Mutual interference during nest-prospecting in the shelduck, Tadorna tadorna

Anim. Behav., 1979,27,522-535

MUTUAL

INTERFERENCE DURING NEST-PROSPECTING SHELDUCK, TADORNA TADORNA

IN

THE

BY I. J. PATTERSON & M. MAKEPEACE Culterty Field Station, University of Aberdeen, Newburgh, Aberdeenshire Abstract. The shelducks (Tadorna tadorna) of the Ythan estuary, Aberdeenshire, defended feeding territories on the estuary, but in the sand dune nesting area they formed groups while nest-prospecting, although they would be expected to space out their cryptic nests. Within the groups there was frequent aggression and the development of dominance hierarchies. Nesting success decreased significantly with increasing population size. During the selection of nest sites, mutual interference, i.e. a decrease in the amount of time spent nest-prospecting, increased with the number of birds present. There was some indirect evidence that such interference might be the cause of the density-dependent reduction in nesting success. This process will tend to have a regulatory effect on population size. The individual survival value of joining groups may be to reduce the risk of predation on the adults, but is perhaps more likely to be to gain information on the location of good nest sites. The main hypothesis being tested is that there will be an advantage to at least some pairs in aggregating with others while prospecting, but that such grouping will lead to mutual mterference with nest-prospecting and a lower overall nesting success. This hypothesis predicts (a) a density-dependent reduction of nesting success with increasing population size, (b) demonstrable mutual interference with nest-prospecting, increasing with increasing population density and (c) decreasing nesting success with increasing mutual interference. The aims of the present study were, (a) to describe the behaviour and dispersion of shelducks while prospecting for nest sites in the nesting area, (b) to test whether pairs interfered with the prospecting activities of others, (c) to test whether such interference might cause nest losses and (d) to investigate the possible survival value of being in a group.

The nesting of shelducks poses some interesting questions on the survival value of behaviour and on its consequences for population variables. Although normally a hole-nester and, as such, expected to have relatively low predation on nests and high success in hatching broods (Lack 1968), shelducks rarely hatch more than half of the clutches laid (Hori 1964; Patterson et al. 1974). The expected low level of predation seems to occur, but some nest failures result from desertion by the female (Patterson et al. 1974). The pairs defend feeding territories on mudflats, shore or freshwater during the nesting period and a similar spaced dispersion might be expected when the birds leave the water to prospect for nest sites. Their highly cryptic nests should suffer least from predation when they are considerably dispersed (Tinbergen et al. 1967; Taylor 1976), and the simplest way to ensure dispersed nests would be for the pairs to space out while prospecting. In contrast, however, the prospecting pairs form groups, noticeable enough to have been named ‘parliaments’ (Young 1970a) or ‘communes’ (Hori 1964), in which there is frequent display and aggressive interaction between pairs. Jenkins et al. (1975) have suggested that nest failures, among a small sample of shelducks using nestboxes at Aberlady near Edinburgh, were associated with the presence of new eggs laid by a new female in the partly-incubated clutch. This, together with the aggressive interaction between pairs, suggests that mutual interference, defined here as behaviour by one bird or pair which lowers the efficiency of the nesting behaviour of others, might cause a reduction in nesting success.

General Methods The study area was around the Ythan estuary near Aberdeen, with its population of about 170 shelducks (Patterson et al. 1974). The 55 adult breeding pairs in the population defended feeding territories on the estuarine mudflats and nested in rabbit, Oryctolagus cuniculus, burrows on the adjacent sand dunes of the Sands of Forvie National Nature Reserve (Fig. 1). Shelducks used all of this area, but almost all observations were made south of a path between Waterside Bridge and Rockend (Fig. l), a study area which could be covered adequately in the short period each day when the birds were active. Data on the total number 522

PATTERSON

& M-PEACE:

MUTUAL

INTERFERENCE

of shelducks and their dispersion were obtained by walking a standard route through the dunes, designed to keep the observer out of sight of the birds as much as possible. The position of each shelduck seen was recorded by reference to a map prepared from aerial photographs. The total number seen on the ground was used only as an index to the true total present, since the birds moved frequently during the count and an unknown proportion could have been missed or counted twice. The error, however, was likely to have been small. An alternative method, involving counting the birds as they flew from the estuary to the dunes at dawn, was thrice used in April to June, 1972, while another person did a walking count. The flight count totals of 25, 14 and 23 (mean 20.7) were similar to the totals of 23, 12 and 32 (mean 22.3) from the walking count. The behaviour of shelducks in the dunes was at first recorded on the standard route by noting each bird’s behaviour at the instant when it was first seen, or by watching pairs or groups which appeared not to be disturbed. The effect of the observer could not be eliminated from either of these methods, since shelducks and other species flying overhead gave alarm calls to which the observed birds responded. All the behavioural data presented here were collected from

0

Pools

/

HWM

Fig. 1. The Ythan estuary and dune study area.

