74)] and further standardized the taxonomic names via
the Taxonomic Name Resolution Service online tool (102). The GTS employed
the definition of the tree-type growth habit agreed by IUCN’s Global Tree Special-
ist Group, namely “a woody plant with usually a single stem growing to a height
of at least two meters, or if multi-stemmed, then at least one vertical stem five
centimeters in diameter at breast height”(74). An initial list of 54,020 tree spe-
cies was left (94).
Occurrence data for tree species were compiled from five widely used and
publicly accessible occurrence databases: the Global Biodiversity Information
Facility (GBIF; http://www.gbif.org), public domain Botanical Information and
Ecology Network, version 3 [BIEN; https://bien.nceas.ucsb.edu/bien/ (79, 103)],
Latin American Seasonally Dry Tropical Forest Floristic Network [DRYFLOR; http://
www.dryflor.info/ (104)], RAINBIO database [http://rainbio.cesab.org/ (105)], and
Atlas of Living Australia (ALA; https://www.ala.org.au/). Due to well-documented
problems of biases and errors in global plant occurrence datasets (106), we
applied a workflow for occurrence data quality assessment (94) to the initially
gathered 9,032,654 occurrence records. Our final list of species was lowered to
46,752 species with a total occurrence dataset of 7,066,785 records.
We then constructed alpha hulls (107) to estimate the range of each species
with 20 or more occurrence records using the ashape function of the alphahull
package (108) implemented in R [version 3.5.1 (109)]. For species with fewer
than 20 occurrences or with disjunct records, a 10-km buffer was given to each
point record and then merged with the alpha-hull range to estimate species
ranges. Previously, several alpha levels were recommended for the estimation of
species range (e.g., refs. 110–113); four alpha degrees (2, 4, 6, and 10) were
applied to each species here after external validation of the different alpha levels
tested. To validate the range maps using different alpha levels (alpha values of
2, 4, 6, and 10), we performed three types of external validation, as described in
SI Appendix,Methods,External Validation and Figs. S6–S10. Based on those vali-
dations, we selected the alpha-hull range maps with an alpha parameter of 6
degrees for subsequent analyses, as recommended and applied in similar studies
(112, 113), and with the highest R-squared in our validation (SI Appendix,Fig.S6).
The obtained estimated range maps were rasterized to 110-km equal-area grid cells,
a resolution commonly used in global diversity studies (e.g., refs. 99, 114, and
115), using the letsR package (116). Even though the estimated range can reduce
the geographical bias and fill gaps of the occurrence records, it may overestimate
the species’true area of occupancy and they should therefore not be interpreted as
a detailed coverage of species’realized distribution, but rather as an estimate of its
extent of occurrence (cf refs. 41 and 117).
Phylogeny. We extracted phylogenetic information for the tree species with
range maps from the largest seed plant phylogeny that is currently available
(the ALLMB tree in ref. 118). This phylogeny combines a backbone tree (119)
reflecting deep relationships with sequence data from public repositories (Gen-
Bank) and previous knowledge of phylogenetic relationships and species names
from the Open Tree of Life (synthetic tree release 9.1 and taxonomy version 3;
https://tree.opentreeoflife.org/about/synthesis-release/v9.1). We matched this
phylogeny to our tree dataset by first removing any species that were not in our
data, and then manually adding some species that were missing from the phy-
logeny (due to different taxonomic concepts) following the same approach that
ref. 119 used to add missing species. The resulting phylogeny contained 46,752
species (SI Appendix,Fig.S11and Dataset S2).
We calculated phylogenetic eigenvectors (60) to represent the phylogenetic
position of each speciesin our dataset using the PVR package (60, 120). Because
we were mostly interested in the deep structure of the phylogeny, and phyloge-
netic eigenvector calculation for large phylogeniesis computationally prohibitive,
we calculated eigenvectors at the genus level (4,031 genera). To accomplish
this, we randomly chose one species per genus, removed all other species, and
computed phylogenetic eigenvectors using the resulting phylogenetic tree. All
species were then assigned the eigenvector values of their genus. In the follow-
ing analysis, we used the first 15 eigenvectors, excluding those that captured
very little phylogenetic variation (eigenvalues <1%, following ref. 44). These
selected eigenvectors accounted for 40.6% of the total phylogenetic variation,
representing the deep evolutionary history of our study species.
It is possible to use the phylogenetic diversity [like Faith’sPD(121)]directly
in a prioritization algorithm (e.g., Zonation); however, the use of this alpha-
diversity layer will diminish the advantages of the complementarity, which is
derived from beta diversity and makes the priority ranking based on all biodiver-
sity features directly, rather than the species-rich hotspots only (122) (see Prioriti-
zation Analyses). To accommodate this, we adapted from the framework in refs.
42 and 44. First, we evenly divided each eigenvector into 20 bins. Then, we cre-
ated a binary variable for each bin, scoring all species with values within the
range of the bin as 1, and all others as 0. This resulted in 20 binary variables for
each of the 15 eigenvectors, that is, a matrix of 46,752 species ×300 binary var-
iables. We multiplied this matrix with the grid cells ×species matrix to generate
apresence–absence matrix of phylogenetic groups in grid cells (44). This matrix
showed which parts of the phylogeny—as represented by the binary variables
derived from the eigenvectors—were present in each grid cell. We used this
matrix in the following prioritization analysis to find priority regions for the con-
servation of tree phylogenetic diversity.
Functional Trait Data. Twenty-one functional traits (SI Appendix, Table S1)
were compiled from major trait databases (SI Appendix, Functional Traits and
Imputation). As many of the traits had missing values, we imputed trait values
applying a gap-filling technique using Bayesian hierarchical probabilistic matrix
factorization (123) (SI Appendix,Fig.S12). We finally selected eight key func-
tional traits for further functional diversity analyses, including leaf nitrogen con-
centration, wood density, leaf phosphorus concentration, leaf dry matter content,
plant maximum height, seed dry mass, specific leaf area, and leaf area. We used
the beanplot package (124) to visually compare the observed original and
imputed data, and found they generally had similar distribution patterns for
each functional trait (SI Appendix,Fig.S13). We further tested the phylogenetic
signal for each imputed trait with the function phylosig in the phytools package
(125), and found four of the eight traits showed significant phylogenetic signals
(SI Appendix,TableS1).
PNAS 2022 Vol. 119 No. 25 e2026733119 https://doi.org/10.1073/pnas.2026733119 7of11
Downloaded from https://www.pnas.org by Wageningen University & Research - Library on July 20, 2022 from IP address 137.224.252.11.

We performed a similar procedure as for phylogenetic eigenvectors to obtain
trait diversity dimensions for our conservation prioritization analysis. We first split
each trait into 20 equal bins, and then converted it into a binary species ×trait
matrix for each trait (46,752 species ×20 bins). For each of the 20 bins, we mul-
tiplied it by the 110-km grid cells ×species matrix to obtain a trait ×grid cell
matrix, in which each 110-km grid cell contained the number of species of trait
values within the trait interval. In total, we obtained 160 trait ×grid cell matri-
ces, a distribution map was generated for each of them, and then all the 160 dis-
tribution maps were used in the prioritization analysis to locate the priority
regions for trait dimension.
