Assessing Breeding Populations of Ducks by Ground Counts

Census Methods

Because of poor visibility of pairs in the heavily vegetated parkland ponds, and because of the relatively large (10.5 square miles) study block in the grassland, two different census methods were adopted to assess the abundance of breeding pairs.

Roseneath Study Area

Dabbling ducks
Each pond on the 895-acre study block was visited a minimum of four times during a 7- to 10-day period when most of the early nesting pairs of mallards and pintails were in the laying or early incubation stages. Because of yearly variations in phenology, the census period varied, but was usually between May 5 and May 25. The exact locations on ponds of pairs, lone drakes, and groups of five or less drakes were plotted on a base map. If a pair or lone drake was observed on or near the same pond on three out of the four counts, a breeding pair of that species was "assigned" to that area. This method was similar to the one proposed by Evans and Black (1956) to test "constancy" of pond use. It is also similar to a method in use by Kirsch (in litt.) on the Woodward Study Area, North Dakota. Counts were conducted in morning and late evening hours when many hens were off their nests for recess periods. A comparable census was conducted 2 to 3 weeks later when most of the late nesting dabbler species-widgeon, gadwall, blue-winged teal, shoveler, and green-winged teal were also in the laying or early incubation stages. Again pairs were assigned to a particular pond or localized area. Where the number and species of pairs breeding in a locality was doubtful, 2 to 3 hours of observation on 4 or 5 consecutive days helped resolve the questionable count. In short, the accuracy of the census depended on an intimate seasonal knowledge of the pairs continually utilizing a localized area and on the assessment of these birds as indicated breeders. Because the study block was small and home ranges of many pairs would encompass all or parts of it (Sowls, 1955; Dzubin, 1955), I also periodically censused pairs in the quarter-sections surrounding the study area to determine populations. Pairs were arbitrarily assigned to the study area only if the drake's or pair's waiting area (Hochbaum, 1944; owls, 1955; Dzubin, 1955) was located within boundary. Censuses of a small block-type area, such as the 895-acre Roseneath Study Area, do lend themselves to close approximations of breeding pair numbers as the ponds, upland and even pair populations form an integral part of a much larger complex of habitat surrounding the block. As such, the assigned population is an estimate of the pairs utilizing ponds on the study block as waiting areas and does not include the pairs breeding in its immediate environs and using the study area ponds periodically.

Diving ducks
Early in the study I concluded that ground census of diving ducks-canvasback, redhead, lesser scaup, and ruddy ducks-utilizing various ponds would not adequately estimate breeding members. Diver pairs, except ruddy ducks, tended to aggregate on particularly deep ponds that I named "primary waiting areas" (Dzubin, 1955), and to fly to surrounding smaller ponds for nesting, feeding, and loafing. On the study area, two such congregating ponds served 15 to 25 pairs that nested on the block and ponds surrounding it. A census of pairs and lone, unmated drakes on such primary waiting ponds could not be used to estimate pairs actually breeding on the block. Canvasback and redheads have maximum home range sizes of 2 to 4 square miles. Such mobile species with large ranges do not lend themselves to adequate census on a small block. Lesser scaup are not as mobile while ruddy ducks tend to be sedentary. Some redhead hens were semi-parasitic, some were completely parasitic, while others laid normal clutches (see Weller, 1959). The distorted sex ratio in all divers and especially scaup (Bellrose et al., 1961) made counts of diver lone males to indicate pairs, meaningless. The secretive habits of ruddy ducks also made analysis of observational data difficult. Therefore, diver populations were censused through a nesting study, wherein all emergent cover was periodically searched for over-water nests. The maximum number of viable, destroyed, and deserted nests found during the peak breeding period was used to estimate the breeding population on the study block.

