April 1977] General Notes 365

Slightlymore than one-third(35%) of the eggsin three-eggclutches had a crown (Table 2), which was progressivelymore commonfrom the first throughthe third egg,and showeda significantrelationship to the egg'sposition in the hatching sequence(P < 0.005). When a definite crown occurredthis crown occurred6.6 timesmore frequently on the third eggthan on the first and 3.8 timesmore frequently on the third eggthan on the second.A definitecrown occurred1.7 timesmore frequently on the secondegg than on the first. This agreedwith Preston's(ibid.) findingswith the LaughingGull and CommonTern. In two-eggclutches a crown(faint or definite)occurred 2.7 timesmore frequently on the second egg than on the first (Table 2), but a Chi-squaretest showedno significantrelationship between the presenceof a crown and the hatchingsequence in two-eggclutches. Preston (ibid.) alsofound no differencebetween the first and secondeggs of the Common Tern. Over one-quarter(28%) of the three-eggclutches contained one differentlycolored egg with two eggsof the samecolor. The one eggthat differedin colorwas the third to hatch 79% of the time and the first to hatch 21% of the time. The secondegg was never found to be the differently coloredegg of a clutch. Assumingthere was an equalprobability that eachposition in the hatchingsequence might be held by the differently coloredegg, a Chi-squareanalysis was made (P < 0.005), showingthat when a three-egg clutchhad two eggsof one colorand one eggof another color, the differently coloredegg was mostlikely to be the third egg.--MICH^EL L. CHAMBERLIN,Interlochen Arts Academy, Interlochen, Michigan 49643. Accepted 14 Nov. 75.

Sizes of snails eaten by Snail Kites and Limpkins in a Costa Rican marsh.--B0th the Limpkin (Aramusguarauna) and the Snail Kite (Rostrhamussociabilis) feed on freshwatermolluscs. The Limpkin may prey upon a variety of molluscs(Snyder and Snyder1969, Living Bird 8: 177),but the Snail Kite is known to feed almost exclusivelyon prosobranchsnails of the genusPomacea (Sykes and Kale 1974,Auk 91: 818). This paperreports a differencein the distributionsof shellsizes of Pomaceaconsumed by kites and Limpkins at a marsh where both bird speciesappear to feed exclusively on Pomacea. The Limpkin and the kite employdifferent strategiesin capturingsnails and in extractingthem from their shells.The Limpkin wades about in the shallows,probing the bottom until its bill contactsa snail, which it then carries to a suitable elevated, flat spot on the marsh, hammers out the operculum with a sharp blow of the bill, and extractsthe meat. The bill makes a characteristicnotch in the columellaredge of the operculumand frequentlymakes a hole in the dorsalside of the shell oppositethe point of entry. Other foraging techniqueshave been reportedfor the Limpkin (Snyderand Snyderop. cit.), but were not observedat the study site. The kite, on the other hand, searchesvisually while flying over open water. It can forage over water that is too deep for the Limpkin to wade in. The kite takes snailsthat are at or near the surface, carries its prey to a perch and then extracts the meat with a quick stroke of the beak. Generally the kite leaves no mark on the discardedshell or operculum(Snyder and Snyder op. cit.). Empty snail shellsaccumulate at the habitual feeding sitesof individual kites and Limpkins. One can determine which predator specieswas responsiblefor the formation of a particular pile of shellsby the presenceor absenceof perforations in some of the shells. Fieldwork was donein January1975 at the Organizationfor Tropical Studiesfield stationat Palo Verde in Guanacaste,Costa Rica. The study was conductedon the large freshwatermarsh south of the field station. The marsh consistedof extensivepatches of rushes, grasses,water hyacinth, or open water. Parkinsoniatrees were scatteredthroughout the shallowerwetlands. Pomacea of unknown specieswere abundant, as evidenced by their empty shells, which formed an integral part of the sediments.The ubiquity of pilesof snail shellssuggested that Limpkins and kites were the major predatorsof Pomacea. Most of these piles seemedto have been made by Limpkins. I collectedsamples of shellsfrom discretepiles. As the kites I saw invariably perchedin Parkinsonia when feeding, only piles found under thesetrees and having few or no perforatedshells were assumedto be of kite origin. Piles located under Parkinsonia trees and having a noticeablefrequency of perforated shellswere rejectedto avoid samplesof mixed origin. Piles made by Limpkins were sampledfrom various parts of the marsh. A 2- to 4-liter sampleof shellswas taken from the centerof eachpile used.I sampled7 piles accumulatedby kites and 17 attributed to Limpkins. The greatestlength of each shell was measured in millimetersusing a rectangularframe and rule (seeFig. 1, inset).In all I measured1373 shells (320 kite, 1053 Limpkin). The mean length for all Limpkin shellscombined is 34.77 mm, with a standarddeviation of 5.45 mm. Shellsfrom kite piles averaged43.18 mm, with a standarddeviation of 6.03 mm. The differencebetween means, using all piles combined for the same bird species,is significant (P < 0.01) by the approximate 366 General Notes [Auk, Vol. 94 LIMPKIN

• 20

uJ

'" 15

•' 10 uJ

cl u,i

u. 5

GREATEST LENGTH OF SHELL (mm) Fig. 1. Frequencydistributions of greatestlength of shellsfrom pilesaccumulated by Limpkinsand Snail Kites.

