Short-Time-Base Studies of Turnover in Breeding Bird Populations on the California Channel Islands
SHORT-TIME-BASE STUDIES OF TURNOVER IN BREEDING BIRD POPULATIONS ON THE CALIFORNIA CHANNEL ISLANDS
H. LEE JONES AND JARED M. DIAMOND
The number and identities of species coexist- additions were of non-breeding visitors, re- ing at a particular locality are not fixed for- corded principally by us and by other observ- ever, but result from dynamic interplay be- ers since 1968. The reason is that most of the tween local extinctions and immigrations. early ornithologists who visited these islands While this dynamic structure must apply to were collectors and oologists mainly interested any local community, it is particularly con- in obtaining specimens or egg sets of the venient to study on islands, because of their breeding birds, some of which are considered sharp boundaries. In their well-known theory endemic or nearly endemic subspecies. Conse- of island biogeography MacArthur and Wil- quently, the breeding avifauna of all the Chan- son (1963, 1967) proposed that the number of nel Islands was more thoroughly documented species on an island approaches an equilibrium by 1917 than was the non-breeding avifauna between extinctions and immigrations, and (see appendix p. 545 for discussion). that populations are subject to turnover. It is In 1968 one of us (J. D. ) visited each of the likely that large differences in turnover rates Channel Islands l-3 times, plus the nearby will be found if one compares the same taxo- Mexican island group Los Coronados, to sur- nomic groups on islands of different area, iso- vey the breeding avifaunas for comparison lation, latitude, and habitat, or if one compares with the surveys summarized by Howell different species on the same island. The (1917). On each island he found that there measurement and understanding of these dif- had been turnover, reflected both in dis- ferences is beginning to emerge as a major appearances of some former breeding popula- empirical and theoretical problem in ecology tions and in breeding presence of some for- (Maguire 1963, Patrick 1967, Simberloff and merly absent or non-breeding species. A brief Wilson 1969, Diamond 1969, 1971, Terborgh account of the results, and of estimated turn- and Faaborg 1973, Hunt and Hunt 1974, over rates in relation to A, d, and Serl,was pub- Schoener 1974), as a problem in assessing the lished (Diamond 1969). One of Diamonds’ importance of group selection and understand- findings was that some breeding populations ing the evolution of social behavior (Levins had immigrated and become extinct several 1975, Wilson 1975:1X-116), and as a times on the same island between the early major practical problem in conservation strat- 1909s’ and 1968. Thus, surveys separated by egy (Terborgh 1974a, 197413, Diamond 1975, many decades must have underestimated turn- 1976, Wilson and Willis 1975, Sullivan and over rates and might only have revealed the Shaffer 1975). tip of the iceberg of community dynamics A convenient situation for studying these ( see Diamond and May in press for examples). problems is provided by the land and fresh- For organisms as mobile as birds, and islands water breeding birds of the Channel Islands as close to the mainland as the Channel Is- off the coast of southern California. The eight lands, it is likely that there will always be a islands of this group vary in area (A) from long waiting list of potential immigrants; many 2.6 to 249 km2, in distance from the mainland more unsuccessful attempts at colonization (d) from 20 to 98 km, and in number of land than attempts that succeed even briefly; and and non-swimming water bird species breed- many more brief successes and rapid failures ing in a given year (S,,,,) from 7 to 39 (see than foundings of a colonist population that Philbrick 1967, Power 1972, Johnson 1972, and survives a long time. Hence repeated surveys Jones 1975 for maps and table of areas and dis- at one-year intervals were clearly required to tances). As of 1917, when A. B. Howells’ reveal the dynamic structure of the avifauna monograph Avifauna of the Islands off the more accurately. Such studies of turnover at Coast of Southern California was published, one-year intervals have the further advantage 170 species of birds had been recorded from of reducing interpretative problems associated the Channel Islands. By 1976 this number with habitat changes and effects of man which had increased to 325. All but five of these 155 develop over longer intervals.
