In the section above, we considered the key question of how we choose where to make our counts. Now we must consider how to choose between counting methods. Although we have presented survey design as a linear process, in reality, there should be a strong feedback loop in which the sampling strategies and
field methods influence and alter each other, and they will in turn influence and potentially alter the survey objectives (Figure 2.10). For example, if the required survey method for a particular species, or habitat, is labor intensive, this might dictate that a smaller number of census plots could be covered. Equally, if the sampling strategy dictated that survey effort needed to be spread across several potential habitats because of uncertainty over the true habitat requirements of a scarce species, this might lead us to re-define and simplify our survey objectives. There are some general issues to consider in planning fieldwork:
• The season and the time of day the survey is to be carried out.
• The number of visits to be made to each sample plot or area (commonly around 10 visits for territory mapping, 2—4 for transects, see below).
• The recommended search effort, for example, walking speed (this is particularly important for line transects) or count duration (for point counts), and general counting protocol for the observers.
• The recording units and behavior of the birds to be noted (ages, sexes, nests, singing, calling males, etc).
The three most common field methods are mapping, and line and point transects; each of these is discussed in turn below.
During the temperate zone breeding season, many individual birds are restricted to relatively small areas, actively defending a territory or spending much time around a nest. If a number of visits are made to an area, and the exact location of birds plotted on maps, it becomes possible to identify clusters of sightings and so to estimate directly the total number of pairs or territories of each species present.
An essential component of this method is the use of activity codes to describe bird behavior in the field. These allow observers to record simultaneous observations of territory-holding birds, different forms of territorial behavior and other factors that later allow an analyst to approximate the boundaries between adjacent bird territories. This is the method of territory or spot mapping. Examples of these codes, and of the way that maps can be analyzed, are given in Marchant etal. (1990), Gibbons etal. (1996), and Bibby etal. (2000). At first sight, this would appear to be an extremely accurate and precise method, but this is not always the case and one needs to be aware of the underlying assumptions about territoriality. An obvious advantage of the method is that it produces a detailed map of the distribution and size of territories, allowing us to link bird distribution with habitats. For certain purposes, for example, habitat management on a nature reserve, such information can be invaluable. The method does, however, have a number of disadvantages:
• It requires high quality maps of the study area.
• It is time consuming, requiring up to 10 visits to each site to be able to identify territories (though fewer visits could be made if only one species is being surveyed—a minimum is around four). The time required for mapping can be up to seven times that of transects.
• Because of the intensity of recording, only small areas can normally be covered, generally 1—4 km2 (though again this depends on whether a single species is being studied and its ecology, and how much time is available).
• Mapping requires a high level of observer skill in identifying and recording birds.
• Interpretation of the results can be difficult, subjective, and requires the application of consistent rules, particularly when territory densities are high. Territories at the edge of a plot are troublesome and require arbitrary rules.
• It is an inefficient method for recording non-territorial species, semi-colonial species, those that sing for brief periods, or those that are not monogamous.
• It is difficult to use in dense or featureless habitats (e.g. thick forests, flat deserts) or when bird densities are high.
• It is difficult to compare results across studies unless common standards of territory analysis have been applied.
Despite these limitations, territory mapping has proved a useful method of surveying birds in temperate situations and the results have proved a valuable data source for ecological research. In those situations where it is critical to map individual territories, and sufficient resources exist to do this, it is the method of choice. When used appropriately, it allows fine-scale habitat associations to be studied and probably provides relatively accurate estimates of population size (although precision, and especially accuracy, are not easily measured). Mapping methods can also be usefully combined with nest finding, radio telemetry, mist netting etc. in research projects. Mapping has seldom been used in the tropics, largely because breeding is more asynchronous and many species have complex social behaviors.
There are two types of transect most commonly used in bird surveying, line transects and point transects. The latter are often termed point counts. Both are based on recording birds along a predefined route within a predefined survey unit. In the case of line transects, bird recording occurs continually, whereas for point transects, it occurs at regular intervals along the route and for a given duration at each point. There are a number of variations on this theme where birds are recorded to an exact distance (variable distance) or within bands (fixed distance) from the transect point or line. The two methods can also be combined within the same survey. While there are important differences between the line and point transects, and choosing between them is an important decision in survey design, there are also many practical and theoretical similarities.
Line and point transects are the preferred survey methods in many situations. They are highly adaptable methods and can be used in terrestrial, freshwater, and marine systems. They can be used to survey individual species, or groups of species. They are efficient in terms of the quantity of data collected per unit of effort expended, and for this reason they are particularly suited to monitoring projects. Both can be used to examine bird—habitat relationships (though generally less well than territory mapping), and both can be used to derive relative and absolute measures of bird abundance. Transects can be usefully supplemented and, to some degree, verified in combination with other count methods such as sound recording, mist netting, and tape playback (e.g. Whitman et al. 1997; Haselmayer and Quinn 2000).
There are a series of issues to consider when using transects in the field. The recommended walking speed is particularly important for line transects, as are the counting instructions for the observers. A further important consideration is whether to use full distance estimation, that is, estimating distances from the center of the point count or from the transect line, to all birds heard or seen, or to use estimation within distance bands or belts. In the latter case, one needs to decide on the specific distance bands.
