337

Apparent survival, natal philopatry, and recruitment of Barrow’s goldeneyes (Bucephala islandica) in the Cariboo–Chilcotin region of British Columbia, Canada W.S. Boyd, B.D. Smith, S.A. Iverson, M.R. Evans, J.E. Thompson, and S. Schneider

Abstract: We used capture–resight data to evaluate apparent survival, natal philopatry, and recruitment of Barrow’s goldeneyes (Bucephala islandica (Gmelin, 1789)) in British Columbia, Canada. Median ages of first pairing and first breeding for females were 2 years and 3 years, respectively. The Cormack–Jolly–Seber model that best fit our data indicated that apparent survival rates (F) differed according to sex, year, and age class at marking. Estimates were similar for after-hatch-year (AHY) females (0.62) and AHY males (0.58), which was consistent with predictions. However, contrary to predictions, apparent survival rates of hatch-year (HY) females (0.68) were similar to those of AHY females and significantly higher than those of HY males (0.35). We interpret this difference as being primarily related to higher dispersal probabilities by HY males. Also evident was a negative correlation between apparent survival rate during the 1st year after capture for HY birds and their subsequent apparent survival rates, which suggests that probability of dispersal increased after these birds reached reproductive maturity and began to compete for breeding territories. We interpret this as evidence for density-dependent control of access to limited resources such as nest cavities. Re´sume´ : Nous utilisons des donne´es de captures et d’observations subse´quentes afin d’e´valuer la survie apparente, la philopatrie natale et le recrutement du garrot d’Islande (Bucephala islandica (Gmelin, 1789)) en Colombie-Britannique, Canada. Les aˆges moyens de premie`re formation des couples et de premie`re reproduction chez les femelles sont respectivement de 2 ans et de 3 ans. Le mode`le Cormack–Jolly–Seber qui s’ajuste le mieux a` nos donne´es indique que les taux de survie apparente (F) varient en fonction du sexe, de l’anne´e et de la classe d’aˆge au moment du marquage. Les estimations pour les femelles de l’anne´e apre`s l’e´closion (AHY) (0,62) et pour les maˆles AHY (0,58) sont semblables, ce qui s’accorde aux pre´dictions. Cependant, contrairement aux pre´dictions, les taux de survie apparente des femelles de l’anne´e de l’e´closion (HY) (0,68) sont semblables a` ceux des femelles AHY et significativement plus e´leve´s que ceux des maˆles HY (0,35). Nous interpre´tons cette diffe´rence comme e´tant principalement le reflet des probabilite´s plus e´leve´es de dispersion des maˆles HY. Il apparaıˆt aussi une corre´lation ne´gative entre la survie apparente durant la premie`re anne´e apre`s la capture chez les oiseaux HY et leurs taux subse´quents de survie apparente, ce qui indique que la probabilite´ de dispersion augmente une fois que ces oiseaux atteignent la maturite´ reproductive et commencent a` faire compe´tition pour les territoires de reproduction. Nous interpre´tons ces faits comme des indications d’un controˆle de´pendant de la densite´ de l’acce`s aux ressources limite´es, telles que les cavite´s de nidification. [Traduit par la Re´daction]

Introduction As a cavity-nesting species, Barrow’s goldeneyes (Bucephala islandica (Gmelin, 1789)), like the closely related common goldeneyes (Bucephala clangula (L., 1758)) and buffleheads (Bucephala albeola (L., 1758)), often face competition for suitable nest sites in desirable habitats (Eadie et al. 1998). This is particularly true in regions of intensive

forest management (Po¨ysa¨ and Po¨ysa¨ 2002). Moreover, both male and female goldeneyes exhibit territorial behavior during the breeding season (Savard 1984, 1988a) and territoriality can limit the breeding density of hole-nesting birds (Newton 1994). Banding studies indicate that return rates of goldeneyes to breeding areas are generally high compared with other duck species (Nilsson 1971; Dow and Fredga 1983; Savard and Eadie 1989; Ludwichowski et al. 2002),

Received 16 October 2009. Accepted 27 February 2009. Published on the NRC Research Press Web site at cjz.nrc.ca on 31 March 2009. W.S. Boyd1 and B.D. Smith. Science and Technology Branch, Environment Canada, Pacific Wildlife Research Centre, 5421 Robertson Road, Delta, BC V4K 3N2 Canada. S.A. Iverson and M.R. Evans. Centre for Wildlife Ecology, Department of Biological Sciences, Simon Fraser University, Burnaby, BC V5A 1S6, Canada. J.E. Thompson. Ducks Unlimited Canada, 10720 178 Street, Suite 200, Edmonton, AB T5S 1J3, Canada. S. Schneider. Canadian Wildlife Service, Environment Canada, Pacific Wildlife Research Centre, 5421 Robertson Road, Delta, BC V4K 3N2, Canada. 1Corresponding

author (e-mail: [email protected]).

