Northern Prairie Wildlife Research Center
Daily, seasonal, and annual precipitation amounts fluctuated greatly on the WSA. These differences and their temporal distribution also affected wetland habitats, upland-nesting habitats, and trends in waterfowl populations occupying these habitats.
Bellrose (1979) reported that the abundance of ponds in the prairie pothole region is the most important single factor regulating the production of mallards and, no doubt, some other duck species. During 1965-81, percents of the 548 wetland basins that contained water were usually highest in early spring and lowest in late summer (Table 3). The wettest spring was in 1966 when 87% of the basins contained water, and the driest was in 1977 when only 8% of the basins had water. The mean percent of basins with water was 56% during 1-15 May surveys for the 17-year period.
Class III (seasonal) wetlands were the most numerous during 1965-75 (Table 4). Class IV (semipermanent) wetlands were less common than Class III wetlands but maintained more surface water and held water more consistently. During 1965-75, an average of 50% or more of the wetlands with water were 0.2 ha or less, however, an average of 60% or more of the surface water area in wetlands was contained in 6-8% of the wetland basins and these were 4 ha or larger (Table 5).
Some wetlands with similar watersheds, basin morphology, vegetation, and watershed land use contained water while others did not during the same years. Likewise, individual wetlands that were dry or nearly dry during wet years, were wet during dry years. Possibly these phenomena occur because of differences in groundwater flows. According to area soils maps, between 5 and 10% of the basins were situated on porous or gravelly soil substrates. These basins were not of substantial value to ducks because nearly all the runoff water went directly into groundwater storage or flowage.
At the WSA, we categorized years as wet for ducks when at least 65-75% of basins were full or nearly full of water during 1-15 May. Factors affecting the percent of wetland basins with water during the 1-15 May and 1-15 August counts were soil moisture conditions during fall freeze, amount of snow accumulation over winter, amount of snow during March, April, and sometimes May, speed of thawing during spring, and amount of heavy rain fail during March and April. The number of basins holding water in the previous August also had some effect on number of ponds with water in spring. To predict when a potential wet year would occur, several combinations of precipitation measure were projected relative to the percent of basins containing water during early May and early August surveys (Table 6). The percent of wetlands with water during the 1- 15 May counts had higher correlation with conserved soil moisture (Williams and Robertson 1965) for the 21-month period preceding 1 May (r = 0.77, P <0.0003,17 df) than with annual precipitation (Pospahala at al. 1974) in the previous 12-month period of 1 June-31 May (r = 0.50, P <0.04, 17 df). There were no significant relations between precipitation amounts during the snow year, 1 September-1 May, or the crop year, 1 August-31 July (P >0.10), and wet ponds during the 1-15 May counts.
The percent of wetlands with water during the 1-15 August counts (Table 6) was related to conserved soil moisture (r = 0.74, P <0.0008, 17 df). There were no significant relations between the 1-15 August counts of wetlands with water and precipitation during the snow year, the calendar year, or in the 12-month period (crop year) preceding 1 August (P > 0.10). An explanation for the weakness of relations with some of the precipitation variables may be the lack of data for a series of consecutive wet or dry years. During this study, wet and dry years were usually single-year events.
In this region, upland habitats provided nearly all the nesting cover for dabbling duck species and lesser scaup whereas most other diving ducks nested overwater in emergent cover. However, some overwater nesting by dabbling ducks did occur in and near the area (Krapu et al. 1979). Although we did not measure the structural characteristics of the upland habitats, there were obvious differences in height and quantity of cover each year; probably related to precipitation and soil moisture conditions. In general, good-to-excellent nesting cover was available in years with average or above average precipitation and fair-to-poor in below average precipitation years. As with wetland water conditions, upland cover structure varied greatly among months and seasons but was usually better when soil moisture was continuously high during sequences of wet falls and springs. Frequent, small rains were believed better than a large, single rain for nesting cover growth.
Chronology of Arrivals
The dates migrating waterfowl arrived in spring at the WSA were highly variable among years, probably because spring weather over northern latitude prairies is uncertain. The earliest arrival was 15 March for mallards and northern pintails, whereas the latest was 16 May for ruddy ducks (Appendix E). During most springs, the first sighting of a species was usually followed by an influx of pairs within a few days to a week. Mallards and northern pintails were usually followed by redheads, American wigeon, gadwalls, lesser scaup, and ring-necked ducks between 25 March and 10 April; canvasbacks, green-winged teal, northern shovelers, and blue-winged teal between 1 and 20 April; and ruddy ducks between 20 April and 15 May.
Hammond and Johnson (1984), reporting on mallards, gadwalls, blue-winged teal, and redheads, found that arrival dates varied with mean temperatures before and during the usual arrival period for each species. Warm spring weather induced earlier arrival and nesting on the WSA. Arrival dates were most affected by temperature for mallards during 12-25 March, for redheads 12 March-8 April, for gadwalls 26 March-8 April, and for blue-winged teal 26 March-15 April. Warm spring weather also seemed to induce earlier nesting; mallards were affected most by temperature during April and the others by temperature during late April and May.
Pair Count Periods and Species Composition
We made the first pair count's during the first 2 weeks in May and a second count during the last week of May or the first 2 weeks in June. The first count closely coincided with the first blue-winged teal nest initiations and the second with the first gadwall nest initiations. Our census periods approximated those of Dzubin (1969), Smith (1971), and Stoudt (1971).
The mean composition of breeding pairs during 1965-81 was 86% dabblers (range 73-97%) and 14% divers (range 3-27%) (Table 7). These means compare favorably with Evans and Black (1956) for South Dakota, Dzubin (1969) for Manitoba, Henry et al. (1972) from May aerial surveys for the Prairie Provinces and Dakotas, Kiel et al. (1972) for Manitoba, and Trauger and Stoudt's (1978) study areas, one in Alberta and one in Saskatchewan. Dzubin (1969) had more dabbling ducks (97%) in the populations at Kindersley, Saskatchewan, and Trauger and Stoudt (1978) had a smaller percent of dabbling ducks (70%) in their Manitoba study areas.
Major fluctuations in species composition occurred over time; however, percent compositional changes were less pronounced than estimates of pair numbers because all species did not change proportionately within years. Percent composition of dabbling ducks was higher than average for all years 1965-73, whereas, beginning in 1974, dabbling duck percent composition was average or below average for all years 1974-81 (except 1976).
Blue-winged teal were the most abundant species, making up an average of 46% of the pair estimate. The next most abundant (between 10 and 15%) were gadwalls and mallards. All other species (Table 7) average less than 10% abundance; ring-necked ducks were scarcest (<1%). The average species composition at the WSA (Table 7) was nearly the same as that reported by Evans and Black (1956) during 1950-53 at Waubay, South Dakota. More lesser scaup at the WSA was the only minor exception between the two areas. This suggests that the species composition of ducks in the Dakotas has remained rather similar during the past 30-35 years.
Number of Pairs and Wetland Occupancy
Total pairs of ducks on the area averaged 492 and varied from 236 to 692 during 1965 through 1981 (Table 8). The density of duck pairs averaged 40/km² during the 17 years and ranged from 19 to 56/km². The density of basins with surface water during 1-15 May for the same period averaged 25/km² and ranged from 3 to 39/km².
