Avian UV vision enhances leaf surface contrasts in forest environments

Abstract

UV vision is prevalent, but we know little about its utility in common general tasks, as in resolving habitat structure. Here we visualize vegetated habitats using a multispectral camera with channels mimicking bird photoreceptor sensitivities across the UV-visible spectrum. We find that the contrast between upper and lower leaf surfaces is higher in a UV channel than in any visible channel, and that this makes leaf position and orientation stand out clearly. This was unexpected since both leaf surfaces reflect similarly small proportions (1-2%) of incident UV light. The strong UV-contrast can be explained by downwelling light being brighter than upwelling, and leaves transmitting < 0.06% of incident UV light. We also find that mirror-like specular reflections of the sky and overlying canopy, from the waxy leaf cuticle, often dwarf diffuse reflections. Specular reflections shift leaf color, such that maximum leaf-contrast is seen at short UV wavelengths under open canopies, and at long UV wavelengths under closed canopies.

Fig. 1

Spectral sensitivities of avian cones and multispectral camera channels. Solid lines show spectral sensitivities of avian cones and dashed lines show multispectral camera channels. Most terrestrially foraging birds are tetrachromats, having L, M, and either S(U) and U or S(V) and V cones. L, M, S, V, and U stand for Long, Medium, Short, Violet, and Ultraviolet wavelengths, respectively

Fig. 2

Example images showing the difference in leaf-contrast seen by the UV- and green-cones in deciduous habitats in southern Sweden. For each of the three views, the top two images show the same scene as viewed by the UV and green photoreceptor channels. The UV-cone variant that saw the higher contrast is the one that is displayed. The bottom two images are false-color RGB images with three of the four avian cone channels plugged into the R (red), G (green), and B (blue) channels of a digital display. For example, RGB = UMS denotes that the U-cone is displayed by the red channel of an RGB display, the M-cone by the green channel, and the S-cone by the blue channel. Note that the upper row of RGB images contains a UV channel and the lower row does not. For display purposes, pixel values from original captured images have been (1) normalized by the mean pixel value in each image, (2) adjusted to fill the dynamic range of an sRGB digital display, and (3) adjusted to undo the sRGB gamma scaling of most digital display devices. This means that, viewed on an sRGB-calibrated display, within each channel, pixel intensity scales linearly with cone output. If a pixel was overexposed or underexposed in any channel, the corresponding pixel in all channels was set to white (for overexposed) or black (for underexposed), and excluded from all calculations. Note that false-color images cannot replicate what animals actually see, but provide the best approximation available

Fig. 3

Example images showing the difference in leaf-contrast seen by the UV- and green-cones in wet schlerophyll habitats in Queensland, Australia. See Fig. 2 legend for further details

Fig. 4

Example images showing the difference in leaf-contrast seen by the UV- and green-cones in rainforest habitats in Queensland, Australia. See Fig. 2 legend for further details

Fig. 5

The advantage of UV vision for visualizing leaf achromatic and color contrasts. Boxplots show the median, interquartile range (IQR), and lowest and highest data, within 1.5 IQR, below and above the IQR (outliers not shown). a Michelson leaf achromatic contrasts seen by the different cone channels in different habitats. In each habitat, UV-cones saw higher leaf-contrast than all other cone channels at P < 0.0001. b Color contrasts in units of just noticeable differences (JNDs) seen by a LMSU tetrachromat and all possible trichromatic and dichromatic combinations of avian photoreceptors, again separated by habitat. To make our results as general as possible across the animal kingdom, all photoreceptor compositions were assumed to have the same number of each photoreceptor class in an integrative unit. In each habitat, transforming a trichromat in the visible range (LMS) into a tetrachromat with a UV-cone (LMSU) enhanced color contrast at P < 0.0001. All cone compositions containing both a U- and M-cone (highlighted in orange) saw higher leaf color contrast than any other composition having the same number of cones at P < 0.0001. All P values were derived from paired two-sided sign tests, i.e., “signtest” in MATLAB. Details of statistical tests can be found in Table 1

Fig. 6

Multispectral camera data show that optimal spectral tuning of the UV-photoreceptor for visualizing leaf contrast depends on habitat type. Each of the left panels shows the distribution of the difference in within-channel leaf-contrast seen by the two UV-cone variants in a different habitat. The middle panels show the same distribution types for the two S-cone variants. The right panels show the same distribution types for color contrasts seen by LMSV and LMSU tetrachromats. The heavy vertical black line at x = 0 indicates the point at which the two variants see equal contrast. The V-cone was advantageous for visualizing leaf achromatic contrast in rainforest (P < 0.0001) and wet schlerophyll (P = 0.004) habitats, and the S(U)-cone was advantageous for visualizing leaf achromatic contrast in all habitats at P < 0.0001. An LMSU tetrachromatic system was advantageous for visualizing leaf color contrast in deciduous (P < 0.0001) and wet schlerophyll (P < 0.0001) habitats (paired two-sided sign tests, i.e., “signtest” in MATLAB; details of statistical tests in Table 2). In this figure, we used realistic relative photoreceptor abundances from terrestrially foraging birds such that the ratio of L:M:S:U cones in an integrative unit equaled 3:2.5:1.7:1^–

Fig. 7

Optical model validates the effect of habitat observed in multispectral data and reveals the mechanism behind it. a First panel: when only diffuse reflectance and transmittance were modeled, the U-cone always saw higher leaf-contrast, regardless of whether deciduous or rainforest leaf spectra were used. Second panel: when specular reflections were added to the model, the V-cone sometimes saw higher leaf-contrast, but there was no clear separation between deciduous and rainforest leaf types. Third panel: by plotting leaves illuminated by different specular light sources, we found that the V-cone saw higher leaf-contrast only when overlying leaves were the specular light source. Fourth panel: comparing habitats with different amounts of overlying canopy demonstrated a clear separation between open and closed habitats, with the V-cone seeing higher leaf-contrast in closed habitats ( >80% canopy cover), and the U-cone seeing higher leaf-contrast in open habitats ( <80% canopy cover). We infer that the differences seen between habitats were due to differences in habitat openness, and not leaf optical properties. b Median radiance spectra of leaf surfaces modeled with and without specular reflections when the specular light source was overlying leaves. Leaf type (deciduous or rainforest) was randomized. Here, we can see that specular reflections from the upper leaf surface shift the spectral shape of upper leaf surface reflectance more than lower leaf surface reflectance, and that this difference is more pronounced in closed habitats. The result of this in terms of median achromatic contrast across the UV-blue spectrum can be seen in the lower two panels. In closed habitats (lower right panel), specular reflections shift upper leaf surface radiance such that contrast becomes higher at longer wavelengths. Note that all habitat and environmental parameters that go unmentioned in the figure were randomized in each of 10,000 repetitions of the model

Fig. 8

The ratio of specular to diffuse reflections from leaves affects optimal spectral tuning for perceiving leaf-contrast. The optical model revealed that optimal spectral sensitivity for visualizing leaf-contrast depends not only on the specular light source (see Fig. 7a), but also on the ratio of specular to diffuse reflections radiating from the upper leaf surface. This ratio depends on how much of the sky is blocked by overlying canopy. The less light that makes its way through holes in the overlying canopy, the greater the proportion of specular reflections relative to diffuse reflections, and the more beneficial the V-cone for visualizing leaf-contrast. The line at y = 0 indicates the point at which both UV-cone variants see equal contrast