We would always recommend recording some measure of the distance to each bird seen or heard because this provides a useful measure of bird detectability in the habitat concerned and allows species-by-species density estimation (see Detection probabilities). It is always preferable to record the exact distance to birds, or failing this, distance within many belts, but in reality, this will often prove to be impractical. As range-finders become increasingly affordable, they open the way for simple and accurate distance estimation, especially for single species surveys.
At its simplest, a line transect involves traveling a predetermined route and recording birds on either side of the observer. The distance a bird is seen or heard from the transect line is normally recorded as an absolute measure, or in distance bands. Distances should be estimated perpendicular to the transect line (rather than the distance from the bird to the observer). Distance estimation of this kind is key to the estimation of bird densities. Perpendicular distances can be estimated in a number of ways:
1. Distance is estimated by eye from the line, given practice and periodic checking against known distances; fixed distances can also be marked unobtrusively in the field using marker posts or colored tape to aid recording.
2. Observers may be able to visually mark the position of a bird when detected and then use a tape or range finder to measure the distance when they are perpendicular to where the bird was recorded.
3. Bird observations can be plotted on to high quality maps and the distance measured subsequently. This requires good mapping skills and is helped by having fixed markers in the field.
4. Observers can use a sighting compass to estimate the angle (6) between the transect line and a line from the observer to the bird, and use a tape or range finder to measure the distance (d) from that point to the bird. The perpendicular distance is then calculated as d cos 6.
The sampling strategy chosen for a particular survey determines the sample square or unit to be surveyed, but there is still the choice of line transect routes within this area. There are several options, and some flexibility is advisable. For example, a regular or systematic approach could be used with parallel transects orientated north to south, or a series of transects oriented along the long axis of the study area. A random approach, for example, with starting points and directions of transects selected randomly, could be used. One could even use a stratified random approach, for example, with the starting points and direction of transects selected at random, but where each lies within an individual habitat stratum. In reality, topography, watercourses, roads, certain land uses, and access permissions, might all limit access, so that the actual routes counted will differ to some degree from the ideal routes—but such deviation cannot be avoided. In some cases, it might be necessary to substitute a piece of transect for one that cannot be covered, providing it is equivalent in habitat.
The survey design of the Breeding Bird Survey in the United Kingdom, which uses a line transect approach, provides a useful model that can be adopted elsewhere for breeding birds (Gregory 2000; Gregory and Baillie 1998, http:// www.bto.org/bbs/index.htm). This survey is based on two counting visits to a square each breeding season, with one previous visit to set up a route, and uses three distance bands, 0-25, 25-100, and over 100 m. In general, and for ease of comparison across studies of terrestrial breeding birds, we recommend a minimum of two visits to a plot each season and a maximum four visits. We recommend, as a minimum, 2 distance bands, 0-25 and over 25 m for line transects, and preferably three (as above) or more.
Observers often differ in their ability to record birds and other data. If more than one observer is available, bias can be reduced by matching observers to particular tasks they suit (e.g. one spotting and identifying birds, one estimating distances, one acting as data recorder), and by incorporating training. Interobserver differences in bird identification can be monitored and compared (e.g. by plotting the decline in the percentage of bird records unidentified through time).
Line transects are highly adaptable; they have been used to survey seabirds from ships, and waterbirds and seabirds from the air, although these are specialized and expensive applications.
Point transects differ from line transects in that observers travel along the transect and stop at predefined spots, allow the birds time to settle, and then record all the birds seen or heard for a predetermined time, ranging, at the extremes, from 2 to 20 min. Again, we have three choices in deciding where to site point counts within the study plot. There are, of course, many variations on this theme and the counting stations do not need to follow a set route. One could select individual points at random, or by a stratified random design, and access each of them individually—in fact, this is one of the strengths of point transects because they do not require access across the whole survey area. As with line transects, practical barriers might limit the degree to which the ideal routes can be followed, but equivalent points can be substituted with a little care.
If the point transect is the chosen method for a particular survey, then the same set of considerations outlined above would apply. In addition, for point counts one needs to decide on a settling time once the counting station is reached, and on the duration of the count itself. For ease of comparison across studies of terrestrial breeding birds, we recommend the minimum number of visits to a plot is two and a maximum four. We recommend a 5- or 10-min count period plus an initial settling time of 1 min. For the longer period, we suggest that birds recorded in the first and second 5 min are noted separately (allowing some check on double counting, on whether birds are attracted to the observer, and allowing comparison with 5-min counts). We recommend a minimum of two distance bands, 0-30 m and over 30 m, better still would be 3 bands, 0-30, 30-100 and over 100 m. Lastly, we suggest a minimum of 200 m between counting stations. Ralph et al. (1995) review point count methods and provide practical recommendations for their use.
The North American Breeding Bird Survey, which is a continent-wide survey, involves point counts along randomly selected road transects (Sauer et al. 2001; www.mbr-pwrc.usgs.gov/bbs/).
