Evolution, 34(2), 1980, pp. 278-291
CLUTCH SIZE, BREEDING SUCCESS, AND PARENTAL SURVIVAL IN THE TREE SWALLOW (lRIDOPROCNE BICOLOR)
DIANE DE STEVEN Museum of Zoology and Department of Ecology and Evolutionary Biology, University of Michigan, Ann Arbor, Michigan 48109
Received November 13, 1978. Revised October 5, 1979
Lack (1947) hypothesized that clutch populations is largely untested. Observa size in nidicolous birds has evolved by tions from naturally occurring brood sizes natural selection to correspond with the have not yielded any consistent relation maximum number of young that, on av- ship between clutch size and parental sur erage, the parents can feed. Although the vival (e.g., Kluyver, 1963; Perrins, 1965; hypothesis gained wide acceptance in sub- Lack, 1966, p. 109). However, since vari sequent years, the evidence is equivocal ation in parental ability (e.g., efficiency in and inconsistencies remain (Klomp, 1970; gathering food for egg formation or for Cody, 1971; von Haartman, 1971; Hus- feeding nestlings) may contribute to adap sell, 1972). Those cases in which the most tive modification of clutch size (d. Klomp, productive brood size is larger than the 1970), the search for such a relationship most common do not support the implied is confounded. If parents that normally concept of direct limitation of clutch size initiate larger clutches are more capable by food supply. Furthermore, the inter- of rearing them, the young in those broods pretation of brood manipulation experi- are not necessarily disadvantaged (d. Per ments that support Lack's hypothesis is rins and Moss, 1975), nor are those par open to question, since the results do not ents necessarily less likely to survive than distinguish between food supply limits in parents raising smaller broods. Greater the environment and possible adaptive weight losses among parents with larger limits upon parental feeding behavior broods (Hussell, 1972; Winkel and Win (Cody, 1971; Hussell, 1972). Mountford kel, 1976; Bryant, 1979) and lower prob (1968) suggested that incorrect formula- abilities of initiating a second brood after tion of predictions was responsible for a large first brood (Kluyver, 1963; Pin some apparent contradictions. kowski, 1977) provide indirect evidence Lack recognized that selection should that rearing large broods is stressful phys favor the clutch size that maximizes fit- iologically, but have yet to be linked to ness, but overemphasized the direct influ- differential survival. Recently, Bryant ence of environmental factors upon clutch (1979) found differences in survival be size as a single trait. A more comprehen- tween single- and double-brooded female sive perspective incorporates interactions House Martins (Delichon urbica), but between clutch size and other life history these differences were not related to the features as well (Fisher, 1958; see review brood sizes reared. in Stearns, 1976). Pertinent models predict Experimental approaches to testing the a most common clutch that is smaller than hypothesis that birds adaptively limit the most productive, under the assump- clutch size for the sake of enhanced sur tion that rearing larger broods places vival are likely to provide the strongest greater stress upon parents and reduces inferences (cf. Ricklefs, 1973, p. 426; their chances of surviving to breed again Stearns, 1976, p. 42). Artificial manipu (Williams, 1966; Charnov and Krebs, lation can extend brood sizes beyond lim 1974). its currently observed within a popula- The assumption of a trade-off between tion. However, if parental ability is clutch size and adult survivorship within reflected in individual clutch sizes, parents 278 AVIAN CLUTCH SIZE AND PARENTAL SURVIVAL 279 must receive clutches enlarged in relation guish sources of variation within and to the sizes they originally attempted and among broods in addition to evaluating must be compared to parents whose age and treatment effects. clutches are not altered in size. This ap proach tests more directly for a relation METHODS ship between increased parental effort and Bird.-The Tree Swallow is an ame survival than does the random assignment nable subject for manipulative studies be of brood sizes to parents that has charac cause it readily uses nest boxes, breeds re terized previous studies (d. Perrins and peatedly in the same colonies, and Moss, 1975) and that tests for average tolerates considerable disturbance at the ability to rear young under a particular set nest without deserting. A detailed natural of environmental conditions. history is given in Kuerzi (1941); here it is I report an attempt to test the hypoth appropriate to point out several features. esis of an interaction between clutch size Normally only one brood is reared per and parental survival in the Tree Swal year. The breeding system is monoga low, Iridoprocne bicolor. Among known mous, although several times I have sources of phenotypic variation in clutch trapped individual males feeding nestlings size in this species (d. Kuerzi, 1941), year at two different nests (usually in adjacent ling females lay smaller clutches than do nest boxes). Thus there may be occasional females two years and older (De Steven, polygyny, but all of the males included in 1978). Age-specific differences in clutch this study were monogamous. Males assist size have generally been regarded as an females in nest building, nest defense, and adaptive modification that adjusts for the feeding nestlings, but not in incubating inexperience of new breeders (cf. Klomp, eggs. Because many aspects of male biol 1970). Interpretation in the context of life ogy are not well known (e.g., age and age history theory suggests the hypothesis that related fecundity and success) and because by raising smaller broods and exerting less males are not trapped as easily as females, effort, young individuals are holding their the experiments in this study were de own probability of survival high (Perrins signed with respect to and results evalu and Moss, 1974; see also Pianka and Par ated for females only. ker, 1975). I tested this subsidiary hypoth Study site.-I conducted this study in esis as well by enlarging broods of both 1976 on Long Point, Ontario, a large female age groups and comparing them to sandspit extending 29 km eastward from each other as well as to controls of both the northern shore of Lake Erie. The ex age classes. treme eastern end of the Point is charac I evaluate the females' breeding perfor terized by early dune successional vege mance and its subsequent effects upon tation (d. Cowles, 1901) amid both their condition and survival. The hypoth permanent and temporary ponds, a habi esis of environmental limitation of clutch tat that provides a rich food source of size (sensu Lack) would predict slower emerging aquatic insects for aerial-forag nestling growth, lower fledging success, ing birds and that appears similar to nat and/or poorer survival of young from en ural Tree Swallow nesting areas (Forbush larged broods than from normal-sized and May, 1939; Palmer, 1949). A nest-box broods. The cost-of-reproduction hypoth colony for Tree Swallows has been main esis would predict greater weight losses tained in this area since 1969 by the Long (d. Ricklefs, 1974, p. 261) and/or lower Point Bird Observatory (LPBO), which survival for females raising enlarged collects nest-record data and supervises broods as a result of increased stress. In annual banding of adults and young. In ability to rear extra young or inability to 1976, 91 of the 134 available nest boxes tolerate reproductive stress might be more were occupied by nesting Tree Swallow pronounced in yearling females than in pairs. older females. Methods of analysis distin- Brood-size experiments.-Tree Swal- 280 DIANE DE STEVEN
TABLE 1. Design for Tree Swallow brood-size ex in a recipient brood were transferred to periments. that brood.
