3.5 Absolute Abundance

Recommended methods: Spot mapping or Distance-based Point Counts.

Two methods are recommended to obtain absolute abundance estimates for songbirds. Each method will yield different types of information and so the selection of one method over the other will depend on a project's objectives.

Spot mapping is generally a valuable way to collect detailed biological information on a specific area, providing a map of songbird territories as well as a somewhat informal estimate of density. Comparisons of different spot maps can be a useful, if somewhat qualitative way, to compare songbird abundance in different areas. In contrast, if survey objectives require a more rigorous estimate of actual songbird density, project biologists may wish to consider distance-based point counts. Although the application of a distance method and its subsequent analysis may be challenging, repeatedly mapping bird territories in a spot mapping exercise is also difficult and time consuming. It can be argued that spot mapping is best if a project is focused on breeding birds since it may be harder to detect the breeding portion of the population using distance-based methods which are more constrained in terms of time and area surveyed. However, distance methods may be a better point estimator of overall bird density. In addition, density estimates from a distance-based point count may be as good or better for breeding bird census if it is possible to discern breeders and non-breeders efficiently.

3.5.1 Spot Mapping

Many songbirds are territorial, and defend their territory through conspicuous displays of singing, calling, and territorial defence. Knowledge of this behaviour has given researchers the opportunity to study territory size and through this, obtain population estimates. Spot mapping requires repeated surveys of measured plots, where each bird detected within the plot is registered on a base map. The data recorded include exact location within the plot, species, sex, behaviour, and number. After each survey, registrations are transferred to a separate summary map for each species, and clusters of registrations are identified that are assumed to represent individual territories. Each territory is assumed to contain one breeding pair. For a more detailed description of spot mapping see Bibby et al. (1992).

Spot mapping is similar to a total count of breeding individuals in a population. This is not a total count of all individuals since non-breeders or birds without territories can rarely be counted. Therefore, the results of spot mapping should be given as absolute abundance of breeding birds or birds with territories as opposed to the total bird population. In addition, spot mapping counts will be an unbiased and accurate estimate of absolute abundance of breeding birds only if all birds are sighted during a survey effort. This assumption might be violated if birds are sitting tight on the nest, or if heavy foliage influences sightability at some sites. In addition, spot mapping estimates may be biased if the number of breeding birds in areas changes during repeated visits to the area. For example, if floater birds establish territories in the course of spot mapping efforts than the estimate of breeding bird density might be positively biased. In addition, like many field methods which attempt to estimate density, the spot count method is susceptible to edge effects in which edge birds on the periphery of the study are counted therefore inflating estimates further. The probability of overestimate is positively correlated with the overall time in which spot mapping methods are conducted.

Office procedures

• Review the introductory manual No.1, Species Inventory Fundamentals.
• Obtain maps for Project and Study Area(s) (e.g., 1:5,000 air photo maps, 1:20 000 forest cover maps, 1:20 000 TRIM maps, 1:50 000 NTS topographic maps). Any map which is used to record data should be referenced to NAD83.
• Outline the Project Area on a small to large scale map (1:250,000 - 1:20,000).
• Determine Biogeoclimatic zones and subzones, Ecoregion, Ecosection, and Broad Ecosystem Units for the Project area from maps.
• Delineate one to many Study Areas within this Project Area. Study Areas will likely be defined as the portions of the Project Area in which the spot mapping takes place. Study areas should be representative of the Project Area if conclusions are to be made about the Project Area.
• Based on the maps and other knowledge of the Study Area (previous reports, local resource specialists) identify strata which are of most interest for surveying.
• Compile a list which includes all potential songbirds for the Study Area.
• Once spot mapping plots have been selected with strata of interest (see Sampling Design):
• Prepare data collection `base' maps for each plot on which detections will be registered. One map per plot per visit is required.
• Prepare one map per species per plot to serve as a `cumulative' map of detections at each plot.

Sampling design

• Location of Study Areas should be chosen randomly.
• Develop your sample frame based on the objectives of the study (e.g., bird community in riparian areas) and then randomly select the Study Area(s) from this sample frame.
• Within the Study Area stratify the habitat if necessary by site series. Use the Ecosystem Field Form to capture this data.
• Identify spot mapping plots in areas of homogeneous habitat.
• Mark gridlines for plots on maps according to protocol outlined in Establishing a spot mapping plot in the field under the Field Procedures section.
• When possible, randomize or change observers between plots to minimize recurring bias in any segment of a survey. The effect of individual observer bias could be pronounced if some observers are better at identifying and mapping territories than others.

Sampling effort

• The Study Area must be resampled six to eight times throughout the field season, with sampling periods being separated by 7-10 days. Be aware that this long sampling period can potentially inflate density estimates if birds change territories before surveying is complete.
• A 20 ha grassland plot requires about two hours to survey. A 20 ha forest plot requires about three hours to survey. Duration of surveys should be consistent within and among plots.
• The optimal sampling effort will depend on how many birds change territories and would have to be determined adaptively.

