In short: Territory mapping, point counts and line transects are key methods for counting birds. They are suitable for many different groups and species of birds. Additional methods can be found in Sutherland (1996).
Table 1. The use of the methods territory mapping, point counts and line transects for different groups of birds (modified after Sutherland 1996). These methods were selected because they can be applied to most bird species.
*=method usually applicable , +=method often applicable.
|Territory Mapping||Line transects||Point counts|
What the method does
Creating a map of all sampling points for each occasion
Territory mapping is a method used to study populations of territorial breeding species such as some ducks, gamebirds, raptors and most passerines. These species are territorial during the breeding season. Males sing in their territory and defend the borders of their territory by disputes. Therefore, the breeding territory is used as a census unit. Territory mapping determines the number of territories of each species in a given area. Multiple species can be mapped during one visit (1). In the first step, the land cover of the study plot is mapped at a scale of commonly 1:2500. Then, obvious features of the plot, such as houses, isolated trees, hedges, ponds etc. are marked on the map. In closed habitats such as temperate woodlands plot sizes of 15-20 ha are often sufficient. In a tropical forest, even half of this area could already be suitable. However, in more open habitats, an area of 60-80 ha is needed (1).
The length of the plot borders should be kept as short as possible by not selecting a long, thin area. This is because many bird territories would be overlapping if the plot boundary was long. To determine if a territory belongs to the plot or not might be difficult in some cases. Therefore, round or square plots should be preferred. Species-rich features such as hedges should also not be located at the plot boundary (1). In temperate woodlands, 10 sampling visits per breeding season are recommended. In temperate open habitats, 5 visits may already be sufficient. To cover both early and late breeding species, the visits should be equally distributed throughout the whole breeding season. For songbirds, the census should usually be started after the first hour of dawn and continue no longer than midday. The first hour after dawn may commonly be omitted because most birds are then singing, which might be confusing in areas with high bird densities. After midday, most birds stop singing which usually makes this time frame unsuitable as well. Each visit to a plot can be completed during a morning (1).
In practical terms, in the beginning of the breeding season, several plot maps are produced, one for each visit. The map is attached to a clipboard that can be covered with a large polyethylene bag in case of rain. By walking at a slow pace, the plot is covered with several routes in parallel with a 50 m distance in between the routes. Each bird that was heard or observed is marked at the location with a species code. Comments are made for evidence of nesting such as nests or birds carrying nesting material (1).
If only a certain group of birds, such as raptors, is censused, the same area can be covered much faster by ignoring all other species and mapping at a bigger scale (e.g., 1:10.000 (2)).
Analysis of maps
After the end of the breeding season, the results for each species in each single-visit map are transferred to a species map covering all visits. This is done by marking each observation for visit 1 with an A on the species map, for visit 2 with a B, and so on. After the map is finished for each species, circles are drawn around clusters of observations. Each cluster should contain at least two registrations for 5-7 visits and at least three for 8-10 visits. Each nest is automatically considered to be a cluster. There are different methods to deal with edge territories (1). We suggest considering a cluster to be within the plot if more than 50% of its registrations lie within the plot.
Strengths & challenges
Territory mapping is a very work-intensive method, that requires good skills in bird identification by the observer. It cannot be applied to species that are colonial or live in loose groups. However, due to the longer time spent in the field compared to line transects and point counts, there is a lower impact of environmental variation such as the weather or the time of visit (1).
What the method does
Point counts are suitable for assessing the relative abundance of vocal or at least highly visible bird species such as passerines. They can be applied in a wide variety of habitats and at any time of the year, not restricted to the breeding season. Fixed counting stations are placed across the study area, either in a grid or in a random manner. From these locations, birds are observed for a fixed time period, usually between 3 and 10 minutes. Often, five minutes are adequate (3).
The distance between the counting stations should at least be 200 m to prevent counting the same individual twice. If the distance between the points is too large, it takes too much time to travel from one counting station to the next. Per study plot, at least 20 counting stations should be installed. To keep the minimum distance of 200 m in between the points, the plot should not be too small (1).
After arriving at the counting station, wait a few minutes for the birds to resume their normal behavior. Then, during the observation time, all birds seen or heard are counted. As most of the birds will be counted in the first few minutes, spending more than 10 minutes is usually inefficient and may lead to increased double counting. Only in areas with very rich bird fauna or where species are especially hard to detect, spending more than 10 minutes might be necessary.
