Back during the Ecological Consequences of Artificial Night Lighting conference in 2002, the prescient Steven Pauly made a strong argument that blue light should be avoided in outdoor lighting for human health reasons and posed the question whether it would reduce impacts on wildlife too.  My position at the time was, based on the research presented during the conference, a “silver bullet” for spectrum was not yet known.  For example, we had instances of yellow light being best to minimize insect attraction while some salamanders mis-oriented under yellow light.  Certainly the skyglow reasons to avoid blue light were known, but what color would be best for species?

The LED revolution has made this question even more important and with a new paper this week, we have proposed an analytical framework to address the question and provided an analysis of spectrum for a range of existing outdoor lighting products.

Our new paper, part of a special issue on light pollution and its effects edited by Davide Dominoni and Randy Nelson.

It turns out that overall, Dr. Pauly is right.  We should be avoiding blue light for the sake of wildlife, at least for those wildlife species where we have good spectral response curves.  And when you combine wildlife with melatonin suppression and sky glow, the use of 3000-5000 K LEDs looks like a bad idea.

We looked at insect responses, sea turtle response, a response curve for Newell’s Shearwater, and a response curve for juvenile salmon (all previously or concurrently published).  Then we converted the spectral power distributions of a range of existing light sources to quantal flux and intersected them with the response curves.  Low Pressure Sodium turns out to be the most wildlife-friendly lamp, in addition to being the favorite for reducing astronomical impacts.  But LPS will not be manufactured much longer, so among the LEDs, PC Amber and filtered yellow/green LEDs intersected the least with the wildlife responses.

Figure 3 from Longcore, T. et al. (2018) Rapid assessment of lamp spectrum to quantify ecological effects of light at night. Journal of Experimental Zoology A. DOI: 10.1002/jez.2184

These patterns are predicted by Correlated Color Temperature (CCT) sufficiently so that one could use it as a rule of thumb.  Some outliers, like the spectral output of kerosene oil, which has a low CCT but high wildlife impacts, reduce the power of the association.  Some animal groups have higher correlation with CCT than others.  Overall the r² is 0.62, and for insects it is 0.82.  But for sea turtles and Newell’s Shearwaters it is lower.

Relation between Correlated Color Temperature and intersection with sensitivity of wildlife groups.  Higher wildlife scores predict higher effects on the group that would be desirable to avoid.  Data from Longcore et al. (2018).

As we show in the paper, it is possible to have relatively low intersection with sensitive areas of the spectrum for a combination of wildlife, circadian rhythms, and sky glow, while still providing reasonable color rendering for human vision.  If we wanted to minimize impacts from outdoor lighting, LPS is still the best product.  But one of the reasons it has not been widely deployed is the complete lack of color rendering.  What we found is that new LED products that mostly if not entirely avoid the blue end of the spectrum present an opportunity to reduce impacts of outdoor lighting. To do so, however, will take more than switching to 3000 or 2700 K LEDs, products below 2200 K CCT that look more yellow than white are the way to go.  It still isn’t a silver bullet, and a spectral mitigation strategy has to be coupled with directionality and dimming, but we think the results provide a replicable approach to support the use of PC Amber and filtered LEDs for outdoor lighting.

Wait, there’s more.  All of the lamp spectra, spectral response curves, and summary assessments are available on github, with a webpage deployment of the code here: https://fluxometer.com/ecological/ . You can download and insert your own animal response curves or lamp spectral power distributions.  Please, if you have a behavioral response curve for a species or group of species, send it to us and we’ll add it to the webpage.

Finally, here are a couple of graphs that could be used in presentations, showing the relationship between CCT and the wildlife index.

Last year, the National Park Service released our report, Artificial night lighting and protected lands: Ecological effects and management approaches (Longcore and Rich 2016), which had been in the works for quite a while. Our colleague Andrej Mohar, while enthusiastic about the report overall, pointed out that a figure that we had included of natural illumination under various conditions was wrong, and by quite a bit, when it came to the levels under a “quarter moon.”  This is the story of that error, where it came from, and its correction.

The figure we used in the 2016 NPS report was re-printed with very minor adjustments from Paul Beier’s chapter in our 2006 edited book, Ecological Consequences of Artificial Night Lighting (Island Press).

This figure was published in Ecological Consequences of Artificial Night Lighting and re-published with slight adjustment of the icons in our 2016 NPS report.

The “quarter moon” is shown as causing ground-based illumination of 0.1 lux or thereabouts.  This is obviously and very wrong, and in fact is contradicted in Chapter 1 of the book, which indicates 0.01-0.03 lux for a quarter moon. We should have caught it then, but did not.

Furthermore, a quarter moon, technically, is half of the disk of the moon illuminated, which is one quarter of the entire moon. The icon we used for the “quarter moons” in the figure, however, is that of a crescent moon, with closer to one quarter of the disk  illuminated. We had switched the icon for the “quarter moons” from a depiction of three quarters of the disk illuminated to one quarter of the disk illuminated to match this commonsense language.

Beier (2006) had adapted the graphic from McFarland et al. (1999), who cited their sources as Blaxter (1970) and Brown (1952).  It was McFarland et al. (1999) who introduced the error.  They correctly copied the illumination line from a moon with three-quarters of the face illuminated (a gibbous moon) and incorrectly labeled it as “quarter moons.” They  had three-quarters of the face illuminated in the icon, which was correct, but inconsistent with the label.

