Developing a protocol for monitoring the Bicknells Thrush (Catharus bicknelli) and other high elevation bird species in Atlantic Canada
Becky Whittam and Melanie Ball
Bird Studies Canada - Atlantic Region P.O. Box 6227 Sackville, NB E4L 1G6 (506) 364-5047 With funding from Species At Risk Implementation and Recovery Program Parks Canada And Canadian Wildlife Service March 31, 2002
High Elevation Bird Monitoring

page 2

Table of Contents Executive Summary.... 3 Introduction.... 4 Methods.... 4 Analyses and Results.... 6 Route Choice... 6 Number of Routes... 6 Factors affecting bird vocalizations... 7 Likelihood of Bicknells Thrush detection.. 7 10- versus 15-minute point count protocols... 8 Playback.... 8 Habitat... 9 Discussion.... 9 Number of Routes... 9 Factors affecting Bicknells Thrush vocalizations... 9 Likelihood of Bicknells Thrush detection.. 10 10- versus 15-minute point count protocols.. 10 Playback... 10 Habitat.... 11 Recommended Protocol... 11 Acknowledgements... 12 Literature Cited... 13 Tables..... 14 Figures.... 19 Appendix A. Location of all survey stops.. 24

page 3

Executive Summary The goal of this project was to develop a protocol for a volunteer-based survey of Bicknells Thrush (Catharus bicknelli)and other high elevation bird species in Atlantic Canada. The two objectives of the survey would be to: 1) monitor populations of high elevation species; and 2) characterize Bicknells Thrush habitat in Atlantic Canada. Eighteen routes consisting of 5 stops spaced 250 m were run in June 2001 in the highlands of Nova Scotia, in order to test short (10minute) and long (15-minute) protocols, and to determine whether time of day, date, weather, and/or playback affect the detectability of the four target species (Bicknells Thrush, Swainsons Thrush, Blackpoll Warbler and Fox Sparrow). Qualitative habitat data were also collected along each stop for each route. Using the data collected during this study, along with data collected in New Brunswick in 1997, a power analysis indicated that a sufficiently powerful survey protocol (power of 0.90 or more and a declining trend of 3%/year, at a significance level of 0.85) could be accomplished using 65 routes surveyed once per year over 20 years. Results of generalized linear models indicated that neither day of year, route, observer, time of day, temperature, wind, cloud or precipitation influenced vocalizations by any of the four target species. The likelihood of detecting Bicknells Thrush increased on each successive run of a route, although the Probability of Occurrence of Bicknells Thrush on any given run of a point was relatively high, at 0.66. There was no difference in the mean number of individual target species detected using the short versus the long protocol, although the Relative Abundance of Bicknells Thrush was higher using the long versus the short protocol. Bicknells Thrush vocalizations peaked between 4-4:30 am and 8-8:30 pm. While playback did not significant increase the number of Bicknells Thrush detected, there was a tendency for more Bicknells Thrush to be detected for the first time following playback compared with before playback. Bicknells Thrush did not appear to habituate to playback using data from multiple runs of the same route. Stops where Bicknells Thrush were present were categorized more often as having dense conifer stems compared with stops where Bicknells Thrush were absent. Hardwood stems were sparse in all cases. Stops where Bicknells Thrush were present also tended to fall into the shorter canopy height category compared with stops where Bicknells Thrush were not present. All types of stops (whether Bicknells Thrush were present or absent) were categorized as predominantly spruce/fir. We recommend a monitoring program involving 65 routes run once per year, either in the morning or evening in mid-June, with each route consisting of 5 stops spaced 250 m. At each stop, a 10-minute protocol consisting of silent listening and playback should be used. Routes where Bicknells Thrush are not detected should be revisited during the peak survey period to verify the presence or absence of the species for more accurate habitat characterization.

