COSTS OF REPRODUCTION IN THE GREAT TIT

COSTS OF REPRODUCTION IN THE GREAT TIT: INTRASEASONAL COSTS ASSOCIATED WITH BROOD SIZE J.M. TINBERGEN Institute for Ecological Research, Boterhoeksestraat 22,6666 GA Heteren, The Netherlands CONTENTS 1. Introduction.................................. 2. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 3. Results 3.1. Breeding performance before manipulation. .. 3.2. Effects of manipulation on nestling mortality and growth . . . . . . . . . . . . . . . . . . . . .. 3.3. Effects of manipulation on the occurrence of subsequent breeding attempts 3.4. Effects of manipulation on the timing of the second brood. . . . . . . . . . . . . . . . . . . . . . . . .. 3.5. Parental condition. . . . . . . . . . . . . . . . . . . . . ... 4. Discussion................................... 4.1. Why are intervals affected by manipulation? .. 4.2. Manipulation and the probability of a second brood 4.3. Costs associated with manipulation of brood size. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 5. Acknowledgements. . . . . . . . . . . . . . . . . . . . . . . . . . .. 6. Summary.................................... 7. References . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . .. 8. Samenvatting.................................

111 112 112 112 112 114 116 116 117 118 118 119 120 120 120 121

1. INTRODUCTION

The theory of life history is concerned with trade-offs between costs and benefits of reproduction (Stearns 1976). This is studied primarily by comparing patterns of reproduction and survival between species. If however the cost of reproduction has been an important selection pressure in speciation one also would expect trade-offs within species. The question is to what extent current reproductive effort affects future reprOductive output within species, whereby current reproductive effort is defined as anything a parent has to do in order to raise offspring up to independence in the current reproductive cycle, and future reproductive output is defined as the number of offspring or recruits that parents will produce in future (Stearns 1976, Pianka 1978). This problem has been tackled by studying correlations between measurements of parental effort and future reproductive output (Kluyver 1963, 1971, Bryant 1979, Smith 1981, Tinbergen et al. 1985, Ekman & Askenmo 1986, Den Boer-Hazewinkel 1987). However, qualitative differences between individuals or habitats may confound these correlations (Van Noordwijk &

De Jong 1986). Therefore to answer this question general correlative studies cannot expected to be fruitful. Yet, in some cases the data are very convincing, as in the case where Ekman & Askenmo (1986) showed for the Willow Tit Parus atricapillus) that non-breeders survive better than breeders. To find out the costs of the various aspects of reproductive effort, experiments should be performed in which reproductive effort is manipulated and the costs in terms of future reproductive output are measured. However, not all aspects of reproductive effort are accessible to experimentation and consequently a complete answer cannot be expected. Such experiments have recently been undertaken in a number of species. In most cases brood size was manipulated and the chance of the parents returning to breed the next season was measured. Askenmo (1979) and Nur (1984) have shown an effect of manipulation of brood size in the Pied Flycatcher Ficedula hypoleuca and the Blue Tit Parus caeruleus, respectively, on the chance of the parents to return the following season. Parents were less likely to return when their brood sizes had been increased. Others (e.g. De Steven 1980) did not find an effect of manipulation of brood size on recovery rate. R0skaft (1985) found in the Rook Corvus frugilegus that brood size manipulation did not affect parental recovery but did influence the breeding success in the next season, again the birds with enlarged broods being at a disadvantage. Den Boer-Hazewinkel(1987) found in Great Tits no effect of artificial feeding of the parents on their local recovery rate. In an experiment to find out if the re-nesting ability of the parents may be affected by brood size, Slagsvold (1984) manipulated brood sizes in Great Tits Parus major shortly after hatching and removed the nestlings before fledging, measuring the timing and breeding success of the repeat attempt in the same season. His birds were affected in two ways: parents with enlarged Ardea 75 (1987): 111-122

112

[Ardea 75

INTRASEASONAL COSTS OF REPRODUCTION IN THE GREAT TIT

broods commenced repeat clutches later and were more likely to desert the repeat broods. Double-brooded birds may be attractive for this kind of experiment since an estimate of the effect of manipulation can already be found in the same season. However, measurements of subsequent survival are more difficult to interpret, since having or not having a second brood is an extra variable that has to be taken into account. This study has been undertaken to try and unravel the effects of brood-size manipulation on future reproductive output in the doublebrooded Great Tit. In this paper I report on the consequences of manipulation of the first brood for one component of future reproductive output, the occurrence of second broods. Parental and fledgling survival in relation to the manipulation will be dealt with in a future paper.

parents were both caught on the 7th day of nestling life in order to assess their body weight and tarsus length. The females were again caught and weighed on the night following the 12th nestling day. Identical measurements were taken in repeat and second clutches. In 1985 an attempt was made to weigh the females at night on the 5th nestling day. However a high percentage of these females deserted their broods, thereby forcing us to stop. The nestlings were ringed on day 7 or 8, and their tarsus length and body weight measured on day 14. The same measurements were taken during repeat and second broods. For comparison between groups the experimental-design method of least-squares analysis of variance (Overall & Klett 1972) was used, unless stated otherwise. This method takes into account the experimental-design hierarchy of main effects and interactions, adjusting each effect for all other effects at an equal or lower level and ignoring higher order effects. It is applicable to disproportionate-cell-frequency problems. Only significant interaction effects are discussed. Contingency tables have been analysed with the log-linear model method (Fienberg 1977).