To avoid double counting, the approximate location of every bird can be noted on a page of a notebook (1). The page can, for example, be divided into four quarters and birds recorded in these quarters (i.e. front-left, front-right, behind-left and behind-right).
If only abundances are required, birds may be counted irrespective of the distance to the observer. Such approach is not suited, though, if comparisons are to be made between species and/or different habitats. To determine bird densities, a “near belt” and a “far belt” are established. The near belt is usually 25 m (in closed habitats such as forests) or 50 m (in open habitats) around the counting station. Beyond this distance is the “far belt”. For each individual, the observer notes if it was first seen in the near belt or in the far belt (1). Formulae for easy calculation of bird density based on this assessment can be found in (4) and (5).
At every point, at least two observations should be conducted, one in the beginning and one in the end of the period that should be covered, e.g. the breeding season. For each plot, the time of the day should be noted as this influences the activity and detectability of birds (1).
Strengths & challenges
Point counts are suitable for quickly collecting large amounts of data, even outside of the breeding season. This method can be applied for all birds that can easily be detected by song. The observer, however, needs to have a high level of experience in determining different bird species by song. Compared to territory mapping, counting stations can be distributed relatively easily in a random pattern (1). Since point counts often rely on the birds singing (or sometimes detecting them by eye), this method is unsuitable for less detectable species. The approach may also not be suited for open habitats where birds are likely to flee from the observer (1).
What the method does
Line transects are a method for counting birds of extensive open habitats. It is suitable for shrub-steppe, moorland, offshore seabirds and waterbirds. The observer moves freely through the land, sea or air. Observers move along a route and note down all birds they see on either side. In the first step, a route is determined that should be followed. This route should be positioned relatively randomly. It should not follow a path, hedge, stream, road or similar features as the results obtained here may strongly differ from the surrounding area. Moreover, counting seabirds from fishing trawlers should be prevented as they may attract birds (1).
A transect route does not have to be straight, but can also be rectangular allowing the observer to end at the starting point. The route should be planned so that following it during subsequent visits is as easy as possible. Therefore, circular routes are not recommended (1).
It is recommended to split the total length of the transect into smaller distances. These can directly continue into each other or they can be separated. If several transects should be undertaken, they should be at least 150 m apart in closed habitats and at least 250 m apart in open habitats to prevent the observer from counting the same individual twice. Additionally, the observers need to decide how often they want to visit the transects. It makes sense to complete the same transect several times as species detectability varies seasonally (1).
Similar to point counts, to determine bird density, a “near belt” and a “far belt” may need to be established for line transects. Commonly, the near belt is 25 m to either side of the transect line. Beyond this distance is the “far belt”. For each individual, the observer notes if it was first seen in the near belt or in the far belt. Formulae for easy calculation of bird density based on this assessment can be found in (4) and (5).
Walking speed should be around 1 km (forests) to 2 km (open habitats) per hour. In a notebook with the schematic representation of the transect including the near belt and the far belt, the location where the bird was first seen is recorded. Birds landing or singing overhead are recorded in the central belt while birds flying over are recorded in the far belt (1).
Strengths & challenges
Line transects are especially useful for large areas of homogenous habitat and areas where bird populations occur at low density. It is possible to calculate estimates of density. The costs for this method depend on the habitat. Terrestrial habitats can be censused relatively cheaply whereas sea habitats are more expensive as a ship is needed (1).
Territory mapping has a slight bias towards under-estimating the number of breeding pairs as paired males of some species sing less than unpaired males (6). The assumption of this method, that birds live in pairs in non-overlapping territories, is false for some species such as polygynous species and polyterritorial species. As there is a large proportion of subjectivity involved even when a standard protocol is used, inter-observer variation can be an issue (1).
The main bias for point counts is that the length of time spent in the field is rather short. Therefore, the results can be strongly influenced by weather conditions. For this reason, strong winds, rain or cold weather are unsuitable for point counts (1). Moreover, the presence of the observer might repel or attract birds that are close to the observer which can seriously impact calculations of bird density, as the area sampled by a point count increases geometrically with the distance from the observer (7).
