Assessing Breeding Populations of Ducks by Ground Counts

Migration, Inventory, and Population Components

Spring Arrival Dates

Most dabbler species arrived at both the Roseneath and Kindersley districts before the diving ducks (Table 2). Of the dabblers, pintails were the first species noted, usually by the last week in March. They were followed by mallards, widgeon, green-winged teal, shoveler, gadwall, and blue-winged teal during the first to third week of April. Redheads and canvasback arrived at about the same time as green-winged teal, and shovelers were usually observed 1 to 4 days before lesser scaup. Ruddy ducks and white-winged scoters, which rarely nested at Kindersley, were the latest species to migrate. The first migrant pairs of pintails and mallards were associated with the first appearance of snow-water pools in fields. Arrival dates were about 1 week earlier at Kindersley than at Roseneath. Migrants arrived at Roseneath 1 to 2 weeks later than dates given for southern Alberta by Keith (1961: 42) and for Delta, Manitoba, by Sowls (1955: 12). The species sequence is about the same for all areas. Arrival dates of all species given by Ellig (1955: 11) for Montana are generally 1 to 2 weeks earlier than the Kindersley arrivals. Keith's arrival dates for all species in 1956 are 5 to 8 days earlier than dates from Kindersley.

Peak influxes of all species were generally 1 to 2 weeks later than first arrivals. Major migrations were associated with favourable weather, i.e., southerly winds and temperatures above 30°F. Prolonged April cold periods tended to dampen movements until early May. In the cold spring of 1954, Gollop and Lynch (1954: 47) recorded flocked mallards and pintails as late as May 10, after which they dispersed. Gollop (1954: 65) has also described the delaying effect of a mid-April cold snap on migration and nesting in the Kindersley district. Except for exceptionally warm springs with few intervening cold snaps when migrants moved into the study areas, en masse, over 7 to 10 days, e.g., 1955, two or three influxes of migrants occurred over a 3-week period. Few migrants were recorded as late as 30 days after the first arrivals were noted. At Kindersley from 1956 to 1959, most mallards and pintail pairs were settled by May 5 whereas all other species, except lesser scaup and ruddies, terminated migration by May 20. At Roseneath, a few migrating mallard flocks were recorded as late as May 12. The last migrant gadwall and blue-winged teal were noted by May 25 while a small number of lesser scaup moved through until early June.

Sex ratios

Sex ratios taken on the Kindersley Study Area before the first clutches were found showed only superficial differences among all dabbler species (Table 3). Utilizing the binomial probability distribution (Steele and Torrie, 1960), I found no significant difference (p = >0.05) in the sex ratio means among all seven dabblers. Similarly, no significant difference existed among the four diver sex ratio means. Yearly comparisons were not particularlly valid because of small sample sizes. However, there was a significant difference (p = <0.05) between years for lumped samples of lesser scaup, i.e., 1956 plus 1957 vs. 1958 plus 1959; the percentage of males ±95 per cent confidence interval was 56.8 ± 2.6 and 63.3 ± 1.6. For all species (Table 3), the sex ratios do not differ significantly (p = >0.05) from those shown for Manitoba by Bellrose et al. (1961: 416). I also found no significant difference (p = >0.05) for all species between the sex ratios gathered at Kindersley and those presented by Bellrose et al. (1961: 428) for the mid-continent breeding grounds, except for lesser scaup, i.e., 61.1 ± 1.4 vs. 66.1 ± 2.1, and assuming sample sizes in Bellrose et al. were in the order of 2,000.

On an Alberta grassland breeding ground, sex ratios for five late nesting species (Keith, 1961: 43) were slightly higher for gadwall (113:100), blue-winged teal (144:100), lesser scaup (163: 100) and ruddy duck (203:100) but lower for redhead (127:100) than comparable data in Table 3. None of these ratios was significantly different (p = >0.05) from those taken at Kindersley.

All dabblers including mallards and pintails, the most common breeders at Kindersley, did not show any significant departure (p = >0.05) from a 50:50 ratio. Yet, Bellrose et al. (1961) showed a consistent preponderance of drakes in spring counts taken of these two species throughout North America. Other authors have also shown consistent deviations favouring males. Sex ratio means for mallards and pintailsfrom Kindersley are close to those given for Delta, Manitoba, by Hochbaum (1944), i.e., 102:100 for mallards and 109:100 for pintails. For Montana, Ellig (1955: 12) gave ratios of 127:100 for mallards and 107:100 for pintails, in birds censused prior to April 14, 1952. Sowls (1955: 24), summarizing early spring mallard and pintail sex ratios from Delta, gave a mean ratio of 108:100, in favour of males for both species. Counts made by Bue (in Bellrose et al., 1961: 418) in eastern South Dakota prior to April 15, 1950 and 1951 show percentages of mallard drakes of about 53 per cent (113: 100) and pintails of 57 per cent (132: 100).

Spring sex ratios for dabblers and divers have been published by a number of other authors. Comparisons of published figures with the data shown in Table 3 are not particularly meaningful as published ratios were taken through the spring migration period and are not confined to the breeding grounds. Sex ratio data for many waterfowl species were presented in Bennett (1938), Furniss (1938), Erickson (1943), Hochbaum (1944), Beer (1945), Low (1945), Sowls (1955), Ellig (1955), Johnsgard and Buss (1956), and Moyle (1964). A discussion of the errors involved in gathering and comparing "piece-meal" sex ratio counts is given by Bellrose et al. (1961). The Kindersley 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.

