---

---

Introduction

Size of the mallard breeding population in North America is correlated positively with production rate the previous year (Reynolds and Sauer 1991). Using data from 1955 to 1970, Johnson and Shaffer (1987) presented a model in which May ponds in the PPR predicted annual changes in continental mallard breeding population size. The authors found that the mean correlation coefficient between mallard population size and May pond counts during the 2 previous years diminished from 0.47 in 1955-70 to 0.27 in 1971-85. Efforts to understand the changing relation between mallard recruitment and number of May ponds in the PPR have been hampered by limited information on wetland habitat preferences, wetland needs of females during reproduction, and effects of land use on settling patterns of pairs. Such information is needed to assess the significance of natural and human-induced changes in prairie pothole landscapes on mallard reproduction and to help focus wetland and upland conservation and management on areas that will contribute most toward increasing recruitment rates.

We hypothesized that breeding females would prefer temporary and seasonal ponds over semipermanent ponds and lakes because macroinvertebrates (the primary food of egg-laying females) are abundant and more available in the shallower waters of the former wetland types (Swanson et al. 1985). Availability of macroinvertebrates is important to mallard recruitment because protein requirements for egg formation are met largely from exogenous sources (Krapu 1981). Shallow basins in cropland in North Dakota usually are tilled and planted to crops when dry in spring (Stewart and Kantrud 1973) causing a decline in organic matter (Martin and Hartman 1987) and a decrease in aquatic macroinvertebrate density (Swanson et al. 1974). Thus, we hypothesized that number of mallard pairs that settle in prairie pothole habitat would decline as percent of landscape in cropland increased.

To test our hypotheses, we determined female wetland use by basin class, tested for wetland preferences during prenesting and egg production, and tested whether distribution of breeding pairs varied with percent of landscape in cropland. We also evaluated effects of pond density, reproductive stage, and year on number of ponds used by individual females during prenesting and egg production.

We are grateful to the Central Flyway Council for financial support; the Mississippi Flyway Council for personnel to conduct breeding pair surveys; and administrative, maintenance, technical staff at the Northern Prairie Science Center (NPSC) for assistance with logistics, fieldwork, and data processing. In particular, we thank S. D. Haseltine, D. H. Johnson, and R. C. Stendell, for assistance in expediting our work. We thank J. A. Beiser, R. R. Cox, W. E. Newton, and T. L. Shaffer for advice on study design and statistical analyses. We appreciate cooperation of landowners and operators for allowing access to their properties. We are grateful to technicians W. Jensen, M. Johnson, D. Licht, W. Mulvaney, R. Speer, and L. Stevenson, who supervised field collection of data; to M. Albrecht, T. Albro, M. Boyer, D. Brandt, J. Edwards, V. Greer, G. Falk, G. Fisher, S. Friedhof, G. Garhartz, J. Grant, P. Hess, J. Huseby, K. Harris, S. Hart, R. Jensen, T. Jones, J. Koloszar, C. Lapp, M. Larson, C. Louisias, J. Lutes, S. Manley, G. McDaniel, K. McDowell, A. Morkill, M. Nelson, B. Norling, W. Norling, L. Peskin, P. Raferty, J. Russell, D. Sausville, Jr., B. Sauve, J. Turner, and K. Wood for monitoring radioequipped female mallards; and to individuals who participated in counts of breeding mallard pairs during May 1987-91. We thank H. T. Sklebar for quantifying wetland water conditions on the 9 study areas through aerial videography, and processing pond availability data for wetland preference analysis; D. A. Brandt for assistance in statistical analyses and preparing the figures; K. M. Lahlum, A. D. Kruse, A. H. Berner, and N. H. Euliss, Jr. for assistance in obtaining documents; and R. R. Cox, R. M. Kaminski, J. R. Keough, R. E. Reynolds, M. A. Sovada, and 2 anonymous reviewers for comments on the manuscript. All capture and marking procedures were approved by the NPSC Animal Care and Use Committee and conformed to recommendations of the American Ornithologists' Union (1988).