2. METHODS The experiments were carried out in the Hoge Veluwe area near Arnhem, The Netherlands. The study area of ca. 160 ha consists of a mosaic of coniferous and deciduous plots on poor sandy soil. Nestbox density amounted to about 22 per 10 ha. Breeding densities of Great Tits in these boxes ranged from 5 tolD pairs per 10 ha (see Van Balen 1973, 1980). In this intensively studied population a large fraction of the adult birds are ringed both with aluminium and coloured rings. This enabled us to identify the females individually during incubation without actually handling them. Unringed birds were ringed during the nestling phase of the first brood. Brood-size experiments were performed in three successive years: 1983, 1984 and 1985. The experimental procedure was as follows: per hatching date groups of three nests were selected which had the same clutch size and approximately the same hatching success. The birds were matched for clutch size because earlier work in other populations (Kluyver 1963) had revealed correlations between clutch size and the subsequent chance of a second brood, indicating that clutch size may be correlated with quality differences of either the individual birds or their circumstances. From one of a set of three selected nests half of the chicks were removed on day two of nestling life and, after marking, added to the second nest. The third nest was kept as a contro'!. In 1983 and 1984 some controls were left out on days that fewer than three nests were available. In this way 66 broods in 1983, 64 in 1984 and 22 in 1985 were manipulated with respectively 28, 22 and 11 control broods. Details on hatching date, clutch size and the number of young hatcbpd for these nests are given in Table 1. Females producing repeat or second clutches were again manipulated, in the same way as they were in their first brood, in an attempt to enlarge or decrease the total number of young reared per season per female. To follow breeding success, nests were checked at least once a week. The

3.1. BREEDING PERFORMANCE BEFORE MANIPULATION

3. RESULTS

Significant differences in hatching date and clutch size occur between years, but not between manipulation categories (Table 1, year resp. manipulation effect on clutch size: F(2,205) = 23.3, P < 0.0001 and F(2,205) = 1.49, n.s.; year resp. manipulation effect on hatching date: F(2,205) = 56.0, P < 0.0001 and F(2,205) = 0.68, n.s.). Hatching success did not differ, neither between years nor between manipulation categories ( year effect: F(2,203) = 0.312, and manipulation effect: F(2,203) = 0.575, both n.s.). In 1985, the year that nesting was earliest, clutches were largest. It must be noted that in this year breeding density was much lower than in the other two years. 3.2. EFFECTS OF MANIPULATION ON NESTLING MORTALITY AND GROWTH

Both the 1983 and 1984 seasons were cold and wet, causing a substantial fraction of the broods to fail completely during the nestling phase. The desertion rate differed between years but was not affected by the manipulation. This was tested using a log-linear test (Fienberg, 1977) in a 2 x 3 x 2 table of year, manipulation and failure. Only an effect of year on failure could be shown(G = 20.26, df = 2, P < 0.001 ). However the non-manipulated group tended to

1987]

113

INTRASEASONAL COSTS OF REPRODUCTION OF THE GREAT TIT

Table 1. Statistics of hatching date, clutch size and number of young hatched for the experimental groups in the three successive years. Young hatched mean s.d.

Clutch size mean s.d.

n

1983

Hatching date mean s.d.

Reduced Control Enlarged

48.3 48.1 48.5

2.5 3.6 2.2

9.5 9.1 9.0

1.4 1.9 1.3

7.9 7.5 7.6

1.7 2.4 1.8

33 28 33

52.0 51.0 52.0

3.2 3.2 3.2

8.4 8.2 8.0

1.4 1.7 1.6

7.1 7.0 6.6

1.8 2.7 2.0

32 32 32

46.4 46.3 46.3

2.1 1.5 1.8

10.2 10.5 9.7

1.6 1.1 1.0

9.3 7.7 8.6

2.1 2.1 1.9

11 11 11

1984 Reduced Control Enlarged 1985 Reduced Control Enlarged

have a lower desertion rate in 1983 and 1984 (1983: reduced 45.5%, control 39.3%, enlarged 53.1 %; 1984: reduced 22.6%, control 13.6%, enlarged 18.8%). The exeptionally high rate of failure in 1983 may have been caused by the fact that there was a spell of three days of continuous rain during the period in which most nestlings were about seven days old. In 1985 part of the desertion was due to our attempts to weigh females on the 5th day of nestling life. Table 2 shows the basic breeding statistics of the nests that were not deserted. Hatching date, clutch size and the number of young are not significantly different between the three manipulation categories and therefore desertion did not seriously interfere with our experiments (year effect on clutch size, hatching date and hatching success resp. F(2,139) = 16.7, P < 0.0001,

F(2,139) 61.05, P < 0.0001 and F(2,138) 0.686, n.s.; effect of manipulation on clutch size, hatching date and hatching success resp. F(2,139) = 0.808, n.s., F(2,139) = 1.646, n.s., F(2,138) = 0.310, n.s.). In Fig. 1 a summary is given of the breeding results in the different years for the three manipulation categories. The number of fledglings, nestling weight and brood weight are plotted against the number of nestlings just after manipulation. In 1983 and 1984 nestling mortality in the enlarged broods was such that they produced about the same number of fledglings as the control broods did, while in all years the reduced broods showed hardly any mortality. In 1985 the enlarged broods produced more fledglings than the control group, indicating better conditions

Table 2. Statistics of hatching date, clutch size, number of young hatched and number of young added for the experimental groups in three successive years, computed for nests that were not deserted. 1983

Hatching date mean s.d.

