Study Area. We studied Golden Eagles that originated from natal areas across the southern half of the Colorado Plateau physiographic province (Patton et al. 1991) and adjacent (to the east) parts of the Rocky Mountains' southern terminus in southwestern North America (37°N, 109°W). This area included far southeastern Utah, northeastern Arizona, northwestern New Mexico, and southwestern Colorado (Fig. 1), though some eagles we tracked moved far beyond this area. The southwestern half of the study area was arid to semiarid land (mean annual precipitation 25–35 cm), with sparse to moderate plant cover dominated by perennial grasses and low (<3 m high) shrubs. The northeastern half was slightly less arid (annual precipitation 35–40 cm) and was dominated by big sagebrush (Artemisia tridentata) shrub-steppe and conifer woodland (Pinus edulis, P. ponderosa, Juniperus spp.). Parts of the latter area included irrigated pastureland, hayland, and cropland. Small mountain ranges and sandstone ridges, mesas, and outcrops occurred across most of the study area; elevations of these typically were 1700–2700 m in the southwest to 2100–3600 m in the northeast. Land use was mainly livestock ranching, recreation, and gas and oil extraction. Main prey of Golden Eagles on the study area, at least during the eagle breeding season, included black-tailed jackrabbits (Lepus californicus), cottontail rabbits (Sylvilagus spp.), and Gunnison's prairie dogs (Cynomys gunnisoni; Stahlecker et al. 2009). About 90% of nests of Golden Eagles were on cliffs and the rest were in ponderosa pine.

Figure 1. 

Fifty-three Golden Eagle breeding areas in the Colorado Plateau and Southern Rocky Mountains of the southwestern U.S.A, where 66 Golden Eagle nestlings were tagged with PTTs during 2010–2014 and monitored during dispersal from natal areas. Dashed line approximates the boundary between breeding areas in the arid, lower-elevation (1700–2700 masl) part of the study area and those in the relatively less arid, higher-elevation (2100–3600 masl) part of the study area. Disperser strategy type of eagles from different natal areas, as described in text, include: (1) SD – short-distance disperser, (2) MD – moderate-distance disperser, (3) LD – long-distance disperser, and (4) Other – individually unique dispersal strategy. At some natal areas, more than one eagle was tagged; individuals from some such sites exhibited different strategies, as indicated. Natal area locations on lands of Native American tribes are obfuscated (±50 km) for purposes of confidentiality.

Telemetry. During late May through early July each year, we entered Golden Eagle nests when nestlings were 7- to 8-wk old (i.e., near fledging age) and fitted nestlings with platform terminal transmitters (PTTs; Solar Argos/GPS models PTT-45 and PTT-70; Microwave Telemetry, Inc., Columbia, MD U.S.A.) that produced GPS-based fixes with ±20-m accuracy. PTTs were attached via “Y-harness” (Buehler et al. 1995) made with 0.64-cm wide Teflon ribbon (Bally Ribbon Mills, Bally, PA U.S.A.). With harnesses attached, PTT-45 and PTT-70 units weighed approximately 55 g and 90–100 g, respectively (Stahlecker et al. 2015), <3% of the mass of nestlings. Our dataset included GPS fixes (n = 293,960) collected hourly each day during 0900–1600 H.

Dispersal Timing, Distance, and Strategies. We used “Method 7” from Weston et al. (2013) as one of two criteria that best defined the onset of dispersal for Scottish Golden Eagles. Their approach, based on patterns of space use around nests by territorial adults as delineated by McGrady et al. (2002), determined that a pre-breeding-age eagle initiated dispersal when it ventured >9 km from its natal nest and stayed >6 km from the nest for the next 10 d. Initially, we estimated this 10-d distance for each of 35 unique breeding areas where we tagged nestling Golden Eagles with PTTs by measuring the distance from a nest at which we tagged a nestling to the nearest nest occupied by a territorial pair of Golden Eagles. One-half of that mean inter-nest distance was 4.0 km (SE = 0.4). Because this nearest-neighbor distance was derived from the less arid half of our study area where breeding density was comparatively high (D. Stahlecker unpubl. data), our result likely was biased downward. Thus, we adopted the 6-km criterion of Weston et al. (2013) and considered the area ≤6 km from the natal nest to be the natal area.

