ArticlePDF Available

Asynchronous Carbon Sink Saturation in African and Amazonian Tropical Forests

Authors:
  • University College London and University of Leeds.

Abstract and Figures

Structurally intact tropical forests sequestered about half of the global terrestrial carbon uptake over the 1990s and early 2000s, removing about 15 per cent of anthropogenic carbon dioxide emissions. Climate-driven vegetation models typically predict that this tropical forest ‘carbon sink’ will continue for decades. Here we assess trends in the carbon sink using 244 structurally intact African tropical forests spanning 11 countries, compare them with 321 published plots from Amazonia and investigate the underlying drivers of the trends. The carbon sink in live aboveground biomass in intact African tropical forests has been stable for the three decades to 2015, at 0.66 tonnes of carbon per hectare per year (95 per cent confidence interval 0.53–0.79), in contrast to the long-term decline in Amazonian forests. Therefore the carbon sink responses of Earth’s two largest expanses of tropical forest have diverged. The difference is largely driven by carbon losses from tree mortality, with no detectable multi-decadal trend in Africa and a long-term increase in Amazonia. Both continents show increasing tree growth, consistent with the expected net effect of rising atmospheric carbon dioxide and air temperature. Despite the past stability of the African carbon sink, our most intensively monitored plots suggest a post-2010 increase in carbon losses, delayed compared to Amazonia, indicating asynchronous carbon sink saturation on the two continents. A statistical model including carbon dioxide, temperature, drought and forest dynamics accounts for the observed trends and indicates a long-term future decline in the African sink, whereas the Amazonian sink continues to weaken rapidly. Overall, the uptake of carbon into Earth’s intact tropical forests peaked in the 1990s. Given that the global terrestrial carbon sink is increasing in size, independent observations indicating greater recent carbon uptake into the Northern Hemisphere landmass reinforce our conclusion that the intact tropical forest carbon sink has already peaked. This saturation and ongoing decline of the tropical forest carbon sink has consequences for policies intended to stabilize Earth’s climate.
Content may be subject to copyright.
See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/339927383
Asynchronous Carbon Sink Saturation in African and Amazonian Tropical
Forests
ArticleinNature · March 2020
CITATIONS
3
READS
604
106 authors, including:
Some of the authors of this publication are also working on these related projects:
Late twentieth-century patterns and trends in Amazon tree turnover View project
REDD+ Benefit Sharing View project
Wannes Hubau
University of Leeds
64 PUBLICATIONS581 CITATIONS
SEE PROFILE
Simon Lewis
University College London and University of Leeds.
255 PUBLICATIONS29,959 CITATIONS
SEE PROFILE
Oliver Lawrence Phillips
University of Leeds
511 PUBLICATIONS45,752 CITATIONS
SEE PROFILE
Hans Beeckman
Royal Museum for Central Africa
292 PUBLICATIONS4,459 CITATIONS
SEE PROFILE
All content following this page was uploaded by Alan Charles Hamilton on 02 April 2020.
The user has requested enhancement of the downloaded file.
Nature | www.nature.com | 1
Article
Asynchronous carbon sink saturation in
African and Amazonian tropical forests
Structurally intact tropical forests sequestered about half of the global terrestrial
carbon uptake over the 1990s and early 2000s, removing about 15 per cent of
anthropogenic carbon dioxide emissions1–3. Climate-driven vegetation models
typically predict that this tropical forest ‘carbon sink’ will continue for decades4,5.
Here we assess trends in the carbon sink using 244 structurally intact African tropical
forests spanning 11 countries, compare them with 321 published plots from Amazonia
and investigate the underlying drivers of the trends. The carbon sink in live
aboveground biomass in intact African tropical forests has been stable for the three
decades to 2015, at 0.66 tonnes of carbon per hectare per year (95 per cent condence
interval 0.53–0.79), in contrast to the long-term decline in Amazonian forests6.
Therefore the carbon sink responses of Earth’s two largest expanses of tropical forest
have diverged. The dierence is largely driven by carbon losses from tree mortality,
with no detectable multi-decadal trend in Africa and a long-term increase in
Amazonia. Both continents show increasing tree growth, consistent with the expected
net eect of rising atmospheric carbon dioxide and air temperature7–9. Despite the
past stability of the African carbon sink, our most intensively monitored plotssuggest
a post-2010 increase in carbon losses, delayed compared to Amazonia, indicating
asynchronous carbon sink saturation on the two continents. A statistical model
including carbon dioxide, temperature, drought and forest dynamics accounts for
the observed trends and indicates a long-term future decline in the African sink,
whereas the Amazonian sink continues to weaken rapidly. Overall, the uptake of
carbon into Earth’s intact tropical forests peaked in the 1990s. Given that the global
terrestrial carbon sink is increasing in size, independent observations indicating
greater recent carbon uptake into the Northern Hemisphere landmass10 reinforce our
conclusion that the intact tropical forest carbon sink has already peaked. This
saturation and ongoing decline of the tropical forest carbon sink has consequences
for policies intended to stabilize Earth’s climate.
Tropical forests account for approximately one-third of Earth’s ter-
restrial gross primary productivity and one-half of Earth’s carbon
stored in terrestrial vegetation
11
. Thus, small biome-wide changes in
tree growth and mortality can have global impacts, either buffering or
exacerbating the increase in atmospheric CO2. Models2,4,5,7,12, ground-
based observations13–15, airborne atmospheric CO2 measurements3,16,
inferences from remotely sensed data17 and synthetic approaches3,8,18
each suggest that, after accounting for land-use change, the remain-
ing structurally intact tropical forests (that is, those not affected by
direct anthropogenic impacts such as logging) are increasing in carbon
stocks. This structurally intact tropical forest carbon sink is estimated
at approximately1.2PgCyr−1 over 1990–2007 using scaled inventory
plot measurements
1
. Yet, despite its relevance to policy, changes in
this key carbon sink remain highly uncertain19,20.
Globally, the terrestrial carbon sink is increasing2,7,8,21. Between 1990
and 2017 the land surface sequestered about 30% of all anthropogenic
carbon dioxide emissions
1,21
. Rising CO
2
concentrations are thought to
have boosted photosynthesis more than rising air temperatures have
enhanced respiration, resulting in an increasing global terrestrial car-
bon sink2,4,7,8,21. Yet, for Amazonia, recent results from repeated censuses
of intact forest inventory plots show a progressive two-decade decline
in sink strength primarily due to an increase in carbon losses from tree
mortality
6
. It is unclear if this simply reflects region-specific droug
ht impacts
22,23
, or potentially chronic pan-tropical impacts of either
heat-related tree mortality24,25, or results frominternal forest dynam-
ics as past increases in carbon gains leave the system26. A more recent
deceleration of the rate of increase in carbon gains from tree growth is
also contributing to the declining Amazon sink
6
. Again, it is not known
whether this is a result of either pan-tropical saturation of CO2 ferti-
lization, or rising air temperatures, or is simply a regional drought
impact. To address these uncertainties, we (1) analyse an unprecedented
long-term inventory dataset from Africa, (2) pool the new African and
existing Amazonian records6 to investigate the putative environmental
drivers of changes in the tropical forest carbon sink, and (3) project its
likely future evolution.
We collected, compiled and analysed data from structurally intact
old-growth forests from the African Tropical Rainforest Observation
Network
27
(217 plots) and other sources (27 plots) spanning the period
1 January 1968 to 31 December 2014 (Extended Data Fig.1; Supplemen-
tary Table1). In each plot (mean size, 1.1 ha), all trees ≥100 mm in stem
https://doi.org/10.1038/s41586-020-2035-0
Received: 9 June 2019
Accepted: 19 December 2019
Published online: 4 March 2020
Check for updates
A list of authors and afiliations appears at the end of the paper.
2 | Nature | www.nature.com
Article
diameter were identified, mapped and measured at least twice using
standardized methods (135,625 trees monitored). Live biomass carbon
stocks were estimated for each census date, with carbon gains and
losses calculated for each interval (Extended Data Fig.2).
Continental carbon sink trends
We detect no long-term trend in the per unit area African tropical
forest carbon sink over three decades to 2015 (P = 0.167; Fig.1). The
aboveground live biomass sink averaged 0.66 tonnes of carbon per
hectare per year (0.66MgCha−1yr−1 with 95% confidence interval (CI)
of 0.53–0.79 and n=244) and was significantly greater than zero for
every year since 1990 (Fig.1; P < 0.001 for each time period in Table1).
Although very similar to past reports (0.63MgCha
−1
yr
−1
)
13
, this first
estimate of the temporal trend in Africa contrasts with the significantly
declining(P = 0.038) Amazonian trend6 (Fig.1). A linear mixed effects
model shows a significant difference in the slopes of the sink trends for
the two continents over the common time window (pooled data from
both continents, common time window, 1 January 1983 to mid-2011;
P=0.017). Therefore, the per unit area sink strength of the two largest
expanses of tropical forest on Earth diverged in the 1990s and 2000s.
The proximal cause of the divergent sink patterns is a significant
increase (P = 0.002) in carbon losses (from tree mortality, that is, the
loss of carbon from the live biomass pool) in Amazonian forests, with
no detectable trend over three decades in African forests (P = 0.403;
Fig.1; Table1). A linear mixed effects model using pooled data shows
a significant difference in slopes of carbon losses between the two
continents over the common time window (P=0.027; 1 January 1983
to mid-2011). Long-term trends in carbon gains (from tree growth and
newly recruited trees) show significant increases on both continents
(P = 0.037 forAfrica; P < 0.001 forAmazon;Fig.1), and we could detect
no difference in slopes between the continents (P=0.348; carbon gains
from tree growth alone also show no continental difference in long-term
trends, P=0.322). However, an assessment of how underlying environ
-
mental drivers affect carbon gains and losses is needed to understand
the ultimate causes of the divergent sink patterns.
Understanding the carbon sink trends
We first investigate those environmental drivers exhibiting long-term
change that affect photosynthesis and respiration in theory-driven
models: atmospheric CO
2
concentration, surface air temperature and
water availability. Bivariate models (Fig. 2) and a linear mixed effects
model of carbon gains(Extended Data Table 1), with censuses nested
within plots, and pooling the new African and published Amazonian
data, show a significant positive relationship with CO
2
(P = 0.021 in
Fig. 2;P = 0.001in Extended Data Table 1), and significant negative
relationships with mean annual temperature (MAT; P < 0.001in
Fig. 2 and Extended Data Table 1) and drought (P = 0.003 in Fig. 2;
P < 0.001 in Extended Data Table 1), with droughtmeasured as the
maximum climatological water deficit(MCWD)14. These results are
consistent with a positive CO
2
fertilization effect, and negative effects
–0.5
0.0
0.5
1.0
Net carbon sink
(Mg C ha−1 yr−1)
Net carbon sink Africa slope 0.015 P = 0.167
Amazon slope –0.016 P = 0.038
1.5
2.0
2.5
3.0
Carbon gains
Carbon gains Africa slope 0.008 P = 0.037
Amazon slope 0.014 P < 0.001
1.5
2.0
2.5
3.0
Carbon losses
Carbon losses Africa slope −0.008 P = 0.403
Amazon slope 0.023 P = 0.002
1985 1990 1995 2000 2005 2010
2015
Year
a
b
(Mg C ha−1 yr−1)
c
(Mg C ha−1 yr−1)
Fig. 1 | Long-t erm carbon dy namics of str ucturally in tact old-
growthtropic al forests in A frica and Am azonia. ac, Trends in net
abovegroun d live biomass car bon (a), carbon gains to th e system from woo d
producti on (b), and carbon los ses from the sys tem from tree mor tality (c),
measured i n 244 Afric an inventory plot s (blue lines) and con trasting
published6 Amazo nian inventory dat a (brown lines; 32 1 plots). For Africa we
show complet e years with at le ast 25 plots mo nitored; for Amazo nia we show
the published record6.Shading cor responds to the 9 5% CI, with darker sha ding
indicatin g a greater numb er of plots monit ored in that year (the lig htest
shading indi cates the mini mum 25 plots mo nitored). The CI for th e Amazonian
datase t is omitted for clar ity, but can be see n in Fig.3. Slopes an d
P values are fro m linear mixed ef fects model s (seeMethods).
Nature | www.nature.com | 3
of higher temperatures and drought on tree growth, consistent with
temperature-dependent increases in autotrophic respiration, and tem-
perature- and drought-dependent reductions in carbon assimilation. By
contrast, the equivalent models for carbon losses show no significant
relationships with CO2 (P = 0.363 in Fig. 2;P = 0.344in Extended Data
Table1), MAT (P = 0.789 in Fig. 2;P = 0.804in Extended Data Table 1) or
MCWD (P =0.338 in Fig.2; P = 0.325 inExtended Data Table1).
We further investigate the responses of carbon gains and losses (for
which the above analysis has no explanatory power) by expanding
our potential explanatory variables to include five more. These are
the changes in environmental conditions (CO
2
-change, MAT-change,
MCWD-change, see Extended Data Fig.3 for calculation details) and two
attributes of forests that may influence their response to the same envi-
ronmental changes: the plot mean wood density (which in old-growth
forests correlates with belowground resource availability
28,29
) and the
plot carbon residence time (CRT, which measures how long fixed carbon
remains in the system and hence reflects when past increases in carbon
gains leave the system as elevated carbon losses30).
The minimum adequate carbon gain model using our expanded
explanatory variables (best-ranked model using multimodel inference)
has a significant positive relationship with CO
2
-change (P = 0.013), and
significant negative relationships with MAT(P = 0.001), MAT-change
(P < 0.001), MCWD (P < 0.001) and wood density (P = 0.015; Table2;
model-average results are similar, seeMethods and Supplementary
Tables2–4). The retention of both MAT and MAT-change suggests that
higher temperatures correspond to lower tree growth, and that trees
only partially acclimate to recently rising temperatures, which further
reduces growth, consistent with warming experiments
31
and observa-
tions9. The inclusion of higher wood density and its relationship to lower
carbon gains (Extended Data Fig.4), alongside no temporal trends in
wood density (Extended Data Fig.5), suggests that old-growth forests
with denser-wooded tree communities typically have fewer available
below ground resources, or such patterns may also emerge from dis-
turbance regimes lacking large-scale exogenous events, consistent
with previous studies26,28,32.
The minimum adequate carbon gain model using our expanded
explanatory variables also highlights continental differences. Between1
January 2000 and 31 December 2014 modelled African forest carbon
gains increased by 3.1% compared with a 0.1% decline in Amazonia over
the same interval (Table2). In Africa, from 2000 to 2015, the increase
in carbon gains was composed of a 3.7% increase from CO
2
-change,
partially offset by increasing droughts depleting gains by 0.5%, and only
a slight decline in gains of 0.1% resulting from temperature increases
(Table2), because the rate of temperature change (MAT-change) decel-
erated over this time window (Extended Data Fig.5). For Amazonia,
the same 3.7% increase in carbon gains due to CO
2
-change was seen.
Opposing this trend was increasing droughts—and the greater sensitiv-
ity to drought of Amazonian forests—which reduced carbon gains by
2.7% (five times the impact in Africa), and temperature increases at the
same rate as in the past (that is, MAT-change is zero) further reduced
gains by 1.1% (ten times the impact in Africa), leaving a net change in
gains slightly below zero (Table2). Therefore, the stalling of carbon
gain increases in Amazonia in the decade to mid-20116 is a response to
drought and temperature and not due to an unexpected saturation
of CO2 fertilization.
Overall, the larger modelled increase in carbon gains in Africa rela-
tive to Amazonia appear to be driven by slower warming, fewer or less
extreme droughts, lower forest sensitivity to droughts, and overall
lower temperatures (African forests are on average ~1.1 °C cooler than
Amazonian forests, because they typically grow at higher elevations of
~200 metres above sea level). Other continental differences may also be
influencing the results, including higher nitrogen deposition in African
tropical forests due to the seasonal burning of nearby savannahs33 and
biogeographical history resulting in differing contemporary species
pools and resulting functional attributes34,35.
The minimum adequate carbon loss model using our expanded
explanatory variables shows significantly higher losses with CO2-change
(P = 0.026) and MAT-change (P < 0.001) and significantly lower losses
with MCWD (P = 0.030) and CRT (P < 0.001; Table2). Thus, changes in
carbon losses appear to be largely a function of past carbon gains. First,
the greater losses in forests with shorter CRT conform to a ‘high-gain,
high-loss’ forest dynamics pattern
26,28
. Second, wetter plots have a
longer growing season and thus they have higher gains and correspond-
ingly higher losses, explaining the negative relationship with MCWD.
Third, as increasing CO2 levels result in additional carbon gains, after
some time these additional past gains leave the system, resulting in
Table 1 | Carbon sink in structurallyintactold-growth tropical forests in Africa, Amazonia and the pan-tropics, 1980–2040
Period Number of plots Per unit area aboveground live biomass C sink (MgCha−1yr−1)Total C sink (Pg C yr−1)a
Africa Amazon Africa Amazon Pan-tropicsbAfrica Amazon Pan-tropicsb
1980–1990 45 73 0.33 (0.06–0.63) 0.35 (0.06–0.59) 0.35 (0.07– 0.62) 0.28 (0.05–0.53) 0.49 (0.08–0.82) 0.87 (0.16–1.52)
1990–2000 96 172 0.67 (0.43–0.89) 0.53 (0.42–0.65) 0.57 (0.39–0.74) 0.50 (0.32–0.66) 0.68 (0.54–0.83) 1.26 (0.88–1.63)
2000–2010 194 291 0.70 (0.55–0.84) 0.38 (0.26–0.48) 0.50 (0.35–0.64) 0.46 (0.37–0.56) 0.45 (0.31–0.57) 0.99 (0.70–1.25)
2010–2015c184 172 0.66 (0.40–0.91) 0.24 (0.00–0.47) 0.40 (0.15–0.65) 0.40 (0.24–0.56) 0.27 (0.00–0.52) 0.73 (0.25–1.18)
20102020d– – 0.63 (0.36 0.89) 0.23 (0.0 50.50) 0.38 (0.110.65) 0.37 (0.210.53) 0.25 (0.050.54) 0.68 (0.171.16)
20202030d– – 0.59 (0. 240.93) 0.12 (0.290.51) 0.30 (0.080.67) 0.31 (0.130.49) 0.12 (0.290.52) 0.47 (0.151.07 )
20302040d– – 0.55 ( 0.080.99) 0.00 (0.540.49) 0.21 (0.290.67) 0.26 ( 0.040.47) 0.00 (0.500.46) 0.29 (0.460.97)
This table covers 1 January 1980 to 31 December 2014and predictions to 31 December 2039. Mean values are in boldface, future predictions in italics, uncertainties in parentheses: 95% boot-
strapped conidence intervals for 1980–2015, and 2σ for the predictions (2010–2040).
aThe total continental C sink is the per unit area aboveground C sink multiplied by intact forest area (from ref. 1; see Extended Data Table2) and includes continent-speciic estimates of trees with
adiameter at breast height of<100 mm, lianas and roots (seeMethods).
bThe per unit area pan-tropical aboveground live biomass C sink is the area-weighted mean of African, Amazonian and Southeast Asian sink values. Southeast Asian values were from published
per unit area carbon sink data15 (n=49 plots) for 1990–2015, with 1980–1990 assumed to be the same as 1990–2000 owing to very low sample sizes. The pan-tropical total C sink is the sum of
African, Amazonian and Southeast Asian total continental carbon sink values. The continental sink in Southeast Asia is a modest and declining contribution to the pan-tropical sink, owing to the
very small area of intact forest remaining, at 0.11PgCyr−1, 0.08PgCyr−1, 0.07PgCyr−1 and 0.06PgCyr−1 in the 1980s, 1990s, 2000s and 2010s, respectively; hence uncertainty in the Southeast
Asian sink cannot reverse the pan-tropical declining sink trend.
cThe Amazonian sink in the 2010–2015 time window was calculated from 172 plots measured between 1 January 2010 and mid-2011. The lack of temporal coverage later in this period has little
impact on the results; adding modelled results for 1 January 2012 to 31 December 2014 gives a per unit area aboveground sink of 0.25 Mg C ha−1 yr−1 (0.00–0.49), which would increase the
pan-tropical total C sink by 0.01 Pg C yr−1.
dPer unit area total C sink for 2010–2020, 2020–2030 and 2030–2040 was predicted using parameters from Table2, except for the 2010–2020 sink in Africa, which is the mean of the measured
sink from 2010–2015 and the modelled sink from 2015–2020. For the Asian sink we assumed the same parameters as for Africa, because Asian forest median CRT is 61 years, close to the African
median of 63 years.
