USGS - science for a changing world

Northern Prairie Wildlife Research Center

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

Assessing Breeding Populations of Ducks by Ground Counts

Recommendations for Block Study Area Census


On the basis of the behaviour of marked birds, studies of the chronology of nesting and frequent direct counts of pairs, I propose that the following procedures be utilized for ground census on large, square-shaped study areas, e.g., 10 or more square miles, in grassland habitats, where emergent vegetation has little effect on visibility of ducks. Their use should allow estimates which more closely approximate absolute breeding numbers per unit of pond-type breeding habitat. The extrapolation of these recommendations to marsh, parkland, and wooded habitats or to transect counts might require further sampling and modification, as visibility and mobility factors vary among habitats. I assume that all field workers are able to differentiate between waterfowl species and are able to recognize various component groups: lone drakes, grouped drakes, aerial flights, migrating groups, etc.
  1. Censuses should be conducted during an optimum interval of the breeding season, i.e., when most pairs and drakes show maximum site attachment, an indication that the greatest percentage of the population is in the prenesting (including renesting), laying, and early incubation stages. Complementary nest phenology data are required to determine optimum census periods. Breeding chronology can also be crudely deduced from ratios of pairs to lone drakes to grouped drakes taken during mid-morning censuses. For counts made between 0800 and 1200 hours I suggest a simple rule of thumb for optimum census time of mallards and pintails. Ratios of pairs to lone drakes to grouped drakes should be approximately 1:1:1 (i.e., one-third or less of the total indicated population should be enumerated as pairs, with the remaining two-thirds or more as lone or grouped drakes). Phenologically optimum census periods for other dabblers, whose pair bonds and site tenacity are stronger, and whose drakes do not aggregate until after the mid-incubation period, are those in which the pair to lone drake ratio is approximately 1:1 (i.e., one-half or less of the population is counted as pairs). To ensure that pairs or drakes are truly resident and show site attachment, counts for each species must be correlated with time of arrival on the breeding grounds and nesting chronology.

  2. Ground census should be conducted between 0800 and 1200 hours, local standard time, when all species are least mobile and pairs and lone drakes are most likely to be on their waiting sites. As few birds as possible should be flushed. Any birds taking flight should be visually followed to locate their points of landing. These birds are then subtracted from counts if they land on ponds yet to be censused or added if they alight on already enumerated potholes. Birds flushing at some distance from the observer should be recorded as unidentified ducks.

  3. Census should be conducted on sunny or bright, but not heavily overcast days, with temperatures above 40ºF and with winds not in excess of 15 mph, because rain, heavy cloud cover, low temperatures and high winds all affect mobility, dispersion, and visibility of ducks (see Diem and Lu, 1960). Winds increase in velocity in the Canadian prairies in afternoons. Mid-mornings present more optimum counting and light conditions. Counts should be conducted from the south or east edge of ponds to avoid difficulties in identification caused by backlighting and water reflecting sunlight. Replicate counts should be made at the same time of day during the optimum census period, over the same route and under approximately similar weather conditions.

  4. To obtain adequate estimates more than one census should be taken at the optimum period for each species and an average of four to six counts be used rather than maximum or minimum counts of each species. Accuracy and precision are both increased with multiple counts. Average population figures taken from multiple counts give some indication of the magnitude of the error of estimates due to mobility of drakes and the temporary absence of drakes from their waiting sites.

  5. Because a single census can not adequately measure populations of early nesters (mallard and pintail), intermediate nesters (widgeon and shoveler), or late nesters (gadwall and blue-winged teal), a minimum of two different censuses must be conducted in the grasslands to sample a multiple-species breeding population. Under the conditions studied from 1956 through 1959, censuses of mallards and pintails made between May 8 and 20 and censuses of widgeon, shoveler, gadwall, and blue-winged teal made between May 25 and June 5 adequately sampled breeding populations for determination of yearly trends. Counts made after these dates tended to underestimate breeding pairs when drakes abandon home ranges or to overestimate populations when small, postbreeding flocks of drakes moved into the region. Drakes whose hens are incubating congregate and may wander away from their home ranges, beyond the boundaries of the study area.
    Where only a single census can be conducted, the optimum period in an "average" year (i.e., one with no bimodal peaks in nesting effort) could be described as approximately a week before the first mallard or pintail broods are observed and while most of the intermediate and late nesters are in the prenesting, laying, or incubation stages. At Kindersley May 15 to 25 seemed most suitable to estimate early breeders and late breeders. Single censuses conducted after May 20 will tend to underestimate the mallard and pintail segment as some drakes have left the breeding home ranges. For the intermediate and late nesters, single counts taken as early as May 15 may not assess late migrants. Censuses of mallard and pintails taken after May 31 tend to overestimate the population if small groups of five or less post-breeding drakes congregate on favoured loafing sites and are counted as indicated pairs and not as postbreeding males. Censuses may be biased in years of extended nesting. When breeding seasons are staggered, with the first broods appearing when the late breeding pairs are initiating their first nests, some drakes have already abandoned home ranges. Two separate censuses may have to be undertaken in years when cold spells protract the breeding season.

