THE AUK A QUARTERLY JOURNAL OF

VOL. 100 JULY 1983 No. 3

"REPLACEMENTS" OF LANDBIRD SPECIES ON , MEXICO

JOSEPHR. JEHL,JR. • AND KENNETHC. PARKES2 •Hubbs-SeaWorld Research Institute, 1700 S. ShoresRoad, San Diego, California 92109 USA, and 2CarnegieMuseum of NaturalHistory, Pittsburgh, Pennsylvania 15213 USA

ABSTRACT.--Theendemic dove Zenaidagraysoni of SocorroIsland, an oceanicisland in the Pacific south of BajaCalifornia, became extirpated between 1958 and 1978, and the endemic Mimodesgraysoni diminished during the same years and was approachingex- tinction by 1981. The mainland Z. macrourabecame establishedon Socorro between (probably) 1971 and 1978, and the mainland poly- glottosbetween 1978 and 1981. Superficially an instanceof replacementof island endemics by mainland relatives through competitiveexclusion, the caseis actually more complicated. We argue that exterminationof the endemicswas probably causedby by fetal introduced in 1957 or later. There is no evidence of any actual competition between the pairs of species,and their preferred overlap only slightly. We attribute the nearly concurrent establishment of the two invading species,known to have occurred frequently as casualvisitors to Socorroand other islands, largely to the provision of fresh water made available as a result of human settlement.Our data show that simple comparisonof island specieslists compiled at intervals is an inadequate base for theories of island speciesturn- over. Received24 May 1982, accepted4 October1982.

"Whena speciesnative to a largearea is successfully 1969,Power 1972,Hunt and Hunt 1974,Lynch introducedinto a new small area, the relatedspecies and Johnson 1974, Jones and Diamond 1976, whichis nativein thisarea and with whichthe former Marshall et al. 1982) and applied to conserva- comesinto competitionis soonsupplanted."--J. Grin- tion problems (Diamond 1975, Temple 1981). It nell (1925). is important to recognize that the data on which "Thingsare seldomwhat they seem."--Little But- tercup. the model is basedare simply faunal lists made at different periods,of which a disproportion- RECENTinterest in island owes ate number have been based on the ornitho- its origins principally to the work of Mac- logical literature. As Gilbert (1980) has shown, Arthur and Wilson (1967). They proposed (1) however, there are many reasonsfor doubting that the size of an island's fauna is predictable the validity and application of thosedata. Fur- from knowledge of its area and its distancefrom ther, recentpaleontological studies (e.g. Pregill other source faunas and (2) that, although and Olson 1981, Olson and James 1982) have speciescomposition might be in a state of flux castserious doubt on the conceptof an equilib- as a result of the establishment of newly colo- rium number of speciesfor any island. nizing speciesand the extinction of former in- Although the causesof turnover are not an habitants ("turnover"), the number of species explicit part of the MacArthur-Wilson model, comprisingthe fauna tends to remain relative- it is clear that thoseauthors regard competition ly constant ("equilibrium"). as an important causeof extinction (MacArthur The MacArthur-Wilson model has been 1972). It follows, therefore, that the lossor gain widely extended and refined (e.g. Diamond of a speciesmay be directly related to a change

551 The Auk 100: 551-559. July 1983 .!

FRONTISPIECE. (Zenatdagra.¾soni) in its natural on SocorroIsland, 10 March 1957; elevation550 m. Photographby Reid V. Moran. 552 JE•L ANDPARKES [Auk, Vol. 100 in the statusof another, often closely-related, error. Here we review the histories of the en- form. demicsand the invaders, and we argue that the In this paper we documenttwo casesof turn- competitive exclusionhypothesis is untenable over that involve replacements in an island for Socorro Island. avifauna by closely related taxa. These ex- changesare so striking that they would appear THE DOVES to be "textbook" examplesof competitive ex- clusion. Field data obtained during the period The Socorro Dove (Zenaida graysoni)is a of replacement, however, show that the com- highly distinctive . Larger than the well- petitive exclusionhypothesis is untenable. known Mourning Dove (Z. macroura)of the SocorroIsland, largest of the Revillagigedo North American mainland, with ruddy rather Islands of Mexico (ca. 210 km2), lies approxi- than buffy underparts,a different pattern of tail mately 400 km south-southwestof the tip of markings, and different vocalizations, it has the BajaCalifornia peninsula.Formed by vol- been treated by some authors as no more than canic eruptions,it is a true oceanicisland, nev- a race of the Mourning Dove, of which it is an er having been connectedto the mainland. The obvious derivative. It is perhaps 15-20% heavi- resident avifauna, omitting seabirds but in- er than the Mourning Dove, and, in accordance cluding one heron, consistsof only 10 forms, with a frequent pattern of variation in island all of which are endemic at the subspecies(7), , has a longer bill and much longer (29%) species(2), or generic (1) level. The bird life of tarsi than the mainland dove, but shows little the island is well known as a result of several differencein wing and tail length. The Socorro early expeditions,long-term studiesby B. H. Dove was discoveredin 1867by Edward Gray- Brattstrom (Brattstrom and Howell 1956) that son and his father, A. J. Grayson,an early nat- have been continued to the present, and two uralist and explorerof the Mexican islands.The recent expeditions(1978, 1981) by the authors Graysonsfound this dove in the vicinity of the (Jehl and Parkes 1982). One of the most inter- cove that bears their name on the southwestern esting findings of the recent expeditions was corner of Socorro.