Northern Prairie Wildlife Research Center
During 1992-1997, numbers of breeding duck pairs in our study area ranged from 1,479,469 (SE = 80,877) in 1992 to 8,020,051 (SE = 381,023) in 1997, and the 6-year mean was 4,751,664 (SE = 182,104). The mean annual estimates of breeding pairs by species were: mallard = 1,091,034 (SE = 38,234); gadwall = 798,714 (SE = 39,571); blue-winged teal = 1,960,265 (SE = 83,137); northern shoveler = 458,165 (SE = 18,008); and northern pintail = 443,486 (SE = 17,588). Numbers of ponds were lowest in 1992 (659,113; SE = 32,316), moderate in 1993 (1,221,227; SE = 49,120), and unusually high during 1994-1997 (mean = 1,631,037; SE = 69,148).
Duck Nesting Study
From 1992-1995, we located 10,727 duck nests on 98 study replicates or 161 replicate-year combinations (Table 2). Of these nests, 10,063 were useable for analyses and included 6,945 nests in 12,399 ha of CRP cover and 3,118 nests in 6,835 ha of WPA cover (Tables 2 and 3). Overall, species composition was 31% blue-winged teal, 27% gadwall, 20% mallard, 10% northern shoveler, 10% northern pintail, and 2% other (American wigeon [A. americana], green-winged teal [A. crecca], redhead [Aythya americana], and lesser scaup [A. affinis]). We located proportionately fewer nests of blue-winged teal (27% vs. 41%) and proportionately more nests of mallard (22% vs. 14%) in CRP than in WPA cover. Numbers of nests of remaining species were proportionately similar in CRP and WPA cover.
|Table 2. Number of replicatesa, replicate-year combinations, and area of Conservation Reserve Program (CRP) cover and Waterfowl Production Area (WPA) planted cover that were searched for duck nests in the Prairie Pothole Region of North Dakota, South Dakota, and Montana, USA, 1992-1995.|
|State||No. of replicates||No. of replicate-year combinations||Area searched (ha)|
|a Includes CRP cover and WPA cover paired in a study site.|
|Table 3. Number of usablea duck nests found in Conservation Reserve Program (CRP) cover and Waterfowl Production Area (WPA) planted cover in the Prairie Pothole Region of North Dakota, South Dakota, and Montana, USA, for estimating nest success for 5 species, 1992-1995.|
|Year||Mallard||Gadwall||Blue-winged teal||Northern shoveler||Northern pintail||Otherb||Total|
Excludes nests that were terminated when found, showed evidence of depredation
or parasitically laid eggs when found, or were abandoned or damaged
as a result of investigator.
b Includes 107 American wigeon, 31 A. green-winged teal, 20 redhead, and 49 lesser scaup nests.
Differences in estimates of DSR between CRP and WPA cover were small, and confidence intervals included zero (Table 4). Survival of nests was highest for blue-winged teal and lowest for northern pintail. Of the unsuccessful clutches, 5,065 (95.2%) were destroyed by predators, 238 (4.5%) were abandoned by the hen (some of these hens may have died while away from the nest), and 18 (0.3%) failed to hatch due to other causes (e.g., flooding, infertile eggs).
|Table 4. Least squares estimates of daily survival rates (DSR) of duck nests in Conservation Reserve Program (CRP) and Waterfowl Production Area (WPA) planted cover, and lower (LCL) and upper (UCL) 95% confidence limits for the DSR differences for 5 species in the Prairie Pothole Region of North Dakota, South Dakota, and Montana, USA, 1992-1995.|
Variables Affecting Daily Survival Rates
Correlation coefficients among candidate explanatory variables, BPOP, WETPOND, WETAREA, PGRASS, EAST, NORTH, FOXINDX, and COYINDX, were all ≤0.5 in absolute value with 1 exception: EAST and NORTH were negatively correlated (r = -0.68; n = 142), owing to the southeast-northwest orientation of our study area.
Daily survival rates were positively correlated with BPOP and PGRASS, and negatively correlated with EAST and FOXINDX (Table 5). The highest correlations usually included PGRASS and EAST, but absolute values were ≤0.5. Scatter plots of DSR, by species, versus the various explanatory variables did not indicate nonlinear relations or nonconstant variances. In stepwise regression analyses considering all explanatory variables (study plots in North Dakota), PGRASS was the first variable to enter the model for each species, and the only variable to appear in all models (Table 6, suite 1). PGRASS was also present in 4 of 5 species models that included all study plots, but excluded FOXINDX and COYINDX (Table 6, suite 2). Location effects appeared in models for mallard, gadwall, and blue-winged teal. Northern shoveler was the only species to not show an effect of either PGRASS or LOC in suite 2 models. Regression models explained ≤31% of the variability in DSR (Table 6).