Protected Areas. PA distribution was extracted from the December 2019
release of the WDPA via the wdpar package (126, 127). The release includes
244,869 PAs globally. According to previous similar global studies (e.g., ref. 54),
we extracted the PAs from the WDPA database by selecting terrestrial areas
belonging to IUCN PA categories I to VI and having a status of “designated,”
“inscribed,”or “established”and areas not designated as man and biosphere
reserves by the UN Educational, Scientific, and Cultural Organization. We also
excluded the PAs represented as points. A final list of 95,506 PAs was kept. We
then resampled PAs at the 110-km grid level following ref. 38, that is, labeling
all cells intersecting a PA polygon as PAs to include any small or narrow PAs.
Thus, the PA layer used here provides a very optimistic estimate of existing
PA coverage.
Global Biodiversity Conservation Priority Frameworks (NGO
Frameworks). Many NGOs have proposed frameworks for global biodiversity
conservation prioritization, such as the BH by Conservation International (90),
LW by the Wildlife Conservation Institute (128), and G200 by the World Wide
Fund for Nature (91). However, these frameworks vary in both location and cov-
erage, largely due to the emphasis on different facets of nature conservation.
Although irreplaceability and vulnerability, the two central aspects of systematic
conservation planning (28), are equally important, some frameworks concentrate
only on irreplaceability, while others focus more on vulnerability (29). Under the
framework of irreplaceability and vulnerability, ref. 29 summarized nine major
NGO frameworks, dividing them into three groups:prioritizing high vulnerability
(regions of high threat, purely reactive, e.g., BH), low vulnerability (regions of
low threat, purely proactive, e.g., LW), or high irreplaceability (e.g., G200).
We selected three NGO frameworks of global biodiversity conservation priori-
tizations (29) (SI Appendix,Fig.S5) encompassing the three major groups of the
irreplaceability–vulnerability gradient. Specifically, we selected the BH (ref. 90
and Conservation International), G200 (91), and LW (128). BH prioritizes high
vulnerability, LW prioritizes low vulnerability, and G200 prioritizes high irreplace-
ability. A detailed description of the three selected frameworks can be found in
SI Appendix . The BH data layer was obtained from ref. 129; the upgraded LW
data layer was obtained from ref. 130; and the G200 terrestrial ecoregion layer
was from the World Wildlife Fund (https://www.worldwildlife.org/publications/
global-200). We aggregated all the three spatial layers to 110-km grid spatial
resolution.
Human Pressure Data. We used the recently proposed Human Modification
map (61) as a proxy of human pressure. Compared with the commonly used
Human Footprint map (100, 128), the Human Modification map was modeled
with the incorporation of 13 recent global-scale anthropogenic layers (with a
median year of 2016) to account for the spatial extent, intensity, and
co-occurrence of human activities, many of which show high direct or indirect
impact on biodiversity. HMI values were extracted at a resolution of 5 km
2
to
ease the calculation burden. Based on ref. 61, we categorized the HMI into three
groups representing different intensity levels of human pressure: low (0 ≤HMI
≤0.1); moderate (0.1 <HMI ≤0.4); and highto very high (0.4 <HMI ≤1.0).
Prioritization Analyses. We used Zonation systematic conservation prioritiza-
tion software, version 4 (59, 131) to identify global priority areas for tree diver-
sity in each of the three biodiversity dimensions (taxonomic, phylogenetic, and
functional). Zonation is based on the principle of complementarity, by balancing
a set of biodiversity features to jointly achieve the most effective representation
in a given region, and evaluating the spatial priority areas through the priority
ranking (59). We used the core-area Zonation (CAZ) algorithm as a cell removal
rule. The CAZ ranking algorithm emphasizes species (or any biodiversity feature)
rarity to ensure high-quality locations for all features, even if these features occur
in species-poor areas (58).
We ran the Zonation spatial conservation prioritization procedure on species,
phylogenetic groups, and functional trait groups separately to compare the mis-
match or congruence between the resulting priority areas. We first assessed
each of the obtained priority rankings on both the top 17 and 50% of the high-
est conservation value areas (i.e., the cells with ranking values greater than or
equal to 0.83 or 0.50). These percentages were chosen to reflect current and
future targets for Earth’s PAs (33, 35). However, merging the top-priority areas
issued from the separate analysis of the three diversity dimensions led to total
greater PAs than the 17 and 50% targets, namely 26.3 and 61.5%, respectively
(Fig. 2). Thus, to obtain prioritizations made jointly across the three diversity
dimensions but consistent with the two biodiversity targets, we ran a joint analy-
sis using the three diversity dimensions (taxonomic, phylogenetic, functional) as
input layers of Zonation, and selected the top 17 and top 50% priority areas for
further analyses. In addition, we also performed a sensitivity analysis by compar-
ing the resulting priority areas with those derived from simply combining the
priority areas from the prioritizations on the separate diversity dimensions. We
found the two analyses generated similar results, namely similar percentages of
overlap with existing PAs and conservation priority frameworks (Fig. 5 vs. SI
Appendix,Fig.S14), and similar HMI values of the priority areas inside and out-
side existing PAs (SI Appendix,Fig.S15vs.Fig.S16). Thus, we report the joint
results in the main text (Figs. 3, 4, and 5), that is, given that only these adhere
to the 17 and 50% PA targets.
Spatial Analysis. The relationships among the three priority areas were tested
after accounting for spatial autocorrelation using the SpatialPack package (132).
We evaluated the degree of spatial overlap of the top 17 and 50% priority areas
generated from the three dimensions using a Venn diagram. We divided the pri-
ority areas based on the ranking scores into two categories: areas with priority
scores higher than 0.83 (top 17%), and between 0.50 and 0.83 (top 17 to 50%).
The species’protection proportion was calculated as the overlap ratio between
species’range and existing PAs and the top 17 and 50% priority areas, respec-
tively. Then, the protection percentages for each species were grouped into 12
levels, for example, no overlap as the group of “0,”(0, 10] for less than 10% of
species’range within PAs, and “100”for a species’range completely inside the
PAs. Species’HMIs were calculated separately as the mean HMI value for the
whole species’range, outside PAs, and inside PAs. These steps were repeated
also treating the top 17% priority areas and the top 50% areas as PAs, respec-
tively. The Sankey diagrams were plotted via the ggalluvial package (133) in Fig.
3andSI Appendix,Fig.S3.
Congruence between PAs, top-priority areas, and NGO frameworks was
assessed through gap analyses. We overlaid the three layers (i.e., the top-priority
areas, PAs, and each conservation priority framework) to calculate the level of
protection in PAs, potential protection in the priority framework, and shared pro-
tection of the two global high-priority areas (both 17 and 50%).
