SOURCES OF VARIATION IN SURVIVAL OF BREEDING FEMALE WOOD DUCKS

Transcription

1 The Condor 108: # The Cooper Ornithological Society 2006 SOURCES OF VARIATION IN SURVIVAL OF BREEDING FEMALE WOOD DUCKS KEVIN M. HARTKE 1,3,JAMES B. GRAND 2,GARY R. HEPP 1, AND TRAVIS H. FOLK 1 1 School of Forestry and Wildlife Sciences, 3301 Forestry and Wildlife Building, Auburn University, Auburn, AL U.S. Geological Survey, Alabama Cooperative Fisheries and Wildlife Research Unit, 3301 Forestry and Wildlife Building, Auburn University, Auburn, AL Abstract. In waterfowl, reproduction is physiologically demanding and females are exposed to varying risks of mortality at different periods of the breeding cycle. Moreover, differences among females may influence survival within breeding periods. We captured and fitted female Wood Ducks (Aix sponsa) with radio-transmitters before nest initiation during two breeding seasons to estimate survival and investigate sources of variation in survival. We partitioned the breeding season into three periods (preincubation, incubation, postnesting) according to breeding status of individual females, and used informationtheoretic methods to compare models in which daily survival varied among periods, between successful and failed nesting females, and with parameters describing individual heterogeneity. Our analysis suggested that daily survival was best modeled as a function of breeding period, differences between successful and failed nesting females during postnesting, and early incubation body condition of successful females during postnesting. Model-averaged daily survival was (95% CL: ) during preincubation and 1.0 during incubation. Postnesting daily survival was 1.0 for failed nesting females and ( ) for successful females, suggesting a trade-off between current reproduction and survival. Female age, body condition at capture, nest initiation date, and brood size generally were not useful for explaining variation in survival. Only early incubation body condition was important for modeling survival of successful females during postnesting; however, weight of evidence was limited and the effect on survival was weak. Mortality was greatest for females during preincubation and for females that nested successfully. Results support the hypothesis that brood care is costly for females. Key words: Aix sponsa, breeding status, female survival, heterogeneity, mortality risk, reproductive effort, Wood Duck. Fuentes de Variación en la Supervivencia de Hembras Reproductivas de Aix sponsa Resumen. En los anseriformes, la reproducción es fisiológicamente demandante y las hembras están expuestas a diversos riesgos de mortalidad durante diferentes periodos del ciclo reproductivo. Además, diferencias entre las hembras pueden influenciar la supervivencia dentro de los periodos reproductivos. Capturamos hembras de Aix sponsa y les colocamos radio transmisores antes de que comenzaran a nidificar. Esto se realizó durante dos temporadas reproductivas para estimar la supervivencia e investigar fuentes de variación en la supervivencia. Dividimos la temporada reproductiva en tres periodos (preincubación, incubación y post-nidificación) de acuerdo al estado reproductivo de las hembras y usamos métodos de información teóricos para comparar modelos en los cuales la supervivencia diaria varió entre los periodos, entre las hembras que nidificaron de modo exitoso y no exitoso, y con parámetros que describen la heterogeneidad individual. Nuestro análisis sugirió que la supervivencia diaria se modelo mejor en función al periodo reproductivo, a las diferencias entre hembras reproductivas exitosas y no exitosas durante el periodo post-nidificación y a la condición física de las hembras en la parte inicial de la incubación en la etapa post-nidificación. El promedio de supervivencia diaria en el modelo fue de (95% LC: ) para el periodo de preincubación y de 1.0 durante la incubación. La supervivencia diaria para el periodo post-nidificación fue de 1.0 para las hembras no exitosas y de ( ) para hembras exitosas, lo cual sugiere una compensación entre la reproducción actual y la supervivencia. En general, la edad de la hembra, la condición física al tiempo de la captura, el tiempo en que se inicio el nido y el tamaño de la nidada no fueron útiles para explicar la variación en la supervivencia. Únicamente la condición física en el periodo temprano de incubación fue importante para Manuscript received 24 May 2005; accepted 28 October Present address: Texas Parks and Wildlife Department, 1620 FM 2218, Richmond, TX kevin.hartke@tpwd.state.tx.us [201]

2 202 KEVIN M. HARTKE ET AL. modelar la supervivencia de hembras exitosas durante la etapa post-nidificación. Sin embargo, el valor de la evidencia fue limitado y el efecto sobre la supervivencia fue poco convincente. La mortalidad más alta fue para las hembras en el periodo de preincubación y para aquellas que anidaron exitosamente. Los resultados apoyan la hipótesis de que el cuidado de la nidada es costoso para las hembras. INTRODUCTION Changes in behavioral and metabolic requirements of breeding waterfowl put them at greater risk of mortality than nonbreeding individuals. Mortality of ducks (Anatinae) during the breeding season, for example, accounts for a large portion of annual mortality (Blohm et al. 1987, LeMaster and Trost 1994) and can adversely affect population growth rates (Flint et al. 1998, Hoekman et al. 2002). Survival probabilities of female ducks often vary among periods of the breeding season (i.e., prenesting, incubation, and