Breeding Seasonality and Reproductive Success of White-Fronted Bee-Eaters in Kenya

BREEDING SEASONALITY AND REPRODUCTIVE SUCCESS OF WHITE-FRONTED BEE-EATERS IN KENYA

PETER H. WREGE1'2 AND STEPHEN T. EMLEN 1'2 •Sectionof Neurobiologyand Behavior, Cornell University, Ithaca, New York15853 USA, and 2NationalMuseums of Kenya,Nairobi, Kenya

ABSTR•CT.--White-frontedBee-eaters (Merops bullockoides) at Nakuru, Kenya, reproduce in an environmentwhere foodsupply for provisioningnestlings is unpredictableand generally scarce.With data from 8 yr of study on three populationsof bee-eaters,we examinedhow these environmental conditionshave influenced the timing of breeding and patterns of nesting success. Breedingcoincided with the generalpattern of rainfall in the Rift Valley. Kenyaexperiences two rainy seasonsper year, and individual bee-eatersbred with eitherthe long or the short rains,but not both. Membersof a givenpopulation bred coloniallyand quitesynchronously, but membersof adjacentpopulations were often out of phasewith one another.As a result, the distributionof long- and short-rainsbreeding populations formed a spatialmosaic. Reproductivesuccess was highly variablebut typically low. Mean (+SD) fledging success was 0.57 + 0.83 young, and only one nest in four produced an independent offspring (6 monthsof age). Most prefledgingmortality (88%)was due to egg lossesand the starvation of young. Very little (7%) was attributableto predation.Social factors, primarily intraspecific nest parasitism,were responsiblefor nearly half of all egg losses,and representan important costof group living. Starvation,the singlemost important source of prefledgingmortality, claimedthe lives of 48%of all hatchlings.Starvation was greatestat times of low food availability,in nestswith large broods,and in neststended by pairs without helpers. Hatching was asynchronous.This enhancedthe ability of older nestlingsto monopolize limited food supplies,and resulted in the selectivedeath of the smallestnestling first. Such brood reduction, coupled with the ability of nestlingsto slow their developmentrate in responseto food stress,is consideredan adaptationfor copingwith the unpredictablevari- ation in food supply commonly faced by these birds. Received17 August1990, accepted18 January1991.

EVOLUTIONARYprocesses act through vari- of breeding as it relates to the annual pattern. ability in the lifetime reproductive output of We then provide reproductive successstatistics, individuals. An understanding of how selec- quantify sourcesof egg and nestling mortality, tion hasshaped the different componentsof life and discuss the influence of environmental con- histories depends on knowledge of the sources ditions on reproductive output and the evolu- of variability in reproductive success. tion of life history patterns. For 8 yr we have studied the social dynamics and reproductivesuccess of individually marked METHODS White-fronted Bee-eaters(Merops bullockoides) in the central Rift Valley of Kenya. These birds This study was conducted in Lake Nakuru Na- are nonmigratory, cooperativebreeders that live tional Park, Kenya. The bulk of our data come from throughout the year in coloniesof 25-400 in- 8 yr of concentratedstudy of 3 populations:Makalia dividuals. Three populations were studied in- (1977-1979), Badlands (1977-1978), and Baharini tensively, and known individuals were moni- (1980-1984).A fourth population (Enderit) was mon- tored continuously for 3-7 yr (depending on itored in 1977,and thesedata are included in analyses population). Helpers at the nest were a con- of seasonalityand overall reproductive success.Ad- ditional data obtained from the Baharini population spicuousfeature of the social system,and we (1985-1986)as well asother populationsin and around detailed their influence on measuresof repro- the national park are included in someanalyses, as ductive successelsewhere (Emlen and Wrege noted. 1991). White-fronted Bee-eaters are colonial, roosting In this paper we describethe salient features throughout the year, and breeding, in holesexcavated of resourceavailability and examine the timing 1-2 m into vertical earth banks or cliffs. These birds

673 The Auk 108: 673-687. July 1991 674 W•aE ANDEMLEN [Auk, Vol. 108 feed almost exclusively on flying insects,and ex- predatorsat Nakuru to snakesand mongooses,both tended family groups(clans) defend foraging terri- of which are known to consume the entire nest toriesin savannahabitat surrounding the colonysite contentsin one visit. MadagascarHarrier Hawks (for a general descriptionof the socialstructure and (Polyboroidesradiatus) and various small falcons cooperativebreeding behavior, see Emlen 1990).All might take single nestlingsfrom hole entrances, adultswere identified individually by permanentleg but we rarely saw these predatorsat our study bands and plastic wing tags,and were sexedby lap- colonies. arotomywhen in adult plumage. 4. Other eggloss--all lossesof eggs not meeting the criteria above.

