A vision physiological estimation of ultraviolet window marking visibility to birds

Abstract

Billions of birds are estimated to be killed in window collisions every year, worldwide. A popular solution to this problem may lie in marking the glass with ultraviolet reflective or absorbing patterns, which the birds, but not humans, would see. Elegant as this remedy may seem at first glance, few of its proponents have taken into consideration how stark the contrasts between ultraviolet and human visible light reflections or transmissions must be to be visible to a bird under natural conditions. Complicating matters is that diurnal birds differ strongly in how their photoreceptors absorb ultraviolet and to a lesser degree blue light. We have used a physiological model of avian colour vision to estimate the chromatic contrasts of ultraviolet markings against a natural scene reflected and transmitted by ordinary window glass. Ultraviolets markings may be clearly visible under a range of lighting conditions, but only to birds with a UVS type of ultraviolet vision, such as many passerines. To bird species with the common VS type of vision, ultraviolet markings should only be visible if they produce almost perfect ultraviolet contrasts and are viewed against a scene with low chromatic variation but high ultraviolet content.

Keywords: Avian vision; Colour vision; Spectrophotometry; Ultraviolet light; Visual physiology; Window collision.

Figure 1. Sky reflected in the windows of a high-rise building in central Gothenburg, Sweden.

Photo by Anders Ödeen.

Figure 2. A spectroradiogram of clear sky, unaltered (A) & (B) or modified (C) & (D) to simulate the reflectance of a window marking that completely absorbs ultraviolet radiation below 400 nm, superimposed on normalized single-cone sensitivity.

The four curves represent the UV, SWS, MWS and LWS cones, from left to right. See methods for details.

Figure 3. Ultraviolet reduction as a function of cut-off wavelength.

(A) Translation between the reductions in ultraviolet modelled in this study (UV filter density) and long-pass cut-off wavelengths normally used in commercial ultraviolet markings. For example, ORNILUX Mikado is claimed to transmit 0% radiation below 380 nm (ORNILUX Bird Protection Glass, 2014; ISO standard UV is <380 nm), which is equivalent to at least 50% reduction of ultraviolet for a UVS bird (black line to the left) and at least 25% for a VS bird (purple line to the right). (B) Reduction in UV cone stimulus resulting from 100% long-pass filtration at various wavelengths.

Figure 4. Estimated visibility of simulated markings on perfectly reflecting or transmitting window glass.

The simulated markings reduce ultraviolet (UV) by 25, 50 or 100% compared to unmanipulated glass (Control). Visibility is shown in units of just noticeable differences (jnd), i.e., chromatic contrasts between randomly chosen pairs of patches in the scene, one patch viewed through the marking and the other through the clear window, Geometric mean jnd (points) with 95% confidence intervals (bars) are shown in the graphs. Blue colour means that the mean falls outside the confidence interval of the control and is above 1 jnd. Red colour means in addition that the confidence interval is completely outside the confidence interval of the control. Visibility is modelled from birds with UVS (A–C) and VS (D–E) type of colour vision.

Figure 5. Reflectance (A) and transmittance (B) of cumulus cloud spectra in the test windows.

Figure 6. Estimated visibility of simulated ultraviolet markings as in Fig. 4 but in a scene reflected in a real window.

Figure 7. Estimated visibility of simulated ultraviolet markings as in Fig. 4 but in a scene transmitted through a real window.

Figure 8. Relative spectral content in the horizontal direction of the sun before (A) and after (B) sunrise in late April, in Uppland, Sweden.

Figure 9. Estimated visibility of simulated ultraviolet markings to the VS bird in Fig. 4, but comparing visibility in the middle of the day (A)–(C) to conditions at dawn (D)–(E).

Figure 10. Illustration of the effect of background variation on the visibility of a window marking.

A 30% blue-absorbing grid was digitally superimposed on the image. The reduction in human blue cone stimulation is comparable to what the 50% treatment in this study should have on the UV cones of a UVS bird. Against a blue rich and homogenous background (the sky) the grid becomes clearly visible but the introduced chromatic contrast is marginalised against a heterogenous and longwave dominated clutter (the apple tree). Photo by Olle Håstad.

Figure 11. Depiction of the focal range around an object of interest reflected in a window.

Reflection (solid lines) makes the object of interest (the tree) appear to be located behind the window (along the dotted line). The bird focusses on the tree, which is far enough from the window relative to the distance between the bird and the window that anything on the glass itself falls out of the focal range (depth of field: double headed arrow).