The average number of pairs per 2.6 km², by species, was: blue-winged teal, 47; gadwalls, 14; mallards, 11; northern pintails, 7; northern shovelers, 6; lesser scaup, 5; redheads, 4; ruddy ducks, 4; American wigeon, 2; green-winged teal, 2; canvasbacks, 1; and ring-necked ducks, <1. The greatest pair densities per 2.6 km², by species, was: blue-winged teal, 76; gadwalls, 19; mallards, 16; northern shovelers and ruddy ducks, 13; northern pintails, 11; lesser scaup and redheads, 8; canvasbacks, 5; American wigeon, 4; green-winged teal, 3; and ring-necked ducks, 1. The greatest total density of pairs was 56/km² in 1975. This density was about half the pair density of 116/km² Smith (1971) counted at Lousana, Alberta, in 1958.
The percent of wetlands containing one or more pairs of ducks in mid-May counts during 1965-75 varied fron 24.3 to 44.8% and averaged 36.8% (Table 9); late May and early June counts varied from 33.7 to 85.5% and averaged 47.5% (Table 10). The number of wetlands with pairs was directly related to shoreline length (Table 11) and to size class (Table 12) of wetlands. These relations were consistent for all duck species; that is, large wetlands were more likely to have pairs than small ones.
The 11-year mean of 36.8% occupancy of wetlands by duck pairs in mid-May at the WSA was lower than the 14-year mean of 46.0% in Saskatchewan (Stoudt 1971) and the 13-year mean of 55.0% in Alberta (Smith 1971), but the 47.5% average occupancy rates for late May and early June counts compare favorably to those of Smith (1971) and Stoudt (1971).
The WSA data, like those of Smith (1971) and Stoudt (1971), suggested that the percent of pair occupancy of wetlands varied inversely with the number of wet basins; that is, the number of wet basins declined through the summer and the percent occupancy of the remaining wet basins increased until approximately the end of the nesting period (late June-early July). Percent pond occupancy by ducks can vary on small study areas or even on individual wetlands due to time of day (Diem and Lu 1960; Dzubin 1969; Klett and Kirsch 1976), or night (Drewien et al. 1967), observability (Sauder et al. 1971), and social behavior among species such as spatial demands (Bellrose 1979). Pond occupancy rates could seem similar over time if quantities of ponds and ducks were declining at the same time, a condition that is probably happening in the prairie pothole region today.
Smith (1971) contended that low habitat (wetlands) occupancy means little unless the circumstances surrounding it are also understood. He further stated that it may reflect an absence of potholes or an absence of ducks. Smith (1971) found that even when duck numbers were large (116 pairs per square kilometer), only 72% of the wetlands were occupied by ducks during one time.
Some recent studies have revealed that the number of wetlands (Bellrose 1979) or amount of water (Kaminski and Prince 1984) affect duck abundance, and that duck distribution varies among wetlands in relation to temporal and spatial distribution of food items (Swanson 1977; Swanson at al. 1974; Swanson et al. 1979). Thus, locally, individual pond occupancy by ducks is dependent on the total duck density and wetland conditions within a complex of different classes of wetlands varying over time. On a larger scale, pond occupancy rates in one State, Province, or region may be greater or smaller during the same season or year because of differences in water conditions in another part of the country. This may cause ducks to stop further south during wet times and overfly to the north during dry times (Smith 1970; Pospahala et al. 1974).
Findings from our study show that pond occupancy data collected by periodic surveys during several years are, at best, weak indexes of specific selection behavior of ducks to various wetland characteristics. However, we believe the data show some meaningful trends and insights into wetland habitat preference by ducks. We caution against using pond occupancy data from a single study area, such as ours, to project possible national trends in continental waterfowl populations. If pond occupancy rates are to be meaningful on a continental basis, we recommend a large coordinated effort over time and area to support the considerable concern about our declining waterfowl numbers today.
Pair Relations to Wetland Size
An inverse relation was apparent between total estimated duck densities and wetland size (Table 13; Fig. 11 and Fig. 12), indicating greater densities of ducks could be expected on wet ponds of smaller size classes. Larger and deeper wetlands were utilized more by dabbling duck pairs in dry years than in wet years. Dabbling ducks utilized seasonal wetlands more than other wetland classes; however, they also used semipermanent wetlands frequently when those were less than 2 ha. With the exception of lesser scaup, diving ducks generally used semipermanent wetlands of all size classes in greater proportion than dabbling ducks; diving ducks generally accepted wet ponds of greater size, regardless of wetland classes. This was particularly true for ruddy ducks. Differences in feeding behavior offer a possible explanation of the contrasting relations of dabbling and diving duck pair densities to wetland size. Diving ducks utilize the total area of larger wetlands by diving for food, regardless of emergent vegetation cover or water depth, whereas dabbling ducks feed more in smaller and shallower wetlands and seldom use the central or open unvegetated areas of larger wetlands. By using this difference and calculating dabbling duck pair densities in relation to a peripheral band, for example, a 35 m width outward from the shore instead of the whole surface area of wetlands, larger than 6 ha, we suspect the comparison of pair density ratios would be more meaningful. If this concept is true, it has important implications for wetland preservation and management. Figure 13 illustrates opposite relations of pair densities to two different characteristics of wetlands (surface area vs. shoreline length) and this, we believe, suggests that the surface area of wetlands being utilized influences the spatial distribution of pairs more than the effects of the spatial density of pairs (or indicators of pairs such as lone males) to one another.
In summary, wetlands of all sizes and classes were used at some time by one species of duck or another. Smaller (<0.2 ha), temporary and seasonal wetlands were important early in the season for attracting, holding, and regulating the abundance of the duck population that would later nest on the area. Medium-sized and larger wetlands and wet lands of seasonal, semipermanent, and permanent classes provided water throughout the nesting and brood-rearing seasons. Seasonal wetlands were believed to be the most important overall. Species, with the exception of canvasbacks, redheads, and ruddy ducks, required a complex of various sizes and classes of wetlands throughout a production season because time-related activities of ducks, wetland plants and other animal had phenologically different distributions. Seldom does a single wetland provide all the essentials for breeding, nesting, and brood rearing at the right time or during a whole season. Programs for wetland protection, either by lease, easement, or fee title, must consider wetland complexes if all duck species in an area are to benefit. This, of course, affects the strategies of agencies involved in wetland preservation, or waterfowl production, or both.
There was a significant positive linear relation (r² = 0.64, P < 0.001, 16 df) between the total estimated duck pairs per year and the percent of basins with water during 1- 15 May (Fig. 14). This relation was also examined for each species (Table 14). Total dabbling duck pairs were related (r²= 0.81) better with the early May percent of wet basins than diving duck pairs (r²= 0.28). Blue-winged teal were the most responsive to percent of wet ponds. We think the lack of a stronger regression between estimated pairs and percent of basins with water is partially because basin fullness and basin water depth were not taken into account.
Schroeder (1971) and Stewart and Kantrud (1974) also found significant correlations between May surface water and duck pairs in North Dakota. Many studies have shown positive correlations between the number of wet pothole basins and duck pair populations; these studies are summarized by Dzubin (1969) and Stewart and Kantrud (1974).
