USGS - science for a changing world

Northern Prairie Wildlife Research Center

  Home About NPWRC Our Science Staff Employment Contacts Common Questions About the Site

Homing and Reproductive Habits of Mallards,
Gadwalls, and Blue-winged Teal


Field data were collected each year during 1976-81 on both study areas from the time waterfowl returned in April through fledging in late August or early September. Nest searches and trapping of hens were not conducted on the Woodworth study area in 1981.


For most analyses we used a sample of hens defined as residents. We considered a female a resident for a particular year if she was identified with a nasal marker and she nested or was hatched within the study areas, or if she was present for at least 14 days after capture. A female was considered to return if she was a resident 1 year and was resighted at any time on the study area the next year. Hens that first returned 2 or more years after being a resident were not used to calculate return rates or homing rates. A female that returned was considered a resident again if she fulfilled the defined requirements of a resident. A female that was a resident in 1 year and was alive the next year but did not return to the study area was called a pioneer. A female that was marked with a leg band or a web tag in year 1 and was trapped on the study area the next year was termed a recapture. Data from nasal-marked hens, leg-banded hens, and web-tagged birds were treated separately.

We defined return rate as the percent of resident hens marked in 1 year that were resighted on the study areas the next year. Recapture rate was the percent of leg-banded or web-tagged hens marked in year 1 that were retrapped on the study areas the next year. Homing rate was the percent of resident hens marked in year 1 that were resighted on the study areas during the next year given that they were alive at that time. For calculating return rates and homing rates, we used only those resident hens that were observed on the study areas after the date when 90% of the nests had been initiated (6 Jun for mallards, 20 Jun for gadwalls, and 8 Jun for blue-winged teal). These dates were used to maximize the probability that resident hens used in our analyses had nested only on the study areas.

A direct recovery was a report of death of a resident hen during the first hunting season after banding (Anderson 1975). An indirect recovery was the report of death of a resident hen after the first hunting season after banding. A successful hen hatched 1 or more eggs in a nest or was observed with 1 or more dependent young. Ages of resident hens were defined as birds in the second year (SY), after second year (ASY), and after third year (ATY). Adults of uncertain age or SY, ASY, and ATY hens combined were designated as after hatching year (AHY).

Nest Data

We searched seeded nesting cover with a cable-chain drag (Higgins et al. 1969) an average of 5 times each year. Other cover types were searched 4 times each year with a 61-m chain towed between 2 vehicles. All cover types except growing grain were searched at the Koenig study area, whereas only seeded nesting cover was searched at the Woodworth study area. Small areas were examined on foot. We defined a nest as any depression containing 1 or more eggs. A numbered stake was placed 4 to north of each nest. Duck species, number of eggs, estimated stage of embryo development (Weller 1956), and habitat type were recorded for each nest. Nesting success was calculated using the Mayfield 40% method (Johnson 1979). The date of nest initiation was determined by counting back from the date of nest location 1 day for each day of incubation plus 1 day for each egg in the clutch.

Nesting Habitat

Species composition and density of vegetation were measured during 197-81 along 26 linear transects that extended diagonally across all major cover types except stubble and wetland. Transects were started in the corner of each field nearest the road, and measurements were made every 5-20 paces depending on field size. Measurements were taken in spring before plant growth occurred and again in midsummer. Plant density was measured with a height-density pole modified by Kirsch et al. 1978) from that developed by Robel et al. (1970) and expressed as a visual obstruction rating. Nesting habitat on the Koenig study area was classified into 8 cover types (Table 1).

Table 1.   Vegetative composition and density and annual use of 8 primary nesting habitats on the Koenig study area, North Dakota, 1978-81.
Nesting habitat Primary plant species Annual
Seeded nesting
Agropyron intermedium None 1.37 4.11
Agropyron elongatum      
Medicago sativa      
Melilotus officinalis      
Pasture Agropyron smithii Grazed 0.24 1.25
Bouteloua gracilis      
Rosa woodsii      
Symphoricarpos occidentalis      
Hayland Medicago sativa Hayed 0.15 2.80
Roadside Bromus inermis None 0.51 3.20
Dry wetland   None    
Stubble Triticum aestivum Cropped    
Odd areas b Bromus inermis None 1.00 3.25
Canal-side Bromus inermis None 1.27 3.80
Agropyron repens      
a VOR = visual obstruction rating.
b Includes fence rows, farmsteads, shelterbelts, and rock piles.

