This dataset contains data from a field-based experiment described in the paper: "Fraleigh, D., Barratt Heitman, J., Robertson, B.A. (2021) Ultraviolet polarized light pollution and evolutionary traps for aquatic insects. Animal Behaviour. October."

The experiment investigates the ability of different families of aquatic insects to detect and move toward sources of ultraviolet polarized light. The experiment conducted adjacent to the Saw Kill River in Annandale-on-Hudson, NY and wild emergent aquatic insects were simultaneously exposed to five lighting treatments: 1) polarized visible wavelengths of light , 2) polarized visible and polarized ultraviolet light, 3) unpolarized visible light, 4) polarized ultraviolet light and unpolarized visible light, and 5) unpolarized visible and ultraviolet light. Insects attracted to a light source reflecting from a oil-filled tray would touch down upon the oil and become trapped. The main results of the experiments was that one family of aquatic insects seems to have the ability to see ultraviolet polarized light and use it as a nocturnal cue to the location of water bodies, while terrestrial insects showed no indication of responses to any type of polarized light. 

The data set was collected on seven evenings in the September of 2016 on the campus of Bard College, NY, USA next to the Saw Kill River. We measured differential attraction to one treatment vs. another as a function of the relative captures of aquatic insect families in one tray vs. another. Raw capture data is avilable for each family, but we have also converted the data to the percentages of captures for each treatment within each capture session. Finally, we created columns for each family rounding each percentage to an integer for using Poisson and negative binomial regression.

The three types of data for each family of aquatic and terrestrial insect collected in the experiment are described above. To reiterate, they are 1) raw captures (# of individuals) for each family during a single capture session, 2) the conversion of raw capture data to the % of captures for that family during that session as a function of the total numbers of individuals of that family captured in all five treatments. This was done to standardize the relative preference of individuals for each of the five treatments within a sampling session. Finally, 3) each % value was rounded off to an integer value (e.g. 35.4% was converted to 34%) for the purpose of analysis using statistical approaches that rely upon integer data. All three types of data are available for each family, for each of the five treatments and for each of the seven sessions.