page 4

Introduction The main goal of this project was to develop a protocol for a volunteer-based survey of Bicknells Thrush (Catharus bicknelli) and other high elevation bird species in Atlantic Canada. This habitat is under threat by climate change (causing loss of red spruce and balsam fir), development (ski areas, cell tower and wind power development) as well as industry, and is often not well covered by existing bird monitoring programs like the Christmas Bird Count and Breeding Bird Survey. The Bicknells Thrush is one of the rarest songbirds in North America, with a population of between 5,000-15,000 pairs (Nixon 1998). It is listed as a species of "High responsibility, High Concern" by Partners in Flight Canada (Downes et al. 2000), and as a Species of Special Concern by the Committee on the Status of Endangered Wildlife In Canada (COSEWIC). Furthermore, it is listed as Globally Vulnerable by the IUCN (Stattersfield and Capper 2000). These designations are largely a result of the species' fragmented populations, and the fact that very little is known about its population status and breeding biology, because of its shy habits, remote breeding habitat, and only very recent designation as a full species (Ouellet 1992, AOU 1995). A monitoring program is urgently needed to help determine the status of the Bicknells Thrush population: is it increasing, decreasing, or remaining stable? Secondarily, a survey could help provide information on habitat used by the Bicknells Thrush in Atlantic Canada. Therefore, the main objectives of this project were: 1. To use the Bicknells Thrush population in Cape Breton Highlands National Park (CBHNP) and nearby Cape North to develop and test a protocol for a volunteer-based survey of high elevation bird species, with particular emphasis on the Bicknells Thrush, and secondary emphasis on Fox Sparrow (Passerella iliaca), Swainson's Thrush (Catharus ustulatus) and Blackpoll Warbler (Dendroica striata). Specifically, we wanted to determine: a) how many routes are required to achieve appropriate statistical power to detect population change over time; b) how long a point count needs to be; c) how many times a route needs to be surveyed in a season; d) whether time of day (morning versus evening) affects survey results; e) whether date affects results; and f) whether or not playback is necessary to increase the ability to detect Bicknells Thrush. 2. To submit new Bicknells Thrush records to the Parks Canada GIS database for use in mapping Bicknells Thrush habitat in Atlantic Canada. Methods Along with a field crew, we scouted eighteen survey routes between 4 and 11 June in five general areas: Pacquet Lake Trail (CBHNP), Lac a Simeon Trail (CBHNP), Tipover Lake Trail (CBHNP), French Mountain (CBHNP), and Money Point Road, Cape North (Figs. 1-6). We chose routes based on habitat suitability, and almost always placed them in areas known (from previous research; R. Knapton and D. Busby, unpubl. data) to have Bicknells Thrush. Each route consisted of 5 stops separated by 250 m. We waypointed each stop along each route (NAD 83; see Appendix A). We also collected basic qualitative habitat information for each stop using the Mountain BirdWatch protocol (VINS 2001): at each stop, we categorized conifer, hardwood,

page 5

and understorey stem density as dense, moderate or sparse, and we also categorized each stop by tree dominance (conifer & hardwood co-dominant, hardwood dominant, or spruce/fir dominant) and canopy height (1-3m, 3-5m, 5-7m, or 7-9m). Routes were run between 11 - 27 June. We collected weather data (cloud, precipitation, temperature) for every survey, and noted the time we started and finished each route. We attempted to survey each route 8 times: twice in the morning (3:30-6:30 am) and evening (7:0010:00 pm) within a 36 hour period, alternating between long and short survey protocols (see below), then repeated again about one week later. For example, for a single route we might attempt to run it Monday evening (long protocol), Tuesday morning (long protocol), Tuesday evening (short protocol) and Wednesday morning (short protocol). We did not vary the order of protocols (i.e. we almost always ran them long pm, long am, short pm, short am). We then attempted to repeat this series of 4 survey runs approximately one week later. We tested both long and short survey protocols. The long protocol was 15 minutes in total, beginning with 10 minutes of silent listening, followed by a 30 second broadcast of Bicknells Thrush vocalizations and another 5 minutes of silent listening. The short protocol was 10 minutes in total, beginning with 5 minutes of silent listening, and then a 30 second broadcast of Bicknells Thrush vocalizations, then another 5 minutes of silent listening. The Bicknells Thrush broadcast tape consisted of a 30 second sequence of calls and songs, originally recorded by Dan Busby. The same model of sound equipment was used for almost all surveys (Sony CFSE2 Radio-Cassette Recorder; recommended by the Vermont Mountain Birdwatch program). An alternate unit (Sony CFD-S22), which was less portable but equally audible, was used when four researchers were working simultaneously as we only had three CFS-E2s. On each survey stop, we concentrated on detecting the four target species: Bicknells Thrush, Fox Sparrow, Swainsons Thrush, and Blackpoll Warbler. If a target species was detected we recorded the time it began to vocalize, its approximate position in relation to a 100 m radius study plot, and the number of minutes it continued to vocalize following detection (in minute-byminute intervals). We did not distinguish between songs and calls. Whenever possible, we also noted the presence of any non-target bird species seen or heard during the stop. The same person surveyed the same routes as much as possible. Each surveyor surveyed only one route per period (i.e. morning/evening). Five different people conducted surveys throughout the length of the project (Sean Blaney, Beth Flanigan, Nev Garrity, Mike Russell, and Becky Whittam), but generally only three to four people conducted surveys on any given day (BF, MR, BW). Each surveyor was trained to recognize the four target species using the Mountain BirdWatch training tape and by practicing in the field with knowledgeable biologists.

