About line transect sampling of grouse and forest birds


In august every year, a number of larger and smaller stakeholders conduct line transect sampling with standing bird dogs. These sampling efforts generate a knowledge-based foundation for the small game management, and will in the future be important to highlight trends in bird populations. Furthermore, the data will be an important part of research on how gallinaceous birds, and specifically willow ptarmigan, are expected to respond to climate change and other environmental changes. Here, you will find a description of the important parts of line transect sampling. To read more about how the work is organized, look here.

Methods

Monitoring of gallinaceous birds is based on Distance sampling. This method is widely used to estimate abundance and density of several species, and several publications detailing its methodological assumptions and applicability are available. The method is well suited to estimate density of grouse and forest birds based on line transect sampling, and has been thoroughly evaluated in multiple research projects. However, the results will always be an estimate, and never a precise measure of the density in a defined area. The precision (uncertainty) and accuracy (the "true" density) of the estimates depends on the quality of the study design, field sampling and the data.

Read more ... Read less ...

For the methods we use here, the following assumptions apply:

  • The geographical area sampled is representative for the whole area for which the estimates are calculated
  • Line transects are placed correctly, and that there are no deviations from the line when the line transect is sampled
  • A minimum of 40-60 observations within the defined sampling area (see e.g. Buckland et al. 1993). Based on experience, a minimum total sampling length of 70 km per defined area is required to observe 40-60 willow ptarmigan.

It is important to note that the Distance sampling method does not require the standing bird dogs to find all the birds within a defined area to estimate density (and corresponding measures of uncertainty). Field personnel walks a predefined virtual line (defined in the GPS/app) with a standing bird dog. They count the observed clutches and measure the distance from the line to where the birds were observed.

The principle for line transect sampling with standing bird dogs. It is imperative that the perpendicular distances from the line to the observations are measured accurately.

Several assumptions must be met for precise density estimates:

  1. We find all the birds located exactly at the line transect,
  2. The birds are observed where they were when the dog/personnel flushed them,
  3. The distance from the line to the point where the birds were flushed is measured accurately.

The two first assumptions have been met through testing on radio-collared willow ptarmigan. We strive to meet the third assumption by walking a straight line in the landscape and measuring the distance perpendicular from the line to the observation. Assumption three is met by accurate measurements of the distance.

Selecting representative sampling areas

Within the scope of the Hønsefuglportalen, it is not realistic to aim for a completely random selection of sampling areas within a region or part of the country. Selecting proper sampling areas is therefore partly subjective. Still, some general rules of thumb applies. The sampling area should be large enough to cover both rich and poor habitats. Otherwise, the density estimate may not accurately depict the real density of the area. It is the stakeholders responsibility to select a representative sampling area. Stakeholders are encouraged to contact the R&D institutions (NINA or INN) for guidance when selecting a representative sampling area. If the selected sampling areas are placed within rich ptarmigan habitats, the estimates may be representative for the sampling area, but not the general management area.

The aim of the line transect sampling is to give an adequately accurate estimate of the number of birds within the defined area. The total number of line transects should therefore represent an average of the area's habitat quality. Within larger areas, it might be necessary to distribute the lines so that they represent the habitat of the whole area. Ideally, we could distribute the lines systematically within a grid to achieve complete coverage of the area. However, this is rarely achievable in practice within larger areas. Lines should therefore be distributed according to manpower (number of lines vs. field personnel and logistics) while making sure that the line transects represent the area as a whole. In practice, the line transects are often distributed within several sub-areas, which in total is a good representation of the management area.

If you want to perform line transect sampling within a small area, we encourage you to collaborate with neighbouring land owners to generate an area large enough to provide informative data. This is also a good approach to ensure enough line transects and to achieve a minimum total sampling length, as well as getting enough field personnel. Increased sampling length will probably yield more bird observations, which subsequently provides a more precise density estimate.

If you want density estimates representative for large areas (e.g. municipality, county), it is often necessary to sample smaller sub-areas representative for the larger area. If the line transect distribution is not representative for the larger area, the estimate can be skewed (i.e., under- or overestimated) in relation to the "true" population density. It is therefore important that the sampling effort within each sub-area is identical and is sampled at the same time. The sub-area should cover both rich and poor habitats to ensure a density estimate which is representative for the whole area.

Distribution of line transects

There are different approaches to distributing line transects, for example systematic or random placement of lines. The optimal solution is to distribute lines randomly within a defined area, as this will prevent a systematic preference for rich habitats. However, random distribution is challenging due to topography (e.g., line transects crossing water or screes) and logistics (e.g., far from roads). In practice, we most often utilize a semi-systematic approach, where line transects are placed with minimum 500 meters apart, and where some lines are subsequently adjusted to account for topography and logistics. We do this since the methodology assumes independent observations, i.e., that the same bird is only observed on one line transect. To ensure a reasonable number of observations, to include variations in habitat quality, and to account for logistical challenges, we recommend a line transect length of minimum 3-4 km.

