A GEOGRAPHIC-INFORMATION-SYSTEM APPROACH TO ESTIMATING THE ORIGIN OF MIGRATORY RAPTORS IN NORTH AMERICA USING STABLE HYDROGEN ISOTOPE RATIOS IN FEATHERS

Author(s) :Casey A. Lott and Jeff P. Smith Source: The Auk, 123(3):822-835. 2006. Published By: The American Ornithologists' Union DOI: http://dx.doi.org/10.1642/0004-8038(2006)123[822:AGATET]2.0.CO;2 URL: http://www.bioone.org/doi/full/10.1642/0004-8038%282006%29123%5B822%3AAGATET %5D2.0.CO%3B2

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological, and environmental sciences. BioOne provides a sustainable online platform for over 170 journals and books published by nonprofit societies, associations, museums, institutions, and presses. Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance of BioOnes Terms of Use, available at www.bioone.org/page/terms_of_use. Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercial inquiries or rights and permissions requests should be directed to the individual publisher as copyright holder.

BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions, research libraries, and research funders in the common goal of maximizing access to critical research.

The Auk 123(3):822835, 2006 The American Ornithologists Union, 2006. Printed in USA.

A GEOGRAPHIC-INFORMATION-SYSTEM APPROACH TO ESTIMATING THE ORIGIN OF MIGRATORY RAPTORS IN NORTH AMERICA USING STABLE HYDROGEN ISOTOPE RATIOS IN FEATHERS
C A. L1 J P. S
HawkWatch International, 1800 South West Temple, Suite 226, Salt Lake City, Utah 84115, USA

A.Analysis of stable hydrogen isotope ratios in feathers (Df ) is a promising method for investigating population connectivity in migratory birds. Stable hydrogen isotope ratios in precipitation (Dp) vary across North America with respect to latitude, elevation, and seasonal air-mass trajectories. A strong relationship between Df and Dp at locations of feather growth has been documented for several bird species. Some studies have used measurements of Df to plot the origins of migrants on maps of long-term weighted-average, growing-season North American Dp (hereaer Dp maps) using the observed relationship between Df and Dp from a reference sample of known-origin birds. The accuracy of this method depends on the strength of the Df and Dp relationship and accuracy of the Dp maps. Recently, a high-resolution (1-km2) model of North American Dp was published (Meehan et al. 2004) that accounts for the eect of elevation on Dp where previous models did not. We compared Df measurements from a geographically diverse sample of 264 raptor feathers with Dp estimates for feather-sample locations. We documented a strong relationship between raptor Df and Dp across North America. However, we also documented substantial regional variation in this relationship. We created a base map of North American raptor Df that incorporated the regional variation described by our sample. We ploed Df values from migrant Sharp-shinned Hawks (Accipiter striatus) captured in eastern Nevada directly on this map to demonstrate how it can be used to view the origins of migrants. Received 2 December 2004, accepted 10 October 2005. Key words: Accipiter striatus, annual cycle, avian migration, geographic information systems, population connectivity, Sharp-shinned Hawk, stopover.

Un Procedimiento Basado en Sistemas de Informacin Geogrca para Estimar el Origen de las Aves Rapaces Migratorias en Norte Amrica Usando los Cocientes de Istopos Estables de Hidrgeno Presentes en las Plumas
R.El anlisis de los cocientes de istopos estables de hidrgeno presentes en las plumas (Df) es un mtodo promisorio para investigar la conectividad entre poblaciones en las aves migratorias. Los cocientes de istopos estables de hidrgeno presentes en la precipitacin (Dp) varan a travs de Norte Amrica con respecto a la latitud, la elevacin y las trayectorias estacionales de las masas de aire. En varias especies de aves, se ha documentado una fuerte relacin entre Df y Dp en las localidades donde crecen las plumas. Algunos estudios han empleado medidas de Df para ubicar el origen de los migrantes en mapas basados

Present address: 111 Hillwood Drive, Huntington Station, New York 11746, USA. E-mail: [email protected]

822

July 2006]

Stable Isotopes and Bird Migration

823

en promedios ponderados de valores de Dp medidos a travs de varios aos en Norte Amrica (mapas Dp), utilizando la relacin observada entre Df y Dp en una muestra de referencia de aves de origen conocido. La exactitud de este mtodo depende de qu tan estrecha es la relacin entre Df y Dp, y de la exactitud de los mapas Dp. Recientemente se public un modelo de Dp de alta resolucin para Norte Amrica (Meehan et al. 2004) que, a diferencia de los modelos previos, tiene en cuenta el efecto de la elevacin sobre Dp. En este estudio comparamos medidas de Df de una muestra geogrcamente diversa de 264 plumas de aves rapaces con estimados de Dp para los sitios donde las plumas fueron obtenidas. Documentamos una fuerte relacin entre Df y Dp en las rapaces a travs de Norte Amrica. Sin embargo, tambin documentamos que esta relacin vara considerablemente entre regiones. Creamos un mapa base de Df para las rapaces que incopora la variacin regional descrita por nuestra muestra. Ilustramos el uso de este mapa para determinar el origen de los migrantes ubicando sobre ste los valores de Df medidos en individuos migratorios de la especie Accipiter striatus.