page 6

Analyses and Results Route Choice We established 18 routes, all of which had at least one of the four target species: 15 had Bicknells Thrush, 18 had Swainsons Thrush, 16 had Blackpoll Warbler, and 12 had Fox Sparrow. Because of unfavourable weather conditions, limited personnel, and situations that posed a risk to the surveyors (i.e. presence of a bear) we were unable to run all survey routes 8 times. We ran each route an average of 5.8 times (range 4-8; Table 1), and we ran each route during at least 2 mornings (one short one long) and 2 evenings (one short, one long). In total, we completed 105 surveys along these 18 routes. Figure 1 shows the location of our survey routes in CBHNP and Cape North. Figs. 2-6 show in detail each stop along each route at the five main areas under study (Simeon Lake, French Mountain, Pacquet Lake, Tipover Lake, and Cape North), and indicate which stops had Bicknells Thrushes at least once and which did not. The mean proportion of survey runs that had Bicknells Thrushes was 0.710.09 (range 0.25-1.0). In other words, on any given run of a route, we had a 71% chance of encountering a Bicknells Thrush. In addition to those birds detected on survey routes, we also found five previously unrecorded locations for Bicknells Thrush in CBHNP (Appendix A, incidental detections). In total, including both incidental detections and birds detected on survey routes, we recorded 53 different locations for Bicknells Thrush in CBHNP (Appendix A). In order to determine if the 18 routes from the five areas could be considered statistically independent, i.e. no grouping effect of routes by area, we did a nested ANOVA by area and route. Cape North and French Mountain, which each had 4 routes, were tested in one ANOVA, and Simeon Lake, Pacquet Lake, and Lac a Simeon, which each had 3 routes, were tested in another ANOVA. In both groups, routes were significantly different from each other (F=49.25, F0.05 (2),15,3, P<0.0005 and F=49.25, F 0.05(2)6,2=7.26, P<0.0005, respectively) warranting testing them as independent units. Number of Routes Using the count data from Cape Breton in 2001 and data from New Brunswick for 1997 provided by Steve Holmes, we calculated an average mean (0.81) and standard deviation (0.83) of Bicknells Thrush detected. Using MONITOR, a power analysis program developed by James Gibbs (1995), we ran a number of simulations varying the number of routes, the number of runs of a route and the number of years over which the project would take place. We decided that our goal would be the ability to detect a 3% decline in the population at 90% power with a significance level of = 0.15 (Eagle et al. 2000). The power analysis indicated that a sufficiently powerful survey protocol (power of 0.90 or more and a trend of 3%/year, at a significance level of 0.85) could be accomplished using 65 routes, surveyed once per year over 20 years (Table 2).

page 8

heard again in subsequent runs of the same area (Nixon 2001). The Probability of Occurrence of Bicknells Thrush on any given point in Nova Scotia was 0.66; this is quite a bit greater than the probability of occurrence found previously in New Brunswick (0.40, Nixon et al. 2001), indicating that there is a high level of variability between sites and years. 10 versus 15 minute point count protocols We compared the number of individuals heard during 10 minute versus 15-minute point count protocols for all species by route, using a Wilcoxin rank-sum test. None of the resulting p-values were significant (Bicknells Thrush p=0.198, Swainsons Thrush p=0.340, Fox Sparrow p=0.388, Blackpoll Warbler p=0.164) (Table 4), indicating that there may not be a significant difference between the numbers of individuals found on the same route using the short protocol versus the long protocol. We also calculated Relative Abundance for Bicknells Thrush by point and compared 10 minute and 15 minute protocols to determine which captured the greatest Relative Abundance of birds. The result was significant using the Wilcoxin rank-sum test for paired data (T_=145, T0.05 (2),42 =319, p<0.05), indicating that the longer protocol does detect a greater Relative Abundance of Bicknells Thrushes. Graphs of peak vocalization time for Bicknells Thrush in Cape Breton during the 2001 season show a distinct peak in song for a 30 minute period from 4-4:30 am and 8-8:30 pm (Figs. 9 and 10). These short windows for best hearing Bicknells Thrush are similar across the species breeding range and were one of the major reasons the Mountain Birdwatch program chose a 5minute protocol for each point count (VINS 2001). Playback To determine whether or not playback increased the number of Bicknells Thrush detected, we examined individual Bicknells Thrush detections. Using data from all short protocol runs of all routes (n = 49), we calculated the number of birds that were detected before & after playback, before but not after playback, and after but not before playback. Out of a total of 87 detections, most birds (37%) were detected both before and after playback; only 21.8% were detected before, but not after playback, and 35.6% were detected after, but not before playback. We also examined the effects of playback using a slightly different method. We divided the short protocol into two, five-minute periods, and the long protocol into three, five-minute periods. Since the short protocol involved playback before the second five-minute period, whereas the long protocol didnt involve playback until after the second five-minute period, the proportion of Bicknells Thrush detected in that second five-minute period could be compared between the short and long protocols to determine whether playback has any effect (hypothesizing that, if playback is important, a larger proportion of Bicknells Thrush detections would occur in that second five minute period using the short protocol than using the long protocol). Indeed, this appeared to be the case, with 36% of Bicknells Thrush detected in the