 Illuminations produced by the sun and moon as reported by McFarland et al. (1999). Note that “quarter moons” identifies the same line as a three-quarter moon in Brown (1952) once the conversion from footcandles to lux is made.

Compare the figure from McFarland et al. (1999) with the figure produced by Brown for the Department of Navy in 1952. The Brown figure is in footcandles.

Illumination produced by the phases of the moon reported by Brown (1952). The text “1^st and 3^rd Quarters” indicates the “apparent half moon.”

The 1952 Brown report provides curves of illumination from the moon at four phases: full (phase angle 0º), three-quarters full (phase angle 60º), half full (phase angle 90º, which is technically a “quarter moon”) and one-quarter full (phase angle 120º). In the McFarland et al. (1999) diagram, two curves are given: one for the full moon and one for “quarter moons.” The full moon line is correct. The “quarter moons” line is the same as the three-quarters full moon in Brown (1952), meaning that the lunar disk is three-quarters illuminated (phase angle 60º) and not a “quarter moon” in the sense of one quarter of the disk being illuminated or even a quarter moon meaning one quarter of the entire moon visible and illuminated (half of the disk visible and illuminated).

Unfortunately, this error propagated forward to Beier (2006), our 2016 report (which is being reissued with a corrected figure), and Gaston et al. (2014), who changed the icon to a crescent moon (as we did in our report) and might have shifted the line down slightly, but not enough to be accurate, especially given the icon depiction of a crescent moon and textual description.

Illumination from the sun and moon reported by Gaston et al. (2014). The curve for the “quarter moons” is shifted downward slightly but is an order of magnitude higher than a crescent moon (a quarter of the face illuminated) and higher than a quarter moon (half of the face illuminated).

I regret not catching the error when editing Beier’s chapter or when updating the figure in 2016. But Andrej did catch it and now the record can be set straight.

This corrected version of the figure shows maximum values for a full moon (0.3 lux) and quarter moons (0.03 lux) with the proper icon for quarter moon showing half of the face illuminated.

For the record, the approximate values for the maximum clear-sky illumination from the moon directly overhead at its phases are as follows.

These values can vary based on the distance between the sun and the moon and whether the moon is waxing or waning because of the differing characteristics of the face of the moon. Illuminations this high are unlikely to occur under most circumstances, especially at temperate latitudes. A working estimate of illumination from the full moon is closer to 0.1 lux on the ground than the ~0.4 lux potential maximum illumination, a fact that has been recently discussed by Kyba et al. (2017).

Travis Longcore, Ph.D.
August 5, 2017

Literature Cited

Beier, P. 2006. Effects of artificial night lighting on terrestrial mammals. Pages 19–42 in C. Rich, and T. Longcore, editors. Ecological consequences of artificial night lighting. Island Press, Washington, D.C.

Blaxter, J. H. S. 1970. Light, Animals, Fishes. Pages 213–230 in O. Kinne, editor. Marine ecology: a comprehensive integrated treatise on life in oceans and coastal waters. Wiley-Interscience, London.

Brown, D. R. 1952. Natural illumination charts. Research and Development Project NS 714-100. Pages 1–11, 43 plates. Department of the Navy, Bureau of Ships, Washington, D.C.

Gaston, K. J., J. P. Duffy, S. Gaston, J. Bennie, and T. W. Davies. 2014. Human alteration of natural light cycles: causes and ecological consequences. Oecologia 176:917–931.

Krisciunas, K., and B. E. Schaefer. 1991. A model of the brightness of moonlight. Publications of the Astronomical Society of the Pacific 103:1033–1039.

Kyba, C., A. Mohar, and T. Posch. 2017. How bright is moonlight? Astronomy & Geophysics 58:1.31–31.32.

Longcore, T., and C. Rich. 2016. Artificial night lighting and protected lands: Ecological effects and management approaches. Natural Resource Report NPS/NRSS/NSNS/NRR—2016/1213. National Park Service, Fort Collins, Colorado.

McFarland, W., C. Wahl, T. Suchanek, and F. McAlary. 1999. The behavior of animals around twilight with emphasis on coral reef communities. Pages 583–628 in S. N. Archer, M. B. A. Djamgoz, E. R. Loew, J. C. Partridge, and S. Vallerga, editors. Adaptive mechanisms in the ecology of vision. Kluwer Academic Publishers, Dordrecht.

I had the fortune of being able to offer some examples of environmental applications in a paper by Dee Pack and Brian Hardy from Aerospace Corporation for the Small Satellite Conference this summer in Utah.  We (mostly they) show the feasibility of using small satellites to measure upward radiance from Earth at night, with examples ranging from the Middle East to Southern California.  I’m glad to have offered the concept of “darkest path” modeling for wildlife connectivity and perspectives on the usefulness of spectral information at this scale. The work we have been doing with the VIIRS Day-Night Band shows up for the identification of a very bright greenhouse on the Oxnard Plain.  Here is the citation, abstract, and download link.

Pack, D. W., B. S. Hardy, and T. Longcore. 2017. Studying Earth at night from CubeSats. Proceedings of the 31st Annual AIAA/USU Conference on Small Satellites, Leo Missions, SSC17-WK-35.