Project Tanager: The Development of a Large-scale Volunteer-based Research Project

S. K. Gregory1, Kenneth V. Rosenberg2, André A. Dhondt2, Gail P. Senesac2,
James D. Lowe2, and Diane L. Tessaglia-Hymes2

ABSTRACTProject Tanager was a 3-year, volunteer-based research project coordinated by the Cornell Lab of Ornithology. The project had two main goals: (1) to involve the public in scientific research and teach them how it works, and (2) to collect data to determine how the four species of North American tanagers are affected by forest fragmentation. Project Tanager volunteers followed a defined protocol in which they conducted point counts, used playback tapes, and collected detailed observations on birds and habitats. Finding appropriately trained, eager volunteers was a concern, but more than 1,400 volunteers were enlisted across the continent during the second and third years of the study. In 1994 we received data from about 356 teams, representing 700-1,000 individuals. Effects of forest fragmentation were different for each tanager species studied. This finding reduced a suspicion that volunteers may be collecting biased data. Benefits of the project were many: amateur birders became involved in and learned about scientific research; data analysis provided results not possible with other studies; and we built a solid base of volunteers and site coordinators that we are confident can collect scientifically valid data. We are currently using this volunteer base and protocol for new projects such as Birds in Forested Landscapes.


Fragmentation of North American forests has been linked to the decline of some species of Neotropical migratory birds (reviewed in Whitcomb et al. 1981, Wilcove and Robinson 1990). However, studies of sensitivity to fragmentation have been confined to specific regions or portions of the nesting range of individual bird species (e.g., Rosenberg and Raphael 1986, Robbins et al. 1989.) Investigating a response to habitat fragmentation across the entire range of a species is beyond the scope of most individual researchers. One practical way to collect the data necessary for such a study is to use volunteers. Project Tanager was designed by the Cornell Laboratory of Ornithology (CLO) to examine the effects of forest fragmentation on nesting tanagers across the United States and Canada, using volunteers as field teams. Project Tanager was highly successful at recruiting volunteer participants, and collecting large amounts of data on how tanagers respond to forest fragmentation continentwide. This paper describes how this large-scale volunteer-based scientific research project was developed, and outlines solutions to the major problems encountered.

Project Tanager initially was designed to test the hypothesis that tanagers are more likely to breed in large forest patches than in small ones. Several additional objectives accompanied our quest for data: (1) to develop a protocol for use in similar future projects; (2) to establish a network of volunteer birders to help with similar future projects; and (3) to involve the public in scientific research to improve their understanding of how science works.

Many questions arose as the project began: Could volunteers collect the type of data we need; could we ask amateurs to collect complex data; would enough willing participants conduct this type of research; could we trust the data they collect? After three seasons of data collection, more than 1,000 volunteers helped us to answer most of these questions. We also developed a workable protocol, and collected enough data to address our hypothesis.

Tanagers were chosen as the subject of this pilot study for several reasons. First, they are an attractive and easily recognized Neotropical migratory bird. Second, the Scarlet Tanager previously had been shown to be area-sensitive (Robbins et al. 1989) and is declining in parts of its range (Breeding Bird Survey data). Third, at least one of the four species of North American tanagers can be found during the breeding season across much of the United States and southern Canada. Scarlet Tanagers (Piranga olivacea) occur throughout the East, Summer Tanagers (P. rubra) in the South, Western Tanagers (P. ludovisciana) in the West, and Hepatic Tanagers (P. flava) in the Southwest.


Participants were recruited through news releases and direct mailings to bird clubs, professional ornithological contacts, and the membership of CLO. Notices also were included in ornithological newsletters and popular bird-watching magazines. Project volunteers did not need to be expert birders; we provided training materials and detailed instructions. However, those who volunteered turned out to be well educated: 75% had a bachelor's degree, and half of those also had advanced degrees (Hezel Associates 1993). Early volunteers helped us to recruit new participants through their own networks.

