Projected global LULC changes under eight coupled SSP-RCP scenarios
The global trends of future LULC changes over the 21st century are consistent between our projections and the LUH2 projections under the eight SSP-RCP scenarios (Fig. 1; Supplementary Figs. 1–6 for details of the global LULC changes). In both predictions, most terrestrial Earth would be covered by grassland, followed by forested land and then cropland, bare land and lastly, urban land. The minor quantitative differences lie in that our projections of forested land and cropland are smaller than the LUH2 projections whereas our projections of grassland, bare land, and urban land are slightly larger. Notably, qualitative differences do exist for each LULC type under certain scenarios. For example, grassland is projected (by FLUS) to increase under SSP3-RCP7.0 versus the projected decrease by LUH2. Forested land by FLUS is to experience a sharp decline before 2050 under SSP2-RCP4.5, while there is an only gentle decline trajectory by LUH2. Cropland is forecast by FLUS to have a slight increase under SSP3-RCP7.0, in contrast to the drastic growth projected by LUH2. Bare land is forecast by FLUS to decrease sharply by 2050 and then expand under SSP2-RCP4.5, instead of the continued decline trend as projected by LUH2. Particularly, urban land is forecast by FLUS to keep its rapid growth trend after 2050 under the eight scenarios, rather than the slowdown growth or even a decrease trend (under SSP1-RCP1.9 and SSP1-RCP2.6) in LUH2 projections.
In general, both our FLUS projections and the LUH2 projections show that the global trends of future LULC changes are susceptible to the socioeconomic pathways the global community decides to take, and that year 2050 could be the tipping point for critical changes (Fig. 1). For instance, following a sustainable development and low carbon emission pathway (SSP1-RCP1.9 and SSP1-RCP2.6), the global forest would generally increase and the bare land rather stable, while the urban and cropland would gain mild growth at largely the cost of grassland. Alternatively, if a regional rivalry and high-emission pathway were to be taken (SSP3-RCP7.0), the global bare land would experience the most decline and forested land also drastic loss, while urban land would gain the least growth with grassland and cropland mildly increasing. Furthermore, coupling global inequality or fossil-fueled development scenarios (SSPs 4-5) with RCPs adds new complications. For one, in an inequality world with high adaption to climate change in developing countries (SSP4), the global land system would be featured by the increase of urban land and decrease of bare land when coupled with RCP3.4 or RCP6.0. Yet, the future trajectories of global grassland, forest, and cropland would go opposite trends in the two coupled representative concentration pathways. For another, although the global land system under SSP5-RCP3.4 and SSP5-RCP8.5 would be featured by the similar trends of rapid urban growth, stable bare land, and mild forest loss, there would be a drastic conversion from grassland to cropland after 2050 under SSP5-RCP3.4, in contrast to the gentle trends of mild grassland loss and cropland gain under SSP5-RCP8.5. Finally, in the middle of the road scenario (SSP2-RCP4.5), the future global LULC changes would be featured by modest increases of cropland and urban land and the rapid decrease of grassland, while forest and bare land would first decrease until 2050 and then increase for the second half of the 21st century. Although changes in the spatial pattern of the simulated LULC could be barely noticeable at the global scale (Fig. 2), the structural changes of the global land system could still lead to pronounced social-environmental consequences at the regional and global scales, depending on future development pathways.
