USGS - science for a changing world

Northern Prairie Wildlife Research Center

  Home About NPWRC Our Science Staff Employment Contacts Common Questions About the Site

Impact of Red Fox Predation on the Sex Ratio of Prairie Mallards

Validation of the Simulation Results


A crucial step in the evaluation of our simulation study was comparing the simulated results to those obtained from actual field counts. There have been many counts of mallard sex ratios during migration and on wintering areas and nearly all show a preponderance of drakes. For example, in January 1973, counts of mallard sex ratios were conducted in all States of the Central Flyway (Central Flyway Technical Committee, unpublished report 1975). The sex composition of over 59,000 mallards was ascertained. The drake:hen ratio ranged from 110:100 in New Mexico to 365:100 for a small population wintering in North Dakota. The flyway mean, weighted by the wintering population in each State, was 162:100, but it was recognized that the greater coloration and activity of drakes probably biased the counts in favor of drakes. Problems associated with migration and winter sex ratio data, notably differential migration and wintering by age and sex, have been recognized in North America and Europe (e.g., Bellrose et al. 1961; Bezzel 1959; Nilsson 1970).

Although migration and winter sex ratio counts are sufficient to substantiate the existence of a preponderance of males in Central Flyway mallard populations, the data are inadequate to serve as validation of results obtained through operation of our model. The most critical variable we needed to validate was the spring sex ratio. To do this, we needed the actual sex composition of the resident breeding mallard population in the North Dakota pothole region at the onset of the nesting season. In this section we compare the limited pertinent field data on spring sex ratios to the ratio predicted by the simulation model. We also apply techniques for obtaining fall sex ratios from banding data, and compare results of these techniques with those of our model. A further verification is made by comparing total summer mortality rates generated from our model to those appearing in the literature. We also examine the magnitude and sex specificity of other mortality factors to determine if our calculated rates are consistent with the available data.

The Spring Sex Ratio

Although many counts of waterfowl have been made on the prairie breeding grounds during spring, most workers (e.g., Bellrose et al. 1961; Dzubin 1969) recognized that numerous problems are associated with the interpretation of such data. Dzubin (1969:179), in summarizing the opinions of many workers, stated: "The inexactness [of waterfowl inventories taken during the breeding season] stems from a wide spectrum of factors that include weather, breeding phenology, asynchronous nesting periods, vegetative growth, species present and their daily activity, previous field experiences of personnel, plus others."

It is apparent that the sex ratio of mallards in North Dakota can best be determined from counts of prenesting birds in April (M. C. Hammond, unpublished data). Mallards arrive by late March and early April, depending on weather conditions, and nesting usually begins in mid- to late April. During April, however, many transitional migrants are also present. Both M. C. Hammond (unpublished data) and Dzubin (1969) pointed out that in making early counts care should be taken to avoid flocks, which are probably migrants; counts made of dispersed birds are believed to be the most reliable. Also, if yearling birds arrive and initiate nesting later than older birds, the recommended earlier counts may be biased in favor of the older birds.

Once nesting has commenced, unpaired "extra" males in the population are by present criteria indistinguishable from "lone males" (mated drakes waiting on ponds for their laying or incubating hens). In most studies aimed at assessing population densities (e.g., Hammond 1969; Stewart and Kantrud 1973), single or small groups of unattached males were counted as representing pairs. In a population with a preponderance of males this could lead to overestimating the size of the breeding population. Dzubin (1969:189) recognized this potential when he stated: "The Kindersley [Saskatchewan] data, which were gathered in a localized area of the vast breeding grounds, tend to substantiate the views of many workers that in spring populations of most waterfowl species there is a preponderance of unmated males which I suggest can be counted erroneously as indicated pairs." Both Dzubin (1969) and Hammond (1969) recognized this possibility and recommended that indicated pair counts be corrected to account for the presence of "extra" drakes.

Much of the existing mallard sex ratio data, whatever their shortcomings, show a preponderance of males among prairie mallards during the prenesting period (Table 16). The male:female ratio for the various sets of available data ranged from 108:100 to 129:100.

