General Technical Report
The assessment of population size should be an integral part of any monitoring program. Various methods have been employed and thoroughly tested (see Ralph and Scott 1981). Abundance of birds has long been used to measure habitat suitability but is often retrospective, giving trends without any possibility of determining causation, and can even be misleading (van Horne 1983).
It is desirable to use a method that allows the biologist to census as many points as possible in the time available, thus gaining as many independent data points as possible. That is, it is much better statistically to census five points in a 10-day interval, than to count at one point five times. The farther apart each of the five points, the more likely the data can be extrapolated to a larger region.
Below we outline four major methods. Two of these, the point counts and the spot mapping methods, are the most common ones used (for definitions see Ralph 1981b). The point count is probably the best for most surveys and has been adopted as the standard method for monitoring (Ralph and others, in press). The methods for both are taken in part from the excellent book by Koskimies and Vaisanen (1991). In addition, a strip transect count and an area search method are also presented. The latter is especially popular with volunteers.
Time of Day
The best time for censusing at most temperate latitudes during the breeding season is usually between 5 and 9 a.m. Under most circumstances, no counts should be done after 10 a.m. Exceptions could be in the non-breeding period. It is best to start within 15 minutes of local sunrise. Examining pilot data is the best way to determine when detection rates are the most stable. In general, the period between official sunrise and the ensuing 3-4 hours is usually relatively stable. For most species, during the period between dawn (first light) and sunrise, the number and rate of birds singing is somewhat higher than the rest of the morning. For maximum comparability in detection probabilities for species among points, it will be best to start at sunrise rather than at first light.
Breeding season point counts should be run during the time of year when the detection rates of the species being studied are most stable. Within the breeding season, the months of May, June, and the first week in July are best for counting most passerines in North America. However, stable counting periods, when the rate of singing of the various species has stabilized, are as early as April in the Southeast and Southwest and may extend later in the boreal zones. In Latin America the breeding season will be longer, and censuses can profitably be conducted throughout the year.
Birds should not be surveyed when rain or wind interfere with the intensity or audibility of bird sounds, when fog or rain interfere with visibility, or when cold weather shuts down bird song activity.
We suggest two levels of point counts. Extensive point counts are intended for a series of points, placed at a minimum of 250 m apart, largely on roads or trails over an entire region. Intensive point counts are placed within a mist net or nest search plot.
The account below is based on Hilden and others (1991), and the standards are taken from Ralph and others (in press), as adopted by the Point Count Workshop of the Monitoring Group of the Neotropical Migratory Bird Conservation Program, held in Beltsville, Maryland, November 1991.
Background and Aims
In many countries point counts are the main method in monitoring the population changes of breeding landbirds. With the point count method it is possible to study the yearly changes of bird populations at fixed points, differences in species composition between habitats, and abundance patterns of species. The point count method is probably the most efficient and data-rich method of counting birds. It is the preferred method in forested habitats or difficult terrain. Point counts involve an observer standing in one spot and recording all the birds seen or heard at either a fixed distance, or unlimited distance. This method can be conducted one or many times at a given point. The North American Breeding Bird Survey of the U.S.D.I. Fish and Wildlife Service is such a method.
The point count method applied to landbirds does not provide reliable data on waterfowl; however, rails and waders are counted well. Some landbirds also pose problems as they are particularly quiet, loud, nocturnal, or flocking. If these species are of particular interest, the method may be modified to accommodate them.
Equipment and Time Needed
One should not start point counts without good identification skills, including a knowledge of the songs and calls of birds. Details on training for distance estimates are given in Kepler and Scott (1981). In the tropics, learning all the songs and calls of all species at all times of the year is difficult in practice. In many areas it takes an experienced observer 4-8 weeks to identify 80-90 percent of the species. In temperate zones, this can often be done in less than 2 weeks.
For the census one needs a map, a pencil, notebook, a watch that shows seconds, and binoculars. The route and the points are marked on a survey map and, if necessary, in the field with plastic tape or streamers to ensure that the same points are found in the following years. The observer may move from one point to another by foot or with a vehicle.