Mr•SUR•M,wrr oF REPRODUCrIVE SUCCESS 5. Nestlingstarvation--young that disappearedsingly and for which there was prior evidence of mor- Behavioralobservations.--Behavioral aspects of nest- phological retardation of the individual (83% of ing phenology of the breeding colonies were ob- cases) or of nestmates. Often we found the car- served daily (2-5 observationhours per colony).In- cassesof suchnestlings either in the nestchamber formation on nest activity and failure, as well as or discardedbelow the colony. changesin the compositionof groupstending specific 6. Other nestlingloss--losses of nestlingsnot attrib- nests, was obtained from such observations. utable to either starvationor predation. Nestchecks.--We checked all activenests at regular 7. HorusSwifts--occasionally Horus Swifts (Apusho- intervals with a "ripariascope": a tube of adjustable rus)actively took over bee-eaternest chambersby length that contains lenses, lights, and a terminal enteringand pushingout eggsor killing nestlings mirror (Demongand Emlen1975). During egglaying, by puncturing them with their claws. nestswere checkeddaily (Baharini population)or ev- 8. Catastrophiclosses--eggs or young lost to severe ery secondday (Makalia, Badlands,and Enderit pop- environmental conditions (e.g. flooding or cave- ulations).Checks usually beganbefore the first egg, ins causedby heavy rains, rising streamlevels, or with supplementalchecks whenever unusualbehav- both). ior was suspected(e.g. intraspecificnest parasitism). Fledglingsurvival to independence.--Wecalculated Nest contentswere monitored periodically through- survivorshipfrom fledging to independencefor the out incubation but daily throughout the expected Baharini population from monthly censusdata. Al- hatching period. Checkscontinued at intervals of 3- though fledglings reach adult foraging proficiency 5 daysto monitorthe growth and fate of all nestlings. approximately6 weeksafter fledging (2.5 monthsof Beforefledging, eachsurviving nestlingwas removed from its nest hole, banded with a numbered alumi- age), they remain with the family unit and do not molt into the definitive adult plumageuntil 6 months num band, weighed, measured,and returned to the nest. of age. We usedthe plumagecriterion to define full independence.Survivorship was defined as the per- Nestingsuccess.--Actual fledging from the nest was centageof nestlingsthat reachedmorphological age not always observed, primarily becausethe age at 25 that were still alive at 6 monthsof age. fledging varied from a norm of 28-30 daysup to 42 Data from the Makalia and Badlands populations daysfor individuals with slowed development(Em- were not used in survival analyses.From 1978-1979, len et al. in press).Although chronologicalage varied, nestlingswere fitted with wing tagsjust before fledg- all nestlingsthat successfullyfledged first attained ing, and we suspectthat mortality was substantially morphologicalage 25 (i.e. the stageof morphological higher on fledglingswith tagsthan on thosewithout developmenttypically completedin 25 daysby well- (Emlen unpubl. obs.).Consequently, young from the fed nestlings).We thereforeused this morphological criterion as a standardized estimate of successful Baharini population were not given wing tags until approximately6 months of age. fledging. Sourcesof eggand nestling mortality.--All losseswere classifedto presumedcause of mortality accordingto ENVIRONMENTAL MONITORING the following criteria: Insect availability was considereda possiblepre- I. Unhatchedeggs--eggs still present in the nest dictor of reproductive success.Rainfall influenced in- chamberwhen the oldest nestling was 5 days old sect abundance, and local weather conditions (hours (longest span for entire clutchesto hatch was 4 of rain, strength of wind, etc.) affectedboth the flight days). activity of insects and the ability of bee-eatersto 2. Desertion--eggsstill present in the nest chamber capturethem. The availabilty of insectsto foraging when, before full-term incubation, the birds ceased bee-eaterswas therefore representedin analysesby incubation and roosting at that nest. variables of insect abundance, rainfall, and their in- 3. Predation(of eggsor young)--entire nestcontents teraction. that disappearedbetween two successivechecks RainfalL--Monthly total precipitationand number separatedby <6 days. Such losseswere usually of dayswith >0.5 mm of rain were derivedfrom daily detectedwithin 2 days.Hole-nesting limits likely precipitationdata recordedat the national park head- July1991] ReproductiveSuccess inBee-eaters 675 quarters,3 km from the Baharini population (Fig. 1). There was no significantdifference in the amountor pattern of rainfall collected at headquartersand at two sites near the southern study populations. Insect abundance.--We monitored the abundance of flying insects,the primary food of White-fronted Bee- eaters, with a 3 x 5 m Malaise net (Malaise 1937, Marston 1965).Netting stationswere establishedclose to typical foraging territories in each of the three primary study populations.Stations were sampled monthly from 1978-1984and biweekly from 1984- 1986. For each monthly or biweekly sample,daily subsampleswere collected on consecutivedays until a minimum of three (maximumof six) subsampleswere obtained under acceptable weather conditions. Un- acceptableweather conditions were (1) temperatures > 15%below the monthlyaverage, or (2) steadywinds >5 mph, or (3) rain. Eachdaily samplingperiod ran from 0800-1600. Samples were sorted to remove insects<5 mm in body length (smaller insectswere rarely capturedby the birds; Hegner 1981), oven- dried for 12-16 h at 80øC,then weighed to +0.001 g. From thesesubsamples we calculatedthe mean daily biomassof insectsper sampling period. In the following analyses,estimates of insect abundancefor particulartime intervalswere calculatedby integrat- ing under the curve describedby all sample means Fig. 1. The locationof 10 Meropsbullockoides pop- connectedby straightline-segments. ulationsin Lake Nakuru National Park, Kenya, and the surroundingarea. Shadedcircles = populations STATISTICAL ANALYSES that breed during the short rains;shaded squares = thosethat breed during the long rains. The primary We used SAS Statistical Software, version 5.1 (SAS studypopulations were Baharini,Makalia, Badlands, 1985), for most statisticalanalyses. For logistic re- and Enderit. gressionwe usedthe PLR procedure,BMDP Statistical Software (1983). Kenya correspondto the single rainy season RESULTS found elsewhere in the species'range. BREEDING SEASONALITY Populationsof White-fronted Bee-eatersnear Nakuru have a single primary breeding season White-fronted Bee-eaters inhabit wooded sa- each year, roughly coincidingwith either the vannah habitat acrossthe entirety of central long or the short rains. Interestingly, popula- Africa, from the DemocraticRepublic of Congo tions living within a few kilometers of one an- and Angola in the west, through Kenya in the other may be timed to opposite seasons.Pop- northeast, south to Mozambique and Natal ulations studied in the Nakuru area between (South Africa). This habitat, a mosaicof grass- 1977 and 1986 were either "long-rains" or lands with scattered bush and open acacia "short-rains" breeders(Fig. 1). The mosaicna- woodland, is characterizedby pronounceddry ture of the distribution is apparent. The four and rainy seasons.Throughout most of its range, populations studied most intensively, and for there is one rainy seasoncentered around Oc- which we present data on reproductive success tober to December. But close to the equator in below, were Makalia, Badlands, and Enderit in Kenya and Tanzania, the intertropical conver- the southern portion of the national park, and gence causestwo rainy periods each year, the Baharini in the north. "long rains" in March to May, and the "short In spite of the clear seasonality in breeding rains" from October to December, with a pro- by populationsnear Nakuru, the rainfall pat- nounced dry seasonfrom June to September tern in this region differsfrom the Kenyan norm; (Sinclair 1978; see Fig. 2b). The short rains of when averagedover many years,it showsonly 676 WRmE^ND EMLEN [Auk, Vol. 108

A. • 180 a single rainy period and lacks a well-defined dry period between June and September(Fig. 2a, data from Brown and Britton 1980). The bi- modal pattern (Fig. 2b) is typical of the Rift Valley generally, and probably representsthe pattern experiencedby bee-eatersover most of their rangein Kenya.Timing of breedingin the Nakuru area includesthe numberof nestsiniti- ated per month for all primary nestingsof the four study populations(Fig. 2c) and the number of colonies(Fig. 2d) in the larger Nakuru area that initiated breeding in a given month (data from 1973 and 1975-1987). These comparisons suggestthat the timing of breeding by the Na- kuru bee-eaterpopulations reflects the rainfall pattern found elsewhere in the species'range rather than the pattern found locally. The mean monthly rainfall and insect abun- danceat our study site varied over a 9-yr period (Fig. 3). Rainfall washighly variable,both from month to month and between years.The influence of rainfall upon aerial insect abundance also varied. Generally, relative insect abundance tracked changesin rainfall with only a slight lag, a pattern most clearly seen for the years1980 through 1985.The long-rainsbreeders (Makalia, Enderit) generally initiated nesting after Nakuru's annual dry period, whereas short-rainsbreeders initiated before the dry period (Fig. 3: bottom). Other than this very gen- Do 15 eral pattern, however, neither insectabundance nor rainfall showed a consistent trend at the • z2 time of breeding. On three occasions individuals of the Bahari- -•Z9 o• Ouj ni population attemptedto raisea secondbrood (Fig. 3: hatched blocks).This was not observed in the other populations and has been docu- mented in only one other speciesof bee-eater 0 J F M A M J J A S O N D (Meropsoreobates; Fry 1984). We prefer to call MONTH this event secondarynesting because participants included not only previously successfulpairs Fig. 2. Correlationsbetween timing of breeding and rainfall in the Nakuru area, and a comparison (the usual definition of secondbrooding) but with rainfall patternselsewhere within the rangeof alsounsuccessful pairs and individuals that had Meropsbullockoides: (A) averagemonthly rainfall in been helpers in the primary nesting. Although Nakuru over 56 yr (fromBrown and Britton1980), (B) clutcheslost early in incubation were some- averagemonthly rainfall from Naivasha,Magadi, times replaced immediately, such losseswere Amsha, and Moshi in the Rift Valley (Brown and rare (see below). Nests that failed later in the Britton 1980),(C) frequencyof nest initiation in the cycle were not usually replaced. Secondary 4 studypopulations, 1977-1985, and (D) mediandate nesting was more or less synchronizedin the of egglaying for 15populations within a 30-kmradius colony as a whole; pairs that failed relatively of Nakuru National Park. Data from 1973, 1975-1986; early waited until severalweeks after successful the number of observationsper population varies. pairsfledged young from the primary attempt. In 1981 and 1982, secondarynesting followed largely successfulprimary nesting,whereas in 677 ReproductiveSuccessinBee-eaters 1991]

