PHAETHON LEPTURUS) at Cayoluis PEIA, PUERTORICO

NEST-SITE SELECTION AND NESTING SUCCESS OF WHITE-TAILED TROPICBIRDS (PHAETHON LEPTURUS) AT CAYOLUiS PEIA, PUERTORICO

FRED CHARLES SCHAFFNER 1 Departmentof Biology,University of Miami, P.O. Box249118, Coral Gables, Florida 33124 USA

ASSTP,ACT.--I studied nest-siteselection and nesting successof White-tailed Tropicbirds (Phaethonlepturus) at nestingcolonies located in coastalboulder talus on CayoLuls Pefia, in the Culebra National Wildlife Refuge, Puerto Rico, during 1983-1988. Twenty-nine to 38 pairs attemptedto nest at each of two colonies,Punta Cruz and South Peninsula,in each season.Nests were distributedalong different coastlinedistances: 170 m at Punta Cruz and up to 420 m at South Peninsula.Nearest-neighbor distances of nest siteswere 2-4 m at Punta Cruz and 6-10 m at South Peninsula.Nests averaged2 m from the forest edge, within 6 m of the waterline and 2 m above mean high tide. Temperaturesin nest crevicesfluctuated between20øC and 35øC.Incubation times averaged 41 days(range: 40-43). Mean fledging times(+SD) were 73 + 5.25 daysin 1984,71 + 2.65 daysin 1985,and 71 + 1.57days in 1986. Overall nestingsuccess (laying through fledging),estimated by the Mayfield method,for the two coloniescombined was low (0.15 in 1984 to 0.26 in 1986), although it was similar to that reported for White-tailed Tropicbirdsat other coloniesworldwide if estimatedby the traditional method. Most nesting failures occurredearly in the egg stage.The mostimportant causesof nest failure at Punta Cruz were abandonmentand agonisticencounters between conspecificsduring the egg stage,including overt and severefighting. In contrast,the most common causeof nesting failure at South Peninsulawas predation. At South Peninsula, reducedpredation by blackrats (Rattus rattus) and increasedhatching success were concurrent with intensive predator control efforts, especially the use of poisoned rat bait. Received11 December1989, accepted 13 May 1991.

CURRENTknowledge of the status, nesting Refuge, at Isla de Culebra, Puerto Rico (18ø20'N, success,and distribution of seabirds in the Ca- 65ø18'W).Cayo Luls Pefia is locatedabout midway ribbean Sea and adjacent tropical western At- between the main island of Puerto Rico and St. Thom- lantic Oceanis at presentincomplete. Available as,U.S. Virgin Islands.White-tailed Tropicbirds were first reportedin the vicinity of Culebraby Wetmore information suggeststhat Caribbean seabird (1917).The nestingcolonies were f•rst identified and populationshave sufferedsignificant declines describedby Keplerand Kepler(1978), and first cen- due to human activity and introduced animals susedby Furniss(1983). About a half-dozenpairs of (Westerman 1953, Dewey and Nellis 1980, van Red-billedTropicbirds (P. aethereus) also nested at Cayo Halewyn and Norton 1984). I examined the Luls Pefia during 1983-1988. breedingbiology of the CaribbeanWhite-tailed Twonesting colonies were at PuntaCruz and South Tropicbird(Phaethon lepturus), at CayoLu{s Pefia, Peninsula on Cayo Luls Pefia (Fig. 1). Punta Cruz Culebra,Puerto Rico. My objectiveswere to de- consistsof a low-lying volcanicboulder talus along termine the size and nestingsuccess of the nest- the entireshoreline. The adjacentsubtropical dry for- ing coloniesat CayoLuls Pefia,to describethe est (Ewel and Whitmore 1973) occupiesa narrow pen- characteristics and distribution of nest sites, and insulaextending to < 15 m abovethe waterline. This colonyhad a distinctlywindward side, with no shel- to identify causesof nesting failure. ter from the wind, and a sheltered leeward side. South Peninsulaconsists of a low-lying talusof large boul- STUDY AREA AND METHODS derssparsely strewn along the northernextent of the shoreline, changing to smaller more densely strewn I studiednesting colonies on Cayo Luis Pefia,an bouldersat the southernextent. The arrangementof uninhabited islet in the Culebra National Wildlife theseboulders suggests an origin as productsof the erosionof the cliffs and steepslopes of the adjacent forested hillside, which extended to >30 m and pro- i Present address: U.S. Fish and Wildlife Service, vided the colonywith moderateshelter from the wind. CaribbeanIslands National Wildlife Refuge,P.O. Box Midafternoonair temperaturesat Punta Cruz varied 510, Boquer6n,Puerto Rico 00622,USA from 26øC to 44øC, whereas those at South Peninsula

911 The Auk 108: 911-922. October 1991 912 FREDCHARLES SCHAFFNER [Auk, Vol. 108

TheCulebra Archipelago ßb AtlanticOcean 65ø18' W• E •; S

ß • "'•--•-,,,,,,• t8ø20' •Q•. Culebra•,•s% •:• . ,4,• '"•'•C •"•,t/Lu•Ys• "/' VirginPassage SouthPeninsula •0 Pe•a

I I I I I I 0 1 2 3 4 5 KILOMETERS Sonda de Vieques

Fig.1. Mapof theCulebra archipelago, showing location of studyareas (Punta Cruz and South Peninsula) on CayoLuis Pefia.