In the spring of 1993, we ran a pilot test by sending draft Research Kits to about 70 individual birders, some of whom worked in teams to try out the methods and collect data. In the spring of 1994, we sent revised Research Kits to more than 1,400 individuals. In turn, about 356 teams—representing from 700 to 1,000 people—returned data from more than 2,000 census points located on public, private, and corporate lands in 44 states and 4 Canadian provinces. About 70% of the census points were located in the eastern U.S. In 1995 we sent revised instructions to approximately 600 individuals who agreed to continue the project, and sent more than 800 kits to new volunteers

Research Kits contained: (1) instruction booklet; (2) computer-scannable data forms; (3) reference booklet with information on all four species of tanagers; (4) color poster of the tanagers; 5) cassette tape for learning tanager and cowbird songs and calls, and for playback use in the field; (6) transparent grid for determining patch size and percent forest; (7) letters of introduction for gaining access to private land; and (8) a "site coordinator" list of local contacts for help. Volunteers received the packets free of charge, but paid the cost of returning data to CLO.

Volunteers require rewards for their efforts. Our participants told us that one of their main rewards was being an integral part of a continentwide academic research program (a close second was seeing tanagers.) Also, many participants wrote letters or called for guidance and assistance in the field; this personal link with CLO staff was rewarding for participants as well as staff, providing both with direct feedback on the progress of the project.


Project volunteers were essential not only for collecting data; they also provided the feedback necessary to develop a protocol appropriate for this and similar, future studies. For example, in the 1994 season, the comments of the 1993 pilot testers were used to revise the protocol for broader distribution. Then, comments from the 1994 participants—as well as findings from data analyses—led to even further revisions for the 1995 season. The final Project Tanager protocol is described briefly below, followed by specific problems that we encountered and our solutions to them.

First, volunteers selected appropriate study sites in forest patches of various sizes. Within the study sites, they selected census points where they conducted at least two census visits during the breeding season, separated by four to six weeks. During each visit, volunteers conducted a census for the specific tanager or tanagers in their area, using the playback tape in the Research Kit if needed. Presence of cowbirds and potential nest predators were recorded during each census.

Volunteers also measured some habitat characteristics at each site, such as dominant tree species and canopy height, and they determined other habitat and landscape variables from aerial photographs or maps. Volunteers were encouraged to make additional visits to search for tanager nests at sites where tanagers pairs were present. Finally, they sent their data back to CLO for analysis. Results were reported to participants through CLO's quarterly newsletter, Birdscope.


Some of the problems that we encountered, and our solutions, follow. We hope these will be useful to others planning or conducting similar projects.

Site selection

Some participants had trouble finding suitable study sites. Volunteers often needed help finding maps or photos to determine forest patch size, and some needed assistance with completing habitat descriptions (landscape variables and tree species.) To alleviate this problem, we developed a network of more than 200 "site coordinators" whose names were sent to participants in 1994 and 1995, so that most volunteers could find a local source of help.

We developed the network by recruiting interested professionals from land management agencies, conservation organizations, and forest products industries. Over one third of the site coordinators were from the USDA Forest Service, with others from the U. S. Fish and Wildlife Service, Bureau of Land Management, Department of Defense, National Park Service, Bureau of Reclamation, Native American Fish and Wildlife Biologists Association, and several timber companies. We also used a select list of project participants (generally those from nature centers and private research facilities) to recruit potential site coordinators. The success of the site coordinator network was due in part to CLO employing a part-time staff person whose primary role was to recruit and nurture professional involvement in the project.

We asked site coordinators simply to answer questions or to help with site selection when contacted. However, we were pleased that many of them also collected data as project participants, and that some even recruited their own volunteers. For example, Jennifer Kimmich, a naturalist at Lake Metroparks near Cleveland, Ohio, recruited, trained, and coordinated more than 30 Project Tanager volunteers to collect data in the area she manages. Her experience was apparently satisfying, as she wrote us: "Enabling the amateur bird watcher to make a contribution to significant research is a win-win situation. We've seen Project Tanager participants gain feelings of accomplishment and learn skills in observing birds that would not have happened without the project ... they've gained an understanding of forest fragmentation."

Range of patch sizes allowable.

During the pilot year in 1993, participants were asked to find forest patches in four predetermined sizes of 1, 10, 100, and more than 800 hectares. Many participants, especially in the West, complained that these patch sizes were not available in their landscape or were difficult to find. In some cases continuous forests left no isolated fragments (northwestern U. S. and western Canada), and in others, only small fragments or narrow riparian strips of trees were available. Our solution in subsequent years was to allow study sites to be of any size, as long as participants told us what the sizes were. We still encouraged sampling in a variety of forest patch sizes, to the best of each participant’s ability. In the end, we realized that using a continuous range of patch sizes actually strengthened our analysis of area sensitivity.