Clutch size mean s.d.

Young added mean s.d.

n

Reduced Control Enlarged

49.1 48.7 49.3

1.9 3.9 1.7

9.6 9.5 8.9

1.6 2.1 1.5

8.3 7.8 7.2

1.7 2.6 2.0

-4.2 0 +3.9

17 17 15

52.6 51.5 53.1

2.9 3.0 2.5

8.2 8.3 8.1

1.5 1.7 1.6

6.8 7.2 6.7

1.8 2.5 1.9

-3.3 0 +3.4

0.9

46.5 46.4 46.3

2.1 1.4 2.0

10.0 10.3 9.8

0.8 1.1 0.8

9.0 7.8 9.1

2.0 2.4 1.3

-4.9

0.8

Young hatched mean s.d.

1.0 0.9

1984 Reduced Control Enlarged

1.0

24 20 26

1985 Reduced Control Enlarged

()

+4.2

1.0

10 9 9

114


decreased

control

increased

"tl Q)

Cl "tl Q)

;: 10

.8E :J

c::

5

0

~

19

.I:. Cl

'0; ~

Cl

.= 7ii

18 1984

Q)

c::

17

16

15

14

~200 .I:.

[Ardea 75

during that year. The number of fledglings was significantly different between years and manipulation categories (year effect F(2,112) = 7.9, P < 0.001, manipulation effect F(2,112) = 37.3, P < 0.0001). In addition there was an interaction between year and manipulation on the number of young (F(4,112) = 2.69, P < 0.04) meaning that the years did differ significantly in the effect of manipulation on the number of young in the nest on day 14. Nestling weight was strongly affected by manipulation and year (year effect F(2,112) = 8.7, P < 0.003, manipulation effect F(2,112) = 22.2, P < 0.0001). Enlarged broods produced light young. Brood weight tends to flatten off as the brood size increases. Higher nestling weights in the small broods do not fully compensate in terms of brood weight for the lower number of young. Brood weights differed significantly both between years and between manipulation categories (year effect F(2,112) = 15.0, P < 0.0001, manipulation effect F(2,112) = 29.7 P < 0.0001). In the brood weights there was a significant interaction between year and manipulation (F(4,112) = 4.09, P < 0.04), presumably mainly caused by the interaction effect between year and manipulation on the number of young as was described above. These results show that in all three years the parents did not provide enough food for their nestlings in the larger broods to keep nestling mortality low and nestling weight high .

Cl

'0;

1985

~ "tl 0

e

150

.s:J

100

50

O+-------r-------r-----

o

5 10 manipulated brood size

Fig. 1. First brood breeding results for non-deserted nests. Number of young fledged, nestling weight and brood weight are plotted against the brood size directly after manipulation, for each year and manipulation category separately. Nestling weight and brood weight are based on subsamples. Bars indicate 95% confidence intervals.

3.3. EFFECTS OF MANIPULATION ON THE OCCURRENCE OF SUBSEQUENT BREEDING ATTEMPTS

All nestboxes in the wood were checked once a week in order to spot subsequent breeding attempts. Since almost all females were known individually from the first breeding attempt a good estimate of relaying probability could be made for groups of females with different experimental histories. Of the females whose first brood failed completely 72% were known to relay in 1983 (n = 43), 88% in 1984 (n = 16) and 60% in 1985 (n = 5). Log-linear analysis of the effect of year· and manipulation on relaying probability in a 3 x 3 x 2 table revealed no significant effects of year and manipulation.

1987] ,,100..,--

-=-de""c'.=eas:..:.ed=--_ _c:..:.on-=-'co-=-'_ _'-=-nc-=-,eas:..:.e-=-d

---,

~

.c "0

1985

l:

o

8l :5

'i lfl

III

115


50

iij

E oS!

"EIII

Gi

Q.

0+--.--r----r---,r--.,---.--r---..-......-.,----r----.~___1

o

5 10 number of nestlings in first brood

Fig. 2. Percentage of second broods produced per experimental category and per year, plotted against the number of nestlings in the first brood. Both the maximum number of nestlings (directly after manipulation) and the minimum (number of fledglings) are given, resulting in a hatched area for each year.

