Northern Prairie Wildlife Research Center
During 1976-81 we weighed several thousand Mallard, Gadwall, and Blue-winged Teal eggs, young, and adults. The work was conducted in central North Dakota in conjunction with a study of homing and reproductive strategies of these three species (Lokemoen et al. 1990). The study area encompassed 22.6 km² of moderately rolling landscape. About 10% of the surface area was wetland, and the uplands were about equally divided between native grassland and cropland.
Two measures of environmental conditions were made. Wetland area was measured as the total hectares flooded in May plus the total hectares flooded in July of the same year. The total hectares flooded in May and July better represent the average wetland conditions for a year than a single annual survey. Wetland change was the difference between wetland area of a year and wetland area of the previous year. We used the variable wetland change to indicate the hectarage of wetland that had dried or been flooded since the previous year. Wetland area in hectares and wetland change for the six years were as follows:
Wetland area | Wetland change | |
1976 | 88 | Unknown |
1977 | 30 | −58 |
1978 | 164 | 134 |
1979 | 199 | 35 |
1980 | 109 | −90 |
1981 | 102 | −7 |
Duck weights were obtained from arrival of the birds in late March or early April until late August or early September. Egg weights were measured at nests during laying and at the pipping stage. Weights of young were recorded at nests and on wetlands when young were captured with drive-traps (Cooch 1953), by night-lighting (Cummings & Hewitt 1964), or with decoy-hen traps (Sharp & Lokemoen 1987). Most adults were captured with rocket nets (Dill & Thornsberry 1950) or decoy-hen traps. Hens on nests were captured with long-handled dip nets or nest traps adapted from Salyer (1962). Weights were taken with Pesola spring scales, whose accuracy was regularly checked with known weights. Weights less than 100 g were recorded to the nearest gram and measurements over 100 g were made to the nearest 5 g.
The exact age, in days, of many young was known because the birds had been marked at hatching with web-tags (Alliston 1975). The age of unmarked young was estimated by feather-age classes described by Gollop & Marshall (1954). These birds were assigned an age equal to the midpoint of the appropriate feather-age class.
The age of many adult females was known because they had been marked as young or adults in a previous year. For unmarked females we distinguished those in their second year (SY) from those after their second year (ASY) using age classification systems developed by Krapu et al. (1979) for Mallard, by Blohm (1977) for Gadwall, and by Dane (1968) for Blue-winged Teal. We could not distinguish ages of adult males of any species.
Weights of flightless ducklings were fitted to growth curves by using nonlinear regression (procedure NLIN of SAS (SAS Institute, Inc. 1987)). Ages were estimated from feather-age classes. To determine the asymptotes of the curves, weights of some adults were used in the model-fitting, and an age of 320 days was ascribed to them. We used only early-breeding-season adult weights: before 15 April for Mallard and before 5 May for Gadwall and Blue-winged Teal. Resulting curves gave a general pattern of duckling growth. We examined departures from the pattern by computing residuals from the model (observed weights minus predicted values) by species and sex, and used these in an analysis of variance to seek effects due to wetland area, wetland change, and year.
For Gadwall, we had enough data to compare actual weights and predicted weights of known-age ducklings. Predicted values were taken from the growth curve model based on ages estimated from feather-age classes. We also calculated expected weights for young flying Mallard (average age 56 days) and Gadwall (50 days). To determine if the attainment of flight had any effect on weight gain, we compared actual weights of flying young with those predicted from the growth curve.
Weights of adults were grouped by 10-day intervals beginning with 27 March-5 April. Each 10-day interval was coded by its midpoint. Intervals with < 10 measurements were grouped with adjacent ones and the weighted average date was used for the coded date.
We used the General Linear Models Procedure (SAS Institute, Inc. 1987) to test for relationships between adult weights and several explanatory variables. These variables included age (for females only), a variable indicating whether or not the bird had been captured on a nest (females only), 10-day interval, year, wetland area, and wetland change. Models were developed by successively eliminating non-significant (P > 0.05) variables. Averages of adult weights were expressed as estimated population marginal means, which accounted for the unbalanced sample sizes (SAS Institute, Inc. 1987:600).