4 | Nature | www.nature.com
Article
greater carbon losses, which explains the positive relationship with
CO
2
-change. Finally, in addition to these relationships with carbon
gains, the inclusion of MAT-change (P<0.001) indicates tree mortality
induced by heat or by increased vapour pressure deficit24. Overall, our
results imply that chronic long-term environmental change factors,
temperature and CO
2
, rather than simply the direct effects of drought,
underlie longer-term trends in tropical forest tree mortality, although
other changes such as rising liana infestation rates seen in Amazonia
36,37
cannot be excluded.
The minimum adequate carbon loss model using our expanded
explanatory variables replicates the continental trends (Fig.3). The
overall modelledlower loss rates in Africa reflect their longer CRT (69
years, 95% CI, 66–72), compared with Amazonian forests (56years, 95%
CI, 54–59) while over the 2000–2015 window the much smaller mod-
elledincrease in loss rates in Africa compared to Amazonia results from
a slower increase in warming and a stable CRT in Africa compared to con-
tinued warming at previous rates and a shortening CRT in Amazonian
forests (Table2;Extended Data Fig.5). Furthermore, given that losses
appear to lag behind gains, they should relate to the long-term CRT of
plots. This is what we find: the longer the CRT the smaller the increase
in carbon losses, with no increase in losses for plots with CRT≥77 years
(Extended Data Fig.6). Consequently, owing to the typically longer
CRT of African forests, increasing losses in Africa ought to appear
10–15 years after the increase in Amazon losses began (around 1995).
Strikingly, in Africa the most intensely monitored plots suggest that
losses began increasing from about 2010 (Extended Data Fig.7), and
plots with shorter CRT are driving the increase (Extended Data Fig.8).
Thus, a mortality-dominated decline of the African carbon sink appears
to have begun very recently.
Future of the tropical forest carbon sink
Our carbon gain and loss models (Table2) can be used to make a tentative
estimate of the future size of the per unit area intact forest carbon sink
(Fig.3). Extrapolations of the changes in the predictor variables from
1983–2015 forward to 31 December 2039 (Extended Data Fig.5) show
declines in the sink on both continents (Fig.3). By 2030 the carbon sink in
aboveground live biomass in intact African tropical forest is predicted to
decline by 14% from the measured 2010–15 mean to 0.57MgCha
−1
yr
−1
(2σ
range, 0.16–0.96; Fig.3). The Amazon sink continues to rapidlydecline,
reaching zero in 2035 (2σ range, 2011–2089; Fig.3). Our estimated sink
strength on both continents in the 2020s and 2030s is sensitive to future
CO
2
emissions pathways (CO
2
-change)
38
, resulting temperature increase
(MAT, MAT-change) and hydrological changes (MCWD), plus changes
in forest dynamics (CRT), but the sink is always lower than levels seen
in the 2000s (seeMethods and Supplementary Table5). Therefore, the
carbon sink strength of the world’s two most extensive tropical forests
have now saturated, albeit asynchronously.
Slope 0.003
a
Carbon gains (Mg C ha−1 yr−1)
6
Slope 0.003
d
Carbon losses (Mg C ha−1 yr−1)
CO2 (ppm)
340
Slope –0.111 P < 0.001
b
Slope 0.01
e
MAT (°C)
22
Slope –0.00058 P = 0.003
c
Drier
Slope –0.00041
f
MCWD (mm)
0
Drier
350 360 370 380 390 24 26 28 100 200 300 400 500
CO2 (ppm)
340
MAT (°C)
22
MCWD (mm)
0350 360 370 380 390 24 26 28 100 200 300 400 500
5
4
3
2
1
P = 0.021
6
5
4
3
2
1
0
7
6
5
4
3
2
1
6
5
4
3
2
1
0
7
6
5
4
3
2
1
6
5
4
3
2
1
0
7
P = 0.363 P = 0.789 P = 0.338
Fig. 2 | Poten tial environm ental driver s of carbon gai ns and losse s in
structurally intact old-growth tropical forestsin Africa and Amazonia.
Abovegroun d carbon gains , from woody produ ction (ac), and abovegroun d
carbon lo sses, from tre e mortalit y (df), are presented as tim e-weighted me an
values for eac h plot, that is, eac h census withi n a plot is weighted by i ts length,
against th e correspond ing values of atmo spheric carb on dioxide
concentration (CO2), temperature (MAT) and drou ght (MCWD), for Africa n
(blue) and Ama zonian (brown) invento ry plots. For v isual clarity e ach data
point there fore represent s an inventory plot, a nd the shading rep resents the
total monitoring length, with empty circles corre sponding to plots monitored
for ≤5 years and s olid circles for plot s monitored for >20 years . Solid lines show
significant trends and dashed lines show non-significant trends calculated
using linea r mixed effect s models wit h census inter vals (n=1,566) neste d
within plots (n=565), using an empir ically derived w eighting bas ed on interval
length an d plot area, on the un transformed p ooled Afric a and Amazon dat aset
(seeMethods). Slop es and P values are fr om the same linea r mixed effect s
models. C arbon loss dat a and models are pre sented untr ansformed for
comparis on with carbo n gains, but tran sformation is ne eded to fit n ormality
assumpti ons; performin glinear mixed effe cts models o n transformed c arbon
loss data d oes not change th e presented si gnifica nce trends, nor do es
including all t hree paramete rs and transform ed data in a model (s ee Extended
Data Table1).
Nature | www.nature.com | 5
Scaling results to the pan-tropics
Scaling our estimated mean sink strength by forest area for each
continent signifies that Earth has passed the point of peak carbon
sequestration into intact tropical forests (Table1). The continental
sink in Amazonia peaked in the 1990s, followed by a decline, driven
by sink strength peaking in the 1990s and a continued decline in for-
est area (Table1). In Africa the per unit area sink strength peaked later,
in the 2000–2010 period, but the continental African sink peaked in
the 1990s, owing to the decline in forest area in the 2000s outpacing
the small per unit area increase in sink strength. Including the modest
uptake in the much smaller area of intact Asian tropical forest15 indicates
that total pan-tropical carbon uptake peaked in the 1990s (Table1).
From the peak pan-tropical intact forest uptake of 1.26PgCyr−1 in the
1990s, we project a continued decline reaching just 0.29Pg C yr
−1
in the
2030s (multi-decade decline of ~0.24PgCyr−1 per decade), driven by
(1) reduced mean pan-tropical sink strength decline of 0.1MgCha
−1
yr
−1
per decade and (2) ongoing forest area losses of ~13.5 million ha yr
−1
(see
Extended Data Table2 for forest area details). Critically, climate-driven
vegetation model simulations have not predicted that the peak net
carbon uptake into intact tropical forests has already been passed2,4,5.
Discussion
Our method of scaling to arrive at a pan-tropical sink estimate—in
common with other studies using similar datasets1,6,13—is limited. Yet,
pervasive net carbon uptake is expected given that we find a strong and
ongoing CO2 fertilization effect. Using our CO2 response in Table2, we
find an increase in aboveground carbon stocks of 10.8±3.7MgCha
−1
per
100ppm CO2, equivalent to 6.5±2.2% (±standard error; using an area-
weighted pan-tropical mean aboveground C stock of 165MgCha
−1
).
This is comparable to the 5.0±1.2% increase in tropical forest C stocks
per 100ppm CO
2
derived from a recent synthesis of CO
2
fertilization
experiments, despite a lack of data from old-growthtropical forests
39
.
Our result is within the range of climate-driven vegetation models
2,7
,
although it is greater than results from a number of recently pub-
lished models that include potential nutrient constraints, reported
as 5.9±4.7MgCha−1 per 100ppm CO2 (ref. 40). We find that the CO2
fertilization-driven uptake is currently only partially offset by the
negative impacts of similarly widespread rising air temperatures
(−2.0±0.4MgCha−1°C−1, from Table2), consistent with models7, limited
experiments
31
and independent observations
9
, plus negative responses
to drought41,42. Long-term and extensive increases in satellite-derived
greenness in tropical regions that have not experiencedmajor changes
in land-use management
17,43
, particularly in central Africa in the past
decade
44
, indicate increases in tropical forest net primary productivity,
providing further evidence that the sink is a widespread phenomenon.
Nonetheless, our analyses suggest that this pervasive intacttropical
forest sink in live biomass is in long-term decline, having peaked first in
Amazonia, and more recently followed by African forests, explaining
the prior Africa–Amazon carbon sink divergence as part of a longer-
term pattern of asynchronous saturation and decline. Over time, the
continued CO
2
fertilization effect is being increasingly counteracted
by the impacts ofhigher temperatures and droughts on tree growth
and mortality, which aremodulated by internal forest dynamics, with
forests with the shortest CRTsaturating first. From an atmospheric
perspective, the full impacts of the contribution to the saturation of
the sink from slowing carbon gains are experienced immediately, but
the contribution from rising carbon losses is delayed because dead
trees do not decompose instantaneously. Decomposition of this dead
tree mass is about half complete in 4 years, and about 85% complete
in 10 years, so rising carbon losses result in delayed carbon additions
to the atmosphere45. Hence, from an atmospheric perspective, the
intact tropical forest biomass carbon sink probably peaked a few years
later than ourinventorydata indicate and the full impacts are not yet
realized. The pan-tropical carbon sink in live biomass declined by
0.27PgCyr−1 between the 1990s and 2000s (Table1), but accounting
for dead wood decomposition
45
shows a smaller 0.17PgCyr
−1
reduction
from an atmospheric perspective (seeMethods).
Given that the overall global terrestrial carbon sink is increasing, a
weakening intact tropical forest sink implies that the extra-tropical
carbon sink has increased over the past two decades. Independent
observations of interhemispheric atmospheric CO
2
concentration
indicates that carbon uptake into the Northern Hemisphere landmass
has increased at a greater rate than the global terrestrial carbon sink
Table 2 | Minimum adequate models to predict carbon gains and losses in African and Amazonian forests
Carbon gains (MgCha−1yr−1)
Predictor variable Parameter value Standard error t value P value 2000–2015 change in gains (%)a
Intercept 5.255 | 5.395 0.603 | 0.614 8.7 | 8.8 <0.001
CO2-change (ppm yr−1)b0.238 0.096 2.5 0.013 3.69% | 3.71%
MAT (°C) −0.083 0.025 3.3 0.001 −0.67% | −1.07%
MAT-change (°C yr−1)c−1.243 0.233 −5.3 <0.001 0.58% | 0.00%d
MCWD (mm×1,000) −0.405 | −1.391 0.381 | 0.24 −1.1 | −5.8 0.289 | <0.001 −0.52% | −2.73%
Wood density (g cm−3)−1.295 0.530 −2.4 0.015 0.05% | 0.00%
Carbon losses (MgCha−1yr−1)e
Predictor variable Parameter value Standard error t value P value 2000–2015 change in losses (%)a
Intercept 1.216 0.086 14.1 <0.001
CO2-change (ppm yr−1)b0.130 0.059 2.2 0.026 11.38% | 14.81%
MAT-change (°C yr−1)c0.766 0.162 4.7 <0.001 −1.56% | 0.00%
MCWD (mm×1,000) −0.232 0.107 −2.2 0.030 −1.21% | −2.42%
CRT (years) −0.003 0.001 −6.1 <0.001 −0.57% | 1.39%
This table shows the best-ranked gains and loss models. Where continental values differ, those for Africa are reported irst, followed by ‘|’, then the Amazonian values.
aThe 1 January 2000 to 31 December 2014 change in gains/losses for each predictor variable was estimated allowing only the focal predictor to vary; this change was then expressed as a
percentage of the annual gains/losses in the year 2000, allowing all predictors to vary.
bChange over the past 56 years(see Extended Data Fig. 3).
cChange over the past 5 years(see Extended Data Fig. 3).
dA positive value for Africa indicates that MAT increased more slowly over 2000–2015 compared to the mean increase over 1983–2015, therefore contributing to an increase in gains; a zero
value for Amazonia indicates that the rate of MAT increase was the same over 2000–2015 as the mean increase over 1983–2015.
eCarbon loss values were normalized via power-law transformation, with power parameter λ=0.361.
6 | Nature | www.nature.com
Article
since the 1990s, with a further disproportionate increase in the 2000s
10
.
The interhemispheric analysis suggests a weakening of the tropical
forest sink by ~0.2PgCyr−1 between the 1990s and 2000s10, which is
similar to the 0.17PgCyr−1 weakening over the same time period that
we find. This reinforces our conclusion that the intact tropical forest
carbon sink has already saturated.
In summary, our results indicate that although intact tropical forests
remain major stores of carbon and are key centres of biodiversity11,
their ability to sequester additional carbon in trees is waning. In the
1990s intact forests removed 17% of anthropogenic CO
2
emissions.
This declined to an estimated 6% in the 2010s, because the pan-tropical
weighted average per unit area sink strength declined by 33%, forest
area decreased by 19% and anthropogenic CO2 emissions increased by
46%. Although tropical forests are more immediately threatened by
deforestation46 and degradation47, and the future carbon balance will
also depend on secondary forest dynamics
48
and forest restoration
plans49, our analyses show that they are also affected by atmospheric
chemistry and climatic changes. Given that the intact tropical forest
carbon sink is set to end sooner than even the most pessimistic climate-
driven vegetation models predict4,5, our analyses suggest that climate
change impacts in the tropics may become more severe than predicted.
Furthermore, the carbon balance of intact tropical forests will only
stabilize once CO2 concentrations and the climate stabilizes.
Continued on-the-ground monitoring of the world’s remaining intact
tropical forests will be required to test our prediction that the carbon
sinkin live trees will continue to decline, particularly as future changes
in thetree species composition may alter the resilience of the sink and
becausewe cannot exclude the possibility of decadal-scale climate
impacts on these forests. Such direct ground-based measurements
also provide a constraint on estimating the size, location and climate
sensitivity of the terrestrial carbon sink. In addition, our conclusion
that tree mortality and internal forest dynamics are important controls
on the future of the tropical forest carbon sink may assist in improving
the vegetation components of Earth System Models
50
and contribute to
reducing terrestrial carbon cycle feedback uncertainty
19,20
. Our findings
also have policy implications. At the individual country level, given that
intact tropical forests are a carbon sink but the rate of reduction will
differ continentally and probably regionally (for example, aseasonal
Amazon forests are less affected by droughts), national greenhouse gas
reporting will require careful forest monitoring. At the international
level, given that tropical forests are likely to sequester less carbon in
the future than Earth System Models predict, an earlier date by which to
reach net zero anthropogenic greenhouse gas emissions will be required
to meet any given commitment to limit the global heating of Earth.
Online content
Any methods, additional references, Nature Research reporting sum-
maries, source data, extended data, supplementary information,
acknowledgements, peer review information; details of author con-
tributions and competing interests; and statements of data and code
availability are available at https://doi.org/10.1038/s41586-020-2035-0.
1. Pan, Y. etal. A large and persistent carbon sink in the world’s forests. Science 333,
988–993 (2011).
2. Sitch, S. etal. Recent trends and drivers of regional sources and sinks of carbon dioxide.
Biogeosciences 12, 653–679 (2015).
1.5
Amazond
e
f
1.5
Net carbon sink
(Mg C ha−1 yr−1)
Slope –0.003 Slope –0.004 P < 0.001
Africaa
3.0
Carbon gains
b
3.0
Carbon losses
1990 2020
Year
c
–0.5
0.0
0.5
1.0
P = 0.547
Slope 0.007 Slope 0.003 P < 0.001P = 0.003
Slope 0.01 Slope 0.008 P < 0.001P = 0.006
(Mg C ha−1 yr−1)
1.5
2.0
2.5
(Mg C ha−1 yr−1)
1.5
2.0
2.5
2000 2010 2030 2040
Slope –0.018 Slope –0.013 P < 0.001P = 0.036
–0.5
0.0
0.5
1.0
Slope 0.001 Slope –0.001 P < 0.001P = 0.645
3.0
1.5
2.0
2.5
3.0
1.5
2.0
2.5
1990 2020
Year
2000 2010 2030
2040
Slope 0.019 Slope 0.012 P < 0.001P = 0.001
Fig. 3 | Mode lled past and f uture carbo n dynamics of s tructurall y intact old-
growthtropic al forests in A frica and Am azonia. af, Predic tions of net
abovegroun d live biomass car bon (a, d), carbon gains (b, e), and carbo n losses
(c, f), for African (lef t panels) and Amazo nian (right panels) plot i nventory
network s, based on CO2-ch ange, MAT, MAT-change, droug ht (MCWD), plot
wood densi ty, and plot CRT, using obser vations in Afr ica until 31 Dec ember
2014 and Amazo nia until mid-2011, and ex trapolation s of prior trends to
31 Decemb er 2039. Model predic tions are in blue (Af rica) and brown (A mazon),
with solid li nes spanning th e window when ≥75% of plo ts were monitore d to
show model co nsistency w ith the obser ved trends, and sh ading showing up per
and lower confidence intervals accounting for uncertainties in the model (both
fixed an d random effec ts) and uncert ainties in the pre dictor varia bles. Light-
grey lines an d grey shading are th e mean and 95% CI of the o bservatio ns from
the Afric an and Amazoni an plot network s.
Nature | www.nature.com | 7
3. Gaubert, B. etal. Global atmospheric CO2 inverse models converging on neutral tropical
land exchange, but disagreeing on fossil fuel and atmospheric growth rate.
Biogeosciences 16, 117–134 (2019).
4. Huntingford, C. etal. Simulated resilience of tropical rainforests to CO2-induced climate
change. Nat. Geosci. 6, 268–273 (2013).
5. Mercado, L. M. etal. Large sensitivity in land carbon storage due to geographical and
temporal variation in the thermal response of photosynthetic capacity. New Phytol. 218,
1462–1477 (2018).
6. Brienen, R. J. W. etal. Long-term decline of the Amazon carbon sink. Nature 519, 344–348
(2015).
7. Piao, S. etal. Evaluation of terrestrial carbon cycle models for their response to climate
variability and to CO2 trends. Glob. Change Biol. 19, 2117–2132 (2013).
8. Schimel, D., Stephens, B. B. & Fisher, J. B. Effect of increasing CO2 on the terrestrial
carbon cycle. Proc. Natl Acad. Sci. USA 112, 436–441 (2015).
9. Anderegg, W. R. L. etal. Tropical nighttime warming as a dominant driver of variability in
the terrestrial carbon sink. Proc. Natl Acad. Sci. USA 112, 15591–15596 (2015).
10. Ciais, P. etal. Five decades of northern land carbon uptake revealed by the
interhemispheric CO2 gradient. Nature 568, 221–225 (2019).
11. Lewis, S. L., Edwards, D. P. & Galbraith, D. Increasing human dominance of tropical
forests. Science 349, 827–832 (2015).
12. Pugh, T. A. M. etal. Role of forest regrowth in global carbon sink dynamics. Proc. Natl
Acad. Sci. USA 116, 4382–4387 (2019).
13. Lewis, S. L. etal. Increasing carbon storage in intact African tropical forests. Nature 457,
1003–1006 (2009).
14. Phillips, O. L. etal. Drought sensitivity of the Amazon rainforest. Science 323, 1344–1347
(2009).
15. Qie, L. etal. Long-term carbon sink in Borneo’s forests halted by drought and vulnerable
to edge effects. Nat. Commun. 8, 1966 (2017).