To characterize 1st-yr dispersal strategies, we considered hatch date for all eagles to be 15 April because this approximated the median hatch date (14 April) and individual hatch dates diverged little from this (SE = 1.2 d; n = 66). Next we considered 1 yr from 15 April to be the end of each eagle's 1st yr of life, with 16 March–15 April representing the early spring period ending the 1st yr. We plotted the maximum distance from the natal nest for each eagle each day through the 1st yr and visually examined these distance graphs for evidence of distinct temporal-spatial patterns possibly representing various dispersal strategies. We then assigned each eagle to one of four 1st-yr dispersal strategy categories: (1) short-distance dispersers (SD) remained <120 km from natal nest; (2) moderate-distance dispersers (MD) dispersed 120–500 km from natal nest; (3) long-distance dispersers (LD) dispersed >500 km; or (4) Other individuals, which demonstrated unique dispersal patterns. Exceptions included some SDs that made brief excursions up to 192 km from natal nests and some MDs that made brief excursions to as close as 12 km from natal nests. Individuals that made excursions typically did so only one to three times, with each excursion lasting 1–4 d.

Influence of Area of Origin. We considered that 1st-yr dispersal strategies of Golden Eagles could be driven by prey availability via positive linkages with precipitation and primary productivity (e.g., leporids in western North America: McKay and Verts 1978, Ernest et al. 2000, but see Lightfoot et al. 2011). Specifically, we hypothesized that eagles from natal areas in arid landscapes would disperse farther than those from natal areas in moister landscapes; assuming that prey would be more abundant in the latter. To test this, we calculated total precipitation within 120 km of the natal nest in the year each eagle was tagged as a nestling, plus total precipitation from the previous year. Daily precipitation records were extracted from Oak Ridge National Laboratory's Daymet service (Thornton et al. 2014) for each 1-km × 1-km grid cell covering the 120-km-radius area. We used 120 km because it was the upper limit of 1st-yr dispersal distance for SDs.

First-year Survival. We estimated 1st-yr survival rate for Golden Eagles that lived >3 wk after the onset of dispersal, which we deemed long enough to reveal their dispersal strategy. To analyze survival by dispersal strategy type with minimal bias, we combined data from three eagles that died approximately 4 mo after the onset of dispersal with those surviving longer and dispersing relatively short distances because all exhibited roughly similar early dispersal movement, especially distance from natal nest. We also combined data from two eagles classified as Other with those that dispersed moderate distances because all moved within the same range of distance from respective natal nests through the 1st yr. Results of the survival analysis were unchanged if data from these five individuals were excluded.

All eagles in the overall sample retained functional PTTs through death or age 1 yr, such that we could estimate true annual survival (S) directly from the data. We generated estimates in program R (R Core Team 2015), using a logit-link generalized linear model (GLM) in a Bayesian framework (Kéry 2010), with the binary response variable alive (1) or not (0) at age 1 yr and with fixed categorical effects for dispersal strategy. We compared GLMs that estimated 1st-yr survival of (1) SDs, MDs, and LDs separately; (2) pooled SDs and MDs (SDs–MDs) versus LDs; (3) pooled MDs and LDs (MDs–LDs) versus SDs; and (4) an intercept-only (null) model. We used the Markov chain Monte Carlo method implemented in program WinBUGS (Lunn et al. 2012) to estimate posterior distributions of annual survival probabilities. For all models we ran three chains for 60,000 iterations each with a burn-in of 10,000 iterations per chain. We inspected history plots and used R̂ to estimate convergence; models with values of R̂ < 1.1 were considered to have converged (Gelman and Hill 2006). We compared models based on their deviance information criterion (DIC) scores (Burnham and Anderson 1998, Hooten and Hobbs 2014). We considered models within two DIC units of each other as having equal support (Burnham and Anderson 1998), in which case we selected the model with the fewest estimated parameters as the preferred model. We also examined the relationship between 1st-yr survival of Golden Eagles and maximum distance from natal nests by (1) arranging maximum distances of individual eagles from shortest to longest and (2) calculating the percentage of eagles that survived to 1 yr and had maximum distances greater than each focal eagle's maximum distance. Eagles that moved ≥800 km were placed in one group.

