Northern Prairie Wildlife Research Center
We used spotting scopes to observe marked canvasbacks during daily searches of the study block from mid-April through July. Systematic counts on the study area were conducted each year in mid-May for breeding pairs and in mid-July for broods. Data recorded during each observation included date, time, location, female identity, and size, age, and species composition of broods.
We conducted 2 complete searches of all over-water nesting habitat in the study block between late April and late June. Additional efforts to locate nests of specific marked females were made between searches. Information recorded for each nest followed the methods outlined by Klett et al. (1986) and included date, time, location, female identity, stage of incubation (Weller 1956), estimated initiation date, and number of eggs (canvasback or others) in and out of the nest bowl. All nests were revisited periodically to monitor their status. We searched for canvasback eggs underwater adjacent to each nest and assumed these to be laid by the resident female. Nest initiation date was estimated by back-dating the total number of canvasback eggs in and around the nest (assuming that 1 egg was laid/day), including those found underwater, and according to the stage of incubation. Widely differing initiation dates among eggs provided evidence of intraspecific nest parasitism (parasitic egg-laying by a female canvasback other than the host). We used standard terminology in reference to interspecific nest parasitism (Weller 1959, Sugden 1980, Sorenson 1990) by redheads (Aythya americana) and others, regardless of whether negative impacts were detected.
Successful nests contained ≥1 hatched egg. Unsuccessful nests were categorized as abandoned, destroyed by predators, or unknown. Each nest, egg remains, and surrounding area were examined to determine the number of eggs hatched and to identify nest predators, or other causes of nest failure.
The number of May ponds per km² in the Minnedosa Study Area each year was obtained using roadside transects methods described by Stoudt (1982). Similar counts of July ponds were not conducted, so annual changes in wetland numbers were based on May and July pond numbers derived from the closest transect segment of the U.S. Fish and Wildlife Service's aerial surveys of May breeding populations and July production (Off. of Migr. Bird Manage., Laurel, Md; stratum 40, transect 02, segment 04), which lies immediately south of the study area. Years were arbitrarily categorized as wet (1974, 1979), moderately wet (1975, 1976, 1978), or dry (1977, 1980) based on the range of pond conditions (Table 1). Temperature and precipitation data were obtained from the Environment Canada Atmospheric Environment Service at Brandon, Manitoba, 45 km south of the study area.
|Table 1. Environmental conditions in the Minnedosa Study Area, Manitoba, 1974-80.|
|Apr min. temp (C)||-5.5||-5.7||-2.4||-2.7||-3.4||-7.5||-3.2|
|May min. temp (C)||1.9||2.9||1.8||7.7||4.7||0.3||4.0|
|Seasonal change in wetland no.b (%)||-52||-36||-37||-17||-68||-66||-49|
|First arrival of canvasbacks||1 May||6 May||18 Apr||20 Apr||27 Apr||1 May||24 Apr|
|a Years categorized:
wet (1974, 1979), moderately wet (1975, 1976, 1978), and dry (1977,
b May-July change in wetland numbers based on aerial surveys (U.S. Fish and Wildl. Serv., Off. of Migr. Bird Manage., Laurel, Md.).
We assigned birds to 4 age classes: second-year (SY), after-second-year (ASY), third-year (TY), and after-third-year (ATY). The latter 2 classes were for birds marked in earlier years. We tested for differences among SY, TY, and ATY cohorts, and between ASY and ATY females. Few differences (P < 0.05) were found between TY and ATY females for all response variables, so we pooled TY and ATY females with ASY females for analysis (hereafter referred to as ASY's). Analyses were conducted with the Statistical Analysis System (SAS Inst. Inc. 1985).
Arrival date was the date that a marked individual was first observed on the Minnedosa Study Area and thus, actual arrival dates may have been earlier. We assumed that birds not observed until after 20 May had been missed and they were not used to calculate arrival dates.