Of the females that produced at least one fledgling in their first brood 35 % were known to start a second brood in 1983 (n = 50), 26% in 1984 (n = 69) and 78% in 1985 (n = 28). Analysis of the relations between year, manipulation and the occurrence of second broods (3 x 3 x 2 table) revealed highly significant relations between manipulation and the occurrence of second broods (G = 12.42, df = 2, P < 0.005) as well as between year and the occurrence of second broods (G = 23.76, df = 2, P < 0.005, Fig. 2). The effect of manipulation does not seem to be symmetrical in that the effect of enlarging brood size on the occurrence of second broods seems more pronounced than the effect of decreasing brood size. I will return to this in the discussion. As the tendency of a female to move away may be affected by the manipulation, the chance that a subsequent breeding attempt occurs within the study area may also be affected. Therefore the distances between the locations of the nestboxes of the first and the subsequent breeding attempts were compared for the experimental categories separately. No significant effect of manipulation could be found, neither on distance between first and subsequent breeding attempts nor between the years (year effect: F(2,48) = 0.124, n.s., manipulation effect F(2,48) = 1.38, n.s.). Movements on a scale

within the study area are thus not very likely to affect the conclusion that manipulation of brood size does affect the chance that a second brood will be produced. Clutch size and hatching success in the second brood did not differ between the manipulation categories (Fig. 3, clutch size: year effect F(2,48) = 5.93, P < 0.03, manipulation effect F(2,48) = 0.451, n.s.; hatching success: year effect F(2,48) = 1.835, n.s., manipulation effect F(2,48) = 0.02, n.s.). Fledging success from the second brood can only be given for the 1985 season since in both the 1983 and 1984 season the brood size of the second brood had been manipulated. In the 1985 season there were no differences in fledging success between the birds in relation to the manipulation of their first brood

'8

10

e

.c

1

-g o

~

III

lfl

mereased

_IJ

1984

T

,,

con ro

ecreased

I 1983

'--
-

control

increased

ca

~

iO 25 > Gi ~

20

15+------~-----,__------I

o


Fig. 4. Interval (days between hatching of the first brood and laying of the first egg of the second clutch) plotted against size of first brood (after manipulation), for years and manipulation categories separately. Bars give 95% confidence intervals.

(one way analysis of variance, manipulation effect F(2,19) = 0.258, n.s.). Concluding, we can say that manipulation affected the occurrence of second broods, but the performance of the birds while tending the second brood was not related to the way the first broods had been manipulated. 3.4. EFFECTS OF MANIPULATION ON THE TIMING OF THE SECOND BROOD

The length of the interval between the first and the second brood could be estimated by subtracting the hatching date of the first brood from the date the first egg of the second brood was layed. The hatching date was chosen as opposed to the day of the last egg (see Kluyver et a/. 1977) as not all birds start incubation immediately after the first clutch has been completed. Direct estimates of fledging date are not available for 1983 and 1984 as it involves daily checks of all the nests about to fledge. However direct estimates of intervals for 1985 reveal that the nestling period did not differ between manipulation categories (F(2.21) = 1.255, n.s.). Under the assumption that this was also the case in the other two years we can use the estimate as given above. The duration of the interval appeared to be related to both year and manipulation category

[Ardea 75

(year effect F(2,48)=5.46, P < 0.008 and manipulation effect F(2,48) = 5.80, P < 0.006, Fig. 4). As in 1983 and 1984 only two enlarged broods were followed by second broods, estimation of the rate at which intervals increase as the brood size increases are therefore preferably based on reduced and normal broods for these two years. Intervals for these groups and for the 1985 data increase with experimentally increased numbers of young in the nests by approximately 1.2 days per extra young fledged. This value is very similar to the one based on annual averages of experimentally reduced and normal nests in the Vlieland population (Kluyver et at. 1977). In 1985 the intervals were very short, resulting in about half the birds starting their second clutch when the young of the first were still in the nest. This has been reported earlier for the Great Tit and other tit species (Kluyver eta/. 1977). Concluding we can say that manipulation has an effect on the interval between the first and the second brood. Increasing the number of nestlings increases the time between the first and the second brood. Summarizing the effects of manipulation, we can state that increasing the brood size of the first brood of a Great Tit has an effect on the probability that the parents will start a second clutch. For those birds that did start a second clutch the interval between the first and the second clutch was affected by the manipulation. However, the relaying probability for birds that failed in their first brood was not affected by manipulation. 3.5. PARENTAL CONDITION

One hypothesis to explain the effects found above is that the condition of the parent birds is affected by raising a large brood and that a lowered condition may retard the production of second broods. In this study we use body weight as an estimate of condition of the parents. Females and males were captured around day 7 of nestling life. They were weighed and the tarsus length was measured. In addition the females were weighed on the night following the 12th day of nestling life. Normally females roost in the nestbox until around day 13 (Van Balen pers. comm.). The food requirements of the

1987]

