Parental investment in nest defence by stonechats (Saxicola torquata)

Anita. Behav., 1980, 28, 604-619 PARENTAL INVESTMENT IN NEST DEFENCE BY STONECHATS

(SAXICOLA TORQUATA) BY P. W. GREIG-SMITH*

School of Biological Sciences, University of Sussex, Falmer, Brighton BN1 9QG Abstract. Breeding st0nechats (Saxicola torquata) made mixed sequences of twe calls when a human intruder entered territories. 'Whits' are modulated notes with a small frequency range, and iv laboratory tests caused nestlings to stop begging. 'Chacks' cover a wide range of frequencies, and in the field were combined with flights made so as to distract an intruder from the nest. On average male and female call-rates were similar, but varied greatly according to the intruder's distance from the nest, and at different stages of the nesting cycle. Rates increased rapidly after hatching, and this correlated most closely with the cumulative total of parents' visits to feed nestlings. This suggests that the level of defence may be adjusted to the value of the offspring to their parents. Call-rates declined about one week after fiedging. A smaller peak by some pairs at the start of incubation was apparently related to probable poor condition after a previous breeding attempt, and after laying large clutches. Rates of Whits were higher at nests with larger broods, up to an asymptote, but rates of Chacks were independent of brood size. Birds suffering nest-predation showed lower call-rates before the event than equivalent successful birds, suggesting that the calls do reduce the risk of predation.

fled by measuring call-rates. I therefore examined these rates in various circumstances in order to answer the following questions. (i) Does calling change during the course of a breeding cycle ? (ii) Are there differences between successive broods within a season ? (iii) Does the number of young in a brood influence call-rates ? (iv) Do male and female parents call at different rates ? All these possibilities are suggested by Trivers' theory of parental investment (Trivers 1972; Barash 1975). I also sought to determine how the calls protect the birds' young and what consequences attend the failure to provide adequate defence. Most of the information was obtained from the birds' responses to my presence, as a potential nest-predator (cf. Erpino 1968; Barash 1975; Ricklefs 1977; Andersson et al. 1980), rather than to natural predators. This should not reduce the validity of the arguments about parental investment. However, the results must be regarded as specific to a certain kind of predator, since in common with many species (see e.g. Kruuk 1964; Curio 1975), stonechats make different responses to various dangers (see below, and unpublished data).

In many animal species, parents actively care for their offspring. Our understanding of this behaviour has been greatly clarified by Trivers' (1972) concept of parental investment, defined as 'any investment by the parent in an individual offspring that increases the offspring's chance of surviving (and hence reproductive success) at the cost of the parent's ability to invest in other offspring'. Natural selection should shape a species' behaviour so that in allocating care to a particular offspring, parents take account not only of its needs, but also those of possible future offspring. This may lead to a level of care which is less than the maximum possible, and less than the optimum level for the survival of the present offspring (Trivers 1972, 1974; Dawkins 1976; Maynard Smith 1977; O'Connor 1978). This paper reports a study of parental investment by stonechats (Saxicola torquata), which are small, predominantly insectivorous passerine birds. They are representative of many species in breeding seasonally, holding exclusive breeding territories, being largely monogamous within seasons, and investing heavily in care of offspring. Of the many components of care (Skutch 1976; Lazarus & Inglis 1978), I chose to examine nest defence behaviour, some aspects of which have been investigated in other species by Barash (1975), Curio (1975), Weatherhead (1979), and Andersson et al. (1980). As shown below, the stonechats' defence consists largely of calls, which can be readily quanti*Present address: M.A.F.F., Tangley Place, Worplesdon, Surrey.

Methods

The study was carried out in 1977 and 1978 at Ashdown Forest, Sussex, U.K. This area consists largely of patches of heathland (Gimingham 1972) surrounded by deciduous woodland. Parts of the heathland include isolated mature birch (Betula spp.) and pine trees (Pinus sylvestris), and 604

GREIG-SMITH: NEST DEFENCE BY STONECHATS are fringed by successional stages of birch and pine regeneration. A large population of stonechats inhabits the heathland, males defending territories between March and August (Greig-Smith, in preparation). During the study, most pairs made two or three breeding attempts in a season. Nests were well concealed on the ground, and clutch size varied from three to six (Greig-Smith, in preparation; see also Johnson 1971; Fuller & Glue 1977). Information was gathered by field observation of undisturbed birds or birds deliberately disturbed by the observer. Details of experimental procedures are given below. In addition, one series of experiments was conducted in the laboratory, on young birds temporarily removed from their nests. Tape recordings were made using a Nagra III tape-recorder with microphone and 60 cm parabolic reflector, and analysed with a Kay Electric Co. 6061A Sonagraph (using wide and narrow band filters), and a Tektronix R M 561A oscilloscope. Further recordings, for measurement of call-rates only, were made with a cassette recorder. Nestling weights, feeding behaviour, and details of breeding cycles are taken from GreigSmith (in preparation). Unless stated, statistical tests are two-tailed. I. The Nature of the Defence Calls Differences in Response to Raptors and NestPredators

Two classes of predators might affect breeding stonechats: those which are a threat to adult birds, and those preying only on eggs or nestlings. Table I shows the behaviour of adults on 100 occasions that a potential nest-predator entered territories (taken from experiment 5 below), and 48 occasions that a bird of prey was present. Most records of raptors were of kestrels (Falco tinnunculus), but there were also hobbies (Falco subbuteo), sparrowhawks (Accipiter nisus), hen harriers (Circus cyaneus), buzzards (Buteo buteo) and great grey shrikes (Lanius excubitor). It is clear from Table I that stonechats tended to be almost silent, and hidden in vegetation, when raptors were present, whereas the arrival of a nest-predator led to long series o f calls from exposed perches. On the five occasions that there were many calls in the presence of raptors, the stonechats had already called in response to me, and the calls were probably not a reaction to

605

the raptor. There was no indication of any difference between male and female parents in their response, and all raptor species were treated similarly. When stonechats did give a few calls on a raptor's first appearance, they tended to do so more often if they had fledglings present in the territory (4/5 occasions, 80 ~ ) than if there were none (6/37 occasions, 1 6 ~ ; Fisher exact test, P -~ 0.008). Acoustic Properties of the Calls

