Seasonal Predator Removal Relative to Hatch Rate of Duck Nests in Waterfowl Production Areas

Study Design and Areas

We collected data from 12 pairs of WPA's: 3 in both 1987 and 1988, 5 in 1989, and 4 in 1990 (Table 1). We treated each study area-year (1 WPA during 1 year, hereafter called an AREA) as an independent datum. Eight pairs of AREA's were in west-central Minnesota (MN region) in Becker, Clay, Grant, and Ottertail counties. Four pairs were in southeastern North Dakota (ND region) in Barnes, LaMoure, McIntosh, and Stutsman counties. The AREA's of each pair had similar habitat with adequate ponds to attract many ducks. One of each pair was randomly selected to have predators removed (removal AREA); the other was a control (control AREA). Study constraints stipulated (1) predators would be removedfrom WPA's only, (2) removal methods would include only those approved by the U.S. Fish and Wildlife Service (USFWS) for widespread use, and (3) removal schemes would be suitable for use by USFWS managers.

Figure 1. Location of removal (predators removed) and control (no predators removed) study areas in Minnesota and North Dakota used to evaluate effects of seasonal predator removal on hatch rate of duck nests, 1987-1990.

The AREA's were in rolling topography with abundant temporary, seasonal, and semipermanent wetlands (Stewart and Kantrud 1971). Most land adjacent to AREA's was farmed to raise row crops, small grains, or livestock. The climate is continental with cold winters and hot summers. Annual precipitation averages about 58 cm in the MN region and about 43 cm in the ND region (U.S. Geol. Surv. 1970). Snowmelt occurs during late March-early April.

The AREA's ranged in size from 61-301 ha (Table 1). Each AREA had ≥32 ha of upland, mostly of idle grasses and fortes planted to attract nesting ducks (e.g., Duebbert et al. 1981). Removal AREA's were ≥5 km from other AREA's during the same year.

Wetland conditions at AREA's were good for breeding ducks during 1987; nearly all wetlands were flooded. Drought began that summer. In the ND region, many wetlands were dry during spring 1988. By spring 1989, water remained only in semipermanent wetlands, but levels were low. We conducted no field work in the ND region during 1990 because of continued drought. In the MN region, wetland conditions were fair for breeding ducks during 1988-1989, but water levels in most temporary and seasonal wetlands were low. During 1990, wetland conditions were fair for breeding ducks when the nesting season began and improved during the season.

Predator populations at AREA's were primarily mammalian (based on tracks, observations, and trapping; Sargeant et al. 1993, Sovada et al. 1995). In general, red foxes (Vulpes vulpes), striped skunks (Mephitis mephitis), and raccoons (Procyon rotor) were abundant; minks (Mustela vison) were common; and coyotes (Cants latrans) and badgers (Taxidea taxus) were uncommon. Franklin's ground squirrels (Spermophilus franklinii) occurred in nearly all AREA's in the ND region and in about half of AREA's in the MN region. American crows (Corvus brachyrhynchos) were present but not common in the MN region, and were rare in the ND region. Weasels (primarily ermine [Mustela erminea] in MN region and long-tailed weasel [M. frenata] in ND region) may have been present in all AREA's.

Searching for Nests

Duck nests were found by a crew of ≤4 persons searching upland of each AREA 3 times at 3-week intervals beginning the first week of May and ending the last week of June. Crews used chain-drags pulled by jeeps or ATV's to find most nests (Klett et al. 1986). People walking and beating vegetation with switches to flush hens found nests in habitat that could not be searched using vehicles. Habitat of AREA's too large to be completely searched was randomly sampled. Nests were marked for relocation with a flagged willow (Salix spp.) stake placed 4 m away. Data recorded for each nest included duck species and number and age (Weller 1956) of eggs. Nests were revisited at 7- to 10-day intervals until fate (hatched, abandoned, or destroyed) was determined (Klett et al. 1986). Cause of destruction was recorded as predation, weather, or other (e.g., flooding).

Predator Removal

Predator removal was conducted by trappers skilled in catching carnivores. It began in each removal AREA during 20 March- 12 April and continued until terminated during 28 June-15 July. Trappers were instructed to concentrate primarily on removing foxes, skunks, raccoons, badgers, and Franklin's ground squirrels, and secondarily on removing minks, weasels, and crows. Trappers were not instructed to remove coyotes because we assumed coyotes would reduce predation by foxes (Johnson et al. 1989).

The predator removal scheme during 1987-1988 was to employ a different trapper each year who visited each of 3 removal AREA's 2 consecutive days/week. The trappers were assisted by trap-attenders, who examined traps on other days. Predator removal schemes during 1989-1990 were primarily those used by Arrowwood, Detroit Lakes, and Fergus Falls USFWS Wetland Management Districts. Removal was conducted by 2 or 3 trappers (1/day) in each district.