DURING

NEST-PROSPECTING

523

observation hides on high dune ridges with a wide view as well as a concealed access. They overlooked dune ridges, covered mainly by marram grass, Ammophila arenaria, where almost all nest sites were found, and also relatively flat meadows of short grazed grass. The behaviour of each shelduck visible from the hide was recorded every 5 min if the bird was on a meadow and every 1 min if it was on a dune, where active nest-prospecting occurred. The birds’ behaviour was divided into seven distinct categories: sleep, with the head resting on the back and eyes shut; sit/stand, with the neck bent in a relaxed posture; preen, grooming the feathers; walk, any locomotion; alert, with neck straight and vertical; prospect, looking into and entering burrows; and display, any agonistic posture. Each bird was examined in turn and its behaviour at that instant was recorded. These data were analysed in two ways. To estimate the proportion of time spent in each kind of behaviour, the periodic observations were accumulated over all individuals of each sex in each habitat (or other subdivision). The proportion of observations in which the birds were showing a given kind of behaviour was taken as an estimate of the proportion of time spent in that behaviour. Differences in proportion, however, could not be tested statistically since the accumulated totals contained runs of non-independent observations (e.g. successive 1 min checks on the same bird). The same objection applied to successive landings by one bird. For statistical testing the data had to be totalled separately for the different marked individuals, which became the basic units of data. It was then possible to count the number of individuals showing one kind of behaviour and the number showing another. The Fisher exact probability test and Chisquared test were used to test differences in the proportion of individuals showing a given kind of behaviour between two situations, and the Sign test was used to test whether the proportion in one situation differed from equality. All aggressive interactions in each 5 min period were counted and, since over 80% of the birds were individually marked with coloured leg rings (Patterson 1977), the identities of aggressor, victim and eventual winner could usually be determined to test for dominance relationships. Decoys, used for experiments, were made by removing the viscera and major muscles from

524

ANIMAL

BEHAVIOUR,

any shelducks found dead, injecting the body with concentrated formalin and drying it while it was wired in the required posture. Three males and three females were used, with two males in the alert posture, two females with necks extended in the prospecting posture and the remaining two in the resting position. The decoys were grouped around burrows in the centre of a 20-m length of dune ridge overlooked by an observation hide in April 1977. An equivalent area, 20 m further along the same ridge, was used as a control. This second area was used as the experimental section on alternate observation days. The decoys were positioned soon after dawn, watched continuously for 2 to 3 h and removed at the end of the session. All shelducks hovering over or landing in the 10 m radius around the decoys or the control area were recorded. Nesting success was estimated by daily observation on the estuary to find which of the territorial pairs were seen with broods of newlyhatched ducklings. This method may have missed a few broods if they survived less than a day after hatching, but it avoided the great difficulty of finding nests and the considerable risk of disturbance to those found (Young 1964). Shelduck Occurrence and Behaviour in the Nesting Area Shelducks visited the dunes from mid-March, around the time when they started to defend feeding territories on the mudflats. Williams (1973), observing the birds on the estuary several times daily over the period of the dune observations, found that of 45 pairs, 16% were first seen in the dunes before, 44% during and 40% just after the short period of taking up territory. The diurnal pattern of occurrence was determined by counting the birds passing from estuary to dunes and back, across the west boundary of the dunes, from before dawn until movement ceased. The birds moved rapidly into the dunes just after dawn and most remained at least 90 min before starting a gradual return to the estuary, with the last birds leaving some 6 h after the first arrivals (Fig. 2). Occasionally some birds were seen in the dunes in the afternoon. About 50 z of the ducks still remained 3 h after the first arrivals (Fig. 2), and all counts and observations were concentrated in these 3 h. The seasonal pattern of occurrence showed a distinct peak in mid-April and another in late May or early June, very few shelducks being seen in the dunes m early May (Fig. 3). This

27,

2

period of low numbers coincided with a peak in egg-laying, estimated by Williams (1973) by calculation from observed hatching dates. The first peak of attendance was thus about 2 weeks before the peak of laying and some individually-marked pairs were seen in the dunes foa’ot,“tleast a month before their estimated laying Age and sex composition were determined from the distinctive plumage of the yearlings and a marked sexual dimorphism. All the birds seen in March and April were in adult plumage (at least 2 years old) and had an equal sex ratio (Table I). In May and June, about 8% of the birds were yearlings, all of them females, but the sex ratio among the adults remained equal. Over the same period, Williams (1973) determined the status of the marked individuals on the estuary. Most of the pairs attending in the April peak were territorial and many succeeded in hatching broods. The adults attending in May and June, however, included two- and threeyear-old birds without territories. For example, in 1975, all of 12 pairs seen before mid-April, but only 5 of 11 first seen later, were territorial

Fig. 2. The number of shelducks in the dune study area, from counts of birds flying in and out over the west boundary, 1st June, 1972.

Fig. 3. Seasonal variation in the number of shelducks seen on the ground from the standard counting route, described in the text.

PATTERSON

& MAKEPEACE:

MUTUAL

INTERFERENCE

(Fisher exact test, P = O*OOS).Those territorial pairs which arrived late in the dunes were very unsuccessful; in data from 1970, 1971, 1972 and 1975, none of 10 pairs which were first seen after 30th April succeeded in hatching a brood, whereas 31 (43 %) of 72 pairs which appeared before then were successful (x 2 = 5.21, P < 0.05). The late May peak contained a number of pairs which had been seen in April but which had apparently lost their clutches. Group size of birds flying to and from, and also among, the dunes was most commonly two, as pairs left their territories and moved around independently of others. Single birds were mainly females leaving the nest to feed and males returning to their territory after accompanying the female back to the nest. Among birds on the ground in the dunes, the modal group size was also two, but larger groups (up to 16) were common as some pairs joined others. In all situations there was a tendency towards even-numbered groups, reflecting the pre-ponderance of mated pairs among the birds. Dispersion within the dunes was similar throughout the nesting season and between seasons. Birds tended to concentrate in particular meadows, such as several round the perimeter of a large central valley, some in a valley in the south of the area and one in a grass field adjacent to the dunes (Fig. 4). Pairs and small groups also occurred throughout the dunes. Sightings of those marked birds seen more than once (Fig. 5) showed that each pair had a restricted range within the dunes, usually centred on one of the favoured meadows. Each meadow was attended by a number of pairs with widely overlapping ranges and most pairs were seen occasionally at meadows other than their main one, although usually adjacent to it (Fig. 5). Marked birds seen during surveys of the north part of the Table I. The Ages and Sexgf

DURING

NEST-PROSPECTING

525

dunes were only exceptionally seen in the main study area in the south. Timing of first arrival, in relation to the position of the range within the dunes, was

Fig. 4. The dispersion of shelducks seen on the ground. Data from 12 counts in March to April, 1972, have been superimposed and the numbers seen at each place have been totalled.

