Visual Biology of Hawaiian Coral Reef Fishes. III. Environmental Light and an Integrated Approach to the Ecology of Reef Fish Vision
In the previous two papers in this three-part series, we have examined visual pigments, ocular media transmission, and colors of the coral reef fish of Hawaii. This paper first details aspects of the light field and background colors at the microhabitat level on Hawaiian reefs and does so from the perspective and scale of fish living on the reef. Second, information from all three papers is combined in an attempt to examine trends in the visual ecology of reef inhabitants. Our goal is to begin to see fish the way they appear to other fish. Observations resulting from the combination of results in all three papers include the following. Yellow and blue colors on their own are strikingly well matched to backgrounds on the reef such as coral and bodies of horizontally viewed water. These colors, therefore, depending on context, may be important in camouflage as well as conspicuousness. The spectral characteristics of fish colors are correlated to the known spectral sensitivities in reef fish single cones and are tuned for maximum signal reliability when viewed against known backgrounds. The optimal positions of spectral sensitivity in a modeled dichromatic visual system are generally close to the sensitivities known for reef fish. Models also predict that both UV-sensitive and red-sensitive cone types are advantageous for a variety of tasks. UV-sensitive cones are known in some reef fish, red-sensitive cones have yet to be found. Labroid colors, which appear green or blue to us, may be matched to the far-red component of chlorophyll reflectance for camouflage. Red cave/hole dwelling reef fish are relatively poorly matched to the background they are often viewed against but this may be visually irrelevant. The model predicts that the task of distinguishing green algae from coral is optimized with a relatively long wavelength visual pigment pair. Herbivorous grazers whose visual pigments are known possess the longest sensitivities so far found. “Labroid complex colors” are highly contrasting complementary colors close up but combine, because of the spatial addition, which results from low visual resolution, at distance, to match background water colors remarkably well. Therefore, they are effective for simultaneous communication and camouflage.Abstract
Light habitat and water color in a variety of habitats and microhabitats near Kaneohe bay, Oahu, Hawai'i. All values are in relative photons/sr/nm for radiance and relative photons/cm2/nm for irradiance. (A) Depth series in a channel 400 m west of Coconut Island, using an upward-pointing 180° irradiance collector on a cloudless day close to 1200 h (Station 3 below). Values are presented for surface irradiance (dashed line with sensor just submerged) and depths 3, 6, and 10 m in decreasing thickness. (B) Horizontal radiance (collection angle of sensor = 6°) at Stations 1, 2, and 3 shown in the map below. All were taken at a depth of 1 m with the detector directed out to sea. The bottom depth at each position was (1) 30 m; (2) 4 m; and (3) 7 m. (C) Radiance at 1 m depth over 30 m of water in 3 directions: up, thick line; horizontal, medium thickness; and down, thin (Station 1 below). (D) Radiance at 1 m over 4 m depth on a reef flat. Horizontal, thick line; down, medium, thickness. Thin line, radiance from the back of a cave/hole on the reef at 4 m depth. The hole was recessed approximately 1 m (Station 2 below). (E) Map showing station locations in Kaneohe Bay where light measurements were taken
Fish colors (thin lines) as normalized reflectance and backgrounds against which they may be viewed (thick lines). For (A), (B), (E), and (F), the background is normalized radiance in relative photons/sr/nm (Fig. 4). For (C) and (D), the backgrounds are normalized reflectance. Each graph shows a selection of similar colors and known natural backgrounds to which they are well or approximately matched. (A) Blue, Blue/UV and Blue/UV hump colors (Table 1; for color category explanation, see Marshall, 2000a) and background, radiance measured upward at a 1 m (thick solid line) background radiance measured horizontally at 1 m in deep (30 m) water (thick gray line). (B) Green, green/UV, and labroid green colors normalized to green peak and background, radiance measured horizontally at 1 m in shallow (3 m) water over a reef flat. (C) Yellow colors and background of “average reef” reflectance (average of seven coral and 10 algae, Marshall, 2000b). This curve has been normalized to its green peak to emphasize similarity in the 500 nm change in the reflectance of average reef (the green we see in chlorophyll) and fish yellows (Fig 3). (D) Blue/red, blue/far red, and labroid green colors and average reef reflectance (same as in C) normalized to the far red component of chlorophyll which we do not see. (E) Red of the cave/crevice-dwelling fish holocentrids and Priacanthus sp. and the radiance from the back of a cave at 3 m. (F) Blue/red, blue/far red, and labroid green colors and cave radiance
