Accuracy of rubber-related deforestation maps

Accuracy of rubber-related deforestation maps | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Accuracy of rubber-related deforestation maps Nguyen Tien Hoang, Peter Potapov, Pontus Olofsson, Keiichiro Kanemoto This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5834237/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 12 Aug, 2025 Read the published version in Nature → Version 1 posted You are reading this latest preprint version Abstract Accurate mapping of rubber plantations is essential to understanding where deforestation due to rubber production occurs. Wang et al . 1 produced 10-meter resolution maps of mature rubber with a reported high overall classification accuracy (OA = 0.95 ± 0.02), indicating that more than 4 million hectares of forests have been converted to rubber plantations in Southeast Asia and parts of China since 1993. However, we have serious concerns about the accuracy of these maps, as our analysis indicates that the area estimates by Wang et al . 1 are significantly inflated. This observation underscores the need for caution in using these high-resolution maps for specific applications, such as assessing deforestation linked to rubber consumption under the EU regulation on deforestation-free products. Forestry Environmental Economics Environmental Policy Hevea brasiliensis mapping accuracy Sentinel-2 EUDR Figures Figure 1 Figure 2 Full Text The questionable accuracy of the maps published by Wang et al . 1 arises from their approach to area estimation and accuracy assessment. Despite their claim of adhering to “good practices,” 2 we contend that these practices were misapplied. First, estimating areas by pixel-counting is biased and produces erroneous estimates because it ignores map classification errors 3 . Second, the sample data used for estimating map accuracy were not selected using probability sampling—a critical component for robust statistical evaluation 4 . Wang et al . 1 pooled over 3,800 sample points from three independent sources—the World Agroforestry Centre, Lan et al. 5 , and Hurni and Fox 6 —but this aggregation does not adhere to probability sampling rules across the entire study area. Probability sampling requires that each unit in the population has a known, non-zero inclusion probability 2,4 , which was not ensured. For example, one sample was drawn only from areas identified as rubber and evergreen forest, resulting in an inclusion probability of zero for all other units in the study area. Moreover, the in-situ observations from Xishuangbanna (China) and the Greater Mekong Subregion do not appear to have been selected under probability sampling. The geographical distribution of their sample data—predominantly in mainland Southeast Asia and notably absent in large areas of Indonesia, Myanmar, and the North Central Coast to North Vietnam—further supports this argument. Because the accuracy measures reported by Wang et al . 1 rely on a design-based inference framework that requires probability sampling 4 , their lack of adherence invalidates the area and map accuracy results they provided. Wang et al . 1 defend their rubber area estimates by citing two key points: first, they attribute the discrepancy with the harvested rubber area reported by the Forest and Agriculture Organization of the United Nations (FAO) to unharvested plantations resulting from low global rubber prices. However, this claim lacks empirical support, such as statistical data comparing harvested and mature rubber areas. Second, they argue that their estimates align with those from two other recent remote sensing studies 6,7 . Yet, the maps published in these two studies—referred to by Wang et al . 1 as reference data —are of lower resolution than their own maps, which results in a mismatch in data quality that undermines the validity of using these maps as a standard for comparison. Furthermore, reference data should represent the best available assessment of actual conditions on the ground 3 . Hence, using wall-to-wall maps as reference data is problematic as it implies that the maps constitute a complete census of reference conditions of the study area. To investigate the accuracy of Wang et al .’s rubber maps, we utilized comprehensive and comparable official statistical data from the top four rubber-producing countries in Southeast Asia: Indonesia, Thailand, Vietnam, and Malaysia (see Supplementary Tables 1-3). These countries report total rubber planted areas annually, with Indonesia, Thailand, and Vietnam providing detailed data at the Level 2 administrative units (districts and equivalents). Notably, Indonesia and Thailand report areas of mature rubber trees (ready for tapping), while Vietnam provides harvested rubber area data. For Malaysia, only regional total rubber planted areas are available. In addition, we examined mapping accuracy in Laos—the country with the smallest rubber area in Wang et al .'s estimate—using semi-official data on rubber planted and mature areas. It is important to note that statistical data, grounded in a land-use perspective, can provide higher area estimates than remote sensing data, which are rooted in a land-cover perspective. The discrepancy arises because the land use-based estimate tends to align with parcel boundaries rather than actual tree cover 8 . Wang et al . 