VISUAL VERSUS RANDOM-POINT PERCENT COVER ESTIMATIONS - OBJECTIVE IS NOT ALWAYS BETTER

We compared the accuracy and repeatability of 2 common methods of estimating percentage covers of sessile organisms: visual estimates and random-point-quadrats (RPQ). Comparisons of estimates were made using both quadrats in the rocky intertidal zone and simulated quadrats drawn on a computer, where estimates could be compared with true, digitized percent cover values. In each case, visual estimates were found to be more repeatable (less within- and among-observer variation) and more accurate (closer to the true value as determined by digitizing) than the RPQ method. RPQs using 100 points were more accurate and less variable than those using 50 points, but were still less accurate (and much slower to carry out) than visual estimates. The RPQ method often missed rare species (< 2 % cover) altogether, but when it 'hit' them, values were usually overestimated. Visual estimates also tended to overestimate percent covers of species (although less than the RPQ method), especially uncommon ones. Thus although the probabilistic RPQ method is supposedly more objective and is statistically valid, visual estimates may give a more accurate representation of relative coverage of sessile organisms, and can reduce overall sampling error because they make increased sample sizes possible. Use of small subdivisions in quadrats, pre-field observer training, and a conscious effort to avoid bias are necessary to make the visual method valid and accurate.