Anderson and Glover (1967) found a strong relation between the number of duck pairs and early spring water at Monte Vista National Wildlife Refuge, Colorado. Boyd (1981) found a stronger relation between duck numbers and a soil moisture index than with the number of ponds estimated during aerial surveys. We agree with Boyd (1981), that for ducks, the effects of sequences of dry or wet years are greater than those of a single season. To have a series of wet years, an area would need to have consistently greater than normal precipitation and high spring runoff with an average or above average ratio of precipitation to evaporation. Winter and Carr (1980) recently demonstrated that persistence in basin water conditions is also strongly related to groundwater flows and closeness of the groundwater table to terrain surface, further complicating prediction of wet year wetland conditions.
More intensive studies are needed of the connection between duck populations and water conditions in the prairie pothole region. This area can be in a drought during 1 month or season and saturated with water the next. Long-term studies of temporal variables of precipitation and soil moisture indices, similar to that of Boyd (1981), seem appropriate and timely, especially for agencies involved in future intensive regulation of the annual sporting harvest of waterfowl in North America.
Similar to Boyd's (1981) work with wetlands, Hanson et al. (1982) investigated the linear dependence between the current year's precipitation (1 August-31 July) and the previous 2 year's precipitation they found no dependence between forage yield and seasonal precipitation, suggesting that the dependence between yields must be associated with other factors such as soil water, plant vigor, and other biological factors. They did find an indication that during years of below-average yield there was a reduced probability of a good yield the next year. During the 51-year (1930-80) herbage yield study, Hanson et al. (1982) found only two occurrences of three consecutive below-average yields, both in the early 1960's, and four occurences of 3 sequential years of above-average yields, one group during the mid-1950's and the other during the late 1960's. These periods coincide with Boyd's (1981) lowest projected number of dabbling ducks in May 1962 and the highest number in May 1956. We point this out to demonstrate that two independent researchers, one working on wetland habitats (Boyd 1981) and the other on range forage yield (nesting cover; Hanson et al. 1982), had similar conclusions based on historical climatic records. The condition of the two habitats is probably related and most probably to soil mositure or precipitation, or to a precipitation-evaporation ratio, or a combination of these.
Ducks, eggs, and ducklings suffer an array of fatal factors on the breeding grounds. We believe the long-term nature of this study and the relatively large sample of nests and brood observations each year provide useful data on the relations of duck nesting, brood production, types of habitat use, and the influence of habitat on predation.
During the study, 3,832 duck nests were found in grassland and grass-shrubland habitats: 56% belonged to blue-winged teal, 18% were gadwalls, 11% mallards, 6% northern pintails, 5% northern shovelers, 2% American wigeon, 2% green-winged teal, and 1% lesser scaup (Table 15). Pair estimates and nests found were in similar proportion to species present except for blue-winged teal and lesser scaup. The percent of blue-winged teal nests was higher than the percent for pairs (56 vs. 46%) and the percent of lesser scaup nests was lower than the percent of pairs (1 vs. 5%). Because blue-winged teal have a relatively low reflushing rate (70%) with our nest searching device, but make up a large portion of the duck population, it seems probable that the discrepancy in proportions of percent composition by teal may be more related to counting and tabulation of blue-winged teal pairs than finding their nests. On the basis of some marked bird observations and years of counting experience, we propose the hypothesis that many lone drake blue-winged teal do not truly represent a pair but are in fact just bachelor drakes. With lesser scaup, we believe the relative discrepancy probably occurs in the pair estimates because not all lesser scaup seen in spring represent nesting pairs, specifically the yearling females (Trauger 1971, 1975).
The nesting period for dabbling ducks from first egg laid (Fig. 15) until the last egg hatched was from 7 April to 12 August. There was little fluctuation in the nesting chronology from year to year. Overall, nest initiations inclusive of renests as well as initial nesting attempts, ranged from 7 April to 14 July. This range is probably conservative because it is based on backdating only active nests.
Spring migration and early nest initiations at the WSA were invariably interrupted by weather. An extreme example of this occurred in 1967 when nesting species (except ruddy ducks) arrived between 23 March and 12 April, an early and short arrival period. However, during 1-3 May 1967, a blizzard occurred; winds reached 97 kph, temperatures dropped to -9° C and 15 cm of snow fell. By the morning of 2 May, most of the smaller (<0.4 ha) potholes were frozen solid enough to support a man. The northward migration of waterfowl and some early nests were disrupted by this storm. The only ducks remaining in the area congregated on the unfrozen, larger potholes and lakes. The physiological effects to and deaths of waterfowl at Woodworth and in other parts of the storm area were reported by Bry (1967, 1970), and Dane and Pearson (1971).
Mallards, northern pintails, blue-winged teal, and northern shoveler were the earliest nesting species (Fig. 15). In fact, at our location, blue-winged teal and northern shovelers should probably be considered early-nesting species. Next in order were green-winged teal, gadwalls, American wigeon, and lesser scaup. Mallards, northern pintails, and blue-winged teal continued initiating nests over longer time periods than other ducks. American wigeon and lesser scaup nest initiations occurred over a shorter period than all other ducks.
The total nesting period was usually shorter during years of drought than during years with wet, warm springs and summers. Warm spring weather with an abundance of surface water areas induced early nesting, whereas extremely hot, dry weather reduced either early or late nesting activities. Furthermore, we believe the nesting period was prolonged more often in years when rainfall occurred frequently and was distributed throughout the nesting season, compared to years of infrequent, light or intense rainfall.
Hatching Chronology and Land Management
Hatch dates for all upland-nesting species during the 16 years varied between 12 May and 12 August (Table 16). Overall mean dates for earliest and latest hatches were 1 June and 2 August, respectively. In the absence of a severe predation problem, much of the hatch curve would be perhaps a month earlier. Length of the hatching season interval varied between 43 and 88 days and averaged 63.
The distribution of potential hatch dates indicated that an average of 43% of the active nests would have been disturbed or destroyed by landuse treatments initiated on 10 July, 33% on 15 July, 22% on 20 July, 15% on 25 July, and 9% on 1 August (Table 16). Furthermore, in 9 of the 16 years, 50% or more of the clutches were still active on 10 July and 25% or more were still active on 15 July. Similarly, Duebbert and Frank (1984) reported that <10% of nests still being incubated would be affected by a mowing date of 21 July.
Vegetation at Nest Sites
The average size of a nest containing a full clutch of eggs, down, and miscellaneous cover is 30 cm in diameter with an area of 706.5 cm². Cowardin et al. (1985) stated that a hen probably selects a specific nest site within an area of home range that was selected from the entire breeding range. Why a hen selects a particular nest site is still unknown. We suspect that security has a lot to do with it. Security provided by the vegetation within a meter radius or so of the nest bowl may be critical. Cover may also be a visual relocation cue for the hen during laying and incubation, and for habitat imprint-recognition for the ducklings.
We used a 50% threshold to separate heterogenous vegetation of nest sites into cover classes. For example, if vegetation was a mixture of 40% brush, 5% forbs, and 55% grasses, it was classified as a grassy nest site. At the WSA, 54% of 3,429 clutches (Table 17) were laid in sites concealed by grasses, 18% by forbs, 14% by brush, and 1% by marshy vegetation (marshy vegetation was searched only in dry sites of wetland basins). Blue-winged teal, northern shoveler, and lesser scaup nests occurred in grassy sites in much greater proportion than other species. Mallards, gadwalls, American wigeon, and green-winged teal nested in brushy sites more readily than other species. Forbs were similarly used by all species except blue-winged teal, northern shovelers, and lesser scaup. Six of eight upland-nesting species used dry marsh sites, but mallards, American wigeon, and lesser scaup used marshes for nesting in proportionately greater frequency than did the other species.