brood rearing) that are associated with changes in body condition, behavior, and habitat use (Kirby and Cowardin 1986, Robb and Bookhout 1990, Devries et al. 2003). Prior to incubation, female ducks engage in a variety of activities including courtship, feeding, searching for nest sites, and producing eggs. These activities are energetically demanding and can increase mortality risks of females. Nutrient reserves and developing eggs increase body mass of females before incubation, which may make them more vulnerable to predation (Lee et al. 1996). Incubating females must balance the competing demands of maintaining good physical condition while keeping an optimum nest environment for developing eggs (Deeming 2002). Females use both endogenous and exogenous nutrients to meet energy demands of incubation. They spend most of the day on the nest but take short recesses to feed (Afton and Paulus 1992). In some species, nutrient reserves of incubating females are important for helping them meet costs of incubation (Gloutney and Clark 1991, Blums et al. 1997). Predation risk for incubating females also can be high, but birds incubating eggs in open nests are more vulnerable than cavity nesting birds (Sargeant et al. 1984, Martin 1995). After incubation, successful females are faced with the challenge of caring for young. Parental care of precocial ducklings generally is considered less demanding than raising altricial young; however, female ducks are solely responsible for care and will put themselves at risk to defend young from predators (Afton and Paulus 1992). In contrast, postnesting survival of females without broods is expected to be high because of less risky behavior. Individual variation among females associated with age (Dufour and Clark 2002) and body condition (Hepp et al. 1990) also may influence survival. We used radio-telemetry to study survival of female Wood Ducks (Aix sponsa) captured before nest initiation. We used a novel modeling approach to estimate daily survival during the breeding season, whereby female survival was partitioned into three periods (preincubation, incubation, and postnesting) of the breeding season according to an individual s breeding status. A benefit of this method was that it facilitated the use of individual covariates in models. We used an information-theoretic approach to compare models that estimated daily survival during each period of the breeding season and examined the influence of intrinsic differences among individuals on survival at specific periods of the breeding season. Prior work has estimated survival rates of female Wood Ducks during incubation and brood rearing (Robb and Bookhout 1990, Davis et al. 2001), but survival of breeding females has not been estimated before incubation (prebreeding to egg-laying). Influence of individual heterogeneity on breeding season survival of Wood Ducks also has not been examined. Because reproductive costs are important to the evolution of life-history traits, we explored the relationships between breeding season survival and individual covariates, such as nest success, body condition, brood size, laying date, and female age. Spring return rates of female Wood Ducks have been used previously to examine influences of body mass and brood size on annual survival (Hepp et al. 1990, Rohwer and Heusmann 1991). In contrast, we used radiotelemetry and known fate models to investigate the effect of current reproductive effort on survival of female Wood Ducks during the breeding season.

3 SURVIVAL OF BREEDING FEMALE WOOD DUCKS 203 METHODS STUDY AREA This study was conducted at Eufaula National Wildlife Refuge and the surrounding area (32uN, 85uW). The refuge (4520 ha) is located in southeast Alabama and southwest Georgia and is centered on a portion of Lake Eufaula, an impoundment of the Chattahoochee River. Dominant aquatic habitats included open water on the lake, farm ponds, beaver ponds, managed impoundments, and littoral areas of the lake. Wood Ducks primarily nested in nest boxes on Eufaula National Wildlife Refuge because forested habitats on the study area were dominated by young trees and few natural cavities were available. CAPTURE AND RADIO- MARKING PROCEDURE Early in the prebreeding seasons of 1999 and 2000 (January February), we installed automatic nest-box traps modified from Zicus (1989) in nest boxes (1999: n 5 25; 2000: n 5 28) to capture female Wood Ducks. During this period, resident females were visiting nest boxes to select a suitable nest site (Bellrose and Holm 1994). Migrant, nonresident Wood Ducks also were present during this time, so the use of nestbox traps allowed us to target and capture only resident females. All females caught in next-box traps during January-February each year were selected for the study. We weighed captured females to the nearest 5 g and measured tarsus, wing chord, and culmen length (mm). Females were aged as yearling (hatch-year females entering their first breeding season) or adult (second year of breeding or older; Harvey et al. 1989) and, if needed, banded with a U.S. Fish and Wildlife Service leg band. We used radio-telemetry to monitor mortality of females during the breeding season. Transmitters (model RI-2BM, Holohil Systems, Ltd., Carp, Ontario, Canada) weighing 6.6 g were implanted subcutaneously posterior to the nape. Each transmitter had a life expectancy of 180 days and was equipped with a motion sensitive (4-hr delay) mortality sensor. Administration of anesthesia and surgical procedures are described