Forests play a critical role in stabilizing Earth’s climate. Establishing Protected Areas (PAs) represents one approach to forest conservation, but PAs were rarely created to mitigate climate change. The global impact of PAs on the carbon cycle, through avoided emissions and/or enhanced growth, has not previously been quantified due to a lack of accurate global carbon stock maps. Here we used ~412 million lidar samples from NASA’s GEDI mission to estimate a total of 19.7 +/- 1.8 Gt of additional Aboveground Biomass (AGB) associated with PA status. These higher C stocks are primarily attributed to avoided emissions, and are roughly equivalent to annual global fossil fuel emissions. The total measured PA AGB was 125.3 Gt (+/- 0.63), 26% of all mapped terrestrial woody AGB. These results underscore the importance of conservation of high integrity1, high biomass forests for avoiding carbon emissions and preserving future sequestration.
Article
The Effectiveness of Global Protected Areas for Climate Change Mitigation
https://doi.org/10.21203/rs.3.rs-2318743/v1
This work is licensed under a CC BY 4.0 License
Journal Publication
published 01 Jun, 2023
Version 1
posted
You are reading this latest preprint version
Earth’s ecosystems play a critical role in the carbon cycle, with estimates of global terrestrial Aboveground Biomass (AGB) of ~522 Gt in 2010 2,3 and annual uptake of ~8 Gt CO2 4. The primary causes of AGB loss are deforestation and forest degradation, while vegetation carbon sinks are associated with afforestation and forest recovery. The UN Sustainable Development Goals (SDGs), decisions under the United Nations Framework Convention on Climate Change (UNFCCC) and the Convention on Biological Diversity (CBD), emphasize that habitat conservation and restoration should contribute simultaneously to biodiversity conservation and climate change mitigation 5. To support goal setting and the implementation of international strategies and action plans, guidance is needed to identify how well protected areas contribute to maximizing synergies between conserving biodiversity and other ecosystem services such as climate change mitigation 6.
Forest conservation is a crucial mechanism for forest management, climate change mitigation, and curbing biodiversity loss 7,8. Protected areas are a foundation for global forest conservation efforts and are key indicators for achieving the SDGs 9. While most efforts to establish protected areas have been focused on biodiversity protection 10, there are clear co-benefits of biodiversity and carbon conservation efforts 6,7,11. Protected areas effectively avoid forest cover loss 12,1314,regulate temperature and local climate 15, may boost carbon sequestration capacity 16,17and are therefore likely critical to climate change mitigation18. Intact forests, especially tropical forests, can sequester twice as much carbon than more human-impacted forests and planted monocultures 17,19. Protected forest areas are thought to contribute a large fraction (~27%) of the net global GHG sink 4 but large uncertainties remain in the magnitudes of AGB stocks and fluxes in terrestrial ecosystems 20,21. As a result, the degree to which protected status contributes to avoided carbon emissions or enhanced sequestration at a global scale remains highly uncertain.
Here, we analyze millions of spaceborne lidar derived estimates of AGB from NASA’s Global Ecosystem Dynamics Investigation (GEDI, 22) to spatially quantify the carbon effectiveness of Protected Areas (PAs) and test the assumption that protected areas provide disproportionately more ecosystem services through carbon storage and sequestration than non-protected areas. Previous attempts to quantify PA biomass content had high uncertainties and/or biases, as past satellite biomass products are known to saturate in high biomass forests 23, such as old growth protected areas. GEDI is the first satellite lidar system designed specifically to map forest structure, and provides orders of magnitude more 3D samples of Earth’s forests than have previously been available, capable of collecting accurate data in even the densest tallest forests 23. Thus, GEDI provides a richer dataset to quantitatively address questions of forest carbon stocks and fluxes than have been previously available. We use GEDI’s data to quantify the additional vegetation AGB stocks attributed to the existence of protected areas (termed ‘carbon effectiveness’ of PAs) at a global scale. This is achieved through matching each PA to ecologically similar unprotected areas (based on climate, human pressure, land type, country, and other factors). Our analysis is based on the unprecedented richness of GEDI samples within globally distributed terrestrial PAs and their ecologically matched unprotected areas. This comparison allows us to assess the enhanced value of carbon sequestration in protected areas relative to their unprotected counterparts.
We estimate the GEDI-domain (N/S of 51 degrees latitude) total protected area woody AGB is 125.3 Gt (+/- 0.62 Gt), equivalent to approximately 225.4 Gt (+/- 1.11) CO2eq. While protected areas represent approximately 11% of the measured forested area (16.2 Mkm2, Fig. 1), they store 26% (125.3 Gt) of the total estimated AGB (480 Gt, 24). This represents all PA biomass, not just that attributed to PA status (termed additionally preserved AGB). Areas with PA status have, on average, 28% more AGB than their matched unprotected sites, for a total of 19.7 Gt (+/- 1.84 Gt) of additionally preserved PA AGB.
