Annual productivity is an important parameter for the management of waterfowl populations. Fall age ratio (juveniles:total birds) is an index of productivity of the preceding breeding season. However, differences in the timing of migration between family groups and nonbreeding birds may bias age-ratio estimates. We examined temporal variation in age ratios of midcontinent greater white-fronted geese Anser albifrons frontalis from interior and northwestern Alaska at a northern autumn staging area near Delta Junction, Alaska. Photographic sampling conducted near Delta Junction resulted in an annual age ratio of 0.388 ± 0.004 (mean ± SE) in 2010 and 0.390 ± 0.001 in 2011. Our study demonstrated temporal variation in age ratios over the duration of the migration period during August and September. We recommend that sampling be conducted for 3-d periods at the beginning, middle, and end of the migration period to account for temporal variation in migration of family groups.
Waterfowl surveys that measure the proportion of hatch-year birds in a population are used to assess annual production and contribute to our understanding of population dynamics (Lynch and Singleton 1964; Prevett and MacInnes 1980; Ebbinge et al. 1990; Cowardin and Blohm 1992; Alisauskas and Lindberg 2002). Age ratio is the number of hatch-year birds (juveniles) relative to the total population of after-hatch-year birds (adults) and juveniles combined. In North America, most goose populations are subject to harvest by hunters (U.S. Fish and Wildlife Service [USFWS] 2016). Therefore, goose populations must be monitored to assess effects of harvest on abundance. Because geese have high fidelity to breeding, wintering, and staging sites, resource agencies have developed population-specific management plans for most identifiable goose populations (e.g., midcontinent greater white-fronted goose management plan). These plans typically include data on population size, harvest distribution, harvest rate, survival rate, and productivity. This information is used by managers when setting harvest regulations, while ensuring population conservation.
Harvest strategies are usually based on indices of population abundance, but better-informed management decisions require information on various demographic parameters including annual production of juveniles (Hestbeck et al. 1990). In addition to annual production, annual survival rates also have a significant influence on growth rate of goose populations (Schmutz et al. 1997; Gauthier et al. 2007). Although harvest has a direct effect on annual survival of some goose populations (Schmutz and Ely 1999), management of harvest alone may not yield defined conservation goals if demographic parameters other than survival influence population growth rates. For example, variation in annual production has been shown to affect population growth in geese irrespective of harvest (Cooch et al. 2001; Bromley and Rothe 2003); thus, estimates of age ratio contribute to determining causes of change in abundance.
Midcontinent greater white-fronted geese Anser albifrons frontalis nest in arctic and subarctic areas from the west shore of northern Hudson Bay in Canada west to the Seward Peninsula of western Alaska (Baldassare 2014). Despite this expansive breeding range and variation in breeding habitat types, migration timing, and winter distribution, these geese are managed as a single population (Ely et al. 2013). During fall migration, birds from this wide-ranging breeding area funnel south through central Canada along the Central and Mississippi flyways. Age-ratio data for midcontinent greater white-fronted geese are collected annually in Alberta and Saskatchewan (Sullivan 2010; R.T. Alisauskas, Canadian Wildlife Service, personal communication). For waterfowl managers specifically focused on midcontinent greater white-fronted geese from interior and northwestern Alaska, these data are of limited value as this subpopulation commingles with birds from the overall population. Unlike midcontinent greater white-fronted geese that breed in tundra habitats, geese from interior and northwestern Alaska nest in boreal and taiga habitats. This subpopulation is of management interest because of historical declines, low annual survival rates, and unique life history characteristics (Spindler et al. 1999; Spindler and Hans 2005; Sullivan 2010). Interior and northwestern Alaska midcontinent greater white-fronted geese differ from other midcontinent greater white-fronted geese in that they initiate fall and spring migrations earlier and winter in the most southern and western part of the winter range for the overall population (i.e., Mexico; Ely et al. 2013). Also, besides nesting in taiga vs. tundra habitats, all stages of their breeding cycle are approximately 3–4 wk earlier than the tundra-nesting component (Ely et al. 2013; J.A. Schmutz, U.S. Geological Survey, personal communication). Hence, there is reason to suspect that this subpopulation differs in other aspects of its biology and may require subpopulation specific adjustments in management to ensure their conservation.