t-test with Welch's estimated degreesof freedom (Kirk 1968, Experimental design:Procedures for the biologicalsciences, Belmont, California, Wadworth, pp. 106-107). Shell lengthsranged from 22 to 55 mm for Limpkin pilesand 26 to 61 mm for kite piles. Fig. 1 showsshell size distributions, using 3-ram intervals for clarity. The observeddifference in the sizesof prey taken by thesetwo birds may reflect evolutionaryadjust- ments in their feeding mechanisms,either as carry-oversfrom their phyletic historiesor as character displacementcaused by competitionwith eachother. The largersize of preytaken by kitesmay be a result of largersnails being easier for them to see.It is alsopossible that the kite'stalons are bettersuited to larger snails,or that largesnails are difficultfor Limpkinsto open,but the extremesof preysizes taken by either speciesare quite similar. Both shellsize distributions are distinctlyskewed, as indicatedby Fisher'sg• statistic(Bliss 1967, Statistics in biology,vol. 1, New York, McGraw-Hill, pp. 144-145).The distribution for Limpkin pilesis positivelyskewed, with g, = 0.478, whereas,the distributionfor kite pilesis nega- tively skewed, with g, = -0.131. The observeddeviations are significantlydifferent from normality at levelsof P < 0.001for Limpkinpiles and P < 0.1 for kite piles.The overlappingskewed tails of these distributionssuggest that eitherspecies may take practicallyany sizeof snailavailable, but on the average they take prey that are distinctlysmaller or larger than thosetaken by the other snail predator. Whetheror not kites and Limpkinsare in closecompetition for Pomaceaat Palo Verde is a difficult questionto test.Data areneeded on size distributions of availablesnails and how these distributions vary spatiallyand temporally. The differencein sizesof preymay also reflect age-specific variation in theuse by Pomaceaof different habitats. If older, larger snailswere more prevalent in deep water, they would be moreavailable to kitesthan to Limpkins.A studyof the distributionby sizeof live Pomaceain the marsh couldilluminate this questionbut wasnot undertakenin the presentstudy for lack of time and becauseof the difficultyof locatingsubmerged live snailsin the turbid marsh.Moreover, the amountof openwater availableto kitesforaging on the marshvaries greatly between the dry season(December through April), whenpiles were sampled,and the wet season(May throughNovember). How thesechanges affect the area of the marsh available to the Limpkins could not be evaluated during this brief study. The two speciesdid not establishexclusive foraging territories: kites and Limpkinswere often seen foraging at the sametime in the shallowerchannels of the marsh. It shouldbe recalledthat althoughthe kite seemsto be dependenton Pomacea, the Limpkin may switchto otherprey when the snailsare scarce(see Snyder and Snyderop. cit.). Giventhat the marshcontains only one species of Pomacea, and giventhe highdegree of April1977] GeneralNotes 367 specializationbythe kite for feeding onPomacea, during times ofresource scarcity theLimpkin, bytaking thesmaller snails, may greatly reduce the future availability ofthe larger snails that the kite selects. Myparticipation onOTS course 1975-2 was supported byawards from the College ofNatural Science at MichiganState University and the Latin American Studies Center, MSU. I thankthe instructors and studentsofthat course for their helpful discussions, especially D. Paulson, S.Smith-Stiles, D. Pool, J. Ewel,and S. Pickett.R. W. Hill andG. A. Dawsoncritically reviewed an earlierversion of themanu- script.Calculations were performed on theWang 600-14 calculator in theMSU Museum.--S.F. COL- LETT,The Museum and Department ofZoology, Michigan State University, East Lansing, Michigan 48824. Accepted25 Nov. 1975.

Abnormalnest building in theEastern Phoebe.-- Several instances ofbirds abnormally build- ingmultiple nests have been reported. Welty (1975) cited instances of the American Robin (Turdus migratorius)andthe European Blackbird (Turdus merula) beginning numerous nests on horizontally hung ladders,girder spaces between roof rafters, stacks of pipes,and pigeon-holes. Herrick (1935) told of an AmericanRobin that began five nests on steps of a fireescape, completed two, and divided the clutch betweenthem. Ashmole (1968) similarly reported anEastern Phoebe (Sayornis phoebe) that built two completenests and divided the clutch between them; two young were ultimately fledged from one of these nests.The usual explanation forthis aberrant behavior revolves around the repetition in man-made structuresand the inability of birdsto identify the proper site. The observations forthis paper were made from1970 to 1972 on Crane Naval Ammunition Depot, Indiana, in conjunction witha studyof Eastern Phoebeand Barn Swallow (Hirundo rustica) breeding ecology. On21 April 1970 I foundEastern Phoebe nesting material (i.e. fresh moss and mud) forming a continuousthinmat for some 2 m along the west side of a centralI-beam under bridge 25, a smallwood andsteel I-beam bridge that extended north-south over an east-flowing stream (Fig. 1). The date of the beginningofthe mat's construction isunknown, butmany first nests were begun before 10April that year. On27 April the depth ofthe mat had been substantially increased, butthe length was only slightly greater. By7 May, a typicalnest cup had appeared atthe south end of the mat. No further construction wasnoted. Atthis time the nest mat was 2.2 m long and 2.50-3.75 cm thick for 1.0 m, and 1.25-2.50 cm thick for the remainderofthe length. The first egg of a five-eggclutch was !aid on 8 May,and five young fledged successfullyon13 June. All aspects ofthe nesting cycle after nest building appeared normal. This nest was destroyedduring the winter, and a normalnest was built and young fledged during the 1971 nesting season.A single visit to the bridge on 1 June 1972 revealed an abnormal nest very similar to the 1970 nest althoughofsmaller dimensions. Thisnest, which contained fiveeggs, was on a west-facing I-beam, and the cup was at the north end of the mat. A secondtype of aberrant nest was found on 2 June1972 under bridge 49, also a woodand I-beam bridge,8 km from the former. This nest, also with greater than normal dimensions, wasuniquely composedof5 separate cups, 2 cups descending assteps ineach direction from the complete central cup

- i

Fig.1. Abnormalnest with extensive base built in 1970at Crane Naval Ammunition Depot, Indiana.