16. Gatti, L. V. etal. Drought sensitivity of Amazonian carbon balance revealed by
atmospheric measurements. Nature 506, 76–80 (2014).
17. Nemani, R. R. etal. Climate-driven increases in global terrestrial net primary production
from 1982 to 1999. Science 300, 1560–1563 (2003).
18. Keenan, T. F. etal. Recent pause in the growth rate of atmospheric CO2 due to enhanced
terrestrial carbon uptake. Nat. Commun. 7, 13428 (2016).
19. Booth, B. B. B. etal. High sensitivity of future global warming to land carbon cycle
processes. Environ. Res. Lett. 7, 024002 (2012).
20. Lombardozzi, D. L., Bonan, G. B., Smith, N. G., Dukes, J. S. & Fisher, R. A. Temperature
acclimation of photosynthesis and respiration: a key uncertainty in the carbon cycle-
climate feedback. Geophys. Res. Lett. 42, 8624–8631 (2015).
21. Le Quéré, C. etal. Global carbon budget 2018. Earth Syst. Sci. Data 10, 2141–2194
(2018).
22. Lewis, S. L., Brando, P. M., Phillips, O. L., vander Heijden, G. M. F. & Nepstad, D. The 2010
Amazon drought. Science 331, 554 (2011).
23. Feldpausch, T. R. etal. Amazon forest response to repeated droughts. Glob. Biogeochem.
Cycles 30, 964–982 (2016).
24. McDowell, N. etal. Drivers and mechanisms of tree mortality in moist tropical forests.
New Phytol. 219, 851–869 (2018).
25. Aleixo, I. etal. Amazonian rainforest tree mortality driven by climate and functional traits.
Nat. Clim. Chang. 9, 384–388 (2019).
26. Lewis, S. L. etal. Concerted changes in tropical forest structure and dynamics: evidence
from 50 South American long-term plots. Phil. Trans. R. Soc. Lond. B 359, 421–436
(2004).
27. Lewis, S. L. etal. Above-ground biomass and structure of 260 African tropical forests.
Phil. Trans. R. Soc. Lond. B 368, 20120295 (2013).
28. Quesada, C. A. etal. Basin-wide variations in Amazon forest structure and function are
mediated by both soils and climate. Biogeosciences 9, 2203–2246 (2012).
29. Malhi, Y. etal. The above-ground coarse wood productivity of 104 neotropical forest
plots. Glob. Change Biol. 10, 563–591 (2004).
30. Galbraith, D. etal. Residence times of woody biomass in tropical forests. Plant Ecol.
Divers. 6, 139–157 (2013).
31. Reich, P. B. etal. Boreal and temperate trees show strong acclimation of respiration to
warming. Nature 531, 633–636 (2016).
32. ter Steege, H. etal. Continental-scale patterns of canopy tree composition and function
across Amazonia. Nature 443, 444–447 (2006).
33. Bauters, M. etal. High ire-derived nitrogen deposition on central African forests.
Proc. Natl Acad. Sci. USA 115, 549–554 (2018).
34. Parmentier, I. etal. The odd man out? Might climate explain the lower tree alpha-
diversity of African rain forests relative to Amazonian rain forests? J. Ecol. 95,
1058–1071 (2007).
35. Slik, J. W. F. etal. Phylogenetic classiication of the world’s tropical forests. Proc. Natl
Acad. Sci. USA 115, 1837–1842 (2018).
36. Phillips, O. L. etal. Increasing dominance of large lianas in Amazonian forests. Nature 418,
770–774 (2002).
37. Schnitzer, S. A. & Bongers, F. Increasing liana abundance and biomass in tropical forests:
emerging patterns and putative mechanisms. Ecol. Lett. 14, 397–406 (2011).
38. Meinshausen, M. etal. The RCP greenhouse gas concentrations and their extensions from
1765 to 2300. Clim. Change 109, 213–241 (2011).
39. Terrer, C. etal. Nitrogen and phosphorus constrain the CO2 fertilization of global plant
biomass. Nat. Clim. Chang. 9, 684–689 (2019).
40. Fleischer, K. etal. Amazon forest response to CO2 fertilization dependent on plant
phosphorus acquisition. Nat. Geosci. 12, 736–741 (2019).
41. Jiang, Y. etal. Widespread increase of boreal summer dry season length over the Congo
rainforest. Nat. Clim. Chang. 9, 617–622 (2019).
42. Gloor, M. etal. Recent Amazon climate as background for possible ongoing and future
changes of Amazon humid forests. Glob. Biogeochem. Cycles 29, 1384–1399 (2015).
43. Kolby Smith, W. etal. Large divergence of satellite and Earth system model estimates of
global terrestrial CO2 fertilization. Nat. Clim. Chang. 6, 306–310 (2016).
44. Chen, C. etal. China and India lead in greening of the world through land-use
management. Nature Sustain. 2, 122–129 (2019).
45. Chambers, J. Q., Higuchi, N., Schimel, J. P., Ferreira, L. V. & Melack, J. M. Decomposition
and carbon cycling of dead trees in tropical forests of the central Amazon. Oecologia 122,
380–388 (2000).
46. Hansen, M. C. etal. High-resolution global maps of 21st-century forest cover change.
Science 342, 850–853 (2013).
47. Pearson, T. R. H., Brown, S., Murray, L. & Sidman, G. Greenhouse gas emissions from
tropical forest degradation: an underestimated source. Carbon Balance Manag. 12, 3
(2017).
48. Schwartz, N. B., Uriarte, M., DeFries, R., Gutierrez-Velez, V. H. & Pinedo-Vasquez, M. A.
Land-use dynamics inluence estimates of carbon sequestration potential in tropical
second-growth forest. Environ. Res. Lett. 12, 074023 (2017).
49. Lewis, S. L., Wheeler, C. E., Mitchard, E. T. A. & Koch, A. Regenerate natural forests to store
carbon. Nature 568, 25–28 (2019).
50. Yu, K. etal. Pervasive decreases in living vegetation carbon turnover time across forest
climate zones. Proc. Natl Acad. Sci. USA116, 24662–24667 (2019).
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional afiliations.
© The Author(s), under exclusive licence to Springer Nature Limited 2020
Wannes Hubau1,2,3,95 ✉, Simon L. Lewis1,4,95, Oliver L. Phillips1, Koi Affum-Baffoe5,
Hans Beeckman2, Aida Cuní-Sanchez4,6, Armandu K. Daniels7, Corneille E. N. Ewango8,9,1 0,
Sophie Fauset11, Jacques M. Mukinzi8,12,1 3, Douglas Sheil14, Bonaventure Sonké15,
Martin J. P. Sullivan1,16, Terry C. H. Sunderland17,18, Hermann Taedoumg15,19 , Sean C. Thomas20,
Lee J. T. White21,22,23, Katharine A. Abernethy22,23, Stephen Adu-Bredu24,
Christian A. Amani17,25, Timothy R. Baker1, Lindsay F. Banin26, Fidèle Baya27,28,
Serge K. Begne1,15, Amy C. Bennett1, Fabrice Benedet29,30, Robert Bitariho31,
Yannick E. Bocko32, Pascal Boeckx33, Patrick Boundja17,34 , Roel J. W. Brienen1, Terry Brncic34,
Eric Chezeaux35, George B. Chuyong36, Connie J. Clark37, Murray Collins38,39,
James A. Comiskey40,41, David A. Coomes42, Greta C. Dargie1, Thales de Haulleville2,
Marie Noel Djuikouo Kamdem36, Jean-Louis Doucet43, Adriane Esquivel-Muelbert1,44,
Ted R. Feldpausch45, Alusine Fofanah46, Ernest G. Foli24, Martin Gilpin1, Emanuel Gloor1,
Christelle Gonmadje47, Sylvie Gourlet-Fleury29,30, Jefferson S. Hall48, Alan C. Hamilton49,
David J. Harris50, Terese B. Hart51,52, Mireille B. N. Hockemba34, Annette Hladik53,
Suspense A. Ifo54, Kathryn J. Jeffery23, Tommaso Jucker55, Emmanuel Kasongo Yakusu2,3,1 0,
Elizabeth Kearsley2,56, David Kenfack48,57, Alexander Koch4,58, Miguel E. Leal59,
Aurora Levesley1, Jeremy A. Lindsell60,61, Janvier Lisingo62, Gabriela Lopez-Gonzalez1,
Jon C. Lovett1,63, Jean-Remy Makana62, Yadvinder Malhi64, Andrew R. Marshall6,65,66,
Jim Martin67, Emanuel H. Martin57,6 8, Faustin M. Mbayu10, Vincent P. Medjibe37,69,7 0,
Vianet Mihindou21,70, Edward T. A. Mitchard38, Sam Moore64, Pantaleo K. T. Munishi71,
Natacha Nssi Bengone21, Lucas Ojo72, Fidèle Evouna Ondo70, Kelvin S.-H. Peh73,74,
Georgia C. Pickavance1, Axel Dalberg Poulsen50, John R. Poulsen37, Lan Qie1,75, Jan Reitsma76,
Francesco Rovero77,78, Michael D. Swaine79, Joey Talbot1,80, James Taplin81, David M. Taylor82,
Duncan W. Thomas83, Benjamin Toirambe2,84, John Tshibamba Mukendi2,10,85,
Darlington Tuagben7,86, Peter M. Umunay8 7,88 , Geertje M. F. van der Heijden89,
Hans Verbeeck56, Jason Vleminckx90,91, Simon Willcock92, Hannsjörg Wöll93, John T. Woods94
& Lise Zemagho15
1School of Geography, University of Leeds, Leeds, UK. 2Service of Wood Biology, Royal
Museum for Central Africa, Tervuren, Belgium. 3Department of Environment, Laboratory of
Wood Technology (Woodlab), Ghent University, Ghent, Belgium. 4Department of Geography,
University College London, London, UK. 5Mensuration Unit, Forestry Commission of Ghana,
Kumasi, Ghana. 6Department of Environment and Geography, University of York, York, UK.
7Forestry Development Authority of the Government of Liberia (FDA), Monrovia, Liberia. 8DR
Congo Programme, Wildlife Conservation Society, Kinshasa, Democratic Republic of Congo.
9Centre de Formation et de Recherche en Conservation Forestière (CEFRECOF), Epulu,
Democratic Republic of Congo. 10Faculté de Gestion de Ressources Naturelles
Renouvelables, Université de Kisangani, Kisangani, Democratic Republic of Congo. 11School
of Geography, Earth and Environmental Sciences, University of Plymouth, Plymouth, UK.
12Salonga National Park, Kinshasa, Democratic Republic of Congo. 13World Wide Fund for
Nature, Gland, Switzerland. 14Faculty of Environmental Sciences and Natural Resource
Management, Norwegian University of Life Sciences, Ås, Norway. 15Plant Systematic and
Ecology Laboratory, Higher Teachers’ Training College, University of Yaounde I, Yaounde,
Cameroon. 16Department of Natural Sciences, Manchester Metropolitan University,
Manchester, UK. 17Center for International Forestry Research (CIFOR), Bogor, Indonesia.
18Faculty of Forestry, University of British Columbia, Vancouver, British Columbia, Canada.
19Bioversity International, Yaounde, Cameroon. 20Faculty of Forestry, University of Toronto,
Toronto, Ontario, Canada. 21Ministry of Forests, Seas, Environment and Climate, Libreville,
Gabon. 22Institut de Recherche en Écologie Tropicale, Libreville, Gabon. 23Department of
Biological and Environmental Sciences, University of Stirling, Stirling, UK. 24Forestry Research
Institute of Ghana (FORIG), Kumasi, Ghana. 25Université Oficielle de Bukavu, Bukavu,
Democratic Republic of Congo. 26UK Centre for Ecology & Hydrology, Penicuik, UK.
27Ministère des Eaux, Forêts, Chasse et Pêche (MEFCP), Bangui, Central African Republic.
28Institut Centrafricain de Recherche Agronomique (ICRA), Bangui, Central African Republic.
29Forêts et Sociétés (F&S), Centre de Coopération Internationale en Recherche Agronomique
pour le Développement (CIRAD), Montpellier, France. 30Forêts et Sociétés (F&S), Université de
Montpellier, Montpellier, France. 31The Institute of Tropical Forest Conservation (ITFC),
Mbarara University of Science and Technology (MUST), Mbarara, Uganda. 32Faculté des
8 | Nature | www.nature.com
Article
Sciences et Techniques, Laboratoire de Botanique et Écologie, Université Marien Ngouabi,
Brazzaville, Republic of Congo. 33Isotope Bioscience Laboratory-ISOFYS, Ghent University,
Ghent, Belgium. 34Congo Programme, Wildlife Conservation Society, Brazzaville, Republic of
Congo. 35Rougier-Gabon, Libreville, Gabon. 36Faculty of Science, Department of Botany and
Plant Physiology, University of Buea, Buea, Cameroon. 37Nicholas School of the Environment,
Duke University, Durham, NC, USA. 38School of GeoSciences, University of Edinburgh,
Edinburgh, UK. 39Grantham Research Institute on Climate Change and the Environment,
London, UK. 40Inventory and Monitoring Program, National Park Service, Fredericksburg, VA,
USA. 41Smithsonian Institution, Washington, DC, USA. 42Department of Plant Sciences,
University of Cambridge, Cambridge, UK. 43TERRA, Forest is Life, Gembloux Agro-Bio Tech,
University of Liège, Liège, Belgium. 44School of Geography, Earth and Environmental
Sciences, University of Birmingham, Birmingham, UK. 45Department of Geography, College of
Life and Environmental Sciences, University of Exeter, Exeter, UK. 46The Gola Rainforest
National Park, Kenema, Sierra Leone. 47National Herbarium, Yaounde, Cameroon. 48Forest
Global Earth Observatory (ForestGEO), Smithsonian Tropical Research Institute, Washington,
DC, USA. 49Kunming Institute of Botany, Chinese Academy of Sciences, Kunming, China.
50Royal Botanic Garden Edinburgh, Edinburgh, UK. 51Lukuru Wildlife Research Foundation,
Kinshasa, Democratic Republic of Congo. 52Division of Vertebrate Zoology, Yale Peabody
Museum of Natural History, New Haven, CT, USA. 53Département Hommes et Environnement,
Muséum National d’Histoire Naturel, Paris, France. 54École Normale Supérieure (ENS),
Département des Sciences et Vie dela Terre, Laboratoire de Géomatique et d’Écologie
Tropicale Appliquée, Université Marien Ngouabi, Brazzaville, Republic of Congo. 55School of
Biological Sciences, University of Bristol, Bristol, UK. 56Department of Environment,
Laboratory of Computational & Applied Vegetation Ecology (Cavelab), Ghent University,
Ghent, Belgium. 57Tropical Ecology, Assessment and Monitoring (TEAM) Network, Arlington,
VA, USA. 58Department of Earth Sciences, University of Hong Kong, Hong Kong, China.
59Uganda Programme, Wildlife Conservation Society, Kampala, Uganda. 60A Rocha
International, Cambridge, UK. 61Centre for Conservation Science, The Royal Society for the
Protection of Birds, Sandy, UK. 62Faculté des Sciences, Laboratoire d’Écologie et
Aménagement Forestier, Université de Kisangani, Kisangani, Democratic Republic of Congo.
63Royal Botanic Gardens, Kew, London, UK. 64Environmental Change Institute, School of
Geography and the Environment, University of Oxford, Oxford, UK. 65Tropical Forests and
People Research Centre, University of the Sunshine Coast, Sippy Downs, Queensland,
Australia. 66Flamingo Land Ltd, Kirby Misperton, UK. 67Fleming College, Peterborough,
Ontario, Canada. 68Udzungwa Ecological Monitoring Centre, Mang’ula, Tanzania.
69Commission of Central African Forests (COMIFAC), Yaounde, Cameroon. 70Agence
Nationale des Parcs Nationaux, Libreville, Gabon. 71Sokoine University of Agriculture,
Morogoro, Tanzania. 72University of Abeokuta, Abeokuta, Nigeria. 73School of Biological
Sciences, University of Southampton, Southampton, UK. 74Department of Zoology,
Conservation Science Group, University of Cambridge, Cambridge, UK. 75School of Life
Sciences, University of Lincoln, Lincoln, UK. 76Bureau Waardenburg, Culemborg, The
Netherlands. 77Department of Biology, University of Florence, Florence, Italy. 78Tropical
Biodiversity Section, MUSE—Museo delle Scienze, Trento, Italy. 79Department of Plant & Soil
Science, School of Biological Sciences, University of Aberdeen, Aberdeen, UK. 80Institute for
Transport Studies, University of Leeds, Leeds, UK. 81UK Research & Innovation, Innovate UK,
London, UK. 82Department of Geography, National University of Singapore, Singapore,
Singapore. 83Biology Department, Washington State University, Vancouver, WA, USA.
84Ministère de l’Environnement et Développement Durable, Kinshasa, Democratic Republic of
Congo. 85Faculté des Sciences Appliquées, Université de Mbujimayi, Mbujimayi, Democratic
Republic of Congo. 86Friends of Ecosystem and the Environment, Monrovia, Liberia. 87Yale
School of Forestry and Environmental Studies, Yale University, New Haven, CT, USA. 88Wildlife
Conservation Society, New York, NY, USA. 89School of Geography, University of Nottingham,
Nottingham, UK. 90International Center for Tropical Botany, Department of Biological
Sciences, Florida International University, Miami, FL, USA. 91Faculté des Sciences, Service
d’Évolution Biologique et Écologie, Université Libre de Bruxelles, Brussels, Belgium. 92School
of Natural Sciences, University of Bangor, Bangor, UK. 93Independent Researcher, Bad Aussee,
Austria. 94W.R.T. College of Agriculture and Forestry, University of Liberia, Monrovia,
Liberia. 95These authors contributed equally: Wannes Hubau, Simon L. Lewis.
e-mail: whubau@gmail.com
Methods
Plot selection
Closed canopy (that is, not woody savannah) old-growth mixed-age
forest inventory plots were selected using commonly used crite-
ria6,13,27: structurally intact (thatis,free of fire and industrial logging);
all trees with diameter at reference height ≥100 mm measured at least
twice; area ≥0.2 ha; altitude <1,500m above sea level; MAT≥20.0 °C51;
annual precipitation ≥1,000mm
51
; located ≥50 m from anthropogenic
forest edges. Of the 244 plots included in the study, 217 contribute
to the African Tropical Rainforest Observatory Network (AfriTRON;
www.afritron.org), with data curated at www.ForestPlots.net
52,53
. These
include plots from Sierra Leone, Liberia, Ghana, Nigeria, Cameroon,
Gabon, Republic of Congo, Democratic Republic of Congo, Uganda and
Tanzania52,53 (Extended Data Fig.1). Fifteen plots are part of the TEAM
network, from Cameroon, Republic of Congo, Tanzania and Uganda54–57.
Nine plots contribute to the ForestGEO network, from Cameroon and
Democratic Republic of Congo58 (9 plots from Democratic Republic of
Congo, with codes SNG, contribute to both AfriTRON and ForestGEO
networks, included above in the AfriTRON total). Finally, three plots
from Central African Republic are part of the CIRAD network59,60. The
large majority of plots are sited in terra firme (not inundated by river
water) forests and have mixed species composition, although four are in
seasonally flooded forest and 14 plots are in Gilbertiodendron dewevrei
monodominant forest, a locally common forest type in Africa (Supple-
mentary Table1). The 244 plots have a mean size of 1.1 ha (median, 1 ha),
with a total plot area of 277.9 ha. The dataset comprises 391,968 diam-
eter measurements on 135,625 stems, of which 89.9% were identified
to species, 97.5% to genus and 97.8% to family. Mean total monitoring
period is 11.8 years, mean census length 5.7 years, with a total of 3,214
hectare years of monitoring. The 321 Amazon plots are published and
were selected using the same criteria6, except in the African selection
criteria we specified a minimum anthropogenic edge distance and
added a minimum temperature threshold.