(Place names on Socorromay the discoverythat the statusof two endemics, be found on the map presented by Jehl and the Socorroor Grayson'sDove (Zenaidagray- Parkes 1982.) A. J. Grayson (1872) called his soni) and the • (Mimodes probable new speciesthe "Solitary Dove," as graysoni),has changeddramatically in the past he never saw more than one or two at a time; 25 yr. The dove has evidently becomeextinct, the Mourning Dove, except when nesting, is and the mockingbird appearsto be on the verge highly gregarious.Grayson and most subse- of extinction.Concurrently, two mainland rel- quent observers commented on the incredible atives of the island endemics, the Mourning tamenessof the SocorroDove; apparently,it Dove (Zenaida macroura) and the Northern was not uncommon to have one walk across Mockingbird (Mimus polyglottos),have become one's shoesor enter one's tent (R. Moran pers. established and are now thriving. comm.), and Grayson as well as Villa (1960) This situation, in which island forms have noted that doves could be capturedby hand. disappearedwhile their closerelatives from the Anthony (1898) reported that in May 1897 mainland have become established,represents the dove "did not seem to be at all common, the most dramatic case of replacement known but was perhaps more abundant in the higher in birds, and we have little doubt that ecolo- parts of the island." Kaeding (1905), who ac- gists comparing faunal lists made in 1958 and companiedAnthony but publishedseparately, 1981 would invoke the competitive exclusion stated that it was "not common, evidently hypothesisto explain it. But they would be in breeding May 2nd." This evaluation as uncom- mon was probably basedentirely on observa- tions at lower elevations. McLellan (1926), who publishedinformation on the birds studiedby •This speciesis known in much of the literature as "Socorro ," but it bears little external re- the California Academyof SciencesExpedition, semblanceto the true thrashersof the Toxos- which visited Socorro 2-12 May 1925, stated torna;its generalappearance and behavior (including from the field notes of the taxidermist of the vocalizations) are more like those of the true mock- expedition that Zenaidagraysoni was "very nu- ingbirds (Mimus). merous.... being particularly abundanton the July1983] "Replacements"onSocorro Island 553 higher wooded levels, and ranging almost to island, and even climbed above the forest to the top of the island"; the central peak, Mount the crest of Mount Evermann, we found no trace Evermann, reaches about 1,130 m. Thirteen of of the species.Furthermore, the personnel of the doves were captured alive "with an ordi- the military base on Socorro were unfamiliar nary fisherman'slanding net tied on the end with this dove. We have concluded that Zena- of a stick about six feet long" (Hanna 1926). ida graysoniis state. Brattstromand Howell (1956), in a comprehen- In 1981 Mourning Doves seemed even com- sive review of the avifauna of the Revillagi- moner than in 1978 in the dry lowlands, and gedo Islands,noted that the SocorroDove was over 100 were seen at one time in trees bor- "common around rocks at low elevations in dering Laguna Escondida,where they flocked March 1953, but in November 1953 it was rare and drank. In 1978 we saw no Mourning Doves in such placesand commonabove 1500 feet." above 500 m, but our observations there were Data from specimen labels in several museums very brief; in 1981a few were presentand call- also indicate that most doves were collected at ing in open areaswithin the forest up to about 500 m or higher. Villa (1960) visited several 650 m. localities and habitats on Socorro in January 1958 but did not specify exactly where he saw THE the dove. This is unfortunate, as he was the last to report this species.R. Moran had seen and The Socorro Mockingbird (Mirnodesgraysoni) photographedseveral in March 1957 (seeFron- is even more distinctive than the Socorro Dove. tispiece);neither he nor Villa had any indica- The latter is rather clearly derived from the tion that the dove was declining. mainland Mourning Dove, whereas the mock- In April 1978, our field party of six, includ- ingbird, placed in its own monotypic genus, ing R. Moran and B. Brattstrom,failed to find has no obvious mainland relative. Anatomical- any trace of the SocorroDove, but our searches ly it