Can. J. Zool. 87: 337–345 (2009)

doi:10.1139/Z09-018

Published by NRC Research Press

338

and successful nesters frequently reuse nest cavities in successive years (Savard 1988b). However, many details, including the ecological correlates and age-specific dynamics of survival, dispersal, and recruitment of Barrow’s goldeneyes, remain poorly understood. To address this information gap, we used mark–resight data to evaluate return rates of Barrow’s goldeneyes in the Cariboo–Chilcotin region of interior British Columbia, Canada, and tested a set of predictions derived from ecological theory. Our study follows on investigations of goldeneyes (Barrow’s and common) conducted during the 1980s in the same study area (Savard and Eadie 1989). We expanded these investigations to incorporate statistical advances in the evaluation of live-encounter data, including model development based on goodness of fit and likelihood ratio testing (Lebreton et al. 1992) and model selection using information– theoretic approaches (Burnham and Anderson 2002). In our analysis, we used data collected over an 8-year study period between 1995 and 2002, and used a series of Cormack–Jolly–Seber (CJS) models in program MARK (White and Burnham 1999). CJS models correct for the resighting rate (p) and yield estimates of apparent survival (F) that are the product of the probabilities of true survival and site fidelity. Our candidate set of models allowed explicit consideration of model variation and the explanatory power of a number of individual covariates, including sex, age, study site, and cohort. We predicted that (i) apparent survival rates for Barrow’s goldeneyes should be high for both after-hatch-year (AHY) males and AHY females owing to benefits of territorial defense and nest-cavity familiarity of females, but that rates may differ between sexes; (ii) apparent survival rates of AHY Barrow’s goldeneyes should be higher than those of hatch-year (HY) birds for both sexes owing to the combined effects of lower overwinter survival of HY birds and dominance interactions, which should lead to higher dispersal rates by young compared with adults; and (iii) future apparent survival should be positively correlated with current apparent survival for established breeders and negatively correlated among potential recruits attempting to establish territories owing to competition for limited key resources such as nest cavities, which in turn force a higher proportion of individuals to disperse.

Materials and methods Study area Our study was conducted in the Cariboo–Chilcotin region of British Columbia near Riske Creek (52807’N, 122827’W). The study area encompasses 117 ponds and small lakes over a 300 km2 area and is located on a rolling plateau northwest of the confluence of the Chilcotin River and Fraser River. It is characterized by extensive open grasslands and dry forest consisting of Douglas-fir (Pseudotsuga menziesii (Mirbel) Franco), lodgepole pine (Pinus contorta Dougl. ex Loud.), and trembling aspen (Populus tremuloides Michx.) (Krajina 1969). The Riske Creek area is recognized as one of the most important breeding areas for waterfowl in British Co-