page 9

second five-minute period when it was preceded by playback (i.e. short protocol) compared to only 27% when it was not (i.e. long protocol) (Table 5). However, this difference was not significant (chi-square = 1.70, df = 1, p > 0.10). In order to determine if individuals might be becoming habituated to playback over time we did a Friedmans test comparing the number of birds heard at the same points of each route over time (i.e. on successive runs of the same route). Out of the 96 tests done, we had one significant result for Swainsons Thrush on Route 4, which was no longer significant after a Bonferroni correction, indicating that playback did not cause a decline in the number of birds heard over time.
Habitat Stops where Bicknells Thrush were present were categorized more often as having dense conifer stems compared with stops where the birds were not present (Table 6). Understorey stem density of stops where Bicknells Thrushes were present tended to be either sparse or dense compared with stops where the birds were not present (Table 6). Hardwood stems were sparse in all cases (Table 6). Stops where Bicknells Thrushes were present also tended to fall into the shorter canopy height category compared with stops where Bicknells Thrush were not present (Table 7). Canopy dominance did not appear to differ between stops with and without Bicknells Thrush; all types of stops were characterized as predominantly spruce/fir (Table 8). Discussion Number of Routes The preliminary results from the power analysis suggest that 65 routes surveyed once per year should provide sufficient power for detecting population change. However, because both sets of data used in this study were from non-randomly chosen sites (i.e. chosen because of the known presence of Bicknells Thrush), this power analysis will not reflect the results more likely obtained from the stratified random sampling that will occur during the actual monitoring program. Given this caveat, the power analysis will be redone over the next two years of the monitoring program in order to determine more appropriate estimates of power. Factors affecting Bicknells Thrush Vocalizations Generalized linear modeling suggests that most of the environmental variables measured while doing the counts are not significantly affecting vocalizing by any of the four species. Route was important, however, indicating that there is some variation among routes in different areas. The relatively poor fit of the model indicates that another factor, or factors such as pairing status, may be influencing the vocalization behaviour of these species.

page 10

Likelihood of Bicknells Thrush detection While variation was high, the maximum number of Bicknells Thrush recorded on any given route increased slightly, although not significantly, with each additional run of a route, suggesting that multiple runs are necessary in order to exhaustively survey the number of birds on a route. Nixon et al. (2001) also found that the cumulative number of points where Bicknells Thrush were detected increased steadily, with no leveling off, over six visits, leading to a low probability of detection (0.40) on any given visit. It is difficult to imagine a volunteer survey requiring six or more visits to a site, however, and because population trends are the goal of this monitoring program rather than population size, one or two runs of a route would suffice. Given that fewer Bicknells Thrush are detected, in general, on the second run of a particular protocol (separated by 6.5 days), one run of each route should suffice. Neither Bicknells nor Swainsons thrush appeared to be detected more often in the morning compared with the evening, either in terms of absolute numbers heard or relative abundance. This is an important result, as it suggests that surveys can be run during either of these two periods, making it easier for volunteers to choose a convenient survey time. Blackpoll Warbler and Fox Sparrow, however, appeared to be detected more often in the evening compared to the morning, although these differences were not significant. 10 versus 15 minute point count protocols Despite the fact that there was no statistical difference in the absolute number of Bicknells Thrush detected in the short and long protocols, our results do show that on average, the long protocol detected 1.5 to 2 times as many Bicknells Thrush compared with the short protocol and the Relative Abundance of Bicknells Thrush was greater on long vs. short protocols. The other target species (Swainsons Thrush, Blackpoll Warbler, Fox Sparrow) showed very little difference between numbers detected on short and long protocols. These species may be more likely to be singing constantly during the survey period, whereas Bicknells Thrush may sing less predictably. One potential problem with using the longer protocol is an increased chance of missing birds on the last few stops of the survey. The window of time for peak Bicknells Thrush vocalizations in the morning is 4-5:30 and 7:30-9:00 in the evening (Figs.9,10). Only the short protocol allows for all points to fall within the 1.5 hour time frame when Bicknells Thrush vocalizations are greatest. Playback Our results do not definitively prove that playback significantly increases the number of Bicknells Thrush detections, although playback may improve the detectability of Bicknells Thrush to a small extent. Our results, however, may be confounded by the fact that we always