Why is there more Biomass in Protected Areas than similar forests?
Most global forested PAs (62.7%) had significantly higher biomass in 2020 than matched unprotected areas. Matching between PAs and unprotected areas was conducted for a baseline year of 2000, assuming that forest structure was equal between PAs and matches for the year 2000 based on a suite of ecological, anthropogenic pressure, and climate variables. Our results therefore represent approximately 20 years of change in PAs and their matched counterparts. The observed differences in 2020 structure could be explained by (i) less AGB loss in PAs compared to matched unprotected areas resulting from deforestation and/or forest degradation between 2000 and 2020, (ii) increased forest growth in PAs compared to matched unprotected areas between 2000 and 2020, or (iii) PAs being preferentially established in higher biomass areas before 2000.
Forest cover dynamics from the Landsat data record were also analyzed from 2000–2020 25, and showed more than half of PAs with > 5 Mg/ha higher mean Aboveground Biomass Density (AGBD) also had lower rates of forest loss within PAs than in matched unprotected areas (Fig S4). Thus, the observed higher concentrations of AGB in PAs were attributed primarily to avoided carbon emissions from deforestation, which is supported by the optical data record (hypothesis i). In ~ 18% of PAs, the forest cover change data does not detect loss, while GEDI still observes higher AGB in PAs compared to matches. In these cases, we speculate that degradation is occurring outside of PAs, but is not visible to passive optical sensors that form the basis of the forest cover loss data (e.g. small-scale logging, understory loss, etc.). This apparent degradation signal demonstrates the importance of datasets such as GEDI to detect subtle changes in carbon stocks that were not detectable with previous satellite datasets 26. Indeed, avoided degradation associated with PAs has likely been missed in past studies analyzing reduced carbon loss rates in PAs, and thus underestimated. Although we attribute PAs where we see higher AGB without reduced forest cover losses as avoided degradation, PA vegetation in these cases may also be exhibiting enhanced growth (ii) 27. Our assertion of enhanced growth is supported by local and regional studies assessing PA forest growth 28,29. For these 18% of PAs without apparent forest loss in matches, we cannot definitively attribute the signal to degradation, enhanced growth, or a combination of both. Overall, we therefore attribute our findings primarily to avoided emissions from deforestation and secondarily to a combination of enhanced growth and/or avoided degradation (hypotheses i and ii).
We found little evidence that PA were placed in higher Carbon density forests (hypothesis iii). If PAs were being established in higher AGB areas, their baseline (year 2000) AGB should be higher than matches. We used a pre-existing year 2000 AGB map to test this, and found that recently established PAs (established in or after 2000) had little differences in AGB between PA and matched unprotected area in the year 2000 (Fig S3). Older PAs did have significantly higher 2000 AGB values than matched areas, and this difference increased with time since establishment. These findings are in line with expectations of PAs adding additional AGB through time, rather than being preferentially located in high AGB locations. We therefore conclude that preferential establishment in high AGB areas does not explain our observations of higher AGB densities in PAs.
The Amazon Dominated the Global Signal
The starkest contrast between protected and matched unprotected forests was found in South America, specifically the Tropical and subtropical moist broadleaf forest biome in the Brazilian Amazon (Fig. 2). This supports recent results related to the effectiveness of PAs in Brazil for avoiding deforestation 30,31, and quantifies the climate change impact of Brazilian PAs at 7.22 Gt AGB more than matched unprotected areas, representing 36.6% of the global signal. Again, this supports our hypothesis that differences between PA and unprotected AGB are primarily associated with avoided emissions from forest loss and degradation, as Brazil experienced the highest national forest loss rates of any country during the analysis period 13. It is also noteworthy that while South American PAs have the greatest avoided carbon emissions (additionally preserved AGB), they cover roughly the same geographic extent as African PAs (Table 1). Avoided emissions are a factor of both (a) deforestation rates outside of PAs and (b) the biomass densities of forests being lost (Fig. 4). Therefore while Africa hosts a similar total area of PAs, these have, on average lower biomass densities than in Latin America or tropical Asia 32, and are smaller (the median PA area in Africa is 85 km2 compared to South America’s 127 km2), which may result in increased disturbance. Additionally, there is a larger increase in anthropogenic pressure in both PAs and matched unprotected areas in the Afrotropics than in South America, which may be reflected in the relatively lower carbon effectiveness we saw in African forests 33. Indeed, Africa has the largest proportion of PAs with no additionally preserved AGB (Fig S5).
