In the North American bird banding program, recaptures (including resightings) constitute the vast majority of post-release data from banded or otherwise marked birds. A powerful suite of analytical models and software is available for analyzing recapture data to estimate parameters such as survival, population size, rate of population change, movement probabilities, and age-specific breeding probabilities. Many banders deliberately collect and analyze recapture data for their project-specific needs. Others cumulatively collect millions of recapture data suitable for use in mark-recapture models but do not use them. The potential use of these data in population research and management is great. Actual use is limited, however, because recapture data currently are not maintained in a central repository. To remedy this, the US Bird Banding Laboratory and the Canadian Bird Banding Office are developing a bird banding recapture database.
Recapture data are not collected uniformly, and their "value" to an analysis can vary depending on questions asked and models used to analyze them. Most banders record basic data such as date and location of recapture. Others record additional data such as weight or breeding condition that add value and enable additional analyses with covariates. Banders also record recaptures with varying frequency and effort, ranging from only one per year to multiple recaptures on a single day. To allow maximum possibilities for analyses, all recapture data from select projects where data collection procedures can be adequately described, and banders are willing to contribute, will be included in the database.
Because all recapture cannot be treated equally, the potential analyst will need to know the circumstances under which the data were collected in order to pose relevant questions and select appropriate models. Thus, recapture data will be stored as project-specific sets accompanied by data set profiles that include detailed temporal, spatial and operational information about the collection of the data (metadata). Given that most data sets will have a unique combination of descriptors, the profiles will be indexed to facilitate searches for sets that meet the analyst's criteria. Profiles for historical (project completed) data sets will be developed at the time they are added to the database. Profiles for continuing projects will be updated periodically as data are added to the set. In most cases, contributing banders will be able to develop their Project/data set profiles on line.
Individual recapture records will contain standard, fixed fields for universally collected data (e.g., capture date and location); standard fields for frequently collected data (e.g., weight); and user-defined, variable fields for project-specific data (e.g., bird mated/unmated, disease present/absent). Most importantly, recapture records will include a project code to link the record with project profile data.
The variable format of the recapture record and its link to project profiles requires a relational database. For this and other reasons, the BBL computer system is being re-engineered to accommodate a relational database on the client/server model. Initially BBL will act as a clearinghouse for metadata on recapture data available from banders and other sources. Several recapture databases with accompanying metadata sets have already been developed. Once its new computer system is up, BBL will begin serving actual recapture data.
BANDMANAGER, software used by banders to manage their banding data and submit them electronically to banding offices, has been pre-programmed to allow banders to store recapture data. Banders eventually will be able to submit recapture data along with banding data to the banding offices via disk or Internet FTP. BANDMANAGER also contains edits that allow the bander to detect and correct erroneous data before they are submitted to the banding offices. The recapture database will be accessible via Internet and banding office web sites. Data analysts will be able to search the database and download data sets. Links to analytical literature and software will be provided. Data release policies and guidelines will be developed to acknowledge banders' interests and to protect threatened, endangered or otherwise sensitive species.
Historical and current data sets are being sought for testing and inclusion in the recapture database. To date, recaptures from the US/Canadian pre-hunting season duck banding program have been obtained and analyzed to estimate and test hypotheses concerning fidelity of Mallards (Anas platyrhynchos) to breeding areas. Recaptures of Red-tailed Tropicbirds (Phaethon rubricauda) banded in the Pacific are being analyzed to estimate survival, movement rates, and age-specific breeding probabilities. Approximately 100,000 recaptures of Laysan (Phoebastria immutablis) and Black-footed Albatrosses (Phoebastria nigripes) banded in the 1950s and 1960s have been obtained and should provide valuable reference material relevant to current issues surrounding the conservation of these species. Unfortunately, like many older data sets, the Albatross recaptures exist as paper records that must be computerized before they can be included in the new recapture database.
The potential for obtaining additional recapture data sets is great. Data sets from large-scale cooperative banding projects could be included, e.g., the millions of resighting records from goose neck collaring programs, recaptures from the Monitoring Avian Productivity and Survivorship (MAPS) program, and recaptures and resightings of birds marked in threatened and endangered species restoration programs. Individual banders have the largest variety and number of data sets that could be included in the recapture database. A recent survey of US banders found one bander having 40 years of Chipping Sparrow (Spizella passerina) recaptures, and another with more than 100,000 recaptures of Cliff Swallows (Hirundo pyrrhonota). The list goes on and far exceeds expectations.
Results from the same survey also revealed that, while there is popular interest in developing a recapture database, many individual banders and custodians of programmatic data sets have proprietary-based reservations about contributing their particular data sets. Since banders will not be required to contribute recapture data, incentives must be developed. One possibility would be to offer banders assistance in obtaining analytical help in exchange for contributing their data sets. The recapture database will take a long time to develop, and considerable negotiation for data sets will be required. An outreach program promoting the idea of a national recapture database and soliciting data sets will be developed.