  6. All censuses should be conducted from a vehicle which is driven to a point overlooking each pond, but not close enough to flush birds. To assure that all birds are visible and not sleeping on shore some minor commotion, slamming of car door or sounding of horn, should be used to alert them. Counts should be conducted in vegetated ponds before new growth becomes dense. In late June, birds may have to be flushed by walking through ponds choked with emergents.

  7. All lone pairs and lone drakes should be considered resident, indicated pairs if they are spaced 15 or more feet from other pairs. Before the start of nesting in April and again in June, all aggregated pairs are to be considered migrants, displaced birds, or postbreeding groups. Lone females are not to be considered pairs. Practically all dabbler hens are paired and disassociation of drake and hen is invariably temporary. Late June and July counts of lone hens, after drakes have left the breeding grounds, may be used as evidence of incubation and continued breeding but because of variable daily recess periods among hens, no estimate of total number of hens incubating can be made from single censuses taken during one time interval of the day. The enumeration of lone hens as indicated breeding pairs should be restricted to uncommon or rare breeding species, e.g., at Kindersley, green-winged teal made up less than 1 per cent of the breeding population. Lone hens found over one-half mile from the next nearest drake of an uncommon breeding species could be assessed as a pair.

  8. All groupings of males from 2 to 10 should be considered indicated pairs except for the following stipulations:
    1. Mallard and pintail grouped drakes of up to 10 should be considered resident pairs until approximately 1 week before the first two to three broods are observed, i.e., about May 20 at Kindersley. Thereafter, to the first week of July, only groups of five or less should be considered resident pairs. Stage of body moult and behaviour may aid in separation of apparent breeding drakes and those in postbreeding flocks. Prior to mid-incubation, groups of two to five, inclusive, mated mallard drakes can be considered the same as lone males on waiting sites, as periodic shifts toward aggregation and then dispersion occur during the day.

    2. Widgeon and shoveler grouped drakes of five or less in number should be considered resident pairs until the first appearance of two or three broods, i.e., approximately June 5 at Kindersley. Rarely do mated drakes of these species associate before their hens are in mid-incubation. If grouped drakes are observed before May 10 they are invariably unmated but corrections for these individuals can be made using the prebreeding sex ratio.

    3. Gadwall and blue-winged teal grouped drakes of five or less should be considered resident pairs until the first appearance of broods, i.e., about June 15 at Kindersley. Again, rarely do drakes of these species associate before their hens are in mid-incubation. Grouped drakes observed before May 15 are usually unmated, although an unmated blue-winged teal drake may occasionally associate with a pair or with a mated drake.

    The validity of those dates depends on time of spring migration and time of nesting of each species. They can be either 1 or 2 weeks earlier or later depending on nest chronology. For mallards and pintails, counts might be initiated 2 to 3 weeks after the first hens begin to lay in mid-April. For intermediate and late breeding dabblers, counts should be started 1 to 2 weeks after the first clutches are noted. An attempt should be made to complete counts before the first nesting hens of any species are in their third week of incubation.

  9. Drakes in groupings of 2 to 30 males and one hen, in group flights associated with courtship (GFAC) or attempted rape flights (i.e., aerial flights temporarily on ponds) should be considered resident pairs as both mated and unmated drakes join such flights. Groupings of several pads or aggregations of five or more males and two females in apparent postbreeding groups (usually after June 1) should not be enumerated as resident pairs. A drake initiating a three-bird flight ("territorial chase") from a pond should be considered a resident pair even though he may land elsewhere (see Dzubin, 1957; Lebret, 1961; and McKinney, 1965, for descriptions of all aerial flights). Drakes or pairs flying over an area are not to be counted as resident pairs. 10. Because an unknown, but apparently large, proportion of unmated males remain on the breeding grounds and because sex ratios of all dabbler species are unbalanced toward males, a correction factor should be used to reduce the error arising when all drakes are considered potential pairs. Provided all lone and grouped drakes are counted as pairs (under 8 and 9, above) a correction factor to account for unmated males should be applied. Such factors based on prelaying sex ratios for each species as given in this paper (Table 3) and by Bellrose et al. (1961) should be utilized to obtain a "sex ratio corrected population". Sex ratios may fluctuate yearly and may also be different in pond and large marsh habitats. Hammond (in litt.) has shown that sex ratios of mallards and pintails can vary yearly, especially following a poor production year, e.g., 1961. An attempt should be made to gather these ratios yearly in each census region. Sex ratio corrected populations are important in determining accurate productivity rates of pairs. If sex ratios do not appear to vary yearly in one habitat type, uncorrected indicated pair figures can be used for determining trends on transects. In summary, all dabbler drakes should be enumerated as indicated pairs but a sex ratio correction factor should then be applied to account for the unmated segment of the breeding population.

