Sample Unit Selection for Studies of Herbaceous Oldfield Vegetation

The size and number of sampling units needed to sample herbaceous oldfield vegetation must be determined for each situation. Prior to obtaining data for a long-term study of early oldfield plant succession at the Waterloo Mills Field Research Station, Pa. statistical tests and theoretical considerations were evaluated. Of the five sizes of circular quadrats tested, the 0.03125 m size was determined to be suitable to obtain detailed data efficiently and yet sample within the estimated smallest scale of heterogeneity. A sample of 100 quadrats, the minimum number needed to compare frequencies, was determined to be adequate (within 10%) from calculations of the standard error of the mean, running means, and species-area curves. Subsequently, annual samples were obtained at 100 points from a stratified grid of 570 quadratcenter points in each 6.40 x 9.12 m subplot studied. OHIO J. SCI. 76(4): 185, 1976 The trend toward broad conceptual thinking on the nature of successional plant communities (e.g. Drury and Nesbit, 1973; Horn, 1974; McNaughton, 1974; and Mellinger and McNaughton, 1975) emphasizes the importance of evaluating the adequacy and accuracy of data sets before using them to make sweeping generalizations. The use of quadrats to sample oldfield herbaceous vegetation necessitates decisions as to the size, shape, and number of the quadrats in the sample. The initial experimental design of each investigation should begin with some form of pre-testing to assure the adequacy and accuracy of the sampling methods. The shape of observational units has long been known to have an effect on the sampled data (Clapham, 1932). Although rectangular, square, and circular quadrats have been used in various vegetational studies, several theoretical considerations favored the use of circular quadrats in the Devon study. The occurrence of a species may be expressed as the probability of occurrence within a given distance (radius of the sample Manuscript received October 30, 1975 and in revised form May 17, 1976 (#75-64). Present address: Taylor University, Upland, Indiana 46989. unit) from a random point. Because a basic goal of the Devon Project was to investigate spatial relationships between species, isodiametric quadrats were preferable to insure that the maximum distance between any two species was minimized relative to the area of the quadrat. Circular quadrats required only a single random center-point for location whereas, if square or rectangular quadrats were used randomness would have been sacrificed by orienting all quadrats in a standardized way or else two random points had to be selected: one to locate and one to orient each quadrat sampled. Circular quadrats were preferred for theoretical reasons and it was found that data from the square quadrats did not differ substantially. The purpose of this paper is to describe the strategy used in sample unit selection for the Devon Project, a long-term study of early oldfield plant succession at the Waterloo Mills Field Research Station at Devon, Pennsylvania. The study was designed from the onset to test succession theory by creating a detailed data base suitable for statistical interpretation.