The calls given on the approach of nestpredators were of two kinds, Whit and Chack (Johnson 1971). They are commonly given in mixed sequences, although both occur singly, and occasionally in unmixed bouts (Greig-Smith, in preparation). Sonagrams of the calls (Fig. 1) show that Whits are brief, modulated notes, their energy confined to a narrow range of frequencies, while Chacks are of similar duration, but cover a much wider range of frequencies. This difference carries implications for the relative ease with which a listener can locate the caller (see below). Chacks appeared to be slightly louder than Whits, but this was not measured. Since duration and pitch of the notes were not very variable (Greig-Smith, in preparation), these parameters are unlikely to permit any signal grading. This is more likely to involve the loudness or repetition rates o f the calls. I have relied on differences in call-rate rather than loudness, which besides being difficult to measure, is less reliable because its effects depend on the distance to the listener. My impression was that loudness and rate o f calling were correlated, loud calls being rapidly repeated. Table L Differences in the Responses of Stonechats to the Presence of Birds of Prey and Nest-Predators in their Territories

Birds

Stonechat perched exposed Many loud calls: 0-5 quiet calls: Stonechat hidden Many loud calls: 0-5 quiet calls:

Nest-

of prey

predators

5 7

85 2

0 36

0 13

Type of predator vs. exposure: G = 56.7, df = l, P < 0.001. Type of predator vs. calls: G = 126.2, df = 1, P < 0.001. Exposure vs. calls: G = 69.9, dr= 1, P < 0.001. Interaction: G = -- 44.5.

606

ANIMAL

BEHAVIOUR,

Temporal Pattern in Mixed Series of Calls Eighteen mixed series of calls made as I entered territories were tape-recorded, and the sequence of Whits and Chacks written down. A onesample runs test was carried out on each series, in order to determine whether the two calls occurred in a random order. In 13 series, the number of switches was significantly greater (P < 0.05) than expected in a random sequence, and in three other series was nearly so. Therefore there was a tendency for very few Whit-Whit and Chack-Chack sequences, producing an approximate alternation of calls. In some contexts, however, birds switched to long unmixed sequences of either Whits or Chacks (see below, and GreigSmith, in preparation). Possible Effects of the Calls In a broad sense, these responses can be regarded as 'alarm' calls, which may be of selective advantage in several ways (see Sherman 1977; Harvey & Greenwood 1978). Because the predators concerned represent little personal risk to the adult stonechats, some possible effects can be eliminated. Thus, manipulation of other vulnerable individuals (Charnov & Krebs 1975) and confusion of a predator's attack (Curio 1976) do not apply, while reciprocal altruism (Trivers 1971) is improbable because of the usually wide spacing of neighbours' nests. Communication that the prey has seen the predator (Smythe 1970) and inducement of mobbing (Curio 1975) are unlikely, since the birds normally remained at least 15 m from the 'predator', without any direct aggression. This leaves two possibilities: the calls might warn the caller's mate or offspring of danger; or they might distract the predator away from the vulnerable offspring (Sherman 1977; Harvey & Greenwood 1978).

7-

Y:5-

f

o~ 4c

_6

I

t~r

210-

7". . . . . . . . . . . . . . .

0

o5 Time (sec)

IO

Fig. 1. Diagrammatic s o n a g r a m of a W h i t - C h a c k - W h i t sequence of calls by an adult stonechat in response to a h u m a n intruder near its nest.

28,

2

Tests of Warning Effects In many species, a single call is sufficient to convey a warning of danger (e.g. Powell 1974), and the stonechats usually warned fledglings of the presence of raptors with fewer than five calls (see Table I). This suggests that a long series of calls is unlikely to function solely as a warning. However, the lack of risk to the parents from nest predators relaxes constraints on calling, while the amount of calling appropriate as a warning also depends on the action taken by those receiving it. If the latter respond by assessing the danger themselves (as adult stonechats did), or make an immediate escape (cf. Powell 1974), then a single call might suffice. On the other hand, if they hide, then the caller might provide continual information about the danger, through prolonged calling. This argument suggests that the first few calls might act to warn both the caller's mate and offspring, but the long series that followed are more ]ikely to be directed only at the offspring, or the predator. The first call as I approached was a Whit on 87/105 occasions (83 %), implying that Whits are more likely than Chacks to have a warning function. Such warnings during the incubation period must be directed at the bird's mate, and should be most frequent when the female was on the nest and unable to watch for danger herself. I found that males gave a few calls on 10/32 (31 ~o) occasions that the female was on the nest, and only on 4/28 (14 %) occasions that she was feeding elsewhere, but the difference is not significant (Z~ : 2.34, P > 0.10). A few field observations suggest that long series of Whits and Chacks might cause fledglings to remain silent and hidden. On four occasions, I entered territories where recently-fledged young were loudly begging from their parents, before the latter responded to my approach. As soon as the parents began rapid Whit-Chacks, the fledglings fell silent and moved out of sight. As I left the area, the rates of calling declined, and the fledglings again started begging. The response of nestlings to Whit-Chack calls was not observed in the field and therefore a series of laboratory experiments was run. Under controlled conditions it was possible to investigate separately the effects of Whits and Chacks. Experiment 1. During August 1978, two chicks were removed from each of three nests for several hours, for a series of indoor tests. They soon learned to beg noisily for pieces of mealworm presented with a pair of forceps. Each nestling was isolated, and given three tests, in which I

GREIG-SMITH: NEST DEFENCE BY STONECHATS measured the time (during a 20 s period) that it begged for a mealworm offered in this way. In the second test, a tape-recording was played in the background. This consisted of 10 Whits in 25 s, 10 Chacks in 25 s, or a control of background noise for 25 s (edited from mixed series of Whits and Chacks recorded in the field). Tests were repeated with the other two tapes, until all chicks had been tested with the three sounds, in random order. The results of this experiment are summarized in Fig. 2. There was considerable variation in the levels of begging by different individuals in their first test of each trio, perhaps related to hunger or age. All six nestlings greatly reduced their amount of begging while hearing Whits, resuming a high rate after the calls ceased. Some birds responded in this way to the control tape or the tape of Chacks, but others were evidently little affected. Analyses of variance showed that the effects of Whits on begging were highly significant (F (2, 3) = 38.02, P < 0.01), but those of Chacks and background noise were not (Chacks: F ( 2 , 3 ) ~ 1.72, P > 0 . 2 0 ; background noise: F ( 2 , 3 ) : 2.08, P > 0.20). This experiment demonstrates that the Whit calls of adults cause nestlings to remain silent. Since the chicks soon resumed begging, this warning would have to continue while the danger persisted, if it were to reduce the chicks' conspicuousness effectively. Chack calls did not produce a similar effect, and presumably do not function as a warning. A further difference between Whits and Chacks is revealed by the next experiment, which investigated the influence of the 'predator's' activity on calling. Control

Whits

Chacks

100 ~, 80

~ 40 ~

2O

Before During After tope t a p e tape

Before During After tope t a p e tape

BefOreDurin~l After tape tope ,tape

Fig. 2. Results of a laboratory:experiment to determine the effectsof adult calls on begging by stonechat nestlings (see text). Chicks from the same nest have symbols of the same shape.