Habitat descriptions.

During the pilot year we also asked for detailed descriptions of study-site habitat, including measures of foliage species and stem density at each site. Most participants found that collecting this information was tedious and complex, and we realized that not all of the vegetation data were necessary to the project. In 1994 we streamlined habitat descriptions, but we added an optional measure, the percentage of forested area in the surrounding landscape (1000 ha, or 2,500 acres, around each point). This information turned out to be so important to our analyses that we made it a mandatory measure for 1995, also adding a measure of forest-patch isolation. Most of the 1995 participants provided this landscape information.

No birds, no data.

Volunteers were reluctant to report results when they felt they did not have anything to report. We received numerous letters that said "Sorry, I didn't see any birds this year, so I didn't send you any forms." Clearly, this lack of negative data could create a bias. Our solution was to add a specific place on the data form where participants could record "no birds seen." We also rewrote the instruction booklet to explain the importance of following the protocol and completing the data forms when no tanagers were seen.

Coincidentally, we expanded the project by having participants record any potential nest predators noticed while conducting their census. This may have had a secondary effect of providing more data to report and validating (to the volunteer) the effort of completing and returning the data forms even if no tanagers were seen.

Two species at same point.

Data forms used in 1994 allowed participants to record two species of tanager on one form; however, an unforeseen problem occurred. In areas where the range of two or more species of tanager overlapped, it was not always apparent from the data forms whether censuses had been conducted for all potential species of tanagers. Two problems ensued: (1) both species were censused, but data were recorded on only one, or (2) a second tanager species was recorded only if it was encountered opportunistically. For example, in states such as West Virginia, where Scarlet and Summer tanagers both are present, Scarlet Tanager was recorded primarily as a "second species" if it was encountered during censuses for Summer Tanager. We suspect that this latter problem resulted in inflated values for the presence of Scarlet Tanager at the southern edge of its range. We were forced to remove these data from our geographic analyses. In 1995 we altered the protocol and asked participants to perform separate censuses for each species, timed at least one hour apart. Data forms also were redesigned, limiting each form to a single species.

"Inverse fragmentation"

In different regions of North America, forest fragmentation is displayed across the landscape in different patterns. In many parts of the East and Midwest, fragments of forest dot an open landscape. Much of the Northwest contains a continuous forest with scattered openings, creating a "Swiss cheese" pattern. In such areas, volunteers quickly pointed out that isolated forest fragments are difficult to find, and when they are found, often are an anomaly. By requiring participants to measure the percent forest in the 2,500 acres surrounding each point, we allowed many western areas with varying amounts of fragmentation to be differentiated, even if the individual study sties were in large or continuous forest tracts.

Tanagers were not the ideal species.

Tanagers offer many advantages for a large-scale volunteer study. In particular, they often are a favorite among birders, owing to their bright coloration. Indeed, we received many letters from excited participants who had lured the handsome males into clear view using the playback tape during their census. At the same time, tanager songs are easy to confuse with other species, and can be difficult to learn. Furthermore, tanagers nest in the forest canopy, making nests difficult to find. As a result, we received fewer data on nest success and cowbird parasitism than we desired. A species that nests on the ground, or in the sub-canopy, would likely have provided more data on nest success. This is the reason why the newer project Birds in Forested Landscapes, which evolved from Project Tanager, centers on the study of thrushes.


One potential criticism of a volunteer-based protocol is that data from non-scientists will be biased or unreliable. Although validating volunteer data can be difficult, we were able to address a few questions. Most importantly: Do amateurs collect the same quality of data as professional biologists, or will they see and hear fewer birds?

During the pilot year of Project Tanager, we were able to compare data collected by professionals and amateurs. CLO researchers completed tanager surveys at 19 sites that also were surveyed by amateur birders who were members of a local bird club. We were encouraged to find that at 17 of the 19 sites, results were identical. We also compared Project Tanager results for Scarlet Tanagers in the state of Maryland with results from another study of patch-size selection by migratory birds (Robbins et al. 1989). Project Tanager data collected by volunteers were gratifyingly similar to results from the academic study that Robbins and his colleagues completed.