Age-ratio estimates can be biased due to temporal and spatial variation in age classes (Ward et al. 2018). Breeding success in geese has been shown to affect timing of migration (Reed et al. 2003). Patterson and Hearn (2006) found that the percentage of juvenile pink-footed geese Anser brachyrhynchus decreased from September through November and speculated that differences in productivity throughout the breeding range could influence age-ratio estimates if the timing of migration varies among geese from different breeding areas. Similarly, Lambeck (1990) found that age ratios of brent geese Branta bernicla varied among seasons and sampling areas, speculating that it was due to differences in the timing of migration between family groups and failed or nonbreeders.
We examined temporal variation in age ratios of midcontinent greater white-fronted geese from interior and northwestern Alaska in a northern staging area. We sought to detect and mitigate potential bias in age-ratio estimates of geese near Delta Junction, Alaska, where interior and northwestern Alaska midcontinent greater white-fronted geese stage during fall migration. Our primary objective was to obtain age-ratio estimates specific for midcontinent greater white-fronted geese from interior and northwestern Alaska. In addition, we sought to estimate how the proportion of juveniles changed during fall staging to recommend appropriate timing for annual data collection efforts.
Midcontinent greater white-fronted geese stage near Delta Junction from early August through late September. We collected data at two sites near the confluence of the Tanana River and Clearwater Creek (2010–2011) and from grain fields approximately 25 km to the southeast of the confluence of the Tanana River and Clearwater Creek (2011). These sites serve as staging areas for fall migrating midcontinent greater white-fronted geese from breeding areas in interior and northwestern Alaska, before their movement southeast through central Alaska, as they begin their southern migration down the Central and Mississippi flyways (Bellrose 1980, Spindler and Hans 2005).
We sampled geese to determine age ratios at roosting sites and feeding field sites from August 16 through September 4, 2010, and from August 7 through September 19, 2011. We sampled geese by using photography for 5–7 d each week timed when geese were flying between their roosting areas on or near the Tanana River and nearby grain fields. Daily sampling efforts were similar within and among the 2 y and were not dependent on weather.
We determined age ratios by plumage characteristics observable in photographs by using single-lens reflex digital cameras with the following settings: center weighted or spot focusing, 800 ASA, 1/1,000-s shutter speed, and exposure compensation +2 stops. We took photographs each day of as many flying groups of geese as possible. When we took multiple photographs of the same group, we selected a single photo with the largest number of age-identifiable birds to estimate age ratio in that group. The sum of the total number of geese counted in all photographed groups provided an index to the actual number of geese in the fall staging population on that day. In some cases, inadequate lighting, poor focus, distance, and obstructions prevented age-class distinction of all birds photographed. In these cases, we only considered individual geese as identifiable if we had a view of breast plumage. Adults have black patches on the belly whereas juveniles have a plain belly, so we classified each bird photographed as juvenile, adult, or unidentifiable.
To verify that the birds we monitored were from interior Alaska, we analyzed the origin of banded birds harvested by hunters near Delta Junction. Of the 65 recoveries of banded birds in the vicinity of Delta Junction, 63 (96.9%) were from birds banded in interior and northwestern Alaska (U.S. Geological Survey, Bird Banding Report, retrieved January 26, 2012). These recoveries represented geese from the Innoko River drainage (60.3%), Koyukuk River valley (14.2%), Kanuti River basin (3.1%), Selawik lowlands (12.7%), Noatak River valley (3.1%), and Seward Peninsula (6.3%). Banding efforts have focused on the largest concentrations of molting geese, so recovery proportions are likely reflective of the distribution of molting birds.
It was not known whether families of geese with young and geese without young stay in the fall staging area for the same duration. Observations of a few neck-collared and radio-marked geese indicate birds do not stay for more than a few days in the area near Delta Junction (USFWS 2008, unpublished data). We assumed that many different groups of geese were present for short times during the fall staging period and duration of stay was independent of productivity status of the geese.