Study Area

Research was conducted on 9 50.8-km² circular plots, 3 each in 3 glacial land forms, the Missouri Coteau (study areas 1-3 near Kulm, N.D.), a region of dead-ice moraine in central North Dakota (Bluemle 1977); the glaciated drift plain of eastern North Dakota (study areas 4-6 near Jamestown, N.D.); and terminal and ground moraine in westcentral Minnesota (Leverett 1932); (study areas 7-9 near Detroit Lakes, Minn.; Table 1). Each study area was centered on a waterfowl production area (WPA) owned and managed by the U.S. Fish and Wildlife Service (USFWS), and was surrounded by lands mostly in private ownership.

Wetland basin classes were temporary, seasonal, semipermanent, or lake (Cowardin et al. 1988a), which are about equivalent to classes II-V of Stewart and Kantrud (1971). Number and area of wetland basins (within class) varied among study areas (Table 1). Uplands on all study areas were used for production of cereal grains, row crops, and livestock grazing. All study areas, except 1 and 2, were predominantly cropland (Table 1). On average, cropland, native grassland, and wetland accounted for 55, 14, and 15% of the land within study areas. The remainder (16%) consisted primarily of planted cover, hayland (mostly alfalfa), right-of-way, and odd areas.

Methods

Wetland Terminology

We assigned wetland basin class based on the most permanent water regime present within the basin. Ponds were defined as basins that contained water (Cowardin 1982) and were categorized according to basin class, (e.g., water within a basin with a seasonally flooded water regime was termed a seasonal pond). To reduce the bias in female preference estimates when a large portion of a wetland basin was not flooded and was unsuitable as foraging habitat, only pond area was considered available to ducks in analyses of wetland habitat preference.

Pond Availability

Numbers of wetland basins and their areas on the 9 study areas were obtained from digitized maps prepared by the USFWS National Wetland Inventory (NWI) as a special project from high-altitude (1:63,000) color-infrared photographs taken before the onset of the study. Pond density and pond area were updated annually in early May from aerial videography (Cowardin et al. 1988b) taken from a Cessna 172 aircraft equipped with a floor-mounted video camera. Video scenes were converted to digital images by computer with the feature-mapping process of the Map and Image Processing System software (MIPS; Miller et al. 1990). The NWI basin class was attached to each feature map output providing information on basin and pond area by NWI wetland basin class for each study area by year.

Monitoring Wetland Use

We captured female mallards in decoy hen traps placed in wetland basins (Sharp and Lokemoen 1987) on the 9 study areas from mid-April to early May 1988-90 (Table 1). At capture, each female was banded and radioequipped with an adjustable harness transmitter having a mass of about 23 g (Dwyer 1972). Transmitters had an effective range of 2 km to receiving sites on the ground and a life expectancy of 4 months. We released each female at her capture site when marking was completed; if captured with a male, both members of the pair were released simultaneously.

Females were monitored daily with the aid of vehicles equipped with null/peak antenna systems. Locations of females were estimated in the field by plotting multiple bearings at close range on habitat base maps (1:12,000). We visually assessed how close the bearings converged (White and Garrott 1990:72). A grid system of roads or trails, spaced at 1.6 km intervals, usually permitted observers to maintain a transmitter to receiver distance ≤0.8 km and allowed bearings to be obtained from multiple sites. When it was uncertain if a marked female was in a basin, the observer left the vehicle and verified the bird's location using a hand-held antenna and receiver. The order in which we searched for females was changed daily to avoid potential temporal biases. We located individuals between 0600 and 2200 hours daily. A minimum interval of 1 hour between observations was used to maintain independence of relocations (Swihart and Slade 1985). Only relocations that occurred within wetland basins were included in analyses.

Determination of Reproductive Stage

Stage of reproduction of radioequipped females was based on information gained from nests. Using number of eggs laid and stage of incubation (Weller 1956), we backdated from the date each nest was found to establish date of nest initiation. We assumed that 1 egg was laid daily through completion of the clutch. We defined 4 reproductive stages: pre-nesting, egg production, incubation, and post-nesting. Pre-nesting was from 3 days after the female was captured and marked to the day rapid follicular growth began (i.e., 7 days before the date the first egg was laid; Phillips and van Tienhoven 1962). The 3-day interval was used to allow time for females to adjust to capture and marking. Egg production was from the estimated onset of rapid follicular growth to date of clutch completion, corresponding to the interval of increased intake of protein, lipids, and calcium for egg production. To minimize the interval between when a radioequipped female began nesting and when the nest was found, females were monitored daily starting at about 0600 hours so nests could be found at an early stage. If a bird was located in potential nesting cover for 3 consective days, the bird was flushed to determine whether a nest had been initiated. In cases where a nest was located, stage of nest was determined and no further visits were made while the female was present to minimize effect of disturbance. Incubation was from the day after the last egg was laid to hatch or date of nest destruction. Post-nesting began on the date a clutch was destroyed and continued until the bird joined a flock or left the area, or until the time when we estimated follicular growth began again.