Another potential criticism was that our volunteers would somehow bias the data to show that tanagers need large tracts of forest. Here we let the project test itself. Our results showed effects of forest fragmentation were different for each tanager species studied, and within some species, we could show regional variation. If our volunteers were biased, all species would have shown a dependence on large tracts of forest.


Project Tanager was a success on two levels: First, as a conservation-education tool, and second, as a scientific project.

Volunteers enthusiastically participated in the project, and told us that they learned about bird conservation and scientific research. Their excitement and dedication was evident in the numerous letters and phone calls that we received thanking us for allowing them to participate. One participant wrote: "It was most rewarding to see the male, female, and two young sweep down and land within 20 feet of where I was standing." Another wrote: "My daughter and I had a great time, she took her assignment quite seriously and did well for 4 1/2 years old ... I learned a lot about bird behavior; the more I learned, the more interested I became." Several participants instigated newspaper articles in their local areas about bird conservation and Project Tanager. Also, the project was used by high schools and colleges, conservation clubs, numerous bird clubs, and government agencies as a way to increase awareness of bird conservation and to encourage involvement in scientific research.

At the same time, the scientific results have been impressive; the data have allowed us not only to test our hypotheses, but the results have far exceeded our scientific expectations (see Rosenberg et al, 1999). For example, although Scarlet Tanager was the most area-sensitive species, we discovered significant regional variation in the relative importance of patch size per se and percent forest in the surrounding landscape.

A major implication of this finding is that extrapolating results from small-scale studies to new areas is often difficult, and potentially dangerous. Natural resource managers who must make urgent management decisions need more large-scale studies such as Project Tanager to address variation in habitat use across broad areas, or a large number of specific, small-scale studies. Due to logistics and expense, such studies will require the use of volunteers. In the future, we anticipate additional collaborative projects, building on our network of volunteers, and applying our successful protocol to address questions of conservation importance to other bird species.


First, we would like to thank several thousand volunteer participants whose efforts made Project Tanager a success. We also thank Greg Butcher for helping to conceive and initiate Project Tanager. This project was funded through grants to CLO from the National Science Foundation, the National Fish and Wildlife Foundation, the Wallace Genetic Foundation, and Kenentech/Windpower, all in cooperation with Partners in Flight. Many participants also voluntarily contributed funds. Additional support was provided by the USDA Forest Service and the New York Cooperative Fish and Wildlife Unit at Cornell University (Biological Resource Division).


Hezel Associates, 1993. "Project Tanager Evaluation: Pilot Participants’ Profile and Pretest." 1201 East Fayette Street, Syracuse, New York 13210.
17 pp.

Robbins, C. S., D. K. Dawson, and B. A. Dowell. 1989. Habitat area requirements of breeding forest birds of the Middle Atlantic States. Wildlife Monographs 103:1-34.

Rosenberg, K. V. and M. G. Raphael, 1986. Effects of forest fragmentation on vertebrates in Douglas-fir forests. Pages 263-272 in J. Verner, M. L. Morrison, and C. J. Ralph, eds. Wildlife 2000: Modeling habitat relationships of terrestrial vertebrates. Univ. Wisc. Press, Madison.

Rosenberg, K. V., J. D. Lowe, and A. A. Dhondt. 1999. Effects of forest fragmentation on breeding tanagers: A continental perspective. Conservation Biology 13:568-583.

Whitcomb, R. F., C. S. Robbins, J. F. Lynch, B. L. Whitcomb, M. K. Klimkiewicz, and D. Bystrak. 1981 Effects of forest fragmentation on the avifauna of the eastern deciduous forest. In Forest Island Dynamics in Man Dominated Landscapes, R. L. Burgess and D. M Sharpe, eds. New York: Springer Verlag, pp 125-205.

Wilcove, D. S. and S. K. Robinson. 1990. The impact of forest fragmentation on bird communities in eastern North America. Pages 319-331 in Biogeography and Ecology of Forest Bird Communities, A. Keast, ed. SPB Acad. Pub., The Hague, Netherlands.

1 Dept. of Natural Resources
   Cornell University
   Ithaca, NY 14850

2  The Cornell Laboratory of Ornithology
159 Sapsucker Woods Road
    Ithaca NY 14850