appears in some ways to link the true were made at elevations below 600 m. Instead, mockingbirds (Mirnus)with the (Tox- we found that the Mourning Dove had become ostorna)(Gulledge 1975) and may be a relict of establishedsince 1958 (and perhapssince 1971, a primitive form ancestralto both. Its voice is when Brattstromhad last previouslyvisited the more like that of a mockingbird (J. Barlow pers. island); these were the first records for Socorro. comm.), as is its general appearance. It was commonand widely distributed in open Like the dove, the SocorroMockingbird was dry wooded areas and adjacent scrub and formerly widespreadand tame.McLellan (1926) meadows on the south and east sides of the wrote that it was "the most abundant and most island as far north, at least, as the airstrip. We widely distributed species on Socorro. They saw scatteredflocks of up to 40 Mourning Doves were particularly numerousabout the spring at between BahiaBraithwaite and the airstrip, and Grayson's Cove, and in the heavily wooded some were seen as high as 500 m. Some were carlons."Hanna (1926),a participantin the 1925 courting, and their calls, heard commonly in expedition,wrote that in the denselyforested the early evening, were indistinguishablefrom canyonson the upper slopes of Mount Ever- those of mainland birds. The calls of the So- mann, "the trees were teeming with bird life corro Dove are quite different (Baptista et al. ... the curiosity of the mockingbirdswas un- MS) and would have been easily recognizable. controllable.They would hop up and peck, jay- We also saw courtship flights and found two like, at our knapsacksor boots.... This fear- active nestsand one desertednest. The species lessnessof the birds is one of the most striking was unequivocally established. featuresof far away, uninhabited islands such Our failure to find Zenaidagraysoni prompted as this." In March 1953, Brattstrom found this a second expedition, in April 1981. Not only species"common at lower elevationson Socor- did we revisit many areas studied in 1978, but ro ... at that time they seemedto prefer open our party of four (including R. Moran and a areas with a few shrubs." In November of the local guide) spent two daysand nights camped same year, Mimodes"appeared to be rare at in the forested areas of Mount Evermann, at lower elevations on the south side of Socorro 500 m and 650 m, where the specieshad been but common in forestedareas at higher eleva- most abundant. Although we made thorough tions and in canyonson the north side of the searchesin each of the major habitats of the island" (Brattstrom and Howell 1956). Villa 554 JEHLAND PARKES [Auk,Vol. I00

(1960) reported that during his 1958 visit the corro Dove and the impending disappearance mockingbirds, like all of the other Socorro of the SocorroMockingbird? An obvious"ex- species,were extraordinarily tame, to the ex- planation" is that the "fragile" island endemics tent that one pair fed from his hand, and one were replaced by more "vigorous" mainland flew to his shoulder and sang. He stated that stocksvia the processof competitive exclusion. mockingbirds were abundant in the higher The theory underlying this explanationis dis- wooded parts of Socorro, and "siempre nada cussedby Mayr (1963: 74-76), with citationsof huranos con el hombre" [= never shy towards various examples. If the competitive exclusion man]. explanation were appropriate to Socorro, it By 1978 Mimodesgraysoni appeared to be al- would be remarkable that the replacement most entirely restrictedto the vicinity of large events took place virtually simultaneously and fig groves near the coast.Away from the coast congruently in two very different types of birds. our party found only two near the airstrip in 2 Full data for the critical years are unavailable, days of field work, whereas in a relatively small but there is sufficient information about the area of fig groves at Playa Bianca there were as birds and the island ecologyto supportan al- many as five, somewhat wary but readily seen. ternative explanation. At least one and perhaps This favored habitat is very restricted.In 1981 both endemicswere largely gone prior to the most of our time was spent searching for the arrival of the mainland species.Thus, we think dove, and we spent little or no time in most of that the processinvolved was a simple one of the areaswhere mockingbirdshad been pres- immigration and colonization that had no di- ent in 1978, except for Playa Bianca. Here rect effect on the insular forms. Parkes, using playback of songs recorded in 1978,elicited a distant responsefrom one point, FACTORS PROMOTING EXTINCTION and later lured into the open what may have been the same individual. But for the lure of Although the causesof the declineof the en- the playback, we would not have detected this demic taxa on Socorro can never be known with speciesat all. The SocorroMockingbird now is certainty, a number of possiblecontributing absentor exceedinglyrare in most of its former factors can be analyzed. range, and we consider the prospectsfor its Human factors.