Northern Prairie Wildlife Research Center
Our results indicate that the CRP provided substantial benefits to upland nesting ducks in the U.S. Prairie Pothole Region during 1992-1997. It would be most simple to infer then, that CRP cover was more attractive to nesting hens and provided greater security from nest predators relative to most other cover types in the study area, as was hypothesized. However, the effect of the CRP on duck recruitment in the Prairie Pothole Region appears to be more complex. One of our objectives was to compare DSR of nests found in CRP cover with that of nests found in WPA cover. Klett et al. (1988) reported nest success rates of 9-12% (for the same species we studied) in WPA cover in North Dakota for 1980-1984. These nest success rates are lower than rates believed necessary for duck population stability in this region (15-20%; Cowardin et al. 1985, Klett et al. 1988). When we initiated our study, we expected to find nest success rates in WPA cover to be similar to those reported by Klett et al. (1988). Kantrud (1993) studied duck nest success in WPA and CRP cover in Minnesota and North Dakota during 1989-1991 and found higher nest success in CRP cover than in WPA cover and nest success in WPA cover was similar to that reported by Klett et al. (1988) for 1980-1984. We did not find a difference in DSR for nests found in CRP cover compared to those found in WPA cover during 1992-1995. In our study, nest success in both types of cover was higher than reported for WPA cover by either Klett et al. (1988) or Kantrud (1993). We believe an explanation for our dissimilar results is related to our finding that DSR in CRP cover was positively related to the percent of perennial grass cover on study plots. Because WPA cover is similar to CRP cover, DSR in WPA cover presumably also was related to the amount of grass near the fields studied. We could not test this assumption because some WPA were located off the 10.4-km² study plots where we did not have land-cover information. However, CRP cover was distributed broadly throughout our study area and presumably contributed to improvements in nest success we observed in WPA cover during the CRP period. We speculate that DSR in many other cover types were influenced by the percent of perennial grass cover in the surrounding area. This is consistent with our findings of higher overall nest success during 1990-1994 (CRP period) than 1980-1984 (pre-CRP period). Greenwood et al. (1995) found that success of upland nesting ducks in grassland areas of prairie Canada (all habitats included) was negatively related to the percent of cropland in the landscape (approximately the inverse of perennial grass). In our study, DSR increased from northeast to southwest for areas of equal PGRASS, or, stated another way, to achieve similar DSR a higher percentage of perennial grass would be necessary in the northeast compared to the southwest.
Several possible explanations may explain why nest success in CRP and WPA cover was higher during our study compared to WPA cover during the pre-CRP period, and why DSR in CRP cover was positively related to PGRASS. Because predation was the principal cause of nest failure in this and other studies in the Prairie Pothole Region (Klett et al. 1988, Sargeant and Raveling 1992), any factor that significantly influences nest success likely influences predator foraging activity. The CRP resulted in the conversion of 1.9 million ha (7% of the total land area) of cropland to undisturbed grass cover in our study area. In some counties, 20-25% of cropland was enrolled in the CRP. Perhaps this large scale increase in idle grass cover provided nesting ducks and predators with increased nesting and foraging options, respectively, that reduced predator contact with nests. Another factor may be availability of prey other than ducks and duck eggs for predators. CRP cover in the northern plains provides suitable habitat for small mammals such as deermouse (Peromyscus maniculatus), vole (Microtus spp.; Lysne 1991), and numerous species of grassland nesting birds (Johnson and Schwartz 1993). Many of the predators that prey on ducks and duck eggs (Sargeant et al. 1993) also prey on small mammals and other birds or their eggs (McAtee 1935, Jones et al. 1983). Presumably, individual predators are capable of taking a finite number or mass of prey. As prey availability increases, the rate of predation should decline, assuming predator numbers do not increase in response to the increased availability of prey. The availability of food resources has been shown to influence predation rates on waterfowl nests in other areas (Pehrsson 1985, Summers 1986, Crabtree and Wolfe 1988).
With the large amount of perennial grass cover, including CRP, that occurred in some study plots, some fields or partial fields of CRP and WPA cover may not have been frequently visited by predators. These undisturbed conditions may have allowed hens to settle and find security long enough for clutches to hatch. Interestingly, several CRP fields contained high densities of nests (>2.5 nests found/ha searched), while densities in adjacent or nearby fields were much lower. These "hot spots" were usually characterized by intervals of apparently high DSR (i.e., most nests survived to late stages of laying or incubation and/or nest success was high). High densities of duck nests have been observed by others on areas protected from predators, such as islands (Duebbert 1982, Duebbert et al. 1983) or fenced exclosures (Cowardin et al. 1998).
Our models that included PGRASS and LOC explained only 10-31% of the observed variation in nest DSR. We included a priori other variables (Table 5) in our analyses, which we expected might help explain variability in nest DSR.
Density-dependent recruitment in North American Anatinae has been of interest for some time (Dzubin 1969, Dzubin and Gollop 1972, Pospahala et al. 1974, Weller 1979, Kaminski and Gluesing 1987). Kaminski and Gluesing (1987) reported some compelling evidence for density-dependent recruitment rates in mid-continent mallards. However, their study did not identify the mechanism(s) by which density dependence operates. If density-dependent recruitment exists, it must operate by influencing some component of the reproductive process. The definition of density is a complex issue, and in the past, density has been defined in various ways, such as: (1) breeding pairs/wetland in certain study areas in prairie Canada (Dzubin 1969), (2) breeding population/specified survey area (Kaminski and Gluesing 1987), or (3) breeding population distribution influenced by availability of wetland habitat (Pospahala et al. 1974). We used breeding pairs/10.4-km² (BPOP) as a measure of population density for each species on each study plot-year and included it as an explanatory variable in our stepwise regression analyses of DSR. Daily survival rate consistently was positively correlated with BPOP for the 5 species we studied, but the relationship was not strong enough to retain in 4 of 5 final regression models. Thus, we conclude that nest DSR in CRP cover were not strongly related to breeding population density.