were typically 29øCto 37øC.Lowest midafternoon egg (Schaffner1988, 1990a, b; Schaffnerand Swart temperaturesoccurred on the windward sideof Punta 1991).I madeadditional visits to the nestingcolonies Cruz, while the highest temperatureswere on its lee- during late March of 1987 and 1988, 25 May to 1 ward side. Surfacetemperatures on the rocksranged Augustof 1987,and 14 May to 2 Juneof 1988.How- up to 66øC.Nighttime ambienttemperatures could ever, the number of visits madeduring the egg stages fall to 18øCat all locations.The colonies experienced of 1987 and 1988 was inadequatefor reliable esti- strongtrade winds (6.8 m/s, gustingto 9.2 m/s), which mationof nestingsuccess, and duringvisits to Cayo were easterlyto northeasterlyearly in the nesting Luis Pefiaduring the chickstages of 1987and 1988, season(February to March), and shifted to south- I did not alwaysexamine all nests(see Fuller et al. easterlyfrom April to May onwards, with the pro- 1989,Pennycuick et al. 1990). gressionof the IntertropicalConvergence (see Fuller I paintedtranset markers at 10-mintervals on boul- et al. 1989,Pennycuick et al. 1990,Schaffner 1990a). dersalong the coastlineof eachcolony by suspending Nest sites.--Data for 1983 were collected by U.S. a 10-m cord horizontally from marker to marker Fishand Wildlife Servicepersonnel (Furhiss, Taylor, throughthe colony.The heightsof the markerlo- and Griffen-Taylor MS). During February to June cationsapproximately bisected the verticaldistribu- (1984)and Februaryto August(1985 and 1986),I vis- tion of the nest sites in each 10-m interval. These lines ited each colony two or three times per week. On providedmeasures of the linear extentof the nesting each visit all likely looking creviceswere searched colonies,which could be directly comparedbecause for evidenceof nestingactivity. From a blind located of the standardized10-m step lengths (see Pennycuick in the adjacentforest, or from a boat, I also found and Kline 1986). likely nestingsites by observingadults entering and I measurednearest-neighbor distances of nest sites leaving the colony.All known visitation sites(where by meansof a tape measuresuspended between the an adult was capturedor observed,but no egg was site entrances.Sites sharing an entrancewere record- laid--most became nest sites in subsequentseasons) ed ashaving 0.0 m nearest-neighbordistances. Using and nestsites were uniquely marked,and I captured a clinometermounted on a tripod (Fig. 2), I measured and banded adults in all accessible locations. Periodic theheight above the approximate mean high-tide line visits allowed me to determine the nesting successof (waterline) and the distanceinland from the water- the breeding birds, observe evidence of predators, line for eachsite. I suspendeda tape measureto de- and documentother phenomenaat the colonies.Al- termine the distance(d•) from the site entrance to the though CaribbeanWhite-tailed Tropicbirdsdo not clinometer,and I suspendeda tapemeasure from the exhibit the obvious sexual dimorphism reported by waterline to the clinometer to determine that distance Stonehouse(1962) for South Atlantic (Ascension Is- (d2)as well. I alsorecorded the respectiveangles (01, land) individuals,I usuallywas able to infer the sex 02)between the horizontal plane of the clinometer of nestingindividuals by incubationorder (males take and dl and d2.Thus, H, =dl sin 0•; H2 = d2 sin 02;L• the first full [ > 36 h] incubationshift after egg laying) = d• cos 01,and L2 = ds cos 02. and by observationof which pair member laid the Nest temperaturesand ambient temperatureswere October1991] TropicbirdBreeding Biology 913 recordedon many visits. Three nestsin 1985 and four in 1986 were fitted with remote reading thermometers for at leasthalf of the nesting period. Thermom- eters at six of these nestsalso provided readingsof the highest (max.)and the lowest(min.) temperatures in these nest crevices between visits, as well as the temperatureat the time of my visit (usually late af- d2 crevice ternoon). The thermometers were then reset. H,•oncli ....tripod ter Nestingsuccess.--I calculated nesting successby the traditional estimationmethod and by Hensler's(1985) modificationof the Mayfield method. The traditional .... method consideredall known eggs and chicks, regardlessof ageat discovery,and assumedsurvival for all offspring remaining in the colony at time of ob- serverdeparture. Hatching and fiedging (normal de- partureof the chickfrom the nest)success of the nests waterline (H3) and distance inland from waterline (L3) at eachcolony were calculatedas number of eggslaid of nest sites and visitation sites. Distances d• and d2 or chicks hatched minus the number known failed, were measureddirectly with a tape measure. divided by the number of eggslaid or chickshatched. Hatching and fiedging successwere multiplied together for an estimateof overall success. nobitaclypeatus). I usedsnap trapsand poisonedbait For events (such as hatching, fledging, or preda- (dipthacinone-impregnatedbaited paraffin blocks, i.e. tion) that occurred between consecutive visits, I oc- "Eatoh's All-Weather Bait Blocks") to control rats, and casionallyused circumstantialinformation (such as I removed large land crabsand hermit crabsphysi- degreeof decompositionof the remainsof failed eggs cally from the colony areas.I made the first bait block or dead chicks,or very high or very low body masses applicationin April of 1985,after some egg losses had of incubating adults) to judge when the event oc- already occurred.I placed bait blocksnear entrances curred. Most often, however, I assumed that an event of previouslydestroyed nests, and at regularintervals occurredon the exact middle day for intervals of odd along the forest edgesat both colonies.In June and numbersof daysor on the earlier of the middle two Augustof 1985,I alsoplaced bait blocksinside empty days for intervals of even numbers of days, because nest crevices,and I applied poisonedbait upon my failures may be more likely to occurearly in the cycle arrival at Culebrain February,March, April, and Au- (Miller and Johnson1978). For long intervals, as in gust of 1986, and March and July of 1987. 1983, I assumedthat an event occurred on the day closestto the 40%point of the interval (Johnson1979). RESULTS Nest exposure(number of days of observation)was taken to be the period from discoveryto fate or my Colony and nest-sitecharacteristics.--White- last visit of the season. For chicks that were also known tailed Tropicbirds nested along 170 m of con- from eggs shortly before hatching, exposurebegan tiguouscoastline at the Punta Cruz colony dur- on the estimatedday of hatching. Becausenests often were inspectedtwice in a single day (early morning ing 1984-1988. At the South Peninsula colony and late afternoon), nestsknown from a single day nestsoccurred along 220 m of coastlinein 1984, 330 m in 1985, and 420 m in 1986-1988. were allotted an exposureof 1 day. I assumeda typical incubationtime to be 41 days Nestswere situatedin crevicesthat provided for all years,and I assumedthe typicalchick fledging shelter from wind and direct sunlight. Nest time to be 72 days for 1983, 73 days for 1984, and 71 scrapeswere placed from immediately adjacent daysfor 1985and 1986.A successfulnest for a period to the crevice opening to a few meters within (e.g. egg stageor chick stage)is one in which at least the crevice. Some nest sites shared entrances. oneyoung survives through the specifiedperiod. Thus, The birds brought no nest material to the crev- nest survival is the same as the survival of individual ices.They placed their single egg on bare rock young becauseWhite-tailed Tropicbirds always lay or soil, or on whatever litter may have fallen only one egg per nest. into the crevice.Daily variation in ambient tem- PredatorcontroL--During 1984 I controlledpreda- peratureswithin the crevicesranged from 20ø torswith snaptraps set at the colony-forestedges. In 1985-1987, in responseto managementconcerns re- to 35øC(Fig. 3). garding predation on eggs and chicks,I continued Nearest-neighbor distances(NNDs) for nest predator control, aimed primarily at black rats (Rattus sites at Punta Cruz were one-half to one-third rattus), and to a lesser extent land crabs (Gecarcinus those at South Peninsula (Table 1). The number spp.) and large land hermit crabs(Coenobitidae: Coe- of nest sitesper unit coastlinedistance at Punta 914 FREDCHARLES SCHAFFNER [Auk, Vol. 108

