Geospatial data layer of seminatural grasslands with century-old or longer temporal continuity

Geospatial data layer of seminatural grasslands with century-old or longer temporal continuity | Authorea try { document.documentElement.classList.add('js'); } catch (e) { } var _gaq = _gaq || []; _gaq.push(['_setAccount', 'G-8VDV14Y67G']); _gaq.push(['_trackPageview']); (function() { var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); })(); Skip to main content Preprints Collections Wiley Open Research IET Open Research Ecological Society of Japan All Collections About About Authorea FAQs Contact Us Quick Search anywhere Search for preprint articles, keywords, etc. Search Search ADVANCED SEARCH SCROLL This is a preprint and has not been peer reviewed. Data may be preliminary. 8 January 2026 V1 Latest version Share on Geospatial data layer of seminatural grasslands with century-old or longer temporal continuity Authors : Shogo Ikari 0000-0002-1318-7105 [email protected] , Mahoro Tomitaka , Yasuhiro Kubota , and Kenta Tanaka 0000-0002-6234-1017 Authors Info & Affiliations https://doi.org/10.22541/au.176784129.94226004/v1 271 views 96 downloads Contents Abstract Information & Authors Metrics & Citations View Options References Figures Tables Media Share Abstract Grasslands in Japan, once covering over 10% of the national land, have rapidly declined due to land-use change and the abandonment of traditional management, resulting in the loss of biodiversity and associated ecosystem services. Because the ecological value of grasslands is closely tied to their temporal continuity, identifying long-persisting sites is crucial for effective conservation and restoration planning. However, such assessments have so far been limited to local or regional scales due to the lack of comprehensive historical datasets. This study presents a national-scale geospatial dataset of century-old seminatural grasslands in Japan, constructed by integrating multiple sources of historical and contemporary spatial data. We combined long-term land-use and vegetation information to trace changes in grassland distribution over more than a century, linking modern high-resolution datasets with historical land-use maps and expert knowledge on long-maintained grasslands. By assessing the overlap between historical and recent grassland distributions, we identified areas that have persisted for over 100 years and remain maintained as seminatural systems through human management. The resulting dataset comprises 2,654 polygons representing seminatural grasslands with over a century of continuity across Japan. This dataset provides the first nationwide distribution records of grasslands with long term temporal continuity, supporting conservation prioritization, biodiversity monitoring, and landscape-scale restoration efforts of grasslands. The data are openly available under non-commercial license in GIS-compatible formats to facilitate broad research and policy applications. 1. INTRODUCTION Degradation of grassland environments has led to the loss of biodiversity and associated ecosystem services worldwide (Bardgett et al. 2021). Grassland ecosystems are maintained under conditions where dry climates, natural disturbances such as wildfire, flooding and herbivory, or continuous human management prevent the establishment of woody species (Pärtel et al. 2005; Veldman et al. 2015). Consequently, grasslands are ecosystems that are highly susceptible to loss when once the balance of such continuous disturbance regimes is altered (Veldman et al. 2015; Bardgett et al. 2021). Despite such vulnerability, grasslands remain underrepresented in conservation efforts and are at risk of disappearing in many parts of the world (Pillar & Overbeck 2025). Grassland environments are especially vulnerable to loss in humid regions, where climatic conditions favor tree growth (Habel et al. 2013; Ohwaki 2018). In such regions, seminatural grasslands maintained through continuous human management play an essential role in sustaining grassland biodiversity and the benefits they provide to humans (Ushimaru et al. 2018). Among them, seminatural grasslands in Japan have provided crucial habitats for grassland flora and fauna composed of continental relics and island endemics (Koyama et al. 2018; Ohwaki 2018; Ushimaru et al. 2018; Nakahama et al. 2023). However, seminatural grasslands in Japan, which once covered more than 10% of the national land a century ago, have been rapidly disappearing and fragmented (Ogura 2006) and the estimate of recent-day grassland cover of Japan ranges from 1~3% (Tomitaka et al. 2025). Abandonment of management due to decline of resource