Competition for Nesting Space Among Peruvian Guano Birds

COMPETITION FOR NESTING SPACE AMONG PERUVIAN GUANO BIRDS

DAVID CAMERON DUFFY PercyFitzPatrick Institute of AfricanOrnithology, University of CapeTown, Rondebosch7700, South Africa

ABSTRACT.--Inthe mid-1940's, Peruvian managers greatly increased the nesting space available to the three principal surface-nestingspecies of the Peruvian CoastalCurrent: the Guanay Cormorant (Phalacrocoraxbougainvillii), the Peruvian Booby (Sula variegata),and the Peruvian Brown Pelican (Pelecanusoccidentalis thagus). The combined populations of these three speciesincreased from 8 to 20 million birds. The annual rate of increase of the population rose from 8 to 18%.The three speciesappear to have evolved in the face of a shortage of nesting space.They have not diverged in their respectivebreeding seasons.Each species has habitat preferences for nesting, but the overlap is great. The booby and cormorant compete through a "scramble" to occupy space before it is settled by the other species. Neither can displacethe other from nest sites.The pelican is dominant over the other two in aggressiveinteractions and frequentlyusurps their nests.Pelicans are apparentlyconfined to nesting in level areas, whereas the other two speciescan nest on a greater range of gradients. Despite the factsthat nesting spaceis limited and that its scarcityhas a demonstratedeffect upon the combinedpopulations, interspecific competition for nestingspace was difficult to document. Interspecific aggressive interactions were few and involved only a small percentageof the three populations. The individuals most affected by competition, those denied nesting space,were displacedfrom the area of competitionand were thus lessaccessible for study. Received7 June1982, accepted14 January1983.

FOR several millennia, seabirds have nested to nest on cliffs during the 19th century but on islands off the coast of Peru (Hutchinson that boobies outcompeted pelicans when hu- 1950). The Guanay Cormorant (Phalacrocorax mans visited the islands. Hutchinson (1950) bougainvillii),Peruvian Booby (Sula variegata), concludedthat, before exploitation by humans, and Peruvian Brown Pelican (Pelecanusocciden- Guanay Cormorants lost out in competition talis thagus)are the principal nesting species. with boobies and pelicans, but, with the dis- Ten other speciesalso nest (Murphy 1925, 1936; turbance caused by human exploitation of Galarza 1968) but tend to occupyburrows, caves, guano deposits,Guanays were competitively or other habitats not used by the three major superior.Vogt (1942) appearedto believe that species. differencesin tolerancesto high temperatures There are fewer than 40 guano islandsalong and human disturbance, as well as the use of the Peruvian coast;many are small, and only a cliffsby boobies,led to habitatpartitioning with portion of the surfacesof the remainder ap- minimal interspecific interactions. Nelson pears suitable for nesting becauseof human (1978) believed that, although boobies could disturbance,heat, or ectoparasites(LaValie 1918, displacecormorants and pelicanswere domi- Vogt 1942, Duffy 1983a). Where the birds do nant over both cormorants and boobies, few nest,densities are very high, generally 2-3 nests birds came into direct interspecific contact and per m2 (Vogt 1942;Table 1). competition. He suggested(p. 589) that "... Murphy (1925), Vogt (1942), and Nelson meaningful competition,if it occurs,takes the (1978) have suggestedthat spacesuitable for form of 'swamping our'--one species,once es- nesting was limited comparedwith the num- tablished and undisturbed by man, more or less bers of birds that could be supported by the continuously occupying traditional areas and high productivityof the Humboldt or Peruvian merely by its presenceexcluding others." This, CoastalCurrent. Opinions differ as to how space however, does not explain how dominant was partitioned. Murphy (1925) suggestedthat specieswere excludedfrom nesting in areasoc- competition with pelicans had forced boobies cupied by subordinatespecies.