in detail by Hepp et al. (2002). Females were held 1 2 hr in a dark, ventilated holding box to allow full recovery from the anesthesia before being released near the nest box in which they were captured. Radiomarked females returned to Eufaula National Wildlife Refuge the year following implant surgery at a similar rate as females without radio implants, indicating that the method of marking did not negatively influence survival (Hepp et al. 2002). DATA COLLECTION We monitored radio-marked females from January to June in 1999 and We attempted to monitor females once daily from 1 hr before sunrise to 1 hr after sunset; however, observation intervals were unequal. We continued monitoring females until either the transmitter failed (signal loss), females emigrated from the study area, females died, or the study ended (30 May 1999, 2 June 2000). We used hand-held Yagi antennas on foot for daily monitoring and fixed-wing aircraft fitted with antennas as described by Gilmer et al. (1981) to locate missing radio-marked females. We made a visual observation to check female status when transmitters emitted a mortality signal. Mortality was assumed when we found female remains (e.g., carcass, feathers, bones) near the transmitter, bite marks on the transmitter, or a damaged antenna. Survival may be underestimated if radio-marked individuals lose their transmitter, and it is not uncommon for individuals to lose implanted transmitters (Mulcahy et al. 1999, Brook and Clark 2002). We documented evidence of females losing their implanted transmitters between monitoring periods at Eufaula National Wildlife Refuge. Thirteen females that were implanted with transmitters in 1999 were recaptured in Four (31%) of these females had lost their transmitters, and one female had a transmitter partially extruding through the skin. While monitoring females during the 1999 and 2000 breeding seasons, we recovered five undamaged transmitters in mortality mode. On each occasion, we found no female remains in the immediate area; three of these females were recaptured during incubation later that year or the following year. We assumed the remaining two females also lost their transmitters and survived. We treated these cases similar to transmitter failure and right-censored the data starting the day after the last observation. Nest boxes (n 5 71) at Eufaula National Wildlife Refuge were checked approximately

4 204 KEVIN M. HARTKE ET AL. weekly from January to June in 1999 and 2000 to monitor nesting activity of radio-marked females. We counted and marked new eggs with a unique number during each visit. Wood Ducks lay eggs at a rate of one per day (Drobney 1980); therefore, date of nest initiation was estimated by subtracting the number of eggs in the nest from the ordinal date when the active nest was first found (Hepp et al. 1990). If the number of eggs exceeded the number of days between nest box checks (i.e., brood parasitism), we assumed that egg-laying began on the day after the nest box was last checked. Nesting radio-marked females were recaptured in nest boxes during early incubation (#9 days) and weighed. Next, we candled eggs (Weller 1957) to determine stage of incubation and installed a temperature data logger in the nest box (Manlove and Hepp 2000). Before returning females to the nest box, we administered propofol to reduce the chance of abandonment (Hepp and Manlove 2001). The first day of incubation was estimated by subtracting the number of days of egg development (Hanson 1954) from the ordinal date when the incubating female was captured on the nest. We used 24-hr records of nest temperature (i.e., temperature recorded every 6.4 min for 35 days) from data loggers to determine the day that successful females left the nest box with ducklings. We assumed females left the nest box when nest temperature declined during a 24-hr period. Within five days after hatching, we returned to the nest box and estimated the number of ducklings that left the nest by examining the remains (e.g., whole eggs, membranes, dead ducklings) in the nest box (Davis et al. 1998). We used daily records of data loggers and radio-telemetry to determine the day when unsuccessful females abandoned nests. STATISTICAL ANALYSES We used the nest survival model (Dinsmore et al. 2002) in program MARK (White and Burnham 1999) with the logit link function to estimate daily survival probability of females. The nest survival model was appropriate for this application because it allows the estimation of daily survival rates when the fate of nests or individuals can be determined similar to known fate models, and the length of the observation interval varies. It also allows for staggered entry of individuals into the at-risk group, rightcensoring when the probability of survival is unrelated to detection (Pollock et al. 1989), and the use of individual covariates (Dinsmore et al. 2002). If individual heterogeneity is not incorporated into the analysis, estimates of survival (regardless of estimator type) from field studies may be overestimated (Zens and Peart 2003). Therefore, we included several covariates that have been shown to describe variation in survival among female waterfowl. Females that died within five days after implant surgery or emigrated from the study area were excluded from all analyses. We assumed that daily survival rate was constant while females were of a given breeding status. We created an a priori set of candidate models (Table 1) restricting the number of models relative to the sample size of females, and only considering models with an additive relationship between selected parameters that were biologically meaningful and interpretable (Burnham and Anderson 2002). We used information-theoretic methods to direct model selection and parameter estimation (Burnham and Anderson 2002). In addition to a constant survival model {S(.)