ESTIMATING THE ABUNDANCE OF BENTHIC INVERTEBRATES - A COMPARISON OF PROCEDURES AND VARIABILITY BETWEEN OBSERVERS

Different procedures for underwater sampling of epifaunal organisms we re compared for their robustness to bias due to observers and precisio n using multifactorial sampling designs. Variability among 3 observers was tested in relation to: (1) the method employed to estimate the pe rcent cover of organisms (visual vs point-intercept technique); (2) th e size of quadrats (50 x 50 vs 20 x 20 cm); (3) stress (sampling at th e beginning vs the end of the dive), and (4) random factors that were likely to change from dive to dive. Precision was expressed in terms o f standard errors (over 3 replicates) obtained with each method and si ze. Two cnidarians (Astroides calycularis and Leptosammia pruvoti) and 2 sponges (Petrosia ficiformis and Geodia cydonium) were considered i n this study because of their abundance at the study site (a submarine cave). Much of the variability was related to significant differences between observers that changed from dive to dive for estimates of the cover of A. calycularis, and from dive to dive and with method for L. pruvoti. The small quadrats were more precise than the large ones whe n used to estimate the percent cover of A. calycularis and L. pruvoti, irrespective of method. In contrast, for P. ficiformis the small quad rats were more precise if sampled with the visual method, while the re verse occurred for the large quadrats. A trend toward a greater precis ion of the large quadrats sampled with the visual method was evident f or G. cydonium, although no significant effects were found. Pilot stud ies and cost-benefit analyses were also used to determine the optimal allocation of resources (time) for sampling epifaunal organisms at dif ferent spatial scales in the cave, for each method and size of quadrat s. The small units in conjunction with the visual method were most eff icient. This procedure offered the best compromise between repeatabili ty among observers, precision and maximization of replication for a fi xed amount of resources. The large quadrats in conjunction with the vi sual method were probably more adequate for G. cydonium, given the app arent greater precision provided by this procedure for the large spong e. Caution is recommended in employing different researchers working o n the same sampling project in extreme environments, at least before d ivergence among observers to changing environmental conditions is acco unted for.