680 The Auk 100: 680-688. July 1983 July 1983] GuanoBird Competition 681

In this paper I examine the assumption that likely to have been representative. Atypical wind spacewas limited and investigatehow the three conditions are associatedwith E1 Nifio phenomena major speciespartition available nesting space (Vogt 1942), which usually produce massdesertions with referenceto interspecificinteractions and of nests(Vogt 1940). behavioral and morphologicallimitations. Species'interactions.--I originally planned to watch interspecificinteractions over set periods of time, but METHODS interactions were so rare that I made most observations on an opportunisticbasis during the 7 months If a resourcelimits a population, then an increase of fieldwork. I noted the speciesand whether antag- in the resource should produce an increase in the onists were defending nest sites or were nonbreed- population.Birds and other mobile organismsare not ers.A bird was consideredto have won if it displaced usually conduciveto testing this hypothesis,but Pe- the other from the scene of the conflict. ruvian management practicesallow us to examine Although the three specieshave elaborate reper- suchan increasein nesting space. toires of agonisticbehavior (Vogt 1942,Nelson 1978; Jordan and Fuentes (1966) estimated the combined pers. obs.), I used interspecific jabbing--with or populations of the three speciesbetween 1909 and without contact--asthe only behavioral criterion for 1964 basedon guano harvests.In the 1940's,follow- an interaction, as more subtle aggressiveor territorial ing a sharp decreasein the population during a series displays do not necessarilyconstitute communica- of E1 Nifio events,when upwelling stopsand food tion between species. becomesscarce (Vogt 1940), Peruvian authoritiesbe- Settlementpatterns.--On Mazorca, I observedwhere gan fencing off coastal headlands in the hopes of each speciesfirst establishednest sites in relation to providing more nesting space,a larger population, wind, slope,substrate, and "cliff" edges(where "cliff" and a greater production of guano. Here, I compare was any edge with a vertical drop of over 1 m). population levels and the percentageof annual in- I monitored four booby nesting areas containing creases before and after the creation of the new sites. both cliff and interior nest sitesto seeat which type Studysites.--Most fieldwork was conducted on Isla of nest site eggswere laid first. West Platform had a Mazorca (I 1.6 ha; 1Iø23'S, 77ø45'W) between Septem- maximum of 42 sites(26 edge); Lee Platform, 43 (14); ber 1977 and March 1978. Other observations were East Platform, 58 (12); and Barranca, 79 (12). I waited made on Isla Guafiape Norte (25 ha; 08ø32'S,78ø58'W), for turnovers or shifts in positions to determine nest Isla Macabi (8 ha; 07ø47'S, 79ø30'W), and Islas Balles- contents.No birds flushed from their nests during tas (32 ha; 13ø44'S,76ø24'W) in February and March these observations, so I believe I had minimal dele- 1979.All of theseislands are protectedby guardsand terious effects. remain essentiallyundisturbed during the breeding To determinewhether nestingsuccess was greater season. in cliff-edge or interior nests,I comparedthe number Nest characteristics.--Nestshape and materiMs can of 4-10-week-old booby young in nests in the two determine such things as density, permissibleslopes habitatson Mazorca,Macabi, GuafiapeNorte, and the of underlying substrates,and the number of young Ballestasislands. Only nests with young were sam- that can be reared in the nest. On Mazorca, I mea- pied. This would have produced a bias if nests in sured nest diameter, outside height, and distanceto either habitat were more likely to suffer complete nearest neighbor for the three species. After the nesting failure. This did not appear to be the case. breedingseason, using an inclinometerattached to a Overlapbetween species.--Guards resident on the is- 16-cm straightedge,I measuredthe maximum Mope landsmap birds and their stateof breedingfortnight- under nests. Using a 2-m straightedgeand inclino- ly. I chosethe islandsof Mazorca and Macabi Grande, meter, I determined the Mope of ground adjacentto becauseI was most familiar with their topography; nests, using the mean of three measurements.Den- both are small (11.6 ha and 5.4 ha, respectively)and sity was computed by counting all complete nests relatively flat. within 2 X 2-m squaresfor 38 cormorant, 49 booby, I partitioned the maps from Macabi into 450 quad- and 2 pelican samples. rats, each representing approximately 121 m2 (11 X Exposure to wind and nest temperature are be- I1 m), and did the samefor Mazorca, dividing it into lieved to be important in partitioning nesting space 530 quadrats of approximately 225 m2 (15 X 15 m). (Vogt 1942). To investigatethis, I measuredthe tem- For each year of available records (Mazorca: 30 nest- peratures of guano crusts on the rims of occupied ing seasons;Macabi: 31 seasons),I chosethe census nestson Isla Macabi on 20 February 1979 over 2 h. showingthe greatestarea of nestswith eggsin order The sky was partially overcast,and there were south- to plot the quadrats used by each species,because erly winds of 20 kph and a shaded ambient air tem- competitive pressure should be most severe then. perature of between 24 and 26øC. These conditions From these, I obtained maps of frequency of use for are typical of breeding-seasonconditions (e.g. Vogt each island. Three decades of records reduce the 1942;pers. obs.),so that resulting temperatureswere mapping errorsthat occurduring any one year. I also 682 DAVIDCAMERON DUFFY [Auk, Vol. 100