Can. J. Zool. Vol. 87, 2009

lumbia (McKelvey and Munro 1983)2 and is a core breeding area for Pacific populations of Barrow’s goldeneyes (Eadie et al. 2000). Wetlands within the study area range in size from 0.1 to 54.0 ha, most of which are shallow, alkaline, devoid of fishes, and contain diverse and abundant aquatic invertebrate communities (Boyd and Smith 1989). The study area is divided into three regions: Bald Mountain (BM), Becher’s Prairie (BP), and Stack Valley (SV). Bald Mountain is located approximately 15 km southwest of Riske Creek and characterized by dense coniferous and deciduous forest interspersed with grassland. Becher’s Prairie is adjacent to and immediately north of Riske Creek and is characterized by prairie pothole-like wetlands and grasslands. Forested areas in Becher’s Prairie are less dense and more naturally fragmented than those of Bald Mountain and Stack Valley. Stack Valley is located along the northern perimeter of Becher’s Prairie at the interface between prairie grasslands and dense coniferous forest. It is characterized by naturally fragmented coniferous forest interspersed with deciduous tree species. Densities of Barrow’s goldeneyes in the study area have been enhanced through a series of nest-box programs beginning in the early 1980s (Savard 1986). The boxes have periodically been restored and replaced (Thompson 1996; Evans et al. 2002) to counteract natural deterioration or damage caused by black bears (Ursus americanus Pallas, 1780). In addition to periodic increases in nest cavities, 97 adult female Barrow’s goldeneyes were collected from 1993 to 1994 (1993: n = 49; 1994: n = 48) as part of a study on reproductive energetics (Thompson 1996). Hence, Barrow’s goldeneyes at Riske Creek have experienced several ‘‘experimental’’ perturbations, which may have affected competition for limiting resources during the course of our study. Field methods Barrow’s goldeneyes were captured as both AHY adults of unknown age and as HY ducklings during six breeding seasons from 1995 to 2000. Most AHY females were captured on their nests during incubation in late May and early June, whereas most AHY males were captured in May using underwater gill nets or decoy traps. Ducklings (class IIc and III) were captured with their brood hens in late July using drive traps and sexed by cloacal examination. All birds were fitted with uniquely colored and shaped nasal disks (Lokemoen and Sharp 1979). Each year from 1996 to 2002, approximately 12–13 surveys were conducted on a weekly basis from early May to late July to record the abundance of Barrow’s goldeneyes and locations of nasal-marked individuals. The behavioral response of goldeneyes when approached is to swim to the centre of the lake, as opposed to dabbling ducks which often hide in emergent vegetation near shore or flush. Moreover, most ponds were small enough that nasal markers could easily be seen using a 60 telescope from the shoreline, which gave us confidence that detection rates were high when marked birds were present. Supplemental data were collected to determine if marked individuals were paired and whether or not they had bred. Paired birds were identified based on their behavior, which

2 R.

McKelvey and W. Munro. 1983. Cooperative waterfowl management plan for British Columbia. Unpublished report for the Canadian Wildlife Service, Pacific and Yukon Region, Delta, British Columbia, and B.C. Fish and Wildlife Branch, Victoria, British Columbia. Published by NRC Research Press

Boyd et al.

339 Table 1. The recovery matrix (mi,j, where i is 1995–2001 and j is 1996–2002) for hatch-year (HY) and unknown age after-hatch-year (AHY) Barrow’s goldeneyes (Bucephala islandica) banded and resighted in subsequent years at Riske Creek, British Columbia. Year j Year i

Ni

Ri

1996

1997

1998

1999

2000

2001

2002

2 26

4 0 54

3 0 4 48

0 0 0 4 36

0 0 0 0 2 24

0 0 1 1 0 1 11

2

1 3

0 0 8

0 2 1 5

0 0 0 0 2

0 0 0 1 0 1

0 0 0 0 0 0 1

7

1 26

0 5 53

1 0 6 39

1 4 3 11 34

0 0 0 2 3 57

2 0 0 0 2 10 33

0

0 4

0 1 18

0 0 2 8

0 0 0 3 5

1 0 0 1 2 4

0 0 0 0 2 3 0

AHY females resighted 1995 68 68 30 1996 11 41 1997 53 81 1998 26 84 1999 20 75 2000 11 51 2001 0 26 AHY males resighted 1995 7 7 1996 15 17 1997 17 21 1998 0 8 1999 4 12 2000 1 3 2001 0 2 HY females resighted 1995 39 39 1996 67 74 1997 67 94 1998 70 128 1999 40 86 2000 90 143 2001 0 62 HY males resighted 1995 38 38 1996 86 86 1997 78 82 1998 83 102 1999 54 64 2000 99 107 2001 0 8

included courtship displays, proximity to individuals of the opposite sex, mate guarding, and territoriality. A marked female in a nest with eggs or on a pond with ducklings was interpreted as evidence of breeding. Statistical analysis We used data on the number of marked animals released (Ni) and resighted (Ri) in year i to create a recovery matrix for birds resighted in future years j (mi,j; Table 1). This recovery matrix informed the CJS models that we analyzed in program MARK (Pollock et al. 1990; Lebreton et al. 1992; White and Burnham 1999). Our candidate model set was guided by a priori predictions to explain variation in apparent survival and resighting probability. We organized our mark–recapture data into 12 ‘‘groups’’ (White and Burnham 1999), recognizing all combinations of site (BM, BP, SV)  sex (male, female)  age class at marking (HY, AHY). Our data set included 8 encounter occasions (1995–2002 breeding seasons), which allowed for 7-year-specific parameter