Within stands of seeded nesting cover, 52% of 852 nest sites were in grasses, 45% in forbs, 2% in brush (mostly Rosa), and 1% in marshy vegetation. In these stands, mallards, gadwalls, American wigeon, green-winged teal, and northern pintails nested in sites dominated by alfalfa. Blue-winged teal, northern shovelers, and lesser scaup nested predominantly in grassy sites. Within stands of seeded cover, 70% of gadwall nests were in forb sites whereas 70% of the blue-winged teal nests were in grassy sites.
Within stands of native prairie vegetation, 67% of 1,702 nest sites were in grasses, 22% in brush, 9% in forbs, and 2% in marshy vegetation. In these stands, mallards, gadwalls, and American wigeon nested in sites dominated by brush, whereas, the other species nested mostly in grassy sites. Within native prairie, approximately 88% of northern shoveler nests, 85% of lesser scaup nests, 81% of blue-winged teal nests, and 60% of northern pintail nests were in grassy sites. In these stands, 55% of both gadwall and American wigeon nests, and 57% of the mallard nests were in brushy sites.
Of 852 nests in stands of seeded nesting cover, most were associated with the following plant species: alfalfa, 36%; quackgrass (Agropyron repens), 11%; smooth brome, 11%; Kentucky bluegrass, 8%; intermediate wheatgrass, 4%;, and sweet clover, 3%. With the exception of lesser scaup, nests of all species were more frequently associated with alfalfa than with any other single vegetative species in stands of seeded nesting cover.
Of 1,702 nests found in stands of native prairie, most were associated with the following plant species: Kentucky bluegrass, 29%; wolfberry, 20%; smooth brome, 8%; quackgrass and needlegrass (Stipa spp.), 4% each; upland sedges (Carex spp.), 3%; and goldenrod (Solidago spp.), 2%. Of the nests in native prairie stands, 41% occurred in quackgrass, smooth brome, or Kentucky bluegrass, all of which are introduced cool-season invaders. Kentucky bluegrass was the dominant species at 40% of the blue-winged teal nests and 36% of the northern shoveler nests in native prairie stands. Wolfberry was the dominant species at 50% of the American wigeon nests, 47% of the gadwall nests, 41% of the green-winged teal nests, 36% of the mallard nests, and 21% of the northern pintail nests. In wetlands, Carex spp. were the dominant plants at 35% of the lesser scaup nests.
Nest Site Cover Quality, Physiognomy, and Concealment
On the basis of subjective cover quality ratings, 58% of 3,402 nests (Table 18) were in fair nesting cover, 28% in good nesting cover, 7% in excellent nesting cover, and 7% in poor nesting cover. More than 55% of the mallard, gadwall, and American wigeon nests were in good-to-excellent quality cover whereas 50% or more of the blue-winged teal, northern shoveler, and northern pintail nests were found in poor-to-fair quality cover. Green-winged teal and lesser scaup nests were found in nearly equal proportions in good-to-excellent and poor-to-fair quality cover.
Vegetative cover in the immediate area surrounding nest sites was grouped into eight physiognomic categories (Fig. 16) according to shape of the vegetation actually concealing the clutches. Mallards, gadwalls, American wigeon, northern pintails, lesser scaup, and to a lesser extent, northern shovelers and blue-winged teal selected more nest sites with erect cover and a open or closed canopy than in other sites (Table 19). Brush, forbs, and some of the taller cool-season grasses provided most of the erect cover. Blue-winged teal had more nests in grassy cover, which provided tent-shaped canopies, than the other categories, but they also commonly nested in sites with erect vegetation. Green-winged teal nesting sites were less distinctive in relation to other species, but they nested more in cover that was tented, and erect and under dominant plants, for example, alfalfa.
Nest sites were grouped into three categories according to the amount of vegetation that concealed the top of each nest at the time of discovery: no top concealment, between none and one-half of the nest concealed, and between half and all of the nest concealed. Vegetative cover surrounding the sides of nest sites was grouped into five categories: no concealment, one-fourth concealment, one-half concealment, three-fourths concealment, and complete nest concealment. Except for northern pintails, all species selected sites with top concealment of one-half or greater (Table 20) rather than sites with no top concealment. Northern pintails nested more readily in intensively cultivated fields, where there are more open nesting sites, than did other species (Higgins 1977). These fields provided only sparse, open cover for nesting sites. All species selected nest sites with one-half or greater side concealment (Table 21).
Tradition of Nest Site Selection
With probable over simplification, we submit that the tradition of nest site selection starts with the successful hatching of a clutch. Newly hatched ducklings then have two subjects on which to imprint: the hen (true imprinting) and the vegetative cover (imprint-recognition) immediately surrounding the nest. Brooding may occur for a few hours up to 24 h at the nest. Thus, there is ample time for ducklings to associate with the nest site, after which the brood is usually guided by the hen to a nearby wetland and, subsequently, ducklings are exposed to an ever increasing area of habitat. Ducklings that reach flight stage have usually been exposed to more than one wetland and a rather large area. Movements of mallard and gadwall broods as far as 5.5 km from the nest site were noted by Lokemoen et al. (1984). Beginning with the first flight, ducklings soon become exposed to a greater area on the breeding grounds until they migrate in fall. The following spring, surviving hens attract mates and migrate back to or near where they hatched or to a previously successful nest site, even though extreme drought or other perturbations may have drastically changed the character of these sites.
Evidence of the importance of successful nest sites is borne out in the rates of hens homing back to a particular habitat type or even to the same nest sites (Doty and Lee 1974). During early phases of this study, we became aware that some nest bowls were used in more than 1 year, but we did not know if the reuse of these sites was by the same or different hens. In one instance, the senior author found blue-winged teal nesting in the same nest bowl during 3 consecutive years. The clutch hatched during years 1 and 2 but did not hatch during year 3 and no further nesting occurred at this nest bowl in following years. Reuse of some nest bowls in natural sites has also been reported by Weller et al. (1969), Schamel (1974), and Duebbert at al. (1983), and in nest baskets by Doty and Lee (1974). Duebbert et al. (1983) found that 73% of 252 mallard and 49% of 168 gadwall nests on an island were placed in previously used nest bowls and in six instances, bowls of previously successful nests were reused that same season by later-nesting hens, some of different species. Doty and Lee (1974) found 46% of 113 marked mallard hens homed back to nest baskets and successful hens homed at a significantly (P < 0.01) higher rate than unsuccessful hens. In their study, 7 adult hens from 140 web-tagged female ducklings homed and nested in baskets in the vicinity where they were hatched. In brief, adult and juvenile female ducks have an innate ability to home back (Hochbaum 1955) and select a specific nesting site (±30 cm in diameter). The rate of homing is apparently directly related to nest success (whether in natural sites or in human-made baskets). Therefore, recognition or imprint-recognition of nesting site features or any manipulations of nesting habitat that enhance the success of nesting waterfowl are important to waterfowl management.