|Table 5. Correlation of daily survival rates of nests for 5 species of ducks with breeding population size, numbers of ponds, area wet, percent of landscape in perennial cover, location (Universal Transverse Mercator easting and northing)a, and indicesb to red fox and coyote abundance. Sample sizes (n) are the number of study plot-years in North Dakota, South Dakota, and Montana, USA, 1992-1995.|
|Variable||Mallard||Gadwall||Blue-winged teal||Northern shoveler||Northern pintail|
|No. breeding pairs||0.14||141||0.08||148||0.30||142||0.24||113||0.26||109|
|Area wet (ha)||0.06||141||-0.09||148||0.07||142||0.05||113||0.05||109|
|% perennial cover||0.34||142||0.28||149||0.43||143||0.21||114||0.36||109|
|Easting (1,000 m)||-0.16||142||-0.21||149||-0.32||143||-0.15||114||-0.07||109|
|Northing (1,000 m)||-0.05||142||0.05||149||-0.17||143||-0.02||114||-0.06||109|
|Red fox indexb||-0.16||95||-0.13||95||-0.12||94||-0.23||75||-0.13||74|
coordinates projected in zone 14.
b Data from North Dakota Game and Fish Department. Similar data not available for South Dakota and Montana.
|Table 6. Results of stepwise regression models using daily survival rates of duck nests in CRP cover by species as the response variable. Two suites of explanatory variables for each study plot were considered: (1) breeding population size (BPOP), number of ponds (WETPOND), area wet (WETAREA), percent of study plot in perennial cover (PGRASS), location (LOC [consisting of Universal Transverse Mercator easting, northing and their arithmetic product]), and indices to red fox (FOXINDX) and coyote (COYINDX) abundance, and (2) the above variables excluding FOXINDX and COYINDX.|
|Species||Suite (1) a||Suite (2) b|
|Variables in final model c||n d||R2||Variables in final model c||n d||R2|
|Mallard||PGRASS, WETPOND||94||0.17||PGRASS, LOC||141||0.17|
|Gadwall||PGRASS, LOC||94||0.19||PGRASS, LOC||148||0.11|
|Blue-winged teal||PGRASS||93||0.18||LOC, PGRASS||142||0.31|
|Northern shoveler||PGRASS, COYINDX||74||0.16||BPOP||113||0.06|
|Northern pintail||PGRASS, BPOP||74||0.25||PGRASS||109||0.13|
Included data from study plots in North Dakota only.
b Included data from all study plots in North Dakota, South Dakota, and Montana.
c Significance levels for adding variables and retaining existing ones in the models were P ≤ 0.15. Variables in the final models are listed in the order in which they entered the model.
d Number of study plot-year combinations.
Model for Estimating Daily Survival Rates in CRP
Regressions of DSR on PGRASS and LOC were significant (P < 0.01) for all species except northern shoveler (mallard: F4,137 = 6.9, P < 0.001, R2 = 0.17; gadwall: F4,144 = 4.7, P = 0.001, R2 = 0.12; blue-winged teal: F4,138 = 15.4, P < 0.001, R2 = 0.31; northern shoveler: F4,109 = 2.9, P = 0.03, R2 = 0.10; northern pintail: F4,104 = 4.5, P = 0.002, R2 = 0.15). In analyses of residuals from these regressions, we failed to detect a year effect for any species (ANOVA: P = 0.04-0.68), nor were we able to detect species differences in regression coefficients for PGRASS, EAST, NORTH, or E × N (ANCOVA: F16,632 = 0.78, P = 0.72). We did, however, detect marginally significant differences in the constant terms for individual species (ANCOVA: F4,648 = 2.73, P = 0.03). Our final model was DSR = ai + b1(PGRASS × 10-2) + b2(EAST × 10-5) + b3(NORTH × 10-5) + b4(E × N × 10-12), where ai depended on species (mallard = 1.634, SE = 0.163; gadwall = 1.636, SE = 0.162; blue-winged teal = 1.639, SE = 0.163; northern shoveler = 1.640, SE = 0.163; northern pintail = 1.629, SE = 0.162), b1 = 0.0305 (SE = 0.0062), b2 = -0.0885 (SE = 0.0321), b3 = -0.0125 (SE = 0.0030), and b4 = 0.1529 (SE = 0.0604).
