ARE DAWN VOCALIZATION SURVEYS EFFECTIVE FOR
MONITORING GOSHAWK NEST-AREA OCCUPANCY?
SARAH R. DEWEY,1, 2 U.S. Forest Service, Ashley National Forest, Vernal Ranger District, 355 North Vernal Avenue, Vernal, UT
84078, USA
PATRICIA L. KENNEDY, Eastern Oregon Agricultural Research Center, Union Experiment Station Oregon State University, 372
South Tenth Street, P.O. Box E, Union, OR 97883, USA
ROBERT M. STEPHENS, Wyoming Cooperative Wildlife Research Unit, University of Wyoming, Laramie, WY 82071, USA
Abstract: In 1999 and 2000, we used dawn vocalization surveys (Penteriani 1999) to confirm nest-area occupancy by
northern goshawks (Accipiter gentilis atricapillus) breeding in montane forests of northeastern Utah, USA. We visited 19–20 historic nest areas in 1999 (n = 20) and 2000 (n = 19; 15 were sampled both years) during the courtship
period (mid-Mar through late Apr), and recorded the type and duration of calls heard. We detected vocalizations
at 51% of the nest areas surveyed with this technique (45% in 1999 and 58% in 2000). Overall accuracy rate (nest
areas in which we correctly determined occupancy status) was 90%. On average, goshawks vocalized over a 62-min
period at dawn, but durations ranged from 3 to 117 min. We found dawn vocalization surveys to be an effective
technique to determine the status of known goshawk nest areas during the courtship/pre-incubation periods,
although the technique requires considerable effort when applied to remote, montane forest locations. With only
minor modifications to the Penteriani (1999) protocol, the technique can be easily adapted to monitor status of
known goshawk nest areas in any area where local knowledge of nesting chronology is available.
JOURNAL OF WILDLIFE MANAGEMENT 67(2):390–397
Key words: Accipiter gentilis, acoustic lure, courtship, nest-site occupancy, raptor monitoring, Utah, vocal survey
techniques.
Some land management agencies currently
monitor occupancy of known northern goshawk
(hereafter, goshawk) nest areas as a way to track
population trends (Kennedy and Andersen
1999). Because the goshawk is secretive and often
difficult to locate, monitoring can be time-consuming and expensive. Several recent papers
have suggested using dawn vocalizations to detect
breeding raptors, including accipiters (Stewart et
al. 1996, Penteriani 1999, Penteriani et al. 2000).
This technique holds promise for monitoring
goshawk nest-area occupancy because of its efficiency, relatively low cost, and simplicity. Proper
implementation of dawn vocalization surveys also
has potential to reduce some of the bias inherent
to monitoring occupancy. Specifically, vocalization surveys are initiated prior to egg laying,
which improves detection of nests that fail early
in the season or where no nest attempt was made
(S. Patla, Northern Rockies Conservation Cooperative, personal communication). While this
method has been successfully used to detect
European goshawks (A. g. gentilis) in France
(Penteriani 1999), it has not been tested for
goshawks in North America. Vocalization surveys
1 Present address: U.S. Forest Service, Bridger-Teton
National Forest, Blackrock Ranger Station, P.O. Box
278, Moran, WY 83013, USA.
2 E-mail: [email protected]
in Europe mainly have been used to monitor raptor populations in fairly accessible urban and
rural areas. We expand the use of vocal surveys to
remote, high elevation forests where goshawks
live in western North America. In 1999 and 2000,
we used this method in a western, forested landscape to determine the (1) optimal time to conduct surveys, (2) efficacy of this method, and (3)
types of calls heard and duration of calling bouts.
STUDY AREA
Our study area was located in the Uinta Mountains of northeastern Utah (described in detail in
Dewey and Kennedy 2001). All nest sites were
located on National Forest System lands administered by the Ashley National Forest (ANF). Nest
sites were in homogenous lodgepole pine (Pinus
contorta) stands or mixed stands of aspen (Populus
tremuloides) and lodgepole pine. Nest-site elevations ranged from 2,335 to 2,800 m. Average
annual precipitation throughout the Uinta
Mountains is 70 cm (range 40–90 cm), with
roughly equal precipitation from winter snowfall
(Nov–Apr) and summer rains (May–Oct; Ashcroft et al. 1992). In 1999, snowfall was very heavy
during April. Weather stations at the eastern end
of the Uinta Mountains recorded snowpack
depths 70–110% of average on 1 March and
>130% of average on 1 May (Snotel Data,
http://www.wcc.nrcs.usda.gov/water/snow/snow
390
J. Wildl. Manage. 67(2):2003
_rpt.html). In 2000, snowpack depths were closer
to average. A network of snowmobile trails provided winter access to the more remote portions
of our study area. Nest sites were then approached
via cross-country skis or snowshoes.