MONTHLY(dr•ginsMEAN / daY)INSEC:T$ 678 WREGEAND EMLEN [Auk, Vol. 108

T^BLE1. Reproductivesuccess (• + SD)for fourpopulations of White-frontedBee-eaters in Nakuru,Kenya. Bahariniand Badlandspopulations bred with the shortrains; the Makaliaand Enderetpopulations bred with the longrains. Samples include all nestswith accuratedata for the phaseof reproduction;sample sizesare in parentheses.

% nests Clutch size No. hatched No. fledged successful Baharini 1980 2.60 + 0.74(53) 1.70+ 1.23(46) 0.82 + 0.82(51) 61.1(54) 1981 3.16 + 1.75(44) 2.32 + 1.04(31) 0.80 + 0.84(45) 57.7(52) 1982 2.62 + 0.82(34) 2.06 + 1.15(36) 1.20 + 0.99(35) 72.5(40) 1983 2.00 + 0.71(25) 1.08+ 1.15(25) 0.28 + 0.46(25) 38.2(34) 1984 2.91 + 1.61(22) 2.24 + 1.34(21) 0.80 + 0.77(20) 71.4(28) Total 2.70 + 0.82(178) 1.89 + 1.23(161) 0.81 + 0.85(176) 60.1(208) Badlands 1977 2.91 + 0.73(23) 2.70 + 0.70(23) 1.50+ 1.14(32) 79.4(34) 1978 2.64 + 0.58(22) 1.95 + 1.05(20) 0.65 + 0.75(20) 50.0(20) Total 2.78 + 0.67(45) 2.35 + 0.95(43) 1.17+ 1.08(52) 68.5(54) Makalia 1977 2.62 + 0.89(37) 1.03 + 1.30(38) 0.03 + 0.16(39) 02.6(39) 1978 2.68 + 0.69(85) 1.95 + 1.22(84) 0.31 + 0.62(86) 23.0(87) 1979 2.43 + 0.80(42) 1.86+ 1.23(44) 0.26 + 0.57(46) 19.6(46) Totall 2.60 + 0.77(164) 1.45+ 1.34(166) 0.23 + 0.55(171) 17.4(172) Total2a 2.59 + 0.77(159) 1.62+ 1.31(141) 0.45 + 0.69(88) 33.7(89) Enderit 1977 2.50 + 0.76(26) 2.00 + 1.05(21) 0.27 + 0.60(40) 20.0(40) 1977a 2.50 + 0.76(26) 2.00 + 1.05(21) 0.37 + 0.67(30) 26.7(30) Baharini secondarynestings 1981 2.29 + 0.76(7) 0.80 + 1.30(5) 0.00 + 0.00(7) 12.5(34) 1982 2.50 + 0.62(34) 1.91+ 1.01(33) 0.74 + 0.95(35) 47.2(36) 1983 2.17 + 0.58(12) 1.00+ 1.28(12) 0.08 + 1.28(13) 07.7(13) Total 2.40 + 0.63(53) 1.58+ 1.18(50) 0.49 + 0.84(55) 33.3(57)

Without catastrophiclosses.

