The global greening continues despite increased drought stress since 2000

Full text

Global Ecology and Conservation 49 (2024) e02791
Available online 25 December 2023
2351-9894/© 2023 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
The global greening continues despite increased drought stress
since 2000
Xin Chen
a
, Tiexi Chen
a
,
b
,
*
, Bin He
c
, Shuci Liu
d
, Shengjie Zhou
a
, Tingting Shi
e
a
School of Geographical Sciences, Nanjing University of Information Science and Technology, Nanjing 210044, Jiangsu, China
b
Qinghai Provincial Key Laboratory of Plateau Climate Change and Corresponding Ecological and Environmental Effects, Qinghai University of
Science and Technology, Xining 810016, Qinghai, China
c
College of Global Change and Earth System Science, Beijing Normal University, Beijing, China
d
Department of Environment and Science, Queensland Government, 4102 Brisbane, QLD, Australia
e
School of Applied Meteorology, Nanjing University of Information Science and Technology, Nanjing, China
ARTICLE INFO
Keywords:
Leaf area index
Growth rate
Global greening
Climate change
Drought trend
ABSTRACT
Increases or decreases in remote sensing-based vegetation greenness are usually referred to as
greening or browning. The CO
2
fertilization along with land management determined that
greening is dominant. However, recently global browning signals due to drought stress have also
been widely reported. In this study, We used the four latest leaf area index (LAI) datasets to
explore this controversial topic, and found that global greening was not only present (trend be-
tween 3.1–6.4 ×10
−3
m
2
m
−2
yr
−1
) but also continued (growth rate trend between 3.3–6.4 ×10
−4
m
2
m
−2
yr
−2
) during 2001–2020. Greening acceleration occurred in 55.15% of the globe (positive
trend and positive growth rate trend), while browning acceleration occurred in only 7.28%
(negative trend and positive growth rate trend). Combined with meteorological variables, we
found that CO
2
change dominated the LAI trend, while climate change largely determined the LAI
growth rate trend. Importantly, our study highlighted that drought trend did not necessarily
trigger vegetation browning, but slowed down the rate of greening.
1. Introduction
Vegetation is one of the basic components of terrestrial ecosystems and a regulator of climate change (Alkama et al., 2022; Griscom
et al., 2017). Since the 1980 s, the global leaf area index (LAI) based on satellite observations has shown a signicant increasing trend,
which is widely known as greening (Piao et al., 2020a). Numerous studies have conrmed the greening phenomenon, investigated the
drivers and corresponding inuences (Chen et al., 2019a; Chen et al., 2022b; Zhu et al., 2016). These studies found that global greening
could reduce climate warming by increasing terrestrial carbon sequestration and cooling the surface (Chen et al., 2019b; Zeng et al.,
2018).
However, the conclusion of greening has recently been challenged, with some studies nding that global greening stagnated or
even turned browning after 2000 (Chen et al., 2022a; Liu et al., 2023; Pan et al., 2018; Yuan et al., 2019). Therefore, more and more
attention has been paid to whether global vegetation is continuously greening or turning browning in recent years (Jiang et al., 2017;
* Corresponding author at: School of Geographical Sciences, Nanjing University of Information Science and Technology, Nanjing 210044, Jiangsu,
China.
E-mail address: txchen@nuist.edu.cn (T. Chen).
Contents lists available at ScienceDirect
Global Ecology and Conservation
journal homepage: www.elsevier.com/locate/gecco
https://doi.org/10.1016/j.gecco.2023.e02791
Received 9 October 2023; Received in revised form 29 November 2023; Accepted 22 December 2023

Global Ecology and Conservation 49 (2024) e02791
2
Wang et al., 2022). However, the results seem to be sensitive to the chosen datasets with different versions or sources. For example,
Enhanced Vegetation Index (EVI) of MODIS showed opposite trends in version 5 and version 6 (Zhang et al., 2017). AVHRR-LAI
illustrated global browning after 2000 (Chen et al., 2022a), which was in contrast to previous ndings based on MODIS-LAI (Chen
et al., 2019a).
The inconsistency of these ndings has also led to widespread controversy among studies on the identication of key drivers.