1 used Sentinel-2 imagery from 2020 to 2022 to map rubber plantations in 2021. Official statistics from 2019 to 2021 showed stable total planted, mature, and harvested rubber areas in the top four rubber-producing countries—Indonesia, Thailand, Vietnam, and Malaysia (Fig. 1a). Based on this stability, we used the official 2021 data as our benchmark. Theoretically, the remotely sensed-based mature rubber area should be smaller than the reported planted or mature areas and larger than the harvested area. However, Wang et al .'s estimates were significantly higher—exceeding official total planted areas by 26% in Indonesia and 73% in Vietnam. Remarkably, their estimated mature rubber area for Indonesia alone surpassed the combined official total planted areas of Indonesia and Vietnam. Despite uncertainties in the Lao data (see Supplementary Methods), their estimate was more than twice the reported figures, even assuming that immature rubber in 2018 (over 50% of the total planted area) matured by 2021. While deviations from official national data in Thailand and Malaysia were smaller (ranging from 4% to 13%), detailed comparisons at the Level 1 administrative divisions revealed substantial errors with considerable regional variation (Fig. 1b). We identified numerous administrative regions —8 in Indonesia, 8 in Thailand, and 32 in Vietnam— where Wang et al . 1 reported the presence of rubber plantations, but official data suggest they do not exist. For example, over 162,000 hectares of mature rubber were mapped in Nusa Tenggara Timur, Indonesia—a region known for lontar palm plantations 9 and lacking any rubber cultivation. This highlights frequent commission errors, where other trees are misclassified as rubber. Specifically, in 27 of 34 provinces in Indonesia, 57 of 77 in Thailand, 53 of 63 in Vietnam, and 14 of 18 in Laos, the mapped rubber areas exceeded statistical data, leading to positive percentage errors (Fig. 1b); remarkably, 55% of these errors exceeded a 100% discrepancy. In Vietnam, the estimated areas were compared with official rubber planted areas—which are typically larger than mature rubber areas—suggesting the actual errors are even greater. Conversely, omission errors (undercounted rubber) ranged from 9% to 71%. A clear pattern emerges: commission errors are prevalent in regions with limited rubber cultivation, while omission errors prevail in areas where rubber is a primary crop (see Supplementary Table 2). Wang et al . 1 adopted an approach to categorize the European Space Agency's (ESA) global tree cover map into rubber and non-rubber, which were used to define the rubber and evergreen forest strata. However, their methodology excluded other monoculture plantations—rapidly expanding crops in Southeast Asia 6 —from their sampling design and accuracy assessment. Our field surveys in Central Vietnam, focusing on acacia plantations (a commonly grown exotic species), revealed consistent misclassification of these plantations as rubber in their map (Fig. 2). The broad distribution of acacia sample units across mountainous and plain regions provides robust evidence of these commission errors. The geographic coordinates of these sample units enable identification of canopy shapes and patterns of the acacia plantations in very high-resolution satellite imagery available on platforms like Google Earth Pro. The spatial scale of acacia misclassified as rubber can be recognized by the continuous connectivity of acacia parcels around the survey sites (as analyzed in Extended Data Fig. 1). Moreover, our analysis, supported by a visual interpretation protocol 10 of high-resolution satellite images, also revealed misclassifications in Indonesia, where coconut, oil palm, acacia, and tropical dry forests were incorrectly identified as rubber (see Supplementary Methods). We suggest that the limitations in Wang et al .'s rubber mapping methodology stem primarily from two factors: the heterogeneity of rubber samples and the spectral resemblance of rubber to other land covers, such as short-rotation plantations and shifting cultivation. A key aspect of this heterogeneity is their presumption of a uniform leaf phenology across mainland Southeast Asia—specifically, leaf shedding in January–February and regrowth in March–April—but our field surveys in Vietnam reveal regional variations, including differences between mountainous and plain areas within the same region (Extended Data Fig. 2). Notably, rubber trees in Southeast Vietnam begin defoliation about a month earlier and have a shorter leaf-changing cycle compared to Central Vietnam. Yearly weather variations and different rubber clones further complicate spectral differentiation. Additionally, inaccuracies in GPS positioning 2 of rubber sample locations (specifically not collected for mapping 5 ) and the presence of understory vegetation (Extended Data Fig. 2c and d) can lead to other trees' spectra being mistaken for rubber. The second major challenge arises from fast-growing trees like acacia, which are clear-cut every 4-7 years and exhibit rubber-like spectral features. For instance, acacia plantations harvested in late 2020 and subsequently replanted or naturally regenerated from March 2021 presented challenges in spectral analysis. Although Wang et al . 1 utilized the 2001 primary forest layer 11 and forest loss data 12 to filter out plantation removal, this primary forest mask was developed explicitly for mapping tropical forests, not fast-growing plantations. While Wang et al . 1 recognize that their map of rubber-related deforestation might include conversions other plantations, agricultural crops, and even rubber plantations established in the 1980s, our analysis suggests that this factor does not fully account for the overestimation. We estimate that in five selected countries, over 2.6 million hectares of non-rubber trees were misclassified as rubber, leading to their inclusion in rubber-related deforestation. The approach of Wang et al . 