MARINE ECOLOGY PROGRESS SERIES versus random-point 'objective' is not - PDF document

MARINE ECOLOGY PROGRESS SERIES versus random-point 'objective' is not always differ significantly each other. digitized values illustrates differences among the observer showed the the true value, the results also differed widely Observers' bar, resents the multiple comparisons indicated that included in substantially lower (50 points) was was the least. Other analyses tested how well different proto- species (Table Species that covered estimation (Table 5A). The rare species-occurrences RPQ, 45 When a protocols in among the 3 replicate from all test, p last column significantly different Mean standard dev. (SD) Observer 1 point did rare species it usually resulted in points) or estimation was more reliable 'finding' rare species cover values (Table 5A). both visual estimation (Table 5B). method continued expected (Table 4 protocols differed their frequency the greatest degree, and most often 5. Comparison 4 estimation proto- species occurrences overestimated values, or for include species-occurrences that in the no point that species). always non-integers, making observer values Overestimations Underestimations Number Mean (SD) 1.9 (1.4) 1.1 (1.0) 0.28 (0.23) 70 Observer 1 0.33 (0.29) 148 0.34 0.09 (0.05) 0 4.0 (3.6) (0.83) 68 0.76 (0.62) 134 0.91 (0.41) 0 These data sampling (Fig. than were the comparable amounts results suggest visual estimation and repeatable same observer, method, although theoretically objective, was visual estimation. the true value, although probabilistic sampling counts converged once per sampling not just be important time in permanent quadrats), cial in assessing the strengths and weaknesses 2 methods. visual estimation better than the to 100 a standard 50 to 100 calculations show large numbers a species cover, only to distinguish from 50 are needed a species often missed method was noting all species present (1992) found method was Researchers who use point-intercept methods field usually species (those the field, however, the dom point picked up the upper understory [more sophisticated frames et al. problem]. Photographic estimation niques encounter estimated without and then held out the number in the required a substantial in- sampling time. Since time in field especially in intertidal research, must be weighed against the points still estimates that variable, and biased towards overestimation ways more difficult perform, since concentrate on rather than the species under each larger than difficult to mentally (written filled could facilitate this explicitly, our use a large number censusing scattered not give accurate enough reference addition, a method should plex decision (see 'Methods'); Pre-fieldwork training images used this study (Fig. technique. Meese Tomich (1992) observers than either digitizing, but describe the use we consider must be results show that mates that racy from