FORAGING BEHAVIOR, REPRODUCTIVE SUCCESS, AND COLONIAL NESTING IN OSPREYS

JOHN M. HAGAN lip AND JEFFREYR. WALTERS Departmentof Zoology,Box 7617, North CarolinaState University, Raleigh,North Carolina27695-7617 USA

ABSTRACT.--Wetested several hypotheses about the evolution of colonial nesting in Os- preys.We examined foraging behavior and reproductive success of membersof a densecolony in coastalNorth Carolinato test(I) if the colonyfunctioned as a centralplace for exchanging information about the location of food patches(Information Center Hypothesis), (2) if the colonywas locatedcentrally relative to foragingsites, which would minimize foragingtravel costs,(3) if the colony functioned to promote synchronizedbreeding, which would swamp predatorswith more young than they could consume,(4) if nestingin the centerof a group decreasedpredation pressure,and (5) if safe nest-sitedistribution controlsactual breeding distribution.The hypothesisthat the colony functionedas an information center was sup- portedby synchronyin departure,but contradictedby evidenceof fidelity to foragingareas. The colony did not function as the geometric center of an individual's foraging locations. Colony memberstraveled 1.5-7 timesas far as they would have traveled had they nestedat their geometric centers.There was a negative correlation between the proportion of eggs lost in the colony and the number of eggsavailable to predators,suggesting some benefits to synchronousnesting in a dose spatialgroup. However, there was no numericalresponse in predation relative to the total number of eggsavailable in the colony.Birds nesting toward the centerof the colonyproduced significantly more fledglingsthan peripheral nesters,but strong nest-sitefidelity prevented shifting to more central locationsby peripheral nesters. Spatialdistribution of nestsreflected the distribution of potential nest trees,both within the lake and within the region. This and the plasticity of nesting density shown in Ospreys supportLack's (1968) Nest-siteHypothesis. We concludethat the spatialdistribution of safe nest sitesand advantagesrelated to predation maintain coloniality in Ospreys.Received 22 May 1989,accepted 15 January1990.

COLONIALnesting in birds is a common yet 1968,Lack 1968). The firsttwo explanationspos- poorly understoodphenomenon. A colonymay tulate that individualsnesting near othersde- be defined as a group of animals that nest at a rive a higher fitness as a result of interaction. centralized location, from which they recur- The third explanation postulatesno such ad- rently depart in search of food (Wittenberger vantage. For most speciesit is not clear whether and Hunt 1985). Members of a nesting colony nestingnear othersis beneficialor is a by-prod- of birds are constrainedby the immobility of uct of a habitat constraint. their clutch or brood. The costsof traveling to The Information Center Hypothesis (Ward and from such a central location are obvious, andZahavi 1973) relates colonial nesting to for- but the benefitsare not. Explanationsof colonial aging success.Individuals may learn about the breeding fall into three categories(Alexander locationof patchilydistributed, ephemeral foods 1974).Colonial breeding may enhanceforaging from other colonymembers. Information need (Crook 1965, Emlen 1971, Fisher 1954, Lack 1968, not be exchangedactively. Simply following or Ward and Zahavi 1973), reduce the probability cuing in on the return direction of an individ- of predation (Burtonand Thurston 1959,Kruuk ual known to have been a successfulforager 1964, Horn 1968, Burger 1974, Hoogland and qualifiesas "exchange."Ward and Zahavi's In- Sherman 1976), or result from resource con- formationCenter Hypothesis has become widely straints(e.g. nestsites, food distribution) (Horn cited (ISI 1983), but clear evidence that infor- mationtransfer induces colonial nesting is lack- ing. It is a difficult hypothesisto test and in- ' Presentaddress: Mahomet Bird Observatory, P.O. volves distinguishing between successfuland Box 936, Mahomet, Massachusetts 02345 USA. unsuccessfulforagers, and then demonstrating 506 The Auk 107: 506-521. July 1990 July1990] ColonialNesting inOspreys 507 that unsuccessfulforagers improve their suc- similarlyenhanced by groupingamong nesting cessby obtaininginformation on food location birds. Nests in the center of colonies suffer less from successfulbirds (Mock et al. 1988). predationthan thoseon the peripheryin many Nevertheless, evidence that information ex- species,which is consistentwith the proposed changeseems to occurin colonies(or roosts) effectsof groupmobbing or predatordetection has recently been obtained for three avian (e.g.Patterson 1965, Tenaza 1971, Siegfried 1972, species(Brown 1986, Rabenold 1987, Greene Feare1976, Siegel-Causey and Hunt 1981). 1987). The role of the information exchangein Synchronousbreeding of an avian assem- the evolution of colonies remains debatable, blagecan also produce benefits to members.By however. Information centers could be wide- swampingpotential predators with a synchro- spreadphenomena, critical to the evolutionof nized productionof young, colonymembers coloniality,or they couldsimply be a secondary can reducethe probability that their offspring adaptationthat sometimesoccurs as a result of will be preyed upon (Patterson1965, Nisbet group living (colonialityhaving evolved for 1975, Feare 1976, Veen 1977). Emlen and De- other reasons). mong(1975) suggested that synchronous breed- To addressthis question,we examinedanother ing in BankSwallow (Riparia riparia) colonies colonyof a speciesthat hasalready been shown increased opportunities for social foraging, to exchangeinformation. Demonstration of in- therebyincreasing foraging efficiency. We tested formation exchange at an independent site this hypothesisby relating breedingchronol- would imply that it can be a widespreadphe- ogyto reproductivesuccess in the colony. nomenon, and therefore may be an important Finally,we testedtwo hypothesesrelated to factor leading to, or at leastmaintaining, group resource constraints'. Horn's Geometric Center living. Its relative importanceto colonymain- Hypothesisholds that colonyformation could tenance should be correlated to its common- be due to food distribution (Horn 1968). By ness. nestingin the centerof the distributionof a We testedthe Information Center Hypothesis patchy,ephemeral food supply,birds can min- for an Osprey(Pandion haliaetus) colony in coast- imize overall foraging distancetraveled. No al North Carolina. This site is distant from the benefit of socialityis implied; rather, foraging colonyin NovaScotia where Greene (1987) doc- considerationsthat act on many individuals in umentedinformation exchange, and the strong an identicalway leadto colonyformation. The natal-sitefidelity of Ospreysmakes it extremely evidenceto supportthis hypothesiscomes from unlikelythat individuals of thesecolonies would Horn's(1968) study of Brewer'sBlackbird (Eu- mix (Poole 1989: 137). Male Ospreys supply phaguscyanocephalus). nearly all of the food for the female and nest- The Nest-siteHypothesis (Lack 1968)holds lings.Males in the North Carolinacolony must that colonialityis relatedto the distributionof travel long distancesto any of severalforaging safe nest sites. If safe nest sites are clumped, locations, and exploit schoolsof marine fish, nestswill also be clumped. For those colonial which fluctuatespatially and temporally. The speciesthat feed solitarily, Lack (1968) main- colony is large (50-60 pairs) and dense,and tainedthat safenesting sites--rather than feed- returning foragersare readily visible. If infor- ing habitats--arethe criticaldeterminants of mationexchange is a prominentfeature of group nestingdistribution. It has never been dem- living in Ospreys,it shouldbe evident at this onstrated that nest-site limitation alone can in- site, where conditionstheoretically conducive duce coloniality. to informationexchange are nearly ideal. We used a comprehensiveapproach to the We testedother hypothesesrelated to pre- studyof colonyformation and functionin Os- dation reduction as a result of group living in preys,recognizing that many factorsmay op- Ospreysby examinationof spatialand temporal erate to maintain colonial nesting. Although patternsof reproductivesuccess in the colony. not commonlyconsidered a coloniallynesting Detectionof predatorsis improved by flocking species,Ospreys historically nested in colonies. (Powell 1974, Siegfried and Underhill 1975, Coloniesbecame rare along the U.S. Atlantic Kenward 1978, Lazarus 1979, Caraco et al. 1980, seaboard in the 1950s and 1960s as a result of Barnard 1980, Bertram 1980, Jennings and Ev- extensiveuse of DDT and its effectson eggshells ans 1980,Thompson and Barnard1983), and it (Amesand Mersereau1964, Ames 1966, Hickey seemslikely that detectingpredators would be and Anderson 1968). Before this time, colonies 508 HAGANAND WArTEES [Auk, Vol. 107