1 to consider any forest loss detected before the establishment of rubber plantations as deforestation is not justified. Additionally, in Vietnam and Laos (and possibly in other countries), rubber could be planted in areas previously deforested due to war, logging, or slash-and-burn agriculture 13,14 ; thus, rubber planting is not the initial cause of deforestation. While we acknowledge the significance of rubber in zero-deforestation policies, the actual contribution of rubber cultivation to deforestation, especially its extent and geographical distribution, cannot be accurately determined using the results of Wang et al . 1 Consequently, we call for caution in utilizing these maps and highlight the necessity for their revision to ensure more accurate and reliable data for future deforestation analyses. Declarations Data availability All data are taken from the previously published articles referenced in the text. The official statistical data used in this study are available from the sources listed in the Supplementary Information. Code availability For calculating the area, we employed the same Google Earth Engine script provided in the code availability section of the original paper. Acknowledgements This work was supported by Cross-ministerial Strategic Innovation Promotion Program (JPJ012290), the Environment Research and Technology Development Fund (JPMEERF20234004), and JSPS KAKENHI Grant Number JP22K18055. Contributions K.K. and N.T.H. designed the research. N.T.H. performed the field survey and data analysis. N.T.H., K.K., P.P., and P.O. wrote the manuscript. Competing interests The authors declare no competing interests. References Wang Y et al (2023) High-resolution maps show that rubber causes substantial deforestation. Nature 623:340–346 Olofsson P et al (2014) Good practices for estimating area and assessing accuracy of land change. Remote Sens Environ 148:42–57 Global Forest Observations Initiative (2016) Integration of remote-sensing and ground-based observations for estimation of emissions and removals of greenhouse gases in forests: Methods and guidance from the Global Forest Observations Initiative, edition 2.0. UN Food Agric Organ 224:1–224 Stehman SV, Czaplewski RL (1998) Design and Analysis for Thematic Map Accuracy Assessment: Fundamental Principles. Remote Sens Environ 64:331–344 Lan G et al (2022) Main drivers of plant diversity patterns of rubber plantations in the Greater Mekong Subregion. Biogeosciences 19, 1995–2005 Hurni K, Fox J (2018) The expansion of tree-based boom crops in mainland Southeast Asia: 2001 to 2014. J Land Use Sci 13:198–219 Xiao C et al (2021) Latest 30-m map of mature rubber plantations in Mainland Southeast Asia and Yunnan province of China: Spatial patterns and geographical characteristics. Prog Phys Geogr Earth Environ 45:736–756 Chazdon RL et al (2016) When is a forest a forest? Forest concepts and definitions in the era of forest and landscape restoration. Ambio 45:538–550 Cunningham AB, Ingram W, Kadati W, Maduarta IM (2017) Opportunities, barriers and support needs: micro-enterprise and small enterprise development based on non-timber products in eastern Indonesia. Aust For 80:161–177 Petersen R, Goldman ED, Harris N, Sargent S, Aksenov D (2016) Mapping Tree Plantations with Multispectral Imagery: Preliminary Results for Seven Tropical Countries . 1–18 https://www.wri.org/research/mapping-tree-plantations-multispectral-imagery-preliminary-results-seven-tropical Turubanova S, Potapov PV, Tyukavina A, Hansen MC (2018) Ongoing primary forest loss in Brazil, Democratic Republic of the Congo, and Indonesia. Environ Res Lett 13:074028 Hansen MC et al (2013) High-Resolution Global Maps of 21st-Century Forest Cover Change. Science 342:850 Kiyono Y et al (2014) Can converting slash-and-burn agricultural fields into rubber tree (Hevea brasiliensis) plantations provide climate change mitigation? A case study in northern Laos. Bull FFPRI 13:79–88 Cochard R, Ngo DT, Waeber PO, Kull CA (2017) Extent and causes of forest cover changes in Vietnam’s provinces 1993–2013: a review and analysis of official data. Environ Rev 25:199–217 Additional Declarations The authors declare no competing interests. Supplementary Files ExtendedDataFig.1.png Extended Data Fig. 1: A selected example site of Wang et al . 1 ’s failures in classifying rubber plantations. a, Map of a site in Thua Thien Hue, Vietnam, showing Wang et al .’s rubber pixels overlaid on an ArcMap Imagery Hybrid basemap, highlighting correctly and incorrectly classified areas. b, A field photograph taken in March 2023 using a Mavic 2 Enterprise Advanced, covering locations 1, 2, 3, and 4 identified in (a). Location 2 features defoliated immature rubber classified as mature rubber, while locations 1, 3, and 4, consisting of mature acacia and logged acacia (2022), are misclassified as rubber. c,d,e,f, Field photos from November 2019 captured by a Mavic 2 Pro, corresponding to locations 5, 6, 7, and 8 in (a). These photos illustrate misclassifications: acacia and a fragment of natural forest (top right) in (c), acacia in (d) and (f), and acacia mixed with native timber trees in (e), all incorrectly identified as rubber plantations in Wang et al .’s map. ExtendedDataFig.2.jpg Extended Data Fig. 2: Observing seasonal rubber leaf phenology in Vietnam. a,b,c, Photos from March 6–7, 2023, in Quang Nam province, illustrating varying stages of leaf phenology in rubber plantations. In (a), leaves have completely fallen with no new growth; in (b) and (c), early stages of leaf regrowth are visible. d,e, Photos from March 8–9, 2023, in Quang Tri province, further demonstrating phenological variation. Leaves are just beginning to sprout in (d) and have reached about 30% leaf coverage in (e). f, Photo from March 5, 2023, in Dong Nai province, showing rubber trees that have fully regained foliage. Notably, understory vegetation is visible in the rubber plantations depicted in (c) and (d). SupplementaryInformation.pdf Supplementary Information Cite Share Download PDF