The time needed for censusing one point count route is usually no more than four morning hours, depending on the distance between the points and the method of travel.
Choosing a Counting Route for Extensive Point Counts
An extensive point count route should encompass all the habitats of a region, if possible. In addition, it should include any mist net or nest searching plots in the region. In choosing a route and laying out the points for census, use a systematic rather than random sampling design, either on roads or off roads. Systematic gridding of points is preferable to the random placement of points in most cases. Systematic placement can include placing points at designated distances along roads. Do not stratify by habitat, unless separate estimates for a habitat are required. If the goal is to estimate population trends for an entire management unit, then point counts should be spaced evenly throughout that unit, or along the road system in an area, without regard to current habitat configurations.
Observers should attempt to carry out censuses primarily on tertiary roads, then secondary roads, and should avoid wide, primary roads. Off-road censuses should be carried out on trails, if possible, in major habitats not covered by road systems. Using roads, travel time can be reduced to as little as 1-2 minutes between sampling points. Under optimal road conditions, up to 25 5-minute point counts can be conducted in one morning. In an off-road situation, the number of point counts one observer can conduct during a morning varies between 6 and 12. Roadside habitats usually do not sample all of the available habitats. In this situation, a collection of both on- and off-road surveys can be created that best fits local conditions. Although a road modifies the surrounding habitats, we feel that tertiary road systems (i.e., narrow dirt roads) allow for birds to be counted in approximately the same proportions as off-road surveys.
The minimum distance between point counts in wooded habitats is 250 m. Birds previously recorded at another sampling point should not be recorded again. In virtually all habitats, more than 99 percent of individuals are detected within 125 m of the observer. In open environments, this minimum distance should be increased because of the greater detectability of birds. Along roads, where travel by vehicle is possible, distances of 500 m or more should be used.
Choosing Points for Intensive Point Counts
The intensive point counts are conducted within a study plot for mist nets or nest searches. We suggest between 9 and 16 points in a grid of 3 by 3, 3 by 4, or 4 by 4 points. For most analyses, the birds counted from these points will be combined into a single mean. Therefore, the distance between points is less critical than for extensive point counts where each point is intended to be statistically independent. The points on an intensive census grid should be adjusted to fit within the netting array or nest search plot so as to fully census the area. It is very important not to include areas much beyond the array or plot boundaries. These are covered by the extensive point count censuses. For example, a census grid of nine points, 100 m apart, would cover 4 ha. Allowing for an effective radius of censusing of perhaps 50 m outside this grid, the area covered expands to about 9 ha. A grid of 12 points 150 m apart would have an effective area of about 22 ha. Thus a census grid should have points that are between 75 and 150 m apart, depending upon the area to be covered and the number of points to be included. Under most circumstances 9-12 points should be more than adequate.
The censuser should approach the point with as little disturbance to the birds as possible. Counts should begin immediately when the observer reaches the census point. Time spent at each count point should be 5 minutes if travel time between counting points is less than 15 minutes (for greater efficiency) and 10 minutes if travel time is greater than 15 minutes. If a survey is primarily for inventory and few points will be surveyed, then 10 minutes is appropriate. Data should be separated into those individuals seen or heard during the first 3 minutes (for comparison with Breeding Bird Surveys) and those additional individuals heard in the remaining 2 and 8 minutes.
The details of each point are recorded: the reference number, name of the point, date, and the time. The species are written down in the order they are observed. For each species, the number of individuals is recorded separately for those within a circle of 50 m around the censuser and for all those outside the circle, out to an unlimited distance. In noisy environments, dense foliage, or in tropical forests, observers have found that 25 m was preferred. The distance is that at which the individual was first observed. For birds near the 50-m border, the category may be confirmed by measuring paces to the border when the counting is over. If a bird flees when the censuser arrives at the point, the bird should be included according to its take-off place. Birds that were detected flying over the point, rather than detected from within the vegetation, should be recorded separately.
Estimating distances requires experience, so a new censuser should measure the length of steps in different terrain, and then check the distance to several singing birds in order to make the estimating of distances routine. Estimating may be eased by either natural or artificial landmarks.
