The Role of Parents in Sibilicidal Brood Reduction of Two Booby Species

Home , Bank cormorant, Crowned cormorant, Siblicide

The Auk 112(4):860-869, 1995

THE ROLE OF PARENTS IN SIBILICIDAL BROOD REDUCTION OF TWO BOOBY SPECIES

DAVID J. ANDERSON Departmentof Biology,Leidy Laboratory, University of Pennsylvania, Philadelphia,Pennsylvania 19104, USA; and Departmentof Biology,Wake Forest University, Winston-Salem, North Carolina27109 USA •

ABSTRACr.--Parentalregulation of the probability and timing of avian brood reduction is expectedto existand evolve becausenatural selectionshould favor parentsthat matchbrood size to food availability mostefficiently. Since food availabilityvaries among species, interspecificvariation in this regulation is expected.Comparative study of specieswith different brood reduction systemsis one way to test the hypothesisthat these mechanismsevolve as adaptations.Previous work on obligately siblicidal Masked Boobies(Sula dactylatra)and facultativelysiblicidal Blue-footed Boobies (S. nebouxii) has shown that their differenthatching asynchroniescannot fully explain the qualitativedifference in their brood reductionsystems. In this study,I report interspecificdifferences in nestshape that appearto contributeto early siblicide in Masked Boobies,but that suppressearly siblicide in Blue-footedBoobies. Blue- footedBooby nest shape is closelyregulated by parents.Differences in eggsizes of nestmates do not appear to contributeto the differencein socialsystems. Received 19 January1995, accepted2 July 1995.

FATAL SIBLING AGGRESSION,known as "sibli- 1987). Thus, parents may benefit from regulat- cide," often reduces the brood size of a number ing siblicide. Specifically, they are expected to of bird species(Stinson 1979,Mock et al. 1990). have experiencednatural selection for regula- Lack's(1954) brood-reduction hypothesis is the tory mechanismsthat rank offspring for sur- basisfor the common evolutionary interpreta- vival within a competitive hierarchy, but sup- tion of siblicide, and of brood reduction (Rick- press siblicide that is not in the parents' best lefs 1965) in general. When food supply for interests.Whether the first or secondobjective nestlings cannot be predicted reliably at the is emphasizedduring selectionwill depend on time of laying, selectionis expectedto favor a the extent of brood reduction that is optimal strategyof hatching as many nestlingsas can for parents,and is expectedto vary acrossspe- be raised in the best conditions, and then of cies. eliminating nestlingsafter hatching, if neces- Past interest has focused on two mechanisms sary,to bring demand into line with supply (see to regulate sibling competition. Variation in the also Temme and Charnov 1987, Forbes and interval between successivehatchings of nest Ydenberg 1992). Siblicide provides parents an mates (hatching asynchrony) and in egg size optimizing tool with a notable advantage: in- explain much of the variation in competitive dividual nestlings are best able to gauge their ability within a brood (Ricklefs 1965, Parsons own physiological condition and can use that 1970, 1975, Hahn 1981, Braun and Hunt 1983, information,in conjunctionwith a competitive Slagsvoidet al. 1984, Hebert and Barclay1986, hierarchy, to maximize efficiency of brood re- Plogerand Mock 1986,Skagen 1987, Amundsen duction (Mock et al. 1987a). and Stockland 1988, Hebert and Barclay 1988, Reliance on dominant offspring to conduct Anderson1989a, Magrath 1992,Jover et al. 1993), siblicide also carries a notable disadvantage. with hatching asynchrony being the more im- Offspring may commit siblicide under less- portant (Stokland and Amundsen 1988, Ma- stringentconditions than is optimal for parents, grath 1992). In specieswith sibling aggression, especially if parent-offspring conflict exists available evidence shows that food distribution (Trivets 1974, O'Connor 1978, Dickins and Clark parallels the competitive hierarchy within the brood (Poole 1979, Braun and Hunt 1983, Fu- jioka 1985, Inoue 1985, Mock 1985, Cash and Evans1986, Drummond et al. 1986,Hagen 1986, Present address. Ploger and Mock 1986). Parents create acom-