Status: Published Journal Publication published 12 Aug, 2025 Read the published version in Nature → Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5834237","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":402477767,"identity":"f9a4a67b-ad03-4961-b188-9a294e4ea029","order_by":0,"name":"Nguyen Tien Hoang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABBElEQVRIiWNgGAWjYPCCAxAqgcEGSDI2HiBFSxpISwMJWhgYDqNysQFz/jVmEh933JHjn91++cWDP+ft1rYfBtpSYxONS4vljDdmkjPPPDOWuHOmzCKx7XbytjOJQC3H0nIbcGgxuHHGTJq37XBiw42cNIPEhtvJZgeAWhgbDuPX8heoZT5IS8Kfc8lm5x8S0HK+x0yaEahlw430ww8S2A7Ymd0gaAtbsWVv22Fjwxs5bAyJbckJZjeAtiTg88v5wxtv/Gw7LCd3I/3xxx9/7OzNzqc/fPChxganFgaJBBYJCIvHDMRIBKtMwKUcBPgPMH+AsNgfgxj2+BSPglEwCkbByAQAArhxUA/K2MYAAAAASUVORK5CYII=","orcid":"","institution":"Graduate School of Environmental Studies, Tohoku University","correspondingAuthor":true,"prefix":"","firstName":"Nguyen","middleName":"Tien","lastName":"Hoang","suffix":""},{"id":402478564,"identity":"8506e6b9-2685-4cfe-ad17-85cf486f3b65","order_by":1,"name":"Peter Potapov","email":"","orcid":"","institution":"Department of Geographical Sciences, University of Maryland","correspondingAuthor":false,"prefix":"","firstName":"Peter","middleName":"","lastName":"Potapov","suffix":""},{"id":402478565,"identity":"3b59849c-71a4-4a0e-bdde-1761d736b566","order_by":2,"name":"Pontus Olofsson","email":"","orcid":"","institution":"NASA Marshall Space Flight Center","correspondingAuthor":false,"prefix":"","firstName":"Pontus","middleName":"","lastName":"Olofsson","suffix":""},{"id":402478566,"identity":"01431558-9efb-499f-88c8-79d829bf4c6c","order_by":3,"name":"Keiichiro Kanemoto","email":"data:image/png;base64,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","orcid":"","institution":"Graduate School of Environmental Studies, Tohoku University","correspondingAuthor":true,"prefix":"","firstName":"Keiichiro","middleName":"","lastName":"Kanemoto","suffix":""}],"badges":[],"createdAt":"2025-01-15 12:06:33","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-5834237/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5834237/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41586-025-08847-w","type":"published","date":"2025-08-13T00:00:00+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":74016231,"identity":"75ec7598-b718-4e62-871f-3c88abe0a186","added_by":"auto","created_at":"2025-01-17 04:00:47","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1403539,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eRubber area comparison between official statistics and Wang \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003eet al\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e.’s map for five example countries. a,\u003c/strong\u003e Wang \u003cem\u003eet al\u003c/em\u003e.’s estimated rubber areas versus official statistical data (2019–2021) for Indonesia, Thailand, Vietnam, Malaysia, and semi-official data (2018, 2020) for Laos. \u003cstrong\u003eb,\u003c/strong\u003e Spatial distribution of percentage errors in Wang \u003cem\u003eet al\u003c/em\u003e.’s rubber map at Level 1 administrative division (provinces in Indonesia, Thailand, Vietnam, and Laos; states in Malaysia), using official 2021 statistics as benchmarks (see Supplementary Tables 2 and 3). Benchmarks include mature rubber areas for Indonesia and Thailand, and planted areas for Vietnam and Malaysia. In Malaysia, states within the same region share identical percentage errors due to regional reporting. For Laos, we assumed all rubber trees immature in 2018 matured by 2021. Administrative boundaries are from the Global Administrative Areas (GADM) Version 4.1 database, offering updated boundaries compared to the Global Administrative Unit Layers used in Wang \u003cem\u003eet al\u003c/em\u003e.’s study.\u003c/p\u003e","description":"","filename":"FIG1.png","url":"https://assets-eu.researchsquare.com/files/rs-5834237/v1/c1bfed053884d65c8d8ab88d.png"},{"id":74016235,"identity":"fb3741e6-b1dd-4626-84b0-bc62b2edcea9","added_by":"auto","created_at":"2025-01-17 04:00:47","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":9839152,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMisclassification of acacia plantations as rubber in Wang \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003eet al\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e.’s map. \u003c/strong\u003eSpatial distribution of acacia survey points (see Supplementary Table 6) overlaid on Wang \u003cem\u003eet al\u003c/em\u003e.’s rubber map layer and Shuttle Radar Topography Mission (SRTM) shaded relief. Each point represents an acacia monoculture plantation incorrectly classified as a rubber plantation. The survey area covers Thua Thien Hue and Quang Nam provinces and Da Nang city in Central Vietnam. Additionally, field photos from a representative survey site can be found in Extended Data Fig. 1, offering a visual example of the misclassified areas.\u003c/p\u003e","description":"","filename":"FIG2.png","url":"https://assets-eu.researchsquare.com/files/rs-5834237/v1/29217ed1cdbc886a3146f3aa.png"},{"id":89692954,"identity":"941f713c-6af2-433b-970c-2c82dbb24705","added_by":"auto","created_at":"2025-08-22 17:15:52","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":10742424,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5834237/v1/923e2cd4-b56b-4437-b282-654681df2ae6.pdf"},{"id":74016234,"identity":"293f8274-724f-40b0-91f7-05cf5deba390","added_by":"auto","created_at":"2025-01-17 04:00:47","extension":"png","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":9236869,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eExtended Data Fig. 1: A selected example site of Wang \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003eet al\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e.