noon sampleswere followed by subsequentmorning samplesto obtain this equivalent of a full day of sam- pling of a pair. We sampled three pairs twice (two daysfor eachpair), but at differenttimes during the breeding cycle. We observedpairs for a total of 170h, and assigned behaviorsonly to broad categoriesbecause we were interestedin qualitativepatterns in foraging by the males,which provide most of the food for their mates and young until fledging.The categoriesincluded (1) perched over nest, (2) perched on nest (includes in- cubation),(3) perchedin territory (not nest tree), (4) away from territory, (5) flying within territory, (6) fishingin lake, and (7) other. Category4 includedall foragingaway from the lake. Becauseforaging males rangedfar, we couldnot subdividethis categoryfur- ther. Foragingpatterns.--During the incubation and nest- ling stagesof the breedingseason (April throughJune) in 1984and 1985,we observedforaging patternsfrom a blind 12 m high, constructedof scaffolding,near the center of the lake. Arrivals, departures,and the foragingsuccess of maleswere recordedevery 15 min Fig. 1. Lake Ellis Simon and vicinity. from sunrise to sunset, for each nest that could be seen from the tower. Sampling was done in two full were common. One on Gardiner's Island, New (dawn to dusk) consecutiveday sessions.Intervals of this span were unlikely to result in missedarrivals York, supportedapproximately 300 nestingpairs and departuresbecause males typically spent 30 min of Ospreys (Puleston 1977). Ospreys are re- to severalhours perchedin the nest territory after an markably flexible in their nestingdensity, from arrival, and severalhours away after a foraging de- remoteand solitarynesting to coloniesin which parture. Although we could monitor 15-20 nestsfor active nests are only 20 m apart (Poole 1989). fish deliveries with an 83 x spotting scope,we could determine male presence/absencefor only 10 nests STUDY AREA AND METHODS in 1984, and 11 nests in 1985. Other nest territories had visual obstructions between the tower and the We studiedan Ospreycolony locatedat Lake Ellis male's perch tree. Simon in coastal North Carolina (Craven County, In 1985 we attached transmitters to 16 males so that 35ø50'40"N,76ø59'06"W) (Fig. 1) from March through vanishing and returning directionscould be observed August, 1983 to 1985. This 600-ha, shallow, marshy accurately.Only 8 provided consistentdata; the other lake containshundreds of live cypresstrees over water, 8 malesbit off the antennas,which drasticallyreduced in which 50-60 pairs of Osprey's nested each year. signal range. We attachedtransmitters of 2-3% of the The density and number of nesting pairs has in- male'sbody massto the baseof the two centralretrices creasedsteadily from the 17 nestsin 1967(Henny and with hot-melt glue. The height of the transmitter on Noltemeier 1975). a flying Osprey and of the 14-elementreceiving an- Activity budgets.--From15 May to 9 August in the tenna atop the blind producedsignal rangesof 6-14 1983 breeding season,we sampled activity budgets km. We deduced foraging location from vanishing of 23 nesting pairs. We restricted observationspri- and return directions, and used these azimuths to marily to the nestlingand postfledglingstages of the compute geometriccenters of foraging. We deter- breedingcycle. From the pool of breedingbirds, we mined optimum nesting locationsfor eachindividual randomlychose pairs for observation,excluding those by calculatingthe geographicpoint where foraging whose nestshad already failed. We took observations travel distance,based on telemetry data, would have in 0.5-day samples(ca. 7 h duration, depending on been minimized. day length), eachof which involved focal samplesof We testedthe Information Center Hypothesisby a single pair, unlesstwo nestswere closeenough that comparing probabilities of departures of successful adultsof both pairscould be watchedsimultaneously. and unsuccessfulforagers. When male Ospreys re- We observed birds from a portable blind on stilts, turned to their nestsafter being away for a minimum located ca. 100 m from the nest, and we completed of 45 min (3 sample intervals), we classifiedthem as the equivalent of one day of observation(sunrise to being "successful"or "unsuccessful"foragers based sunset) before we moved to a different nest. After- on whether or not they returned with a fish. We chose July1990] ColonialNesting in Ospreys 509

45 min as the criterion because observation indicated the first day of a 2-day sampling sessioncould not be that a foray outside the lake could not be accom- classified as successful or unsuccessful. This biased plished in less time. the analysistoward synchronousdepartures because Greene (1987) found that unsuccessful birds cued someintervals in which no departureswere recorded on the arrival of a successful bird. We examined the actuallyincluded departures, but of unclassifiedbirds. distributions of departures of unsuccessful birds Therefore, we repeated all analysesusing only data (DEPv) and departures of successfulbirds (DEPs) rel- from the secondof two consecutivedays of sampling. ative to the distribution of arrivals of successful birds On the second day, all initial departures could be (ARRs) among 15-min sampling intervals. If depar- classifiedbased on successof the final foray on the tures (or arrivals) were independent, their distribu- previous day. tion among intervalsshould fit a Poissondistribution Temporaland spatial patterns of fiedging success.--From whose mean (œ)equals the number of departuresdi- 1983 to 1985, we checked all nests in the colony at vided by the number of intervals. If departures of weekly intervalsfor eggs,nestlings, or fledglings(from unsuccessfulbirds were triggered by arrivals of suc- beforebreeding began to after all young had fledged). cessfulbirds, but successfulbirds departed indepen- All nests were less than 9 m above the water and dently of arrivals, then the following should hold: could be checked from a boat with a mirror attached to the end of a telescopingaluminum pole. Checking 1. The distributionsof DEPv among intervalswith at all nests required I-3 h (depending on water level least one ARRsand among intervals with no ARRs and weather conditions). Parent Ospreys were dis- should both fit a Poisson distribution, but œ will placed from their nests for only I-2 min during a be greater for the former distribution. check. 2. The distributionsof DEPsamong intervals with at We designatedall active nestsas "central" or "pe- leastone ARRsand among intervals with no ARRs ripheral" according to their location within the col- should both fit a Poisson distribution, and œshould ony. We considereda nest peripheral if no other ac- be the same for both distributions. tive nests stood within 200 m of either side of a line 3. The distributionsof DEPv among intervalswith at connecting the nest to the nearest point on shore. least one unsuccessfularrival (ARRv) and among We plotted the locationsof all nestsand potential intervals with no ARRv should fit a Poisson dis- nest sites on aerial photographs. Potential nest sites tribution, and œ should be the same for both dis- were identified subjectivelyas cypresstrees that ap- tributions. Differencesin œwere evaluated using peared to be physically able to support a nest. To t-tests. examine whether Osprey nestswere distributed in a Unsuccessfulbirds might follow successfulbirds pattern similar to available nest sites, we compared when they leave the colony rather than depart in Morisita's index of dispersion (Morisita I959) for ac- responseto the arrival of a successfulbird. If this was tual nestswith that for potential nest sites. the mechanism of information exchange, then the For each clutch, we recorded the week that eggs following predictionsshould hold: first appeared in the nest relative to the week that eggsfirst appearedin the colony that year (i.e. week 1. The distributionsof DEPvamong intervals in which of laying, where week of laying of first nesting at- at least one DEPs and among intervals with no tempt = 1). Laying was sufficientlyasynchronous that DEPs should both fit a Poisson distribution, but œ we could employ regressiontechniques to examine will be greater for the former distribution, and the relation between reproductivesuccess and week 2. the distribution of DEPs among intervals with at of laying. We usedthe mean reproductivesuccess for leastone DEPv and among intervalswith no DEPv should both fit a Poisson distribution. all pairs laying in a given week in the regression analysis (GLM procedure in SAS [SAS 1985]). So as In the caseof joint departures, it is not possibleto not to give undue weight to weeks with low sample determine whether DEPsdepends on DEPv or DEPv sizes,we weighted the regressionsby weekly sample dependson DEPs,only that they are not independent. sizes (number of pairs that laid in a particular week We tested for independenceof distributions in two of laying) using the GLM WEIGHT statement.To ex- ways. First, we performed standardChi-square tests amine advantagesof nesting synchronously,we cor- of independence.Second, to better illustrate where related the number of eggs in the colony during a deviations from independence occurred, we deter- given week of the breeding seasonto the number of mined an expected joint distribution based on the eggs lost during that week. assumptionsthat each distribution was Poisson,and that the two distributionswere independent.We tested for deviationsfrom this predicted distribution using RESULTS the Chi-square analysis. These analyseswere per- formed for the joint distributionsof DEPv and DEPs, Activitybudgets.--Male Ospreys spent 54%of DEPv and ARRs, DEPv and ARRu, and DEPs and ARRs. their time away from the nest territory during Birds that departed the colony at the beginning of the nestlingphase of the breedingseason (Table • l 0 HAGANAND WALTERS [Auk, Vol. 107