860 October1995] Siblicidein Boobies 861 petitive hierarchyand, by not distributing food gatelysiblicidal. If both eggshatch, the A-chick ejects counter to the hierarchy, reinforce it through the second-hatched chick ("B-chick") from the nest positive feedback. scrape,and the victim dies from exposure or preda- Only rarely have comparativestudies tested tion. The timing of siblicidealso differs;the average age at death of victim Blue-footedBoobies is 18 days the hypothesisthat differencesbetween species (Drummond et al. 1986),while that of Masked Booby in siblicidal phenomena are explained by dif- victims is 1.8 days (Anderson 1989a).Masked Booby ferencesin regulatorymechanisms. Ideally, such parentsare preventedfrom fledging two offspringby studies should compare speciesthat differ in unrelenting sibling aggression(Nelson 1978,Ander- outcomeof siblingcompetition, they shouldfo- son 1989a); the insurance value of the second egg cuson interspecificvariation in traits known to againstthe first egg'sfailure to hatch apparently ex- influence competitive hierarchies, and they plains why they lay two eggs (Dorward 1962, An- should eliminate to the extent possible con- derson1990a). Some data suggestthat parent Masked founding effectsof phylogeneticand ecological Boobiescould provide enoughfood for two surviving dissimilarity. Two studies have compared the nestlings (Anderson 1990b, Anderson and Ricklefs 1992),but A-chicks do not permit parents to try. Even hatching asynchroniesof obligately siblicidal if the parents could suppresssiblicide behaviorally, eagles(Edwards and Collopy 1983)and boobies both parentsare absentfrom the nest site for up to 8 (Anderson1989a) with thoseof faculativelysib- h daily in experimentally-managedtwo-chick broods licidal congeners.Longer hatching asynchron- (Anderson 1990b). During these periods a subordi- ies (presumablyunder at leastpartial control of nate has no protectionfrom its sibling'sattacks. Par- parents)give greater competitive advantageto ents avoid investing in a doomed offspring and A- elder chicks of many siblicidal species (see chicksexpend less effort when siblicideoccurs shortly above);both studiesfound that hatching asyn- after hatching of the B-chick. chronies of obligately siblicidal specieswere A previous experimental study showed that the de- longerthan thoseof facultativelysiblicidal con- gree of hatching asynchronyinfluenced the probability and timing of siblicidein MaskedBoobies (An- geners.Mock and associatesfound that the de- derson 1989a).Broods hatching at intervals of at least gree to which food deliveredby parentscan be 4 daysvirtually alwayslost the B-chickwithin 20 days monopolized explains variation between two (total nestling period ca. 100 days);the averagetime ardeid speciesin the probability of siblicide to brood reduction was 1.8 days. Early brood reduc- (Mock 1984, 1985, Mock et al. 1987a, Mock et tion becameincreasingly unlikely as hatching asyn- al. 1987b);however, whether parentsactively chrony was decreasedfrom three to one days in ex- manipulatefood monopolizabilityas a regula- perimental broods.As expectedunder the "doomed- tory mechanism is unknown. offspring"scenario above, the meanhatching interval In this study,I presentcomparative data from in Masked Booby broods was 5.4 days (range 3-10 two siblicidal booby speciesof potential influ- days),above the "early reductionthreshold" (Ander- son 1989a)of 3 days. Anderson (1989a)also showed encesof sibling aggression.My goal is to iden- that hatching intervals of Blue-footed Boobies(:• = tify mechanismsthat parentsuse to regulatethe 3.5 days) were significantly shorter than those of outcomeof sibling aggression,and to test the Masked Boobies, but that more than 50% of Blue- hypothesisthat differencesin regulatorymech- footed Boobybroods hatched at intervals exceeding anisms contribute to differences in the social the three day early reductionthreshold. A Masked systemsof these species. BoobyB-chick hatches four days after its sibling and can expectto be killed within three daysof hatching. THE STUDY SYSTEM However, a Blue-footed Booby B-chick in a similar situationin the Galfipagoswill probablyfledge, if it Masked Boobies (Sula dactylatra)and Blue-footed escapespredators (Anderson 1989a, 1991, Anderson Boobies(S. nebouxii)are colonial, ground-nesting, pi- and Hodurn 1993) and bad weather (this paper). The scivorousseabirds that breed sympatricallythrough- degree of hatching asynchronyis the primary regu- out most of the Blue-footed Booby'srange (Nelson latoryvariable investigated in studiesof aviansibling 1978). Blue-footedBoobies lay one to four eggsper competition, but it cannot explain the qualitative dif- clutch (usually two) and are faculatively siblicidal; ference between the socialsystems of these two sib- first-hatchedchicks ("A-chicks") kill youngersiblings licidal species. in times of food stress(Drummond et al. 1986, Drum- A MaskedBooby A-chick ejectsits sibling by grasp- mond and Garcia Chevelas 1989). Nonetheless, more ing in its beak the sibling'sneck, appendage,or skin than one offspring often (Nelson 1978) or usually and extending its neck to thrust the B-chick across fledge(66% in caseof Drummond et al. 1986).Masked the nest scrape.Then, an A-chick