Several basic assumptions were made in arbitrarily assigning breeding pair numbers to the study block:

That lone drakes or pairs of dabbling ducks localize their breeding activity to one or more ponds and are consistent in their use of waiting areas. Previous studies on marked birds had shown that all breeding pairs restrict their activity during the prenesting, laying, and early incubation periods (Sowls, 1955; Dzubin, pers. obs.). However, much individual variation in activity localization within any species occurs, and species differ in hourly and daily mobility and home range sizes (Sowls, 1955; Dzubin, 1955; Evans and Black, 1956). Sowls (1955: 54-57) has reported that on a ditch near a marsh much interchange of pairs between the same waiting site occurred through the day. I noted more interchange of waiting sites under dense pair populations of the grassland than under less dense populations of the parkland.
That any ingress of pairs onto the block was counterbalanced by a similar egress of the same number and species of pairs out of the study area. This is the most difficult assumption to assess, in the light of the wide home range size of pintails, mallards, and divers. However, under the low population levels with which I worked, assumption (1), above, was considered to be valid.
That all pairs counted on the block remained there to nest and that all species present bred. I noted that pairs were occasionally displaced by other pairs or remained on the study area for 1 to 2 weeks without any nesting attempt. This was especially true of a small number of late nesting gadwall and widgeon pairs that tended to move off the area as soon as the nest site was chosen away from the study block. There was no evidence of nonbreeding pairs of any species in the parkland, except in the case of parasitic redheads.
That turnover of pairs was minimal. Hochbaum (1944: 158) and Smith and Hawkins (1948) discussed the possibility of late nesting pairs moving into an area and not being enumerated by a census conducted during one interval. Almost yearly I noted an influx of five or six mallard pairs in late May or early June. Drakes of such pairs were brightly plumaged, unlike the drab males that had been seen in the area for the previous 2 weeks. These pairs appeared to be late breeders, nesting for the first time. In the population assessment, they were not considered to be breeding pairs over and above those censused in mid-May. As I could not determine if they were new breeders or renesters, I again assumed that an equal number of renesting mallards left the area to breed elsewhere. Acceptance of this assumption probably led to underestimation of seasonal population sizes each year.
That all diver nests were located in the emergent vegetation of study area ponds. This assumption was considered valid as all emergent vegetation was checked for nests at 2-week intervals. Any diver pairs that utilized the study area as a part of their home range, and nested on a pond immediately off the area, were not counted.

At best, estimates taken from direct ground counts of pair numbers breeding on a small block of parkland habitat should be considered relatively imprecise approximations of seasonal breeding pair populations. The estimation of absolute breeding pair numbers per unit area remains an inexact technique, subject to many vagaries of species behaviour, visibility, mobility, and seasonal nesting chronology.

Coding of Population Components--Kindersley

All data were coded by a system adopted after a U.S. Bureau of Sport Fisheries and Wildlife ground-census code, i.e., 1/0 = pair, 0/1 = lone male, 3/2 = 3 pairs and 2 lone males, 3:0 = 3 grouped males, 5:5 = 5 grouped pairs, 0/F = lone female, 0:4 = 4 grouped females (W.H. Kiel, in litt.).

Although Hochbaum (1944: 85) recommended that only territorial pairs and lone, waiting drakes be censused as breeding birds, he described seven categories of ducks found in a breeding marsh from April to early July: (1) unmated ducks not yet courting, (2) unmated ducks in prenuptial courtship, (3) mated pairs, (4) novice drakes, (5) sexually active unmated males, (6) summering drakes, and (7) unsuccessful or nonbreeding females.

For counts of dabbler species on the potholes of the Kindersley Study Area the following 10 categories were adapted for use after Hochbaum's components:

Resident pairs-lone pairs on ponds, or pairs spaced over 15 feet apart along sections of shore line. These were apparently settled, dispersed nonmigratory birds.
Grouped pairs and drakes-aggregated pairs or pairs and drakes that behaved as flocks and were not spaced. They were apparently migrating individuals, not yet settled or dispersed.
Breeding-season groups-aggregations of 2 to 20 + males and one hen on ponds. Birds were assigned to either (a) "group flights associated with courtship" (GFAC), in April and early May, if males were giving displays and hens were "inciting", i.e., "spring courting flight" of Dzubin (1957); or (b) "attempted rape flights" (ARF), if drakes harassed lone hens which gave the repulsion call (Dzubin, 1957; Raitasuo, 1964; McKinney, 1965). In mallards and pintails these latter flights were seen after May 5, at a time hens start to incubate. For other species they were generally recorded after May 25. Birds in "three-bird flights" (TBF), i.e., territorial pursuits, were also noted. If the pursuing drake returned to the pond under observation, he was recorded as a lone drake. If the chased pair landed in an already censused pond, it was also recorded.
Postbreeding-season groups-aggregations of males and two or more hens which behaved as a unit. These groups were usually observed after June 15. Such flocks of drakes and pairs were considered to be in postbreeding condition and not part of the breeding population.
Lone drakes-drakes that were spaced over 15 feet from other drakes on waiting stations. These drakes were generally observed through the laying and early incubation periods. The distance separating drakes varied with phonology of season and the species.
Grouped drakes, five or less- drakes associated with other drakes in small, cohesive aggregations of two, three, four, or five. These groups were observed in mallards and pintails from late in the laying period through the mid-incubation period, April 24 through June. In other dabblers they were seen from mid- to late incubation, beginning May 20 through June. Small groups of unmated drakes, two to five in number, of all species, were occasionally recorded through April and early May.
Grouped drakes, more than five-aggregations of more than five drakes. They were usually observed in the late incubation or postbreeding periods.
Lone Hens-hens not associated with drakes. This category included hens that had just left their nests, after laying or during incubation recesses, and had not yet rejoined their drakes. They made up less than 4 per cent of any population count in May and early June but were more common in late June after drakes had left for the moulting grounds.
Grouped Hens-two or more hens in aggregations that behaved as units. They were observed in the postbreeding period after June 15 and to July 10. They included hens that had either lost clutches or had abandoned near-flying broods. Rarely would two or more incubating hens on recess from the nest be seen together. They were not considered pairs, even though their drakes may have already abandoned the home range.
Drake-hen Ratios-where divers, and occasionally migrating dabblers, were associated in loose aggregations, and pairs and lone drakes could not be separated, counts were lumped and recorded as a ratio of drakes to hens, e.g., 12:9.

Kindersley Study Area

Direct counts of all ducks were used to estimate population levels on the grassland study block Because of staggered breeding seasons and differential times of migration and nesting of each species, a number of counts were conducted through April, May, and June. Few breeding pairs or lone drakes of any species were recorded after July 15.

In 1956, the 10.5-square-mile study area was divided into three sections. All of the "indicated" pairs on the ponds of each section were counted by two men who walked together to each pond. Although an attempt was made not to flush pairs, it was unavoidable on small, open ponds where some pairs tended to flush as far as 200 to 300 yards away. Pair counts made on ponds in the western third of the study area showed that from 20 to 45 per cent of the pairs were flushed. Pairs or drakes seen to land in a pond not already tallied were subtracted from the pond totals. However, some censusing of previously counted pairs undoubtedly occurred. The magnitude of the duplication error was unknown and variable. However, the consistency of counts taken in mid-May was high, and comparison with nests found, especially for the major breeding species (i.e., mallard), showed no wide discrepancies. Counts were conducted from 0530 to 1100 hours, M.S.T., at approximately 7-day intervals from May 3 to June 11.

From 1957 to 1959, inclusive, breeding pair counts were made with binoculars or a 20x telescope from a vehicle that was driven to a point overlooking each pond. Fewer than 5 per cent of all ducks were hushed. Approximately 22 miles were driven during the census of the 10.5-square-mile block. Censuses were generally conducted on bright days when wind velocities were below 15 mph. Two censusers working together were able to survey all wet ponds between 0800 and 1200 hours. I had determined that pairs of most species were least mobile during this period. Fewer than 15 of the 114 pond basins contained dense stands of dried emergents (Carex, Scirpus, Beckmannia, or Glyceria) in which pairs could secrete themselves. It was generally after May 20 when emergent vegetation grew to a 6- to 8-inch height, and therefore no attempt was made to "beat-out" ponds. Near emergent-rimmed ponds, the slamming of the car door or sounding of a horn was sufficient to alert the birds enough to make them visible for censusing. Again counts were made at 5- to 10-day intervals from early April to mid-June.

As on the Roseneath Study Area, census of pair and drake divers was not considered an adequate measure of the number of pairs breeding on the study block. Therefore, data on breeding divers were obtained from an associated nesting study. A viable, hatched, or predator-destroyed nest was considered evidence of a breeding pair. To ensure that dabbler censuses were conducted at an optimum period, a similar nesting study was simultaneously conducted to determine breeding season chronology. Census dates were arbitrarily chosen to ensure coverage during optimum breeding periods for early nesters, mallards and pintails; intermediate nesters, widgeon and shovelers; and late nesters, blue-winged teal and gadwall.