METHODS
Definition and Delineation of Nest Sites
and Areas
Our sampling unit was a goshawk nest site, which
we define following Siders and Kennedy (1996)
as the area surrounding the nest tree, including
vegetation and topographic features used by a
nesting pair. A nest site may include >1 alternate
nest. We define nest area as the landscape area
encompassing all known nests used by an individual or goshawk pair (Siders and Kennedy 1996).
We had a limited number of banded birds, so we
used Siders and Kennedy’s (1996) approach to
identify which nest sites belonged to a nest area.
Distances between known neighboring active
nests were calculated by year (1993–1998) for an
intensively surveyed block (Siders and Kennedy
1996). The shortest mean distance between neighboring active nests was 1,977 m in 1993. We also
examined the mean distances among all alternate
nests within a nest area for all nests found between
1991 and 1998. The greatest mean inter-alternate
nest distance was 1,582 m, which is less than the
minimum separation between adjacent nest
areas. Thus, during 1999 and 2000, all historic
nest trees were considered to be within the same
nesting area if their separation was <1,977 m.
The goshawk nest areas sampled in our study
were part of a large pool of known nest areas
(>50) that have been intensively monitored since
the early 1990s. Initially, nest sites (within nest
areas) were located by crews conducting goshawk
surveys or based on reports of goshawks or nests
from agency personnel or the public between
1991 and 1998. The known nest sites do not represent a random or systematic sample or census
of those on the ANF.
We conducted dawn surveys at individual nest
sites within nest areas and extended occupancy
status to the nest area as whole as multiple nest
sites were surveyed. In 1999, we conducted dawn
surveys at 27 nest sites located within 20 nest
areas. In 2000, we sampled 23 nest sites located
within 19 nest areas. We sampled 31 nest sites
within 24 unique nest areas; 15 of these nest areas
were sampled during both years. We treated the
15 nest areas sampled in both years as indepen-
GOSHAWK DAWN VOCALIZATION • Dewey et al.
391
dent and examined results for years separately.
We believe that this treatment is justified because
(1) our sample is drawn from an intensively monitored pool of nests, where we had considerable
prior knowledge of historic nest sites used and
occupancy status in both years of this study; (2)
only 1 misclassification occurred at a nest site not
previously discovered and then sampled in the
subsequent year; (3) the error rate calculated
from a random sample of 1 of the 2 data points
from each of the 15 nest areas surveyed in both
years compared to the combined error rate for
both years; and (4) the sample of nests was randomly drawn each year.
Adjustment of Survey Timing
We conducted dawn vocalization surveys at
known goshawk nest sites from mid-March
through late April 1999 and 2000. Our sampling
period differed from that recommended by Penteriani (1999; Feb–Mar) because data from 2
studies of radiomarked birds in the Uinta Mountains indicated that most goshawks did not return
to their nest areas until mid-March (Stephens
2001; K. M. Paulin and S. R. Dewey, U.S. Forest
Service, unpublished data). Some goshawks
remain on or close to their nest areas year-round,
but most appear to move down in elevation and
usually return to their nest areas in March
(Stephens 2001; K. M. Paulin and S. R. Dewey,
U.S. Forest Service, unpublished data). In neighboring Wyoming, radiomarked goshawks that left
their nest areas for the winter, returned to them
between 23 March and 12 April (Squires and Ruggiero 1995). Our nest phenology data also indicated that a later sampling period would be appropriate. To estimate hatching and clutch initiation
dates, we back-calculated from the age at which
nestlings were banded during the 1996 and 1997
breeding seasons (S. R. Dewey, U.S. Forest Service; and P. L. Kennedy, Colorado State University; unpublished data). Eggs hatched between 31
May and 20 June in 1996 and between 6 June and
20 June in 1997. Using Palmer’s (1988) estimate
of 32–34 days for incubation, egg-laying would
have occurred from late-April through mid-May
(roughly 27 Apr–16 May in 1996 and 3 May–16
May in 1997) and courtship activity peaks during
the 30 days prior to egg laying (Møller 1987).
Dawn Vocalization Survey Protocol
We either selected an observation point within
100–200 m of the last used nest site or, if the nest
area contained several alternate nest sites, cen-
392
J. Wildl. Manage. 67(2):2003
GOSHAWK DAWN VOCALIZATION • Dewey et al.
trally located an observation point among the
nest sites, but within 100–200 m from each known
site. If alternate nest spacing precluded locating
observation points within 200 m of all known
alternate sites, either 2 observers conducted the
dawn vocalization survey (each positioned
100–200 m from 1 or more of the sites), or surveys
were conducted on different days starting with
observations at the last occupied site. Given our
desire to test the dawn survey method using a
larger sample of nest areas (vs. sampling all alternate nest sites within a nest area), not all known
nest sites within nest areas were surveyed.