S measured stable hydrogen isotope ratios in feathers (Df) of birds at migratory stopover or wintering sites to estimate the areas where these feathers were grown (Chamberlain et al. 1997; Hobson and Wassenaar 1997, 2001; Hobson et al. 2001; Meehan et al. 2001; Wassenaar and Hobson 2001; Kelly et al. 2002; Rubenstein et al. 2002; Smith et al. 2003, 2004; DeLong et al. 2005). This method has been described as a major breakthrough in the ability to establish connectivity among breeding, migrant, and wintering populations (Kelly and Finch 1998, Hobson 1999, Webster et al. 2002, Rubenstein and Hobson 2004). However, precise methods for estimating the origins of migrants and for representing these data spatially are still in the early stages of development. Estimating migrant origins on the basis of Df relies on two basic paerns in biogeochemistry. First, stable hydrogen isotope ratios in weighted-average growing-season precipitation (Dp) show strong geographic variation across the ranges of many species of migratory birds (Chamberlain et al. 1997, Hobson and Wassenaar 1997). Second, the local environmental signal of Dp is transferred to the feathers of birds through their diet during feather growth (Chamberlain et al. 1997, Hobson and Wassenaar 1997, Hobson et al. 1999a). Thus, measurements of Df from migrants will be indicative of the Dp of the area where the feathers were grown (Chamberlain et al. 1997, Hobson and Wassenaar 1997). In the case of

hatching-year birds, this will be the natal area, because their feathers are typically grown on or near the nest site. Growing-season Dp values decrease with increasing latitude and altitude because of the strong negative relationship between temperature and Dp (Dansgard 1964). At a continental scale, the coarse spatial paern of decreasing Dp values with increasing latitude is aected by topography, such that high-latitude, lowelevation areas can have Dp values similar to those of lower-latitude, high-elevation areas (Dansgard 1964). Regional paerns in growing-season Dp are further modied in complex ways by more diuse eects, such as seasonal air-mass trajectories and local evaporation rates (Rozanski et al. 1993, Bowen and Wilkinson 2002). Large-scale maps of spatial paerns in Dp have been created that account for variation in Dp in relation to latitude, altitude, and other sources of variation (Meehan et al. 2004, Bowen et al. 2005). Previously, stable-isotope studies of animal movement have relied on simple models of Dp that did not account for these eects. Elevation-explicit models of large-scale paerns in Dp should help rene the accuracy of stable hydrogen isotope studies of animal movement (Hobson et al. 2004, Meehan et al. 2004). Several studies have described a strong relationship between Df and Dp within various groups of birds across dierent regions of North America and Europe (Chamberlain et al. 1997; Hobson and Wassenaar 1997; Wassenaar

824

L S

[Auk, Vol. 123

and Hobson 2000; Hobson et al. 2001, 2004; Meehan et al. 2001; Rubenstein et al. 2002; Lo et al. 2003). Many of these studies are not directly comparable because of dierences in laboratory methods used to measure Df and in the models used to estimate Dp for feathercollection sites. Some of these studies took a two-step approach in using measurements of Df to estimate the origins of migrants. First, the relationship between Df and estimated Dp was quantied using a reference sample of known-origin birds from throughout the source range of the study species. Second, the Df and Dp relationship was used to estimate the origins of unknown-origin birdsfor example, birds sampled during migration or at a wintering site (Chamberlain et al. 1997, Hobson and Wassenaar 1997, Meehan et al. 2001, Wassenaar and Hobson 2001). Predictive regression models were constructed from the reference sample using measurements of Df as the dependent variable and estimates of Dp as the independent variable. The regression equations from these models were then used to inversely predict Dp from measurements of Df from migrants of unknown origins. Finally, distributions of predicted Dp values for migrants were ploed on maps of Dp to illustrate their potential origins (Hobson et al. 1999b, Meehan et al. 2001, Wassenaar and Hobson 2001, Smith et al. 2004, DeLong et al. 2005). Several published studies have constructed models from reference samples with limited geographic coverage. If the relationship between Df and Dp varies regionally, limited or inappropriate spatial distribution of reference samples may violate the important regression assumption that reference samples are representative of the source regions of a migrant population, compromising the accuracy of predictions. Here, we (1) quantify the relationship between Df and Dp for North American raptors, using a large sample of known-origin birds with extensive geographic coverage; and (2) use this relationship to create a geographic information system (GIS) base map of Df that can be used to estimate the origins of migrant raptors from measurements of Df at stopover or wintering sites anywhere in North America. We assembled a reference sample of 264 feathers from 12 species of raptors from known locations across most of North America to document the strength of the relationship between raptor

Df and Dp and to describe regional variation in this relationship across the continent. We then used this reference sample and an elevation-explicit model of North American Dp (Meehan et al. 2004) to create a GIS base map of continental paerns in Df that accounted for regional variation in the Df and Dp relationship. We demonstrate how this base map can be used to plot the origins of migrants using Df values from Sharp-shinned Hawks (Accipiter striatus) captured during fall migration in the Goshute Mountains, Nevada. Our base map of North American raptor Df is available (see Acknowledgments). M Feather-sample collection.We collected one to two contour feathers from 264 individual raptors of 12 species from 255 locations across most of North America (Fig. 1). We obtained feathers from 16 dierent ornithology collections in the United States and Canada, and six feathers from two researchers (see Acknowledgments). We selected samples to achieve broad spatial coverage within the breeding ranges of most North American species of diurnal raptors and to achieve representative coverage of all habitats within these ranges. All samples were taken from nestlings or birds in juvenal plumages that were assumed to be collected at or near their natal territory on the basis of collection date. Within-year sample-date criteria varied across species and locations. We used Johnsgard (1990) to choose dates that minimized the chance that specimens were collected during spring migration, during fall migration, or aer postedging dispersal. In most cases, this resulted in collection dates in June or July, with some exceptions ranging into May, August, or, in a few cases, September. Original specimen collection years ranged from 1874 to 2003. No samples were taken from birds in downy plumages, because very young individuals may have Df values related to their parents wintering site rather than the local food web (Duxbury et al. 2003). We only collected feathers with relatively exact location, such as cities, small and well-dened geographic features, or counties in areas with regionally low Dp variance. Data for each of the reference population feather samples used here are available (see Acknowledgments).