Plot inventory and tree biomass carbon estimation
Tree-level aboveground biomass carbon is estimated using an allomet-
ric equation with parameters for tree diameter(in mm), tree height(in
m) and wood mass density (in g cm−3)61. The calculation of each is dis-
cussed in turn. All calculations were performed using the R statistical
platform, version 3.2.1 (ref.
62
) using the BiomasaFP R package, version
0.2.1 (ref. 63).
Tree diameter. In all plots, all woody stems with ≥100 mm diameter at
1.3 m from the base of the stem (‘diameter at breast height’, DBH, in mm),
or 0.5 m above deformities or buttresses, were measured, mapped and
identified using standard forest inventory methods64,65. The height of
the point of measurement (POM) was marked on the trees and recorded,
so that the same POM is used at the subsequent forest census. For stems
developing deformities or buttresses over time that could potentially
disturb the initial POM, the POM was raised approximately 500 mm
above the deformity. Estimates of the diameter growth of trees with
changed POM used the ratio of new to old POMs, to create a single tra-
jectory of growth from the series of diameters at two POM heights
6,13,65
.
We used standardized protocols to assess typographical errors and
potentially erroneous diameter values (for example, trees shrinking
by >5 mm), missing values, failures to find the original POM, and other
issues. Where necessary we estimated the likely value via interpolation
or extrapolation from other measurements of that tree, or when this
was not possible we used the medianor mean growth rate of trees in
the same plot, census and size-class. We used the median growth rate
forsize classes of DBH=100–199 mm or 200–399 mm.We used themean
growth rate for asize class with DBH > 400 mm, as there were fewer trees
in the largest size class
65
. We interpolated measurements for 1.3% of
diameters, extrapolated 0.9%, and used median growth rates for 1.5%.
Tree height. Height of individuals from ground to the top leaf, hereafter
H
t
, was measured in 204 plots, using a laser hypsometer (Nikon forestry
Pro) from directly below the crown (most plots), a laser or ultrasonic
distance device with an electronic tilt sensor, a manual clinometer, or
by direct measurement, that is, climbing the tree. Only trees where
the top was visible were selected
66
. In most plots, tree selection was
similar: the 10 largest trees were measured, together with 10 randomly
selected trees per diameter from five classes: 100–199 mm, 200–299
mm, 300–399 mm, 400–499 mm, and 500+ mm trees, following stand-
ard protocols
66
. We measured the actual height of 24,270 individual
trees from 204 plots. We used these data and the local.heights function
in R package BiomasaFP63 to fit 3-parameter Weibull relationships:
Ha=(1−e) (1)
b
−DBH
We chose the Weibull model (with Weibull parameters a, b and c)
because it is known to be robust
66,67
. We parameterized separate H
t
-
DBH relationships for four different combinations of edaphic forest
type and biogeographical region: (1) terra firme forest in West Africa,
(2) terra firme forest in Lower Guinea and the Western Congo Basin, (3)
terra firme forest in Eastern Congo Basin and East Africa, (4) seasonally
flooded forest from Lower Guinea and the Western Congo Basin (there
were no seasonally flooded forest plots in the other biogeographi-
cal regions). The parameters are: (1) terra firme forest in West Africa,
a=56.0; b=0.0401; c=0.744; (2) terra firme forest in Lower Guinea and
the Western Congo Basin, a=47.6; b=0.0536; c=0.755; (3) terra firme
forest in the Eastern Congo Basin and East Africa, a=50.8; b=0.0499;
c=0.706; and finally (4) seasonally flooded forest from Lower Guinea
and the Western Congo Basin, a=38.2; b=0.0605; c=0.760. For each
of these combinations of forest type and bioregion, the local.heights
function combines all height measurements from all plots belonging to
that forest type/bioregion and fits the Weibull model parameters using
nonlinear least squares (nls function in R with default settings), with
starting values of a=25, b=0.05 and c=0.7 chosen because they led
to regular model convergence. We fitted these models either treating
each observation equally or with weights proportional to each tree’s
basal area. These weights give more importance to large trees during
model fitting. We selected the best fitting of these models, determining
this to be the model that minimized prediction error of stand biomass
when calculated with estimated heights or observed heights. In this
way, we selected the non-weighted model for terra firme forests in
Lower Guinea/Western Congo Basin and for flooded forests in the Lower
Guinea/Western Congo Basin; we selected the weighted model for the
other two biogegraphical regions (West Africa and Eastern Congo Basin/
East Africa).The parameters were used to estimate H
t
from DBH for all
tree DBH measurements for input into the allometric equation. Median
measured individual total tree height is 20.5 m; the height range is 3.1
to 72.5 m. The root mean squared error (RMSE) between the full dataset
of measured heights and the predicted heights is 5.7 m, which is 8.0%
of the total range. Furthermore, RMSE is 5.3 m in terra firme forest in
West Africa (7.5% of the range; n=9,771 trees); RMSE is 6.4 m in terra
firme forest in Lower Guinea and the Western Congo Basin (8.7% of the
range; n=10,838 trees); RMSE is 4.8 m in terra firme forest in the Eastern
Congo Basin and East Africa (8.8% of the range; n=3,269 trees); and
RMSE is 4.1 m in seasonally flooded forest from Lower Guinea and the
Western Congo Basin (12.5% of the range; n=392 trees).
Wood density. Dry wood density (ρ) measurements were compiled
for 730 African species from published sources and stored in www.
ForestPlots.net; most were sourced from the Global Wood Density
Database on the Dryad digital repository (www.datadryad.org)
68,69
.
Each individual in the tree inventory database was matched to a taxon-
specific mean wood density value. Species in both the tree inventory
and wood density databases were standardized for orthography and
Article
synonymy using the African Plants Database (www.ville-ge.ch/cjb/bd/
africa/) to maximize matches
13
. For incompletely identified individuals
or for individuals belonging to species not in the ρ database, we used the
mean ρ value for the next-highest known taxonomic category (genus or
family, as appropriate). For unidentified individuals, we used the mean
wood density value of all individual trees in the plot13,52.
Allometric equation. For each tree we used a published allometric
equation61 to estimate aboveground biomass. We then converted this
to carbon, assuming that aboveground carbon (AGC, in MgCha
−1
) is
45.6% of aboveground biomass70. Thus:
ρHAGC=0.45(0.067DB))/1,000 (2
with DBH in mm, dry wood density ρ in g cm−3, and total tree height Ht
in m (ref.
61
). Aboveground carbon in living biomass for each plot at
each census date was estimated as the sum of the AGC of each living
stem, divided by plot area (in hectares).
Carbon gain and carbon loss estimation
Net carbon sink (in MgCha
−1
yr
−1
) is estimated as carbon gains minus
carbon losses. Carbon gains (in MgCha
−1
yr
−1
) are the sum of the above-
ground live biomass carbon additions from the growth of surviving
stems and the addition of newly recruited stems (recruits are stems
reaching a DBH ≥ 100 mm during a given census interval), divided
by the census length (in years) and plot area (in hectares). For each
stem that survived a census interval, carbon additions from its growth
(MgCha−1yr−1) were calculated as the difference between its AGC at the
end census of the interval and its AGC at the beginning census of the
interval. For each stem that recruited during the census interval (that
is, reaching DBH≥100 mm), carbon additions were calculated in the
same way, assuming DBH=0 mm at the start of the interval, follow-
ing standard procedures
6,65
. Carbon losses (in MgCha
−1
yr
−1
) are esti-
mated as the sum of aboveground biomass carbon from all stems that
died during a census interval, divided by the census length (in years)
and plot area (in hectares). Both carbon gains and carbon losses are
calculated using standard methods
6
, including a census interval bias
correction, using the SummaryAGWP function of the R package Bio-
masaFP63,64,68.
As carbon gains (and losses, see below) are affected by a census
interval bias, with the underestimate increasing with census length,
we corrected this bias by accounting for (1) the carbon additions from
trees that grew before they died within an interval (unobserved growth)
and (2) the carbon additions from trees that reached 100 mm DBH (that
is, were recruited) and then died within the same interval (unobserved
recruitment)65,71.
The first component, the unobserved growth of a stem that died
during a census interval, is estimated as the difference between AGC
at death and AGC at the start of the census. These are calculated using
equation (2), from DBHdeath and DBHstart, respectively. The latter is part
of the data, the first can be estimated as: DBHdeath=DBHstart×G×Ymean,
where G is the plot-level median diameter growth rate (in mm yr
−1
) of the
size class the tree was in at the start of the census interval (size classes
are defined as DBH<200 mm, 400 mm>DBH≥200 mm and DBH≥400
mm) and Ymean is the mean number of years that trees survived in the
census interval before dying. Ymean is calculated from the number of trees
that are expected to have died in each year of the census interval, which
is derived from the plot-level per capita mortality rate (ma; as percent-
age of dead trees per year) calculated following equation (5) in ref. 71.
The second component, the growth of recruits that were not
observed because they died during the census interval, is estimated
by calculating the number of unobserved recruits and diameter at death
for each unobserved recruit. The number of unobserved recruits in a
given year (stems ha−1 yr−1) is estimated as: Nu.r=RaPsurv×Ra, where Ra
(number of recruited stems ha
−1
yr
−1
) is the per-area annual recruitment
calculated following equation (11) in ref. 71 and Psurv is the probability of
each recruit surviving until the next census: Psurv=(1−ma)T, where T is the
number of years remaining in the census interval. Summing N
u.r
for each
year in a census interval gives the total number of unobserved recruits
in that census interval. We then estimate diameter at death for each
unobserved recruit, which is given in millimetres by DBH
death,u.r
=100 +
(G
s
×Y
mean-rec
), where G
s
is the plot-level median diameter growth rate (in
mm yr−1) of the smallest size class (that is,DBH <200 mm) and Ymean-rec
is the mean number of years that unobserved recruits survived in the
census interval before dying. Ymean-rec is calculated as follows: from ma
we can calculate the number of recruits in a given year that died in each
subsequent year, and from this calculate the mean lifespan of recruits
in a given year that died before the next census; Y
mean-rec
is then the mean
of each year’s recruit-lifespan, weighted by the number of unobserved
recruits in each year.
The census interval bias correction (components one and two com-
bined) typically add <3% to plot-level carbon gainscalculated for each
plot census interval. Carbon losses are affected by the same census
interval bias, so we corrected this bias by accounting for the additional
carbon losses from the trees that were recruited and then died within
the same interval, and the additional carbon losses resulting from the
growth of the trees that died in the interval
6,15,63
. These two components
are calculated in the same way as for carbon gains and typically add
<3% to plot-level carbon losses.
Carbon gains include both gains from the growth of surviving stems
and new recruits. Separating carbon gains from the tree growth of sur-
viving stems and newly recruited stems shows that carbon gains from
recruitment are small overall, and are significantly lower in Africa than
in the Amazon (in Africa, 0.17MgCha−1yr−1; CI: 0.16–0.18 versus in the
Amazon, 0.27MgCha
−1
yr
−1
; CI: 0.25–0.28, P<0.001; two-way Wilcoxon
test), but this is compensated by carbon gains from survivors being
significantly larger in Africa (2.33MgCha−1yr−1; CI: 2.27–2.39) than in
the Amazon (2.13MgCha
−1
yr
−1
; CI: 2.09–2.17, P=0.014). Therefore,
gains overall (sum of gains from surviving stems and newly recruited
stems) are indistinguishable between the continents (in Africa,
2.57MgCha
−1
yr
−1
; CI: 2.51–2.67 versus in the Amazon, 2.46MgCha
−1
yr
−1
;
CI: 2.41–2.50, P=0.460; two-way Wilcoxon test). The lower carbon
gains from recruitment in Africa are probably due to the lower stem
turnover rates and longer CRT.
Long-term gain, loss and net carbon sink trend estimation
The estimated mean and uncertainty in carbon gains, carbon losses and
the net carbon sink of the African plots from 1 January 1983 to 31 Decem-
ber 2014 (Fig.1, Extended Data Fig.7 and Extended Data Fig.8) were
calculated following ref.
6
to allow direct comparison with published
Amazonian results. First, each census interval value was interpolated for
each 0.1-year period within the census interval. Then, for each 0.1-year
period between 1 January 1983 and 31 December 2014, we calculated
a weighted mean of all plots monitored at that time, using the square
root of plot area as a weighting factor6. Confidence intervals for each
0.1-year period were bootstrapped.
Trends in carbon gains, losses and the net carbon sink over time were
assessed using linear mixed effects models (lmer function in R, lme4
package
72
), providing the linear slopes reported in Fig.1. These models
regress the midpoint of each census interval against the value of the
response variable for that census interval. Plot identity was included
as a random effect, that is, by assuming that the intercept can vary ran-
domly among plots. We did not include slope as a random effect, con-
sistent with previously published Amazon analyses6, because models
did not converge owing to some plots having too few census intervals.
Observations were weighted by plot size and census interval length.
Weightingfor the Africa data was derived empirically, by assuming
apriori that there is no significant relation between the net carbon sink
and census interval length or plot size, following ref. 13. The following
weighting removes all pattern in the residuals:
Weight =length+plotsize −1 (3
int
where length
int
is the length of the census interval, in years. Significance
was assessed by regressing the residuals of the net carbon sink model
against the weights (P=0.702).Similar published weighting was used
for the Amazon plots6.
Differences in long-term slopes between the two continents for
carbon gains, carbon losses and net carbon sink, reported in the main
text, were also assessed using linear mixed effects modelsand weight-
ing, as described above, but performed on the combined African and
Amazonian datasets and limited to their common time window, 1 Janu-
ary 1983 to mid-2011. For these three tests on the pooled data (gains,
losses and net sink) we included an additional interaction term between
census interval date and continent, where a significant interaction
would indicate that the slopes differ between continents. The statisti-
cal significance of continental differences in slope were assessed using
the F-statistic (ANOVA function in R, car package
73
). Shortening the
common time window to the 20 years when the continents are best-
sampled, mid-1991 to mid-2011, gave very similar results, including a
divergent continental sink (P=0.04).
Continental and pan-tropical carbon sink estimates
The per unit area total net carbon sink (in MgCha−1yr−1) for each time
period in Table1 (each decade between 1 January 1980 and 31 December
2009; and between 1 January 2010 and 31 December 2014) is the sum
of three components. The first component is the per unit area above-
ground carbon sink from living trees and lianas with DBH≥100 mm.
For Africa we use the per unit area net carbon sink values presented
in this paper. For Amazonia, we use data in ref.
6
. For Southeast Asia,
we use inventory data collected using similar standardized methods
from 49 plots in ref.
15
. For each time window, we use all plots for which
census dates overlap the period, weighted by the square root of plot
area, as for the solid lines in Fig.1. The second component is the per
unit area aboveground carbon sink from living trees and lianas with
DBH <100 mm. This is calculated as 5.19%, 9.40% and 5.46% of the first
component (that is, aboveground carbon of large living trees) in Africa,
Amazonia and Southeast Asia respectively
74
. The third component is
the per unit area belowground carbon sink in live biomass, that is, roots.
This is calculated as 25%, 37% and 17% of the aboveground carbon of
living trees with DBH ≥100 mm in Africa
13
, Amazonia
6
and Southeast
Asia75 respectively.
For each time period in Table1 we calculated the continental-scale
total carbon sink (Pg C yr
−1
) by multiplying the per unit area total net
carbon sink described above by the area of intact forest on each con-
tinent at that time interval (in ha) reported in Extended Data Table2.
Decades are calculated from 1 January 1990 to 31 December 1999. For
comparability with previous continental-sink results, we used conti-
nental values of intact forest area for 1990, 2000, 2005 and 2010 as
published in ref.
1
, that is, total forest area minus forest regrowth. We
used the 1990–2010 data to fit an exponential model for each continent
and used this model to estimate intact forest area for 1980 and 2015.
Finally, in the main text we calculated the proportion of anthro-
pogenic CO
2
emissions removed by Earth’s intact tropical forests, as
the total pan-tropical carbon sink from Table1 divided by the total
anthropogenic CO
2
emissions. Total anthropogenic CO
2
emissions
are calculated as the sum of emissions from fossil fuel and land-use
change and are estimated at 7.6PgCyr
−1
in the 1990s, 9.0PgCyr
−1
in
the 2000s, and 11.1PgCyr−1 in the 2010s (ref. 21, assuming 1.7% growth
in fossil fuel emissions in 2018 and 2019, and mean 2010–2017 land-use
change emissions for 2018 and 2019).
Carbon sink from an atmospheric perspective
To estimate the evolution of the carbon sink from an atmospheric
perspective, we assumed that the contribution to the atmosphere
from carbon gains are experienced immediately, while the contribu-
tion to the atmosphere from carbon losses must take into account
the delay in decomposition of dead trees. We did this by calculating
total forest carbon loss (MgCha−1yr−1) for each year in the period
1 January 1950 to 31 December 2014, using the mean 1 January 1983 to
31 December 2014 records from Fig.1 and assuming constant losses
before 1983 (1.9MgCha−1yr−1 and 1.5MgCha−1yr−1 for Africa and Ama-
zonia respectively). Then, for each focal year in the period 1950–2014,
we calculated how much carbon was released to the atmosphere in the
subsequent years as follows: y
i
=x
0
×e
−0.17(i− 1)
x
0
×e
−0.17i
, where x
0
is the
total forest carbon loss of the focal year; y
i
is the carbon released to the
atmosphere at i years from the focal year; and −0.17 yr−1 is a constant
decomposition rate calculated for tropical forests in the Amazon
45
.
For example, carbon loss was 1.95MgCha
−1
in 1990 in African forests
(Fig.1), from which 0.31MgCha
−1
was released to the atmosphere in
1991; 0.26MgCha
−1
in 1992; 0.22MgCha
−1
in 1993; 0.07MgCha
−1
in
2000 and 0.01MgCha−1 in 2010. Hence, of the full 1.95MgCha−1 dead
tree biomass from 1990, ~50% was released to the atmosphere after 4
years, ~85% after 10 years, and ~97% after 20 years. Finally, for each year
between 1983 and 2014, the total contribution to the atmosphere from
carbon losses was calculated as the sum of all carbon contributions
released at that year, including all carbon loss pools from previous
years that are released during the focal year(an approach similar to
ref. 6). We then calculated decadal-scale mean contributions to the
atmosphere from carbon lossesto estimate the carbon sink from an
atmospheric perspective, reported in the main text.
Predictor variable estimates (1983–2015)
For each census interval of each plot, we examined potential predictor
variables that may explain the long-term trends in carbon gains and
carbon losses, reported in Table2 and Extended Data Table1. First,
the environmental conditions during the census interval; second, the
rate of change of these parameters; and third, forest attributes that
may affect how different forests respond to the same environmental
change. The predictor variable estimates for each census need to avoid
bias due to seasonal variation, for example the intra-annual variability
in atmospheric CO
2
concentration. We therefore applied the following
procedure to avoid seasonal variability impacts on long-term trends:
(1) the length of each focal census interval was rounded to the nearest
complete year (for example, a 1.1-year interval became a 1 year interval);
(2) we computed dates that minimized the difference between actual
fieldwork dates and complete-year census dates, while ensuring that
subsequent census intervals of a plot do not overlap. The resulting
sequence of non-overlapping census intervals was used to calculate
interval-specific means for each environmental predictor variable to
remove seasonal effects. The mean difference between the actual field-
work dates and the complete-year census dates is 0.13 decimal years.
The first group of potential predictor variables, estimated for each
census interval of each plot, are theory-driven choices: atmospheric
CO
2
concentration, MAT and drought intensity, which we quantified
as MCWD14,20,76,77.
Atmospheric CO2 concentration. CO2 (in ppm) is estimated as the
mean of the monthly mean values from the Mauna Loa record78 over the
complete year census interval. While atmospheric CO2 concentration is
highly correlated with time (R
2
=0.98), carbon gains are slightly better
correlated with CO
2
(R
adj2
=0.0027) than with time (R
adj2
=0.0025), as
expected from theory.
