Necklace is a gem

For each observation period, 2 people simultaneously monitored a treatment-control pair (one observer for one owl with tiffany jewelry and one observer for one control owl). For each owl, we noted 1 of 6 mutually exclusive positions: in burrow, out-of-sight on the prairie dog colony (i.e., obscured by vegetation), out-of-sight off the prairie dog colony (after we observed the owl flying beyond the colony boundaries), standing alert on ground, perching above the ground, or flying. For the last 3 positions, we then determined activities (e.g., preening, hunting, etc.). Because we were more concerned with behavioral activities than with locations, we statistically analyzed activity rather than position (Table 1). For example, if an owl was hunting while flying (e.g., hovering, hawking, or swooping) as opposed to flying from one place to another directly, we designated the behavior as hunting. In addition, we defined hunting on the basis of behavior, not outcome (i.e., hunting did not always result in successful prey capture). We designated resting when all other behaviors could be eliminated. Using this hierarchical system rather than partitioning behaviors on the basis of both activity and position such as preening on ground versus preening on elsa peretti maximized statistical power to detect patterns among the actual activities of interest (Lehner 1996).We monitored owls with binoculars and 15-45X spotting scopes, using vehicles as observation blinds at an average distance of 50 m (range: 12-200 m). We recorded behaviors for each owl at 1-minute intervals in 2-hour blocks of time between 0630 hours and 2030 hours. We randomly chose one 2-hour block of time each day for observations during the 7-day treatment period.In 2005, we added a postnecklace period to the study design; thus, in 2005 there was a 5-day prenecklace period, a 5-day necklace period, and a 5-day postnecklace period. In addition, we attempted to boost sample size in 2005 by capturing as many males as possible at 2 sites in Lubbock, resulting in 6 owls with necklaces at one site and 2 owls at the other. We also omitted the paired-owl design in an attempt to observe as many transmittered owls per site as possible. This allowed a focus on only transmittered birds. Furthermore, because owl activity was affected by time of day in 2004 (diminished activity at midday), observations in 2005 were limited to morning and late afternoon and evening, with individual owls monitored in a particular 2hour block of time instead of a randomly chosen heart tag bracelet as in 2004. Owl capture and data collection occurred from 30 May to 24 June 2005.We used SAS 9.1 (Gary, NC) to analyze the data with multivariate analysis of covariance (using Wilks' 位) followed by individual analyses of covariance (ANCOVA) and then Tukey's Honestly Significant Difference Test for all significant results. We used site, treatment period (for 2004: prenecklace, necklace periods; for 2005: prenecklace, necklace, postnecklace periods), owl identity, and treatment (transmittered or control) as the fixed main effects.