Search

Modern soil erosion prediction technology, such as the Water Erosion Prediction Project, requires detailed information about plant status in the interrill and rill erosion processes. The total amount of rain that reaches the soil and is available for erosion under vegetated surfaces is a function of rain throughfall and stemflow. The objective of our research was to determine the effects, including the production of stemflow of canopies, of corn (Zea maysL.) and sorghum (Sorghum bicolor[L.] Merr.) on interrill soil erosion. A series of rainfall simulations of similar intensity (6.35 cm h‐1) was performed in 1989 and 1990 on experimental plots of Cargill SX 383 corn and Garst 5503 sorghum grown on Cecil soil (clayey, kaolinitic, thermic Typic Kanhapludult). Runoff and sediment were collected at 5‐min intervals over a 1‐h period for two sets of treatments each under corn and sorghum: canopy and canopy minus stemflow. Stemflow, plant‐height, and leafarea data were collected from a total of 18 corn and 16 sorghum plants and used to estimate stemflow. Rain throughfall was estimated as the difference between total rainfall and calculated stemflow and was compared with observed runoff. Both corn and sorghum were observed to have high stemflow, the amount being highly correlated to total leaf area. Runoff under sorghum indicated that all the stemflow was potentially available for infiltration. Under corn, however, about one‐third of the stemflow may have contributed to runoff. The average sediment concentration was not sensitive to canopy type. Erosion due to stemflow appeared negligible compared with that caused by throughfall.

The relationship between plant roots and interrill soil erosion is important in dynamic soil erosion predictions. Rainstorm simulations of similar intensity (63.5 mm h−1) were conducted in the summer of 1989 on 1 by 1 m field plots of Cecil sandy loam (clayey, kaolinitic, thermic Typic Kanhapludult) to study the effect of different root length densities, Lv(cm cm−3), of corn (Zea maysL.), during vegetative, preanthesis, and anthesis plant developmental stages on interrill soil erosion. Sediment loss and runoff data were assigned to treatments, Lva, on the basis of Lvrange to reduce rooting variability associated with plant developmental stage. To avoid confounding by canopy cover, corn plants were cut at the stem base and removed for the first four sets of simulations. A reference set of simulations was performed on a fallow plot containing no roots, Lva0. Runoff and detached sediment were collected in buckets from each plot during successive 5‐min intervals over a 1‐h period. Thirty cores, 5 cm in diam. and 5 cm deep, were taken from each 1‐m2plot and roots were washed from the cores and measured. Means for runoff and detached sediment were generally not significantly different for the high Lvor Lva, and Lva0. When Lvawas <1.5 cm cm−3, runoff and detached sediment were significantly lower during the first 30 min of simulated rainfall. High densities of live corn roots did not reduce interrill soil erosion from a moldboard‐plowed Cecil sandy loam.

Previous experience in southern and western Australia has shown that tree clearing alters the water balance in a landscape, leading to increased deep drainage and consequent rises in phreatic water tables. Assessment of the risk of regional salinization involves integration of hydrology, hydrogeology, soil and land management issues. An example of the use of soil survey data, integrated with water resources and digital elevation data in a GIS, to estimate the risk of salinization after tree clearing is illustrated for the upper Burdekin river basin in the wet/dry tropics of North Queensland. Soil map unit descriptions include information on soil‐landform relations, parent material, soil thickness, depth to bedrock and dominant and subdominant soil types. Such information can be used to qualitatively estimate permeability and drainage classes (high, medium, or low) and to determine likely recharge and discharge areas in the landscape. Analysis of the spatial association between surveyed electrical conductivity measurements and estimated recharge and discharge areas was used to cross‐check this qualitative classification. The distributions of saline soils and shallow water tables are used to assess the salinity hazard of the region. Where % total soluble salts (TSS) are >0.25% and water table >6 m deep, a salinity hazard is present. This hazard can translate into a salinity problem if the hydrologic balance is altered so that the salt is remobilized and redeposited in the landscape. An estimate of the likelihood and amount of increased recharge to groundwater after tree clearing was obtained using water balance models with soil physical parameters measured at key sites. Where physical data were not available, surrogate physical properties were estimated from the soil survey data. Results suggest that tree clearing may increase deep drainage by up to 10 times and that in watersheds where shallow unconfined aquifers are present and where %TSS > 0.25%, recharge areas should not be cleared.

Saline or alkaline soils present a strong stress on plants that together may be even more deleterious than alone. Australia's soils are old and contain large, sometimes overlapping, areas of high salt and alkalinity. Acaciaand other Australian plant lineages have evolved in this stressful soil environment and present an opportunity to understand the evolution of salt and alkalinity tolerance. We investigate this evolution by predicting the average soil salinity and pH for 503 Acaciaspecies and mapping the response onto a maximum-likelihood phylogeny. We find that salinity and alkalinity tolerance have evolved repeatedly and often together over 25 Ma of the Acaciaradiation in Australia. Geographically restricted species are often tolerant of extreme conditions. Distantly related species are sympatric in the most extreme soil environments, suggesting lack of niche saturation. There is strong evidence that many Acaciahave distributions affected by salinity and alkalinity and that preference is lineage specific.

Where map legends and map unit definitions reflect the mental models used by soil surveyors to map soils in the past, the association between soil map units and other environmental spatial data can be re-modelled to infer formal survey rules. These rules then can be used to guide subsequent re-interpretations of spatial information for the same area or for another area judged to be similar. Classification trees and Bayesian statistical modelling were used to extract soil mapping rules from an existing map using the Toowoomba area in south-eastern Queensland, Australia, as a case study. In the Toowoomba map area, regional soil-landscape rules could be extracted by combining geology and DEM-derived attributes. The two approaches achieved comparable success. classification trees, Expector method, spatial predictive modelling.