--Although there had been a survival dim. few temporary settlements on Isla Socorro On 10 April 1978, Brattstromsaw a Northern (Richardsand Brattstrom1959), by 1956 Bratt- Mockingbird (Mimus polyglottos)at Grayson's strom and Howell described it as uninhabited. Cove, and Parkes and Brattstrom saw two, about In 1957 the Mexican Government established a 300 m apart, in open country below the airstrip smallmilitary garrisonand weather stationnear 2 days later. These were the first recordsof this BahiaBraithwaite. The presenceof humanshad mainland species. By April 1981, Northern no direct effect on the avifauna; in fact, Villa Mockingbirds were abundant on the southeast (1960) reported that in 1958 the soldiers were side of the island, in open and scrubby areas protective of the local birds, with the exception up to about 400 m. On 6 April we saw at least of the Red-tailedHawk (Buteojamaicensis socor- 30 individuals and found a nest containing a roensis),which they persecutedbecause it killed large juvenile. other birds for food. The , nevertheless, persists in fair numbers to this day (Jehl and DISCUSSION Parkes 1982). The military garrison expanded in the late 1970's,in part to accommodatecon- The successfulestablishment of immigrant struction of an airstrip on the island, which birds on Isla Socorro, 400 km from the nearest was completed in 1980. The base commander mainland, must be a rare event, as all 10 of the in 1978-1981 was particularly solicitous of the resident nonmarine speciesare endemic, from wildlife on Socorroand took actionsto help in well-marked subspeciesto one distinct genus. its preservation. It is thus especially notable that two, or 20%, were introduced on Socorro in 1869 of the resident speciesappear to have been re- and are still abundant. Although parts of the placed by mainland relatives in a period of less island appear severely overgrazed,creating than two decades. new, open field environments that were prob- What events led to the extinction of the So- ably not present a century ago, the impact of July 1983] "Replacements"on SocorroIsIand 555 the sheep on the endemic avifauna has been several specimens collected in 1978 and on minimal, as there was no apparent change in careful scrutiny of hundreds of doves at close either abundance or wariness of birds from range in 1981. The mockingbirds are so distinct Grayson'stime through 1958. Soldiers are per- that we doubt that they would hybridize under mitted to hunt sheep in many parts of the any circumstances. island, even to the crest of Mount Evermann, We cannot exclude the possibility that the but do not disturb birds during their hunting endemicssuccumbed to diseasesintroduced by trips, and we certainly have no reason to im- the immigrants, but that seems unlikely be- plicate either the sheep or their hunters in the causethe ranges and habitat preferencesof the decline of the dove and the mockingbird. species involved are so different (see below) The Socorro Dove has been maintained in that contact, if any, would have been minimal. aviariesfor many years(and now survivesonly Predation.--There are no native on as an aviary bird). To the best of our knowl- SocorroIsland and only two raptorial birds, one edge, the last wild-trapped doves were re- of which is the tiny insectivorousElf Owl (Mi- moved from the island in the mid 1940's. In crathenewhitneyi graysoni). The other is the en- view of the tameness of the doves, it is con- demic subspecies of Red-tailed Hawk. Al- ceivable that somewere caught by soldiersand though this hawk on the mainland is primarily maintained as pets. In view of the species' for- a -feeder, on Socorro it feeds on birds, mer wide distribution, the small size of the gar- land crabs, and probably lizards. Villa (1960) rison, the infrequency of collecting trips, and reported that the hawk was adept at catching the species'persistence until the late 1950's,we Common Ground-Doves (Columbinapasserina see no seriouspossibility that overtrapping de- socorroensis),and the crop of one collected by pleted the dove population. There is no evi- the 1925 expedition contained a Socorro Dove dence that the Socorro Mockingbird was ever (McLellan 1926). In April 1981, Parkes found kept in aviaries,although it is possiblethat some remains of a Mourning Dove and a Common were trapped by local inhabitants; in 1981 we Ground-Dove at Laguna Escondida; both had witnessed a soldier capturing a fledgling probably been killed by . The Red-tailed Northern Mockingbird for a pet. Trapping is Hawk, the Socorro Dove, and the Socorro even lesslikely to have affected the population Mockingbird long coexisted successfully,as of Mimodesgraysoni than might be true of the evidenced by their taxonomic distinctnessfrom dove. mainland relatives. It is unlikely that any re- Environmentalfactors.