Differential Sex Ratios and Migration

Bellrose et al. (1961: 412-416) discussed changes in sex ratio related to times of northward migration in many species. Sex ratios are generally more unbalanced in favour of drakes in late spring than in early spring. In the present study with its small samples no such marked seasonal changes in prebreeding sex ratios were noted with any species except lesser scaup. In 1958 when only three lesser soaup nested on the Kindersley area and in 1959 when no breeding was observed, sex ratios taken before May 15 were significantly different (chi square, p = <0.05) from those collected in late May and through June (Table 4). The data, although not affected by increasing number of hens nesting, show an increase in proportion of migrant drakes through early and mid-June. The change may reflect an increasing number of drakes abandoning early nesting hens away from the study block or may suggest that unmated males migrated later, perhaps remaining farther south than mated males.

Changes in sex ratios in favour of males as spring migration progresses have been reported for widgeon and shoveler by Erickson (1943), for green-winged teal by Beer (1945), and for pintail and lesser scaup by Hammond (in Bellrose et al., 1961: 402). In Manitoba, Hochbaum (1944: 16) noted that although April sex ratios were nearly balanced in mallards and pintails there was an influx of unmated drakes into the marsh in early May, as evidenced by large numbers of "courting parties". In Illinois, Bellrose et al. (1961: 414) report that sex ratios were heavily weighted to drakes in mallard, pintail, canvasback, and ring-necked ducks in late February but that the preponderance of drakes declined in March and April. Lesser scaup sex ratio counts given for Manitoba by Kiel (in Bellrose et al., 1961: 416) do not show an upward swing in drakes through June as noted in the present study. However, local differences in sex ratios determined from small samples may not reflect similar changes in the population as a whole over the entire migration or breeding habitat.

Species Composition-Roseneath

Mallard, blue-winged teal, and widgeon were the predominant dabbler species, making up an average of 82 per cent of the 105 breeding pairs (Table 5). Ruddy ducks and canvasbacks were the major diving duck species nesting on the area, comprising an average of 75 per cent of the 28 breeding diver pairs. Green-winged teal, gadwall, and lesser scaup were the least numerous breeding species. Mallards dropped from 54 to 33 pairs over the 4-year span of the study, while blue-winged teal and pintails showed somewhat erratic yearly fluctuations. Major increases in numbers were noted for ruddy ducks, with a recorded increase from 4 breeding pairs in 1952 to 15 pairs in 1955. The ring-necked duck was an uncommon breeder in the area; one pair established a nest in 1954. Species composition of the study area and pairs observed per square mile are similar to those given for the same block by Evans, Hawkins, and Marshall (1952: 38) for the 1951 breeding season. In their study, of the 88 breeding pairs recorded per square mile, mallard, blue-winged teal, and canvasback were again the most common species. For the entire Newdale-Erickson district of west-central Manitoba, Kiel (1951: 56) showed the blue-winged teal, mallard, and lesser scaup as the three most numerous breeding species. For the same area, Pospichal, Cram, and Parsons (1954: 86, 87) showed that from 1949 to 1954, mallard was the predominant breeder followed by blue-winged teal, lesser scaup, and pintail.

Species Composition-Kindersley

On the grassland study block, mallard and pintail were the predominant breeding species (Table 6). They made up 72 per cent of the mean breeding population for the 4-year period. Dabblers made up 97 per cent of the entire population while divers, chiefly lesser scaup, made up the remainder. No ruddy ducks were found breeding on the area, while one nesting pair of white-winged seaters was recorded in 1958. This area contained an average of 52 breeding pairs per square mile, in contrast to the average of 95 pairs per square mile at the Roseneath study area (Table 5). Although 2 to 12 pairs of white-winged scorers were regularly censused on a 90-acre pond in late May and early June 1956 through 1959, only one pair was recorded nesting near this pond in June 1958.

Major yearly fluctuations in pair numbers occurred with all species. Peak numbers of mallard and pintail pairs (358 and 269, respectively) I were found on the area in 1957, a year of drought I which probably forced many pairs onto the relatively well-watered study block from the surrounding drought-stricken regions. While mallards and pintails increased in abundance in 1957, other breeding species decreased. Decreases in breeding pair numbers occurred generally in all species in 1958 and 1959, associated with continuing drought and poor production of young.

Gollop (1954: 71) conducted pair surveys on a sample of 20.5 square miles in the Kindersley-Eston district and determined indicated breeding populations of 75 pairs per square mile in 1952 and 40 pairs per square mile in 1953. Pintails, mallards, shovelers, and blue-winged teal were the predominant dabbler species recorded, while lesser scaup and canvasback made up the greatest portion of divers.