Systematically placed, parallell, and straight lines across the isoclines yields the best result. However, in areas with large variations in topography, this design usually needs to be modified. The distance between lines should not be less than 400 meters, and preferably 500 meters or more. If the topography allows, line transects can be placed along the UTM grid in the M711 series of maps (1:50000; for Norway), following the north-south or east-west directions. This approach will also be beneficial when using a GPS/app while tracking the lines.

Performing line transect sampling in the field

Most of the line transect sampling is carried out by volunteer field personnel, based on an agreement between the stakeholders organising the work. Most of the field personnel are organized in local departments of the Norwegian Association of Hunters and Anglers (NJFF) or local/regional bird dog associations. Most field personnel are qualified for the sampling work through Distance sampling courses, although this is currently not a requirement. Currently, a revised line transect sampling course, is being developed, although it is currently undetermined if the course should become mandatory for field personnel in the future.

Read more ... Read less ...

A field team consists of two people; a line transect/field sampler and a dog handler, along with one or more dogs. A field team should be able to sample up to 8-10 km per day in regular mountain terrain. If the terrain is rugged, steep or consisting of bogs/peat, the sampling length should be reduced. For practical reasons, the end point of the sampling should be as close to the starting point as possible. A good solution is therefore to place line transects with a length of maximum 4-5 km, so that multiple field teams can operate in parallell at separate lines. Only one dog should be active at a time. Use a map/compass or GPS/app (preferably) to follow the virtual line. The precision of the field sampling is dependent on correct observation of the number of birds flushed, and that the distance perpendicular from the line to the observation is measures accurately.

Field personnel

The quality of the density estimates depends on the performance during field sampling. Field personnel should receive some form of training prior to line transect sampling, and must be shown how the final estimates depend on accurate work. The intensity of search along the line, particularly in the proximity to the line, must be even and without stops. The dog handler must not be too fast and thus stress the dog so that it skips areas which have not been thoroughly searched.

As the precision of the density estimates depends on the quality of the field sampling, field personnel must be adequate at identifying bird species and sub-species (e.g. willow and rock ptarmigan), and accurately determine their age (juvenile and adult). Similarly, field personnel must be able to identify species of forest birds, and determine their age, if the sampling is carried out in the forest. Other requirements are adequate knowledge on how to use a map, compass and GPS. Field personnel must be informed of the most important aspects of the work pertaining to high quality data. There are limits and regulations to the field sampling; e.g., dispensation from the dog leash law (in Norway) allows free movement of the dogs along the line transects, but is not a general dispensation for the whole area.

Quality requirements of dogs

Line transect sampling can be carried out when a dog leash dispensation has been granted, and when the dogs have been tested for disease and confirmed to be weaned off sheep. The dog is our tool to locate birds, and must therefore have displayed an affinity for locating birds prior to field sampling. Field sampling is conducted within a period of the year where chicks are small and vulnerable. Out of consideration for the birds, the dog must be properly trained and be able to stop its actions in any situation. Therefore, only properly trained and experienced dogs should be used for this activity.

A good standing bird dog searches an area of 150-200 meters to each side of the line transect, and is controllable in its actions. It is particularly important that the area close to the line is searched thoroughly, and that the dog cooperates well with its handler. In total, the quality of the final density estimates depends on the quality of the dog, the field personnel, and the conducted field sampling. Field personnel should receive training, and inexperienced dogs should not be used during field sampling.

Statistical analysis and reports to stakeholders

When sampling is concluded within an area, the data will be analyzed and a report with the results will be sent to the stakeholder. The work on the analysis and report is currently divided between researchers at NINA and INN. The report summarizing the results is focused on three parameters:

  1. Density (birds per km2)
  2. Adult density (birds per km2)
  3. Chick production (number of chicks per pair for ptarmigan, and number of chick per hen for forest birds)

As the precision of the density estimates depends on the quality of the field sampling, field personnel must be adequate at identifying bird species and sub-species (e.g. willow and rock ptarmigan), and accurately determine their age (juvenile and adult). Similarly, field personnel must be able to identify species of forest birds, and determine their age, if the sampling is carried out in the forest. Other requirements are adequate knowledge on how to use a map, compass and GPS. Field personnel must be informed of the most important aspects of the work pertaining to high quality data. There are limits and regulations to the field sampling; e.g., dispensation from the dog leash law (in Norway) allows free movement of the dogs along the line transects, but is not a general dispensation for the whole area.

Read more ... Read less ...

We have previously described the assumptions for line transect sampling. For density estimation based on Distance sampling we include three parameters:

  1. Length of the transect line
  2. Perpendicular distance from the line to each observation
  3. Clutch size per observation

When all distances from the transect line to the bird observations are measured precisely and without systematical errors, we can create a function to describe how the probability of detecting birds declines with increasing distance from the transect line. When transect lines are randomly distributed in the landscape, there should be an even distribution in the number of birds detected close to, and far from, the line. In principle, all birds located directly on the line should be detected, whereas no birds should be detected further from the line than the dog is searching. The figure below/on the left illustrates the concept of Distance sampling.

When distances to the transect line are measured correctly, and there has been an adequate number of bird observations, we expect that the number of observations decline with increasing distance to the transect line. This information is used to calculate a detection function, which is a central concept in the estimation of the number of birds per square kilometer.