estimates of apparent survival for AHY adults of unknown age, 7-year-specific estimates for HY birds, and 6-year  age-specific parameter estimates for known age adults because they were marked as ducklings. Our predictions led us to structure our CJS analysis using design matrices that challenged the predictions emerging from our hypotheses. Age-related variation in apparent survival or resighting probability was evaluated using a design matrix structure that allowed apparent survival to differ in the years following the initial year of capture. Likewise, program MARK was structured to test if apparent survival differed among cohorts (i.e., birds fledged in the same year). For example, such an effect may reflect the quality of breeding conditions in a given year that may persist in future years through mechanisms such as density-dependence. We defined our most saturated model to be full timedependency in apparent survival and resighting rates for all 12 sex  site  age-class combinations, while additionally distinguishing HY birds from AHY birds (204 parameters). Published by NRC Research Press

340

Goodness of fit was evaluated with the parametric bootstrap procedure in program MARK, using 100 randomizations to estimate the variance inflation factor b c as the ratio of deviance from the observed data to the mean deviance from the bootstrap replicates. We used our estimate of b c to compute the quasi-Akaike’s information criterion adjusted for sample size (QAICc). We chose our best approximating model based on the lowest QAICc score, this model, in theory, being the one best describing the data when also considering model parsimony (number of free parameters). Akaike weights, wi, were calculated as indicators of relative support for each model, with models having DQAICc £ 2 being considered well supported by the data.

Can. J. Zool. Vol. 87, 2009 Fig. 1. Return rates of (a) female and (b) male Barrow’s goldeneyes (Bucephala islandica) banded as ducklings and resighted in subsequent years at Riske Creek, British Columbia, 1995–2002.

Results Encounter histories were available for 373 HY female and 438 HY male Barrow’s goldeneyes, and for 189 AHY females and 44 AHY males. Return rates for females banded as ducklings averaged 38% (range 18%–52%) in the 1st year after capture (i.e., the hatch year), and declined to 28% in the 2nd year after capture, 10% in the 3rd year, and £3% in the 4th year and thereafter. For males banded as ducklings, return rates were much lower, averaging 6% in the 1st year after capture, 4% in the 2nd year, and £2% thereafter (Fig. 1). Among the females banded as ducklings, 86 (23% of the total number banded) were observed paired at Riske Creek during subsequent years. The median age of first breeding for females was 2 years and the cumulative frequency distribution indicated that nearly all females (94%) were paired by their 3rd year (Fig. 2). Six females banded on the study site (2% of the total number banded) were eventually discovered in a nest box with clutches or on a pond with broods. The median age of first breeding was 3 years, with the earliest observed age of first breeding at 2 years and the latest at 5 years (Fig. 2). Based on model selection using QAICc, sex, year, age class at marking, and year hatched were all important predictors of apparent survival or resighting rate. The model best supported by the data (rank 1 — AHY: FGt HY: FGSty AHY: pt HY: pgt; for variable definitions see Table 2) indicated that apparent survival probabilities differed according to sex, year, and age class at marking. Sitespecific differences in apparent survival also were detected for HY birds during subsequent years, but they were not detected during the hatch year itself or for AHY birds marked as adults of unknown age (Table 2). There was significant annual variation in apparent survival evident for all 12 groupings, as there was for resighting probabilities, which were also group-dependent. Model fit was acceptable for the top-ranked model, with a model deviance (uncorrected for b c ) of 556.6, occurring at the 95th percentile among bootstrap replicates. Although marked birds generally appeared to have behaved independently, we neverthec = 1.29 obtained from less calculated QAICc values using b our global model (rank 5; Table 2). Our second best model (rank 2; Table 2) differed from the best model only by the addition of a parameter to test the hypothesis that apparent survival for AHY birds was a linear function of age. Support for this additional model complex-