usage from grasslands and afforestation. promoted by the government have been considered as a driver of grassland loss (Ushimaru et al. 2018). Consequently, many of grassland species in Japan are threatened (Koyanagi & Furukawa 2013). Temporal continuity has been increasingly recognized as a promising surrogate for representing the ecological value of grassland ecosystems (e.g., Pornon & Andalo 2023). Studies have revealed that seminatural grasslands with long-term temporal continuity serve as critical habitats for numerous threatened species (Uchida et al. 2016; Inoue et al. 2021). Further, while the effects of long-term continuity remain untested, recent studies indicate that grassland continuity over time is linked to ecosystem services such as drought tolerance (Iepema et al. 2022) and soil carbon sequestration (Iepema et al. 2022; Elias et al. 2023). Moreover, it has been shown that biodiversity of grasslands once subjected to soil alteration remains degraded for several decades (Tsutsumi et al. 2022), and that 75 years of management in seminatural grasslands is insufficient to restore species richness and interactions to levels comparable to those of centuries-old grasslands (Hirayama et al. 2025). Those findings are indicative of the limited reversibility of species-rich grassland ecosystems once degraded, advocating the prioritization of remaining grasslands with long temporal continuity as key targets for conservation. Accordingly, knowing spatial distribution of long-persisting grasslands can enhance the effectiveness of various conservation and restoration actions of grasslands, enabling (1) the prioritization of areas that retain old grasslands, which can guide the strategic allocation of conservation resources, and (2) the planning of restoration efforts in regions where old grasslands are concentrated, as these areas may offer higher potential for the recovery of adjacent new grasslands. Although those relevance in conservation plannings, studies that visualize grasslands with long-term temporal continuity have been conducted only at regional scales (e.g., Inoue et al. 2021; Ohta et al. 2022), and maps of this kind do not exist at the national level. National-scale land-use maps are a valuable data source for tracing the historical spatial distribution of grasslands. In Japan, government-produced land-use maps have existed for decades to monitor land utilization (Geospatial Information Authority of Japan; https://nlftp.mlit.go.jp/ksj/gml/datalist/KsjTmplt-L03-b-2021.html). However, seminatural grasslands in Japan are maintained for various purposes, such as pastures, hayfields, ski resorts, and military training grounds (Ushimaru et al. 2018). Since land-use maps do not aim to capture distributions of wildlife habitats, seminatural vegetations have been classified differently across datasets. Alternatively, Japanese Ministry of the Environment has made vegetation maps available as GIS data, providing spatial information from around 1980 and from the 2000s onward, which enables the identification of seminatural grassland vegetation distribution at two time points. Nevertheless, these two timepoints alone are insufficient to determine whether a grassland has been continuously maintained over a long period. Recent efforts to visualize vegetation based on satellite imagery, which can trace change in grassland distribution with shorter time intervals (Parente et al. 2024), also face limitations: they cannot reconstruct conditions from periods before satellite data became available, and training data for regions like Japan, where grasslands are relatively limited, remains sparse (Parente et al. 2024) hence limited accuracy in the region. Thus, none of a single existing data source is sufficient to identify the distribution of seminatural grasslands with long-term continuity in Japan. Here, the current study integrated different available data sources to reveal the distribution of seminatural grasslands that have persisted for over a century. We primarily relied on nation-wide land-use maps provided by Geospatial Information Authority of Japan for nine time points since 1976 to visualize a time-series of the distribution of grasslands in Japan at a 100-meter grid scale, verifying the absence of forests through satellite-based canopy cover data (Hansen et al. 2013) and absence of artificial land cover with high-resolution land cover map as of 2020 (Hirayama et al. 2022). Vegetation information that could not be obtained from land-use maps alone (distributions of grazing fields and naturality/seminaturality of grasslands) was supplemented using vegetation maps. Additionally, ski resorts, military training grounds, and pastures, which are not categorized as grasslands in the standard land-use classification, were separately identified and integrated to visualize seminatural grasslands. Furthermore, to assess the century-scale continuity of grasslands, we utilized historical land-use maps (Arizono 1995; Himiyama 1995) available at a 2-kilometer grid scale for the years 1850 and 1900 as well as expert-driven knowledge on grassland continuities across Japan. 