117


nestlings increase during the first 8-10 days and level off after that. It is therefore expected that the work load of the parents is maximal from day 8 onwards. The moment chosen for the night weight measurement of the female was a compromise between the lower chance of recapturing her as the chicks got older and a higher chance of finding out the effects of the work load. Apart from the data on the 12th night body weight, the weight changes between the 7th day of nestling life and the 12th night are analysed. Weights of individual Great Tits are related to their body size (Garnett 1981) which can be estimated by tarsus length. It was therefore decided to analyse body weight as a function of manipulation, year and tarsus length, looking for the effects of manipulation independent of year and tarsus length effects. The weight of the female was dependent on tarsus length only (year effect F(2,98) = 0.193, n.s.; manipulation effect F(2,98) = 0.627, n.s.; tarsus effect F(1,98) = 8.991, P < 0.004). No indication of an effect of manipulation on weight is apparent (Fig. 5). The same applies to the weight of the males on the 7th day (year effect F(2,76) = 0.18, n.s.; manipulation effect F(2,76) = 0.943, n.s.; tarsus effect F(1,76) = 13.8, p < 0.0005, Fig. 5). Analysis of the weight changes per day of the females between day 7 and day 12 reveals essentially the same pattern (year effect F(2,60) = 0.88, n.s.; manipulation effect F(2,60) = 0.93, n.s.; tarsus effect F(2,60) = 0.063, n.s.). The average weight change amounted to +0.023 g/day (n = 85) and was not significantly different from zero. Concluding we can say that condition, measured as body weight on the 12th day and corrected for size, is not related to the manipulation of the brood size. In females the change of body weight between day 7 and night 12 also did not correlate with the nature of the manipulation. This means that the cause of the manipulation effect on the probability of having a second brood must not be sought in the loss of condition (in terms of weight) of the female and probably also not in that of the male. A comparison between the weights of the birds that were going to produce a second clutch and those that were not, including tarsus length

§ ~

decreased

control

increased

19

ell

'iii ~

Q)

18

(ij

E

oS!

17 16 15 14 0

10

5

§ 19

~

ell

~h~~r

.~

Q)

18

OJ E

I'

17

1984

I

16 15 14 0


Fig. 5. Body weight of females on the 12th mght and males on the 7th day of nestling life as a function of brood size after manipulation, for years and manipulation categories separately. Bars give 95 % confidence intervals.

in the analysis to correct for body size, revealed no significant differences in weight (two way analysis of variance, females: second brood effect F(1,112) = 2.98, n.s., tarsus effect F(1,112) = 6.84, P < 0.01; males: second brood effect F(1,90) = 0.127, n.s.; tarsus effect F(1,90) = 24.40, P < 0.0001). No correlations between weight during the first brood and the breeding results of the second brood could be found nor a relationship with the duration of the interval. Thus there are also no indications of a significant role of body weight during the first brood in predicting breeding success of the second brood. 4. DISCUSSION

In order to discuss intra-seasonal costs of reproduction in Great Tits I will firstly go into

118


more detail about the possible mechanisms causing the effect of manipulation on the timing and the occurrence of the second brood. Finally I will discuss the results with the emphasis on the intra-seasonal cost of reproduction. 4.1. WHY ARE INTERVALS AFFECTED BY MANIPULAnON?

It is known from correlative studies that, in a number of species, adults that raise large broods take more time before starting the next clutch (House Sparrow Passer domesticus McGillivray (1983), Song Sparrow Melospiza melodia Smith & Roff (1980), Budgerigars Melopsittacus undulatus in captivity Stamps et al. (1985), Eastern Bluebird Sialia sialis Pinkowski (1977)). Experimental studies point in the same direction (Great Tits: Kluyver et al. (1977), Slagsvold (1984), this study). Most authors argue that the condition of the parents rearing the young is lowered as the brood size increases (Drent & Daan 1980), and that this lowered condition would cause the inter-brood interval to increase. However, in this study no influence of condition on differences in inter-brood interval could be detected. In this respect it is very interesting to compare the data gathered by Slagsvold (1984) on Great Tits in Scandinavia with our own data. Slagsvold manipulated brood sizes just after hatching and removed all young on day 15 of nestling age. He found that his birds had a high relaying probability, which was not related to the manipulation of the brood size (Slagsvold pers.comm.). In our case birds that had fledged young differed greatly in their relaying probability. However there were no differences when all the nestlings died. Assuming that we can combine Slagsvolds and our own data, as if they were taken from the same population, we must conclude that the causes of the lowered chance of a second brood after the brood size was increased must be strongly associated with the presence of young in the last part of the nestling phase. Direct comparison of his data with that of 1985 reveals that for the Hoge Veluwe data the slope of the regression of interval on brood size on day 14 was significantly steeper (comparison of regressions with different variances: t' = 3.73, P < 0.01, Fig. 6). This also indicates

[Ardea 75

that the number of young present in the last period of the nestling phase and possibly after fledging has an effect on the duration of the interval. To elaborate on this I analysed for the Hoge Veluwe data the dependence of the interval of individual birds on the number of nestlings on day 14 and their average weight for the different years separately. In 1983 the interval was negatively related to the average weight of the nestlings (p < 0.05, r 2 = 0.31, n = 13). In 1984 and 1985 the interval was positively related to brood size (1984: p < 0.03, n = 16, r 2 = 0.29; 19-85: p < 0.02, n = 20, r2 = 0.27). We therefore advance the hypothesis that parents that have to care for a large number of young at one time or for lighter young need more time to lead their young to independence and that this in itself may cause longer inter-brood intervals. Ceasing to care for these young too soon may mean a much lower chance of survival of the young involved. However, the number of fledglings or their condition is not necessarily the only factor involved, as appeared from Slagsvolds results. Although these birds did not have to tend any fledglings they still showed a significant trend in the intervals between broods, although it was small in absolute terms. A possible cause of this last difference may be the parental condition during the last days of the nestling phase, after the weight measurements were taken. Alternatively a food depletion effect around the nestboxes of the enlarged broods may playa role. However data on this aspect are not yet available. 4.2. MANIPULATION AND THE PROBABILITY OF A SECOND BROOD