If there are several males of the same species around a point, one may sketch in the margin the directions and distances of each singing male with an arrow to ensure that they are not confused. Juvenile birds or birds that fledged during the current breeding season should be recorded separately.
A bird flushed within 50 m of a points center as an observer approaches or leaves a point should be counted as being at the point if no other individual is seen during the count period. It is advisable that this be recorded separately.
If a flock is encountered during a census period, it may be followed after the end of the period to determine its composition and size. An observer should follow such a flock for no more than 10 minutes. This is especially useful during the winter. A bird giving an unknown song or call may be tracked down after the count period for confirmation of its identity.
No attracting devices or records should be used, except in counts for specialized groups of birds.
Filling in the Forms
The data taken at point counts are of two types, the location information and the census data. The location data are contained in the first three lines of the Location and Vegetation Form (fig. 15, described below) and contain information about each census point. We also suggest that the vegetation data be taken (see Habitat Assessment, below). We suggest two types of census data forms. One involves mapping and the other direct recording.
Mapping Point CountsThis method of taking data involves the recorder placing on a map (fig. 13) the location of each bird detected (D. Welsh, pers. comm.). We suggest that species codes be used on the map, with a single letter for the most common species, and the full 4-letter code for other species. The birds activities can be recorded by the various mapping symbols given in figure 13. The circle on the map can be the 50-m radius, enabling the observer to keep track of individuals easily. The orientation of the observer (DIR) should be entered on each form by placing the compass direction in the box at the top. Separate time periods are easily kept by using different colored pencils, e.g., birds seen in the first 3 minutes in black, and those seen within 3-5 minutes in red.
Figure 13A recording form for mapping the location of birds during point counts with some mapping symbols. Taken from Welsh (pers. comm.)
The data are then transcribed onto the Point Count Data Form (fig. 14), described below.
Figure 14An example of a data form for recording point count data. Birds are recorded separately within or outside a 50-m circle around the observer, and in the first three minutes or later in the census. The data are recorded as tick marks in each box; then later the actual numbers of birds, as derived from the data, are summarized and recorded.
Figure 15The Location and Vegetation Form. The upper portion should be filled out at all study locations. The bottom portion quantifies vegetation.
Direct Recording Point CountsThis method involves a single-step process of the observer recording the observations directly on the Point Count Data Form (fig. 14). Many observers do not think that it is necessary to map the location of birds in order to keep track of individuals. Using this method, an observer tallies in pencil each individual detected by placing a tick mark (a single line), or another code, in the appropriate column. Codes, for example, can be used to separate out singing vs. visual-only birds (S and V) and age categories. When field work is over, the actual number in each distance and time category can be written in ink for data entry.
The specific data suggested are as follows:
- State or provinceThe 2-column code for each.
- RegionAn 8-column code, designated by the investigator. Often, the name of the USGS quad, a prominent landmark, or a nearby town will provide the best code name.
- StationFor intensive point counts, we suggest a 4-letter code, the same as that used for the mist net array or nest search plot. For extensive point counts, we suggest a code relating to the general area or road. In general, we expect that a given station will have no more than 50 points.
- Month, day, and year.
- Visit numberIndicate how many visits this year will have been made to these points at the end of this days census.
- Point NumberThe 2-column census point number.
- TimeUsing the 24-hour clock.
- SpeciesThe 4-letter species code.
- Tally of individualsThis is a series of five fields. The major subdivisions are those birds detected at less than, and more than, 50 m, and those birds flying over, but not landing within detection of, the observer. Within the two distance categories, observers can separate out those detected in the first three minutes, and the next two minutes. Observers wishing to separate out behavioral, age, or sex categories can note them with an appropriate letter code. Otherwise, tick marks (e.g., 3 = ///) can be used.
Repeating the Count
In general, a station should be sampled only once each season. Counts can be repeated if the goal is good estimates of the community at certain, specific points, such as a small area of rare wetland habitat.
The timing of the census of each route should be kept constant from year to year; it should not differ by more than seven days from the date of the first count. If the phenology of the spring differs, then the date can be changed. The start of the count should not differ by more than 30 minutes from that of the first year. If possible, the same observer should census the route every year.