Analyses of Census Results
Census results from each survey made during the optimum breeding interval for each species (i.e., when the greatest proportion of hens of pairs were laying, incubating, or in revesting breeding phases) were lumped and an average breeding pair figure calculated. All mallard and pintail pairs, lone drakes, grouped drakes, and grouped drakes in group flights associated with courtship and attempted rape flights, counted before May 20, were considered breeding pairs. For widgeon and shoveler, all of the above categories were considered breeding pairs if censused before June 5, while for blue-winged teal and gadwall all of the above categories were considered breeding pairs if noted before June 10. Lone hens were not added to the population counts as I assumed all hens to be paired to one of the lone or grouped drakes already enumerated on the study block. Since fewer than four lone hens of any species, or less than 4 per cent of the indicated pairs counted from 1956 to 1959, were ever encountered on any census, any error in deleting this component would be small. The average number of pairs was termed the "mean indicated breeding population".

Sex ratio data from Bellrose et al. (1961) had shown that among most dabbler species there is a preponderance of drakes immediately prior to the breeding season. I collected similar data during the prenesting interval. To ensure that pair census data were not weighted with the unmated-male segment, a correction factor using the average prelaying drake-to-hen ratio for each species was applied to the indicated population. The breeding-pair population figure was then termed the "sex-ratio corrected population". Sex ratio corrections were applied only to dabbler figures as diver populations were assessed from nesting studies. Murdy (1962) has more recently applied such corrections to spring censuses of lesser scaup and ring-necked ducks at Yellowknife, N.W.T.

In 1956 and 1959, in order to test stability of population. and reproducibility of census counts, four or five censuses were made during the period when the greatest proportion of pairs were in the prenesting, laying, or incubation periods. In 1959, periodic censuses were conducted three times a day on May 11, 15, and 16 to determine daily variability in the census and percentages of pairs, lone males, and grouped males during any one time interval.

Definitions
To clarify further what population components were used in breeding-pair counts, the following definitions will be followed throughout this paper:

Assigned breeding population-The dabbler breeding population assigned to a unit of habitat, usually 160 acres. The population was determined by plotting the location of pairs and lone drakes on a map during four or five censuses. If a pair or lone drake of any species was seen on a particular pond three or four times, it was assigned to that pond or quarter-section.
Indicated breeding population-The population estimated from counts of various components.
1. Mallards and pintails: Prior to May 20, all lone males, pairs, all grouped males, and males in aerial flights temporarily on ponds censused were considered pairs. Thereafter, all lone males, pairs, grouped drakes of five or less, and aerial flights temporarily on ponds were considered evidence of breeding pairs. After May 20, groups of six or more males were considered postbreeding birds.
2. Widgeon and shoveler: Prior to June 5, all lone males, pairs, all grouped males, and all males in aerial flights on ponds censused were considered pairs. Thereafter, only grouped males of five or less in number, lone drakes, drakes in aerial flights, and pairs were considered breeding pairs. Groups of mated males, usually in two's and three's, were rarely observed until mid- or late incubation. Groups of unmated males were, however, more common.
3. Gadwall and blue-winged teal: Same as widgeon and shoveler except cut-off date of June 10 was used. Again, associations of mated males were uncommon until mid-incubation.
4. Divers: All viable, hatched, destroyed, or deserted nests found, which were initiated prior to June 15, were considered evidence of a breeding pair unless field observation indicated a nest to be a renest. With ruddy ducks the cut-off date was extended to June 20.
Sex-ratio corrected population-This was the mean indicated breeding population taken during four or five censuses, during the time most breeding pairs were in the prelaying, renesting, laying, or incubation stages, to which a sex-ratio correction factor as found in Table 3 was applied. The correction factor was determined from spring counts of the sex ratio of each species before the first eggs were found. It reduced the indicated pair population by the proportion of unmated drakes found in the species. The sex-ratio corrected population was considered to be the best estimate of the absolute breeding population of a study area.

Previous Section -- Study Areas
Return to Contents
Next Section -- Migration, Inventory, and Population Components