2. DATA DESCRIPTION 2.1 Identifier ####### 2.2 Contributor 2.2.1 Dataset owner Think Nature Inc., Okinawa, Japan 2.2.2 Dataset creator Shogo Ikari, Think Nature Inc. Yasuhiro Kubota, Think Nature Inc. 2.2.3 Contact person Shogo Ikari, Think Nature Inc. 202 Ocean Current, 3 -15-10 Maeda, Urasoe City, Okinawa, 901-2102, Japan Yasuhiro Kubota, Think Nature Inc. 202 Ocean Current, 3 -15-10 Maeda, Urasoe City, Okinawa, 901-2102, Japan 2.3 Geographic coverage 2.3.1 Geographic description Japan 2.4 Temporal coverage 2.4.1 Begin 2022 2.4.2 End 2022 2.5 Taxonomic coverage None 2.6 Method 2.6.1 Overview 2.6.1.1 Main data sources Figure 1 describes overview of the procedure used to extract century-old seminatural grasslands. The land-use datasets used to identify the distribution of century-old grasslands include the 2022 land-use data (hereafter ALOS-LU; 10 m resolution, derived from satellite imagery https://www.eorc.jaxa.jp/ALOS/jp/dataset/lulc/lulc_v2312_j.htm); the time-series land-use subdivision mesh data published by the Japanese Ministry of Land, Infrastructure, Transport and Tourism for the years 1976, 1987, 1991, 1997, 2006, 2009, 2014, 2016, and 2021 (hereafter MILT-LU; 100 m resolution https://nlftp.mlit.go.jp/ksj/gml/datalist/KsjTmplt-L03-b-2021.html) and the historical land-use data for years of 1850 and 1900 (hereafter LUIS-LU; 2km resolution; Arizono 1995; Himiyama 1995; https://db.cger.nies.go.jp/dataset/luis/ja/). Table 1 shows the main data sources for land use layers. In addition to these land-use maps, we also integrated vegetation maps (from the 1980s and 2000s) and forest distribution maps (for 1992 and after 2000). To exclude mining areas, visual inspection using aerial photographs was used. Further, to supplement the old-time distribution of grasslands, we relied on online data sources, which list century-old grasslands based on expert knowledge. Usage of these data sources are described below. 2.6.1.2 Spatial resolution for data processing We adopted a spatial resolution of approximately 1 ha (100 m × 100 m). This scale balances the precision of high-resolution datasets with the operational relevance for policy and land management decisions. In addition, the most frequently available MILT-LU land-use data have a spatial resolution of 100 m, which further supports the appropriateness of this resolution in terms of information accuracy. 2.6.1.3 Temporal coverage The temporal continuity that could be traced in this dataset extends back only about 100 years (since circa 1900). However, among the grasslands identified here are sites, such as those in the Aso region, for which a grassland history spanning nearly 10,000 years has been suggested (Kawano et al. 2012). 2.6.2 Extraction of Land-Use Types with Potential for Seminatural Grasslands Using Land-Use Maps 2.6.2.1 Current land cover The 2022 land-use map was used to identify the current distribution of grasslands; we extracted grids classified as grassland or wetland in the ALOS-LU. Wetlands were included because preliminary visual inspection revealed that many areas known to be seminatural grasslands were classified as wetlands. Since the dataset has a resolution of 10 m, we aggregated it to 100 m resolution to match the land-use maps and extracted grids in which the combined proportion of grassland and wetland exceeded 50%. Note that this dataset does not distinguish between natural grasslands, seminatural grasslands, and pastures. 2.6.2.2 Time series of land covers From MILT-LU data available for 9-year points, we extracted the following categories as land-use types with potential for seminatural grasslands: wasteland (includes non-forest/non-built land cover such as grasslands and open-pit mines, excluding agricultural land) and other land use (includes ski resorts and military training areas). Each category contains land types other than grasslands and does not distinguish between natural grasslands and seminatural grasslands. Furthermore, omission and misclassification also occur in the map generation. Specifically, in MILT-LU maps, land use types were determined based on map symbols and the color tones of satellite imagery. Among areas identified as vegetated, distinguishing between forest and grassland is prone to error because both appear green in tone. In addition, map symbols may fail to represent small habitat patches. Thus, we applied the below-mentioned procedures to identify time series distributions of grasslands. 