Brood sizes Rather than repeatedly disturbing the Brood category represented broods in order to construct curves for Control: Yearling S' sa, 6 nestling growth, I compared weights and Control: Older S' e-, 7 primary feather lengths attained at specif Experimental: Yearling S' 7, 8 ic ages as a means for assessing rapidity Experimental: Older S' 8, 9 of growth and the size and condition of a Most common clutch for the age group (De Steven, 1978). nestlings. Since hatching is sometimes asynchronous in clutches of six or seven, growth data for nestlings in such broods low nests were visited daily during the were taken on different days in order to laying period to determine precise laying make the ages of measurement equivalent. dates and clutch sizes for each pair. The Weights were taken to the nearest 0.1 g age of the female (yearling or older) was with a Pesola spring balance at 3, 13, and also ascertained by observation of plum 16 days of age, and the length of the outer age during this time (d. Kuerzi, 1941, on primary feather from sheath base (where plumage differences between yearling and it emerges from the skin) to feather tip was older females; see also De Steven, 1978). measured to the nearest mm at 13 and 16 Daily visits were resumed close to the days also. There is no external feather de hatching time of each brood to determine velopment at three days of age. Three hatch dates and ages for individual nest days was chosen partly for convenience, lings. In each brood used in the study, all as it was necessary to renew the temporary nestlings were individually marked on the claw marks, and partly for an early as day of hatching by clipping the tip of one sessment of growth. Thirteen days is the claw in a prearranged sequence. Nestlings approximate age of a peak in weight for added to experimental broods (see below) nestling Tree Swallows (Paynter, 1954), were marked so as to be distinguishable after which weight declines to adult levels from the original nestlings of those broods. as a probable result of water losses accom The clips required one renewal when the panying feather maturation (Ricklefs, young were about three days old and sub 1968). Measurements at 16 days estimate sequently remained visible until the young the size and condition of the young near were banded with U.S. Fish and Wildlife fledging. It is the latest age at which mea Service bands at 12 days of age. surements can be obtained without risk of A control and an experimental category premature fledging, since the young are were designated for broods of both year capable of limited flight prior to the nor ling and older female parents. In control mal fledging age of 18 to 20 days (Kuerzi, broods, nestlings were added only to re 1941). Brief visits to the nests after 16 days place hatching losses (dead eggs) so that confirmed final fledging dates for the brood size equalled original clutch size; broods. Those nests that had provided the these broods represent the clutch sizes additional nestlings for the experiments commonly observed for each parental age were followed only for purposes of band group (d. De Steven, 1978). In experi ing and confirmation of fledging, and no mental broods, nestlings were added both growth data were taken. to replace hatching losses and to increase Broods chosen for the study hatched the original clutch sizes of each pair by during the main breeding period of the two young. The result was an overall colony and differed in hatching dates by range of brood sizes from five to nine (Ta only slightly more than a week (Fig. 1), ble 1). All transfers were performed within thus minimizing the effects of seasonal one day (occasionally two days) after variations in nesting success. Thirteen of hatching, and only nestlings that had the 17 completed clutches of yearling fe hatched on the same day(s) as the young males were used in the study, the rest AVIAN CLUTCH SIZE AND PARENTAL SURVIVAL 281 being late nests. Of the 72 completed clutches of older females, 34 were initially 15 used for the study, five hatched either too ALL NUTS early or too late, and the remaining 33 o N = 89 provided the additional nestlings for STUDY NUTS CIl transfer. Normally two or three young I- N= 41 were left in the donor nests to prevent de ~ 10 • sertion by the parents. z u.. Metal shields were placed around the o nest-box poles to exclude predators; how ci ever, six broods were still partially or z 5 wholly preyed upon by snakes after the study had commenced. These broods were not included in the analyses. Parents.-Since the females roost on the nest at night for about one week after 6 16 26 the young hatch, each could be caught and June weighed easily once on the first night of DATE OF HATCHING her brood's hatching and again five nights later. This permitted standardization of FIG. 1. Hatching dates of Tree Swallow broods selected for study in relation to the hatching period weighing times from the first to the sec of the Long Point colony in 1976. ond, since body weight can vary widely with time of day (Baldwin and Kendeigh, 1938). Females were weighed to the near ysis System (MIDAS). Significant results est 0.5 g with a Pesola spring balance. In were those giving a probability of Type I a few cases, females had ceased brooding error of 5% or less. by the sixth night and a second weight Analysis of nestling growth presents was not obtained. Most male parents were special problems. A set of growth mea captured once for identification and band surements normally contains several ing; however, no male was caught at four sources of variation: 1) that from nestlings of the study nests, perhaps indicating male within a single brood, 2) that from broods desertion. of a given size reared by different parents, Survivorship.-To determine returns of and 3) that from different brood sizes. adults and especially young from the 1976 Surprisingly, few studies of nestling study, in 1977 and 1978 nearly all breed growth have taken all these sources of ing adults were captured while feeding variation into account. Either the first is nestlings. Two consecutive years of trap ignored and the measurements are aver ping effort are important for recovery of aged for each brood, or the second is ig fledglings of a given year, since some may nored and all measurements from a given return to the colony in their second year brood size are combined. Either approach but fail to breed there until their third may seriously bias any subsequent tests of year. In contrast, fewer adults skip breed the effects of brood size on nestling ing in the colony between years. Among growth. An analysis that evaluates all the group under study in 1976, no females sources of variation is both more correct were recovered in 1978 that had not re and more informative. turned in 1977. The experiments in this study introduce Return rates are minimum estimates of two additional sources of variation: paren survival, since emigration and death are tal age class and "treatment" effect (con indistinguishable. trol vs. experimental). Since these are not Statistics.-Data analysis was per independent of brood-size variation, the formed with use of statistical programs analysis is simplified; nevertheless the re from the Michigan Interactive Data Anal- quirements of a model allowing simulta- 282 DIANE DE STEVEN
TABLE 2. Breeding success (±SD) of Tree Swallow study broods.