PA additionally preserved AGB varied considerably by biome, and while the signal was unsurprisingly highest in tropical moist forests, the dominant biome varied by continent (Fig. 2). The Amazon dominates the global signal in carbon effectiveness. Tropical moist forests also dominate Asia's signal, given high forest loss rates in Southeast Asia. Conversely, African PA effectiveness was dominated by temperate and tropical/subtropical grasslands, savannas, and shrublands. Indeed, Africa is the only continent where the PA effectiveness is not highest in a forest dominated biome, suggesting that PAs in woodlands, grasslands, savannas and shrublands may be reducing land conversion, e.g. to agriculture, reducing charcoal degradation 34, or bolstering woody encroachment 35,36 and thus curbing net carbon emissions in these systems. Yet tropical dry forests and woodlands are under protection both in Africa (less than one fourth protected) and worldwide (less than one third protected)37. With an estimated population of 320 million inhabiting such landscapes in the 2000s and an average of 2.4% increase per annum in sub-saharan Africa (Eva et al, 2006; Chidumayo and Gumbo, 2010), these ecosystems are facing higher human development pressure than humid forests. Therefore, our results substantiate the critical roles of protected areas in dry forest and woodland ecosystems.
At a global scale, forests dominate the carbon effectiveness of PAs (Table 1, Fig. 2). A singular exception is mangrove forests, which show a near zero effect of PA on Carbon stocks. This may be due to a few factors. First, many mangroves globally are below 5 m in height, therefore GEDI will miss a large portion of mangrove biomass both in protected areas and outside of them, this will skew the results. Secondly mangrove PAs may either be ineffective at protecting AGB as we know that mangroves are extremely vulnerable to human pressure. Specifically, we found lower AGB in protected mangrove areas in Indonesia and Malaysia, which also harbor most of the mangrove cover and PA’s worldwide (25% of Global mangrove extent and C is in Indonesia alone, as well as most of the deforestation). Finally, deforestation rates have declined in all mangroves since the year 2000 38 and 50% or more of global mangrove cover was already lost by 2000, limiting remaining unprotected mangrove areas available for cutting. While this result contrasts with studies demonstrating effective PAs for curbing mangrove loss in Indonesia 39, it may be related to complicated and challenging mangrove management 40, and is supported by results of a global mangrove study 38 which indicated increased pressure on Indonesian protected mangroves in particular, providing evidence that some PAs may be ineffective at protecting mangrove AGB. It is possible that mangrove AGB is being degraded while canopy cover remains intact, but further research specifically into carbon-rich mangrove ecosystems in PAs is critical. Finally, as GEDI does not collect data north of ~ 52 latitude, results related to boreal forests are underrepresented here, and should be interpreted as only the most southern boreal forests.
Table 1. The “additionally preserved AGB” is the difference between the AGB observed in PAs and ecologically similar unprotected areas. This AGB is aggregated at a biome, continental, and global scale. The total PA AGB stock and total area of PAs in million km2 (Mkm2) are also reported.
Most countries (78%) in the GEDI domain have reduced carbon emissions in protected areas (higher AGB compared to matched unprotected areas). The top 20 countries in terms of PA effectiveness at preserving carbon (Fig. 3) are either (a) geographically large, (b) host forests with high AGB, and/or (c) have high forest loss rates of unprotected forests (Fig. 4). Many of the top 20 countries fall in tropical dense forested areas such as the Amazon (Brazil, Venezuela, Peru, Bolivia), the Congo Basin (DRC), or Southeast Asia (Thailand, Indonesia, Cambodia, Malaysia). Outside the tropics, highly ranked countries tend to be geographically large (Australia, USA, Canada, France, Spain), or clustered in Eastern or Southern Africa where our results show biggest impacts outside of forests (Tanzania, Mozambique, Zimbabwe).
PAs are characterized by taller, denser, higher biomass forests
While we focused primarily on an analysis of forest AGB, similar trends were found for other GEDI-based forest structure variables, including maximum and mean canopy height, Plant Area Index (PAI), and canopy cover (Fig. 5). As GEDI predicts AGB as a function of height metrics 41, similar effectiveness was anticipated between AGB and height. However, canopy cover and PAI, which are independent structural data products, were also higher within PAs than outside PAs. This suggests forest structure beyond carbon is being negatively affected in the absence of protected status, which correlates with habitat suitability and biodiversity 42. Further, forest structure (e.g. complexity, cover) is known to influence regional hydrology 43 and be tightly linked to climate and soil characteristics 44. These results therefore highlight co-benefits of PAs between carbon and biodiversity 11.