}, we considered models including effects of breeding status, nest success, capture body condition (cbc), early incubation body condition (eibc), nest initiation date (init; ordinal date), female age, and brood size (bs). We modeled the influence of nest success on female survival by estimating postnesting survival rates separately for successful (spn) and failed (fpn) females. We performed principal component analysis on the measurements of tarsus, wing chord, and culmen length taken from each female (PROC PRINCOMP; SAS Institute 1999). The first principal component for the three variables accounted for 48% of the total variance, and each of the component loadings (tarsus: 0.70; wing chord: 0.58; culmen: 0.42) were positive. We interpreted this positive covariation as variation in overall body size, and used the first principal component scores for each female as a measure of body size. Body condition indices (cbc and eibc) were residuals from the regression of body mass on body size. Breeding status represented three segments (preincubation, incubation, and postnesting) of the breeding season. Rather than classifying breeding status according to the general chronology (i.e., mean or median dates of nest

5 SURVIVAL OF BREEDING FEMALE WOOD DUCKS 205 TABLE 1. Candidate set of a priori models used to examine the effects of breeding status, nest success, body condition (at capture and early incubation), age, nest initiation date, and brood size on daily survival of female Wood Ducks breeding at Eufaula National Wildlife Refuge, Alabama and Georgia in 1999 and Models were ranked using Akaike s information criterion corrected for small sample size (AIC c ); D i is the increase in AIC c over the lowest observed value of AIC c. Akaike weights (w i ) indicate the likelihood of a given model; log(l) is the maximized log-likelihood value, and K is the number of parameters. Model a log(l) K D i b w i S(pi in spn fpn eibc spn ) S(pi in pn eibc pn ) S(pi in spn fpn) S(pi in spn fpn cbc pi eibc spn ) S(pi in pn cbc pi eibc pn ) S(pi in spn fpn bs spn ) S(pi in pn) S(pi in spn fpn init spn ) S(pi in spn fpn cbc pi ) S(pi in spn fpn age pi ) S(pi in spn fpn age spn ) S(pi in pn cbc pi ) S(pi in pn age pi ) S(.) S(pi in pn age pn ) S(pi in pn init pn ) S(pi in spn fpn age pi age spn ) S(pi in pn age pi age pn ) a Parameter notation: S 5 daily survival; pi 5 preincubation; in 5 incubation; spn 5 successful postnesting; fpn 5 failed postnesting; eibc 5 early incubation body condition; pn 5 postnesting; cbc 5 capture body condition; init 5 nest initiation date; bs 5 brood size; age 5 female age (yearling or adult). Parameters with italicized subscript indicate the variable s effect on daily survival during a specific breeding period. b The lowest AIC c value was initiation, incubation, and brood rearing) of the population at Eufaula National Wildlife Refuge, we coded daily observations according to an individual s breeding status for that day. Preincubation (pi) started 24 hr after implant surgery and ended on the day prior to the onset of incubation. Survival data for females (n 5 6) that did not incubate a clutch and remained on the study area were included in preincubation. Because brood parasitism is common in Wood Ducks (Bellrose and Holm 1994), it is likely these females were brood parasites and were exposed to the same mortality risks as females in the preincubation stage that subsequently incubated their own clutch. Incubation (in) started on the first full day of incubation and ended the day before ducklings departed the nest. The start of postnesting status (pn) differed between successful and failed nesting females. For successful nesting females, postnesting started on the day when the female and ducklings left the nest. Among unsuccessful incubating females, postnesting started on the day after the nest was abandoned. For the purpose of this study, postnesting ended on the last day of the study (30 May 1999, 2 June 2000). Some postnesting females (n 5 5) attempted a second nest; we excluded data from these females starting on the estimated first day of rapid follicular growth (Alisauskas and Ankney 1992) of the second nest attempt. We estimated the beginning of rapid follicular growth by subtracting seven days from the initiation date of the second nest (Drobney 1980). We estimated the influence of female age and continuous covariates on survival within breeding status groups (pi, pn, spn). For example, cbc pi represented the influence of capture body condition on survival of females during preincubation. We did not estimate the influence of covariates on survival of females in incubation or unsuccessful females in postnesting status because no deaths occurred in these groups. Because nest success can be related to age, body condition, and nest initiation date of female waterfowl (Rohwer 1992), we conducted univariate tests of multicollinearity (Neter et al.