2.6.3. Supplementing Overlooked Grassland in MILT-LU Using Other Data Sources 2.6.3.1 Vegetation maps Grasslands that are overlooked in MILT-LU maps may be captured in vegetation maps. In Japan, the Ministry of the Environment has published vegetation maps for two time periods: the 1980s (1:50,000 scale) and after 2000 (1:25,000 scale). Since these datasets are provided as polygon data, we rasterized them by using the center points of 100 m grids. In the vegetation map, each vegetation type is represented by a ten-level naturalness index ranging from 1(lowest) to 10(highest). According to the legend of the vegetation map, seminatural grasslands are classified as having a naturalness level of 8 or lower. Therefore, to extract seminatural grasslands from the vegetation map, we extracted grids representing grassland environments with a vegetation naturalness index of 8 or lower for both time periods. The classification of vegetation category was based on vegetation classification in Tomitaka et al. (2025). Finally, we identified grids that were classified as grassland environments in both time periods. 2.6.3.2 Forest cover maps Furthermore, to capture small-scale grasslands that are not represented in neither MILT-LU maps nor vegetation maps, we utilized satellite-based forest distribution data. Specifically, we used the forest distribution derived from the 1992 land cover map (300 m resolution) published by ESA CCI-LC (https://www.esa-landcover-cci.org/), as well as the 30 m resolution forest distribution data by Hansen et al. (2013) available for 2000 onward. The former provides discrete values (forest vs. non-forest), while the latter represents the proportion of forest cover. For 1976, 1987, and 1991, grids that were classified as forest in MILT-LU maps but identified as non-forest in the 1992 ESA CCI-LC data were considered grassland grids. For subsequent years, we used the Hansen et al. dataset, resampled to 100 m resolution. For 1997, we substituted the data with those from 2000. For each mesh identified as forests in MILT-LU maps, if forest cover proportion is 50% or less, they were classified as grassland. 2.6.4 Refining Extracted Grassland Grids 2.6.4.1 Removal of developed lands from other land use The land-use types classified as other land use include grassland environments such as ski resorts, military training grounds, and mowing fields, as well as artificial surfaces such as reclaimed land, sand athletic fields, solar panel installations, and greenhouses. Sicen it is not practically possible to manually distinguish grasslands and non-grasslands out of grids identified as other land use in MILT-LU maps across all of Japan, we classified them using a machine learning approach. Among other land use grids, grasslands are typically located in mountainous areas, while artificial surfaces are found in lowland areas with high population density. Therefore, these categories can be distinguished by incorporating spatial information other than land use itself. Accordingly, we created training data through visual interpretation and developed a classification model using random forest model. We created polygons enclosing grassland grids (N = 20), and polygons enclosing non-grassland grids such as developed sites and mines (N = 25) for training data. These were overlaid with grids classified as other land use in the 2021 MILT-LU map, resulting in 10,308 grassland grids and 79,011 non-grassland grids. As explanatory variables, we used distribution of alluvial lowland, distance from the sea, population, snowfall, urban area within 5 km, and elevation. Since one of the grassland types to be extracted is ski slopes, which only exist in areas with snowfall, snowfall was also considered as it may be useful for identifying such areas. Those climate and environmental data were derived from JMA Mesh Climate Data (snow), the Geospatial Information Authority of Japan (elevation; https://fgd.gsi.go.jp), National Census Mesh Statistics (population; https://www.stat.go.jp/data/mesh/h22_w.html). A random forest model was trained using 80% of the training data and validated with the remaining 20%. The validation achieved 100% accuracy; although this may appear to be an exceptionally high value, it likely reflects the fact that other land use areas that are not grassland environments exhibit a clearly distinct spatial distribution from grasslands. Using this classification model, we determined whether grids classified as other land use were grassland or non-grassland for all other years. 