Projected future global GPP dynamics driven by LULC changes
The global trend of projected GPP dynamics over the 21st century is strongly affected by future LULC changes under the eight coupled scenarios (Fig. 3). Historically, our simulation shows that the LULC changes during 2000-2015 reduced the global GPP by 568.10 Tg C, a 0.35% loss. In a sustainable future, the trend would remain largely stable during 2015-2100, as the Mann–Kendall test detects statistically significant yet minor temporal changes in SSP1-RCP1.9 and SSP1-RCP2.6 (+2.99 Tg C yr-2, p = 0.006; +0.97 Tg C yr-2, p = 0.08). In the other six unsustainability scenarios, contrastingly, the total amount of the global GPP would experience dramatic changes over the 21st century, and the year 2050 could be a tipping point of shifting trends. The most dramatic decline of the global GPP occurs in a future of inequality with a stringent mitigation policy (SSP4-RCP3.4), while the total amount of the global GPP in the other five scenarios either shows a slowdown of decrease during 2050-2100 or shifts to an increasing trend after 2050 mainly due to the forest restoration. Generally speaking, under these six scenarios, the projected global GPP loss during 2015-2050 ranges from −30.27 to −11.84 Tg C yr-2, and the change during 2050-2100 ranges from −40.78 to +22.96 Tg C yr-2. As compared to 2015-2050, the impact of future LULC changes on the global GPP trend has remarkable uncertainties and even directional differences during 2050-2100. This indicates potentially huge environmental degradation risks and, simultaneously, a vast window of human interventions for sustainability in the second half of the 21st century.
Because of the possibility of nonlinear GPP trajectories, the nonparametric Sen’s slope was computed for each grid to assess its median annual GPP change. The computation was based on the GPP projections of each grid for every five years from 2015 to 2100. The slope shows strong spatial heterogeneity of the projected future global GPP dynamics driven by LULC changes (Fig. 4). Although most of the world would experience minor GPP changes (i.e., gray areas) in all the LULC scenarios, noticeable regional variations exist among the areas with significant GPP changes (e.g., blue and red areas). For example, in the scenarios of no aggressive development but with strict policy enforcement regarding vegetation protection or vegetation regrowth (i.e., SSP1-RCP1.9, SSP1-RCP2.6, and SSP2-RCP4.5), the projected global GPP at 2100 is slightly higher than the 2015 GPP (Fig. 3). In these three scenarios, the main GPP gainers include the Amazonia, Equatorial Afrotropics, southeast Asian forests, and west boreal forests, and the main GPP losers include the northeast American forests, southeast U.S. savannas and forests, Greater European forests, central east Asian forests, Brazil Cerrado and Atlantic coast, South American grasslands, and central Afrotropics (Figs. 4 a, b, and e). In the other five unsustainability scenarios with net global GPP loss during 2015-2100 (Fig. 3), the distribution of the GPP gainers and losers shows similar spatial patterns (Figs. 4 d and f-h). Yet, the area of these gainers slightly shrinks and that of the losers noticeably expands, which is particularly dramatic in a future world of inequality with relatively low carbon emissions (Fig. 4c, SSP4-RCP3.4).
These main gainers and losers—which form five GPP change hotspot areas, i.e., Eastern North America, Central South America, Europe, Central Africa, and East and Southeast Asia—were further examined to compare their simulated GPP trends under the eight LULC scenarios (Fig. 5). Our results show that these hotspot areas also have high GPP. The estimated GPP in 2015 shows that the five regions have an average GPP of around 2000 g C m-² yr-1 or above, except that Europe has close to 1000 g C m-² yr-1 GPP. Unfortunately, in most of the eight LULC scenarios, the five hotspot areas would experience GPP loss of varying degrees. The ranges of simulated GPP dynamics in the Eastern North America, Central South America, Europe, Central Africa, and East and Southeast Asia are −85.06 ~ −8.82 g C m-² (−4.11% ~ −0.43%), −32.14 ~ +7.29 g C m-² (−3.34% ~ +0.76%), −185.20 ~ +19.21 g C m-² (−6.49% ~ +0.67%), −33.60 ~ +80.03 g C m-² (−1.28% ~ +3.05%), and −49.28 ~ +8.19 g C m-² (−2.58% ~ +0.43%), respectively. The worst case of a −185.20 g C m-² (−6.49%) GPP loss would occur in East and Southeast Asia due to the biofuel plant under the SSP4-RCP3.4 scenario. Under this very scenario, contrastingly, Central South America would potentially gain the highest GPP of +80.03 g C m-² (+3.05%) through vegetation regrowth. This contrast again suggests that there is no panacea but place-based actions to mitigate GPP loss and even increase regional GPP.