often repeatsthe Boobieslay one- or two-egg clutches,and are obli- thrust after moving its own body 2 to 10 cm toward 862 DAVIDJ. ANDERSON [Auk, Vol. 112 the displacedB-chick. A seriesof thrustsoften moves gressionor of their own disorientation.Thus, hatch- the B-chick from the shade cast by the parent. Be- ing asynchronyand time required for brood reduc- havioral data, both observational (Nelson 1978:565, tion were measuredin increments of one day. this study) and experimental(Lougheed 1995),show Both specieslay their eggsdirectly on the ground that hatchlingBlue-footed Booby A-chicks display the in circular nest scrapesthat have been cleared of de- same behavior, but at lower frequency and without bris. I measured the diameters of 75 Blue-footed Boo- lethal outcome.Experimental cross-fostering of chicks by and 45 Masked Boobynest scrapesby placing a of one speciesinto nests of the other speciesdem- meter stick acrossthe scrapeat its maximum diameter onstratedthat both parentsand chicksplay regulatory and recordingthe distancebetween points of contact roles vis-a-visexpression and outcomeof sibling ag- with the ground. A secondmeasurement was made gression(Lougheed 1995). In the presentstudy, I fo- horizontally perpendicular to the first, and the av- cuson parentalinfluences that facilitatethe B-chick's erageof the two wastaken asthe nest'sdiameter. The ejectionin MaskedBooby broods and suppressit in maximum depth was measured at the center of the Blue-footedBooby broods. scrapewith another ruler held perpendicular to the horizontal meter stick. The nest scrape'smaximum steepness(r, the angle in degreesmade by horizontal MATERIALS AND METHODS plane and tangent to surfaceof scrapewhere surface nearsground level) is approximatedby Data presented here were collected from 1984 through 1986 as part of a long-term study of the I • = 180ø - 2[tan-•(R/D)], (2) breeding ecologyand behavior of Masked and Blue- where R is the radius (cm) and D is the maximum footed boobiesat Punta Cevallos, Isla Espafiolain the depth (cm; see Appendix for a proof). GallpagosArchipelago (for detailsof studysite, see Statisticalanalyses were done with SYSTAT (SYS- Anderson and Ricklefs 1987). This period fell between the E1 Nifio-Southern Oscillation events of TAT, Inc. 1984), CSS:STATISTICA (StatSoft, Inc. 1991), and STATISTICA (StatSoft,Inc. 1994) software. Step- 1982-1983 and 1986-1987 (Anderson 1989b). My as- sistants and I recorded nest histories and measured wise multiple regressionsof dependent variableson hatchinginterval and egg-volumeratio attemptedto chick growth of approximately250 Masked Booby include hatching interval in the model first, at a-to- and 100Blue-footed Booby breeding attempts in each enter = 0.15, where a is determined by an indepen- season.We checked nests daily between 1200 and dent variable'st-value (e.g. Neter and Wasserman1974: 1430,marked newly laid eggs,and weighedand mea- 386). suredchicks daily until the age of 10 days.We mea- suredthe maximumlength and breadthof eachegg, andapproximated the egg'svolume with the equation RESULTS V = •rLB2/6, (1) Egg-volumedifferences.--Variation in egg size where V is volume (cc), L is length (cm), and B is explainssignificant variation in the posthatch- breadth (cm; Preston 1974). We measuredchick mass- ing growth rate of somebird species,and dif- es with Pesolaspring scales;wing length was mea- ferencesin egg size within a clutch could in- sured by holding the metacarpalsat a right angle to fluence the outcome of siblicidal behavior the radius and ulna, and straightening the manus or throughan effecton body-sizedifference (An- longest primary. The A- and B-chickswere individ- derson 1989a). To do so, egg volume must in- ually identified using ink spots we placed on the head, plumagedevelopment, and, after about30 days fluencebody size. In order to explain interspe- cific differences in siblicidal behavior, the two of age,their U.S. Fish and Wildlife Servicealuminum leg bands. I excludedfrom analysespresented here speciesmust have different egg-volumeratios. three three-eggBlue-footed Booby clutches, and two Eggvolume explains 46.4 and 50.9%of variation Masked Boobybroods in which brood reduction did in hatching mass of Masked and Blue-footed not occurby 20 days after the B-chickhatched (see boobies,respectively, and the effectdecays with Anderson 1989a). increasingage but remains significantat least A chick was recordedas hatchedat a given midday througheight daysof age(Table 1).Egg-volume nestcheck when the chickwas completely outside its ratios(i.e. volumeA-egg/volume B-egg) of the egg shell, or the chick was still attachedto the egg two speciesdiffer in associationwith the brood- shell but had split the eggshell into two halves.Brood reduction was recorded when a chick was absent from reduction system: Masked Booby A-eggs usu- its nest scrapeat a particular day's nest check and, ally were largerthan B-eggs(Anderson 1990a), subsequently,did not return;frequently, chicks of while 24.4%of Blue-footedBooby A-eggs were both speciesreturned to their nestscrape after being the smallerof the two (Fig. 1, Table 2). A two- recordedoutside the scrapeas a result of sibling ag- way ANOVA, with speciesand year as main October 1995] Siblicidein Boobies 863

