On the Errors in the Ordinary use of Subjective Frequency Estimations in Grassland

Abstract

THE tests described in this paper were made primarily in connexion with work on chalk grasslands which it is hoped to publish later. The results of these tests have, however, a general bearing on ecological method, and are therefore given separately. Subjective estimations of frequency, employing symbols such as d. (dominant), a. (abundant), r. (rare), etc., have often been used in ecological studies, especially in Great Britain. On the Continent, subjectively estimated frequency is more often recorded numerically on a 1-5 scale. When such methods are employed, the work is usually in the nature of a descriptive survey requiring no high standard of accuracy. If fine distinctions have to be made, methods of counting individuals, weighing the crop of each species, or calculating percentage cover, are much more satisfactory, provided that a large enough sample can be studied, than any method depending wholly on the judgement of the observer. In many sorts of natural vegetation strictly quantitative sampling methods are very laborious, because the community can only be considered homogeneous over an area so large that the slow and accurate methods take an almost prohibitive length of time. For certain purposes, therefore, a rapid method of estimation applicable to large areas is very desirable, and the method of subjective frequency estimation is an obvious solution if it is accurate enough for the purpose in view. It has already been shown by Tansley & Adamson (1925), and this paper shows again from another angle, that the errors involved in the method of subjective frequency estimation are very considerable, and it is important to examine how these errors compare in magnitude with the differences which one wishes to study, and what types of plants are most affected. The tests examined in this paper were all done on the principle of making records twice and comparing the differences. All the examples are from chalk grassland, but similar results might be expected in any herbaceous community composed of many species in close association. Even within the chalk grassland community, however, the amount of error may vary according to the type of grassland, especially according to the length of the grass. While the errors recorded 'here are those of a particular person who may be more, or less, accurate and consistent than the average, other observers' records are subject to similar errors. Different observers may, however,