July 2006]

Stable Isotopes and Bird Migration

825

F. 1. A map of North America showing sample locations of feathers analyzed in this study. Detailed location data for each sample are available (see Acknowledgments). Estimates of stable hydrogen isotope ratios in precipitation for each feather-sample location.For samples with exact location information (e.g., the name of a city), we created a circular buffer with a 10-km radius around the location for estimates of Dp for that site. We used buers because it was oen dicult to tell from specimen data whether the specimen was collected at or near the reported location. We took this approach, rather than using point estimates for each location, because specimens were collected by many dierent collectors across many years, without a standard protocol for reporting location. For locations that reported only county as geographic location, we used the county layer in ARCVIEW, version 3.2 (ESRI, Redlands, California), to acquire county polygons for each sample. For specimens with geographic-feature locations such as small lakes or creeks, we created polygons with 10-km buers around each feature. We used the Meehan et al. (2004) map of North American weighted-average, growing-season Dp (hereaer the MPM; see Acknowledgments). Finally, we acquired mean estimates for Dp from the MPM for each sample-location polygon using the summarize zones feature of the ARCVIEW SPATIAL ANALYST, version 2.0, extension (McCoy and Johnston 2001). Stable-isotope laboratory methods.The basemap and Goshute-migrant feather samples were analyzed for stable hydrogen isotope ratios (Df ) between 19 February and 14 April 2004 at the Stable Isotope Hydrology and Ecology Lab at the National Water Research Institute of Environment Canada in Saskatoon, Saskatchewan, following the detailed analytical protocols described in Wassenaar and Hobson (2003). All Df results are reported in delta () notation, in per-mil units () and normalized on the standard VSMOW-SLAP scale. Repeat analysis of standards during feather analyses yielded a laboratory repeatability of 1.0 for homogenized keratin samples. However, repeatability of feathers reanalyzed several months to a year aer their original analysis showed statistically signicant shis in Df values between analyses (paired t-test: P values < 0.03 for ve of six sample groups; n = 2840 per sample group). Systematic shis in Df values between analyses spanned from 9.2 to 6.4 per sample group (a range of 15.6), and the direction and magnitude of these shis diered signicantly among groups (F = 20.44, df = 5 and 179, P < 0.0001). Because Df values from repeat analyses were strongly correlated with original analyses (r2: 0.900.98), we standardized original Df values among studies using regressions from repeat analyses for all six sample groups. For example, the equation used to standardize Df values for the migrant Sharp-shinned Hawk feathers reported here was: standardized Df value = 13.90 + 0.90 original Df value. This step standardized results among several concurrent studies, because all corrections were made on the basis of the same repeat analysis session. Use of our base map in future studies will require that a subset of the feathers used

826

L S

[Auk, Vol. 123

to create the map are reanalyzed at the same time as the migrant samples. Limited supplies of original base-map feathers are available for this purpose by request from HawkWatch International (HWI; see Acknowledgments). Quantifying the relationship between stable hydrogen isotope ratios in feathers and in precipitation.We used least-squares regression to quantify the relationship between measured Df and estimated Dp for our reference population samples. We explored several models of the Df and Dp relationship that included covariates for species, feather-sample collection year, or within-season feather-collection date. None of these covariates contributed signicantly to the models (all P > 0.05), and they were not considered further. We mapped the regional trend in regression residuals to illustrate spatial variation in the relationship between Df and Dp. We used inverse distance weighting (Johnston et al. 2001) to interpolate Df and Dp regression residual values for each 1-km2 grid cell for North America. We set several interpolation input parameters to account for the fact that each of our reference feather-sample locations typically had only a single Df value, and Df is known to vary within raptor populations (Meehan et al. 2001). First, we used a large number of nearest neighbors (n = 20) for interpolations, so that regional paerns of residuals would reect multiple samples within a region. Second, we set the weighting coecient for each sample to a value of 1.0 to reduce the eect of any single observation on regional trends and to limit the aect of distant points on local values (Johnston et al. 2001). We conducted all statistical analyses using JMPIN, version 3.2.6 (Sall and Lehman 1996). We performed all spatial analyses and interpolations using the SPATIAL ANALYST extension in ARCVIEW (McCoy and Johnston 2001). Geographic-information-system base maps of stable hydrogen isotope ratios in feathers of North American raptors.We created a map of continental paerns in raptor Df using a two-step process. First, we used the regression equation from our reference population sample to create an initial map of estimated raptor Df values using the map calculator feature in SPATIAL ANALYST with the MPM as a base layer. This step multiplied each 1-km2 grid cell of the MPM by the slope of the regression equation and subtracted the intercept (e.g., Df = 0.908 MPM Dp 5.619). This produced