Shelducks in the Nesting

Adult

Yearling

Male

Female

Male

1972

118 205

118 195

;

I-%7? 1972

188 130

188 132

+ 0” *

Female

*Sign test, against an equal sex ratio, P -c 0401.

Fl

E

Fig. 5. The dispersion of six marked pairs in March to June, 1972. The larger symbols show the number of sightings of a pair at that place.

526

ANIMAL

BEHAVIOUR,

determined for marked birds by dividing the pairs into groups in order of the date when they were first seen and each pair was assigned to the east, west or south of the study area, according to which meadow they used most (Fig. 5). In both 1971 and 1972 the earliest pairs tended to settle in the west part of the area nearest to the estuary and the last pairs tended to settle in the east (Table II). All but one of 13 ringed pairs, and all of seven pairs with one bird ringed, returned to the same part of the dunes in two successive years in 1970 to 1972. Two females which changed their mates between years also returned to their previous areas, but one male with a new mate changed his area. Fidelity to the previous nesting area (by 19 out of 20 pairs) was significantly higher than the approximately one-third expected if the birds changed randomly over the three areas available (~2 = 8.88, P -=cO-01). Dispersion of territories on the estuary, as determined by Williams (1973), was associated with dispersion in the dunes. The pairs found mainly in the south of the study area tended to have territories on the adjacent south part of the estuary (Fig. 6). The pairs using mainly the west dunes had territories in the mid-part of the estuary, and included a group of 4 to 5 adjacent territory holders. The pairs using mainly the east dunes had territories scattered more widely

27,

2

and both the ‘east’ and ‘south’ birds included a few with territories in the upper part of the estuary, from where most pairs went to the north of the study area. Behaviour in the Nesting Area The proportion of time spent in each kind of behaviour was estimated by accumulating the 5 min observations on meadows and the 1 min observations on dunes for all individuals, dividing only between the sexes and between habitats. While on meadows, males and females behaved similarly, spending most of their time asleep, sitting or standing (Fig. 7, A). On dunes, however, their behaviour differed considerably; the males spent most time alert while the females also stood, walked and prospected (Fig. 7, B). Walking and prospecting tended to occur together as the ducks walked between burrows. When the same bird was compared between dunes and meadows, four of five marked males in 1975 and all of ten in 1976 (Sign test, P < 0.01) showed a higher proportion of time alert in the dunes than in the meadows. Similarly, all of nine marked females in 1975 and all of nine in 1976 (Sign test, P < 0.01) showed a higher proportion of walking and prospecting in the dunes than in the meadows. This difference

Table II. The Position of Ranges within the Nesting Area In Relation to Order of Arrival during the Season Number

of pairs with ranges in each part of area

Order of arrival

West

1971 1st 6 2nd 6 3rd 6 last 7

3 2 3 0

2 3 i0

7

4

2 2 3

0 0 3

0

6

1972 1st 6 2nd 6 3rd 6 last 7

4 0 1

South

East 1 1 2

For each set of birds, the preferred area is underlined. (1972, first 12 versus the rest, west versus south + east, Fisher exact test, P = OGO3).

Sands 01 Forvle

i--

.i 500m

I

Fig. 6. Territory (centre) sites of marked pairs with ranges in the west (W), east (E) and south (S) of the dune study area.

PATTERSON

& MAKEPEACE:

in behaviour between habitats throughout the season.

MUTUAL

INTERFERENCE

was present

Discussion Visits to the nesting area occurred mainly just after dawn, as described by Hori (1964) and Young (1970a). This synchrony should enhance the possibility of grouping since the maximum number of pairs will be in the dunes at the same time, and dawn is a convenient timer commonly associated with other kinds of social behaviour such as bird song. Possibly there is selection for grouping, and the survival value of this will be considered later. Other possible functional explanations of the timing of visits seem less likely; the birds do not feed at a particular time of day but at particular states of tide, the nest sites are unlikely to change during the day, and the risks from predators are unlikely to be less in the early morning. Indeed almost all sightings of foxes, Vulpes vulpes, hunting in the dunes were made just n=820

n=1060 50 40 %

2

?

30 20 10 0

lI!dLL si st pr al w pp d

si st pr al w pp d

A. Meadows

n.1136

n-1052

8. Dunes

Fig. 7. The percentage of observations made up by each type of behaviour, in male and female shelducks on A, meadows and B: dunes. The types of behaviour: sl-sleep, St-stand or sit, pr-preen, al-alert, w-walk, ppprospect and d-display, are defined in the text.