page 11

ran the short protocol after the long protocol, and fewer Bicknells Thrush were detected using the short protocol (87) than in the first ten minutes of the long protocol (111, or 97 when corrected for the slight difference in number of runs of the long protocol compared to the short protocol; see Table 3). However, there was no significant decrease in the numbers of Bicknells Thrush detected on a point over time, indicating that habituation of Bicknells Thrush to playback may not be an issue. Rimmer et al. (1996) found that playback increased the efficiency of detecting Bicknells Thrush. Consequently, playback is used by participants in Mountain BirdWatch, although only when birds are not detected by point counts alone (Lambert et al. 2001). Furthermore, Bicknells Thrushes detected after playback in Mountain Birdwatch are not included in trend analysis (Lambert et al. 2002). Habitat More detailed habitat measurements at point counts are required in order to better elucidate the habitat use of this species. Our qualitative measures of habitat did not show any clear trends, except that conifer stem density was greater at sites that had Bicknells Thrush compared with those that didnt. This may reflect a preference for overall stem density as well as conifer stem density. Our study involved routes that were chosen because they were known or suspected to have Bicknells Thrush. The habitat did not vary greatly between sites, and we did not include any controls in the habitat analysis. As a result, there was very little variation between our habitat measurements on sites that did and did not have Bicknells Thrush. Furthermore, our qualitative measures were so broad that any variation that might have existed was likely not recorded. Routes must be chosen randomly from available habitat, and must not be chosen based on knowledge of Bicknells Thrush presence, in order to gather unbiased data on population trends and habitat selection by this species. Because the monitoring protocol is designed to detect population trend and not absolute presence or absence of Bicknells Thrush in an area it will not provide enough data to allow a proper habitat analysis unless routes where Bicknells Thrush were not detected are revisited at least once. Therefore, during years when habitat analyses are done (see below), technicians will attempt to re-run routes where no Bicknells Thrush were found in order to get a better estimate of presence/absence of the bird. This information will then be used in the habitat analysis, but not for the monitoring program. Technicians will then do quantitative measurements on habitat following Nixon (2001) and Connolly (2000), and in conjunction with the methods chosen by the Mountain Birdwatch program. Recommended Protocol In 2002, suitable areas in Nova Scotia, New Brunswick and PEI will be chosen using a GIS model based on current information on the elevation and habitat preferences of Bicknells Thrush (Nixon et al. 2001 and D. Busby pers. com.). Additional constraints on suitable areas will include: availability of roads or trails; areas at, or exceeding, the 1-km transect length and

page 12

50m radius of each transect; and access to land. Using stratified random sampling, final routes will be chosen that satisfy the GIS criteria as well as practical limitations (VINS 2001). We recommend the following protocol: 1. 65 or more routes monitored once, starting 30 minutes before dawn or at dusk, during mid-June in order to provide data for population trends. 2. Each 1 km route should consist of 5 points with 250 m between points. 3. Each point along each route should consist of 5 minutes of silent listening, a playback component of 30 seconds and then 5 more minutes of silent listening. 4. On routes where no Bicknells Thrushes are found during the first visit, technicians will attempt to re-run the route during the peak period to better determine presence/absence of the bird for habitat analyses. Field technicians will measure habitat in July every 2-5 years, or volunteers will be asked to record basic habitat characteristics each year. Because routes were chosen non-randomly for this study, this recommended protocol should still be considered tentative. 2002 should be considered another pilot year during which sites will be chosen using stratified random sampling and the power analysis will be re-run with these new independent data; it is likely that the number of routes required for sufficient power will increase. Acknowledgements Thanks to Parks Canadas Species At Risk Implementation and Recovery Program and the Canadian Wildlife Service (CWS) for financial support. Dan Busby (CWS) provided the initial impetus for the study. James Bridgland, Sheldon Lambert, and Derek Quann of Cape Breton Highlands National Park provided scientific and logistical support while in the park. Thanks to Richard Knapton and Rob Boone of the University College of Cape Breton for discussions on Bicknells Thrush fieldwork, for sharing the Big Intervale cabin with us, and for keeping us safe from Big Red. Kim Mawhinney provided a liaison between the Parks Canada Species At Risk program and other parks programs. Thanks to Dawn Allen of Parks Canada for creating the maps used in this report and providing GIS layers required for random route selection in 2002. Thanks to intrepid field workers Beth Flanigan and Mike Russell for their excellent surveying and hiking skills. Nev Garrity and Dan Busby spent several days helping to train the field crew in survey techniques, and Nev Garrity and Sean Blaney helped with fieldwork for one week. Thanks to the Canadian Wildlife Service for providing logistical support to us while completing this report. Yves Aubry, J.P. Savard, Erica Dunn, Brian Collins, Dan Lambert and Steve Holmes provided constructive comments on earlier drafts of this document.