Stoudt (1971) demonstrated that unless nest searching was conducted periodically throughout the nesting season, average clutch size figures would be in error because clutches laid later in the season averaged smaller than earlier ones. This was also true at the WSA where average clutch size gradually decreased for all species as the season progressed (Fig. 17). Average seasonal clutch size decreases within species were lesser scaup, 42%; northern pintail, 41%; gadwall, 36%; American wigeon, 35%; blue-winged teal, 32%; northern shoveler, 30%, mallard, 25%; and green-winged teal, 19%. The total average completed clutch size for all species was 29% smaller at the end of the nesting season than it was at the beginning, a substantial reduction in potential duckling production. Nesting later in the season was mainly a contribution from immature or late nesters and renesters, which results from predation, weather, habitat conditions, or combinations of these. Slight variation in mean sizes of hatched clutches was found between nests in native prairie, seeded cover, and cropland habitats and inconsistently among species (Table 22). However, average sizes of hatched clutches for all species inclusive were apparently smaller in burned tracts than in either grazed or idled tracts of native prairie or seeded cover habitats (Table 23). This was expected because nearly all burning was conducted during May and June, which usually eliminated all nesting cover, and with few exceptions, destroyed all active nests. Thus, nearly all successful hatches that occurred during the same year of spring fires would have been late initiations or renests, both of which produce smaller clutch sizes.
Samples of hatched clutch sizes in cropland habitat were small, but the data available indicated a mix of early- and late-laid clutches in standing grain crops and only early-laid clutches in summer fallow (Table 23). Standing crops on the area included both spring- and fall-seeded grain crops. Fall-seeded crops provided early nesting cover whereas spring-seeded crops usually provided mid- to late-season nesting cover. Summer fallow was left uncropped and was tilled several times during the nesting season beginning in late May to early June. Therefore, early-laid clutches had a better chance to hatch.
Mammalian predators caused 87% of 2,395 nest losses during the study (Table 24), less than 1% was attributed to weather. Most unknown predations probably were caused by either ground squirrels or red fox. Clues to nest predation by these two species are difficult to assess, especially when inspections of nests occur at long intervals.
Destruction of nests by humans and machines was low on this area because only about 12% of the land was annually hayed or cultivated during an average year on nearby study areas that were 84% or more annually tilled, Higgins (1977) found 19% of the nest destruction was caused by men and machines, and on the cropland portion farm machinery destroyed 93% of active nests during tillage operations.
We found it difficult to assign nest loss to a specific predator. Of 2,374 nest failures, the specific predator causing a failure was attributed at only 1,595 (67%) of the destroyed nests. Most (64%) predation nest failures were caused by red fox. Mink, weasels, and cattle trampling caused less than 1% (Table 25).
Predation by badgers, raccoons, and gulls was fairly consistent throughout the study whereas predation by red fox, skunks, and ground squirrels varied greatly among years. Although we had no definite means of enumerating these species, the frequency that they were seen and annually trapped by the public, suggested that the amount of predation by these three species was largely related to their densities.
Overall Duck Nest Densities, Success, and Production
The density of 3,821 duck nests on 8,464 ha of searched uplands averaged 45km². This calculated to an average annual density of 0.07 nest per hectare or 2.2 ha per nest during 1966-81. Nest success for 3,517 nests with complete histories averaged 16.3% Mayfield. In all, 11,430 duck eggs hatched, yielding an annual average of 1 duckling per 0.7 ha or 55.6 ducklings per 40.5 ha of upland cover.
Duck nest success did not differ among species for all years combined (X² = 13.1, 7 df, P = 0.70, Table 26), but it did differ among years for all species in combination (X² = 122.0, 15 df, P = 0.001; Table 27).
Effects of Cover Height
Significantly greater duck nest success occurred as cover increased in height, or conversely, predation was significantly reduced by taller vegetative cover at nest sites (Kirsch et al. 1978). Cover height effect, even though significant overall, was not constant among duck species. Our data showed that as cover height increased, the daily mortality of nests decreased for mallards, gadwalls, and northern shovelers; remained the same for blue winged teal and green-winged teal; and increased for northern pintails (Fig. 18). Possibly the high proportion of blue-winged teal among the duck population during recent years is related, in part, to the fact that their nest success is independent of cover height. Meltofte (1978), Duebbert and Lokemoen (1980), and Lokemoen et al. (1984) reported ducks used nest sites with taller and denser cover even when predation rates were low. Our results are in general agreement with those of Heiser (1971); Dwernychuk and Boag (1972); Kirsch et al. (1978); Livezey (1981a, 1981b); and Hines and Mitchell (1983).
Using part of our data (1974-76), Kirsch et al. (1978) showed a direct relation between nest densities, as well as nest success for five species of upland-nesting ducks, and readings of visual obstruction along transects in cover fields before new spring growth affected readings in the residual cover. Height and density of vegetative cover at the nest sites, throughout the season, and of total residual cover in early spring before new growth, are important to duck nest success and nest densities.
Effects of Habitat Type
Waterfowl populations fluctuate in the Missouri Coteau as habitats change due to annual moisture regimes. The amount of precipitation that comes before and during the growing season influences growth of upland vegetation and water levels in dry potholes.
There were four broad types of upland-nesting habitats on the area: marshy vegetation in dry wetlands, annually tilled croplands including summer fallow, native mixed-grass prairie, and former croplands currently sown to mixtures of cool-season grasses and legumes. All upland habitat classes were used for nesting by all species upland-nesting ducks at some time during the study. The small sample of nests (33) in dry wetlands, averaging only about three per year and consisting primarily of mallards, gadwalls, and lesser scaup, curtailed further comparisons among years and species for this habitat. Most duck nesting is completed by the time most basins dry up; which probably accounts for the infrequent occurrence of duck nesting in dry wetland basins. Thus, we include comparisons only of crop lands, native prairie, and seeded grassland habitats in the remainder of this report.
Although duck nesting success averaged slightly higher in seeded grasslands (18%) than in native prairies (16%) or croplands (13%, Table 28), nest success did not differ (X² = 0.02, 1 df, P > 0.90) significantly between seeded and native prairie habitats for all species and years combined, but it differed significantly among years (X²= 134.3, P < 0.001); interactions between years and habitats were not significant (Table 28). Nest success did not differ significantly among field size classes ( X² = 4.6, 3 df, P > 0.10; Table 29).
Ducks nested in greater densities in seeded grasslands (33 nests per 40.5 ha) than in native mixed-grass prairie (16 nests per 40.5 ha), or in annually tilled croplands (about 3 nests per 40.5 ha, Table 30). For all species and habitats combined, the average annual nest density during 1966-81 was 18.2 nests per 40.5 ha, or 45 nests per square kilometer. All species except lesser scaup showed the highest nest densities in seeded grasslands (Table 31). Lesser scaup had slightly greater nest densities in native prairie habitat, particularly, near wetland margins, than in seeded grassland (0.3 vs. 0.2 nests per 40.5 ha). We did not find any nests of American wigeon or green-winged teal in annually tilled croplands. Nest densities were much greater in fields of smaller size (Table 29) for all habitat types.
Among habitat types, annual duckling production was greater in seeded grasslands (100 ducklings per 40.5 ha) than in native prairie (35 ducklings per 40.5 ha), or in annually tilled croplands (7 ducklings per 40.5 ha; Table 30).