Indicated Pairs and Nonbreeding in Divers

As previously discussed, there is some difficulty in assigning diver pairs to a study area as indicated breeding pairs from ground census. Also, there is no way to differentiate migrating pairs from residents and nonbreeding from breeding pairs (Smith and Hawkins, 1948; Ellig, 1955). A comparison of the mean indicated population with numbers of nesting pairs found on the 10.5-square-mile block showed that only 48 per cent of the indicated population of diver pairs nested in 1956, 38 per cent in 1957, 39 per cent in 1958, and 3 per cent in 1959 (Table 7). Most of the breeders were lesser scaup. Only one canvasback and six redhead pairs nested, while no ruddy duck nests were found. I concluded that indicated pair populations taken only from pair counts (i.e., omitting all lone or grouped males) during the period when divers should be nesting do not adequately approximate pair numbers actually nesting. Some of the indicated pairs were undoubtedly late migrants, especially ruddy ducks, but observations showed that many pairs of lesser scaup, and to a lesser extent, canvasback and redhead, remained on the area through late June without making any attempt to nest.

Some of the lesser scaup censused may have been nonbreeding yearlings, although McKnight and Buss (1962), after histologically examining 16 ovaries, concluded that most, but not all, yearlings are physiologically capable of breeding. In Manitoba nonbreeding in lesser scaup has been associated with deteriorating habitat conditions and nonflooding of nesting cover (Rogers, 1964). Rogers (1963) also noted that the proportion of hen scaup nesting around four intensively studied potholes decreased from 64 per cent in 1958 to 8 per cent in 1959. In 1960 with a recovery of water levels, 60 per cent of the resident pairs nested. Each year some of the hens failed to nest. In Montana, Smith (1953: 286) also noted an absence of lesser scaup broods in late summer even though pairs were present earlier on his study reservoirs. Similarly, in British Columbia, Munro (1941) suspected that a proportion of the lesser scaup population did not breed but retained their bright breeding plumage into mid-summer. I suggest that, owing to poor habitat conditions, deteriorating water levels, and decrease in pothole numbers, many divers also became nonbreeders in the Kindersley area from 1957 through 1959, with the lowest ratio of nesting pairs to indicated pairs in 1959.

Nonbreeding in Dabblers

Nonbreeding in mallards and pintails has been reported for Alberta by Smith (1961) who concluded that pairs did not breed under deteriorating water conditions due to "physiological and psychological shock". I obtained no direct evidence for nonbreeding of dabbler species because of the difficulty in separating breeders from nonbreeders. However, in 1957, of some 358 indicated mallard pairs censused on the Kindersley Study Area, only 300 nests were located during two complete "beat-outs" of the upland nesting cover. Approximately 20 more nests were estimated to be in stubble fields and fence rows. Either the coverage of nesting habitat was less than 100 per cent efficient, or some pairs nested at over 1 mile from the study block, or 35 to 40 of the immigrant pairs failed to nest. Also, in 1959, a periodic census of gadwall and widgeon pairs indicated that as many as 20 of the 35 widgeon pairs and 14 of the 23 gadwall pairs failed to nest. Loose aggregations of these pairs were seen on two ponds through late May and early June with no evidence of dispersion or laying. Six gadwall hens were collected from flocked pairs outside the study area on June 5. On internal examination four of the hens' ovaries showed ova in various stages of atresia, with no evidence of ovulated follicles present. The other two had apparently attempted to lay, as regressed follicles were noted. The mechanisms involved in nonbreeding under conditions of poor quality habitat and high pair densities are not known.

Breeding Season Dynamics of Dabblers-1958

Periodic censuses conducted from 0800 to 1200 hours on the Kindersley Study Area throughout the 1958 breeding season showed wide changes in pair, lone drake, and grouped drake categories of each dabbler species (Tables 8a to 8f). A nesting study conducted concurrently gave known reference points for start of nesting, peak laying, first broods, and peak of hatching. The major difference between the early nesting mallard and pintail population components (Tables 8a and 8b, respectively) and later nesting widgeon (Table 8c), shoveler (Table 8d), gadwall (Table 8e), and blue-winged teal (Table 8f) is the near absence of a grouped drake component in the last four species until the mid- or late incubation period. Also, no major posthatching influxes of the last nesting species were recorded.

A graphic comparison of one example of an early nester, mallard, with an intermediate nester, widgeon, shows the lack of grouped widgeon males until the June 12 count, while mallard groupings are evidenced as early as April 24 (Figure 3). Optimum census periods overlapped in the two species during mid-May.

Figure 3. April to June population components from periodic censuses on the Kindersley Study Area, 1958. (Note the lack of grouped male widgeons but the prevalence of grouped male mallards.)

Graphs displaying the number of mallards and widgeons in the
categories of pairs, lone, and grouped on the various census dates.

Mallard
Even before the start of egg laying between April 12 and 15, a number of lone and grouped drake mallards were observed (Table 8a). These were probably unmated males. On April 29 the ratio of pairs to lone males was about 1:1, but thereafter pair numbers tended to decrease and lone males increased until a ratio of 1:1.5 was recorded at the hatching peak, about May 21. The grouped drake component increased during this 22-day interval as more hens started to incubate.