Observers arrived at the observation point 30–45
min prior to sunrise and remained approximately 1–1.5 hr after sunrise. If goshawks were still
vocalizing at the end of the observation session,
observers continued to record calls until birds
had been quiet for 5 min. If goshawks were not
detected during the first observation session, we
returned to some nest sites for 1–2 additional
observation periods, as time allowed (6 nest areas
were not revisited each year). We separated repeat
visits to nest areas by ≥1 week and did not conduct
surveys during inclement weather conditions. We
recorded the observation point location; weather
conditions; survey start and end time; and call
type (chattering or wail; Penteriani 2001), time,
and duration during the dawn survey.
When we did not detect goshawks during
courtship, we revisited the nest area during the
incubation period (May–mid-Jun) to verify status.
We visited all old nest sites and examined an area
within 200 m of each nest for signs of use. If no
goshawks or sign implying use (e.g. reconstructed nests, fresh prey remains, molted goshawk
feathers, copious amounts of mute) were observed, we conducted systematic broadcast surveys
(Kennedy and Stahlecker 1993, Joy et al. 1994),
either during the nestling or fledgling periods,
within at least an 800-m radius of the last used nest
site. If we did not detect goshawks during broadcast surveys or find sign indicating current use of
the nest site by goshawks, we considered the nest
site unoccupied. We used these follow-up surveys
to assess accuracy of the nest-area occupancy estimates obtained from the dawn surveys.
RESULTS
Dawn Vocalization Surveys
We conducted a total of 56 dawn vocalization
surveys during 1999 (n = 32) and 2000 (n = 24). We
sampled 20 nest areas in 1999 and 19 in 2000 (24
unique nest areas). Only 1 dawn survey was conducted within 10 nest areas in 1999 and 14 in 2000.
Occupancy was confirmed based on the results of
single surveys in 4 nest areas in 1999 and 8 in 2000.
Multiple or repeat visits to nest sites occurred within 5 nest areas in 1999 and 2 in 2000. Two observers
conducted simultaneous visits to separate nest
sites within 5 nest areas in 1999 and 3 in 2000.
We conducted surveys 16 March–27 April 1999
and 24 March–27 April 2000 (Table 1). At 20 of the
39 nest areas (51%), we heard goshawk vocalizations, and we considered these nest areas occupied.
We later confirmed nesting activity in all of the
nest areas identified as occupied and also in 4 nest
areas where no vocalizations were heard during
dawn surveys (Table 2). Of the 39 nest areas sampled over 2 years with this technique, 24 (62%)
were occupied by goshawks. Occupancy rates
were 60% in 1999 and 63% in 2000. We correctly
identified occupancy status (either occupied or
unoccupied) at 35 of the 39 (90%) nest areas
using the dawn survey technique, but would have
incorrectly identified 4 (10%) as unoccupied had
we not performed follow-up surveys (Table 2).
Three of the misclassifications occurred in 1999
and 1 in 2000 (Table 2), and all occurred in the
first survey year for a nest area. Because some
goshawks may have gone undetected, the above
success and error rates represent the maximum
and the minimum, respectively.
We conducted only 1 dawn survey at each of the
4 nest areas initially identified as unoccupied.
Table 1. Results of goshawk dawn vocalization surveys conducted in the Uinta Mountains, Utah, USA, mid-Mar–late-Apr
1999 and 2000.
Year
na
1999 20
2000 19
Total 39e
No.
No.
Date first detected No. visits
detected
prior to
dawn detectionsb
surveys (occupancy Earliest Median detectiond
32
24
56
9 (0.45)
11 (0.58)
20 (0.51)
25 Mar 17 Apr
28 Mar 10 Apr
25 Mar 13 Apr
1.7 ± 0.24
1.0 ± 0.00
1.3 ± 0.13
a Number of nest areas at which dawn surveys were performed.
b Total number of nest areas in which goshawks were detected during dawn surveys.
c Predicted nest-area occupancy rate based on results of
dawn surveys.
d When 2 nest-site surveys within a nest area were conducted at widely separated sites simultaneously and a detection occurred, the visit on which the detection occurred was
assumed to be the earliest (e.g. visit 1 if first-time simultaneous visits, visit 3 if a repeat simultaneous visit).
e Sample size represents the number of nest areas sampled
over 2 years. We sampled 24 unique nest areas during our
study, 15 of which were sampled during both years.
J. Wildl. Manage. 67(2):2003
GOSHAWK DAWN VOCALIZATION • Dewey et al.
Table 2. Observed minimum error rates in predicting northern
goshawk nest-area occupancy using dawn vocalization surveys in the Uinta Mountains, Utah, USA, mid-Mar–late-Apr
1999 and 2000.