an elevation-explicit map of estimated Df values for each 1-km2 grid cell in North America by incorporating the MPMs treatment of the eect of elevation on Dp. Next, we used a map calculator to add our interpolated regression residuals to this layer to adjust the general relationship between Df and Dp by the regional variation in this relationship that we documented in our sample of raptor feathers (sensu Bowen and Wilkinson 2002). This produced a nal map of estimated North American raptor Df (with 1-km2 resolution) using the equation: Df = 0.908 MPM Dp 5.619 + interpolated Df/Dp regression residual. To assess the accuracy of predictions made by this map, we acquired map-based predictions of Df for all 264 of our original feather sampling sites and compared measured Df values with map-predicted Df values. Predicted and measured Df values were strongly correlated (r2 = 0.94). The mean dierence between predicted and measured Df values was (mean SD) 0.1 8.2. Ninety-ve percent of all predicted Df values were within 17.6 of measured values. Example application: Estimating the origins of migrant Sharp-shinned Hawks.We used breeding habitat associations and previous bandrecovery information to geographically limit our estimations before estimating the origins of migrants using Df (Smith et al. 2004, DeLong et al. 2005). We used a GIS coverage for North American biomes (Brown et al. 1998) to eliminate a large number of areas within the breeding range of Sharp-shinned Hawks that did not have suitable breeding habitat (e.g., deserts and grasslands). Therefore, our nal estimations of the origins of migrant Sharp-shinned Hawks include only forested areas where Sharpshinned Hawks are known to breed. We applied further constraints based on Homan et al. (2002), who demonstrated that Sharp-shinned Hawks banded in the Intermountain West only infrequently originate from breeding areas east of the eastern front of the Rocky Mountains or west of the Sierra and Cascade ranges within the continental United States. For each of the 173 measured Df values from our migrant sample, we ploed the measured value 8. This incorporated our assessment of map accuracy in the description of origins of migrants. For each Df value on our base map, we calculated the percentage of individuals in our migrant sample whose range of estimated

July 2006]

Stable Isotopes and Bird Migration

827

Df values included this value, creating a relative-frequency histogram. For example, 36 of 173 migrants in our sample (21%) had ranges of estimated Df values that included a Df value of 97. We considered all Df values that included <5% of our migrant sample to represent rare sources of migrants and did not plot these values. We used natural breaks in the data to split the resulting distribution into three dierent categories of relative abundance (abundant: >28% or more of all samples per Df value; fairly common: 1726%; and uncommon: 513%) that indicate the relative contribution of any predicted Df value on our feather map to the migrant sample. We assigned colors to each of these categories and displayed each Df value on our map of North American raptor Df with its corresponding relative-abundance color to produce a spatial histogram of the relative contribution of dierent source populations to our migrant sample. Thus, the relative abundance of migrants was represented spatially, without reference to the overall distribution of migrant Df values, and in proportion to abundance for each Df value on the map. This approach will work equally well for migrant Df samples that are normally distributed (Meehan et al. 2001) or skewed. The laer may be expected when more migrants at a given location are locally dispersing birds (DeLong et al. 2005) or when a small number of birds at a site are shortdistance migrants and a larger number are long-distance migrants from farther north (e.g., Peregrine Falcons [Falco peregrinus] captured in the Florida Keys; HWI unpubl. data). R Relationship between stable hydrogen isotope ratios in feathers and in precipitation in North American raptor feathers.The broad geographic extent of our reference sample produced a wide range of estimated Dp values for feather-sample locations (13 to 149) that were well distributed along the x-axis of our regression model (Fig. 2). Values of Dp in our sample are representative of the full range of Dp values encountered across most North American raptor breeding populations. Reference samples represented a similarly wide range of altitudes, from sea level to 3,180 m. However, some geographic regions had small feather sample sizes or were missing samples (e.g., northern Quebec,

F. 2. Regression model showing the relationship between D in raptor feathers (Df ) and D in precipitation (Dp) across North America. The model is based on a sample of 264 raptor feathers from 12 species from 255 different locations. Estimates of Dp are from Meehan et al. (2004). Nunavut, and the southern and southeastern United States; Fig. 1). Still, the present study represents the most geographically extensive reference sample of feathers used to investigate the relationship between Df and Dp for any North American bird group. The overall relationship between Df and Dp for North American raptors was strong (F = 426.95, df = 1 and 245, r2 = 0.62, P < 0.0001). The slope of the regression equation was 0.908, and the 95% condence interval (CI) for the slope was slightly lower than 1 (0.820.99). Our regression model had a y-intercept of 5.6 that did not dier signicantly from 0 (95% CI for the intercept: 13.1 to 1.9). Mapped residuals from the DfDp regression showed strong paerns of regional variation, which indicates that the relationship between Df and estimates of Dp may not be the same across all of North America (Fig. 3). In general, regression residuals were relatively high (indicating more positive Df values than expected from local Dp) along the Pacic Coast and within the arid Southwest, and residuals were relatively low within the northern continental interior. North American maps of estimated stablehydrogen isotope ratios in feathers.Several important geographic paerns in Df values are apparent in Figure 4. As in previous coarse maps of stable hydrogen isotopes in precipitation, Df values tend to decrease with both latitude and

828

L S

[Auk, Vol. 123

F. 3. Interpolated map of regional variation in the DfDp relationship across North America. Positive residual values indicate feather samples with higher D values than expected from local precipitation. Negative residual values indicate feather samples with lower D values than expected from local precipitation.

F. 4. Map of Df for North American raptors. Given standardized lab analysis, Df values from raptors of unknown origin captured during migration or at wintering sites can be plotted directly on the map to view their origins. In general, Df values decrease with increasing latitude. However, regional complexity in spatial patterns of Df increases in areas of topographic relief and along the Pacific coast, making interpretation of results based solely on Df problematic for some species, particularly in western North America. This map should be considered carefully in relation to the geographic range of a study species before sampling Df at migration or wintering sites to evaluate whether measurements of Df can clearly differentiate among potential source areas. elevation. And as in other elevation-explicit maps of Dp, Df values are more homogeneous in areas with lile topographic relief (e.g., the central and eastern United States) and become increasingly complex in areas with greater topographic relief (e.g., most of the western United States and, to a lesser extent, the Appalachians and Ozarks in the east). In coastal areas (particularly within western North America), Df values tend to be higher than in inland areas at similar latitudes, and there is a relatively steep gradient of decreasing Df values moving from