Studies supporting global browning primarily attribute browning to increased drought stress, and report potential mechanisms such as
a sharp increase in atmospheric saturated vapor pressure difference (VPD) limiting vegetation growth (Yuan et al., 2019), excessive
optimal temperatures inhibiting vegetation photosynthesis (Chen et al., 2022a), increasing water restriction on vegetation (Jiao et al.,
2021), and reduced CO
2
fertilization effects due to water and nutrient availability (Wang et al., 2020). However, process-based models
and observations suggest that global vegetation is still positively affected by CO
2
fertilization and land management, making greening
dominant (Chen et al., 2019a; Zhu et al., 2016). These controversies not only increase the uncertainty in the estimation of global
terrestrial carbon sources and sinks, but also hinder the better understanding of processes within terrestrial carbon cycle, which is of
critically importance for the development of models to describe these processes.
Several key LAI datasets have been signicantly updated recently, providing an opportunity to re-examine global vegetation
change trends and their drivers over nearly two decades. The widely used GIMMS-LAI3g has been further updated to GIMMS-LAI4g,
which solves a series of problems such as sensor drift (Cao et al., 2023). GLASS-LAI has also been updated to version 6, which shows
higher accuracy compared to other products (Ma and Liang, 2022). The update of these products may help to reduce the uncertainty in
the analysis of global vegetation change trends after 2000.
In addition, linear regression is often used to analyze the vegetation change trend in the current study. However, the vegetation
change trend in the later period may be covered up in the case of widespread greening (or browning) in the earlier period (Pan et al.,
2018). Some studies use piecewise linear regression to characterize differences in vegetation change over different time periods, but
this limits the application of some data sets with shorter time series (Chen et al., 2020; Wang et al., 2011). Therefore, another potential
problem is how to determine a reliable indicator of whether the vegetation change trend has changed over a short period of time.
In order to explore these issues, in this study, we analyzed the global vegetation change trends from 2001 to 2020 based on the
latest version of LAI datasets. Importantly, we also introduced the concept of LAI growth rate to analyze the rate of greening
(browning). Finally, we further analyzed the drivers of LAI trend and LAI growth rate trend.
2. Method
2.1. Satellite data
LAI has clearer physical meaning compared to other vegetation greenness, which characterizes the area of green leaves on the
ground (Fang et al., 2019). Therefore, we collected four widely used LAI datasets, including MODIS-LAI (C61), GLASS-LAI (V6),
GIMMS-LAI4g (V1), and GLOBMap-LAI (V3), and these datasets are available during 2001–2020. MODIS-LAI is generated based on the
spectral information of red band and near infrared band using look-up table (Knyazikhin et al., 1998). The spatial resolution of the
dataset is 500 m, and the temporal resolution is 8-day. GLASS-LAI is generated based on MODIS surface reectance data using
bidirectional long short-term memory model, and have high accuracy compared with observation (Ma and Liang, 2022). The spatial
resolution of the dataset is 0.05◦, and the temporal resolution is 8-day. GIMMS-LAI4g is generated based on PKU GIMMS NDVI product
(data source from AVHRR and MODIS) and 3.6 million high-quality Landsat LAI samples using back propagation neural network,
which eliminates the effects of satellite orbit drift and sensor degradation (Cao et al., 2023). The spatial resolution of this dataset is
1/12◦, and the temporal resolution is half a month. GLOBMap-LAI is generated based on MODIS surface reectance data using
GLOBCARBON-LAI algorithm (Liu et al., 2012). The spatial resolution of this dataset is 1/13.75◦, and the temporal resolution is half a
month. In addition, we used the vegetation greenness dataset, including MODIS-NDVI (0.05◦, 8-day), MODIS-EVI (0.05◦, 8-day) and
PKU GIMMS-NDVI (1/12◦, half month) (Li et al., 2023a) as auxiliary data to determine the global vegetation change trends. All pixels
in each 0.5◦grid cell with the center of each LAI and vegetation greenness datasets were averaged to 0.5◦for global upscaling, and all
data were aggregated to monthly scale to satisfy spatio-temporal consistency.
2.2. Meteorological data and CO
2
concentration data
Meteorological data were obtained from ERA5-land (Munoz-Sabater et al., 2021), and the selected variables included air tem-
perature, precipitation, radiation, dew point temperature and surface soil moisture. The spatial resolution of these data is 0.1◦, and the
temporal resolution is monthly. To align with the LAI datasets, all pixels of meteorological data within 0.5◦grid cell were upscaled. It is
worth noting that VPD was not used in our study, although it is a very important meteorological driver for vegetation, we found that it
has multicollinearity with other meteorological variables (the global average variance ination factor was greater than 10), so in order
to avoid covering up the contribution of other drivers to vegetation change, VPD was not used. The CO
2
concentration data were
obtained from the CarbonTracker dataset (CT2022) with a spatial resolution of 3◦×2◦at monthly temporal scale (Jacobson et al.,
2023), which was downscaled to 0.5◦using nearest neighbor resample method.