TABLE1. Hours spent in variousactivities and associatedpercentages of total time observed.Data are pooled from 23 different breeding pairs observedin the pre- and post-fledglingportion of the breeding season.

Breedingmales Breedingfemales Prefiedgling Postfledgling Prefledgling Postfledgling Activity Total % Total % Total % Total % Away from territory 61.76 53.9 24.74 40.1 13.44 13.5 17.41 22.8 Fishing in lake 0.10 0.1 0.02 0.0 0.00 0.0 0.00 0.0 Flying in territory 1.00 0.9 2.07 3.4 1.82 1.8 3.40 4.5 Perched on nest 5.52 4.8 0.07 0.1 51.53 51.8 14.48 19.0 Perched over nest 2.44 2.1 0.28 0.4 16.97 17.1 13.25 17.4 Perched on tree 42.65 37.2 34.39 55.7 15.27 15.4 27.43 36.0 Other 1.17 1.0 0.12 0.2 0.39 0.4 0.22 0.3

1). They rarely foraged in the lake itself, but significantlyhigher after fish delivery than be- instead foraged in 1 of 3 estuaries,each ca. 14 fore (Table 2). These data indicate that conspe- km away (Fig. 1). If weather conditions were cific intrusions were related to food deliveries. suitable, males kettled in thermal air currents Foragingpatterns.--In 1984and 1985,we spent over the lake, and then glided toward the for- 28 days (14 two-day sessions)in the tower sam- aging area of choice. When thermals were not pling presence/absenceand foraging success available, they used powered flight. Rarely, data. For 144 foraysin which we obtained both males flew to an open section of the lake, for- departure and arrival directions, the difference aged there for a few minutes, and then pro- in direction averaged 15.2 + 15.1ø. When we ceeded to the estuaries. misseddeparture directions,we used return di- Males typically made 1-3 forays during the rections to indicate foraging location. Teleme- half-day samples.They tended to make only 1 try produced 211 directions of forays on 8 in- trip per sample after the young fledged, which dividuals. We recorded 68 directions for other reduced the amount of time spent away from birds before their antennas were detached (a the lake and increasedtime spent perched on total of 279 observations). a tree in the nestterritory (Table 1). Of 27 forays Fish deliveries were not spread uniformly that began and ended within the half day sam- over the day (Fig. 2). Instead, there were four ple, the mean time away was 2.25 + 0.98 h (œ peaksin deliveriesseparated by 2-3 h intervals. + SD; range = 1.15-5.33 h). Most forays lasted This pattern probably resulted from the mean between 1.5 and 2.5 hours. When a male re- foray length of 2.46 h (SD = 1.49, n = 659 for turned to the colony, he typically delivered the tower data), and the fact that dawn imposed an fish to the female at the nest; he then perched initial foraging synchrony each day. By pro- on a nearby tree, where he usually remained jecting these departure directions to the for- until the next foray. aging areas,we determined that Ospreysused Becauseof the nesting density, territorial in- trusions by other Ospreys were common. In- TABLE2. Rates of conspecificintrusion before and trusions within 100-150 m of the nest rarely after fish delivery. Ratesrepresent the mean for 22 elicited a chase,but nearly always provoked a fish deliveries. characteristic vocal response by the resident. During the 170 h of observation,we recorded Conspecificintrusion rate 761 vocalizations resulting from territorial in- (per hour) trusions (4.48 per hour). Often, intrusions ap- Time span After Before peared to be not agonisticbut rather incidental (min) delivery delivery pa (as a result of normal movements). To deter- 10 5.45 1.68 0.012 mine if these intrusions were related to fish 20 5.27 2.52 0.029 30 5.38 2.41 0.002 deliveries by the males, we calculated an intru- 40 5.18 2.24 0.001 sion rate for time intervals after a fish was de- 50 5.03 2.07 0.001 livered to the nest, and compared it with intru- 60 4.97 2.13 0.001 sion rates for an equal interval before fish ' Significancelevel for differencesbetween rates for Wilcoxontwo- delivery. For each period, intrusion rates were sampletest. July1990] ColonialNesting in Ospreys 511

3O port the GeometricCenter Hypothesis(Fig. 4). Individual maleshad strongpreferences for one or the other of the two broad directions avail- able for foraging. Individuals that foragedin a northeasterlydirection (Neuse River) rarely or neverdeparted toward Bogue Sound/White Oak River (southwest), and vice versa. We calculatedwhere each of the eight indi- viduals should have placed their nests,accord- ing to the GeometricCenter Hypothesis,to min- Time of Delive• imize foraging costs(Fig. 5). For each bird, the Fig. 2. The frequenciesof fish deliveriesat 15-min optimum locationwas close to the foragingareas intervalsthroughout the day(Eastern Daylight Time). used,rather than the Lake Ellis Simon colony. If optimum points had been chosen,minimum the White Oak River, BogueSound, and the round-trip distancesof forays would have av- NeuseRiver intensivelyas foragingsites (Fig. eraged3.80-21.84 km. Observedminimum mean 3). distancesranged from 27.56-33.96km. The geo- GeometricCenter Hypothesis test.--Lake Ellis metric center for all eight radio-equippedin- Simon is geographicallycentral relative to for- dividuals combined occurredjust south-south- aging areasused by the colonyas a whole. How- east of Lake Ellis Simon (Fig. 5). Only 20 ever, for this site to function as a geometric randomly selectedazimuths from each individ- center, each individual would have to exploit ual were used in this calculation so that the all three of the major foraging areasequally. value would be equally dependenton all eight. Preferencefor a given area would causea shift InformationCenter Hypothesis test.--In the 2 yr in the optimum location for placementof the of study, we recorded 402 successfulbird for- nest away from the lake and toward that for- aging departuresand 338 unsuccessfulbird de- agingarea. Directions of departurefor the eight partures. The mean duration of absence from transmitter-equippedindividuals do not sup- the territory for trips scoredas unsuccessfulre-

N

AO

Fig. 3. The intensityof foraginguse of the estuariesand riversin the vicinityof LakeEllis Simon (LES). Peaksrepresent usage intensity, based on departureand arrivalazimuths obtained at the colonyon all individuals.Dark areas are identified as NR (NeuseRiver), BS (Bogue Sound), AO (AtlanticOcean), and WOR (White Oak River). 512 HAGANAND WALTERS [Auk,Vol. 107

w

w

Fig. 5. The geometricoptimum nesting locations based on known foraging locations of transmitter- equippedmales. Dots representgeometric centers for malesindividually (associatednumber indicatesmale ID), and the enclosed star indicates the geometric center for all individuals combined (GC).