to 20 30 40 50 60 70

YEARS 1909- 1977

Fig. 1. Three-year running meansof guano bird numbersbased on guano harvests(after Duffy 1980). calculatedthe percentageof area used by one species bies, which had only low rims of guano and that was never used by a secondspecies and the per- pebbles.Pelicans scraped depressions into mats centageof quadratsused by two or more speciesover of feathers and guano, which they gathered the years. from outside their colonies. Pelicans also usurped booby and cormorant nests but rap- RESULTS idly reduced them to shallow depressions. Pelicans tended to nest on fiat surfaces; boo- Did nest spacelimit the popuIation?--Between bies and cormorants used areas with consider- 1909 and 1940, the combined populations of able slopes(Table 1), and boobieseven nested the three speciesincreased as more islandscame on the ledges of perpendicular cliffs. Boobies under the protection of Peruvian authorities and cormorants showed no difference in the (Jordanand Fuentes1966). The populationthen overall angles of nesting slopes (excluding leveled off at about 8 million birds (Fig. 1). The cliffs), but slopes of nest sites were twice as mean increase per year was 8% (SE mean = 4.2; steep for cormorants.Cormorants and pelicans n = 23; years of E1 Nifio excluded). After 1946, nested at densities of around three nests/m 2, with the creation of breeding sites free of ter- whereas booby nests were only two-thirds as restrial predators on the mainland, the popu- dense (Table 1). lation of the three species rose as high as 20 The three speciesappeared to differ in their million and the percentage annual increase temperature tolerances (Table 2). Cormorants more than doubled to 18% (SE mean = 4.5; n = preferred the coolest sites and pelicans the 18; excluding E1 Nifio years). warmest. The differences were significant (ex- More recently, becauseof overfishing, the tension of the median test, X2 = 8.758, df = 2; bird populations have fallen sharply (Nelson P < 0.01; Siegel 1956) and agreed with earlier 1978, Tovar 1978, Valdivia 1978, Duffy 1980), work of Vogt (1942). but the two islands studied here were both cov- Species'interactions.--During the 6 months of ered with nesting or roosting birds during at fieldwork on Mazorca and 3 weeks on other least part of my study, so conditionswere pre- islands, I saw only 193 interspecific aggressive sumably similar to those in effect when space interactions on land. There were tens or was in short supply throughout the islands. hundreds of thousandsof nesting birds on the Nest characteristics.--Nest structure and loca- islands, but they nested in monospecific sub- tion differed among the three species(Table 1). colonies, and very few pairs came into inter- Cormorants made substantial nests of guano, specificcontact (Nelson 1978). Watchesat con- feathers, and debris. Their nests were twice as tiguous subcolonies showed that aggressive high on the downhill side as were nestsof boo- interactions were primarily directed at conspe- July 1983] GuanoBird Competition 683

TABLE1. Nest and nest-sitecharacteristics of the three major nesting specieson Isla Mazorca.

