Monday, 11 February 2013

How many Swifts are there in GB and the UK?

The February 2013 issue of British Birds has a paper titled 'Population estimates of Birds in Great Britain and the United Kingdom' by Andy Musgrove et al. This of course includes estimates for the Swift population. As Swift is such a famously difficult species to census, we thought it worthwhile taking a closer look, as well as coming up with our own "fag-packet" estimate to see how it compares.

Written by Dick

The population estimates are produced by the Avian Population Estimates Panel (APEP), composed of representatives from the BTO, GWCT, RSPB, WWT and JNCC. Their estimate of the population of the Common Swift in 2009 is:

GB: 87,000 pairs (63,000 - 111,000)
UK: 87,000 pairs (64,000 - 111,000)
[Note the UK includes the island of Great Britain and Northern Ireland].

The authors believe this to be an underestimate and that the reliability of the result is 'poor'.

The numbers are rounded to 2 digits (so the nearest 1000), implying that there are less than 1000 pairs in Northern Ireland. This seems low, given, for example, the huge numbers of Swifts sometimes seen feeding over Lough Neagh.

The estimate is based on the results of the Breeding Bird Survey (BBS), a survey designed for monitoring population change, rather than absolute population levels. The method calculates the density of birds within a defined distance of a transect, and then extrapolates across the country.

It is not stated whether account has been made for non-breeders and incubating breeders. Swifts do not breed until about 4 years old, implying that roughly 40% must be non breeders. If haIf the breeders are occupied in a nest site, then only ~70% will be at large at any one time for BBS participants to see them.

The final paragraph of the paper invites people to get involved in gathering data and coming up with their own ideas, so here goes!

Fag-packet estimate

Swifts nest in houses, so one might expect the number of pairs in any area to be proportional to the number of houses. In what follows, we do not always know how many houses there are, but we have found figures for the human population, so we have divided this by 2.5 to get the number of houses. From this we calculate the number of pairs per 1000 houses, then we can extrapolate across the UK.

There are a number of published local estimates of the breeding Swift population (see table below). One of the earlier ones was Perrins estimate in 1971 who thought that there were between  20 and 50 times as many breeding Swifts in Oxford as there were in the museum tower, which numbered 40 pairs at the time. So we have taken his lowest number, 800 pairs in 1971. Likewise for Northants we have used a minimum estimate for the population in 1998 by taking the minimum from the 1978 survey times the maximum percentage decline estimated in 1998 (Richardson 1978 and 1998).

Since 1994, Swifts have been declining at a rate of about 3% per annum (BBS).  So, to estimate the population in 2009, we need to make an adjustment to account for the decline since the survey was made. As there are no figures before BBS, we have used the same rate of decline to adjust numbers obtained before 1994 (i.e. Oxford).

The number of households in the UK is ~25 million in 2009. So, the APEP estimate of 87,000 pairs in 2009 translates into 3.5 pairs of Swifts per 1000 houses. In the table below, most surveys come up with a density larger than this:

Analysis of some Swift survey data.
The minimum (Northants) and maximum (Oxford villages) densities translate into overall breeding population estimates between 78,000 and 265,000 pairs. Note though that the Oxford villages sample size is small.

To calculate an estimated mean density across the UK, we can use either a calculated overall mean sample density, or the mean of the sample densities.

Mean sample density - total pairs / total households:
Total estimated pairs in 2009: 6,580
Total Households: 958,252
Average pairs per 1000 households: 6.87
UK population (25,000,000 households): 171,671 pairs

Mean of sample densities
Average of sample densities: 6.28 pairs per 1000 households. 
UK population (25,000,000 households): 156,894 pairs

The first estimate is effectively weighted by sample size and the second is unweighted. The difference is because the largest sample (Cheshire) recorded a relatively high density. 

Based on this latter, lower figure, the estimate for Northern Ireland, with 1.8 million people, would come out at about 4,700 breeding pairs, but we did not use any data from Northern Ireland!

So how many Swifts are there in the UK in total?

Using our lower figure, rounded to 2 figures, of 160,000 as an estimate of UK breeding pairs in 2009, and given that  ~40% are non breeders, then the total number of individual Swifts in the UK during the breeding season in 2009 may have been of the order of 530,000. If each pair of Swifts yields on average 1.5 young, then the number of Swifts leaving the UK in the autumn is 770,000 (minus a small amount of mortality during the breeding season).

Comparison with some other urban species

The paper includes an estimate for House Sparrow of 5,300,000 pairs in the UK, and for Starling the number is 1,900,000 pairs. It is ironic that these two hole-nesting species are red-listed, but Swift, at an official estimate of 87,000, is only amber listed. There must be scope for a huge increase in the Swift population in the UK if we could provide nesting sites for them.


This analysis is a fag packet exercise, and there is scope for more refinement. For example, we could apply the regional BBS changes; we have not accounted for changes in number of households over the years; all of these surveys were in England, and there are not enough of them. Maybe larger samples should be given more weight as should more recent samples. 

With more data from more recent local surveys of this type one could investigate other variances. Wotten et al 2002 showed that houses in rural areas are more likely to hold Swifts than in urban areas; older houses, especially those constructed before 1919 contain more breeding Swifts than houses built since then and there are regional differences in density.

Our result comes within a factor of 2 of the APEP estimate so there could be some mileage in this approach. After all, APEP say that they think that their number is an underestimate, and our estimate is bigger!


Thank you to Andy Musgrove and Stuart Newson of the BTO for encouragement to publish this. Thank you to Jake for digging out old papers with much of the data from his Concern for Swifts days and to Chris Mason for providing estimates for 6 Oxford villages.


Hornbuckle, J. 1984. Survey of Swifts breeding in the Sheffield areaMagpie 3: 29-33. Sheffield Bird Study Group.

Roberts, Graham C.M. 2001. The 1999-2000 Sussex breeding Swift survey, comparison with the 1968-70 survey and conservation issues The Sussex Bird Report 53

Martin, B. 1998. A survey of summering swifts in Cheshire and Wirral and their conservation status. Cheshire and Wirral Ornithological Society.

Musgrove A., Aebischer N., Eaton M., Hearn R., Newson S., Noble D., Parsons M., Risely K. and Stroud D. 2012.
Population estimates of birds in Great Britain and the United Kingdom. British Birds 106 * February 2013 * 64-100

Perrins, C. 1971Age of first breeding and adult survival rate in the Swift. Bird Study 18: 61-70.

Richardson P. W. 1979. A Survey of Breeding Swifts (Apus apus) in Northamptonshire in 1978. Northants Bird Report 1979

Richardson P. W. 1998. Swifts in Northamptonshire 1998 - a repeat sample survey. In litt.

Wotton S. R., Field R., Langston R. H. W. and Gibbons D. W. 2002. Homes for birds: the use of houses for nesting by birds in the UK. British Birds 95 * November 2002 * 586-592

No comments:

Post a Comment