6 206 KEVIN M. HARTKE ET AL. 1985) prior to analysis, but found no relationship between nest success and any of the previously mentioned covariates (all P. 0.05). All continuous covariates were standardized ([x 2 x ]/SD) to improve optimization of the maximum-likelihood function in program MARK. We examined two global models (candidate models with the most parameters; Table 1) for goodness-of-fit by calculating a Pearson chi-square statistic following Hosmer and Lemeshow (2000:148). Lack of fit (P # 0.05) indicates a structurally inadequate model (e.g., omission of an important explanatory variable) or lack of independence among observations. Each global model {S(pi in spn fpn cbc pi eibc spn ), S(pi in spn fpn age pi age spn )} fit the data well (x 2 7 ~ 4:4, P ; x 2 2 ~ 1:1, P , respectively). We calculated Akaike s information criterion corrected for bias due to small sample size (AIC c ) for each of the a priori candidate models (Burnham and Anderson 2002). The model with the lowest AIC c is considered the best approximating model given the data, and differences between models are represented by D i (increase in AIC c over the lowest observed value of AIC c ). Models with D i # 2 have greater support compared to other models in the candidate set (Burnham and Anderson 2002). We used the total number of females monitored for survival as the sample size (n) when computing AIC c. We assessed model likelihood using Akaike weights (w i ; Burnham and Anderson 2002). If the best approximating model had a low w i (less than approximately 0.9) relative to other models, we used multimodel inference, and calculated model-averaged estimates of survival and unconditional SE based on w i (Burnham and Anderson 2002). RESULTS In total, 62 females were radio-marked, 29 in 1999 and 33 in Two females (one each year) apparently died of surgical trauma within five days of being released, and four females (two each year) emigrated from the study area. Fifty-six females (yearling:adult ratio 5 19:37) in preincubation status were monitored for survival. Predators killed four of these females (7%; two each year) before they initiated a nest. Forty-eight (15:33) of 52 (92%) radio-marked females initiated nests (1999: n 5 21; 2000: n 5 27), and nests of 16 females (33%) were unsuccessful (i.e., failed to hatch at least one egg). All females survived during incubation; however, three transmitters failed. Two females lost their transmitter during incubation, and four females completed incubation after the study ended. Therefore, we monitored survival of 14 (5:9) failed and 25 (6:19) successful postnesting females. All females of failed nests survived to the end of the study, whereas predators killed four females (16%; two each year) of successful nests. There was little support for the model with constant daily survival, and the best approximating model of daily survival included variation associated with breeding periods, nest success, and early incubation body condition (Table 1). However, model likelihood (w i ) was low for the best model relative to other models in the candidate set; therefore, we computed model-averaged estimates of daily survival for each breeding period. Because modelaveraged estimates and confidence limits for survival were nearly identical between yearlings and adults, we present model-averaged estimates without the effect of female age. Daily survival was (95% CL derived from unconditional SE: ) during preincubation and 1.0 during incubation. Three of the top four models (D i # 2; Table 1) included variation associated with nest success (spn and fpn), indicating greater support compared to models that estimated a single postnesting survival probability. Postnesting daily survival was (95% CL: ) for successful and 1.0 for unsuccessful nesting females, suggesting an increased mortality risk for female Wood Ducks attempting to rear a brood. Early incubation body condition of successful females during postnesting (eibc spn ) also was included in the best approximating model (Table 1). Successful females that survived the postnesting period started incubation in better condition (x SE, n 5 21) than successful females that died (x , n 5 4); however, effect of body condition on postnesting survival was weak and 95% confidence limits for the estimated slope included zero (b5 1.3, 20.2 to 2.9 logit scale). In general, the influence of early incubation body condition on postnesting survival (eibc pn and eibc spn ) had strong support, as it was included in three of the top four models (Table 1).

7 SURVIVAL OF BREEDING FEMALE WOOD DUCKS 207 DISCUSSION Population dynamics and life-history tradeoffs are influenced by patterns of variation in survival. In ducks and geese, population growth rate is directly related to female survival (Schmutz et al. 1997, Flint et al. 1998, Hoekman et al. 2002). The breeding season is especially hazardous for female ducks, and it is likely that exposure to mortality varies temporally with respect to breeding status of females. A recent study in the Prairie Pothole region of Canada found that mortality of female Mallards (Anas platyrhynchos) was greatest during the period of maximum nesting effort (i.e., egg-laying and incubation; Devries et al. 2003), which supports other studies of Mallards and American Black Ducks (A. rubripes, Kirby and Cowardin 1986, Dwyer and Baldassarre 1993). Mortality of female Wood Ducks using natural cavities in southern Indiana also was greater during the nesting period (preincubation and incubation activities combined) than in the brood-rearing or postbreeding periods (Robb and Bookhout 1990). We partitioned females according to their daily breeding status and found that survival of females was higher during incubation than in either the preincubation or brood-rearing