40 m in 1984, 32 nest sites/330 m in 1985, and 33 nest sites/420 m in 1986). 30 Despite the differences in topography and nesting density of the two colonies,nest place- 20 ments were strikingly similar. Distance from (15)(38) min. (18)min. (50) 0 min. the forestedge was approximately 2 m for nests o lO (29) (6) (22) at both colonies in 1984-1986, and horizontal -1985- distancesfrom the waterline ranged from 4.09 • o to 6.49 m (Table 1). All nest sites were less than 10 m above the water, and heights above the I•' 30 waterline were between 1.73 m and 2.83 m for the 3 yr data were collected(Table 1). 20 Populationsize.--The total number of nest sites (26)(26) min. (9) (9) min.(9) (74) at CayoLuls Pefiawas similar in 1984-1986(63, (26) (77)min. 61, 62) as was the total number of clutches ini- 10 (74) -1986- tiated (69, 69, 68) (Table 2). The smaller number of nest sitesand clutchesin 1983 (Table 2) reflect fewer surveysof the coloniesthat year; clutches Incubation Brooding Chick-rearing that started and failed between observer visits Fig. 3. Temperature regimes of 3 White-tailed were overlooked (Table 3). Tropicbird nest sites at Punta Cruz in 1985 and 4 in The number of nest sites known to be active 1986,during the incubation,brooding, and the post- midway through the nestingseasons (mid-May) brooding chick-rearing phasesof the nesting cycle. was similar at Punta Cruz for the years1971 and Measurementswere obtained using maximum-min- imum thermometers with thermistors installed in the 1983-1988 (Table 3), whereas the number of known active sites mid-season at South Pen- nest crevices.For each stage (incubation, brooding, or chick-rearing),the middle reading is the midday insula was lower for 1983-1984 vs. 1985-1988 temperature at the time I visited the nest. Designa- (Table 3). tions "min." and "max." indicate lowest and highest More frequent visits in 1984-1986 revealed temperaturewhich occurredbetween my visits.Hor- neststhat startedand failed in periodsof 2 days izontal lines represent the mean temperatures, ver- to 4 weeks. The between-yeardifference in the tical lines representthe ranges,and boxesrepresent number of known nesting attempts(including standard deviations. Total number of meaurements failures)at mid-seasonwas greaterthan the dif- within each categoryis indicated below the points. ferencein the number of remaining active nests (Table 3). The difference varied with the num- Cruz (28-30 nests/170 m) was greater than at ber of observer visits to each colony at mid- South Peninsula, where similar numbers of nest season(Spearman rank correlation[Siegel 1956]; sites were dispersed over increasing coastline Punta Cruz rs = 0.995, P = 0.0193, n = 7; South distances from 1984 to 1986 (33 nest sites/220 Peninsula r• = 0.902, P = 0.0273, n = 7). Simi-

TABLE1. Placement(m) of nestsites at Cayo Luls Pe•a, 1984-1986.Abbreviations: NND = nearest-neighbor distances, DNV = distance from nearest vegetation, DWL = horizontal distance from waterline, and HTL = height above mean high tide line. Data are œ+ SD; samplesizes are in parentheses.

NND DNV DWL HTL

1984 Punta Cruz 3.96 + 7.03 (31) 2.09 + 1.41 (30) 5.34 + 2.75 (30) 2.44 + 1.28 (30) South Peninsula 6.32 + 14.00 (33) 2.37 + 1.31 (32) 4.09 + 2.59 (32) 1.73 + 1.24 (32) 1985 Punta Cruz 3.24 + 2.28 (31) 2.18 + 1.50 (28) 6.46 + 2.62 (28) 2.83 + 1.60 (29) South Peninsula 9.84 + 14.19 (32) 2.25 + 1.60 (31) 5.27 + 3.13 (31) 2.11 + 1.26 (31) 1986 Punta Cruz 2.24 + 1.65 (30) 2.11 + 1.48 (28) 5.92 + 1.96 (27) 2.37 + 1.26 (28) South Peninsula 7.80 + 9.28 (35) 1.94 + 1.68 (33) 5.46 + 3.16 (33) 2.16 + 1.19 (33) October1991] Tropicbird Breeding Biology 915