1983 fledging successin the primary breeding pearto havelower mean success rates than nests was extremely low. in the correspondingprimary breeding seasons. Poolingall data,fewer than half of the breed- GENERAL PATTERN OF REPRODUCTIVE SUCCESS ing attemptssucceeded in fledgingyoung (41% of 531 nests),and the overall mean (_+SD) num- Breeding in each population was character- ber of chicksfledging was 0.57 _+0.83. When ized by high varianceand low success(Table catastrophiclosses are omitted, 50% of 438nests 1). Nests initiated during the short rains (Ba- were successful,and the mean number of chicks harini and Badlandspopulations) averaged 0.89 fledgingwas 0.70 _+0.87. Of thesefledglings, fledgedyoung, while nestsfrom the long-rains 58%survived to reachindependence (6 months populations (Makalia, Enderit) averaged 0.24 of age).The cumulativeprobability of survival fledglings.The long-rains(but not the short- to each major stageof developmentis shown rains) populations suffered high catastrophic in Fig. 4. lossesdue to floodingof the river systemsalong whose banks the colonies were located. When SOURCES OF VARLa•TION IN we excluded these lossesfor purposesof com- REPRODUCTIVE SUCCESS parison, average nesting successin the long- rains populationswas 0.43 fledgedyoung, still High variabilityoccurred in both hatching considerablylower than the averagefor short- and fledgingsuccess between years and popu- rains populations. Although secondarynest- lations(Table 1). An understan din g of thisvari- ings in Baharini involved few nests,they ap- ationbegins with an examinationof the deter- July1991] ReproductiveSuccess inBee-eaters 679 minants of clutch size and the subsequent sourcesof egg and nestling mortality. Clutchsize.--Mean clutch size, averagedacross all populations, was 2.62 eggs, and it varied significantlybetween years (ANOVA: Fyear= 6.54, df = 9/409, P << 0.01). Long-rains and short-rains populations did not differ signifi- cantly (t = 1.04, df = 348, P > 0.3), but clutches laid during the secondarynestings in Baharini were significantlysmaller than thoseof primary Fig. 4. Cumulative transition probabilities from nestings (t = 2.59, df = 238, P < 0.02). egg laying to independence.Probability of survival Severalfactors might contributeto clutchsize through eachbreeding phase, from eggsto fledging, variability: (1) the condition or quality of the was calculatedfrom all populationsand all years.Sur- breeding female, (2) the availability of re- vival from fledgingto 6 monthsof agewas estimated sourcesnecessary to produceeggs, (3) the pres- from Baharini primary nesting data only. Barsindi- ence of helpers at the nest (which allofeed the cate the standard error of cumulative probabilities, calculated for each population-year. breeding female), (4) the time of laying within the season,and (5) the anticipatedfuture availability of food for feeding nestlings (Klomp correlation, and repeatability of clutch size 1970, Cody 1971, O'Connor 1984). To examine within females was low (r = 0.12). clutch-sizevariation, we used a multiple AN- Sourcesof eggand nestling mortality.--All non- COVA model that included variables repre- catastrophicsources of prefledgingmortality are senting each of these factors.The first was rep- diagramedin Fig. 5. Nestslost to flooding were resentedby femaleidentity and age;the second excluded to facilitate interpopulation compar- by insectavailability in the month of laying, as ison. The two largest causesof mortality were well as averagedover the 3 months before lay- egglosses and starvation of nestlings.Predation ing; the third by groupsize at the time of laying; on eggs and young was of only minor impor- the fourth by date of laying relative to the me- tance to White-fronted Bee-eaters.These pat- dian for that season;and the fifth by insect terns were consistentin most years and in all availability during the nestling period. The populations. analysis was limited to Baharini females for Of all eggs laid, 28% never hatched. This ac- whom we had data on a minimum of three counted for 40% of all prefledging losses.At clutchesfrom three different years (n = 33 fe- least 18% of these losses were due to the active males). removal of eggsby bee-eaters.Removed eggs Clutch size was correlatedsignificantly with frequently were found under the Baharini and insectavailability at and before laying, but not Badlands colonies, which confirmed that pre- with female quality, group size, date of laying, dation was not involved in such losses. Similar or insectabundance during the nestlingperiod. confirmation was not available for many Maka- Two measuresaffecting insect availability (in- lia coloniesbecause they were locatedover wa- sectabundance in the month before laying, and ter. This may partly explain the higher value rainfall over the 3-month period before laying) for egg predation in Makalia (Fig. 5), because accounted for 16% of the variation in clutch size entire clutches removed by bee-eaterswould (F...... = 10.67, df = 1/76, P << 0.01; Frei,= 13.75, erroneouslybe assignedto predationunless the df = 1/76, P • 0.01). The correlation with insect eggswere actually found below the colony. abundance was negative. This indicates that Intensive behavioral observations and fre- clutch sizeswere actually smaller in yearswith quent nest checksat the Baharini primary nest- higher insect densities.The biologicalsignifi- ing coloniesallowed us to examine in detail the cance of this correlation is unclear. Most of the causesbehind this high rate of egg loss.Social variation in clutch size (32%) was attributed to factors accounted for nearly half (48%) of all characteristicsof the individual breeding fe- egg losses(Table 2). The single largest source males (Ffem•,e= 1.48, df = 32/76, P = 0.08); how- was intraspecific nest parasitism (24% of egg ever, we were not able to identify any particular losses).Females parasitizing nests sometimes characteristic(age, breeding experience, time tossedout a host egg before laying their own, with current mate) that was responsiblefor the and parasiticallylaid eggsoften failed to hatch 680 WRECEAND EMI, EN [Auk, Vol. 108

BAHARINI - 1980-84 BADLANDS - 1977-78 501 EGGS 93 EGGS

egg predafion 1% (2%) nestlingpredation 3% (4%) unknownnestling loss

MAKALIA 1978-79 BAHARINI RENESTS- 1981-83 123 EGGS 112 EGGS

2% (2%)

3% (3%)

egg predation 3% (3%) g% 01%)

unknownnestling loss 5% (6%) Fig. 5. The causesof egg and nestling mortality in three populationsof bee-eaters.All primary nestings are pooled for each population (all secondarynestings are pooled in the relevant plot for Baharini). Each sectorof the diagram representsthe percentageof all eggslaid that were lost to a particular mortality source. The percentageof all lossesattributed to eachsource is indicatedin parentheses.Makalia population colonies that sufferedthe lossof virtually all eggsduring floodingwere omitted. becausethey were laid after the host clutch had ther a single young fledged or none at all; and been completed. More frequently eggs were only 4% of nestswith young had 3 young suc- tossedout by the host female herself. Parasitic cessfullyfledge (n = 322 nests). eggslaid before the host'sown first egg were Five lines of evidence indicate that what we removedby the hostfemale (Emlen and Wrege have called starvation did indeed result from 1986). On four occasionsbreeding femalesre- the inability of the breeders(plus any helpers) moved their entire clutches shortly after at- to sufficiently provision nestlings.First, most tempted or successfulparasitism, presumably nestlingsshowed signs of retarded develop- becausethey detected the parasitism in some ment before they were found dead or disap- way (n = 12 eggs in Table 2). peared from the nest. Competition between two females over the Second, starvation losses followed a consis- same mate or for the same nest hole, and the tent pattern in which the smallestnestling in suddenloss of a matethrough conscription (when the brood died first. Nestlings hatched asyn- a breeder is socially harassedinto abandoning chronously (at approximate 24-h intervals). its mate to help at the nest of closekin; Emlen Consequently, chicks within a brood were 1981, Emlen et al. unpubl. obs.), together ac- clearlydifferentiated by sizeand morphological countedfor an additional 15%of egg losses. development within 3 or 4 days. We assume The singlelargest source of prefledgingmor- that the size hierarchyat this early age reflects tality was nestling starvation. Starvation ac- the hatching sequence.A similar pattern was counted for 48% of all noncatastrophiclosses reportedfor the Red-throatedBee-eater (Merops and was a conspicuousaspect of bee-eater re- bullocki)by Dyer (1979). He found that under productionin every population in every year; conditions of food stress,asymmetries in the 48% of all hatchlingsstarved before they reached weight of asynchronouslyhatched broodmates fledging age. Although modal clutch size was increasedwith increasingage, and mortalityalso 3 eggs,in three-quarters(76%) of all nestsei- followed the sequenceof youngestfirst. In the July1991] ReproductiveSuccess in Bee-eaters 681

lOO TAnrE 2. Causesof egg mortality during primary 2O nestingsof the Baharinipopulation. All egg losses that could be attributed to some particular source 138

of mortality are included (17 eggslost to unknown 80 causes are omitted). z

No. of % egg Causeof egg loss eggs losses 24O Sociallyinduced losses Parasitism 35 24.3 Mate competition/hole takeover 12 8.3 4o Conscriptiona 9 6.3 Other social causes 13 9.0