During the peak laying interval of April 24 to 29, most grouped drakes were found in aggregations of two. As the incubation period continued, more groups of three to five were observed. During early May when first nesters were incubating, few aggregations of more than five drakes were seen. Before May 21 only three groups of more than five males were observed, one each on April 29, May 10 and May 21. During the censuses prior to May 24 no group flights associated with courtship or attempted rape flights were recorded, although a number were noted on the area in the afternoons when laying and incubating hens returned to waiting areas.

The validity of enumerating the grouped drake component, prior to the main hatching peak, with the recorded pair and lone male components as indicative of the breeding pair population, was substantiated by the seasonal censuses. Indicated breeding populations, taken from an enumeration of the pair, lone drake, and grouped drake components before the hatching peak, were consistent and varied between 182 and 212 pairs (Table 8a). After the peak of hatching between May 21 and 24, the number of indicated pairs based on grouped drake counts rose and there was no way in which resident grouped males and transient males which had left their breeding home ranges elsewhere could be differentiated. A marked influx of grouped males in aggregations of over 5 and up to 40 was noted after May 31. Such postbreeding aggregations of drakes and hens were recorded until the end of June.

From the present data and complementary data on marked pairs, I concluded that drake mallards congregated while their hens were in the last stages of laying or in the first stages of incubation. Further, they remained on the breeding grounds, although not necessarily on the waiting site, until the second or third week of incubation (first nesting only). Small aggregations of males were associated with the waning of the drake-hen pair bond primarily through the incubation period.

Because some pairs continued to renest through June, the indicated population, after the first influx of postbreeding groups on May 31, was considered to include the pairs, lone drakes, and only those groups of males five or less in number. Observations of marked drakes from pairs nesting for the first time had indicated that mated males rarely congregated into groups of five or more before their hens were in the third or fourth week of incubation. Therefore, an arbitrary division was made to include only groups of five or less after the hatching peak. A similar decision was made by Jessen, Lindmeier, and Farmes (1964) and by Hammond (1966).

The lone female component made up less than 3 per cent of the total indicated pair count for all May censuses, i.e., rarely were more than two lone hens recorded per species per census. A few more lone hens were recorded in June after drakes had abandoned their home ranges and left the study area, but the maximum number noted during one census was five pintail. Only five lone mallard hens were recorded, one from each census from May 3 to 31. Smith (1956: 36) reported that lone hens of each species made up less than 4 per cent of the populations on four Alberta study areas. At Kindersley the first small postbreeding groupings of predominantly hens were recorded by June 12. They were usually associated with a number of males. Small flocks of hens from 3 to 38 were periodically noted in the district during the first 2 weeks of July. I assumed these were hens that had lost renest clutches or abandoned nearly flying young and were retiring to the moulting lakes.

Seasonal sex ratios of mallards recorded on the study area showed a progressive increase in the percentage of drakes, as absences by laying and incubating hens reduced the proportion of visible hens. Ten days after the hatching peak, 90 per cent of the population was made up of drakes. Thereafter, a downward trend in percentage of drakes was noted as more drakes left the breeding grounds and postbreeding groups of hens moved into the region. By examination of sex ratio data from census and a comparison with prebreeding sex ratios, time of laying and incubation can be deduced. The first appearance of lone drakes, i.e., when lone drakes compose over 10 per cent of the indicated pairs, is invariably a good indicator of start of laying by hens, while the first appearance of groups of two or three drakes indicates start of incubation.

Pintail
The seasonal changes in population components paralleled those of mallards (Table 8b). Lone drakes made up a small proportion of the prebreeding population prior to April 12, the start of laying, but the number grew as more nests were initiated. Grouping of males was most evident as soon as incubation started, although prebreeding association between mated and unmated males is much more common in this species than in mallards (Smith, 1963). A major difference between the two species is the frequent association of pintail drakes with pairs after April 19, i.e., in groups of three or more males and one hen. These were primarily composed of males harassing females, temporarily on ponds, in attempted rape flights. No pintail pairs were recorded after June 6, indicating an earlier abandonment of home ranges by renesting individuals than by mallards. As with mallards, a major influx of postbreeding grouped males was noted after May 24. The indicated breeding population based on pairs, lone males, and grouped males from April 24 to May 21 inclusive (the optimum census interval) varied between 150 and 181 pairs.

The percentage of drakes in the breeding population increased to 88 per cent 10 days after the peak of hatching, but did not decrease through June, as in mallards, as no postbreeding groups of hens moved into the area.

Widgeon
The indicated breeding pair population was composed of pair and lone drakes for the greatest portion of the breeding season (Table 8c, Figure 3). The lack of the grouped male category, until after the hatching peak, reflects the stronger pair bond and site attachment in this species than in mallards and pintails.

Two to seven lone drakes were observed on each census prior to the start of egg laying on May 8, and a group of three unmated males was recorded only once on April 12. Lone drakes became more common after May 10 with the peak of laying reached between May 16 and 20. Sex ratios became progressively weighted to males as more hens started laying and initiated incubation. The percentage of drakes reached a maximum of 83 per cent some 10 days after the hatching peak. Few males apparently remained on the study area for more than a week after the hatching period. There was also no major influx of grouped widgeon drakes into the area as was recorded for mallards and pintails.