Nesting Statistics

We compared characteristics of the last nest of successful and unsuccessful AHY mallard and gadwall hens in 1 year with first nests the next year and first nests of SY or TY hens to their natal nests. Mallard nests initiated before 17 May and gadwall nests initiated before 4 June were assumed to be first nests of the year because no earlier renests of marked mallard or gadwall hens were found before these dates. Analysis of variance (ANOVA) tests were used to examine distances between nest sites of the same hen or a hen arid tier progeny between years as related to success and age. If ANOVA indicated a significant difference, the Newman-Keuls method (Snedecor and Cochran 1980) for pair-wise comparisons was used to determine where differences occurred. Statistical differences between nest site physiognomy and plant species at nest were determined by standard chi-square tests.

Analysis of nest survival rates followed the method outlined by Johnson (1979) for testing differences in daily mortality rates between 2 groups of nests. We extended this procedure by performing an ANOVA on the daily mortality rates using exposure days as a weighting factor. Chi-square test statistics were obtained from the sum-of-squares for each effect in the model. In examining the interactions among years, cover types, and species, only the relationship between year and species was significant. To investigate this interaction, we conducted an additional analysis by species and tested for main effects due to year and cover.

Nest initiation periods for all nests and arrival dates (date the marked hen was first seen) for resident hens were examined by comparing the points at which 10, 50, and 90% of the respective observations occurred. We used correlation coefficients to compare the relationship between the visual obstruction ratings obtained on the Koenig study area with duck nest densities found in 6 cover types in 1979-81.

Capturing, Aging, and Marking

We attempted to capture all mallard and gadwall hens on the Koenig study area. On the Woodworth study area we marked hens only at nests in seeded nesting cover. Blue-winged teat hens were captured at nests only in 1977 through 1980. Hens were captured with rocket-propelled nets (Dill and Thornsberry 1950, Sharp and Lokemoen 1980), floating bail traps (Thornsberry and Cowardin 1971), and decoy hen traps (Anderson et al. 1980, Sharp and Lokemoen 1987). We captured hens on nests with long-handled dip nets and bow-net traps adapted from Salyer (1962), which at times were triggered by remote-controlled systems (Shaiffer and Krapu 1978). Hens with broods were captured by drive trapping (Cooch 1953) or at night from airboats with spotlights (Cummings and Hewitt 1964).

Ages of flightless ducklings were estimated from plumage patterns described by Gollop and Marshall (1954). In spring, ages of captured yearling and older females were identified by using classification systems developed by Krapu et al. (1979a) for mallards, Blohm (1977) for gadwalls, and Dane (1968) for blue-winged teal. The age of many hens was known because marked hens returned in subsequent years.

We attached nasal markers to all adult females and all juvenile mallard and gadwall females that weighed ≥250 g (Lee 1960, Bartonek and Dane 1964, Lokemoen and Sharp 1985). All hens captured also were banded with standard United States Fish and Wildlife Service (USFWS) leg bands. In addition, mallard and gadwall ducklings at nests were marked in a web of the foot with numbered tags (Alliston 1975).

We observed all wetlands on each study area once or twice each week from early April to mid-August to detect nasal-marked birds. Also, we intermittently examined larger wetlands off the study areas for nasal-marked hens. For these observations, we used binoculars, 20x spotting scopes, and 50-80x mirror scopes. Date, location, and behavior were recorded during each observation. Capture females were examined to identify leg-band and web-tag numbers. We calculated direct recovery rates from mallard and gadwall hens banded on both study areas using models developed by Johnson (1974). Banded birds were separated into 3 age classes (HY, SY, ASY), and a chi-square comparison was used to determine differences in recovery rates among the 3 age classes.

Return Rate and Homing Rate Statistics and Survival Estimates

Tests of significance for differences in hen return rates and homing rates were performed using the FUNCAT procedure (SAS Institute, Inc., 1982). Variables initially examined included the previous year breeding success, study areas, study years, hen age, and the interactions between these effects. Sample sizes for study years were insufficient for valid testing, so results for all years were combined. Study area effects were not significant for mallards (P = 0.59) or for gadwalls (P = 0.87); therefore, we combined study areas. Blue-winged teal ages were not obtained at the Woodworth study area, so return rates were tested separately for the 2 study sites. Only SY, ASY, and ATY hens were included in the final analysis because HY hens did not have a previous breeding history.