Years
sampled
Year
na
1-year only 1999
5
2000
4
Both years 1999 15
2000 15
Total
1999 20
Total
2000 19
Total years combined 39c
15
Random sampled
Predicted Occupancy
Classification
Occ.Unocc.- Unocc.- Error
Correct Correct Incorrect rateb
1
1
8
10
9
11
20
8.9
4
2
4
5
8
7
15
4.6
0
1
3
0
3
1
4
1.4
0.15
0.05
0.10
0.10
a Number of nest areas at which dawn surveys were performed.
b Error rate defined as the proportion of misclassified nest
areas.
c Sample size represents the number of nest areas sampled
over 2 years. We sampled 24 unique nest areas during our
study, 15 of which were sampled during both years.
d Represents a repeated sampling (n = 50) of 1 of the 2 data
points from each of the 15 nest areas sampled in both years.
Surveys at these sites occurred prior to 15 April.
In 1 of these nest areas, we later confirmed occupancy by goshawks at the nest site where the
dawn survey was performed. However, in the
other 3 nest areas, goshawks used alternate nests
(all but 1 of the nest sites were known from previous monitoring efforts), widely separated from
the surveyed site (x– = 695 m ± 253).
At 5 of 9 confirmed occupied nest areas in 1999,
2 surveys were required to correctly determine
occupancy status. Most (80%) of the second surveys (when goshawks were detected) occurred
after 15 April. In 2000, all detections were recorded on the first visit (Table 1). Dawn surveys at
sites where occupancy was determined by 1 visit
were conducted 25 March–19 April 1999 and 28
March–27 April 2000. In 1999, we first detected
goshawks during a dawn vocalization survey on 25
393
March and continued to hear dawn vocalizations
until the last observation session on 27 April
(Table 1). The peak in dawn vocalizations
occurred in late-April (median date 17 Apr; Table
1). Between 16 and 27 April, we heard vocalizations during 5 observation sessions. In 2000, we
first heard dawn vocalizations on 28 March and
also heard them during the last observation session on 27 April (Table 1). Median date of first
detection was 10 April in 2000 (Table 1).
In nest areas where goshawks were detected, we
first heard vocalizations 32 min before to 2 min
after sunrise. On average, we first detected vocalizations approximately 20 min before sunrise (0615
hr; Table 3). Vocalizations spanned 3–117 min (x– =
62), with roughly equal amounts of vocalizations
occurring before and after sunrise (Table 3).
Goshawks uttered a variety of calls during
courtship activities, but these could generally be
considered variations of the chattering (slow and
fast kek kek keks) or wail (whee-oo…wheeoo)
calls (Penteriani 2001). During observation sessions, we frequently observed goshawks copulating, and these events were usually accompanied
by intense chattering bouts by both birds. Both
males and females uttered calls, though knowing
which sex was calling was often difficult, unless
birds were marked and the marker could be observed. Vocalizations varied from short, 1-note
call segments of the chattering or wail, lasting
roughly 1 sec, to a series of chattering calls and
wails lasting up to 10 sec. Calls could be clearly
heard from our observation positions (up to 200
m), and could probably be heard at even greater
distances under ideal listening conditions.
DISCUSSION
Patterns in our accuracy rates (higher in year
2), occupancy rates (small annual difference relative to accuracy rates), and misclassifications
rates (all misclassifications occurring in the first
survey year and no misclassifications the second
Table 3. Timing and span of dawn vocalizations given by goshawks breeding in the Uinta Mountains, Utah, USA, mid-Mar–Apr
1999 and 2000.
Year
1999
2000
Years combined
a
b
Sample
size
na
9
10b
19
Time of first
vocalization (hr)
(x– ± SE)
0610:48 ± 0005:00
0622:19 ± 0007:08
0615:27 ± 0004:27
First vocalization relative
to sunrise (min)
(x– ± SE)
–0:25 ± 0:03
–0:16 ± 0:03
–0:21 ± 0:02
Number of dawn surveys during which goshawks were heard.
Nest area where distant vocalizations heard, not included in exploratory statistics.
Vocalization
span (hr:min)
(x– ± SE)
0:54 ± 0:05
1:13 ± 0:11
1:03 ± 0:06
394
J. Wildl. Manage. 67(2):2003
GOSHAWK DAWN VOCALIZATION • Dewey et al.
year) for nest areas sampled both years are consistent with what would be expected from nonindependent samples. Our sample was drawn
from intensively monitored nest areas, where we
had considerable prior knowledge of historic
nest-site use and occupancy. Of the 4 nest areas
that were initially misclassified as unoccupied,
only 1 misclassification occurred at a nest site not
known previously. Misclassifications at the other 3
sites were the result of goshawks using alternate
nest sites >400 m from the survey point near the
last used nest site. In these cases, the alternate
nest site was known but not surveyed. Thus, the
result of the previous year’s dawn survey (i.e. new
nest-site location) was only helpful the following
year in 1 instance. Even with the limitation of
non-independent samples, we believe that data
from nest areas sampled in both years is useful to
illustrate that dawn vocalization surveys require
follow-up monitoring to verify results. The error
rates should be interpreted cautiously, and we
acknowledge that they may be biased low.