TAI•LE2. Nest sitesand clutchesat CayoLuls Pefia. to 81 days (œ= 73.3 + 5.25, n = 4 [1984]; 71.2 Designations"1 clutch," "2 clutches,"and "3 ñ 2.65, n = 17 [1985]; 71.1 ñ 1.57, n = 18 [1986]. clutches" indicate the number of clutches laid (i.e. Nestingsuccess and predatorcontroL--By the the numberof nestingattempts) at a particularsite. traditional method, I estimated nesting success 1983 1984 1985 1986 during the eggstages at CayoLuls Pefia: 55.9%, n = 34 (1983); 50.7%, n = 69 (1984); 42.0%, n = Punta Cruz 69 (1985); and 53.7%, n = 67 (1986). I estimated Clutches Discovered as chicks 2 7 1 2 nestingsuccess during the chick stage (79.0%, Discoveredas eggs 19 24 33a 32b n = 19[1983]; 81.8%, n = 33 [1984];93.3%, n = Total known 21 31 34 34 30 [1985],and 80.6%,n = 36 [1986]),with overall With egg and chickdata 7 8 12 14 nestingsuccesses (hatching x fledging)of 44.1% Sites (1983), 41.5% (1984), 39.1% (1985), and 43.3% 1 clutch 21 29 26 25 (1986). 2 clutches 0 1 3 3 Nesting success(probability of survival) es- 3 clutches 0 0 0 1 timatesfor the egg stagesat both colonies,es- Total 21 30 29 29 timatedby the Mayfield method, were (ñSD) South Peninsula 0.3609 ñ 0.09497 for 1983, 0.2168 ñ 0.05685 for Clutches 1984, 0.2724 ñ 0.05601 for 1985, and 0.3883 _+ Discovered as chicks I 9 5 7 0.06598 for 1986, basedon 31, 52, 63, and 58 Discoveredas eggs 12 28 • 30 27 Total known 13 38 35 34 nestsfor which sufficientegg data were record- With egg and chick data 7 8 11 13 ed. Nestingsuccess during the chickstage was 0.6298 + 0.1456, 0.7075 + 0.09993, 0.9190 ñ Sites ! clutch 13 29 29 32 0.07763,and 0.6742 + 0.08405,based on 19, 33, 2 clutches 0 3 2 I 29, and 33 nestswith chicks,for 1983-1986. 3 clutches 0 1 1 0 Overall nesting successwas estimated to be Total 13 33 32 33 0.2273 _+0.08081 (1983), 0.1534 _+0.04604 (1984), Total Luis Pefia 0.2497 ñ 0.05369 (1985), and 0.2648 ñ 0.05515 Clutches (1986). There was a trend towards increased Discovered as chicks 3 16 6 9 overall nesting success,primarily due to in- Discoveredas eggs 31 53• 63a 59 creasedegg survival, at South Peninsulafrom Total known 34 69 69 68 1984to 1986,while nestingsuccess at Punta With egg and chickdata 14 17 23 28 Cruz remained relatively constant(Fig. 5). Sites Ages of eggsand chicksat time of discovery 1 clutch 34 57 56 2 clutches 0 4 5 574 indicateincreased proficiency in locating nest 3 clutches 0 1 1 1 sitesearly after both egg laying and hatching Total 34 63 61 62 in successiveseasons (Table 4). The exception,

a ! no data. egg age at time of discoveryat South Peninsula b 2 no data. in 1986,is consistentwith my later arrival that seasonand the greater nesting activity at South Peninsula than at Punta Cruz in mid-February larly, at both coloniesthe number of nesting of all seasons(Fig. 4). attemptsknown for the entire seasonvaried Most egg lossoccurred early in the cycle (Ta- with the number of observer visits for the entire ble 4), usually at the end of the first full incu- season(Spearman rank correlation;Punta Cruz bation shift (of the male) and before the return rs = 0.991, P = 0.0152, n = 7; South Peninsula of his mate (see also Schaffner 1988). Predation rs = 0.821, P = 0.0442, n = 7. on chicks by land crabs or hermit crabs (see Nestingphenology.--Nesting'activity was un- below) always occurred< 10 days after hatch- synchronized,continued over 6 months, and ing. Important causesof nestingfailures during peaked in mid-May (Fig. 4). Incubation times the egg stagewere nest abandonment, agonistic (laying to hatching) ranged from 40 to 43 days encountersat the nest site between parents and (means[+SD] = 40.7 ñ 1.21, n = 6 [1984];41.2 conspecificintruders, and predationby rats,land ñ 0.86, n = 15 [1985]; and 41.2 ñ 0.86, n = 18 crabs, or hermit crabs (Table 5). [1986].Fledging times (hatching to normal de- Two egg losses due to agonistic encounters parturefrom the nest)of chicksranged from 66 between adults occurred at Punta Cruz in 1984, 916 FREDCHARI•ES SCHAFFNER [Auk,Vol. 108

TABLE3. White-tailed Tropicbirdnesting attempts (including failures),observer visits, and activenest sites at CayoLuls Pefiain mid-May of 1971,aand 1983-1988.a Note the relationshipbetween number of observer visits and number of known nesting attempts.

1983 1988 1971 (mid- 1984 1985 1986 1987 (14 & 15 (14 May) May) (15 May) (15 May) (15 May) (25 May) May) Punta Cruz Active nests on above date 11 14 13 11 14 14 14 Known attempts to above date 11 19 25 24 28 23 20 Known attempts for season 11 21 31 34 34 25 22 No. observer visits b to above date 1 4 38 28 31 8 4 Season total observer visits b 1 6 51 65 80 22 9 South Peninsula Active nests on above date 6 9 10 15 15 15 16 Known attemptsto above date 6 13 35 32 28 28 22 Known attemptsfor season 6 13 38 35 34 30 23 No. observer visits • to above date 1 6 34 23 25 7 3 Season total observer visits b 1 7 45 46 62 15 5 Total Luls Pefia Active nests on above date 17 23 23 26 29 29 30 Known attemptsto above date 17 32 60 56 56 51 42 Known attempts for season 17 34 69 69 68 55 45 No. observer visits • to above date 2 10 72 51 56 15 7 Season total observer visits b 2 13 96 111 142 37 14

' Sources:for I971, Kepler and Kepler 1978;for 1983,Furniss, Taylor and Griffen-Taylor MS; for I984-88, this study, Schaffner1988, unpubl, data. • Observer visit defined as any visit during which any censusdata were recorded, regardlessof number of observers. four in 1985, and six in 1986. There were only At a nest site at Punta Cruz in 1986, a fight three such losses at South Peninsula for 1984- between the resident female and a female in- 1986 (Table 5). I witnessed overt and severe truder resulted in the destruction of the egg. fighting, resultingin injuriesand lossof blood, The intruder displacedthe resident female and at nest and visitation sites in 1984 and 1986. The nestedwith the original male later that season. first such encounter in 1984 occurred at a nest I observedwounds on another incubating fe- site shortlyafter egg laying and resultedin the male at Punta Cruz in 1986. This female and abandonment of the intact egg and nest site. ! her egg disappeared4 days later and a second found two bloodied birds (later identified as female nestedat that site with the original male males) exhausted and immobile in a nest crev- a few weeks later. Other losses believed to have ice, their bills locked together. This site re- resulted from agonistic encounters were fol- mained unused until one of the combatants and lowed by nestingat the samesite by one or two his mate from 1984 nestedthere successfullyin new individuals within 30 days. 1986. During 1984 1 observedblack rats (Rattusrat- A second event at Punta Cruz in 1984 in- tus)in the tropicbirdcolonies and adjacentfor- volved three birds at a visitation site. ! discov- est, as well as rat fecal material in nest crevices. ered two of the birds strugglingwith their bills With snaptraps (1984) I collected12 ratsat Pun- interlocked. All three birds exhibited minor lac- ta Cruz and 15 at South Peninsula. ! trapped 7 erations and blood stains. One of the inter- rats in 1985 and 4 in 1986 in the two colonies. locked birds (a male) and the third bird nested In April and May of 1985,ca. 2 weeksafter the together at that site one month late. first application of poisonedbait, I found dead In a third incident at Punta Cruz in 1984, one rats in the coloniesand adjacentforest edges. pair of nesting White-tailed Tropicbirds was Dead ratswere found in both colonies5-15 days displacedby conspecifics.The original inhabi- after the first applicationof rat bait in 1986,and tants abandoned the site and the intact egg, but by the end of that seasonI had discovered4 the site remained unused, until it was occupied dead at Punta Cruz and 6 at South Peninsula. by Red-billed Tropicbirdsin 1987 and 1988. In both 1985 and 1986, the presenceof other October1991] TropicbirdBreeding Biology 917