For validation of our model the best source of data on the sex ratio of breeding mallards in the pothole region of North Dakota was the largely unpublished files of M. C. Hammond, which contained intermittent data for 1939-64 (Table 16). We used his data for April, here assumed to represent the prenesting period. The overall male:female ratio for the prenesting period for all years combined was 117:100. Although there was considerable variation among years, the data suggest that the population was more imbalanced during the 1960's than in the earlier years. The overall male:female ratio for 1939-52 was 108:100, but the ratio for 1959-64 was 129:100. In the early 1960's the North Dakota red fox population was reportedly at one of its highest levels and the mallard population was relatively low. More recent data were collected near Woodworth, North Dakota, during 1964-69 (L. M. Kirsch, personal communication). The April censuses there revealed a sex ratio of 120:100 in 551 mallards.

The above data suggest that the sex ratios generated by our model are reasonable, but it remains to be determined how accurately field counts made during the prenesting period represent the actual sex ratio among breeding mallards.

The Fall Sex Ratio

Anderson (1975) presented two estimates for the average fall sex ratio of adult mallards in North America. The first, a straightforward application of the method of Wight et al. (1965), was based upon average survival and production figures for 1961-70 and furnished an estimate of 121:100. Anderson's second estimate was based on a complex stochastic model of the mallard population. Although intended to explore the consistency of the many parameter estimates, rather than to fit the predictions to observed data, his model generated an average fall sex ratio of 127:100.

A preliminary method for estimating the sex ratio of mallards before the hunting season was presented earlier by Anderson et al. (1970), who used the sex composition of mallards harvested, adjusted for differences in hunting vulnerability between males and females. The harvest data were obtained from the collection of waterfowl parts voluntarily sent in by hunters who were selected to participate in the annual survey. The difference in vulnerability was based on band recovery rates during the hunting season of normal wild birds. The estimated sex ratios, which apply to the population just before the hunting season, were 152:100 in adults and 100:100 in immatures for 1967-69. The even sex ratio in immatures is in substantial agreement with other studies (e.g., Bellrose et al. 1961).

Fall sex ratios, which pertain to the mallard population after female-selective fox mortality has taken place and before male-selective hunting mortality comes into play, should be more highly distorted than spring sex ratios. The various estimators described above although they apply to a broader population than ours, are seen to be consistent with or exceed the spring sex ratio of about 126:100 generated by our model. Thus, while the relationship between the sex ratios is tenuous, the fact of more highly distorted fall sex ratios is confirmed.

Summer Mortality Rates

The estimates of the mortality rates of mallards during the summer, as produced by the simulation model, are important as a check of the model and in their own right. So little is known of such rates that it seems prudent to discuss our results, even though they were somewhat influenced by our decision to set the summer rate of "other" mortality at three times the winter rate.

By multiplying the rate of survival from fox mortality by the rate of survival during summer, we obtained the spring through summer survival rate. For males, we have, using the data in Table 14,

(1 - r7) x (1- r9)
= (1 - 0.051) x (1 - 0.119)
= 0.836.

Thus, the mortality rate during spring and summer (approximately April-September) for males was 16.4% (1.00-0.836). The corresponding figure for females was 28.5%. These rates pertain to the approximate 6-month period from arrival in spring to the beginning of the fall hunting season.

We have very little real data with which to compare our estimates of summer mortality. Keith (1961) calculated mortality rates for all duck species nesting on a study area in southeastern Alberta as 2% for males and 8% for females. He noted that the rates, based on discoveries of dead ducks, are conservative. The rates applied to a 10-week period, 6 May-15 July, although the female rate may have included losses that occurred earlier in the nesting season.

Dzubin and Gollop (1972) calculated mallard death rates on two Canadian study areas, one each in grassland and aspen parkland, from searches for dead birds. The estimated rates which apply to the April-June period, were 2% and 4% for drakes and 5% and 7% for hens, in the grassland and parkland, respectively.

Gilmer et al. (1974) provided estimates of mortality based on losses of radio-equipped mallards and wood ducks (Aix sponsa) in a forested region of north-central Minnesota. Female mortality rates for a 120-day breeding period were 25% for mallards and 29.4% for wood ducks. Male rates, less reliable because of fewer radio-equipped birds and shorter tracking periods, were 0% for mallards and 10.2% for wood ducks.