Mean annual temperature. MAT (in °C) was derived from the tempo-
rally resolved (1901–2015) dataset of monthly mean temperature from
the Climatic Research Unit (CRU TS version 4.03; ~3,025-km
2
resolution;
released 15 May 2019; https://crudata.uea.ac.uk/cru/data/hrg/)
79
. We
downscaled the data to ~1-km
2
resolution using the WorldClim v2 data-
set
51,80
, by subtracting the difference in mean monthly temperature,
Article
and applying this monthly correction to all months
81
. We then calcu-
lated MAT for each complete year census interval of each plot using
the downscaled monthly CRU record.
Maximum climatological water deicit. MCWD (in mm) was derived
from the ~3,025-km2 resolution Global Precipitation Climatology Centre
dataset (GPCC version 6.0) that includes many more rain gauges than
CRU in tropical Africa82,83. Because GPCC ends in 2013 we combined it
with satellite-based Tropical Rainfall Measurement Mission data (TRMM
3B43 V7 product, ~757-km2 resolution)84. The fit for the overlapping time
period (1998–2013) was used to correct any systematic difference be-
tween GPCC and TRMM: GPCC′=a+b×GPCC, with GPCC′ the adjusted
GPCC record and a and b being different parameters for each month
of the year and for each continent. Precipitation was then downscaled
to ~1-km
2
resolution using the WorldClim dataset
51,80
, by dividing by
the ratio in mean monthly rainfall, and applying this monthly correc-
tion to all months
81
. For each census interval we extracted monthly
precipitation values and estimated evapotranspiration to calculate
monthly climatological water deficit (CWD), a commonly used metric
of dry season intensity for tropical forests14,76,77. Monthly CWD values
were calculated for each subsequent series of 12 months (complete
years)77. Monthly CWD estimation begins with the wettest month of
the first year in the interval, and is calculated as 100 mm per month
evapotranspiration (ET) minus monthly precipitation (P). Then, CWD
i
values for the subsequent 11 months (i) were calculated recursively
as: CWDi=ET − Pi + CWDi−1, where negative CWDi values were set to
zero77 (no drought conditions). This procedure was repeated for each
subsequent complete 12 months. We then calculated the annual MCWD
as the largest monthly CWD value for every complete year within the
census interval, with the MCWD of a census interval being the mean
of the annual MCWD values within the census interval. Larger MCWD
indicates more severe water deficits.
We assume evapotranspiration is 100 mm per month on both conti-
nents, based on measurements from Amazonia
76,77
, more limited meas-
urements from West Africa summarized in ref.
85
, predictive skill
86
, and
use in past studies on both continents
14,87
. MCWD therefore represents
a precipitation-driven dry season deficit, given that evapotranspiration
remains constant. An alternative assessment, using a data-driven evapo-
transpiration product88,89, gave a mean evapotranspiration of 95mm
and 98 mm per month for the African and Amazonian plot networks
respectively(mean for the 1982–2008 period). Using these values did
not affect the results.
To calculate the environmental change of potential predictor vari-
ables, CO
2
-change (in ppm yr
−1
), MAT-change (in °C yr
−1
) and MCWD-
change (in mm yr−1), we selected an optimum period over which to
calculate the change, derived empirically by assessing the correlation
of carbon gains (all plots, all censuses) with the change in each envi-
ronmental variable, using linear mixed effects models (lmer function
in R, lme4 package72). The annualized change in the environmental
variable was calculated as the change between the focal interval and
a prior interval (termed the baseline period) with a lengthening time
window ranging from 1 year through to 80 years before the focal interval
(that is, 80 linear mixed effects models per variable). We calculated
Akaike’s Information Criterion (AIC) for each model and selected the
interval length with the lowest AIC. Thus, MAT-change=(MAT
i
−MAT
b
)/
(date
i
−date
b
), where MAT
i
is the MAT over the focal census interval
calculated using the procedure described above, MAT
b
is the MAT over
a baseline period before the focal interval, datei is the mid-date of the
focal census interval and dateb is the mid-date of the baseline period.
The lmer results show that the baseline period for MAT-change is 5 years
and for CO2-change it is 56 years, while MCWD showed no clear trend,
so MCWD-change was not included in the models (see Extended Data
Fig.3). All three results conform to apriori theoretical expectations.
For CO
2
a maximum response to an integrated 56 years of change is
expected because forest stands will respond most strongly to CO
2
when
most individuals have grown under the new rapidly changing condition,
which should be at its maximum at a time approximately equivalent to
the CRT of a forest stand30,90 (mean of 62 years in the pooled dataset).
For MAT, 5 years is consistent with experiments showing temperature
acclimation of leaf- and plant-level photosynthetic and respiration
processes over half-decadal timescales
31,91
. MCWD has no overall trend
suggesting that once a drought ends, its impact on tree growth fades
rapidly, as seen in other studies14,92. Furthermore,in the moist tropics
wet-season rainfall is expected to recharge soil water, so lagged impacts
of droughts are not expected.
We calculated estimates of two forest attributes that may alter
responses to environmental change as potential predictor variables:
wood density and CRT. In intact old-growth forests, mean wood density
(in g cm
−3
) is inversely related to resource availability
28,93,94
, as is seen in
our dataset (carbon gains and plot-level mean wood density are nega-
tively correlated; Extended Data Fig.4). Wood density is calculated for
each census interval in the dataset, as the mean wood density of all trees
alive at the end of the census interval, to be consistent with the previous
Amazon analysis6. Carbon residence time (CRT, in years) is a measure
of the time that fixed carbon stays in the system. CRT is a potential
correlate of the impact of past carbon gains on later carbon losses
30
.
To avoid circularity in the models, the equation used to calculate CRT
differed depending on the response variable. If the response variable
is carbon loss, the CRT equation is based on gains: CRT=AGC/gains,
with AGC for each interval based on AGC at the end of the interval, and
the gains for each interval calculated as the time-weighted mean of
the gains in the interval and the previous intervals (that is, long-term
gains). If the response variable is carbon gains, the CRT equation is
based on losses: CRT=AGC/losses. The equation employed for use in
the carbon loss model (based on gains) is the standard formula used
to calculate CRT and is retained in the minimum adequate model (see
below and Table2). The non-standard CRT equation (based on losses)
used in the carbon gain model is not retained in the minimum adequate
model (see below).
Statistical modelling of the carbon gain, loss and sink trends
We first constructed two models including those environmental driv-
ers exhibiting long-term change that impact theory-driven models
of photosynthesis and respiration as predictor variables: CO
2
, MAT
and MCWD. One model had carbon gains as the response variable, the
other had carbon losses as the response variable (both in MgCha
−1
yr
−1
).
Models were fitted using the lme function in R, with maximum likeli-
hood (NLME package
95
). All census intervals within all plots were used,
weighted by plot size and census length (using equation (3)). Plot iden-
tity was included as a random effect, that is, assuming that the intercept
can vary randomly among plots. All predictor variables in the models
were scaled without centring (scale function in R, RASTER package
62
).
Carbon gain values were normally distributed but carbon loss values
required a power-law transformation (λ=0.361) to meet normality
criteria. Multi-parameter models are: carbon gains=intcp + a×CO2 +
b×MAT + c×MCWD (model 1); carbon losses=intcp + a×CO2 + b×MAT +
c×MCWD (model 2); where intcp is the estimated model intercept, and
a, b and c are model parameters giving the slope of relationships with
environmental predictor variables. For multi-parameter model outputs
see Extended Data Table1, for single-parameter relationships, Fig.2.
The second pair of models include the same environmental pre-
dictors (CO2, MAT, MCWD), plus their rate of change (CO2-change,
MAT-change, but not MCWD-change, as explained above), and forest
attributes that may alter how forests respond to the same environmen-
tal change(wood density, CRT), as described above. We also evaluated
the possible inclusion of a differential continent effect of each variable
in the full model. We first constructed models with only a single pre-
dictor variable, and allowed different slopes in each continent. Next,
if removal of the continent-specific slope (using stepAIC function in
R, MASS package
96
) increased model AIC then the continent-specific
slope was included in the full model for that variable. Only MCWD
showed a significant differential continent-specific slope (P < 0.001).
This implies that forests on both continents have common responses
to CO2, CO2-change, MAT, MAT-change, wood density and CRT, but
respond differently to differences in MCWD. This may be because wet-
adapted species are much rarer in Africa than in Amazonia as a result of
large differences in past climate variation34. Last, we allowed different
intercepts for the two continents to potentially account for differing
biogeographical or other continent-specific factors. For the carbon
loss model, we applied the same continent-specific effects for slope
as for the carbon gain model. Carbon loss values were transformed
using a power-law transformation (λ=0.361) to meet normality criteria.
For both carbon gains and losses we parameterized a global model
including the significant continent-specific effect of MCWD, select-
ing the most parsimonious simplified model using all-subsets regres-
sion
97,98
. To do so, we first generated a set of models with all possible
combinations (subsets) of fixed effect terms in the global model using
the dredge function of the MuMIn package in R
99
. We then chose the
best-ranked simplified model based on the second-order Akaike Infor
-
mation Criterion (known as AICc), hereafter called the ‘minimum
adequate carbon gain/loss model’, reported in Table2. The minimum
adequate models are: carbon gains=intcp×continent + a×CO
2
-change
+ b×MAT + c×MAT-change + d×MCWD×continent + e× wood density
(model 3); carbon losses=intcp + a×CO2-change + b×MAT-change +
c×MCWD + d×CRT (model 4). Wood density was retained in the carbon
gain model, probably because growth is primarily affected by resource
availability, whereas CRT was retained in the carbon loss model, prob-
ably because losses are primarily affected by how long fixed carbon is
retained in the system.
Table2 presents model coefficients of the best-ranked gain model
and best-ranked loss model selected using all-subsets regression.
These best-ranked gain and loss models have weights of 0.310 and
0.132 respectively, which is almost double the weight of the second-
rank models (0.152 and 0.075 respectively). In Supplementary Table2
we also used the model.avg function of the MuMIn package to calcu-
late a weighted mean of the coefficients of the models that together
represent a cumulative weight-sum of 0.95 (that is, a 95% confidence
subset). Supplementary Table2 (model-averaged) and Table2 (best-
ranked) model parameters are very similar. Supplementary Tables3
and 4 report the complete sets of carbon gains and loss models that
contribute to the model average results.
The model-average results show the same continental differences in
sensitivity to environmental variables as the best-ranked models. From
1 January 2000 to 31 December 2014, carbon gains increased owing to
CO2-change (+3.7% in both the averaged and the best-ranked models,
both continents), whereas temperature rises led to a decline in gains,
which especially had an effect in the Amazon (−1.14% and −1.07% due to
MAT and MAT-change together in the averaged and best-ranked model
respectively). Finally, both model-average and best-rankedmodels
result in similar predictions of the net carbon sink over the 1 January
1983 to 31 December 2039 period: the future net sink trend in Africa is
−0.004 and −0.003 in the best-ranked and averaged models, respec-
tively; in Amazonia the future net sink trend is −0.013 and −0.011 in
the best-ranked and averaged models, respectively. The Amazon sink
reaches zero in 2041 using model-averaged parameters compared to
2035 using the best-ranked models.
Estimating future predictor variables to 2040
To calculate future modelled trends in carbon gains and losses
(Fig.3), we first estimated annual records of the predictor variables
(CO2-change, MAT, MAT-change, MCWD, wood density and CRT) to 31
December 2039 (Extended Data Fig.5).
To do so, we first calculated annual records for the period of the
observed trends for each plot location (that is, from 1 January 1983 to 31
December 2014 in Africa and 1 January 1983 to mid-2011 in Amazonia).
For CO2-change, MAT, MAT-change and MCWD we extracted monthly
records as described in theMethods section ‘Predictor variable esti-
mates (1983–2014)’. For wood density and CRT we interpolated to a
0.1-year period within each census interval (as in Fig.1). Then, we cal-
culated the mean annual value of each predictor variable from the 244
plot locations in Africa, and separately the mean annual value of each
predictor variable from the 321 plot locations in Amazonia (solid lines in
Extended Data Fig.5). For each predictor variable, we calculated annual
records of upper and lower confidence intervals by respectively adding
and subtracting 2σ to the mean of each annual value (shaded area in
Extended Data Fig.5). Second, for each predictor variable we param-
eterized a linear model for each continent using the annual records for
the period of the observed trends. Then for each predictor variable,
the continent-specific linear regression models were used to estimate
predictor variables for each plot location from 1 January 2015 to 31
December 2039 in Africa and from mid-2011 to 31 December 2039 in
the Amazon (dotted lines in Extended Data Fig.5).
Estimating future carbon gain, loss and sinktrends
We used the minimum adequate models (Table2) to predict annual
records of carbon gain, carbon loss and the carbon sink for the plot
networks in Africa and Amazonia over the period 1983 through to 2040
(Fig.3). We extracted predicted carbon gain and loss values using the
mean annual records for each predictor variable (predictSE.lme func-
tion, AICcmodavg package100). Upper and lower confidence intervals
were calculated accounting for uncertainties in the model (both fixed
and random effects) and predictor variables using the 2σ upper and
lower confidence interval for each predictor variable (using predictSE.
lme). Finally, the net carbon sink was calculated by subtracting the
losses from the gains. To obtain sink values in the future, reported in
Table1, annual per unit area sink predictions (from Fig.3) were aver-
aged over each decade and multiplied by the future forest area, as
described above.
To test the sensitivity of the future predictions in Fig.3, we reran the
analysis by modifying future trajectories of predictor variables one
at a time, while keeping all others the same, to assess the mean C sink
over 2010–15 and 2030 (averaging at 2030 is not necessary as trends
in MAT-change and MCWD, which largely drive modelled inter-annual
variability, are estimated as smooth trends in the future). For each pre-
dictor variable, we exploredthe potential impacts of the likely bounds
of possibility: (1) by taking the steepest slope of either continent from
the extrapolated trends, doubling this slope and applying it on both
continents; and (2) by taking the steepest slope of either continent
from the extrapolated trends, taking the additive inverse of this slope
and applying it on both continents. These bounds represent deviations
of >2σ from observed trends. Change in MAT also alters MAT-change,
so we present the sensitivity of both parameters together.
Additionally, for CO
2
-change and MAT, we also calculated future
slopes under three future Representative Concentration Pathway (RCP)
scenarios38 with different radiative forcing in 2100: RCP2.6, RCP4.5 and
RCP8.5. Future RCP CO2-change slopes (ppm yr−1) were calculated using
RCP CO
2
concentration data for the years between 2015 and 2030 inclu-
sive. Future RCP MAT and MAT-change slopes were obtained from plot-
specific MAT values extracted from downscaled ~1-km
2
resolution data
for current80 and future51 climate from WorldClim, and averaged over
19 CMIP5 models. We subtracted the mean 2040–2060 climate MAT
(that is, 2050) from the mean 1970–2000 climate MAT (that is, 1985),
divided by 65 years to give the annual rate of change. We then calculated
a mean slope over all plots per continent. Finally, to avoid mismatches
between RCP-derived values of CO
2
and MAT and the observed records,
we removed any difference in intercept between the RCP trends and
observed trends, so that the RCP trends were a continuation of the
end-point of the observed trajectory (31 December 2014). We did not
estimate the sensitivity of MCWD under the RCP scenarios, because the
mean of theCMIP5 models do not show drought trends for our forest
Article
plot networks, unlike rain gauge data for the recent past41,42, and thus
would show little or no sensitivity to MCWD. For each modified slope,
Supplementary Table5 reports the absolute decline in the sink in each
continent in 2030 compared to the 2010–15 mean sink. This shows that
the future sink strength is sensitive to future environmental conditions,
but within both RCP scenarios and our bounds of possibility we show a
decline in the sink strength in both continents over the 2020s.
Reporting summary
Further information on research design is available in theNature
Research Reporting Summary linked to this paper.
Data availability
Source data to generate figures and tables are available from https://
doi.org/10.5521/Forestplots.net/2019_1.
Code availability
R code to generate figures and tables is available from: https://doi.
org/10.5521/Forestplots.net/2019_1
51. Hijmans, R. J., Cameron, S. E., Parra, J. L., Jones, P. G. & Jarvis, A. Very high resolution
interpolated climate surfaces for global land areas. Int. J. Climatol. 25, 1965–1978 (2005).
52. Lopez-Gonzalez, G., Lewis, S. L., Burkitt, M. & Phillips, O. L. ForestPlots.net: a web
application and research tool to manage and analyse tropical forest plot data. J. Veg. Sci.
22, 610–613 (2011).
53. Lopez-Gonzalez, G., Lewis, S. L., Burkitt, M., Baker, T. R. & Phillips, O. L. ForestPlots.net
Database http://www.forestplots.net (2009).
54. Sheil, D. & Bitariho, R. Bwindi Impenetrable Forest TEAM Site https://www.wildlifeinsights.
org/team-network, TEAM-DataPackage-20151201235855_1254 (2009).
55. Kenfack, D. Korup National Park TEAM Site https://www.wildlifeinsights.org/team-network,
TEAM-DataPackage-20151201235855_1254 (2011).
56. Rovero, F., Marshall, A. & Martin, E. Udzungwa TEAM Site https://www.wildlifeinsights.org/
team-network, TEAM-DataPackage-20151130235007_5069 (2009).
57. Hockemba, M. B. N. Nouabalé Ndoki TEAM Site https://www.wildlifeinsights.org/team-
network, TEAM-DataPackage-20151201235855_1254 (2010).
58. Anderson-Teixeira, K. J. etal. CTFS-ForestGEO: a worldwide network monitoring forests in
an era of global change. Glob. Change Biol. 21, 528–549 (2015).
59. Gourlet-Fleury, S. etal. Tropical forest recovery from logging: a 24 year silvicultural
experiment from Central Africa. Phil. Trans. R. Soc. Lond. B 368, 20120302 (2013).
60. Claeys, F. etal. Climate change would lead to a sharp acceleration of Central
African forests dynamics by the end of the century. Environ. Res. Lett. 14, 044002
(2019).
61. Chave, J. etal. Improved allometric models to estimate the aboveground biomass of
tropical trees. Glob. Change Biol. 20, 3177–3190 (2014).
62. R Development Core Team R: A Language and Environment for Statistical Computing
http://www.R-project.org/ (2015).
63. Lopez-Gonzalez, G., Sullivan, M. & Baker, T. BiomasaFP. R package version 0.2.1
http://www.forestplots.net/en/resources/analysis (2017).
64. Phillips, O., Baker, T., Brienen, R. & Feldpausch, T. RAINFOR ield manual for plot
establishment and remeasurement. http://www.rainfor.org/upload/ManualsEnglish/
RAINFOR_ield_manual_version_2016.pdf(Univ. Leeds, 2016).
65. Talbot, J. etal. Methods to estimate aboveground wood productivity from long-term
forest inventory plots. For. Ecol. Manage. 320, 30–38 (2014).
66. Sullivan, M. J. P. etal. Field methods for sampling tree height for tropical forest biomass
estimation. Methods Ecol. Evol. 9, 1179–1189 (2018).
67. Feldpausch, T. R. etal. Tree height integrated into pantropical forest biomass estimates.
Biogeosciences 9, 3381–3403 (2012).
68. Chave, J. etal. Towards a worldwide wood economics spectrum. Ecol. Lett. 12, 351–366
(2009).
69. Zanne, A . E. etal. Towards a Worldwide Wood Economics Spectrum https://doi.
org/10.5061/dryad.234 (Dryad Digital Repository, 2009).
70. Martin, A. R., Doraisami, M. & Thomas, S. C. Global patterns in wood carbon
concentration across the world’s trees and forests. Nat. Geosci. 11, 915–920 (2018).
71. Kohyama, T. S., Kohyama, T. I., Sheil, D. & Rees, M. Deinition and estimation of vital rates
from repeated censuses: choices, comparisons and bias corrections focusing on trees.
Methods Ecol. Evol. 9, 809–821 (2018).