We had complete histories on 2,081 nests found in native mixed-grass prairie during 16 years of searches. Plant communities were numerous within this habitat type and seldom distinct from one another. For example, the potential natural vegetation community for the area is Agropyron-Stipa-Andropogon (Küchler 1964), but it was often invaded by Kentucky bluegrass, smooth bromegrass, buckbrush, silverberry, and quackgrass or combinations of these.
Dix and Smeins (1967) classified the native mixed-grass prairie into high prairie, mid-prairie, and low prairie. Relative to these classes, we found few duck nests in high prairie, low numbers in low prairie, and high numbers in mid-prairie. In general, native mixed-grass prairie provided the greatest diversity of nesting sites among the four nesting habitats.
Higgins (1981) reported that 37,864 ha of native prairie occurred on 1,746 federal waterfowl production areas in five states in the glaciated prairie pothole region. Based on a density of 34.5 ducklings per 40.5 ha from our study, the U.S. Fish and Wild life Service could hatch 32,239 ducklings per year from these areas; however, the potential is much higher with lower predation effects.
The complete histories on 1,366 nests found in seeded grasslands were used as data. The nesting cover among fields in this type was similar except when sweetclover became dominant, which was usually during the second growing season (Higgins and Barker 1982). On average, seeded grasslands were the most productive nesting habitat on the WSA. In other studies, seeded grasslands also supported high densities of nesting ducks relative to other cover types (Duebbert 1969; Miller 1971; Oetting and Cassel 1971; Duebbert and Kantrud 1974; Nelson and Duebbert 1974; Duebbert and Lokemoen 1976; Cowardin and Johnson 1979; Livezey 1981a, 1981b).
Most nesting in cropland on the WSA was in standing, small-grain crops rather than stubble or summer fallow. Cowardin et al. (1985) analyzed mallard preferences for nesting among available habitats on a nearby study area and found mallards nested in croplands less than expected relative to its availability. Our data agree with their results for all upland-nesting species.
Effects of Land-use Treatments
Land-use treatments apparently affected nest success more than the habitat types. Among land-use treatments within native prairie, duck nesting success averaged higher in burned fields (18.3%) than grazed pastures (13.0%) or idle fields (12.9%), but the difference was not significant (X2 = 2.35, 2 df, P = 0.309) nor consistent among species (Table 32). Another analysis comparing nest success among land-use treatments within years yielded a marginally significant effect. Except for blue-winged teal and gadwalls, ducks experienced lower nest success in grazed pastures. In grazed pastures, gadwalls nested predominantly in wolfberry patches that generally offer taller and denser cover than other pasture sites. American wigeon, green-winged teal, blue-winged teal, northern shovelers, and lesser scaup had the highest nest success in burned native prairie, but mallards had the best success in idle native prairie (Table 32). Northern pintails had nearly equal nest success in burned or idle native prairie and were least successful in grazed pastures.
Among land-use treatments of seeded grasslands, nest success was higher in burned (22.7%) than idle (16.7%) or grazed (9.3%) fields. Among duck species and treatments within seeded grasslands, nest success was highest in burned fields for all species except mallards, northern shovelers, and northern pintails; mallard nest success in burned fields was nearly equal that of idle fields.
We suggest that duck nest success and densities in seeded grasslands could be enhanced by treatment with fire at 3-4 year intervals. We burned seeded grasslands during March-June; however, we achieved best plant regrowth with burns in March and April, and weather permitting, also in May for the cool-season grass-alfalfa mixtures in our plantings. Early spring burns allowed total field restoration and regrowth during the burn year and a longer fire-free nesting and renesting interval after the fire event in the burn year.
Among land-use treatments in annually tilled croplands, nest success was greater in standing grain crops (20.7%) than in summer fallow (12.5%), mulched stubble (3.7%), or standing stubble (2.3%). Except for blue-winged teal, nest samples among species and treatments were too small for meaningful comparisons of nest success rates among species. Blue-winged teal had their highest nest success in growing grain.
In seeded grasslands, nest densities were greater in idle cover (46.3 nests per 40.5 ha) than burned fields (22.8 nests per 40.5 ha) or grazed fields (17.8 nests per 40.5 ha; Table 33). Nest densities were greater in idle cover than in burned or grazed cover for all species except lesser scaup.
Within native prairie, nest densities were greater in idle fields (26.7 nests per 40.5 ha) than grazed pastures (15.7 nests per 40.5 ha) or burned fields (14.7 nests per 40.5 ha).
In annually tilled cropland, nest densities were greater in summer fallow (5.2 nests per 40.5 ha) and standing green crops (4.2 nests per 40.5 ha) than mulched (2.7 nests per 40.5 ha) or standing stubble (1.0 nests per 40.5 ha). These density results are in general agreement with earlier results reported by Higgins (1977).
Within native prairie, duckling production was greater in idle fields (70.7 ducklings per 40.5 ha) than burned fields (46.8 ducklings per 40.5 ha) or grazed pastures (40.7 ducklings per 40.5 ha; Table 34). Duckling production averaged higher in idle fields of native prairie than grazed or burned fields for all duck species except American wigeon.
In seeded grasslands, duckling production was higher in idle cover (141.5 ducklings per 40.5 ha) than burned (83.9 ducklings per 40.5 ha) or grazed fields (22.0 ducklings per 40.5 ha).
Duckling production averaged higher in summer fallow fields (20.8 ducklings per 40.5 ha) than in growing grain crops (8.9 ducklings per 40.5 ha) or standing stubble (0.9 duckling per 40.5 ha). No ducklings hatched from fields of stubble mulch. The high number of cropland fields with no nests, or in which no ducklings hatched, prevented comparing duckling production among species by land-use treatments within croplands.
The eight species of upland-nesting ducks used fields of native mixed-grass prairie regardless of whether the fields were burned, grazed, or idled. Gadwalls, American wigeon, green-winged teal, and blue-winged teal had average nest success rates greater than or equal to 15% in burned treatments, whereas only gadwalls and green-winged teal had such rates in grazed pastures. In idled fields, gadwalls, American wigeon, green-winged teal, blue-winged teal, northern shovelers, northern pintails, and lesser scaup all had average nest success rates greater than or equal to 15%.
Few studies have been conducted in these same habitats on the effects of fire on habitats and subsequent production of nesting ducks. Fritzell (1975) reported lower densities of duck nests on burned cover (0.5 nest per hectare) than on unburned cover (1.2 nest per hectare), but higher apparent nest success in burned cover (38.5%) than in unburned cover (15.0%) during a 2-year study in southern Manitoba. Kirsch and Kruse (1973) similarly reported a higher apparent success for duck nests on burned grasslands during the second growing season after a fire (52%) than on undisturbed (33%) or grazed areas (23%). Their results were based on the analysis of the 1966-71 portion of this data. Nest success rates observed by them on burned areas were higher than the overall apparent nest success rates on burned areas for the years 1966-81 (52 vs. 34%).
All species of upland-nesting ducks used seeded grassland habitat regardless of land-use treatment, except for American wigeon and green winged teal, two species that were sparsely distributed in the WSA. Mallards, gadwalls, northern pintails, American wigeon, blue-winged teal, and northern shovelers had average nest success rates greater than or equal to 15% in nonused fields of seeded grassland, whereas in burned fields only mallards, gadwalls, American wigeon, and blue-winged teal had average rates greater than or equal to 15%, and in grazed areas none had average rates greater than or equal to 15%.