ity was weak. Our third best model (rank 3; Table 2), which was identical to our best model except that we excluded cohort effects, was poorly supported by our data. No other structural variants relevant to our key hypotheses received meaningful support by our data (Table 2). Apparent survival estimates were similar for AHY females and AHY males and annual patterns of variation between the sexes were indistinguishable (Fig. 3). Annual estimates averaged 0.62 (95% CI: 0.47–0.76) for AHY females compared with 0.58 (95% CI: 0.39–0.76) for AHY males. Resighting rates were reasonably high for AHY females, but resighting probabilities for males were more uncertain. This is likely due to the smaller sample sizes for males and to their tendency to be present on breeding areas for a much shorter time period than most AHY females, which would reduce their resighting opportunities. In contrast to AHY birds, apparent survival rate among HY females averaged 0.68 (95% CI: 0.46–0.84) and only 0.35 (95% CI: 0.06–0.57) for HY males (Fig. 4). As with AHY males, less certain resighting probabilities among HY males led to lower confidence in estimates of apparent survival. Site-related differences appeared to play a role in the apparent survival rate of HY females, which was noticeably higher in SV (0.71; 95% CI: 0.48–0.86) and BM (0.75; 95% CI: 0.52–0.89) than in BP (0.59; 95% CI: 0.37–0.77). Sitespecific variation was not supported in the best fitting model for AHY females (Fig. 3). Annual variation in apparent survival rate was pronounced Published by NRC Research Press

Boyd et al.

341

Fig. 2. Age at which female Barrow’s goldeneyes (Bucephala islandica) were first observed (a) pairing and (b) breeding, as well as the cumulative proportion of females that were either (c) paired or (d) showed evidence of breeding according to age.

Table 2. Quasi-Akaike’s information criterion adjusted for sample size (QAICc) scores and Akaike weights (wi) of the models best explaining variation in apparent survival (F) and resighting probability (p) of Barrow’s goldeneyes (Bucephala islandica) at Riske Creek, British Columbia, 1996–2002. Model* AHY: FGt HY: FGSty AHY: pt HY: pgt AHY: FGt HY: FGStya AHY: pt HY: pgt AHY: FGt HY: FGSt AHY: pt HY: pgt AHY and HY: Fgt AHY and HY: pgt AHY: Fgt HY: Fgt AHY: pgt HY: pgt

Rank 1 2 3 4 5

K{ 43 44 38 168 204

k{ 42 43 37 108 128

cˆ 1.11 1.13 1.16 1.25 1.29

QDev§ 431.4 431.2 451.3 372.9 348.8

QAICc 1983.49 1985.36 1992.86 2070.32 2092.82

DQAICc 0.0 1.87 9.38 86.84 109.33

wi 0.71 0.28 0.01 0.00 0.00

Note: Models having DQAICc £ 2 are considered well supported by the data. *AHY is after-hatch-year birds of unknown age; HY is birds first marked during their hatch year; G is gender (male or female); S is site (Bald Mountain, Becher’s Prairie, or Stack Valley); t is year; a is age; y is HY cohort; g is 12 groups (S  G  C), where C is the age class at marking (AHY or HY). { K is the number of structural parameters. { k is the number of estimable parameters. § QDev is the cˆ adjusted model deviance.

and appeared not to be random. Rather, it followed a pattern of gradual change, implying that it is probably modulated either by gradual temporal changes in the abundance of Barrow’s goldeneyes, nest-cavity availability, environmental conditions, or a combination of these factors. Also evident was a declining trend in apparent survival rate within cohorts among birds banded as ducklings (Fig. 5). Figure 6 portrays the yearly relationship between the 1st year survival after hatch (from Fig. 4 but expressed in logit form) against the Beta estimate (logit) from program MARK for the contribution of the HY (y) effect to apparent survival rate. The significant relationship (p = 0.001) in Fig. 6 indicates that apparent survival during the 1st year after hatch is negatively correlated with future annual apparent survival rates for birds in that cohort.

Discussion The life history of Barrow’s goldeneyes is characterized

by delayed reproductive maturity, relatively high adult survival rates, and competition for nest cavities and pair–brood territories. Return rates were expected to be high for AHY Barrow’s goldeneyes of both sexes and our results corroborate this expectation. However, our analysis indicates additional dynamics at play with respect to survival, dispersal, and recruitment. The CJS modeling approach that we used to estimate apparent survival and resighting probability incorporates two important considerations: first, that mortality and permanent emigration rates are confounded in the estimates of apparent survival and, second, that resighting and temporary emigration rates are confounded in the estimates of resighting probability. Thus, results must be interpreted in light of the underlying components for each variable. Our first prediction, that return rates would be high for both AHY males and AHY females, was met. The apparent survival rate for AHY males averaged 0.58 compared with 0.62 for AHY females. True survival is probably higher for males than females as evidenced by male-biased adult sex Published by NRC Research Press

342

Can. J. Zool. Vol. 87, 2009

Fig. 3. Apparent survival (F) and resighting (p) rates for after-hatch-year (AHY) Barrow’s goldeneyes (Bucephala islandica). Symbols represent study sites at Riske Creek, British Columbia: &, all sites combined; *, Bald Mountain (BM); *, Becher’s Prairie (BP); !, Stack Valley (SV). There were insufficient data to estimate apparent survival for AHY males at Bald Mountain. Error bars are standard errors.