2.3. Trend and growth rate trend
Trend that represent interannual changes in vegetation and meteorological variables at the global and grid scale were estimated
X. Chen et al.

Global Ecology and Conservation 49 (2024) e02791
3
using linear least squares and two-tailed t-test. In addition, we further estimated the growth rate of all data, a concept commonly
applied to atmospheric CO
2
concentration (Keenan et al., 2016), which is used to represent the change rate of vegetation and
meteorological variables, as follows:
Datagr =Datat+1−Datat
where Data
gr
represents the growth rate and t represents a year from 2001 to 2020. The growth rate trends in time series were also
analyzed. Therefore, there are four combinations of LAI: a positive trend with a positive growth rate trend (PP) indicates that greening
is accelerating; a positive trend with a negative growth rate trend (PN) indicates that greening is slowing down; a negative trend with a
positive growth rate trend (NP) indicates that browning is accelerating; a negative trend with a negative growth rate trend (NN)
indicates that browning is slowing down.
2.4. Attribution analysis
Multiple linear regression model was used to quantify the contribution of all drivers to LAI trend and LAI growth rate trend, and this
method has been widely used in quantifying the drivers of vegetation change trend (Jung et al., 2017; Li et al., 2023b; Song et al., 2022;
Zou et al., 2023). Specically, the meteorological variables, i.e., CO
2
, air temperature (Airt), precipitation (P), radiation (Srad) and
surface soil moisture (SM) were used as predictor variables and the LAI as well as the LAI growth rate as response variables. Taking the
LAI trend as an example, the model can be expressed as
LAIobs =βCO2×CO2+βAirt ×Airt +βP×P+βSrad ×Srad +βSM ×SM +δ
where LAI
obs
represents the LAI observed by the satellite during 2001–2020, β
CO2
, β
Airt
, β
P
, β
Srad
and β
SM
represent the sensitivity
factors of CO
2
, temperature, precipitation, radiation and soil moisture to LAI trend respectively, and δ is random error term. Therefore,
the relative contribution of different drivers to the LAI trend can be obtained using the sensitivity factor multiplied by the interannual
changes of different drivers, i.e
LAIobs =LAICO2+LAIAirt +LAIP+LAI Srad +LAISM +δ
where LAI
CO2
, LAI
Airt
, LAI
P
, LAI
Srad
and LAI
SM
represent the relative contributions of CO
2
, temperature, precipitation, radiation and
soil moisture to LAI trend, respectively. Similarly, we calculated the contribution of each driver to the LAI growth rate trend.
In order to reduce the uncertainty of multiple linear regression model in attribution analysis, we also used partial correlation
analysis to identify the inuence of different drivers on LAI trend and growth rate trend.
3. Result
All four LAI datasets showed signicant global greening (p<0.05) with a trend between 3.1 ×10
−3
m
2
m
−2
yr
−1
and 6.4 ×10
−3
m
2
m
−2
yr
−1
(Fig. 1a). Meanwhile, LAI growth rate showed a slight increasing trend with values between 3.3 ×10
−4
m
2
m
−2
yr
−2
and
6.4 ×10
−4
m
2
m
−2
yr
−2
(Fig. 1b), although none of them were signicant (p>0.05). Fig. S1 shows the change trend of vegetation
greenness. Similar to LAI, NDVI and EVI also showed a signicant increasing trend, and the growth rate was increasing. Therefore, at
the globe scale, vegetation continued greening during 2001 - 2020.
The spatial pattern of LAI trend and LAI growth rate trend during 2001–2020 was further investigated. The distribution of trends
was inconsistent between these datasets, especially in the tropical areas (Fig. S2). We counted areas with the consistent trend and
Fig. 1. Trends (a) and growth rate trends (b) of four LAI datasets from 2001 to 2020.
X. Chen et al.

Global Ecology and Conservation 49 (2024) e02791
4
growth rate trend in four LAI datasets. As shown in Fig. S3, 60.29% of the areas in the globe were inconsistent, however in the
consistent areas, 64.06% (25.44%/39.71%) of the areas showed accelerated greening (PP group, see the method 2.3), mainly
distributed in India, European plain and East Africa. The areas that browning was accelerating was only 2.07%, and most of them were
distributed in Eastern area in Brazil. We further took the mean of the four data sets as a reference, and analyzed the trend and growth
rate trend. As shown in Fig. 2, the greening was accelerating in 55.15% of the globe, among which the greening acceleration of India
and European plains was the most obvious, while the greening of China and North America plains was slowing down. Only 14.44% of
the globe was browning, with the accelerating (7.28%) and slowing down (7.16%) roughly equal.