Our results highlight the critical importance of Protected Areas to help mitigate climate change. Preserving aboveground biomass is only one element of climate forcing, and changes to forest albedo, evaporative flux, belowground biomass etc. are also likely impacted by protected status 27. This study focused specifically on the avoided carbon emissions associated with preservation of woody aboveground carbon stocks. While our findings support results from past studies which analyze the effectiveness of PAs for preventing forest loss 11,28,45,46, our results uniquely extend past studies using next generation satellite data to directly quantify enhanced stocks and/or avoided carbon emissions from deforestation and degradation with great consistency and accuracy. The vast majority of PAs exhibited avoided emissions compared to unprotected matches. In some areas we see a relatively small difference between PA and matched AGB levels, which suggests either reduced effectiveness of PAs, or lower pressure on forests regardless of PA status. In these countries, our data cannot determine whether PAs are ineffective or anthropogenic pressure is low, and results should be interpreted in the context of national data. The correspondence between high effectiveness of PAs for AGB preservation in countries with high forest loss rates (e.g. Brazil 47) clearly highlights that maintaining and expanding PAs, particularly in high biomass forests, is essential for achieving global carbon mitigation goals.
Our analysis used a novel satellite lidar dataset to quantify the carbon effectiveness of protected areas, which has not previously been possible with the accuracies or spatial resolution enabled by the GEDI mission. Further, GEDI is sensitive to structural change in forests that do not cause full canopy cover removal, and thus past studies likely have not accounted for avoided degradation in PAs. Time series satellite records 4 indicate that these findings are related to avoided emissions from deforestation, but several important caveats should be noted. First, we do not account for leakage (i.e. deforestation pressures that move away from protected areas), and it is possible that background deforestation rates would be lower if not enforced in protected areas 48. However, evidence shows that leakage is unlikely to negate our findings 49,50. Secondly, we do not analyze which type of PA is the most effective, analyze PAs by governance type, or subdivide PAs into multiple classes, but instead focus on a necessarily simplified global analysis. Results from Tanzania suggest that multiple designations of PAs bolster carbon effectiveness, but that analysis of the impact of PA designation likely varies by nation 51GEDI data might be crucial for Indigenous Territories, which may be at higher risk for degradation 52, but also may be more resilient to deforestation pressure 53.
Our analysis focused only on aboveground woody biomass but does not account for belowground biomass (BGB) or the future added sequestration from maintaining higher AGB within PAs 27. Therefore, our AGB totals only account for part of PAs' full current and future effectiveness for AGB preservation. Additionally, we do not account for the future protection and enhancement of carbon sequestration potential related to leaving PA AGB intact. Our results summarize the aggregate of avoided emissions and forest growth over the past two decades (post 2000), but these protected forests will continue to absorb carbon in the future, and thus we have made conservative estimates of their importance for avoiding carbon emissions, provided they remain intact 54.
Protected areas are effective for preventing carbon emissions related to deforestation and degradation, and as they store a substantial proportion of Earth’s forest aboveground biomass, preservation of these regions is an essential Natural Climate Solution 8. These areas are particularly critical in regions of the world experiencing high rates of deforestation. Our results highlight synergies in efforts and resources into protected areas, and a pathway towards effective carbon and biodiversity conservation monitoring for the next decade. Our work demonstrates that the protected area targets highlighted by the United Nations Convention on Biological Diversity will benefit the UNFCCC goals such as the Glasgow COP26 Leaders’ Declaration on Forests and Land Use commitment to end deforestation by 2030. Additionally, accurately quantifying the carbon offset capacity of protected areas with respect to a baseline condition addresses the additionality requirement and evidence that protected areas should be included to benefit from the reduced emissions from deforestation and degradation (REDD+) 55. We provide a quantitative and globally consistent demonstration of the success of PAs to mitigate climate change and reiterate the multiple benefits of expanding these areas.
We used NASA’s Global Ecosystem Dynamics Investigation (GEDI) data to quantify the effectiveness of PAs at preserving forest structure and aboveground biomass. We used statistical matching to link PA 1 km pixels to pixels outside of PAs with similarity in terms of their ecology, environment, and human pressure using data layers from the year 2000 51. These control pixels, or matched unprotected areas, allowed us to compare 2020 conditions between areas that would theoretically have the same biomass distributions in 2000 except for the presence of forest conservation through PA designation. We extracted distributions of GEDI forest structure measurements within PAs and controlled matches and quantified the differences between the two. Figure 6 provides a visual representation of the methods applied in this paper.