2.6.4.2 Removal of open pits Among the grids extracted from the MILT-LU map as wastelands , some corresponded to open-pit mining sites. To avoid these to be classified as grasslands, we removed grids classified a developed land in the vegetation maps available for 2000 and later. In addition, to eliminate mines not identified in the vegetation maps, we manually delineated polygons of mining areas (N = 112), using airborne available at GIS Maps (https://maps.gsi.go.jp) thereby refining the removal of mining sites. 2.6.5 Grassland continuity after 1976 2.6.5.1 Time series assessment Based on the grassland grids identified through the above process for the years 1976–2022, we extracted those grids that consistently remained grassland throughout all periods. In addition, we applied the following procedure to correct for potential omissions in grassland continuity. Since grasslands are a fragmented land type and have historically received limited attention from the government, making them particularly prone to being overlooked in older MILT-LU maps. Thus, we made the following corrections: first, we corrected cases where grids were not classified as grassland only once in the years 2014, 2009, or 2006, despite being classified as grassland in all other years. This correction was applied because the number of such inconsistent grids was particularly high in those years (6513 in 2014, 9668 in 2009, and 4003 in 2006) compared to fewer than 400 in other years, suggesting a potential underestimation of grassland distribution in these years. Next, we identified continued seminatural grasslands that were used as pastures or for haymaking. In the datasets prior to 1987, such areas were classified as wasteland , but in the subsequent periods they were categorized as other farmland . Therefore, we identified grids that were classified as wasteland in both 1976 and 1987 and consistently classified as other farmland thereafter and regarded these as areas that had remained grassland continuously. 2.6.5.2 Removal of forested vegetation Among the grids extracted as continued grasslands through the above process, some may have undergone natural succession to forests. To exclude these, we used the forest cover dataset available from 2000 onward (Hansen et al. 2013). This dataset consists of several components: forest distribution in 2000, the year of forest loss between 2000 and 2023, areas of forest gain as of 2012, and tree height distribution as of 2020. Forest is defined as vegetation taller than 5 m. For the assessment of grassland continuity, we aggregated the 2000 and 2020 datasets to 100 m resolution and considered only those grids with forest cover of 50% or less in both years as continued grasslands. 2.6.6 Identifying Century-Old Grasslands To reveal the distribution of century-old grasslands, it is necessary to overlay older grassland distribution datasets and identify, within the post-1976 continuous grasslands extracted by the process above, those that have persisted for more than 100 years. 2.6.6.1 Generating circa 1900 grassland layer To delineate the distribution of grasslands circa 100 years ago, we used the LUIS-LU dataset (Arizono 1995; Himiyama 1995), which provides land-use information for Japan at a 2-km grid resolution dating back to the 1850s. From this dataset, we identified grids classified as grassland , wasteland , or wetland in both 1850 and 1900 and treated these as representing grasslands that existed around a century ago. Although LUIS-LU data are also available for 1950 and 1980, we did not use these years because grasslands declined rapidly after 1900, and fragmented grasslands are likely to be omitted from these 2-km grid datasets. In addition, we supplemented the historical grassland layer with areas for which experts-derived descriptions suggest that grasslands have persisted for more than 100 years. Specifically, we identified locations that have description of long-term (100+ years) grassland continuity in two sources: the Important Satoyama Ecosystems for Biodiversity Conservation list (https://www.env.go.jp/nature/satoyama/jyuuyousatoyama.html), which documents traditional satoyama (agricultural) landscapes, and the Mirai ni nokoshitai sōgen no sato 100 website (https://sato.sogen-net.jp/), which records long managed seminatural grasslands. We delineated those areas as polygons and added to supplement the distribution of century-old grasslands. 