DURING

NEST-PROSPECTING

521

after dawn and there was little human disturbance at any time of day; unlike Hori’s (1964) study where farm workers disturbed the birds later in the morning. The seasonal distribution of visits by the birds to the nesting area suggested that there were two largely separate groups of birds. Those visiting the dunes in April were exclusively adult pairs and the low numbers seen in the dunes in early May were presumably due to the onset of the incubation among these early pairs. Some of those whose nests apparently failed resumed their visits to the nesting area, as also described by Hori (1964). However, most of the pairs visiting the area in May and June were new, non-territorial and young pairs (which possibly come into breeding condition later than older ones), and very few of these nested successfully. Young (1964) did not record the arrival date of each pair, but found that among 39 marked birds, only 13% hatched a brood. However he made only one observation before April 15 and 70% of his sightings of birds were in May and June when the young unsuccessful pairs were commonest in the present study. The marked preponderance of females among yearlings visiting the dunes (Table I) suggested either that the sexes mature at different ages or, more likely, that visits to the nesting area are made primarily by females and that males visit only when they are pair-bonded to a visiting female. Yearling males were rarely paired and were so rarely seen in the dunes that none appeared in the counts shown in Table I. Both Hori (1964) and Young (1970a) described pairs arriving singly in the dunes and then joining others, and showed that the same marked birds occurred in the same places over a period, as in the present study. Hori (1964) also found that three pairs in one dune ‘commune’ had adjacent feeding territories and that two pairs in another group were also territorial neighbours. These results are confirmed by the present findings which show sub-groups within the nesting birds, with the same pairs associating in the same dune meadow morning after morning and having a tendency to be territorial neighbours. Since 79 to 88% of surviving territorial pairs retain the same feeding territory from year to year (Young 1970b; Williams 1973), there was likely to be a high degree of individual recognition among such dune groups, an excellent basis for dominance relationships to develop.

528

ANIMAL

BEHAVIOUR,

While in the nesting area the shelducks were surprisingly inactive. Most of those seen on the standard counting route were on meadows (Fig. 4) where both sexes spent most of their time sleeping and standing (Fig. 7, A), as Young (1970a) also described. It is unlikely that these birds had already selected their nest sites since most groups on meadows eventually prospected again later in the same morning and on subsequent days. Pairs which already had a site might draw the attention of predators to the nest by spending time near it. However, they might be defending it against other pairs or be assessing the frequency of visits by predators to the area. Possibly the inactive pairs were watching for others to begin prospecting, or had been deterred from prospecting by being joined by others, as described in the next section. Mutual Interference Nesting Success in Relation to Population Size The size of the Ythan population of shelduck has changed over the years, as measured by Patterson et al. (1974) in 1962 to 1970, Williams (1973) in 1970 to 1972 and the present authors in 1973 to 1977. Nesting success was measured by expressing the total number of broods seen to arrive on the estuary each year as a percentage of the number of territorial pairs present in May of the same season. The success of territorial pairs in hatching broods was significantly lower with increasing number of territorial pairs on the estuary (Fig. 8), showing a densitydependent effect of numbers on hatching. Williams (1973) showed that the seasonal peak of nest failure coincided approximately with the late-May peak of visits to the dunes, largely by non-territorial birds (which seem never to breed), suggesting that the presence of these birds might be associated with the nest failures. However, the number of failures in each 5-day period in 1971 was not significantly correlated with the number of birds seen in the dunes in the same period or in the preceding period. There was also no correlation between nesting success in any year and the number of non-territorial birds present on the estuary in that year. Nesting success was thus more closely related to the number of territorial pairs than to the size of the non-territorial flock. Interference with Prospecting Often a female shelduck was seen visiting burrows accompanied only by her alert mate, but such pairs were often joined by others and this allowed a comparison of prospecting

27,

2

by single females and by groups. Prospecting activity by one female attracted others quickly. Groups on meadows spent long periods sleeping, sitting or standing, but if one female walked or flew to a dune to visit burrows, another female in the group frequently followed her to the same dune within 1 min (79.8 % of 94 cases, 1975 and 1976). The two females usually approached within a few metres of each other and frequently looked into the same burrows. Some females interacted aggressively (in 21.7 % of 120 cases of at least two together, 1975 and 1976), usually showing only mutual head pumping display (a rapid forward circling of the head and neck) and, rarely, a short lunge and peck between birds very close together. However, as in the winter flock (Patterson 1977), more interactions were between males than between females, and male-male encounters occurred in 29.2% of 120 cases of at least two pairs together in 1975 and 1976. The mean rate of interaction was similar in different years (0.04 f 0009 interactions per bird per minute in 1970, O-038 & 0.013 interactions per bird per minute in 1971 and O-043 f O-009 interactions per bird per minute in 1972) and it did not change consistently with group size, habitat or date. Interactions were, however, much more frequent immediately after a new pair arrived to

‘“o*20

1

l::l 45 50

55

60

6’5

70

75

Fig. 8. The proportion of territorial pairs hatching broods in relation to the number of such pairs. The numbers give the year for each point. The line is the calculated linear regression, y = 104%1.015x, Y= O-812, P < 0.01.

PATTERSON

& MAKEPEACE:

MtJTtJAL

INTERFERENCE

join a group (0.392 f 0.057 interactions per bird per minute in the first 5 min after an arrival in 1974). The rate declined steadily to O-093 & O-033 interactions per bird per minute after 25 min and was only O-006 f 0.008 interactions per bird per minute in groups where there had been no arrivals since the start of the observation period. When a pair arrived in a group, attacks by the arriving male towards others already present were as common as the reverse. Most interactions were supplant attacks in the head-down, with the head and neck extended forward parallel to the ground. There was no indication that males were attempting to defend a particular area, although in a few cases males appeared to attack others only when they approached their prospecting females. Occasionally a male was seen to attack a female of another pair. Aggressive interaction thus was fairly common, probably more so than the data indicate since the birds were often out of sight in the uneven dunes and so some interaction may have been missed. The time spent in the dune ridges, which contained almost all the nest burrows, was measured from the time a female landed in or walked into marram grass until she flew up or walked out on to a meadow again. There was no significant difference between the time spent on a dune by single females (mean 5-2 f O-7 min, n = 24, 1975; mean 7.8 rt 1.4 min, n = 31, 1976) and by females accompanied or later joined by one or more others (mean 5.1 6 0.8 min, n = 33, 1975; mean 12-l -& 2.2 min, n = 71, 1976). However, in 17% of 112 cases where one female joined another already on a dune, the original female left the dune immediately (less than 10 s) after the arrival of the other. There was no consistent difference in the time spent on the dune when interaction occurred between pairs, compared with instances when no interaction was seen. Females accompanied by other females spent a smaller proportion of time prospecting compared with single females (Fig. 9, A and B). Most of this difference was due to females spending a lower proportion of time actually inside burrows when other females were present (Fig. 9, B). Accompanied females spent more time standing and displaying than did single ones, The reduction in prospecting by a-mpanied females was shown consistently by individual marked birds; all of three females in 1975 and all of seven in 1976 (Sign test, P <