Removal methods allowed in all AREA's were (1) padded-jaw foothold traps, (2) small (15 × 15 × 48 cm) and large (25 × 30 × 81 cm) wire-mesh live traps, (3) small (12 × 12 cm) and large (18 × 18 cm) quick-kill body-gripping traps placed in wire-mesh (for small traps) or wooden box (for large traps) cubbies or at burrow entrances, (4) shooting, and (5) hand-capture at dens. Snares and unpadded-jaw foothold traps were used in AREA's 712 and 722. Live traps were used only where traps were attended daily.

Trappers selected baits and determined how many traps to set and where to set them. Traps were attended daily except in AREA's 712 and 722 (attended on average every 3.4 days). Trappers and trap-attenders recorded time spent in each AREA and data (e.g., status of trap) for each visit to each trap-site. Uninjured nontarget animals were released at capture sites. Target animals and injured nontarget animals were killed humanely by shooting or injection of euthanizing agent.

Data analyses

We used HR to measure effects of predator removal because it is the most important determinant of recruitment rates of prairie ducks (Cowardin and Johnson 1979,Johnson et al.1992) and it could be accurately and economically measured. We excluded from analyses of HR nests with broken eggs or hens injured by investigators, nests abandoned because of investigator disturbance, and nests parasitized by another species (Klett et al. 1986).

We used SAS (SAS Inst., Inc. 1987) for statistical analyses; tests of significance were at P = 0.05. To calculate HR, we estimated daily survival rates (DSR's) using Mayfield's (1961, 1975) method as modified by Johnson (1979). We combined nests of species to increase sample sizes. We used DSR's for statistical tests, but for ease of interpretation, converted DSR's to HR (HR = DSR³⁴; Klett et al. 1986). We calculated confidence intervals for HR according to Johnson (1979). We used analysis of variance (ANOVA) in a randomized block design (Milliken and Johnson 1984) to examine effects of predator removal on DSR's. We treated pairs of AREA's as blocks. We fitted the ANOVA model using weighted least squares (Neter et al. 1990), weighting by the number of exposure days/AREA. We used least-squares means to estimate the effect of predator removal on DSR's. We used correlation analyses to examine the effect of number of predators removed on DSR's.

Outcomes

Destroyed duck nest shwoing the effect of mammalian predation in Waterfowl Production Areas in the prairie pothole region.

Duck Populations and Hatch Rates

We searched all (26 AREA's) or ≥50% (4 AREA's) of upland of AREA's (Table 1). We found 947 duck nests suitable for analyses of HR (57 nests excluded from analyses). Suitable nests/AREA ranged from 1-186 (n = 646) in removal AREA's and from 2-48 (n = 301) in control AREA's (Table 1). Species composition of nests was 58% blue-winged teal (Anas discors), 18% mallard (A. platyrhynchos), 10% gadwall (A. strepera), 7% northern shoveler (A. clypeata), 4% northern pintail (A. acuta), and 3% other species. Estimated initiation dates of nests ranged from 7 April-24 June; the median date was 20 May.

Mean DSR of nests was higher for removal AREA's than for control AREA's (DSR = 0.943, SE = 0.010 vs. DSR = 0.919; SE = 0.012; F = 4.74; 1, 14 df; P = 0.047). Mean HR was 13.5% for removal AREA's and 5.6% for control AREA's, but varied much among removal and among control AREA's (Table 1). The percent of nest failures from predation was slightly lower for removal AREA's than for control AREA's (91%, SE = 0.015 vs. 96%, SE = 0.013; X²₁, = 5.47, P = 0.019). Abandonment accounted for all other nest failures.

Predator Removal Effort

Predator removal was conducted for an average of 97 days/AREA (SD = 12.7) and involved 30,872 trap-days (1 functional trap for 24-hours). Trappers and trap-attenders spent an average of 1.5 hours/AREA (SD = 0.6) on days when traps were attended. Numbers of trap-days in each AREA averaged 634 (SD = 306) for foothold traps and snares combined, 968 (SD = 417) for large live traps and large body-gripping traps combined, and 456 (SD = 472) for small live traps and small body-gripping traps combined.

