Precision and accuracy of visual estimates of aquatic habitat

Abstract

Visual estimates of habitat features are often used in aquatic ecology because of their relatively low cost, ease of collection, and minimal equipment requirements. However, the precision and accuracy of these methods have rarely been quantified. We assessed the precision and accuracy of visual estimates of aquatic habitat features by comparing independent estimates made by 4 observers and by comparing observer estimates with values based on more thorough and objective methods (measured). Percent areal cover and % water-column cover of aquatic macrophytes and algae, aquatic macrophyte and algal species richness, and % composition of river substrates (mud, sand, and gravel) were estimated at 50 points for each of 3 strata in the Niagara River (wetland, water depth ≤1.0 m, and water depth >1.0–2.0 m). Estimates between pairs of observers differed on average by 10% for areal cover, 9% for water column cover, 0.79 for species richness, 11% for mud, 11% for sand, and 3% for gravel. Mean absolute differences (MADs) of observer estimates from measured values were 16% for water-column cover, ∼1.6 species for species richness, 20% for mud, 20% for sand, and 5% for gravel. Visual estimates of the aquatic habitat features examined were relatively precise (i.e., repeatable among observers) but often inaccurate (i.e., biased from the measured values). If visual estimates are implemented, we recommend training of observers based on samples with known values to provide more accurate and precise estimates.