Multiple linear regression model could explain (p<0.05) LAI trend and LAI growth rate trend in most areas of the globe, and the
explanatory power of the model was generally poor in middle central Africa and high latitude regions such as northern North America
(Fig. S4). Soil moisture led negative LAI trends in most areas of the globe, while temperature and CO
2
had positive contributions,
especially in China, India and European plain (Fig. S5). In contrast, precipitation and radiation had almost no contribution to LAI
trend. We further calculated the dominant drivers of LAI trend in each grid, and we found that CO
2
dominated the LAI trend of 75.63%
of the globe, and temperature and soil moisture only could reach 11.34% and 7.30% respectively, which were mainly concentrated in
the high latitude areas of the northern Hemisphere and western Australia. However, other meteorological factors could only dominate
LAI trend in a few areas (Fig. 3a).
For the LAI growth rate trend, the contribution of each driver varied with regions without a clear spatial distribution rule (Fig. 3b).
In general, soil moisture and radiation contributed more in most areas of the globe, with a signicantly higher contribution of soil
moisture than other drivers in areas such as the tropics and southeast Australia, and a very clear negative contribution of radiation in
southern China (Fig. S6). In contrast, the contribution of temperature, precipitation and CO
2
to the LAI growth rate trend was relatively
small in most areas. When we counted the dominant driver of each grid in the globe, we found that CO
2
could only dominate the LAI
growth trend in 30.89% of areas, while the proportion of areas dominated by meteorological factors increased signicantly compare to
their relative contribution to LAI trend, among which soil moisture and radiation reach 25.43% and 17.09% respectively.
Partial correlation analysis was used to further identify the dominant factors for LAI trend and growth rate trend at each grid in the
globe. Similar to the results of multiple regression model, the spatial distribution of the two was roughly the same (Fig. S7). However,
the areas dominated by CO
2
were all decreasing. The LAI trend was 39.54% and the growth rate trend was 11.56%, which was mainly
caused by the differences in high latitudes in the northern Hemisphere. Partial correlation analysis suggested that the LAI trend and
growth rate trend in this area were mainly dominated by air temperature and radiation.
As a comprehensive index of temperature and precipitation, soil moisture can be used to measure the degree of dryness and wetness
of an area. We noted a decreasing trend in soil moisture in most parts of the globe, indicating that drought stress had increased in recent
Fig. 2. Spatial distribution of trend and growth rate trend based on the mean of four LAI datasets. Yellow and blue indicate that LAI shows a positive
trend, yellow indicates a negative trend in LAI growth rate, and blue indicates a positive trend in LAI growth rate; Red and green indicate that LAI
shows a negative trend, red indicates a negative trend in LAI growth rate, and green indicates a positive trend in LAI growth rate.
X. Chen et al.

Global Ecology and Conservation 49 (2024) e02791
5
years (Fig. 4). The globe was divided into four groups according to LAI trend and growth rate trend as suggested in the methods section,
i.e., PP, PN, NP and NN. The soil moisture in the four sub-regions showed a decreasing trend, The trend of PP, PN, NP and NN were
−4.47 ×10
−4
m
3
m
−3
yr
−1
(p<0.01), −3.57 ×10
−4
m
3
m
−3
yr
−1
(p<0.01), −1.12 ×10
−4
m
3
m
−3
yr
−1
(p>0.01) and
Fig. 3. Dominant drivers of LAI trend (a) and growth rate trend (b), dened as the drivers that contributed most to LAI trend or growth rate trend
within each grid.
X. Chen et al.

Global Ecology and Conservation 49 (2024) e02791
6
−0.94 ×10
−4
m
3
m
−3
yr
−1
(p>0.01), respectively. We further calculated the LAI trend and growth rate trend of these four sub-
regions. As shown in Fig. 5, similar to the results of the spatial analysis, soil moisture showed negative contributions to LAI trend
in all sub-regions, indicating that drought trend seriously affected vegetation greening. In areas with positive LAI trend, the contri-
bution of CO
2
(7.32 ×10
−3
and 6.48 ×10
−3
) was much higher than that of other drivers, making CO
2
dominant global greening.