Data Availability. All species occurrences reported in this article have been
deposited in the BIEN database (https://bien.nceas.ucsb.edu/bien/)andcanbe
accessed via the RBIEN package. In addition, species’alpha-hull ranges at the
110-km resolution, which were the input data for analyses in the study, have
been deposited in GitHub (https://github.com/wyeco/TC_conservation), together
with the relevant R codes. The phylogeny and imputed trait data and phylogeny
are available in Datasets S2 and S3.
All other study data are included inthe article and/or supporting information.
ACKNOWLEDGMENTS. W.-Y.G., J.M.S.-D., and J.-C.S. acknowledge support from
the Danish Council for Independent Research jNatural Sciences (Grant 6108-
00078B) to the TREECHANGE Project. J.-C.S. also considers this work a contribution
to his VILLUM Investigator Project “Biodiversity Dynamics in a Changing World”
funded by VILLUM FONDEN. We thank Brad Boyle for valuable database and infor-
matics assistance and advice, and TRY contributors for sharing their data. This work
was conducted as a part of the BIEN Working Group, 2008 to 2012. We thank all
the data contributors and numerous herbaria who have contributed their data to
various data-compiling organizations for the invaluable data and support provided
to BIEN. We thank the New York Botanical Garden, the Missouri Botanical Garden,
Utrecht Herbarium, UNC Herbarium, GBIF, REMIB, and SpeciesLink. The staff at
CyVerse provided critical computational assistance. We acknowledge the herbaria
that contributed data to this work: A, AAH, AAS, AAU, ABH, ACAD, ACOR, AD, AFS,
8of11 https://doi.org/10.1073/pnas.2026733119 pnas.org
Downloaded from https://www.pnas.org by Wageningen University & Research - Library on July 20, 2022 from IP address 137.224.252.11.

AK, AKPM, ALCB, ALTA, ALU, AMD, AMES, AMNH, AMO, ANGU, ANSM, ANSP, AQP,
ARAN, ARIZ, AS, ASDM, ASU, AUT, AV, AWH, B, BA, BAA, BAB, BABY, BACP, BAF,
BAFC, BAI, BAJ, BAL, BARC, BAS, BBB, BBS, BC, BCMEX, BCN, BCRU, BEREA, BESA,
BG, BH, BHCB, BIO, BISH, BLA, BM, BOCH, BOL, BOLV, BONN, BOON, BOTU,
BOUM,BPI,BR,BREM,BRI,BRIT,BRLU,BRM,BSB,BUT,C,CALI,CAN,CANB,
CANU, CAS, CATA, CATIE, CAY, CBM, CDA, CDBI, CEN, CEPEC, CESJ, CGE, CGMS,
CHAM, CHAPA, CHAS, CHR, CHSC, CIB, CICY, CIIDIR, CIMI, CINC, CLEMS, CLF,
CMM, CMMEX, CNPO, CNS, COA, COAH, COCA, CODAGEM, COFC, COL, COLO,
CONC, CORD, CP, CPAP, CPUN, CR, CRAI, CRP, CS, CSU, CSUSB, CTES, CTESN, CU,
CUVC, CUZ, CVRD, DAO, DAV, DBG, DBN, DES, DLF, DNA, DPU, DR, DS, DSM,
DUKE,DUSS,E,EA,EAC,EAN,EBUM,ECON,EIF,EIU,EMMA,ENCB,ER,ERA,ESA,
ETH,F,FAA,FAU,FAUC,FB,FCME,FCO,FCQ,FEN,FHO,FI,FLAS,FLOR,FM,FR,
FRU,FSU,FTG,FUEL,FULD,FURB,G,GAT,GB,GDA,GENT,GES,GH,GI,GLM,
GMDRC, GMNHJ, GOET, GRA, GUA, GZU, H, HA, HAC, HAL, HAM, HAMAB, HAO,
HAS, HASU, HB, HBG, HBR, HCIB, HEID, HGM, HIB, HIP, HNT, HO, HPL, HRCB,
HRP, HSC, HSS, HU, HUA, HUAA, HUAL, HUAZ, HUCP, HUEFS, HUEM, HUFU, HUJ,
HUSA, HUT, HXBH, HYO, IAA, IAC, IAN, IB, IBGE, IBK, IBSC, IBUG, ICEL, ICESI, ICN,
IEA, IEB, ILL, ILLS, IMSSM, INB, INEGI, INIF, INM, INPA, IPA, IPRN, IRVC, ISC, ISKW,
ISL, ISTC, ISU, IZAC, IZTA, JACA, JBAG, JBGP, JCT, JE, JEPS, JOTR, JROH, JUA, JYV,
K, KIEL, KMN, KMNH, KOELN, KOR, KPM, KSC, KSTC, KSU, KTU, KU, KUN, KYO, L,
LA, LAGU, LBG, LD, LE, LEB, LIL, LINC, LINN, LISE, LISI, LISU, LL, LMS, LOJA, LOMA,
LP, LPAG, LPB, LPD, LPS, LSU, LSUM, LTB, LTR, LW, LYJB, LZ, M, MA, MACF, MAF,
MAK, MARS, MARY, MASS, MB, MBK, MBM, MBML, MCNS, MEL, MELU, MEN,
MERL, MEXU, MFA, MFU, MG, MGC, MICH, MIL, MIN, MISSA, MJG, MMMN,
MNHM,MNHN,MO,MOL,MOR,MPN,MPU,MPUC,MSB,MSC,MSUN,MT,
MTMG, MU, MUB, MUR, MVFA, MVFQ, MVJB, MVM, MW, MY, N, NA, NAC, NAS,
NCU, NE, NH, NHM, NHMC, NHT, NLH, NM, NMB, NMNL, NMR, NMSU, NSPM,
NSW, NT, NU, NUM, NY, NZFRI, O, OBI, ODU, OS, OSA, OSC, OSH, OULU, OWU,
OXF, P, PACA, PAMP, PAR, PASA, PDD, PE, PEL, PERTH, PEUFR, PFC, PGM, PH,
PKDC, PLAT, PMA, POM, PORT, PR, PRC, PRE, PSU, PY, QCA, QCNE, QFA, QM,
QRS,QUE,R,RAS,RB,RBR,REG,RELC,RFA,RIOC,RM,RNG,RSA,RYU,S,SACT,
SALA,SAM,SAN,SANT,SAPS,SASK,SAV,SBBG,SBT,SCFS,SD,SDSU,SEL,SEV,
SF, SFV, SGO, SI, SIU, SJRP, SJSU, SLPM, SMDB, SMF, SNM, SOM, SP, SPF, SPSF,
SQF,SRFA,STL,STU,SUU,SVG,TAES,TAI,TAIF,TALL,TAM,TAMU,TAN,TASH,TEF,
TENN, TEPB, TEX, TFC, TI, TKPM, TNS, TO, TOYA, TRA, TRH, TROM, TRT, TRTE, TU,
TUB, U, UADY, UAM, UAMIZ, UB, UBC, UC, UCMM, UCR, UCS, UCSB, UCSC, UEC,
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, USM, USNC, USP, USZ, UT, UTC, UTEP, UU,
UVIC,UWO,V,VAL,VALD,VDB,VEN,VIT,VMSL,VT,W,WAG,WAT,WELT,WFU,
WII, WIN, WIS, WMNH, WOLL, WS, WTU, WU, XAL, YAMA, Z, ZMT, ZSS, and ZT.
