Satellite measurements of New York and Washington against Newark and Philadelphia suggest light-reduction policies pulled urban plant seasons back toward normal.
Urban trees keep a different calendar from rural ones. They leaf out earlier in spring and hold their leaves later in autumn, and while the heat island explains much of that, artificial light at night does some of the work — plants read day length, and a city never gets properly dark.
A study in PNAS Nexus asks whether policies intended to reduce that light have had a measurable effect, and finds that in two American cities they appear to have.
The work is led by Lin Meng of Vanderbilt University.
A natural experiment, and a new satellite
The design pairs cities. New York City was compared with Newark, New Jersey, and Washington, DC with Philadelphia — in each pair, the first city has light-mitigation measures in place and the second does not. Nearby cities share climate, latitude and growing season, which is what makes the comparison informative.
The measurements come from SDGSAT-1, a satellite in orbit since late 2021 and designed for sustainable development monitoring, used here to measure artificial light at night from 2022 to 2024. Its advantage over the long-serving night-lights sensors is resolution and spectral range: it can separate colour bands, and the blue band is the one that matters.
Blue light is the part of the spectrum that most strongly affects biological timing, because the photoreceptors that register day length in plants and the melanopsin system in animals are both most sensitive there. A policy that replaces older lamps with shielded or warmer fixtures should show up as a reduction in blue specifically rather than in total brightness, which is the kind of distinction the earlier generation of satellites could not make.
New York City and Washington both showed reduced artificial light at night after their policies were implemented, and most clearly in the blue band. Newark and Philadelphia did not.
The plants responded
The ecological measurement comes from the same source: satellite images of new green growth, used to date the start of the growing season, and of autumn colour.
In the cities with mitigation policies, the start of the growing season was several days less advanced than it would otherwise have been, and the turn to autumn colour was less delayed.
Both of those are movements back toward a natural calendar rather than improvements in any absolute sense, and several days is a real effect at the scale of a growing season. A tree that leafs out early is exposed to late frosts, and one that holds its leaves late keeps photosynthesising into a period when the insects and birds that depend on its timing have already moved on. Phenological mismatch of that kind is one of the better-documented ecological consequences of urbanisation.
The authors’ stated conclusion is more modest than the ecology: that high-resolution satellite images can identify relative differences in artificial light at night, which could support urban ecological research. The phenology result is offered as a demonstration of what the measurement makes possible.
Four cities is four cities
The caveats are structural and the release does not raise them.
Two city pairs is a sample of four, and paired-city comparisons attribute any difference to the policy because the pairing is supposed to control for everything else. New York and Newark are adjacent and not equivalent — they differ in density, building height, tree species composition, park area, income and the heat island itself, all of which affect both how much light a satellite sees and when leaves appear. With four cities there is no way to adjust for any of it.
The time series is also short. Light was measured from 2022 to 2024, which is three growing seasons, and interannual weather variation moves the start of a growing season by more than several days on its own. A warm March advances leaf-out across a whole region, and distinguishing a policy effect from a mild spring requires either more years or more cities.
The release reports no numbers beyond this. “Several days” is the only quantity given for the phenology result — no confidence interval, no magnitude for the light reduction, no statistical test. For a paper whose claim is that a satellite can detect relative differences, the absence of the differences it detected is a notable gap.
There are no quotations from any researcher on the page either, which leaves the team’s own framing of the result unreported beyond what the text paraphrases.
Why the measurement is the point
Read as the authors frame it, the study is about instrumentation rather than policy evaluation, and that framing holds up better.
Light pollution has been hard to study at the scale policies operate on. Ground measurements are local and sparse; the older satellite record is coarse and panchromatic, which means it registers total brightness and cannot tell a city that replaced its lamps with warmer ones from a city that simply has fewer of them. A sensor that resolves the blue band at urban scale makes it possible to ask whether a specific lighting ordinance changed the specific part of the spectrum that affects living things.
That capability is what would allow the four-city result to be tested properly. The same analysis across dozens of cities with and without ordinances, over enough years to absorb weather variation, is a tractable study now in a way it was not before — and the honest status of this paper is as the first demonstration that it could be run.
The release does not name the other authors, state funding sources or report any conflicts of interest.
Notes
- Meng, L., et al. “Mitigating artificial light at night: effectiveness of policies and implications for plant phenology.” PNAS Nexus, 2026. DOI: 10.1093/pnasnexus/pgag237.
- PNAS Nexus. “Reducing night lights in cities has ecological effects.” EurekAlert!, 4 August 2026. https://www.eurekalert.org/news-releases/1137931.
Photo by Carlos Oliva on Pexels.com