Crissey (1969), Kaminski and Gluesing (1987), and Reynolds (1987) found positive relationships between large-scale indices to mallard production and pond numbers in the Prairie Pothole Region. This implies that some component(s) of productivity is related to wetland conditions on the breeding grounds. Nesting effort (i.e., nests/hen) and nest success are the principal components of hen success (i.e., proportion of hens that produce a brood during the breeding season). Krapu et al. (1983) and Cowardin et al. (1985) found that mallard nesting effort in North Dakota was positively related to abundance of ponds. Greenwood et al. (1995) reported similar relationships for 5 species of upland nesting ducks in prairie Canada. The influence of water conditions on nest success is equivocal. Johnson et al. (1988) reported that predation rates on early-season duck nests in the Prairie Pothole Region of Canada were lower in areas and years in which larger fractions of seasonal wetlands contained water. They found a similar relationship between late-season nests and semipermanent wetlands. Cowardin et al. (1985) found that mallard nests were more successful when ponds were more abundant at the time of nesting. Conversely, Greenwood et al. (1995) included wetland variables in their analyses of duck nest predation rates in the Prairie Pothole Region of Canada and found no effect, and Beauchamp et al. (1996) found no evidence that nest success was associated with conserved soil moisture (index to wetness). WETPOND entered into 1 of 5 species models in suite 1 and none of the species models in suite 2 (Table 6). WETAREA did not enter into any of the models. We conclude that numbers and area of wet basins did not influence nest success in our study plots.
Because predation was the primary cause of duck nest failure, the number of individuals and species composition of the predator community on or near our study plots could have influenced nest success. Red foxes have been identified as the most important duck nest predator in much of the Prairie Pothole Region (Johnson and Sargeant 1977, Johnson et al. 1988, Higgins et al. 1992, Sargeant et al. 1993). Coyotes also are a predator of duck nests (Sargeant et al. 1993) but are considered to pose less of a threat to nests than red foxes. Sovada et al. (1995) found that nest success of upland nesting ducks was about 15 percentage points higher on study areas where coyote was the dominant canid predator compared to areas where red fox dominated. In general, red fox abundance likely decreased and coyote abundance increased from east to west across our study area (Sargeant et al. 1993). For these reasons, we included indices to red fox and coyote abundance in our analyses of nest DSR for study plots in North Dakota. We found no evidence of a relationship between DSR and red fox abundance after other variables were considered and little influence of coyote abundance (Table 6). However, our indices of abundance were crude and were based on county-scale survey data as opposed to more preferred study-plot survey data.
We acknowledge that factors other than the CRP may have been responsible, at least in part, for the increase in duck nest success observed between the pre- and post-CRP periods. Sovada (1993) provided evidence that coyote populations expanded in much of our study area during a time partly coinciding with our study. If so, an argument could be made that increased nest success should be expected even if no change in the landscape had occurred. However, our finding that nest DSR was positively related to the percent of grass cover on our study plots supports the premise that the increase in CRP cover was at least partly responsible for the increase in nest success between the pre-CRP and CRP periods. In an attempt to check the logic of this finding, we used our regression models to estimate the average nest success in WPA cover for the combined 5 duck species using cover compositions that existed on our sample plots during the pre-CRP period (1980-1984). We then compared these expected nest success estimates with the observed nest success for that same cover type and period. Our expected nest success estimate was 17.26 (SE = 4.06) compared to the observed 17.53 (SE = 4.44). This finding supports our conclusion that the relationship between PGRASS and nest success is legitimate. Furthermore, in a 3-year (1992-1994) study of duck nest success in planted nesting cover in southern Saskatchewan (McKinnon and Duncan 1999), nest success was lower than we found in CRP and WPA planted cover but similar to that reported by Klett et al. (1988) for planted cover in our study area during pre-CRP periods. Canada does not have a landscape-level program similar to the CRP; therefore McKinnon and Duncan's (1999) results might reflect expectations of nest success in planted cover in the absence of such a program.
We conclude that the CRP has significantly benefitted populations of upland nesting ducks in the Prairie Pothole Region of the U.S. by providing attractive, secure nesting cover that is available to a large portion of nesting hens in the region. Johnson et al. (1992) concluded that nest success was the most important component of the reproductive process for mallards and other dabbling ducks in the Prairie Pothole Region. We estimated that 30% of the recruits hatched in our study area were from nests in CRP cover, and our results suggested that hatch rates in other cover types improved because CRP increased the amount of perennial cover in the landscape. Overall, we estimated that 12.4 million additional ducks (average of 2.07 million per year) were produced in our study area during 1992-1997 with CRP cover on the landscape compared with predicted production which simulated cropland in place of CRP cover. For the 5 common duck species, nest success in our study area for all nesting habitats combined was above levels considered necessary for population stability (Klett et al. 1988). During the pre-CRP period, we estimated that average nest success was below maintenance level for all species except gadwall. The combined impact of high nest success and a strong nesting effort due to ample availability of ponds across most of our study area in most years resulted in high production during the 1992-1997 period. Beauchamp et al. (1996) presented evidence that a widespread decline in duck nest success had occurred across the Prairie Pothole Region between 1935 and 1992, and suggested that a large-scale solution would be required to reverse the trend. We believe that the CRP contributed substantially to such a solution for the U.S. portion of the region.