Sampling Design

• Non-random. Set-up banding sites in areas that will ensure that the banded sample is representative of the population of interest. If assumptions are met this will allow estimate of population size for a defined area. However, the estimation of standard error will be inaccurate.

Sampling Effort

• Band a very large sample of (non-breeding) birds and band different age and sex classes in proportions that allow for comparisons of their recovery rates (see below).
• Distribute banding effort among several sites.
• Sample the population repeatedly over short time intervals (the assumption is that sampling is instantaneous).
• Rapid sampling can be done by organizing a large group of trained volunteers to resight bands over an extensive area on the same day.
• Plan to band in the same areas consistently over several years.

Time of survey

• Banding:
• For inventories of localized wintering populations, band after migration is complete and wintering numbers and distributions have stabilized (between mid November and February in the parts of the province where birds winter).
• For inventories of resident populations of Canada geese, band flightless moulters and young (usually early June, but may vary locally).
• Resightings:
• Resightings of wintering populations should occur between mid November and February. Resightings of resident geese can occur between June and February.
• The location and timing of resightings may depend on local diurnal, tidal, or other bird use patterns.

Sample sizes

It is difficult to recommend sample sizes (numbers of observations or animals sampled) which are appropriate for every situation, as this will depend on the level of precision needed. Of course, the larger the sample size is, the more precise the abundance estimates will be. In general, sample sizes can be increased by increasing sampling effort. However, there are always limits to the amount of sampling effort that can be afforded during a study. The solution then becomes a compromise between available resources and levels of precision required to meet the objectives of the study. The simulation modules provided in CAPTURE and NOREMARK can be very helpful in determining the sampling effort needed to get adequate estimates. In addition, Pollock et al. (1990) provide sample size tables for the Jolly- Seber model.
It must be remembered that sampling effort will not remove consistent sampling errors or violations of the assumptions of the model, for example, when a non-random sample is recaptured/resighted.

The "robust" study design of Pollock et al. (1990) is recommended if density estimates, and survival, and other demographic rates are an objective of inventory efforts. With this design, a series of five-day samples (capture session) are conducted at equal intervals (e.g. every month) during the time period of interest. The data from the five day sessions is used to estimate density using program CAPTURE. (See section on data analysis, and Species Inventory Fundamentals, No. 1, Appendix G for more details on program CAPTURE.) In addition, these data are pooled and used with the Jolly-Seber model to estimate survival and other demographic parameters (using JOLLY or JOLLYAGE). This design has the following advantages:

• Theoretically robust estimates of population size and survival are possible.
• Temporary emigration from the study area can be estimated from the data set allowing for further demographic inference, and less biased survival estimates if a subset of the population is not available for resampling in a given period. A new program, RDSURVIV, has been designed for this purpose when the robust design is used (see Species Inventory Fundamentals, No. 1, Appendix G).
• The data should also allow further demographic inference and model fitting of survival rates using programs MARK, SURGE, and POPAN (see Species Inventory Fundamentals, No. 1, Appendix G).

Methods are available for biologists to determine appropriate sample sizes for the various mark-recapture estimators. It is recommended that project biologists consult the following sources for sample size calculations (Table 3).

Table 8. Sources for sample size calculation.

Estimator	Source for optimal sample size calculation:
Lincoln-Peterson estimator.	Krebs (1989 page 22)
Jolly Seber estimates	Pollock et al. (1990, page 72) Simulation: POPAN
CAPTURE	White et al. (1982) Simulation: CAPTURE

The above references include graphs, and discussions of needed sample sizes for estimators. The determination of optimal sample sizes for program CAPTURE is complex. An easy to use simulation module is available as part of program CAPTURE to allow biologists to explore sample size issues.

Personnel

• Two or three experienced biologists and a large crew of volunteers for banding.
• Hunters to provide band recoveries where the resighting method is not used.
• A large crew of trained volunteers to resight bands, neck collars, nasal tags, etc.

Equipment

• Boats to corral moulting birds for banding
• Boots/chest-waders
• Capture equipment: Herring nets, aluminium rods, lead line rigged up for drive-trapping
• Catch boxes and/or capture pen
• Banding tools, e.g., pliers
• Special markers (e.g., plastic leg bands, patagial tags, neck collars, nasal saddles, radio transmitters)
• Binoculars and spotting scope with tripod for resighting bands.

Field Procedures

• Avoid poor weather when capturing birds.

Capture and Marking

• Capture and band the birds (See Protocol at the end of this section for the various capture methods).
• Band a very large sample of birds and band different age and sex classes in proportions that allow for comparisons of their recovery rates.
• Take precautions to reduce mortality in captive sample, i.e., do not let birds suffer from hypothermia or heat prostration.
• Record all relevant information about the bird (e.g., age, sex, status, location) on banding schedules.
• Record the band numbers and/or other marking ID of individuals banded during previous banding efforts

Recapture/resight

• Record the number of marked individuals and unmarked individuals in recaptured or resighting samples.
• Record locations of marked and unmarked individuals.
• Sample the population repeatedly over short time intervals (the assumption is that sampling is instantaneous).
• Rapid sampling can be done by organizing a large group of trained volunteers to resight bands over an extensive area on the same day.