Strip transects are very similar to point counts, but the observer records all birds seen or heard while traversing each section of a trail. Each section is then the unit of measurement, and can be 100 m or 250 m long. This method is best used in very open terrain where the observer can devote his or her full attention to the birds, and not worry about footing.
In this method the observer should attempt to cover a given amount of trail in a fixed amount of time, e.g., 100 m in ten minutes.
Background and Aims
The area search method has been adopted for a nation-wide survey, the Australian Bird Count (Ambrose 1989), and was chosen over several others because of its appeal to volunteers. It uses a method that, while quantitative, mimics the method that a birder would use while searching for birds in a given area. Essentially this is a series of three 20-minute point counts in which the observer can move around in a somewhat restricted area. In this way unfamiliar calls can be tracked down and quiet birds can be found.
The observer should be reasonably familiar with most (if not all) bird species likely to be encountered at the plot. This method allows the observer to track down unfamiliar birds, but walking the plot before a survey with a person familiar with the birds allows the observer to be more efficient.
Choosing a Plot
The plot should allow relatively easy detection and identification of birds (by sight or calls) and allow the observer to move about freely. The plot should be sufficiently large to provide three separate search areas (or plots), each about 3 ha in forest or dense woodland, but larger areas of 10 ha or more can be used in more open habitats. In very dense forest, smaller areas of 1-2 ha can be used. The search areas can have adjoining boundaries or can be in completely separate regions of the plot. More than three search areas can be established within a plot, but the same search areas must be used on each visit.
Time of Day
Because of the intensive nature of this method, it can be carried out longer into the morning than other methods. However, it should continue no later than five hours after dawn.
Walk throughout the plot for exactly 20 minutes in each search area, stopping or moving to investigate sightings or calls when appropriate. Record numbers of birds of each species seen, heard, or both seen and heard in the search area during this time. Record birds outside the search area separately, but concentrate on finding as many birds as possible within the plot. The observer may find it easier to tape record observations and then transfer results onto paper soon after the survey. An accompanying person can serve as a recorder. A single survey is completed after at least three areas have been searched at a plot.
Filling in the Form
A standard form is suggested, listing the species found and a running tally of the number of birds, both on and off the area. These tallies can be totaled on the right of each area for each species.
Background and Aims
The mapping method is based on the territorial behavior of birds. By marking the locations of observed birds on a detailed map during several visits within a breeding season, it is possible to count the number of territories in an area and estimate the density of birds. Spot mapping is not usually used as a general method for broad-scale monitoring of breeding landbirds, because it requires more time and field work than single-visit point counts and line transects. However, the method should be applied when fairly precise pair numbers and densities as well as the distribution of territories in small study areas or patchy habitats are to be studied. The standard mapping method is less suitable for species that live in colonies or loose groups, or species with large or no territories.
In general, one or two observers make repeated visits (a minimum of 8) to specific plots during the breeding season. Some habitat analysis is also required. Standard methodology as described by Robbins (I.B.C.C. 1970) is also used by The Cornell Laboratory of Ornithologys (CLO) resident bird counts. The latter program, known as the Breeding Bird Census (BBC), is a continent-wide program that welcomes contributors and publishes results of North American plots annually in the Journal of Field Ornithology. The CLO also encourages Winter Bird Population Studies (WBPS) on the same plot. For more information, write to: CLO, Resident Bird Counts, 159 Sapsucker Woods Road, Ithaca, NY 14850; Telephone (607) 254-2441. The basics of the method are contained in Koskimies and Vaisanen (1991). We present here enough information for a biologist to evaluate the technique. The methods of especially data recording, evaluation, and analysis are complex and detailed.
Equipment and Time Needed
One needs 30-40 copies of a very detailed map (preferably 1:2000, or, in open areas, 1:3000 may be acceptable), a compass, and flagging for marking the area.