72. Bates, D., Maechler, M., Bolker, B. & Walker, S. lme4: linear mixed-effects models using
Eigen and S4. R package version 1.0-4http://www.inside-r.org/packages/lme4/
versions/1-0-4 (2013).
73. Fox, J. Applied Regression Analysis and Generalized Linear Models 2nd edn (Sage
Publishing, 2008).
74. Chave, J. etal. Assessing evidence for a pervasive alteration in tropical tree communities.
PLoS Biol. 6, 0455–0462 (2008).
75. Yuen, J. Q., Ziegler, A. D., Webb, E. L. & Ryan, C. M. Uncertainty in below-ground carbon
biomass for major land covers in Southeast Asia. For. Ecol. Manage. 310, 915–926 (2013).
76. Aragão, L. E. O. C. etal. Spatial patterns and ire response of recent Amazonian droughts.
Geophys. Res. Lett. 34, L07701 (2007).
77. Aragão, L. E. O. C. etal. Environmental change and the carbon balance of Amazonian
forests. Biol. Rev. Camb. Phil. Soc. 89, 913–931 (2014).
78. Tans, P. & Keeling, R. Trends in Atmospheric Carbon Dioxide for Mauna Loa, Hawaii
http://www.esrl.noaa.gov/gmd/ccgg/trends/ (ESRL, 2016).
79. Harris, I., Jones, P. D., Osborn, T. J. & Lister, D. H. Updated high-resolution grids of monthly
climatic observations – the CRU TS3.10 Dataset. Int. J. Climatol. 34, 623–642 (2014).
80. Fick, S. E. & Hijmans, R. J. WorldClim 2: new 1-km spatial resolution climate surfaces for
global land areas. Int. J. Climatol. 37, 4302–4315 (2017).
81. Ramirez-Villegas, J. & Jarvis, A. Downscaling Global Circulation Model Outputs: The Delta
Method. Decision and Policy Analysis Working Paper No. 1https://cgspace.cgiar.org/
handle/10568/90731 (International Center for Tropical Agriculture (CIAT), 2010).
82. Schneider, U. etal. GPCC Full Data Reanalysis Version 6.0 at 0.5°: Monthly Land-Surface
Precipitation from Rain-Gauges built on GTS-based and Historic Data https://opendata.
dwd.de/climate_environment/GPCC/html/fulldata_v6_doi_download.html (Global
Precipitation Climatology Centre (GPCC) at Deutscher Wetterdienst, 2011).
83. Sun, Q. etal. Review of global precipitation data sets: data sources, estimation, and
intercomparisons. Rev. Geophys. 56, 79–107 (2017).
84. Huffman, G. J. etal. The TRMM Multisatellite Precipitation Analysis (TMPA): quasi-global,
multiyear, combined-sensor precipitation estimates at ine scales. J. Hydrometeorol. 8,
38–55 (2007).
85. Kume, T. etal. Ten-year evapotranspiration estimates in a Bornean tropical rainforest.
Agric. For. Meteorol. 151, 1183–1192 (2011).
86. Zelazowski, P., Malhi, Y., Huntingford, C., Sitch, S. & Fisher, J. B. Changes in the potential
distribution of humid tropical forests on a warmer planet. Phil. Trans. R. Soc. A 369,
137–160 (2011).
87. James, R., Washington, R. & Rowell, D. P. Implications of global warming for the climate of
African rainforests. Phil. Trans. R. Soc. Lond. B 368, 20120298 (2013).
88. Jung, M. etal. Recent decline in the global land evapotranspiration trend due to limited
moisture supply. Nature 467, 951–954 (2010).
89. Jung, M. etal. Global patterns of land-atmosphere luxes of carbon dioxide, latent heat,
and sensible heat derived from eddy covariance, satellite, and meteorological
observations. J. Geophys. Res. 116, https://doi.org/10.1029/2010JG001566 (2011).
90. Lloyd, J. & Farquhar, G. D. The CO2 dependence of photosynthesis, plant growth
responses to elevated atmospheric CO2 concentrations and their interaction with soil
nutrient status. I. General principles and forest ecosystems. Funct. Ecol. 10, 4–32 (1996).
91. Aspinwall, M. J. etal. Convergent acclimation of leaf photosynthesis and respiration to
prevailing ambient temperatures under current and warmer climates in Eucalyptus
tereticornis. New Phytol. 212, 354–367 (2016).
92. Bonal, D., Burban, B., Stahl, C., Wagner, F. & Hérault, B. The response of tropical
rainforests to drought—lessons from recent research and future prospects. Ann. For. Sci.
73, 27–44 (2016).
93. Quesada, C. A. etal. Variations in chemical and physical properties of Amazon forest soils
in relation to their genesis. Biogeosciences 7, 1515–1541 (2010).
94. Baker, T. R., Swaine, M. D. & Burslem, D. F. R. P. Variation in tropical forest growth rates:
combined effects of functional group composition and resource availability. Perspect.
Plant Ecol. Evol. Syst. 6, 21–36 (2003).
95. Pinheiro, J. C. & Bates, D. M. Mixed-Effects Models in S and S-PLUS 1st edn 528 (Springer,
2000).
96. Venables, W. N. & Ripley, B. D. Modern Applied Statistics with S 4th edn 498 (Springer,
2002).
97. Olejnik, S., Mills, J. & Keselman, H. Using Wherry’s adjusted R2 and Mallow’s Cp for model
selection from all possible regressions. J. Exp. Educ. 68, 365–380 (2000).
98. Whittingham, M. J., Stephens, P. A., Bradbury, R. B. & Freckleton, R. P. Why do we still use
stepwise modelling in ecology and behaviour? J. Anim. Ecol. 75, 1182–1189 (2006).
99. Bar toń, K. MuMIn: Multi-Model Inference. Tools for performing model selection and
model averaging. R package version 1.43.6 (2019).
100. Gelman, A. & Hill, J. Data Analysis Using Regression and Multilevel/Hierarchical Models
(Cambridge Univ. Press, 2007).
101. Mayaux, P., De Grandi, G. & Malingreau, J.-P. Central African forest cover revisited: a
multisatellite analysis. Remote Sens. Environ. 71, 183–196 (2000).
102. Mayaux, P. etal. The Land Cover Map for Africa in the Year 2000 GLC2000 database,
https://forobs.jrc.ec.europa.eu/products/glc2000/products.php (European Commission
Joint Research Centre, 2003).
Acknowledgements This paper is a product of the African Tropical Rainforest Observatory
Network (AfriTRON), curated at ForestPlots.net. AfriTRON has been supported by numerous
people and grants since its inception. We sincerely thank the people of the many villages and
local communities who welcomed our ield teams and without whose support this work would
not have been possible: Sierra Leone (villages: Barrie, Gaura, Koya, Makpele, Malema, Nomo,
Tunkia; teams in protected areas: the Gola Rainforest National Park), Liberia (villages: Garley
town, River Gbeh, Glaro Freetown), Ghana (villages: Nkwanta, Asenanyo, Bonsa, Agona,
Boekrom, Dadieso, Enchi, Dabiasem, Mangowase, Draw, Fure, Esuboni, Okumaninin, Kade,
Asamankese, Tinte Bepo, Tonton), Nigeria (Oban village), Gabon (villages: Ekobakoba,
Mikongo, Babilone, Makokou, Tchimbele, Mondah, Ivindo, Ebe, Ekouk, Oveng, Sette Cama;
teams in protected areas: Ivindo National Park, Lope National Park, Waka National Park; teams
in concessions: Ipassa station, Kingele station, Leke/Moyabi Rougier Forestry Concession),
Cameroon (villages: Campo, Nazareth, Lomié, Djomédjo, Alat-Makay, Somalomo, Deng Deng,
Eyumojok, Mbakaou, Myere, Nguti, Bejange, Kekpane, Basho, Mendhi, Matene, Mboh,
Takamanda, Obonyi, Ngoïla; teams in protected areas: Ejagham forest reserve), Democratic
Republic of Congo (villages: Yoko, Yangambi, Epulu, Monkoto), Republic of Congo (villages:
Bomassa, Ekolongouma, Bolembe, Makao, Mbeli, Kabo, Niangui, Ngubu, Goualaki, Essimbi).
We thank the ield assistants whose expertise and enthusiasm is indispensable to successful
ieldwork, including: M. E. Abang, U. P. Achui, F. Addai, E. J. Agbachon, J. Agnaka, A. J. Akaza,
G. Alaman, G. Alaman, A. E. Alexander, K. Allen, M. Amalphi, D. Amandus, J. Andju,
L. A. Limbanga, S. Asamoah, T. M. Ashu, M. Ashu, J. Asse, B. Augustine, H. Badjoko, M. Balimu,
J. Baviogui-Baviogui, S. Benteh, A. Bertrand, A. Bettus, A. Bias, A. Bikoula, A. Bimba,
P. Bissiemou, M. Boateng, E. Bonyenga, M. B. Ekaya, G. Bouka, J. Boussengui, D. B. Ngomo,
C. Chalange, S. Chenikan, J. Dabo, E. Dadize, T. Degraft, J. Dibakou, J.-T. Dikangadissi,
P. Dimbonda, E. Dimoto, C. Ditougou, D. Dorbor, M. Dorbor, V. Droissart, K. Duah, E. Ebe, O. J. Eji,
E. B. Ekamam, J.-R. Ekomindong, E. J. Enow, H. Entombo, E. M. Ernest, C. Esola, J. Essouma, A.
Gabriel, N. Genesis, B. Gideon, A. Godwin, E. Grear, D. J. Grear, M. Ismael, M. Iwango, M. Iyafo,
N. Kamdem, B. Kibinda, A. Kidimbu, E. Kimumbu, J. Kintsieri, C. K. Opepa, A. Kitegile, T. Komo,
P. Koué, A. Kouanga, J. J. Koumikaka, I. Liengola, E. Litonga, L. Louvouando, O. Luis, N. M. Mady,
F. Mahoula, A. Mahundu, C. A. Mandebet, P. Maurice, K. Y. Mayossa, R. M. Nkogue, I. D. Mbe, C.
Mbina, H. Mbona, A. Mboni, A. Mbouni, P. Menzo, M. Menge, A. Michael, A. Mindoumou, J.
Minpsa, J. P. Mondjo, E. Mounoumoulossi, S. Mpouam, T. Msigala, J. Msirikale, S. Mtoka, R.
Mwakisoma, D. Ndong-Nguema, G. Ndoyame, G. Ngongbo, F. Ngowa, D. Nguema, L. Nguye, R.
Niangadouma, Y. Nkrumah, S. Nshimba, M. N. Mboumba, F. N. Obiang, L. Obi, R. Obi, E. L.
Odjong, F. Okon, F. Olivieira, A. L. Owemicho, L. Oyeni-Amoni, A. Platini, P. Ploton, S. Quausah,
E. Ramazani, B. S. Jean, L. Sagang, R. Salter, A. Seki, D. Shirima, M. Simo, I. Singono, A. E. Tabi,
T. G. Tako, N. G. Tambe, T. Tcho, A. Teah, V. Tehtoe, B. J. Telephas, M. L. Tonda, A. Tresor, H.
Umenendo, R. Votere, C. K. Weah, S. Weah, B. Wursten, E. Yalley, D. Zebaze, L. Cerbonney, E.
Dubiez, H. Moinecourt, F. Lanckriet, S. Samai, M. Swaray, P. Lamboi, M. Sullay, D. Bannah, I.
Kanneh, M. Kannah, A. Kemokai, J. Kenneh and M. Lukulay. For logistical and administrative
support, we are indebted to international, national and local institutions: the Forestry
Department of the Government of Sierra Leone, the Conservation Society of Sierra Leone, the
Royal Society for the Protection of Birds (RSPB, UK), The Gola Rainforest National Park (Sierra
Leone), the Forestry Development Authority of the Government of Liberia (FDA), the University
of Liberia, the Forestry Commission of Ghana (FC), the Forestry Research Institute of Ghana
(FORIG), University of Ibadan (Nigeria), the University of Abeokuta (Nigeria), the Ministère des
Eaux, Forêts, Chasse et Pêche (MEFCP, Central African Republic), the Institut Centrafricain de
Recherche Agronomique (ICRA, Central African Republic), The Service de Coopération et
d’Actions Culturelles (SCAC/MAE, Central African Republic), The University of Bangui (Central
African Republic), the Société Centrafricaine de Déroulage (SCAD, Central African Republic),
the University of Yaounde I (Cameroon), the National Herbarium of Yaounde (Cameroon), the
University of Buea (Cameroon), Bioversity International (Cameroon), the Ministry of Forests,
Seas, Environment and Climate (Gabon), the Agence Nationale des Parcs Nationaux de Gabon
(ANPN), Institut de Recherche en Écologie Tropicale du Gabon, Rougier-Gabon, the Marien
Ngouabi University of Brazzaville (Republic of Congo), the Ministère des Eaux et Forêts
(Republic of Congo), the Ministère dela Rercherche Scientiique et de l’Innovation
Technologique (Republic of Congo), the Nouabalé-Ndoki Foundation (Republic of Congo),
WCS-Congo, Salonga National Park (Democratic Republic of Congo), The Centre de Formation
et de Recherche en Conservation Forestière (CEFRECOF, Epulu, Democratic Republic of
Congo), the Institut National pour l’Étude et la Recherche Agronomiques (INERA, Democratic
Republic of Congo), the École Régionale Postuniversitaire d’Aménagement et de Gestion
intégrés des Forêts et Territoires tropicaux (ERAIFT Kinshasa, Democratic Republic of Congo),
WWF-Democratic Republic of Congo, WCS-Democratic Republic of Congo, the Université de
Kisangani (Democratic Republic of Congo), Université Oficielle de Bukavu (Democratic
Republic of Congo), Université de Mbujimayi (Democratic Republic of Congo), le Ministère de
l'Environnement et Développement Durable (Democratic Republic of Congo), the FORETS
project in Yangambi (CIFOR, CGIAR and the European Union; Democratic Republic of Congo),
the Lukuru Wildlife Research Foundation (Democratic Republic of Congo), Mbarara University
of Science and Technology (MUST, Uganda), WCS-Uganda, the Uganda Forest Department, the
Commission of Central African Forests (COMIFAC), the Udzungwa Ecological Monitoring
Centre (Tanzania) and the Sokoine University of Agriculture (Tanzania). We thank C. Chatelain
(Geneva Botanic Gardens) for access to the African Plants Database. Grants that have funded
the AfriTRON network including data in this paper are: a European Research Council Advanced
Grant to O.L.P. and S.L.L. (T-FORCES; 291585; Tropical Forests in the Changing Earth System), a
NERC grant to O.L.P., Y.M., and S.L.L. (NER/A/S/2000/01002), a Royal Society University
Research Fellowship to S.L.L., a NERC New Investigators Grant to S.L.L., a Philip Leverhulme
Award to S.L.L., a European Union FP7 grant to E.G. and S.L.L. (GEOCARBON; 283080), Valuing
the Arc Leverhulme Program Grant to Andrew Balmford and S.L.L., a Natural Environment
Research Council (NERC) Consortium Grant to Jon Lloyd and S.L.L. (TROBIT; NE/D005590/),
the Gordon and Betty Moore Foundation to L.J.T.W and S.L.L., the David and Lucile Packard
Foundation to L.J.T.W. and S.L.L., the Centre for International Forestry Research to T.S. and
S.L.L. (CIFOR), and Gabon’s National Parks Agency (ANPN) to S.L.L. W.H. was funded by
T-FORCES and the Brain programme of the Belgian Federal Government (BR/132/A1/AFRIFORD
grant to Olivier Hardy and the BR/143/A3/HERBAXYLAREDD grant to H.B.). O.L.P., S.L.L., M.J.P.S,
A.E.-M., A.L., G.L.-G., G.P. and L.Q. were supported by T-FORCES. Eight plots (codes ANK, IVI,
LPG, MNG) included in AfriTRON are also part of the Global Ecosystem Monitoring network
(GEM). Additional African data were included from the consortium MEFCP-ICRA-CIRAD (Centre
de Coopération Internationale en Recherche Agronomique pour le Développement), the
Tropical Ecology Assessment and Monitoring Network (TEAM), and the Forest Global Earth
Observatory Network (ForestGEO; formerly the Center for Tropical Forest Science, CTFS). The
TEAM network is a collaboration between Conservation International, the Missouri Botanical
Garden, the Smithsonian Institution and the Wildlife Conservation Society, and funded by the
Gordon and Betty Moore Foundation and other donors. The ForestGEO Network is a
collaboration between the Smithsonian Institution, other federal agencies of the United States,
the Wildlife Conservation Society (WCS) and the World Wide Fund for Nature (WWF), and
funded by the US National Science Foundation and other donors. The paper was made
possible by the RAINFOR network in Amazonia, with multiple funding agencies and hundreds
of investigators working in Amazonia, acknowledged in ref. 6, providing comprehensive
published data and code and assisting in the onward analysis of their data; see ref. 6. Data from
AfriTRON and RAINFOR are stored and curated by ForestPlots.net, a long-term cyber-
infrastructure initiative hosted at the University of Leeds that unites permanent plot records
and their contributing scientists from the world’s tropical forests. The development of
ForestPlots.net and curation of most data analysed here was funded by many sources,
including grants to O.L.P. (principally from ERC AdG 291585 ‘T-FORCES’, NERC NE/B503384/1
and the Gordon and Betty Moore Foundation 1656 ‘RAINFOR’), T.R.B. (the University of Leeds
contribution to ‘AMAZALERT’, NERC (NE/I028122/1) with T. Pennington, the Gordon and Betty
Moore Foundation (‘MonANPeru’) and a NERC Impact Accelerator grant for the initial
development of the BiomasaFP R package), E.G. (‘GEOCARBON’ and NE/F005806/1
‘AMAZONICA’) and S.L.L. (Royal Society University Research Fellowship, NERC New
Investigators Award, NERC NE/P008755/1). We acknowledge the contributions of the
ForestPlots.net developers (M. Burkitt, G. Lopez-Gonzalez) and the steering committee (T.R.B.,
A.L., S.L.L., O.L.P., L.Q., E. N. H. Coronado and B. S. Marimon) for advice on database
development and management.
Author Contributions S.L.L. conceived and managed the AfriTRON forest plot recensus
programme, O.L.P., T.C.H.S., L.J.T.W. and Y.M. contributed to its development. W.H., S.L.L.,
O.L.P., B.S. and M.J.P.S. developed the study. W.H., S.L.L., O.L.P., K.A.-B., H.B., A.C.-S., C.E.N.E.,
S.F., D.S., B.S., T.C.H.S., S.C.T., K.A .A., S.A.-B., C.A.A., T.R.B., L.F.B., F. Baya, S.K.B., F. Benedet,
R.B., Y.E.B., P. Boeckx, P. Boundja, T.B., E.C., G.B.C., C.J.C., M.C., J.A.C., D.C., A.K.D., G.C.D., T.d.H.,
M.D.K., J.-L.D., T.R.F., A.F., E.G.F., M.G., C.G., S.G.-F., J.S.H., A.C.H., D.J.H., T.B.H., M.B.N.H., A.H.,
S.A.I., K.J.J., T.J., E.K.Y., E.K., D.K., M.E.L., J.A.L., J.L., J.C.L., J.-R.M., Y.M., A.R.M., J.M., E.H.M.,
F.M.M., V.P.M., V.M., E.T.A.M., S.M., J.M.M., P.K.T.M., N.N.B., L .O., F.E., K.S.-H.P., A.D.P., J.R.P., L.Q.,
J.R., F.R., M.D.S., H.T., J. Talbot, J. Taplin, D.M.T., D.W.T., B.T., J.T.M., D.T., P.M.U., G.v.d.H., H.V., J.V.,
L.J.T.W., S.W., H.W., J.T.W. and L.Z. contributed data (with larger ield contributions by S.L.L.,
W.H., A.C.-S., B.S., H.T., A.K.D., C.E.N.E., J.M.M., K.A.-B. and S.F.). O.L.P., T.R.B., S.L.L. and G.L.-G.
conceived and managed forestplots.net; O.L.P., T.R.B., S.L.L., E.G., G.L.-G., G.C.P., A.L., R.J.W.B.,
T.R.F. and M.J.P.S. developed it. W.H., M.J.P.S., S.L.L., O.L.P., R.J.W.B., A.L., G.L.-G., A.E.-M., A.K.,
E.G., T.R.B., A.C.B. and G.C.P. contributed analysis tools. W.H. and S.L.L. analysed the data (with
important contributions from M.J.P.S.). S.L.L. and W.H. wrote the paper. All co-authors read and
approved the manuscript (with important insights provided by O.L.P., S.F., R.J.W.B., E.G., H.B.,
D.S., M.J.P.S., S.G.-F., P.B., H.V. and S.C.T).