Progressive Distancing and Space Use. We generated 95% kernel density estimates of home ranges by using a 0.7 proportional modifier in place of least-squares cross-validation for bandwidth (Worton 1989). These home ranges, hereafter termed HRs, were estimated in AnimalMovement tools for ArcGIS (https://github.com/regan-sarwas/AnimalMovement). The centroid of a given HR was derived from all polygons and represented the kernel density area's center of mass. We explored evidence of 1st-yr settling by measuring (1) the distance from each eagle's natal nest to centroids of successive monthly HRs; (2) month-by-month cumulative HR size, corresponding with total area explored over time; and (3) percentage overlap between successive monthly HRs. A horizontal asymptotic trend in any of these cases would suggest settling. We began with November, as most dispersal was initiated by the end of this month. Eagles that died before the end of the 1st yr were excluded from the dataset. Next, we conjectured that if an eagle shifted among temporary HRs or wandered randomly before reaching breeding age, there would be little or no overlap between its 1st- and 2nd-yr HRs. Thus, we measured overlap between HRs of individuals at approximately 12 mo and 24 mo of age. For 1st-yr eagles, this 1-mo, early spring period of comparison was the interval between the last dispersals from natal areas and the first northward summer migrations by about one-third of the eagles, starting at the beginning of their 2nd yr (R. Murphy and D. Stahlecker unpubl. data). To convey simple examples of 1st- and 2nd-yr home range overlap and support comparisons among studies, we also estimated home ranges by using 95% minimum convex polygons.

Statistical Analyses. Our dataset included several pairs of nest-mates. We considered members of individual pairs independent of one another in our analyses because we did not observe consistent, paired links in either dispersal timing, strategy, or distance and direction from natal nests at the end of the 1st yr; however, we acknowledge that nest-mates' shared genetic and environmental histories may engender some nonindependence. Throughout, we convey central tendency as means ± SEs and medians. We used unequal variance, unpaired or paired t-tests for simple univariate comparisons after checking data for normality. The t-tests were one-tailed when we had an a priori notion about the direction of a relationship as conveyed in this Methods section; otherwise tests were two-tailed. We used the nonparametric Mann-Whitney U-test for unpaired, univariate comparisons if data were not normally distributed and to test for differences in dispersal timing where the median was the most appropriate metric of central tendency. We used Pearson's r to evaluate correlation between age and dispersal date, and survival and maximum distance traveled from natal nest. We considered α = 0.05 to indicate statistical significance.

Dispersal Timing, Distance, Direction, and Strategies. Sixty-six Golden Eagle nestlings tagged with PTTs during 2010–2014 (4, 10, 14, 23, and 15 eagles respectively; total n = 40 males and 26 females, including seven pairs of nest-mates) survived at least until the onset of dispersal from their natal areas. All were represented in our analyses of dispersal timing and survival. Three of the eagles died <3 wk after dispersing; data for these were omitted in other analyses.

All eagles dispersed from natal areas before age 1 yr. Onset of dispersal by individual eagles spanned 32 wk (4 August–18 March) but was concentrated (66.7% of individuals) during 16 September through 21 November (Fig. 2). Median date for the onset of dispersal was 22 October (mean ± SE = 8 November ± 7.1 d). Annual median dates were congruent in 3 of the 4 yr with ≥10 dispersal records (17, 13, and 15 October in 2011, 2013, and 2014) and about 3 wk later in the other year (8 November in 2012). Before dispersing, 78.8% of eagles made one to five brief (1–4 d) excursions up to 150 km from the natal nest (Fig. 3a); others remained <9 km from the natal nest until dispersing or made an excursion immediately before dispersing (Fig. 3b). Only six eagles that we classified as dispersed returned to natal areas during their 1st yr. All such returns were temporary: (1) two eagles returned for <1 d each; (2) three returned six, seven, and nine times, respectively, each return visit lasting <1–4 d; and (3) one eagle returned 10 times during a 6-wk period that began 2 wk after dispersing, with three returns of <1 d each to <3 km of its natal nest and the rest 5–6 km from the natal nest for parts of 1- to 9-d periods.