Projected future global GPP dynamics driven by climate and LULC changes
The global trend of projected GPP dynamics over the 21st century is strongly dependent on future climate policies we shall take (i.e., RCPs), as indicated by the contrasting GPP trends in the simulation experiments with constant LULC and different climate forcing (Fig. 6 dashed lines). Interestingly, the GPP trends simulated with both climate and LULC changes (Fig. 6 solid lines) are similar and close to the results based on only changing climate forcing (dashed lines). The two observations suggest that future global GPP dynamics would be more influenced by climate change than LULC change. Among the three scenarios of concurrent climate and LULC changes (i.e., Clim45-LULC245, Clim60-LULC460, and Clim85-LULC585), the most dramatic GPP loss would occur under the climate forcing of RCP8.5 and LULC of SSP5-RCP8.5 (i.e., Clim85-LULC585), which could reduce the global GPP by −17.32 Pg C from 2015 to 2100, −10.75% of the 2015 global total. The dramatic GPP loss in a future with high energy demand and carbon emissions yet without any climate mitigation policy (i.e., RCP8.5) would become increasingly divergent from the GPP trajectories under the other two scenarios from 2050 onward, indicating a potential risk of locked-in environmental degradation that may be addressed with fewer policy efforts before 2050.
Although the global GPP is projected to decrease over the 21st century at the three levels of climate forcing (i.e., RCP4.5, RCP6.0, and RCP8.5), strong spatial heterogeneity stands out in terms of the climate change impact on future GPP dynamics, as measured by the Sen’s slope denoting the annual GPP change at a grid-scale (Fig. 7). Unlike the spatial pattern of the LULC change impact on future GPP dynamics, which is mostly minor (Fig. 4), the majority of the world would experience non-trivial GPP changes at the three levels of climate forcing (Fig. 7 non-gray areas), with GPP-stable areas mostly located at the desert areas (gray areas). This contrast corroborates the finding that future climate change would be more impactful than LULC change on global GPP dynamics. Another corroboration emerges from two other observations—LULC change would not notably affect the global pattern of projected annual GPP change (compare Fig. 7 subfigures horizontally), while climate change would expand GPP loss areas markedly (vertical comparisons). Generally speaking, the global patterns of projected annual GPP change under the three levels of climate forcing are consistently featured by GPP gains in those temperature-limited high-latitude/altitude areas (e.g., Russia, Canada, American West, and Andes & Pacific Coast) and humidity-limited semi-arid grasslands (e.g., Mongolian Grasslands and Australian Savannas), contrasting to GPP losses in mostly the tropical rain forest and temperate forest areas (e.g., Amazonia, Central America, Equatorial Afrotropics, Greater European Forest, Southeast Asian Forests, Malaysia & Western Indonesia, and Australian Islands & Eastern Indonesia). In particular, with the highest considered climate forcing (i.e., under RCP8.5), the areas with dramatic GPP loss would expand remarkably in the Amazonia and Central America (Fig. 7 dark blue areas).
