Stand observations coincided with the courtship, egg laying, and early incubation nesting stages and consisted of looking for goshawk activity from a fixed location 100-500 m from a known or suspected nest stand. Observation sessions lasting 2-4 hr were conducted 1-3 times per historic site between 1 April and 13 May. Upon detecting goshawks during stand observations, we estimated the age, sex, and behavior of the bird and estimated and mapped any likely activity center. We did not reexamine stands where goshawks were detected until egg hatching to reduce possible disturbance during egg laying and incubation, when nest desertion is most likely (Grier and Fife 1987). If no activity was detected from the fixed location, we quietly walked through the stand looking for goshawks, nests, molted feathers, feces, and prey remains. The observer typically covered =100 ha at this time by walking informal survey routes.
Alarm call broadcasts consisted of playing adult female goshawk alarm calls at stations 1300 m apart (Kennedy and Stahlecker 1993) and were conducted between 14 May - 10 July (from late incubation through fledging). These dates represent a trade-off: detection rates at successful nests are consistently lowest during incubation (Watson et al. 1996), however, early detection of occupancy in historic nest stands is critical for investigation of goshawk habitat use. To balance these needs, we sometimes conducted the first broadcast trial in a nest stand before some or all of the eggs hatched. At each call station the observer broadcast adult female goshawk alarm calls for 60-sec during 3-min observation trials (Kennedy and Stahlecker 1993). Alarm calls were amplified using a cassette tape, Sony Walkman, and a modified Realistic bullhorn to broadcast 85 decibels, the loudest volume that did not discomfort to the observers' ears (Watson et al. 1996). During a trial, three 10-sec broadcasts were directed at 60°, 180°, and 300° from the direction of travel followed by a 30-sec observation period. The surveyor then broadcast a 10-sec call at 60° and observed for 30-sec finally repeating this sequence at 180° and 300°. A call station was completed after the final 30-sec observation period. Breeding goshawks consistently respond to alarm call broadcasts within 150 m of a nest (Kennedy and Stahlecker 1993) and have been shown to aggressively respond to alarm call broadcasts in western Washington (Flatten and Swingle 1989, Watson et al. 1996). However, the probability and intensity of response from non-breeding goshawks has yet to be quantified. Thus, while walking between stations the surveyor looked for raptors, nests, feces, molted feathers, pluck posts, and prey remains.
Beg call broadcasts consisted of playing juvenile goshawk food-begging calls at stations =300 m apart (Kennedy and Stahlecker 1993) and were conducted between 11 July - 20 August (during the fledgling stage). Procedures at each station during beg call broadcasts were the same as those used during alarm call broadcasts.
Upon contacting a goshawk during an alarm or beg call broadcast survey, we estimated its age, sex, behavior, and aggressiveness and plotted its location on a 1:24000 scale U.S. Geological Survey (USGS) map. Following a response we attempted to determine the birds' breeding status and locate any nests. Detection of aggressive adults or fledglings resulted in intense local searches for nests. Detection ofnon-aggressive adults led the observer to search along the bird's flight path. If no further contacts or discovery of a breeding attempt occurred the surveyor resumed the protocol survey.
Watson et al. (1996) reported the cumulative probability of detecting a goshawk at successful nests was 100%, at a call station 100 m from an active nest after four broadcast trials and that the greatest probability of locating a breeding site occurred during the fledgling stage. Our survey results supported this finding. Thus, we define a historic stand as completely surveyed after four broadcast trials, including one trial conducted during the fledgling stage.
Initially, in 1996, we attempted a broad-based survey, visiting many historic goshawk sites but searching small irregular survey plots. During alarm call broadcasts in 1996, call stations were not mapped a priori but determined and completed during observations. Our goals during this time were to become familiar with the historic site, search for historic nests, and increase our chances of finding feathers, feces, or prey remains along with eliciting goshawk responses to the broadcasts. To achieve these objectives surveyors walked informal routes, at times detouring to examine pluck posts, snags, or perch sites for raptor sign. Call stations were always = 300 m apart and typically located within or near the edge of mature forest. Stands of younger seral stage (< 50 yrs) were at times bypassed to increase surveys in mature forests.