  10. Enumeration of populations of divers (i.e., Aythya sp.) is complicated by unbalanced sex ratios and the congregation of breeding pairs on deep ponds used as preferred waiting sites. A1though dispersal, through periodic movements of diver breeding pairs to nearby ponds to nest, does occur, divers show contagious and non-random distribution patterns. For portions of any day through the breeding season, diner pads and drakes are found loosely associated. Canvasback and redhead (and to a minor extent, lesser scaup) are also highly mobile during the breeding season, the maximum extent of the home range being some 2 to 4 miles. Unless counts of all preferred waiting sites are made on a study block and consideration given to home range sizes, census of divers will be inconsistent. Also, as Weller (1959) has pointed out, a varying proportion of redhead hens are completely parasitic and do not lay in their own nests. They might, therefore, not be considered true breeding pairs. With divers, only visible pairs or lone drakes known by their behaviour to be on waiting sites should be enumerated as indicated pairs. Nest cover searches should encompass all pond-edge and upland habitats for lesser scaup and all pond-emergents plus nearby shrub uplands for canvasback, redhead, and ruddy duck. For block-type study areas I concluded that a better estimate of diver populations could be made during the mid- to late incubation period of the early nesting pairs, through enumeration of nests, including all those which are viable, hatched, or destroyed.
    As with divers, ruddy ducks are best enumerated by nest counts as their secretive habits do not lend themselves to accurate ground census. Also, some pairs appear to be nonbreeding, summering birds. It is acknowledged that nesting studies can also tend to distort estimated population sizes per unit area as localization of prime nesting cover may attract hens from 1 to 2 miles away. Also, all nests may not be located and some late nesting pairs may be considered renesters. Enumeration of observed pairs and lone drakes on waiting sites should be used to complement nest counts and aid in estimating pair numbers. I suggest that census errors are smaller using nest estimates than those obtained from a ground count of indicated pairs which includes all drakes. Where nesting studies are not feasible the enumeration of all diver pairs and drakes and subsequent correction of these estimates with prebreeding sex ratios may give crude population estimates useful in measuring trends (Murdy, 1962). On block areas, lone diver hens need not be assessed as indicated pairs as they are invariably mated with a nearby drake, which may be enumerated as a lone male on a waiting site.
Censuses can be further complicated by bimodal nesting peaks which reflect April or early May cold snaps, by differential and late migrations due to inclement weather factors south of the breeding grounds, and by major shifts of transient pairs in mid-breeding season from drought-stricken areas. Although measures of an influx of late migrants can be made, their breeding status while on a study area is still unclear. At present, little measure can be made of percentage of such pairs which have already attempted to nest and are nonbreeders. On the basis of species behaviour and examination of gonads, few nonbreeding pairs were found in the parkland habitat but more were noted in the grasslands in drought years. Also, no adequate estimates have yet been made of the seasonal turn-over of breeding pairs on an area (Smith and Hawkins, 1947; Ellig, 1955). Presently utilized counts aid in estimating maximum populations during one time period only, whereas in a dynamic breeding population the actual number attempting to breed on one area may be much higher if the total seasonal population is considered. Mortality of hens and drakes on the breeding ground is largely unknown. Keith (1961: 44) estimated a summer mortality of less than two per cent for males and eight per cent for females of all species. If these are representative figures, pair estimates should be further corrected to account for summer mortality of adults. In all, waterfowl censuses which attempt to measure absolute numbers of breeding pairs utilizing a unit of habitat through the spring and summer remain inexact, but closer approximations can be made of resident pair numbers by studying species behaviour, noting nesting phenology, and utilizing the above recommendations.


