Spectral Composition as Infrastructure Policy: Nocturnal Invertebrate Mortality and LED Procurement in Detroit's Hybrid Lighting Grid

The temperature sensor on the coffee counter at the corner of Michigan and Trumbull logged 98.4°F, then 104.2°F, then 98.4°F again across a span of roughly one second. In Timeline 7, a similar oscillation in a café's building management system preceded a compressor fault that took the block's commercial refrigeration offline for six hours. Here, the reading stabilized. The discrepancy was logged, marked for follow-up, and left open.

Two minutes before its own timestamp, an email from the city's infrastructure department arrived on the field device. The server clock was running ahead of the network time protocol sync. In Timeline 3, the same drift caused a maintenance crew to miss a scheduled window because the dispatch system and the crew's devices were operating in different versions of Tuesday morning. The servers were lying, not dramatically, but consistently, and the lie was load-bearing.

The streetlight timer on Michigan Avenue cycled through intervals that matched no current municipal ordinance. The photometric sensor at ground level reported 0.4 lux, within the human visibility specification. The insects registered something the sensor was not designed to measure.

Detroit's current lighting grid is not a unified system. It is a sequence of procurement decisions made across different budget cycles, each optimizing for a different metric. Main arterials run warm white. Commercial districts run cool white 4000 K LEDs, selected in 2021 when the budget cycle weighted energy efficiency over biological impact. A stretch of the amber corridor near the Corktown test zone runs 1800 K, the result of a supplier substitution that was never flagged because no one filed a complaint about the light quality, and the permitting office had no record of approving it.

In Timeline 3, the 4000 K deployment was a deliberate choice. In Timeline 7, the original sodium lamps remained because the replacement schedule was deferred during a revenue gap. In four of the six accessible Detroits, the amber corridor does not exist. This is not one of those Detroits, and that accident is worth documenting.

Research from the Leibniz Institute for the Analysis of Biodiversity Change, conducted at six railway platforms inside the Westhavelland Dark Sky Reserve in Germany, tested the biological outcomes of spectral composition directly. The study replaced existing 2000 K high-pressure sodium lamps with either 4000 K cool white LEDs or 1800 K amber LEDs, monitoring adjacent unlit control sites throughout. The findings were not ambiguous: the 1800 K amber LEDs attracted significantly fewer individual insects, fewer species, and fewer threatened taxa than either the cool white LEDs or the sodium baseline. The mechanism was the emission curve. Narrow spectral range, minimal short-wavelength blue light output. Moths orient toward blue-range wavelengths the way a guidance system locks onto a beacon frequency. Remove the beacon, reduce the lock.

The Leibniz team estimated one billion insects affected per summer night in Germany alone, scaled to illuminated zone density. The number is not a projection. It is a count, adjusted for geography.

The amber corridor in Corktown is running below predicted moth counts. The models, built on sodium-lamp baseline data, expected higher attraction rates. The amber LEDs are not performing to the model's specification because the model's specification was written for a different bulb. The system is operating outside its documented parameters in a direction that is ecologically favorable, and nobody in the permitting office knows why, because the procurement error that caused it was never entered into the record.

This is the part of the trace where the failure mode becomes visible. The 1800 K amber LED costs the same to operate as the 4000 K equivalent. Energy savings over sodium are identical across both variants. The difference is in the emission curve, which is a line item in the procurement specification, not a capital expenditure. The city is currently paying for the outcome in the 4000 K districts and receiving it accidentally in the amber corridor, because the specification language for the 2021 LED rollout did not include spectral requirements.

The Leibniz study noted its own limits. Only nocturnal flying arthropods were sampled. Ground-dwelling species may respond differently to spectral composition, and reduced attraction to a light source does not automatically mean reduced ecological harm across the full range of insect behavior. The intervention is documented, not complete. What is complete is the finding that spectral composition is a measurable policy variable with biological outcomes that can be specified, procured, and audited.

In Timeline 3, the procurement language was updated in 2022 after a biodiversity review flagged the 4000 K rollout. In Timeline 7, the review was scheduled and deferred twice. Here, the review has not been initiated. The amber corridor exists as evidence of what the specification could require, installed by mistake, performing correctly, undocumented.

The moth counters in the amber zone continued to log lower-than-model counts as the field observation closed. The Michigan Avenue timer continued its undocumented cycle. The photometric sensor continued to report 0.4 lux. None of these systems were communicating with each other. Each was accurate within its own parameters and blind to the others.

A field note from the Corktown corridor, logged at the junction of Michigan and 14th, reads: "The procurement error is doing the work the policy hasn't written yet. This is not a recommendation. It is a description of what is already happening, in one Detroit, by accident."
, G., Michigan and 14th, Timeline 4

Advisory: The amber bulb is in the socket. The specification line that requires it is not in the document. Find the procurement template before the next replacement cycle. The accident does not repeat itself automatically.