Detection Probabilities of Woodpecker Nests in Mixed Conifer Forests in Oregon

—Accurate estimates of Black-backed (Picoides arcticus) and Hairy Woodpecker (P. villosus) nests and nest survival rates in post-fire landscapes provide land managers with information on the relative importance of burned forests to nesting woodpeckers. We conducted multiple-observer surveys in burned and unburned mixed coniferous forests in Oregon to identify important factors influencing detection rates of woodpecker nests. We found 21 Black-backed Woodpecker nests and 38 Hairy Woodpecker nests in burned forest, and three Hairy Woodpecker nests in unburned forest. Competing models of detection probability in Program MARK indicated that nest-detection probability differed by nest stage. We found no evidence to indicate that detection rates of nests were associated with survey timing during the nesting season. Raw nest counts in burned coniferous forests may underestimate nest numbers, especially for nests in early stages of development. Blackbacked Woodpecker nests were slightly more detectable than those of Hairy Woodpeckers in burned forests, and observers may differ in their abilities to detect nests. Received 12 February 2008. Accepted 29 May 2008. The importance of estimating probabilities of detection for individuals, species, or age classes has been noted by many authors (Nichols 1992, Buckland et al. 2001, Williams et al. 2001, MacKenzie et al. 2006), and computer programs are available for estimating detection probabilities for a wide variety of sampling scenarios. Program MARK (White and Burnham 1999) and Program PRESENCE (MacKenzie et al. 2006), for example, are based on generating estimates of detection and/or recapture probabilities (p) to adjust observed counts of individual animals or occupied sites, and account for differences in detection probabilities (Nichols 1992). Despite the volume of literature emphasizing potential biases that can be induced by ignoring detection rates, few studies have quantified detection rates for nests of avian species. In the few cases where avian researchers have estimated detection probabilities for nests, p has varied by observer, species, or habitat type (Erwin 1980, Ferns and Mudge 1981, Wanless and 1 Montana Fish, Wildlife, and Parks, 1400 South 19th Avenue, Bozeman, MT 59718, USA. 2 USDA Forest Service, Rocky Mountain Research Station, 1648 South 7th Avenue, Bozeman, MT 59717, USA. 3 Department of Ecology, Montana State University, Bozeman, MT 59717, USA. 4 USDA Forest Service, Rocky Mountain Research Station, 322 East Front Street, Suite 401, Boise, ID 83702, USA. 5 Corresponding author; e-mail: [email protected] Harris 1984, Rivera-Milán 2001, McPherson et al. 2003, Barbraud et al. 2004, Barbraud and Gélinaud 2005). Woodpeckers perform an important ecosystem function by creating cavities which are subsequently used by a number of different wildlife species (Raphael and White 1984, Murphy and Lehnhausen 1998, Hutto 2006, Saab et al. 2007). Previous studies have identified post-fire forests as important nesting habitat for several woodpeckers and other cavity-nesting species (Bock et al. 1978, Raphael and White 1984, Saab and Powell 2005, Saab et al. 2007). Accurate counts of nests and estimates of productivity are necessary to provide land managers with information regarding the effects of management practices (such as salvage logging) on these species. We selected two woodpecker species, Hairy Woodpeckers (Picoides villosus) and Blackbacked Woodpeckers (P. arcticus), as our focal species. We conducted a study using a survey design with independent observers to examine whether probability of nest detection varied as a function of nesting stage, observer, species, season, and/or habitat (burned or unburned). Our objectives were to: (1) identify important factors affecting nest detection rates of the two woodpecker species, (2) quantify the detection rates of woodpecker nests and examine how the probability of detection increases from one to three surveys, and (3) provide recommendations to improve the probability of detecting woodpecker nests. 83 Russell et al. • ESTIMATING DETECTION RATES OF WOODPECKER NESTS We expected nest detection would be highly associated with nest stage. Late-stage nestlings close to fledging are often more conspicuous and vocal (Best and Petersen 1982, Pietz and Granfors 2000). Johnson and Shaffer (1990) noted the number of Mallard (Anas platyrhynchos) nests found increased with age, indicating nests were missed in early stages for this species. Mayfield (1961, 1975) methods for estimating nest success were developed because successful nests have a higher probability of being detected than failed nests due to longer exposure times, and estimates of apparent nest success are biased high. We expected the day in the season the survey was conducted should be related to nest stage (more late stage nests found later in the season), and potentially would be associated with increasing nest detection rates later in the season. Hairy and Black-backed woodpeckers are similar with regard to calling rates, size, foraging behavior, and visual appearance (Dixon and Saab 2000, Jackson et al. 2002), and we expected detection rates to be similar for both species. We also expected nest detection probability to be lower in unburned landscapes due to denser vegetation than in forests burned by wildfire, and that detection probability would vary by observer.