\u003c/strong\u003e\u003csup\u003e\u003cstrong\u003e1\u003c/strong\u003e\u003c/sup\u003e\u003cstrong\u003e’s failures in classifying rubber plantations. a,\u003c/strong\u003e Map of a site in Thua Thien Hue, Vietnam, showing Wang \u003cem\u003eet al\u003c/em\u003e.’s rubber pixels overlaid on an ArcMap Imagery Hybrid basemap, highlighting correctly and incorrectly classified areas. \u003cstrong\u003eb,\u003c/strong\u003e A field photograph taken in March 2023 using a Mavic 2 Enterprise Advanced, covering locations 1, 2, 3, and 4 identified in (\u003cstrong\u003ea\u003c/strong\u003e). Location 2 features defoliated immature rubber classified as mature rubber, while locations 1, 3, and 4, consisting of mature acacia and logged acacia (2022), are misclassified as rubber. \u003cstrong\u003ec,d,e,f,\u003c/strong\u003e Field photos from November 2019 captured by a Mavic 2 Pro, corresponding to locations 5, 6, 7, and 8 in (\u003cstrong\u003ea\u003c/strong\u003e). These photos illustrate misclassifications: acacia and a fragment of natural forest (top right) in (\u003cstrong\u003ec\u003c/strong\u003e), acacia in (\u003cstrong\u003ed\u003c/strong\u003e) and (\u003cstrong\u003ef\u003c/strong\u003e), and acacia mixed with native timber trees in (\u003cstrong\u003ee\u003c/strong\u003e), all incorrectly identified as rubber plantations in Wang \u003cem\u003eet al\u003c/em\u003e.’s map.\u003c/p\u003e","description":"","filename":"ExtendedDataFig.1.png","url":"https://assets-eu.researchsquare.com/files/rs-5834237/v1/2ded8cddb4df78e5741e3119.png"},{"id":74016242,"identity":"503ad020-1f7a-43b2-8929-28b161143876","added_by":"auto","created_at":"2025-01-17 04:00:47","extension":"jpg","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":8200110,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eExtended Data Fig. 2: Observing seasonal rubber leaf phenology in Vietnam. a,b,c,\u003c/strong\u003e Photos from March 6–7, 2023, in Quang Nam province, illustrating varying stages of leaf phenology in rubber plantations. In (\u003cstrong\u003ea\u003c/strong\u003e), leaves have completely fallen with no new growth; in (\u003cstrong\u003eb\u003c/strong\u003e) and (\u003cstrong\u003ec\u003c/strong\u003e), early stages of leaf regrowth are visible. \u003cstrong\u003ed,e,\u003c/strong\u003e Photos from March 8–9, 2023, in Quang Tri province, further demonstrating phenological variation. Leaves are just beginning to sprout in (\u003cstrong\u003ed\u003c/strong\u003e) and have reached about 30% leaf coverage in (\u003cstrong\u003ee\u003c/strong\u003e). \u003cstrong\u003ef,\u003c/strong\u003e Photo from March 5, 2023, in Dong Nai province, showing rubber trees that have fully regained foliage. Notably, understory vegetation is visible in the rubber plantations depicted in (\u003cstrong\u003ec\u003c/strong\u003e) and (\u003cstrong\u003ed\u003c/strong\u003e).\u003c/p\u003e","description":"","filename":"ExtendedDataFig.2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5834237/v1/fe19f1930802530af1cf1d7a.jpg"},{"id":74016245,"identity":"e520c47f-5355-413a-97fd-c9233398ba3e","added_by":"auto","created_at":"2025-01-17 04:00:47","extension":"pdf","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":652292,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSupplementary Information\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"SupplementaryInformation.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5834237/v1/e30920ebed68c27e1032f2c7.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003e\u003cstrong\u003eAccuracy of rubber-related deforestation maps\u003c/strong\u003e\u003c/p\u003e","fulltext":[{"header":"Full Text","content":"\u003cp\u003eThe questionable accuracy of the maps published by Wang \u003cem\u003eet al\u003c/em\u003e.\u003csup\u003e1\u003c/sup\u003e arises from their approach to area estimation and accuracy assessment. Despite their claim of adhering to \u0026ldquo;good practices,\u0026rdquo;\u003csup\u003e2\u003c/sup\u003e we contend that these practices were misapplied. First, estimating areas by pixel-counting is biased and produces erroneous estimates because it ignores map classification errors\u003csup\u003e3\u003c/sup\u003e. Second, the sample data used for estimating map accuracy were not selected using probability sampling\u0026mdash;a critical component for robust statistical evaluation\u003csup\u003e4\u003c/sup\u003e. Wang \u003cem\u003eet al\u003c/em\u003e.\u003csup\u003e1\u003c/sup\u003e pooled over 3,800 sample points from three independent sources\u0026mdash;the World Agroforestry Centre, Lan et al.\u003csup\u003e5\u003c/sup\u003e, and Hurni and Fox\u003csup\u003e6\u003c/sup\u003e\u0026mdash;but this aggregation does not adhere to probability sampling rules across the entire study area. Probability sampling requires that each unit in the population has a known, non-zero inclusion probability\u003csup\u003e2,4\u003c/sup\u003e, which was not ensured. For example, one sample was drawn only from areas identified as rubber and evergreen forest, resulting in an inclusion probability of zero for all other units in the study area. Moreover, the in-situ observations from Xishuangbanna (China) and the Greater Mekong Subregion do not appear to have been selected under probability sampling. The geographical distribution of their sample data\u0026mdash;predominantly in mainland Southeast Asia and notably absent in large areas of Indonesia, Myanmar, and the North Central Coast to North Vietnam\u0026mdash;further supports this argument. Because the accuracy measures reported by Wang \u003cem\u003eet al\u003c/em\u003e.\u003csup\u003e1\u003c/sup\u003e rely on a design-based inference framework that requires probability sampling\u003csup\u003e4\u003c/sup\u003e, their lack of adherence invalidates the area and map accuracy results they provided.