--Despite two decades cent shift in predation by the hawk could have of human occupancyon Socorro, we have ob- been involved in the decline of the other two served no environmental changesof sufficient species. magnitude to have caused the decline of the Human settlement of the island in the late dove and mockingbird. The island is large and 1950's was responsible for the introduction of extremely rugged; human activity near the air- domestic , and cats have become feral. strip, the garrison, and the small agricultural No cats were reported on Socorro prior to es- areas is so restricted as to leave by far the ma- tablishment of the military base.The size of the jority of the island undisturbed. present feline population is not known, but it Biologicalfactors.--One might postulate that may be considerable.We found a nest of cats the mginland Mourning Dove was able to in 1978, far from settlement, and in 1981 we swamp the endemic form by interbreeding, in found the remains of several Townsend's effecthybridizing it out of existence.At a lower Shearwaters(Puffinus auricularis) high on Mount taxonomic level, this is what happened to the Evermann, within the woodlands formerly oc- endemic subspeciesof MadagascarTurtle-Dove cupied by the Socorro Dove. This shearwater (Streptopeliapicturata rostrata) on most islands spends the day at sea, returning to the island of the Seychellesafter the introduction of the only after dark, and is thus vulnerable only to mainland Madagascarsubspecies S. p. picturata a nocturnal predator while ashore. The only (Penny 1974). The Socorro Dove has, in fact, nocturnal predator capable of killing a shear- hybridized with the Mourning Dove under water or a dove is the . The tame, ground- captive conditions. There is no evidence, how- dwelling doves and mockingbirds would have ever, of introgression in the present dove pop- been easy prey for cats,by night or day. ulation. This is based on the examination of We infer that feral cats were responsible for 556 JEHLAND PARKES [Auk, Vol. 100 the extinction of Zenaidagraysoni and the near- Mockingbird in 1978, Miraodesgraysoni was al- extinction of Miraodesgraysoni. We base this ready confined almost exclusively to dense fig conclusionin part on a broader examination of groves and canyonsalong the coast.The invad- the fauna (Jehl and Parkes 1982), in which we ing species,like the Mourning Dove, is found show that the relative abundance of tree-dwell- in open farmland and fields with scattered ing birds on Socorrohas increasedwhereas all grovesor isolatedtrees. Only once,in 1978,did the ground-nestershave declined. The arboreal we seea Northern Mockingbird in the habitat endemic wren (Thryomanessissonii) and warbler of the SocorroMockingbird; not only is it un- (Parula pitiayumigraysoni) are not only abun- likely that the recent immigrant competesin dant but retain the traditional fearlessness of any way with the endemic species,but it is the insular avifauna. By contrast,the ground- probablethat the populationsof the two species nesting towbee (Pipilo erythrophthalmussocor- have not even been in contact since the arrival roensis),although still present in fair numbers, of the Northern Mockingbird. is neither as abundant nor as tame as reported It is thus our contention that feral cats have by all observersprior to 1978. depleted or eradicatedthe ground-nesting en- Competitiveexclusion.--Although we accept demic birds on Socorro and that the invasion the possibility that Zenaidamacroura invaded by two related mainland forms bears no causal Socorroprior to the disappearanceof the en- relationship to the decline of the endemic in- demic Z. graysoni(we lack firm data for the pe- sular species.Indeed, under present environ- riod between January 1958 and April 1978), and mental conditionson Socorroit seemspossible we know that Miraodesgraysoni was declining that the native and invading speciescould co- during the three-year period in which Mimus exist. polyglottosbecame established, we do not be- It would be logical to ask how the Mourning lieve that competitive exclusionwas involved. Dove and Northern Mockingbird were able to Indeed, we agreewith Pregill and Olson (1981: withstand cat predation, especiallyimmediate- 91) that "ecologicaldoctrine and good sense ly after their immigration while their popula- revolt at the idea that a specieswith a long tions were still low. Unlike their endemic rel- history of to a particular environ- atives, both species nest in trees--and the ment would be at a competitivedisadvantage arboreal birds of Socorro Island remain as with newly arriving colonists." abundant as ever. Although the Northern In Zenaida,direct physical exclusionof the Mockingbird and, especially, the Mourning island speciesby the mainland speciesis un- Dove forage on the ground, the open habitats likely for several reasons. First, the Socorro they prefer would make stalking them by day- Dove is considerablylarger than the Mourning hunting cats more difficult than would have Dove and, in the experienceof aviculturists(L. been true of the SocorroMockingbird and So- Baptistapers. comm.), is dominant to the small- corro Dove. er speciesin all interactions.It is a very feisty In a similar study, Diamond and Veitch (1981) speciesand aggressivetoward