Study site
The study was conducted in the Tone river basin, central Japan (Fig. 1). The Tone river is Japan’s second-longest river, running through the entire Kanto plain in central Japan. The Tone river basin is covered mainly by rice paddies and also contains arable fields other than rice, seminatural grasslands, coppice forests, farm villages, and urban areas29. The Tone river basin is located in the Kanto plain which is the largest plain field in Japan (approximately 170,000 km2), including large floodplains. Thus, this area have a variety of both terrain conditions and agricultural modernization works.
Location of the Tone river basin and monitoring sites.
Plants community data
The Institute for Agro-Environmental Sciences, NARO, Japan conducted the program for monitoring biodiversity, including birds29 and plants34, in each of the thirty-two 1-km2 grids in the Tone river basin in 2002. In this program, the Tone river basin was initially divided into one hundred 1-km square grids (hereafter, 1-km grid), and each square was classified into one of four major land use types in the region: (1) midstream paddy; (2) downstream lowland paddy; (3) plateau and valley-bottom paddy; and (4) urban fringe29. Then, eight grids were selected randomly as study sites from each land use type, making a total of 32 grids (Fig. 1). The grids were more than 5 km apart (Fig. 1), so they were spatially independent of each other. In this study we used the plant monitoring records from the program. In the plant monitoring program there were three terms—2002, 2007, and 2012—of vegetation survey based on the Braun–Blanquet approach in each 1 km grid 35. In each survey, approximately 20 quadrats measuring 1 m2 were placed randomly in each 1-km grid in each survey term and the coverage ratios of all plant species in four hierarchies—(1) tall tree, (2) semi-tall tree, (3) shrub, and (4) grasses—were recorded. In this study, we used only the grasses class without abundance and the presence or absence of species records in the grasses class. We pooled all the species records within each 1-km grid for analysis. All plant monitoring data are available as Open Data (CC BY 4.0) at github space own by Dr. N. Iwasaki who was the member of this monitoring program (https://github.com/wata909/RuLIS_monitoring, accessed at 25, May 2020).
Dividing wetland plants and non-wetland plants
To test our hypothesis, we needed to divide the plants that typically grow in wetlands (hereafter, wetland plants) and those that typically grown in non-wetlands (hereafter, non-wetland plants) to evaluate the habitat quality of paddy fields as wetland. To this end, we used a published checklist of wetland plants in Japan (Shutoh et al. 2019; https://wetlands.info/tools/plantsdb/wetlandplants-checklist/, accessed at 25, May 2020). This checklist defined 8,358 Japanese vascular plants as wetland and aquatic plants according to their habitat requirements and the “wetland” definition of the Ramsar Convention (Ramsar Convention Secretariat 2016, https://www.ramsar.org/sites/default/files/documents/library/manual6-2013-e.pdf, accessed at 25, May 2020). We used this checklist to identify the wetland plant species in the monitoring records.
Land use, terrain condition, and human activity
A digitized land use map for paddy fields in 2009 that relatively matched the plant monitoring terms (2002, 2007, and 2012) was prepared from the National Land Numerical Information (National Land Information Division, MLIT of Japan: https://nlftp.mlit.go.jp/ksj-e/index.html, accessed at 25, May 2020). These map data were developed using both topographic maps and satellite imaging data, with the land use labeled on the basis of nationwide land use classifications, including paddy fields, at approximately 100-m grid resolution (National Land Information Division, MLIT of Japan: https://nlftp.mlit.go.jp/ksj-e/index.html, accessed at 25, May 2020).
A FAV, which was ascertained by accumulating the weights of all cells that flowed into each downslope cell, was used to define the concave areas (ESRI, https://pro.arcgis.com/en/pro-app/tool-reference/spatial-analyst/how-flow-accumulation-works.htm, accessed at 25, May 2020); lower elevations and valley areas had a higher FAV because they could potentially store more water, whereas higher ridge areas had low FAVs (Fig. 2). We used FAV to define the wetland potential, as this value could reflect the water accumulation from upper areas to lower areas, which strongly relates to the natural process of wetland formation36. We considered that terrain variable could reflect the geographical conditions of paddy field namely potentially wetland habitat for their intact ecosystem. We considered high FAV areas to have high potential of wetland habitat for their intact ecosystem. We calculated FAV value on a whole for mainland Japan; therefore, that range could cover the entire basin which overlapped with our target areas. The FAV was calculated using ArcGIS 10.5 with Spatial analyst (ESRI, Redlands, CA, USA) using a 50-m digital elevation model from the Japanese Map Centre (https://www.jmc.or.jp/, accessed at 25, May 2020). The FAV and paddy field maps were overlaid, and the total FAVs for paddy fields in each 1-km grid were calculated to determine the potentiality of the paddy fields in the 1-km grid being wetland. If a paddy field had an extremely high FAV within the basin which included the paddy field, that paddy field could have been a wetland because that area could store a large amount of water naturally.
Conceptual image of the flow accumulation value to indicate the potential of wetland.
The proportional area of field consolidation as current human activity was calculated for each grid square using digital polygon data on the shape of farmland, as derived from aerial imagery collected in 2001 by the Ministry of Agriculture, Forestry, and Fisheries (MAFF), Japan. We obtained data on land leveling in agricultural areas from MAFF (https://www.maff.go.jp/j/tokei/porigon/, accessed at 25, May 2020) and used these data as an index of consolidated farmland because land leveling is one of the important components of agricultural consolidation in Japan4,23. Generally, agricultural consolidation in Japan involves land leveling, which integrates small, patchy farmland areas. Each polygon was assigned a status of “leveled” or “not leveled” according to its current status. Using ArcGIS, we calculated the ratio of consolidation for paddy fields in each 1-km grid that had survey sites.
Statistical analysis
We performed the two types of analysis used in this study with the statistical package R version 3.5.2 (R development core Team, https://www.r-project.org/, accessed at 17, Feb. 2020). First, we tested the species number in each 1-km grid using GLM with Poisson distributions (log link) and a Wald test37. The response variables were total species number, number of wetland plants, and number of non-wetland plants in each 1-km grid in each survey term. Explanatory variables were the log-transformed FAV values for the paddy fields and consolidation ratio of the paddy field within the 1-km grid. The aim of this analysis was to assess the effects on species diversity of both the original environmental condition of and current human activities in the paddy fields. Prior to the GLM analysis, all explanatory variables were tested for multicollinearity by calculating the variance inflation factors (VIFs)38; no significant multicollinearity was found (VIF < 10 for all variables).
Second, we tested the number of both nested and turnover species from previous survey terms using GLMs with Poisson distributions (log link) and a Wald test37. We set three multi-temporal combinations: 2002–2007, 2002–2012, and 2007–2012. We used species number which observed both term as response variables, and used species number of latter term as offset term. Thus, we tested the ratio of species that were nested from the previous community, namely, non-turnover rates. Our analysis included all plants, both wetland plants and non-wetland plants, in each 1-km grid. Explanatory variables were the same as the species number analysis. We predicted that grids that were potentially wetland grids, namely, high FAV grids, would have a large ratio of nested wetland species. Furthermore, we predicted that heavy consolidated grids would have a large ratio of species turnover that could be negatively influenced.
Source: Ecology - nature.com