GEDI Mission Overview and Datasets
NASA’s GEDI mission launched December 5, 2018, and is collecting full waveform lidar samples from the ISS between ~ 52°N and 52°S under the ISS orbit (Fig. 1). GEDI has three lasers operating at 1064 nm, each illuminated ~ 25m circular ‘footprints’ to produce billions of high resolution samples of surface elevation, vegetation height, and foliage distribution. These lidar waveforms were processed to generate estimates of vegetation height metrics, terms ‘Relative Height’ (RH) metrics representing the vertical distribution of foliage in each 25 m footprint 56. RH100 represents the maximum canopy height per sample, while RH50 roughly represents mean height. Waveforms are also processed to estimate %canopy cover, and Plant Area Index (PAI), which is roughly the same as Leaf Area Index (LAI) but includes woody material 57. Finally, AGBD is estimated for each footprint by fitting models between thousands of globally distributed sets of waveform metrics and field plot measurements 41. These models are parametric Ordinary Least Squares models that predict AGBD as a function of RH metrics. They are divided by continent and Plant Functional Type (PFT), and are broadly consistent, typically predicting AGBD as a function of RH98 (maximum height) and a lower RH metric, representative of roughly mean canopy height. The height, cover, PAI and AGBD products used in this study are from the first 18 months of mission data, between April 2019 and September 2020, and represent a total of > 400 million 3D structure samples. GEDI presents the first ever global-scale satellite dataset designed specifically for measuring forest structure, and overcomes limitations for forest height, cover, and AGB mapping in dense forests where previous maps have little sensitivity to AGB 20.
IUCN Datasets
The World Database of Protected Areas (WDPA) is a joint project between the United Nations Environment Programme (UNEP) and the International Union for Conservation of Nature (IUCN), and represents the most comprehensive global database of terrestrial and marine protected areas (iucn.org). The database is compiled and managed by the World Conservation Monitoring Centre of UNEP (UNEP-WCMC), in collaboration with governments, NGOs, academia, and industry. Each protected area in the WDPA must meet the IUCN definition of “a clearly defined geographical space, recognised, dedicated and managed, through legal or other effective means, to achieve the long term conservation of nature with associated ecosystem services and cultural values” 58 and the Protected Planet data standards (see WDPA User Manual 1.6). IUCN classifies protected areas based on management category (I-VI, with increasing human intervention), governance type (i.e. who holds authority and accountability), status year, designation, and various other attributes. The WDPA is available online through ProtectedPlanet.net and is updated monthly. In this study we used the September 2020 version of the WDPA, with a total of 262,804 protected areas (240,713 polygons and 22,091 points) covering 245 countries and territories (Fig. 7). We only used PAs from terrestrial, vegetated ecosystems within GEDI’s area of coverage. The WWF biomes corresponding to the PAs analyzed in this study are shown on Fig. 2 and Table 1. In the accounting of total AGB in protected areas, overlapping protected areas are only counted once.
Matching Algorithm
To determine the effectiveness of PAs, we matched 1km sample pixels within PAs to 1km sample pixels outside of any PAs by controlling for a series of geophysical and socioeconomic characteristics (Table S1). A 10 km buffer around any PA borders was also implemented to remove potential spillover effects e.g. encroachment at the PA boundaries, and no locations within the 10 km buffer were considered for matching 45. By comparing the vegetation structure in PAs and their matched unprotected areas, we addressed the fact that PAs were usually non-randomly distributed and assessed the conservation strategy’s efficacy in storing biomass and carbon. We required controls to be in the same land cover category, country, ecoregion (dropped if returns no matches), and biome as the protected 1km samples sites, termed exact matching. In addition, we used propensity score matching to further identify 1km control pixels for each treatment (one to one matching) that yielded the highest propensity score value. Matched unprotected pixels were removed from the potential control dataset (without replacement). The propensity scores were based on a logistic regression model built with geophysical, climatic and social variables, in elevation, slope, mean precipitation, min and max temperature, distance to city, distance to roads, travel time to city, population count, population density. The matching procedure was adopted from the methods described in 59–61.We evaluated our matching algorithm performance by inspecting the distributions of propensity scores after each step of matching following 51. To reduce computational time in the propensity score matching, we set a caliper to constrain the degree of difference between potential controls and treatment by requiring an overlap in the range of propensity scores for treatment and potential control datasets.
GEDI Data Processing
We used data from GEDI’s footprint-level height product (GEDI02_A), cover product (GEDI02_B), and biomass product (GEDI04_A). We applied quality filters to each of these products to only include high quality forest structure data. We filtered any data where the GEDI02_A quality_flag was equal to 1 and the sensitivity metric was at least 0.95 (i.e. GEDI’s lasers were capable of penetrating more than 95% canopy cover for all data used in this study).
We then extracted forest structure metrics from quality filtered GEDI data from all matched 1 km pixels with GEDI data within PAs and within matched unprotected areas. The total number of filtered GEDI shots used in this analysis was 412,100,767. For each GEDI sample, four structure metrics (RH98, canopy cover, PAI, and AGBD) were aggregated for comparison at a national, continental, and biome level. The difference in mean AGB between protected areas and matched unprotected areas was used to indicate the carbon effectiveness of PAs at the given level of aggregation. The distribution of GEDI samples over a single protected area in the Brazilian Amazon, with the associated matched pixels, illustrates the sampling nature of this comparison (Fig. 8).