The sampling unit was the best photograph from each group of geese. The number of birds identifiable to age and the size of groups sampled varied among photographs. Due to changing abundance of geese in the staging area during our sampling period, the number of photographs taken each day varied. For the sampled estimated age ratio to reflect the average age ratio from the entire fall population, we sought to use equal field effort throughout. We considered that the daily sums of total birds in our photographs were proportional to abundance of geese for each day throughout fall staging. The total number of birds photographed on any given day provided an index to population size in the staging area for that day and thus the relative contribution or weight of each day toward the total age ratio for the entire fall season.
We calculated the mean and variance of age ratio on a given day using a ratio estimator. The ratio estimate provides the weighted average, calculated as the sum of all juveniles divided by the sum of all birds. Thus, photos with more birds received more weight as is appropriate for cluster sampling for proportions with unequal size clusters (Cochran 1963). We calculated the estimated proportion of juveniles as follows:
where ji is the number of juveniles in each photo, ni is total number birds aged in each photo, and m is the total number of photos.
We calculated variance as follows:
We obtained an equivalent estimate with a generalized linear model (glm) function in R (R Development Core Team 2011) by using glm(cbind(juv,adult) ∼ −1+dayf, data=data2, family=quasibinomial), where dayf is a factor for each day and −1 indicates no intercept term. The quasibinomial distribution allows for variance inflation due to overdispersion related to excesses of 0's and 1's in the observed proportions. Standard error of daily age ratios from the glm averaged approximately 30% lower on average than from the ratio estimator. A similar glm used a polynomial fit to smooth the trend in the age ratio with day and day^2 as numeric factors instead of dayf. This quadratic trend model estimated standard errors (SEs) as 76% smaller than the ratio estimator for each day.
We calculated the mean age ratio for the fall staging population each year in two ways. For the first method we simply pooled all the photographs across days. The implicit assumption was that sampling was proportional to the abundance of geese throughout the time period and that any change over time in age ratio was representatively sampled. The second method of calculation relied on estimating smoothed trends for both the age ratio and abundance of birds over individual days of the fall season. Smoothing any trend in age ratio and weighting daily estimates of abundance provides more defendable estimates if sampling is more sporadic or influenced by factors not related to the abundance of geese.
We constructed a full glm for age ratio each day as a function of additive factors for day, day-squared, total birds, hour, location, and proportion-identified. We progressively eliminated the nonsignificant factors of hour, time of day, location (river, two grain fields, high migrating flocks), and proportion of birds identifiable (photographic quality). Significant factors were day, day-squared, and total birds. The glm included appropriate weighting for the amount of data each day (number of photographs and number of identified birds) and used a quasibinomial for the expected distribution of sampling error. We used backward elimination (R procedure “Drop1”) to reduce the model to those significant factors. We also verified the best-fit model by individually adding each factor back into the model (R procedure “Add1”). We analyzed each year of data separately.
In 2010, we took approximately 1,700 photographs of 580 groups of midcontinent greater white-fronted geese. In 2011, we took approximately 2,500 photographs of 980 groups. We only selected photographs with the largest number of identifiable birds from each group for analysis. This resulted in 441 photographs with 7,572 total birds and 5,892 age-identifiable birds in 18 d in 2010 and 838 photographs with 13,410 total birds and 10,410 age-identifiable birds in 28 d in 2011.
The glm with all significant coefficients included day, day-squared, and total birds in each photograph. Photographs with fewer total birds often contained only a few family groups of paired adults and their young, therefore having higher age ratios. Groups of larger size often included failed or nonbreeding adult geese, hence lower age ratios. Although total birds was a significant factor in the generalized linear model, the magnitude of its influence was small compared to variation over days. The significant coefficients were the same for both backward and forward selection, and for 2010 and 2011.