Compared with other drivers, soil moisture contributed more to the LAI growth rate trend, especially in NP and PN areas. In NP areas,
soil moisture had positive contributions (6.33 ×10
−4
), indicating that the changes of soil moisture contributed to the browning ac-
celeration. On the contrary, soil moisture showed negative contributions (−3.45 ×10
−4
) in PN areas, indicating that changes in soil
moisture slowed down vegetation greening.
4. Discussions
4.1. Controversy on vegetation change trend after 2000
Earlier studies have conrmed the fact of global greening, but most of them considered the long-term trend since 1982, that is, all
studies agree on global greening during 1982 – 2000 (Piao et al., 2020b; Zhu et al., 2016), however there is no widespread agreement
on global greening after 2000 (Liu et al., 2023; Pan et al., 2018; Yuan et al., 2019). In this study, based on the latest remote sensing
data, we try to answer the key question of whether the globe has been greening or browning since about 2000. Our results showed that
the vegetation across the globe was greening and that the greening had maintained a slight acceleration (Fig. 1), which supports
previous nding (Chen et al., 2019a). Therefore, the rst thing that needs to be discussed here is the potential causes of the paradoxical
phenomenon of global greening and global browning after 2000. The signal of the sensor used in the datasets is usually the rst
consideration. MODIS information is also used in the algorithm of the other three LAI datasets, which lead these datasets that we used
are not absolutely independent. As we all know, MODIS sensors have long exceeded their designed service life, showing a series of
problems such as sensor degradation in version 5 (Tian et al., 2015; Zhang et al., 2017). Fortunately, MODIS has been calibrated in
subsequent versions, which greatly reduces the uncertainty of all kinds of datasets released. In addition, a study by Yan et al. (2021)
conrmed the effectiveness of the MODIS calibration algorithm by comparing the LAI of MODIS with subsequent launched VIIRS,
ruling out the possibility of false greening caused by sensor degradation.
In contrast, the global browning identied in most current studies is based on ndings from AVHRR data sources (Chen et al.,
2022a; Yuan et al., 2019), which should be used in caution. It is well known that AVHRR-based NDVI and LAI have multiple sources of
uncertainty. There are obvious articial signals from the orbital drift in the widely used GIMMS-NDVI3g and GIMMS-LAI3g based on
AVHRR (Tian et al., 2015; Zhu et al., 2013). An important study by Wang et al. (2020) on the continued decline of the global CO
2
fertilization effect is also due to the fact that the quality of the AVHRR data source has also been widely questioned (Frankenberg et al.,
Fig. 4. Spatial distribution of soil moisture trend from 2001 to 2020 (Unit: m
3
m
−3
yr
−1
).
X. Chen et al.

Global Ecology and Conservation 49 (2024) e02791
7
2021; Zhu et al., 2021). Given that the global browning in GIMMS-NDVI3g and GIMMS-LAI3g after 2000 translates into global
greening in PKU GIMMS-NDVI and GIMMS-LAI4g after 2000 only due to algorithmic improvements and the addition of MODIS in-
formation, so one possible reason is that potential problems with AVHRR sensors trigger vegetation browning (Cao et al., 2023; Li
et al., 2023a).
4.2. Indicators of the vegetation change rate
The introduction of the concept of LAI growth rate provides a new perspective for the analysis of global vegetation change and
overcomes the limitations of traditional piecewise regression and breakpoint methods to a certain extent. Based on the LAI trend and
LAI growth rate trend, we have some novel ndings. Similar to the ndings of Chen et al. (2019a), India and China were responsible for
the overall global greening, but in terms of the rate of greening, the two countries showed opposite directions (Fig. 2). Greening was
accelerating in India while it was slowing down in China, which has rarely been reported in previous studies. Multiple linear regression
models attributed both phenomena to the effects of soil moisture, temperate and radiation, however it is clear that in China and India,
two countries with signicant land management, it is obviously impractical to attribute vegetation changes to meteorological factors
alone. Therefore, another possible explanation is that different land management are responsible for the difference in the rate of
greening. In China, after a massive afforestation program and agricultural modernization, greening may gradually reach saturation
(Sha et al., 2022). And in India, irrigated agriculture, which mitigated atmospheric and soil drought and made vegetation less sus-
ceptible to moisture pressure, may have further enhanced greening (Ambika and Mishra, 2020).
4.3. The synergistic phenomenon of drought trend and greening
Without considering human activities, LAI trend is mainly determined by the positive effect of CO
2
fertilization and the negative
effect of drought stress (Yuan et al., 2019; Zhu et al., 2016). A growing body of research shows that vegetation growth is enhanced by
moisture constraints due to increased VPD and decreased soil moisture caused by climate warming (Jiao et al., 2021; Liu et al., 2020).