was not higher in intervals with at least one ARRs than in intervals with no ARRs. In fact, the trend was in the opposite direction, al- though the difference was not significant. The distribution of DEPswas similarly related to the distribution of ARRs, as was DEPu to ARRv. The Fig. 4. Frequency of various foraging directions for eachmale providing consistenttelemetry data (in- distributions of DEPu and ARR s,DEP s and ARRs, dividual bird ID lower right of circle). The numbers and DEPv and ARRv were all independent by below the north marker indicate the frequency rep- Chi-squareanalysis (P > 0.05). However, only resentedby the outermostconcentric circle. The inner the joint distribution of DEPs and ARRs was circle representshalf that value. accuratelypredicted by assumingindependent Poissondistributions (Chi-square, P < 0.05 for DEPu and ARRs, DEPv and ARRu). This was due primarily to a tendency for intervals with mul- turns (2.39 h, SD = 1.45, n = 219) was very tiple DEPv to have no other departures, and similar to that of successfulreturns (2.50 + 1.50 intervals with multiple departuresto have no h, n = 443), which indicates that these absences DEPv. Overall, the analyses indicate that de- were also related to foraging behavior. Based partures tend to be asynchronous relative to on these sample sizes, foraging Ospreys re- arrivals rather than synchronousas predicted, turned to the colonywithout a fish 33.1%of the and that relationshipis unaffectedby the status time. (successfulor unsuccessful)of either departing Departuresof unsuccessfulbirds clearly were birds or arriving birds. The only change that unrelated to arrivals of successful birds. Results resulted from including all the data was that of analysesof seconddays' data were contrary for DEPu, the mean was much greater in inter- to predictions (Table 3). For DEPu, the mean vals with at least one ARRv than in intervals July1990] ColonialNesting in Ospreys 513

TABLE3. Summaryof analysesto test the Informa- That DEPu and DEPs were not independent tion Center Hypothesis. Second day's data only. wasconfirmed by Chi-squareanalysis (P < 0.01), DEPv= unsuccessfulbird departure, DEPs = suc- and by our inability to predict their joint dis- cessful bird departure, ARRs = successfulbird ar- rival, ARRv = unsuccessful bird arrival. "Mean" is tribution based on assumedindependent Pois- the mean number of indicated events per 15-min son distributions (P < 0.01). Both intervals with sampling period. multiple departures (especially intervals with 3 or more departures) and intervals with no Distribution Fits Poisson a Mean b departures were much more common than ex- DEPu no*** 0.203 pected.If all data are included in the analysis, DEP s no** 0.242 DEPsno longer hasa Poissondistribution, either ARRs no** 0.348 in intervals with at least one DEPu or intervals ARRu yes 0.285 DEPu [ DEPs = 0 no* * 0.225 with no DEPu (P < 0.05 in both cases). Other- DEPu I DEPs -> 1 no** 0.377A wise results were unaffectedby including the DEPslDEPu = 0 yes 0.311 first day's data. DEPsI DEPu -> 1 yes 0.442 DEPu I ARRs = 0 no** 0.289 Temporalpatterns of reproductivesuccess.--In DEPu[ARRs >- 1 no*** 0.206B the three years of study 130 nesting attempts DEPsI ARRs = 0 yes 0.366 produced at least one egg. We excluded from DEPslARRs-> 1 yes 0.263B regression analysis five renesting attempts DEPuI ARRu = 0 no** * 0.283 which producedno fledglings.There was a sig- DEPuIARRu -> I no*** 0.212B nificant differencebetween yearsin clutch size ' Tested by Chi-square, ** = P < 0.01, *** = P < 0.001. and number of fledglings, but not brood size bFor pairs of meanscompared using t-tests,"A" indicatesthose that differed (P <: 0.01), and "B" indicates those that did not differ (P > (Table 4). The lack of difference in brood size 0.05). (at hatching) is likely due to a large variance resulting from including brood sizesof 0 (nests with no ARRu. This, too, is contrary to the pre- where no eggs hatched). The differences in dictionsof the Information Center Hypothesis. fledglingsper nestresulted from a dramaticin- The datawere consistentwith the hypothesis creasein survival of nestlings in 1985. that unsuccessful birds follow other birds leav- Ospreysbegan to arrive at Lake Ellis Simon ing the colony.Using seconddays' data only, in late February,and mosthad returnedby mid- two predictions were realized. First, the mean March. The earliest breeders laid eggs by 20 was significantlygreater for DEPuin intervals March, and one pair laid a single egg as late as with at least one DEPscompared with intervals early June(Fig. 6). Most pairs laid 2 weeksafter with no DEPs. Second, DEPs had a Poisson dis- the first pair(s), whereasonly two pairs laid as tribution both amongintervals with at leastone late as 10 weeksafter the first pair (Fig. 7). There DEPuand amongintervals with no DEPu(Table was a strong negative linear relationship be- 3). However, contraryto prediction,DEPu did tween the week of laying (WOL) and clutch size not have a Poisson distribution either in inter- (r2 = 0.784,P = 0.0007)(i.e. early nesting pairs vals with DEPs or intervals with no DEPs. The laid significantlymore eggsthan late nesters). unsuccessful birds followed both successful Regression of brood size on WOL indicated a birds and other unsuccessful birds from the col- similar trend (r 2 = 0.788, P = 0.0006). Week-1 ony, whereas successfulbirds departed inde- layers did not produce as many nestlings as pendently of one another. This conclusion is those that laid in the two subsequentweeks, also evident from analysis of DEPs relative to but this pattern was not sufficientlystrong to ARRs (see Table 3). warrant a quadratic regression.The negative

TABLE4. Summary(t + SD) of reproductionfor eachyear of study.P-values represent the probabilitiesthat therewas no differencebetween years for the indicatedparameter. Sample sizes are in parenthesesafter year.

Year Clutchsize Broodsize Fledglings 1983 (41) 2.59 + 0.745 1.65 + 1.277 0.90 + 0.860 1984 (41) 2.75 + 0.537 1.78 + 1.173 1.02 + 0.961 1985 (48) 2.86 _+ 0.461 1.98 + 1.082 1.48 + 1.111 P 0.039 0.850 0.016 514 HAGANAND WALTERS [Auk, Vol. 107

I 2 3 4 5 6 7 81012 Week of Laying relationship between WOL and numbers of Fig. 7. Newly active nestsfor each week of the fledglings producedwas also significant(r 2 = laying season.Week of laying is relative to the week 0.755, P = 0.001). eggsfirst appearedin the colony. Although decreasein clutch size was an ob- vious factor in the seasonal decline in fledg- lings, other factorsaffected egg or nestlingloss. viability. Although later nesterslost more eggs, We defined eggfailure as the number of eggs comparisonof nestlingloss and week of laying that failed to hatch (clutch size minus brood indicatesthat they did not lose more nestlings size at hatching). There was a significant posi- (r 2 = 0.003, P = 0.899). tive relationshipbetween week of laying and Lost eggs, represented as the proportion of egg failure (r2 = 0.628, P = 0.006). Later layers total eggs in the colony during a given week, had smaller clutches, and more of their eggs wassignificantly negatively correlated with the failed to producenestlings, which exacerbated absolute number of eggs in the colony (r • = the seasonaldecline in offspringproduction. -0.46, P << 0.01). This indicates that the more Eggsmay fail to hatch becauseof predation, eggsavailable to predators,the lessthe proba- accidentsin the nest, or inviability due to in- bility any single egg would be lost. However, fertility, overheating,or overcooling.To distin- the absolute number of eggs lost in a given guishbetween egg lossdue to predationor ac- week was not related to absolute number of cident and loss due to inviability, we defined eggsavailable (r• = 0.02, P >> 0.10) (i.e. there an inviableegg as one that remained in the nest was no evidence of a numerical response of for 6.5 weeks or longer, and a lost eggas one predation in relation to the number of eggs that disappearedbefore that time. The normal available). incubation period for Ospreysis 4.5-5 weeks. If reproductivesuccess is measuredon a pro- No significantrelationship was found between portional scale,losses are measuredin termsof mean number of inviable eggsand week of lay- proportionof reproductiveeffort (clutch size) ing (r2 = 0.146, P = 0.276), or the mean pro- rather than in absolute terms. The proportion portionof the clutchthat was inviable and week of eggsthat becamenestlings significantly de- of laying (r2 = 0.293, P = 0.106). However, a creasedwith week of laying (r2 = 0.727, P = significantrelationship was found betweenthe 0.002) (i.e. late nesterslost a significantlylarger mean number of lost eggs per nest and week proportion of their smaller clutch size). Simi- of laying (r• = 0.456, P = 0.032). Thus the in- larly, later nestersfledged a significantlysmall- creasedegg failure for later nesterswas due to er proportion of their clutches(r • = -0.696, P egg predationor accidentalloss rather than in- = 0.003). The later nesters produced fewer July1990] ColonialNesting in Ospreys 515