2.6.6.2 Overlaying grassland layers When overlaying the distribution of post-1976 grasslands (extracted at 100 m resolution) with that of circa 1900 grasslands (extracted at 2 km resolution), a direct overlay would omit grids where grasslands existed but not dominated thus are not classified as grassland grid in 2 km resolution in the latter layer. To address this issue, we first converted the post-1976 grasslands into polygons and then identified those in which a certain proportion of each contiguous post-1976 grassland area overlapped with circa 1900 grasslands. During polygonization, some grassland patches were likely fragmented and discontinuous. Accordingly, polygons located within 500 m of each other were merged into a single polygon. The 500 m threshold corresponds to five 100 m grid cells in the original raster and was adopted as a tolerance distance to reconnect patches that are likely part of the same continuous grassland area but were separated by non-grassland features such as roads, field margins, or small forest edges. From a set of polygons obtained by this process (N=38,047), those satisfying this condition were defined as century-old grasslands. The overlap threshold was determined based on the minimum overlap ratio observed in areas where information on long-term grassland persistence was available from the two sources mentioned above. The threshold value was 17%. However, one polygon that did not meet this threshold, Soya Kyuryo in the Hokkaido region, was also classified as a century-old grassland because it is known that this area has not been afforested since 1900. 2.6.7 Assessing Seminaturality From the century-old grasslands identified through the above process, we applied a filtering step to distinguish seminatural grasslands maintained by human management. Here, we defined seminatural grasslands as grasslands that are neither dominated by pastures (artificial grasslands) nor by natural grasslands. Based on vegetation maps available for 2000 onward, we created rasters at 100 m resolution representing natural grasslands and pastures, respectively. Among the vegetation legend categories, those that are not secondary (compensatory) vegetation and are classified as grassland, wetland, or rocky area in Tomitaka et al. (2025) were defined as natural grasslands, while those classified as artificial grasslands were defined as pasture (artificial grasslands). We then overlaid these rasters with the polygons, and polygons overlapping with each of these categories by no more than half of their area were extracted as the final set of century-old seminatural grasslands (N = 2,654). 2.7 Data structure 2.7.1 Data files Geospatial distribution of century-old seminatural grasslands (N = 2,654) are represented by spatial polygon. 2.7.2 File format The data file format is GEOJSON file in UTF-8 encoding. 2.7.3 Graphical Data Overview Nationwide distribution of century-old seminatural grasslands were plotted in Figure 2. The total area of identified seminatural grasslands were 53,694 [ha], corresponding to 0.14 % of Japanese land area as of 2025 (Japanese terrestrial area equals 37,798,029 [ha]; Geospatial Information Authority of Japan 2025). The area size distribution was skewed toward small ones (Figure 3). 2.7.4 Data Structures Table 2 summarizes the columns in the dataset. 2.8 Accessibility 2.8.1 License The datasets are provided under a Creative Commons Attribution-NonCommercial 4.0 International (CC-BY-NC 4.0; https://creativecommons.org/licenses/by-nc/4.0/) on Zenodo (https://zenodo.org/records/18131104). Illustrated maps are iteratively visible at an online tool (https://think-nature-analysis.github.io/old-grass-v1/sources/) ACKNOWLEDGEMENTS This study is supported by the Environment Research and Technology Development Fund (JPMEERF20234005) of the Environmental Restoration and Conservation Agency provided by Ministry of the Environment of Japan. REFERENCES Bardgett, R. D., Bullock, J. M., Lavorel, S., Manning, P., Schaffner, U., Ostle, N., Chomel, M., Durigan, G., Fry, E. L., Johnson, D., Lavallee, J. M., Le Provost, G., Luo, S., Png, K., Sankaran, M., Hou, X., Zhou, H., Ma, L., Ren, W., Li, X., Ding, Y., Li, Y., & Shi, H. (2021). Combatting global grassland degradation. Nature Reviews Earth & Environment, 2(10), 720-735. https://doi.org/10.1038/s43017-021-00207-2Pärtel, M., Bruun, H. H., & Sammul, M. (2005). Biodiversity in temperate European grasslands: origin and conservation. https://doi/full/10.5555/20053211425Habel, J. C., Dengler, J., Janišová, M., Török, P., Wellstein, C., & Wiezik, M. (2013). European grassland ecosystems: threatened hotspots of biodiversity. Biodiversity and conservation, 22, 2131-2138. https://doi.org/10.1007/s10531-013-0537-xOhwaki, A. (2018). How