However, It is not clear that the current drought trend reaches a threshold to exceed the positive effect of CO
2
fertilization. The
introduction of the concept of growth rate provided additional explanation, as we found that the drought trend could only have a
partial negative impact on vegetation, slowing down the vegetation greening and accelerating the vegetation browning. However,
drought trend could not lead to global browning as it could not overtake the positive effect of CO
2
fertilization that contributed to
global vegetation (Fig. 5). Our study explains the synergistic phenomenon of drought trends with greening, which is similar to recent
ndings on gross primary productivity, that is, the rise in VPD offsets only a small fraction of the increase in productivity caused by
warming and CO
2
, and gross primary productivity is still increasing globally (Song et al., 2022).
4.4. Limitations and prospects
In the attribution analysis of LAI trend and growth rate trend, there were some differences between multiple linear regression
model and partial correlation analysis at high latitudes in the northern hemisphere. Partial correlation analysis showed that LAI trend
and growth rate trend in this region were mainly affected by temperature and radiation, while multiple linear regression suggested that
CO
2
was the dominant driver. The results of the partial correlation analysis seem more reasonable because some previous studies have
shown that vegetation change in the high latitudes of the northern Hemisphere is mainly positively affected by climate warming
(Berner et al., 2020; Keenan and Riley, 2018). Multiple linear regression model performed relatively poorly in this region and might
Fig. 5. Relative contributions of the drivers of LAI trend and growth rate trend in different areas. PP represents areas with positive LAI trend and
positive growth rate trend, PN represents areas with positive LAI trend and negative growth rate trend, NP represents areas with negative LAI trend
and positive growth rate trend, and NN represents areas with negative LAI trend and negative growth rate trend. Relative contributions were
obtained by an area-weighted average of the contributions of drivers within different areas.
X. Chen et al.

Global Ecology and Conservation 49 (2024) e02791
8
not fully identify the contribution of individual drivers. In addition, our model does not include the contribution of land management,
which has been highlighted in some previous studies (Chen et al., 2019a; Chen et al., 2022c; Chen et al., 2023). Of course, the
contribution of land management to LAI trend and growth rate trend may be implied in other drivers, such as the positive contribution
of CO
2
to LAI trend in PP and PN areas.
Previous studies have demonstrated potential mechanisms by which different drivers affect LAI trend, such as climate warming
promoting greening through lengthening the growing season in high latitudes (Keenan and Riley, 2018), CO
2
fertilization effects
promoting LAI increase (Zhu et al., 2016), and afforestation and agricultural modernization leading to greening in China and India
(Chen et al., 2019a). However, as for the LAI growth rate trend, this is a new concept. Although we have quantied the contribution of
different drivers to the LAI growth rate trend on a global scale, there is no further study on its potential mechanism. In future studies,
we can try to explore this problem, especially in China and India, two countries with opposite LAI growth rate trends.
5. Conclusion
In conclusion, based on the latest remote sensing data, we explored the important issue of global vegetation change trends after
2000. Importantly, we introduced the concept of growth rate to characterize the rate of greening/browning. Our results showed that
the global greening was still present in 2001–2020, with 55.15% of areas greening at an accelerated rate, mainly concentrated in India
and the European plains, compared with 7.28% of browning. Multiple linear regression and partial correlation analysis agreed that
CO
2
dominated the LAI trend, while climate change determined the LAI growth rate trend. By analyzing different sub-regions of the
globe, we found that the drought trend only slowed down global greening, but was far from triggering browning. These ndings will
improve our understanding of the processes within carbon cycle and narrow the research gap of better dening if the status of global
vegetation is greening or browning in recent two decades.
CRediT authorship contribution statement
Liu Shuci: Validation, Writing – original draft. He Bin: Validation, Writing – original draft. Chen Tiexi: Conceptualization,
Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing. Chen Xin: Conceptualization, Data
curation, Formal analysis, Methodology, Software, Validation, Writing – original draft, Writing – review & editing, Visualization. Shi
Tingting: Validation, Writing – original draft. Zhou Shengjie: Validation, Writing – original draft.
Declaration of Competing Interest
The authors declare that they have no known competing nancial interests or personal relationships that could have appeared to
inuence the work reported in this paper.
Data Availability
Data will be made available on request.
Acknowledgments
This study was supported by the National Natural Science Foundation of China (No. 42130506, 42161144003 and 31570464) and
the Postgraduate Research & Practice Innovation Program of Jiangsu Province (No. KYCX23_1322).