Relative contributions of climate and LULC changes to global GPP dynamics
To quantitatively compare the respective impacts of climate and LULC changes on global GPP dynamics during 2000-2100, we adopted a contribution decomposition methodology (see Methods) to disentangle their relative contributions (and residual effects) at each 0.5° grid under three matched scenarios (i.e., Clim45-LULC245, Clim60-LULC460, and Clim85-LULC585; Fig. 8). Unsurprisingly, the contribution of LULC change to GPP dynamics during 2000-2100 is little or minor (i.e., contribution less than 10%) for 54% ~ 56% of the global terrestrial area in the three scenarios. Further, 7% ~ 9% (9,054,884 ~ 12,092,465 km2) would have LULC change dominating the global GPP dynamics (i.e., contribution over 50%, Figs. 8a, b, and c). Contrastingly, 75% ~ 77% of the terrestrial area would have climate change as the dominant driver (Figs. 8d, e, and f), and 12% of the global terrestrial area, mostly deserts, has stable negligible GPP changes—suggesting that although secondary to the impact of climate change on global GPP dynamics, the impact of LULC change is by no means trivial (cf., 7%). Spatially, the places with LULC change dominating GPP dynamics are rather fragmented, and seem to be mostly located at the margins of global deserts or the frontiers of global urbanization in especially coastal areas and central Africa. The area of such places does not increase proportionately with the increasing level of climate forcing, nor does their spatial pattern notably change with the three scenario settings (Figs. 8a, b, and c). The relatively stable spatial distribution of such LULC-change-dominated places suggests that proactive land management policymaking in these areas could be fruitful.
Of equal importance for human interventions are the five GPP change hotspot areas, i.e., East North America, Central South America, Europe, Central Africa, and East and Southeast Asia (same as those in Fig. 5), about which there are two observations worth-noting. For one, the contribution of LULC change to GPP dynamics in each hotspot is generally smaller in 2000-2050 than in 2050-2100 under the three matched climate-LULC change scenarios (Table 1). There are five exceptions, though: Europe and East and Southeast Asia under Clim60-LULC460, and Central South America, Europe, and Central Africa under Clim85-LULC585. For the other, the general temporal pattern of these five areas is opposite from that of the globe, for which LULC change is more impactful on GPP dynamics in the first half of the 21st century under all three scenarios. The contrast suggests that LULC change monitoring and management in these GPP change hotspot areas alone is insufficient for maintaining terrestrial carbon balance. Nonetheless, LULC change in these five hotspots could make a striking impact on GPP dynamics in a few cases. For instance, the relative contribution of LULC change in East and Southeast Asia could hit 56.08% in 2050-2100 under Clim45-LULC245, Central Africa for 34.99% under Clim60-LULC460, Central South America for 28.49% and 24.56% under Clim45-LULC245 and Clim60-LULC460, respectively. For the globe, however, the contribution of LULC change is estimated to reach 16.16%, 14.22%, and 10.92% in 2000-2050 under Clim45-LULC245, Clim60-LULC460, and Clim85-LULC585, respectively—in stark contrast to the corresponding 14.57%, 1.41%, and 1.74% in 2050-2100. The message is clear: the window of LULC-based solutions to build a carbon-neutral world is quickly vanishing, and the GPP change hotspot areas are insufficient for land policy interventions.
Table 1 Relative contributions of LULC change to GPP dynamics (%) under Clim45-LULC245, Clim60-LULC460, and Clim85-LULC585 in five GPP change hotspots and for the globe, for 2000-2050 and 2050-2100. Note that the five hotspots are consistent with those in Fig. 5.
|
SSP2-RCP4.5
|
SSP4-RCP6.0
|
SSP5-RCP8.5
|
2000-2050
|
2050-2100
|
2000-2050
|
2050-2100
|
2000-2050
|
2050-2100
|
|
Eastern North America
|
2.19
|
5.56
|
0.25
|
15.4
|
1.58
|
8.73
|
|
Europe
|
4.34
|
6.91
|
4.15
|
2.02
|
7.08
|
4.09
|
|
East and Southeast Asia
|
17.58
|
56.08
|
15.35
|
1.37
|
13.19
|
14.31
|
|
Central South America
|
7.96
|
28.49
|
0.69
|
24.56
|
4.97
|
0.67
|
|
Central Africa
|
0.08
|
5.14
|
5.19
|
34.99
|
3.97
|
3.6
|
|
Globe
|
16.16
|
14.57
|
14.22
|
1.41
|
10.92
|
1.74
|
|