Figure 2. 

Date of onset of dispersal from natal areas for 66 juvenile Golden Eagles tagged as nestlings with PTTs. Dates were determined by using Method 7 in Weston et al. (2013). Forty-four (66.7%) of the eagles dispersed during the 10-wk period bounded by dashed lines. All eagles dispersed by age 1 yr.

Figure 3. 

Movement patterns of Golden Eagles before and just after onset of dispersal from natal nests: (a) typical pre-dispersal pattern including multiple, 1- to 4-d excursions to 20–150 km from natal nest; (b) less common pattern, with no excursions before dispersing or just one to two excursions 2–12 d before dispersing. Vertical dashed lines represent date of the onset of dispersal, determined by using Method 7 in Weston et al. (2013).

Maximum straight-line distances of Golden Eagles from natal nests during their 1st yr of life ranged from 55 km to 1379 km (n = 63). These were highly skewed toward shorter distances (Fig. 4); thus, the mean maximum distance (279 ± 39 km) was nearly twice that of the median (151 km). Included were three eagles that died approximately 2 mo after the onset of dispersal. Maximum distances <350 km were in relatively random directions from natal nests (Fig. 5). Greater distances were mostly (71.4%; n = 14) south or southeast or east, to Mexico, east-central and southeastern New Mexico, and west central Texas. Most distant (>900 km) points reached by the eagles extended from northern Sinaloa to western Nuevo Leon, Mexico (southernmost latitude 25°45′N; n = 4), plus far western Arkansas (easternmost longitude 94°05′W; n = 1).

Figure 4. 

Maximum distance (km) from natal nest during 1st-yr dispersal for 63 Golden Eagles tagged with PTTs as nestlings in the Colorado Plateau and Southern Rocky Mountains of the southwestern U.S.A. during 2010–2014. Each distance category value (x-axis label) is ±50 km.

Figure 5. 

Maximum distances (km) moved and direction to point of maximum distance moved by 63 Golden Eagles from the Colorado Plateau and Southern Rocky Mountains during their 1st yr of life. Each eagle is represented by a line extending from the central point; maximum distances of eagles that died before 1 yr of age are indicated by numeric values at ends of their respective lines. Radii of inner to outer circles are 120 km, 500 km, and 1000 km.

First-year dispersal strategies of 60 Golden Eagles were 66.7% SD, 16.7% MD, 13.3% LD (Fig. 6), and 3.3% Other. The latter included one eagle that alternated between short- and moderate-distance movements through its 1st yr and another that moved back and forth between an area 150–200 km from its natal nest and an area approximately 375 km away. MDs–LDs (n = 18) dispersed earlier than SDs (medians = 1 October and 29 November respectively; U = 155, P < 0.001; means = 4 October ± 5.4 d and 29 November ± 10.1 d respectively). All MDs–LDs, but only 57.5% of SDs dispersed by 23 November.

Figure 6. 

Examples of dispersal onset and subsequent movement through ca. the 1st yr of life for Golden Eagles exhibiting three dispersal strategies, based on maximum daily distances from the natal nest: (a) two short-distance dispersers, (b) two moderate-distance dispersers, and (c) three long-distance dispersers. For each strategy type, time-distance graphs are accompanied by movement tracks of the respective individual eagles to provide spatial context; black dots represent natal nests. Onset of dispersal (vertical lines) is based on changes in distance from natal nest per Method 7 in Weston et al. (2013). Maximum daily distance is derived from GPS locations collected hourly during 0900–1600 H.

Influences of Age, Sex, Nest-mates, and Area of Origin. The eagles' median hatch date was 14 April (mean = 13 April ± 1.2 d; interquartile range = 6–19 April; range = 16 March–13 May). Thus, 4.3 mo lapsed between median dates of fledging and dispersal onset (22 October), assuming a fledging age of approximately 10 wk (O'Toole et al. 1999, Watson 2010). All eagles dispersed before age 1 yr. We found no overall relationship between age and date of the onset of dispersal (r = 0.040, P = 0.37, n = 66). However, MDs–LDs dispersed at younger ages than SDs (174.2 ± 6.4 d and 230.2 ± 10.2 d respectively; df = 56, t = 3.54, P < 0.001). All MDs–LDs, but only 65.0% of SDs dispersed by 250 d of age.