Our results support Penteriani’s (1999) conclusions that dawn vocalization surveys are an efficient
technique for detecting the presence of breeding
goshawks at known goshawk nest sites. His accuracy rate with 21 goshawk nest areas tested in 1 year
was 100%, and our accuracy rate with 39 nest
areas tested over 2 years was 90% (Table 4). The
differences in accuracy rates may be a reflection of
differences in migratory status, nesting chronology, and influences of weather on breeding activities
and occupancy rates (Kennedy 1997, McClaren et
al. 2002). Nevertheless, accuracy is high.
Our observations confirm that vocalizations are
frequently heard at dawn during the courtship
period and support Penteriani’s (1999) recom-
mendations for survey timing, both diurnally and
seasonally. We successfully exploited the peak in
vocalization events during the courtship period
to detect goshawks. Given differences in migratory status and nesting chronology, actual survey
dates for goshawks breeding in western North
America will have to be adjusted to coincide with
local courtship phenology.
Similar to Penteriani (1999), we consistently
heard vocalizations in the interval from 30 min
before to 5 min after sunrise, suggesting that surveys covering that time period should be sufficient to detect the presence of goshawks. Penteriani (1999) suggested that unoccupied status
could be confirmed if no detections are recorded
after 2 observation sessions performed during
the month most distant to egg-laying or after 1
observation session during the 2 months preceding egg-laying. At the 4 misclassified nest areas
(Table 2), 1 dawn vocalization survey was conducted between 2–15 April; all dates within the 2month period preceding egg-laying. Using the
dawn vocalization survey method, we successfully
determined the status of the nest site closest to
the observation point (i.e., unoccupied) in 3 of
the 4 nest areas, but erroneously classified the
nest-area status. That we didn’t hear goshawks in
the above nest areas is not a failure of the technique per se; rather 3 important caveats were
highlighted in regard to using this technique to
document nest-area status.
First, since our surveys occurred early in April,
some birds possibly had not yet returned to their
nest areas from their wintering areas. Unlike
France, where birds are year-round residents (V.
Penteriani, Estación Biológica de Doñana, personal communication), and vocalizations can be
Table 4. Comparison of detection rates of dawn vocalization surveys conducted for breeding accipiters.
Study area—species
Survey
timing
Vancouver Island, Canada— Mar–Apr
Cooper’s hawk
Burgundy Region, France— Feb–Mar
European goshawk
Northeastern Utah, USA—
Mar–Apr
northern goshawk
a
Landscape
Migratory
status
Small tract,
Residenta
urban woodlands
Large tract,
Residentc
rural woodlands
Remote, montane Partial
forest
migrante
Mean call
duration Detection Occupancy
(min)
rate
rate
Source
17
Not tested
0.46b Stewart et al. 1996
N/Ad
1.00
1.00
Penteriani 1999
62
0.65
0.62f
This study
A. Stewart, Ministry of Sustainable Resource Management–Government of British Columbia, personal communication.
Based on an inventory of potential nest sites rather than known nest sites used in the other studies.
c V. Penteriani, Estación Biológica de Doñana, personal communication.
d Comparable data not available.
e Most goshawks move down in elevation (Stephens 2001).
f Estimate includes nest areas initially misclassified as unoccupied that were later determined to be occupied.
b
J. Wildl. Manage. 67(2):2003
heard near the nest throughout the year (Penteriani 1999), most goshawks in our study area probably are partial migrants. In 1 case where we
recorded a false negative, we conducted the dawn
vocalization survey on 3 April and heard no vocalizations. Follow-up monitoring surveys revealed a
breeding attempt, by a color-marked female
goshawk that had formerly occupied an adjacent
nest area for at least the 4 previous years (K. M.
Paulin and S. R. Dewey, U.S. Forest Service, unpublished data). An unbanded goshawk pair
occupied her former nest area during the 1999
breeding season. She possibly lost her original
nest area and did not find a mate or vacant nest
area until relatively late in the courtship period.
As a result, she may have made a relatively late
breeding attempt in the new nest area. Boal
(2001) observed that age of Cooper’s hawks
(Accipiter cooperii) was significantly related to timing of breeding with adult pairs nesting earliest,
pairs with 1 subadult later, and pairs with both
subadults nesting last. Weather also may influence
the timing of breeding activities. Our study area
received heavy snowfall during the courtship period in 1999, which may have resulted in courtship
activities being initiated later relative to 2000.