5O &0*

Z

.20 E z=1o

•0 0 r-'-i 2 7 I I 2 2 3 3 4 4 5 5 5 2 7 2 7 2 7 2 7 2 5 5 5 5 5 5 5 5 5 GridSize (m) --Bolh • Unumd -- IJ•:l Distance to Neighbor (m) •iõ. 9. Morisita's [•d.ex of Dispersio• for used., Fig. 8. Number of nests with its nearest active potential, and both used and potential nest sitescom- neighbor falling in the indicated distancecategory. bined, at various quadrat sizes.Grid size value rep- Values on the abscissarepresent the midpoint of the resentsthe length of the quadrat edge in meters. distance category.

tween reproductive successand mean distance fledglings and achieved less reproductive suc- to nearestfour active neighbors(r 2 = 0.13, P = cessper unit effort than early nesters. 0.307) (i.e. proximity to neighbors did not en- The number of pairs that initiated laying at hance or diminish fledgling production). Lake Ellis Simon was greatest in week 3 (Fig. Because used sites are members of both "used" 7), while reproductivesuccess expressed as mean and "potential" sites,we calculatedMorisita in- number of young fledged peaked at week 2. A dices for three categories:used sites, unused Wilcoxon two-sampletest did not indicate any sites,and all sites (used or unused). Index values difference in reproductive success between for both "used" and "unused" sites were both week-2 and week-3 breeders (P = 0.218), but above 1, which indicates clumping. However, week-2pairs produced significantly more young the ratio of the "used" site index to the "all" than week-5 pairs(P = 0.050),even though the site index is the important value becausethis samenumber of pairs initiated laying in both revealswhether actual nestsare more clumped weeks (Fig. 7). We suggestthat the number of or less clumped than available nest sites. The pairsat a similar stageof the breeding cyclewas index for "used" sitestended to be slightly low- unrelated to reproductive success,but rather er than the indices for "all" sites or "unused" time of year had an overriding influence. sites (Fig. 9). The great disparity between the Spatial patterns of reproductivesuccess.--The indicesat quadratsizes of 100 m may reflect a mean distance to nearest active neighbor was minimum inter-nest distance threshold for this 183.9 + 75.254 m, and the range was 89-435 m. population. Although "used" sitesare slightly The modal distance fell in the 101-150 m cat- lessclumped than "all" sites, the pattern is not egory (Fig. 8), and the median distancewas 158 sufficientlystrong to suggestavoidance of con- m. The mean distance to a nest's nearest four specifics(other than that someminimum inter- active neighborswas only 280.7 m, which in- nest distance exists). dicatesthat mostnests were surroundedby oth- The dispersion indices show used sites and er active nests. We used exact distances between potentialsites to be similarlydispersed, but they nests and found no relation between week of do not indicate whether the locations of the two laying and distanceto nearestactive neighbor- distributions are in the same space. The nest ing nest (r2 = 0.028, P = 0.641). This indicates sitescould all occur on one edge of an area of that early nesters were not any closer to, or available sites,yet still be dispersedsimilarly farther from, other active nests than late nest- to available sites.To examine this possibility, ers. There was no significant relationship be- we used a multivariate regressionwith the x 516 HAGANAND WALTERS [Auk,Vol. 107

TABLE5. Summarystatistics (œ +_ SD) and comparisonsof reproductivevariables of central and peripheral nests.Wilcoxon two-sample tests were usedfor all comparisons.Sample sizes are in parentheses.Means shown are on a per nest basis.

Nests Variables Central Peripheral P-valuea Clutch size 2.79 + 0.543 (78) 2.61 + 0.661 (52) 0.120 Brood size 2.06 + 1.097 (78) 1.44 + 1.194 (52) 0.003* Brood size• 2.47 + 0.640 (65) 2.14 + 0.772 (35) 0.023* Fledglings 1.30 +__1.308 (78) 0.92 + 0.967 (52) 0.032* Egg failure 0.73 + 0.832 (78) 1.17 + 1.004(52) 0.011' Inviable 0.38 + 0.540 (78) 0.73 + 0.866 (52) 0.029* Nestling loss 0.91 + 0.896 (65) 0.77 + 0.942 (35) 0.383 Egg loss 0.35 + 0.718 (78) 0.44 + 0.725 (52) 0.275 Week of laying 4.10 + 1.592(78) 4.15 + 2.304 (52) 0.625 * = significantat alpha = 0.05. Adjusted brood size (brood sizesof 0 excluded).

and y coordinates of the nest sites (used and pared with foraging durations reported previ- unused) as the dependent variables, and a class ously for Ospreys.This appeared to be a result variable with the value of "used" or "unused" of the substantialdistance males flew to acquire as the independent variable. This analysistests food.Stinson (1978) reported that foragingtrips for statistically similar centers of gravity be- of Ospreysin the ChesapeakeBay region typ- tween points representing used and unused ically lasted < 1 h, as did Greene (1987) for Os- sites,regardless of dispersionpattern. No sig- preysin Nova Scotia,and Poole(1982) for widely nificant differences were indicated by this separatepopulations in Florida and New York. multivariate comparison (Wilk's criterion = The consequencesof conditionsat Lake Ellis 0.99356, P = 0.435, F2,258) (SAS 1985). Thus the Simon were an unusually low rate of fish de- two sets of points occurred in the same space. livery and extensive manifestation of brood re- Central nestsproduced clutch sizessimilar to duction through sibling aggression (Hagan peripheral nests(Table 5). However, peripheral 1986). Under these conditions, food location in- nestsproduced significantly smaller brood sizes, formation should have been at a premium to and consequentlyfledged fewer young (Table Ospreysnesting at Lake Ellis Simon. In addi- 5). The relatively poor successof peripheral nests tion, the apparent central location of Lake Ellis could be attributed to higher egg inviability, Simon relative to foraging sites presentedan becauseegg losswas otherwise similar. Thus, equally suitablecondition to test Horn's Geo- brood sizes of peripheral nests were smaller metric Center Hypothesis. than central ones(Table 5). Peripheral nestsdid Foragingpatterns and colonialnesting.--Mean not losemore nestlingsthan central nests.Once flight distancesare less for average colony an egg hatched, the nestling had as good a members than for average members of a dis- chanceof surviving in a peripheral as in a cen- persedpopulation, as long as the colony is dis- tral nest. Spatial effects thus contrastedwith placed from the center of a circular foraging chronologicaleffects, which were relatedto egg area, with random patch location, by no more lossrather than egg inviability. Consistentwith than 70% of the radius (Wittenberger and Dol- this, peripheral nesters did not lay any later linger 1984).By using known foraginglocation than did central ones (Table 5). The decreased data, as in this study, rather than a theoretical productivity of peripheral nestsappeared to be distribution, the cost of acentric location can be independent of the overall seasonaldecline in measureddirectly. The travel distancesof Lake productivity. Ellis Simon males were from 1.5 to more than 7 times as great as birds could have achieved DISCUSSION by nesting in the geometriccenter of their own foraginglocations. This impliesthat substantial The mean length of time Lake Ellis Simon increasein foragingcosts resulted from colony breeding males spent away from the nest on a membership over the study period. foraging trip (2-5 h) was unusually long com- One does not expect the computed optimum July1990] ColonialNesting in Ospreys 517