Personnel

• The crew leader should be a qualified biologist with experience in spot mapping survey method.

Equipment

• Maps
• GPS (use NAD83)
• Compass
• Hip chain (Remember to gather all the string between stations; failure to do so can result in the death of many birds)
• Flags or stakes
• Data forms or tape recorder
• Clipboard
• Digital watch
• Thermometer
• Binoculars
• Field guide(s) for species confirmation
• Bird song tapes for training

Field procedures

Establishing a spot mapping plot in the field

• Locate plot(s) in areas of homogeneous habitat.
• Plots should be 20 ha in size and rectangular or square in shape to facilitate gridding. Smaller patches of habitat down to 10 ha can be mapped, but several biases may begin to influence results and should be accounted for. Larger plots can be mapped as well, but plots over 30 ha often cannot be adequately surveyed because of time constraints, and are not recommended.
• Multiple plots should be of similar area and spaced so that there is little chance of individual territories overlapping between plots.
• Plot borders should be >50 m from the edge of a different habitat type to reduce the likelihood of residual edge effects.
• Mark edges of plots with flags or stakes.
• Mark gridlines, 25 m apart in forested habitats, and 50 m apart in more open habitats, with flags or stakes along one axis of the plot.

Surveying at a spot mapping plot

• Note: surveys must be completed within four hours after sunrise.
• Select a gridline adjacent to a plot boundary and a direction of travel along the gridline.
• Walk the series of gridlines in an 'S' pattern until all gridlines have been traversed. On each successive visit, reverse the starting point and direction the plot is surveyed.
• Walk at a rate of 0.5 km/hr, stopping when appropriate to record data, identify birds, or listen.
• Record plot data on a `base' map (one map per plot per visit is required).
• Register on the base map all birds when they are first detected, including species, number, and behaviour. Use codes for notes (e.g. , B-SASP (Savannah Sparrow) SM (singing male), but mark location as precisely as possible with an X. Record movement (if any) of the bird(s) with arrows to help delineate territory or home range boundaries. A sample of a spot-map and a system of symbols is given in Figure 4.
• Do not register an individual more than once unless that factor is noted.
• When a nest of interest is located refer to the Nest Site Description Form to determine the data attributes to be collected.
• If there are multiple plots to survey during a day, reverse the order the plots are surveyed each day. Two plots per day per person should be a maximum effort to minimize the effects of time of day. Generally this will not be practical except in open grassland habitats.
• After each visit to a plot, add data from the `base' map used for that day to the appropriate species `cumulative' map for each plot. Use as many cumulative maps as required, particularly if there are many species and individuals. Ensure that registrations for any given species occur on one `cumulative' map only. Also ensure that summary information has been recorded onto the Songbird Spot Mapping Summary Form.

Data Entry

• The Design Component for this survey is a Plot. Currently Spot Mapping data cannot be entered into the species inventory (SPI) provincial database. However, Wildlife Features (such as nests) and Incidental Observations of note (such as a rare bird) can be included.

Figure 4. Sample spot-map and associated symbols for delineating bird territories during the breeding season (from Bibby et al. 1992, Arbib 1970*).

Data analysis

• Determination of territories and number of territories
• Ensure that all information that has been recorded onto a `base' map from each visit to each plot has been transferred onto the appropriate species `cumulative' maps for each plot. Also ensure that summary information has been recorded onto the Songbird Spot Mapping Summary Form.
• After the survey program is completed, circle clusters of registrations for the same species that are considered to represent one territorial individual, based on the map data and your field observations.
• Clusters with two or three registrations (depending on the number of valid visits) can be considered to represent one territorial male.
• For territories that occur along the edges of the plot, estimate the proportion of the territory enclosed within the plot.
• Analysis
• Calculate abundance of birds per unit area by adding the total number of territories per species of territorial males within the plot (include whole and fractions of territories), multiplying by two (which assumes that each male has one female associated with it) and divide by the area of the plot. This assumption may not be valid for some species. This calculation assumes that all territories occur within the plot area. Edge territories which border on the plot area will be overcounted. It may be possible to partially adjust for this problem by approximating the percentage of a birds territory in the plot area and adjusting the count accordingly. For example, if half of a sparrow's territory is in the plot area then the actual count of the bird would be one half, or the count of a pair of birds would be one as opposed to two birds. Note that the degree of bias with uncorrected counts will be dependent on sampling area size, and the size of individual species territories.
• To obtain a mean abundance from multiple plots, sum abundance estimates for each species in each plot within a homogeneous habitat type and divide by the number of plots.
• For statistical purposes, the abundance per species per area for one plot represents only a single data point, regardless of the number of registrations.
• The statistical analysis of spot mapping data will be identical to the analysis of count data for relative abundance methods.
• Note: Statistical comparisons can only be made when multiple plots are sampled.