Predators Removed and Relation to Hatch Rate

Trappers removed 666 predators; 92% (n = 611) were adults and 8% (n = 55) were juveniles. Species composition of adults was skunk (47%), raccoon (20%), Franklin's ground squirrel (18%), fox (11%), badger (2%), weasels (2%), and other species (1%; minks [n = 4], crow [n = 1]). Mean numbers of adults removed/AREA were 18.9 (SD = 9.9) skunks; 8.0, (SD = 4.9) raccoons; 7.3 (SD = 8.8) Franklin's ground squirrels; 4.5 (SD = 5.3) foxes; 1.0 (SD = 1.6) badgers; and 1.1 (SD = 1.9) other species. Of the adults, 76% (n = 217) of skunks, 42% (n = 50) of raccoons, 58% (n = 63) of Franklin's ground squirrels, 75% (n = 50) of foxes, 67% (n = 10) of badgers, and 37% (n = 6) of other species were removed by 20 May, the estimated median date of duck nest initiations. Hatch rate was not related to total adult carnivores removed (r = -0.08, 13 df, P = 0.76), or to total predators removed (r = 0.02, 13 df, P = 0.95).

Trappers also removed 179 and released 62 nontarget animals. The removed nontarget animals were 97% mammals, primarily thirteen-lined ground squirrels (Spermophilus tridecemlineatus, n = 127). The released animals were mostly domestic cats (n = 26) and woodchucks (Marmota monax, n = 17).

Interpretations

Hatch rates

Our finding that intensive seasonal predator removal in scattered WPA's increased HR by 8 percentage points was unexpected because of greater effects of predator removal on HR reported by others (Balser et al. 1968, Duebbert and Kantrud 1974, Duebbert and Lokemoen 1980, Greenwood 1986, Doty and Rondeau 1987). Moreover, mean HR in our removal AREA's (13.5%) was below the 15-20% threshold suggested for stability of populations of the 5 most common species of ducks nesting in our AREA's (Cowardin et al. 1985, Klett et al. 1988). These findings demonstrate the difficulty of accomplishing effective predator removal to increase HR in small scattered wildlife management tracts that have diverse nammalian predator communities.

Our finding of 5.6% mean HR in uplands of control AREA's is consistent with findings of others (e.g., Greenwood 1986, Johnson et al. 1987, Klett et al. 1988, Greenwood et al. 1990), and further demontrates the severe effects of predation on duck proluction. However, HR among the control AREA's varied greatly; 2 had much higher HR (36%, 63%) than was needed for stability of populations. This variation illustrates the uncertainty associated with determining need for predator removal on individual WPA's. Our finding that number of adult carnivores removed or number of all predators removed was unrelated to HR demonstrates the inadequacy of using counts of removed predators to infer success of predator removal programs.

Factors hindering success

We think several factors hindered success of our predator removal efforts, especially (1) restricting the removal to WPA's only, (2) restricting choice of removal methods, and (3) inflexible work schedules. By restricting predator removal to the WPA's, we think we delayed or prevented capture of many predators. Many predators using our AREA's may have been more vulnerable to capture outside the WPA's (e.g., at dens or buildings). Also, by restricting predator removal to the WPA's, we ensured that all removed predators were entering or in habitats being evaluated for HR when captured. These factors may explain, in part, our failure to increase HR above that obtained by Greenwood (1986) and Doty and Rondeau (1987), who targeted fewer predator species but simultaneously trapped in WPA's and in surrounding lands.

Balser et al. (1968) suggested using a variety of methods to effectively remove predators because individual predators may be wary of a particular method. Our trappers had few methods available for use compared with previous studies. Predacides, extensively used in the studies of Balser et al. (1968), Duebbert and Kantrud (1974), and Duebbert and Lokemoen (1980), were not used because they were banned by federal and state governments. Snares, which are effective for capturing foxes and raccoons, were banned by state government in the MN region, and for consistency, were not used in the ND region except in 1989. For humane reasons, the foothold traps used in nearly all AREA's had padded jaws, even though padding increased the likelihood of animals escaping (Linhart et al. 1986). We think these limitations of removal methods reduced effectiveness of our trappers because many predators, especially carnivores, using our AREA's may have been wary of baits, traps, or humans. All adult carnivores at our AREA's had survived >1 fur-harvest season. Fur-trapping and predator-hunting are popular in WPA's and private land in our regions and create wariness among predators.

Inflexible 8-hour workdays and daily attending of traps generally required of trappers in our study by federal, state, or district regulations decreased trapper effectiveness. In the Fergus Falls District, each day the trapper on duty visited all traps in 4 AREA's during an 8-hour workday that began and ended at District headquarters. This resulted in an average of 0.9 hours/day spent in each AREA, and little time for reconnaissance or trapping hard-to-reach sites.