Mean no. of Mean fledging Category Brood size No. of nests fledged young rate (%)8
Control: 5 2 5.0 ± 0.0 100.0 ± 0.0 Yearling <;> 6 3 5.7 ± 0.6 94.4 ± 9.6 Comb. 5 96.7 ± 3.3 Control: 6 10 5.9 ± 0.3 98.3 ± 5.3 Older <;> 7 5 6.6 ± 0.5 94.3 ± 7.8 Comb. 15 97.0 ± 1.6 Experimental: 7 3 6.3 ± 0.6 90.5 ± 8.2 Yearling <;> 8 4 7.8 ± 0.5 96.9 ± 6.2 Comb. 7 94.1 ± 2.8 Experimental: 8 8 7.5 ± 0.8 93.8 ± 9.4 Older <;> 9 6 8.3 ± 0.8 92.6 ± 9.1 Comb. 14 93.2 ± 2.4
a G-test (Sokal and Rohlf, 1969, p. 601) on combined categories (Comb.) = 4.14, df ~ 4, P > .10 (n.s.).
neous evaluation of both factors were not the differences are not significant (Table met by the data. I performed separate 2). The increased incidence of mortality analyses in parallel for the two age groups should be the result of starvation if food of females and used brood size as the main supply or parental feeding response is in effect; the resulting model is a one-way adequate for rearing larger than normal nested (hierarchical) analysis of variance broods. I analyzed nestling mortality in (d. Sokal and Rohlf, 1969, p. 253). Using more detail and considered starvation a brood size in this manner incorporates the probable cause of death if a nestling died "treatment" of brood enlargement. The at an early age C~3 days old) or was ob model represents the best approximation viously underweight relative to its siblings given that data collected under field con at the last weighing prior to death. One ditions often fail to meet all the criteria of control nest and four experimental nests statistical models. Graphical presentations showed evidence of starvation, but these summarize the data and permit visual in account for only five of the 13 nests in spection. which deaths occurred. Nestlings that died were not necessarily transferred young (4 of 15 deaths) nor late-hatched RESULTS young (6 of 15). In many instances there Breeding Performance was no obvious reason for death. Overall nesting success of female Tree The consistently low total mortality as Swallows in each of the brood categories reflected in mean fledging rates greater is summarized in Table 2. Productivity than 90% shows that food stress, if it ex figures are subdivided according to brood isted, was not sufficient to prevent the size and show clearly that the largest fledging of most young regardless of brood broods fledge the most young. Even category. though yearling females normally fledge fewer young than older females because Nestling Growth of their smaller average clutch size (De As Perrins (1965) has shown for the Steven, 1978), they can successfully rear Great Tit (parus major), the number of additional young. fledged young is an incomplete measure of The slightly lower average fledging reproductive success, since the condition rates from experimental broods relative to of the young at fledging may influence control broods suggest some negative ef subsequent survival. Nestlings in enlarged fect of brood enlargement, even though Tree Swallow broods may suffer retarded AVIAN CLUTCH SIZE AND PARENTAL SURVIVAL 283 growth if food supply or parental feeding YEARLING ~ OLDER ~ ability is limiting (but not severely) and 13-0AY 13-0AY 45 may have poorer chances of survival after leaving the nest. Primary feather growth was thought to 'i 35 be a meaningful index of overall nestling :E f t 17 growth rate, since swallows do not fledge i= 25 ~ until adequate flight capability is attained. , 54 34 w f 3. t Primary measurements for 13- and 16-day ~ old nestlings are presented in summary ~ 15 lB 50 :E form in Figure 2. Most obvious is the large ~ 59 16-0AY 16-0AY variability in growth. Overall, "experi- ffi 57 mental" nestlings are significantly more ~ ~ variable in primary length than "control" -ill- nestlings (F test for equality of variances, ~ 47 10 " P < .001 for all four combinations of E 30 nestling age and parental age class). Min- Z 37 t54 f imum measurements also ranged much * • lower in experimental broods. When the 19 27 61 t43 data are partitioned for the nested ANOVA, a less obvious result emerges. There is no 5 6 a 9 5 6 a 9 significant influence of brood size upon aROOO SIZE nestling growth; most of the variation is FIG. 2. Outer primary feather lengths of Tree due to differences in average growth Swallow nestlings at 13 and 16 days of age in relation among individual broods of a given size to brood size. Data are presented separately for (Table 3). Thus experimental broods did broods reared by yearling females and by older fe not show poorer average growth but rath males. Within each age category, the two smaller brood sizes are control broods and the two larger er greater variability in growth. Member brood sizes are the experimental broods created by ship in a particular brood was more im addition of two young each to control brood sizes (cf. portant than was the size of the brood. Methods). Vertical line represents the range, hori Greater variability in growth implies zontal line the mean, and rectangle one standard that young in at least some experimental deviation on either side of the mean. Sample size (number of nestlings) is given below each symbol. broods were growing more slowly. If fledging age remains constant, a small pri- mary feather length at 16 days would in dicate a disadvantage in size at fledging. The relationship still holds (r = -0.81, However, duration of the nestling period