We detected no overall sex bias in timing of dispersal onset (median = 14 October and 11 November for males and females respectively; U = 443, P = 0.29, n = 40 and 26; mean = 27 October ± 8.0 d and 26 November ± 12.6 d). Among SDs, however, onset of dispersal was earlier for males than females (median = 24 October and 3 January respectively; U = 116, P = 0.04, n = 24 and 16; mean = 12 November ± 12.1 d and 25 Dec ± 15.5 d). Sex bias in dispersal timing was not evident among MDs–LDs, but the sample was small (n = 10 males and eight females). We found no overall sex bias in maximum distance traveled from the natal nest during the 1st yr (276.7 ± 53.9 km for males and 283.6 ± 53.3 km for females, respectively; n = 37 and 26; df = 61, t = 0.08, P = 0.93). Males composed nearly identical proportions of SDs and MDs–LDs (0.60 and 0.56 respectively; n = 40 and 18 total eagles, sexes combined).

Golden Eagles we monitored included seven pairs of nest-mates. The difference in dispersal date between individuals of nest-mate pairs ranged from 3 d to approximately 3–5 mo (Table 1). Of four nest-mate pairs that survived through age 1 yr, two (nest locations named “Angus” and “Chama”) included SDs that remained closely associated through early spring despite initiating dispersal approximately 3.5 and 1 mo apart, respectively. Another pair of SD nest-mates (“Grapevine”) occupied early spring HRs in opposite directions of their common natal nest. The fourth nest-mate pair monitored through age 1 yr (“Kittyhawk”) included a LD eagle that by early spring had returned close to its natal area and thus nearer to its SD nest-mate.

Table 1.

Characteristics of 1st-yr dispersal by Golden Eagle nestmates from natal areas in the Colorado Plateau and Southern Rocky Mountains of the southwestern U.S.A., monitored via satellite telemetry during 2010–2015.

Table 2. 

Models evaluated to explain variation in the probability of survival to age 1 yr by dispersal strategy for Golden Eagles from natal nests in the Colorado Plateau and Southern Rocky Mountains of the southwestern U.S.A., monitored via satellite telemetry during 2010–2015. Survival probabilities were estimated using a generalized linear model with a logit-link function in a Bayesian framework (see Methods for more details on the modeling approach). Model ranking was based on deviance information criterion (DIC).

Golden Eagles from the more arid, southwestern half of our study area were more likely to be MDs–LDs; seven of eight LDs originated there (Fig. 1). Mean annual precipitation within 120 km of the natal nest during the hatch year plus previous year for individual eagles was lower among MDs–LDs (325.9 ± 26.5 mm) than SDs (381.8 ± 14.7 mm; n = 18 and 40, df = 23, one-tailed t = 1.85, P = 0.04). Moreover, the coefficient of variation in annual precipitation within 120 km of natal nests during the same respective 2-yr periods was lower among MDs–LDs (mean = 31.1 and 41.0 respectively; df = 27, one-tailed t = 4.13, P < 0.001).

First-year Survival. Fifty-two of 66 Golden Eagles survived the period between the onset of dispersal and end of the 1st yr of life. Causes of mortality included electrocution and/or collision with electrical distribution lines (five); shooting (two); starvation/disease in winter (two); killed by mountain lion at deer carcass (one); unrecoverable (due to human safety concerns) but presumed dead based on circumstances associated with data record (three); and unknown cause (one). Overall survival rate between the onset of dispersal and end of the 1st yr of life was 78.8% (95% CI = 68.2%–87.6%). Only four of 43 eagles and none of 12 eagles categorized for this survival analysis as SDs and MDs (respectively) died before the end of their 1st yr, but seven of eight LDs died. The most competitive model grouped SDs and MDs and separated LDs (Table 2). First-year survival was 92.6% (84.4%–97.9%) for the SD–MD group and 13.2% (5.0%–42.0%) for LDs (P difference ≠ 0 = 1.0). We plotted 1st-yr survival rate against maximum distance moved from natal nest and found an inverse relationship between the two (r = 0.98, P <0.001, n = 66; Fig. 7).