Several studies have been conducted on duck nesting in seeded grasslands of the types we studied, but few have dealt with the effects of land-use treatments on duck production in seeded grasslands. Livezey (1981a) compared duck nesting among land-use treatments in seeded grasslands at Horicon NWR in Wisconsin and reported that mowing and haying significantly reduced duck nest densities until the second year after mowing, when fields seemed fully restored to pretreatment vegetation structure. He also found duck nest densities and success to be similar among hayed fields and long-retired fields (5+ years). Our results relative to treatments are in agreement with Livezey (1981a) in that the effects reduced duck nesting during the treatment year and the subsequent year, particularly with burning treatments.
Sample sizes for individual duck species were small among treatment divisions within annually tilled cropland. However, there was one obvious difference among ducks nesting in the cropland treatments. Fields of standing grain were used by all species of upland-nesting ducks except American wigeon and green-winged teal, and were comparatively more attractive and productive to ducks than fields of stubble or summer fallow. Winter rye provided earlier standing grain cover than did spring-seeded wheat, barley, or oats. Winter wheat was not available during this study but became available as a potentially good nesting cover crop during the 1980's. Gadwalls, blue-winged teal, and lesser scaup nested with better success in growing grain (55.5, 33.8, and 25.0%, respectively) than did northern pintails, northern shovelers, or mallards (5.5, 3.2, and 2.5%, respectively). This was probably because the former three species are mid-to-late nesters and growing grain is taller and denser later in the year.
Generally, our results relative to annually tilled croplands were in agreement with results reported by other duck studies (Milonski 1958; Moyle 1964; Evans and Wolfe 1967; Smith 1971; Stoudt 1971; Higgins and Kantrud 1973; Duebbert and Kantrud 1974; Higgins 1977; Cowan 1982). Further, we think annually tilled croplands will give greater duck production when a substantial proportion of this habitat type within the prairie pothole region is managed for winter wheat and winter rye, or is under minimum or no-till cultivation systems. A study by Cowan (1982) comparing nest success in zero tillage and conventional tillage grain cropping supports our belief.
Brood Census Periods
Two brood counts were made each year except in 1978 when only one was made. Usually the initial count was begun when broods were first seen in Age Class II-A (Gollop and Marshall 1954), between 25 June and 15 July, and the second count was made between 25 July and 15 August. We think that estimates of the overall brood production are best based on two field counts because species have different hatching dates. Incidental sightings of Class I age broods between the July and August counts, and after the August count, can be used either to augment the total brood count or to eliminate duplication of broods already counted during earlier regular counts, using cross matching broods by age, species identification, brood size, and location. This method of brood censusing and tallying takes considerable field effort by several people each year, possibly 40 work hours per 200 wetlands for two censuses and 2 work hours per 200 wetlands to tally and remove duplicate brood sightings between counts and between counts and incidental sightings. Overall, we believe this method gave us a fairly reliable estimate of the total duck production.
Number of Broods
The number of duck broods estimated on the study area each year varied from 47 to 301 during 1965-81 and averaged 148 (Table 35). The two highest counts of broods occurred in 1966 and 1969. These were years of below average total precipitation and above average winter snowfall. From 2 to 5 March 1966, an estimated 76 cm or more of snow accumulated during a severe blizzard and at least 122 cm of snow accumulated during the winter of 1968-69.
The lowest number of broods was censused in 1973, a year of near-average total precipitation but below-average winter snowfall. As with pairs, there was no direct correlation between the number of broods and the total amount of annual precipitation.
Brood counts are only an index to production because they are affected by the same wetland and climatic conditions that affect pair counts. In addition, accuracy of counts is influenced by behavioral characteristics of hens seeking shelter and protection for broods and by the movements of broods between wetlands. We assumed that with brood movements on a block study area, ingress equaled egress.
Pond Occupancy by Broods, Species Composition of Broods, and Brood Densities
During 1965-75, the percent of wetland basins occupied by broods of all species varied from 11-36% and averaged 17% for July censuses. In August, the percent varied from 14-50% and averaged 20%. Generally, percent occupancy of wet basins by broods increased as the number of wet basins decreased during late summer. Broods, more than pairs, tended to congregate on certain wetlands but avoided others. This tendency was apparent during consecutive years.
During 1965 through 1981, dabbling duck broods averaged 86% and diving ducks 14% of the brood populations, the same proportions as with pair populations (Table 36).
Blue-winged teal were the most abundant species in all years, averaging 46% of the pair population and 50% of total broods in July plus August counts.
Brood densities per 2.6 km², from 1965 through 1981, varied from 10 to 63, and averaged 31 (0.02 broods per hectare), in an area of 19% wetlands and 81% uplands. Mean brood densities per 2.6 km² per species for this same period were blue-winged teal, 16; gadwalls, 4; mallards, 3; ruddy ducks, 2; American wigeon, canvasbacks, green-winged teal, lesser scaup, northern pintails, northern shovelers, redheads, 1 each; and ring-necked ducks, lesser than or equal to 1. Highest brood densities by species were blue-winged teal, 35; gadwalls, 8; mallards and northern pintails, 5; northern shovelers and ruddy ducks, 4; canvasbacks, 3; American wigeon, green-winged teal, lesser scaup and redheads, 2; and ring-necked ducks, lesser than or equal to 1.
The relation of brood densities to wetland size was generally different between diving duck broods and dabbling duck broods (Fig. 19). Seemingly, diving duck broods show a direct relation between densities and increasing wetland size, but for both diving ducks and dabbling ducks, the density of broods dropped off sharply for wetlands > 10 ha in size.
Dabbling duck broods generally used wetlands of all sizes. Exceptions were wetlands lesser than or equal to 0.04 ha, where broods of northern shovelers and blue-winged teal only were seen (Table 37). In contrast, with the exception of ruddy ducks, diving duck broods seldom used wetlands < 0.4 ha in size.
During the 11-year sampling period, no broods were counted on Class I, IIt, T-2, T-3, or T-4 wetlands and, with the exception of blue-winged teal, no other broods were counted on Class It or II wetlands (Table 38). Class III and IV wetlands and wetlands with dams were generally used by all species of broods similar to their occurrence in the total brood population. Class IIIt wetlands were used by broods of dabbling but not diving ducks, and only mallard broods were counted on wetland dugouts. Class V wetlands were generally used by all species of duck broods but were not used in proportion to their availability. Class V wetlands were approximately 20 ha in size.
Brood-to-wetland Relations and Brood Sizes
There was a significant relation (r² = 0.41, P < 0.01, 16 df) between the estimated total of all species of duck broods and percent of basins with water during 1-15 May (Fig. 20). The relation between estimated numbers of broods per year and percent of basins with water during 1-15 May were also examined for each species (Table 39). As with pairs, dabbling duck broods related better with early May water (r² = 0.62) than did diving duck broods (r² = 0.28). Canvasbacks, American wigeon, and lesser scaup showed a negative relation to May water counts; whereas, northern pintails, blue-winged teal, northern shovelers, and ruddy ducks showed a positive relation.