607

Experiment 2. On 20 visits to a total of 10 territories, I approached to 30-50 m from the nest, and remained motionless for 5 min. Then, while stationary, choosing either the male or female parent, I counted the numbers of Whits and Chacks in 10 s, followed by the numbers in the next 10 s, during which I walked rapidly towards the nest. On all 20 occasions, the rate of Whits was greater as the 'predator' moved, whereas the rate of Chacks increased on 8 occasions, remained the same on six, and decreased on six (comparing Whits and Chacks, combining equal and decreased rates, Z~ = 17.14, P < 0.001). On average, the rates of Whirs (no./10 s) were 1.8 ~ 0.40 (mean :[: SE) before the predator's movement and 6.4 A: 0.93 after (t38 = 4.54, P < 0.001), while the rates of Chacks were 4.0 :k 0.95 before and 4.7 :k 0.94 after (t38 = 0.52, P > 0.50). The increase in danger represented by the predator's sudden movement towards the nest is reflected in an increase in the rate of warning, which would probably reinforce the nestlings' silence (cf. experiment 1). The rate of Chacks did not change consistently in this context, again implying that they do not provide a warning. Tests of Distraction Effects Experiment 3. If calls are given to distract a predator, the parents would be more likely to call in contexts that would cause the predator to move away from the young. I tested this prediction by allowing the behaviour of parents to determine my route on visits to territories during the late nestling period. On entering a territory, I walked slowly in a straight line towards the nest. As soon as one of the parents called, I changed direction to walk towards the bird, and repeated this every time it moved. Changes were scored as (i) causing me to move away from the nest (i.e. increasing the angle between my line of approach and a straight line to the nest), (ii) causing me to move towards the nest, or Off) making no difference. Throughout the visit I continually measured the bird's rate o f calling (calls~5 s), counting either Whits or Chacks in each visit. Figure 3 shows an example of the movements of a male stonechat and my corresponding movements during one test. I was led well away from the nest, as o n 14 other tests, while my route was little changed on the remaining six tests. The call-rates of both Whits and Chacks increased during tests, as I penetrated further into the territory. However, considering changes

608

ANIMAL

BEHAVIOUR,

28,

2

Short-term changes in the rate of Whits (experiment 2) could provide detailed information on the behaviour of the predator, comparable to the findings of Morton & Shalter (1977) on Carolina wrens (Thryothorus ludovicianus). Both calls were also given in contexts other than the approach of a predator, when these effects would not apply. However, the rates of calling, and the sequence and proportions of Whits and Chacks on these occasions were different (Greig-Smith, in preparation). Using these contexts to identify the 'messages' (Smith 1968) of the calls suggested that Whits always occur in contexts where they communicate danger, while Chacks were used more widely, apparently emphasizing the caller's presence. The different acoustic properties of the calls (Fig. 1) suggest that, because of their narrow frequency range, Whits make the caller more difficult to locate than Chacks (Marler 1955, 1967; Konishi 1973). Selection should favour the use of easily locatable calls for purposes such as distraction displays, when the caller's position must be revealed. Functions such as warning could be better served by poorly locatable calls if exposure of the caller is unnecessary. The ability of avian predators to locate the source of calls with narrow frequency ranges is better than previously supposed (Shalter & Schleidt 1977; Shalter 1978). However, this does not affect the argument about the kinds of calls appropriate to different functions, since the important factor is the relative properties of alternative calls. The uses of Whirs and Chacks as warnings and distraction respectively are consistent with this argument.

ceu~rv~

Fig. 3. Example of distraction by an adult stonechat. Open squares indicate the bird's successive perches, and filled squares the positions at which the observer changed direction in response to its movements (see text). in rates which occurred at each of the stonechats' moves, there was a tendency to decrease the rate of Chacks more than expected on moves towards the nest, but no comparable tendency in the rate of Whits (Table II, all tests combined). These results indicate that predators which follow one of the parents would be drawn away from the nest, and probably also from fledglings. Judging from their changing rates when the bird moves, Chacks are presumably more effective than Whirs as distraction. Discussion of the Roles of Whirs and Chacks

Adult stonechats make several responses to the presence of a human 'predator' during breeding. A few preliminary calls (usually Whits) warn the caller's mate, and once the eggs have hatched, the nestlings as well. This was followed by long mixed series of Whits and Chacks, the former causing the young to remain silent and hidden, the latter used to distract the predator away from them. Johnson (1971) suggested that Whits function as distraction and Chacks as warnings, but his proposal was unsupported by evidence.

TaMe II. Changes in the Rates of Whit and Chaek Calls by Stonechats Moving Towards and Away from their Nests

Change in rate of Chacks

Change in rate of Whits

Decrease

Increase or equal

Movement away from nest, or along same line

11

40

12

16

Movement towards nest

11

10

5

9

~r = 6.18 P < 0.02

Increase Decrease or equal

X~ = 0.22 P > 0.50

GREIG-SMITH: NEST DEFENCE BY STONECHATS II. Spatial Variation in Rates of Calling

This section examines how the rates of calling by parents vary according to the location of danger within their territory. The observations were made late in the nestling period, when average call-rates were high (see below). Experiment 4. On entering a territory, I walked in a straight line towards the nest, and stopped every 20 paces (1 pace ~ 1 m) to count the rates (caUs/30 s) of Whits and Chacks by the parents. Male and female calls could not always be reliably distinguished at long distances, and therefore combined rates were measured. A few separate counts indicated that the sexes responded similarly. After reaching the nest, the procedure was repeated as I walked away from the nest, usually following the same route out of the territory. The results of nine tests are summarized in Table III, showing that in all tests the rate of Whits increased as I approached and decreased as I departed. The rate of Chacks also increased as I approached, but on departure the proportion of tests on which there was a decline was not significant. This meant that the percentage of calls that were Whits did not change on approach, but declined on departure (Table 1II). Figure 4 presents the average values of callrates at various distances from the nest. Because it was not possible to standardize the distances at which calls were counted, averages have been obtained by interpolating between some data points, and it is not appropriate to examine the slopes of the curves statistically. It appears that the decline of Whits once I moved away was

609

more rapid than their increase on my approach. This is consistent with their role as a warning signal. The pattern of gradual increase indicates that the birds raise their investment in warning as the danger increases, thus ensuring their offspring's concealment. The rate of Chacks did not fall as I moved away, presumably as distraction was continued (see also experiment 3). Their increase is very similar to that of Whits (Fig. 4), suggesting that the birds' tendency to start distraction, like warnings, is dependent on the degree of danger. Nestlings and fledglings could be made aware of increasing danger by changes in the rate of either Whits or Chacks, but could only assess its relaxation from a fall in the rate of Whits. They could also obtain this information from a fall in the proportion of Whits in call sequences (Fig. 4), but I have no evidence to suggest that they do. III. Seasonal Variation in Rates of Calling Average Patterns of Change