Total Biomass, Uncertainty Estimation and Carbon Expansion
We estimated the mean and total PA biomass and associated uncertainty in each PA, and for the aggregation of all PAs at a country, continent and biome scale using the statistical hybrid inference approach developed for the GEDI mission 24. This approach accounts for uncertainty resulting from footprint biomass model parameter error and the GEDI sampling design to produce an estimate of the mean biomass and the standard error of the mean at any scale ≥ 1 km2 (50). This algorithm was applied for each PA to estimate a mean PA AGB density, with an associated standard error. The total AGB for the area used the mean multiplied by the non-overlapping PA area, with a confidence interval generated using the standard error of the mean and weighted by the size of each strata. The set of matched 1 km pixels for all PAs and unprotected matches in each biome and within every country was aggregated. The mean difference and associated standard error of the mean difference was computed at a country by biome scale. This mean difference was also multiplied by the PA area to estimate the total additionally preserved AGB attributed to the PA status. Results were aggregated to a country, continent, biome, and global scale by totaling the PA-level estimated additionally preserved biomass and associated uncertainties. For unmatched PAs, where there was insufficient GEDI data, we extrapolated from country by biome results to estimate total additionally preserved biomass. If county by biome results were unavailable (e.g. in the case of rare entirely protected biomes), extrapolation was from a continent by biome level. Conversion of all AGB estimates to carbon used a conversion factor of 0.49, the IPCC global average for conversion factor for woody biomass 62.
Acknowledgments: The authors gratefully acknowledge the large number of contributors of field and airborne lidar data that enabled the creation of the empirical biomass models that underpinned the biomass products analyzed in this study. We also gratefully acknowledge the continued support of the GEDI mission from the NASA Terrestrial Ecology program.
Funding:
National Aeronautics and Space Administration grant 80NSSC21K0196 (LD, ML, VL)
National Aeronautics and Space Administration Contract #NNL 15AA03C (RD, JA)
National Aeronautics and Space Administration grant 80NSSC19K0186 and 80NSSC21K0189 (SG)
National Science Foundation grant OAC1934389 (BE, CM, LD, SM)
Author contributions:
Conceptualization: LD, LF, ML, KT, PR
Methodology: LD, ML, VL, JA, SC, MG, AZ
Investigation: LD, ML, VL, SM
Visualization: ML, VL, LD
Funding acquisition: BE, CM, LD
Writing – original draft: LD, ML, VL, RD, BE, LF, SG, CM, PR, KT, JA, SC, SM
Writing – review & editing: LD, ML, VL, RD, SM, SC, BE, LF, SG, CM, PR, KT, AZ
Competing interests: Authors declare that they have no competing interests.
Data and materials availability: All data, code, and materials used in the analysis are publicly available. The satellite data are available from LPDAAC & ORNLDAAC, and all analysis code is available here: https://github.com/lauraduncanson/GEDI_PA.git.
- Grantham, H. S. et al. Anthropogenic modification of forests means only 40% of remaining forests have high ecosystem integrity. Nat. Commun. 11, 5978 (2020).
- Spawn, S. A., Sullivan, C. C., Lark, T. J. & Gibbs, H. K. Harmonized global maps of above and belowground biomass carbon density in the year 2010. Sci Data 7, 112 (2020).
- Santoro, M. et al. The global forest above-ground biomass pool for 2010 estimated from high-resolution satellite observations. Earth Syst. Sci. Data 13, 3927–3950 (2021).
- Harris, N. L. et al. Global maps of twenty-first century forest carbon fluxes. Nat. Clim. Chang. 1–7 (2021).
- Lee, B. X. et al. Transforming Our World: Implementing the 2030 Agenda Through Sustainable Development Goal Indicators. J. Public Health Policy 37 Suppl 1, 13–31 (2016).
- Jung, M. et al. Areas of global importance for conserving terrestrial biodiversity, carbon and water. Nat Ecol Evol 5, 1499–1509 (2021).
- Dinerstein, E. et al. A Global Deal For Nature: Guiding principles, milestones, and targets. Sci Adv 5, eaaw2869 (2019).
- Griscom, B. W. et al. Natural climate solutions. Proc. Natl. Acad. Sci. U. S. A. 114, 11645–11650 (2017).
- Lu, Y., Nakicenovic, N., Visbeck, M. & Stevance, A.-S. Policy: Five priorities for the UN Sustainable Development Goals. Nature 520, 432–433 (2015).
- Watson, J. E. M., Dudley, N., Segan, D. B. & Hockings, M. The performance and potential of protected areas. Nature 515, 67–73 (2014).