N = 15, P < .01). Thus, those young ex for each brood (computed as the time from hibiting slower primary feather growth hatching to final fledging of all young) is remained longer in the nest while, pre inversely related to mean primary length sumably, making up their growth deficit. for each brood at 16 days(r = -0.71,N = Even if Tree Swallow young in enlarged 38, P < .01). The nestling period is prob broods may attain adequate size (as re ably overestimated slightly where the flected in primary feather growth) at members of a brood do not fledge simul fledging, they may be lower in weight and taneously, but it was not feasible to record in poorer condition. Nestling weight vari time in the nest for individual young. ation may reflect variation in fat stores (cf. Since the occurrence of unusually windy O'Connor, 1976), which may in turn in and cool weather from July 1 to 3 may fluence survival after leaving the nest have delayed fledging of some broods that (e.g., Perrins, 1965; Howe, 1976). Weights would normally have left the nest during of Tree Swallow nestlings at three, 13, that time, I repeated the analysis for only and 16 days of age are presented in sum those broods due to fledge after July 4. mary form in Figure 3. Among broods 284 DIANE DE STEVEN
TABLE 3. Analyses of variance for nestling outer primary lengths in Tree Swallow study broods.
Yearling females Older females
Source 55 dj. M5 F 55 dj. M5 F Thirteen-day primary length Brood size 85.1 3 28.4 0.2 263.2 3 87.7 0.7 Broods within brood size 1,156.6 8 144.6 20.3** 2,930.7 24 122.1 17.0** Nestlings within broods 461.9 65 7.1 1,210.1 169 7.2 Total nestlings 1,703.6 76 4,404.0 196 Sixteen-day primary length Brood size 190.5 3 63.5 0.5 397.2 3 132.4 1.1 Broods within brood size 835.2 7 119.3 20.9** 2,848.2 23 123.8 20.6** Nestlings within broods 341.1 60 5.7 985.3 165 6.0 Total nestlings 1,366.8 70 4,230.7 191
** p ~ .01. reared by yearling females, a tendency for variation in weights was not consistently lower nestling weights in larger broods larger in the experimental categories as appears as early as three days of age, al compared to the control categories (F test, though at this stage the differences among P > .10 at 3 and 16 days) and at 13 days brood sizes are not significant (Table 4). the reverse was true (F = 1.45; d.f. = 87, However, by 13 days, young in broods of 109; P < .05). five attain significantly higher peak After 16 days, nestling weights may weights than young in all larger broods, continue to decline if peak weights were and this difference is maintained as the attained slowly, or they may stabilize at young approach fledging age (Table 4 and adult levels (cf. Methods). While an ex Fig. 3). The nested analysis of variance tension of the nestling period may allow (Table 4) shows that the high interbrood feather maturation to be completed, the variation that characterized primary differences in weight are unlikely to feather growth does not occur in the change appreciably and may even become weights attained in these broods. Tenden accentuated as weight recession is com cies for slower growth in experimental pleted in more slowly growing broods. broods are also reflected in the failure of Thus, observed differences in weight at 16 11 nestlings from three different enlarged days estimate differences in the condition broods to reach peak weights at 13 days. of the young at fledging several days later. In contrast to broods reared by yearling Brood enlargement influenced fledging females, nestling weights in broods reared weights negatively among broods reared by older females show large variability but by yearling females, but not among broods no clear trends with respect to brood size reared by older females. (Fig. 3). The nested analysis of variance shows most of the variation to be due to Fledgling Survivorship differences among individual broods (Ta Reproductive success should also be ble 4). Only at 16 days of age is there a evaluated in terms of the survival of suggestion of a decline in weight with in young to breeding age, but this is often creasing brood size, but this may be due the factor for which the least information to a number of nestlings in broods of six is available. While 30-40% of adult Tree that continued to increase in weight from Swallows may return annually to the Long 13 to 16 days. Emphasizing the large vari Point colony, return rates for juveniles ation in weights in these broods, 12 young fledged from the colony are in the order from 4 control broods and 9 young from of 5-10% (D. Hussell, pers. comm.; De 4 experimental broods had failed to reach Steven, 1978). The low number of return peak weights at 13 days of age. In general, ing juveniles probably reflects dispersal to AVIAN CLUTCH SIZE AND PARENTAL SURVIVAL 285 other breeding sites as well as high post YEARLING 2 OLDER 2 fledging mortality (Chapman, 1935). As 3-DAY 3-DAY suming that every surviving Tree Swallow 10 juvenile is equally likely to return to the breeding colony, return rates will estimate survivorship. Since all fledglings from the ~ 10 t,. 1976 experiments hatched within a limited 4 f t 32 time period (Fig. 1), seasonal variations in survival probabilities should be minimal tit f52 (cf. De Steven, 1978). 1 Of 417 young fledged from the colony, 13-DAY 13-DAY 27 (6.5%) returned in 1977 and/or 1978. 