The fact that Whits are apparently given principally to warn nestlings or fledglings implies that they would not be necessary before hatching, while the levels of both Whits and Chacks would be expected to decline as the young attained independence. I therefore investigated the patterns of change in call-rates through the nesting cycle. Also, since many of the stonechats made three breeding attempts during the season, I searched for differences between successive broods. Experiment 5. Territories were visited regularly and the rates of Whits and Chacks of each parent were measured (calls/30 s). When more than one

Table Ill. Changes in Rates of Whit and Chack Calls as a Human 'Predator' Approaches and Departs from the Nest, during the Late Nestling Period

Whits Whits (a) Predator approaching nest

Chacks Whits + Chocks

No. of occasions on which rate increased

9

8

6

No. of occasions on which rate decreased

0

1

3

0.004

0.039

0.520

No. of occasions on which rate increased

0

2

1

No. of occasions on which rate decreased

9

7

8

0.004

0.179

0.039

P (Binomial test) (h) Predator leaving nest

P (Binomial test)

610

ANIMAL

BEHAVIOUR,

count was made on a particular day, the totals were averaged. During most of the nesting cycle, I made counts while standing near the nest, where call-rates were highest (Fig. 4), but after fledging the parents were followed until the maximum call-rate was reached, presumably when I was near to hidden fledglings. This procedure minimized variations resulting from the location and activities of the 'predator' (see above). Preliminary data were collected between 20 June and 11 September 1977, and more extensive records were made throughout the 1978 breeding season (26 February to 16 September). Observer

Observer

opprooching

departing

3o-

;R~}

20-

"~ (.3 10O-

.

~

Averagerate

~

28,

2

Figure 5 illustrates how call-rates changed through the breeding season in one territory, where there were three successful breeding attempts. Overall, there is a clear similarity between Whirs and Chacks, and between the male and female. Three major peaks are discernible, broadly coinciding with the three nesting cycles, but there was considerable day-to-day variation. In order to examine the pattern in relation to the stage of breeding, I constructed average curves from data for 48 nesting cycles, of 30 pairs. Each nesting cycle was divided into four stages: egglaying (3 to 6 days); incubation (about 14 days); nestling stage (about 14 days); and fledgling stage (arbitrarily defined as a 15 day period after fledging). Too few observations of nest-building were made to allow the inclusion of pre-laying stages. To reduce individual variation, and permit comparison of pairs in which the length of stages varied, I calculated average values for the three thirds of each stage. Results for different birds were then averaged, considering first, second and third broods separately. These final values are presented in Fig. 6.

t ........

120

0

100 40

20

~ : e

to

20 10 0

401

cks rate

0

1(~0

Whits

50 20

0

120

o# Whirs

9

10 0

60-

40-

hits @

20-

o-

. . . . . .

50

12~io'o8~ 6'o,~ ~ o 2'o4'0~ ~o,~o

~ Chacks

10 0 i

Distance from nest ( m )

Fig. 4. Changes in the r~ttes of Whit and Chack calls, and the percentage of calls that were Whits, as a human intruder approached and departed from the nest. The

curves are averages of nine occasions, with male and female calls combined.

r

April

i

May

J

June

i

July

i

August

Fig. 5. Seasonal changes in the rates of Whit and Chack calls by male and female stonechats in one territory in 1978. Three breeding cycles are indicated by egg-laying period (close shading), incubation period (unshaded), and nestling period (wide shading).

GREIG-SMITH: NEST DEFENCE BY STONECHATS

Figure 6 shows very similar patterns for male and female call-rates, in both Whits and Chacks (Spearman rank correlation coefficients between 0.77 and 0.94, P < 0.01). Whir and Chack calls were also highly correlated within each sex (rs between 0.82 and 0.96, P < 0.01). Two major features in the pattern of call-rates are evident from Fig. 6. First, both Whits and Chacks increased rapidly after the eggs hatched, to peak late in the nestling stage or early in the fledgling stage, after which they rapidly declined. This change was very similar in first, second and third broods, and the greatest level of calling reached did not vary through the season. Second,

611

in some broods there is a smaller peak early in the cycle, around the start of incubation. This is most obvious in third broods, but also occurred widely in second broods, and in a few first broods (e.g. Fig. 5). High rates of calling early in a second or third cycle might be attributable in part to the tail-end of the major peak from the previous cycle, since the interval between fledging and the next egglaying was usually no more than about 15 days ( Morgan & Davis 1977; Greig-Smith, in preparation). However, in most cases the peak at the start of incubation followed an absence of calling

WHITS

CHACKS

30-

First broods 0-

-O---Q

-

20~II

1

10

I

I

I

1

I

I

I

I

I

I

I

i

I

I

I

I

1

(

I

I

I

Second broods

t

i '~ 0

I

I

t

I

I

I

I

1

1

I

I

I

I

1"

I

I

~\0

,/~'"

,

I

I

I

I

I

I

I

Q, /

Third broods

20t

!

/r

P---O --z3"

I

I

1

--

0

till

10

0 I

I

I

I

I

I

t

1

2

3

1

2

3

1

Egg-laying

Incubation

"

I '

"~"

j '

~

" I

2

3

1

2

3

Nestlings

Fledglings

I

1

I

I

I

I

'f

I

t

I

1

I

I

2

3

1

2

3

1

2

5

1

2

3

Egg-IQying

Incubation

Fig. 6. Average male and female call-rates through the first, second and third breeding cycles of the season. Each stage of the cycle is divided into thirds (see text), and average rates calculated for each period (sample sizes between two and 16 birds). Males are indicated by O . . . . O, females by 9 -9

Nestlings

Fledglings

612

ANIMAL

BEHAVIOUR,

during egg-laying, so that an additional factor is evidently involved, as it must be in first broods. The only differences apparent between male and female call-rates are that females on average achieved a slightly higher rate of Whits, and males a higher rate of Chacks in the major peak at fledging. In second broods, males show higher rates of Chacks early in the cycle, probably due to their care of previous fledglings while the females incubated the next clutch. This effect is not evident in third broods, for two reasons. Many pairs attempting a third brood had failed to raise young from their second attempt, and their calling was curtailed (see below). Also, the average interval between first and second broods was shorter than between second and third broods (Greig-Smith, in preparation). Reasons for the Peak at Fledging