- Soto-Navarro, C. et al. Mapping co-benefits for carbon storage and biodiversity to inform conservation policy and action. Philos. Trans. R. Soc. Lond. B Biol. Sci. 375, 20190128 (2020).
- Yang, H. et al. A global assessment of the impact of individual protected areas on preventing forest loss. Sci. Total Environ. 777, 145995 (2021).
- Wolf, C., Levi, T., Ripple, W. J., Zárrate-Charry, D. A. & Betts, M. G. A forest loss report card for the world’s protected areas. Nat Ecol Evol 5, 520–529 (2021).
- Shi, H. et al. Global protected areas boost the carbon sequestration capacity: Evidences from econometric causal analysis. Sci. Total Environ. 715, 137001 (2020).
- Chen, J. et al. Microclimate in Forest Ecosystem and Landscape Ecology: Variations in local climate can be used to monitor and compare the effects of different management regimes. Bioscience 49, 288–297 (1999).
- de la Fuente, B. et al. Land productivity dynamics in and around protected areas globally from 1999 to 2013. PLoS One 15, e0224958 (2020).
- Huang, Y. et al. Impacts of species richness on productivity in a large-scale subtropical forest experiment. Science 362, 80–83 (2018).
- Goldstein, A. et al. Protecting irrecoverable carbon in Earth’s ecosystems. Nat. Clim. Chang. 10, 287–295 (2020).
- Baccini, A. et al. Tropical forests are a net carbon source based on aboveground measurements of gain and loss. Science 358, 230–234 (2017).
- Araza, A. et al. A comprehensive framework for assessing the accuracy and uncertainty of global above-ground biomass maps. Remote Sens. Environ. 272, 112917 (2022).
- Duncanson, L. et al. The Importance of Consistent Global Forest Aboveground Biomass Product Validation. Surv. Geophys. 40, 979–999 (2019).
- The Global Ecosystem Dynamics Investigation: High-resolution laser ranging of the Earth’s forests and topography. Science of Remote Sensing 1, 100002 (2020).
- Joshi, Mitchard, Brolly & Schumacher. Understanding ‘saturation’of radar signals over forests. Sci. Rep. (2017).
- Dubayah, R. et al. GEDI Launches a New Era of Biomass Inference from Space. Preprint at https://doi.org/10.31223/x52w68.
- Hansen, M. C. et al. High-resolution global maps of 21st-century forest cover change. Science 342, 850–853 (2013).
- Ahrends, A. et al. Detecting and predicting forest degradation: A comparison of ground surveys and remote sensing in Tanzanian forests. Plants People Planet 3, 268–281 (2021).
- Melillo, J. M. et al. Protected areas’ role in climate-change mitigation. Ambio 45, 133–145 (2016).
- Andam, K. S., Ferraro, P. J. & Hanauer, M. M. The effects of protected area systems on ecosystem restoration: a quasi-experimental design to estimate the impact of Costa Rica’s protected area system on forest regrowth. Conserv. Lett. (2013) doi:10.1111/conl.12004.
- N. Alves-Pinto, H. et al. The role of different governance regimes in reducing native vegetation conversion and promoting regrowth in the Brazilian Amazon. Biol. Conserv. 267, 109473 (2022).
- Gonçalves-Souza, D., Vilela, B., Phalan, B. & Dobrovolski, R. The role of protected areas in maintaining natural vegetation in Brazil. Sci Adv 7, eabh2932 (2021).
- Soares-Filho, B. et al. Role of Brazilian Amazon protected areas in climate change mitigation. Proc. Natl. Acad. Sci. U. S. A. 107, 10821–10826 (2010).
- Saatchi, S. S. et al. Benchmark map of forest carbon stocks in tropical regions across three continents. Proc. Natl. Acad. Sci. U. S. A. 108, 9899–9904 (2011).
- Geldmann, J., Manica, A., Burgess, N. D., Coad, L. & Balmford, A. A global-level assessment of the effectiveness of protected areas at resisting anthropogenic pressures. Proc. Natl. Acad. Sci. U. S. A. 116, 23209–23215 (2019).
- Sedano, F. et al. Monitoring forest degradation from charcoal production with historical Landsat imagery. A case study in southern Mozambique. Environ. Res. Lett. 15, 015001 (2020).
- Venter, Z. S., Cramer, M. D. & Hawkins, H.-J. Drivers of woody plant encroachment over Africa. Nat. Commun. 9, 2272 (2018).
- McNicol, I. M., Ryan, C. M. & Mitchard, E. T. A. Carbon losses from deforestation and widespread degradation offset by extensive growth in African woodlands. Nat. Commun. 9, 3045 (2018).
- Miles, L. et al. A global overview of the conservation status of tropical dry forests. J. Biogeogr. 33, 491–505 (2006).
