Aa. Thompson et Bd. Mapstone, OBSERVER EFFECTS AND TRAINING IN UNDERWATER VISUAL SURVEYS OF REEF FISHES, Marine ecology. Progress series, 154, 1997, pp. 53-63
Visual survey techniques are used widely to estimate abundances of tar
get organisms in terrestrial and aquatic environments. There are a num
ber of methodological 'errors' in almost all applications of visual su
rveys. Given the dependence of all visual survey data on the skill and
technique of the observer, one potentially important source of imprec
ision and/or bias is variation among and within observers. In studies
involving large amounts of fieldwork over great geographic range and m
any years, it is inevitable that observers will change from place to p
lace and through time at any single site as they are replaced or gain
experience. We present the results of 3 observer training/calibration
exercises that indicate that observational studies in which multiple o
bservers must be employed may be subject to considerable observer-rela
ted biases and imprecision. We found that careful training and calibra
tion of observers ameliorated such effects for most taxa, but non-triv
ial levels of bias for some taxa and imprecision in estimates for seve
ral taxa remained even after thorough training. It is essential that t
he influence of observer bias and imprecision be well documented in mu
lti-observer monitoring studies, so that (spurious) patterns related t
o differences among observers can be distinguished from real spatial o
r temporal patterns in the environment.