To help explain the major peak of call-rates the observed pattern of increase can be compared to changes predicted under several alternative hypotheses. These arguments must take into account both the benefits and costs of defence. Cost should ultimately be measured in terms of decreased ability to raise future offspring (Trivers 1972; Dawkins & Carlisle 1976; Maynard Smith 1977), but is expressed in loss of energy or time needed to produce future young, or by the risk of incurring injury, death or loss of breeding status. Caution is required, because the relationship of such short-term effects to future reproductive success may not be simple or obvious (see e.g. Maynard Smith 1977). Three hypotheses predict changes in the level of parental investment through a breeding cycle. (i) For a given danger, the value of the young birds to their parents increases through the cycle, and parents might therefore become willing to defend their brood more strongly. This could arise because the further effort (feeding, etc.) required to raise the present brood to independence becomes progressively less relative to that demanded by a replacement brood in the event of predation (Barash 1975). Since investment in a replacement brood generally involves repeating the same course of investment given to present offspring, their value to parents will be related to the cumulative total already invested in them (see e.g. Dawkins & Carlisle 1976). Since feeding is the principal component of investment after egg-laying and incubation, cumulative investment can be estimated by considering rates of feeding nestlings (see Fig. 9).

28,

2

Andersson et al. (1980) developed an alternative model which predicts an increase in the value of offspring through the nestling stage. This is based on the assumption that the offspring's chances of survival to fledging progressively increase with age, so that parents should become more willing to accept the risks of defending them. For stonechats, these risks are probably small, but there may be other costs of defence, and the prediction of Andersson et al.'s model remains unchanged when the parents' risk of death is negligibly small. I measured the nestlings' probability of survival from the proportion of chicks dying on each day after hatching (see Fig. 9). Hypothesis (ii) is that attacks by predators become more dangerous as the young birds grow, because the predators value the larger food items more highly, and stronger defence is required to deter them. Nestling growth was almost independent of brood size, and the pattern was approximately sigmoid, maximum weight being reached shortly before fledging (Greig-Smith, in preparation). A brood's total weight can be used to describe its changing value for the predator, and the predicted pattern of defence (see Fig. 9). This assumes that the predator can assess the stage of breeding before finding the nest, which may be unlikely (see below). If it can not, the predicted pattern is a constant level throughout the cycle. Under hypothesis (iii), the risk of predators detecting the nest increases because of a rise in its conspicuousness. This differs from hypothesis (ii) in the lack of any change in the predator's persistence. Conspicuousness could be influenced by the rate at which the parents visit the nest, the rate at which chicks beg, the loudness of their begging calls, and disturbance of the nest area. I exclude the last possibility, because disturbance to vegetation was minimal, and the nestlings' Visits by pQrents

III1[

i

Begging by nestlings

|I.[I [

II

I

Time (rain)

Fig. 7. Begging by nestling stonechats in relation to feeding visits by the adults. Begging calls were taperecorded with a recorder hidden beside the nest while the parents' visits were simultaneously observed from

over 150 m away.

GREIG-SMITH: NEST DEFENCE BY STONECHATS faecal sacs were removed by the parents to at least 50 m from nests. The other factors were quantified as follows. The rate of visiting the nest rose during the first half of the nestling stage, but declined in the second half (see Fig. 9). The begging rates of nestlings were measured by placing a tape-recorder close to the nest and recording their calls while observing the parents' behaviour from a distance. The example in Fig. 7 (confirmed by further unpublished data) shows that chicks only begged loudly and persistently when a parent brought food to them although there were occasional brief calls at other times. This result means that changes in conspicuousness due to the frequency of begging would follow the same pattern as those due to feeding rates. Thus visual and aural predators would experience similar changes in information. To determine whether the volume of begging calls changed, I taperecorded under standard indoor conditions the calls of the three pairs of chicks used in experiment 1, respectively 7, 11 and 13 days after hatching. The relative loudness of calls was determined from the height of spikes produced on an oscilloscope screen (Fig. 8). Very young chicks ( < 2 days after hatching) often gaped silently when I visited nests, and these points have been included in Fig. 8. Clearly, older nestlings beg more loudly, To determine which hypothesis best predicts the observed patterns of increase in defence calls, average rates of Whits and Chacks over the first 11 days of the nestling stage were compared to five predicted patterns (Fig. 9). Table IV shows that the rates of both calls were highly correlated with the cumulative total of feeding visits, the probability of nestling survival, and the weight of nestlings. They were also significantly correlated, though less strongly, with the frequency of begging weighted by the changing loudness of the begging calls. Correlations with the observed rate of feeding were poor. These results imply that hypotheses (i) and (ii) are almost equally good predictors of the level of investment in defence, while hypothesis (iii), based on the frequency and loudness of begging, also predicts it well. This does not necessarily mean that all three factors are involved, since the predicted curves in Fig. 9 are interrelated, and some fortuitous correlations are to be expected. However, it is unlikely that hypothesis (iii) provides the principal influence on call-rates, because the weighted frequency of begging continues to decline after the 10th day, whereas

613

parental call-rates continue to rise. The importance of hypothesis (ii) depends on the feasibility of predators being able to assess the stonechats' stage of breeding without actually finding the young. Observing the parents' rate of visiting the nest, or listening to the rate of begging would not be sufficiently accurate guides (see Fig. 9). It would probably be difficult to assess the loudness of chicks' calls if their distance was unknown, while the parents' rates of calling vary so much depending on the location and activities of intruders that they would not provide a simple indication of the stage of breeding. Further, if the defence calls themselves were used by the predator, a compensatory depression of their rate might be expected. Therefore, it seems likely that hypothesis (i) may be the best explanation of changes in call-rate, implying that, as Barash (1975) and Andersson et al. (1980) suggest, the value of the nestlings is the principal determinant of the level of defence. Nevertheless, it seems unlikely that the dramatic change in the rate of nest visits by parents after hatching is entirely unimportant. 0/ / / / / / / / r t--

/ /

//

9

_o 10-

/ /

0 CU

/ / / / / / / / / / /

Oi

i

!

i

i

i

!

0

2

4

6

8

10

12

t4

Days Fig. 8. Relative loudness of the beggingcalls of stonechat nestlings in relation to their age. Birds of ages 0-1 days were assessed in the field; others were tape-recorded and analysed with an oscilloscope(see text). The relationship is described by y = 1.25x--0.80 (FI.6 = 19.7, P < 0.01).

ANIMAL

614

BEHAVIOUR,

The fall in rates of calling after fledging (see Fig. 6) is probably dictated by several effects. As the young birds approach independence, the danger posed by nest-predators becomes less, and Parental call rate

~,

28,

2

distraction and warnings would be less necessary. Also, the conspicuousness of each birds' location is lowered, because they scatter, move frequently, and are fed at lower rates than nestlings.