30 :<;: Two of the 27 were not recovered in the -t- study colony but instead in another colony ~ 25 under LPBO supervision approximately ~ 29 km distant. Fourteen returns were ~ ~ from broods used in the study. From these 2 20 ! t en t 18 recoveries it appears that enlarged broods ~ 59 t54 ttt recruited young as well as normal-sized IS broods; half of the 14 had come from 16-DAY 16-DAY broods of eight (Table 5). Among the 30 study broods, 6.5% of the young reared by yearling females were recovered in con 25 trast to 4.5% of the young of older females (X2 = 0.14, dj. = 1, P > .10). From the 23 broods that were eventually reduced in 20 !, f, ~ size to provide extra nestlings for the ex 12 19 t tt43 periments, 13.5% of the young fledged re 15 turned to the colony (N ~ 52). Most of 6 9 5 6 8 9 these broods contained two or three BROOD SIZE young, and recruitment from these may FIG. 3. Weights of Tree Swallow nestlings at 3, be particularly high. 13, and 16 days of age in relation to brood size. Data Mean 16-day primary lengths for the are presented separately for broods reared by year broods that recruited young averaged no ling females and by older females. Interpretation of symbols as in Fig. 2. higher than those for broods from which no young returned (t = 0.84, d.f. = 36, P > .10), nor did mean 16-day weights differ between the two groups (t = 0.78, a treatment category, for yearlings to lose dj. = 36, P > .10). Thus young that re more weight than older females. Analysis turned did not come from broods that of weight loss as a percentage of initial grew most slowly, but neither did they weight produced the same results, indi necessarily come from broods that grew cating that initial weights had been similar fastest or were heaviest near fledging. in every category. Percent loss was thus proportional to net loss. Among nine older Female Weight Loss and Survivorship females whose broods had been reduced Weight losses of female parents during in size to provide extra nestlings for the the first part of the nestling period are pre experiments, weight losses averaged 1.1 sented in Table 6. Although the differ g (SE = 0.2) and did not differ signifi ences were not significant, there was a cantly from the losses of older females in tendency for females raising enlarged the two treatment categories (F = 1.52; broods to lose more weight than females dj. = 2,31; P > .10). raising normal-sized broods, and, within If these losses do reflect physiological 286 DIANE DE STEVEN
TABLE 4. Analyses of variance for nestling weights in Tree Swallow study broods.
Yearling females Older females
Source SS d.f. MS F SS dj. MS F Three-day weight Brood size 5.3 3 1.8 2.2 17.5 3 5.8 1.1 Broods within brood size 6.9 8 0.8 1.3 126.5 24 5.3 6.6** Nestlings within broods 44.7 69 0.6 131.6 175 0.8 Total nestlings 56.9 80 275.6 202 Thirteen-day weight Brood size 75.6 3 25.2 5.0* 1.6 3 0.5 0.05 Broods within brood size 39.6 8 5.0 1.8 260.3 24 10.8 6.3** Nestlings within broods 177.0 65 2.7 299.0 170 1.7 Total nestlings 292.2 76 560.9 197 Sixteen-day weight Brood size 36.5 3 12.2 7.2* 50.0 3 16.7 1.8 Broods within brood size 12.2 7 1.7 0.7 213.8 23 9.3 5.8** Nestlings within broods 140.8 61 2.3 259.2 164 1.6 Total nestlings 189.5 71 523.0 190
* .01 < P ::s:;: .05; ** P ~ .01. stress, the trends are at least consistent enlarged broods influenced subsequent with the hypothesis that brood enlarge survival. Returns are presented for all fe ment increases stress upon parents and males of the 1976 experiments as well as that yearlings are less tolerant of this in for that subset for which weights were crease than older females. However, even successfully obtained (Table 7). Overall, the maximum difference in average loss 57.9% of the control females returned in between any two groups is less than 1 g, contrast to 66.7% of the experimental fe although the largest average loss (1. 9 g) is males (X2 = 0.34, d.f. = 1, P > .10). near 10% of body weight. Yearlings had somewhat higher return The final question is whether rearing rates (72.7%) than older females (58.6%) in general tv'' = 0.66, d.f. = 1, P > .10), but these values may be inflated due to TABLE 5. Returns in 1977 or 1978 of Tree Swallows the smaller sample of yearlings. Among 23 fledged from 1976 study broods. females whose broods were reduced in size, 78.3% returned, and of the nine from No. No. re No. No. young turned that group that had been weighed, 56% Brood of young returned per Category size broods fledged (%)' brood returned. Control: 5 2 10 1 0.50 Yearling '( 6 3 17 1 0.33 Comb. 5 27 2 (7.4) TABLE 6. Weight losses of female parents of Tree Control: 6 10 59 2 0.20 Swallow study broods over a six-day period. Older '( 7 5 33 1 0.20
Comb. 15 92 3 (3.3) Mean weight loss (g) No. of Experimental: 7 3 19 1 0.33 Category ± SE females Yearling '( 8 4 31 2 0.50 Control: Yearling '( ± Comb. 7 50 3 (6.0) 1.6 0.6 4 Control: Older '( 1.3 ± 0.2 11 Experimental: 8 8 60 5 0.62 Older '( 9 6 49 1 0.17 Experimental: Yearling '( 1.9 ± 0.3 5 Experimental: Older '( 1.6 ± 0.2 14 Comb. 14 109 6 (5.5) ANOVA: F (yearling vs. older) = 1.4; d.f. = 1,30; P > .10 (n.s.). a G-test on combined categories (Comb.) = 3.78, df. = 4, P > .10 F (control vs. experimental) = 1.1; df. = I, 30; P > .10 (n.s.). F (n.s.). (interaction) = 0.01; dJ. = 1,30; P > .10 (n.s.). AVIAN CLUTCH SIZE AND PARENTAL SURVIVAL 287
TABLE 7. Returns in 1977 of female parents of Tree Swallow study broods.