Shoveler
Most shovelers were observed as either pairs or lone drakes until midway through the incubation period (Table 8d). Eight lone males were observed on May 3, the first indication of laying in the species. Groups of males were first recorded on May 24, 2.5 weeks before the hatching peak. The percentage of males increased after May 10, and reached a maximum of 95 per cent after the hatching peak.

Gadwall
The seasonal population components are similar to those of widgeon. Only one lone male was observed on each count taken prior to May 21 whereas after the first laying commenced, about May 16, the number of lone males steadily increased (Table 8e). Grouped males were not observed in morning counts until just before the hatching peak, again indicating a strong pair bond and site attachment in this species. However, several neck-banded males were observed to associate with other males, for varying periods of the day, during the third and fourth week of incubation. Sex ratios in favour of drakes increased after laying started and peaked at 90 per cent males, 7 days after the hatching peak. No large influx of postbreeding grouped males was recorded in June.

Blue-winged teal
The pair and lone drake component made up the greatest proportion of the indicated pair population until the hatching peak (Table 8f). Seven groups of males, primarily of two, were noted 10 days before the hatching peak. The first nest was recorded on May 12 and only one group of two males was observed on May 21. Four aggregated males were seen on May 31. These may have been associations of mated and unmated drakes or of unmated drakes only. As with other species the percentage of males censused in the population progressively increased until a high of 89 per cent males was noted on June 6, 10 days before the hatching peak.

Optimum Census Periods

For all 1958 counts, plottings of weekly, indicated pair populations show wide seasonal variability (Fig. 4). Most dabbler species showed an ever increasing number of pairs in residence until the beginning of nesting, when a relatively stable "plateau" of pair numbers occurred. This plateau can be correlated with the initiation of laying and extends for 3 to 4 weeks while other hens start laying and incubating. Optimum census periods of the early nesters, mallard and pintail, overlapped with the intermediate nesters, widgeon and shoveler, but did not overlap with the late nesters, gadwall and blue-winged teal (Fig. 4). For the 1958 breeding season the optimum census period was April 24 to May 21 for early nesters, May 10 to June 6 for intermediate nesters, and May 24 to June 12 for late nesters.

Figure 4. Seasonal indicated pair numbers of six dabbler species. (Note the differences in optimum census periods reflecting variations in time of nesting.)

GIF -- Graph showing the number of indicated pairs of mallards,
pintails, widgeons, shovelers, gadwalls, and blue-winged teals over a three month period.

The optimum census periods can be a week earlier or 2 weeks later than dates given above because of yearly variations in spring break-up, migration of species, and dates of nest initiation. Late April cold periods affect migration and nest-initiation dates and may lead to double peaks of hatching in late May and early June. Adequate censuses during such years are extremely difficult as mallard and pintail drakes from early nestings have abandoned home ranges when late nestings are only being initiated.

Estimates Based on Different Components

Various authors (Hochbaum, 1944; Sowls, 1947; Williams, 1948; Smith and Hawkins, 1948; and others) have recommended that only the drake and pair components of a censused population be considered indicated breeding pairs. At Kindersley, a wide discrepancy in pair estimates occurred with two species, mallard and pintail, if only these two components were considered (Table 9). A comparison of indicated pair populations of mallards and pintails taken from a single count on May 10 with a mean sex-ratio corrected population (cf. Methods section) taken from five counts suggested that by enumerating lone males and pairs only 52 per cent of the estimated breeding population of mallards and 54 per cent of the pintail pairs were assessed. Chronology of nesting markedly affects the com- ponent parts. During the optimum census period, mallard lone drakes plus pairs made up 158 of the 182 (or 87 per cent) indicated pairs counted on April 24, but had dropped to only 101 of the 212 (or 48 per cent) indicated pairs noted on May 21 (Table 8a). The pintail lone drake plus pair components made up 135 of the 174 (or 78 per cent) indicated pairs on April 24, but had dropped to 76 of 150 (51 per cent) indicated pairs on May 21 (Table 8b).

Breeding population estimates of other dabbler species show less distortion when only pair and lone drake components are considered because few drakes (primarily unmated ones) associate with each other until after mid-incubation. However, at Kindersley, population levels of all other dabbler species, viz., widgeon, shoveler, gadwall, and blue-winged teal, were considerably lower than those recorded for mallard and pintail. Where pair populations occur at densities in excess of five pairs per square mile (Table 5) drakes may associate much earlier in the incubation period into small groups of two's and three's. Such groups should be enumerated as indicated pairs, and the sex-ratio correction factor applied to delete any groups of unmated males.

Daily Change in Population Components

By hourly observation, I established that variations occur in component parts of a population breeding around a single pond. Censuses conducted during five intervals of the day also showed this variation (Table 10). Counts of indicated mallard pairs conducted after 0530 hours show pairs making up but one-quarter of the population whereas comparable counts started at 0800 hours show 46 per cent pairs, at 1300 hours 57 per cent pairs, at 1530 hours 63 per cent pairs, and at 1800 hours 69 per cent pairs (Table 10). On May 16, only 18 mallard pairs were recorded on the 0530 hour count while 42 were noted on the 0800 hour count. I concluded that most pairs were visiting nesting cover or feeding on upland grain fields in the early morning and were not seen on ponds. Other observations on general daily activity of pairs confirmed this view. At 0530 hours when many hens would be laying, the lone drake component of the population was 30 per cent. It decreased from 23 per cent at the 0800 hour count to 18 per cent at both 1300 and 1530 hours and was only 15 per cent at the 1800 hour count. As the counts were taken between May 11 and 16, when many of the hens were either laying or incubating, I concluded that there is a simple change of component parts from lone drakes and grouped drakes to pairs progressively throughout the day as more hens return to waiting sites from laying and as more hens take their recesses from incubation in late afternoons.

