Assessing terrestrial carbon sink potential from vegetation under optimal land management

The global temperature could increase over 1.5 or even 2 °C by the middle of 21st century due to massive emissions of greenhouse gases (GHGs) — of which carbon dioxide (CO2) is the largest component1. Human activities emit more than 10 PgC (1PgC=1015gC) per year into the atmosphere1, which is regarded as the primary reason for increased atmospheric CO2 concentration and global warming2. Global vegetation sequesters 112–169 PgC each year3, about half of which is released back into the atmosphere through autotrophic respiration while the rest, termed as net primary production (NPP), is for balancing the CO2 emissions from human activities, microbial respiration, and decomposition4. Carbon sequestration from vegetation varies under different environmental conditions5 and could also be signicantly altered by land management practices (LMPs)6. Adopting optimal land management practices (OLMPs) helps sequester more CO2 from the atmosphere and mitigate climate changes. Understanding the extra carbon sequestration with OLMPs, or termed as carbon gap, is an important scientic topic that is rarely studied. Here we propose an integrated method to identify the location-specic OLMPs and assess the carbon gap by using remotely sensed time-series of NPP dataset, segmented landscape-vegetation-soil (LVS) zones and distance-constrained zonal analysis. The ndings show that the carbon gap from global land plants totaled 13.74 PgC per year with OLMPs referenced from within a 20km neighborhood, an equivalent of ~1/5 of the total sequestered net carbon at the current level; half of the carbon gap clusters in only ~15% of vegetated area. The carbon gap ux rises with population density and the priority for implementing OLMPs should be given to the densely populated areas to enhance the global carbon sequestration capacity.


Introduction
Keeping the global temperature increase below 1.5 °C in accord with the Paris Agreement would require prompt and substantial reductions in GHGs emissions on the global scale 1 . Despite considerable effort internationally, many countries are likely to miss the emission control targets proposed in the Paris Agreement, and the world is on track for more than 3°C of warming by the end of the century 7,8 . More severely, global warming will be irreversible if the concentration of GHGs continues to rise in the atmosphere 8 . To cut down GHGs emissions, numerous policies have long been proposed, including resorting to renewable energy, popularizing electric vehicles, optimizing land-use policies, as well as other GHGs cutting policies or programs 1,2 . Unfortunately, those policy options seem not being effectively deployed, given the continuous increase of the emissions of GHGs 9,10 . As the largest part of GHGs, the atmospheric CO 2 has risen beyond 400 ppm in recent years, a level not experienced over the past 800,000 years 3,11 . The rise of CO 2 levels in the atmosphere is mostly due to fossil fuel emissions and responsible for a 1.5°C increase in air temperature since the 1880s 3,12 . The unrelenting uxes and pools of the terrestrial carbon cycle are widely out of equilibrium from pre-historical conditions owing to human activities 3 .
Though cutting fossil fuel consumption provides a direct option to reduce carbon emissions, it is hampered by the fact that the economy in many countries is still powered by fossil energy 2,13 . Vegetation dominates most terrestrial ecosystems (e.g., forests, grasslands, croplands, shrublands, and savannas) and absorbs substantial CO 2 from the atmosphere through a biochemical process called photosynthesis 14 . Hence, biotic measures by improving carbon intake from vegetation provide a viable measure to counteract excessive carbon emissions 5,14 . The net amount of carbon captured by plants through photosynthesis over a given period is called net primary production (NPP) 4,15 . As a key component of energy and mass transformation in terrestrial ecosystems, NPP depends on a variety of factors, such as the supply of nutrients, water availability, soil pro le characteristics, and landscape attributes (e.g., terrain and drainage) 16,17 . Those factors can be broadly categorized as the following three groups 18 : 1) climate impact (e.g., precipitation and temperature); 2) non-climatic physical environmental factors such as soil property, landforms and biomes; and 3) human-related land management practices (HUMAN).
First, vegetation will not attain the saturation of carbon sequestration capacity without appropriate climatic conditions 5 . Lack of rainfall will cause physiological stress and limit vegetation photosynthesis 16 . The dependence of vegetation carbon xation on climatic factors is re ected by most NPP models 4 . For example, the climate-driven Miami model, which has been widely applied to map largescale NPP 5,19,20 , highlights the importance of the climatic factors as vital drivers to carbon sequestration.
Second, an ecosystem cannot maximize vegetation carbon sequestration without favorable phyical environmental conditions such as landforms 17 , soil properties 21 , and biome groups 5 . Lastly, carbon absorption can be updated by land management practices (LMPs) 14 , i.e., one LMP may result in higher or lower carbon sequestration than the other, under the same environmental contexts (the climatic and nonclimatic conditions).
Land use and management has massive effect on carbon sequestration from vegetation 6,20,22 . An optimal land management practice (OLMP) refers to an LMP that is capable of achieving full capacity or a higher target carbon sequestration level given the current climatic and non-climatic conditions. Once LMPs are replaced with OLMPs, vegetation can be expected to sequester more carbon. Modeling the increase potential or the difference in carbon sequestration (carbon gap hereafter) with-and without-OLMPs is valuable for making improved land management policies and mitigating global climate changes. In a broad sense, OLMPs could mean the removal of negative human-related disturbances in natural vegetated areas 22 , or the adoption of human-related programs/practices that are useful for restoring previously degraded vegetation 23,24 . OLMPs pertain to local environmental and socioeconomical contexts. Practically, OLMPs and the carbon gap at a particular site can be evaluated through eld experiments. However, considering manifest spatial variations of the environmental conditions across the globe, there exist no universally applicable OLMPs for all locations. There is also a need for mapping carbon gaps through a globally and locally compatible approach so that both global comparison and local policy enactment are supported 25 . The effectiveness of implementing OLMPs for enhancing vegetation productivity has been widely documented (see Supplementary Table 1). Herein, this work addresses the following questions: how much more carbon could be further sequested from global land vegetation with OLMPs? How can the location-dependent OLMPs be decided from the local abiotic contexts at each location? Where are the most sensitive areas with carbon sequestration potentials?