During 2010, surveys began August 16 and ended September 4. Considering days in which we sampled 10 or more groups, the daily age ratio varied from 0.272 ± 0.040 (mean ± SE, n = 27) on September 3, 2010, to 0.507 ± 0.039 (n = 18) on August 23, 2010. We found a lower age ratio during the beginning and end of the staging period with the highest age ratios occurring during the middle of the staging period (Figure 1A). During 2011, surveys began earlier and continued longer than in 2010 (August 7 through September 19, 2011). On days in which we sampled 10 or more groups, the daily age ratio varied from between 0.191 ± 0.045 (n = 34) on August 18, 2011, to 0.557 ± 0.042 (n = 12) on September 17, 2011. We found age ratios increased from the beginning to the end of the staging (Figure 1C).
From the glm coefficients, we calculated a smoothed predicted age ratio for each day, including those days we did not sample (Figures 1A and 1C). We set total birds per photograph equal to 20 birds, standardizing total birds per photograph that averaged 26 in 2010 and 21 in 2011 (Figures 1B and 1D). The contribution of each seasonal total, the weight was the predicted total number photographed each day divided by the total of predicted numbers over all days (Figures 1B and 1D). We estimated the annual age ratio of the fall staging population by summing over all days the age ratio multiplied by weight for each day. We calculated the seasonal mean and variance (SE) of age ratio to be 0.388 ± 0.004 in 2010 and 0.390 ± 0.001 in 2011.
Our photographic surveys conducted near Delta Junction during 2010 and 2011 found little variation in the overall age ratio between years. Previous years observations at Delta Junction yielded age ratio estimates of 0.29, 0.40, 0.35, 0.10, and 0.35 for 2005, 2006, 2007, 2008, and 2009, respectively (Marks and Fischer 2012). Further south on the fall migration corridor in Alberta and Saskatchewan, Canada, midcontinent greater white-fronted geese from interior and northwestern Alaska commingle with midcontinent greater white-fronted geese from arctic tundra breeding areas across the Arctic Coastal Plain of Alaska and Canada (Ely et al. 2013). There, ground surveys are conducted annually in Saskatchewan throughout the fall staging period from September 1 through early November, which coincides with the hunting season. In 2010, the age ratio was 0.13 ± 0.03 (n = 50) and 0.13 ± 0.03 (n = 53) in 2011 (R.T. Alisauskas, Canadian Wildlife Service, unpublished data). For both years, no change in age ratio was detected during the fall staging period. The higher age ratio estimates we observed in Alaska compared to Saskatchewan could be due to regional differences in production and survival, survey methodologies (photography vs. ocular counts), migration mortality of juveniles, and the high susceptibility of juveniles to harvest.
Age-ratio data have yielded important population level insights for several goose species. Researchers conducted age-ratio surveys on lesser snow geese Anser caerulescens that breed on Wrangel Island, Russia, during their fall migration down the Pacific Flyway and found that juveniles experienced little mortality during the second half of their autumn migration (Ely et al. 1993). Other researchers have used age-ratio data to measure density-dependent effects on arctic breeding grounds, revealing that factors other than hunting influence population dynamics of geese (Alisauskas 2002). Timm and Dau (1979) analyzed production, mortality, distribution, and inventory data for Pacific greater white-fronted geese to determine factors possibly regulating population size. They used age-ratio data from hunter-contributed federal waterfowl parts survey data compared to age ratios from field surveys and found that immature geese were more than twice as prevalent in hunter bags than adults because of the greater vulnerability of young birds to harvest (Timm and Dau 1979). The proportion of juveniles harvested in the entire Central Flyway based on parts collection was 0.41 in 2010 and 0.46 in 2011 (Kruse 2014). These values are not directly comparable to field counts and would be expected to yield a greater age ratio because juveniles are more susceptible than adult geese to hunter harvest (Chapman et al. 1969; Grieb 1970). It is thus apparent that collecting age ratios of geese at staging sites near breeding grounds before the hunting season provides less biased productivity estimates.