Guanay Cormorant Peruvian Booby Brown Pelican œ _+ SD n œ _+ SD n œñSD n

Overall slope 21ø ñ 7ø 100 21 ø_+ 13 ø 105 2 øñ 2 ø 20 Microsite slope 12ø _+12 ø 400 6 ø + 6 ø 420 2 ø -+ 2 ø 26 Density (n/m 2) 3.0 _+0.5 38 1.9 _+ 0.6 49 2.75-3.0 2 Height of downhill side of nest 17 _+4 cm 100 9 -+ 3 cm 100 -- -- Nest diameter 32 + 4 cm 100 31 _+ 4 cm 100 39 ñ 4 cm 11 Nest area (•rr2) 0.08 m 2 -- 0.07 m 2 -- 1.2 m 2 -- Area occupied by two young 0.06 m2 -- 0.05 m 2 ------

cifics, even if pairs of another specieswere ad- Settlementpatterns.--Cormorants and boobies jacent. nested at approximately the same time of the The actual sequenceof events in an interspe- year, pelicansslightly later (Vogt 1942). Strong cific interaction varied greatly. Cormorants or correlationsamong the monthly frequenciesof boobies on nests would often jab at a passing breeding of the three species(data from Vogt bird with no preliminary display. More fre- 1942; n = 12; boobies and cormorants, r• = 0.90, quently, jabbing was precededby "yes head- P < 0.01; boobies and pelicans, rs= 0.71, P < shaking" (Nelson 1978) by boobies or extend- 0.05; pelicans and cormorants, rs= 0.89, P < ed-neck threats by cormorants. Both these 0.01) indicate that the three specieshave not behaviors occasionally led to "sky pointing" diverged in their breeding seasons. (Nelson 1978), which itself might precede or All three speciesnested in monospecificsub- follow jabbing. Jabbing occasionallyescalated colonies,but the mechanismsby which neigh- to prolonged fighting in intraspecificencoun- borhoodswere formed differed. Although there ters, but most int•rsvecific aggressions were was a degree of local synchrony among Peru- brief. vian Boobies(Nelson 1978), a pair's egg laying Pelicans rarely jabbed and usually ignored did not depend heavily on the activities of jabbing by the two smaller species.Pelicans neighbors. Breeding began slowly; pair for- would walk obliviously over a jabbing booby mation and nest building took more than 1 or cormorant defending its nest or continue month (Vogt 1942, Nelson 1978;pers. obs.). Egg preening while surrounded by jabbing cor- laying within a local group spread over more morants. than I month (Fig. 2, see also Nelson 1978:fig. Pelicanswere clearly dominant over the oth- 263). The first sites occupiedby boobies were er two species.Pelicans displaced another bird in all encounters with boobies (n = 18; P < 0.001; binomial distribution) and in 99% (n • TABLE 2. Numbers of nests, the rims of which were 107) of encounters with cormorants. In inter- a particular temperature, of the three species.Mea- actions between boobies and cormorants (n = surements were made on Isla Macabi on 20 Feb- 68), boobieswere the displacers("won") in 31% ruary 1979; ambient air temperature was 24-26øC. of the encounters, and cormorants won 10%, but most (59%) ended as standoffs. The behav- Temperature Guanay Peruvian Brown (Cø ) Cormorant Booby Pelican ioral statesof the antagonistswere important. Cormorants defending a territory against boo- 25 ! ! 0 26 6 2 0 bies without territories displacedthem or tied 27 10 7 ! in all seven interactions (P = 0.008, binomial 28 9 5 6 distribution). With the roles reversed (n = 11), 29 3 2 2 boobiesdisplaced cormorants or tied in all en- 30 3 6 7 counters (P < 0.001). When both species were 31 1 0 2 34 0 ! 0 on territories (n = 17), standoffs occurred 95% Mean 27.6 ø 28.2 ø 29.2 ø of the time. When neither was defending a ter- SE mean 0.24 0.39 0.28 ritory (n = 33), standoffs occurred 66% of the n 33 24 18 time, boobies won 21%, and cormorants 12%. 684 DAVIDCAMERON DUFFY [Auk, Vol. 100