locations of nests to correspondexactly with It is possiblethat departuresof unsuccessful nest sites. The former are subjectto sampling birds were influencedby interactionswith oth- errors. Sampling considerationsaside, an in- er birdsnot includedin the sample.We sampled dividual cannot predict his optimum nest lo- only interactionswithin a neighborhoodof 10 cation before the breeding seasonbecause he to 11 pairs.We expectedpatterns consistent with hasno way of knowing what his exactforaging information exchangeto be strongerthan they locations will be. Furthermore, some of the in- were, despitethis constraint.The departuredata dividual geometriccenters did not fall in suit- are arguably inconclusive. The data obtained able nesting habitat. However, the general lo- from radio-equippedmales are moredifficult to cation of the optimum nest site may be reconcilewith the Information Center Hypoth- predictable.Given the restricteddistribution of esis. The lack of variation in each male's de- male Ospreys'individual foraging sitesthat we parture direction indicatesthat information on found, it seemsthat individualsanticipated the food locationwas not used. Rather, familiarity generalvicinity where they would forage.From with a specificsite apparently was of greater males outfitted with transmitters in 1984 and importance. Moreover, males that foraged 1985, it appeared that general foraging-site northeasterlyhad reproductive successsimilar preferenceremained constantyear to year. A to those that foraged southwesterly,which in- more realisticdetermination of optimum nest- dicates that these two foraging areas were site location might be to calculateeach individ- equally capableof supplying food during the ual's geometric center, and then choose the breeding period. Differencesamong individu- nearestlocation to that point which represents als for preferred foraging sites have been doc- suitablenesting habitat. Without exception,that umented for other central-placeforagers (e.g. point would lie near the foragingsites, not near Cook 1978, Morris and Black 1980, Gorke and Lake Ellis Simon (see Fig. 5). Brandl 1980),and among communally roosting Becausesuccessful and unsuccessfulforagers species(Caccamise and Morrison 1986). Many were identified, our study representsa direct speciesmay lack the variability in individual approachto testingthe Information CenterHy- foraging locations assumedin the Information pothesis.We found only weak evidencefor in- Center Hypothesis. A postulated explanation formationexchange. It wasclear that departures for such behavior is that spatial memory de- of previously unsuccessfulforagers were not creaseswith increasingquantity of information influenced by arrivals of successfulforagers. for processing(Olson et al. 1981).When a large Unsuccessfulbirds tended to departwith other foraging radius is necessary(as for Ospreys birds,but did not discriminatebetween previ- nesting at Lake Ellis Simon), minimization of ously successfuland previously unsuccessful spatialinformation (i.e. a narrow departurearc) birds. The nonrandomdistribution of depar- may enhance foraging efficiency.This reason- tures may have been due to independent re- ing does not offer an explanation for colonial sponsesto environmental factors rather than nesting. dueto interactionsbetween birds. For example, These results contrast with Greene's (1987) wind conditionsfavorable to gaining the alti- studyof a smallcolony of Ospreys(11 pairs)in tude necessaryto accomplishlong flightsto the Nova Scotia. In that study birds foraged pri- foraging areasmay have triggered departures. marily within visual range of the colony (<2 Also,dawn and duskmay have imposeda weak km), and foraging trips were of short duration diurnal rhythm on departures.Departures of (most <30 min). At Lake Ellis Simon, where an successfulbirds were distributed randomly, averageforaging trip was longer than 2 h in which indicatesthat interactionsamong birds, duration, food locationinformation may be of rather than environmental factors, were re- little value, as food patchesmay have shifted sponsiblefor the nonrandomdistribution of de- spatiallyin that interval. Birdsin the Nova Sco- parturesof unsuccessfulbirds. However, among tia colony may have receivedmore precise,cur- successful birds, there was a trend toward an rent information on food location because the excessivenumber of intervals with no depar- colonywas near the foragingareas. This is par- tures or with multiple departures(0.05 > P < ticularly true if arrivals were used as a cue, as 0.1). The pattern was the sameamong both suc- in Nova Scotia. At Lake Ellis Simon, arrival di- cessful and unsuccessful birds, but it was rection provided imprecise information about stronger among the latter. locationof food 10-15km distant.Perhaps Lake 518 HAGANAND WALTERS [Auk, Vol. 107