Author contributions
X.C. compiled the data, conducted analysis, prepared gures. X.C. and T.X.C. wrote the manuscript. X.C. and T.X.C. conceived the
scientic ideas and designed this research. B.H., S.C.L, S.J.Z., and T.T.S. gave constructive suggestions for improving the manuscript.
Appendix A. Supporting information
Supplementary data associated with this article can be found in the online version at doi:10.1016/j.gecco.2023.e02791.
References
Alkama, R., et al., 2022. Vegetation-based climate mitigation in a warmer and greener World. Nat. Commun. 13 (1), 606.
Ambika, A.K., Mishra, V., 2020. Substantial decline in atmospheric aridity due to irrigation in India. Environ. Res. Lett. 15 (12), 124060.
Berner, L.T., et al., 2020. Summer warming explains widespread but not uniform greening in the Arctic tundra biome. Nat. Commun. 11 (1), 4621.
Cao, S., et al., 2023. Spatiotemporally consistent global dataset of the GIMMS leaf area index (GIMMS LAI4g) from 1982 to 2020. Earth Syst. Sci. Data 15 (11),
4877–4899.
X. Chen et al.

Global Ecology and Conservation 49 (2024) e02791
9
Chen, B., et al., 2022a. Inhibitive Effects of Recent Exceeding Air Temperature Optima of Vegetation Productivity and Increasing Water Limitation on Photosynthesis
Reversed Global Greening. Earth’S. Future 10 (11), e2022EF002788.
Chen, C., et al., 2019a. China and India lead in greening of the world through land-use management. Nat. Sustain. 2 (2), 122–129.
Chen, J.M., et al., 2019b. Vegetation structural change since 1981 signicantly enhanced the terrestrial carbon sink. Nat. Commun. 10 (1), 4259.
Chen, T., et al., 2020. The Greening and Wetting of the Sahel Have Leveled off since about 1999 in Relation to SST. Remote Sens. 12 (17), 2723.
Chen, T., et al., 2022b. Land Management Explains the Contrasting Greening Pattern Across China-Russia Border Based on Paired Land Use Experiment Approach.
J. Geophys. Res.: Biogeosciences 127 (6), e2021JG006659.
Chen, T., et al., 2022c. Land Management Contributes signicantly to observed Vegetation Browning in Syria during 2001-2018. Biogeosciences 19 (5), 1515–1525.
Chen, X., Cai, A., Guo, R., Liang, C., Li, Y., 2023. Variation of gross primary productivity dominated by leaf area index in signicantly greening area. J. Geogr. Sci. 33
(8), 1747–1764.
Fang, H., Baret, F., Plummer, S., Schaepman-Strub, G., 2019. An overview of global leaf area index (LAI): methods, products, validation, and applications. Rev.
Geophys. 57 (3), 739–799.
Frankenberg, C., Yin, Y., Byrne, B., He, L., Gentine, P., 2021. Comment on "recent global decline of CO
2
fertilization effects on vegetation photosynthesis" COMMENT.
Science 373 (6562).
Griscom, B.W., et al., 2017. Natural climate solutions. Proc. Natl. Acad. Sci. USA 114 (44), 11645–11650.
Jacobson, A.R. et al., 2023. CarbonTracker CT2022. NOAA Global Monitoring Laboratory.
Jiang, C., et al., 2017. Inconsistencies of interannual variability and trends in long-term satellite leaf area index products. Glob. Change Biol. 23 (10), 4133–4146.
Jiao, W., et al., 2021. Observed increasing water constraint on vegetation growth over the last three decades. Nat. Commun. 12 (1), 3777.
Jung, M., et al., 2017. Compensatory water effects link yearly global land CO
2
sink changes to temperature. Nature 541 (7638), 516–520.
Keenan, T.F., et al., 2016. Recent pause in the growth rate of atmospheric CO
2
due to enhanced terrestrial carbon uptake. Nat. Commun. 7 (1), 13428.
Keenan, T.F., Riley, W.J., 2018. Greening of the land surface in the world’s cold regions consistent with recent warming. Nat. Clim. Change 8 (9), 825–828.
Knyazikhin, Y., Martonchik, J.V., Myneni, R.B., Diner, D.J., Running, S.W., 1998. Synergistic algorithm for estimating vegetation canopy leaf area index and fraction
of absorbed photosynthetically active radiation from MODIS and MISR data. J. Geophys. Res.: Atmos. 103 (D24), 32257–32275.