To determine homing rates for the 3 species, we needed to estimate survival of resident hens from the last observation in 1 year until 1 May of the next year. Survival of resident hens from the last observation in summer until 30 September was calculated by multiplying the number of intervening days times the estimated daily survival rate. Survival of resident hens of all species from 1 October until 1 May was derived by dividing the annual survival rate by a summer survival rate of 0.806 calculated for mallards by Cowardin et al. (1985).

For AHY mallard hens we used a late summer daily survival rate of 0.999 estimated by Johnson and Sargeant (1977:15) and Kirby and Cowardin (1986:41). We used the annual survival rate of 0.59 for AHY mallard hens banded in eastern North Dakota (Anderson 1975). Similarly, we used the annual survival estimate of 0.56 for HY mallard hens banded in eastern North Dakota (Anderson 1975).

For resident AHY gadwall hens we used the same late Summer daily survival rate of 0.999 used for resident AHY mallard hens. We used 0.68 annual survival for AHY gadwall females (estimated by R. J. Blohm, Office of Migratory Bird Management, Laurel, Md.) using model M-1 of the Brownie et al. (1985) program for hens banded in eastern North Dakota and southwestern Manitoba. The estimate of survival we used for HY gadwalls from 1 summer until the next 1 May was 0.54. This was an annual survival rate calculated from hens banded at the Koenig study area using the best fitting model (Johnson 1974) where adult and young survival rates are held constant.

To determine homing rates for AHY resident blue-winged teal hens we used the same late summer daily survival rate of 0.999 that was used for AHY resident mallard hens. The annual survival rate used for blue-winged teal AHY hens was 0.54 (estimated by J. D. Nichols, Patuxent Wildlife Research Center, Laurel, Md.) using birds banded before the hunting season in eastern North Dakota in 1964-70.

To estimate mortality of female mallards between hatching to near fledging, we divided the recapture rate of young marked with web tags at hatching by the return rate of female mallards identified with nasal markers at an average of 44.6 days of age. We made the same estimate for female gadwalls but calculated mortality for the period between hatching and birds marked at an average of 33.8 days of age. We assumed that survival of the 2 groups was the same after the age of 44.6 and 33.8 days and the differences in return rates reflected mortality for the intervening period. We considered one-half of the total web-tagged birds were females assuming a 50:50 sex ratio (Bellrose et al. 1961).

Recapture, Habitat Conditions, and Age-Related Analyses

D. H. Johnson (Northern Prairie Wildlife Research Center, Jamestown, N.D.) devised a test to determine if recaptures of web-tagged young were independent of those of brood mates. If we assume that recaptures are independent and there is the same probability of recovery (P) of individuals from all broods, then the expected fraction of broods from which n were tagged, that have exactly 1 recapture is

Thus, if there is dependence, one would obtain more recaptures of 0, 2, 3, or more young from a brood and fewer recaptures of exactly 1. The test was based on this statistic where the expected number of young recaptured from a brood was compared with actual number of broods where 1 young was recaptured.

We used the General Linear Models procedure of SAS Institute, Inc. (1982) to compare arrival dates of hens of different age and success classes. For ASY and ATY hens we performed an analysis that included effects for nesting success the previous year, age, and the interaction. For SY hens we compared arrival dates with those of ASY and ATY hens that were unsuccessful the previous year. Correlation coefficients were calculated to compare the relationship of May pond numbers to the ratio of hens captured and nasal marked on the Koenig study area that became residents.

We compared clutch sizes from all nests of resident hens between SY and ASY birds arid among years. Because clutch sizes declined as the nesting season progressed, we used initiation date as a covariate to clutch size and performed an analysis of covariance. We found only 6 known-age SY gadwall hens with complete clutches, and thus we did not attempt statistical comparisons of hen age and clutch size for this species.

In the analysis of hen age with recruitment, we did not use ANOVA because variances among classes were not homogeneous. In these situations, we determined differences with the Kruskal-Wallis test.

Previous Section -- Study Area
Return to Contents
Next Section -- Results

Accessibility FOIA Privacy Policies and Notices

Take Pride in America logo logo U.S. Department of the Interior | U.S. Geological Survey
Page Contact Information: Webmaster
Page Last Modified: Friday, 01-Feb-2013 19:31:53 EST
Sioux Falls, SD [sdww55]