We examined precision and accuracy of stream habitat measurements in relation to transect spacing, by using seven study reaches, where wetted width, water depth, and current velocity were measured on the basis of 25-cm (<0.1 width) interval transects. Means and SDs of wetted width, water depth and current velocity based on the 25-cm interval transects were considered the true values, and these six variables were calculated on the basis of different transect spacing from 0.25 to 3.0 wetted width. We assessed (1) decline of the precision, expressed by coefficient of variation (100SD / mean) of repeated measurements, (2) decline of correlation coefficient of the true values with the values based on each spacing, and (3) increase of differences in frequency distribution between the true values and values based on each spacing, due to increasing transect spacing. We also conducted a literature survey to examine (4) transect spacing used by previous studies. Mean wetted width, mean depth and mean velocity based on 0.5-width interval spacing were sufficiently precise and accurate (precision <10%; correlation coefficient with the true value: r ≈1.0). Overall results suggested that reasonable precision and accuracy would be expected up to 1.5-width interval spacing (precision<15%; r with the true value >0.95), when mean wetted width, mean depth and mean velocity are used as habitat variables for among-reach analysis. However, the precision and accuracy for SDs of wetted width, depth and velocity were lower than those for means of the three variables, suggesting that, when SDs (or other measures of variance) of these variables are used to represent within-reach habitat variations, finer spacing would be needed. Further, when a detailed analysis is conducted on within-reach habitat variation (e.g., microhabitat preference by fishes), <0.5-width interval spacing was suggested to be ideal and 1.0-width interval would be necessary.
View full abstract