C.B. was supported by a National Research Foundation of Korea (NRF) grant
funded by the Korean government (MIST) (2022R1A2C1003504). A.S.M. was sup-
ported by the Environment Research and Technology Development Fund (S-14) of
the Ministry of the Environment, Japan. J. Pisek was supported by Estonian
Research Council Grants PUT 1355 and PRG 1405. J. Pe~
nuelas was funded by
European Research Council Synergy Grant ERC-2013-SyG-610028 IMBALANCE-P.
A.G.G. was funded by National Fund for Scientific and Technological Development
(FONDECYT) grant 1200468 and Agencia Nacional de Investigaci
on y Desarrollo
(ANID/BASAL) FB210006. V.D.P. was funded by Conselho Nacional de Desenvolvi-
mento Cient

ıfico e Tecnol
ogico (CNPq), Brazil (grant 307689/2014-0). The BIEN
Working Group was supported by the National Center for Ecological Analysis and
Synthesis, a center funded by NSF EF-0553768 at the University of California,
Santa Barbara and the State of California. Additional support for the BIEN Working
Group was provided by iPlant/CyVerse via NSF DBI-0735191. B.J.E. and C.M. were
supported by NSF ABI-1565118 and NSF HDR-1934790. B.J.E. was also supported
by a Global Environment Facility Spatial Planning for Protected Areas in Response
to Climate Change Project grant (GEF-5810). B.J.E., C.V., and B.S.M. are
partly supported by the Fondation pour la Recherche sur la Biodiversit
e
(FRB) and Electricit
e de France (EDF) in the context of the CESAB (Centre for
the Synthesis and Analysis of Biodiversity) project “Causes and consequen-
ces of functional rarity from local to global scales”(FREE). N.A.S. was
supported by Vidi Grant 016.161.318 issued by the Netherlands Organiza-
tion for ScientificResearch.
Author affiliations:
a
Center for Biodiversity Dynamics in a Changing World (BIOCHANGE),
Department of Biology, Aarhus University, 8000 Aarhus C, Denmark;
b
Section for
Ecoinformatics & Biodiversity, Department of Biology, Aarhus University, 8000 Aarhus C,
Denmark;
c
Zhejiang Tiantong Forest Ecosystem National Observation and Research
Station, School of Ecological and Environmental Sciences, East China Normal University,
200241 Shanghai, People’s Republic of China;
d
Researc h Center f or Globa l Change a nd
Complex Ecosystems, School of Ecological and Environmental Sciences, East China Normal
University, 200241 Shanghai, People’s Republic of China;
e
UMR Silva, Universit
ede
Lorraine, AgroParisTech, and INRAE, 54000 Nancy, France;
f
School of Geography,
University of Nottingham, Nottingham NG7 2RD, United Kingdom;
g
Department of Ecology
and Evolutionary Biology, University of Arizona, Tucson, AZ 85721;
h
Eversource Energy
Center, University of Connecticut, Storrs, CT 06268;
i
Department of Ecology and
Evolutionary Biology, University of Connecticut, Storrs, CT 06268;
j
CEFE, Uni Montpellier,
CNRS, EPHE, IRD, 34293 Montpellier Cedex 5, France;
k
School of Environmental Sciences,
University of Guelph, Guelph, ON N1G 2W1, Canada;
l
Centre for Forest Research,
D
epartement des Sciences Biologiques, Universit
eduQu

ebec 
aMontr

eal, Montr
eal, QC
H3C 3P8, Canada;
m
Department of Biology, University o f Copenhagen, 2100 Copenhagen
Ø,Denmark;
n
Department of Biological Sciences and Biotechnology, Andong National
University, Andong 36729, Korea;
o
Department of Geography, King’sCollegeLondon,
London WC2B 4BG, United Kingdom;
p
Department of Biotechnology and Life Sciences,
University of Insubria, I-21100 Varese, Italy;
q
Escuela de Biolog

ıa, Universidad de Costa
Rica, 11501-2060 San Jose, Costa Rica;
r
Department of Biology, University o f Pisa, 56126
Pisa, Italy;
s
Department of Ecological Science, F aculty of Science, Vrije Universiteit, 1081 HV
Amsterdam, The Netherlands;
t
University of Science, 700000 Ho Ch i Minh City, Vietnam;
u
Vietnam National University, 700000 Ho Chi Minh City, Vietnam;
v
German Centre for
Integrative Biodiversity Research (iDiv), 04103 Leipzig, Germany;
w
Institute for Physical
Geography, Goethe University, 60438 Frankfurt am Main, Germany;
x
Departamento de
Ensino, Instituto Federal de Educac¸~
ao, Ci^
encias e Tecnologia do Cear
a, Crate
us 63708-260,
Brazil;
y
Departamento de Ciencias Ambientales y Recursos Naturales Renovables, Facultad
de Ciencias Agron
omicas, Universidad de Chile, Santa Rosa 11315, La Pintana, Santiago,
Chile;
z
Institute of Ecology and Biodiversity (IEB), Barrio Universitario, 4070374 Concepci 
on,
Chile;
aa
Global Systems and Analytics, Nova Pioneer, Paulshof, Gauteng, 2191, South
Africa;
bb
School of Molecular and Life Sciences, Curtin University, Perth, WA 6845,
Australia;
cc
College of Science, Health, Engineering and Education, Murdoch University,
Murdoch, WA 6150, Australia;
dd
Institute of Botany, University of Natural Resources and
Life Sciences, 1180 Vienna, Austri a;
ee
Meadow Run Environmental, Leavenworth, WA
98826;
ff
Institute of Systematic Botany and Ecology, Ulm University, 89081 Ulm, Germany;
gg
Max Planck Institute for Biogeochemistry, 07745 Jena, Germany;
hh
Department of Plant
& Environmental Sciences, Weizmann Institute of Science, 76100 Rehovot, Israel;
ii
Centre
d’Estudis de la Neu i la Muntanya d’Andorra, Institut d’Estudis, Andorrans (CENMA–IEA),
AD600 Sant Juli
adeL

oria, Principality of Andorra;
jj
Department of Ecology and
Evolutionary Biology, University of California, Los Angeles, CA 90095;
kk
Forest Ecology and
Management Group, Wageningen Universi ty, 6700 AA Wageningen, The N etherlands;
ll
Land Life Company, 1092AD Amsterdam, The Netherlands;
mm
Laboratoire d’Ecologie
Alpine, LECA, UMR UGA-USMB-CNRS 5553, Universit
e Grenoble Alpes, 38058 Grenoble
Cedex 9, France;
nn
Environmental Research Institute, University of Wa ikato, Hamilton
3240, New Zealand;
oo
Department of Physical and Environmental Sciences, University of