Blue-lootedBoobies the hatching-mass ratio on hatching interval MaskedBoobies 70 and egg-volume ratio showed,that both inde- pendent variables accounted for significant 60 variation in the hatching-massratio in Masked Boobies (hatching asynchrony, t = 5.54, P < 0.001; volume ratio, t = 2.91, P = 0.005), but not 4o in Blue-footed Boobies (volume ratio did not enter model at P = 0.15). Thus, the egg-volume 3o ratio appearsto favor A-chicks in Masked Boo- bies, but favors neither chick consistentlyin • 2o Blue-footedBoobies. However, the egg-volume lO ratio had no significant effect on an indicator of competitiveadvantage (i.e. number of days o Masked Booby A-chicks required to kill their siblings). In a multiple regressionof days to brood reduction on hatching asynchrony and egg-volumeratio, the effectof egg-volumeratio Egg-volume Ratio was not significant (t = 0.18, P = 0.86). Fig. 1. Frequencyhistogram of egg-volume ratios Rather than indicating adaptive differences (volume of A-egg/volume of B-egg) for Masked (n = in the brood-reduction system, the differences 171) and Blue-footed (n = 74) booby clutches. between the species in the egg-volume ratio may be proximate phenotypic effectsof differencesin food supply. Blue-footedBoobies lay effects,revealed a significantspecies effect (F•,239 larger clutches,raise larger broods, and make = 24.91,P < 0.001).Masked Boobies had a larger shorter foraging trips than do Masked Boobies mean egg-volume ratio than did Blue-footed (Nelson 1978, Anderson and Ricklefs 1987), all Boobies (œ= 1.11 + SD of 0.11, and œ= 1.03 ___ consistentwith greaterfood availability for fe- of 0.07, respectively;see Table 2). male Blue-footed Boobies than for female Hatching interval explains significant varia- Masked Boobieswhen allocating physiological tion in the ratio of the A-chick's mass to the resourcesfor secondeggs. If so, Blue-footed B-chick'smass on the day of the B-chick'shatch- Booby B-eggs should vary less in size across ing (hatching-massratio; r 2 = 0.41 for Masked yearsthan thoseof MaskedBoobies; data from Boobies, r 2 = 0.39 for Blue-footed Boobies; An- 1984through 1986support this prediction.The derson1989a). Stepwise multiple regressionof volume of A-eggs in two-egg Masked Booby clutchesdid not vary significantly acrossthe three years (F2,•68= 1.66, P > 0.05), while that TABLE1. Proportionsof variance (r2-values)in body of B-eggsdid (F•,•s = 8.36, P < 0.001);neither size(as indicated by massor wing length) explained by variation in egg size. Only A-chicks from two- A-/nor B-egg volumes of Blue-footedBoobies egg clutches used so as to minimize variation due were heterogeneousacross years (A-eggs, F•,• to parent quality. = 0.46, P > 0.05; B-eggs,F•,• = 0.33, P > 0.05). As a result, the egg-volume ratio was hetero- Masked Booby Blue-footedBooby geneousacross years for Masked Boobies(F2,]• Age Wing Wing = 6.40, P < 0.01), but not for Blue-footed Boo- (days) Mass length Mass length bies (F•;• = 0.16, P > 0.05). 0 a 0.464* 0.050* 0.509* 0.179' Nest-shapedifferences.--Both species nest on 2 0.163' 0.083* 0.722* 0.449* the ground, but while Masked Boobiessimply 4 0.110' 0.062* 0.387* 0.090* clear debris from a nest site and deposit eggs 6 0.139' 0.032 0.444* 0.120 8 0.154' 0.050* 0.382* 0.400* there, Blue-footedBooby eggs and small chicks 10 0.068* 0.042 0.021 0.266* are containedwithin a bowl-shapeddepression