We defined annual return rate as the proportion of females marked or observed in the study block in 1 year and resighted within 2 km of the block the next year. We examined relationships among age, previous breeding success, and return rate, with all years pooled using log-linear models (Feinberg 1981). The effects of age and previous nesting success on return rates were evaluated using Chi-square tests. Individuals with unknown breeding histories were excluded from the latter tests. We evaluated the influence of habitat conditions on return rates using simple linear regression of May pond numbers on return rates for each of the 3 age classes.
We lacked enough evidence for some individuals to classify a nest as a first, or a renest. Instead, we classified nests found during the first 15 days of the known nest initiation period each year as "early" nests and the remaining nests as "late." Early nests were likely first nests, given the 8- to 15-day renesting interval for canvasbacks (Doty et al. 1984). Those classified in the later period, however, may have included some first nests. Only 3 nests were found in 1977, and these were excluded from nest initiation data because of the extremely dry habitat conditions and delayed nesting.
We defined nest success as the proportion of all nests that successfully hatched at least 1 egg and hen success as the proportion of females that successfully hatched 1 or more eggs in a year. Hen success was determined from nest records and brood observations. Clutch size was determined from all eggs found inside and adjacent to the nest bowl. Nests known to contain canvasback eggs laid by a female other than the host female were omitted from analysis of clutch size.
We calculated apparent nest success for each year and age class. Nests that were deserted because of observer disturbance were excluded from the calculations. We used apparent nest success rather than the Mayfield method (Johnson 1979, Klett et al. 1986) because we believe that failed and active nests were equally likely to be located. Between 32% and 57% of the nests found each year had been terminated previously, indicating that we found most of the nests present.
We used contingency table analyses to evaluate differences in nest and hen success, rates of interspecific nest parasitism, rates of nest predation, and hen success relative to years of nesting experience. Differences were similarly compared for nest success between parasitized and unparasitized nests, among age-year groups, and relative to the number of parasitic eggs (0, 1-3, and ≥4) deposited in the nest. Age-year groups refer to data for age classes by calendar years. For comparisons among age classes, we used only those years when data for more than 1 age class were available.
We determined hatching success (proportion of eggs hatching/nest) and weighted each nest's hatching success with the clutch size for that nest by each species. Weighted hatching success for age-year groups for both canvasback and redheads predominantly, was compared with analysis of variance (ANOVA) on transformed values (arcsine square root transformation). Hatching success of parasitized and unparasitized nests was similarly compared. To compare hatching success of canvasbacks and redheads in the same nests, we used ANOVA with nests treated as random blocks. Also, we tested for differences in rates of nest parasitism among early (first 15 days of nesting), mid (second 15 days), and late season (days remaining) nests with Chi-square analysis.
We estimated the number of ducklings fledged/brood from the number of ducklings observed with a marked hen ≥20 days after hatching. Most duckling mortality occurs in the first 10 days of life (Leonard 1990; C. E. Korschgen, Northern Prairie Wildl. Res. Cent. unpubl. rep.). Differences in the number of ducklings fledged each year among age-year groups were tested using 2-way ANOVA.
We examined the influence of previous nesting experience on hen success using nesting histories among years via a means model ANOVA test. Years in which no apparent nesting occurred were excluded. The number of years of known previous experience varied from 0 to 4 years.
We evaluated effects of May pond numbers, April and May mean minimum and maximum daily temperatures, and precipitation on initiation date of early nests, nest and hen success, and clutch size using forward stepwise multiple regression. Differences in these variables among age classes were subsequently tested by analysis of covariance using the significant covariates. We evaluated the effect of age and minimum April temperature on arrival dates by a means model ANOVA approach (Milliken and Johnson 1984). A significant ANOVA (F = 4.02; 15, 143 df; P < 0.001) was followed by specific within-year comparisons among age groups and polynomial contrasts (Kirk 1982) across minimum April temperatures.