\u003c/p\u003e\n\u003cp\u003eWang \u003cem\u003eet al\u003c/em\u003e.\u003csup\u003e1\u003c/sup\u003e defend their rubber area estimates by citing two key points: first, they attribute the discrepancy with the harvested rubber area reported by the Forest and Agriculture Organization of the United Nations (FAO) to unharvested plantations resulting from low global rubber prices. However, this claim lacks empirical support, such as statistical data comparing harvested and mature rubber areas. Second, they argue that their estimates align with those from two other recent remote sensing studies\u003csup\u003e6,7\u003c/sup\u003e. Yet, the maps published in these two studies\u0026mdash;referred to by Wang \u003cem\u003eet al\u003c/em\u003e.\u003csup\u003e1\u003c/sup\u003e as reference data \u0026mdash;are of lower resolution than their own maps, which results in a mismatch in data quality that undermines the validity of using these maps as a standard for comparison. Furthermore, reference data should represent the best available assessment of actual conditions on the ground\u003csup\u003e3\u003c/sup\u003e. Hence, using wall-to-wall maps as reference data is problematic as it implies that the maps constitute a complete census of reference conditions of the study area.\u003c/p\u003e\n\u003cp\u003eTo investigate the accuracy of Wang \u003cem\u003eet al\u003c/em\u003e.\u0026rsquo;s rubber maps, we utilized comprehensive and comparable official statistical data from the top four rubber-producing countries in Southeast Asia: Indonesia, Thailand, Vietnam, and Malaysia (see Supplementary Tables 1-3). These countries report total rubber planted areas annually, with Indonesia, Thailand, and Vietnam providing detailed data at the Level 2 administrative units (districts and equivalents). Notably, Indonesia and Thailand report areas of mature rubber trees (ready for tapping), while Vietnam provides harvested rubber area data. For Malaysia, only regional total rubber planted areas are available. In addition, we examined mapping accuracy in Laos\u0026mdash;the country with the smallest rubber area in Wang \u003cem\u003eet al\u003c/em\u003e.'s estimate\u0026mdash;using semi-official data on rubber planted and mature areas. It is important to note that statistical data, grounded in a land-use perspective, can provide higher area estimates than remote sensing data, which are rooted in a land-cover perspective. The discrepancy arises because the land use-based estimate tends to align with parcel boundaries rather than actual tree cover\u003csup\u003e8\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eWang \u003cem\u003eet al\u003c/em\u003e.\u003csup\u003e1\u003c/sup\u003e used Sentinel-2 imagery from 2020 to 2022 to map rubber plantations in 2021. Official statistics from 2019 to 2021 showed stable total planted, mature, and harvested rubber areas in the top four rubber-producing countries\u0026mdash;Indonesia, Thailand, Vietnam, and Malaysia (Fig. 1a). Based on this stability, we used the official 2021 data as our benchmark. Theoretically, the remotely sensed-based mature rubber area should be smaller than the reported planted or mature areas and larger than the harvested area. However, Wang \u003cem\u003eet al\u003c/em\u003e.'s estimates were significantly higher\u0026mdash;exceeding official total planted areas by 26% in Indonesia and 73% in Vietnam. Remarkably, their estimated mature rubber area for Indonesia alone surpassed the combined official total planted areas of Indonesia and Vietnam. Despite uncertainties in the Lao data (see Supplementary Methods), their estimate was more than twice the reported figures, even assuming that immature rubber in 2018 (over 50% of the total planted area) matured by 2021. While deviations from official national data in Thailand and Malaysia were smaller (ranging from 4% to 13%), detailed comparisons at the Level 1 administrative divisions revealed substantial errors with considerable regional variation (Fig. 1b).\u003c/p\u003e\n\u003cp\u003eWe identified numerous administrative regions \u0026mdash;8 in Indonesia, 8 in Thailand, and 32 in Vietnam\u0026mdash; where Wang \u003cem\u003eet al\u003c/em\u003e.\u003csup\u003e1\u003c/sup\u003e reported the presence of rubber plantations, but official data suggest they do not exist. For example, over 162,000 hectares of mature rubber were mapped in Nusa Tenggara Timur, Indonesia\u0026mdash;a region known for lontar palm plantations\u003csup\u003e9\u003c/sup\u003e and lacking any rubber cultivation. This highlights frequent commission errors, where other trees are misclassified as rubber. Specifically, in 27 of 34 provinces in Indonesia, 57 of 77 in Thailand, 53 of 63 in Vietnam, and 14 of 18 in Laos, the mapped rubber areas exceeded statistical data, leading to positive percentage errors (Fig. 1b); remarkably, 55% of these errors exceeded a 100% discrepancy. In Vietnam, the estimated areas were compared with official rubber planted areas\u0026mdash;which are typically larger than mature rubber areas\u0026mdash;suggesting the actual errors are even greater. Conversely, omission errors (undercounted rubber) ranged from 9% to 71%. A clear pattern emerges: commission errors are prevalent in regions with limited rubber cultivation, while omission errors prevail in areas where rubber is a primary crop (see Supplementary Table 2).\u003c/p\u003e\n\u003cp\u003eWang \u003cem\u003eet al\u003c/em\u003e.