macrourawhen compared the abundance of native and exotic they are cagedtogether. Second,the absenceof (introduced by man) bird speciesin disturbed direct competitionis indicated by the very dif- forest on the mainland of New Zealand and in ferent habitat preferencesof the two species. relatively undisturbed forest on islands in The SocorroDove is a ground-nesting species, Hauraki Gulf. They concludedthat "exoticbird and its principal habitat was the fig forestsat specieswere not able to achieve their present higher elevations.That habitattoday is not used penetration into New Zealand forest until the by the tree-nesting Mourning Doves, which forest structurehad been disturbedby brows- occur in abundance at lower elevations in more ing and logging, or until native specieshad arid situations in open woodlands, similar to been decimatedby predation,disease, and these their preferred habitat on the mainland. We habitat changes." On Socorro, as we have see no possibility here for any competition. shown, reduction of the endemic avifauna is Similarly, the preferred habitatsof the island not related to the limited habitat disturbance and mainland mockingbirdsare dissimilar.Al- createdby sheep.On the other hand, both the thoughthe islandendemic species formerly oc- Northern Mockingbird and Mourning Dove curred over most of Socorro,it preferred for- chiefly inhabit thoseareas that have been most ested areas. Before the arrival of the Northern heavily grazed.Thus, the sheepmay have cre- July 1983] "Replacements"on SocorroIsland 557 ated suitable habitat for the invaders, resulting tributed to the failure of the mainland birds to in additions to the avifauna but without any colonize Socorro until 1958-1978, but another correlated loss. event was probably much more important. The establishmentof the military base, though of FACTORS PROMOTING SUCCESSFUL minimal significanceto the ecologyof the is- COLONIZATION land as a whole, did affect the immediate en- vironment of the settlement.Not only were do- Mourning Doves are very strong fliers and mestic animals introduced, but also food and are among the land birds most frequently seen ornamental plants, and this required the estab- off southernCalifornia, often many miles at sea lishment of a fresh-water system.Socorro is a (Jehl pers. obs.).The specieshas been recorded desertisland; the only regular sourcesof fresh from all of the islands off California and Baja water are several intertidal springs, which are California, and we have no doubt that it has exposedfor only brief periods each day. The reachedSocorro frequently, even though it had Graysonssurvived being shipwreckedonly be- not been seen there by ornithologistsprior to causethey landed, by chance,near the spring 1978. The species became established twice at Grayson'sCove. Other than the springs, the previously on the Revillagigedos, having only natural sourcesof drinkable water (there evolved into the endemic Zenaidagraysoni on are some hot springs high on Mount Ever- Socorro and Z. macroura clarionensis on Clari6n. mann) are rain-filled pools; these are tempo- Mourning Doves have been seen on San Be- rary and seasonal,and may be absent in dry nedicto but cannot become established there years. because of the absence of fresh water. Similar- The principal dispersal of land birds is that ly, there are recordsof the Northern Mocking- of immature individuals in the autumn, a well bird from most of the California and Mexican known phenomenon that accountsfor many islands, including Guadalupe and Clari6n, in- "out of range" and unusual distributional rec- dicating a fair dispersalability over the open ords. During migratory flights "water require- seas. ments rather than energy reservesin some in- Given the dispersal potential of these two stancesmay limit flight range" (Blem 1980:204). species,especially the dove, how can we ex- Furthermore, "long-range migrants unable to plain the colonization oœboth on $ocorro in an obtain fresh water may have a severetendency interval of two decades? There is no evidence to dehydrate" (Blem loc. cit.). Therefore, the that either specieswas introducedby man, and rapid discovery of fresh water within a few this was confirmed by our conversationswith hours of reaching landfall is probably critical military personnel. to the survival of migratory or lost birds. At One factor is competiveexclusion, but rather Socorroin fall the rain pools are dry, and the than the invading mainlandersprevailing over possibility that waifs will discover the local- the insular forms, the invaders may have been ized sourcesof intertidal water during the brief prevented from colonizing by the locally hours they become available each day seems adapted(and, in both cases,larger) island forms. very small. Some desertspecies are able to sur- This situation would persist for millenia, until vive for long periods without fresh water, but (a) the insularendemics were in very low num- Mourning Doves, despite their widespread dis- bers or had already become extinct through tribution