page 13

Literature Cited Downes, C.M., E. H. Dunn and C. M. Francis. 2000. Canadian Landbird Monitoring Strategy: monitoring needs and priorities into the new millennium. Partners In Flight-Canada, Ottawa. Eagle, P.C., J.P.Gibbs, S. Droege. 2000. Power Analysis of Wildlife Monitoring Programs: Exploring the trade-offs between survey design variables and sample size requirements. Available online at: http://www.pwrc.usgs.gov/resshow/droege3rs/salpower.htm Lambert, J. D., K. P. McFarland, C. C. Rimmer, and S. D. Faccio. 2001. Mountain BirdWatch 2000. Final Report to the U.S. Fish and Wildlife Service by the Vermont Institute of Natural Science. Available online at: www.vinsweb.org/conservation/citizenscience/2000fullreport.html. Lambert, J.D., S.D. Faccio, B. Hanscom. 2002. Mountain BirdWatch 2001. Final Report to the U.S. Fish and Wildlife Service by the Vermont Institute of Natural Science. Available online at: www.vinsweb.org/conservation/citizenscience/2001fullreport.html. Nixon, E.A. 1998. Status report on Bicknells Thrush, Catharus bicknelli, in Canada. COSEWIC, Environment Canada, Ottawa. 53pp. Nixon, E.A., S. B. Holmes, and A. W. Diamond. 2001. Bicknells Thrushes (Catharus bicknelli) in New Brunswick clear cuts: their habitat associations and co-occurrence with Swainsons Thrushes (Catharus ustulatus). Wilson Bulletin 113: 33-40. Ouellet, H. 1993. Bicknells Thrush: taxonomic status and distribution. Wilson Bulletin 105: 545-754. Rimmer, C. C., J. L. Atwood, K. P. McFarland, and L. R. Nagy. 1996. Population density, vocal behavior, and recommended survey methods for Bicknells Thrush. Wilson Bulletin 108: 639-649. Stattersfield, A.J. and D. R. Capper. 2000. Threatened birds of the world: the official source for birds on the IUCN red list. Cambridge: BirdLife International. VINS 2001. Mountain BirdWatch protocol. Unpublished survey protocol by the Vermont Institute of Natural Sciences. Available online at: www.vinsweb.org/conservation/citizenscience/mb_procedures.html

page 14

Table 1. Number of morning and evening surveys, along with total number of surveys, run for each survey route in 2001. max BITH refers to the maximum number of Bicknells Thrush detected on any given run of a route. See Appendix A for specific route locations.
route # CN - 1 CN - 2 CN - 3 CN - 4 BL - 5 RT - 6 LT - 7 SL - 1 SL - 2 SL - 3 PL - 1 PL - 2 PL - 3 RT - 14 TO - 1 TO - 2 TO - 3 SK - 1 name # am # pm total max BITH Cape North 16 Cape North 9 Cape North 11 Cape North 3 Benjies Lake Radio Tower La Tourbiere Simeon Lake 0 Simeon Lake 3 Simeon Lake 1 Paquet Lake 11 Paquet Lake 4 Paquet Lake 1 Bog/Radio Tower Tipover Lake 1 Tipover Lake 6 Tipover Lake 7 Skyline comments Lost 1 evening to rain Lost 1 evening to rain Lost 1 evening to rain Lost 1 evening to rain Lost 1 evening to lack of time Lost 1 evening to lack of time Lost 1 evening to lack of time Lost 1 AM & 1 PM to bear Lost 1 AM & 1 PM to bear