Figure 7. 

Relationship between percentage of 66 Golden Eagles surviving through their 1st yr of life and “greater than” maximum distance traveled from the natal nest. This distance includes all eagles that moved farther than any particular x-axis value, e.g., 12 eagles moved >400 km, and six (50%; y-axis) of these survived through the end of their 1st yr of life. Data are truncated at 800 km to simplify the illustration; five individuals had maximum distances of 941 to 1379 km and none of these survived to age 1 yr.

Progressive Distancing and Space Use. SDs appeared to settle by November–December, on average, based on constant to slightly increasing distance from natal nests from then through March (Fig. 8a); distance increased in April, however, compared with November (55 ± 5 km and 33 ± 6 km, respectively). Correspondingly little change occurred in month-to-month, cumulative HR size (Fig. 8b). MDs exhibited a roughly similar pattern for both metrics albeit with much individual variation (Fig. 8a, b). Successive monthly HRs of SDs overlapped, on average, by roughly one-half during February–April while those of MDs increased to about three-fourths by April (Fig. 8c). Relatively high overlap between October and November by SDs reflects the fact that many initiated dispersal during those months. Month-to-month comparisons based on minimum convex polygon estimates of home ranges (Appendix) generally showed parallel patterns, especially in distance from nest and cumulative home-range size, albeit with much greater within-month variation.

Figure 8. 

Progressive distancing and spatial use by 1st-yr Golden Eagles exhibiting short-distance (SD; n = 38) or moderate-distance (MD; n = 10) strategies for dispersing from natal nests through the end of their 1st yr (April): (a) mean distance between natal nests and centroids of monthly home ranges (95% kernel density estimates); (b) mean month-to-month, cumulative home-range size; (c) percentage overlap between successive monthly home ranges. Only individuals that survived ≥1 yr are included. Error bars are ±SE.

Distances from natal nests to centroids of 1st- and 2nd-year, early spring HRs did not differ in a paired between-year comparison for individuals tracked through at least their 2nd year, both for 34 SD–MDs (1st year 76.4 ± 15.6 km, median = 43.2 km; 2nd year 66.4 ± 16.9, median = 33.4; paired t = 0.96, df = 33, P = 0.33) and for 26 SDs (1st year 42.2 ± 4.8, median = 34.5; 2nd year 35.1 ± 4.8, median = 32.7; t = 1.48, df = 25, P = 0.15); between-year distances for 16 (61.5%) of the SDs differed by only 0.8–24.3 km. Between-year distances of the eight MDs tracked through their 2nd year varied more (1st year 187.5.0 ±48.2, median = 158.0; 2nd year 168.1 ± 59.4, median = 90.7), including one eagle with a three-fold decrease and another with a five-fold increase.

Average overlap of 1st- and 2nd-year, early spring HRs was 31.8% (±4.8; median = 25.2%, n = 34). Half of SDs exhibited >40% overlap versus only one of eight MDs, though overall HR overlap did not differ strongly between the two groups, due in part to the relatively small number of MDs; we had predicted greater overlap for SDs (SDs 42.2% ± 6.4%, median = 37.2%, n = 26; MDs 26.5% ± 9.5%, median = 21.9%, n = 8; one-tailed t = 1.22, df = 32, P = 0.12). Based on 95% minimum convex polygon estimates, 2nd-year, early spring home ranges of SDs typically were smaller than and often nested entirely within 1st-year home ranges, although size and spatial configuration of the home ranges varied extensively among individuals and between years for individuals (Fig. 9).

Figure 9. 

Spatial overlap between 1st-yr and 2nd-yr, early-spring home ranges of eight Golden Eagles classified as short-distance dispersers and one (GE 121733) classified as a moderate-distance disperser, and relationship to their respective natal nest locations. For simplicity, home ranges are estimated by 95% minimum convex polygons. Arrows link natal nests of three eagles to their respective home ranges. Early spring is 16 March–15 April, corresponding roughly to age 12 mo in the 1st yr and 24 mo in the 2nd yr. Some natal nest locations are intentionally inexact to protect sensitive areas.