The next question is how this ties in with the lowered probability of a second brood for the parents with enlarged broods. It is known that generally speaking nestlings that are born later in the season have a lower chance of being recovered locally in the next breeding season (Perrins 1965, Kluyver et al. 1977). The value of an egg, in terms of recruits produced, will therefore be lower the later it is laid (see also Daan & Dijkstra 1982). In addition it is possible that the parents have to stop reproductiDn at a certain time of the year, for instance to be able

1987]

en


30..-------------------,

>-

Cll

~

~

.=.!

o

25

o o

o o

5

to perform moult before territoriality starts (Dhondt 1973, Drent 1984). Presumably the net value of an egg in terms of recruits, corrected for parental loss in reproductive value, will therefore decrease even more steeply as the laying date increases. After a particular point in time it will no longer be profitable to start a second brood. Considered from the viewpoint of the first brood fledglings, I expect that the chances of the young to be recruited into the next breeding season will increase as the time spent by the parents in tending them increases, but with a decreasing rate of increase. Formulated in terms of optimality the parent should "decide" to start a second brood at the time when the sum of the expected pay-off of the first and the second brood is maximal. Enlarging the brood may lower the expected pay-off curve against the time spent tending the young of the first brood, as more young of lower condition have to be fed simultaneously. This may thereby shift the optimal interval to a higher value. Assuming that, as a consequence of this delay, the net pay-off of a second brood decreases below the gain of tending the first brood young any further, parents should at this point "decide" not to have a second brood, and go on tending the fledglings

119

of the first brood. Davies (1978) has shown clearly that in Great Tits the time of becoming independent can be affected. Great Tits may split tasks between the sexes. After fledging the male primarily guards and feeds the young (Kluyver 1951), and I suspect that during the last part of the nestling phase and after fledging the female will be able to take a smaller share in tending the young if the brood size is small and the condition of the young is better. This will set her free to start her second brood and may explain why in particular years birds are able to start the second brood well before the first one is fledged. Summarizing I would like to stress that whether or not a particular pair will relay after manipulation may largely depend on a trade-off between tending the first brood young and starting a second brood. Division of labour between the sexes will enable the birds to start second broods earlier. The condition of the parents in this particular study does not seem to be of importance in this decision. 4.3. COSTS ASSOCIATED WITH THE MANIPULATION OF BROOD SIZE

I will now return to the original topic of this paper, the cost of reproduction, or more specifically the cost, in terms of future reproductive output within the same season, associated with brood manipulation. The chance that a second brood will be produced decreases with the size of the brood at an increasing rate (Fig. 2), especially when brood size is measured by the number of young fledged. This can be considered as a concave-up cost function of brood size in terms of future fI~Pro ductive output within the same season. The causes for this increased cost should not be sought primarily in the lowered condition of the parents during the first brood but in a tradeoff between tending the first brood and starting a second. Increasing brood size may cause the parents to spend more time tending the nestlings and, as time for breeding is limited, this will lower the probability that there will be a second brood at all. To predict optimal clutch size in double-brooded birds like the Great Tit we have to incorporate these effects. Viewed over a longer period, having a second

120


brood may incur a survival cost to the parent, as suggested by the work of Bryant (1979) on the House Martin Delichon urbica and Ekman & Askenmo (1986). This trend is not always confirmed in correlative studies of the Great Tit (Den Boer-HazewinkeI1987, but see Tinbergen et at. 1985). Analysis of the costs and benefits of reproduction, including possible survival costs, will be dealt with in a future paper. The fact that the weight of the adult tits was not affected by brood size is interesting and possibly differs between species and/or habitats. Nur (1984) found a relationship between brood size and parental weight in the Blue Tit in Wytham Wood, a rich broad-leaved wood, where the visitation rate is very high. There it is possible that the number of visits required increases steeply with the size of the brood. In the Hoge Veluwe area visitation rate of the parents is positively affected by the size of the brood, but mortality in the nest, especially in the larger broods, is much higher, indicating that there is less food available. Consequently the visitation rate of the enlarged broods does not much exceed that of the normal broods. The birds in the Hoge Veluwe may be food limited to a large extent, and may therefore spend less time in transport of food to the nest, which in turn could be the reason that they do not show the drop in weight as the size of the brood increases as was the case in the Blue Tit (see also Norberg 1981). 5. ACKNOWLEDGEMENTS A number of students took part in this project as part of /their biology study including K. Albers, J. Beens and H. Reeders. My colleagues Dr. J. H.van Balen, F. Hage, D. Westra and J. Visser helped in collecting the data. Further I would like to acknowledge the stimulating discussions with my colleagues of the Institute for Ecological Research, their statistical advice and careful reading of the manuscript. 6. SUMMARY In order to estimate within season reproductive cost (in terms of future reproductive output) associated with brood size in the Great Tit, brood size experiments were performed in three successive seasons. Broods with the same hatching date and approximately the same number of young were selected. Half of the young were taken from one brood (reduced) and added to another (enlarged) while one brood was kept as a control. The adults were individually marked. Subsequent breeding attempts of the same parents were identified within the study area (160 ha). Nestbox checks were done every week. Nestling weight was taken on day 14 of nestling life. Nest failure was not de-

