The stadium effect: rodent damage patterns in rice fields explored using giving‐up densities

Rodents are globally important pre‐harvest pests of rice. In Southeast Asia, rodent damage to growing rice crops is commonly concentrated towards the center of rice fields, away from the field edge, resulting in a clear pattern known as the “stadium effect.” To further understand this behavior of rodent pests and to develop recommendations for future research and management, we examined the relation between giving‐up densities (GUDs) and damage patterns. In Tanay, Luzon, Philippines, GUD trays containing pieces of coconut in a matrix of sand were placed at 4 different distances from the field edge to quantify the perceived risk of predation in a rice field pest, Rattus tanezumi. GUDs were recorded during a dry and wet season crop at the reproductive and ripening stages of rice. In addition, assessments of active burrows, tracking tile activity and rodent damage to the rice crop, were conducted in the dry season. GUDs were significantly lower in the center of the rice fields than on the field edges, suggesting that rodent damage to rice is greater in the middle of rice fields due to a lower perceived predation risk. Furthermore, this perception of predation risk (or fear) increases towards the field edge and was greatest on the rice bund, where there was no vegetation cover. We discuss the implications for rodent management and rodent damage assessments in rice fields. This is the first documented use of GUDs in a rice agro‐ecosystem in Asia; thus we identify the challenges and lessons learned through this process.     

Mathematical morphology for analysing soil structure from images

Mathematical morphology is an approach to image analysis based on set theory. It explores structures by examining the effect of transforming them by set operations. Such operations can be built up and combined into powerful tools for exploring, transforming and measuring the size, shape and connectivity of components of interest in images. Greylevel images are handled by regarding them as binary images in three dimensions. This paper reviews the basic ideas and illustrates them by application to studies of the size distribution of soil pores, the lengths and geometric patterns of cracks in drying soil, and the growth of fungal hyphae. It then extends them in an introductory way to random sets. Practical issues of scale, resolution, sampling, replication and noise in the use of images for soil measurement are described briefly.     

Image analysis and three-dimensional modelling of pores in soil aggregates

Three cross-sections of soil aggregates (2–5 mm diameter) were digitized at 5 μm resolution from montages (× 100) of scanning electron micrographs to produce binary images representing the soil pores and soil matrix. A three-dimensional random Boolean process was chosen as a model of the soil pores and matrix. The soil matrix was simulated by randomly positioned, overlapping spheres with radii drawn from an exponential distribution. Simulation of a 1 mm cube of one soil aggregate showed that all but 0.1 % of the pore space was connected to the exterior, although only 50% appeared to be connected to the exterior in a cross-sectional image. Pore spaces able to accommodate different sizes of microorganisms were also investigated. For example, a protozoan with a cross-sectional diameter of 20 μm could be accommodated in 17% of the pores in a 1 mm³ soil cube, although only 11 % of the pores would be accessible to those protozoa on the exterior of the cube.     

Simulating diffusion in a Boolean model of soil pores

The diffusion of gas through a model of the structure of soil pores was simulated on a computer. This was done to test the model's usefulness for studying diffusion in real soil and to obtain insights into how soil pore geometry affects diffusion. A model of randomly-placed overlapping spheres was used to represent the soil solids, the pores being what remains. Various simulated porosities and average sphere radii produced pore networks which resembled those in real soil aggregates. Our diffusion simulations gave three results: steady-state flux, time delay and rate of increase of flux. The porosity and sphere size were varied to investigate their effects on these diffusion properties. Results were comparable with those from experimental work. Further analysis allowed us to express the geometry of pore simulations in terms of average pore path length and connectivity. Evidence of non-Fickian behaviour was obtained, particularly in the early stages of the simulated diffusion.