Effect of Land Use on Pair Distribution

To assess variation in numbers of breeding pairs in relation to temporary/seasonal pond area and percent cropland, we annually determined number of mallard pairs and pond area by wetland basin class on about one-half of the land surface of each study area. Annual surveys of breeding pairs were conducted on study areas 1, 3, 4, and 6 during 1-15 May 1987-91 and on study areas 2 and 5 during 1-15 May 1988. In preparation for a pair count on a study area, each section (259 ha), or a smaller unit when containing multiple land ownerships or at the edge of the study area, was assigned a random number and landowners were contacted in the numerical order drawn to obtain permission for access to their wetlands. When permission was refused, the next tract on the list was selected until all ponds on about one-half of each study area had been surveyed. Birds were recorded by social group (Dzubin 1969) with observed pairs, lone males, and flocks ≤5 converted to estimated breeding pairs. Percent of cropland was determined for the entire study area. Temporary/seasonal pond area and percent cropland were updated annually from May videography and ground truthing in conjunction with May pair counts.

Statistical Analyses

We examined May pond density in relation to temporary and seasonal pond density using least squares regression (PROC GLM; SAS Inst. Inc. 1989). Within study areas and years, pond density was the total number of ponds divided by area of site.

We compared the proportion of female locations by wetland basin class (i.e., use) with the proportion of pond area by wetland basin class (i.e., availability) for each study area using the method of Johnson (1980), with analyses conducted separately for pre-nesting and egg production. Non-flooded basin was analyzed as a separate category. Pond area considered available to a female by wetland basin class was determined by the time interval when a female was in a reproductive stage. Habitat availability for females from 15 April (when monitoring began) through 31 May came from aerial videography taken during 30 April-5 May 1988, 2-12 May 1989, and 2-3 May 1990.

To test for effects of reproductive stage and year on number of individual ponds used by female mallards, we used the general linear models procedure (PROC GLM) in a repeated measures design with unbalanced data (Milliken and Johnson 1984:378-407, SAS Inst. Inc. 1989). The female mallard was considered the whole- unit and the same female at different reproductive stages was considered the subunit. To accommodate a significant positive linear relation between observed number of individual ponds used per female and number of relocations per female (P < 0.01), number of relocations per female was used as a covariate. Females with fewer than 10 relocations were excluded from the analyses. Fisher's protected least significant difference (Milliken and Johnson 1984:33) was used to isolate differences in means for significant effects.

We used regression analysis (PROC REG; SAS Inst. Inc. 1989) to test for the effects of pond density on the number of ponds used during egg production by females, with number of relocations as a covariate. We defined the pond where a female was located most frequently as the "most-used pond," but excluded relocations at the nest site if it was in a wetland basin. If 2 ponds had the same maximum number of female relocations, both were deemed most-used ponds. Therefore, number of most-used ponds can exceed number of females.

We evaluated whether number of mallard pairs changed relative to number of May ponds from 1963-70 to 1971-85 in eastern North Dakota (stratum 46), using breeding pair/pond count data from the annual aerial waterfowl breeding ground population and habitat surveys conducted by the USFWS (Smith 1995). We used multiple regression (PROC GLM; SAS Inst. Inc. 1989) to examine variation in numbers of breeding pairs in relation to temporary/seasonal pond area and percent cropland by year on the 6 North Dakota study areas (plots 1-6), and their interaction during 1987-91. We tested residuals from the above model for year and study area effects. Significance for all analyses was deemed at α < 0.05 unless otherwise indicated.