Anderson (1975), noting the importance as well as the dearth of realistic summer mortality information, presented a procedure to obtain crude estimates based on annual survival rates and subjective knowledge. He derived mortality values, pertaining to the 3-month period 15 May-15 August, of 8-9% for male mallards and 16-18% for female mallards. These rates are more reasonable than the conservative figures of Keith (1961) and Dzubin and Gollop (1972), and, recognizing the different geographic regions and lengths of time to which they apply, are closely comparable to the rates determined in our simulation model.

The calculated summer mortality rates were thus consistent with comparable data collected elsewhere, and lend additional credence to the validity of the results of our model.

The Sex Specificity of Other Mortality Factors

Our results indicate that the sex specificity of fox predation was sufficient to cause a distorted sex ratio similar to those appearing in wild mallard populations. Mortality rates from other causes, except hunting, were shown to be on the whole slightly higher in female than male. mallards. We now examine the available information on rates of mortality from other causes in order to determine if, in fact, near parity between the sexes is reasonable.

Several mortality factors other than fox predation and hunting can be identified, notably accidents, weather, disease, and predation by species other than fox. Evidence relating to these factors is scattered throughout the literature, but much of it was reviewed by Bellrose et al. (1961) and Stout (1967). Stout also examined returns of banded birds and questionnaires completed by 225 observers throughout the United States and Canada. The diverse origins of his data preclude estimation of mortality rates from the various causes, but it ispossible to identify potentially important causes and to determine their sex specificity.

Collisions, with automobiles or overhead wires, and agriculturally related accidents have been singled out as important causes of mortality among dabbling ducks. During 1969-74, data were collected on the composition of ducks found dead in east-central North Dakota in spring and summer (A. B. Sargeant, unpublished data). Field biologists were requested to record and remove from the location all dead ducks or duck remains they found. Of 86 mallards found dead along roads, 35 were males, for a male:female ratio of 69:100 (Table 17).

In contrast, Stout (1967) reported more males than females among waterfowl killed by automobiles. His questionnaire data (for all waterfowl) resulted in a sex ratio of 128:100, while the recoveries of banded mallards totaled 59 males and 31 females, for a sex ratio of 190:100. This latter figure is misleading, however, because more male than female mallards are ordinarily banded so that the expected return would include more males than females, even if the sexes were equally susceptible. It seems that, outside of the breeding season, automobile collisons operate with about equal intensity on the sexes.

Available data do not suggest that automobiles kill significant numbers of mallards. The data in Table 17 represent all mallards found during searches in many thousands of miles of driving. This mortality source appears to be selective for hens in spring and summer and nearly nonselective at other times of the year.

Collisions with telephone and power lines are also a source of mortality among ducks, but few data are available concerning the number and sex composition of mallards killed in such a manner. Cornwell and Hochbaum (1971) and Krapu (1974) suggested that such losses go largely unnoticed; these birds would likely be scavenged by foxes or other predators. Krapu (1974) opined that courtship flights, which often involve several males and only one female, probably result in some deaths from collisions with wires, and Boyd (1961) showed that male ducks in Great Britain were more prone to strike overhead wires than were females. Sowls (1955) observed four drakes and no hens killed by telephone lines. Krapu (1974) reported two mallards killed by wires in North Dakota, one of each sex. Stout (1967) reported the sex of 56 banded mallards that collided with various objects in 1930-63; the ratio of males to females was 250:100. In that study most of the reported losses were of wire-killed mallards in the Central Flyway during fall and winter. Losses from striking overhead wires are probably not a serious cause of mortality and the limited evidence suggests that losses are greater among males than females in all seasons.

Agricultural activites, particularly haying, result in the loss of some mallards, mostly nesting hens. Stout (1967) reported that 66 of the 73 banded mallards reported as killed by farm machinery were females. Respondents in his questionnaire survey reported 74 female waterfowl killed versus no males.