page 15

Table 2. Power analysis for High Elevation Bird Monitoring project. Routes monitored are the number of routes considered in the study, within year effort is the number of times each route is surveyed per year, and number of years is the length of time over which the study is conducted. MONITOR requires a mean and standard deviation from count data over multiple years. On each route we calculated the mean and then subtracted that mean from each count along the same route; using the results we pooled the data and calculated the standard deviation for all routes. By de-meaning the counts the spatial variation is removed, i.e. differences due to habitat and random error in counting methodology, and the temporal variation is left which allows pooling of the data. The significance level was 0.15. Routes Monitored 75 Within Year Effort 1 Number of Years 20 Increasing Trend 5% 3% 2% 5% 3% 2% 5% 3% 2% 5% 3% 2% 5% 3% 2% Power Decreasing Trend 0.87 0.92 0.92 0.89 0.91 0.92 0.89 0.93 0.96 0.93 0.95 0.95 0.93 0.96 0.96 10% 7% 3% 10% 7% 3% 10% 6% 3% 10% 5% 3% 10% 5% 3% Power 0.78 0.89 0.85 0.78 0.93 0.89 0.84 0.86 0.91 0.84 0.85 0.93 0.86 0.86 0.95

page 16

Table 3. Generalized Linear Model t-value results for Bicknells Thrush (BITH), Swainsons Thrush (SWTH), Fox Sparrow (FOSP) and Blackpoll Warbler (BLWA). Highlighted values are significant (<0.05). Measurements Day Route Observer Time Start Time End Temp. Start Temp. End Wind Start Wind End Cloud Start Cloud End Precipitation Start Precipitation End BITH -0.32 -4.13 -1.68 1.17 1.31 0.58 -0.42 -2.28 1.86 0.66 0.79 2.52 -2.32 SWTH -0.72 -5.90 3.29 -0.45 0.42 -0.07 0.62 -4.60 4.71 2.54 -2.82 -0.FOSP 0 BLWA 2.07 -5.28 0.02 -0.86 0.85 0.41 -0.15 -1.05 1.81 -0.22 1.33 0.81 -0.36
Table 4. Mean ( SE) number of target species detected per protocol, using data from the first run only. BITH = Bicknells Thrush, SWTH = Swainsons Thrush, BPWA = Blackpoll Warbler, FOSP = Fox Sparrow. Sample sizes are given in brackets. BITH 3.28 0.88 3.39 1.03 1.5 0.49 2.11 0.60 SWTH 2.89 0.60 4.67 1.06 4.06 0.90 3.94 0.96 BPWA 1.56 0.51 1.44 0.36 1.61 0.57 1.17 0.46 FOSP 1.11 0.40 0.78 0.37 0.72 0.24 0.50 0.29
Long pm (N = 21) Long am (N = 18) Short pm (N = 15) Short am (N = 18)

page 17

Table 5. The number (and percent) of Bicknells Thrush detections (corrected for the number of runs; see below) occurring in five minute intervals for the short protocol (playback after 5 minutes) and the long protocol (playback after 10 mintues). A detection is defined as the first time an individual Bicknells Thrush was detected on any given run of a route. The total number of stops where the long protocol was employed was 280 (56 routes x 5 stops), and the total number of stops where the short protocol was employed was 245 (49 routes x 5 stops). A correction factor of 0.875 (245/280) was therefore applied to the number of detections under the long protocol. The chi-square test compares the number of detections in the two periods of the short protocol with the number of detections in the first two periods of the long protocol. Short Protocol N % Long Protocol % (first two % (all 3 periods only) periods) 27

1-5 minutes 6-10 minutes 11-15 minutes Chi-square =
N/A N/A 46 1.70, df = 1, p > 0.10
Table 6. Number (percent in brackets) of stops with Bicknells Thrush present and absent for which conifer, hardwood and understorey stem densities were categorized as dense, moderate or sparse. Conifer stem density Present Absent Dense 26 (55.3) 15 (34.9) Moderate 10 (21.3) 21 (48.8) Sparse 11 (23.4) 7 (16.3) Chi-square = 7.58, df = 2, p = 0.02 Hardwood stem density Present Absent 4 (8.5) 2 (4.7) 11 (23.4) 6 (14) 32 (68.1) 35 (81.4) 2.1, df = 2, p = 0.35 Understorey stem density Present Absent 23 (48.9) 12 (27.9) 4 (8.5) 17 (39.5) 20 (42.6) 14 (32.6) 12.41, df = 2, p = 0.002
Table 7. Number (percent in brackets) of stops with Bicknells Thrush present and absent that fell into four categories of canopy height. BITH present 1-3 m 18 (38.3) 3-5 m 24 (51.1) 5-7 m 4 (8.5) 7-9 m 1 (2.1) Chi-square = 6.19, df = 3, p = 0.10 BITH absent 9 (20.9) 22 (51.2) 11 (25.6) 1 (2.3)

page 18

Table 8. Number (percent in brackets) of stops with Bicknells Thrush present and absent that fell into three categories of canopy dominance. BITH present BITH absent Conifer & hardwood codominant 11 (23.4) 6 (14) Hardwood dominant 2 (4.3) 2 (4.7) Spruce/fir dominant 34 (72.3) 35 (81.4) Chi-square = 1.31, df = 2, p = 0.52

page 19

Figure 1. Our five study areas in Cape Breton Highlands National Park (grey border) and Cape North, NS. Stops along survey routes are illustrated by filled circles (Bicknells Thrush present) and open circles (Bicknells Thrush absent). Figures 2-6 (below) show each of the five study areas in detail.