Fig. 4. Apparent survival (F) and resighting (p) rates for hatch-year (HY) Barrow’s goldeneyes (Bucephala islandica). Symbols represent study sites at Riske Creek, British Columbia: *, Bald Mountain; *, Becher’s Prairie; !, Stack Valley. Error bars are standard errors.

Published by NRC Research Press

Boyd et al. Fig. 5. Relationships within cohorts in apparent survival rate (F) for Barrow’s goldeneyes (Bucephala islandica) marked as ducklings and resighted during subsequent years at Riske Creek, British Columbia. Error bars are standard errors.

ratios both in coastal British Columbia (1.4 males per female; Iverson et al. 2006) and across the species range (Bellrose et al. 1961). We speculate that breeding site fidelity by AHY males is likely to be slightly lower than that of AHY females at Riske Creek, given the observed similarity in apparent survival coupled with the predicted differences in true survival. Savard and Eadie (1989) similarly estimated return rates (defined as the percentage marked that were observed in subsequent years) of 60% and 63% for male and female Barrow’s goldeneyes, respectively. These rates were similar to our own estimates and those for closely related common goldeneyes in other study areas (Nilsson 1971; Dow and Fredga 1984; Ludwichowski et al. 2002). Savard and Eadie (1989) argued that based on the extremely high resighting probabilities and limited movement within the study area these return rates approximated true survival. However, using joint live resighting and dead recovery data, Ludwichowski et al. (2002) were able to make independent estimates of survival and site fidelity of female common goldeneyes in northern Germany. They determined that true survival was markedly higher (83%) and that site fidelity of females of breeding age with previous nesting experience was lower (91%) than predicted by Savard and Eadie (1989) for Barrow’s goldeneyes at Riske Creek. Our second prediction, that apparent survival would be lower for HY than AHY Barrow’s goldeneyes, was strongly supported for males. However, among females, apparent survival rates were very similar for HY and AHY birds. We in-

343 Fig. 6. Negative correlation between the Beta estimate (logit) from program MARK for the contribution of the hatch-year effect on apparent survival rate (F) in relation to after-hatch apparent survival by site and sex. BM, Bald Mountain; BP, Becher’s Prairie; SV, Stack Valley.

terpret this to be related primarily to variations in natal philopatry between the sexes. HY Barrow’s goldeneyes occur in mixed flocks and the sexes experience similar selection pressure during the nonbreeding season, making it unlikely that sex-specific differences in annual survival are pronounced in prebreeding age classes. In terms of agerelated differences in site fidelity, it has been proposed that kin recognition can promote inclusive fitness gains through conspecific brood parasitism, cre`ching, and other behaviors that facilitate higher than expected breeding densities. However, theoretical and experimental investigations of Barrow’s and common goldeneyes have found little support for such mechanisms (Eadie and Lyon 1998; Po¨ysa¨ 2004). Instead, we believe that the high apparent survival rates that we observed for HY females relates primarily to their delayed sexual maturity and the fact that young females do not compete for nest sites until at least their 2nd year after hatch. We estimated the median age of first pairing for females at 2 years and the median age of first confirmed breeding at 3 years. Although yearling female goldeneyes are known to prospect for potential nest sites (Eadie and Gauthier 1985; Zicus and Hennes 1989), our data suggest that a relatively small proportion of these birds eventually recruits into the local population. Nest sites can be limiting to Barrow’s goldeneyes (Savard 1988b) and the best fitting CJS model among those that we investigated indicated that apparent survival in the 1st year after hatch is negatively correlated with future annual apparent survival rates within cohorts. Our interpretation is that a relatively high rate of apparent survival during the 1st year after hatch within a cohort leads to density-dependent effects on access to breeding territories, forcing a higher proportion of birds to be less likely to return to the same territories or geographic locations in subsequent years. Alternatively, a relatively low survival rate in the 1st year after hatch within a cohort would lead to less competition and a higher probability of these same birds recruiting into the local population. Published by NRC Research Press