3.5.2 Distance-based Point Counts

Distance-based point counts may be a viable alternative to labor intensive territory mapping in some situations. Point count distance methods allow an estimate of absolute density from point counts. They require that the distance from the bird to the observer at the center of the point count be accurately recorded, which should generally be less time consuming then recording coordinates for spot mapping. The advantage of point count distance methods are:

• Point counts are much less labor intensive than spot mapping, and therefore will allow a greater number of replicates to be collected per unit time.
• Breeding and non-breeding birds can both be counted which may provide a less biased overall population density estimate.
• The decrease in sightability as distance of bird from observer increases can be accounted for statistically using the methods in program DISTANCE.

The disadvantage is the complexity of the analysis, which can be overcome if a statistician is consulted for both the design of the study and for the analysis of data.

The use of point counts for relative abundance is similar to the use of a strip transect for transect methods. In some limited situations, it is questionable whether point count methods can be extended to allow estimates of density using distance methods. Distance methods allow one to estimate the probability of detection, rather than assuming it is constant throughout the surveys. The major difference is that the actual distance of birds from the center of the point count must be exactly recorded (Buckland et al. 1993). Point count distance methods have the following assumptions:

Objects on the point are detected with certainty. This assumption is relaxed if the methods introduced in program DISTANCE are used. The actual shape of the detection function is flexible. Many times, the number of species is low closest to the observer and increases as the distance from the observer gets greater, then decreases as the detectability of species diminishes with distance.

Objects are detected in their initial location. This assumption will generally be met as long as the observer does not flush birds or they are not attracted to the observer. Buckland et al. (1993) introduces various "fleeing" models to also relax this assumption. However, the utmost care should be given to minimizing disturbance of birds prior to surveys. Models which incorporate "fleeing" behavior are more complex than simpler models and therefore will require larger sample sizes.

Measurements are exact. This may be a difficult assumption to meet in some brushy habitats. Buckland et al. (1993) suggest marking and measuring the distances to the bird's location after the observation interval of the point count is finished. Note that distance of birds from the objects center can also be marked in groups (i.e. 0-1 m, 1-2 m, etc.) using the concentric rings approach introduced in the relative abundance section. The actual intervals should be tightest towards the center of the point count and get larger as the distance from center increases. See Buckland et al. (1993) for further discussion on this technique.

A common reason for not attempting to use distance methods is sample size constraints. (Buckland et al. 1993) states that most distance models need at least 60 observations to ensure adequate precision. However, if the sightability of birds is high and factors affecting detection rate are not complex (due to proper sampling design), it is possible to obtain adequate results with less than 60 observations (S. Buckland, Mathematical Institute, Univ. St. Andrews, Scotland, pers. comm.).

Program DISTANCE can be used for analyzing data from distance-based inventory methods. The program is quite powerful, and employs parsimonious and robust modeling techniques to allow density estimates from point count data. However, it also has restrictive sampling assumptions and is statistically complex. It is strongly recommended that a statistician trained in distance methods be contacted if this method is considered. See manual No.1 (Appendix G) for more details on program DISTANCE.

Procedures

* See the Relative Abundance section for complete protocol. Only differences in procedures from those found in the Relative Abundance section will be given below.

Sampling design

• The major constraint to this method is sample size. Buckland et al. (1993) states that at least 60 animals should be sighted for the use of distance models in program DISTANCE. However, as noted earlier, if the study is designed properly, and factors determining bird detection are minimized, it is possible to obtain reliable estimates with less than 60 observations (S. Buckland, pers. comm.). If extra parameters are added to models to account for "fleeing" then this number will have to be larger. In addition, if birds occur in groups or flocks, additional modeling considerations will have to be employed.
• The best way to determine the applicability of this method is to conduct a well-designed pilot study as recommended by Buckland et al. (1993). The data from this study can then be used to determine optimal field efforts and sample sizes.
• Another method to determine the applicability of this method to a particular population is to suggest the proposed study to the DISTANCE Discussion Group, made up of bird researchers using DISTANCE throughout the world. This is an excellent way to get an initial approximation of field efforts needed to employ this method. The DISTANCE Discussion Group can be accessed via the DISTANCE web page. See the Species Inventory Fundamentals manual (Appendix G) for details.

Field procedures

• Record birds at their initial location. Record the exact distance from the center of the Point Count Station to the bird (this may be a difficult assumption to meet in some brushy habitat, especially when birds are audible but not visible). Note that distance of birds from the objects center can also be records using distance classes (i.e., 0-1 m, 1-2 m, etc.) based on concentric rings around the observer (introduced in Relative Abundance section). The actual intervals of the rings should be tightest towards the center of the point count, getting larger as the distance from center increases. See Buckland et al. (1993) for further discussion on this technique.

Data analysis

• Use the program DISTANCE for analysis. Point counts should provide better statistical data then spot mapping because detection probabilities can be estimated and more points can be sampled per unit effort. See manual No.1, Appendix G, for more details on the use of DISTANCE.