There are different hypotheses as to how autumn age-ratio estimates of geese may vary over time. Successful breeding geese may initiate migration earlier and move faster in the fall than nonbreeding geese (Prevett et al. 1982). Barry (1967) observed that nonbreeding midcontinent greater white-fronted geese in Northwest Territories, Canada, regained flight after molt up to 3 wk earlier than family groups. If the difference in timing of molt affects the timing of migration, then nonbreeding adults could migrate earlier than family groups, which supports findings by Miller et al. (1968), who found a lower percentage of juvenile midcontinent greater white-fronted geese in flocks first arriving in Saskatchewan. Similarly, Reed et al. (2003), who studied greater snow geese Chen caerulescens atlantica, showed that nonbreeding geese and geese that had failed in their nesting attempts molted earlier and initiated fall migration before breeding geese. Another hypothesis is that observed age ratios change at fall staging sites as groups of birds from different breeding areas with varying levels of breeding success arrive at different times (Lambeck 1990).
Our results were equivocal concerning arrival of family groups relative to nonbreeders near Delta Junction. The within-season pattern of age ratio differed between 2010 and 2011. In part, the trend was better estimated by the longer duration of sampling in 2011, but these data indicated considerable interannual differences in pattern of migration of midcontinent greater white-fronted geese near Delta Junction. In 2010, family groups arrived at the same time as nonbreeders. In 2011, the proportion of juveniles increased into September, indicating that nonbreeding or failed breeding adults departed earlier, followed by increasing numbers of family groups later in the migration period. Inference and conclusions from this study would be strengthened by additional years of data collection, knowledge of annual variation in timing of nesting that can influence migration timing, and better understanding of the breeding derivations of geese staging near Delta Junction.
Two untested assumptions are critical components of our analysis and sampling recommendations. First is that the total number of birds photographed on each day of sampling is approximately proportional to the daily abundance of birds in the fall staging population. The second assumption is that duration of stay in the fall staging area was independent of productivity status of the geese. Our use of the smoothed age ratio over days weighted by relative abundance alleviated the need to assume that family groups arrive at the same time as non- or failed breeders.
Timing, frequency, and duration of photographic sampling periods are significant variables in standardized age-ratio sampling. To account for variation in timing in migration of midcontinent greater white-fronted geese near Delta Junction, we recommend photographic surveys to determine age ratios during 3-d periods at the beginning, middle, and end of the total migration period. The average proportion from three or four consecutive days of data fit much closer to the seasonal patterns. The seasonal pattern differed markedly between years; therefore, estimating the pattern of change each year was important. Adequate data for a smoothed fit of the pattern could be accomplished by completing one early, a few mid, and one late cluster of sampling days. The methods and sampling effort each day should be approximately equal so that the number of photographed birds provides a valid index to the population size of staging geese. The smoothed trend in age ratio each day weighted by the relative daily population size provides a practical, design-based, overall annual estimate of productivity.
Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any archived material. Queries should be directed to the corresponding author of the article.
To cite this archived material, please cite both the journal article (formatting found in the Abstract section of this article) and the following recommended format for the archived material.
Schock WG, Fischer JB, Ely CR, Stehn RA, Welker JM, Causey D. 2018. Data from: Variation in age ratio of midcontinent greater white-fronted geese during fall migration. Journal of Fish and Wildlife Management 9(1):xx–xx. Archived in Dryad Digital Repository: https://doi.org/10.5061/dryad.28660
Data A1. Microsoft Excel file that contains raw survey data for midcontinent greater white-fronted geese Anser albifrons frontalis migrating near Delta Junction, Alaska, during 2010 and 2011, organized in a single tab, “Data A1 – 2010–2011 Raw Data.” The “2010–2011 Raw Data” tab contains date, time, total adults, total juveniles, total not identified, total sample, proportion of juveniles, age ratio, percentage of birds identified, and location.
Found at DOI: https://doi.org/10.5061/dryad.28660 (451.3 KB XLSX).
Reference A1. Bromley RG, Rothe TC. 2003. Conservation assessment for the dusky Canada goose (Branta canadensis occidentalis Baird). U.S. Department of Agriculture Forest Service, General Technical Report PNW-GTR-591, Pacific Northwest Research Station, Portland, Oregon.