Mallard hens lay most of their eggs during the morning (Hochbaum, 1944; Sowls, 1955; Dzubin, MS.). The time spent on the nest per egg is highly variable between successive eggs and among hens. The shortest time I recorded was 2 hours 3 minutes on the nest, while the longest was 13 hours 8 minutes. A few hens may also fly to nest sites in evening and remain on the nest overnight. Incubating hens also vary in the times recesses are taken, but most take an afternoon rest period. Peak recess times occurred between 0300 and 0600 hours and 1500 and 1900 hours (Dzubin MS.). The average length of morning recesses in May was about 47 minutes, (N = 71) while late afternoon recess lengths averaged 89 minutes (N = 200). After leaving the nest, a laying mallard hen flies to the waiting area of her drake. Hens in early and mid-incubation periods, i.e., up to 18 days, also continue to fly to the activity centre of the home range and rejoin the drake. The exact date drakes leave hens and abandon home ranges varies with the individual pair (McKinney, 1965) and with the phonology of the season. During recesses, after mid-incubation, hens may not return to the activity centre to rejoin the drake but may take recesses elsewhere on the home range. The pair bond may be completely broken at this time. Hens feed alone or, uncommonly, join other hens on recess.

Censuses conducted in the morning, when most hens were laying or incubating, showed a greater preponderance of lone males and grouped males than those taken in the afternoon and evening. As a greater proportion of hens start to incubate, fewer pairs will be observed in morning counts while late afternoon and evening counts will again show a greater percentage of pairs, reflecting the re-establishment of pair bonds by incubating hens with their waiting drakes.

A comparison of ratios of pairs: lone drakes: grouped drakes showed some major changes through the day. Counts made after 0530 hours showed the greatest proportion of grouped drakes with progressively fewer noted at 0800, 1300, and 1530 hours (Table 10). The lowest percentage of grouped drakes was noted in the 1800-hour count. Drake aggregations tend to disperse in late afternoon when incubating hens take their recesses. Drakes return to waiting area sites periodically throughout the day but are more commonly seen as lone drakes in late afternoon hours. The ratio of pairs to lone drakes to grouped drakes noted from periodic counts taken through the day can be used to crudely determine percentages of population laying and incubating. Since the time laying hens spend on nests varies and since mallard hens have one or two recesses a day, precise numbers of hens in each reproductive stage cannot be accurately determined from pair to drake ratios. However, proportions of pairs observed in the 0800- to 1200-hour interval when compared to the 1200- to 1530-hour interval can be used as an index to pairs laying, as most hens lay eggs before noon. In morning counts a large proportion of lone drakes, of the total indicated population, is good evidence of the early laying period for their hens. Associations of two or three drakes may denote late laying and early incubation period while groupings of four, five, or more drakes suggest mid- to late incubation or postbreeding periods. The use of changes in component parts would become more complicated in areas where high nest losses led to enumeration of many renesting pairs throughout the day. The seasonal changes in component parts as assessed from comparable morning counts have been previously shown for mallards (Table 8a and Fig. 3) and other species (Tables 8b to 8f). Daily and seasonal changes in population components of mallards and pintails were similar in that drakes form associations with other drakes during the laying and early incubation periods, while other dabbler males do not show the same degree of association until later in the incubation period. Under the low densities studied most widgeon, shoveler, gadwall, and blue-winged teal drakes were enumerated as lone males until 7 to 10 days before the hatching period.

Seasonal Variability of Walking and Vehicle Census

Walking census
Population estimates of six species taken from a series of four or five walking censuses showed wide variability between each count (Table 11). All estimates were made at a period when the greatest proportion of the mallard and pintail population was known to be in the prebreeding (i.e., pairs spaced and showing activity localization) or breeding period with the remaining species in the migration (i.e., pairs grouped) and postmigration (i.e., pairs spaced but not showing activity localization) period. Estimates of the indicated breeding population of mallards on the Kindersley Study Area were the most consistent, showing a coefficient of variability of 4.8 per cent while shoveler estimates were least consistent with a coefficient of variability of 31 per cent. It should be recognized that consistency need not reflect constancy of population or accuracy of counts. A balance between egress and ingress on the area may be occurring, with the same pairs not being enumerated during each count if turnover is constant. Balanced turnover rates would occur rarely. Biases may also be consistent.