The aim is to record all birds identified by sight or sound with an estimate of distance when first detected. It might be helpful to indicate whether a bird is detected by sight or sound on a recording form. Birds that are seen flying over the census area (aerial species) are recorded separately because they cannot be included in standard density estimation. For such mobile species, it is best to make an estimate of their numbers along each section of transect, or at each point. If birds fly away as you are counting, record them from the point you first saw them. We recommend that birds flushed as you approach a point count station should be recorded from that point and included in the point count totals (but you must make this plain in the write-up). Try to avoid double-counting the same individual birds at a point count or within a transect section by using careful observation and common sense. It is, however, correct to record what are likely to be the same individual birds when they are detected from subsequent point counts or transect sections.
There is little to choose between line and point transects because they are so adaptable to species and habitats, but each is better suited to particular situations (Table 2.1). The strengths and weaknesses of the methods need to be matched against your survey objectives.
Both methods require a relatively high level of observer skill and experience because a large proportion of contacts and identifications will be by song or call.
42 | Bird census and survey techniques Table 2.1 A comparison of line and point transects
Suit extensive, open, and uniform habitats
Suit mobile, large or conspicuous species, and those that easily flush
Suit populations at lower density and more species poor
Cover the ground quickly and efficiently recording many birds
Double counting of birds is a minor issue, as the observer is continually on the move
Birds are less likely to be attracted to the observer
Suited to situations where access is good
Can be used for bird-habitat studies
Errors in distance estimation have a smaller influence on density estimates (because the area sampled increases linearly from the transect line)
Suit dense habitats such as forest and scrub Suit cryptic, shy, and skulking species
Suits populations at higher density and more species rich
Time is lost moving between points, but counts give time to spot and identify shy birds
Double counting of birds is a concern within the count period—especially for longer counts
Birds may be attracted to the presence of observers at counting stations
Suited to situations where access is restricted
Errors in distance estimation can have a larger influence on density estimates (because the area sampled increases geometrically from the transect point)
Some thought needs to be given to surveying birds that are non-territorial, semi-colonial species, those that sing for brief periods, and those that have unusual mating systems; but this is less of a concern than in territory mapping. A potential disadvantage of both transect methods for some purposes is that they tend to follow paths, tracks, or roads and so may not be representative of the area as a whole. A practical way around this using point counts is to establish counting stations at right angles to the transect, and say 30 or 50 m into the habitat.
Having conducted a survey of a species in a particular habitat, it makes sense to compare the results with those of other similar studies in order to place your findings in context. This is often easier said than done, however, because to do so using the raw, or "unadjusted counts," you must assume that the probability of detecting birds is the same for each data set that is compared. It is an inescapable fact that some birds present in your study area will go undetected regardless of the survey method and how well the survey is carried out. Detectability is a key concept in wildlife surveys and we neglect it at our peril. Thus, comparison of "unadjusted counts" will only be valid if the numbers represent a constant proportion of the actual population present across space and time. This assumption is often questionable and has been a matter of much debate (Buckland et al. 2001; Rosenstock et al. 2002; Thompson 2002). To be clear, this could affect comparisons between different habitats surveyed at the same time, and between the same or different habitats surveyed at different times.
The solution is to "adjust" counts to take account of detectability, and a number of different methods have been proposed (Thompson 2002). For example, the "double-observer" approach uses counts from primary and secondary observers, who alternate roles, to model detection probabilities and adjust the counts (Nichols et al. 2000). The "double-sampling" approach uses the findings from an intensive census at a subsample of sites to correct the unadjusted counts from a larger sample of sites (Bart and Earnst 2002). The "removal model" assesses the detection probabilities of different species during the period of a point count and adjusts the counts accordingly (Farnsworth et al. 2002). Finally, "distance sampling" models the decline in the detectability of species with increasing distance from an observer and corrects the counts appropriately.
Distance sampling is a specialized way of estimating bird densities from transect data and of assessing the degree to which our ability to detect birds differs in different habitats and at different times (Buckland et al. 2001; Rosenstock et al. 2002). The software and further information to undertake these analyses are freely available at: www.ruwpa.st-and.ac.uk/distance/. Distance sampling takes account of the fact that the number of birds we see or hear declines with distance from the observer. The shape of this decline, the distance function, differs among species, among observers and, importantly, among habitats—birds within open grassland are detectable over greater distances than those within dense forest— even when they occur at the same densities. Distance sampling models the "distance function" and estimates density taking into account both the birds that were observed, plus those that were likely to be present but were not detected. This method is strongly recommended.
Distance sampling provides an efficient and simple way of estimating bird density from field data. It allows for differences in conspicuousness between habitats and species (though not observers), enabling comparisons to be made between and within species, and across different habitats at different times. Density estimates improve with the number of birds recorded—a minimum of about 80 records is recommended. The method relies on a number of assumptions which need to be evaluated carefully in the field and steps taken to lessen their effects (Buckland et al. 2001). The key assumptions of distance methods are that all the birds actually on the transect line or at the counting station are recorded (for cryptic and shy species this may not be true), and that birds do not move in response to the observer prior to detection.
Was this article helpful?