The article below appeared in the Spring 1998 issue of Birdscope, the Cornell Lab of Ornithology's newsletter:

BFL Analyzes Early Results
BY Stefan Hames

Veeries show a marked sensitivity to habitat fragmentation

The 1998 BFL season is already upon us, so we thought we would share insights from our preliminary 1997 results. In some cases, these early results reflect the patterns of response to forest habitat fragmentation that we had expected to see in thrush and accipiter species. In others, the results differ from our expectations, ranging from the subtle to the surprising, with some species showing high sensitivity to fragmentation, others showing sensitivity to vegetation structure in combination with fragmentation, and others showing no significant sensitivity to our analysis variables. Finally, these data suggest even closely related species with similar ranges react differently to habitat fragmentation.

Take the Veery. Based on our preliminary results, this relatively widespread species of the northern United States and southern Canada shows a marked sensitivity to habitat fragmentation. Figure 1, based on data from 449 study sites, plots on the vertical (z) axis the probability of finding a singing Veery on one or more of the two required site visits (this is scored as 'Possible Breeding' on the Visit Form). As shown in Figure 1, the probability of finding a 'Possible' breeding Veery increases from a low of near zero in small, isolated, high-altitude forest patches with an open canopy located in a fragmented landscape, to a high near 75 percent for large, low-altitude forest patches with a closed canopy that are located in a landscape with a high proportion of contiguous forest. These results are very much in line with our expectations and agree well with the habitat description included with the BFL Participants' Kit (see Reference Section 3.2.7 of the BFL Instruction Manual). With other species, however, the results are not so straightforward.

Based on our preliminary analyses, the Varied Thrush shows no sensitivity to fragmentation. Instead, as shown in Figure 2, the probability of finding a Possible Varied Thrush increases slowly as Factor 4 (a measure of the density of the forest understory or low vegetation) increases. Conversely, the probability of detection increases sharply with increases in Factor 5 (a measure of canopy height). This result seems to indicate a preference on the part of this thrush for forests with large trees and a well-developed understory, such as the coniferous forests of the Northwest, where this species is found (see Reference Section 3.2.19 of the BFL Instruction Manual). This validation of our expectations is heartening, but our conclusion is based on data from only 42 study sites, less than a tenth of the sample size for our analysis of the Veery's habitat preferences. Because of the small sample size, we are not as confident about this conclusion; it is entirely possible that this result would change if data from more study sites were available.

Another species for which we lack data is the Cooper's Hawk. This elusive forest raptor occurs in low densities all across the United States and southern Canada. This wide distribution makes it an excellent choice for BFL, but the fact that it occurs in low densities means that the probability of detecting this hawk at any given point is much lower than that of detecting one of the thrush species. Hence, we need a larger number of point counts to find the number of 'Possible' breeders we need for our analysis. For example, Figure 3 (on page 14) is based on only 34 detections out of a total of 437 study sites. It is clear that the probability of finding a calling Cooper's Hawk does not vary with fragmentation (Factor 1) or with elevation/canopy closure (Factor 2). Our preliminary analysis suggests that none of the included variables (Factors 1Ð5) have a significant effect on the Cooper's Hawk.

Several possible explanations exist for the apparent lack of effect. One could be the effects of habitat fragmentation and vegetation structure, but our current sample size is too small to detect these effects. Another is the 1,000-hectare scale at which participants calculate measures, such as percentage of forest, amount of linear edge, and isolation, is too small for a wide-ranging predator such as the Cooper's Hawk, although it is adequate for species with smaller home ranges, such as the Veery. A third is that fragmentation has no effect on the Cooper's Hawk.

We hope to address the question of sample size by ensuring that participants receive instruction materials in plenty of time to allow them to be in the field prior to the beginning of the Cooper's Hawk's incubation period, when the hawks are most easily detected. We are also working to discover the appropriate scale for measuring the effect of habitat fragmentation on the Cooper's Hawk. Lab scientists are developing methods to measure fragmentation with computers using the Lab's Geographic Information System facilities, which will enable us to make these measurements on a number of scales. These measurements, coupled with data from BFL participants, will help us pinpoint the size of the landscape we need to consider when studying raptors such as the Cooper's Hawk.