Haying generally begins in early June in North Dakota. Losses that occurred then would likely have been included in the fox den samples, but those after late June would be largely additive to our fox-attributed mortality. Field studies bearing on hen mortality from haying led to somewhat conflicting conclusions. Labisky (1957) found five mallard and blue-winged teal (Anas discors) hens killed out of a total of 122 active nests in Wisconsin haylands. Milonski (1958) studied 608 nests of seven duck species on diverse farmland in Manitoba. Of 110 nests in hay fields, he found hens killed by mowing at 2. These two studies generated estimates of hen mortality due to mowing of about 4% and 2%, respectively. In contrast, Ordal (1964) suggested that mowing mortality of mallard hens was at least 15% based on a comparison of three mallard hens killed by mowers to the indicated mallard population. B. E. Burkett (personal communication) investigated hayland nests in the Jamestown, North Dakota, vicinity and found that none of seven mallards nesting in hay was destroyed, although eight hens of other species involved in 34 nests were killed.

Susceptibility of hens to mowing probably depends upon the species involved, the phenology of the season, the manner of mowing and, most importantly, the extent of hayland available for duck nesting. Significantly, relatively little land (about 8% for North Dakota in 1973 - Anonymous 1974) is devoted to the raising of hay in much of the Prairie Pothole Region of North Dakota. Bellrose et al. (1961), who cited the Labisky and Milonski studies concluded that the available data indicate that losses of nesting hens resulting directly from agricultural operations do not contribute importantly to imbalance in adult sex ratios. Based on the additional recent evidence cited above, we believe that haying can inflict a significant mortality on female mallards under certain circumstances, but that the average effect in our reference area is minor, because of the restricted acreage of hayland available for nesting.

Considering the three major types of collisions together, automobiles probably take more females than males in spring and the sexes in nearly equal numbers at other times, overhead wires likely kill males selectively, and agricultural operations selectively kill females. The net result of these kinds of mortality, we believe, is a small but significant rate of mortality, somewhat higher in females than in males.

Losses due to weather are probably not selective for either sex. Lebret (1950) stated that, in Europe, female wigeons (Anas penelope) and mallards were less resistant to severe winter weather than were males of the species. Females compensated for this by generally wintering in more mild climates. Stout (1967) identified weather-related mortality factors as ice, cold, wind, hail, and heat, and concluded that they were apparently taking males and females in proportion to their numbers in the exposed populations.

Diseases constitute another source of mortality among mallards; botulism has been most closely identified with significant losses of ducks in the prairie region during our study period. Botulism has been suggested as possibly bringing about excesses of males in waterfowl populations (Hammond 1950). The disease usually appears during middle to late summer when marshes are drying, and often affects relatively large numbers of birds. Of 1,638 adult mallards afflicted with botulism on four wildlife refuges in North Dakota during 1937-47,69% were males; the male: female ratio was 219:100 (Hammond 1950). Hammond concluded that males and females were about equally susceptible to the disease and the sex differential in the kill was related primarily to differential use of the types of marshes where outbreaks of botulism were most frequent. Cooch (1949) reported the sex ratio of 2,570 mallards killed by botulism at Whitewater Lake, Manitoba, in 1949. Weekly samples contained more males than females throughout the 14 July-4 September period. At first, the sex ratio was highly unbalanced in favor of males but was almost even by the end of the period. This change probably reflected differences in the time of molting of males and females, and illustrates how an intrinsically non-sex-selective disease can actually bear more heavily on one sex than the other.

Mallards also succumb to other diseases, but these losses are believed to be of lesser magnitude than that of botulism and probably reflect the sex composition of the population, as stresses and different sexual behavior associated with the breeding season are largely absent. A 1973 outbreak of duck virus enteritis also caused the death of many thousands of wintering mallards in South Dakota but, because it was virtually unknown in wild populations during our reference period, it could not have affected the sex ratio then. Diseases resulting in a weakened condition of birds, however, would predispose them to other mortality factors, especially predation, but such birds would be included in those samples.

Stout's (1967) report included many disease casualties, the sex ratio being 141:100. He felt this ratio indicated a slight selection for females because he believed the sex ratio in the population was more highly distorted. In summary, it appears that disease, on the average, operates with equal force on both sexes.

