Soybean production in Japan shows divergent regional yield trajectories under recent climate change, but the extent to which in-season meteorology explains these trajectories has not been systematically quantified across regions. Using municipality-level soybean yield statistics from the 15 main producing prefectures for 1993-2024, we evaluated how much in-season meteorology could explain yield anomalies by combining stage-aggregated weather variables with linear and nonlinear statistical models under strict random year-based holdout validation. The objective was not operational yield forecasting but rather a quantitative assessment of where in-season meteorology was sufficient or insufficient to account for yield variability. Nationwide, late-vegetative temperature was positively associated with yield anomalies, whereas late-vegetative precipitation was negatively associated with yield anomalies, and a nonlinear low-temperature response emerged below approximately 20°C. Region-specific analyses revealed substantial heterogeneity in the dominant predictors and response shapes, indicating that identical meteorological variables corresponded to different yield risks depending on the regional context. Out-of-sample explainability was generally limited; however, improvements over a training-mean baseline demonstrated that in-season meteorology contained non-zero explanatory information. Decomposition of yield anomalies into meteorologically explained and unexplained components revealed that the national yield decline was reflected almost entirely in the unexplained component. This result does not imply that meteorological influences were negligible but rather indicates that the decline could not be consistently explained by the stage-aggregated meteorological variables used in the present framework. Among the regions with significant yield increases, the increase in Obihiro, Hokkaido, was largely meteorologically explained, whereas that in Miyagi was associated with a comparatively larger unexplained contribution. These results delineate the limits of weather-only explanations for Japanese soybean yields and suggest that future improvements will require more detailed representations of meteorological conditions and greater consideration of non-meteorological influences on soybean production.

　It is important to determine tree mortality rates to develop a more accurate understanding of forest carbon dynamics. This study compared the tree mortality rate estimated from ground surveys with that obtained by repeated lidar observations to evaluate the applicability and limitations of lidar estimation and to identify the causes of canopy height depletion in a cool-temperate conifer-broadleaf mixed forest in the northernmost part in Japan. The results indicated that more than half of the trees showing canopy height depletion were Abies sachalinensis, and that a broken trunk accounted for more than half of the observed height decreases. Although the average mortality rates estimated by both estimation methods over the 25-ha study plot were similar, lidar estimation tended to over- and underestimate the 1-ha mean mortality rates in the lower and higher parts, respectively, of the range of the mortality rate relative to that determined by ground surveys (r² = 0.63). Repeated lidar-based mortality estimation showed promise as a method of estimation, although a reduction in spatial resolution would be necessary to obtain more reliable estimates.

　This study proposes a method for measuring the dimensions of edamame pods using three-dimensional (3D) models acquired with a smartphone. Mesh models of edamame pods were obtained using an iPhone and converted to point clouds for analysis. The two seeds within each pod were automatically separated based on plane detection, height map generation, peak detection, and watershed segmentation. When this peak-based segmentation was unstable, k-means clustering was introduced as a fallback procedure, enabling stable separation in all 80 samples. After segmentation, geometric features, including length, width, height, and volume, were calculated for each seed. Comparison with manual measurements showed that the proposed method estimated the width and length with Root Mean Square Error (RMSE) of 0.642 and 1.320 mm, respectively. In addition, the 3D shape index, defined as width × length × height, exhibited a strong positive correlation with the convex-hull-based estimated volume (r = 0.970, p < 0.001), and a moderate negative correlation with moisture content (r = -0.416, p < 0.01). The latter correlation may suggest an association between 3D shape information and maturity-related traits, although moisture content was only partially explained by the shape index. These results demonstrate the potential of smartphone-based 3D measurements as a simple and nondestructive approach for the quantitative evaluation of edamame pod morphology.

　Forest soils are an important sink for atmospheric methane (CH₄), yet the factors controlling spatial variability of CH₄ uptake among forest types remain insufficiently understood. In this study, CH₄ fluxes were measured along an elevational transect within a geographically limited mountainous region of Takayama, central Japan. Eleven sites representing deciduous broadleaf forests and evergreen coniferous stands, including cedar and cypress plantations, were investigated using manual chamber measurements during six campaigns over two growing seasons. Despite a mean temperature range of approximately 10－22°C during measurements, CH₄ fluxes did not exhibit a clear seasonal pattern. No significant correlation in CH₄ uptake was obtained along the elevational gradient itself. In contrast, deciduous broadleaf forests showed significantly greater CH₄ uptake than adjacent evergreen conifer forests. Soil pH, carbon and nitrogen contents, porosity, and soil moisture were similar between forest types. However, C:N ratios tended to be higher in coniferous forests, whereas soil respiration tended to be higher in deciduous forests. Estimated annual CH₄ uptake based on growing-season measurements was within the range previously reported for Japanese forest soils. Within this landscape, forest type appeared to exert a stronger influence on CH₄ uptake than elevation.