40 ChickStage 1.0'

30 20 1984••J• 10 0.4-

0 0.2'

South • hatchedJayed Peninsula Egg Stage 40' a, fledged 0.6'

30' 1985• 0.4- Punta-----o,.-- layed Cruz • hatched 20' ----a--- fledged

10' 0.0 Overall 0.4- 0

0.2- 10-

0.0 30-

1986 Fig. 5. Nestingsuccess estimated by the Mayfield 20- method(Hensler 1985). Open symbols represent data 10- for the PuntaCruz colony and closedsymbols rep- resentdata for the SouthPeninsula colony. Vertical 0 lines representthe estimatedstandard deviation of Jan. Feb. Mar. Apr. May Jun. Jul. Aug. Sep. the Mayfieldestimate of nestingsuccess. Month Fig.4. Sequenceof laying,hatching, and fledging PuntaCruz. At SouthPeninsula 12 eggsin 1984 at CayoLuls Pe•a in 1984-1986.Number of eggsand chicks are cumulative for each season. and 16 in 1985 were lost to predatorsand unknown causescombined, but just5 in 1986(Ta- ble 5). deadrats deep in the rock crevicesand adjacent Total counts of chicks lost at both colonies in brushwas revealed by the odorsemanating from 1986were higher than in previousyears (Table those areas. 5). Predationon chicksalways occurred when ! also observedlarge land crabs (Gecarcinus chickswere lessthan 2 weeksof age.The Punta spp.;8-12 cmcarapace diameter), and largeland Cruz chickslost to predation in 1985 and 1986 hermit crabs(Coenobitidae: Coenobita clypeatus; were killed at <5 daysof ageby large (>7 cm >7 cm shell length) in the coloniesand in the shell diameter) hermit crabs,which fed on the vicinity of recently damagedeggs and dead or freshly killed chicks. In one case(Punta Cruz moribund chicks (for crab identifications, see in 1985),I founda largehermit crab feeding on Voss1976). More thanhalf of the damagedeggs a freshlykilled, newly hatchedchick. The par- hadlarge oval holes, which implicated rats. Ap- ent wasstill sitting in the neston both the chick proximatelyone third of the apparentlypred- and the hermit crab. ! also found hermit crabs ator-destroyedeggs had pairedlarge and small in nestswith newly hatchedchicks and brood- holes,corresponding to thelarge and small claw ing adults in 1984 and 1986 at Punta Cruz, and tips of large Gecarcinusland crabs.In mostcases twice in 1984 at South Peninsula. ! observed the eggswere pushed against a rockand punc- Sally Lightfootcrabs (Grapsidae: Grapsus grap- turedwith the clawtips. One adult maletrop- sus) feed on a chick carcass at Punta Cruz in icbirdwith a severeneck wound was killed by 1984 and 1986, and on fish or rat carcasses at rats at Punta Cruz in 1985. Rat feces were found both coloniesin all years.I alsoobserved Grap- in thisnest, and the eggdisappeared. The pro- susand Coenobitascavenge previously damaged portion of eggslost to predatorsand unknown eggs and dead chicks. causesdid not changegreatly in 1984-1986at Two chicks at Punta Cruz and three chicks at 918 FREDCHARLES SCHAFFNER [Auk, Vol. 108

TABLE4. Estimatedage (days)of all eggsand chicksat the time of discoveryand failure (destruction,death, or disappearance),at CayoLuls Pefia during 1983-1986.Data are œ+ SD; samplesizes are in parentheses. Timesof occurrenceof laying and hatchingevents were estimatedin the samefashion as events in Mayfield analyses.

1983 1984 1985 1986

Egg age at discovery Punta Cruz 14.0 + 9.62 (14) 5.4 + 9.86 (24) 3.6 + 5.59 (32) 3.4 + 6.25 (29) South Peninsula 24.6 + 6.20 (8) 6.2 + 10.16 (29) 6.1 + 9.31 (30) 8.7 + 6.25 (27) Total Luls Pefia 17.9 + 9.88 (22) 5.8 + 9.94 (53) 4.8 + 7.65 (62) 5.9 + 8.41 (56) Egg age at failure Punta Cruz 27.4 + 9.13 (5) 15.8 + 16.17 (13) 17.6 + 14.07 (20) 12.6 + 15.03 (16) South Peninsula -- 14.0 + 11.30 (21) 13.7 + 11.26 (17) 23.5 + 13.04 (13) Total Luls Pefia 27.4 + 9.13 (5) 14.7 + 13.17 (34) 15.8 + 11.80 (37) 17.5 + 14.98 (29) Chick age at discovery Punta Cruz 35.5 + 14.19 (8) 13.6 + 20.51 (18) 2.0 + 1.29 (13) 2.2 + 2.15 (15) South Peninsula 20.6 + 7.13 (9) 18.6 + 21.96 (15) 15.8 + 22.3 (16) 8.7 + 17.03 (18) Total Luls Pefia 27.6 + 13.28 (17) 15.9 + 20.99 (33) 9.7 + 17.18 (29) 5.7 +_12.92 (33) Chick age at failure Punta Cruz 25.0 + 0.0 (1) 18.7 + 20.13 (3) 31.0 + 0.0 (1) 21.8 + 25.24 (5) South Peninsula 35.7 + 40.04 (3) 38.0 + 10.0 (3) 9.0 + 0.0 (1) 42.2 + 20.36 (5) Total Luls Pefia 33.0 + 6.27 (4) 28.3 + 17.73 (6) 20.0 + 15.56(2) 32.0 + 24.15 (10)