periods. Most (92%) females at Eufaula National Wildlife Refuge used nest boxes equipped with predator guards, so predator access to eggs and incubating females was limited. Nest box programs have the potential to increase productivity of local Wood Duck populations, because nest success tends to be greater in nest boxes than in natural cavities (Bellrose and Holm 1994, Ryan et al. 1998). Besides protection provided by nest boxes, changes in activity of incubating females also may result in reduced mortality risk. For example, cavity-nesting female Pied Flycatchers (Ficedula hypoleuca) died more often during nest building and egg-laying than during incubation (Slagsvold and Dale 1996). Slagsvold and Dale (1996) suggested that female Pied Flycatchers were more susceptible to predation prior to incubation while collecting nest materials and acquiring food for egg production; risk declined during incubation because females spent less time outside the nest cavity. Female Wood Ducks before incubation also are engaged in activities such as nest searching and foraging; however, incubating females spend.80% of the day on the nest (Manlove and Hepp 2000, Folk and Hepp 2003), which may limit their exposure to predators. In west-central Alabama and east-central Mississippi, daily survival (converted from product-limit survival) of brood-rearing Wood Ducks was for two populations that used nest boxes (Davis et al. 2001). In comparison, in southern Indiana, daily survival of brood-rearing Wood Ducks that nested in natural cavities was (Robb and Bookhout 1990). At Eufaula National Wildlife Refuge, we monitored postnesting females of successful and failed nests. Females of successful nests had lower daily survival than females of failed nests, providing evidence that females attempting to rear broods increase their risk of mortality. We found no evidence supporting a relationship between brood size and female postnesting survival {D i. 2 for S(pi in spn fpn bs spn )}. Manipulations of offspring number have been used in other studies to assess effects of variation in reproductive effort on parental survival. Among experiments done with precocial species, over-winter survival was not influenced by clutch or brood size (Lessells 1986, Milonoff and Paananen 1993, Wiebe and Martin 2000). In a study of unmanipulated brood size, Rohwer and Heusmann (1991) also found no evidence that brood size influenced survival of female Wood Ducks. However, female Canada Geese (Branta canadensis) caring for large broods molted later at a lower body mass and nested later the following year than females with small broods (Lessells 1986). Regardless of brood size, caring for young has been found to be an important reproductive cost. Seabirds with reduced reproductive effort in year t survive better to year t + 1 and have greater reproductive success in year t + 1 than parents with high reproductive effort in year t (Pugesek and Diem 1990, Wernham and Bryant 1997, Golet et al. 1998). Among ducks, females that cared for offspring experienced higher within-season mortality (Bergmann et al. 1994, this study) and lower reproductive effort the following year (Milonoff et al. 2004) than females of failed nests. Activity budgets of geese and ducks indicate that parents caring for young have less time for self-maintenance activities, such as feeding, compared to parents without young (Seddon and Nudds 1992,

8 208 KEVIN M. HARTKE ET AL. Ruusila and Pöysä 1998). Caring for broods may influence female survival by inhibiting their ability to quickly restore body mass. In South Carolina, body mass at the end of incubation was positively related to survival of female Wood Ducks to the next breeding season in one of three years (Hepp et al. 1990). We found a similar relationship between body condition and postnesting survival of females at Eufaula National Wildlife Refuge, but evidence was weak. Females caring for broods also may engage in risky behavior. Wood Ducks, for example, often feign injury to divert attention of predators from offspring (Bellrose and Holm 1994). Parents may risk more if the value of current offspring (i.e., chance of recruitment) is greater than the expected value of offspring from future breeding attempts (Winkler 1987, Montgomerie and Weatherhead 1988). Tests of nest defense theory in ducks suggest that risk-taking is related to clutch size, day of incubation, and habitat quality (Forbes et al. 1994, Mallory et al. 1998, Gunness et al. 2001). During brood rearing, offspring value is expected to be high for early-hatched and large broods (Winkler 1987). Therefore, females caring for earlyhatched and large broods should adopt more risky behaviors that might reduce their probability of survival. In our study, postnesting survival of Wood Ducks was independent of brood size and nesting date. These results might be expected if costs of parental care are not related to brood size (Lazarus and Inglis 1986, Ruusila and Pöysä 1998), and if recruitment probability is independent of hatch date, as it was for Wood Ducks in South Carolina (Hepp et al. 1989). In conclusion, we partitioned survival according to breeding status of individual females. Using this approach, we estimated survival of female Wood Ducks during preincubation, incubation, and postnesting, and assessed the influence of individual covariates on survival during specific periods of the breeding season. At Eufaula National Wildlife Refuge, where most Wood Ducks used nest boxes equipped with predator guards, mortality was greatest