[Ardea 75

pendent on manipulation but differed greatly between years. The number of fledglings and the body weight of those broods that produced at least one fledgling differed between years and between manipulation categories. The fraction of birds that relayed after nest failure was not dependent on the brood size manipulation. However, for the birds that fledged young there was a definite effect of manipulation on the production of second broods. Parents of enlarged broods were less likely to produce second clutches than were those of control broods, while parents with reduced broods were only slightly more likely to produce a second clutch. The relationship between the fraction of second broods produced and the manipulated brood size suggests a concave-up cost function of brood size within the season. Clutch size and hatching success of the second brood were not related to the manipulation of the first brood. However, there was a relationship between the manipulation and the interval between the first and the second brood. The birds that had most young fledged started their second clutch later. Parental condition, in terms of body weight corrected for size, could not be shown to be related to manipulation, and thus was unlikely to be an important cause of the above effects. Comparison of the results of Slagsvolds (1984) experiments with this study and analysis of the correlations between both nestling weight and the number of young with the inter-brood intervals, suggest that the number of nestlings and/or their body weight in the very last phase of the nestling period or after fledging is of importance in determining the inter-brood interval, and possibly the occurrence of second broods. The effect of manipulation on the occurrence of second broods is consistent with the hypothesis that parents are faced with a trade-off between tending the fledglings of the first brood and starting a second one. The intra-seasonal cost, in terms of the future reproductive output, of having a large brood is shown to be a lowered probability to produce a second clutch. No indications were found that manipulation affects the ability of the parents to produce a second clutch. 7. REFERENCES Askenmo, C. 1979. Reproductive effort and the return rate of male Pied Flycatchers. Am. Nat. 114: 748-753. Balen, J. H. van. 1973. A comparative study of the breeding ecology of the Great Tit, Parus major, in different habitats. Ardea 61: 1-92. Balen, J. H. van. 1980. Population fluctuations in the Great Tit and feeding conditions in winter. Ardea 68: 143-164. Boer-Hazewinkel, J. den. 1987. On the costs of reproduction: Parental survival and the production of second clutches in the Great Tit. Ardea 75. Bryant, D. M. 1979. Reproductive cost in the House Martin (Delichon urbica). J. Anim. Ecol. 48: 655--'675. Daan, S. & C. Dijkstra. 1982. Annual cycles of breeding in the Kestrel: individual tuning to vole availability. Ibis 124: 389-390. Davies, N. M. 1978. Parental meannes and offspring independence: an experiment with handreared Great Tits Parus major. Ibis 120: 509-514. Dhondt, A. A, 1973. Postjuvenile and postnuptial moult in

1987]


a Belgian population of Great Tits, Parus major, with some data on captive birds. Gerfaut 63: 187-209. Drent, P. J. 1984. Mortality and dispersal in summer and its consequences for the density of Great Tits (Parus major) at the onset of autumn. Ardea 72: 127-162. Drent, R. H. & S. Daan. 19SiO. The prudent parent: energetic adjustments in avian breeding. Ardea 68: 225-252. Ekman, J. & C. Askenmo. 1986. Reproductive cost, agespecific survival and a comparison of the reproductive strategy in two European tits (genus Parus). Evolution 40: 159-169. Fienberg, S. E. 1977. The analysis of cross-classified categorical data. M.LT. Press, Cambridge. Hpgstedt, G. 1981. Should there be a positive or negative correlation between survival of adults in a bird population and their clutch size? Am. Nat. 118: 568-571. Kluyver, H. N. 1963. The determination of reproductive rates in Paridae. Proc. XI11th Int. Omit. Congr.: 706-716. Kluyver, H. N., J.H. van Balen & A. J. Cave. 1977. The occurrence of time-saving mechanisms in the breeding biology of the Great Tit, Parus major. In: B. Stonehouse & C. Perrins (eds.). Evolutionary ecology. Kluyver, H. N. 1971. Regulation of numbers in populations of Great Tits (Parus m. major L.). Proc. Adv. Study Inst. Dynamics Numbers Popu!. (Oosterbeek 1970). McGillivray, W. B. 1983. Interseasonal reproductive costs for the House Sparrow (Passer domestieus). Auk 100: 25-32. Noordwijk, A. J. van & G. de Jong. 1986. Acquisition and allocation of resources: Their influence on variation in life history tactics. Am. Nat. 128: 137-142. Norberg, R. A. 1981. Optimal flight speed in birds when feeding young. J. Anim. Eco!. 50: 473-477. Nur, N. 1984. The consequences of brood size for breeding Blue Tits. I: Adult survival, weight change and cost of reproduction. J. Anim. Eco!. 53: 479-496. Overall, J. E. & C. J. Klett. 1972. Applied multivariate analysis. McGraw-Hill, New York. Perrins, C. M. 1965. Population fluctuations and clutch size in the Great Tit, Parus major L. J. Anim. Eco!. 34: 601-647. Pinkowski, B. C. 1977. Breeding adaptations in the Eastern Bluebird. Condor 79: 289-302. Rpskaft, E. 1985. The effect of enlarged brood size on the future reproductive potential of the Rook. J. Anim. Eco!. 54: 255-260. Slagsvold, T. 1984. Clutch size variation of birds in relation to nest predation: on the cost of reproduction. J. Anim. Eco!. 53: 945-953. Stamps, J. , A. Clark, P. Arrowood & B. Kus. 1985. Parentoffspring conflict in Budgerigars. Behaviour 94: 1-40. Steven, D. de. 1980. Clutch size, breeding success, and parental survival in the Tree Swallow (Iridoproene bic%r). Evolution 34: 278-291. Smith, J. N. M. 1981. Does high fecundity reduce survival in Song Sparrows? Evolution 35: 1142-1148. Smith, J. N. M. , & D. A. Roff. 1980. Temporal spacing of