All females Females weighed
No. returned No. returned Category n (%) n (%)
Control: Yearling a One female that had been weighed died from unknown causes just prior to the successful fledging of her brood. Her weight loss was not abnormal. This bird was omitted from the table. DISCUSSION The effects of brood enlargement upon nestling growth were neither extreme nor The increased productivity of enlarged consistent. Fledging weights were signifi broods, the variability in nestling quality cantly lower in large broods reared by among broods, and the absence of brood yearling females, but this was not true size dependent mortality after fledging among broods reared by older females. suggest that larger clutches in the Long Tendencies for slower growth in larger Point population would not be limited by broods were in part compensated for by food availability, nor by the ability of at extended nestling periods. However, since least some adults to rear additional young. nest predation was deliberately prevented, These results are similar to those from its potential selective importance in favor studies of other passerines (e.g., von ing rapid nestling growth and short nest Haartman, 1967; Russell, 1972; Bryant, ling periods could not be assessed (d. Per 1975, Perrins and Moss, 1975) and do not rins, 1977). Mostinteresting was the result conform to Lack's predictions. Each mer that much of the variation in growth could its brief discussion. be attributed to differences among indi The most productive brood size appears vidual broods. Variation among broods in to be larger than the most common among the quality of young may reflect substan naturally occurring Tree Swallow broods tial variation in parental competence. The (Paynter, 1954; De Steven, 1978; but see male parent's contribution is potentially Stocek, 1970), but in the present study important but could not be evaluated in productivity also increased as brood size detail in the present study. Of the four was enlarged beyond the normal maxi nests at which I could not capture and mum of seven. This is probably not the thus ascertain the presence of a male par result of unique conditions in 1976. In two ent, three were experimental broods separate series of experiments performed reared by older females, and two of these at Long Point from 1970 to 1972 by M. were among the slowest growing in that Bradstreet and by D. Russell, broods of category. The other two broods, the third eight and nine young were successfully experimental and one control, showed av reared to fledging despite yearly variations erage growth. In the Rouse Martin, a sim in weather conditions (and, presumably, ilar swallow species, male feeding rates in food supply) (Russell, in LPBO Annual increased faster than female rates as brood Reports 1970-72 and Russell, pers. size increased; additional evidence em comm.). The question remains as to phasized the importance of the male to whether particularly unfavorable years successful breeding (Rails and Bryant, that may limit optimal brood size occur 1979). more infrequently than the time span over There is no clear evidence of differential which populations of this species have post-fledging mortality with respect to been studied to date. It is a pertinent ques brood size, even for experimentally en tion for other studies as well. larged broods (cf. De Steven, 1978). Row- 288 DIANE DE STEVEN ever, chance events may be important Swallows were not paralleled by subse with juvenile return rates as low as 6%, quent changes in survivorship. There was and more data on yearly variations in re no indication that females rearing abnor cruitment are needed. If conditions influ mally large broods survived less well than encing juvenile survivorship are variable females raising normal-sized broods. and unpredictable at the time that eggs However, return rates were higher than are laid, parents raising smaller broods the 30-40% of previous years (Hussell, in may eventually be favored over parents LPBO Annual Report 1973; De Steven, that risk failure by attempting large unpubl. data). This may be a function of broods in poor years (Perrins and Moss, increasing colony age and more efficient 1975; see also Murphy, 1968; Wilbur et trapping efforts in recent years (Hussell, al., 1974). Gillespie (1977) formalized a pers. comm.), but it could also indicate an similar idea to demonstrate that selection unusually favorable year for adult surviv can act on the variance in offspring pro al. Increased weight losses resulting from duction as well as on average production. rearing larger broods could influence sur Thus a smaller mean brood size is pre vival in poor years if females are stressed dicted if the success of larger broods is while molting and storing migratory fat more variable than that of small broods. reserves in the period following breeding. Results of this study suggest such a trend, Absence of a demonstrable reduction in but more data are needed. survival of female Tree Swallows as a re The hypothesis of an inverse relation sult of brood enlargement prompts two in ship between clutch size and adult surviv terpretations. First, demographic consid al (Williams, 1966; Charnov and Krebs, erations suggest that the adjustment of 1974) is an intuitively attractive modifi clutch size in direct relation to adult mor cation of Lack's hypothesis and is often tality may not be as important a selective assumed despite inadequate evidence. factor in species with high nonreproduc This study failed to support this hypoth tive mortality as in species with low non esis and may illustrate potential problems reproductive mortality (Ricklefs, 1973, p. in testing it experimentally. 