One exciting area of investigation is to determine whether two closely related, co-occurring species differ in their responses to fragmentation. By surveying for multiple species at a single study site, BFL participants help us make the direct comparisons we need to answer this question. For example, the Swainson's and Hermit thrushes appear to respond differently to fragmentation, with the Swainson's Thrush showing a significantly greater response to Factor 1, which is an overall measure of fragmentation. This intriguing interim result stresses the importance of censusing for more than one species at the same location.

Our early look at the data from the 1997 BFL field season is exciting, indicating the project's great potential. Although we need to improve (by increasing the number of participants, increasing the completeness of data forms submitted by participants, and increasing the number of sites at which multiple species are censused), this partnership between the thousands of volunteers and the Lab's scientists is off to a great start. We hope for even greater things from the 1998 field season.

top of page

This article appeared in the Winter 1998 issue of Birdscope:

Birders in Forested Landscapes

by SARA E.BARKER and JAMES D.LOWE

In its first season, BFL attracts more participants
than expected

What is hot and sweaty, covered in bug repellent, and plays the finest tunes in town?  A Birds in Forested Landscapes (BFL) participant, of course.  BFL took flight this past spring and summer (1997), in its first full field season.  As the successor to Project Tanager, BFL looks at habitat and landscape characteristics that are critical to the successful breeding of certain thrushes and hawks.  This project will serve as the basis for conservation recommendations and habitat management strategies that will help forest-dwelling birds.

The response to BFL was overwhelming.  More than 1,700 people signed up--twice the number of sign-ups for Project Tanager in any single year.  We are also impressed with the quality of the data from the participants and the extent to which they have completed their forms.   Time and dedication--that's what our participants are giving, and that's why this project is already a success.  We really appreciate this great effort.  Thus far, we have received data from 335 participants in 46 states and 4 Canadian provinces.   these participants studied 1,600 sites, an average of nearly 5 sites per person.   At this early stage, BFL is matching the Project Tanager returns, and we expect BFL numbers to grow even higher.  More avid citizen scientists are joining our core Project Tanager participants every day.

Seventy-four percent of BFL participants studied both hawks and thrushes, which will enable Lab biologists to study the effects of habitat fragmentation on two different landscape levels.  We hope to use our new Geographic Information System (GIS) to obtain in-depth habitat information from digital maps, satellite imagery, and aerial photographs.  With this added flexibility, we can analyze data not included on participants' data forms as well as decrease the amount of habitat data participants need to record in the future.

How do nest predation and cowbird parasitism rates vary according to forest patch size, habitat characteristics, and geographic regions?  Unfortunately, we have received little detailed information about nesting thrushes and hawks.  Searching for nests can be one of the most time-consuming parts of the project, but it is one of the most rewarding.   Next year, we hope to include more information in our literature about how to confirm nesting success or failure.

One third of the data forms scanned to date represent participants who studied Wood Thrushes, whereas another third studied both Sharp-shinned and Cooper's hawks.  The last third incorporates the remaining six thrush species, which breed for the most part in areas where we have few participants (see map).  We need more participants throughout the Midwest, the Canadian provinces, and Alaska; we are, however, gladly signing up participants all across North America where thrushes and hawks breed.

We will present our first results in future issues of Birdscope, so keep your eyes peeled!

BFL Participants Pass Canopy Cover Estimate Test

BY STEFAN HAMES

Not many people would doubt the birding abilities of participants in the Lab's citizen-science projects.   But, we also ask our volunteers to do things such as estimate vegetation parameters, which may be new to many of them.  How well do our participants do?   Lab biologists take to the field each year to answer that question.

last summer, for example, lab biologists tested how well BFL participants could estimate the percentage of canopy cover at their study sites.  To do this, we first estimated the percentage of canopy cover at 30 census points using a chart included with the BFL participants' kit.   We then compared our estimates to detailed measurements obtained using a standard vegetation sampling method (100 random sightings of the canopy at each point through an optical device).

The results, as shown in the graph below, are quite good.  Overall, estimates of canopy cover were not significantly different from the measured values (statistically speaking, using a goodness-of-fit test, chi-square=3.599, df=3, p=0.309).  Additionally, our estimates of the canopy cover were within 5 percent of the measured value at nearly half of the points (43%).  Not bad at all!