Average brood sizes were based on 1,425 sightings during July and August censuses over 17 years (Table 40). These data include only those sightings where all ducklings were believed seen. Overall, we found the August brood counts yielded about the same number of broods as the July counts. Mean brood size ranged from 7.5 for lesser scaup to 4.4 for northern pintails and averaged 6.4 for all species. Dabbling duck broods averaged one duckling per brood larger than diving ducks (6.6 vs. 5.6). This difference was largely accounted for by the smaller clutch and brood sizes of ruddy ducks.
Without attrition or with a constant brood survival, mean brood sizes would increase in number relative to increasing age of broods because early clutches are generally larger than later clutches. For July and August counts combined, mean brood size generally declined from Class I (6.7) to Class III (5.6) ducklings when all species were grouped. This change suggested duckling attrition occurred within broods. Small samples prevented a more intensive analysis among the independent age classes among species. However, when broods were grouped by three major age classes (I, II, III) and July and August counts were combined, brood sizes averaged consistently higher for July counts than for August counts. These data demonstrate an important bias (Cowardin and Johnson 1979; M. C. Hammond, Northern Prairie Wildlife Research Center, unpublished report) in reported brood sizes if broods are not sampled throughout the nesting and brood-rearing seasons. This bias can be further confounded by differential nesting chronology among species. For example, samples of Class III broods existed for only 3 of 12 species in July counts and 10 of 12 species in August counts; 3 of 12 species were not represented among Class II broods in July counts.
Declines in Duck Production
The number of ducklings to reach flight age also was affected by the decrease in average clutch size between early- and late-laid clutches (11.0 vs. 7.8; Fig. 17), which again was reflected by the decrease in average brood size between July and August counts (7.2 vs. 5.7; Table 40). This subtle and often unnoticed decline in clutch sizes during renesting efforts throughout the nesting season was indirectly caused by high rates of predation of early clutches. Unlike other types of production loss, management strategies can be employed to reduce this loss; specifically, habitats can be managed to reduce predator effectiveness; predator numbers or access to nests can be reduced to achieve higher rates of success for early clutches. Our data suggest that average annual recruitment can be enhanced by about 30% (Table 41) if early nesting attempts are successful as compared to later, renesting efforts. Obviously, any attempt to effectively manage for greater nest success will necessitate some means of reduction in overall nest predation or destruction.
Estimates of Duck Loss
We were able to assess duckling attrition between nesting and fledging periods with our methods of counting pairs and broods and evaluating nest success. Because of the apparent bias between mean brood sizes during early July and late August brood counts (Table 40), duckling loss was estimated as the difference between average clutch sizes and average brood sizes for July and August counts combined. Duckling loss between the mean clutch size and average brood size ranged from 44% for northern pintails to 23% for American wigeon and averaged 29% for all species (Table 41).
An average 24% of the duckling mortality (2.6 young per brood) was estimated to occur between hatching and age Class Ib, whereas only 4% (0.4 young per brood) occurred between age Class Ib and the average total brood size. Thus, approximately 85% of duckling mortality, an average loss of 2.2 ducklings per brood, occurred between hatching and age Class Ib, or approximately 2 weeks of age, and 15% between Class Ib and mean brood size. This agrees with Ball et al. (1975), Reed (1975), Cowardin and Johnson (1979), Talent et al. (1983), and Duebbert and Frank (1984). The latter reported a difference of about 2.5 ducklings between average clutch size of nests and average Class I brood size. Talent et al. (1983) found that 85% of the mortality, of 13 mallard broods in which all young were lost, occurred within the first 2 weeks after hatching and 52% mortality for entire broods during a 2-year study. Ball et al. (1975) found that most total brood loss by mallards also occurred during the first 2 weeks after hatching in Minnesota. Cowardin et al. (1985) reported that 26% of radio-marked mallard hens lost broods between hatching and fledging. Our data also indicated that similar duckling loss occurred among all species, except possibly northern pintails for which we had only a small sample of brood observations.
Duckling loss between Class I and III broods 16%, or 1.1 young per brood, during the 17-year study (Table 42). Attrition estimates in Table 43 do not include losses of entire broods between hatching and counting, thereby biasing these duckling loss rates downward.
We did not specifically attempt to assess duckling mortality between their nest and first pond visited; however, instances of Class Ia duckling deaths were known to have occurred on the study area or nearby. We think the amount of duckling loss between the nest and first pond visit, usually during the first 24 h after hatching, is probably small, relative to the overall duckling mortality that occurs between hatching and fledging. Furthermore, even if there was large loss, it would be extremely difficult to apply any practical management strategy, except possibly intensive predator control, to ameliorate this duckling mortality.
Duckling loss between hatching and fledging is hard to assess and, in our opinion, is one of the greater problems of waterfowl management in need of immediate research. At present, not enough information exists to tell managers what proportions of duckling loss are due to habitat conditions, disease, weather, predation, or starvation.
A hen was termed successful when at least one duckling was fledged from one or more nesting attempts. We used the brood to pair ratio as a measure of hen success and assumed that brood ingress and egress were equal, and that all broods and pairs were seen. During the 17 years, an average of 30% of the hens were successful in hatching a clutch, getting the ducklings to water and having them survive long enough to be counted (Table 43). Our estimate of 30% average hen success agree with the estimates of 32% hen success from long term studies in southern Canada (Smith 1971; Stoudt 1971). At Woodworth, only 7 of the 17 years had hen success >30% and there were only two instances (1966-67,1976-77) of consecutive year with hen success >30%. Among the 12 duck species at the WSA, green-winged teal, blue-winged teal, American wigeon, canvasbacks, ring-necked ducks, and ruddy ducks had hen success estimates averaging >30%. The poorest year for hen success was 1973 (15%) and the best was 1971 (64%).
The brood to pair ratio is only an index to success of the nesting population. Mortality among hens during the nesting and brood-rearing seasons, and total loss of broods before they are counted, are factors that strongly affect hen success estimates. Although we did not attempt to assess adult duck mortality, a concurrent study in the same region estimated an average annual loss of 13.5% of hens and 4.5% of drakes to red foxes (Sergeant at al. 1984). Furthermore, the capability of observers to count and separate the nesting population from migrants and to observe all broods are also factors that affect estimates of hen success.
Cowardin and Johnson (1979) estimated mallard hen success (H) from an estimate of nest success rates (P) using the formula H = Pe(1-P). Application of their formula to our nest success rates (P = 0.16) yielded a hen success estimate of 32%, which agrees with our hen success estimate of 30% based on brood to pair ratios.
Recruitment rates were estimated from direct counts of the breeding population, number of broods produced, and brood size. Our estimates of average recruitment included 0.93 for ring-necked ducks down to 0.40 for ruddy ducks and averaged 0.59 for 12 species (Table 44). Cowardin at al. (1985) indicated that mallard populations in North Dakota would remain stable with a nest success rate of 15.2%, a hen success rate of 0.31, and a recruitment rate (R) of 0.53; where R = HzB/2 (Cowardin and Johnson 1979). The mallard recruitment rate at the WSA averaged 0.50, suggesting a nearly stable to slowly declining population. The recruitment rate for mallards at the WSA was about midway between the highest and lowest estimated rates for the 12 species of ducks nesting on the area.
The number of fledged ducklings produced per year averaged 579 (Table 44), from an average duck population of 492 pairs, approximately 1.18 fledglings per pair per year.