Nonsocial losses 20 268 Desertion after breeder death 26 18.1 Other desertion 4 2.8 Sterile eggs 19 13.2 Predation 10 6.9 FIRST SECOND THrRD FOURTH Single egg lossesb 16 II.I HATCH ORDER Total c 144 I00.0 Fig. 6. The probability of starvationas a function • Conscriptionis when a nestingattempt is abandonedby onebreeder of competitionfrom older nestmates.Pooled data are to becomea helper at another nest,usually of closekin. Typically such conscriptionis precededby socialharassment from an older member from primary nestingsof the four studypopulations. of the clan. Above each bar is the number of nests that included • Single eggs that disappearedduring the incubation period were a nestling hatching first, second,third, or fourth (i.e. probably removed by the breedersand/or helpers. These eggswere the secondbar includes all nestswith ->2 nestlings, frequentlyfound underthe nestingcolony. cThe total sampleof eggsincluded in this analysisand resulting but not neststhat hatchedonly one young). percentagesdiffer from Figure 5 because(a) lossesdue to unknown causesare omitted here, and (b) Figure 5 includesonly nestswith completelyaccurate data on both clutchsize and fledgingsuccess. crease in the number of older nestmates(Fig. 6). The first-hatched chick had an overwhelming competitive advantage over its broodmates. Baharini population, 100%of starvationlosses First-hatchedchicks suffered a 19%probability in broodswith multiple young tookthe smallest of starving before fledging. The likelihood of nestling.For the oldestnestling, the early es- starvation increased to an overwhelming 84% tablishmentof a hierarchyin sizeand mobility and 95% for chicks with 2 and 3 older nestmates, may be a direct advantagein competitionwith respectively. siblingsfor accessto food. In 1987, we dug an Fourth, starvation losses were reduced when underground observationpit from which we helpers were present (Emlen and Wrege 1991). could directly view an active nesting chamber Helpers play full roles in provisioning nest- (for technique, see Fry 1972). Limited obser- lings and bring food to nestlingsat 78%the rate vation of nestling interactions and adult feed- of the breeders. When the influences of food ing behaviorshowed extreme aggression by the availability and the quality of the breederswere oldest nestling toward its smaller siblings,in- controlled statistically,helper number was a cluding vigorouspecking and violent shaking highly significantpredictor of both starvation while holding the neck.Food items brought by rate (F = 17.92,df = 1/126, P < 0.01) and fledg- adults were monopolized by the oldest nesting success(F = 32.46, df = 1/178, P

