The factors affecting the distribution of wet-paddy, ill-drained during the non-irrigation season, which are important spawning grounds for Rana frogs were estimated. The field study was conducted in Toyota City, Aichi Prefecture, Japan in 2019. We traveled along roads and recorded the presence or absence of puddle in the rice fields, cultivation conditions, and farmland consolidation. DEM (10 m), vegetation maps (1:25,000), and GIS data of the road were used to extract geographic factors. The response variable was the presence or absence of puddles in each paddy block, the explanatory variables were categorical variables such as yatsuda (valley paddy field) or not, farmland consolidated or not, forest edge or not, cultivated or abandoned, and quantitative variables such as elevation, slope, TWI, percentages of forest and and paddy area, and road density within concentric circles. The concentric circles of seven different radii ranging from 50 to 600 meters were used. Hierarchical Bayesian models with spatial random effects were fitted using the INLA and SPDE approaches to account for spatial autocorrelation of residuals. In the model with the smallest DIC, the percentage of forest area with a radius of 200 m, yatsuda type, no farmland consolidation, and forest edge increased the wet-paddy ratio, and intermediate elevation decreased the ratio. A spatial effect with a range of 10.5 km was observed, with higher rates of wet-paddy field in the northern mountains and southern foothills, and lower rates in the middle region.
View full abstract