Toronto Scarborough, Toronto, ON M1C 1A4, Canada;
pp
ICREA, 08010 Barcelona, Spain;
qq
CREAF, Universidad Autonoma de Barcelona, 08193 Barcelona, Spain;
rr
Department of
Botany,UniversityofBritishColumbia,Vancouver,BCV6T1Z4,Canada;
ss
Biodiversity
Research Centre, University of British Columbia, Vancouver, BC V6T 1Z4, Canada;
tt
Department of Biology, Vrije Universiteit Brussel, 1050 Brussels, Belgium;
uu
Institute for
Biology and Environmental Sciences, University of Oldenburg, 26129 Oldenburg, Germany;
vv
Graduate School of Environment and Information Sciences, Yokohama National
University, Hodogaya, Yokohama 240-8501, Japan;
ww
Estonian University of Life Sciences,
51006 Tartu, Estonia;
xx
Division of Forest and Biomaterials Science, Graduate School of
Agriculture, Kyoto University, Oiwake, Kitashirakawa, Kyoto 606-8502 Jap an;
yy
CREAF,
Cerdanyola del Vall
es, Barcelona, 08193 Catalonia, S pain;
zz
CSIC, Global Ecology U nit
CREAF, CSIC–UAB, Bellaterra, Barcelona, 08193 Catalonia, Spain;
aaa
Department of Ecology,
Universidade Federal do Rio Grande do Sul, Porto Alegre 91501-970, Brazil;
bbb
Tartu
Observatory, University of Tartu, T~
oravere, 61602 Tartumaa, Estonia;
ccc
Aquatic Ecology &
Environmental Biology Group, Radboud Institute for Biological and Environmental
Sciences, Faculty of Science, Radboud Universi ty Nijmegen, 6525 AJ Nijmegen, The
Netherlands;
ddd
Department of Biology, Algoma University, Sault Ste. Marie, ON P6A 2G4,
Canada;
eee
Smithsonian Tropical Research Institute, Apartado 0843-03092, Balboa, Anc
on,
Republic of Panama;
fff
Embrapa Clima Temperado, 96010-971 Pelotas, RS, Brazil;
ggg
Centre
for Environmental Sciences, Hasselt University, 3500 Hasselt, Belgium;
hhh
Canadian Wood
Fibre Centre, Natural Resources Canada, Qu
ebec City, QC G1V 4C7, Canada;
iii
Institute of
Environmental Sciences, Leiden University, 2333 CC Leiden, The Netherlands;
jjj
Plant
Ecology and Nature Conservation G roup, Wageningen University, 6700 AA W ageningen,
The Netherlands;
kkk
Hawkesbury Institute for the Environment, Western Sydney University,
Penrith, NSW 2751, Australia;
lll
School of Natural Sciences, Macquarie University, North
Ryde, NSW 2109, Australia;
mmm
Beijing Key Laboratory for Forest Resources and
Ecosystem Processes, Beijing Forestry University, Beijing 100083, People’s Republic of
China; and
nnn
The Santa Fe Institute, Santa Fe, NM 87501
Author contributions: W.-Y.G. and J.-C.S. designed research; W.-Y.G. performed
research; J.M.S.-D., F.S., W.L.E., B.S.M., C.M., C.V., M.A., M.B., H.H.B., C.B., J.A.C., 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., K.G., A.G.G., W.H., T.H., P.H., N.H.-S.,
S.J., J.K., T.K., B.K., N.J.B.K., K.K., S.L., C.H.L., A.R.M., M.M., S.T.M., V.M., A.S.M.,
€
U.N., Y.O.,
J. Pe~
nuelas, V.D.P., J. Pisek, B.J.M.R., B.S., M.S.,
^
E.E.S., N.A.S., N.T., P.v.B., F.v.d.P., I.J.W.,
W.-B.X., J.Z., and B.J.E. contributed new reagents/analytic tools; W.-Y.G. analyzed data
with the help of J.M.S.-D., K.G., and W.-B.X.; and W.-Y.G., J.M.S.-D. and J.-C.S. wrote the
paper with the contributions of all authors.
1. Y. Pan et al., A large and persistent carbon sink in the world’s forests. Science 333, 988–993
(2011).
2. B. J. Enquist, A. J. Abraham, M. B. J. Harfoot, Y. Malhi, C. E. Doughty, The megabiota are
disproportionately importantfor biosphere functioning. Nat. Commun. 11, 699
(2020).
3. J. F. Bastin et al., The global tree restoration potential. Science 365,76–79 (2019).
4. E. G. Brockerhoff et al., Forest biodiversity, ecosystem functioning and the provision of ecosystem
services. Biodivers. Conserv. 26, 3005–3035 (2017).
5. L. Gibson et al., Primary forests are irreplaceable for sustaining tropical biodiversity. Nature 478,
378–381 (2011).
PNAS 2022 Vol. 119 No. 25 e2026733119 https://doi.org/10.1073/pnas.2026733119 9of11
Downloaded from https://www.pnas.org by Wageningen University & Research - Library on July 20, 2022 from IP address 137.224.252.11.

6. T. R. E. Southwood, The number of species of insect associated with various trees. J. Anim. Ecol.
30,1–8 (1961).
7. L. Tedersoo et al., Tree diversity and species identity effects on soil fungi, protists and animals are
context dependent. ISME J. 10, 346–362 (2016).
8. A. D. Barnes et al., Direct and cascading impacts of tropical land-use change on multi-trophic
biodiversity. Nat. Ecol. Evol. 1, 1511–1519 (2017).
9. R. Pillay et al., Tropical forests are home to over half of the world’s vertebrate species. Front. Ecol.
Environ. 20,10–15 (2022).
10. M. Barrufol et al., Biodiversity promotes tree growth during succession in subtropical forest. PLoS
One 8, e81246 (2013).
11. Y. Huang et al., Impacts of species richness on productivity in a large-scale subtropical forest
experiment. Science 362,80–83 (2018).
12. X. Liu et al., Tree species richness increases ecosystem carbon storage in subtropical forests. Proc.
Biol. Sci. 285, 20181240 (2018).
13. A. S. Mori, Biodiversity and ecosystem services in forests: Management and restoration founded
on ecological theory. J. Appl. Ecol. 54,7–11 (2017).
14. F. Achard et al., Determination of deforestation rates of the world’s humid tropical forests. Science
297, 999–1002 (2002).
15. M. C. Hansen, S. V. Stehman, P. V. Potapov, Quantification of global gross forest cover loss. Proc.
Natl. Acad. Sci. U.S.A. 107, 8650–8655 (2010).
16. M. C. Hansen et al., High-resolution global maps of 21st-century forest cover change. Science
342, 850–853 (2013).
17. R. S. Defries, T. Rudel, M. Uriarte, M. Hansen, Deforestation driven by urban population growth
and agricultural trade in the twenty-first century. Nat. Geosci. 3, 178–181 (2010).
18. H. K. Gibbs et al., Tropical forests were the primary sources of new agricultural land in the 1980s
and 1990s. Proc. Natl. Acad. Sci. U.S.A. 107, 16732–16737 (2010).