344

Our model results also indicated that HY female apparent survival differed by site. For HY females, the lowest apparent survival rates were observed at the Becher’s Prairie study site, where the landscape is more naturally fragmented, nesting densities of goldeneyes are lower, and a higher proportion of nest sites of goldeneyes are natural cavities rather than nest boxes. Evans et al. (2002) reported larger clutch sizes but lower nesting success in nest boxes compared with natural cavities. He cautioned that although nest boxes are an important management tool for species experiencing population declines, there may be hidden costs to these programs. More specifically, Evans et al. (2002) speculated that nest boxes could create population sinks for Barrow’s goldeneyes, attracting large numbers of breeding birds that experience unnaturally high levels of predation and brood parasitism. Our findings support the mechanism underlying this supposition and suggest that younger females would have higher return rates to areas with high nest-box availability. However, higher return rates alone do not indicate that local populations have been enhanced by management actions. Pacific flyway populations of Barrow’s goldeneyes appear to be stable throughout most of their range; however, a relatively small global population (estimated <200 000 birds) and restricted distributions (60% of the world population breeds and winters in British Columbia) suggest that careful monitoring is necessary (Eadie et al. 2000). We found return and survival rates of Barrow’s goldeneyes to be relatively high for both AHY and HY birds of both sexes, but our analysis suggested additional dynamics at play, including potential density-dependent interactions. Understanding the roles that sex, age, and site-specific factors play in local population dynamics are important to the management of territorial birds such as Barrow’s goldeneyes.

Acknowledgements We thank the Canadian Department of National Defense (Chilcotin Military Training Area), Toosey First Nations, and Brian Durrell (Wine Glass Ranch) for allowing access to the study area. D. Ankney, F. Cooke, and D. Lank supervised parts of the fieldwork and the following people collected field data: B. Andres, B. Andres, B. Arquilla, T. Dahl, M. Einsworth, G. Ferguson, M. Grimm, G. Holte, J. Kenyon, S. Laurence, S. Ogle, T. Matthews, M. Mossip, N. Schmidt, P. Seccombe-Hett, D. Soulin, D. Thompson, S. Timmermans, C. Tucker, C. Williams, and P. Yen. This research was supported by scholarships and grants from the Natural Sciences and Engineering Research Council of Canada, Ducks Unlimited Canada, Science Council of British Columbia, Environment Canada, Simon Fraser University, and Forest Renewal British Columbia.

References Bellrose, F.C., Scott, T.G., Hawkins, A.S., and Low, J.B. 1961. Sex ratios and age ratios in North American ducks. Ill. Nat. Hist. Surv. Bull. 27: 391–474. Boyd, W.S., and Smith, D.W. 1989. Summary of aquatic invertebrate data collected from wetlands at Riske Creek, British Columbia, 1984 and 1985. Can. Wildl. Serv. Tech. Rep. Ser. 60. Burnham, K.P., and Anderson, D.R. 2002. Model selection and