Predation-relatedlosses of eggsand nestlings different colony sites;median dates of egg lay- were very low (7%) in all populations. The low ing between the two subpopulationsdiffered predation rate was presumably due to the bee- by as much as 4 weeks (Fig. 3). eater's habit of excavatingtheir nests 1-1.5 m Dittami and Gwinner's (1985, Dittami pers. deep into vertical cliffs and river banks. The comm.) work on several passerinespecies in principal(and infrequent)predators were snakes Kenya'sRift Valley providesa possiblemech- (Boomslang,Dispholidus typus; Egg EatingSnake, anisticexplanation for why White-fronted Bee- Dasypeltisscaber; Spitting Cobra, Naja sp.; Rock eaterpopulations cannot breed with both rainy Python, Pythonsebae) and mongooses(Egyptian, seasonsor easily shift from one seasonto the Herpestesichneumon; Slender, H. sanguineus; other in successiveyears. They suggest that EasternDwarf, Helogaleundulata). All had dif- breeding initiation does not depend on envi- ficulty gaining accessto nests,but all consumed ronmental cues per se, but rather on the occur- the entire nest contents when they did so. rence of "permissive" environmental conditions following a fairly rigidly programmed annual molt (seealso Fogden 1972, Fogden and DISCUSSION Fogden 1979,Stiles 1980).The result may be a BREEDING SEASONALITY somewhatflexible onset of breeding from year to year,depending on whether sufficientlyharsh White-fronted Bee-eatersin Kenya inhabit an environmental conditions inhibit nesting once unpredictable environment in which there the molt cycle is completed and breeding is generally are two rainy seasonsper year, the physiologically possible.In the absenceof in- long rains in March-May and the short rains hibitory environmental conditions, timing of of October-December (Brown and Britton 1980). breeding may be influenced more by social In the Nakuru area, breeding by each popula- stimulation or the carry-over effect of timing tion of Meropsbullockoides roughly coincideswith the previousseason (Snow and Snow 1964,Fog- either one of these two rainy periods. Long- den and Fogden 1979). rainsbreeding populationsare interspersedwith White-fronted Bee-eatersinitiated a complete short-rains breeding populations, creating a molt once annually, beginning ca. 4 months spatial mosaic of the two types (Fig. 1). This after the onset of breeding. Opportunistic rec- separation into contiguous but "out-of-syn- ords of molt suggest that its timing was rela- chrony"breeding populations is unusualamong tively fixed and synchronouswithin popula- vertebrates and, to our knowledge, has been tions;short-rains breeders began molt in March- reported only for Hyraxes(class Mammalia, Or- April, while long-rains breedersdid not begin der Hyracoidea)living on isolatedkopies in the molt until July-August.Annual molt cyclesthat Serengeti plains of Tanzania (Hoeck 1982). are timed more rigidly than breeding initiation In spite of periodicity reflectingthe rainy sea- have been reported for a number of tropical sons, nesting did not appear to be proximally species(Snow and Snow 1964, Fogden 1972, tied to local environmental cues. Several lines Stiles and Wolf 1974, Dittami and Gwinner of evidencesupported this conclusion.First, nest 1985). If the molt cycle of White-fronted Bee- initiations did not correlate with any obvious eatersis of sufficientduration, each population patterns of local rainfall or insect abundance would be physiologicallyincapable of breeding (Fig. 3). Second, different long-rains (or short- in the "alternate" season. This still leaves un- rains) populations residing within a few kilo- answered the question of how certain popula- meters of one another (and therefore presum- tions come to cycle on April breedings, while ably experiencing similar environmental neighboring populationscycle on October-De- conditions) often initiated breeding weeks or cember breedings. even months apart. Third, short-rains breeders in Nakuru began nesting during October-De- SOURCES OF EGG AND NESTLING MORTALITY cember despite the absenceof the increasein rainfall typically experiencedelsewhere in the Nesting successwas generally low among the Rift Valley (Figs.2 and 3). In addition, breeding populations of White-fronted Bee-eatersat Na- was sometimes offset even within subsections kuru. Averaging the data from 14 population- of the same population. From 1980 to 1982, the seasons(including second nestings), we found Baharini population subdivided to breed at two only 41% of 531 nests from which young July1991] ReproductiveSuccess in Bee-eaters 683 fledged. Average (+SD) productivity was 0.57 Acorn Woodpeckers (56% of all nesting mor- + 0.83 fledglings per nesting attempt. Incor- tality) and that 26% of such losseswere due to porating the 42%mortality that occursbetween egg destructionby competingfemales. Among fledging and attainment of independenceat 6 anis, Vehrencamp and her co-workers reported months of age, we found only 1 nest in 4 pro- that 19% of eggslaid were lost due to egg toss- duced a surviving independent juvenile. ing (Vehrencamp et al. 1988). The two long-rainspopulations (Makalia and In the present study, 28% of eggs were lost Enderit) nestedin coloniesalong rivers and lost before hatching, half of them due to various a major proportion of nests to floods. The po- forms of socialcompetition (Table 2). The most tential for flooding may have been exacerbated common causeswere associatedwith intraspe- by deforestation of the watershed (C. van- cific nest parasitism,because both host and par- Somerenpers. comm.).However, rapid changes asite tend to remove eggsfrom parasitizednests. of water level and seasonalflooding are typical White-fronted Bee-eaters are both colonial and of mostsavannah river systems.Although such cooperative breeders. Colonial nesting facili- catastrophic loss was not characteristic of the tates opportunities for nest parasitismby un- short-rainspopulations in our study,major losses related females.Cooperative breeding provides to colony cave-ins, subterranean safari ants further opportunitiesfor kin-parasitism,in which (Doryussp.), and Horus Swiftswere occasionally the perpetratoris an integrated female member observedat the coloniesof peripheral popula- of the same clan, most frequently a daughter, tions. We do not know whether such losses are who adds to the parental clutch an egg of her a consistent,though infrequent, characteristic own that she then helps to rear (Emlen and of bee-eaternesting mortality, or are character- Wrege 1986). Becausekin-parasitism was diffi- istic only of certain populations with vulnera- cult to detect, some of the egg lossesattributed ble nesting sites. to other sourcesof mortality may have been, in In all populations,the primary causesof low actuality, due to this cause.If true, the impor- reproductivesuccess were high ratesof egg loss tance of socially induced egg losseswould be and nestling starvation. Nest predation was even greater than reported. negligible, only 2-9% of the eggs laid in any The largestsingle sourceof prefledging mor- colony. Similar low rates of egg and nestling tality was starvation. Starvation losses were predation have been reported from Red-throat- highestwhen food availability was low and the ed Bee-eatersin Nigeria (5%, Fry 1984) and Car- number of adults provisioning the nest was mine Bee-eaters(Merops nubicus) in Zimbabwe small. Further, the nestlings at greatestrisk of (11%, Emlen and Demong unpubl. data). Low starvationwere the youngestmembers of large predation ratesare typical of hole-nesting spe- broods,because hatching was asynchronous and cies in general (Nice 1957, Lack 1968, Ricklefs produced a size hierarchy among broodmates 1969), and cooperativebreeders are no excep- that enabledthe largestsiblings to monopolize tion (e.g. Koenig and Mumme 1987, Lennartz food brought to the nest.Clearly, White-fronted et al. 1987). Bee-eatersat Nakuru frequently had difficulty Nearly one-half of the egg losswas socially obtaining sufficient food to provision their induced. This is of special interest becauseit nestlings.An astonishing48% of all hatchlings documentsa costof group living. Brown (1987) starvedbefore reaching fledging age. Food stress noted that, among cooperativebreeders, such of this magnitude has not previouslybeen de- behavior has only been reported from species scribed for any other speciesof cooperatively exhibiting plural breeding (thosein which two breeding bird. or more females may breed within the same socialgroup). These are preciselythe situations ADAPTATIONS TO A VARIABLE ENVIRONMENT in which competitionamong potential breeders would be expected. Various features of avian reproductive biol- In Acorn Woodpeckers(Melanerpes formicivo- ogy have been interpreted as adaptations to rus) and Groove-billedAnis (Crotophagasulci- maximize the number of young that can be rostris),joint-nesting females routinely tossone reared under a given set of environmental con- another's eggs out of the nest (Mumme et al. ditions. Three features, in particular, involve 1983, Vehrencamp1977). Koenig and Mumme specializationsto cope with conditions of food (1987)found that egg losseswere high among shortage:adjustments of clutch size relative to 684 WREGEAND EMLEN [Auk, Vol. 108 food availability, brood reduction, and slower eggs laid. Significant year-to-year variation in growth ratesof young (e.g. Lack and Lack 1951; clutch size was attributable to factors relating Lack 1968; Ricklefs 1965, 1968, 1973, 1976; to food availability during laying. Clutch size O'Connor 1977). Two features of the nestling was not correlated with insect availability dur- food supply, its predictability at the time of egg ing the nestling phase.Regardless of what cue laying and its stability during the nestling pe- females might use at the time of laying to pre- riod, are important determinants of the most dict food resourcesfollowing hatching, the hy- appropriate responseto food stress(O'Connor pothesis of clutch-size adjustment predicts a 1977). Adjustmentsof clutchsize will be adap- correlation with future insect abundance. No tive when conditions before or at the time of such correlation was observed. egg laying accuratelypredict conditionsduring When facedwith high but variable starvation the time of provisioning nestlings(Perrins 1965, lossesacross years, bee-eaters in Nakuru showed Cody 1966). A female could then lay the number no food-relatedadjustments in clutch size. The of eggsthat she and her mate (and any helpers) reason for this presumably lies in the unpre- could raise optimally at predicted food levels. dictability of feeding conditions. At egg laying, If food levels at the time of provisioningcannot bee-eatersare unable to predict how much food be predictedduring or before egg laying, clutch will be available between hatching and fledg- size should not be adjusted.Instead, femalesare ing of young. Changesin two factors,one en- expected to lay the clutch size appropriate to vironmental and one social, might improve good years, but to have a means of adjusting feeding conditions following the completion of brood size downward if conditions deteriorate the clutch. First, large fluctuationsoccur in ac- (Ricklefs 1965, O'Connor 1977). Brood reduc- tual insect abundance (Fig. 3). Laying a rela- tion refers to the situation where nestlings de- tively fixed clutch would allow the breedersto velop a size-graded hierarchy of competitive take advantage of any improvement in food ability such that the effects of food shortage conditions that might occur; in the absenceof impact asymmetrically on the smallest (i.e. the such improvement, rapid brood reduction can youngest) member of the brood (Lack 1954, still be achieved by other means(see below). Ricklefs 1965). Under harsh conditions, such Second,the flexible helper structureof White- chicks starve. The brood size is brought into fronted Bee-eaters can increase the number of line with food supplies without seriously jeop- provisioners. The number of adults tending a ardizing the survival probabilities of the older nest often increasedas the breeding seasonpro- chicks.The third strategy,that of slowednest- gressed,mostly due to nesting failure by some ling development rates, is predicted to occur breeders who subsequently"redirected" their when conditionsduring the nestling period are effortsto ongoing nestsof closekin (Emlen and not only unpredictable but also variable over Wrege 1988). Nearly 20% of nestsgained late- relatively short time periods (Ricklefs 1976, joining helpers in this manner. The presence Howe 1976). The ability of nestlings to retard of helpers leads to an increased feeding rate further growth and to conserveenergy would and dramatically increased nestling survival enable them to withstand temporary periodsof (Emlen and Wrege 1991). The possibility of food shortage.If conditions improved with suf- gaining additional late helpers providesa social ficient speed,such nestlingscould recoverand parallel to the improvement of environmental fledge normally. The ability to alter the rate of conditions. In both cases the specter of food morphological development within the nest- stresson the nestlingsis reduced.In both cases, ling period is seen as distinct from the overall premature reduction of the initial clutch could slow development that characterizes many lead to decrease rather than an increase in the tropical frugivores and oceanic seabirds. number of young ultimately fledged. O'Connor (1977) formalized the dependence of Broodreduction.--Whereas it is relatively in- these strategieson the predictability of food expensiveto tend an extra egg throughout in- resources,expanded the set of relevant predic- cubation, the costof maintaining an extra nest- tions for each, and tested them with data from ling during its development is much higher. three representative speciesof birds. We ex- Lack (1954) was the first to suggestthat asyn- amined the three specializationsas they apply chronoushatching was an adaptation to food to White-fronted Bee-eaters. stressbecause it led to the development of sib- Adjustmentsof clutchsize.--White-fronted Bee- ling hierarchies among nestmates which, in eatersfledged only a small fraction (27%)of the turn, facilitated brood reduction to match food July1991] ReproductiveSuccess in Bee-eaters 685 availability. Asynchronoushatching may also them to survive temporary periods of food de- stagger the period when the greatest food de- privation (Emlen et al. in press).Normal (well- mands are made on adult provisioners(Hussell fed) chicksattain the prefledging stageof mor- 1972, Bryant 1978; but see Lessellsand Avery phological developmentca. 25 daysof age. But 1989). Since their inception, these ideas have others take as long as 42 days to achieve the been applied in a large number of avian studies same stage. As a result, different nestlings (see Clark and Wilson 1981). Recent models, fledged at very different ages.We argue else- however, demonstratethat predation pressure where (Emlen et al. in press)that the ability to can also lead to asynchronous hatching (e.g. slow development and withstand morpholog- Clark and Wilson 1981, Magrath 1988). These ical retardation is an important adaptation, hypothesesare not mutually exclusive,but teas- which enables nestlings to reduce energy re- ing apart their relative importance is difficult. quirements over the short term, until environ- Such a detailed examination of the selective mental or socialconditions improve. factorsunderlying asynchronoushatching was White-fronted Bee-eatersfit nicely into the beyond the scopeof this study. But tropical bee- predictive framework outlined by O'Connor eaters appear to be good candidates for Lack's (1977). Breeding females did not adjust clutch brood-reduction hypothesis.In six tropical spe- size to future feeding conditions. Instead they cies for which incubation and hatching syn- laid a modal clutch that was considerablylarger chrony data are known, all appear to begin par- than the number of young that could be reared tial incubationwith the first egg,and successive in most years. Consequently, large scale brood eggshatch at approximate 24-h intervals (Dyer reduction was commonplace. Asynchronous 1979, Fry 1984, this study). Predation rates are hatching of the clutch and rapid development extremely low in all species studied and are of a size hierarchy among nestmatesfacilitated similar on both eggsand nestlings (3% each for a patterned reduction of broods such that the White-fronted Bee-eaters;see also Dyer 1979, youngest was the first to starve. Committing Fry 1984). Yet partial brood loss(i.e. presumed available resourcesto chicks preferentially by starvation) claims the lives of 48% of all hatch- age results in the least amount of energy lost ling White-frontedBee-eaters, 28% of hatchling with the death of eachsubsequent nestling, and Carmine Bee-eaters,and 17%of hatchling Red- increasesthe probability of survival for at least throated Bee-eaters(Dyer 1979,this study). The someof the brood (Lack 1954).Finally, nestlings European Bee-eater(Merops apiaster), a Palearc- exhibit an ability to slow development during tic breeder, has been studied extensively in periodsof food shortage.If foraging conditions southern Franceand alsoexhibits asynchronous improve, or if additional adults join as helpers hatching, with 21-30% of hatchlings lost to star- at the nest, retarded nestlingscan recoverand vation (Lessellsand Avery 1989). fledge successfully. Whatever the selective pressuresthat main- This suite of featuresmatches those hypoth- tain asynchronoushatching in bee-eaters,one esized to be adaptive for avian speciesin areas effect was that nestlings were quickly sorted where food levels fluctuateunpredictably dur- into a hierarchy of size and development, such ing the breeding season,producing frequent that the oldest (i.e. largest) could monopolize but temporary times of food shortage (Howe availablefood when it waslimiting. In our study, 1976, Ricklefs 1976, O'Connor 1977). These are all starvationlosses in broodswith • 1 nestling precisely the conditions in the Rift Valley area actedfirst on the smallestnestling, and the like- of East Africa. lihood of any given nestling starving increased dramatically with the number of older siblings ACKNOWLEDGMENTS in the nest (Fig. 6). Dyer (1983) also reported that 95% of brood reduction in Red-throated We thank the Governmentof Kenya, the National Bee-eaters and 100% in Carmine Bee-eaters were Museumsof Kenya, the wardens and staff of Lake of the youngestnestling, and that most reduc- Nakuru National Park for their support of this re- searchthroughout its tenurein Kenya.For assistance tion occurredin largebroods. Lessells and Avery in the field, we thank N. Demong, D. Emlen, G. Far- (1989) describea similar increasedprobability ley, R. Hegner, P. Kabochi,E. Kellogg,M. Kinnaird, that the youngestnestlings die first in European M. Kipellian, B. Lyon, C. Miller, S. Moore, M. Read, Bee-eaters. H.-U. Reyer, G. Schwan, T. Schwan, D. Schmidl, G. White-fronted Bee-eaternestlings also show Tabor, and R. Wagner. We also thank W. D. Koenig, highly variable development rates that enable C. M. Lessells,R. L. Mumme, and R. J. O'Connor for 686 w•eu AND E/a•N [Auk, Vol. 108