Global Carbon Gap Modeling And Mapping
Remote sensing technique facilitates large-scale NPP mapping 19 . The Moderate Resolution Imaging Spectroradiometer (MODIS) enables high-temporal and long-term Earth observations; the MODIS MOD17A3 NPP is one of the most highly used products for quantifying vegetation carbon sequestration 26,27 . In this work, the carbon gap is assessed based on the time-series MOD17A3 NPP in the following steps. To derive NPP for global and regional comparisons, homogeneous zones are rst partitioned from landform (L), vegetation (V), and soil (S) so that any LVS zone presents no internal NPP variation from the environmental factors (see Methods). To differentiate the impacts of human management practices (or LMPs) on vegetation carbon sequestration and to quantify the carbon gap, a distance-constrained (DC) zonal analysis is then devised to compute climate-recti ed NPP (NPP CR ) and derive the 90 th percentile statistic (NPP CR 90th ) of NPP CR inputs within a distance-constrained circular window with a radius ~20km on an annual basis (Supplementary Fig. 2 14 . Previous study showed that multiple cropping, fertilizer application and irrigation in agriculture land of China and India had contributed signi cantly to the "greening Earth" 28 . Those LMPs proved to be effective in capturing more NPP, once transferred/copied to neighboring locations with identical natural environments and lower NPP CR , will enhance carbon sequestration from vegetation. Conversely, in grassland ecosystems where degraded vegetation showed decreased productivity due to over-grazing 29 , destocking rate or rotational grazing in a neighborhood with high vegetation productivity could be referenced to be the OLMPs 14,30 .

The Carbon Gap Differs Among Continents And Biomes
Considerable differences in the ux density and in the totals of the carbon gap and NPP were observed among the 12 continents/regions (see Fig. 2 vs. Supplementary Fig. 3 America ranks the third largest in vegetated area (16.58 million km 2 ); however, because of its relatively high carbon gap ux (167.6 gC m -2 yr -1 ), the carbon gap totals 2.78 PgC yr -1 , leading all the continents/regions. The total carbon gap is strongly correlated to the total NPP at the continental/regional level (r=0.98). Theoretically, at pixel scale the carbon gap should also be correlated to NPP, as it is computed from the target carbon level (NPP CR 90th , within the local extent) and the NPP CR .
However, the carbon gap ux (density) was found to be inconsistent with the NPP density at the pixelscale or even within each continent/region. At the pixel level, the carbon gap ux and NPP ux are compared by the ratio, i.e., the carbon gap ux divided by the NPP ux at pixel basis, which shows clearly the high spatial variations ( Supplementary Fig. 4). Thus the carbon gap ux cannot be simply explained by NPP at pixel-scale. Within each continent/region, the carbon gap density (Fig. 1) and NPP density ( Supplementary Fig. 3) do not correlate well either, which is con rmed by the scatter plots between the carbon gap density vs. NPP density (Supplementary Fig. 6) and illustrated particularly by the two typical rainforests, i.e., the Amazon rainforest located within South America and the African rainforests ( Supplementary Fig. 7 Table 2 and Supplementary Fig. 1).
Forests shows NPP of 1106.2 gC m -2 yr -1 , much higher than the other biomes ( Fig. 3 and Supplementary  Fig. 5). Transforming land cover from one type to another, e.g., afforestation from croplands, may help to sequester more carbon 23 . However, while afforestation has been suggested for degraded croplands 23,24 , concerns arise from decreased food supply due to reduction in croplands. Hence, improving carbon sequestration through land conversions are not widely acceptable. The mapped carbon gap could be achieved by tailoring land management practices (i.e., by adopting OLMPs) without changing land use or land cover types, which has both ecological and social bene ts.
Grasslands and croplands lead the carbon gap density, reaching 138.3 and 137.2 gC m -2 yr -1 , respectively (Fig. 3) The high carbon gap uxes in grasslands and croplands indicate that they have a quicker response to human activities (i.e., updated LMPs), compared to the forests and OWV. For grasslands and croplands, improved (or degraded) vegetation may take a shorter period to appear (e.g., within a calendar year) once LMPs are updated. Conversely, apart from abrupt intervention from human activities such as massive reforestation or deforestation 14 , forests and OWV vegetation may take a longer time to show up the effect from updated LMPs. Thus, grasslands and croplands have the advantage of showing immediate bene t, i.e., increased carbon sequestration with implementation of OLMPs; on the other hand, it requires a longterm plan to improve carbon sequestration for forest and OWV ecosystems.
The carbon gap ux presents a strong clustering pattern, which is re ected by the comparison between the accumulative carbon gap and the accumulative sliced vegetated area using the carbon gap ux from low to high ( Fig. 4 and Supplementary Fig. 8). On average, ~50% of the total carbon gap comes from onlỹ 15% of the global vegetated area. The nding implies that, in terms of OLMP implementation, only a small part of prioritized vegetated aream i.e., ~15% of total vegetated area, could be implemented to add more 6.87 PgC yr -1 (i.e., 50% of the total carbon gap) intake from vegetation.