Results

Pond Density and Area 1987-91

Changes in density of temporary and seasonal ponds during 1987-91 explained most of the variability in pond density (r² = 0.93; Fig. 1). Number of temporary and seasonal ponds on study areas 1-6 in North Dakota were highest during spring 1987 following high precipitation and runoff in 1986, declined markedly in spring 1988 with the onset of a major drought, temporarily increased during spring 1989 as a result of runoff from snowmelt, and declined to exceptionally low densities in spring 1990 and 1991 with limited spring runoff (Table 2). Annual precipitation at Jamestown during 1987-91 was 21, 38, 7, 21 and 4% below the long-term mean of 46.0 cm (Natl. Oceanic and Atmos. Adm. 1987-91). With below average precipitation during fall 1989 and during the 1989-90 winter, the number of ponds in all basin classes except lakes were abnormally low by May 1990 (Table 2). In Minnesota, the number of ponds peaked in spring 1989 with runoff of snowmelt from a deep snowpack that accumulated during the 1988-89 winter (Natl. Oceanic and Atmos. Adm. 1988-90). Pond densities declined in 1990 and 1991 on Minnesota study areas, but less than in North Dakota (Table 2).

Total pond area differed by basin class and year, with the largest decrease occurring from 1987 to 1988 and from 1989 to 1990 (Table 2). Pond area was more stable at Minnesota sites than in North Dakota; limited temporary and seasonal pond habitat existed at North Dakota sites in 1990-91 (Table 2).

Wetland Use by Reproductive Stage

For 91 radioequipped females that nested in ≥1 reproductive stage, we obtained ≥10 relocations (Table 1). Females were relocated 7,247 times: 1,969 in 1988 (39 birds), 3,213 in 1989 (36 birds), and 2,065 in 1990 (16 birds). Mean number (± SE) of relocations per female by reproductive stage was: pre-nesting, 44.7 ± 3.3 (58 birds); egg production, 35.9 ± 2.5 (56 birds); incubation, 33.4 ± 5.6 (20 birds); and post-nesting, 41.2 ± 4.0 (48 birds).

Wetland basin use.-- Temporary basins, seasonal basins, semipermanent basins, and lakes accounted for 4-9, 18-26, 41-59, and 11-25% of wetland use over the 4 reproductive stages (Table 3). Semipermanent basins were the most used wetland basin class in all 4 reproductive stages, with the lowest use occurring during pre-nesting. Temporary and seasonal basins were used least during post-nesting when females were located primarily on semipermanent ponds and lakes. Among females during egg production, use of temporary and seasonal basins varied from 6 and 24% in May to 10 and 29% in June.

Wetland preference.-- The 4 wetland basin classes and dry basins were not preferred equally by females within and among the reproductive stages of prenesting and egg production during 15 April to 31 May 1988-90 (Table 4). Temporary and seasonal ponds were preferred over other available wetland habitats and were the only wetland habitats used more than expected during pre-nesting and egg production (Table 4). Semipermanent ponds were ranked higher than lakes and dry basins during both pre-nesting and egg production, but differances were not statistically significant during egg production.

Number of Ponds Used

After accounting for number of relocations per female, there was no year by reproductive stage interaction (F_2,58 = 2.62, P = 0.08) but number of ponds used through May differed among the 3 years (F_2,58 = 4.79, P = 0.01). The number of ponds used in 1990 (Least Squares Mean [LSM] = 1.72, SE = 1.35) was smaller than the number of ponds used in 1989 (P = 0.005, LSM = 6.00, SE = 0.57), and 1988 (P = 0.05, LSM = 4.62, SE = 0.59). Accounting for number of relocations, there was a difference in number of ponds used during pre-nesting and egg production (F_1,58 = 7.69, P = 0.007). The number of ponds used during prenesting was higher (LSM = 5.56, SE = 0.37) than during egg production (LSM = 2.67, SE = 0.97).

Accounting for number of relocations, the number of ponds used by a female during egg production through May increased with pond density (F_1,21 = 22.69, P < 0.0001). During pre-nesting, egg production, incubation, and post-nesting, females were located 50, 53, 67 and 56% of the time in the most-used basin. Of 32 females that most often used a semipermanent basin during egg production, only 1 (3%) was located exclusively in that basin.

Effect of Land Use on Pair Distribution

The number of breeding mallard pairs in stratum 46 (where study areas 1-6 were located) was correlated with May ponds from 1963 to 1985 (F = 4.18; 1, 21 df; P = 0.05; r² = 0.17) but response of mallard pairs to May ponds differed from 1963-70 to 1971-85 (P = 0.07). On average, an estimated 33,659 fewer breeding pairs were present in stratum 46 during 1971-85 than in 1963-70 (P = 0.006).