Competing interests The authors declare no competing interests.
Additional information
Supplementary information is available for this paper at https://doi.org/10.1038/s41586-020-
2035-0.
Correspondence and requests for materials should be addressed to W.H.
Reprints and permissions information is available at http://www.nature.com/reprints.
Article
Extende d Data Fig. 1 | Map s howing the loc ations of the 2 44 plots inc luded in
this study. Dark green represe nts all lowland clo sed-canopy fore sts,
submont ane forests and fore st-agriculture mos aics; light gree n shows swamp
forests and m angroves, blue circl es represen t plot clusters, re ferred to by
three-let ter codes (see S upplementa ry Table1 for the full list of plots). Clust ers
<50 km apar t are shown as one p oint for display only, with t he circle size
corresp onding to samplin g effort in ter ms of hectar es monitored. L and cover
data are from T he Land Cover Map for A frica in the Year 200 0 (GLC2000
database)101,102. T his map was creat ed using the R st atistical pla tform, version
3.2. 1 (ref. 62), which is under the GNU Pub lic License.
Extende d Data Fig. 2 | Lon g-term abovegrou nd carbon dyn amics of 24 4
African structurallyintactold-growth tropical forest inventory plots.
Points in the s catterpl ots indicate th e mid-census in terval date, w ith horizont al
bars conne cting the st art and end date for e ach census in terval for net
aboveground biomass carbon change (a), carbon gains (fro m woody
producti on from tree grow th and newly recr uited stems) (b), and car bon losses
(from tree mor tality) (c). Examples of t ime series for th ree individual p lots are
shown in purp le, yellow and gree n. Associat ed histogram s show the
distribu tion of the plot-level net ab oveground bioma ss carbon (with a t hree-
paramete r Weibull probabilit y density dis tribution f itted in blue , showing that
the carbo n sink is signif icantly larger t han zero; one-taile d t-test: P<0.001)(d),
carbon gai ns (e) and carbon loss es (f).
Article
Extende d Data Fig. 3 | AI C from correla tions bet ween the carbon g ain in
tropical fo rest inventor y plots and chan ges in atmos pheric CO2,
tempera ture (MAT) or droug ht (MCWD), each ca lculated over ever-
longer prior intervals. Panels show th eAIC from linear mi xed effects
models of c arbon gains fro m 565 Africa n and Amazonianplot s and
correspondingchanges in atmospheric CO2 (CO2-c hang e) (a), M AT (M AT-
chan ge) (b), and drought(MCWD-change) (c). For CO2 the AIC minimum wa s
obser ved when predic ting the carbo n gain from the chan ge in CO2 calculated
over a 56-year-long prio r interval leng th. We use this le ngth of time to c alculate
our CO2-chan ge parameter. Such a value i s expected b ecause forest s tands will
respond mo st strongly to CO2 w hen most indiv iduals have grown und er the new
rapidly chan ging conditi on, which should be a t its maximum at a t ime
approximate ly equivalent to th e CRT of a forest stan d30,90 (mean of 62 years in
this poole d African an d Amazonian dat aset). For MAT the AIC mini mum was
5 years, which we u se as the prior in terval to calcu late our MAT-change
parameter. This length is consistent with experiments showing temperature
acclimati on of leaf- and plant-level photos ynthetic and re spiration pro cesses
over approximately half-decadal timescales31 ,91. For MCWD t he AIC minimum i s
not obviou s, while the slope of t he correlation, s hown in panel d, show s no
overall trend an d oscillates be tween posit ive or negative valu es, meaning th ere
is no relatio nship betwee n carbon gain s and the change in MC WD over intervals
longer than 1 ye ar; therefore MCWD -change is not i ncluded in our mod els. This
result sug gests that on ce a drought ends, i ts impact on t ree growth fade s
rapidly, as seen in o ther studies14,9 2. Furthe rmore,in the moist t ropics wet-
season r ainfall is expec ted to recharge s oil water, and hence lag ged impact s of
droughts are not expected.
Extende d Data Fig. 4 | Pote ntial forest dy namics-re lated driver s of carbon
gains and losses in structurally intact old-growthAfrican and Amazonian
tropical forest inventory plots. The above ground carbon g ains, from wood y
producti on (a, b), and abovegro und carbon los ses, from tre e mortalit y (c, d),
are plotte d against the CRT, and wood de nsity for Afr ican (blue) and
Amazonia n (brown) inventory p lots. Linear mi xed effect s models were
perform ed with censu s intervals (n=1,56 6) nested within p lots (n=565) to
avoid pseudo -replication, u sing an empiric ally derived weig hting based o n
interval l ength and plot a rea (seeMethods). Sign ificant re gression line s from
the linear mi xed effect s modelsfor the comple te dataset are s hown as a solid
line; non-sig nificant r egression s are shown as a dashe d line. Each dot
represen ts a time-weigh ted mean plot-level value; th e shading of the dot
represen ts total monit oring lengt h, with empty c ircles corresp onding to plots
monitored for ≤ 5 years and solid ci rcles for plots moni tored for >20 years.
Carbon los s data are prese nted untransfo rmed for compari son with carb on
gains; linea r mixed effect s models on tra nsformed data t o fit normali ty
assumpti ons do not change th e signific ance of the resul ts. Note that CR T is
calculate d differently for th e carbon gains a nd losses mod els (seeMethods).
Article
Extende d Data Fig. 5 | Trends in pr edictor var iables use d to estimat e long-
term tren ds in abovegroun d carbon gain s, carbon los ses and the re sulting
net carbon sink in African and Amazonianstructurally intact old-
growthtropical forest inventoryplot networks. Mean annual CO2-change (a),
MAT (b), MAT-change (c), MCWD (d), CRT (e) and wood density (f) for Af rican
plot locations in blue, and corresponding variables forAmazon plot locations
in brown (gl). So lid lines repres ent obser vational data w here >75% of the plots
were monitore d; long-dashed lin es are plot mean s where <75% of plots were
monitored . Dotted line s are future values e stimated fro m linear trends f rom
the 1 Januar y 1983 to 31 Decemb er 2014 (Africa) or 1 Ja nuary 1983 to mid-2011
(Amazon) dat a (slope and P value repo rted in each p anel), seeMethods for
details . Upper and lower con fidence in tervals (shaded a rea) for the past are
calculate d by respecti vely adding and sub tracting 2 σ to the mean of eac h
annual value . Upper and lower con fidence in tervals for the fu ture (Africa: 1
January 2 015 to 31 Decemb er 2039; Amazoni a: mid-2011 to 31 Decemb er 2039)
were estima ted by adding and su btracting 2 σ from the slope of t he regressio n
model.
Extende d Data Fig. 6 | Th e change in carb on losses ver sus CRT of long-t erm
structu rally intac t old-growth for estinventory plo ts in Africa a nd
Amazonia. For plots wit h two census int ervals, we calc ulated the chang e in
carbon lo sses (‘∆losses’) a s the carbon los ses (in MgCha−1yr−1) of th e second
interval m inus the carbo n losses of the f irst inter val, divided by th e difference
in mid-inter val dates. For plot s with more than t wo intervals , we calculated th e
change in car bon losses for e ach pair of subse quent interv als, then calcul ated
the plot-level mean over all p airs, weighted by t he time leng th between mi d-
interval d ates. This an alysis include s only plots with a t least two cen sus
interval s that were monitore d for ≥20 years (that is, r oughly one-third o f the
mean CRT of the p ooled Afri can and Amazon d ataset; n=116). Breakpoin t
regress ion was used to as sess the CRT le ngth below whi ch forest carbon l osses
begin to in crease. Plot s with CRT <77 years sho w a recent long-ter m increase in
carbon lo sses; longer CRT pl ots do not. Blue p oints are Afri can plots, brow n
points are A mazonian plot s.
Article
Extende d Data Fig. 7 | Trends in n et aboveground l ive biomass ca rbon,
carbon gains and carbon lossesfrom intensively monitored structurally
intact o ld-growth tro pical forest i nventory plots i n Africa. Trends are
calculate d for the last 15 year s of the twentie th century (ac) and the f irst 15
years of the t wenty-firs t century (df). Plots we re selected f rom the full data set
if their cen sus intervals c over at least 50% of th e respecti ve time windows, th at
is, they are inte nsely monitor ed (n=56 plots for 1 January 198 5 to 31 Decembe r
1999, and n=134 plots for 1 Janu ary 2000 to 31 D ecember 2014, resp ectively).
Solid lines s how mean values, a nd shading corre sponds to the 95% C I, as
calculate d in Fig.1. Dashed li nes, slopes an d P values are from lin ear mixed
effect s models, as in Fi g.1. The data show s a difference co mpared to Fig.1,
notably th e sink decline af ter about 2010 dri ven by rising carb on losses. T his is
becaus e in Fig.1 we include all availa ble plots over the 1 Jan uary 1983 to
31 Decemb er 2014 window, which inclu des clusters o f plots monitore d only in
the 2010s, of ten monitored for a s ingle census in terval, that ha d low carbon
loss and hig h carbon sink valu es.
Extende d Data Fig. 8 | Twenty-f irst-centur y trends in ab oveground biom ass
carbon lo sses from st ructurally i ntact old- growthAfrica n tropical fore st
inventory pl ots with eithe r long or shor t CRT. a, b, All plots, t hat is, as in Fig.1,
but split into a l ong-CRT group (a) and a shor t-CRT group (b), each con taining
half of the 24 4 plots. c, d, Plots areres tricted to th ose spanning >50% o f the
time window, tha t is, intensely m onitored plots , as in Extende d Data Fig.7, but
split into a lon g-CRT group (c) and a short-CRT g roup (d), each containi ng half
of the 134 p lots. Solid lin es indicate me an values, shadi ng the 95% CI, as for
Fig.1. Dashe d lines, slope s and P values are from l inear mixedeffec ts models, a s
for Fig.1. Carbo n losses incre ase at a higher r ate in the short-CRT th an the long-
CRT group of plot s, in both data sets, althou gh this increas e is not statis tically
significant.
Article
Extended Data Table 1 | Models to predict carbon gains and losses in structurally intact old-growth African and Amazonian
tropical forests
Models to predict carbon gains and losses in structurally intact old-growthAfrican and Amazonian tropical forests, including only environmental variables, show long-term trends that affect
theory-driven models of photosynthesis and respiration. Carbon loss values were normalized via power-law transformation, λ=0.361.
Extended Data Table 2 | Forest area estimates used to calculate total continental forest sink
Intact forest area for 1990, 2000, 2005 and 2010 is published in ref. 1 (that is, the total forest area minus forest regrowth). To estimate intact forest area for the other years in this table, we itted
exponential models for each continent using the published data1.
1
nature research | reporting summary October 2018
Corresponding author(s): Wannes Hubau
Last updated by author(s): 26/09/19
Reporting Summary
Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency
in reporting. For further information on Nature Research policies, see Authors & Referees and the Editorial Policy Checklist.
Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.
n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement
A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly
The statistical test(s) used AND whether they are one- or two-sided
Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested
A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons
A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient)
AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)
For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted
Give P values as exact values whenever suitable.
For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings
For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes
Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated
Our web collection on statistics for biologists contains articles on many of the points above.
Software and code
Policy information about availability of computer code
Data collection No software was used for data collection.
Data analysis All calculations were performed using the R statistical platform, version 3.2.1 (R Development Core Team, 2015) using the BiomasaFP R
package v0.2.1 (Lopez-Gonzalez, Sullivan, & Baker, 2017). Source data and R-code to generate figures and tables are available from:
https://figshare.com/s/60f48673202283421f43.
References:
Lopez-Gonzalez, G., Sullivan, M., & Baker, T. 2017. BiomasaFP package. Tools for analysing data downloaded from ForestPlots.net. R
package version 0.2.1. Available at http://www.forestplots.net/en/resources/analysis.
R Development Core Team. 2015. R: A Language and Environment for Statistical Computing. Available at http://www.R-project.org/.
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers.
We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.
Data
Policy information about availability of data
All manuscripts must include a data availability statement. This statement should provide the following information, where applicable:
- Accession codes, unique identifiers, or web links for publicly available datasets
- A list of figures that have associated raw data
- A description of any restrictions on data availability
Source data and R-code to generate figures and tables are available from: https://figshare.com/s/60f48673202283421f43. This data and code package allows
2
nature research | reporting summary October 2018
reproducing the main Figures and Table 2. All permanent inventory plot data is bound to data-use restrictions defined on Forestplots.net. To avoid unauthorized use
of data, the input files on figshare provide only the information that is necessary to reproduce the figures and the tables using the r-scripts.
Field-specific reporting
Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.
Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences
For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf
Ecological, evolutionary & environmental sciences study design
All studies must disclose on these points even when the disclosure is negative.
Study description We reconstruct the evolution of the per unit area African tropical forest carbon sink (in Mg C ha-1 yr-1) over three decades to 2015
(Figure 1). To do so, we collected, compiled and analysed data from 244 repeatedly measured permanent forest inventory plots in 11
African countries. Selected plots are situated in structurally intact old-growth forests and are part of the African Tropical Rainforest
Observation Network (AfriTRON; www.afritron.org; 217 plots) and other sources (27 plots). Plot monitoring periods span 2 to 40
years, between 1968 to 2015 (Extended Data Figure 1). In each plot (mean size, 1.1 ha), all trees 100 mm in stem diameter were
identified, mapped and measured on at least two occasions using standardized methods (135,625 trees monitored) and live biomass
carbon stocks were estimated for each census date, with carbon gains and losses calculated for each interval (Extended Data Figure
2). We compared trends in the per unit area African tropical forest carbon sink with published long-term trends in the Amazonian
carbon sink (Brienen, et al. 2015). We pooled the new African and existing Amazonian plot inventory data together to investigate the
putative environmental drivers of changes in the tropical forest carbon sink, and project its likely future evolution.
Aboveground Carbon (AGC, in Mg C ha-1) in living biomass for each plot at each census date was estimated as the sum of the AGC of
each living stem, then divided by plot area (in hectares).
Carbon Gain is the sum of the aboveground live biomass carbon additions from the growth of surviving stems and the addition of
newly recruited stems, using standard methods (Brienen, et al. 2015). For each stem that survived a census interval, carbon additions
from its growth (Mg C ha-1 yr-1) were calculated as the difference between its AGC at the end census of the interval and its AGC at
the beginning census of the interval. For each stem that recruited during the census interval (i.e. reaching DBH100 mm), carbon
additions were calculated in the same way, assuming DBH=0 mm at the start of the interval (Talbot, et al. 2014).The carbon additions
in an interval, from surviving and newly recruited stems, were summed, then divided by the census interval length (in years), and
scaled by plot area (in hectares) (Talbot, et al. 2014). As carbon gains are affected by a census interval bias, with the underestimate
increasing with census length, we corrected this bias by accounting for (i) the carbon additions from trees that recruited and then
died within the same interval (unobserved recruitment), and (ii) the carbon additions from trees that grew before they died within an
interval (unobserved growth) (Talbot, et al. 2014). These typically add <3% to plot-level carbon gains.
Carbon Loss (in Mg C ha-1 yr-1) is estimated, using standard methods (Brienen, et al. 2015), as the sum of aboveground biomass
carbon from all stems that died during a census interval, divided by the census length (in years) and scaled by plot area (in hectares).
Carbon loss is also affected by the same census interval bias, hence we corrected this bias by accounting for (i) the additional carbon
losses from the trees that were recruited and then died within the same interval, and (ii) the additional carbon losses resulting from
the growth of the trees that died in the interval (Kohyama, et al. 2018; Talbot, et al. 2014). Calculation details of both components
are explained in Supplementary Methods.
Net Carbon Sink (in Mg C ha-1 yr-1) is estimated as carbon gains minus carbon losses.
The estimated mean carbon gains, carbon losses and the net carbon sink of the African plots from 1983-2014, the solid lines in Figure
1, were calculated following (Brienen, et al. 2015) to allow direct comparison with published Amazonian results. First, each census
interval value was interpolated for each 0.1-yr period within the census interval. Then, for each 0.1-yr period between 1983 and
2014, we calculate a weighted mean of all plots monitored at that time, using the square root of plot area as a weighting factor.
Finally, confidence intervals for each 0.1-yr period are bootstrapped.
References:
Brienen, R. J. W., et al.
2015 Long-term decline of the Amazon carbon sink. Nature 519(7543):344-348.
Kohyama, Takashi S., et al.
2018 Definition and estimation of vital rates from repeated censuses: Choices, comparisons and bias corrections focusing on trees.
Methods in Ecology and Evolution 9(4):809-821.
Talbot, Joey, et al.
2014 Methods to estimate aboveground wood productivity from long-term forest inventory plots. Forest Ecology and Management
320:30-38.
Research sample We use data from 244 plots in 11 African countries to present the first assessment of the temporal evolution of the tropical forest
carbon sink in Africa. It represents 10 years of new field campaigns in Africa, extending sampling into extremely remote and
previously unsampled regions. This is the first new manuscript using long-term inventory plots to estimate the intact forest carbon
sink in Africa since (Lewis, et al. 2009) was published in Nature.
Plot selection: 244 permanent inventory plots were selected from 11 countries. These plots are situated in closed canopy (i.e. not
3
nature research | reporting summary October 2018
woody savanna) old-growth mixed-age forests and were selected using commonly used criteria (Brienen, et al. 2015; Lewis, et al.
2009; Lewis, et al. 2013): free of fire and industrial logging; all trees with diameter at reference height 100 mm measured at least
twice; 0.2 ha area; <1500 m.a.s.l. altitude; MAT 20.0 °C (Hijmans, et al. 2005); annual precipitation 1000 mm; located 50 m from
anthropogenic forest edges.
References:
Brienen, R. J. W., et al.
2015 Long-term decline of the Amazon carbon sink. Nature 519(7543):344-348.
Hijmans, Robert J., et al.
2005 Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology 25(15):1965-1978.
Lewis, S. L., et al.
2009 Increasing carbon storage in intact African tropical forests. Nature 457(7232):1003-1006.
Lewis, Simon L., et al.
2013 Above-ground biomass and structure of 260 African tropical forests. Philosophical Transactions of the Royal Society B:
Biological Sciences 368(1625):20120295-20120295.
Sampling strategy No sample size calculation was performed. We selected all available plots meeting the criteria described above. All African tropical
forest regions (West Africa, Lower Guinea, Congo Basin, East Africa) are adequately represented. This is the largest dataset of
repeatedly measured plots ever used to calculate long-term trends in African forest carbon dynamics.
Data collection Plot inventory data was collected by teams led by at least one of the 104 researchers co-authoring this paper. All permanent
inventory plots are part of one or several networks. Of the 244 plots included in the study, 217 contribute to the African Tropical
Rainforest Observatory Network (AfriTRON; www.afritron.org), with data curated at www.ForestPlots.net. These include plots from
Sierra Leone, Liberia, Ghana, Nigeria, Cameroon, Gabon, Republic of Congo, Democratic Republic of Congo (DRC), Uganda and
Tanzania (Lopez-Gonzalez, et al. 2011; Lopez-Gonzalez, et al. 2009) (Extended Data Figure 1). Fifteen plots are part of the TEAM
network, from Cameroon, Republic of Congo, Tanzania, and Uganda (Hockemba 2010; Kenfack 2011; Rovero, et al. 2009; Sheil and
Bitariho 2009). Nine plots contribute to the ForestGEO network, from Cameroon and DRC (Anderson-Teixeira, et al. 2015) (9 plots
from DRC, codes SNG, contribute to both AfriTRON and ForestGEO networks, included above in the AfriTRON total). Finally, three
plots from Central African Republic are part of the CIRAD network (Claeys, et al. 2019; Gourlet-Fleury, et al. 2013).