80 WEST PLATFORM LEE PLATFORM

40 III--• I I i I

0 I i I I I/// I I W 29 7 14 28 29 7 14 28 NOV DEC NOV DEC

EAST PLATFORM BARRANCA Z W 80

n• W -

I /111// il I I I I I I i i I 0 29 7 14 28 26 28 ;506 ? NOV DEC NOV DEC

EDGE ---CENTER

Fig. 2. Comparativesettlement at edge and interior rtestsites by Peruvian Boobiesat four study sites(see Methods for samplesizes). on the edges of drop-offs and ledges of cliffs. seemed to begin whenever food became suffi- Later pairs nestedaway from the edges(Fig. 2). ciently abundant that a number of males re- Pairs rarely were able to establishterritories in mained on the island insteadof departing with occupied areas. Neither habitat consistently the fishing flock in the morning. These birds producedmore nestlingsper active nest on four stayed to defend sites and to build closely islands (Table 3). Nesting groups of boobies packed nests. Males gathered next to each oth- ranged from 2 or 3 pairs on isolatedcliff-ledges er to form a colony nucleus on a windswept to tens or hundreds of thousands nesting on slope. Gradually other males set up territories smoothly sloping hillside. on the periphery, so that the colony expanded In contrast, nesting groups of Guanay Cor- witl• the newestnests always on the outside morants always exceeded1,000 pairs (Vogt, cit- (Vogt 1942). Pairs never settled into occupied ed by Hutchinson 1950, said 10,000 was the areas. minimal size). The breeding of cormorants Pelican nesting groups were generally small, July1983] GuanoBird Competition 685

::• 26-50% [] 51-75ø/o FREQUENCY OF NESTING ß 76- I00%

Fig. 3. Frequencyof nesting by the specieson Isla Mazorca (530 quadrats;30 breeding seasons). ranging from 10 to several hundred pairs. Sub- the combined populations. Sufficient food and colonies were highly synchronized; groups other resources were available to support the made up of mated pairs arrived overnight and subsequentpopulations. settled in the midst of nesting boobiesand cor- Several millennia or more (Hutchinson 1950) morants, expropriating their nests. of limited nesting spacewould have produced Overlap betweenspecies.--While the three a strong interspecificcompetition for accessto speciesshowed preferences for certain areas breeding sites. Alternatively, the speciescould year after year on Mazorcaand Macabi (Figs.3 have diverged to reduce competitionby choos- and 4), the overlap was high (Table 4), sug- ing different habitatsor by nesting at different gesting that most areasare suitable for nesting times of the year. by more than one species.On Mazorca, exclu- The speciesdiffer in their apparent nesting sive use of nesting areas was only 19% for cor- preferences.Boobies prefer edges,presumably morants, 21% for boobies, and 4% for pelicans. to facilitate take-offsand landings. Pelicansre- On Macabi, the levels were 21% for cormorants quire flat areas, perhaps becausetheir large and zero for the other two. wingspansdo not permit safelandings on edges or steep slopes.Their nest scrapesmay also be DISCUSSION insufficientto retain eggsand young on slopes. Cormorantsseem to prefer the middle portions The increasein safenesting spaceduring the of windy hillsides, perhaps to facilitate take- 1940's resulted in an increase in the combined offs, to cool their dense colonies, or to allow populations of the three species and in their large continuous colonies to form the "pha- annual rate of increase. This clearly demon- lanx" effect of cormorant breaks against aerial strated that usable nesting space had limited predators such as the Kelp Gull (Larus domini-

TABLE3. Comparisonof the mean number of nestling Peruvian Boobiesin nestson cliffs and on flat areas at four nesting islands.