By July of 1996, we developed familiarity with the topography and access at many historic sites and refined our methods to reduce sampling bias and provide a more quantitative description of our survey efforts. Although goshawks in Washington are known to nest in mature forests (Fleming 1986. Wagenknecht et al. 1998), other data indicate that breeding pairs may select forests as young as 40 years old for breeding (Bosakowski and Vaughn 1996). Moreover, description of habitats used by goshawks for foraging and other diurnal activities are lacking in western Washington. therefore, we focused on conducting standardized broadcast surveys across all habitat types surround historic nest sites. Beginning with beg call broadcasts, on 11 July 1996, we mapped a Survey plot at each site using a known nest or activity center as a focus point. Beg call Survey plots in 1996, were defined as a circular area of I70ha, the estimated Post-fledging Family Area (PFA, Reynolds et al. 1992). Call stations were established at points 300 m apart on parallel transects spaced 260 m apart (Kennedy and Stahlecker 1993). Stations on adjacent transects were staggered by 130 m to optimize broadcast coverage (Joy et al. 1994).
In 1997, we continued to use an area approach but enlarged our survey plots to 314 ha circles. Since goshawks may use alternate nest stands up to 0.8 km apart within their breeding home range (Reynolds et al. 1992) we surveyed a circular area of 314 ha (1 km radius) to estimate goshawk occupancy with greater confidence and to examine many potential alternate sites for each pair. Each call station in the survey plot was completed during at least one survey trial. The entire 314 ha plot was surveyed between 14 May and 20 August 1997 using alarm or beg call broadcasts to coincide with the breeding chronology.
Productivity Counts Territories determined to be productive were revisited at least once in August to count the number of fledglings. During these visits, we remained in the nest stand for a minimum of two hours and broadcast beg calls to elicit responses from fledglings. We examined the ground surrounding the nest for any sign of fledgling mortality and for other signs of fledgling presence such as: down feathers, prey caches, and feces. Prey remains were collected at each site for future analysis and cursory habitat measurement was conducted during these visits.
Capture. Handling and Marking To capture, band, color mark, and radio-tag goshawks in the study area we used dho-gaza nets (Hamerstrom 1963, Bloom et al. 1992), bal-chatri traps (Berger and Mueller 1959), Swedish goshawk traps (Meng 1971), bow nets (Meredith 1943), and a modified bal-chatri using a dead lure (Table 4). We attempted to trap adult goshawks using dho-gaza nets in six nest stands in 1996 and 10 stands in 1997. A live Great-homed Owl (n = 16 stands) (Bubo virginianus) and a Rock Dove (n = 1 stand) were used as lures. We employed 10.2 cm mesh dho-gaza nets measuring 2.5 x 2.5 m attached by clothespins to 3-m tall poles. Nets were weighted with 4-oz weights (2 per net) attached to the bottom corners of each net with 20-ft long drag lines. All nets, poles and attachments were painted black. We placed 2-3 nets at 90° angles in a suitable location = 50 m from a nest during the nestling or post-fledging stage. Nets were set between 0-2 m above the ground and the lure bird was staked to the ground approximately 1.5 m in front of the nets to a perch near ground level (Bloom et al. 1992). Typically, we set the dho-gazas in a stand before first light but on several occasions set up the net during midday. The nets were monitored continuously from a blind of camouflage cloth and natural cover within 20 m of the trap.