Thus, misclassification can occur for pairs whose
courtship phenology is different from median
dates used to design the sampling periods.
Second, many goshawk nest areas contain widely spaced alternate nests that pairs use on a rotational basis (Speiser and Bosakowski 1991, Woodbridge and Detrich 1994). In California, mean
distance between alternate nests was 273 m and
ranged 30–2,066 m (Woodbridge and Detrich
1994). In Arizona, mean inter-alternate nest distance was 489 m and the range was 21–3,410 m
(Reynolds and Joy 1998). Average distance
between alternate nests used by goshawks in the
Uinta Mountains, for 1992–1998, was 432 m. In 3
of the nest areas we later confirmed as occupied in
our study, all alternate nest sites were >400 m from
the observation point. If the goshawks had been in
their nest areas and vocalizing during dawn vocalization surveys, these long distances may have prevented us from hearing vocalizations. However,
during 1 visit in 2000, we heard distant goshawk
vocalizations from an observation point near the
previous year’s nest site. The observer approached
the vocalizing birds and found a new alternate nest
approximately 400 m from the observation point.
The maximum distance at which goshawks are
detectable during dawn vocalization surveys is
unknown and requires further investigation.
GOSHAWK DAWN VOCALIZATION • Dewey et al.
395
Finally, dawn vocalization surveys may miss lone
birds in a nest area if call rates are influenced by
the presence of a mate. Single goshawks may be
more detectable with broadcast surveys.
Our results confirm Penteriani’s (1999) assertion that nest-site occupancy can be determined
from 1 visit during the 2 months closest to egglaying with fairly high accuracy. However, we caution against determining nest-area occupancy status from 1 dawn vocalization survey, especially in
areas where the goshawk population is not resident, or where goshawks maintain widely separated alternate nests. We found that 2 visits/nest
area allowed us cumulatively to detect 79% of
occupied sites.
Detection rates, expressed as the number of
detections per number of surveys conducted,
provide a measure of the efficacy of the dawn survey technique, which affords a useful comparison
with detection rates of other survey techniques.
We used data from occupied nest areas only in
our calculations of detection rates because these
rates reflect the ability to detect something that is
actually present. Our detection rates varied from
47% in 1999 to 92% in 2000 and were 65% overall
(Table 4). Although our detection rates were
lower than Penteriani’s (1999), they compare to
goshawk detection rates performed during
courtship with broadcast experiments (alarm call
within 100 m of nest; New Mexico 66.7%
[Kennedy and Stahlecker 1993]; Vancouver
Island 40% [McClaren et al. in press]).
In comparison to conducting broadcast surveys
or checking known nest sites for sign during the
courtship period to determine occupancy, the
dawn vocalization survey method has some
strengths and limitations. Similar to what would
be encountered using those methods, accessing
nest areas presents logistical challenges. Accessing some of our nest sites required travel via several modes of transportation (e.g., truck, snowmobile, cross-country skis, snowshoes) well
before dawn. Navigating to nest sites in the dark
can be a challenge, particularly when traveling
cross-country through large, homogenous forest
stands. Inexpensive, handheld Global Positioning System (GPS) units were very helpful for
locating nest sites in the dark. Camping near survey sites also is an option. Penteriani (Estación
Biológica de Doñana, personal communication)
indicated that snowfall was very limited at his
study area in France, and in many years there was
no snow. Thus, spring access did not present the
same challenges in France as in our study area.
396
GOSHAWK DAWN VOCALIZATION • Dewey et al.
Similar to the broadcast methodology, dawn
vocalizations can be successfully used by observers
with limited training (only the ability to recognize
goshawk vocalizations). However, a tradeoff exists
between the number of nest sites that can be
checked per day and the ability to confirm unambiguously nest site status. A single observer can only
check 1 nest site/day using the dawn vocalization
survey technique. Whereas, observers could check
several nest sites in a day and confirm breeding
activity using broadcast surveys or sign. However,
the latter method may not always yield data as
convincing as actually hearing vocalizations
because old goshawk nests can be used by a variety
of other raptors (Patla 1997). Perhaps the greatest strength of dawn vocalizations is that fewer
goshawks will go undetected because occupied nest
sites are identified early. When surveys are conducted after courtship, goshawks that failed their
breeding attempt early may go undetected because
they are no longer tied to their nest area and/or
may be less apt to respond to broadcast calls.