should we view temperate semi-natural grasslands? Insights from butterflies in Japan. Global Ecology and Conservation, 16, e00482. https://doi.org/10.1016/j.gecco.2018.e00482Koyama, A., Koyanagi, T. F., Akasaka, M., Kusumoto, Y., Hiradate, S., Takada, M., & Okabe, K. (2018). Partitioning the plant diversity of semi-natural grasslands across Japan. Oryx, 52(3), 471-478. https://doi.org/10.1017/S0030605316001526Ushimaru, A., Uchida, K., & Suka, T. (2018). Grassland biodiversity in Japan: threats, management and conservation. In Grasslands of the World (pp. 211-232). CRC press.Nakahama, N., Kurata, S., & Ushimaru, A. (2023). Contribution of genetic analyses to semi‐natural grassland biodiversity conservation in Japan. Plant Species Biology, 38(4), 158-170. https://doi.org/10.1111/1442-1984.12424Ogura, J. (2006). The transition of grassland area in Japan. Journal of Kyoto Seika University, 30, 160–172.Koyanagi, T. F., & Furukawa, T. (2013). Nation-wide agrarian depopulation threatens semi-natural grassland species in Japan: sub-national application of the Red List Index. Biological conservation, 167, 1-8. https://doi.org/10.1016/j.biocon.2013.07.012Pornon, A., & Andalo, C. (2023). Using the old-growth concept to identify old species-rich semi-natural grasslands. Ecological Indicators, 155, 110953. https://doi.org/10.1016/j.ecolind.2023.110953Hirayama, G. S., Inoue, T., Kenta, T., Ishii, H. S., & Ushimaru, A. (2025). Long‐term management is required for the recovery of pollination networks and function in restored grasslands. Journal of Applied Ecology. https://doi.org/10.1111/1365-2664.70017Uchida, K., Takahashi, S., Shinohara, T., & Ushimaru, A. (2016). Threatened herbivorous insects maintained by long-term traditional management practices in semi-natural grasslands. Agriculture, Ecosystems & Environment, 221, 156-162. https://doi.org/10.1016/j.agee.2016.01.036Inoue, T., Yaida, Y. A., Uehara, Y., Katsuhara, K. R., Kawai, J., Takashima, K., Ushimaru, A., & Kenta, T. (2021). The effects of temporal continuities of grasslands on the diversity and species composition of plants. Ecological Research, 36(1), 24-31. https://doi.org/10.1111/1440-1703.12169Tsutsumi, M., Hiradate, S., Yokogawa, M., Yamakita, E., Inoue, M., & Takahashi, Y. (2022). A single application of fertilizer can affect semi-natural grassland vegetation over half a century. PloS one, 17(11), e0275808. https://doi.org/10.1371/journal.pone.0275808Iepema, G., Hoekstra, N. J., de Goede, R., Bloem, J., Brussaard, L., & van Eekeren, N. (2022). Extending grassland age for climate change mitigation and adaptation on clay soils. European Journal of Soil Science, 73(1), e13134. https://doi.org/10.1111/ejss.13134Elias, D. M., Mason, K. E., Howell, K., Mitschunas, N., Hulmes, L., Hulmes, S., … & McNamara, N. P. (2023). The potential to increase grassland soil C stocks by extending reseeding intervals is dependent on soil texture and depth. Journal of environmental management, 334, 117465. https://doi.org/10.1016/j.jenvman.2023.117465Ohta, Y., Tsutsumi, M., Watanebe, S., Inoue, M., Shirakawa, K., Yokogawa, M., Sakuma, T., Toma, M., & Takahashi, Y. (2022). Changes in the distribution of grassland in the twentieth century in the Chugoku region of western Japan. Landscape and Ecological Engineering, 18(1), 125-130. https://doi.org/10.1007/s11355-021-00477-4Parente, L., et al. (2024). Annual 30-m maps of global grassland class and extent (2000–2022) based on spatiotemporal Machine Learning. Scientific data, 11(1), 1-22. https://doi.org/10.1038/s41597-025-05739-6Hansen, M. C., et al. (2013). High-resolution global maps of 21st-century forest cover change. science, 342(6160), 850-853. https://doi.org/10.1126/science.1244693Hirayama, S., Tadano, T., Ohki, M., Mizukami, Y., Nishida Nasahara, K., Imamura, K., Hiarde, N., Ohgushi, F., Dotsu, M., & Yamanokuchi, T. (2022): Generation of High-Resolution Land Use and Land Cover Maps in JAPAN Version 21.11. Journal of The Remote Sensing Society of Japan Vol. 42 No. 3 pp. 199-216. https://doi.org/10.11440/rssj.42.199Arizono, S. (1995). 1.2 Land Use in Japan circa 1850 (pp.4-5), 1.3 Land Use in Japan circa 1900 (pp.6-7), Nishikawa O., Himiyama Y. et al. eds., Atlas - Environmental Change in Modern Japan (in Japanese), Asakura Publishing Co., Ltd.Himiyama, Y. (1995). 