The time needed depends on the size and terrain of the census area as well as on bird density, with higher densities requiring the mapping of more individuals. Usually about 10-30 hectares in a wooded area or 50-100 ha in an open area may be counted in one morning. Thus, in forest it takes 10 mornings to census 30 ha by the ordinary 10-visit version of the mapping method (about 50-60 hours of field work). In addition, it can take as many as 40 hours (4 hours per census morning) to prepare the species maps, and about 5-10 hours to analyze them. In total, one could spend as many as 100 hours censusing 30 ha of forest during one breeding season. Marking the 50- by 50-m plot in the field takes about 25 hours before the first census season.
Drawing a Map and Marking the Area
The census area should be as round or square as possible in order to minimize border length, because territories along edges are difficult to analyze. After the area has been chosen, a detailed map (known as a visit map) is drawn of it before the first census. The recommended scale for the map is 1:2000. A survey map (1:20,000) and field experience should be used in drawing. Boundaries of the area and landmarks such as edges between habitats, streams, roads, paths, buildings, big rocks, and trees are marked on the map. There should be enough landmarks on the map so that the observer can locate the positions of birds accurately on the map. One copy of the map is needed for each visit, and enough copies should be reserved for making the species maps. If there are only a few natural landmarks, a grid of 50-m squares can be established with the corners of the squares marked with plastic flagging with coordinates written on them.
Census Period and Number of Visits
Because of differences in phenology of arrival and nesting, the visits should cover a period long enough to ensure that each species is easily observable on at least three visits. There should be 10 visits in a standard mapping of forest birds. If the bird density is very high and the nesting period of the community is long, 12 visits are recommended. The visits ought to be evenly distributed over the census period. Fewer visits can suffice in open habitats, where bird densities are usually lower than in forests, or where the season is short (e.g., tundra or alpine grasslands).
Time of Day
The main census time is 5 a.m. to 10 a.m. when the birds sing most actively. After a very cold night counting can be delayed. During very warm weather it should be prolonged because of the lower activity of birds. Two visits should be made in the evening: the first in the beginning of the census period (especially for counting thrushes), the second about two or three weeks later (especially for counting nocturnal singers). If there are several nocturnal or dusk-active species breeding in the area, these two censuses should be added to the ordinary program of 10 morning visits, for a total of 12. In northern temperate zones, owls, woodpeckers, and crossbills breed early and should be censused by extra visits in March and April.
A clean map is reserved for each visit. Each visit should cover the area as evenly as possible, and no place should be farther from the route than 25 m (dense vegetation or high density of birds), 50 m (sparse vegetation, few birds) or 100 m (open habitats). The route you follow through the plot should be on a grid twice the size of the distances above, for example, 50 m in dense habitat. Successive visits should be started at different points, especially if you think that a part of the area is getting attention at the expense of the rest. Simultaneous observations of two individuals of the same species singing or seen must always be recorded carefully so that birds can still be separated from their neighbors after they have moved, which frequently happens during a census visit.
Even while you are busy censusing, you should not walk very slowly, because then, for example, a bird uttering alarm calls may attract other birds to congregate nearby. Therefore, walk with moderate speed and record the birds all the time. Stop frequently to hunt for simultaneous observations of different individuals of the same species, to listen, and to mark the birds on the map. If you are not sure whether there is only one bird or two, you can return to the area censused already to make sure which is the case. In open areas it is often useful to search for the birds with binoculars.
The ordinary speed of censusing is 10-12 min/ha, or 5-6 ha/hour when the bird density is about 300-500 pairs/km2. If the density is very high, the censusing speed slows down to 3-4 ha/hour (15-20 min/ha). When the density is very low or only some of the species are being censused early in the spring, one may walk a little more rapidly; however, at least eight minutes should be allowed for each hectare.
There are many advantages to slow and thorough censusing: (1) one can gather simultaneous observations effectively by following the movements of individual birds in different parts of their territories; (2) one can pay special attention to species difficult to detect; and (3) one can search for nests and check those found earlier. All observations are marked on a map using standard codes which are given in the detailed instructions in Koskimies and Vaisanen (1991). All observations are transferred from the field maps to exactly the same locations on the species-specific maps. There should be a separate map for each species.