We used bal-chatri traps to capture unmarked fledglings in 1996. The bal-chatri's were constructed of 1/2" hardware cloth formed into a 12 x 8 x 6" quonset-shaped cage and painted flat black. Colorless monofilament nooses (20 lb. test, 5-cm diameter) were tied to the top of the trap at 5-cm intervals using a slip knot for the noose. We used European Starlings (Sturnus vulgaris) or Rock Doves (Columba livia) as lures. Initially, we attempted to trap in the morning and late afternoon but later used bal-chatris throughout the day after our first capture occurred in the late morning. During our early attempts to trap, we positioned 1-3 bal-chatris around a blind sometimes in conjunction with 1-2 bow nets and remained stationary throughout the trap session. When this strategy (35 trap hours) failed to attract the fledglings we attempted a-more mobile strategy by presenting the trap and lure to the fledgling. The trapper would move through the forest toward a fledgling and place the live lure and trap (20 trap hours), sometimes along with the dead lure bal-chatri (8 trap hours), in view of the bird and then hide nearby. We placed the traps on an elevated log or on the ground in an accessible location and weighted each with three 4-oz weights secured by 20-ft drag lines. When set, traps were continuously monitored from a permanent blind or a combination of camouflage cloth and available natural cover within 50 m of the trap(s).
A Swedish goshawk trap was set in three stands (Burnt Mountain, Donkey Creek, and the Hole) in 1996 where an adult goshawk was detected in May or June and where no further evidence of occupancy was detected. The Swedish goshawk trap was constructed according to the design described by Meng (1971).
We employed two bow nets (13 trap hours) to capture unmarked fledglings in 1996 using a European Starling or Rock Dove as bait. The bow nets were made of an circular aluminum frame l-m in diameter painted black and covered with 2.5 cm mesh gill net. One bow net was triggered manually and the other was equipped with a remote control firing mechanism. Traps were placed on accessible open ground in a stand with confirmed presence of fledgling goshawks and monitored continuously from a blind < 30 m away.
We attempted to capture fledgling goshawks in two stands (8 hr) on a bal-chatri trap with a dead lure. This trap consisted of a dead rock dove, plucked and eviscerated, and placed in a 1/2" hardware cloth 'sandwich' which was painted black. We tied 5-cm diameter nooses of colorless monofilament (20 lb, test) 5-cm apart on the top of the trap using a slip knot as the noose. We used this trap in conjunction with a bal-chatri containing a live dove. We placed and anchored both traps in view of the fledgling on accessible substrates, then hid 15-20 m away. We monitored traps continuously until it was clear that the fledgling had left the vicinity of the traps.
We climbed nine active nest trees (two in 1996, seven in 1997) to band the nestlings and measure the nest structure. Nest trees at Raney Creek and Dungeoness were climbed using tree climbing spurs. The climber ascended the nest tree using girth hitched slings and a 150 m climbing rope for protection. All other trees were climbed using direct aid. A small lead line was propelled through the mid-canopy with a crossbow and used to haul a 150 m climbing rope into a secure position near the active nest. The climber then ascended the rope using locking ascenders.
When an individual was captured, it was immediately restrained and hooded Bloom 1987, Maechtle 1998). Each goshawk was banded with a USFWS aluminum band on one leg and colored alpha-numeric band on the other. Color bands were alternated on individuals of a breeding pair to facilitate identification. We measured each individual's weight (g, using a Pesola scale), unflattened wing chord (cm), tail length (cm), tarsus depth, halux length, middle toe length, footpad, and bill depth (mm). Age and sex was estimated based on plumage, wing length, and tail length (USDI 1977). We measured wingspan and body circumference and traced the right wing of 17 adults in 1997. We attached 18 g radio transmitter backpack packages (transmitter, and Teflon strap harness following the design of Buehler et al. 1995) to 18 adults (two in 1996, 18 in 1997) and to two fledgling goshawks at the Raney Creek breeding territory in 1996. Two adults (at Raney Creek) were refitted with backpack transmitters in 1997 because they bit off the antenna during the winter of 1996-97. The transmitters were designed to last 2-3 yr and were equipped with a motion sensor mercury switch that emitted a fast pulse when the transmitter unit was horizontal and a slow pulse when the unit was vertical. Harnesses were secured with dental floss stitches anteriorly and with copper tube clamps adjacent to the transmitter on the back. This backpack design has been shown to have no negative influence on Prairie Falcon (Falco mexicanus) productivity (Vecasy et al. 1996). Typically, goshawks were released as soon as measurement and attachment was completed, although when attempting to capture both members of a breeding pair the first individual caught was held until its mate was caught or the trapping session ended.