In 1987-91, the number of pairs on study areas located in stratum 46 showed no relation to percent of study area in cropland when temporary/seasonal pond area was low under drought conditions, but the negative effect of cropland on number of breeding pairs increased as temporary/seasonal pond area increased (cropland-by-pond area interaction; F = 3.71; 1, 18 df; P = 0.07; r² = 0.77; Fig. 2). Using residuals from this model as the response variable, we found no effects due to year (P = 0.44) and study area (P = 0.42).

Discussion

Wetland Use and Preferences

Water conditions in prairie wetland complexes are dynamic as reflected in wide annual variation in pond numbers, due primarily to changes in numbers of temporary and seasonal ponds. Preference we observed for temporary and seasonal ponds by female mallards during pre-nesting and egg production presumably reflects more productive foraging opportunities than existed in semipermanent ponds and lakes. Temporary ponds in the PPR are present principally in early spring following snowmelt or rains (Stewart and Kantrud 1971), so attract pairs and are a source of food to early-nesting females (Swanson et al. 1985). Seasonal ponds are the principal habitat of breeding pairs when available (Stewart and Kantrud 1973) and a source of food into or through the breeding season, depending on year and location. At sites where precipitation is high in late spring and summer, temporary and seasonal ponds hold water throughout the nesting season.

Predominant use of semipermanent ponds by female mallards in this study resulted because a high proportion of pond area was in semipermanent basins because of drought. A lower preference for semipermanent ponds than temporary and seasonal ponds probably resulted, in part, because of limited availability of shallow temporarily- and seasonally-flooded habitat within the drought-affected semipermanent basins. Seasonally-flooded wetland habitat has been found to support higher densities of several benthic macroinvertebrates that are important mallard foods than does semipermanently-flooded habitat (Neckles et al. 1990). Semipermanent basins in wet years provide extensive seasonally-flooded habitat for breeding pairs.

Moderate use of lakes resulted primarily from 31% of monitored females being on a plot where lakes formed the principal available habitat. Use of lakes was less than predicted presumably because most lake area was too deep for females to forage efficiently. As a result, lakes supported lower densities of breeding pairs than did the other wetland basin classes studied. Part of dry basin use probably was by females searching for nest sites particularly early in the season before other cover became plentiful. Mallards and other dabbling ducks commonly nest in dry basins during years of drought (Greenwood et al. 1995). Also, some basins probably contained ponds on dates of duck use, but not on the dates when pond areas were documented through videography.

Number of Ponds Used

The higher number of ponds used by radioequipped females during pre-nesting than during egg production probably reflects pre-nesting females spending more time searching for suitable foraging sites not already occupied. When a female located a pond containing adequate food resources to meet needs for egg production and began nesting, most of her time off of the nest was spent in that basin until she was displaced.

The decline in number of ponds used during egg production as pond density decreases reflects fewer options existing for females attempting to breed as pond density declines, whether from drought or man-induced change. Those pairs that stay to breed during drought make fewer nesting attempts (Krapu et al. 1983), and those nesting attempts that are successful result in fewer fledged young (Rotella and Ratti 1992, G. L. Krapu and P. J. Pietz, unpubl. data).

The positive correlation between number of ponds used and pond density (ponds/km²) probably reflects females having greater access to suitable foraging sites as number of temporary and seasonal ponds increase. Under high pond densities, competition declines and pairs readily find alternate feeding sites when displaced from ponds by intraspecific competition. As a result, females in wet years breed longer than in years when most temporary and seasonal basins are dry (Krapu et al. 1983). In the summer of 1993, with most temporarily- and seasonally-flooded habitat filled to capacity throughout summer in eastern North Dakota, some female mallards continued to re-nest through August, apparently in response to exceptionally favorable foraging conditions (G. L. Krapu, unpubl. data).