Numerous predatory animals other than red foxes inhabit North Dakota's pothole region. They include the coyote (Canis latrans), badger (Taxidea taxus), raccoon (Procyon lotor), striped skunk (Mephitis mephitis), mink (Mustela vison), long-tailed weasel (Mustela frenata), house cat (Felis catus), domestic dog (Canis familiaris), great horned owl (Bubo virginianus), and several species of hawks. Only the red fox (Sargeant 1972) and mink (Eberhardt 1974) have been implicated in substantial losses among adult ducks in this region but few data are available on most other predatory species. Fox predation during their denning season has already been discussed. Foxes also prey on ducks after the denning season. Late-nesting ducks would be vulnerable as would some hens leading broods overland and some drakes and hens on marsh shorelines. We believe that fox predation rates after the denning season are much less intense and less selective toward females than earlier in the season.

In a study of mink on North Dakota marshes, Eberhardt (1974) found a total of 71 adult dabbling ducks, 17 of which were mallards, at mink dens during spring and early summer. The sex of nine mallards, of which three (33%) were females, could be ascertained. Although the mallard sample is small, the sex ratio was the same as that obtained for all 46 dabbling ducks whose sex was determined (Table 18). The apparent selectivity for male dabbling ducks was believed to reflect their greater presence on semipermanent marshes occupied by mink during the nesting and molting periods. Mink predation on mallards is of a magnitude considerably lower than that attributed to foxes and is selective for drakes.

Other predators probably have relatively little impact on mallard sex ratios. Most interact with waterfowl primarily by destruction of nests and use of scavenged birds. Coyotes, like foxes, probably prey heavily on mallards, but their numbers were very low during the period of this study. Occasional observations such as those of Hanson and Eberhardt (1971) document their predation on waterfowl. Keith (1961) reported that long-tailed weasels killed approximately 3% of the nesting hens in his study in Alberta, but predation by weasels is probably most successfully directed at small waterfowl species and long-tailed weasels are only sporadically abundant in North Dakota. Great horned owls take mallards (Errington 1932; Houston 1960) but we have no information on the predation rate or its sex specificity.

Predation also occurs while mallards are migrating or wintering. We have no evidence, however, of any sex selectivity among predation losses outside the breeding season, and believe that most losses during migration and wintering periods are nearly proportional to the sex composition.

Physiological stress is an additional mortality factor that has been suggested as a cause of disproportionate sex ratios (Bellrose et al.1961). Females engage in egg-laying, incubation, brood-rearing, and molting in rapid succession and the stress associated with these activities may occasionally be the proximate cause of death (e.g., Bolen 1970). More often, we think, stress heightens the bird's vulnerability to other mortality factors (Harris 1970), particularly predation. Most birds dying directly or indirectly from stress will appear as predation losses. Those taken by foxes would have been accounted for in our model; those taken by other predators would have been included in "other" summer mortality.

The overall magnitude and effect of the other mortality factors affecting mallards in North Dakota are unknown, but it is unlikely that any single cause of mortality averaged more than a small percentage of the annual breeding population. In summary, it appears that the approximately 15-16% average annual loss attributed to factors other than red foxes and hunting (Table 14) is a reasonable approximation of the magnitude of such losses. It also appears that some of the losses, particularly those due to mowing, are selective for females while others such as collisions with wires and mink predation are selective for males. Thus, there is a tendency for the sex selectivity of these mortality factors to cancel each other, and the evidence suggests that the overall effect on North Dakota mallard sex ratios is considerably less than that caused by fox predation. Additional work, however, is needed before the actual effect of the individual mortality factors on sex ratios can be satisfactorily determined.


Previous Section -- Results of the Simulations
Return to Contents
Next Section -- Concluding Remarks

Accessibility FOIA Privacy Policies and Notices

Take Pride in America logo USA.gov logo U.S. Department of the Interior | U.S. Geological Survey
URL: http://www.npwrc.usgs.gov/resource/mammals/sexratio/valid.htm
Page Contact Information: Webmaster
Page Last Modified: Saturday, 02-Feb-2013 05:56:55 EST
Menlo Park, CA [caww55]