South Peninsula were abandoned in 1986. These the depression.On my next visit to this site, chicks were the first hatched at these nest sites, the same White-tailed Tropicbird adult was although the nesting attemptsof these parents present,and the Red-billed Tropicbird chick was failed during the egg stagein previousyears. missing. Three weeks later this White-tailed Abandonments occurred when chicks were be- Tropicbird adult (a male) was incubatingan egg tween 40 and 60 days of age. In nearly all cases, at this site. White-tailed Tropicbirds used this 10-18 days passedbetween the last feeding of site in 1985 and 1986, but the site was reclaimed the chickby one parent (female or unidentified) by Red-billed Tropicbirdsin 1987 and 1988. and the last feeding by the secondparent (male or unidentified). DISCUSSION One Red-billed Tropicbird egg at South Pen- insula (of the two nests there that season)was Wetmore (1917) observed"six or eight trop- destroyedby White-tailed Tropicbirds in 1984. icbirdscircling about a rockypoint on CayoLuls A pair of White-tailed Tropicbirds was found Pefia," on 11 April 1912. A similar observation just inside the entrance of a crevice where a could have been made at Punta Cruz midmorn- Red-billed Tropicbird was incubating its egg. ing on 11 April, 1984-1988. Kepler and Kepler All three birds called loudly. I subsequently (1978) censusedboth coloniesin May 1971and found the Red-billedTropicbird egg broken and located 17 nesting pairs on their single visit to rolled out of the nest crevice,and the samepair each colony. My more frequent visits in sub- of White-tailed Tropicbirdsoccupying the nest sequent seasonsrevealed nests that started and depression. Within a month these White-tailed failed in periods of 2 days to 4 weeks, which Tropicbirds produced an egg, and the site was resulted in larger numbers of known nesting usedby White-tailed Tropicbirdsin subsequent attempts. Thus, different estimates of nesting years. population size are, at least in part, artifacts of During 1984 one Red-billed Tropicbird chick the frequency and number of observer visits. at Punta Cruz (of the only nest there that sea- The resultsof my study do not suggestdramatic son) was killed by White-tailed Tropicbirds.The changesin nestingpopulation size, and the Cayo 3-week-old chick was found with a severeopen Luis Pefia nesting population appearsto have wound on its back, displacedfrom the nest de- been fairly stablesince 1971 (and perhapssince pression; and a White-tailed Tropicbird adult, 1912) through 1988. However, there was a sub- with a blood-stainedbill and breast,occupied stantial turnover among individuals that at- October1991] TropicbirdBreeding Biology 919

TABLE5. Causesof nestingfailures during the egg stageand chick stage.Abbreviations: Abd = egg/chick abandoned;Agn = lossas a resultof agonisticinteractions between adults; Prd = destroyedby a predator; Ukn = disappearancedue to an unknowncause; Ald= eggfound rotten while still beingincubated; DH = died while hatching;Fid = nest flooded;Bkn = egg broken.

Egg stage Chick stage Abd Agn Prd Ukn Ald DH Fld Bkn Abd Agn Prd Ukn Fld Punta Cruz 1984 3 2 a 3 3 0 1 0 1 1 0 1 b 0 1 1985 9 4 2 3 3 0 0 0 0 0 1c 0 0 1986 9 6 a 0 2 1 0 0 0 2 a 0 2 c 0 1

South Peninsula 1984 8 0 7 5 1 0 0 0 0 1e 0 1 0 1985 I I 10 6 I 0 0 0 1 0 0 0 1 1986 1 2 2 3 2 1 2 0 3 a 0 0 1 f 1

* Includesone failure eachin 1984and 1986 due to Red-billed Tropicbirds,as inferred by the changein site ownership. b Due to rats. ß Due to hermit crabs. a PuntaCruz--one chickfed in the field, one hand-reared,both of thesefledged. South Peninsula--one chick hand-reared and fledged. ' Due to Red-billed Tropicbirds. • Emaciated,apparently starved but not abandoned.

tempted to nest in a given year. Approximately The estimatesof nestingsuccess from Aldabra 10 nest sites discovered in 1983 were not reused Atoll (seealso Phillips 1987), AscensionIsland, in 1984-1988. During 1984-1988 I discovereda and Cayo Luls Pefia in 1983 must be considered total 104 active nest sites and identified 110 maximabecause they unavoidablyexcluded any nesting pairs, but only about two thirds of this nests that started and failed between the number actually nested at Cayo Luls Pefia in monthly observer visits, whereas estimates at any single year (see Tables 2 and 3). Cayo Luls Pefia in 1984-1986are derived from Estimatesof nesting successcan be affected data collected at more frequent intervals. The profoundly by samplesize, frequency of sam- majorityof White-tailed Tropicbird nestingfail- pling, and method of calculation. Diamond ures at Cayo Luls Pefia occurredfrom 6 to 21 (1975) calculated overall nesting successfor days after laying, and it seemsreasonable to White-tailed Tropicbirds at Aidabra Atoll as the supposethat nesting failures at other locations number of successfulnests per number of nests would also tend to occur early in the cycle. of known fate. This resulted in 50.0% (7/14) for Therefore many nesting attempts would be 1967-1968, and 42.9% (12/26) for 1969. On As- missedby monthly sampling only. cension Island, Stonehouse (1962) calculated Monthly sampling for 1984-1986 cannot be overall nesting successof White-tailed Tropic- reliably simulated by simply consideringonly birds similarly for 1957-1959as 30.3% (249/821). recordsat intervals of 30 or 31 days.Although Prys-Jonesand Peet (1980) reported an overall somenests would be excludedcorrectly because nesting successof 46% for White-tailed Trop- they startedafter one visit and failed before the icbirds at Aidabra Atoll for 1976-1977. They next visit, I discoverednests by many different usedthe number of chickssurvived per number means. Often nest discoverywas the result of of eggslaid, which is the sameas (number laid several days of prior observationof displaying - number failed)/number laid, assumingthat birds, or due to observing birds entering or all large chickspresent at the end of the study leaving a crevice. This was especially true in period survived to fledging. By this method, 1984,when I alsodiscovered nests that had pre- overall nestingsuccess at CayoLuls Pefia would viously escaped my attention until the incu- be 44.1% (15/34) for 1983, 39.7%(27/69) for 1984, bating or brooding adult inside the crevicecalled 39.1% (27/69) for 1985 and 38.2% (26/68). Kep- loudly as I passed. ler and Kepler's (1978) data for 1971 would es- The Mayfield method yields realistic esti- timate 100%(17/17) overall nesting success,for mates of nesting success.It involves few as- their single visit. sumptions, yet accommodatesincomplete data 920 FREDCHARLES SCHAFFNEg [Auk, Vol. 108