19. A. B. Adams, J. Pontius, G. Galford, D. Gudex-Cross, Simulating forest cover change in the
northeastern U.S.: Decreasing forest area and increasing fragmentation. Landsc. Ecol. 34,
2401–2419 (2019).
20. H. Escobar, Deforestation in the Brazilian Amazon is still rising sharply. Science 369, 613 (2020).
21. W. F. Laurance et al., Averting biodiversity collapse in tropical forest protected areas. Nature 489,
290–294 (2012).
22. J. Alroy, Effects of habitat disturbance on tropical forest biodiversity. Proc. Natl. Acad. Sci. U.S.A.
114, 6056–6061 (2017).
23. N. A. Soudzilovskaia et al., Global mycorrhizal plant distribution linked to terrestrial carbon stocks.
Nat. Commun. 10, 5077 (2019).
24. J. Liang et al., Positive biodiversity-productivity relationship predominant in global forests.
Science 354, 196 (2016).
25. M. T. Coe et al., The forests of the Amazon and Cerrado moderate regional climate and are the key
to the future. Trop. Conserv. Sci. 10,1–6 (2017).
26. A. B. Harper et al., Land-use emissions play a critical role in land-based mitigation for Paris
climate targets. Nat. Commun. 9, 2938 (2018).
27. G. M. Mace et al., Aiming higher to bend the curve of biodiversity loss. Nat. Sustain. 1, 448–451
(2018).
28. C. R. Margules, R. L. Pressey, Systematic conservation planning. Nature 405, 243–253 (2000).
29. T. M. Brooks et al., Global biodiversity conservation priorities. Science 313,58–61 (2006).
30. J. Geldmann, A. Manica, N. D. Burgess, L. Coad, A. Balmford, A global-level assessment of the
effectiveness of protected areas at resisting anthropogenic pressures. Proc. Natl. Acad. Sci. U.S.A.
116, 23209–23215 (2019).
31. B. Soares-Filho et al., Role of Brazilian Amazon protected areas in climate change mitigation. Proc.
Natl. Acad. Sci. U.S.A. 107, 10821–10826 (2010).
32. O. Venter et al., Targeting global protected area expansion for imperiled biodiversity. PLoS Biol.
12, e1001891 (2014).
33. P. Visconti et al., Protected area targets post-2020. Science 364, 239–241 (2019).
34. S. Woodley et al., A bold successor to Aichi target 11. Science 365, 649–650 (2019).
35. E. O. Wilson, Half-Earth: Our Planet’s Fight for Life (W. W. Norton, 2016).
36. S. L. Pimm, C. N. Jenkins, B. V. Li, How to protect half of Earth to ensure it protects sufficient
biodiversity. Sci. Adv. 4, eaat2616 (2018).
37. J. Schleicher et al., Protecting half of the planet could directly affect over one billion people. Nat.
Sustain. 2, 1094–1096 (2019).
38. E. Dinerstein et al., A global deal for nature: Guiding principles, milestones, and targets. Sci. Adv.
5, eaaw2869 (2019).
39. E. Dinerstein et al.,A“global safety net”to reverse biodiversity loss and stabilize Earth’s climate.
Sci. Adv. 6, eabb2824 (2020).
40. Convention on Biological Diversity, Synthesis of views of parties and observers on the scope and
content of the post-2020 global biodiversity framework. https://www.cbd.int/doc/c/de9c/8c12/
7c0cb88a47f9084e5d0b82eb/post2020-prep-01-inf-01-en.pdf. Accessed 9 August 2021.
41. F. Montesino Pouzols et al., Global protected area expansion is compromised by projected land-
use and parochialism. Nature 516, 383–386 (2014).
42. A. L. Strecker, J. D. Olden, J. B. Whittier, C. P. Paukert, Defining conservation priorities for
freshwater fishes according to taxonomic, functional, and phylogenetic diversity. Ecol. Appl. 21,
3002–3013 (2011).
43. A. Arponen, L. Zupan, “Representing hotspots of evolutionary history in systematic conservation
planning for European mammals”in Biodiversity Conservation and Phylogenetic Systematics,
P. Roseli, P. Grandcolas, Eds. (Springer International, 2016), pp. 265–285.
44. F. T. Brum et al., Global priorities for conservation across multiple dimensions of mammalian
diversity. Proc. Natl. Acad. Sci. U.S.A. 114, 7641–7646 (2017).
45. V. Devictor et al., Spatial mismatch and congruence between taxonomic, phylogenetic and
functional diversity: The need for integrative conservation strategies in a changing world. Ecol.
Lett. 13, 1030–1040 (2010).
46. L. Zupan et al., Spatial mismatch of phylogenetic diversity across three vertebrate groups and
protected areas in Europe. Divers. Distrib. 20, 674–685 (2014).
47. J. F. Gonz
alez-Maya, L. R. V

ıquez-R, A. Arias-Alzate, J. L. Belant, G. Ceballos, Spatial patterns of
species richness and functional diversity in Costa Rican terrestrial mammals: Implications for
conservation. Divers. Distrib. 22,43–56 (2016).
48. F. Mazel et al., Prioritizing phylogenetic diversity captures functional diversity unreliably. Nat.
Commun. 9, 2888 (2018).
49. F. Mazel, A. O. Mooers, G. V. D. Riva, M. W. Pennell, Conserving phylogenetic diversity can be a
poor strategy for conserving functional diversity. Syst. Biol. 66, 1019–1027 (2017).
50. B. H. Daru et al., Spatial overlaps between the global protected areas network and terrestrial
hotspots of evolutionary diversity. Glob. Ecol. Biogeogr. 28, 757–766 (2019).
51. G. Rapacciuolo et al., Species diversity as a surrogate for conservation of phylogenetic and
functional diversity in terrestrial vertebrates across the Americas. Nat. Ecol. Evol. 3,53–61 (2019).
52. L. Shi, W. Feng, J. Xu, Y. Kuzyakov, Agroforestry systems: Meta-analysis of soil carbon stocks,
sequestration processes, and future potentials. Land Degrad. Dev. 29, 3886–3897 (2018).
53. S. Kay et al., Agroforestry creates carbon sinks whilst enhancing the environment in agricultural
landscapes in Europe. Land Use Policy 83, 581–593 (2019).
54. K. R. Jones et al., One-third of global protected land is under intense human pressure. Science
360, 788–791 (2018).
55. K. Schulze et al., An assessment of threats to terrestrial protected areas. Conserv. Lett. 11, e12435
(2018).
56. J. R. Allan et al., The minimum land area requiring conservation attention to safeguard
biodiversity. bioRxiv [Preprint] (2021). https://doi.org/10.1101/839977 (Accessed 12 August
2021).
57. B. Mappin et al., Restoration priorities to achieve the global protected area target. Conserv. Lett.
12, e12646 (2019).
58. A. Moilanen et al., Prioritizing multiple-use landscapes for conservation: Methods for large multi-
species planning problems. Proc. Biol. Sci. 272, 1885–1891 (2005).