* P < 0.05. (Nelson 1978:plate 12). On two occasionsdur- ßRegression equations of mass(M, in g) on eggsize (E, in cc)for age ing nestwatches, I observedBlue-footed Booby 0 days:Masked Booby, M = 0.820E 5.081(significance of slope,Ft,• A-chickspushing their hatchlingsiblings from = 76.2,P < 0.001;of intercept,t = 0.78,P > 0.08);Blue-footed Booby, M = 0.583E- 9.867(significance of slope,Fx,•t = 32.2, P < 0.001;of under the brooding parent in the manner of intercept,t = 1.60, P > 0.08). Masked Boobychicks (unpubl. data). However, 864 D^VID J. ANDERSON [Auk, Vol. 112

TABLE2. Mean egg volumesand egg-volumeratios (+SD) for two-eggbooby clutches.Mean egg-volume ratiosdiffer from ratio of meanA-volume to meanB-volume because A- andB-egg volumes covary (Anderson 1989c; see Welsh et al. 1988).

Volume (cc) Egg-volume Year n A -egg B-egg ratio Masked Booby 1984 66 70.68 + 5.44 65.64 + 5.43 1.079 + 0.058 1985 46 68.82 + 4.53 62.16 + 6.12 1.113 + 0.089 1986 59 69.80 + 5.78 61.40 + 6.92 1.148 + 0.156 Blue-footed Booby 1984 45 59.24 + 5.30 57.42 + 5.03 1.034 + 0.072 1985 18 60.15 + 5.70 58.48 + 5.55 1.030 + 0.030 1986 11 60.84 + 5.83 58.29 + 5.25 1.045 + 0.068

in both casesthe Blue-footed Booby B-chicks was most marked preceding and during the 44- rolled down the steep sidesof the nest into the to 53-day age class,when Masked Boobysibli- shade of the parent when releasedby the A- cide occurs. Blue-footed Booby chicks in this chick, suggestingthat parents could use nest age classwere in wider (radius of 175 vs. 127 shape to influence the outcomeof the A-chick's mm, t = 4.36, df = 14, P < 0.001) and deeper efforts.Specifically, obligately siblicidal Masked (44.7 vs. 21.6 mm, t = 4.71, df = 14, P < 0.001) Boobieswill facilitate early siblicideby using a nest scrapesthan were Masked Booby chicks. flat nest,and facultativelysiblicidal Blue-footed This is a meaningful differenceto a highly al- Boobieswill suppressearly siblicidewith a steep- tricial (i.e. feeble) nestling with a body length sided, bowl-shaped nest. of 15 mm. A chick attemptingto evict its sibling A two-way ANOVA of I', the maximumsteep- from a Blue-footed Booby nest would face a nessof the nest scrape,with speciesand nest- slope 64% steeper and 38% longer than would age class(10- to 13-dayintervals sincelaying of a chick in a Masked Boobynest. first egg;see Fig. 2) as main effects,showed that In comparisonto ground-nestingseabird spe- Blue-footed Booby nests were significantly cies that are not obligately siblicidal, Masked steeperthan were Masked Boobynests (species Boobies had the flattest nests measured (Table effect, F•,•32= 64.1, P < 0.001). This difference 3), further suggestinga link between nest steep- nessand the brood-reductionsystem. Masked Boobyparents used existing slight depressions as nest-scrapelocations, but Blue-footedBooby 33 ß Blue-lootedBooby parents actively excavatedand regulated bowl 30 ß Masked Booby depth. This point was demonstrated experimentally with eight Blue-footedBooby nests at 27 the eggstage by filling them with dirt and pack- 24 ing the fill. The nests'dimensions were mea-