For a breeding duck census period, five conditions are desirable: (1) that the population is resident and not migrating; (2) that no pairs move into the area during the census interval; (3) that approximately the same number of birds are flushed and duplicate counts are minimal; (4) that there is no influx of mated or unmated lone drakes onto the study area; (5) that mortality is not removing part of the population during the census period. Assumption (1) was invalid for all species except mallard and possibly pintail. There was no way of assessing assumption (2), but perusal of the data suggested that a portion of the pintails enumerated on May 3 and the shovelers on May 28 were migrants, as the indicated population showed peaks at this time. These peaks tended to increase the variances. Furthermore, mobility and home range size of pintails, their lack of strong site attachment and their erratic long-distance wanderings could have posed a sampling error. There is, however, some consistency of estimates for the mallard, pintail, widgeon, and shoveler taken on 3 successive days. The low densities of widgeon, shoveler, gadwall, and blue-winged teal on the 10.5-square-mile block may have also affected the variability. Under low densities home ranges may be larger. There might be more pairs with home ranges partly off the study area leading to greater variability of counts.

Vehicle census
In 1958, pair counts showed increased precision of estimates for pintail and shoveler but decreased precision of estimates for widgeon, gadwall, and blue-winged teal (Table 12). The coefficient of variability was almost the same in mallards for the two counting methods, viz., 4.8 per cent for walking counts versus 5.4 per cent for vehicle counts. As the same population was not sampled and a 2-year span separated the two series of counts, the data were inconclusive as to what method of census showed the least variation. Other sources of sampling error may tend to increase variability, masking any differences attributable to the two methods. Factors such as water and vegetation condition, population level, and phase of breeding season were not weighed and they may colour any valid conclusions. Because fewer birds were flushed by the vehicle, one might expect higher rates of precision with this method. The 1958 estimates from vehicle counts for mallards and pintails showed marked consistencies of pair estimates for the April 24 to May 21 interval. The data for these two species tended to substantiate the view that each population reaches a plateau of numbers for a 3- to 4-week period every year. Again, low densities of the other four species may have affected variability as there was little consistency among counts.

Time of Day and Variability of Estimates

In 1959, population estimates from 11 vehicle counts taken on 3 days of 1 week showed wide variation (Table 13, middle). The coefficient of variability was lowest for mallards, 14.6 per cent, and highest for blue-winged teal, 38.3 per cent. Five mid-day counts arbitrarily chosen from censuses taken after 0800, 1300, and 1530 hours showed lower coefficients of variability for all species except shoveler, when compared to the variation of the 11 counts (Table 13, bottom). I observed that pairs left the census ponds in early morning to fly to nesting cover. Many were not on ponds during the census period. In the evening and to a lesser extent in the morning, a number of mallard, pintail, and widgeon pairs fed in grain-stubble fields and were again not available on ponds for census. Therefore, estimates made from counts between 0800 and 1530 hours were probably more representative of the absolute breeding population than estimates made before or after these times. Population estimates of blue-winged teal were low for May 11 but much higher on May 15 and 16, indicating an influx of birds in this interval. Such migratory influxes naturally increase variance of estimates. For all species except mallard and pintail, the wide variation in counts taken after 0800, 1300, and 1530 hours suggested that all counted pairs were not resident or were nonbreeders with no firm site attachments, moving on and off the study block at various periods of the day and over the 5-day period.

An analysis of variance of dabbler counts for 0800, 1300, and 1800 hours of 3 days in 1959 showed that there was a significant difference (p = < .01) in the variances of shoveler and gadwall estimates (Table 14). There was no difference existing between the variances of the morning, mid-day and late afternoon pair estimates of mallard, pintail, widgeon, and blue-winged teal. Even so, the test can be biologically misinterpreted since field observations showed that mallards and pintails, especially, are prone to leave ponds after 1800 hours and be found in wheat-stubble fields. Replication of counts over a longer period would better corroborate whether time of day has an influence on countableness. On two days, May 15 and 16, counts of mallards and pintails started at 1800 hours were lower than mid-day counts. The May 11 counts do not show this decrease but point out the need for further extensive series of replicate daily and hourly counts. The present data are too few for valid conclusions.

Counts should be conducted at times when pairs of all species are most regularly found on ponds and not in nesting cover or fields, i.e., between 0800 and 1800 hours. Other supporting data show that wind velocity generally increased in the afternoons. Also, more laying and incubating hens left clutches and returned to waiting sites after the noon hour. The return of hens on recess to ponds invariably led to increased pair contacts, chases, and mobility. For these reasons counts night better be restricted to the 0800- to 1200-hour interval when pairs and drakes are least mobile and most likely to be found on ponds.