Bird densities that are calculated based on line transects can be biased by several factors. For instance, birds can easily be missed by walking too fast or counted twice by walking too slowly. Furthermore, errors in the estimate of distances have a strong impact on the calculation of bird densities. Also, it might happen that one bird is not detected because of being alarmed by other birds. This highlights that this method relies on assumptions that are, in practice, often not met (1).
The development of bird census data sets is increasingly impacted by the growing movement of citizen science. Especially during the Covid-19 pandemic where many other leisure activities were not available due to restrictions, many people turned to outdoor activities such as bird watching.
This trend in citizen participation is supported by technical development making bird identification easier through the usage of identification apps. Smartphone apps such as “eBirds” help to identify species and also upload photos and the location where the individual was observed. This leads to a growing global online-community of bird watchers who create new data points in observation databases that scientists and environmentalists can make use of for conservation research or planning. For instance, in Canada, the number of people submitting photos to eBirds increased by 30% between 2019 and 2020 (8).
Because many bird watchers are novices, bird identification is not always carried out correctly. This is a general problem of citizen science. In the case of eBirds, bird experts volunteer as reviewers who check photos for correct determination, especially in the case of rare species that are uncommon in a certain area. This increases the quality of the data (8). However, a study conducted in the UK has shown that citizen science data may be reliable only for widespread and common species, even with many data points and a good coverage of the area (9).
Scientists hope that citizen science data may fill data gaps, for instance in the tropics. However, comparisons to data from Bird Life International have shown that the bird abundances for rare species in the tropics are strongly over-estimated. A likely reason is that bird observers are very determined to find these rare species and thereby overlook or under-record more common species and focus on rare ones (10). Nevertheless, as technology progresses, identification apps are becoming better and easier to use. Moreover, scientists are working on new approaches to include such data into their research in meaningful ways. This trend of increasing citizen participation is thus likely to continue.
Fuller, R. J., & Langslow, D. R. (1984). Estimating numbers of birds by point counts: how long should counts last?. Bird study, 31(3), 195-202. Sutherland, W. J., Editor (1996). Ecological Census Techniques - a Handbook. p. 227-259. Cambridge University Press. Robertson, J. G. M., & Skoglund, T. (1985). A method for mapping birds of conservation interest over large areas. Bird census and atlas work. British Trust for Ornithology, Tring.
Gibbs, J. P., & Wenny, D. G. (1993). Song Output as a Population Estimator: Effect of Male Pairing Status (El Canto Utilizado para Estimar el Tamaño de Poblaciones: El Efecto de Machos Apareados y No-apareados). Journal of Field Ornithology, 316-322.
(1) Sutherland, W. J., Editor (1996). Ecological Census Techniques - a Handbook. p. 227-259. Cambridge University Press.
(2) Robertson, J. G. M., & Skoglund, T. (1985). A method for mapping birds of conservation interest over large areas. Bird census and atlas work.
(3) Fuller, R. J., & Langslow, D. R. (1984). Estimating numbers of birds by point counts: how long should counts last?. Bird study, 31(3), 195-202.
(4) Bibby, C. J., Burgess, N. D., Hillis, D. M., Hill, D. A., & Mustoe, S. (1992). Bird census techniques. Elsevier.
(5) Buckland, S. T., Anderson, D. R., Burnham, K. P., & Laake, J. L. (1993). Distance sampling: estimating abundance of biological populations. Chapman & Hall, London.
(6) Gibbs, J. P., & Wenny, D. G. (1993). Song Output as a Population Estimator: Effect of Male Pairing Status (El Canto Utilizado para Estimar el Tamaño de Poblaciones: El Efecto de Machos Apareados y No-apareados). Journal of Field Ornithology, 316-322.
(7) Verner, J. (1985). Assessment of counting techniques. Current Ornithology: Volume 2, 247-302.
(8) CBC (2021). How birding’s pandemic popularity is expanding data collection for science. https://www.cbc.ca/news/science/science-birding-pandemic-data-wildlife-1.6113333 (accessed on 06.03.2023)
(9) Boersch-Supan, P. H., Trask, A. E., & Baillie, S. R. (2019). Robustness of simple avian population trend models for semi-structured citizen science data is species-dependent. Biological Conservation, 240, 108286.
(19) Science Daily (2020). Community science birding data does not yet capture global bird trends. https://www.sciencedaily.com/releases/2020/07/200707084012.html (accessed on 06.03.2023)
The author of this entry is Anna-Lena Rau.