\u003csup\u003e1\u003c/sup\u003e adopted an approach to categorize the European Space Agency's (ESA) global tree cover map into rubber and non-rubber, which were used to define the rubber and evergreen forest strata. However, their methodology excluded other monoculture plantations\u0026mdash;rapidly expanding crops in Southeast Asia\u003csup\u003e6\u003c/sup\u003e \u0026mdash;from their sampling design and accuracy assessment. Our field surveys in Central Vietnam, focusing on acacia plantations (a commonly grown exotic species), revealed consistent misclassification of these plantations as rubber in their map (Fig. 2). The broad distribution of acacia sample units across mountainous and plain regions provides robust evidence of these commission errors. The geographic coordinates of these sample units enable identification of canopy shapes and patterns of the acacia plantations in very high-resolution satellite imagery available on platforms like Google Earth Pro. The spatial scale of acacia misclassified as rubber can be recognized by the continuous connectivity of acacia parcels around the survey sites (as analyzed in Extended Data Fig. 1). Moreover, our analysis, supported by a visual interpretation protocol\u003csup\u003e10\u003c/sup\u003e of high-resolution satellite images, also revealed misclassifications in Indonesia, where coconut, oil palm, acacia, and tropical dry forests were incorrectly identified as rubber (see Supplementary Methods).\u003c/p\u003e\n\u003cp\u003eWe suggest that the limitations in Wang \u003cem\u003eet al\u003c/em\u003e.'s rubber mapping methodology stem primarily from two factors: the heterogeneity of rubber samples and the spectral resemblance of rubber to other land covers, such as short-rotation plantations and shifting cultivation. A key aspect of this heterogeneity is their presumption of a uniform leaf phenology across mainland Southeast Asia\u0026mdash;specifically, leaf shedding in January\u0026ndash;February and regrowth in March\u0026ndash;April\u0026mdash;but our field surveys in Vietnam reveal regional variations, including differences between mountainous and plain areas within the same region (Extended Data Fig. 2). Notably, rubber trees in Southeast Vietnam begin defoliation about a month earlier and have a shorter leaf-changing cycle compared to Central Vietnam. Yearly weather variations and different rubber clones further complicate spectral differentiation. Additionally, inaccuracies in GPS positioning\u003csup\u003e2\u003c/sup\u003e of rubber sample locations (specifically not collected for mapping\u003csup\u003e5\u003c/sup\u003e) and the presence of understory vegetation (Extended Data Fig. 2c and d) can lead to other trees' spectra being mistaken for rubber. The second major challenge arises from fast-growing trees like acacia, which are clear-cut every 4-7 years and exhibit rubber-like spectral features. For instance, acacia plantations harvested in late 2020 and subsequently replanted or naturally regenerated from March 2021 presented challenges in spectral analysis. Although Wang \u003cem\u003eet al\u003c/em\u003e.\u003csup\u003e1\u003c/sup\u003e utilized the 2001 primary forest layer\u003csup\u003e11\u003c/sup\u003e and forest loss data\u003csup\u003e12\u003c/sup\u003e to filter out plantation removal, this primary forest mask was developed explicitly for mapping tropical forests, not fast-growing plantations.\u003c/p\u003e\n\u003cp\u003eWhile Wang \u003cem\u003eet al\u003c/em\u003e.\u003csup\u003e1\u003c/sup\u003e recognize that their map of rubber-related deforestation might include conversions other plantations, agricultural crops, and even rubber plantations established in the 1980s, our analysis suggests that this factor does not fully account for the overestimation. We estimate that in five selected countries, over 2.6 million hectares of non-rubber trees were misclassified as rubber, leading to their inclusion in rubber-related deforestation. The approach of Wang \u003cem\u003eet al\u003c/em\u003e.\u003csup\u003e1\u003c/sup\u003e to consider any forest loss detected before the establishment of rubber plantations as deforestation is not justified. Additionally, in Vietnam and Laos (and possibly in other countries), rubber could be planted in areas previously deforested due to war, logging, or slash-and-burn agriculture\u003csup\u003e13,14\u003c/sup\u003e; thus, rubber planting is not the initial cause of deforestation. While we acknowledge the significance of rubber in zero-deforestation policies, the actual contribution of rubber cultivation to deforestation, especially its extent and geographical distribution, cannot be accurately determined using the results of Wang \u003cem\u003eet al\u003c/em\u003e.\u003csup\u003e1\u003c/sup\u003e Consequently, we call for caution in utilizing these maps and highlight the necessity for their revision to ensure more accurate and reliable data for future deforestation analyses.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll data are taken from the previously published articles referenced in the text. The official statistical data used in this study are available from the sources listed in the Supplementary Information.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCode availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFor calculating the area, we employed the same Google Earth Engine script provided in the code availability section of the original paper.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by Cross-ministerial Strategic Innovation Promotion Program (JPJ012290), the Environment Research and Technology Development Fund (JPMEERF20234004), and JSPS KAKENHI Grant Number JP22K18055.