Northern Prairie Wildlife Research Center
Spring and early summer nesting intervals and nesting success. I used spring and early summer nesting records to test for interspecific differences among the five species in dates of nest initiation and to compare rates of nesting success. The sample of nests was obtained by searching a stratified random sample of fields enrolled in the Conservation Reserve Program (CRP) in the PPR of North Dakota and north-central South Dakota. Fields searched were distributed across 128 10-km² sites, each of which contained at least 16.2 ha of CRP cover. The sites form a subset of those used in studies described by Reynolds et al. (1994). Data on nesting duration and nesting success were obtained through a minimum of three systematic nest searches of each plot between early May and early July 1993 to 1995 following standard methods (Klett et al. 1986). Laying or incubating females were identified to species when flushed, the number of eggs in each nest was determined, and the stage of egg development was established by candling (Weller 1956). The date of initiation of egg laying was estimated by backdating, taking into account stage of incubation and clutch size, and assuming that one egg was laid daily until the clutch was complete. Nests were revisited every 7 to 10 days until at least one egg hatched, the nest was abandoned, or the nest was totally destroyed. On each revisit to a nest, species identity and date were recorded, number of eggs was identified, and completed clutch size was recorded, if known. Nest fate was recorded on the last visit. A nest was deemed successful if at least one egg hatched, indicated by the presence of shell membranes (Klett et al. 1986) or ducklings in the nest bowl. Nesting success was estimated using the modified Mayfield technique (Johnson 1979).
Frequency and distribution of summer nesting. September brood surveys were conducted to evaluate which species continued to nest into mid- and late summer and to evaluate temporal chronology of mid- and late summer nesting. September brood surveys were timed to allow nests initiated before about 20 August to have hatched prior to the surveys. To assess relative frequency and duration of nesting by the five species in mid- and late summer, I measured numbers of broods of each species seen by age class within transects located on the FBSA which was centrally located within the SNSA. Initially, the FBSA was divided into 30 plots of equal size (13.7 × 9.4 km), but one was dropped when inspection revealed a large reservoir that encompassed most of the plot. Within the remaining 29 plots, 19 km × 0.4 km transects (0.2 km on each side of roads) were established randomly within the existing road system. We had enough people to survey 18 transects in random order. Transect surveys were conducted on 20 to 21 September 1993, 14 to 15 September 1994, and 13 to 14 September 1995. Coverage of transects was limited to wetlands visible from roads. Broods were identified to species and brood ages were estimated from plumage development (Gollop and Marshall 1954). Brood visibility during fall 1993 to 1995 was excellent because water levels were high and vegetation was defoliated by frost and senescence. In 1993, the initial survey was followed by a survey of broods during 22 to 30 September on 6 of the 29 plots. Similar data were gathered on plots and transects, and coverage of plots was limited to wetlands visible from roads. For each brood, I estimated nest-initiation date by backdating as follows: I summed the age of the brood in days (age-class midpoint; Gollop and Marshall 1954), the average incubation period of the species (Bellrose 1980), and the number of days required for laying of a clutch of average size (Bellrose 1980) assuming a laying rate of one egg daily until the clutch was completed, and then subtracted this sum from the Julian date of the survey.
Random surveys of the bag of duck hunters provided information to assess the relative frequency and distribution of mid- and late-summer breeding across the United States. Frequency of occurrence of flightless juveniles shot on or after 1 October provided a measure of which species nested in mid- and late summer and the distribution of summer breeding. Data were obtained from the annual fall duck harvest survey from states in the Central Flyway (North Dakota, South Dakota, Nebraska, Oklahoma, Texas, and parts of Montana, New Mexico, Colorado, and Wyoming) during 1993 and throughout the United States in 1994 to 1995. Selected hunters were asked to remove a wing from each duck harvested during the fall and submit wings to the Migratory Bird Management Office. Wings of flightless juveniles were separated from the rest of the wing sample and identified to species. Only young with primaries not having developed beyond the "blood quill" stage were included so that the sample was limited to young that were locally reared from nests initiated in mid- to late summer.
Statistical analyses. I used randomized block analysis of variance (ANOVA), with year as a blocking factor and species as a classification factor, to compare percentiles of nest-initiation dates. To maintain an error rate of α = 0.05 for each percentile, I employed Bonferroni's correction, rejecting null hypothesis of no differences among species when I found evidence of differences at P ≤ 0.01. When ANOVA revealed evidence of differences among species, I used Fisher's protected LSD test to conduct multiple comparisons to determine which species differed. I used S-PLUS 4.0 (Mathsoft 1997) to perform statistical analyses.