helpful reviewsof the manuscript.This researchwas ing decisionsamong White-fronted Bee-eaters. supportedby John Simon GuggenheimFoundationß Behav. Ecol. Sociobiol. 23: 305-315. the Chapman Fund of the American Museum of Nat- ß& . 1991. Breeding biology of White- ural Historyß grants 2778-84 and 3025-85 from the fronted Bee-eaters at Nakuru: the influence of National GeographicSociety, and BNS-7924436,BNS- helperson breederfitness. J. Anita. Ecol.60: 309- 8216776, and BNS-8517725 from the National Science 326. Foundation to Emlen. ß --, N.J. DEMONG,& R. E. HEGNER. 1991. Flexiblegrowth ratesin nestlingWhite-fronted LITERATURE CITED Bee-eaters:a possibleadaptation to short-term food shortage.Condor 93. In press. BMDP. 1983. BMDP statisticalsoftware. Berkeleyß FOGDEN,M.P. L. 1972. The seasonalityand popu- Univ. California Press. lation dynamicsof equatorialforest birds in Sa- BROWNßJ.L. 1987. Helping and communalbreeding rawak. Ibis 114: 307-343. in birds. Princetonß Princeton Univ. Press. ß & P.M. FOX;DEN. 1979. The role of fat and BROWN,L. H.ß & P. L. BRITTON.1980. The breeding protein reservesin the annual cycleof the Gray- season of East African birds. Nairobi, East Africa backed Camaroptera in Uganda. J. Zool. London Nat. Hist. Soc. 189: 233-258. BRYANT,D.M. 1978. Establishmentof weight hier- FRY,C.H. 1972. Thesocialorganizationofbee-eaters archies in the broods of House Martins Delichon (Meropidae) and co-operativebreeding in hot- urbica. Ibis 120: 16-26. climate birds. Ibis 114: 1-14. CLAm<ßA. B.ß& D. S. WILSON. 1981. Arian breeding 1984. The Bee-eaters. Verraillion, Buteo adaptations:hatching asynchronyß brood reduc- Books. tion, and nest failure. Q. Rev. Biol. 56: 253-277. HœGNER,R. E. 1981. Territoriality, foraging behav- COPYßM. L. 1966. A general theory of clutch size. ior, and breedingenergetics of the White-fronted Evolution 20: 174-184. Bee-eater in Kenya. Ph.D. dissertationßIthaca, --. 1971. Ecologicalaspects of reproduction.Pp. New YorkßCornell University. 461-512 in Arian biologyßvol. I (D. S. FarnetßJ. HOECK,H. N. 1982. Populationdynamicsß dispersal R. Kingßand K. C. ParkesßEds.). New YorkßAc- and geneticisolation in two speciesof hyrax(Het- ademic Press. erohyraxbrucei and Procaviajohnstoni) on habitat DEMONG,N.J., & S. T. EMLEN. 1975. An optical scope islandsin the Serengeti.Z. Tierpsychol.59: 177- for examining nest contents of tunnel-nesting 210. birds. Wilson Bull. 87: 550-551. HowE, H. F. 1976. Egg sizeßhatching asynchrony, DI•"'rAMI,J.P., & E. GWINNER.1985. Annual cycles sex, and brood reduction in the Common Grack- in the African Stonechat and their relationship le. Ecology57: 1195-1207. to environmental factors.J. Zool. (London) 207: HUSSœLL,D. J.T. 1972. Factorsaffecting clutch size 357-370. in arcticpasserines. Ecol. Monogr. 42: 317-364. D•R, M. 1979. The adaptive significanceof co-op- KLOMP, H. 1970. The determination of clutch size erative breeding in the Red-throatedBee-eater in birds. Ardea 58: 1-124. Merops bulocki(Vieillot) and other bee-eaters KOENIG,W. D., & R. L. MUMME. 1987. Population (Aves:Meropidae). Ph.D. dissertationßAberdeenß ecology•t• the cooperativelybreeding Acorn Univ. Aberdeen. Woodpecker.Monogr. Popul. Biol., no. 24 (R. M. 1983. Effect of nest helpers on growth of May, Ed.). PrincetonßPrinceton Univ. Press. Red-throated Bee-eaters. Ostrich 54: 43-46. LACKßD. 1954. The natural regulation of animal EMLœN,S.T. 1981. Altruism, kinship, and reciprocity numbers. Oxford, Clarendon Press. in the White-fronted Bee-eater.Pp. 217-230 in 1968. Ecologicaladaptations for breedingin Natural selection and social behavior (R. D. Al- birds. Londonß Methuen. exander, and D. Tinkleß Eds.). Londonß Chiron ß & E. LACK. 1951. The breeding biology of Press. the swift Apusapus. Ibis 93: 501-546. ß 1990. The White-fronted Bee-eater:helping LENNARTZ,M. R., R. G. HOOPER,& R. F. HARLOW. 1987. in a colonially nesting species.Pp. 489-526 in Socialityand cooperativebreeding in Red-cock- Cooperativebreeding in birds:long-term studies aded WoodpeckersßPicoides borealis. Behav. Ecol. of ecologyand behavior(P. Staceyand W. Koe- SociobioL 20: 77-88. nig, Eds.).Cambridge, Cambridge Univ. Press. LESSœ•KS,C. M., & M. I. AVERY.1989. Hatching asyn- ß& P. H. WREGœ.1986. Forcedcopulations and chrony in EuropeanBee-eaters Merops apiaster. J. intra-specificparasitism: two costsof socialliving Anim. Ecol. 58: 815-835. in the White-fronted Bee-eater. Ethology 72: 2- MAGRATI-I,R. D. 1988. Hatching asynchronyin al- 29. tricial birds: nest failure and adult survival. Am. , & . 1988. The role of kinship in help- Nat. 131: 893-900. July1991] ReproductiveSuccess in Bee-eaters 687