Human Activities Strongly Affect The Carbon Gap
We examine the relationship between the carbon gap and the population, as human density is a good indicator of human activities 32 . The world population varies substantially over the space, between and within continents/regions (Fig. 5 and Supplementary Table 5). South Asia, East Asia and South-East Asia present the most densi ed population, followed by moderately populated South-east Asia, Europe, and Africa. Though there exists no clear link between the carbon gap ux and population density at the continental level (Pearson's correlation coe cient r=0.27; see Supplementary Table 5), they present a strong consistency within-continental/regional scale, as revealed by the statistically signi cant and positive slope (β coe cient of the independent variable) for all the continents (Table 1 and Supplementary Fig. 9), suggesting that areas with intensive human domination are more likely to observe reduced NPP. The high carbon gap ux is primarily observed in the densely populated areas ( Fig. 1 and Fig. 5), meaning high potential of carbon sequestration enhancement. As a case of human activities, updating land management practices (LMPs) can alter vegetation NPP in terrestrial ecosystems, either positively or negatively 23,30 . The study veri ed that human activities have generally induced negative effect on vegetation NPP rather than improving it in the past two decades and thereby policy interference should focus more on those highly populated areas. Half of the total carbon gap is clustered in ~15% of the vegetated area, a small proportion that needs particular policy-making attention. The carbon gap is signi cantly and positively correlated to population density, con rming that intensi ed human activities have led to degraded vegetation.
The carbon gap ux is computed as the difference from the actual NPP CR to the target NPP CR level (NPP CR 90th ) which is equivalent to be the 90 th percentile of the NPP CR within ~20km neighborhood. The impact on vegetation NPP from LMPs is decoupled from that of the climatic and non-climatic factors.
Three steps are taken to assure that the result of carbon gap ux is comparable aross the global and regional scales and considers the contextual differences in identifying OLMPs. First, non-climatic physical environmental factors (landscape -L, vegetation -V and soil -S) are intersected to construct homogeneous LVS zones for global and regional comparison. Second, the varied impact on NPP from climatic factors is recti ed at LVS zone level. Third, each LVS zone is further apportioned into distanceconstrained neighborhood to compute the target carbon sequestration (NPP CR 90th ) and carbon gap.
LMPs at locations showing high climate-recti ed NPP are optimal ones (OLMPs) that are transferable/duplicable to their neighboring locations presenting less NPP CR but similar environmental conditions. Hence, the identi ed OLMPs are referenced from LMPs with proven effectiveness of vegetation productivity in a local and environmentally homogenious neighborhood within ~20km, which makes implementation of the OLMPs more practical at the local contexts. The identi ed OLMPs and the carbon gap suggest a prospective way for mitigating atmospheric CO 2 concentration through adjusting LMPs and highlight a global view of the carbon gap variation across continents/regions and biomes. The biotic measure through implementation of the localtion-dependent OLMPs to ll the carbon gaps, particularly in the ~15% prioritized areas possessing high carbon gap ux, can be one of the key focuses in response to global climate changes.