Investigations into the Properties of Quantitative Vegetational Data: II. Further Data Type Comparisons

Publisher Website
Google Scholar
Abstract
The work reported in the 1st paper of the series (Smartt et al. 1974) is extended to encompass 13 data types, comprising qualitative data, 2 frequency measures, 4 estimates of percentage cover with an without an arcsin transformation, and an estimate of bulk with and without proportionalization. These are tested under more searching ecological conditions than those offered by the previous test community. The qualitative and quantitative components of the raw data matrices are examined, leading to comparisons of the relative performance of the different data types under a given classificatory technique. The results of the classifications are assessed against an independent classification of relevant habitat factors. The main findings are that qualitative data provide the best ecological groupings in terms of the major habitat divisions, but quantitative data can become important at a lower level; for quantitative data, more interpretable ecological results are likely to be obtained from direct measures of the amount of plant material, suitably transformed, than from frequency measures; since the main function of the transformation is to increase the proportional contribution of subsidiary but discriminating species, such transformations should be sought in monotonic functions which implement such increases without destroying the rank relationships of the individual records; and approximations in methods of field recording, unless exceptionally crude, have less effect on the results than changes in the nature of the measure.