59. J. Lehtom€
aki, A. Moilanen, Methods and workflow for spatial conservation prioritization using
Zonation. Environ. Model. Softw. 47, 128–137 (2013).
60. J. A. F. Diniz-Filho, C. E. R. de Sant’Ana, L. M. Bini, An eigenvector method for estimating
phylogenetic inertia. Evolution 52, 1247–1262 (1998).
61. C. M. Kennedy, J. R. Oakleaf, D. M. Theobald, S. Baruch-Mordo, J. Kiesecker, Managing the
middle: A shift in conservation priorities based on the global human modification gradient. Glob.
Change Biol. 25, 811–826 (2019).
62. R. Cazzolla Gatti et al., The number of tree species on Earth. Proc. Natl. Acad. Sci. U.S.A. 119,
e2115329119 (2022).
63. M. Rivers, “European Red List of trees”(IUCN,2019). https:/doi.org/10.2305/iucn.ch.2019.erl.1.en
(Accessed 9 August 2021).
64. W. B. Xu et al., Human activities have opposing effects on distributions of narrow-ranged and
widespread plant species in China. Proc. Natl. Acad. Sci. U.S.A. 116, 26674–26681 (2019).
65. R. Aerts, O. Honnay, Forest restoration, biodiversity and ecosystem functioning. BMC Ecol. 11,29
(2011).
66. T. Fremout et al., Mapping tree species vulnerability to multiple threats as a guide to restoration
and conservation of tropical dry forests. Glob. Change Biol. 26, 3552–3568 (2020).
67. H. Gaisberger et al., Tropical and subtropical Asia’s valued tree species under threat. Conserv.
Biol., 10.1111/cobi.13873 (2021).
68. R. Crouzeilles et al., Ecological restoration success is higher for natural regeneration than for
active restoration in tropical forests. Sci. Adv. 3, e1701345 (2017).
69. R. L. Chazdon, Beyond deforestation: Restoring forests and ecosystem services on degraded
lands. Science 320, 1458–1460 (2008).
70. C. N. Jenkins, S. L. Pimm, L. N. Joppa, Global patterns of terrestrial vertebrate diversity and
conservation. Proc. Natl. Acad. Sci. U.S.A. 110, E2602–E2610 (2013).
71. C. N. Jenkins, K. S. Van Houtan, S. L. Pimm, J. O. Sexton, US protected lands mismatch
biodiversity priorities. Proc. Natl. Acad. Sci. U.S.A. 112, 5081–5086 (2015).
72. A. Santangeli et al., Priority areas for conservation of Old World vultures. Conserv. Biol. 33,
1056–1065 (2019).
73. F. Forest et al., Preserving the evolutionary potential of floras in biodiversity hotspots. Nature
445, 757–760 (2007).
74. E. Beech, M. Rivers, S. Oldfield, P. P. Smith, GlobalTreeSearch: The first complete global database
of tree species and country distributions. J. Sustain. For. 36, 454–489 (2017).
75. M. A. Batalha, I. A. Silva, M. V. Cianciaruso, G. H. de Carvalho, Trait diversity on the phylogeny of
Cerrado woody species. Oikos 120, 1741–1751 (2011).
76. E. Dinerstein et al., An ecoregion-based approach to protecting half the terrestrial realm.
Bioscience 67, 534–545 (2017).
77. Botanic Gardens Conservation International, Global tree assessment. https://www.bgci.org/our-
work/projects-and-case-studies/global-tree-assessment/. Accessed 9 September 2021.
78. G. M. Mace et al., Quantification of extinction risk: IUCN’s system for classifying threatened
species. Conserv. Biol. 22, 1424–1442 (2008).
79. B. J. Enquist et al., The commonness of rarity: Global and future distribution of rarity across land
plants. Sci. Adv. 5, eaaz0414 (2019).
80. S. P. Hubbell et al., Colloquium paper: How many tree species are there in the Amazon and how
many of them will go extinct? Proc. Natl. Acad. Sci. U.S.A. 105 (suppl. 1), 11498–11504 (2008).
81. F.-Y. Hua et al., Tree plantations displacing native forests: The nature and drivers of apparent
forest recovery on former croplands in southwestern China from 2000 to 2015. Biol. Conserv.
222, 113–124 (2018).
82. D. A. Jensen et al., The potential for using rare, native species in reforestation—A case study of
yews (Taxaceae) in China. For. Ecol. Manage. 482, 118816 (2021).
83. M. Chapman et al., Large climate mitigation potential from adding trees to agricultural lands.
Glob. Change Biol. 26, 4357–4365 (2020).
84. C. Soto-Navarro et al., Mapping co-benefits for carbon storage and biodiversity to inform
conservation policy and action. Philos. Trans. R. Soc. Lond. B Biol. Sci. 375, 20190128 (2020).
85. D. F. Doak, V. J. Bakker, B. E. Goldstein, B. Hale, What is the future of conservation? Trends Ecol.
Evol. 29,77–81 (2014).
86. W. Xu et al., Strengthening protected areas for biodiversity and ecosystem services in China. Proc.
Natl. Acad. Sci. U.S.A. 114, 1601–1606 (2017).
87. J. M. Melillo et al., Protected areas’role in climate-change mitigation. Ambio 45, 133–145
(2016).
88. M. Jung et al., Areas of global importance for conserving terrestrial biodiversity, carbon and
water. Nat. Ecol. Evol. 5, 1499–1509 (2021).
89. R. L. Pressey et al., The mismeasure of conservation. Trends Ecol. Evol. 36, 808–821
(2021).
90. N. Myers, R. A. Mittermeier, C. G. Mittermeier, G. A. B. da Fonseca, J. Kent, Biodiversity hotspots
for conservation priorities. Nature 403, 853–858 (2000).
10 of 11 https://doi.org/10.1073/pnas.2026733119 pnas.org
Downloaded from https://www.pnas.org by Wageningen University & Research - Library on July 20, 2022 from IP address 137.224.252.11.

91. D. M. Olson, E. Dinerstein, The Global 200: Priority ecoregions for global conservation. Ann. Mo.
Bot. Gard. 89, 199 (2002).
92. P. Keil, J. M. Chase, Global patterns and drivers of tree diversity integrated across a continuum of
spatial grains. Nat. Ecol. Evol. 3, 390–399 (2019).
93. J. E. M. Watson et al., The exceptional value of intact forest ecosystems. Nat. Ecol. Evol. 2,
599–610 (2018).
94. J. M. Serra-Diaz, B. J. Enquist, B. Maitner, C. Merow, J. C. Svenning, Big data of tree species
distributions: How big and how good? For. Ecosyst. 4, 30 (2017).
95. L. J. Pollock, W. Thuiller, W. Jetz, Large conservation gains possible for global biodiversity facets.
Nature 546, 141–144 (2017).
96. J. P. Schmidt, T. J. Davies, M. J. Farrell, Opposing macroevolutionary and trait-mediated patterns
of threat and naturalisation in flowering plants. Ecol. Lett. 24, 1237–1250 (2021).