Dawn vocalization surveys may also have some
potential as an inventory tool (Stewart et al. 1996)
for goshawks. Observers could conduct dawn
vocalization surveys in areas where previous sightings have suggested occupancy, but where no nests
have been found. The technique could also be
used to locate unknown alternate nest sites within
known nest areas. However, the utility of dawn
vocalizations for conducting random surveys of
large expanses of forest is unknown. Before dawn
vocalizations are used for this type of survey,
goshawk detectability would have to be tested at
observation points located at various distances
from active nest sites, similar to the approaches
used to test the effectiveness of using broadcast
surveys to locate breeding goshawks (Kennedy and
Stahlecker 1993, Watson et al. 1999). In addition,
dawn vocalization detection rates should be compared to detection rates of other techniques (e.g.,
broadcast surveys or checking sign at known nests)
using a design similar to that of Reid et al. (1999).
The area-occupied technique (Geissler and
Fuller 1986, Iverson and Fuller 1991) also could
be used to determine detection rates, although it
may require numerous visits per nest to obtain an
adequate number of replicates.
We recommend that managers use dawn vocalization surveys to determine the status of known
goshawk nest sites. As the status of multiple nest
sites within a nest area are confirmed, status of the
nest area as a whole can be determined. To use
this technique to monitor nest-site occupancy by
J. Wildl. Manage. 67(2):2003
goshawks in montane forests of western North
America, we suggest that surveys follow Penteriani’s (1999) protocol with 5 minor changes.
First, conduct dawn observation sessions during
the 45 days preceding egg laying. Survey times will
need to be based on knowledge of local breeding
phenology. Second, conduct observation sessions
during the peak vocalization period (45 min
before to 90 min after sunrise). We think being
settled at the observation location 45 min before
sunrise is important to reduce the potential for
goshawks altering their behavior in response to
the observer’s presence or movements. Third, if
no vocalizations are heard during the first observation session, return for a second survey at least
1 week but preferably 2 or more weeks later.
Fourth, select an observation position 100–200 m
from the last used nest. If several alternate nests
are clustered within 200 m of one another, choose
a central observation location. We suspect that
vocalizations may not be clearly or easily heard
over distances >200 m in dense coniferous forests.
Finally, if inter-alternate nest distance is >200 m,
several observers should be placed simultaneously
at separate observation points so that all observers
are within 200 m of an alternate nest structure.
MANAGEMENT IMPLICATIONS
Current goshawk monitoring programs on public lands emphasize broadcast surveys and nestsite searches during the nestling and fledgling
periods to monitor occupancy and productivity
of known goshawk nest areas. Timing of these
surveys does not overlap the courtship period
when goshawks are highly detectable. Thus, nestarea status can never be determined with confidence because some areas may be misclassified as
unoccupied, when in reality goshawks were present but either failed early or did not attempt to
nest, even if follow-up broadcast or nest-site
searches are performed. Some goshawk researchers have detected goshawks during dawn surveys
and then failed to document a nest attempt (S.
Patla, Northern Rockies Conservation Cooperative, personal communication). In combination
and when implemented according to a standardized sampling protocol, dawn surveys, nest-site
searches, and broadcast surveys can form the basis
for an effective monitoring approach. We recommend that land managers monitoring goshawk
populations use a nest-site-based approach (and
aggregate up to the nest area as multiple nest
sites are surveyed) and a standardized set of techniques, including dawn vocalization surveys. Ben-
J. Wildl. Manage. 67(2):2003
efits of using dawn vocalization surveys include a
reduction in bias of occupancy estimates and
ability to compare goshawk nest-area occupancy
across regions.
ACKNOWLEDGMENTS
We are indebted to T. Morris for his assistance
and persistence with follow-up monitoring and
broadcast surveys. We are also grateful to K. Paulin
for support of this project and her continuing support of our goshawk research, and S. Lewis for valuable input and discussions on this topic. C. Boal, E.
McClaren, S. Patla, K. Titus, and B. Woodbridge
provided useful comments on the manuscript.
Funding for this project was provided by U.S.
Department of Agriculture, U.S. Forest Service,
Ashley National Forest. Publication costs were generously contributed by Eastern Oregon Agricultural Research Center at Oregon State University.
LITERATURE CITED
ASHCROFT, G. L., D. T. JENSEN, AND J. L. BROWN. 1992.
Utah Climate. Utah Climate Center, Utah State University, Logan, Utah, USA.
BOAL, C. W. 2001. Nonrandom mating and productivity
of adult and subadult Cooper’s hawks. Condor
103:381–385.
DEWEY, S. R., AND P. L. KENNEDY. 2001. Effects of supplemental food on parental-care strategies and juvenile
survival of northern goshawks. Auk 118:352–365.
GEISSLER, P. H., AND M. R. FULLER. 1986. Estimation of
the proportion of an area occupied by an animal species. Pages 533–538 in Proceedings of the section on
survey research methods of the American Statistical
Association. American Statistical Association, Alexandria, Virginia, USA.
IVERSON, G. C., AND M. R. FULLER. 1991. Area-occupied
survey technique for nesting woodland raptors. Pages
118–124 in B. G. Pendleton, D. L. Krahe, M. N.