Pond Numbers and Breeding Pair Populations

Discussions of the correlation between spring mallard breeding populations and May or July pond numbers have been presented by a number of authors (Evans and Black, 1956; Bellrose, Scott, Hawkins, and Low, 1961; Salyer, 1962; Lynch, Evans, and Conover, 1963; Crissey, 1963a, 1963b, 1969; Drewien, 1967). Evans and Black (1956), Drewien (1967), and Bellrose et al. (1961) show strong positive correlations between May pond numbers and breeding populations of blue-winged teal and mallards, while Crissey (1968) has shown a significant correlation between July pond numbers and number of young mallards produced, and also the subsequent spring breeding population. I have pointed out that supplementary data on pond quality, size, and densities should also be considered in any such correlation attempts (Dzubin, 1969). Little is yet known of the effects of social interactions of duck, pairs in limiting breeding population levels or whether "saturation points" of waterfowl occupancy are yearly reached or exceeded on habitat units in which water levels and pond numbers are constantly changing. Over-harvests of local populations (Moyle, 1964) and shifts in populations from one waterfowl stratum to another because of droughts (Lynch, 1949; Crissey, 1957) tend to make inferences from yearly pond-pair correlations difficult. Also, much of the fluctuation in pond numbers revolves about filling and drying of small transient potholes, with the larger, deeper ponds generally holding some water through each summer or until major droughts occur (Dzubin, unpublished). For example, Smith (1949) recorded a 77 per cent decrease in pond numbers in the Alberta parklands in 1949 but a 59 per cent Increase in duck populations, indicating that ponds, perhaps the larger, deeper potholes, had not yet reached critical levels of occupancy. A number of smaller ponds may in fact be superfluous to some basic number of large, deep ponds required by pairs in any breeding home range.

Direct counts of indicated breeding pairs utilizing the 10.5-square-mile Kindersley Study Area from 1956 through 1967 show a trend downward from 1957 to 1963 and a partial recovery thereafter (Table 15). Comparable counts were made only once during each season at the optimum census period for mallards and pintails and therefore the data presented do not lend themselves to particular consideration of fluctuations of other species. From brood surveys I calculated that production of young mallards in any of the four summers, 1956 through 1959, was not sufficient to balance annual mortality.

In a study of mortality of flightless young mallards, banded throughout the Kindersley district from 1954 to 1959, Gollop (1965) showed a loss of 32 per cent of young from 3 to 7 weeks of age. For flying young an average mortality of 61 per cent was calculated for the year following September 1. Mean annual adult mortality was 47 per cent. Assuming that the 1956 adult and immature segments were subject to these mortality rates, there should have been a marked reduction in the returning population in the spring of 1957. Yet breeding populations in 1957 rose markedly over those in 1956. I concluded that in 1957 mallard and pintail pairs moved onto the study area from the surrounding drought-stricken regions. Thereafter all pair populations continued to decrease to a low about 1963. These decreases probably reflected low production rates and their subsequent effects on spring adult populations homing to the area. Also poor May pond quality, i.e. low water levels and extensive mud flats, may have deterred settling of pairs and led to their emigration northward. If production of young was low and if hunting and natural mortality yearly reduced the adult component, the population traditionally homing to the area would be quickly reduced. the study area was completely dry on July 25, 1963, except for a one-quarter-acre, spring-fed pond, and few young of any species were successfully fledged that year. Since 1964 there has been a yearly recovery of the breeding populations of all species, associated with higher May pond numbers, a greater pond acreage, a longer, total shore-line distance, and higher July pond numbers for broods. The population may have also experienced higher survival, or pond quality so improved that it attracted pioneering pairs (cf. Hochbaum, 1946). A lack of adequate supporting data on habitat requirements of each species, young produced yearly, homing rates, extent of nonbreeding, and spatial relationships of pairs precludes any knowledgeable discussion of correlations between pond and breeding pair numbers.

A comparison of the number of May and July ponds with mallard and pintail populations from 1956 through 1967 again showed yearly decreases after 1957 to 1963 and 1964, and an immediate recovery after 1964 (Fig. 5) . The yearly decreases from 1957 to 1959 were associated with decreasing May and July pond numbers. Thereafter, the recovery was associated with increases in both May and July pond numbers. Increasing numbers of pioneering pairs and higher production rates possibly led to increased numbers of pairs breeding in 1965 through 1967. Pintail populations dropped faster than mallard populations from 1957 to 1960 but recovered faster from 1964 to 1967. Neither species showed an increase in breeding populations from 1959 to 1962 in spite of a slight increase in May and July pond numbers in these years.

Figure 5. May - July pond numbers and yearly mallard and pintail pair populations, 1956 to 1967. (Note that pintail numbers decreased more sharply than mallard numbers from 1957 to 1959 but increased more rapidly from 1964 to 1967.

graph of pond numbers and yearly duck pair populations in Kindersley study area.

From 1964 through 1967, I observed that pintails and shovelers showed a marked propensity toward using newly flooded depressions. This attribute may be a characteristic of species with strong pioneering tendencies and weaker homing tendencies (Sowls, 1955). Smith (1949) has also commented on the population shifts of these two species. In Alberta, where there was a grassland drought in 1949, he noted that pintails and shovelers showed the greatest individual population losses after 1948, indicating a movement elsewhere. Lynch (1949) reported a major shift of pintails from drought-stricken Saskatchewan grasslands to areas beyond the parklands, even though mallards, widgeon, and blue-winged teal moved into the better watered parkland. Pintails and shovelers may have evolved in habitats containing ephemeral ponds. Any predisposition to quickly shift breeding grounds to better watered areas may hold some adaptive significance, especially where it fosters brood survival. If these two species have evolved in relatively unstable environments, emigration may be a major density regulatory mechanism whereas in mallards and perhaps blue-winged teal, self-regulatory mechanisms concerned with density effects on reproductive rates or behavioural spacing mechanisms controlling density may be more prominent (Dzubin, 1969).

Previous Section -- Census Methods
Return to Contents
Next Section -- Duck Census Techniques