As previously noted, two different methods were used to arrive at dabbler population estimates on the grassland and parkland study areas. In the grassland, periodic counts were averaged and a sex-ratio correction factor applied to account for unmated males. In the parkland, fewer direct counts were made but many ponds were visited daily, which led to a more intimate knowledge of pairs and species using a particular portion of the block. Pair numbers were assigned to the block if pairs were observed on or near a particular pond during three out of four censuses.

The minimum block size, its configuration, pair numbers, species make-up, pond numbers, vegetation, etc., required to obtain statistically adequate estimates for accurately measuring yearly changes in population size are still largely unknown. Small blocks of 1 to 2 square miles have the advantage of being quickly censused and are most amenable to replication of counts but the assessment of pairs is subject to wide sampling error and various biases because of bird mobility, overlapping home ranges, and small sample sizes of each species. On larger blocks of 10 square miles or over, sample sizes are increased, but mobility on the edges still persists and no differentiation between late migrants and residents can be made.

Many of the above grassland recommendations can be modified for use on small sample plots of parkland habitat, i.e., 600 to 900 acres containing 50 to 100 pairs of 10 species per square mile. A weekly census of dabbling ducks on all ponds of the area, throughout the breeding season, can be used to determine peak indicated numbers of each species. Optimum census periods can then be calculated for the early, intermediate, and late breeders. For each of these three groups, counts might be conducted daily, from 0800 to 1200 hours, for 4 or 5 successive days and all pairs, lone drakes, grouped drakes, aerial flights on ponds, lone hens, and aggregated pairs or mixed-sex groups be plotted on a base map. Such data can be used to determine any localization of activity of indicated pairs. As noted under grassland recommendations, migration and nesting chronology vary with species. Counts of mallards and pintails should be conducted 2 weeks before those of other dabblers, usually in early or mid-May, i.e., about 2 to 3 weeks after the first hens start to lay. All counts can then be used to aid in estimating breeding pair abundance of dabblers, if the assumption is made that the average indicated population taken from four or five counts during the optimum census period is in fact an accurate representation of the seasonal breeding population. If late migrant pairs are noted moving into the area through the summer, periodic counts from April through July may be necessary.

It should be recogmzed that any counts conducted over a 5- to 7-day period will not adequately assess the seasonal turn-over of pairs on a block and will not measure late season migrants or breeders. Peak indicated populations may occur in June, after late season influxes of pairs (Jessen, Lindmeier, and Farmes, 1964; Evans and Black, 1956; Kirsch, in litt.). More accurate estimates of breeding pairs on small study areas can be made by use of the above census recommendations supplemented by an intimate seasonal knowledge of the behaviour, biology, and nesting success of each pair using the block. By observing pair behaviour and mobility and by locating nests, a better estimate of resident pairs can be made. Such intense field work requires that a worker confine his summer activities to only one block area. Objectives of any study will determine the degree of accuracy or precision required of any population estimate.

As in the grasslands, counts of diver breeding pair populations are complicated by irregular congregations of drakes and pairs on favoured areas. Seasonal counts of viable, hatched, or destroyed diver nests obtained through periodic nest searches, supplemented by periodic counts of pairs and lone drakes on waiting stations, can be used to assess diver populations on a small block. Census of ruddy ducks is difficult because of their secretive habitat and use of congregating ponds by mated and unmated drakes. A number of hens arrive on the breeding ground unpaired. Early morning (0400 to 0600 hours) and late evening (1900 to 2100 hours) counts on preferred congregating and nesting ponds, in which the observer quietly watches one pond for one-half to 1 hour, can be used to assess breeding pair numbers. Again, a nesting study should complement any census.

Behaviour and census

Waterfowl census, although requiring an appreciation of statistical methods and an adequate knowledge of species biology, also requires an intimate knowledge of species behaviour under a variety of population densities and environmental conditions. The decision whether to include a group of drakes and a single hen of an aerial flight temporarily on a pond or a group of pairs, in the potential breeding pair column of a census sheet, will depend on how well the census-taker knows the seasonal behaviour of each species. Waterfowl census then becomes, in part, an art for those knowledgeable workers who can appreciate key attack, sexual, and escape patterns in drakes and hens and apply such observations to decisions as to whether to include a bird, an aerial flight, or group of birds in a count. Similar sentiments have been echoed by Stokes and Balph (1965) who stress the need for an appreciation of species behaviour for a better understanding of all population ecology phenomena and by Diem and Lu (1960) who point out that "accurate interpretation of census data requires more basic knowledge of the behaviour and physiology of the individual bird".