Probobilityof nestling survival

Rate of feeding nestlings

__.+

+.,+._

50-

Whits~

~o.9

f

to

20-

o-9-

o~

o

g

10-

~o.7-

s nO. 8-

0-

o

a

m 03 C

Rate of begging weighted by loudness of calls

8-

8 x

.E E tO

>0"5I

24

I

I

8

b

I

lb

Cumulative total of feeding visits

6-

-~ 500- ~

4-

"- 200>

2O-

l

I

i

I

I

I

0

2

4

6

8

10

d

I

I

Weight of nestlings 16122 8c9

o~

g

I n

100

4-

I

0

I

I

I

2 4 6 8 e Nestling age (days)

0-

I

10

I

I

0

2

I

4f

I

I

l

6

8

10

Fig. 9. Average changes over the first ten days of the nestling stage in" (a) the rate of Whits and (;hacks, with male and female calls combined; (b) the probability of nestling survival; (c) the frequency of feeding nestlings; (d) the frequency of begging calls, weighted by their relative loudness; (e) the cumulative total of feeding visits; and (f) the weight of nestlings. Table IV. Correlations Between the Rates of Whits and Chacks on the First 11 Days after Hatching, and Predicted Levels of Parental Investment Based on Hypotheses (i)-(iii) (See Text)

Correlation with rate of Whits r

Correlation with rate of Chacks

P

r

P

Hypothesis (i): changing value of offspring to parents Cumulative total of feeding visits: Probability of chicks' survival:

0.99 0.98

<0.001 <0.001

0.95 0.94

<0.001 <0.001

Hypothesis (ii): changing value of offspring to predator Total weight of the brood:

0.99

<0.001

0.96

<0.001

Hypothesis (iii): changing conspicuousness of the nest Rate of feeding visits by parents: Frequency of begging by nestlings, weighted by loudness of calls:

0.17 0.87

>0.10 <0.01

0.36 0.91

>0.10 <0.001

GREIG-SMITH: NEST DEFENCE BY STONECHATS Reasons for the P e a k at tile Start o f Incubation

Can the sanae hypotheses explain the peak in call-rates around the start of incubation, which differed in being smaller, and occurring in only some cycles ? There is no sudden change in brood weight at that time, to suggest an increase in value to predators. Conspicuousness of the nest site does increase, owing to females starting to incubate in stints of 30.1 ~ 3.78 rain (mean i sB, n ----9), separated by feeding intervals of 30,6 ~: 2.48 min (n = 25); this leads to about 2 visits/ h, compared to a single visit each day during egg-laying. However, this rate is then maintained until hatching, and does not explain the peak in call-rate. This leaves the possibility that the value of the young to their parents is temporarily raised, a change which is difficult to predict by a prospective model of parental investment. The data in Table V suggest that this peak may be dependent on the body condition of the female, which is likely to be impaired following a successful breeding attempt that demanded heavy investment in feeding young. Table V shows that there was a peak in call-rates during 36 ~ of first broods, which did not follow an earlier nesting attempt; during 4 3 ~ of later broods that followed a nesting failure; and during 70 ~ of broods that followed a successful attempt. Comparing the last against the first two categories combined, the tendency to have a peak after successful attempts was significant (Z~ ---19.1, P < 0.001). Body condition might further be low after laying a large clutch. Within each category in Table V, the average dutch size of broods with peaks in calling was greater than those of broods without peaks; the difference is significant among first broods, and almost so in broods after failures.

615

Therefore it appears that females with poor body condition due to recent expenditure in breeding activities are more likely to defend their offspring around the start o f incubation than earlier or later in this part o f the cycle. Females in better condition are less likely to provide any defence. Male behaviour follows a similar pattern, due probably to his dependence on a single mate within the season (see below). IV. Rates of Calling in Relation to Brood S i z e

Some pairs of birds reached very different levels of calling at comparable stages of successive nesting cycles (e.g. Fig. 5). This might be caused in part by differences in brood size, since the value of a brood should be proportional to the number of nestlings in it assuming, as their comparable growth rates make likely, that individuals from large broods do not survive markedly less well than those from small ones. To test this, I calculated the average rates of calling during the final third of the nestling stage, for pairs having between one and six chicks (Fig. 10). Pairs without chicks did not call (see below), while the relationship between brood size and call-rate differed between Whirs and Chacks. The rate of Whits was significantly dependent on brood size (y ~ 4.6x q- 14.8, F(1, 4) = 8.6, P < 0.05), whereas the rate of Chacks was not (F(1,4) = 0.0002, P > 0.50) (Fig. 10). Assuming that predators would be unlikely to assess brood size without finding the nest, these patterns might reflect either the greater conspicuousness of large broods (due to louder total begging, and the higher rate of visiting the nest by parents), or the greater value of a large brood. Under these explanations, both Whits and Chacks are expected to increase with brood size.

Table V. Comparison of Nesting Cycles in which the Parents Did or Did Not Show a Peak in Call-Rates at the Start of Incubation

MannWhitney With No peak U-test peak in in call (one-tailed) call-rates rates on clutch size First broods of the season Broods following a failed attempt Broods following a successful attempt

4 (5.00) 3 (5.67) 7 (5.33)

7 (4.15) 4 (4.67) 3 (5.00)

P = 0.003

4o

B~

o"

~o

P = 0.057 P = 0.131

Figures are the numbers of broods and, in brackets, their average clutch sizes.

Fig. 10. Average call-rates in the final third of the nestling stage, by stonechats with different brood sizes. Whirs are indicated by 9 . . . . O, Chacks by 9 O. Male and female calls are combined.

ANIMAL

616

BEHAVIOUR,

However, it is possible that the effectiveness of distraction is not improved by increasing the callrate above a certain level, which might be attained by the late nestling stage in all broods. A similar kind of explanation could apply to the apparent asymptote reached in the rate of Whits (Fig. 10), although it is also possible that there are constraints on the maximum rates which the birds can produce. It could be argued that brood size and the parents' call-rates are independent consequences of some other variable, such as the 'quality' of the habitat. This objection is unlikely to apply here, since many of the broods included in Fig. 10 were reduced from larger clutches by infertility o f eggs, chilling, or other causes (Greig-Smith, in preparation). However, a more direct test can be made by searching for temporary reduction in call-rates when single chicks were lost. This occurred naturally in three broods, while in three others I temporarily removed pairs of chicks. The call-rates of all six males and five of the female parents were measured before and after the change in brood size. Though call-rates are expected to increase daily during the nestling stage (Fig. 9), nine of the 11 birds actually reduced their rates of both Whits and Chacks (Sign test, P = 0.033). This indicates that parents adjusted their call-rates according to the number of chicks defended. The adjustment in the rate of Whirs was expected from the relationship in Fig. 10, but Chacks also apparently fell, despite the lack of a significant trend in Fig. 10. Not surprisingly, on 15 occasions when predators removed whole broods, the parents immediately stopped calling in response to my visits.