in arid regions, require water daily some agency other than competition--in this and are unable to subsist on 50% sea water (Bar- case,predation by the newly introduced cats-- tholomew and MhcMillen 1960). No data on or (b) habitat changes,or the restriction of is- the osmoregulatoryabilities of the Mocking- land forms to certain habitats, created a vacant birds are available, but there is no reason to "niche." As Pregill and Olson (1981: 92) put it, suspectthat they are exceptional. "we see no reasonwhy a speciesadapted to a With the establishmentof the military base, particular set of conditions on an island would with its human families and their domestic an- not persist indefinitely, provided that its en- imals, gardens,and orchards,fresh water that vironment did not change." This is essentially could be used by migrants became available what Diamond and Veitch (1981) found in their year round. Survivorshipof newly arrived birds New Zealand study. would be greatly increased, and species that Exclusionby the endemicsmight have con- could find an appropriate vacant "niche"--such 558 JEFILAND PARKES [Auk,Vol. 100 as the open field habitatscreated by overgraz- clude much of Socorro in the near future. Had ing sheep--would have a much improved op- we not visited the island in 1978 and 1981 but portunity to becomeestablished. 20 years hence, we might have found that the We have no direct evidence from Socorro for size of the avifauna remained constant but that the critical water theory, but there appearsto two endemic forms had disappearedand their be a comparablesituation on Guadalupe Island. mainland counterparts had become wide- The Mourning Dove was considered to be ac- spread. This simple comparison of specieslists cidental there prior to 1952 (Howell and Cade would have led us to the "acceptable,"but in- 1954), having been recorded only once, in the correct, conclusion that the Socorro situation 1880's. Subsequently, doves were reported in was a classicalcase of replacement via compet- August 1956 (1), November 1964 (1), and Feb- itive exclusion. ruary 1967 (24), datescompatible with the tim- In an important review of the zoology of West ing of the fall dispersal. By 1970 the species Indian vertebrates, Pregill and Olson (1981) was widespread and nesting, and in August documented the inadequaciesof theories of is- 1981 the population numbered in the low land biogeographythat were constructedwith- hundreds (Jehl and Everett MS). The coloni- out an historical and paleontologicalperspec- zation was approximately synchronous with tive. Our data from Socorro confirm their that on Socorro,but probably not becauseof skepticism and show that an historical per- any change in the abundanceor dispersalpat- spective of as little as a decade may be critical terns of mainland populations.In the late 1960's for interpreting turnover on oceanic islands; the military garrisonat the south end of Gua- for near-shore islands annual censuses are nec- dalupewas enlarging,and leaksfrom the fresh- essary(Diamond and Jones 1980). water systemwere commonin the village, pro- So long as the testing of theories of island viding an attraction for the endemic House biogeography is based primarily on compari- Finch (Carpodacusmexicanus amplus), often con- sonsof faunal lists from different periods and sidered a full species(Jehl pets. obs.,photos). does not incorporate information about ecolog- Habitat conditionson Guadalupeare such that ical conditions,habitat preferencesof the en- doves are among the very few species that demic and "replacement" species, and extra- might be expected to become established.Al- neous factors such as, on Socorro, the though the Northern Mockingbird has not be- establishmentof feral catsand the provision of come established there, it has reached that is- supplementary water supplies, interpretation land. One was convincingly described to Jehl of turnover data will be hazardous, if not im- on 22 February 1969, the first record since two possible(see also Lynch and Johnson1974). Be- were seen by Bryant in March 1886 (Jehl and cause such data are not available for most is- Everett MS). lands, we advocate great caution in extrapolating comparisons of faunal lists to CONCLUSIONS general descriptionsor theoriesregarding pro- cessesof faunal change. Double invasions of islands by mainland stocksof birds are not uncommon (Mayr 1963), ACKNOWLEDGMENTS but they must almost always be studied infer- entially, long after the invasionsoccurred. We Most of the cost of the 1978 expedition was cov- were fortunate to visit Socorro during the ac- ered by grantsfrom the M. GrahamNetting Research tual establishmentof one species,the Northern Fund and the Edward O'Neil Field Fund of the Car- Mockingbird, and have been able to narrow negie Museum of Natural History. Additional fund- the establishmentof the Mourning Dove to a ing was provided by Hubbs-SeaWorld ResearchIn- period of 20 years or (probably) less. Further- stitute. The 1981 expedition was sponsoredby Tom more, we have first-hand knowledge of the and Dorothy Hawthorne of San Diego, California, with additional support from the M. Graham Netting ecologicalconditions on the island and of the Research Fund. Permission to conduct these studies habitat preferencesof the speciesinvolved in was granted by the Mexican government. We thank the turnover. Dr. JorgeCarranza Frazer for his interest in our stud- At their present rate of population growth, ies and Capt. F. Urtaza of the Mexican Navy for sup- Mourning Doves and Northern Mockingbirds port and hospitality on our visits to Socorro. can be expected to expand their range to in- This paper hasbenefitted from the commentsof E. July1983] "Replacements"onSocorro Island 559