Found at DOI: https://doi.org/10.5061/dryad.28660 (538.2 KB PDF).
Reference A2. Kruse KL, compiler. 2014. Central flyway harvest and population survey data book. U.S. Fish and Wildlife Service, Lakewood, Colorado.
Found at DOI: https://doi.org/10.5061/dryad.28660 (722 KB PDF).
Reference A3. Lynch JJ, Singleton JR. 1964. Winter appraisals of annual productivity in geese and other water birds. Waterfowl Trust Annual Report 15:114–126.
Found at DOI: https://doi.org/10.5061/dryad.28660 (2.492 MB PDF).
Reference A4. Marks D, Fischer JB. 2012. Data from: Alaska midcontinent greater white-fronted geese-Project updates, 2010–2011 Field Seasons. Unpubl. U.S. Fish and Wildlife Service report, Migratory Bird Management, Anchorage, Alaska.
Found at DOI: https://doi.org/10.5061/dryad.28660 (635.4 KB PDF).
Reference A5. Miller HW, Dzubin A, Sweet JT. 1968. Distribution and mortality of Saskatchewan-banded white-fronted geese. Transactions of the North American Wildlife and Natural Resources Conference 33:101–119.
Found at DOI: https://doi.org/10.5061/dryad.28660 (2.582 MB PDF).
Reference A6. Spindler MA, Lowe JM, Fujikawa JY. 1999. Trends in abundance and productivity of white-fronted geese in the taiga of northwestern and interior Alaska. Report to the Central Flyway Technical Committee. U.S. Fish and Wildlife Service, Koyukuk/Nowitna NWR Complex, Galena, Alaska.
Found at DOI: https://doi.org/10.5061/dryad.28660 (1.343 MB PDF).
Reference A7. Spindler MA, Hans MR. 2005. Nesting biology and local movements of female greater white-fronted geese in west-central Alaska. U.S. Fish and Wildlife Service, Galena, Alaska.
Found at DOI: https://doi.org/10.5061/dryad.28660 (1.924 MB PDF).
Reference A8. Sullivan B. 2010. Management plan for midcontinent greater white-fronted geese. Central Flyway Waterfowl Technical Committee, Denver, Colorado.
Found at DOI: https://doi.org/10.5061/dryad.28660 (353.7 KB PDF).
Reference A9. U.S. Fish and Wildlife Service. 2016. Waterfowl population status, 2016. U.S. Department of Interior, Washington, D.C.
Found at DOI: https://doi.org/10.5061/dryad.28660 (4.288 MB PDF).
This project was a collaborative effort among the USFWS, U.S. Geological Survey, and University of Alaska Anchorage. We thank the USFWS for funding and logistical support throughout the project. In addition to project support, Richard Seward, Steve Dubois, David Davenport, John Haddix, Jeffrey Mason, Ronald Robinson, Michael Schultz, and Scott Schultz provided valuable insight on midcontinent greater white-fronted goose migration near Delta Junction. Special thanks to Charles Schock, Diane Schock, Delphine Dura, Ellen Clark, Elizabeth Goldsmith, Karen Mailer, and Lydia Schock for enduring early mornings and long days in the marshes to collect data for this project. We thank the anonymous reviewers and the Associate Editor who substantially improved the manuscript. Lastly, we are grateful to Ray Alisauskas (Canadian Wildlife Service) for providing fall age-ratio data from Saskatchewan. W.G.S. also extends thanks to the University of Alaska Anchorage Graduate Student Association for providing grants for this project.
Any use of trade, product, website, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government.
Citation: Schock WG, Fischer JB, Ely CR, Stehn RA, Welker JM, Causey D. 2018. Variation in age ratio of midcontinent greater white-fronted geese during fall migration. Journal of Fish and Wildlife Management 9(1):340–347; e1944-687X. doi:10.3996/112015-JFWM-117
The findings and conclusions in this article are those of the author(s) and do not necessarily represent the views of the U.S. Fish and Wildlife Service.