426; Charnov and Krebs, 1974). The ar Although differences in weight loss gument is most easily applied to birds, were not significant, female Tree Swal since their annual survival appears to be lows raising experimentally enlarged relatively constant after maturity (Deevey, broods tended to lose more weight than 1947; but see Botkin and Miller, 1974). A females with normal-sized broods. The small reduction in a high annual adult sur trend is consistent with similar observa vival rate (e.g., from 96% to 92%, as in tions that have been interpreted as indic some seabirds) would correspondingly re ative of increased stress associated with duce life expectancy (and hence potential rearing larger broods (Hussell, 1972; Win future breeding opportunities) from 24 kel and Winkel, 1976). However, female years to 11.5 years (cf. Wooller and Coul weight loss during the breeding season son, 1977), whereas a similarly small re may partially reflect normal regression of duction in a lower annual survival rate ovary and oviduct tissue (Ricklefs, 1974 (e.g., from 75% to 72%) would reduce life and 1977), although this is more likely to expectancy from 3 years to 2.6 years, a be true early in the nesting cycle. This barely detectable change. Adult survival suggests more cautious use of weight rates of temperate passerines such as the changes as sufficient evidence of energetic Tree Swallow range from 40 to 60% stress due to reproduction (d. Ricklefs, (Ricklefs, 1973); thus the results of this 1974, p. 261), but neither are such changes study could support the contention that a irrelevant, as males may also lose weight trade-off between clutch size and adult during the nesting cycle (e.g., Bryant, survival is not important in short-lived 1979). species. Weight changes among female Tree However, a second and related issue is AVIAN CLUTCH SIZE AND PARENTAL SURVIVAL 289 that even if one wished to argue that small ification of Lack's hypothesis; extended changes in survival could be selectively studies of specific populations will be important, one is faced with the problem needed to evaluate other hypotheses. Fur of statistically demonstrating a difference ther complications will arise because the between very similar survival rates, which hypotheses are not mutually exclusive, would not be possible without extremely and proposed factors may act simulta large samples. For example, a sample of neously. The task ahead is to devise clear >200 birds in each category would be tests that allow distinction between alter needed to demonstrate a significant differ natives, where possible. Satisfactory ex ence between the 58% return rate of Tree planations are unlikely to be unitary, but Swallow "control" females and the 67% will involve a multiplicity of factors that return of "experimental" females (X2 test); may vary in relative importance in partic a sample of 500 birds in each category is ular cases. not sufficient to yield a significant differ ence between the 75% return of yearling SUMMARY "control" females and the 71% return of Modifications of Lack's hypothesis for yearling "experimental" females (d. Re the evolution of clutch size in nidicolous sults). Thus there may be practical limi birds predict modal clutches smaller than tations to testing the hypothesis of an in the most productive, under the assump teraction between clutch size and parental tion that large broods stress parents and survival in many species. reduce subsequent parental survival. This The absence of age-specific differences hypothesis is tested for the Tree Swallow in breeding success and survival in re by comparison of breeding success, weight sponse to brood enlargement despite clear loss, and subsequent survival between fe differences in clutch size and laying date males raising experimentally enlarged (De Steven, 1978) presents an interesting broods and females raising normal-sized paradox. The success with which yearling broods. As age-specific differences in female Tree Swallows reared extra young clutch size and laying date occur in this suggests that foraging inefficiency may be species, the effects of brood enlargement insufficient to explain the observed differ upon yearling females and upon older fe ences in breeding biology between year males are examined separately in order to ling and older females (d. De Steven, test the related hypothesis that stress im 1978). Other breeding features, such as posed by large broods should be more se the acquisition and defense of favorable vere for inexperienced parents. nest sites, may also be influenced by ex More young were fledged from enlarged perience and may exert indirect effects on broods than from normal-sized broods, age-specific breeding parameters (d. Fin but fledging rates of yearling and older fe ney and Cooke, 1978). male parents were similar. Brood enlarge In addition to the hypothesis tested ment resulted in greater variability in here, other hypotheses attempt to account nestling growth rather than in poorer av for mean brood sizes smaller than would erage growth. Methods of analysis attrib be possible if food were the primary lim ute most of the variation in growth to dif iting factor as Lack originally believed. ferences among individual broods rather These include predation favoring rapid than among different brood sizes; this may growth and short nesting periods (Skutch, reflect considerable variation in parental 1949), selective advantages of reduced abilities. Among broods reared by yearling variance in breeding success (Gillespie, females, nestlings in larger broods were 1977), and intergenerational effects result lighter in weight near fledging than in the ing in poorer reproductive performance of smallest broods, but no such differences adults reared from large broods (Fretwell, occurred among broods reared by older 1969; Andersson, 1978). This study points females. Mortality after fledging appeared out possible difficulties in testing one mod- to be independent of brood size. 290 DIANE DE STEVEN Females raismg experimentally en ---. 1979. Reproductive costs in the House larged broods did not show significantly Martin (Delichon urbica). J. Anim. Eco!. 48:655 greater weight loss or poorer subsequent 675. CHAPMAN, L. B. 1935. Studies of a Tree Swallow survival in comparison to females raising colony. Bird-Banding 6:45-57. normal-sized broods, nor did yearling fe CHARNOV, E. L., AND J. R. KREBS. 