28,

2

V. Consequences of L o w Calling Rates

The results presented so far suggest that the parents' calls reduce predation. This can be tested directly by comparing the call-rates of parents whose young were later taken by predators with those of birds that raised young successfully. In six territories I measured call-rates shortly before predation; these are listed in Table VI, together with the average call-rates of equivalent pairs (i.e with the same brood size, and measured at the same stage of the nesting cycle). I f predation was independent of parental callrates, equal numbers of successful pairs should have higher rates as lower rates compared with those pairs suffering predation. However, for Whits, 30 out of 42 successful pairs had higher rates, and 12 had equal or lower rates (Z~ 7.71, P < 0.01). For Chacks, the trend is in the same direction, but is not significant (23 out of 42 had higher rates, and 19 were lower or equal, Z~ = 0.76, P > 0.30). On average, the pairs suffering predation made 0.42 times as many Whits and 0.70 times as many Chacks as successful pairs. Since there was no indication that low rates of calling were correlated with poor parental care of other kinds, these findings imply that birds making lower than average rates are selected against by being more liable to suffer nest predation. It is difficult to separate the importance of Whits and Chacks, since rates of the two calls were normally correlated (see above). However, the apparently stronger selection on Whits (Table VI) suggests that failure to keep young birds silent and hidden is more likely to aid a predator than failure to make adequate distraction displays.

Table VL Differences in the Rates of Defence Calls between Parents whose Broods were Taken by Predators, and Those Rearing Young Successfully

Nests sufferingpredation Whits Nest A Nest B Nest C Nest D Nest E Nest F

0 17 20 26 1 0

Average for 'equivalent' nests without predation*

Chacks

Whits

Chacks

n

0 30 11 18 6 0

6.8 23.3 33.4 29.0 3.2 2.5

5.3 14.8 21.8 18.0 9.1 2.3

9 11 10 3 4 5

Figures are rates of calling (calls/30s), with male and female calls combined. *'Equivalent' nests had the same number of nestlings, and were assessed at the same stage of breeding as the nest which suffered predation. Sample sizes indicate the number of 'equivalent' nests in each case.

GREIG-SMITH: NEST DEFENCE BY STONECHATS General Discussion This study has shown that parent stonechats defend their young by warning them of the approach of danger, and distracting predators away from them. These ends are accomplished by separate calls, and although less economic than combining both functions in one display, the possession of two calls is probably adaptive, for two reasons. Although rates of Whits and Chacks were often correlated (see e.g. Fig. 6), there were contexts in which high rates of one, but not the other, were appropriate (see e.g. Table II, Fig. 4). Also, the effectiveness of distraction and warnings would be enhanced by different acoustic properties, affecting the locatability of the caller. The rates of the defence calls are highly variable. In some cases, high call rates appeared to be associated with a high level of predation risk (e.g. calls at various distances from the nest: experiment 4), but in other eases were more easily explained as a consequence of the changing value of the young birds to their parents (e.g. changes in call rates during the nesting cycle). The latter results are well predicted by the theory (Trivers 1972, 1974) that the willingness of parents to protect their young is influenced by the needs of both present and future offspring. Viewing this defence as 'parental investment' in Trivers' sense requires that its costs should be measured in terms of decreased ability to invest in future offspring. However, in a field study it is much easier to assess immediate costs, such as expenditure of time or energy (cf. Lazarus & Inglis 1978), or incurring risk of mortality. In the present case, the energy costs involved in the calls are unlikely to be great or difficult to recover, except perhaps when calling is prolonged. The birds may devote time to defence which would otherwise be given to feeding, and might thereby delay or reduce their fitness when attending later broods. The high vigilance of stonechats (GreigSmith, in preparation) means that the parents would be unlikely to increase greatly their risk of attack by birds of prey when they perched exposed giving defence calls. Overall, it seems likely that defence is a low cost activity. This means that predicted differences and relationships could be masked, if birds were able to call at an optimum rate despite differences in the value placed on offspring. The results in Fig. 10, showing a levelling-off in the rate of Whits, and a constant rate of Chacks in broods of more chicks, can perhaps be accounted for in this way. Low costs might also explain why no change was found in the pattern of defence calls through

617

the breeding season, first, second and third broods receiving similar levels of defence at comparable stages (experiment 5). It can be predicted that in species such as temperate passerine birds, in which there is a low probability of surviving the winter to breed in future years, the costs of investing in late broods are less than for early broods which could be replaced, and therefore a higher level of defence would be seen (Weatherhead 1979). This argument predicts a critical point during the season, after which a replacement brood is not possible. However, if the parents were able to provide an optimum level of defence for early broods, no such change would appear. It is also possible that the consequences of diverting energy to defence from processes such as deposition of body fat, moult, etc., would be different at different stages of the breeding season. There were no major trends in the frequency with which birds of prey occurred in stonechat breeding areas through the season (Fig. 11), suggesting that the personal risk of calling is unlikely to cause differences between early and late broods. I found no consistent differences between male and female parents in any aspect of nest-defence, except for responses during the incubation of second broods (Fig. 6), which reflects a division of labour rather than different levels of defence in one context. At first sight, this is unexpected, since males are thought to be predisposed to provide less parental care than females because they allocate fewer resources to gamete production (see e.g. Dawkins & Carlisle 1976). However, this difference would be outweighed if the male's prospects of attracting an alternative mate were

.c o

2

0 29 Augusl

1 March 7irneof season (14 day periods)

Fig. 11. Seasonalchanges in the numbers of birds of prey seen in stonechat breeding areas during 1978.