- Goldberg, L., Lagomasino, D., Thomas, N. & Fatoyinbo, T. Global declines in human-driven mangrove loss. Glob. Chang. Biol. 26, 5844–5855 (2020).
- Miteva, D. A., Murray, B. C. & Pattanayak, S. K. Do protected areas reduce blue carbon emissions? A quasi-experimental evaluation of mangroves in Indonesia. Ecol. Econ. 119, 127–135 (2015).
- Friess, D. A. et al. Policy challenges and approaches for the conservation of mangrove forests in Southeast Asia. Conserv. Biol. 30, 933–949 (2016).
- Duncanson, L. et al. Aboveground biomass density models for NASA’s Global Ecosystem Dynamics Investigation (GEDI) lidar mission. Remote Sens. Environ. 270, 112845 (2022).
- Hansen, A. et al. Global humid tropics forest structural condition and forest structural integrity maps. Sci Data 6, 232 (2019).
- Lovejoy, T. E. & Nobre, C. Amazon Tipping Point. Sci Adv 4, eaat2340 (2018).
- Ehbrecht, M. et al. Global patterns and climatic controls of forest structural complexity. Nature Communications vol. 12 Preprint at https://doi.org/10.1038/s41467-020-20767-z (2021).
- Joppa, L. N. & Pfaff, A. Global protected area impacts. Proc. Biol. Sci. 278, 1633–1638 (2011).
- Scharlemann, J. P. W. et al. Securing tropical forest carbon: the contribution of protected areas to REDD. Oryx 44, 352–357 (2010).
- Nogueira, E. M., Yanai, A. M., de Vasconcelos, S. S., de Alencastro Graça, P. M. L. & Fearnside, P. M. Carbon stocks and losses to deforestation in protected areas in Brazilian Amazonia. Regional Environ. Change 18, 261–270 (2018).
- Ford, S. A. et al. Deforestation leakage undermines conservation value of tropical and subtropical forest protected areas. Glob. Ecol. Biogeogr. (2020) doi:10.1111/geb.13172.
- Lui, G. V. & Coomes, D. A. Tropical nature reserves are losing their buffer zones, but leakage is not to blame. Environ. Res. 147, 580–589 (2016).
- Villoria, N., Garrett, R., Gollnow, F. & Carlson, K. Leakage does not fully offset soy supply-chain efforts to reduce deforestation in Brazil. Nat. Commun. 13, 5476 (2022).
- Liang, M., Duncanson L., et al. Assessing Protected Area’s Carbon Stocks and Ecological Structure at Regional-Scale Using Gedi Lidar. In press, Global Environmental Change. Available at SSRN 3978725.
- Walker, W. S. et al. The role of forest conversion, degradation, and disturbance in the carbon dynamics of Amazon indigenous territories and protected areas. Proc. Natl. Acad. Sci. U. S. A. 117, 3015–3025 (2020).
- Nolte, C., Agrawal, A., Silvius, K. M. & Soares-Filho, B. S. Governance regime and location influence avoided deforestation success of protected areas in the Brazilian Amazon. Proc. Natl. Acad. Sci. U. S. A. 110, 4956–4961 (2013).
- Li, Y. et al. Deforestation-induced climate change reduces carbon storage in remaining tropical forests. Nat. Commun. 13, 1–13 (2022).
- Fuller, C., Ondei, S., Brook, B. W. & Buettel, J. C. Protected-area planning in the Brazilian Amazon should prioritize additionality and permanence, not leakage mitigation. Biol. Conserv. 248, 108673 (2020).
- Drake, J. B. et al. Estimation of tropical forest structural characteristics using large-footprint lidar. Remote Sens. Environ. 79, 305–319 (2002).
- Dubayah, R. et al. GEDI L2B Canopy Cover and Vertical Profile Metrics Data Global Footprint Level V001, NASA Earth Data. Preprint at (2020).
- Dudley, N. Guidelines for Applying Protected Area Management Categories. (IUCN, 2008).
- Ho, D. E., Imai, K., King, G. & Stuart, E. A. Matching as Nonparametric Preprocessing for Reducing Model Dependence in Parametric Causal Inference. Polit. Anal. 15, 199–236 (2007).
- Olmos, A. & Govindasamy, P. A Practical Guide for Using Propensity Score Weighting in R. Practical Assessment, Research, and Evaluation 20, 13 (2015).
- Schleicher, J. et al. Statistical matching for conservation science. Conserv. Biol. 34, 538–549 (2020).
- Eggleston, H. S., Buendia, L., Miwa, K., Ngara, T. & Tanabe, K. 2006 IPCC guidelines for National Greenhouse Gas Inventories. https://www.osti.gov/etdeweb/biblio/20880391 (2006).
There is NO Competing Interest.