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eContributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eK.K. and N.T.H. designed the research. N.T.H. performed the field survey and data analysis. N.T.H., K.K., P.P., and P.O. wrote the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eWang Y et al (2023) High-resolution maps show that rubber causes substantial deforestation. Nature 623:340\u0026ndash;346\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOlofsson P et al (2014) Good practices for estimating area and assessing accuracy of land change. Remote Sens Environ 148:42\u0026ndash;57\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGlobal Forest Observations Initiative (2016) Integration of remote-sensing and ground-based observations for estimation of emissions and removals of greenhouse gases in forests: Methods and guidance from the Global Forest Observations Initiative, edition 2.0. UN Food Agric Organ 224:1\u0026ndash;224\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStehman SV, Czaplewski RL (1998) Design and Analysis for Thematic Map Accuracy Assessment: Fundamental Principles. Remote Sens Environ 64:331\u0026ndash;344\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLan G et al (2022) Main drivers of plant diversity patterns of rubber plantations in the Greater Mekong Subregion. \u003cem\u003eBiogeosciences\u003c/em\u003e 19, 1995\u0026ndash;2005\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHurni K, Fox J (2018) The expansion of tree-based boom crops in mainland Southeast Asia: 2001 to 2014. J Land Use Sci 13:198\u0026ndash;219\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXiao C et al (2021) Latest 30-m map of mature rubber plantations in Mainland Southeast Asia and Yunnan province of China: Spatial patterns and geographical characteristics. Prog Phys Geogr Earth Environ 45:736\u0026ndash;756\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChazdon RL et al (2016) When is a forest a forest? Forest concepts and definitions in the era of forest and landscape restoration. Ambio 45:538\u0026ndash;550\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCunningham AB, Ingram W, Kadati W, Maduarta IM (2017) Opportunities, barriers and support needs: micro-enterprise and small enterprise development based on non-timber products in eastern Indonesia. Aust For 80:161\u0026ndash;177\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePetersen R, Goldman ED, Harris N, Sargent S, Aksenov D (2016) \u003cem\u003eMapping Tree Plantations with Multispectral Imagery: Preliminary Results for Seven Tropical Countries\u003c/em\u003e. 1\u0026ndash;18 \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.wri.org/research/mapping-tree-plantations-multispectral-imagery-preliminary-results-seven-tropical\u003c/span\u003e\u003cspan address=\"https://www.wri.org/research/mapping-tree-plantations-multispectral-imagery-preliminary-results-seven-tropical\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTurubanova S, Potapov PV, Tyukavina A, Hansen MC (2018) Ongoing primary forest loss in Brazil, Democratic Republic of the Congo, and Indonesia. Environ Res Lett 13:074028\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHansen MC et al (2013) High-Resolution Global Maps of 21st-Century Forest Cover Change. Science 342:850\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKiyono Y et al (2014) Can converting slash-and-burn agricultural fields into rubber tree (Hevea brasiliensis) plantations provide climate change mitigation? A case study in northern Laos. Bull FFPRI 13:79\u0026ndash;88\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCochard R, Ngo DT, Waeber PO, Kull CA (2017) Extent and causes of forest cover changes in Vietnam\u0026rsquo;s provinces 1993\u0026ndash;2013: a review and analysis of official data. Environ Rev 25:199\u0026ndash;217\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":true,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Hevea brasiliensis, mapping accuracy, Sentinel-2, EUDR","lastPublishedDoi":"10.21203/rs.3.rs-5834237/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5834237/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eAccurate mapping of rubber plantations is essential to understanding where deforestation due to rubber production occurs. Wang \u003cem\u003eet al\u003c/em\u003e.\u003csup\u003e1\u003c/sup\u003e produced 10-meter resolution maps of mature rubber with a reported high overall classification accuracy (OA = 0.95 ± 0.02), indicating that more than 4 million hectares of forests have been converted to rubber plantations in Southeast Asia and parts of China since 1993. However, we have serious concerns about the accuracy of these maps, as our analysis indicates that the area estimates by Wang \u003cem\u003eet al\u003c/em\u003e.\u003csup\u003e1\u003c/sup\u003e are significantly inflated. This observation underscores the need for caution in using these high-resolution maps for specific applications, such as assessing deforestation linked to rubber consumption under the EU regulation on deforestation-free products.\u003c/p\u003e","manuscriptTitle":"Accuracy of rubber-related deforestation maps","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-01-17 04:00:42","doi":"10.21203/rs.3.rs-5834237/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"c6819431-8639-4bfb-b192-2f4f4a0529c2","owner":[],"postedDate":"January 17th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":42902246,"name":"Forestry"},{"id":42902247,"name":"Environmental Economics"},{"id":42902248,"name":"Environmental Policy"}],"tags":[],"updatedAt":"2025-08-22T17:15:40+00:00","versionOfRecord":{"articleIdentity":"rs-5834237","link":"https://doi.org/10.1038/s41586-025-08847-w","journal":{"identity":"nature","isVorOnly":false,"title":"Nature"},"publishedOn":"2025-08-13 00:00:00","publishedOnDateReadable":"August 13th, 2025"},"versionCreatedAt":"2025-01-17 04:00:42","video":"","vorDoi":"10.1038/s41586-025-08847-w","vorDoiUrl":"https://doi.org/10.1038/s41586-025-08847-w","workflowStages":[]},"version":"v1","identity":"rs-5834237","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5834237","identity":"rs-5834237","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}