Friday, 18 July 2014

Swifts do prefer boxes with concaves

There is little in the way of statistics that support what Swifts might prefer in their breeding location, so I was pleased to get this result.

by Dick
Feathers added by a pair of Swifts to a concave
Photo Judith Wakelam
Some years ago, we put 24 nest boxes in St Mary's Ely, with a concave in every other box. So 12 boxes with and 12 without a concave. When we examined the boxes a couple of years later, 10 were occupied, 7 with a concave and 3 without. So, it appeared that Swifts prefer a box with a concave. However, the probability of this being a chance result was 10.7%

This year we put 18 new nest boxes in Worlington church, again with a concave in every other box. So 9 with and 9 without a concave. We have checked the boxes and 6 boxes were occupied, 5 with a concave and just 1 without. The probability of this result by chance is lower at 6.6%.

Neither of these experiments passed the statisticians test for 'confidence' of 5%, but together, they do.

The probability of both of these results occurring by chance is 10.7% x 6.6% = 0.7% 4.18% (I used Fisher's Method to combine the results)

So now we can be sure that Swifts do prefer boxes with concaves.

2 chicks on a concave. Photo Rob Mungovan

Not only that, but 2 of the Worlington boxes contained chicks, (as did 2 first time occupants in Landbeach church) supporting the assertion that breeding in the first year of occupancy is more likely in a nest-box with a concave.

So, on our next visits to St Mary's and Worlington we will insert a concave in all unoccupied nestboxes.

We need more experiments like this e.g.:
Dark interior versus light interior
Oval entrance versus rectangular entrance
Rough exterior versus smooth exterior
Large box versus small box

[For the technically minded, for the statistics I used the Excel HYPGEOM.DIST() function.]

Postscript 2016: Since the above, George Candelin sent me the results after putting 12 concaves into 52 boxes in the Oxford Museum Tower. After the first season 6 boxes were occupied, 3 with concaves and 3 without. The probability of this result occurring by chance, assuming that Swifts have no preference is  12.73%. Combining this with the 2 results above (10.7% and 6.6%) using Fisher's Method gives a significance level of 2.92%

No comments:

Post a Comment