seriesand allows use of nearly the entire data predation)accounted for mostnesting failures. set. This avoids reduction of sample sizes to At PuntaCruz, with a high nestingdensity and unreliably low levels. However, Mayfield es- smallnearest-neighbor distances, agonistic en- timatesare time sensitive,assuming equal prob- counters and abandonment accounted for most ability of successon all days of a given period nestingfailures in 1985and 1986.Tropicbirds (Johnson1979, Henslet 1985).Losses in my study are well-known for hostile interactions at their tended to occur early in a period (egg stageor nesting colonies (Stonehouse1962, Snow 1965, chick stage),and therefore the Mayfield esti- Harris 1966). The relatively constantnesting matesreported here are probablyslightly lower successduring the eggstage at PuntaCruz (Fig. than actual reproductivesuccess. 5), despitepredator control efforts, highlights Although a strictly controlled experimental causesother than predationfor nestingfailures examination of the effectivenessof predator- at that colony. Agonistic encountersbetween control effortswas impossible,the decreasesin adultswere especiallyimportant causes of nest- egg predation (Table 5) and trend towards in- ing failuresat PuntaCruz duringthe eggstage creasingnesting success at SouthPeninsula (Fig. in 1986,and they might alsohave lead to aban- 5), coincident with rat-control efforts in 1985- donments. Punta Cruz and South Peninsula had 1986, are encouraging.Moreover, at South Pen- similar numbers of nests, but the nests were insula, the number of surviving nestsmidsea- more crowded at Punta Cruz (170 m of coastline son in 1985-1988 was 50% greater than in pre- versus up to 420 m at South Peninsula). The treatment years (1983 and 1984) (Table 3). nearest-neighbor distancesat Punta C•'uz dur- Blackrats and land crabscan influencetrop- ing 1984-1986 were only 28-63% of those at ical island avifaunas(Atkinson 1985),and pre- SouthPeninsula (Table 1), which provided more dation by land crabs on birds and their eggs opportunities for nestersand would-be nesters hasbeen reported(Sprunt 1948,Amerson 1969, to interact and competedirectly for sites.Fur- and King 1973). Land hermit crabs destroyed thermore, although abandonmentwas an im- eggsof White-cheeked Pintails (Anasbahamen- portantcause of failure during the eggstage at sis)at Green Cay, St. Croix, in a fashion similar PuntaCruz in all years,it wasimportant at South to Gecarcinuspredation on tropicbirdeggs (Meier Peninsulaonly in 1984 (Table 5), the year in et al. 1989). At Cayo Luls Pefia rats were the which nestsites were most concentrated (along most important predatorson eggs,with Gecar- only 220 m of coastline),and nearest-neighbor cinusof lesserimportance. Many of the disap- distances smallest (Table 1). Chick abandon- pearancesclassified as "unknown" are presum- ment in 1986 also may have been related to ably due to predators,particularly rats, dragging changesin the foodresource that year,as these eggs away from nest sites. I observed adult abandonments were coincident with the almost White-tailed Tropicbirdsvigorously attacking completeabsence of an entirefamily of impor- and killing intruding Gecarcinus.I presume they tant food fishes,the Belonidae (needlefishes), attack rats as well. Some of the rat damage, in the regurgitationscollected that year(Schaff- therefore, must occur when adult tropicbirds ner 1988: appendix 1). leave their egg unattendedor after they aban- During May to July, 1984-1988, nonnesting don eggs.Thus there is someuncertainty as to adultsoften followed provisioning parents back the number of lossesdue to primary predation to the nestingcolony and visitedsites contain- versusthe number due to scavengingon aban- ing chicks,as well as emptysites in other crev- doned eggs. ices. On eight occasionsat three sites at Punta Gecarcinusare likely to destroyeggs only when Cruz during 1985-1987,as many as six birds at they can push the egg beyond the adult'sbill- one time entered a single crevice(Schaffner un- jabbing range or push the egg againsta rock to publ. data).Such piling-in visitsalways led to puncture it, and wait for the adult to abandon hostileinteractions. The first of thesepopular the nest. Both Coenobitaand Grapsusare pri- sites changed ownership and the seconddid marily scavengers,although Coenobitaoccasion- not; but the third site was never used. Most ally enter nestcrevices to feed on newly hatched emptysites visited by nonnestingadults in 1984- chicks,and parents seem unaware of them. 1987 were eventually nestedin by otherindi- At SouthPeninsula, with low nestingdensity viduals one to four seasonslater. Thus, a meth- and large nearest-neighbordistances, predation od of nest-site acquisition for White-tailed and "unknown" causes(which could have been Tropicbirdsat CayoLuls Pefia, and particularly October1991] TropicbirdBreeding Biology 921 at Punta Cruz, may have been to usurp a site Islands.U.S. Dep. Agric. For. Serv. Res.Pap. ITF- already visited or in use by other individuals 18. rather than to locate a completely new site. FULLER,M. g., H. H . OBRECHTIII, C. J. PENNYCUICK, Ironically, this occurred despite an apparent & F. C. SCHAFFNER.1989. Aerial tracking of ra- abundance of suitable crevices (Stonehouse 1962, dio-marked White-tailed Tropicbirds over the CaribbeanSea. Pp. 133-137in Proceedingsof the Snow 1965, Harris 1966, this study). Tenth International Symposiumon Biotelemetry, FayettevilleßArkansas (C. J. Amlaner, Ed.). Fay- ACKNOWLEDGMENTS etteville, Univ. Arkansas Press. FURNISS, S. 1983. Status of seabirds of the Culebra I thank the U.S. Fish and Wildlife Service and, in ArchipelagoßPuerto Rico. Colon. Waterbirds 6: particular, Sean Furniss for permissionto conduct 121-125. fieldwork at the Culebra National Wildlife Refuge. HARRIS,M.P. 1966. Factorsinfluencing the breeding SeanFurniss, John Taylor, and MaggieGriffen-Taylor cycle of the Red-billed Tropicbird in the Gala- graciouslyallowed use of their data for 1983.Culebra pagosIslands. Ardea 57: 149-157. refuge managers,John Taylor (1984-1985)and Henry HENSLER,G. L. 1985. Estimation and comparisonof Morales (1986-1988), provided valuable logistic as- functionsof daily survivalprobabilities using the sistancefor which I am grateful. Alexander Sprunt Mayfield method. Pp. 289-301 in Statisticsin or- IV and the National Audubon Society ResearchDe- nithology (B. J. T. Morgan and P.M. NorthßEds.). partment donated some much-neededequipment. I New York, Springer-Verlag. thank Ave Gaa, Janis Gonzalez-Martinez, Lisa Hilli, JOHNSON,D. H. 1979. Estimating nest success:the Jos• Rodriguez-V&lezand Anton Tucker, as well as Mayfield methodand an alternative.Auk 96: 651- Mark R. Fuller, Holiday H. Obrecht III, Colin H. Pen- 661. nycuick,and SeanKirkpatrick, who assistedwith var- KEPLER,C. B., & A. K. KEPLER. 1978. The seabirds of iousaspects of the fieldwork. I thank C. J. Pennycuick, Culebra and its adjacentislands, Puerto Rico. Liv- T. J. Herbert, J. C. Lee, O. T. Owre, C. R. Robins, and ing Bird. 16: 21-50. R. D. Morris for commentsand suggestionson a pre- KING, W.B. 1973. Conservation and status of birds vious draft of the manuscript.I am especiallygrateful of the central Pacific islands. Wilson Bull. 85: 89- to Ruud van Halewyn, Jim Wiley, and the late Bud 103. Owre for the inspiration to begin field studiesin the MEIER, A. J., R. E. NOBLE, P.M. MCKENZIE, & P. J. Caribbean.Most importantly, I expressmy sincerest ZWANK. 1989. Observationson the nestingecol- thanks and appreciationto Colin Pennycuickfor his ogy of the White-cheeked Pintail. Caribbean J. tirelessencouragement, tolerant friendship, and phe- Sci. 25: 92-93. nomenalgood cheerduring the courseof this study MILLER,n.M., & D. n. JOHNSON.1978. Interpreting and others.Partial funding for this study camefrom the resultsof nesting studies.J. Wildl. Manage. the University of Miami Department of Biology and 42: 471-476. a Maytag Fellowshipin Biology.I alsothank Mark R. PENNYCUICK,C. J., & N. C. KLINE. 1986. Units of Fuller, the U.S. Fish and Wildlife Service Patuxent measurementfor fractal extentßapplied to the Wildlife ResearchCenter, and the Seaand Sky Foun- coastaldistribution of Bald Eagle nests in the dation, for partial support during a portion of the Aleutian Islands,Alaska. Oecologia: 86: 254-258. final writing period. ßF. C. SCHAFFNER,M. g. FULLER,H. H. OBRECHT