97. W.-T. Wang, W.-Y. Guo, J. Scott, J. M. Serra-Diaz, J.-C. Svenning, Anthropogenic climate change
increases vulnerability of Magnolia species more in Asia than in the Americas. Biol. Conserv. 265,
109425 (2022).
98. C. Wyborn, M. C. Evans, Conservation needs to break free from global priority mapping. Nat. Ecol.
Evol. 5, 1322–1324 (2021).
99. B. B. N. Strassburg et al., Global priority areas for ecosystem restoration. Nature 586, 724–729 (2020).
100. O. Venter et al., Global terrestrial Human Footprint maps for 1993 and 2009. Sci. Data 3, 160067
(2016).
101. J. S. Hall et al., The ecology and ecosystem services of native trees: Implications for reforestation
and land restoration in Mesoamerica. For. Ecol. Manage. 261, 1553–1557 (2011).
102. B. Boyle et al., The Taxonomic Name Resolution Service: An online tool for automated
standardization of plant names. BMC Bioinformatics 14, 16 (2013).
103. B. J. Enquist, R. R. Condit, R. K. Peet, M. Schildhauer, B. M. Thiers, The Botanical Information and
Ecology Network (BIEN): Cyberinfrastructure for an integrated botanical information network to
investigate the ecological impacts of global climate change on plant biodiversity. PeerJ 4,
e2615v2 (2016).
104. K. Banda-R et al.; DRYFLOR, Plant diversity patterns in neotropical dry forests and their
conservation implications. Science 353, 1383–1387 (2016).
105. G. Dauby et al., RAINBIO: A mega-database of tropical African vascular plants distributions.
PhytoKeys 74,1–18 (2016).
106. C. Meyer, P. Weigelt, H. Kreft, Multidimensional biases, gaps and uncertainties in global plant
occurrence information. Ecol. Lett. 19, 992–1006 (2016).
107. D. G. Kirkpatrick, R. Seidel, On the shape of a set of points in the plane. IEEE Trans. Inf. Theory 29,
551–559 (1983).
108. B. Pateiro-L
opez, A. Rodr

ıguez-Casal, Generalizing the convex hull of a sample: The R package
alphahull. J. Stat. Softw. 34,1–28 (2010).
109. R Core Team, R: A Language and Environment for Statistical Computing (R Foundation for
Statistical Computing, 2019).
110. M. A. Burgman,J. C. Fox, Bias inspecies rangeestimatesfrom minimumconvex polygons:
Implications for conservation and options for improvedplanning. Anim. Conserv. 6,19–28 (2003).
111. M. C. Rivers, S. P. Bachman, T. R. Meagher, E. N. Lughadha, N. A. Brummitt, Subpopulations,
locations and fragmentation: Applying IUCN Red List criteria to herbarium specimen data.
Biodivers. Conserv. 19, 2071–2085 (2010).
112. E. Garc

ıa-Rosell
oet al., Can we derive macroecological patterns from primary Global Biodiversity
Information Facility data? Glob. Ecol. Biogeogr. 24, 335–347 (2015).
113. A. Mauri, G. Strona, J. San-Miguel-Ayanz, EU-Forest, a high-resolution tree occurrence dataset for
Europe. Sci. Data 4, 160123 (2017).
114. C. Rahbek, G. R. Graves, Multiscale assessment of patterns of avian species richness. Proc. Natl.
Acad. Sci. U.S.A. 98, 4534–4539 (2001).
115. R. Grenyer et al., Global distribution and conservation of rare and threatened vertebrates. Nature
444,93–96 (2006).
116. B. Vilela, F. Villalobos, R. Lets, A new R package for data handling and analysis in macroecology.
Methods Ecol. Evol. 6, 1229–1234 (2015).
117. K. J. Gaston, R. A. Fuller, The sizes of species’geographic ranges. J. Appl. Ecol. 46,1–9 (2009).
118. S. A. Smith, J. W. Brown, Constructing a broadly inclusive seed plant phylogeny. Am. J. Bot. 105,
302–314 (2018).
119. S. Magall
on, S. G
omez-Acevedo, L. L. S
anchez-Reyes, T. Hern
andez-Hern
andez, A metacalibrated
time-tree documents the early rise of flowering plant phylogenetic diversity. New Phytol. 207,
437–453 (2015).
120. J. A. F. Diniz-Filho et al., On the selection of phylogenetic eigenvectors for ecological analyses.
Ecography 35, 239–249 (2012).
121. D. P. Faith, Conservation evaluation and phylogenetic diversity. Biol. Conserv. 61,1–10 (1992).
122. V. Veach, E. Di Minin, F. M. Pouzols, A. Moilanen, Species richness as criterion for global
conservation area placement leads to large losses in coverage of biodiversity. Divers. Distrib. 23,
715–726 (2017).
123. F. Schrodt et al., BHPMF—A hierarchical Bayesian approach to gap-filling and trait prediction for
macroecology and functional biogeography. Glob. Ecol. Biogeogr. 24, 1510–1521 (2015).
124. P. Kampstra, beanplot: A boxplot alternative for visual comparison of distributions. J. Stat. Softw.
28,1–9(2015).
125. L. J. Revell, phytools: An R package for phylogenetic comparative biology (and other things).
Methods Ecol. Evol. 3, 217–223 (2012).
126. UN Environment Programme World Conservation Monitoring Centre; IUCN, Protected Planet: The
World Database on Protected Areas (WDPA, 2019).
127. J. O. Hanson, wdpar: Interface to the World Database on Protected Areas. https://prioritizr.github.
io/wdpar/ (Accessed 14 December 2019).
128. E. W. Sandersonet al., The human footprint and the last of the Wwild. Bioscien ce 52, 891–904
(2002).
129. M. Hoffman, K. Koenig, G. Bunting, J. Costanza, K. J. Williams, Biodiversity hotspots, Version
2016.1. https:/doi.org/10.5281/zenodo.3261807. Accessed 24 December 2019.
130. J. R. Allan, O. Venter, J. E. M. Watson, Temporally inter-comparable maps of terrestrial wilderness
and the Last of the Wild. Sci. Data 4, 170187 (2017).
131. A. Moilanen, J. R. Leathwick, J. M. Quinn, Spatial prioritization of conservation management.
Conserv. Lett. 4, 383–393 (2011).
132. F. Osorio, R. Vallejos, F. Cuevas, SpatialPack: Computing the association between two spatial
processes. arXiv [Preprint] (2016). https://doi.org/10.48550/arXiv.1611.05289 (Accessed 12
August 2021).
133. J. Brunson, ggalluvial: Layered grammar for alluvial plots. J. Open Source Softw. 5, 2017
(2020).
PNAS 2022 Vol. 119 No. 25 e2026733119 https://doi.org/10.1073/pnas.2026733119 11 of 11
Downloaded from https://www.pnas.org by Wageningen University & Research - Library on July 20, 2022 from IP address 137.224.252.11.