Ellis Simon functions as an information center 1971).Most birds breed over a 3-5-week period only when foraging conditionsare worse than within a season(Perrins 1970), but Ospreysat during our study. Fledgling successwas lower LakeEllis Simonlaid over a 12-weekrange dur- (Table 4), brood reduction more common, and ing the 3 yearsof this study. That breeding in fewer foraging trips were successful(60% vs. Ospreycolonies is more prolongedand lesssyn- 74%) in 1984 than in 1985. However, two males chronous (see also Ames and Mersereau 1964, in two seasonsshowed the same affinity for Ogden 1977,Judge 1983, Garber 1972,Kennedy particular foraging locations. 1977, Prevostet al. 1978) than in many species Becausethe Nova Scotiacolony was closerto is in part a consequenceof body size that makes foragingsites and becauseforagers were within predator swamping less likely. Regardlessof visual range of the colony, the apparent ex- the degreeof synchronyachieved, the question changeof food locationinformation may have is whether rateof eggor nestlingloss is reduced representedlocal enhancement. Local enhance- by the presenceof other nests.Our dataindicate ment is describedtypically as unsuccessfulfor- that benefitsderived from swampingpredators agerscuing on successfulforagers at theforaging may have been realized becausethe rate of egg site.When the colony is in visual proximity to losswas lower when eggswere mostabundant. the foraging site, individuals can watch forag- However, the lack of a numerical response of ing birds without leaving the colony. predators(as measured by numberof eggslost) Whether information exchange occurs near during this time indicatesthat the egg-lossrate or far from the foraging site representsan im- had to do with lower egg lossof early-season portant distinction. The original Information breeders(and perhapsmore experiencedbreed- CenterHypothesis (Ward and Zahavi 1973)was ers), and not with predator-swampingeffects intended to explain assemblagesof birds away due to synchrony. from foraging sites, and thus a phenomenon If Ospreysobtained substantialbenefits from distinct from local enhancement.Assemblages information exchange about food patch loca- (coloniesor roosts)near foraging sites could tion, later layerswould be expectedto fare best, result exclusivelyfrom the locationof the food as they would have more birds on which to cue resource, and not from the need to exchange when their young were at the critical 2-4 post- information. Certainly local enhancement, hatching week age (Hagan 1986).We found no which involves information transfer, is an im- suchpattern. Rather, earlier breederswere more portantbiological phenomenon through which successful than late breeders. birds can gain useful information. However, to Among Ospreys,later layers producedfewer explain coloniesor roostssolely on the basisof young for two reasons.First, they laid smaller the Information Center Hypothesis,as intend- clutches,and second,they lostmore eggs before ed by Ward and Zahavi, requiresthat the as- hatching.Eggs may have been lostto Fish Crows semblagebe remotefrom the foragingsites. The (Corvusossifragus), which were abundantlocal- Lake Ellis Simon colony satisfiedthis distinc- ly. Also, intraspecificterritorial encounterswere tion. From our negativeresults, along with pos- frequent in the colony, and often resulted in itive resultsfrom Greene's(1987) study, we con- momentary battles on the nest platform. Con- clude that information transfer is not likely a specificagonistic interactions in Ospreysare not primary causeof coloniality in Ospreys,but well-documented,but a study of more solitary rather a secondarycapability that may be facul- nesters in northeastern Nova Scotia recorded tatively employeddepending on geographyof 99 conspecificencounters in 560 h of observa- the colony in relation to foraging areas. tion (0.17 per hour) (Jamiesonand Seymour Reproductivesuccess and colonial nesting.--Oth- 1983). Conspecificencounters, usually mani- er sociallyderived benefitsof nestingnear oth- fested as vocal warnings, were 26 times more er individuals can be obtained through syn- frequent at Lake Ellis Simon. Egg loss might chronizationof the group'sbreeding effort. The result from occasional, more intense interac- proximatemechanism Darling (1938)proposed tions (seePoole 1989).This seemsmore likely was that of social facilitation, but the evolu- than predation, given the tenaciousnest atten- tionary basisof synchronized reproduction in danceby femalesduring incubation.Intraspe- a colony might be to reduce the probability of cifickleptoparasitism of fish at the colonyalso predation by swamping predators with more occurred,which representsan additional cost young than predatorscan consume(Hamilton to colonial nesting. Such intraspecificinterac- July1990] ColonialNesting inOspreys 519 tions might be expected to increasegradually unlike those of Lake Ellis Simon. Two active as the breeding seasonprogresses, if interac- nestswere found in these foraging areas, but tions increasewith the density of pairs involved in the topsof very tall treesrather than in trees in nesting. over water. Treesover water occurabundantly Chronologicalpatterns in reproductivesuc- only at Lake Ellis Simon.Ospreys appear to be cessdo not indicate any causeof colony for- equallysuccessful nesting alone, or nestingnear mation, but spatialpatterns may. Birds in the each another (Poole 1989: 138). This is consis- centerof the colonyproduced more young than tent with the hypothesisthat colonial nesting peripheralpairs. This was caused by an increase in Ospreysis a consequenceof nest-sitedistri- in number of inviable eggs produced by pe- bution.This hypothesiswas not testeddirectly ripheralpairs. Though younger individuals of (exceptwithin the colony), and thus is only colonial speciesoften occupynest siteson the weakly supportedby our study. colony periphery (Coulson 1968, 1971;Tenaza Colonial nestingOspreys may benefit from 1971), peripheral nestersin our study did not predationreduction. The benefitsappear to be initiate egg laying any later than central nest- restrictedto effectson egg viability rather than ers, as would be expectedif younger individ- on egg or nestlingloss, and to spatialrather uals occupiedthese sites. Moreover, Ospreys than temporal factors.This may be a primary tendedto return to the samenest year after year, or secondarybenefit. Colony formation may also regardlessof age. Of 21 color-bandedmales, be explainedby resourceconstraints (specifi- none switchednesting territories in 3 yr. Of 27 callythe distributionof nestsites) but not food. color-bandedfemales, only 4 changednest sites Osprey colonieswhich exist today are located among years. in trees over water, like Lake Ellis Simon, or on If a factor other than age produced higher predator-freeislands (Greene 1987). Thus, the egg inviability of peripheralnesters, then col- Nest-siteHypothesis, perhaps combined with ony formation may be an evolved feature. This predation reduction, seemsthe most parsimo- factor would have to be one which had an in- nious explanationfor colonial nesting in Os- fluenceonly on the viability of eggs,as repro- preys,although the Nest-siteHypothesis needs ductive successin peripheral and central nests to be testeddirectly. This conclusiondoes not was otherwise equivalent. Disturbance, dis- contradict the fact that information about food placingthe parentsfrom the nesttoo frequently locationcan be exchangedby Ospreys.We sug- or for excessivelengths of time during incu- gestthat this is not crucialto the origin of co- bation, is a possibility. Sourcesof disturbance lonial nesting in Ospreys,but is rather an in- might include Fish Crows, other Ospreys,or terestingadaptation resulting from groupliving. human recreational fishing in the lake. Clear evidencethat the InformationCenter Hy- Nesting toward the center of the colony may pothesisis a generalexplanation of assemblages representa reproductiveadvantage, but pairs of birds is still lacking. did not move gradually toward the center over the years.Perhaps what precludesnest-site shifts ACKNOWLEDGMENTS is the importance of nesting at a familiar site, Fundingfor this studywas provided by the Wey- of nesting with a familiar mate, or of retaining erhaeuserCorporation, Union Camp Corporation,the a prior resident'sadvantage in disputesover North Carolina Wildlife Federation, the Chapman nest sites. Ospreysnest in a distribution that Fund of the American Museum of Natural History, mirrors available nest-site distribution within the Wake County Audubon Society, and the North the colony. This implies that central positions CarolinaAgricultural Research Service at North Car- were not soughtactively and that pairs did not olina StateUniversity. We thank A. Poole,J. Atwood, seekto nest near other pairs on a within-colony K. Parsons,and P. Rabenold, and an anonymous re- scale.In fact,Ospreys are sensitiveto encroach- viewer for helpful comments on an earlier draft. C. ment into their nesting territory by other Os- Brownieand S. Altmannprovided advice on analyz- ing the InformationCenter Hypothesisdata. We give preys.These results reduce support for the hy- specialthanks to the membersof Camp Bryan Farms pothesis that a primary benefit of colony for permittingus accessto LakeEllis Simon, and for formation is a socially mediated reduction in their dedication to the conservation of that extraor- predation. dinary naturalarea. We are alsograteful to all those TheNest-site Hypothesis.--Ospreys apparently who helped us catchand band Ospreysduring this could nest near the foraging sites,but in trees study. 520 HAGANAND WALTERS [Auk, Vol. 107