618

ANIMAL

BEHAVIOUR,

poor (Maynard Smith 1977), in which case he should value his present offspring as highly as does the female. Although polygyny is known in stonechats (Johnson 1961, 1971), its frequency is extremely low, and in the population I studied, no birds mated with more than one partner during a season. Therefore, the male's defence should be adjusted to the same level as the female's, as observed. Nevertheless, it is perhaps surprising to find that in a highly dimorphic species such as the stonechat, there is little division of labour in defence. Finally, we may ask whether my findings are likely to apply to other species. Few detailed studies of this type of behaviour have been attempted, but information is available on changes through the breeding cycle in the defence displays o f several bird species. An increase in defence after hatching occurs widely (e.g. Simmons 1955; Gramza 1967; Erpino 1968; Curio 1975, Weatherhead 1979), though Barash (1975) found a rise in the strength of distraction by Alpine accentors Prunella collaris starting before hatching. In most cases defence consists of calls, or injury-feigning, which might increase for similar reasons to those determining the stonechats' call-rates, as Erpino (1968) and Barash (1975) speculate. In the pied flycatcher Ficedula hypoleuca there is evidence of a peak in mobbing display during nest-building and egg-laying (Curio 1975, page 30), which might reflect some aspect of commitment to egg production, as suggested for the stonechat. Thus among birds at least, change in defence displays may follow a general pattern. Detailed comparison of species would be valuable, particularly between those predicted to show different patterns, e.g. altricial versus precocial (Barash 1975), single versus multiplebrooded, hidden versus exposed nests (Ricklefs 1977), monogamous versus polygamous.

Acknowledgments I am grateful to the Conservators of Ashdown Forest for permission to conduct the study there, and to the Natural Environment Research Council for finance. Tony Martin helped with some fieldwork, and first-year ecology students at Sussex assisted with pilot experiments. P. J. Weatherhead kindly showed me a pre-publication copy of his paper. I received much helpful discussion, and comments on the manuscript, from P. J. Greenwood, J. A. Greig-Smith, J. Lazarus, J. Maynard Smith, D. I. Rubenstein, P. J. B. Slater, and P. H. Harvey, who also suggested several of the analyses.

28,

2

REFERENCES Andersson, M., Wiklund, C. G. & Rundgren, H. 1980. Parental defence of offspring: a model and an example. Anim. Behav., 28, 536-542. Barash, D. P. 1975. Evolutionary aspects of parental behaviour: The distraction display of the alpine aecentor Prunella eollaris. Wilson Bull., 87, 367373. Charnov, E. L. & Krebs, J. R. 1975. The evolution of alarm calls: altruism or manipulation? Am. Nat., 109, 107-112. Curio, E. 1975. The functional organization of antipredator behaviour in the pied flycatcher. A study of avian visual perception. Anita. Behav., 23, 1-115. Curio, E. 1976. The Ethology of Predation. Berlin, Heidelberg, New York: Springer. Dawkins, R. 1976. The Selfish Gene. Oxford: Oxford University Press. Dawkins, R. & Carlisle, T. R. 1976. Parental investment, mate desertion and a fallacy. Nature, Lond., 262, 131-133. Erpino, M. J. 1968. Nest-related activities of Blackbilled Magpies. Condor, 70, 154-165. Fuller, R. J. & Glue, D. E. 1977. The breeding biology of the Stoneehat and Whinchat. Bird Study, 24, 215-228. Gimingham, C. H. 1972. The Ecology of Heathlands. London: Chapman & Hall. Gramza, A. F. 1967. Responses of brooding nighthawks to a disturbance stimulus. Auk, 84, 72-86. Harvey, P. H. & Greenwood, P. J. 1978. Anti-predator defenee strategies: some evolutionary problems. In: Behavioural Ecology: an Evolutionary Approach (Ed. by J. R. Krebs & N. B. Davies). Oxford: Blackwell. Johnson, E. D. H. 1961. The pair relationship and polygyny in the Stonechat. Brit. Birds, 54, 213225. Johnson, E. D. H. 1971. Observations on a resident population of Stonechats in Jersey. Brit. Birds, 64, 201-213 and 267-279. Konishi, M. 1973. Locatable and non-locatable acoustic signals for barn owls. Am. Nat., 107, 775-785. Kruuk, H. 1964. Predators and anti-predator behaviour of the black-headed gull (Larus ridibundus). Behaviour, SuppL, 11, 1-129. Lazarus, J. & Inglis, I. R. 1978. The breeding behaviour of the pink-footed goose: parental care and vigilant behaviour during the fledging period. Behaviour, 65, 62-88. Marler, P. 1955. Characteristics of some animal calls. Nature, Lond., 176, 6-8. Marler, P. 1967. Animal communication signals. Science, N.Y., 157, 769-774. Maynard Smith, J. 1977. Parental investment: a prospective analysis. Anim. Behav., 25, 1-9. Morgan, R. A. & Davis, P. G. 1977. The number of broods reared by Stonechats in Surrey. Bird Study, 24, 229-232. Morton, E. S. & Shalter, M. D. 1977. Vocal response to predators in pair-bonded Carolina Wrens. Condor, 79, 222-227. O'Connor, R. J. 1978. Brood reduction in birds: selection for infanticide, fratricide and suicide? Anita. Behav., 26, 79-96.

GREIG-SMITH: NEST DEFENCE BY STONECHATS Powell, G. V. N. 1974. Experimental analysis of the social value of flocking by starlings (Sturnus vulgaris) in relation to predation and foraging. Anim. Behav., 22, 501-505. Ricklefs, R. E. 1977. Reactions of some Panamanian birds to human intrusion at the nest. Condor, 79, 376-379. Shaltcr, M. D. 1978. Localization of passerine seeet and mobbing calls by Goshawks and Pygmy Owls. Z. Tierpsychol., 46, 260-267. ShaRer, M. D. & Schleidt, W. M. 1977. The ability of barn owls, Tyro alba, to discriminate and localize avian alarm calls, lbis, 119, 22-27. Sherman, P. W. 1977. Nepotism and the evolution of alarm calls. Science, N.Y., 197, 1246-1253. Skutch, A. F. 1976. Parent Birds and their Young. Austin and London: University of Texas Press. Simmons, K. E. L. 1955. The nature of the predatorreactions of waders towards humans; with special reference to the role of the aggressive-, escapeand brooding-drives. Behaviour, 8, 130-182.

619

Smith, W. J. 1968. Message-meaning analysis. In: Animal Communication (Ed. by T. A. Sebeok), pp. 44-60. Bloomington: Indiana University Press. Smythe, N. 1970. On the existence of 'pursuit invitation' signals in mammals. Am. Nat., 104, 491-494. Trivers, R. L. 1971. The evolution of reciprocal altruism. Q. Rev. Biol., 46, 35-57. Trivets, R. L. 1972. Parental investment and sexual selection. In: Sexual Selection and the Descent of Man, 1871-1971 (Ed. by B. Campbell), pp. 136-179. Chicago: Aldine-Atherton. Trivets, R. L. 1974. Parent-offspring conflict. Am. Zool., 14, 249-264. Weatherhead, P. J. 1979. Do savannah sparrows commit the Concorde fallacy? Behav. Ecol. Sociobiol., 5, 373-381.

(Received 10 April 1979; revised 6 July 1979; MS. number: 1885)