DURING

529

NEST-PROSPECTING

0.05) showed a lowered proportion of time prospecting in observations when other females were present than when the same birds were alone. This was evidence that females interfered with the prospecting activity of others. Hori (1964, pages 342 to 3) has also noted that the amount of ‘real prospecting’ was low among groups of shelducks and Young (1970a, page 126) states that ‘As more and more birds join the group, the serious nest hunting gradually tails off and eventually ceases altogether, whereupon the birds usually adjourn to some plot of open ground such as a ploughed field or a grassy meadow between the dunes . . .,’ showing that in these two studies also group formation apparently interfered with prospecting. Interference and Density Since the observations on prospecting were made from fixed hides, it was not possible to make simultaneous complete counts of the A. 1975 n= 161 (alone)n = 283 (group)-

60

- -

1

8.1976 n= 428 (alone)n= 708 (group)-

40

%

- -

20 0

1

?.I (out) (in)

Fig. type pair and

9. The percentageof observationsmade up of each of behaviour in female shelducks in dunes as a single (solid line) or in a group (dotted line) in A, 1975 B. 1976. The tYDes of behaviour are as in Fia. 8

except in B where-pp(out) = prospecting outsid; a burrow and pp(in) = prospecting actually inside a burrow. The shaded portion shows where single females show a higher percentage of a behaviour than females in a group.

530

ANIMAL

BEHAVIOUR,

number of birds in the whole study area. However, the number of marked pairs seen from the hide each day could be used as an index of the total number of shelducks present. The number of recorded prospecting bouts with at least two females present on a dune at the same time was used as a measure of the potential amount of interference on the same day. The number of such incidents of group prospecting increased significantly with increasing number of marked pairs seen in the area (Fig. 10). The increase was roughly proportional to the increase in the number of pairs seen. Interference and Nesting Success Prospecting pairs which were known to have territories and thus to be potential breeders (Patterson et al. 1974) were divided into (a) those seen more often ‘alone’ (i.e. as an isolated pair or initially isolated but later joined by others) and (b) those seen more often ‘grouped (i.e. having joined a pair already prospecting or having arrived on the dune in company with others). Nesting success was determined by daily checks on the estuary for newly hatched ducklings. There was no tendency for pairs which were successful in hatching broods to have been alone during prospecting more frequently than unsuccessful pairs, although the sample was too small for adequate testing. An alternative test, though an indirect one, involved dominance. Since the approach of one pair to others commonly involved aggressive interaction, any dominance relationships within the groups should allow the more dominant pairs to reduce interference with their prospecting, for example by keeping other pairs further away. Aggressive interactions were observed

27,

2

mainly on meadows. By recording the identities of the winner and loser of each aggressive encounter between males it was possible to arrange the birds attending one meadow in 1974 into a linear hierarchy (Table III), and a similar result was obtained in 1976 and 1977. In all three years the more dominant males were much more successful than the subordinates in hatching broods (Table IV). In a smaller sample from 1975, only two pairs seen in the dune study area were successful in hatching broods, but both had dominant males which were never seen to be defeated in their encounters with others. Discussion The results give some support to all three predictions of the hypothesis. Perhaps the most striking result is the density-dependent reduction of nesting success with increasing population size, although there is a need for more data from years with a high population, since the Table III. Dominance Hierarchy, West Group 1974. Tbe Data are the Number of Encounters Won by tk Bird in tbe Row over the Bird in the Column

Winner, bird number

1 2 :

1

Loser, bird number 234567 5 8 2 4 9 1 4 8 7 2

9

3

1

:, 7 Table IV. Nesting Sucwe

2

i&R.?.lation to Dominance of Number

of pairs which: Ditryjdatch

%%z 1974 Dominant 3 pairs Subordinate 4 pairs 5 1 1977 Dominant 4 pairs Subordinate 4 pairs Fig. 10. The frequency of incidents of at least two females prospecting together, in relation to the number of marked pairs seen that day (v = 0.089 + 0-694x, r = 0.481, P < 0.05).

*The three most dominant pairs. Fisher exact test, 1974, P = 0.029; 1976, P = 0.051; 1977, P = 0.071.

