Data from: Vehicle-mounted cameras reveal negative impact of the Fukushima Daiichi nuclear power plant accident on large-bodied bird abundance via paddy field abandonment
Data files
Nov 10, 2025 version files 898.28 MB
-
d0.csv
485.53 MB
-
dmatrix3.csv
2.03 KB
-
farm-polygon.csv
412.75 MB
-
README.md
3.11 KB
Abstract
Populations of birds utilizing farmlands are declining due to farmland abandonment and agricultural intensification. Effective conservation strategies require appropriate monitoring approaches, including efficient and scalable survey methods. In this study, we developed a large-scale monitoring method for herons and egrets (family Ardeidae, hereafter referred to as herons) using vehicle-mounted video cameras and distance sampling models that incorporate location uncertainty. The survey was conducted in and around the evacuation zone of the Fukushima Daiichi nuclear power plant accident. A total of 7031 km of roadside video footage was recorded, covering 24.41 km² of farmland. Herons were observed only outside the evacuation zone and were entirely absent within it. Predicted abundance differed greatly between areas inside (0.0279 ± 0.0307/km2) and outside (4.57 ± 5.36/km2) the evacuation zone. Incorporating location uncertainty into the distance sampling model had little effect on the estimates (4.57 ± 5.36 vs. 4.51 ± 5.29/km2 with and without integrating location uncertainty, respectively). This finding suggests that our video-based roadside survey method is robust to location uncertainty in structured landscapes such as Japanese paddy fields. The accuracy may be attributed to the study system, where levees and roads divide paddy fields, limiting potential error in observer-target distances. Our method can be applied to other open habitats, including natural grasslands and wetlands, especially in areas lacking artificial markers, by incorporating measurement uncertainty into the model. This combination of roadside surveys with vehicle-mounted cameras and distance sampling provides a practical and transferable approach for monitoring large-bodied species in open landscapes, enhancing both the efficiency and spatial coverage of biodiversity assessments.
Dataset DOI: 10.5061/dryad.jq2bvq8pm
Description of the data and file structure
The dataset was collected by the authors to estimate the abundance of herons and egrets in paddy fields using vehicle-mounted video cameras and distance sampling models. The data were obtained from roadside surveys and UAV-based validation, and are used to model bird abundance in relation to farmland abandonment following the Fukushima Daiichi nuclear power plant accident.The dataset consists of the following three .csv files:
Files and variables
1. Grid-level farmland data (d0.csv)
This file contains grid-level data used for modeling heron and egret abundance based on roadside video surveys. Each row represents a grid cell within a farmland polygon.
Variables
poly_id.x, DATE: Polygon and survey identifiersx, y, X_3100, Y_3100: Coordinates (decimal and UTM Zone 54N, EPSG:3100)HubDist: Distance from the camera to the grid cell (meters)water: Area of water bodies within the 1 km mesh (Ogawa et al. 2013) that contains the grid cell (standardized)evac: Evacuation zone status; 1-3 indicates the evacuation zone, 0 indicates around the zone. The evacuation zone is divided into three categories according to radiation levels: Preparing for the Lifting of the Evacuation Order Area (1), Restricted-Residential Area (2), and Difficult-to-Return Area (3)sagi: Presence/absence of heronsarea: Area of the grid cell (km²)
Note: Coordinates are based on the centroids of grid cells derived from farmland polygons. The geographical and projected coordinate systems are JGD2000 (EPSG:4612) and UTM Zone 54N (EPSG:3100), respectively.
2. Location Error Validation Data (dmatrix3.csv)
This file contains data used to quantify location uncertainty in heron observations by comparing vehicle-mounted camera detections with UAV-based ground truth.
Variables:
Distance: Euclidean distance (meters) between the observed location (vehicle camera) and the true location (UAV)cardis: Distance from the camera to the heron (meters)
3. Prediction Mesh Data (farm-polygon.csv)
This file contains mesh-level data for extrapolating heron abundance across the broader study area.
Variables:
pointid, grid_code, Mesh3id: Grid, and mesh identifierX, Y: Coordinates (decimal)water_std: Standardized area of water bodies contained within each 1 km mesh (Ogawa et al. 2013)evac2, evac3: Evacuation zone status (actual and counterfactual); 2 indicates the evacuation zone, 1 indicates around the zone
Access information
Area of water surfaces was derived from:
Ogawa, M., Takenaka, A., Kadoya, T., Ishihama, F., Yamano, H., and Akasaka, M. (2013) Land-use classification and mapping at a whole scale of Japan based on a national vegetation map. Japanese Journal of Conservation Ecology 18:69-76.