Figure 2

page 20

Figure 3

Figure 4

page 21

Figure 5

Figure 6

page 22

Mean # BITH

Long am (8)

Long pm (8)

Short am (9)
Figure 7. Change in the mean ( SE) number of Bicknells Thrushes detected per route on the first and second run of a given protocol (separated by 6.5 days on average). Sample sizes are given in brackets.
Mean of max BITH detected
0 first (14) second (14) third (8) fourth (9) Run Number
Figure 8. Change in the maximum number of Bicknells Thrushes detected per route with successive runs. Data are from the long protocol only, and are averaged across routes. Samples sizes are given in brackets.

High Elevation Bird Monitoring 46.64262 46.64239 46.64246 46.64125 46.7213 46.7192 46.71698 46.71488 46.72345 46.73068 46.73132 46.73217 46.73379 46.73532 46.72903 46.72681 46.72903 46.72681 46.72524 46.72361 46.722 46.72036 46.71875 46.71685 46.73803 46.74025 46.74246 46.74467 46.74554 46.82435 46.82311 46.8216 46.82035 46.82003 46.82593 46.82581 46.82605 46.82494 46.82374 46.82618 46.82767 46.82973 46.83161 46.83267 46.71524 46.71378 46.81313 46.81667 46.81982 -60.9057 SL3-002 -60.9089 SL3-003 -60.9122 SL3-004 -60.9149 SL3-005 -60.8404 RT2-002 -60.8393 RT2-003 -60.8387 RT2-004 -60.8376 RT2-005 -60.8396 RT2-001 -60.5354 TL1-001 -60.5327 TL1-002 -60.5297 TL1-003 -60.5274 TL1-004 -60.525 TL1-005 -60.5378 TL2-005 -60.5378 TL2-004 -60.5401 TL2-003 -60.5425 TL2-002 -60.5447 TL2-001 -60.547 TL3-001 -60.5493 TL3-002 -60.5519 TL3-003 -60.5542 TL3-004 -60.5567 TL3-005 -60.8809 SK1-001 -60.8805 SK1-002 -60.881 SK1-003 -60.8814 SK1-004 -60.8844 SK1-005 -60.4649 PL1-001 -60.4677 PL1-002 -60.4701 PL1-003 -60.4729 PL1-004 -60.4761 PL1-005 -60.4496 PL2-001 -60.4529 PL2-002 -60.4562 PL2-003 -60.459 PL2-004 -60.4618 PL2-005 -60.4463 PL3-001 -60.4439 PL3-002 -60.4427 PL3-003 -60.4409 PL3-004 -60.438 PL3-005 -60.4876 LONG-L -60.4806 LONGL2 -60.4747 LONGL3 -60.463 BRA PON -60.5115 LOI TRA Simeon Lake 3 Simeon Lake 3 Simeon Lake 3 Simeon Lake 3 Radio Tower 2 Radio Tower 2 Radio Tower 2 Radio Tower 2 Radio Tower 2 Tipover Lake 1 Tipover Lake 1 Tipover Lake 1 Tipover Lake 1 Tipover Lake 1 Tipover Lake 2 Tipover Lake 2 Tipover Lake 2 Tipover Lake 2 Tipover Lake 2 Tipover Lake 3 Tipover Lake 3 Tipover Lake 3 Tipover Lake 3 Tipover Lake 3 Skyline Skyline Skyline Skyline Skyline Paquet Lake 1 Paquet Lake 1 Paquet Lake 1 Paquet Lake 1 Paquet Lake 1 Paquet Lake 2 Paquet Lake 2 Paquet Lake 2 Paquet Lake 2 Paquet Lake 2 Paquet Lake 3 Paquet Lake 3 Paquet Lake 3 Paquet Lake 3 Paquet Lake 3 Incidental Incidental Incidental Incidental Incidental no no no no yes yes yes yes no no no no yes no yes yes yes no no no yes yes yes yes no no no no no no yes yes yes yes no no yes yes yes yes no yes no no yes yes yes yes yes

page 25

tipover cabin and TL2-005
Long Lake (Glasgow Lakes) Long Lake (Glasgow Lakes) Long Lake (Glasgow Lakes) Branch Pond Lake of Islands Trail Junction