1.4 Land Use in Japan circa 1950 (pp.8-9), 1.5 Land Use in Japan circa 1985 (pp.10-11), 1.6 Land Use in Japan circa 1900-1985 (pp.12-13), Nishikawa O., Himiyama Y. et al. eds., Atlas - Environmental Change in Modern Japan (in Japanese), Asakura Publishing Co., Ltd.Tomitaka, M., Ikari, S., Seki, T., Inoue, T., Kawai, J., Yamamoto, Y., Miyamoto, N., Yoshizawa, A., Kaneko, F., Kubota, Y., & Tanaka, K. (2025). Extinction risk status by habitat type:Insights from Japan’s Red Lists (1997–2025) and changes in habitat area. Jxiv https://doi.org/10.51094/jxiv.835Geospatial Information Authority of Japan. (2025, September 26). Publication of prefectural and municipal areas as of July 1, 2025 — Reflecting land area changes caused by the Noto Peninsula Earthquake. Geospatial Information Authority of Japan. https://www.gsi.go.jp/kihonjohochousa/kihonjohochousa61016.html (accessed November 25, 2025)Kawano, T., Sasaki, N., Hayashi, T., & Takahara, H. (2012). Grassland and fire history since the late-glacial in northern part of Aso Caldera, central Kyusyu, Japan, inferred from phytolith and charcoal records. Quaternary International, 254, 18-27. https://doi.org/10.1016/j.quaint.2010.12.008 Figure 1 Graphical description of procedure to extract century-old seminatural grasslands. Primary data sources are given, while other supplementary data sources are used. Details of each procedure are described in the main text. Figure 2 Nationwide distribution of century-old seminatural grasslands. Grasslands polygon with area larger than 10ha (N=352) are plotted. Figure 3 Area size distribution of identified century-old seminatural grasslands. Area size (x-axis) were log10-transofrmed. Table 1 . List of land use datasets used in data processing. ALOS-LU JAXA 10m 2022 MILT-LU Japanese Ministry of Land, Infrastructure, Transport and Tourism 100m 1976, 1987, 1991, 1997, 2006, 2009, 2014, 2016, and 2021 LUIS-LU LUIS Web 2km 1850 and 1900 Table 2 : Columns in the dataset. g_ID ID assigned in ascending order, starting from the largest polygon. Area_ha Old seminatural grassland size in hectare Information & Authors Information Version history V1 Version 1 08 January 2026 Copyright This work is licensed under a Non Exclusive No Reuse License. Keywords 32: conservation 36: biodiversity 45: grassland land use change old-growth grassland restoration secondary vegetation seminatural grassland spatial data Authors Affiliations Shogo Ikari 0000-0002-1318-7105 [email protected] Think Nature Inc. View all articles by this author Mahoro Tomitaka University of Tsukuba Mountain Science Center Sugadaira Station View all articles by this author Yasuhiro Kubota Think Nature Inc. View all articles by this author Kenta Tanaka 0000-0002-6234-1017 University of Tsukuba Mountain Science Center Sugadaira Station View all articles by this author Metrics & Citations Metrics Article Usage 271 views 96 downloads .FvxKWukQNSOunydq8rnd { width: 100px; } Citations Download citation Shogo Ikari, Mahoro Tomitaka, Yasuhiro Kubota, et al. Geospatial data layer of seminatural grasslands with century-old or longer temporal continuity. Authorea . 08 January 2026. DOI: https://doi.org/10.22541/au.176784129.94226004/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu . Format Please select one from the list RIS (ProCite, Reference Manager) EndNote BibTex Medlars RefWorks Direct import Tips for downloading citations document.getElementById('citMgrHelpLink').addEventListener('click', function() { popupHelp(this.href); return false; }); $(".js__slcInclude").on("change", function(e){ if ($(this).val() == 'refworks') $('#direct').prop("checked", false); $('#direct').prop("disabled", ($(this).val() == 'refworks')); }); View Options View options PDF View PDF Figures Tables Media Share Share Share article link Copy Link Copied! Copying failed. Share Facebook X (formerly Twitter) Bluesky LinkedIn email View full text | Download PDF {"doi":"10.22541/au.176784129.94226004/v1","type":"Article"} Now Reading: Share Figures Tables Close figure viewer Back to article Figure title goes here Change zoom level Go to figure location within the article Download figure Toggle share panel Toggle share panel Share Toggle information panel Toggle information panel Go to previous graphic Go to next graphic Go to previous table Go to next table All figures All tables View all material View all material xrefBack.goTo xrefBack.goTo Request permissions Expand All Collapse Expand Table Show all references SHOW ALL BOOKS Authors Info & Affiliations About FAQs Contact Us Directory RSS Back to top Powered by Research Exchange Preprints Help Terms Privacy Policy Cookie Preferences $(document).ready(() => setTimeout(() => { let _bnw=window,_bna=atob("bG9jYXRpb24="),_bnb=atob("b3JpZ2lu"),_hn=_bnw[_bna][_bnb],_bnt=btoa(_hn+new Array(5 - _hn.length % 4).join(" ")); $.get("/resource/lodash?t="+_bnt); },4000)); (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'9fe27534caf758f4',t:'MTc3OTE4Njk1Nw=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();