Strong and others 



Chapter 32 



Distribution and Population Estimates in Oregon 



Location was determined by distance travelled through 

 the water between known landmarks on shore, using the 

 speedometer and trip log functions on a sonar fish finder. 

 Speed was maintained at approximately 8 knots at all times. 

 Other variables monitored included water temperature and 

 depth, presence of sonar scattering layers, rip currents, type 

 of shoreline (rocky, sandy beach, adjacent to river mouths, or 

 a combination of the above), association of murrelets with 

 other species, and weather conditions. Observation conditions, 

 as they affected the detectability of murrelets, were categorized 

 as excellent, very good, good, fair, and poor. Observation 

 conditions were classified based on Beaufort sea state, swell, 

 reflections, and fog. Surveys were not initiated at Beaufort 

 state 3 (fair observing conditions), and surveys were terminated 

 at Beaufort state 4 (poor observing conditions). The driver 

 alternated with observers periodically to reduce observer 

 fatigue, and a rest stop was taken at least every 3 hours. 



To quantify distribution along the length of the Oregon 

 coast ("coastline transects"), transect lines parallel to the 

 shore between 250 and 500 m from shore were run, typically 

 covering from 25 to 100 km in a day. 



To quantify distribution in relation to distance from 

 shore ("offshore transects"), repeated transect lines along 

 the same 4 km section of coast were run, each one 300 m to 

 600 m farther out to sea than the previous one (all 1993 

 increments were of 300 m; in 1992 the distance increment 

 was variable). Transects lines were repeated progressively 

 farther offshore until no murrelets were seen on the water for 

 a full 4 km line. In 1992 the outer limit of surveys was 2.5 

 km offshore, in 1993 the outer limit was 6 km. offshore. The 

 sample 4 km coastal sections were selected at various locations 

 between Gleneden Beach and Seal Rocks (except for one 

 survey south of Heceta Head in 1992) in central Oregon. The 

 sample locations were all off sand or mixed sandy and rocky 

 shorelines where murrelets were consistently present. 



All information was spoken into a tape recorder via an 

 external microphone, held by one of the observers. 



Aerial Surveys 



A single engine high-wing Cessna 187 or 206 aircraft 

 was used for aerial surveys. An observer on each side of the 

 plane used a tape recorder with remote microphone to 

 record observations. 



In 1993, the inboard observer (nearest the shoreline) 

 noted when landmark locations were passed. In 1992, a third 

 person recorded time and location on maps. The pilot 

 maintained an altitude of approximately 60 m and a speed of 

 90 knots. Distance from shore was held at between 300 and 

 500 m (the same as for coastline vessel transects), except 

 when passing seabird nesting islands, where a wide berth 

 was given (>800 m) to avoid disturbance. Each observer 

 continuously scanned a 50 m wide corridor of ocean surface 

 which was calculated as an angle between 32 and 57 off 

 horizontal, as measured with a clinometer. While maintaining 

 their scan of the water surface, observers recited the number 

 and species of birds seen and time to the nearest 10 seconds, 

 and reported on observing conditions. We found that at the 



altitude we flew, we were able to identify most birds to 

 species. The 60 m level was recommended in Briggs and 

 others (1985) and by Varoujean and Williams (this volume) 

 as optimal for surveys of small marine birds. Since our 

 aircraft had only a pressure altimeter, our recorded altitude 

 was only approximated. 



Shore-Based Observations 



Additional shore observations were made oppor- 

 tunistically. A 20-45 power telescope was used to carefully 

 scan the sea beyond the surf line to a distance of approximately 

 1 .2 km (using marks on topographic maps a known distance 

 offshore for reference). Information recorded included location, 

 time of beginning and end of survey, weather and observation 

 conditions, number of all seabird species (except in a few 

 instances when time limitations allowed only Marbled 

 Murrelets to be counted), group size of murrelets, and other 

 notes on murrelet behavior or distribution (e.g. fish holding, 

 concentrated in surf line, etc.). 



Data Management and Analyses 



To describe distribution along the Oregon coast, Marbled 

 Murrelets counted from coastline vessel transects were 

 summed in 10 km blocks as measured by landmarks on 

 shore and time elapsed when traveling at known speed (8 

 knots). Currents and variation in speed resulted in location 

 errors of up to 3 km on some long transects without 

 landmarks, but error was usually less than 1 km. The 10 km 

 sums were averaged where counts were repeated on the 

 same section of coast. 



Population Estimates 



We used both line and strip transect analyses to develop 

 population estimates from the vessel coastline transects. 

 This allowed for a more robust conclusion and assessment of 

 the different assumptions underlying each method. For both 

 analyses, the complete transect of each day was treated as a 

 sampling unit, which avoided statistical dependence of 

 adjacent transect legs. Birds flying through the transect area 

 were not included in any calculations. 



Line Transects 



Because distance from the vessel to each murrelet sighting 

 were recorded, these data were amenable to line transect 

 analyses. Data were processed by the program DISTANCE 

 (Laake and others, 1993) which fits a model to the distances 

 at which birds were detected (a detection probability curve) 

 and then includes data on encounter rate (number of detections/ 

 length of transect) and average group size to derive a density 

 of birds per km 2 . This is then multiplied by the length of the 

 region to achieve an abundance estimate for a given area. 

 The resulting models (half-normal or cosine, with polynomial 

 adjustments to the fit) all had their peak detection probability 

 on the transect line, whereas, due to avoidance behavior, 

 peak reported detection distance was typically 20-40 m from 

 the line. To resolve this, we divided the reported distance by 

 2 or 3 for birds seen on the forward quarters and divided by 4 



340 



USDA Forest Service Gen. Tech. Rep. PSW-152. 1995. 