Tree-level aboveground biomass carbon is estimated using an allometric equation (Chave, et al. 2014) with parameters for tree
diameter, tree height and wood mass density. The estimated aboveground biomass of a plot is the sum of the estimated biomass of
all live trees at that census date.
Tree Diameter: In all plots, all woody stems with 100 mm diameter at 1.3 m from the base of the stem (‘diameter at breast height’,
DBH), or 0.5 m above deformities or buttresses, were measured, mapped and identified using standard forest inventory methods
(Phillips, et al. 2016). The height of the point of measurement (POM) was marked on the trees and recorded, so that the same POM
is used at the subsequent forest census. For stems developing deformities or buttresses over time that could potentially disturb the
initial POM, the POM was raised approximately 500 mm above the deformity. Estimates of the diameter growth of trees with
changed POM used the ratio of new and old POMs, to create a single trajectory of growth from the series of diameters at two POM
heights (Brienen, et al. 2015; Lewis, et al. 2009; Talbot, et al. 2014). We used standardized protocols to assess typographical errors
and potentially erroneous diameter values (e.g. trees shrinking by >5 mm), missing values, failures to find the original POM, and
other issues. Where necessary we estimated the likely value via interpolation or extrapolation from other measurements of that tree,
or when this was not possible we used the median growth rate of trees in the same plot, census and size-class, defined as DBH =
100-199 mm, or 200-399 mm, or >400 mm (Talbot, et al. 2014). We interpolate measurements for 1.3% of diameters, extrapolate
0.9%, and use median growth rates for 1.5%.
Tree height: Height of individuals from ground to the top leaf, hereafter Ht, was measured in 204 plots, using a laser hypsometer
(Nikon forestry Pro) from directly below the crown (most plots), a laser or ultrasonic distance device with an electronic tilt sensor, a
manual clinometer, or by direct measurement, i.e. tree climbing. Only trees where the top was visible were selected (Sullivan, et al.
2018). In most plots, tree selection was similar: the 10 largest trees were measured, together with 10 randomly selected trees per
diameter from five classes: 100-199 mm, 200-299 mm, 300-399 mm, 400-499 mm, and 500+ mm trees, following standard protocols
(Sullivan, et al. 2018). We use these data and the local.heights function in R package BiomasaFP (Lopez-Gonzalez, et al. 2017) to fit 3-
parameter Weibull relationships (see Supplementary Methods for a full explanation of this procedure):
H_t=a ×(1-e^((-b ×(DBH/10)^c ) )) (equation 1).
We chose the Weibull model as it is known to be robust when a large number of measurements are available (Feldpausch, et al.
2012; Sullivan, et al. 2018). We parameterize this Ht-DBH relationship for four different combinations of edaphic forest type and
biogeographical region (parameters in parentheses): (i) terra firme forest in West Africa (a=56.0; b=0.0401; c=0.744); (ii) terra firme
forest in Lower Guinea and Western Congo Basin (a=47.6; b=0.0536; c=0.755); (iii) terra firme forest in Eastern Congo Basin and East
Africa (a=50.8; b=0.0499; c=0.706); and finally (iv) seasonally flooded forest from Lower Guinea and Western Congo Basin (a=38.2;
b=0.0605; c=0.760). The parameters were used to estimate Ht from DBH for all tree DBH measurements for input into the allometric
equation.
Wood Density: Dry wood density (ρ) measurements were compiled for 730 African species from published sources and stored in
www.ForestPlots.net; most were sourced from the Global Wood Density Database on the Dryad digital repository
(www.datadryad.org)(Chave, et al. 2009; Zanne, et al. 2009). Each individual in the tree inventory database was matched to a
species-specific mean wood density value. Species in both the tree inventory and wood density databases were standardized for
orthography and synonymy using the African Flowering Plants Database (www.ville-ge.ch/cjb/bd/africa/) to maximize matches
(Lewis, et al. 2009). For incompletely identified individuals or for individuals belonging to species not in the ρ database, we used the
mean ρ value for the next higher known taxonomic category (genus or family, as appropriate). For unidentified individuals, we used
the mean wood density value of all individual trees in the plot (Lewis, et al. 2009; Lopez-Gonzalez, et al. 2011).
Allometric equation: For each tree we use a published allometric equation (Chave, et al. 2014) to estimate aboveground biomass. We
4
nature research | reporting summary October 2018
then convert this to carbon, assuming that aboveground carbon (AGC) is 45.6% of aboveground biomass (Martin, et al. 2018). Thus:
AGC=0.456× (((0.0673×(ρ ×(DBH/10)^2 ×H_t )^0.976))(1000)) (equation 2), with DBH in mm, dry wood density, ρ, in g cm-3, and
total tree height, Ht, in m (Chave, et al. 2014).
References:
Anderson-Teixeira, Kristina J., et al.
2015 CTFS-ForestGEO: a worldwide network monitoring forests in an era of global change. Global Change Biology 21(2):528-549.
Brienen, R. J. W., et al.
2015 Long-term decline of the Amazon carbon sink. Nature 519(7543):344-348.
Chave, Jerome, et al.
2009 Towards a worldwide wood economics spectrum. Ecology Letters 12(4):351-366.
Chave, Jérôme, et al.
2014 Improved allometric models to estimate the aboveground biomass of tropical trees. Global Change Biology 20(10):3177-3190.
Claeys, Florian, et al.
2019 Climate change would lead to a sharp acceleration of Central African forests dynamics by the end of the century.
Environmental Research Letters 14(4):044002.
Feldpausch, T. R., et al.
2012 Tree height integrated into pantropical forest biomass estimates. Biogeosciences 9(8):3381-3403.
Gourlet-Fleury, Sylvie, et al.
2013 Tropical forest recovery from logging: a 24 year silvicultural experiment from Central Africa. Philosophical Transactions of the
Royal Society B-Biological Sciences 368(1625):20120302.
Hockemba, M. B. N.
2010 Nouabalé Ndoki TEAM Site. Data Set Identifier: TEAM-DataPackage-20151201235855_1254.
Kenfack, D.
2011 Korup National Park TEAM Site. Data Set Identifier: TEAM-DataPackage-20151201235855_1254.
Lewis, S. L., et al.
2009 Increasing carbon storage in intact African tropical forests. Nature 457(7232):1003-1006.
Lopez-Gonzalez, G., et al.
2011 ForestPlots.net: a web application and research tool to manage and analyse tropical forest plot data. Journal of Vegetation
Science 22:610–613.
Lopez-Gonzalez, G., et al.
2009 ForestPlots.net Database. www.forestplots.net. Date of extraction [10/11/2017].
Lopez-Gonzalez, Gabriela, Martin Sullivan, and Tim Baker
2017 BiomasaFP package. Tools for analysing data downloaded from ForestPlots.net. R package version 0.2.1. Available at http://
www.forestplots.net/en/resources/analysis.
Martin, Adam R., Mahendra Doraisami, and Sean C. Thomas
2018 Global patterns in wood carbon concentration across the world’s trees and forests. Nature Geoscience 11(12):915-920.
Phillips, O., et al.
2016 RAINFOR field manual for plot establishment and remeasurement. Available at http://www.rainfor.org/upload/
ManualsEnglish/RAINFOR_field_manual_version_2016.pdf.
Rovero, F., A. Marshall, and E. Martin
2009 Udzungwa TEAM Site. Data Set Identifier: TEAM-DataPackage-20151130235007_5069.
Sheil, D., and R. Bitariho
2009 Bwindi Impenetrable Forest TEAM Site. Data Set Identifier: TEAM-DataPackage-20151201235855_1254.
Sullivan, M. J. P., et al.
2018 Field methods for sampling tree height for tropical forest biomass estimation. Methods in Ecology and Evolution
9(5):1179-1189.
Talbot, Joey, et al.
2014 Methods to estimate aboveground wood productivity from long-term forest inventory plots. Forest Ecology and Management
320:30-38.
Zanne, A.E., et al.
2009. Data from: Towards a worldwide wood economics spectrum: Dryad Digital Repository.
Timing and spatial scale The large majority of plots are sited in terra firme forests and have mixed species composition, although four are in seasonally
flooded forest and 14 plots are in Gilbertiodendron dewevrei monodominant forest, a locally common forest type in Africa
(Supplementary Table 1). The 244 plots have a mean size of 1.1 ha (median, 1 ha), with a total plot area of 277.9 ha. The dataset
comprises 391,968 diameter measurements on 135,625 stems, of which 89.9% were identified to species, 97.5% to genus and 97.8%
to family.
Plots were measured at least twice and maximum 10 times, between 1968 and 2015. Plot monitoring periods span 2 to 40 years;
mean total monitoring period is 11.8 years, mean census length 5.7 years, with a total of 3,214 ha years of monitoring. The 321
Amazon plots are published and were selected using the same criteria (ref.6), (Brienen, et al. 2015)except in the African selection
criteria we specified a minimum anthropogenic edge distance and added a minimum temperature threshold.
Brienen, R. J. W., et al.
2015 Long-term decline of the Amazon carbon sink. Nature 519(7543):344-348.
Data exclusions Plots were selected using the criteria described above (section Research sample). Plots that did not meet one or several of these
criteria were not used for analysis.
Reproducibility Our analysis does not include experimental findings.
Randomization Trends in carbon gains, losses and the net carbon sink over time were assessed using linear mixed effects models (lmer function in R,
5
nature research | reporting summary October 2018
Randomization lme4 package (Bates, et al. 2013)), providing the linear slopes reported in Figure 1. These models regress the mid-point of each
census interval against the value of the response variable for that census interval. Plot identity was included as a random effect, i.e.
assuming that the intercept can vary randomly among plots. Observations were weighted by plot size and census interval length.
Weightings were derived empirically, by assuming a priori that there is no significant relation between the net carbon sink and census
interval length or plot size (Lewis, et al. 2009).
References:
Bates, D., et al.
2013 lme4: Linear mixed-effects models using Eigen andS4.Rpackage version, 1.0-4. Available at http://www.inside-r.org/packages/
lme4/versions/1-0-4.
Lewis, S. L., et al.
2009 Increasing carbon storage in intact African tropical forests. Nature 457(7232):1003-1006.
Blinding Blinding was not relevant to our study.
Did the study involve field work? Yes No
Field work, collection and transport
Field conditions All plots are located in African tropical forests receiving at least 1000 mm rainfall annually and with a mean annual temperature
of at least 20 °C.
Location Plots are located at low elevations (<1500 m.a.s.l. altitude). A map showing locations of all plots is presented in Extended Data
Figure 1.
Access and import/export This paper is a product of the African Tropical Rainforest Observatory Network (AfriTRON), the TEAM network, the ForestGEO
network, and the CIRAD network. These permanent inventory plot networks only exists thanks to the support of governments,
local administrations and villages across Africa who have given us permission for, and helped us complete, our fieldwork. A full
list of partner institutions (excluding those in the co-author affiliations) can be found in (on-line only) acknowledgements.
Furthermore, plot inventory data are the product of many field-teams which mainly consisted of local assistants. A full list of
people involved in data collection can be found in (on-line only) acknowledgements, along with a full list of villages and
communities that hosted the field-teams and provided logistical and infrastructural support.
This paper includes 264 plot-censuses (out of 746) that are published for the first time here, including censuses from plots
located in extremely remote areas such as the Salonga National Park in the heart of the Congo Basin. Each plot-census
represents several months of preparation, transport, data collection, digitalisation and data quality assessment.
Disturbance No significant disturbance was caused by our measurements. Trees were tagged using a single aluminum nail (no iron), avoiding
damage to trees due to corrosion.
Reporting for specific materials, systems and methods
We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material,
system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response.
Materials & experimental systems
n/a Involved in the study
Antibodies
Eukaryotic cell lines
Palaeontology
Animals and other organisms
Human research participants
Clinical data
Methods
n/a Involved in the study
ChIP-seq
Flow cytometry
MRI-based neuroimaging
View publication statsView publication stats
... Besides being recognized as a biodiversity hotspot (1), and an important resource pool for the livelihood (e.g. food, wood, medicine) of local communities (2), this region plays a crucial role in the regional circulation of water (3,4), the global carbon cycle (5,6) and the continental greenhouse gas (GHG) balance (7). The Congo basin strongly regulates regional precipitation patterns and dominates global tropical rainfall distribution during transition seasons, tightly influencing regional and global climate (3,8). ...
... The Congo basin strongly regulates regional precipitation patterns and dominates global tropical rainfall distribution during transition seasons, tightly influencing regional and global climate (3,8). It sequesters approximately 0.59 Mg C ha -1 yr -1 making it the tropical region with the largest carbon uptake per unit of area (6,9). Moreover, the net full GHG sink in forests of the Congo basin is approximated at 0.61 Gt CO 2 equivalent yr -1 , which is six times stronger than the Amazon basin, although its surface is smaller in extent (7). ...
Article
Full-text available
The Congo basin is home to the second-largest tropical forest in the world. Therefore, it plays a crucial role in the regional water cycle, the global carbon cycle and the continental greenhouse gas balance. Yet very few field-based data on related processes exist. In the wake of global change, there is a need for a better understanding of the current and future response of the forest biome in this region. A new long-term effort has been set up to measure the exchange of greenhouse gasses between a humid lowland tropical forest in the Congo basin and the atmosphere via an eddy-covariance (EC) tower. Eddy-covariance research stations have been used for decades already in natural and man-made ecosystems around the globe, but the natural ecosystems of Central Africa remained a blind spot. The so-called “CongoFlux” research site has been installed right in the heart of the Congo Basin, at the Yangambi research center in DR Congo. This introductory paper presents an elaborated description of this new greenhouse gas research infrastructure; the first of its kind in the second-largest tropical forest on Earth.
... Pg C y −1 in secondary forests 2 . Tropical regions are particularly important in sequestering atmospheric carbon dioxide (CO 2 ) in both regenerating [3][4][5] and intact forests 1,6,7 . Nevertheless, recent analyses from both temperate 8 and tropical regions 7 have indicated that the magnitude of C sinks in old-growth forests are declining. ...
... Tropical regions are particularly important in sequestering atmospheric carbon dioxide (CO 2 ) in both regenerating [3][4][5] and intact forests 1,6,7 . Nevertheless, recent analyses from both temperate 8 and tropical regions 7 have indicated that the magnitude of C sinks in old-growth forests are declining. ...
Article
Full-text available
Woody tissue carbon (C) concentration is a key wood trait necessary for accurately estimating forest C stocks and fluxes, which also varies widely across species and biomes. However, coarse approximations of woody tissue C (e.g., 50%) remain commonplace in forest C estimation and reporting protocols, despite leading to substantial errors in forest C estimates. Here, we describe the Global Woody Tissue Carbon Concentration Database (GLOWCAD): a database containing 3,676 individual records of woody tissue C concentrations from 864 tree species. Woody tissue C concentration data—i.e., the mass of C per unit dry mass—were obtained from live and dead woody tissues from 130 peer-reviewed sources published between 1980–2020. Auxiliary data for each observation include tissue type, as well as decay class and size characteristics for dead wood. In GLOWCAD, 1,242 data points are associated with geographic coordinates, and are therefore presented alongside 46 standardized bioclimatic variables extracted from climate databases. GLOWCAD represents the largest available woody tissue C concentration database, and informs studies on forest C estimation, as well as analyses evaluating the extent, causes, and consequences of inter- and intraspecific variation in wood chemical traits.
... Visualizing feedback in tropical forests is therefore a key step toward understanding how these ecosystems might respond to global changes. For example, tropical forests have been acting as a net carbon sink, absorbing anthropogenic CO 2 emissions, but they can do that only up until a certain level of CO 2 -induced global climate change, after which they may become a net source, owing to combinations of increased mortality, respiration and fires, and reduced growth (Clark, 2004;Covey et al., 2021;Cuni-Sanchez et al., 2021;Gatti et al., 2021;Hubau et al., 2020;Mitchard, 2018). ...
Article
Full-text available
Tropical forests are complex systems containing myriad interactions and feedbacks with their biotic and abiotic environments, but as the world changes fast, the future of these ecosystems becomes increasingly uncertain. In particular, global stressors may unbalance the feedbacks that stabilize tropical forests, allowing other feedbacks to propel undesired changes in the whole ecosystem. Here, we review the scientific literature across various fields, compiling known interactions of tropical forests with their environment, including the global climate, rainfall, aerosols, fire, soils, fauna, and human activities. We identify 170 individual interactions among 32 elements that we present as a global tropical forest network, including countless feedback loops that may emerge from different combinations of interactions. We illustrate our findings with three cases involving urgent sustainability issues: (1) wildfires in wetlands of South America; (2) forest encroachment in African savanna landscapes; and (3) synergistic threats to the peatland forests of Borneo. Our findings reveal an unexplored world of feedbacks that shape the dynamics of tropical forests. The interactions and feedbacks identified here can guide future qualitative and quantitative research on the complexities of tropical forests, allowing societies to manage the nonlinear responses of these ecosystems in the Anthropocene. Tropical forests are complex systems containing myriad interactions and feedbacks with their biotic and abiotic environments, but as the world changes fast, the future of these ecosystems becomes increasingly uncertain. Our findings reveal an unexplored world of feedbacks that shape the dynamics of tropical forests. The interactions and feedbacks identified can guide future qualitative and quantitative research on the complexities of tropical forests, allowing societies to manage the nonlinear responses of these ecosystems in the Anthropocene.
... While most Earth system model-based climate change studies analyze projections till year 2100, these projections may miss physical-biogeochemical feedbacks that arise later from the cumulative effects of climate warming (Moore et al., 2018). The negative sensitivity of terrestrial carbon cycle to rising temperature will likely have increasing adverse implications on carbon cycle extremes over time (Frank et al., 2015;Hubau et al., 2020). Understanding the direction and strength of these feedbacks is essential for estimating long-term CO 2 concentrations and predicting and mitigating the impact and extent of climate change. ...
Article
Full-text available
The increasing atmospheric carbon dioxide (CO2) mole fraction affects global climate through radiative (trapping longwave radiation) and physiological effects (reduction of plant transpiration). We use the simulations of the Community Earth System Model (CESM1‐BGC) forced with Representative Concentration Pathway 8.5 to investigate climate‐vegetation feedbacks from 1850 to the year 2300. Human‐induced land use and land cover change (LULCC), through biogeochemical and biogeophysical processes, alter the climate and modify photosynthetic activity. The changing characteristics of extreme anomalies in photosynthesis, referred to as carbon cycle extremes, increase the uncertainty of terrestrial ecosystems to act as a net carbon sink. However, the role of LULCC in altering carbon cycle extremes under business‐as‐usual (continuously rising) CO2 emissions is unknown. Here we show that LULCC magnifies the intensity, frequency, and extent of carbon cycle extremes, resulting in a net reduction in expected photosynthetic activity in the future. We found that large temporally contiguous negative carbon cycle extremes are due to a persistent decrease in soil moisture, which is triggered by declines in precipitation. With LULCC and global warming, vegetation exhibits increased vulnerability to hot and dry environmental conditions, increasing the frequency of fire events and resulting in considerable losses in photosynthetic activity. While most regions show strengthening of negative carbon cycle extremes, a few locations show a weakening effect driven by declining vegetation cover or benign climate conditions for photosynthesis. Increasing hot, dry, and fire‐driven carbon cycle extremes are essential for improving carbon cycle modeling and estimation of ecosystem responses to LULCC and rising CO2 mole fractions. Moreover, large aberrations in vegetation productivity represent potential and growing threats to human lives, wildlife, and food security.