|
Article Excerpt
Much debate still centers on methods for obtaining reliable density estimates of passerines (Ralph et al. 1995, Jones et al. 2000, Farnsworth et al. 2002, Buckland 2006, Thogmartin et al. 2006, Diefenbach et al. 2007). Integral to all methods is an estimate of detection probability that varies within and among species, habitat, weather, season, and observers (Wilson and Bart 1985, Sauer et al. 1994, McShea and Rappole 1997, Diefenbach et al. 2003, Confer et al. 2008). Two key components of estimating detection probability are: (1) availability of the subject for detection, and (2) the probability the subject will be detected given that it is available (Diefenbach et al. 2007). Two methods commonly used to estimate avian abundance are distance sampling (Reynolds et al. 1980, Buckland 2006) and double-observers (Nichols et al. 2000); both assume the probability that a subject is available to be detected is 1.0. Farnsworth et al. (2002) recognized that singing frequency has a significant influence on estimating detection probabilities and incorporated this factor into detection probability models by using censuses with multiple time intervals. Farnsworth et al.'s (2002) removal models estimate detection probabilities that include availability and detection, but when only three intervals (e.g., 3-min, 5-min, 10-min in duration) are sampled, all members of the first time interval must be assumed to have a probability detection of 1.0. Singing territorial males are the principal cue for detection in passerines and results from the few intensively studied birds, ranging from grassland to forest species, have demonstrated the assumption the subject is available 100% of the time is seriously flawed (Scott et al. 2005, Staicer et al. 2006, Diefenbach et al. 2007, Confer et al. 2008). The objectives of our study were to: (1) document song rate variation for Cerulean Warblers (Dendroica cerulea) in Ozark riparian forest, and (2) provide an estimate of the number of individuals that might be available for detection during censuses.
METHODS
Song rate monitoring was conducted between 21 May and 4 June in 2003, 2007, and 2008 in the southeastern Missouri Ozarks. We chose the above time period to ensure high singing rates and to exclude both late migrants and young of the year; this period also corresponds to when many passerines are surveyed across North America. Our monitoring was conducted along the Current River within the Ozark National Scenic Riverways from Akers Ferry (37[degrees] 22.5' N, 91[degrees] 33.1' W) to Big Spring Access (36[degrees] 57.3' N, 90[degrees] 59.3' W). Thirty-one males were monitored along the Current and other Ozark rivers, but we limited our song analyses to 24 males (n = 6 in 2003, n = 10 in 2007, n = 8 in 2008) along the Current River to reduce bias among river systems and years with samples of only 1-2 individuals.
[FIGURE 1 OMITTED]
Cerulean Warbler song rates were monitored continuously from 0500 to 0900 hrs (CST) with the entire observation period subdivided into 5-min intervals (n = 48 intervals). Each individual was monitored, using stopwatches and counters, for at least one morning under natural conditions (no playback) with no observer disturbance, as males typically were in the canopy. No birds were marked, but singing males were closely monitored by following them from song perch to song perch, and noting locations of interactions with neighboring males (counter-singing and aggressive interactions). Two observers monitored each male or separate observers simultaneously monitored adjacent males in areas where territorial males were densely...
|