121

broods, brood size and parental care in Song Sparrows (Melospiza me/odia). Can. J. Zool. 58: 1007-1015. Stearns, S. C. 1976. Life-history tactics: a review of the ideas. Quart. Rev. Bio!. 51: 3-47. Tinbergen, J. M. , J. H. van Balen and H. M. van Eck. 1985. Density dependent survival in an isolated Great Tit population: Kluyvers data reanalysed. Ardea 73: 38-48. 8. SAMENVATTING De broedselgrootte van Koolmezen werd gedurende drie seizoenen gemanipuleerd om na te gaan of er kosten, in termen van toekomstig reproductief succes, aan broedselgrootte verbonden zijn. Er werden broedsels met dezelfde uitkomstdatum en ongeveer hetzelfde aantal eieren uitgekozen. De helft van de jongen werd op hun tweede levensdag uit een broedsel gehaald (verkleind) en aan een ander toegevoegd (vergroot), terwijl een derde als controle werd aangehouden. De ouders waren individueel gemerkt. Latere broedpogingen, zoals vervolglegsels of tweede broedsels, werden binnen het studiegebied (160 ha) vastgesteld. In dit artikel worden de effecten van manipulatie op het v'crdere verloop van het broedproces binnen hetzelfde jaar besproken. Effecten op langere termijn zullen in een later stuk naar voren worden gebracht. De nestkasten werden elke week gecontroleerd en de jongen werden op hun 14e levensdag gewogen. Er bleken aanzienlijke verschillen tussen de jaren te zijn in het percentage van de nesten dat geen vliegvlugge jongen opleverde, maar er kon geen effect van de manipulatie op deze kans om te mislukken aangetoond worden. Voor die broedsels, die ten minste een vliegvlug jong produceerden, bestond er een sterke samenhang tussen manipulatie en het aantal uitgevlogen jongen en hun gemiddelde gewicht. Ook jaarverschillen in deze grootheden konden worden aangetoond. De fractie van de paren, die na mislukken een vervolglegsel maakten, hing niet samen met de voorafgaande manipulatie. Bij de paren die ten minste een vliegvlug jong JProduceerden, bestond er een duidelijk effect van manipulatie op de kans dat zij een tweede broedsel maakten. Ouders van vergrote broedsels produceerden minder tweede broedsels dan de controle paren, terwijl de verkleinde broedsels slechts een iets hogere kans op een tweede broedsel lieten zien. Het verband tussen de fractie geproduceerde tweede broedsels en de gemanipuleerde broedselgrootte suggereert dat vergroten sterker remmend op de productie van tweede broedsels werkt dan verkleinen stimulerend werkt. Er is dlis binnen een seizoen een meer dan evenredig toenemende kostenfactor aan broedselgrootte gekoppeld. Legselgrootte en uitkomstpercentage van het tweede broedsel hingen niet samen met de manipulatie van het eerste. Wei was er een relatie tussen manipulatie van het eerste broedsel en het interval tussen het eerste en het tweede. Hoe meer jongen bij een paar uitvlogen, des te langer was het interva!. De conditie van de ouders, uitgedrukt in gewicht gecorrigeerd voor de tarsuslengte, had geen aantoonbaar verband met de manipulatie en kan dus niet als oorzaak van de boven beschreven effecten van manipulatie gelden. Vergelijking van de hier beschreven resultaten met die van Slagsvold (1984) suggereert, dat het aantal jongen en/of hun gewicht in de laatste dagen voor het uitvliegen of net

122


daarna van belang is bij het bepalen van het interval tussen het eerste en het tweede broedsel. Omdat er een beperking is aan de tijd dat er gebroed kan worden, heeft deze verlenging mogelijk zelfs effecten op het al of niet beginnen van een tweede broedsel. De vastgestelde effecten van manipulatie op het voorkomen van tweede broedsels passen in de hypothese dat de ouders in de laatste fase van de jongenverzorging een afwe-

[Ardea 75

ging maken tussen het verder verzorgen van de jongen van het eerste broedsel en het beginnen van een tweede. Er zijn geen aanwijzingen gevonden, dat de ouders niet in staat zouden zijn om een tweede broedsel te maken. De kostenfactor van het hebben van een vergroot broedsel, in termen van verlies aan toekomstige reproductie binnen hetzelfde broedseizoen, bestaat uit een verlaagde productie van tweede broedsels.