July 2006]

Stable Isotopes and Bird Migration

829

west to east along the Pacic Coast. Similar paerns of decreasing Df values between 80 and 130 occur latitudinally within the mountainous west and longitudinally in Alaska, converging at Df values below 130 in the Yukon Territory. The yellow band of Df values in Figure 4 (range: 110 to 120) illustrates the complexity of redundant Df values in the west, showing that migrants with Df values ranging from 110 to 120 could originate from high-elevation areas within the Rocky Mountains in Idaho and Montana, from lower-elevation mountainous areas in southern and central British Columbia, or from across the interior of Alaska. Maps of the estimated origins of migrant Sharpshinned Hawks.Figure 5 displays a map and frequency histogram of the estimated origins of fall migrant Sharp-shinned Hawks sampled in the Goshute Mountains, Nevada. The Df value for the study site on our base map was 73. Estimated origins were most frequent within the range of Df values between 108 and 126, representing two possible major source areas for Goshute migrants: (1) the Northern

Rockies from Idaho and Montana north through British Columbia, (2) the forests of interior Alaska, or both. Knowledge gained from bandrecovery studies suggests that the former is the more likely source area (Homan et al. 2002). Estimated origins were also relatively common in areas with Df values between 82 and 107, representing either (1) high-elevation forests in Central Idaho, eastern Oregon, and eastern Washington; (2) forested areas of central and western Alaska; or (3) coastal forests of western British Columbia. Few migrant Sharpshinned Hawks seemed to originate from forests near the migration banding site in Nevada or from areas to the far north within the western breeding range of Sharp-shinned Hawks in the Yukon Territory. D Relationship between stable hydrogen isotope ratios in feathers and in precipitation in North American raptors.As in previous stable hydrogen isotope studies, we found a strong relationship between raptor Df and Dp across

F. 5. Estimated origins of Sharp-shinned Hawks captured during fall migration in the Goshute Mountains, Nevada (star). Darker colors indicate Df values that contribute larger numbers of birds to the migrant sample. Black lines follow generalized contours of Df values from our base map of North American raptor Df.

830

L S

[Auk, Vol. 123

North America using a large sample of raptor feathers from 12 species. The regression equation describing this relationship diered from regression equations in previous studies with passerines (Hobson and Wassenaar 1997, Wassenaar and Hobson 2000, Hobson et al. 2001, Bowen et al. 2005) and raptors (Meehan et al. 2001, DeLong et al. 2005). We believe that comparative biological interpretations of regression equations in stable-isotope studies should proceed cautiously at this point, because regression equations may dier among studies for many reasons. We propose three nonexclusive hypotheses to explain dierences in regression equations among studies, all of which need further testing. First, dierences may be aributable to a lack of standardization among studies, both in lab methods for measuring Df and in the choice of models used for estimating Dp for feathercollection sites. Our repeat analyses question the comparability of results among studies conducted during dierent periods within the same lab, and questions remain about the comparability of results from dierent labs. Although the virtual equilibration of feather samples with keratin standards is theoretically promising (Wassenaar and Hobson 2003), more empirical data from repeat measurements of feather samples among seasons within the same lab and among labs are necessary to address the comparability of results among studies. In addition, previous studies have used a number of dierent Dp models (Hobson and Wassenaar 1997, revised in Wassenaar and Hobson 2000; Meehan et al. 2004; Bowen et al. 2005) to estimate Dp for feather-collection sites, and systematic differences among these models may aect regression equations. Second, dierences in regression equations among studies may be aributable to ecophysiological dierences among avian taxa. This should be addressed experimentally or by eld studies that compare Df among a number of species representing dierent ecological groupings and trophic positions within the same study area. Complementary local sampling of D in precipitation, plant material, and important prey items for each species would help to explain taxonomic dierences in the relationship between Df and Dp if they exist. All samples should be analyzed in the same lab at the same time to control for

possible lab-related variation when making these comparisons. Finally, dierences in regression equations among studies may be aributable to regional dierences, as shown here, in the Df and Dp relationship. Our reference sample included more samples from the Pacic Coast and the southwestern United States than previous studies. Unlike samples from interior or eastern locations, feathers from these regions tend to be more enriched in deuterium in relation to local precipitation. Thus, the wider spatial distribution of samples in the present study (compared with earlier studies that focused on eastern or central North America) could be responsible for the more positive intercepts observed for our reference sample. Until the consistency of the Df and Dp relationship among taxa is claried, the maps of estimated Df produced here should be used only for studies involving migrant raptors. Similarly, until the comparability of Df measurements among studies and labs is conrmed with empirical data, unknown-origin samples will have to be standardized for use with our base map by reanalyzing a subset of the feathers that were used to create the map. These feathers can be obtained by request from HWI. Regional variation in the relationship between isotope ratios in feathers and in precipitation.This is the rst study to demonstrate regional variation in the relationship between Df and Dp. Feather samples from the arid Southwest and the Pacic Coast had consistently higher Df Dp regression residuals compared with other areas of North America. Residuals were also relatively low in the northern continental interior. These regional paerns may be the result of regional anomalies in the Dp model used to estimate Dp for feather-sample locations or dierences in plant or avian ecophysiology in these regions. Wolf and Martnez del Rio (2000) found that the body-water pool of Whitewinged Doves (Zenaida asiatica) in the southwest was 20 enriched compared with local source moisture, and McKechnie et al. (2004) found that this enrichment may be aributable to the preferential evaporation of light isotopes during evaporative cooling in hot and arid environments. This physiological factor may contribute to relatively enriched values for Df in arid regions. Sternberg et al. (1984) showed that D values in desert plants with the CAM photosynthetic pathway tended to be more

