Abstract

Back, flank (reddish band along the lateral line and below the lateral line), and belly regions of cultured rainbow trout Oncorhynchus mykiss were characterized in terms of skin coloration, spottiness, and darkness by using computerbased image analysis as an approach to separate fish into desirable back color categories for market and export. Three highly represented colors were observed on the back (mean ± SE percentage of area covered by the given color): blue or sky-blue, 18.0 ± 1.8%; yellow, 13.4 ± 1.5%; and green, 66.1 ± 2.0%. Individuals with a high frequency of these colors on the back (at least 50% of area covered by a given color) were reclassified into groups exhibiting a yellowish back color (YB), a bluish back color (BB), or a greenish back color (GB) to determine whether the BB group presented differences in various skin color variables when compared with the other back color groups; an undefined back color (UB) group (fish for which all colors had less than 50% coverage) was also used in these comparisons. The number and size of dark spots (mean ± SE) on the back was significantly smaller in the BB group (2.3 ± 0.3 spots/cm2; 1.3 ± 0.2% of area covered by spots/cm2) than in the YB group (3.4 ± 0.2 spots/cm2; 1.9 ± 0.1% of area covered by spots/cm2); furthermore, skin darkening on the back as measured in grayscale space (where 0 = black, 255 = white) was slightly greater in the BB group (58.2 ± 3.9) than in the GB group (80.7 ± 2.8). In the lateral band along the flank, the percentage of area covered by purple was significantly lower in the BB group (7.1%) than in the UB group (27.5%). Skin darkening of the back and darkening of the belly demonstrated a significant positive correlation in the BB group (r = 0.558, P < 0.05). Discriminant function analysis clearly distinguished the BB group from the other back color groups based on color profile. These results support the fact that the BB group, more commonly referred to as “blue back” rainbow trout, presents a color pattern of high commercial interest that can be used to improve the external appearance of farmed rainbow trout. Furthermore, computer-based image analysis was shown to be an effective technique for objectively measuring skin color to categorize rainbow trout into particular color patterns.