Marbled murrelet surveys: site and annual variation, sampling effort, and statistical power

Populations of the marbled murrelet (Brachyramphus marmoratus) have decline d and become fragmented throughout the Pacific Northwest, apparently becaus e of the loss of older forests from logging. Because clearcut logging conti nues throughout a large portion of the bird's range, there is a pressing ne ed to develop rigorous monitoring protocols. We conducted 10-minute dawn su rveys at fixed point-count stations along 2 roadway segments during mid-Jul y 1991-1996 to quantify annual and with in-year spatial variation in detect ing marbled murrelets. Mean detections/station (pooled across stations) var ied among years and ranged from 11.3 to 33.8. The greatest within-year diff erence among stations was a range of 0-74 detections, which occurred in the same vicinity in 1994; within-year variation between roadway segments was not significant (F-1,F-21 = 0.46, P = 0.51). For a specified effect size of 0.40, we determined that the sample size needed to detect significant diff erences among years, for alpha = 0.05 and power (1-beta) = 0.95, was 22 poi nt-count stations. Number of years required to detect an annual decrease in detection of 10% with one visit to a station/year varied according to the precision of estimates, which varied annually and ranged between 28-46 year s and 12-19 years for alpha = beta = 0.05 and 0.10, respectively. By visiti ng each station 20 or 12 times annually, it may be possible to detect a 10% annual decrease in detection in 6 years (i.e., 50% cumulative decline) wit hout compromising power or error rate. Assuming that dawn survey detections reasonably index habitat distribution or relative abundance, these results provide preliminary estimates of sampling effort necessary to detect local population decline or within-year differences in murrelet habitat distribu tion in the Yakutat Foreland of southeastern Alaska.