July 2006]

Stable Isotopes and Bird Migration

831

enriched in relation to local precipitation than C3 or C4 plants. Thus, more enriched Df values in the Southwest may reect the higher proportion of CAM plants in this region. High residual values along the Pacic Coast and low residual values within the continental interior may be the result of the Dp model not fully accounting for the isotopic depletion of Pacic air masses as they move across the continent from west to east (Yonge et al. 1989, Rozanski et al. 1993). The MPM is based on generalized relationships between temperature, altitude, and Dp across all of North America. This model accounts for regional deviations from these large-scale relationships in the same way that we have in the present study, by adjusting estimated Dp values by interpolated model residuals from nearby Dp sampling stations (Meehan et al. 2004). However, there are less than ve of these stations along the Pacic Coast, which potentially limits the ability of this step to account for this potentially strong eect. Alternatively, there may be regional aspects of plant or avian ecophysiology along the Pacic Coast that led to relatively enriched environmental signals for D that are reected in the feathers of birds. Another possibility may be that growing-season estimates of Dp for this region do not accurately reect local Dp during the exact period when juvenile raptor feathers are being grown (Smith and Duy 2005). Although at this point unexplained, the relationship between Df and estimated Dp along the Pacic Coast is very dierent from this same relationship inland. Increased sampling for both Df and Dp along transects from the Pacic Coast inland at dierent latitudes would help to improve models of Df and Dp in western North America. Ploing migrant distributions on geographicinformation-system base maps of stable hydrogen isotope ratios in feathers.For the present study, we gave an example of one possible approach to ploing the origins of migrants on our base map of Df using Sharp-shinned Hawks sampled on migration in the Goshute Mountains of Nevada. We see this as a heuristic exercise to demonstrate the value of a Df base map and the power of GIS approaches. Together, they allow for spatially explicit estimations of the origins of migrants based on multiple criteria, including measurements of Df , habitat associations, and geographically dened yway boundaries.

We view the development of more robust approaches for ploing migrant distributions on such base maps as an important area for future research. Until methods are developed to truly address the error inherent in this approach, and additional validations of the base map are made by independent data sets, estimations of the origins of migrants using this approach should be viewed as preliminary, particularly in studies with small migrant-population sample sizes. This GIS-based approach could be extended to include additional data for other intrinsic markers that display geographic variation, such as morphology, genetics, or other stable isotopes such as carbon (13C) and nitrogen (15N). It is critical to realize, however, that this will be the case only if these data can be accurately mapped at the appropriate spatial scale to improve our ability to estimate the origins of migrants (e.g., the entire range of possible source origins of a migrant population). Given the spatial heterogeneity of D, 13C, and 15N across the breeding range of most migratory species, continuous and spatially explicit predictive surfaces for stable-isotope distributions (such as the model of North American raptor Df produced here) may produce more accurate descriptions of the origins of migrants than discrete site-classication models that assign samples to only a small number of potential source populations, not the entire range of populations that could be contributing to a migrant sample (Wassenaar and Hobson 2000, Farmer et al. 2003, Hebert and Wassenaar 2005). Classication models assume clinal variation in areas between sampled reference populations. This assumption is likely untenable for stable carbon and nitrogen isotopes, which vary tremendously at local or regional scales because of both natural and anthropogenic eects (Kelly 2000, Hebert and Wassenaar 2001, Dawson et al. 2002, Graves et al. 2002, Cerling et al. 2004), and is not valid for stable hydrogen isotopes in areas with large topographic relief, where Dp may vary between sampling sites because of the eect of elevation (Bowen and Wilkinson 2002, Meehan et al. 2004). Our study and previous studies of large-scale paerns in Dp demonstrate that geographic variation in D can be mapped at the relevant physical scale of the entire range of potential source areas for migrant populations, primarily because of the detailed understanding of the

832

L S

[Auk, Vol. 123

eects of environmental and physical parameters (such as temperature and elevation) on the geographic distribution of D (Dansgard 1964, Rozanski et al. 1993). A similarly mechanistic understanding has not been developed to describe geographic variation in other stable isotopes. To assess whether additional intrinsic markers will be helpful in delineating the origins of migrants, much more geographically widespread sampling is necessary to accurately map variation in these parameters at the relevant spatial scale for their use in estimating the origins of migrants (Webster et al. 2002, Rubenstein and Hobson 2004). Of particular interest would be the discovery of any other marker that could help to discriminate among regions where Df values are redundant (e.g., interior Alaska and the southern Rocky Mountains). A plea for standardization of methods and for increased comparative studies.North American growing-season Dp models have evolved rapidly during the brief history of the use of stable- hydrogen isotopes to track animal movements (Hobson and Wassenaar 1997, Meehan et al. 2004, Bowen et al. 2005). Large-scale models of Dp will most likely continue to evolve as the spatial coverage of Dp sampling stations from which models are built increases (Welker 2000). To avoid confusion in future comparative studies, care should be taken to clearly state which Dp model was used for deriving Dp estimates, and comparative studies should use the same models for Dp estimates for feather-collection sites. As Dp models are revised, Df data from reference population samples could also serve as independent data sets to compare Dp models (Meehan et al. 2004). In addition, we emphasize, as have others, that for direct comparisons among studies, careful aention must be paid to standardization of Df measurements (Wassenaar and Hobson 2003). We are skeptical that current lab methodologies produce comparable results among studies both within and among labs. Only by comparing and eventually pooling standardized results across numerous studies involving dierent taxa and regions will we be able to arrive at the sample sizes and geographic coverage that will be necessary to understand continental paerns in Df. To this eect, we view our base map of Df as a work-in-progress to be improved upon by others as sample sizes grow for standardized measurements of Df for more species and more regions, and as