Li, M., et al., 2023a. Spatiotemporally consistent global dataset of the gimms normalized difference vegetation index (PKU GIMMS NDVI) from 1982 to 2022. Earth
Syst. Sci. Data 15 (9), 4181–4203.
Li, S., et al., 2023b. Vegetation growth due to CO
2
fertilization is threatened by increasing vapor pressure decit. J. Hydrol. 619, 129292.
Liu, L., et al., 2020. Soil moisture dominates dryness stress on ecosystem production globally. Nat. Commun. 11 (1), 4892.
Liu, Q., et al., 2023. Vegetation browning: global drivers, impacts, and feedbacks. Trends Plant Sci. 28 (9), 1014–1032.
Liu, Y., Liu, R., Chen, J.M., 2012. Retrospective retrieval of long-term consistent global leaf area index (1981–2011) from combined AVHRR and MODIS data.
J. Geophys. Res.: Biogeosciences 117 (G4).
Ma, H., Liang, S., 2022. Development of the GLASS 250-m leaf area index product (version 6) from MODIS data using the bidirectional LSTM deep learning model.
Remote Sens. Environ. 273, 112985.
Munoz-Sabater, J., et al., 2021. ERA5-Land: a state-of-the-art global reanalysis dataset for land applications. Earth Syst. Sci. Data 13 (9), 4349–4383.
Pan, N., et al., 2018. Increasing global vegetation browning hidden in overall vegetation greening: Insights from time-varying trends. Remote Sens. Environ. 214,
59–72.
Piao, S., et al., 2020a. Characteristics, drivers and feedbacks of global greening. Nat. Rev. Earth Environ. 1 (1), 14–27.
Piao, S., et al., 2020b. Interannual variation of terrestrial carbon cycle: Issues and perspectives. Glob. Change Biol. 26 (1), 300–318.
Sha, Z., et al., 2022. The global carbon sink potential of terrestrial vegetation can be increased substantially by optimal land management. Commun. Earth Environ. 3
(1), 8.
Song, Y., Jiao, W., Wang, J., Wang, L., 2022. Increased global vegetation productivity despite rising atmospheric dryness over the last two decades. Earth’S. Future 10
(7), e2021EF002634.
Tian, F., et al., 2015. Evaluating temporal consistency of long-term global NDVI datasets for trend analysis. Remote Sens. Environ. 163, 326–340.
Wang, S., et al., 2020. Recent global decline of CO
2
fertilization effects on vegetation photosynthesis. Science 370 (6522), 1295-+.
Wang, X., et al., 2011. Spring temperature change and its implication in the change of vegetation growth in North America from 1982 to 2006. Proc. Natl. Acad. Sci.
USA 108 (4), 1240–1245.
Wang, Z., et al., 2022. Large discrepancies of global greening: indication of multi-source remote sensing data. Glob. Ecol. Conserv. 34, e02016.
Yan, K., et al., 2021. Performance stability of the MODIS and VIIRS LAI algorithms inferred from analysis of long time series of products. Remote Sens. Environ. 260,
112438.
Yuan, W., et al., 2019. Increased atmospheric vapor pressure decit reduces global vegetation growth. Sci. Adv. 5 (8), eaax1396.
Zeng, Z., Peng, L., Piao, S., 2018. Response of terrestrial evapotranspiration to Earth’s greening. Curr. Opin. Environ. Sustain. 33, 9–25.
Zhang, Y., Song, C., Band, L.E., Sun, G., Li, J., 2017. Reanalysis of global terrestrial vegetation trends from MODIS products: browning or greening? Remote Sens.
Environ. 191, 145–155.
Zhu, Z., et al., 2013. Global data sets of vegetation leaf area index (LAI)3g and fraction of photosynthetically active radiation (FPAR)3g derived from global inventory
modeling and mapping studies (GIMMS) normalized difference vegetation index (NDVI3g) for the period 1981 to 2011. Remote Sens. 5 (2), 927–948.
Zhu, Z., et al., 2016. Greening of the earth and its drivers. Nat. Clim. Change 6 (8), 791–795.
Zhu, Z., et al., 2021. Comment on "Recent global decline of CO
2
fertilization effects on vegetation photosynthesis" COMMENT. Science 373 (6562).
Zou, L., Stan, K., Cao, S., Zhu, Z., 2023. Dynamic global vegetation models may not capture the dynamics of the leaf area index in the tropical rainforests: a data-model
intercomparison. Agric. For. Meteorol. 339, 109562.
X. Chen et al.