Connor, J. Diamond, N. K. Johnson, B. G. Murray, and turnover rates: the avifauna of Santa Barbara Jr., and D. M. Power, all of whom have sufficient Island, California. Condor 76: 363-369. confidence and courtesy to sign their reviews. JEHL,J. R., & K. C. PARKES. 1982. The status of the avifauna of the , Mexico. Wilson Bull. 94: 1-19. LITERATURE CITED JONES,H. L., & J. M. DIAMOND. 1976. Short-time- ANTHONY, A. W. 1898. Avifauna of the Revillagi- base studies of turnover in breeding bird popu- gedo Islands. Auk 15: 311-318. lations on the California Channel Islands. Con- BARTHOLOMEW,G. A., & R. E. MACMILLEN. 1960. The dor 78: 526-549. water requirementsof Mourning Dovesand their KAEDING, H. B. 1905. Birds from the west coast of use of sea water and NaC1 solutions. Physiol. Lower California and adjacent islands. Condor Zool. 33: 171-178. 7: 105-111, 134-138. BLEM,C. R. 1980. The energeticsof migration. Pp. LYNCH, J. F., & N. K. JOHNSON. 1974. Turnover and 175-224 in migration, orientation, and equilibria in insular avifaunas, with special ref- navigation(S. A. Gauthreaux,Jr., Ed.). New York, erence to the California Channel Islands. Con- Academic Press. dor 76: 370-384. BRATTSTROM,B. H., & T. R. HOWELL. 1956. The birds MACARTHUR,R.H. 1972. Geographicalecology. Pat- of the Revilla Gigedo Islands, Mexico. Condor terns in the distribution of species. New York, 58: 107-120. Harper and Row, Publ. DIAMOND, J. M. 1969. Avifaunal equilibria and , & E. O. WILSON. 1967. The theory of island speciesturnover rateson the Channel Islandsof biogeography. Princeton, New Jersey,Princeton California. Proc. Natl. Acad. Sci. 64: 57-63. Univ. Press. 1975. The island dilemma: lessons of mod- MARSHALL,L. G., S. D. WEBB,J. J. SEPKOSKI,JR., & D. ern biogeographicstudies for the design of nat- M. RAUP. 1982. Mammalian and the ural reserves. Biol. Conserv. 7: 129-146. Great American interchange. Science 215: 1351- --, & H. L. JONES.1980. Breeding land birds of 1357. the Channel Islands.Pp. 597-612 in The Califor- MAYR,E. 1963. Animal speciesand evolution. Cam- nia islands (D. M. Power, Ed.). Santa Barbara, bridge, Massachusetts,Harvard Univ. Press. California, Santa Barbara Mus. Nat. Hist. MCLELLAN,M. E. 1926. Expedition to the Revilla- --, & C. R. VEITCH. 1981. Extinctions and intro- gigedo Islands, Mexico, in 1925. VI. The birds ductions in the New Zealand avifauna: cause and and mammals. Proc. California Acad. Sci., Ser. 4, effect? Science 211: 499-501. 15: 297-322. GILBERT,F. S. 1980. The equilibrium theory of is- OLSON,S. L., & H. F. JAMES. 1982. Fossil birds from land biogeography: fact or fiction. J. Biogeogra- the Hawaiian Islands: evidence for wholesale ex- phy 7: 209-235. tinction by man before western contact. Science GRAYSON,A.J. 1872. List of Socorro birds collected 217: 633-635. by A. J. Grayson, May, 1867. Proc. Boston Soc. PENNY,M. 1974. The birds of Seychelles and the Nat. Hist. 14: 290-302. [N.B.--This paper is usu- outlying islands. London, Collins. ally cited as "Lawrence 1871," but it was merely POWER,D. M. 1972. Numbers of bird specieson the edited by Lawrence,the wording being entirely California Islands. Evolution 26: 451-463. Grayson's,and it was actuallypublished in 1872]. PREGILL,G. K., & S. L. OLSON. 1981. Zoogeography GRINNELL,J. 1925. Risks incurred in the introduc- of West Indian vertebrates in relation to Pleis- tion of alien game birds. Science61: 621-623. tocene climatic cycles.Ann. Rev. Ecol. Syst. 12: GULLEDGE,J. L. 1975. A study of phenetic and phy- 75-98. logenetic relationships among the mocking- RICHARDS, A. F., & B. H. BRATTSTROM. 1959. Bibli- birds, thrashers and their allies. Unpublished ography, cartography, discovery, and explora- Ph.D. dissertation. New York, City Univ. New tion of the Islas Revillagigedo. Proc. California York. Acad. Sci., Ser. 4, 29: 315-360. HANNA,G D. 1926. Expeditionto the Revillagigedo TEMPLE,S. A. 1981. Applied island biogeography Islands,in 1925.Gen. Rept. Proc.California Acad. and the conservationof endangered island birds Sci., Ser. 4, 15: 1-113. in the Indian Ocean. Biol. Conserv. 20: 147-161. HOWELL,T. R., &T. J. CADE. 1954. The birds of Gua- VILLAR., B. 1960. Vertebradosterrestres. Pp. 203- dalupe Island in 1953. Condor 56: 283-294. 216 in La Isla Socorro, Monogr. Inst. Geofisica, HUNT, G. L., JR.,& M. W. HUNT. 1974. Trophic levels Univ. Nac. Autonoma de Mexico no. 2.