additional maps are created of geographic patterns in Dp. For the sake of future comparative studies, all sample data from the present study, including Df values and sample locations, are released for general use on the HWI website (see Acknowledgments), and we encourage other investigators to do the same. Using the methods outlined here, and web-available Df and Dp data, more accurate maps of Df can be created as future studies collect additional samples in areas where this study had poor geographic or taxonomic representation. A We thank the following collections and individuals for the contribution of feather samples to this study: M. Gosselin (Canadian Museum of Nature); M. Peck (Royal Ontario Museum); P. Sweet, S. Kenney, and P. Campainolo (American Museum of Natural History); D. Willard and J. Bates (Field Museum of Natural History); C. Cicero (Museum of Vertebrate Zoology); G. Shugart (Slater Museum of Natural History); R. Fauce (Burke Museum); B. Alther and R. Ramey (Denver Museum of Nature and Science); C. White (Monte L. Bean Life Science Museum); D. Dyer (Montana State University); A. Kraer (Florida Museum of Natural History); J. Gerwin (North Carolina State Museum of Natural Sciences); R. Dickerman (Museum of Southwestern Biology); Cornell University Museum of Vertebrates; Wildlife Collection at Humboldt State University; University of Kansas Museum of Natural History; and G. Kaltenecker and J. Kirkley for samples from Idaho. All feather samples for the present study were collected and shipped under permit and according to national and international law. We thank J. Heath, G. Bowen, M. Wunder, J. DeLong, T. Meehan, K. Hobson, C. Martnez del Rio, and four anonymous reviewers for their insightful comments on various dras of this paper. We thank L. Wassenaar, P. Healy, and M. Benjamin at the Stable Isotope Hydrology and Ecology Lab of Environment Canada for their careful aention to detail in the preparation and analysis of feather samples. We thank our volunteer banding crews in the Goshute Mountains for their help in collecting migrant feather samples. Primary funding for this project was provided by the National Fish and Wildlife Foundation, Walbridge Fund, Columbus

July 2006]

Stable Isotopes and Bird Migration

833

Foundation, and HawkWatch Internationals (HWI) private donors and members. Additional nancial support for the Goshute Mountains migrant banding operation was provided by the Bureau of Land Management (Elko Field Oce, Bureau of Reclamation) Upper Colorado Regional Oce, Barrick Goldstrike Mines, Placer-Dome North America (Bald Mountain Mine), Bale Mountain Gold Company, Nevada Power Company, and Lahontan Audubon Society. The GIS base map (downloadable as a GIS grid le) and data for each of the referencepopulation feather samples are available at the HWI website: www.hawkwatch.org. Meehan et al.s (2004) map, the MPM, is available at biology.unm.edu/wolf/precipitationD.htm. L C B, G. J., L. I. W, K. A. H. 2005. Global application of stable hydrogen and oxygen isotopes to wildlife forensics. Oecologia 143:337348. B, G. J., B. W. 2002. Spatial distribution of 18O in meteoric precipitation. Geology 30:315318. B, D. E., F. W. R, S. E. F. 1998. A Classication of North American Biotic Communities. University of Utah Press, Salt Lake City. C, T. E., J. A. H, T. B. H. 2004. Stable isotope ecology in the Ituri Forest. Oecologia 138:512. C, C. P., J. D. B, R. T. H, X. F, T. W. S, G. R. G. 1997. The use of isotope tracers for identifying populations of migratory birds. Oecologia 109:132141. D, W. 1964. Stable isotopes in precipitation. Tellus 16:436468. D, T. E., S. M, A. H. P, P. H. T, K. P. T. 2002. Stable isotopes in plant ecology. Annual Review of Ecology and Systematics 33:507559. DL, J. P., T. D. M, R. B. S. 2005. Investigating fall movements of hatchyear Flammulated Owls (Otus ammeolus) in central New Mexico using stable hydrogen isotopes. Journal of Raptor Research 39: 1925. D, J. M., G. L. H, K. M. 2003. Changes in hydrogen isotope ratios in sequential plumage stages:

An implication for the creation of isotopebase maps for tracking migratory birds. Isotopes in Environmental and Health Studies 39:179189. F, A., R. R, G. L, C. B, C. K, I. R. 2003. Tracing the pathways of Neotropical migratory shorebirds using stable isotopes: A pilot study. Isotopes in Environmental and Health Studies 39: 169177. G, G. R., C. S. R, A. R. N. 2002. Stable isotope signature of philopatry and dispersal in a migratory songbird. Proceedings of the National Academy of Sciences USA 99:80968100. H, C. E., L. I. W. 2001. Stable nitrogen isotopes in waterfowl feathers reect agricultural land use in western Canada. Environmental Science and Technology 35:34823487. H, C. E., L. I. W. 2005. Feather stable isotopes in western North American waterfowl: Spatial paerns, underlying factors, and management implications. Wildlife Society Bulletin 33:92102. H, K. A. 1999. Tracing origins and migration of wildlife using stable isotopes: A review. Oecologia 120:314326. H, K. A., L. A, L. I. W. 1999a. Inuence of drinking water and diet on the stable-hydrogen isotope ratios of animal tissues. Proceedings of the National Academy of Sciences USA 96:80038006. H, K. A., G. J. B, L. I. W, Y. F, H. L. 2004. Using stable hydrogen and oxygen isotope measurements of feathers to infer geographical origins of migrating European birds. Oecologia 141:477488. H, K. A., K. P. MF, L. I. W, C. C. R, J. E. G. 2001. Linking breeding and wintering grounds of Bicknells Thrushes using stable isotope analyses of feathers. Auk 118:1623. H, K. A., L. I. W. 1997. Linking breeding and wintering grounds of Neotropical migrant songbirds using stable hydrogen isotopic analysis of feathers. Oecologia 109:142148. H, K. A., L. I. W. 2001. Isotopic delineation of North American migratory wildlife populations: Loggerhead Shrikes. Ecological Applications 11:15451553.

