The importance of detectability
The BBS was introduced to monitor the population trends of a broad range of breeding bird species in the UK. As birds are recorded in distance bands, it is also possible to use BBS data to correct for variation in the detectability of different species. This means that we are able to also estimate breeding densities and population size.
The Breeding Bird Survey (BBS) was the successor to the Common Bird Census (CBC), requiring only two short field visits to count birds at each site in each year, which encouraged many more people to take part. Compared to the CBC, the BBS provides less detailed information about bird populations at each site, but by covering a much greater number of sites, results in more robust estimates of trend. But how do we convert the number of birds seen on two short visits, to estimates of density? The secret is a technique known as distance sampling, which takes advantage of the fact that birds are easier to detect when they are close. When conducting distance sampling, observers record not only how many birds they see, but also how far from the observer the birds are. For the BBS, these distances are pooled into three distance bands: the bird was less than 25m from the observer, 25-100m from the observer or more than 100m from the observer (see Figure, right).
Using this information, we can estimate how detectability of a species changes with distance from the observer. By doing this, we can estimate the proportion of individuals of that species that were observed within a particular distance band, and get a handle on the number of birds of that species we did not see, simply because they were further away.
Clearly bird species differ in how easy they are to observe (their detection probability). This varies with the size, colour and behaviour of a bird, as well as the habitat within which it is seen. For example Kingfishers can be surprisingly difficult to see, and only about one third of birds which are within 100m of a BBS observer are seen or heard. In contrast, Chiffchaffs are small but the distinctive song of the male can be heard from far away enabling about 50% of the birds within 100m to be detected. The ability of being able to convert counts from BBS surveys to densities enhances many of the BTO’s science projects. In particular, it enables us to turn BBS survey data into national population estimates of breeding bird species that are so useful for conservation purposes.
National Population Estimates
Two published studies have evaluated the use of BBS data and distance-sampling for producing abundance estimates. The first, a preliminary analysis, considered 20 species, and suggested that distance-sampling was able to produce useful population size estimates (Newson et al. 2005). A further study by Newson et al. (2008), summarised in more detail below, critically reviewed the reliability of the various methods for estimating population size of UK birds and the overall reliability of the picture that they provide of population size. A secondary aim of this work was to explore the factors promoting differences in national population size estimates between methods in order to improve population estimates in future.
Distance sampling was used to generate estimates of the national population size for all common and widespread non-marine breeding birds in the UK using BBS data. This work found that there is a strong positive relationship between our distance-sampled estimates and estimates generated using other methods. This implies that most methods get the broad picture right, even if one of them is less precise for individual species. For some species, detectability may vary sufficiently between males and females to generate biases in population size estimates if these differences are not taken into account. We found a slight tendency for population estimates from distance sampling to be lower than existing estimates for species with marked sex biases in detectability, which may be a wider problem than is currently acknowledged in distance sampling. We are currently considering options which would allow us to incorporate sex-specific detectability into our analyses in the future.
Newson, S.E., Woodburn, R., Noble, D.G., Baillie, S.R. & Gregory, R.D (2005) Evaluating the Breeding Bird Survey for producing national population size and density estimates. Bird Study, 52 (1), 42-54. http://www.tandfonline.com/doi/abs/10.1080/00063650509461373
Newson, S.E., Evans, K.L., Noble, D.G., Greenwood, J.J.D., Gaston, K.J. (2008) Use of distance sampling to improve estimates of national population sizes for common and widespread breeding birds in the UK. Journal of Applied Ecology, 45, 1330-1338. http://onlinelibrary.wiley.com/doi/10.1111/j.1365-2664.2008.01480.x/abstract
Our ability to be able to account for variation in detectability between species and habitats means that we can produce robust estimates of the comosition of bird communities and look at how those have changed through time. We can produce estimates of density which incorporate how the detectability of species may differ between habitats, in order to map variation in species abundance and population trends across the UK and predict how future scenarios of change may affect bird populations.
As a result of the importance of detectability in making the most of BBS data, we are continuing to develop our analyses and use of the distance data collected by BBS volunteers. This currently includes an appraisal of the potential for variation in species’ detectability through time to influence our estimates of population trends. We are also examining how much detectability varies between species, which will tell us for every individual of a particular species seen or heard, how many are likely to have been missed.
Distance sampling in BTO News
Migration blog (3rd – 9th September)
With the first days of Autumn upon us and the breeding season over for many species, the focus is now on preparing for the coming winter months.
Counting birds and the Wetland Bird Survey (Wednesday 22 September, 10am)
This course involves one online session of about 1 hour 45 minutes, with a trainer:participant ratio of about 1:30. Participants' microphones are muted during the presentations but there is a large interactive component...