Potential for the use of GIS and spatial analysis techniques as tools for monitoring changes in forest productivity and nutrition, a New Zealand example

A scheme for combining forest data (growth, foliar nutrition) and soil and site information in a predictive spatial system is proposed to address the following questions: Can geostatistics be used to produce maps of forest sustainability indicator variables such as tree growth or foliage and soil properties to show spatial and temporal trends? Can soil map units be used to stratify forests according to certain criteria and what are the implications of such a stratification on sampling needs? Is a GIS the logical tool for analysis and presentation of trends? We concluded that geostatistics is a valid tool for estimating and interpreting spatial trends in growth (site index) and foliar data (foliar P), but that in the study area available soil and foliage data was too sparse for confident use of this technique. Soil map units were used to stratify the study forest for growth, foliar P nutrition and soil P concentration, but at the selected map scale this only led to sampling efficiencies for the foliar N data. A modelled example of how the GIS and soil map units could be used to illustrate predicted changes in productivity and P status over space and time was presented. Soils were ranked according to their resilience to changes in P status based on initial soil P concentration, soil volume, and P fixing character. Likely changes in foliar P concentration in P. radiata stands on such sites were estimated over two rotations and changes in overall forest productivity predicted. In this study, based on existing available forest data, it became apparent that monitoring systems for site quality will not be able to depend on such data alone, and structured explicit sampling designs will be needed to address the requirements of long-term monitoring programmes. Tree-based indicators will be easier and cheaper to monitor than soil-based indicators, but a mix of both indicator types is likely in future schemes. # 1999 Elsevier Science B.V. All rights reserved.