834

L S

[Auk, Vol. 123

H, K. A., L. I. W, O. R. T. 1999b. Stable isotopes (D and 13C) are geographic indicators of natal origins of monarch bueries in eastern North America. Oecologia 120:397404. H, S. W., J. P. S, T. D. M. 2002. Breeding grounds, winter ranges, and migratory routes of raptors in the mountain west. Journal of Raptor Research 35:97110. J, P. A. 1990. Hawks, Eagles, and Falcons of North America: Biology and Natural History. Smithsonian Institution Press, Washington, D.C. J, K., J. M. V H, K. K. 2001. Using ARCGIS Geostatistical Analyst. ESRI Press, Redlands, California. K, J. F. 2000. Stable isotopes of carbon and nitrogen in the study of avian and mammalian trophic ecology. Canadian Journal of Zoology 78:127. K, J. F., V. A, Z. D. S, D. M. F. 2002. Insights into Wilsons Warbler migration from analyses of hydrogen stableisotope ratios. Oecologia 130:216221. K, J. F., D. M. F. 1998. Tracking migrant songbirds with stable isotopes. Trends in Ecology and Evolution 13:4849. L, C. A., T. D. M, J. A. H. 2003. Estimating the latitudinal origins of migratory birds using hydrogen and sulfur stable isotopes in feathers: Inuence of marine prey base. Oecologia 134:505510. MC, J., K. J. 2001. Using ARCGIS Spatial Analyst. ESRI Press, Redlands, California. MK, A. E., B. O. W, C. M D R. 2004. Deuterium stable isotope ratios as tracers of water resource use: An experimental test with Rock Doves. Oecologia 140:191200. M, T. D., J. T. G, P. M. C. 2004. GIS-based model of stable hydrogen isotope ratios in North American growing-season precipitation for use in animal movement studies. Isotopes in Environmental and Health Studies 40: 291300. M, T. D., C. A. L, Z. D. S, R. B. S, R. N. R, A. C. S, R. K. M. 2001. Using hydrogen isotope geochemistry to estimate the natal latitudes of immature Coopers Hawks migrating

through the Florida Keys. Condor 103: 1120. R, K., L. A-A, R. G. 1993. Isotopic paerns in modern global precipitation. Pages 136 in Climate Change in Continental Isotopic Records (P. K. Swart, K. C. Lohmann, J. McKenzie, and S. Savin, Eds.). Geophysical Monographs, no. 78. R, D. R., C. P. C, R. T. H, M. P. A, J. R. W, G. R. G, N. C. T. 2002. Linking breeding and wintering ranges of a migratory songbird using stable isotopes. Science 295:10621065. R, D. R., K. A. H. 2004. From birds to bueries: Animal movement paerns and stable isotopes. Trends in Ecology and Evolution 19:256263. S, J., A. L. 1996. JMP Start Statistics: A Guide to Statistics and Data Analysis Using JMP and JMP IN Soware. Duxbury Press, Belmont, California. S, A. D., A. M. D, J. 2005. Variation in the stable-hydrogen isotope composition of Northern Goshawk feathers: Relevance to the study of migratory origins. Condor 107: 547558. S, R. B., E. C. G, B. O. W. 2004. Migratory movements of Sharp-shinned Hawks (Accipiter striatus) captured in New Mexico in relation to prevalence, intensity, and biogeography of avian hematozoa. Auk 121:837846. S, R. B., T. D. M, B. O. W. 2003. Assessing migration paerns of Sharpshinned Hawks Accipiter striatus using stable-isotope and band encounter analysis. Journal of Avian Biology 34:387392. S, L., M. J. D, H. A. 1984. Stable hydrogen isotope ratios of saponiable lipids and cellulose nitrate from CAM, C3 and C4 plants. Phytochemistry 23: 24752477. W, L. I., K. A. H. 2000. Stable-carbon and hydrogen isotope ratios reveal breeding origins of Red-winged Blackbirds. Ecological Applications 10: 911916. W, L. I., K. A. H. 2001. A stable-isotope approach to delineate geographical catchment areas of avian migration monitoring stations in North America.

July 2006]

Stable Isotopes and Bird Migration

835

Environmental Science and Technology 35: 18451850. W, L. I., K. A. H. 2003. Comparative equilibration and online technique for determination of non-exchangeable hydrogen of keratins for use in animal migration studies. Isotopes in Environmental and Health Studies 39:211217. W, M. S., P. P. M, S. M. H, S. B, R. T. H. 2002. Links between worlds: Unraveling migratory connectivity. Trends in Ecology and Evolution 17:7683. W, J. M. 2000. Isotopic (18O) characteristics of weekly precipitation collected across

the USA: An initial analysis with application to water source studies. Hydrological Processes 14:14491464. W, B. O., C. M R. 2000. Use of saguaro fruit by White-winged Doves: Isotopic evidence of a tight ecological association. Oecologia 124:536543. Y, C. J., L. G, H. R. K. 1989. An isotopic study of water bodies along a traverse of southwestern Canada. Canadian Journal of Hydrology 106:245255.

Associate Editor: K. A. Hobson