The recapture database will provide a permanent archive for data sets that otherwise might be lost. Some of these data sets are unique, and represent individual banders' lifetime work with a species. Data sets will be useful for testing new analytical models, and for comparative studies where the analyst has his or her own data, but would like to compare them with others'. Graduate students and other researchers operating within short timeframes would have easy access to long-term data. More banders will be encouraged to use contemporary mark-recapture models to analyze their data. Eventually, the database will contain millions of recapture and resighting records available publicly for a variety of avian research and management endeavors important to bird conservation.
While the initial focus of database development is to capture historic and current data, plans are to evaluate the desirability and feasibility of developing a prescribed recapture program. Banders would be enlisted to routinely recapture and report banded birds, similar to what is done in the British Trust for Ornithology's Recapturing Adults for Survivorship program. Governmental and nongovernmental partners in the North American bird banding program will be consulted.
Since its beginnings about 100 years ago, bird banding has been used to answer questions about the survival of birds. Early investigators focused on "longevity" or age records of individual birds. Later investigators were more interested in "population dynamics", and they commonly used life tables to gain insight into the survival of cohorts or populations of birds. Life tables, however, have a poor foundation in statistical theory. By the early 1970s, ornithologists and wildlife managers were abandoning life tables in favor of statistically sound models developed in the mid-1960s by Richard Cormack, George Jolly and George Seber. Those models, still referred to as the C-J-S models, revolutionized the analysis of bird banding data.
The revolution continues, and today we have a suite of powerful, versatile mark-recapture models that go far beyond survival estimation. These models offer something for most everyone who conducts banding studies. By using them and their accompanying software with appropriate data, one can estimate population size, rate of population change, movement probabilities, age-specific breeding probabilities, and other parameters.
The development of these mark-recapture models represents a major evolutionary step in the history of bird banding, the benefits of which are becoming apparent in many areas. Journal articles are reporting excellent results from studies employing the models. The models feature in the adaptive management approach being used today by state and federal agencies to set waterfowl hunting regulations. The models are used in the MAPS program to analyze data gathered annually from 500 constant-effort banding stations.
The potential for expanded use of mark-recapture models is enormous. Hundreds of banders have thousands, if not millions, of recapture and resighting data suitable for use in these models. More banders could and should use contemporary mark-recapture models to analyze their data, because the models promote good science, and good science is important to bird conservation.
Herein we provide a guide to mark-recapture literature and software with hopes that the possibilities these models afford will become apparent to more banders. We hope that veteran banders who have amassed data will apply these models to them, and that banders who are just starting their research careers will consider these models when they design their studies.
The mark-recapture literature is quite extensive. The newcomer may find several introductory works to be helpful, starting with George Seber's classic and its subsequent updates. Some of these works contain good material on study design.
Williams, B.K., M.J. Conroy and J.D. Nichols. Analysis and management of animal populations. Academic Press. San Diego, CA. in press.
There are two subsets: band-recovery models and capture-recapture models. Band-recovery models use recoveries of dead birds to estimate survival and reporting rates. Capture-recapture models are dependent on live recaptures of animals. Capture-recapture models have undergone the most extensive development recently, with many variations being developed.
Brownie, C., D. R. Anderson, K. P. Burnham, and D. S. Robson. 1985. Statistical inference from band recovery data -- a handbook, 2nd ed. U. S. Fish and Wildlife. Service. Publication 156, Washington, D. C. 305pp.
J-D. Lebreton and Ph. M. North (eds.).1993. Marked Individuals in the Study of Bird Populations. Birkhauser Verlag. Berlin. 397 pgs.
P.M. North and J.D. Nichols. (eds.) 1995. Statistics and ornithology. Journal of Applied Statistics 22:553-1081.
Baillie, S.R., P.M. North, and A.G. Gosler (eds.) 1999. Bird Study 46(supplement):S1- S308.
An extensive body of software has been developed to complement the models referenced above. Dr. Evan Cooch of Cornell University maintains an excellent, comprehensive website with descriptions of software and links to all the major software sites. In addition, there is a mirror site from which to download software program "Mark". Program Mark was programmed by Dr. Gary White of Colorado State University and is widely recognized as the premier software for mark-recapture analysis. Cooch's site also contains his and Whites's electronic manual, "Program Mark - first steps...", and it features a lively on-line forum for discussions of program Mark and mark-recapture analysis.
The main Mark web site maintained by Gary White is another source of information and the main download site for program Mark. White's site also features the class lecture notes for a mark-recapture analysis course co-taught by him and Dr. David R. Anderson. White and Anderson conduct periodic workshops on program Mark and other aspects of mark-recapture.
The USGS Patuxent Wildlife Research Center software page maintained by James Hines has downloadable software not available elsewhere, and links to additional sites.
At the University of Manitoba, Dr. Carl Schwartz' Population Analysis Software Group web site is the home of POPAN, a mark-recapture survivorship program that also gives estimates of population size.