PATTERSON

& MAKEPEACE:

MUTUAL

INTERFERENCE

existing relationship depends heavily on a few of the earlier years of the study when numbers were higher, However, there is no obvious factor, other than the high population size, which can account for the low nesting success in these years. There is evidence of mutual interference with the process of selecting a nest site. Although the total time that a female spent per visit to a dune did not change when other birds were present, the proportion of time spent actually prospecting and especially the time spent inside burrows was greatly reduced in all the birds tested. Thus the presence of others may reduce the efficiency of nest site selection and prolong the time which a pair must spend in the dunes. This, in turn, would increase the risk of predation to which these conspicuous ducks are exposed when entering uneven terrain with thick cover known to be hunted by foxes and stoats, Mustela erminea. There was some evidence that the amount of interference might increase with density, which is at least consistent with the reduction in nesting success with increasing population size and makes it possible that there is a causal relationship between interference and nest failure. Data on this last point were the most difficult to obtain directly since the nesting burrows were very difficult to find and to investigate and the females were liable to desert the nest if it was disturbed (Young 1964). The finding that dominant birds were more successful might suggest that interference could reduce nesting success since dominant pairs should be less vulnerable to interference by others, but there are other possible explanations for their greater nesting success. They may, for example, be able to compete more successfully for ‘better’ nest sites (giving a higher chance of success) or they might have better access to the most productive feeding territories or other important resources. The question of whether interference with prospecting for nest sites has a causal effect on nesting success and, if so, how such an effect is produced must remain open, since it was not possible to study the nests and factors affecting success. It is possible, for example, that a reduction in prospecting efficiency might reduce success directly, or that the attendance of other females at a site during prospecting might increase the chance of later interference by additional egg-laying during incubation, as suggested by Jenkins et al. (1975). Weller (1959) has shown that in the redhead, Aythya americana,

DURING

531

NEST-PROSPECTING

nesting success was reduced by several females laying in the same nest, but Hori (1964) showed that, in his area, such multiple nests of shelducks were more successful than single ones. Whatever the causal mechanism, there are important consequences at the population level. The significant reduction in nesting success as the territorial population increases (Fig. 8) will tend to have a regulatory effect, although the relationship between success and population is not likely to be linear at higher population sizes. Young (1970a) also regarded the interaction between pairs as a major factor limiting the breeding output of the population. A problem in this argument is that the Ythan shelduck population was only just self-maintaining over the years considered (Patterson et al. 1974) and it is difficult to see why individuals should begin to compete strongly enough to produce densitydependent reduction of nesting success, at population levels where breeding output is barely that required to replace the mortality of adults. A possibility being explored is that shelducks may be adapted to breeding in low density populations which may be more successful (Jenkins et al. 1975) and hence maintain the high density estuarine ones. The results also pose problems of individual survival value. The argument, so far, has been that pairs should disperse widely and solitarily while prospecting in the dunes. However, most landings, by shelducks both in meadows and on dunes, were made beside other shelducks rather than in unoccupied areas (Table V). It is difficult to calculate the expected proportions, but they can be estimated from the number of occupied and unoccupied areas available. In both 1971 and 1972 there was a maximum of 13 separate groups of shelducks in any one count in the study area where there were at least 50 places where groups could occur (Fig. 4). Thus at any one time at least three-quarters of the suitable areas were unoccupied. Despite this most shelducks landed in the minority of occupied sites and so Table V. Percentage of Landings which were Beside Other Shelducks (Sample Size in Brackets) Meadows

Dunes

1975

58-2

(67)

40.0 (5)

1976

59.3 (59)

67.6 (34)

Mean

58.7

64-l

(126)

(39)

S32

ANlMAt

REHAVIOUR,

were selecting the occupied areas. The analysis of nearest-neighbour spacing between males confirmed that the birds were significantly closer together than would be expected by chance. Such grouping could result either from social attraction or from a response by each pair to a changing suitability of parts of the habitat. These two possibilities were separated by observing a group of decoys on an arbitrarily chosen section of dune. Almost all of the shelducks hovering over or landing in the area did so over the area currently occupied by the decoys (Table VIII). Since the experimental (decoy) area and control area were exchanged on alternate days, the birds presumably were reacting to the presence of the experimental group. Thus the birds clearly form groups, suggesting that there must be some advantage in this to offset the apparent disadvantages of mutual interference. The next stage of the study was to test some possible benefits of grouping. The SorvivaI Value of Grouping At least two hypotheses can explain the advantage to the individual pair of shelducks in approaching others; (a) that grouping reduces the risk of predation on the adults and (b) that pairs may gain information on the location of Table VI. Percentageof Pairs and Groups wbicb were Alarmed by tbe Observer. (Sample Size in Brackets)

A. %?m 1972

Single pairs

Larger groups

27-3 (44) 36.9 (65)

38.6 (57) 30.7 (75)

:z

Ez

B. Dunes 1971 1972

::i;

{ii{

None of the differences between pairs and groups is statistically significant; Fisher and ~2 tests.

27,

2

those nest sites which give the best chance of success in hatching a brood. The anti-predator hypothesis suggests that although grouping may have the disadvantage of clumping of nests, with its possible consequence of increased nest predation and interference by other ducks, this may be outweighed by the increased vigilance of a group and the possibility of a decrease in the proportion of time that each individual needs to be alert (Mutton 1968). Pairs which join others also avoid the possible risk of being first to land in an unknown area of thick cover. In a long-lived species like the shelduck a risk to the adult represents a greater potential loss of genetic fitness than does the same risk to one year’s clutch. This hypothesis predicts that birds in groups should detect a predator better and spend less of their time alert than single pairs. There should be a preference for landing beside others rather than in unoccupied areas and the female of a single pair arriving in an area should start prospecting later than a female arriving in a group. The second hypothesis is that the various disadvantages of grouping (discussed above) are offset, at least for some individuals, by an increase in nesting success resulting from learning the location of good nest sites, i.e. those which have had successful clutches in the past. The net benefit of grouping will be greatest for females which do not already know a good site, i.e. females breeding for the first time and those which were unsuccessful in the previous year. Pairs which already have a good site could reduce the disadvantage of others joining them by stopping prospecting immediately or by driving others away. Dominant pairs should be most effective at this if it were in their interests to do so, i.e. if they had been successful previTable VIII. Responses of Shelducksto an Experimental

Groap of Decoys

Table VII. Percentageof Observationain which Males were Alert, in Relation to Habitat and Group Size. (Sample Size in Brackets)

Area currently Decoys*

with:

No decoys*

Single pairs

Larger groups

Number of birds hovering

2 77 ~2 = 43.6,P < OGOl

A. Meadows 1975 1976

16.5 (127) 25.0 (288)

21.2 (661) 15.3 (772)

Number of birds landing

0 49 x2 = 30.1, P < O*OOl

B. Dunes 1975 1976

56.4 (117) 59-l (492)

34.7 (167) 40.9 (560)

*The decoy and control areas were used alternately on a total of seven tests. ~2 was calculated against an expectation of equal numbers in the two areas if there was no effect of the decoys.