LeFranc, Jr., K. Titus, J. C. Bednarz, D. E. Andersen,
and B. A. Millsap, editors. Proceedings of the Midwest
Raptor Management Symposium and Workshop, Scientific and Technical Series 15. National Wildlife Federation, Washington, D.C., USA.
JOY, S. M., R. T. REYNOLDS, AND D. G. LESLIE. 1994. Northern goshawk broadcast surveys: hawk response variables
and survey cost. Studies in Avian Biology 16:24–30.
KENNEDY, P. L. 1997. The northern goshawk (Accipiter
gentilis atricapillus): is there evidence of a population
decline? Journal of Raptor Research 31:95–106.
———, AND D. E. ANDERSEN. 1999. Research and monitoring plan for northern goshawks in the western
Great Lakes region. Final report. Submitted to National Council of the Paper Industry for Air and
Stream Improvement, Inc., U.S. Forest Service, U.S.
Fish and Wildlife Service, Wisconsin Department of
Natural Resources, Michigan Department of Natural
Resources, and Minnesota Department of Natural
Resources. http://www.fw.umn.edu/CO-OP/download/default.html.
———, AND D. W. STAHLECKER. 1993. Responsiveness of
GOSHAWK DAWN VOCALIZATION • Dewey et al.
397
nesting northern goshawks to taped broadcasts of 3
conspecific calls. Journal of Wildlife Management
57:249–257.
MCCLAREN, E. L., P. L. KENNEDY, AND P. CHAPMAN. In
press. Efficacy of male goshawk food-delivery calls in
broadcast surveys on Vancouver Island. Journal of
Raptor Research.
———, P. L. KENNEDY, AND S. R. DEWEY. 2002. Do some
northern goshawk nest areas consistently fledge more
young than others? Condor 104:343–352.
MØLLER, A. P. 1987. Copulation behaviour in the goshawk,
Accipiter gentilis. Animal Behaviour 35:755–763.
PALMER, R. S., editor. 1988. Handbook of North American birds. Volume 4. Yale University Press, New
Haven, Connecticut, USA.
PATLA, S. M. 1997. Nesting ecology and habitat of the
northern goshawk in undisturbed and timber harvest
areas on the Targhee National Forest, Greater Yellowstone Ecosystem. Thesis, Idaho State University, Pocatello, Idaho, USA.
PENTERIANI, V. 1999. Dawn and morning goshawk
courtship vocalizations as a method for detecting nest
sites. Journal of Wildlife Management 63:511–516.
———. 2001. The annual and diel cycles of goshawk
vocalizations at nest sites. Journal of Raptor Research
35:24–30.
———, GALLARDO, AND H. CAZASSUS. 2000. Diurnal
vocal activity of young eagle owls and its implications
in detecting occupied nests. Journal of Raptor
Research 34:232–235.
REID J. A., R. B. HORN, AND E. D. FORSMAN. 1999. Detection rates of spotted owls based on acoustic-lure and
live-lure surveys. Wildlife Society Bulletin 27:986–990.
REYNOLDS, R. T., AND S. M. JOY. 1998. Distribution, territory occupancy, dispersal, and demography of northern goshawks on the Kaibab Plateau, Arizona. Unpublished final report for Arizona Game and Fish,
Heritage Project No. I94045.
SIDERS, M. S., AND P. L. KENNEDY. 1996. Forest structural
characteristics of accipiter nesting habitat: is there an
allometric relationship? Condor 98:123–132.
SPEISER, R., AND T. BOSAKOWSKI. 1991. Nesting phenology, site fidelity, and defense behavior of northern
goshawks in New York and New Jersey. Journal of Raptor Research 25:132–135.
SQUIRES, J. R., AND L. F. RUGGIERO. 1995. Winter movements of adult northern goshawks that nested in southcentral Wyoming. Journal of Raptor Research 29:5–9.
STEPHENS, R. M. 2001. Winter ecology of northern
goshawks in the Uinta Mountains. Thesis, University
of Wyoming, Laramie, Wyoming, USA.
STEWART, A. C., R. W. CAMPBELL, AND S. DICKIN. 1996. Use
of dawn vocalizations for detecting breeding Cooper’s hawks in an urban environment. Wildlife Society
Bulletin 24:291–293.
WATSON, J. W., D. W. HAYS, AND D. J. PIERCE. 1999. Efficacy of northern goshawk broadcast surveys in Washington state. Journal of Wildlife Management 63:98–106.
WOODBRIDGE, B., AND P. J. DETRICH. 1994. Territory occupancy and habitat patch size of northern goshawks in
the southern Cascades of California. Studies in Avian
Biology 16:83–87.
Received 29 October 2001.
Accepted 28 January 2003.
Associate Editor: Kelly.