The model implies that at a given location, DSR would increase linearly as PGRASS was increased (Fig. 2), provided that values of PGRASS were within the range of those observed (5-80%). Daily survival rate increased from east to west and from north to south, but the rate of increase in both directions varied spatially (Fig. 3). Visually, the contour lines appeared to roughly follow the eastern or northern edge of the Missouri Coteau (Stewart and Kantrud 1973). The rate of increase from north to south was greatest in northeastern Montana and western North Dakota, and that from east to west was most pronounced in South Dakota.
|Fig. 2. Relationship between daily survival rate of nests for 5 species of ducks in Conservation Reserve Program cover and percent total perennial cover at an arbitrarily selected location (Universal Transverse Mercator easting = 4.75 × 105 and northing = 52.00 × 105) in the Prairie Pothole Region of North Dakota, USA, 1992-1995. Identical regression slopes with varying intercepts occur at other locations in our study area.|
|Fig. 3. Geographic variation in daily survival rate and nest success (in parentheses) of mallard nests in Conservation Reserve Program cover assuming that percent of perennial grass cover for the entire Prairie Pothole Region of North Dakota, South Dakota, and Montana, USA, was constant at 35%. Shaded areas represent gradients. Similar relationships were found for gadwall, blue-winged teal, northern shoveler, and northern pintail.|
Models that included nest age and initiation date provided evidence of effect for both variables, but differences in DSR between these models and those that did not include age or initiation date effects were small (1-2 percentage points for nest success estimates of most species-PGRASS-LOC combinations). Our objectives included comparing results with those of Klett et al. (1988), and because these authors did not adjust for nest age or initiation date, we chose to use models that did not include these variables.
We estimated average nest success (weighted by population size on each sample plot) in CRP cover for each species using our models. Nest success estimates in CRP were mallard 19.3% (SE = 2.0), gadwall 22.1% (SE = 2.4), blue-winged teal 24.4% (SE = 2.9), northern shoveler 26.5% (SE = 3.0), and northern pintail 22.6% (SE = 2.8).
Nesting Habitat Preference and Nest Success
We estimated nesting habitat preference and nest success by habitat, region, period and species using 26,697 nest records from the NPWRC waterfowl nest file. Species composition was 20% mallard, 24% gadwall, 40% blue-winged teal, 7% northern shoveler, and 8% northern pintail. Planted cover was the most preferred habitat and cropland was least preferred by all species (Table 7). Estimated nest success varied greatly among habitats, regions, species and periods, but was generally higher in 1990-1994 (MED = 19%, IQR [interquartile range; SAS Institute 1990] = 20%, n = 160) than in 1980-1984 (MED = 15%, IQR = 14%, n = 160). Habitat rankings, from lowest to highest nest success for all species combined, were right-of-way, wetland, other, grassland, hayland, idle grassland, planted cover (CRP and WPA), and cropland. The relative high nest success ranking of cropland was due to gadwall and blue-winged teal. Nest success in cropland was relatively low for northern pintail and mallards. Species rankings, from lowest to highest nest success, were northern pintail, mallard, northern shoveler, gadwall and blue-winged teal. Nest success was lowest in eastern North Dakota, followed by eastern South Dakota, central North Dakota, and central South Dakota.
|Table 7. Relative preferencea that 5 species of ducks show for 8 habitatsb in the Prairie Pothole Region of North Dakota, USA, 1966-1994.|
|Habitat||Mallard||Gadwall||Blue-winged teal||Northern shoveler||Northern pintail|
Preference for a particular habitat is the probability that a female
duck will select that habitat for nesting, given that all habitats are
b Habitats as defined by Klett et al. (1988), except odd area that included other, woodland, and scrubland as defined by Cowardin et al. (1988).
Impact of CRP on Duck Production
Estimated nest success and recruitment rates of the 5 principal species combined during 1992-1997 were 46% and 30% higher, respectively, with CRP cover rather than cropland on the landscape. Mallard and blue-winged teal showed the largest (38% and 32%, respectively) and gadwall showed the smallest (21%) gains in recruitment rate between the pre-CRP and CRP periods.
The estimated average annual difference in recruits (fledged male and female ducklings) produced during 1992-1997, with and without CRP, was 37,301 (SE = 65,008) in Montana, 989,727 (SE = 223,882) in North Dakota, 1,043,589 (SE = 354,854) in South Dakota and 2,070,617 (SE = 424,583) overall. The percent of total recruits that hatched in CRP was 40 (SE = 13), 36 (SE = 3), 21 (SE = 4), and 29 (SE = 3), respectively, for Montana, North Dakota, South Dakota, and overall. For all species combined, predicted nest success was higher in planted cover (CRP cover and WPA cover, 23%, SE = 3%) than in WPA cover only (18%, SE = 4%), when we modeled a scenario to simulate what would have happened if the CRP had not occurred.