Thursday, 17 October 2019

Dark versus light interior experiment

This post is more about trying to persuade people to do similar experiments than publishing a conclusive result.

A point of debate has been whether it is a good idea to paint the inside of a Swift box black or to leave it unpainted. Two years ago we erected a 10-box triangle (apex box) at Trumpington Meadows Nature Reserve and we painted half of the chambers black inside (see here).

When we recently inspected the boxes, we found evidence that Swifts had explored all 10 chambers, they had built nests in 5 boxes and had bred in 4 of them. With the boxes numbered from the top down and from left to right, the occupied boxes were 1, 5, 7, 9 and 10. They did not breed in 1. Of these 5, 4 are unpainted and only 1 is painted (10). [In addition a pair bred in the Starling box that you can also see in the above picture - this was unpainted inside, but is not included in the analysis below].

The probabilities of randomly distributing 5 pairs of Swifts in 10 nest boxes, half painted black would be:

B = black, W = unpainted
5B + 0W = 0.4%
4B + 1W = 9.9%
3B + 2W = 39.4%
2B + 3W = 39.4%
1B + 4W = 9.9%
0B + 5W = 0.4%

So the probability of 4 or more pairs occupying unpainted boxes is 10.3% (9.9% + 0.4%)
It needs to be less than 5% to be called "statistically significant"

We don't need many more experiments like this with a similar result to get a definitive answer.