Cost-effectiveness

Our trappers and trap-attenders spent an average of 1.5 hours/day (when traps were attended) in each removal AREA. On this basis, 1 trapper could service no more than 4 WPA's/8-hour workday, leaving only 2.0 hours/workday for travel and other activities. By servicing 4 WPA's daily for 97 days (our average removal period), direct salary to individuals conducting predator removal, based on the 1990 wage of $8.56/hour paid to trappers in the Fergus Falls District, was $1,661/WPA ([97 days x 8 hours/day x $8.56/hour]/4 WPA's). Actual costs are much higher when costs of training, equipment, supplies, transportation, supervision, and overhead are included. Clearly, cost/additional hatched clutch from seasonal predator removal in WPA's is high unless HR is greatly increased or many clutches are affected or both. Moreover, for increased HR to result in greater production of fledged young, it must be followed by adequate survival of ducklings (Johnson et al. 1992).

Modifications of our predator removal schemes or methods to increase cost-effectiveness of increasing HR in WPA's by removing predators warrant consideration. However, we are not optimistic about possible outcomes in areas with predator populations like those we encountered. More intensive trapping, both in and near WPA's, might increase HR but would increase costs. Costs could be reduced by attending traps less than daily, but to treat animals humanely, this option generally is unacceptable and may be illegal. Exclusive use of quick-kill traps might reduce need to examine traps daily, but reliance on types we used would be ineffective for removal of some predators, especially foxes. Trapping fewer days would reduce costs but would also reduce effectiveness of the predator removal. Greater use of unpadded foothold traps, snares (especially effective), or use of predacides might increase effectiveness of predator removal without increasing costs. However, for legal, social, and humane reasons, use of these methods probably will be restricted in most predator removal programs.

Summary and Management Considerations

During 1987-1990, we evaluated effects of intensive seasonal predator removal on hatch rate (HR) of duck nests in uplands of waterfowl production areas (WPA's). We obtained data from 15 pairs of removal and control AREA's in Minnesota and North Dakota (9 pairs of WPA's during 1 year and 3 pairs during 2 years). Principal predators were fox, skunk, raccoon, and Franklin's ground squirrel. We searched upland habitat in each AREA 3 times for duck nests and found 947 nests that were suitable for analyses. Mean HR was 13.5% for removal and 5.6% for control AREA's. Predator removal involved 30,872 trap-days of foothold traps, body-gripping traps, live traps, and snares. We removed 611 adult predators (47% skunks, 20% raccoons, 18% Franklin's ground squirrels, 11% foxes, 2% badgers, and 3% other species), 55 juvenile predators, and 179 nontarget animals (97% mammals). Mean HR was unrelated to total adult carnivores removed or to total predators removed. Effective predator removal in scattered WPA's was costly and not easily accomplished.

Managers contemplating initiating predator removal programs in WPA's or similar sites should carefully consider the cost-effectiveness of the programs. Need for predator removal may seem greatest where predation rates are most severe. However, severe predation rates may be characteristic of areas where managers will have greatest difficulty accomplishing cost-effective predator removal, especially if duck populations are low. In such areas, the interactions of an unfavorable landscape and unfavorable predator community may overpower attempts to increase HR to desired levels. In our study, removal AREA's were small and in a highly fragmented agricultural region occupied by predator communities particularly harmful to nesting ducks (Sargeant et al. 1993). Habitat fragmentation may have increased amount of predator activity in WPA's. These factors and the small number of duck nests in many of our AREA's reduced the cost-effectiveness of predator removal. Managers initiating predator removal programs should recognize the difficulty of the task and concentrate on reducing predator activity by using skilled trappers, adopting flexible work schedules for trappers, and providing trappers with the greatest possible choice of removal methods. Programs that stress numbers of predators removed over quality of removal effort are likely to be ineffective.

Acknowledgements

Much financial, logistical, and technical assistance was provided by Arrowwood, Detroit Lakes, Fergus Falls, Kulm, and Valley City USFWS Wetland Management Districts; Central Flyway Council via University of North Dakota Institute for Ecological Studies; Mississippi Flyway Council via Iowa and Minnesota Departments of Natural Resources. We thank 8 trappers and >50 other persons who assisted with field work. We thank numerous staff of Northern Prairie Science Center who supported the study or provided technical services. We thank R. J. Greenwood, J. T. Lokemoen, and L. L. Strong for helpful comments on previous drafts.

Literature Cited

Alan B. Sargeant (left) is a Wildlife Research Biologist at the Northern Prairie Science Center (NPSC) where, for over 25 years, he has focused his research on predator ecology, predation, and predation management in relation to production of prairie ducks. He received his B.S. in Wildlife Management from the University of Minnesota at St. Paul.

Marsha A. Sovada (center) is a Wildlife Research Biologist at the NPSC. She received her M.A. in Biology from the University of Idaho in Moscow and a Ph.D. in Zoology from North Dakota State University. Her research includes predator ecology and predation management.

Terry L. Shaffer (right) is Chief of Support Services at the NPSC, where he provides statistical advice to biologists. He holds an M.S. in Applied Statistics from North Dakota State University. His research focus is on statistical and computer applications in ecological research.