LITERATURE CITED III,& L. STERNBERG.1990. Foragingflights of the White-tailed Tropicbird Phaethonlepturus: radio- AMERSON,A.B. 1969. Ornithology of the Marshall tracking and doubly-labelled water. Colon. Wa- and Gilbert islands. Atoll Res. Bull. 127: 1-348. terbirds 13: 96-102. ATKINSON,I. A. E. 1985. The spread of commensal PHILLIPSßN.J. 1987. The breeding of White-tailed speciesof Rattusto oceanic islands and their ef- Tropicbirds Phaethonlepturus at Cousin Island, fects on island avifaunas. Pp. 1-217 in Conser- Seychelles.Ibis 129: 10-24. vation of island birds (P. J. Moors, Ed.). ICBP PRYS-JONES,R. P., & C. PEET. 1980. Breeding peri- Tech. Publ. No. 3. odicity, nesting successand nest site selection DEWEY, R. A., & D. W. NELLIS. 1980. Seabird research amongRed-tailed Tropicbirds Phaethon rubricauda in the Virgin Islands.Pp. 225-452 in Trans. 45th and White-tailed Tropicbirds P. lepturuson Ai- North American Wildlife and Natural Resources dabra Atoll. Ibis 122: 76-81. Conference.Washingtonß D.C., Wildl. Manage. SCHAFFNER,F. C. 1988. The breeding biology and Inst. energeticsof the White-tailed Tropicbird (Phae- DIAMOND,A.W. 1975. The biology of tropicbirdsat thon lepturus)at Culebra, Puerto Rico. Unpubl. Aidabra Atoll, Indian Ocean. Auk 92: 16-39. Ph.D. dissertation. Coral Gables, Florida, Univ. EWEL,J. J., & J. L. WHITMORE.1973. The ecological Miami. life zones of Puerto Rico and the U.S. Virgin 1990a. Food provisioning by White-tailed 922 FREDCHARLES SCHAFFNER [Auk, Vol. 108

Tropicbirds:effects on the developmental pattern VAN HALEWYN, R., & R. L. NORTON. 1984. The status of chicks. Ecology 71: 375-390. and conservation of seabirds in the Caribbean. 1990b. Feed size and feeding periodicity in Pp. 169-222 in Status and conservationof the pelagicbirds: notes on methodology.Colon. Wa- world's seabirds(J.P. Croxall, P. G. H. Evans, and terbirds 13: 7-15. R. W. Schreiber, Eds.). ICBP Technical Publica- --, • P. K. SWART. 1991. Influence of diet and tion No. 2. environmental water on the carbon and oxygen Voss, G. 1976. Seashore life of Florida and the Ca- isotopicsignatures of seabirdeggshell carbonate. ribbean. Miami, Florida, Banyan Books,Inc. Bull. Marine Science 48(1): 23-38. WESTERMAN,J.H. 1953. Nature preservationin the SIEGEL,S. 1956. Nonparametric statisticsfor the be- Caribbean. A review of literature on the destruc- havioral sciences. New York, McGraw-Hill. tion and preservationof flora and fauna in the SNOW,D.W. 1965. The breeding of the Red-billed Caribbean area. Publ. Found. Sci. Res. in Surinam Tropic Bird in the GalfipagosIslands. Condor 67: and the Netherlands Antilles No. 9. 210-214. WETMORE,A. 1917. The birds of Culebra, Porto Rico. SPRUNT,A., JR. 1948. The tern colonies of the Dry Auk 34: 51-62. Tortugas Keys. Auk 65: 1-19. STONEHOUSE,B. 1962. The tropic birds (Genus Phae- thon) of Ascension Island. Ibis 103: 124-161.