PATTERSON

& MAKEPEACE:

MUTUAL

INTERFERENCE

ously (which is likely, Table IV). The hypothesis, unlike the anti-predator one, predicts a difference between females which hatched broods in the previous year and those which did not. The previously successful females should be less often in groups, should approach others less and should more often stop prospecting when others approach. Dominant pairs should more often be alone or should be in smaller groups than subordinates. The Anti-Predator Hypothesis Foxes and stoats were seen too rarely to be useful for testing, but an observer who followed a concealed route and attempted to stalk the birds was probably an effective substitute. We recorded which pairs or groups flew up in alarm while being approached or observed, and those which remained apparently undisturbed. The distance between the observer and the birds depended on the availability of vantage points and neither distance nor intervening cover could be controlled. On meadows, where the birds spent most time sleeping or sitting, there was no difference between pairs and larger groups in the proportion which flew up (Table VI, A). On dunes, however, larger groups were more likely to fly up than pairs (Table VI, B). The difference was not significant but was very similar in the two years. Single pairs were equally likely to fly up in both habitats. Periodic observation of behaviour, accumulated over all males, showed that on meadows there were no consistent differences between pairs and larger groups (Table VII, A), whereas on dunes the males in groups were less often alert than were the males of single pairs (Table VII, B). This was cotirmed by data on individual marked males. Both of two in 1975 and all of six in 1976 were more often alert when they were alone on dunes than when they were with other pairs (Sign test, P < O-05), whereas individual males observed on meadows showed no consistent differences in alertness with group size. Periodic observation of behaviour, accumulated over all females, showed no consistent differences between groups and pairs in the proportion of alert behaviour on meadows and showed only a slightly higher proportion of time spent alert by females in groups on dunes (Fig. 9). Data from individual females on dunes showed no consistent difference in alert behaviour with group size. Females thus did not reduce the amount of time they spent in alert behaviour when they were in groups and, as was shown

DURING

NEST-PROSPECTING

533

earlier, spent much less time prospecting when in groups (Fig. 9). The females of single pairs arriving in an unoccupied area did not take significantly longer to begin prospecting (mean 13.9 & 6.8 min, n = 12) than did females arriving in the company of others (mean 12.9 f 3.7 min, n = IO). The Nest Site Hypothesis The marked pairs seen landing in the study area were divided into those more often seen landing in an unoccupied area and those more often seen landing beside others. In both 1975 and 1976, more of the territorial pairs which had hatched broods in the previous season landed mainly in unoccupied areas than did territorial pairs which had previously been unsuccessful and pairs and single females which were not territorial in the current year (Table IX). The differences were significant in 1975 and when the years were combined (justined by the sample of successful pairs being independent between the two years). Marked pairs seen Table IX. Proportion of Pairs which More Often Landed in Unoccupied Parts of tbe Nesting Area 1975

1976 Total

%

A.

Hatched brood in previous year

414

l/3

517

71

B.

Territorial in current year but did not hatch brood in previous year

l/6

115

2/11

18

Non-territorial current year

015

o/4

o/9

0

C.

in

1975: A versus B, P = 0.023; A versus C, P = OGO8; 1975 and 1976 taken together: A versus B, P = 0.039; A versus C, P = 0.0048, Fisher exact test. Table X. hop&ion of Pairs wbicb More Often Prospected Alone as Defined in the Text 1975 1976

Total

%

A.

Hatched brood in previous year

314

213

517

71

B.

Territorial in current year but did not hatch brood in previous year

214

317

5/11

46

Nen-territorial current year

l/4

012

l/6

17

C.

in

None of the differences between categories of bird is significant; Fisher exact test.

534

ANIMAL

BEHAVIOUR,

prospecting -were divided into (a) those more often seen alone throughout, or initially alone but later joined by others, and (b) those which more often joined another pair already prospecting or which arrived in a group. Again more of the pairs which had hatched broods in the previous season were seen alone in both 1975 and 1976 than were previously-unsuccessful pairs or non-territorial birds (Table X), although none of the differences were significant. In the decoy experiment in 1977, described above, previously-unsuccessful females landed beside the artificial group on more of their visits to the study area than did previously-successful females and they stayed with the group for longer. Although, again, neither of these differences is significant, the trend is in the same direction as that in the previous observations. There was no consistent difference between the categories of pairs in the time taken to stop prospecting after they were joined by others. Pairs of known dominance in 1976 (described above) were similarly divided into those more often seen alone and those more often seen with others while prospecting. Four of five dominant pairs were more commonly seen prospecting alone compared with only two of five subordinate pairs. The difference is not significant, but an even smaller sample from 1975 was consistent in direction. However the decoy experiment did not confirm this result. The females of dominant males landed beside the decoys on more of their visits than did females of subordinate males. Again this difference is not significant and the trend is inconsistent with that in the previous observations. The group size was recorded on the first time each pair was seen in the area each day, irrespective of the habitat in which they were seen. The percentage of sightings in which a pair were alone was significantly higher in the more dominant pairs (Fig. 11). The groups which included dominant pairs were slightly, but not significantly, smaller than those which contained only subordinate pairs. Discussion There seems to be some support for both hypotheses, in that the data co&r-m some predictions of both. There is possibly more support for the nest site hypothesis, however, in that the predicted differences between categories of pairs were found ; thus previouslyunsuccessful pairs, which would have the greatest net gain in approaching others, landed more