for females prior to incubation and for females raising broods. Our results support the idea that brood care can be costly for females. In addition, brood-rearing females in good condition survived better than those in poor condition, but the relationship was weak. Because females nested primarily in nest boxes at Eufaula National Wildlife Refuge, our results may not be generally applicable to local populations of Wood Ducks that use natural cavities. Management actions such as providing nest boxes with predator exclusion devices may reduce mortality risk during incubation, thereby potentially changing selection pressures that act on local populations. ACKNOWLEDGMENTS We thank R. Bubb and L. de Souza for assistance with fieldwork, and the staff at Eufaula National Wildlife Refuge for logistical support. Funding was provided by Alabama Department of Conservation and Natural Resources. This study was conducted under Project No. W for Alabama Division of Wildlife and Freshwater Fisheries. Wood Ducks were captured and radio-marked under Auburn University Institutional Animal Care and Use Permit No R J. B. Davis, P. L. Flint, D. N. Koons, K. J. Reinecke, and S. G. Somershoe provided helpful reviews on various drafts of the manuscript. The U.S. Geological Survey, Patuxent Wildlife Research Center supported KMH during manuscript preparation. LITERATURE CITED AFTON, A. D., AND S. L. PAULUS Incubation and brood care, p In. B. D. J. Batt, A. D. Afton, M. G. Anderson, C. D. Ankney, D. H. Johnson, J. A. Kadlec, and G. L. Krapu [EDS.], Ecology and management of breeding waterfowl. University of Minnesota Press, Minneapolis, MN. ALISAUSKAS, R. T., AND C. D. ANKNEY The cost of egg laying and its relationship to nutrient reserves in waterfowl, p In. B. D. J. Batt, A. D. Afton, M. G. Anderson, C. D. Ankney, D. H. Johnson, J. A. Kadlec, and G. L. Krapu [EDS.], Ecology and management of breeding waterfowl. University of Minnesota Press, Minneapolis, MN. BELLROSE, F. C., AND D. J. HOLM Ecology and management of the Wood Duck. Stackpole Books, Mechanicsburg, PA. BERGMANN, P. J., L. D. FLAKE, AND W. L. TUCKER Influence of brood rearing on female Mallard survival and effects of harness-type transmitters. Journal of Field Ornithology 65: BLOHM, R. J., R. E. REYNOLDS, J. P. BLADEN, J. D. NICHOLS, J. E. HINES, K. H. POLLOCK, AND R. T. EBERHARDT Mallard mortality rates on key breeding and wintering areas. Transactions of the North American Wildlife and Natural Resources Conference 52: BLUMS, P., A. MEDINS, AND R. G. CLARK Effect of incubation body mass on reproductive success and survival of two European diving

9 SURVIVAL OF BREEDING FEMALE WOOD DUCKS 209 ducks: a test of the nutrient limitation hypothesis. Condor 99: BROOK, R. W., AND R. G. CLARK Retention and effects of nasal markers and subcutaneously implanted radio transmitters on breeding female Lesser Scaup. Journal of Field Ornithology 73: BURNHAM, K. P., AND D. R. ANDERSON Model selection and multimodel inference: a practical information-theoretic approach. 2nd ed. Springer-Verlag, New York. DAVIS, J. B., R. M. KAMINSKI, B. D. LEOPOLD, AND R. R. COX JR Survival of female Wood Ducks during brood rearing in Alabama and Mississippi. Journal of Wildlife Management 65: DAVIS, J. B., R. M. KAMINSKI, AND S. E. STEPHENS Wood Duck eggshell membranes predict duckling numbers. Wildlife Society Bulletin 26: DEEMING, D. C. [ED.] Avian incubation: behaviour, environment, and evolution. Oxford University Press, New York. DEVRIES, J. H., J. J. CITTA, M. S. LINDBERG, D. W. HOWERTER, AND M. G. ANDERSON Breeding-season survival of Mallard females in the Prairie Pothole region of Canada. Journal of Wildlife Management 67: DINSMORE, S. J., G. C. WHITE, AND F. L. KNOPF Advanced techniques for modeling avian nest survival. Ecology 83: DROBNEY, R. D Reproductive bioenergetics of Wood Ducks. Auk 97: DUFOUR, K. W., AND R. G. CLARK Differential survival of yearling and adult female Mallards and its relation to breeding habitat conditions. Condor 104: DWYER, C. P., AND G. A. BALDASSARRE Survival and nest success of sympatric female Mallards, Anas platyrhynchos, and American Black Ducks, A. rubripes, breeding in a forested environment. Canadian Field-Naturalist 107: FLINT, P. L., J. B. GRAND, AND R. F. ROCKWELL A model of Northern Pintail productivity and population growth rate. Journal of Wildlife Management 62: FOLK, T. H., AND G. R. HEPP Effects of habitat use and movement patterns on incubation behavior of female Wood Ducks (Aix sponsa) in southeast Alabama. Auk 120: FORBES, M. R. L., R. G. CLARK, P.J.WEATHER- HEAD, AND T. ARMSTRONG Risk-taking by female ducks: intra- and interspecific tests of nest defense theory. Behavioral Ecology and Sociobiology 34: GILMER, D. S., L. M. COWARDIN, R. L. DUVAL, L. M. MECHLIN, C. W. SHAIFFER, AND V. B. KUECHLE Procedures for the use of aircraft in wildlife biotelemetry studies. U.S. Fish and Wildlife Service Resource Publication 140. GLOUTNEY, M. L., AND R. G. CLARK The significance of body mass to female dabbling ducks during late incubation. Condor 93: GOLET, G. H., D. B. IRONS, AND J. A. ESTES Survival costs of chick rearing in Black-legged Kittiwakes. Journal of Animal Ecology 67: GUNNESS, M. A., R. G. CLARK, AND P. J. WEATH- ERHEAD Counterintuitive parental investment by female dabbling ducks in response to variable habitat quality. Ecology 82: HANSON, H. C Criteria of age of incubated Mallard, Wood Duck, and Bob-white Quail eggs. Auk 71: HARVEY, W. F., IV, G. R. HEPP, AND R. A. KENNAMER Age determination of