21 suredthree times in the subsequent48.5 h, and the measurementscompared with the original e 18 nest dimensions. Nest width was not consis- (/'3 15 tently affectedby the manipulation.Nest depth

12 was initially significantly different from the 0-12 13-22 23-33 34-43 44-53 54-63 64-73 original depth, but returned gradually to the original dimensionswithin 48.5 h (Table 4). Nest Age (days) When I' valueswere classedby time sincethe Fig.2. Maximumsteepness (I•) of Maskedand Blue- eggswere laid (Fig. 2), variation acrossage class footed booby nestsin relation to time since laying of was significant in Blue-footed Boobies (F6,9o= nest'sfirst egg. Eggs hatched at approximately44 days. 3.19, P < 0.01), but not in Masked Boobies(Fs,• Period of Masked Boobysiblicide indicated by shad- = 1.41, P > 0.05). Blue-footed Booby parents ing. Data presentedas œ+ SE. were remarkably attentive to the shapeof their October 1995] Siblicidein Boobies 865

TABLE3. F-values(• + SD with n in parentheses)for ground-nestingseabirds.

Nesting-cycle Species stage P (degrees) Sourcea Single-egg,single-chick species Red-tailedTropicbird (Phaethon rubricauda) Egg 32.4 + 6.0 (18) 1 Hatchling 29.2 + 3.9 (8) 1 Multiple-chick species Cape Cormorant (Phalacrocoraxcapensis) Egg 54.6 + 10.6 (16) 2 Crowned Cormorant (P. coronatus) Egg 64.9 + 3.6 (7) 2 Bank Cormorant(P. neglectus) Egg 63.6 (21) 3 Hatchling 65.6 (21) 3 Blue-footedBooby Egg 24.8 + 4.6 (51) 4 Hatchling 29.5 + 3.7 (48) 4 Obligately siblicidal species MaskedBooby Egg 17.8 + 3.4 (27) 4 Hatchling 18.0 + 4.5 (17) 4 ßSource: (1) D. J. Anderson,unpublished data from JohnstonAtoll, CentralPacific Ocean; (2) R. E. Ricklefs,unpublished data from Malgas Island, South Africa; (3) Cooper 1986;(4) this study.

nests. Parents restored their experimentally lings were partially submergedin a pool of wa- filled neststo a steepnessthat closelymatched ter, and Masked Booby hatchlings were com- the original steepness(Spearman rank corre- pletely dry. In 1986,rain fell on 25 of the total lation, r = 0.93, n = 6, P < 0.01). In nonexper- days that I was present on the island (Fig. 3). imental nests,F increasedimmediately prior to The totalaccumulation during 17 of thosedays hatching, with peak values and lowest vari- was lessthan 10 mm; little mortality occurred ability during the critical period in which sib- of Blue-footedBooby nestlings of 20 days or licide occursin MaskedBooby nests (i.e. during less(three deaths). On the remainingeight days, the first 10 daysafter hatching),but not in Blue- 10 to 43 mm were recordeddaily, filling Blue- footed Boobynests (Fig. 2). I did not attempt to footed Booby nests with water and causing observe the behavior used to form nest bowls. hatchlingbody temperaturesto drop (pets.obs.). However, the soil in Blue-footedBooby nesting On theseeight days,17 Blue-footedBooby nest- areasis dry and fine-grained,and easily canbe lings of 20 days or lessdied (Fig. 3). Causesof excavatedwith a tool resemblinga booby'sfoot mortality other than hypothermia(e.g. preda- (K. Huyvaert unpubl. data). tion and starvation) could be ruled out in most Blue-footedBooby nest steepnessis greatest of these 17 cases.Nestlings older than 20 days at preciselythe time that steepnessis required were not affectedby heavy rainfall: only 2 of for suppressionof hatchling siblicide, and is 24 (0.08) deathsin this age group occurredon less steepand deep at other times (Fig. 2). I dayswith at least10 mm of rain, comparedwith identified a possiblecost to maintenance of a 17 of 32 (0.53) deaths in the 0- to 20-day age deep nest: hypothermia of hatchlings during group. heavy rains, when Blue-footed Booby hatch- I comparedthe mortality rates of nestlings

TAnrE4. Nestshapes (width and length,• + SD; ram)following experimental filling of Blue-footedBooby nest scrapes.An egg hatched in two nestsbefore the secondmeasurement; these nestsnot measuredafter hatching occurred.

Hours Paired Paired after fillingof comparison comparison nest n Width (ram) t P Depth (ram) t P Original 8 298.8 + 41.9 -- -- 29.0 + 5.42 -- -- 4.5 8 284.5 + 36.2 1.35 >0.20 15.6 + 6.55 6.30 <0.001 24.5 6 270.0 + 27.2 2.91 <0.05 24.2 + 6.68 2.36 0.07 48.5 6 293.3 + 28.8 1.38 >0.10 31.8 + 5.91 0.06 >0.90 866 D^VIDJ. ANDERSON [Auk,Vol. 112

50 prior to hatching,Blue-footed Booby nests were still steeperthan Masked Booby nestsduring the first four nest age classes(Fig. 2; two-way •,• 30 ANOVA, nest age classand speciesas main ef- r• fects;species effect F•,70= 32.52, P < 0.001). 10

DISCUSSION • •o Jan 1 Feb 1 Mar 1 Apr 1 May My goal was to identify factorsthat contrib- Date ute to the differences between Masked and Blue- footed boobies in their brood-reduction sys- Fig. 3. Rainfall (above) and Blue-footed booby tems. I detected no contribution of within-clutch nestling mortality for birds in 0- to 20-day age class in 1986. egg-sizedifferences, although interspecificdif- ferencesin this factorare consistentwith a reg- ulatoryrole. However, a novel factor,nest shape, differs between the two species:Blue-footed from the 0- to 20-dayage group that were pres- Boobynests, like thoseof other ground-nesting ent on a given day as a function of rainfall on seabirdssampled, have steepsides that impede that day. I useddata from the period of 6 Jan- ejectionby hatchling chicks, but Masked Boo- uary through 7 March only, becauseafter 7 bies have atypically flat nests,facilitating ejec- March the number of nestlingsin this age group tion. Moreover,nest shapeis closelyregulated never exceededthree, reducing the precisionof by Blue-footedBoobies in a manner that should the mortality-rate estimate. During the 6 Jan- suppresssiblicidal ejection of hatchling off- uary to 7 March period, the number present spring. averaged11.4 + 5.8 nestlingsper day. A higher Comparativestudies allow one to test the hy- proportion of these nestlings present died on pothesisthat regulatory mechanismsare adapt- days with at least I0 mm of rain than died on ed to maximize reproductivesuccess. Two stud- days with lessthan I0 mm (mean arc-sintrans- ies have approachedthis issue by comparing formed daily proportions were 0.39 and 0.03, hatching asynchroniesof obligately and fac- respectively;t = 7.44, df = 68, P < 0.001). This ultatively siblicidal eagles (Edwards and Col- test is conservativewith respectto the conclu- lopy 1983)and boobies(Anderson 1989a).Both sionthat rainfall and mortality are linked, given found that parents of obligately siblicidal spe- that three of the four deaths that occurred after cies hatched eggs at longer intervals (thus es- 7 March happenedon days with at least I0 mm tablishing more biasedcompetitive hierarchies) of rain. than did parentsof facultatively siblicidal spe- Masked Boobies in their flatter nests re- cies.Because obligately siblicidal offspringpre- mained dry during rainfalls, and none of the vent parentsfrom raising more than one chick, 17 deathsof A-chicksin the 0- to 20-day post- even with short hatching intervals (Anderson hatching age group occurredon days with at 1989a),these data are consistentwith a hypoth- least I0 mm of rainfall. Blue-footedBooby par- esis of adaptive adjustmentof hatching asyn- ents that had nests characteristic of Masked Boo- chrony that cutslosses early. However, they are bies might have similarly low hatchling mor- alsoconsistent with a proximate-levelhypoth- tality, but the limited variation in Blue-footed esis that the food limitation that restricts brood Boobynest shapearound the time of hatching size also restrictsthe rate of egg formation. (Fig. 2) did not allow that comparison.How- The contrastingnest shapes of parent Masked ever, nest shapesof Blue-footedBoobies prior and Blue-footed boobies offer clearer evidence to hatching were consistentwith regulation to- of an adapted regulatory mechanismthat has ward flatness:a one-wayANOVA of F showed respondedevolutionarily to both costsand ben- significantheterogeneity across nest age classes efits.The differenceis not explainedby the most (F6,90= 3.19, P < 0.01), and nestswere consis- likely alternative hypotheses. For example, tently flatter prior to hatchingthan during and thermal considerationssuggest that nestlingsin shortly after hatching (Fig. 2). Masked Booby the hotter environment should have a flatter F-valuesdid not showsimilar heterogeneity (Fs,• nest to increaseexposure to air flow. The study = 1.41,P = 0.24). In spite of the relative flatness site is equatorial,and both speciesnest in ther- October 1995] Siblicidein Boobies 867 mally stressfulsituations, but the pattern is the G. Hunt, Jr., D. Mock, R. Ricklefs, D. Rubenstein, J. oppositeof that predicted:Masked Boobieshave Smith, and D. Westneat for comments on an earlier significantly cooler, windiet nest sitesthan do version of the manuscript;and P. Petraitisfor statistical advice. Blue-footed Boobies(Anderson et al. unpubl. manuscript).Another alternativehypothesis in- volves the nature of the substrate:possibly a bowl-shaped nest is optimal for both species, LITERATURE CITED but perhapsthe soil in Masked Boobycolonies AMUNDSEN,T., AND J. N. STOKLAND.1988. Adaptive does not permit formation of a bowl. Masked significanceof asynchronoushatching in the and Blue-footedboobies do show nesting seg- Shag:A test of the brood reductionhypothesis. regation into different areas (Duffy 1984), but J. Anim. Ecol. 57:329-344. the easeof digging in the two areasis not sub- ANDERSON,D. J. 1989a. The role of hatching asyn- stantially different. A similar degreeof digging chrony in siblicidal brood reduction of two boo- effort yields a hole approximately20% shallow- by species.Behav. Ecol. Sociobiol.25:363-368. er in Masked Booby areasthan in Blue-footed ANDERSON,D. J. 1989b. Differential responsesof Booby areas (Anderson et al. unpubl. manu- boobiesand other seabirdsin the Gal•pagosto the 1986-87 E1 Nifio-Southern Oscillation event. script), but the difference in nest depths that I Mar. Ecol. Progr. Ser. 52:209-216. found in this studyapproached 100% at the time ANDERSON,D. J. 1990a. Evolution of obligate sibli- of hatching. cide in boobies. 1. A test of the insurance egg In conclusion,hatching asynchrony clearly hypothesis.Am. Nat. 135:334-350. influences the speed of obligate siblicide in ANDERSON,D.J. 1990b. Evolution of obligate sibli- Masked Boobies,but cannot provide a proxi- cide in boobies.2: Food limitation and parent- mate-level explanationfor the differencein so- offspringconflict. Evolution 44:2069-2082. cial systemsbetween Masked and Blue-footed ANDERSON,D. J. 1991. Apparent predator-limited boobyhatchlings (Anderson 1989a). Hatchlings distributionof Gal•pagosRed-footed Boobies Sula of both speciesare capableof siblicidal behav- sula. Ibis 133:26-29. ior, but Blue-footedBooby A-chicks face a sig- ANDERSON,D. J., m',lD P. J. HODUM. 1993. Predator behavior favorsclumped nesting in an oceanic nificant obstacleto ejectingsiblings in the wid- seabird. Ecology 74:2462-2464. er, steepernest bowl than MaskedBooby chicks ANDERSON,D. J., m,ao R. E. RICKLEFS. 1987. Radio- face. Experimentalmodifications of nest shape tracking Masked and Blue-footed boobies (Sula in thesetwo speciesthat do not disturb parents spp.)in the Gal•pagosIslands. Natl. Geogr.Res. and especially during periods of heavy rain 3:152-163. would complementthis comparativeapproach ANDERSON,D. J., m,ao R. E. RICKLEFS. 1992. Brood to costsand benefits of nest steepness.Manip- size and food provisioningin Masked and Blue- ulation of the "arena" in which sibling aggres- footedboobies (Sula spp.). Ecology 73:1363-1374. sion occurshas not been suggestedpreviously BRAUN,B. M., AND G. L. HUNT, JR. 1983. Brood reduction in Black-leggedKittiwakes. Auk 100:469- as a mechanismfor regulating the outcomeof 476. the interactions;this appearsto have a proxi- CASH, K. J., AND R. M. Evm',ls. 1986. Brood reduction mate effect on brood size in these two species. in the American White Pelican (Pelecanusery- throrhynchos).Behav. Ecol. Sociobiol. 18:413-418. ACKNOWLEDGMENTS COOPER,J. 1986. Biologyof the BankCormorant, part 4: Nest construction and characteristics. Ostrich I am grateful to the staff of the Charles Darwin 57:170-179. ResearchStation and the Servicio Parque Nacional DICKtNS,D. W., ANDR. A. CLARK.1987. Gamestheory Gal•pagos,and in particularto S. Harcourt,for lo- andsiblicide in the kittiwake gull, Rissatridactyla. gisticalsupport during fieldwork and to R. Ricklefs J. Theor. Biol. 125:301-305. for sharing unpublisheddata. Fieldwork was sup- DRUMMOND, H.,AND C. GARCIA-CHq. 1989. Food ported by the National GeographicSociety, Sigma shortageinfluences sibling aggression in the Blue- Xi--the ScientificResearch Society, a Universityof footed Booby.Anim. Behav.37:806-819. MichiganBlock Grant, the FrankM. ChapmanFund, DRUMMOND,H., E. GONZALEZ,ANDJ. L. OSORNO. 1986. and the GeorgeD. Harris Fund. I was supportedby Parent-offspringcooperation in the Blue-footed an NSF Pre-doctoralFellowship during parts of this Booby(Sula nebouxii): Social roles in infanticidal work. I thank: R. Brubaker, L. Hamilton, P. Hodurn, brood reduction. Behav. Ecol. Sociobiol. 19:365- R. Ricklefs,and especiallyS. Fortner for their field 372. assistance;D. Cheney, D. Duffy, A. Dunham, F. Gill, DUFFY,D.C. 1984. Nest site selection by Masked 868 DAVID J. ANDERSON [Auk, Vol. 112

and Blue-footedboobies on Isla Espafiola,Gal•- cific test of the prey-sizehypothesis. Anim. Be- pagos.Condor 86:302-304. hav. 35:1386-1393. EDWARDST. C., AND M. W. COLLOP¾.1983. Obligate NEI_SON,J.B. 1978. The Sulidae. Oxford Univ. Press, and facultative brood reduction in eagles:An ex- Oxford. amination of factors that influence fratricide. Auk NETER,J., AND W. W,•.SSERMAN.1974. Applied linear 100:630-635. statistical models. Richard D. Irwin, Inc., Home- FORnES,L. S., AND R. C. YDENn•RG. 1992. Sibling wood, Illinois. rivalry in a variable environment. Theor. Popul. O'CONNOR,R.J. 1978. Brood reduction in birds: Se- Biol. 41:135-160. lection for fratricide, infanticide, and suicide? FUJIOKA,M. 1985. Sibling competitionand siblicide Anim. Behav. 26:79-96. in asynchronously-hatchingbroods of the Cattle PARSONS,J. 1970. Relationshipbetween egg size and Egret Bubulcusibis. Anim. Behav. 33:1228-1242. post-hatchingchick mortality in the Herring Gull HAGEN,J. M. 1986. Temporal patterns in pre-fledg- (Larusargentatus). Nature 228:1221-1222. ing survival and brood reduction in an Osprey PARSONS,J. 1975. Asynchronoushatching and chick colony.Condor 88:200-205. mortality in the Herring Gull, Larusargentatus. HAHN, D.C. 1981. Asynchronoushatching in the Ibis 117:517-520. Laughing Gull: Cutting lossesand reducing ri- PLOGER,B. J.,AND D. W. MOCK. 1986. Role of sibling valry. Anim. Behav. 29:421-427. aggressionin food distributionto nestlingCattle HEnERT,P. N., AND R. M. R. BARCt.A¾. 1986. Asyn- Egrets(Bubulcus ibis). Auk 103:768-776. chronousand synchronoushatching: Effect on POOLE,A. 1979. Sibling aggressionamong nestling early growth and survivorshipof Herring Gull, Ospreysin Florida Bay.Auk 96:415-416. Larusargentatus, chicks. Can. J.Zool. 64:2357-2362. PRESTON,F.W. 1974. The volume of an egg. Auk 91: HESERT,P. N., AND R. M. R. BARCt.•¾. 1988. Parental 132-138. investment in Herring Gulls: Clutch apportion- RICKLEFS,R.E. 1965. Brood reduction in the Curve- ment and chick survival. Condor 90:332-338. billed Thrasher. Condor 67:505-510. INOUE,Y. 1985. The processof asynchronoushatch- SKAGEN,S.K. 1987. Hatching asynchronyin Amer- ing and sibling competition in the Little Egret ican Goldfinches:An experimental study. Ecol- Egrettagarzetta. Colon. Waterbirds8:1-12. ogy 68:1747-1759. JOV•R,L., X. RU•Z, AND M. GONZaLEz-MARTiN. 1993. SLAGSvOLD,T., J. SANDVIK,G. ROFSTAD,O. LORENTSEN, Significanceof intraclutch egg size variation in AND M. HUSBY. 1984. On the adaptive value of the Purple Heron. Urnis Scand.24:127-134. intraclutch egg-sizevariation in birds. Auk 101: LACK,D. 1954. The natural regulation of animal 685-697. numbers. Clarendon Press, Oxford. STATSOFt, INC. 1991. CSS:STATISTICA, version 3.1. LOUGHEED,L. W. 1995. Proximate level investiga- StatSoft, Inc., Tulsa, Oklahoma. tions of behavioral reproductive tactics of two STATSOFt, INC. 1994. STATISTICA for Windows, Galfipagosbooby species. M.S. thesis,Wake For- version 4.5. StatSoft, Inc., Tulsa, Oklahoma. est Univ., Winston-Salem, North Carolina. STINSON,C. H. 1979. On the selectiveadvantage of MAGRATH,R. D. 1992. Roles of egg massand incu- fratricide in raptors. Evolution 33:1219-1225. bation pattern in establishmentof hatching hi- STOKLAND,J. N., AND T. AMUNDSEN. 1988. Initial size erarchies in the Blackbird (Turdus merula). Auk hierarchyin broodsof the Shag:Relative signif- 109:474-487. icanceof eggsize and hatchingasynchrony. Auk MOCK, D. W. 1984. Siblicidal aggression and re- 105:308-315. sourcemonopolization in birds. Science225:731- SYSTAT,INC. 1984. SYSTAT,the systemfor staffs- 733. tics, versions 1.1 and 1.3. SYSTAT, Inc., Evanston, MOCK, D. W. 1985. Siblicidal brood reduction: The Illinois. prey-sizehypothesis. Am. Nat. 125:327-343. TEMME, D. H., AND E. L. CHARNOv. 1987. Brood size MOCK, D. W., H. DRUMMOND, AND C. H. STINSON. adjustment in birds: Economicaltracking in a 1990. Avian siblicide. Am. Sci. 78:438-449. temporallyvarying environment.J. Theor. BioL MOCK, D. W., T. C. LAMEY,AND B. J. PLOGER. 1987a. 126:137-147. Proximate and ultimate roles of food amount in TmVERS,R. L. 1974. Parent-offspringconflict. Am. regulating egret sibling aggression.Ecology 68: Zool. 14:249-264. 1760-1772. WELSH, A. H., A. T. PETERSON,AND S. A. ALTMANN. MOCK, D. W., T. C. LAM•, C. F. WILLImaS, AND A. 1988. The fallacyof averages.Am. Nat. 132:277- PELLL•rlER.1987b. Flexibility in the develop- 288. ment of heron sibling aggression:An intraspe- October1995] SiblicideinBoobies 869

APPENDIX C

This proof, in conjunctionwith Figure 4, showsthat the angle FAD (which approximatesthe steepnessof a nest) can be calculated when the lengths of lines AD and DE are known from field measurements. The D ••.•//G roun•urfa ce stepsin the proof are as follows: (1) Angle AED = tan-• (AD/DE). (2) Angle ACE = 180ø - 2(AED) be- causetriangle ACE is isocelestriangle. (3) Angle ACB is complementof angle ACD, and angle CAD is com- FJõ.•. Idealized•e•tJcal c•oss-secQo• o• a boobx plement of angle FAD. (4) Angle ACB equalsangle •est. O•IX shaded•e•Jo• existsat actual •est site. CAD becauselines BC and AD parallel, so angle ACD A•ies a•d shapesdescribed = angle FAD = angle ACE. (5) Thus, from 1 and 2, cula•shape •ep•ese•ts assumptJo• that •est-bowl su•- angle FAD = 180ø - 2[tan-• (AD/DE)]. •acesconform to that o• a sectJo•o• a sphere.