27,

2

commonly beside groups and they approached the decoy group slightly, although not significantly, more than did successful pairs. Dominant pairs also showed the predicted tendency to be alone more commonly than subordinates. This latter difference, however, was not confirmed by the results of the decoy experiment. All of the observations on dominant and subordinate pairs must be interpreted cautiously since only one of the differences in behaviour was statistically significant and these data should be treated as preliminary. Although the demonstrated general tendency to join groups would also be predicted by the anti-predator hypothesis, it cannot easily account for the differences between categories of pairs in their tendency to join groups. The anti-predator hypothesis was not strongly supported since although groups on dunes were, as predicted, somewhat more likely than single pairs to fly up when approached by the observer (and so presumably predators), in spite of a decrease in the proportion of time spent in alert behaviour by the individual males in the groups, the difference between the proportions flying up was not significant. In addition, the formation of groups on dunes did not allow more time for prospecting by females, since those in groups spent much less time prospecting than did those in single pairs (Fig. 9). The benefit of grouping in general was thus less than predicted; although it probably reduced the risk from predation, it also decreased the time available for prospecting.

‘0” 0.J III 0

1

2

III 3

4

5

I 6

,,,,,,I 7 6

9

10

11

12

13

Rank order

Fig. 11. Variation in the percentage of sightings of a pair in which they were alone, iu relation to the rank order of the male. The number against each point gives the number of sightings. Only pairs which were sighted more than five times have been included. (Spearman rank correlation coefficient Rs = - 0.719, P < O-05).

PATTERSON

& MAKEPEACE:

MUTUAL

INTERFERENCE

The possible functions of alert behaviour are not, of course, confined to defence against predators. The birds might also be watching for other shelducks (Jenkins et al. 1975) and the greater alertness of males in single pairs might enable them more readily to join, avoid or drive away other pairs depending on their status. The results of the observations on dominant and subordinate pairs were inconsistent. However, they suggest that dominants are alone more often than subordinates (Fig. 11) because they drive others away, since the decoy experiment showed no tendency for them to avoid groups. Drent & Swierstra (1977), suggested that variations in alertness in barnacle geese, Brunta leucopsis, were correlated with variations in food abundance and that the birds were watching other flocks. The data on alertness in shelducks could thus be explained by the nest site hypothesis whereas the differences in behaviour observed between categories of pairs could not be explained easily by the antipredator one. There is also no reason why both hypotheses should not co-exist. Previously unsuccessful pairs might obtain a dual advantage from grouping with others, while for previously successful pairs the disadvantage of being joined by others some of the time would be reduced, to some extent, by lessening the risk from predators. Acknowledgments The study was partly supported by a grant, number GR3/2439, from The Natural Environment Research Council, and the Director and Staff of the Aberdeen University Computing Centre provided facilities and help with the analysis of data, Mr W. Murray of Culterty Field Station helped greatly with trapping and marking of shelducks and Professor G. M. Dunnet, Dr D.

DURING

NEST-PROSPECTING

535

Jenkins, Dr H. Milne, Dr A. Watson and Dr M. J. Williams .made helpful comments on drafts of this paper. REFERENCES Drent, R. & Swierstra, P. 1977. Goose flocks and food finding: field experiments with Barnacle Geese in winter. Wildfowl, 28, 15-20. Hori, J. 1964. The breeding biology of the shelduck Tadorna tadorna. Ibis, 106, 333-360. Jenkins, D., Murray, M. G. Kc Hall, P. 1975. Structure and regulation of a shelduck (Tadorna tadorna (L.)) population. J. Anim. Ecol., 44,201-231. Lack, D. 1968. Ecological Adaptations for Breeding in Birds. London: Methuen. Murton, R. K. 1968. Some predator-prey relationships in bird damage and population control. In: The Problems of Bira% as Pests (Ed. by R. K. Murton & E. N. Wright), pp. 157-180. London: Academic Press. Patterson, I. J. 1977. Aggression and dominance in winter flocks of shelduck, Tadorna tadorna CL.). . Anim. Behav., 25, 447-459. Patterson, I. J., Young, C. M. & Tompa, F. S. 1974. The shelduck population of the Ythan estuary, Aberdeenshire. Wildfowl, 25, 16-28. Taylor, J. 1976. The advantages of spacing-out. J. theor. Biol., 59, 485-490. Tinbereen. N.. Imnekoven. M. & Frank. D. 1967. An ixpkriment on spacing-out as a defence against predation. Behaviour,. 28, 307-321. Weller, M. W. 1959. Pam&c egg laying in the Redhead (Avthva americana) and other North American krkidae. Ecol. Mokogr., 29,333-365. Williams, M. J. 1973. Dispersionary behaviour and breeding of shelduck Tadorna tadorna L. on the River Ythan estuary. Unpubl. Ph.D. Thesis, Aberdeen Universitv. Young, C. M. 1964. An ecological study of the common shelduck (Tadorna tadorna L.) with special reference to the regulation of the Ythan population. Unnubl. Ph.D. Thesis. Aberdeen Universitv. Young, 6. M. 197Oa. Shelduck parliaments. ~krdea, 58, 125-130. Young, C. M. 1970b. Territoriality in the common shelduck Tadorna tadorna. Ibis, 112, 330-335. (Received 13 February 1978; revised 15 July 1978; MS. number: 1721)