MALAISE,R. 1937. A new insecttrap. Entomol. Tidskr. 1976. Growth rates of birds in the humid 58: 148-160. new world tropics. Ibis. 118: 179-207. MARSTON, N. 1965. Recent modifications in the de- SAS INSTITtrrE,INC. 1985. SAS usersguide: statisticsß signof Malaisetraps, with a summaryof the in- version 5.0. Cary, North Carolina, SAS Institute. sectsrepresented in the collections. J. Kansas En- SINCLAIR,A. R. E. 1978. Factorsaffecting the food tomol. Soc. 38: 154-162. supplyand breedingseason of residentbirds and Mt•MME, R. L., W. D. KOFa•IG,& F. A. PrrEL•C^.1983. movementsof Palaearcticmigrants in a tropical Reproductivecompetition in the communalAcorn African savannah. Ibis 120: 480-498. Woodpecker:sisters destroy each other's eggs. SNow, D. W., & B. K. SNow. 1964. Breedingseasons Nature 306: 583-584. and annual cyclesof Trinidad land-birds.Zoolo- NICE,M. M. 1957. Nesting successin altricial birds. gica 49: 1-39. Auk 74: 305-321. STILESßF.G. 1980. The annual cycle in a tropical wet O'CONNOR,R.J. 1977. Growthstrategies in nestling foresthummingbird community. Ibis 122: 322- passerines.Living Bird 16: 209-238. 343. --. 1984. The growth and developmentof birds. ß & L. L. WOLF. 1974. A possible circannual New York, Wiley & Sons. molt rhythm in a tropicalhummingbird. Am. Nat. PERRINS,C. M. 1965. Population fluctuationsand 108: 341-354. clutchsize in the GreatTit, Parusmajor L. J.Anim. VEHRENCAMP,$. L. 1977. Relativefecundity and pa- Ecol. 34: 601-647. rental effort in communallynesting anis, Crotoph- RIC•CL•S, R.E. 1965. Brood reduction in the Curve- aga sulcirostris.Science 197: 403-405. billed Thrasher. Condor 67: 505-510. , R. R. KOFORD, & B. S. BOWEN. 1988. The effect 1968. Patternsof growth in birds. Ibis 110: of breeding-unit size on fitnesscomponents in 419-451. Groove-billedAnis. Pp. 291-304 in Reproductive 1969. An analysisof nestingmortality in success: studies of individual variation in con- birds. Smithsonian Contrib. Zool. 9: 1-48. trastingbreeding systems(T. H. Clutton-Brock, ß 1973. Patternsof growth in birds. II. Growth Ed.). Chicago,Univ. ChicagoPress. rate and mode of development.Ibis 115:177-201.