female Wood Ducks during the breeding season. Wildlife Society Bulletin 17: HEPP, G. R., T. H. FOLK, AND K. M. HARTKE Effects of subcutaneous transmitters on reproduction, incubation behavior, and annual return rates of female Wood Ducks. Wildlife Society Bulletin 30: HEPP, G. R., R. A. KENNAMER, AND W. F. HARVEY IV Recruitment and natal philopatry of Wood Ducks. Ecology 70: HEPP, G. R., R. A. KENNAMER, AND W. F. HARVEY IV Incubation as a reproductive cost in female Wood Ducks. Auk 107: HEPP, G. R., AND C. A. MANLOVE A comparison of methoxyflurane and propofol to reduce nest abandonment by Wood Ducks. Wildlife Society Bulletin 29: HOEKMAN, S. T., L. S. MILLS, D. W. HOWERTER, J. H. DEVRIES, AND I. J. BALL Sensitivity analyses of the life cycle of midcontinent Mallards. Journal of Wildlife Management 66: HOSMER, D. W., JR., AND S. LEMESHOW Applied logistic regression. 2nd ed. John Wiley and Sons, New York. KIRBY, R. E., AND L. M. COWARDIN Spring and summer survival of female Mallards from northcentral Minnesota. Journal of Wildlife Management 50: LAZARUS, J., AND I. R. INGLIS Shared and unshared parental investment, parent-offspring conflict and brood size. Animal Behaviour 34: LEE, S. J., M. S. WITTER, I.C.CUTHILL, AND A. R. GOLDSMITH Reduction in escape performance as a cost of reproduction in gravid Starlings, Sturnus vulgaris. Proceedings of the Royal Society of London Series B 263: LEMASTER, E. T., AND R. E. TROST Summer survival rate estimates of adult Wood Ducks: implications for banding programs. Journal of Wildlife Management 58: LESSELLS, C. M Brood size in Canada Geese: a manipulation experiment. Journal of Animal Ecology 55: MALLORY, M. L., D. K. MCNICOL, R. A. WALTON, AND M. WAYLAND Risk-taking by incubating Common Goldeneyes and Hooded Mergansers. Condor 100:

10 210 KEVIN M. HARTKE ET AL. MANLOVE, C. A., AND G. R. HEPP Patterns of nest attendance in female Wood Ducks. Condor 102: MARTIN, T. E Avian life history evolution in relation to nest sites, nest predation, and food. Ecological Monographs 65: MILONOFF, M., AND P. PAANANEN Egg formation, brood survival, and cost of reproduction as clutch-size-determining factors in Common Goldeneyes. Auk 110: MILONOFF, M., H. PÖYSÄ, P. RUNKO, AND V. RUUSILA Brood rearing costs affect future reproduction in the precocial Common Goldeneye Bucephala clangula. Journal of Avian Biology 35: MONTGOMERIE, R. D., AND P. J. WEATHERHEAD Risks and rewards of nest defense by parent birds. Quarterly Review of Biology 63: MULCAHY, D. M., D. ESLER, AND M. K. STOSKOPF Loss from Harlequin Ducks of abdominally implanted radio transmitters equipped with percutaneous antennas. Journal of Field Ornithology 70: NETER, J., W. WASSERMAN, AND M. H. KUTNER Applied linear statistical models. 2nd ed. Richard D. Irwin Inc., Homewood, IL. POLLOCK, K. H., S. R. WINTERSTEIN, C. M. BUNCK, AND P. D. CURTIS Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7 15. PUGESEK, B. H., AND K. L. DIEM The relationship between reproduction and survival in known-aged California Gulls. Ecology 71: ROBB, J. R., AND T. A. BOOKHOUT Female and juvenile Wood Duck survival and movements in Indiana, p In L. H. Fredrickson, G. V. Burger, S. P. Havera, D. A. Graber, R. E. Kirby, and T. S. Taylor [EDS.], Proceedings of the 1988 North American Wood Duck Symposium. St. Louis, MO. ROHWER, F. C The evolution of reproductive patterns in waterfowl, p In B. D. J. Batt, A. D. Afton, M. G. Anderson, C. D. Ankney, D. H. Johnson, J. A. Kadlec, and G. L. Krapu [EDS.], Ecology and management of breeding waterfowl. University of Minnesota Press, Minneapolis, MN. ROHWER, F. C., AND H. W. HEUSMANN Effects of brood size and age on survival of female Wood Ducks. Condor 93: RUUSILA, V., AND H. PÖYSÄ Shared and unshared parental investment in the precocial Goldeneye (Aves: Anatidae). Animal Behaviour 55: RYAN, D. C., R. J. KAWULA, AND R. J. GATES Breeding biology of Wood Ducks using natural cavities in southern Illinois. Journal of Wildlife Management 62: SARGEANT, A. B., S. H. ALLEN, AND R. T. EBERHARDT Red fox predation on breeding ducks in mid-continent North America. Wildlife Monographs 89. SAS INSTITUTE SAS/STAT user s guide. Version 8. SAS Institute, Inc., Cary, NC. SCHMUTZ, J. A., R. F. ROCKWELL, AND M. R. PETERSEN Relative effects of survival and reproduction on the population dynamics of Emperor Geese. Journal of Wildlife Management 61: SEDDON, L. M., AND T. D. NUDDS The costs of raising nidifugous offspring: brood rearing by Giant Canada Geese (Branta canadensis maxima). Canadian Journal of Zoology 72: SLAGSVOLD, T., AND S. DALE Disappearance of female Pied Flycatchers in relation to breeding stage and experimentally induced molt. Ecology 77: WELLER, M. W A simple field candler for waterfowl eggs. Journal of Wildlife Management 20: WERNHAM, C. V., AND D. M. BRYANT An experimental study of reduced parental effort and future reproductive success in the Puffin, Fratercula arctica. Journal of Animal Ecology 67: WHITE, G. C., AND K. P. BURNHAM Program MARK: survival estimation from populations of marked animals. Bird Study 46:S120 S139. WIEBE, K. L., AND K. MARTIN The use of incubation behavior to adjust avian reproductive costs after egg laying. Behavioral Ecology and Sociobiology 48: WINKLER, D. W A general model for parental care. American Naturalist 130: ZENS, M. S., AND D. R. PEART Dealing with death data: individual hazards, mortality and bias. Trends in Ecology & Evolution 18: ZICUS, M. C Automatic trap for waterfowl using nest boxes. Journal of Field Ornithology 60: