Relationships in soil distribution from digital soil modelling and mapping over eastern Australia under past, present and future conditions

Abstract

This research project applied digital soil modelling and mapping (DSMM) techniques to elucidate relationships between key soil properties and the main soil-forming factors. It attempted to address several broad research issues relating to quantifying the factors that control soil distribution and identifying how these combine together to control soil distribution and their change due to alteration in land use and climate over New South Wales and eastern Australia.«br /» «br /» These broad issues were examined through a number of more specific research issues that were progressively addressed over five chapters, each intended as publishable journal papers. These chapters/journal papers relate to (i) the influence of lithology in soil formation and its application in DSMM (ii) relationships of soil-forming factors to key soil properties and their use in digital soil mapping; (iii) factors controlling the distribution of soil organic carbon stocks (SOC), spatially and with depth; (iv) change in SOC stocks following historic clearing of native vegetation, and (v) change in SOC stocks with projected climate change.«br /» «br /» The strong influence of lithology in controlling soil distribution was demonstrated. Following its classification into 12 classes based on mineral and chemical composition, it was shown to have the highest influence of all soil-forming factors for six key soil properties (SOC, pH, cation exchange capacity (CEC), sum-of-bases, total phosphorous and clay content) examined over NSW. Lithology had similar influence at the scale of eastern Australia; however climate variables were of equivalent or slightly stronger influence for SOC and pH. It was shown to have two to five times more influence than the next highest ranked geophysical covariate such as gamma radiometrics in the models. A marked improvement in the statistical quality of digital models and maps was demonstrated when lithology was applied together with other geophysical covariates.«br /» «br /» Quantitative relationships that are readily interpreted were developed with eight key properties (those listed above plus sand and silt contents) over eastern Australia. These relationships at least partially solve Jenny’s fundamental soil equation in a manner that is more universally applicable and readily interpreted than appears to have been reported previously. Using these relationships, the quantitative influence of the different factors on each soil property is determined, including the unit change per unit variation in the factor, for example a decrease of 0.11 pH units for each 100 mm increase in annual rainfall for the 0-10 cm interval (other factors remaining constant). These relationships were applied together with readily available covariate grids to prepare digital soil maps (DSMs) with 100-m resolution for the eight soil properties over NSW. The predictive ability demonstrated by the maps was broadly moderate, with Lin’s concordance generally between 0.4 and 0.7. They compared well with maps prepared using more sophisticated modelling methods and covariate data. They have the ability to be readily prepared and interpreted and thus have the potential to serve as a useful introduction to the more sophisticated DSMM approaches.«br /» «br /» Systematic patterns of SOC stock levels were graphically demonstrated over 45 different climate-parent material-vegetation cover regimes for upper soils (0-30 cm) and lower soils (30-100 cm) over eastern Australia. There are generally uniform trends of increasing SOC stocks with increasingly moist climate, increasing mafic character of parent material and increasing vegetation cover. Average SOC stocks in the 0-30 cm depth interval range from 16.3 Mg ha-1 (t/ha) in dry, highly siliceous parent material and low vegetation cover environments, up to over 145.0 Mg ha-1 in wet, mafic parent material and high vegetation cover environments. It was demonstrated that the proportion of SOC stored in the subsoil (30-100 cm) relative to the top 100 cm varies systematically from an average of 43% in moist climates to an average of 54% in dry climates.«br /» «br /» Digital soil maps of pre-clearing (pre-European) SOC stocks (100-m resolution) were prepared over NSW. These maps may be used to provide baseline soil carbon levels for carbon turnover models and carbon accounting and trading schemes. They were demonstrated to outperform the existing equivalent maps produced by conventional soil survey methods, with independent validation RMSE values being 33% lower. Comparison of these maps with current SOC stock maps allowed an examination of the change in SOC over NSW following native vegetation clearing. A total SOC loss of approximately 0.53 Gt (530 million Mg or tonnes), or 12.6% over the entire State was revealed. It was demonstrated that the change in SOC stocks following clearing increases (in both absolute and relative terms) with increasingly cool (moist) climate, more mafic parent material and more intensive land use. In the 56 different climate-parent material – land use regimes, the loss varied from less than 1 Mg ha-1 (or 4%) in warmer climates over highly siliceous parent materials under grazing land uses to 44.3 Mg ha-1 (or 50.0%) in cooler (moist) conditions over mafic parent materials under intensive cropping land use.«br /» «br /» Digital soil mapping techniques involving Cubist piecewise linear decision trees, in combination with a space-for-time substitution process (DSM-SFTS), were demonstrated to be effective in mapping the potential change in SOC stocks due to projected climate change over NSW until approximately 2070. Considerable variation in both direction and magnitude of change was demonstrated with application of the 12 different climate change models with their differing climate trajectories. For the mean state-wide change there were some climate models that predicted an increase but others that predicted a decrease over the two depth intervals studied (0-30 and 30-100 cm). Greater consistency between climate change models is required. The predicted SOC changes are primarily controlled by the balance between changing temperatures and rainfall. However, the extent of change is also shown to be dependent on the precise environmental regime, with systematically differing changes demonstrated over 36 current climate-parent material-land use combinations. For example, the projected mean decline of SOC is less than 1 Mg ha-1 for dry-highly siliceous-cropping regimes but over 15 Mg ha-1 for wet-mafic-native vegetation regimes.«br /» «br /» The study has provided quantitative data on the influence of the main soil-forming factors. The necessity of considering the combined influence of multiple soil-forming factors to make meaningful quantitative estimates of current and potential future soil properties is demonstrated. Clear patterns of soil property distribution and change under changing land use and climate conditions are identified, particularly for the vital soil property of SOC. The presentation of relationships that are readily interpreted can assist in their application in natural resource planning and management activities and also in other environment modelling programs. They may thus potentially help to address a range of environmental challenges facing eastern Australia and beyond.