We monitored each goshawk for at least one hour after release to determine if handling or tagging affected the bird's flight or behavior. Birds were tracked and observed again when possible within two days of tagging to determine if the transmitter was still working and if the bird was behaving normally.
Radio-tracking Radio tagged adults at Raney Creek (1996 and 1997) and at Dungeoness and Lily Creek (1997) were tracked intermittently between 1 June and 14 August 1996 and 15 April and 8 August 1997. We tracked and located adults between 0630-2100 to assess use of available habitat during the nestling and post-fledging stages. We used 2- and 3-element, hand held, Yagi antennas with programmable ATS and Telonics receivers. Two attempts were made (at Raney Creek in 1996) to obtain accurate locations of the adults using triangulation from fixed points. Also, two attempts were made to assess accuracy of triangulation on beacon transmitters placed in the Raney Creek territory in 1996. Directionality of signal receptions from fixed points was too inaccurate for remote triangulation due to topography and vegetative screening. Therefore, determination of goshawk locations was achieved by homing toward a bird on foot. We homed in on the target goshawk and determined by visual observation or signal strength the bird's location. if the target bird was not visible, we continued to approach the direction with the strongest signal. We estimated we were 1100 m from the target bird when the signal strength was equal in all directions. This 100 m estimate was derived from field trials using signal reception from beacons set in an elevated position in the nest stand. We determined the behavior of the birds by direct observation where possible. When the target bird was not observed we recorded the signal pulse speed and changes to estimate the goshawks behavior as follows: stationary (steady slow pulse), alternating posture (slow pulse with intermittent fast pulses), alternating flight (fast pulse with intermittent slow pulses), or flying (fast pulses). Locations were then plotted on a 1:24000 USGS quadrangles or on habitat classification maps provided by the timber company, Rayonier) when available. Universal Transverse Mercator (UTM) coordinates were determined for all location estimates. We used RANGES V (Kenward and Hodder 1995) to assess the effectiveness of our telemetry efforts (Figure 3) and to analyze the spatial component of the data.
We measured vegetation and topographic attributes in 16 nest stands using a simplified USFS Region 6 Timber Stand Exam (USDA Forest Service 1989) to determine habitat features within each historic nest stand. We define the nest stand as the homogenous forest patch surrounding a nest or cluster of nests. Stand boundary lines were scribed along ecotones surrounding the nest(s) and ground truthed in the field. Vegetation was measured at 12 - 14 points systematically placed in the nest stand. At each point two concentric plots were established: a variable radius plot to sample trees > 5.0" DBH (poletimber and sawtimber) and a fixed radius plot to sample trees = 5.0" DBH (regeneration). Trees of the appropriate size for sampling on variable radius plots were determined by angle gauge. Each tree in the variable radius plot had the following information recorded: species, DBH, total height, crown ratio, crown class, and level of mistletoe infection. The first tree of each species on each plot < 28" DBH was cored forage and ten-year radial growth. All trees on the fixed radius plot were grouped by one inch diameter class and species, and average values of height, crown ratio, crown class, and mistletoe infection recorded. Crown density was measured at all vegetative plots using a moosehorn. Density was measured at the plot center by facing each cardinal direction. These four measurements were then used to attain an average crown density for that point. Plant association was assigned to all vegetative plots following Henderson et al. (1989). Due to the subjective nature of some measurements, all vegetation was measured by one of two observers, one of whom trained the other. Each measure was averaged per point and per stand for subsequent analysis..