LITERATURE CITED FISHER,J. 1954. Evolution and bird sociality. Pp. 71- 83 in Evolution as a process(J. S. HuxleyßA. C. ALEXANDER,R. D. 1974. The evolution of social be- Hardy, and E. B. Ford, Eds.). London, Unwin. havior. Annu. Rev. Ecol. Syst. 5: 325-383. GARBER,D. P. 1972. Ospreynesting ecology in Las- AMES,P. L. 1966. DDT residuesin the eggs of the sen and Plumas countiesß California. M.S. thesisß Ospreyin the northeasternUnited Statesand their Humboldt, California, California State Univ. relationto nestingsuccess. J. App1. Ecol. 3(Suppl.): GORKE,M., & R. BR•NDL. 1980. How to live in col- 87-97. onies: spatial foraging strategiesof the Black- ß & G. S. MERSEREAU. 1964. Some factors in headedGull. Oecologia70: 288-290. the decline of the Osprey in Connecticut.Auk GREEN,R. 1976. Breedingbehaviour of Ospreys(Pan- 81: 173-185. dion haliaetus) in Scotland. Ibis 118: 475-490. BARNARD,C.J. 1980. Flockfeeding and time budgets GREENE,E. 1987. Individuals in an Osprey colony in the House Sparrow (Passerdomesticus L.). Anim. discriminatebetween high and low quality in- Behav. 28: 295-309. formation. Nature 329: 239-241. BERTRAM,B. C.R. 1980. Living in groups:predators HAGAN,J.M. 1986. Temporalpatterns of pre-fledg- and prey. Pp. 64-96 in Behaviouralecology, an ling survival and brood reductionin an Osprey evolutionary approach (J. R. Krebs and N. B. Da- colony. Condor 88: 200-205. vies, Eds.). Sunderland, Massachusetts,Sinauer HAMILTON,W. D. 1971. Geometry for the selfish Assoc. herd. J. Theor. Biol. 31: 295-311. BROWN,C.R. 1986. Cliff Swallow colonies as infor- HENNY,C. J., & A. P. NOLTEMEIER.1975. Osprey nest- mation centers. Science 234: 83-85. ing populationsin the coastalCarolinas. Am. Birds BURGER,J. 1974. Breedingadaptations of Franklin's 29: 1073-1079. Gull (Laruspipixican) to a marsh habitat. Anim. HICKEY,J. J., & D. W. ANDERSON.1968. Chlorinated Behav. 22: 521-567. hydrocarbonsand eggshellchanges in raptorial BURTON,P. J. K., & M. H. THURSTON. 1959. Obser- and fish-eatingbirds. Science162: 271-273. vationson Arctic Terns in Spitsbergen.Br. Birds HOOGLAND,J. L., & P. W. SHERMAN. 1976. Advan- 52: 149-161. tagesand disadvantagesof BankSwallow (Riparia CACCAMISE, D. F., & D. W. MORRISON. 1986. Avian riparia)coloniality. Ecol. Monogr. 46: 33-58. communalroosting: implications of diurnal ac- HORN,H.S. 1968. The adaptivesignificance of co- tivity centers.Am. Nat. 128: 191-198. lonial nestingin the Brewer'sBlackbird (Euphagus CAR•CO, T., S. MARTINDALE, & H. R. PULLIAM. 1980. cyanocephalus).Ecology 49: 682-694. Avian flockingin the presenceof a predator.Na- ISL. 1983. This week's citation classic,i1 April 1983. ture (London) 285: 400-401. Curr. Contents 14: 26. COOK,D.C. 1978. Grey Herons (Ardeacinerea) hold- JAMIESON,I., & N. g. SEYMOUR.1983. Inter- and intra- ing feedingterritories on the Ythan estuary.Bird specific agonistic behavior of Ospreys (Pandion Study 25: 11-16. haliaetus)near their nest sites. Can. J. Zool. 61: COULSON,J. C. 1968. Difference in the quality of 2199-2202. birds nestingin the centreand on the edgesof JENNINGS,T., & S. M. EVANS. 1980. Influence of po- a colony.Nature 217: 478-479. sition in the flock and flock size on vigilance in --. 1971. Competitionforbreedingsitescausing the StarlingSturnus vulgaris. Anim. Behav.28: 634- segregationand reduced young production in 635. colonial animals. Pp. 257-268 in Dynamics of populations(P. J. den Boer and G. R. Gradwell, JUDGE,D. S. 1983. Productivity of Ospreysin the Gulf of California. Wilson Bull. 95: 243-255. Eds.).Proceedings of the Advanced Study Insti- tute on dynamicsand numbers in populations KENNEDY,R. S. 1977. The status of the Osprey in (Oosterbeek 1970). The NetherlandsßCentre for Tidewater, Virginia, 1970-1971.Pp. 121-134 in AgriculturalPublishing, Wageningen. Transactions North American Osprey research CROOK,J.H. 1965. Theadaptivesignificanceofavian conference(J. C. Ogden, Ed.). Washington,D.C., socialorganizations. Symp. Zool. Soc.London 14: U.S. Dep. Inter. Trans. Proc. No. 2. 181-218. KENWARD,R. E. 1978. Hawks and doves: factors af- DARLING,F. F. 1938. Bird flocks and the breeding fecting successand selectionin Goshawkattacks cycle,a contributionto the study of avian soci- on Woodpigeons.J. Anim. Ecot.47: 449-460. ality. Cambridge,Cambridge Univ. Press. KRUUK,H. 1964. Predatorsandanti-predatorbehav- EMLEN,S. T., & N.J. DEMONG. 1975. Adaptive sig- iour of the Black-headed Gull (Larus ridibundus). nificanceof synchronizedbreeding in a colonial BehaviourSuppl. 11: 1-129. bird: a new hypothesis. Science 188: 1039-1041. LACK,D. 1968. Ecologicaladaptations for breeding FLARE,C. J. 1976. The breeding of the Sooty Tern in birds. London, Methuen & Co., Ltd. Sternafuscata in the Seychellesand the effectsof LAZARUS,J. 1979. The early warning function of experimentalremoval of its eggs.J. Zool. 179: flocking in birds: an experimental study with 317-360. captive quelea. Anim. Behav. 27: 855-865. July1990] ColonialNesting in Ospreys 521

LEVENSON,•I. 1979. Time and activity budgetsof PVLESTON,D. 1977. Osprey population studies on Ospreys nesting in northern California. Condor Gardiner's Island. Pp. 95-99 in North American 81: 364-369. Osprey researchconference (J. C. Ogden, Ed.). MOCK, D. W., T. C. LAMEY, & D. B. A. THOMPSON. 1988. U.S. Natl. Park Serv. Proc. 2. Falsifiability and the Information Center Hy- RABENOLD, P.P. 1987. Recruitment to food in Black pothesis.Ornis Scandinavica 19: 231-248. Vultures--evidence for following from com- MORISITA,g. 1959. Measuring the dispersionof in- munal roosts. Anim. Behav. 35: 1775-1785. dividuals and analysisof the distributional pat- $A$ INSTITUTEINC. 1985. SAS user'sguide: statistics, terns. Mem. Fac. Sci. Kyushu Univ. E (Biol) 2: 1985 ed. Cary, North Carolina, SAS Inst. Inc. 215-235. SIECE•-CAVSE¾,D., & G. L. HUNT JR. 1981. Colonial MORRIS,R. D., & J.E. BLACK.1980. Radiotelemetry defensebehavior in Double-crestedand Pelagic and Herring Gull foraging patterns.J. Field Or- cormorants. Auk 98: 522-531. nithol. 51: 110-118. SIEGFRIED,W. R. 1972. Breeding successand repro- NISBET,I. C.T. 1975. Selectiveeffects of predation ductiveoutput of the Cattle Egret.Ostrich 43: 43- in a tern colony. Condor 77: 221-226. 55. 1977. Courtship-feedingand clutch size in ---, & L. G. UNDERI•I•. 1975. Flocking asan anti- Common Terns Sternahirundo. Pp. 101-110 in Evo- predator strategyin doves.Anim. Behav.23: 504- lutionary ecology (B. Stonehouseand C. Perrins, 508. Eds.). Baltimore, Univ. Park Press. STINSON, C.H. 1978. The influence of environmen- OClVEI%J.C. 1977. Preliminaryreport on a studyof tal conditionson apsectsof the time budgetsof Florida Bay Ospreys.Pp. 139-142 in Transactions breeding Ospreys. Oecologia 36: 127-139. North American Osprey researchconference (J. TENAZA,R. 1971. Behavior and nesting successrel- C. Ogden, Ed.). Washington, D.C., U.S. Dep. In- ative to nest location in Adelie Penguins(Pygos- ter. Trans. Proc. No. 2. celisadeliae). Condor 73: 81-92. OLSON, D. S., G. E. •IANDELMANN, & J. A. WALKER. THOMPSON, D. B. A., & C. J. BARNARD. 1983. Anti- 1981. Spatialmemory and food searchingstrat- predator responsesin mixed-speciesassociations egies.Pp. 333-354 in Foragingbehavior. Ecolog- of Lapwings, Golden Plovers and Black-headed ical and psychologicalapproaches (A. C. Kamil Gulls. Anim. Behav. 31: 585-593. and I. D. Sargent,Eds.). New York, London, Gar- VEEN,J. 1977. Functional and causalaspects of nest land STPM Press. distribution in colonies of the Sandwich Tern PATTERSON,][.J. 1965. Timing and spacingof broods (Sterna sandvicensisLath.). Behav. Suppl. 20: 1- in the Black-headed Gull Larus ridibundus. Ibis 107: 193. 433-459. WARD, P., & A. ZAHAVI. 1973. The importance of PERRINS,C. 1970. The timing of birds' breedingsea- certain assemblagesof birds as "information- sons. Ibis 112: 242-255. centres"for food-finding. Ibis 15: 517-534. POOLE,A.F. 1982. Broodreduction in temperateand WITTENBERGER,J. F., & M. g. DOLLINGER. 1984. The sub-tropical Ospreys. Oecologia 53: 111-119. effect of acentric colony location on the ener- 1989. Ospreys,a natural and unnatural his- getics of avian coloniality. Am. Nat. 124: 189- tory. Cambridge,Cambridge Univ. Press. 204. POWELL,G. V. N. 1974. Experimental analysisof the , & G. L. Hvt•T. 1985. The adaptive signifi- socialvalue of flocking by Starlings(Sturnus vul- cance of coloniality in birds. Pp. 1-78 in Avian garis)in relation to predationand foraging.Anim. biology, vol. 8 (D. S. Farner, J. R. King, and K. C. Behav. 22: 501-505. Parkes, Eds.). Orlando, Florida, Academic Press. PREVOST,Y., R. P. BANCROFT,& N. R. SEYMOUR. 1978. Statusof the Ospreyin Antigonish County, Nova Scotia. Can. Field-Nat. 92: 294-297.