Reply to Gagnon et al.: All color vision is more difficult in turbid water

We appreciate Gagnon et al.’s interest (1) in our paper (2), which proposes chromatic aberration and pupil shape as a mechanism for spectral discrimination. Although Gagnon et al. (1) do state our “mechanism works in theory,” they raise two qualitative concerns regarding the potential reduction of contrast variation with focal length: spectral reflectance structure and water turbidity. The authors also suggest that the color-range ambiguity we described might be particularly challenging for benthic octopuses.

Spatial variation of spectral content in the scene provides the basis for the mechanism we propose, and the spectral discrimination through chromatic aberration will depend on the spectral reflectance profile. Gagnon et al. (1) suggest that the natural fish spectra we chose from Marshall et al. (3) to illustrate the principle are not appropriate, but cuttlefish create (movie S2 in ref. 2) their own high-spectral-contrast displays using iridophores with spectrally sharp structural color across a background of fine black lines (providing an ideal spatial/spectral contrast that maximizes our mechanism’s effectiveness). Our proposed mechanism has limitations in inferring the true spectral shape of objects, but so does color vision by opponency, allowing the use of simple RGB displays to simulate variable natural spectra.

Gagnon et al. (1) note that turbidity would cause image blurring and reduce the spectrally induced contrast variation vs. focal length. This is a quantitative issue: our proposed mechanism would be less effective in turbid water once image blurring as a result of small-angle scattering dominates over chromatic aberration in the retinal image quality budget, or at much deeper depths, where light is spectrally narrow. Thus, Gagnon et al.’s qualitative argument has merit but, as illustrated by figure 2 in ref. 1, long-distance color perception in a mildly turbid aquatic environment is difficult by any mechanism. This is precisely why organisms in deep or very turbid environments often have poor color vision (4). The very paper Gagnon et al. (1) cite for reflectance spectra of substrate materials states that “…in general coral reefs are located in clear tropical waters…” (5) and then adopts a chlorophyll concentration of 0.05 mg/m³, quite close to the value of 0.043 mg/m³ we used for our paper’s (2) numerical calculations.

Gagnon et al. (1) assert that octopuses largely use monocular vision, making distance determination more difficult. There are, however, additional cues available to organisms judging distance with one eye. Wells (6) noted that Octopus vulgaris moved their heads up and down before attacking prey, suggesting that they used parallax to judge distance. Reef octopus typically change color after actually touching the object they are attempting to match (see movie S1 in ref. 2). Presumably, physical contact could also give these intelligent organisms a direct distance metric. Gagnon et al. (1) mention octopus displays but we are unaware of any octopus species that signals to conspecifics with particularly colorful displays like the reef-dwelling squid and cuttlefish discussed in our paper (2).

At this point, we believe behavioral experiments are more likely than numerical calculations or qualitative arguments to determine if cephalopods do exploit our proposed mechanism. We hope this dialogue with Gagnon et al. (1) helps inform those experiments.