Abstract

A comparison was made between analysis methods of archived monitoring data for extrapolating quantitative values or trends from additional partial measurements. These were done from both the ‘species from environment and management’ or ‘environment and management from species’ predictions. The sample data used was from two 19 year annually monitored grazing trials in a tussock grassland environment under 90 different controlled inputs, but treated as if from a random ecological survey. Quantitative estimation of trends from such a reference data set was best by a direct data search for closest matching observation, followed by response surface regression, simple multiple regression, multiple imputation, transformation regression, and least from principal component analysis.

This work is licensed under a Creative Commons Attribution 4.0 License.