Cliff Flat Island Month Age 27 n 2 n Mazorca Feb. 1978 6 weeks 1.97 360 2.03 377 Mar. 1978 10 weeks 1.78 348 1.84 154 Guafiape Norte Feb. 1979 4-8 weeks 1.63 65 1.71 65 Macabi Feb. 1979 8 weeks 1.86 29 1.77 57 Ballestas Mar. 1979 8 weeks 1.47 36 1.30 50 686 DAVIDCAMERON DUFFY [Auk,Vol. 100

CORMORANT

BOOBY

[] o% [] 1-25% ::[] 26-50% [] 51-75% [] 76-100%

FREQUENCY OF NESTING Fig. 4. Frequencyof nesting by the three specieson lsla Macabi (450 quadrats;31 breeding seasons). canus).These differences between the species, the two smaller species.As is common in such however, are not sufficient to prevent large cases(Morse 1974), however, the pelican has a overlaps between the species. narrower range of acceptable nesting habitats All three speciesnest at approximately the than do the other two speciesso that usurpa- sametime of the year. Initially, all three settle tion of the nests of cormorants or boobies can on their preferred nesting habitatsuntil these only occur on level ground. Such terrain is are full. Cormorants begin nesting on the scarceon most guano islands. Even so, pelicans slopes,boobies on the edges.Their coloniesex- have rarely filled all the available flat areason pand toward the habitat of the other. The fail- any one island and have not even nested on ure of either speciesto displacethe other when many islands every year (Fig. 4). This suggests both are defending nest sites (ties = 95%) sug- that some other factor, such as food, keeps the geststhat when they meet, they stop, neither population of pelicansbelow that level at which speciesdisplacing the other. Occupancy of a all the available nesting spacecould be used. particular area depends on which species Nelson (1978: 589) suggested that a species reaches it first. This depends in turn on prox- excludes the others from "traditional" areas imity to the initial nestingareas and the num- "merely by its presence."Occupancy of an area, bers of each speciesseeking to nest. however, would be interrupted annually or In contrast,pelicans are clearly dominant over biennially by guano extraction; by desertions July 1983] GuanoBird Competition 687