1974. On males appear more susceptible to stress clutch-size and fitness. Ibis 116:217-219. than older females. These results are con CODY, M. L. 1971. Ecological aspects of reproduc sistent with suggestions that the hypo tion, p. 461-512. In D. S. Farner and J. R. King (eds.), Avian Biology. Vo!. 1. Academic Press, thetical trade-off between clutch size and N.Y. adult survival plays a minor selective role COWLES, H. C. 1901. The physiographic ecology in shaping the life histories of short-lived of Chicago and vicinity. Bot. Gaz. 31:73-108, species; however, the results may also re 145-187. DE STEVEN, D. 1978. The influence of age on the flect practical problems in the detection of breeding biology of the Tree Swallow Irido small but selectively important changes in procne bicolor. Ibis 120:516-523. survival in response to brood enlarge DEEVEY, E. S., JR. 1947. Life tables for natural ment. populations of animals. Quart. Rev. Bio!. 22:283-314. ACKNOWLEDGMENTS FINNEY" G., AND F. COOKE. 1978. Reproductive habits in the Snow Goose: the influence of female I thank the Long Point Company for age. Condor 80:147-158. permission to work on Long Point, and FISHER, R. A. 1958. The genetical theory of natural the Long Point Bird Observatory for use selection. 2nd ed. Dover, N.Y. FORBUSH, E. W., AND J. B. MAy. 1939. Natural of facilities and access to data collected by History of the Birds of Eastern and Central Observatory personnel, especially in 1978. North America. Houghton Mifflin, Boston. The study could not have been completed FRETWELL, S. D. 1969. The adjustment of birth without the assistance of S. Freyburger, rate to mortality in birds. Ibis 111:624-627. GILLESPIE, J. H. 1977. Natural selection for vari who persevered with extraordinary toler ances in offspring numbers: a new evolutionary ance. G. Miller and A. Rivers also pro principle. Amer. Natur. 111:1010-1014. vided additional help at critical times. D. HAILS, C. J., AND D. M. BRYANT. 1979. Repro Hussell contributed valuable advice and ductive energetics of a free-living bird. J. Anim. support. K. Fiala, P. Grant, H. Howe, D. Eco!. 48:471-482. HOWE, H. F. 1976. Egg size, hatching asynchrony, Hussell, R. Payne, and R. Storer made sex, and brood reduction in the Common Grack helpful comments on the manuscript. The le. Ecology 57:1195-1207. study was supported by a Frank M. HUSSELL, D. J. T. 1972. Factors affecting clutch Chapman Grant from the American Mu size in Arctic passerines. Eco!. Monogr. 42:317 seum of Natural History. General support 364. KLOMP, H. 1970. The determination of clutch size was also provided during the writing of in birds: a review. Ardea 58:1-124. this paper by the Division of Biological KLUYVER, H. N. 1963. The determination of re Sciences, University of Michigan, and by productive rates in Paridae. Proc. 13th Inter. the Smithsonian Tropical Research Insti Ornitho!. Congr., p. 706-716. KUERZI, R. G. 1941. Life history studies of the Tree tute through a Noble Fund Fellowship in Swallow. Proc. Linn. Soc. NY 52-53:1-52. 1978. LACK, D. 1947. The significance of clutch size. Ibis 89:302-352. LITERATURE CITED ---. 1966. Population Studies of Birds. Clar ANDERSSON, M. 1978. Natural selection of off endon Press, Oxford. spring numbers: some possible intergeneration MOUNTFORD, M. D. 1968. The significance of lit effects. Amer. Natur. 112:762-766. ter-size. J. Anim. Eco!. 37:363-367. BALDWIN, S. P., AND S. C. KENDEIGH. 1938. Vari MURPHY, G. 1968. Pattern in life history and the ations in the weights of birds. Auk 55:416-467. environment. Amer. Natur. 102:391-404. BOTKIN, D. B., AND R. S. MILLER. 1974. Mortality O'CONNOR, R. J. 1976. Weight and body compo rates and survival of birds. Amer. Natur. sition in nestling Blue Tits Parus caeruleus. Ibis 108:181-192. 118:108-112. BRYANT, D. M. 1975. Breeding biology of House PALMER, R. S. 1949. Maine birds. Bul!. Mus. Martins Delichon urbica in relation to aerial in Compo Zoo!' Harvard 102:1-656. sect abundance. Ibis 117:180-216. PAYNTER, R. A. 1954. Interrelations between AVIAN CLUTCH SIZE AND PARENTAL SURVIVAL 291 clutch-size, brood-size, prefledging survival, and SKUTCH, A. F. 1949. Do tropical birds rear as many weight in Kent Island Tree Swallows. Bird young as they can nourish? Ibis 91:430-455. Banding 25:35-58, 102-110, 136-148. SOKAL, R R, AND F. JAMES ROHLF. 1969. Biom PERRINS, C. M. 1965. Population fluctuations and etry. W. H. Freeman and Co., San Francisco. clutch-size in the Great Tit Parus major. J. STEARNS, S. C. 1976. Life-history tactics: a review Anim. Ecol. 34:601-647. of the ideas. Quart. Rev. BioI. 51:3-47. ---. 1977. The role of predation in the evolution STOCEK, R F. 1970. Observations on the breeding of clutch size, p. 181-191. In B. Stonehouse and biology of the Tree Swallow. Cassinia 52:3-20. C. M. Perrins (eds.), Evolutionary Ecology. VON HAARTMAN, L. 1967. Clutch-size in the Pied Univ. Park Press, Baltimore. Flycatcher. Proc. 14th Inter. Ornithol. Congr., PERRINS, C. M., AND D. Moss. 1974. Survival of p. 155-164. young Great Tits in relation to age of female par ---. 1971. Population dynamics, p. 391-459. In ent. Ibis 116:220-224. D. S. Farner and J. R. King (eds.), Avian Biol ---. 1975. Reproductive rates in the Great Tit. ogy. Vol. 1. Academic Press, N.Y. J. Anim. Ecol. 44:695-706. WILBUR, H. M., D. W. TINKLE, AND J. P. COL PIANKA, E. R, AND W. S. PARKER. 1975. Age LINS. 1974. Environmental certainty, trophic specific reproductive tactics. Amer. N atur. level, and resource availability in life history evo 109:453-464. lution. Amer. Natur. 108:805-817. PINKOWSKI, B. C. 1977. Breeding adaptations in WILLIAMS, G. C. 1966. Natural selection, the costs the Eastern Bluebird. Condor 79:289-302. of reproduction, and a refinement of Lack's prin RICKLEFS, R E. 1968. Patterns of growth in birds. ciple. Amer. Natur. 100:687-690. Ibis 110:419-451. WINKEL, W., AND D. WINKEL. 1976. Uber die ---. 1973. Fecundity, mortality, and avian de brutzeitliche Gewichtsentwicklung beim Trauer mography, p. 366-435. In D. S. Farner (ed.), schnapper (Ficedula hypoleuca). J. Ornith. Breeding Biology of Birds. Nat. Acad. Sci., 117:419-437. Washington, D.C. WOOLLER, R D., AND J. C. COULSON. 1977. Fac ---. 1974. Energetics of reproduction in birds, tors affecting the age of first breeding of the Kit p. 152-292. In R A. Paynter (ed.), Avian En tiwake Rissa tridactyla. Ibis 119:339-349. ergetics. Publ. Nuttall Ornithol. Club, No. 15. ---. 1977. On the evolution of reproductive Corresponding Editor: H. M. Wilbur strategies in birds: reproductive effort. Amer. Natur. 111:453-478.