Can. J. Zool. Vol. 87, 2009 multimodel inference: a practical information–theoretic approach. 2nd ed. Springer Publishing, New York. Dow, H., and Fredga, S. 1983. Breeding and natal dispersal of the goldeneye duck, Bucephala clangula. J. Anim. Ecol. 52: 681– 695. doi:10.2307/4447. Dow, H., and Fredga, S. 1984. Factors affecting reproductive output of the goldeneye duck Bucephala clangula. J. Anim. Ecol. 53: 679–692. doi:10.2307/4543. Eadie, J.M., and Gauthier, G. 1985. Prospecting for nest sites in cavity-nesting ducks of the genus Bucephala. Condor, 87: 528– 534. doi:10.2307/1367952. Eadie, J.M., and Lyon, B. 1998. Cooperation, conflict and cre`ching behavior in goldeneye ducks. Am. Nat. 151: 397–408. doi:10. 1086/286128. PMID:18811315. Eadie, J.M., Sherman, P.W., and Semel, B. 1998. Conspecific brood parasitism, population dynamics and conservation of cavity nesting birds. In Behavioural ecology and conservation biology. Edited by T.M. Caro. Oxford University Press, Oxford. pp. 306–340. Eadie, J.M., Savard, J.-P.L., and Mallory, M.L. 2000. Barrow’s Goldeneye (Bucephala islandica). In The birds of North America Online. No. 548. Edited by A. Poole. The Birds of North America Inc., Philadelphia, Pa., and Cornell Laboratory of Orthnithology, Ithaca, N.Y. Available from http://bna.birds.cornell. edu/bna/species/548. Evans, M.R., Lank, D.B., Boyd, W.S., and Cooke, F. 2002. A comparison of the characteristics and fate of Barrow’s goldeneye and bufflehead nests in nest boxes and natural cavities. Condor, 104: 610–619. doi:10.1650/0010-5422(2002)104[0610:ACOTCA]2.0. CO;2. Iverson, S.A., Boyd, W.S., Regehr, H.M., and Rodway, M.S. 2006. Sex and age-specific distributions of seaducks wintering in the Strait of Georgia, British Columbia: implications for the use of age ratios as an index of recruitment. Can. Wildl. Serv. Tech. Rep. Ser. No. 459. Krajina, V.J. 1969. Ecology of the forest trees in British Columbia. Ecology of Western North America, 2: 1–146. Lebreton, J.-D., Burnham, K.P., Clobert, J., and Anderson, D.R. 1992. Modeling survival and testing biological hypotheses using marked animals: a unified approach with case studies. Ecol. Monogr. 62: 67–118. doi:10.2307/2937171. Lokemoen, J.T., and Sharp, D.E. 1979. Assessment of nasal marker materials and designs used on dabbling ducks. Wildl. Soc. Bull. 13: 53–56. Ludwichowski, I., Barker, R., and Bra¨ger, S. 2002. Nesting area fidelity and survival of female common goldeneyes Bucephala clangula: are they density-dependent? Ibis, 144: 452–460. doi:10.1046/j.1474-919X.2002.00075.x. Newton, I. 1994. The role of nest sites in limiting the numbers of hole-nesting birds: a review. Biol. Conserv. 70: 265–276. doi:10. 1016/0006-3207(94)90172-4. Nilsson, L. 1971. Migration, site tenacity and longevity of Swedish goldeneyes (Bucephala clangula). Var Fagelvarld, 30: 180–194. Pollock, K.H., Nichols, J.D., Brownie, C., and Hines, J.E. 1990. Statistical inference for capture–recapture experiments. Wildl. Monogr. 107: 1–97. Po¨ysa¨, H. 2004. Relatedness and the evolution of conspecific brood parasitism: parameterizing a model with data for a precocial species. Anim. Behav. 67: 673–679. doi:10.1016/j.anbehav. 2003.08.009. Po¨ysa¨, H., and Po¨ysa¨, S. 2002. Nest-site limitation and density dependence of reproductive output in the common goldeneye Bucephala clangula: implications for the management of cavityPublished by NRC Research Press

Boyd et al. nesting birds. J. Appl. Ecol. 39: 502–510. doi:10.1046/j.13652664.2002.00726.x. Savard, J.-P.L. 1984. Territorial behavior of common goldeneye, Barrow’s goldeneye and bufflehead in areas of sympatry. Ornis Scand. 15: 211–216. doi:10.2307/3675929. Savard, J.-P.L. 1986. Territorial behaviour, nesting success and brood survival in Barrow’s Goldeneyes and its congeners. Ph.D. thesis, University of British Columbia, Vancouver. Savard, J.-P.L. 1988a. Winter, spring and summer territoriality in Barrow’s Goldeneye: characteristics and benefits. Ornis Scand. 19: 119–128. doi:10.2307/3676460. Savard, J.-P.L. 1988b. Use of nest boxes by Barrow’s Goldeneyes: nesting success and affect on the breeding population. Wildl. Soc. Bull. 16: 125–132.

345 Savard, J.-P.L., and Eadie, J.M. 1989. Survival and breeding philopatry in Barrow’s and common goldeneyes. Condor, 91: 198– 203. doi:10.2307/1368163. Thompson, J.E. 1996. Comparative reproductive ecology of female Buffleheads (Bucephala albeola) and Barrow’s Goldeneyes (Bucephala islandica) in central British Columbia. Ph.D. dissertation, University of Western Ontario, London. White, G.C., and Burnham, K.P. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study, 46(Suppl.): 120–138. Zicus, M.C., and Hennes, S.K. 1989. Nest prospecting by common goldeneyes. Condor, 91: 807–812. doi:10.2307/1368064.

Published by NRC Research Press