TABLE4. Percentage of nesting area ever used by namics of the populations than does limited Species A overlapping with that ever used by nesting spaceon any one island. Even before Species B on Isla Mazorca (30 breeding seasons) the expansion of nesting space in the 1940's, and Isla Macabi (31 seasons). limited spaceacted as a damper on the increase Species A of the population, not as an absolutelimit. Cot- The importanceof competitionbetween bird SpeciesB Booby morant Pelican speciesremains a sourceof controversy.It is to be hoped that studies such as this one serve to Isla Mazorca Booby -- 79% 36% Cormorant 71% -- 83% illustrate the complexitiesthat must be under- Pelican 20% 28% -- stood before the issue can be resolved. IslaMacabi Booby -- 80% 74% Cormorant 100% -- 100% ACKNOWLEDGMENTS Pelican 17% 14% -- I would like once again to thank those mentioned in my dissertation(Duffy 1980) for their help and encouragement. This study was funded by The Na- caused by food shortages during El Nifios, tional Science Foundation (DEB-7716077), The Or- which occur at approximately 5-yr intervals ganization of American States, The International Committee for Bird Preservation (Pan American), and (Duffy 1983b);and by desertionscaused by out- the Biology Department of Princeton University. ! breaks of ticks (Duffy 1983a). Partitioning of would also like to thank M. Gochfeld, J. Wiens, and nesting areas under present conditions is thus an anonymous reviewer for comments on the manu- an almostannual event. Even before guano ex- script. traction, repartitioning would have occurred once or twice a decade during E1 Nifios. Com- LITERATURE CITED petition for spaceis thus much lessstatic than suggestedby Nelson (1978). ASHMOLE,N.P. 1963. The regulation of numbers of Studying the effectsof competition on long- tropical oceanic birds. Ibis 103: 458-473. lived, mobile animals in complex environ- DUFFY,D.C. 1980. Comparative reproductive be- havior and population regulation of seabirdsof ments can be difficult. For example, both the Peruvian coastalcurrent. Unpublished Ph.D. Mazorca and Macabi have been covered by dissertation. Princeton, New Jersey, Princeton nesting birds over the years. With large areas Univ. suitable for both cormorants and boobies, oc- 1983a. The ecology of tick parasitism on cupancyby one speciesshould have prevented densely nesting Peruvian seabirds. Ecology 64: nestingby the other. Exceptfor occasionalbouts 110-119. of sparring by pairs at borders between the 1983b. Environmental uncertainty and species,however, there were almost no inter- commercial fishing: effects on Peruvian guano specific interactions, aggressiveor otherwise. birds. Biol. Conserv. 26: 227-238. Agonistic encounterswere a poor measure of GALARZA, N. 1968. Informe sobre estudios ornito- logicosrealizados en el laboratorio de La Puntil- competition. Ia (Pisco) en Setiembre de 1965/66. Inf. Esp. In- From a single year's study of the species' stit. Mar Perd--Callao 31: 1-20. nesting distributions, one might conclude that HUTCHINSON,G. E. 1950. Survey of contemporary cormorants prefer slopes and boobies prefer knowledge of biogeochemistry.3. The biogeo- edges.Without detailed mapsof nesting distri- chemistry of vertebrate excretion. Bull. Amer. butions on Mazorcaand Macabi over the years, Mus. Nat. Hist. New York 96: 1-554. one might never realize the extent of overlap. JORDAN,R., & H. FUENTES.1966. Las poblaciones de Measuring competitionby assessingits effect aves guanerasy su situaci6n actual. Inf. Instit. on the populations of the three speciesis also Mar Peril--Callao 10: 1-31. likely to be unproductive. Birds unable to ob- LAVALLE,J.A. 1918. Estudiosobre los factoresque influyen sobrela distribuci6n de los nidos de las tain nesting spacewould either go to lesssuit- aves productores del guano. Mere. Comp. Ad- able habitats, which were not full, or would mora. Guano 9: 207-213. not breed (Ashmole 1963). The massive adult MORSE, D. H. 1974. Niche breadth as a function of mortalities during E1 Nifio phenomena (Mur- social dominance. Amer. Natur. 108: 818-830. phy 1925,Vogt 1942,Jordan and Fuentes 1966) MURPHY, R. C. 1925. Bird islands of Peru. New York, are likely to have more of an effect on the dy- Putnam. 688 DAVIDCAMERON DUFFY [Auk, Vol. 100

--. 1936. Oceanic birds of South America. New VALDIVIA,J. 1978. The anchoveta and E1Nifio. Rapp. York, Amer. Mus. Nat. Hist. P.-v. Reun. Cons. int. Explor. Mer 173: 196-202. NELSON,J. B. 1978. The Sulidae: gannets and boo- VOGT,W. 1940. Un depresi6necologica en la costa bies. Oxford, University Press. del Perfl. Bol. Comp. Admora. Guano 16: 307- SIEGEL,S. 1956. Non-parametricstatistics for the be- 329. havioral sciences. New York, McGraw-Hill. 1942. Aves guaneras.Bol. Comp. Admora. TOVAR,H. 1978. Las poblacionesde aves guaneras Guano 18: 1-132. en los ciclos reproductivosde 1969/70 a 1973/ 74. Inf. Instit. Mar Perf--Callao 45: 1-13.