Effect of Land Use on Recruitment

A positive correlation between landscape pair density (pairs/km²) and May pond density exists across the primary breeding range of the mallard (Johnson and Grier 1988), including eastern North Dakota, because pairs settle where preferred temporary and seasonal ponds are available. However, number of mallard pairs relative to May ponds declined by an estimated 33,659 pairs or 34% from 1963-70 to 1971-85 in stratum 46, probably due primarily to adverse effects of cropland expansion on habitat used by mallards. Our conclusion is based on lower observed pair use of cropland-dominated landscapes in wet years when tilled temporary and seasonal ponds formed a high percentage of total pond area. We interpret this to indicate that annually-tilled ponds are less productive foraging sites, and thus attract fewer pairs. Among pairs that initially do settle where a high percentage of landscape is in cropland, nest success rate is low because of high nest predation by egg-eating mammals (Greenwood et al. 1995), resulting in low homing rates (Lokemoen et al. 1990). Renesting rate among females that lose their first nests probably also decreases in landscapes where most temporary and seasonal basins are cropped when dry. In central North Dakota during 1977-80, a higher correlation existed between mallard nesting intensity and percent wetness of semipermanent basins than of temporary or seasonal basins (Cowardin et al. 1985). This pattern probably resulted, in part, because macroinvertebrate food resources sought during egg production were more available in wet years in the temporarily- and seasonally-flooded zones of semipermanent wetlands which are less altered by cultivation than are temporary and seasonal wetlands. Wetland basin tillage probably has the greatest influence on mallard pair distribution and on nesting intensity in wet springs immediately following drought periods, when temporary/seasonal pond area is most altered by cultivation.

Annual recruitment in stratum 46 in central and eastern North Dakota probably was less influenced by number of May ponds (primarily temporary and seasonal ponds) in 1971-85 than in 1963-70 when a higher proportion of ponds and uplands were in a non-tilled status. Similarly, the percent of landscape in cropland increased from the 1950s to the 1980s across a large part of the PPR in North Dakota (Krapu, unpubl. data) and Canada (Turner et al. 1987, Bethke and Nudds 1995), causing breeding incidence to decline (this study, Bethke and Nudds 1995) along with nest success rate (Beauchamp et al. 1996). Breeding incidence, nest success rate, and nesting intensity are critical determinants of reproductive output in mallards (Johnson et al. 1992). Declines in all 3 factors probably occurred as cropland area expanded across large areas of the PPR from the 1950s to the 1980s, weakening the correlation between number of May ponds and size of the continental mallard breeding population in following springs.

Management Implications

Management Implications

Female mallards during reproduction in the PPR obtain their nutrient requirements from a wide range of wetland habitats. However, females during pre-nesting and egg production prefer temporary and seasonal ponds causing pair distribution and density to vary with availability of these pond types. Efforts to increase nest success rates in the PPR will "reach" more ducks/ha and be most cost-effective if concentrated where temporarily- and seasonally-flooded wetland habitats are plentiful and land use ensures productive pond conditions. The Conservation Reserve Program (CRP) of the U.S. Department of Agriculture, as reauthorized under the Federal Agricultural Improvement and Reform Act of 1996, pays producers that qualify to plant cropped wetlands and adjacent cultivated land to perennial cover for 10 or 15 years under Conservation Practice 23 (CP 23). In the PPR, waterfowl managers can increase the contribution of CRP to duck production by working with producers to enroll tracts that offer the highest potential for increases in recruitment. Tracts with high densities of cropped wetlands, when occurring near potentially secure nesting sites such as islands, predator exclosures, large blocks of nesting cover (Greenwood et al. 1995), or in areas where the predator community is dominated by coyotes (Canis latrans) (Sovada et al. 1995), are particularly well suited for enhancement of duck production when enrolled in CRP. To shorten the period required for cropped wetlands to revegetate and become productive foraging sites for breeding ducks and other migratory waterbirds, we recommend that appropriate vegetation tolerant of seasonal flooding be established in tilled basins when idled under CP 23. Because annual precipitation and subsequent availability of temporary, seasonal, and semipermanent ponds vary temporally and spatially across the PPR, waterfowl managers should seek to maintain high densities of wetlands throughout this important waterfowl breeding area.

---

Accessibility FOIA Privacy Policies and Notices

U.S. Department of the Interior | U.S. Geological Survey
URL: http://www.npwrc.usgs.gov/resource/birds/wetlndus/index.htm
Page Contact Information: npwrc@usgs.gov
Page Last Modified: August 24, 2006

---