Today, waterfowl censusing remains highly subjective. Counts made by two workers are not strictly comparable for the same or different areas: Census-takers, themselves, can help infuse more objectivity into counts. Recognition of ducks must be instantaneous. They should also be able to determine by sample counts of pair, lone drake, and grouped bird ratios, the optimum census periods. They should readily recognize any biases and potential sampling errors involved. Accuracy and precision of estimates can, in large part, be a reflection of the knowledge and mental alertness of the individual conducting the census.

Seasonal counts and periodic replication of censuses tend to increase accuracy of estimates. A high degree of accuracy and precision is justifiable on special study areas from which pair population estimates are used to calculate the true number of pairs successful in producing broods. Also, increased precision leads to the better detection of the effects of various limiting factors, however minor, on final production. Valid comparison of results of work in various habitat types is facilitated by data with known sampling errors and variability. Statistically describable estimates of seasonal breeding populations based on proven census techniques still remain a basic need of most waterfowl research and management programs.

Supplementary data required

Waterfowl breeding pair census will become more meaningful as data become available on the following topics:
  1. The proportion of any pair population which is nonbreeding and the climatic, density, or habitat factors which lead to nonbreeding: do all yearling lesser scaup and late-hatched mallards of the previous year nest? If they do breed, are they late nesters?
  2. Duration of the pair bond in seasons of varying phonology: how long does the "average" drake of each species remain on his activity centre and when do drakes abandon home ranges, i.e., how long are drakes available for counting as indicated pairs? Does site attachment strength change with increasing density?
  3. Mobility radii and home range sizes of lone drakes, pairs, and grouped drakes: how does pond density and availability of breeding requisites in each habitat affect home range and activity centre size? What are the daily and seasonal patterns of activity for each species and each population component? How do daily activity patterns of lone or grouped drakes affect spatial distribution and census?
  4. Yearly prelaying sex ratios of all species on the breeding grounds: what is a statistically adequate sample to describe prelaying sex ratios of a species? How long is the period of residence of unmated drakes on nesting grounds and when do they leave for moulting marshes? Do unmated drakes migrate earlier or later than mated ones?
  5. Turn-over rates of local populations and the adequate census of total seasonal populations: should indicated pair population estimates obtained in early May be added to mid-June estimates to account for delayed nesting by late migrants, yearlings, and shifting populations? Does the maximum or mean pair population counted during one 3-week period of a 2½-month-long breeding season adequately assess breeding populations actually breeding on a block or transect?
  6. The effect of nonrandom and contagious distribution patterns of ponds, pairs, lone drakes and aggregated drakes, on sampling procedures: how do periods of intense spacing activity followed by increased sociability of drakes affect pair and drake distribution and their countability on strips or blocks of habitat?
  7. Length of breeding home range residence of pairs which have lost clutches: preliminary observations indicate that some mallard pairs may remain on or near their home range for periods up to 3 weeks after their clutch is destroyed. These pairs make no attempt to renest in this interval, and may never renest, but they are enumerated as indicated breeding pairs.
As early as 1951, Murdy (1953) stressed that a number of variations of waterfowl activity patterns affected census results: ( 1 ) some pairs made only one nesting attempt, while others made several, (2) some pairs of a species completed clutches before others had settled, ( 3 ) nesting was asynchronous among species, with pintail drakes deserting hens before blue-winged teal had dispersed for nesting, (4) paired birds were more conspicuous and behaved differently than lone drakes which had deserted hens, (5) shifting of ducks from drought areas during the census interval influenced counts. These same problems are still with us today.

Thirty years after the publication of Bennett's (1938) census recommendations, waterfowl biologists are still using methods which are not being constantly challenged, improved, and tested. Field workers have commendably adapted recommended census methods to varying habitat conditions but too few of them have published or tested their counting techniques. Meaningful comparisons of results of studies by two workers using different methods, with varying error estimates, are almost impossible to make. Apparent differences of yearly estimates may be due more to variations in census methods than to actual population status (Diem and Lu, 1960). Although trend data obtained from population indices are sufficiently accurate for yearly management purposes (Crissey, 1957), trends are not adequate for precise measures and descriptions of species population dynamics.

Previous Section -- Appraisal of Factors Influencing Inventories
Return to Contents
Next Section -- Summary

Accessibility FOIA Privacy Policies and Notices

Take Pride in America logo logo U.S. Department of the Interior | U.S. Geological Survey
Page Contact Information: Webmaster
Page Last Modified: Friday, 01-Feb-2013 19:13:37 EST
Sioux Falls, SD [sdww55]