Yellow of the yellow phase of the flutemouth, Aulostomus chinensis, compared to two backgrounds against which it may appear, average reef color and blue water viewed from a reef edge (i.e., over 30 m Fig. 2C). (A) Normalized reflectance of A. chinensis, gray curve; average reef color (Fig. 3C–D), thick black curve; normalized irradiance (relative photons/cm2/nm) at 3 m depth (Fig. 2A), thin black curve. (B) Three calculated radiances from a reef scene containing A. chinensis. Aulostomus chinensis at 3 m, gray curve (the result of multiplying irradiance and Yellow color in A); average reef color at 3 m, thick black curve (the result of multiplying the average reef color and irradiance curves in A) and background space-light blue water, thin black curve (Fig. 2C) as might be seen near the edge of a reef
Comparison of visual pigment λ-max-values (Losey et al., 2003), visual sensitivities (calculated using Losey et al., 2003; ocular media data and visual pigments), predicted visual pigments from the model (Fig. 5) and fish color peak and R50 points for color steps (see Discussion). All histograms are data in 5 nm bins scored on the y-axis. (A) Histogram of R50 points for steplike reflectances from all fish measured. From 4–6 clusters exist (Marshall et al., 2003). Vertical, lightly shaded boxes indicate the λ-max ranges of known or predicted cone types as shown in (C). The darkly shaded horizontal box indicates the λ-max ranges of known rods. (B) Histogram of maximum wavelengths for peaklike colors from all fish measured. Three or four clusters exist. Bars are the same as those in (A). Shaded boxes are as in (A). (C) λ-max peaks of known single cone visual pigments (McFarland et al., 2003) plotted as a histogram for ease of comparison with (D). (D) Sensitivity peaks of single cones calculated from cone absorptions and, where known, ocular media transmission (Losey et al., 2003). Cones from fish for which no ocular media data exist are not plotted. Gray bars, in this case, are the ranges of predicted sensitivity peaks from model calculations for 10 natural discrimination tasks on the reef
Model predicting ideal visual pigment pairs for an imagined dichromatic visual system in a number of natural situations on a reef. Three examples of the natural tasks chosen are shown here. Data for each example is shown as a best visual pigment pair density plot on the left (see Materials and Methods) and a graph on the right with normalized colors being discriminated (thick black and gray curves) and the visual pigment pair predicted (thin curves). Numbers for density differences from black (best) to light (worst) stippled are on an arbitrary scale for each graph, the black areas encompassing visual pigment pairs of equal and highest efficiency. Best visual pigment pair density plots are mirror symmetrical about the diagonal from the bottom left corner to upper right; however, for clarity only the lower right half is plotted. In all cases, the task is to distinguish two colors, one of which may be a region of fish or a natural background. In some instances (e.g., A and C), one of the predicted visual pigments has a range of good sensitivity either side of the optimum. These are seen as a dark streak in the left side density plots, and the range is then indicated on the right by a double-headed arrow from the optimum peak. (A) A yellow fish (gray curve, from Fig. 2) against blue open water (black curve, from Fig. 4). (B) A blue fish (black curve from Fig. 2) against an average coral background (gray curve as an average of 210 measurements of coral). In this example, a third UV-sensitive visual pigment is predicted almost as good as the 525 nm peak pigment to go with the 435 nm peak pigment. This is shown as a thin dotted black curve on the left. (C) A red fish (gray curve from Fig. 2) against open blue water (black curve)
The complementary colors labroid green and labroid purple/pink of the parrotfish Chlororus sordidus (thin dashed and thin black curves, respectively). When these two colors are combined (thick black curve), the resulting color is a good match to background water radiance over a reef flat (gray curve, Fig. 2D, scaled to summed colors), the habitat in which this fish is frequently found