Insect attraction to light

What kinds of lights are insects most attracted to? Analyze the results of a factorial experiment using light traps.

Motivation

You’ve probably seen that insects seem to be attracted to artificial lights at night. Insects appear to be most attractive to blue and ultraviolet light, and less attracted by red or infrared light.

Previously, most outdoor high-intensity lighting—such as for street lights and parking lots—used sodium vapor lamps, a type of lamp that is very efficient for large lights but also produces a characteristic orange glow. Many cities and countries, however, are moving towards using light-emitting diodes (LEDs) in outdoor lighting, since they have become affordable and highly efficient. LED lights promise to require less maintenance and cost less to operate. Also, they are not limited to the orange glow of sodium vapor lamps—they can produce whiter light. Some research suggests this is beneficial for safety, since whiter light makes roads and sidewalks easier to see. But it can also have some harms.

One possible harmful effect is to insects. If LEDs can produce whiter or bluer light, they may be more attractive to insects at night, disrupting their circadian rhythms and potentially even leading to their deaths as they circle lights endlessly.

This data comes from a study intended to understand how what lights attract insects and what types of insects are most affected, to study the problem. The researchers set up light traps: each trap contained a light bulb and captured any insect attracted by the light to enter the trap.

The researchers conducted their experiment by setting up six light traps, each with a different light bulb. (One trap contained no bulb, as a control.) The traps were set up each evening at sunset and then checked in the morning after sunrise, and the insects collected identified and counted.

Traps were set up at three different locations (one urban, two rural). Relevant weather conditions were recorded. The experiment was repeated 32 times: 16 times at the urban location and 8 times at each rural location. This allowed the researchers to rotate which site (within an individual location) each trap was placed in, and allowed them to collect data under varying weather and moon phases. (Since the moon is bright in the night sky, the presence of a bright moon could presumably affect the results.)

Data

Each row of data represents one trap on one night, and records the light bulb used in that trap, its location, the weather conditions, and the insects collected in that trap.

Data preview

bug-attraction.csv

Variable descriptions

Variable	Description
Light Type	Type of light bulb used to attract insects. No means no bulb at all; A, B, and C are all Philips light bulbs whose color can be adjusted by the user; LED is a commercial white LED bulb; and CFL is a commercial compact fluorescent bulb.
Model Lamp	More detailed description of the light bulb. For A, B, and C, describes the color temperature (K) the bulbs were set to.
Location	Location the light trap was placed. BG = botanical garden, LK = La Kretz field station, Stunt = stunt ranch.
Site	Each location has six separate trap sites, designated by a number. Site codes starting with B are urban (the UCLA Botanical Garden); site codes L and S were rural locations (a UCLA field station and a UCLA ranch).
Standardized Moon	Standardized variable representing the percentage of the moon visible on the night chosen (i.e., the moon phase).
Standardized Mean Temp	Temperature on the night, standardized.
Standardized Mean Humidity	Humidity on the night, standardized.
Standardized Max Wind	Maximum wind speed on the night, standardized.
% Moon Visible	Percentage of moon visible (moon phase).
Start Tem	Temperature at the beginning of the night (Celsius)
End Temp	Temperature at the end of the experiment (Celsius)
Mean Temp	Mean temperature (Celsius)
Humid Start	Relative humidity at the start of the experiment (%)
Humid End	Relative humidity at the end of the experiment (%)
Mean Humidity	Mean humidity (%)
Max Wind	Maximum recorded wind speed (m/s)
Illumination	Illumination produced by the light (lux)
Diptera	Number of insects of the order Diptera captured in the trap
Hymenoptera	Number of insects of the order Hymenoptera captured in the trap
Hemiptera	Number of insects of the order Hemiptera captured in the trap
Psocoptera	Number of insects of the order Psocoptera captured in the trap
Coleoptera	Number of insects of the order Coleoptera captured in the trap
Collembola	Number of insects of the order Collembola captured in the trap
Arachnid	Number of insects of the order Arachnid captured in the trap
Thysanura	Number of insects of the order Thysanura captured in the trap
Isoptera	Number of insects of the order Isoptera captured in the trap
Lepidoptera	Number of insects of the order Lepidoptera captured in the trap
Neuroptera	Number of insects of the order Neuroptera captured in the trap
Larave	Number of insects of the order Larave captured in the trap
Orthoptera	Number of insects of the order Orthoptera captured in the trap
Unident	Number of unidentified insects captured in the trap
Total	Total number of insects captured in the trap
Other Orders	Number of insects of other identified orders captured in the trap

Questions

These are adapted from questions used in notes for 36-707.

1. We are interested in the total number of insects collected in each trap (the Total variable) and how it relates to the light bulb type (Light Type). We also know that the location (Location), moon phase (% Moon Visible), mean night temperature (Mean Temp), and mean humidity (Mean Humidity) might affect insect activity. Conduct an EDA of these variables to determine what relationships the predictors may have with the response.

2. Consider the study design. Because six traps were placed out each night, each with a different bulb (or no bulb), the traps were exposed to the same location, moon phase, temperature, and humidity. No bulb type has data from moon phases that are systematically different from the other bulbs, or was used on nights that were warmer than those for the other bulbs, or anything like that. This is known as a balanced design.

   Comment on what this design choice means. Can location, moon phase, mean temperature, or mean humidity be confounding variables that would bias our estimates of the effect of bulb type? Do we need to incorporate them into our models to obtain an unbiased estimate of the causal effect of bulb type?

3. Obtain the mean number of insects trapped by each bulb. Display the means in a table. Which bulb attracts the fewest insects?

   Note there is one missing count (recorded as NA) that you may need to remove.

4. Select a GLM for the total number of insects collected as a function of bulb type, location, moon phase, temperature, and humidity. Use your EDA to guide your model choice, and produce any necessary diagnostic plots to validate your choice.

   Produce a table of coefficients estimated from your final model, including standard errors. Interpret, in words, the meaning of the coefficient for temperature, giving a 95% confidence interval for the size of the effect.

5. Consider a night where the moon is 60% visible, the humidity is 60%, and the mean temperature is 15 °C. (These are roughly the average values in the data.) Produce a table of the predicted number of insects trapped by each bulb type at the La Kretz station on a night like this, giving 95% confidence intervals for each prediction. Does your result match the table from part 3?

References

Longcore, Aldern, Eggers, Flores, Franco, Hirshfield-Yamanishi, Petrinec, Yan, and Barroso (2015). “Tuning the white light spectrum of light emitting diode lamps to reduce attraction of nocturnal arthropods”. Philosophical Transactions of the Royal Society B 370: 20140125. https://doi.org/10.1098/rstb.2014.0125