[5261 The Condor 78:526-549, 1976 TURNOVER ON THE CHANNEL ISLANDS 527
In 1973, L. J. began a program of annual TABLE 1. Ornithological visits to the Channel Is- breeding surveys of a11eight Channel Islands. lands. This program is now entering its fifth year. - Earlier observers Jones Surveys of some of the islands by other ob- first field servers for certain years between 1969 and Island year visits years visits hours 1972 are also available. The purpose of this San Miguel 1875 30 24 10 241 paper is four-fold: to report the problems en- Santa Rosa 1889 22 13 220 countered and methods used in these short- Santa Cruz 1875 103 ;; 14 274 time-base turnover studies; to illustrate and Anacapa 1899 60 35 27 172 describe the rich spectrum of dynamic be- San Nicolas 1863 24 17 34 1008 Santa Barbara 1863 25 25 317 havior they reveal; to discuss the several ways Santa Catalina 1861 % 43 16 215 in which man has directly and indirectly in- San Clemente 1863 46 33 11 252 creased and decreased turnover rates; and to 150 2699 present estimated turnover rates, evidence for For each island the table gives the year of the first orni- dynamic equilibrium of species number, and thological observations; the approximate number of visits by ornithologists, and number of years in which there were a model of island colonization. Species ac- visits, through 1967; and the number of visits and accumu- counts of the Channel Islands avifauna, with lated field hours by Jones. The number of visits since 1967 by observers other than Jones is too large to tally. details of records and with descriptions of is- land habitats and history, are being prepared for publication in book form elsewhere (Jones J. Atwood visited Santa Cruz a number of times in 1975 and in prep.). An appendix considers 1975 and 1976. As part of our survey program, K. Garrett made four visits to Anacapa, two to Santa critiques of turnover studies by Johnson Barbara, and one to San Clemente in 1975 and three (1972) and by Lynch and Johnson (1973, to Anacapa in 1976; G. and S. Grant, three visits to 1974). Santa Catalina in 1975 and one in 1976; D. and G. Schroeder, one visit each to San Miguel, Santa Rosa, Santa Catalina and San Clemente in 1976 and six METHODS visits to Santa Cruz; P. Unitt, one visit each to San Clemente and San Miguel in 1976; R. Higson, one SOURCES OF INFORMATION visit to Santa Cruz and Santa Catalina in 1976. Many The present paper is based on island visits by L. J. other individuals contributed additional observations. from February 1973 through August 1976. During In addition, we critically evaluated all published this period, which included four breeding seasons, papers known to us that discuss Channel Island birds, he visited the islands 150 times for periods of up to from 1868 to the present. Numerous unpublished 15 days per visit and accumulated approximately 2699 manuscripts and field notes of ornithologists who hours of field observation (table 1, columns 5 and 6). visited the Channel Islands were found, especially in The number of accumulated visits by Jones to any one the libraries and Dickey Collection at UCLA, at the island ranges from 10 (San Miguel) to 34 (San Nico- Museum of Vertebrate Zoology (Berkeley), and at las); of accumulated hours of field observation on one the Western Foundation of Vertebrate Zoology (Los island, from 172 (Anacapa) to 1008 (San Nicolas). Angeles), and were similarly critically evaluated. Visits were made in all months of the year. Most vis- Significant unpublished manuscripts dating back to its were in the breeding season, but some were made at least 1927 were found for all islands, and back to outside the breeding season to assess resident status 1897 for some islands. Published and unpublished of populations. Additional observers accompanied general accounts of the Channel Islands, not dealing Jones on many of these visits. specifically with birds, were studied in order to re- During the 19681976 period many other biologists construct the history of habitat changes and human contributed additional surveys or records, of which disturbance; such information was available for all the most extensive are the following. In the 1968 islands at least from the 1840s.’ Jones examined the breeding season teams of biologists from the Smithson- collections of egg sets and bird specimens from the ian Institutions’ Pacific Ocean Biological Survey Proj- Channel Islands in the following museums and uni- ect visited six of the eight islands up to three times versities: California Academy of Sciences Museum each, and maintained a resident observer on San Nico- (including Stanford University collections), Santa las for most of the breeding season. G. and M. Hunt Barbara Museum of Natural History, Los Angeles lived on Santa Barbara for most of the breeding sea- County Museum of Natural History, Western Founda- son from 1972 to 1976; R. Yeaton on Santa Cruz in tion of Vertebrate Zoology (WFVZ), San Bernardino 1969 and 1970; D. Propst on Santa Catalina from County Museum, San Diego Natural History Museum, 1968 (and earlier) through 1976; L. Laughrin on Museum of Vertebrate Zoology (MVZ), UCLA Dickey Santa Cruz from 1969 through 1976; D. Lees on Collection, and California State University at Long Santa Catalina in 1968; and R. DeLong on San Miguel Beach. Some of the specimens and unpublished manu- in 1972 and 1973. D. Ode11 made numerous visits to scripts that we examined for years prior to 1917 were Santa Barbara in 1968 and to San Nicolas from 1969 evidently not available to Howell (1917) when he to 1971, and lived on San Nicolas for part of this time. prepared his review of the avifauna. Conclusions J. Larson and associates from the Naval Underseas about breeding status reached by Diamond (1969) Center and California State University at San Diego have been reassessed in the light of additional infor- made numerous visits to San Clemente from 1972 to mation encountered for years prior to 1968 and of our 1976; F. Gress, 13 visits to Anacapa in 1970; R. Stew- observations since 1968. Because details of records art and W. Clow, two visits to San Clemente in 1974. will be published elsewhere (Jones, in prep.) and 528 H. LEE JONES AND JARED M. DIAMOND would increase the length of this paper prohibitively, islands (p. 540), which have the advantage of being we cite references only for some key records obtained the easiest islands to survey completely and on which before our observations began in 1968. to prove turnover. The possibility remains for the The years for which it proved possible to compile larger islands that we overlooked one or more species reasonably complete lists of breeding birds to use for of which only one or two pairs bred in single years. turnover calculations (p. 539) are as follows for each We have no reason to suspect having overlooked such island: San Nicolas 1897. 1945. 1962-1963. 1968. cases, but if we did, actual turnover rates for larger 1973, 1974, 1975, 1976; Santa Barbara 1897, ’ 19111 islands would be higher than those we report. For 1912, 1968, 1972, 1973, 1974, 1975, 1976; Anacapa purposes of calculating immigration and extinction 1910-1912, 1939-1941, 1963-1964, 1968, 1970, 1973, rates in the Channel Islands avifaunas, we have con- 1974, 1975, 1976; San Miguel 1939, 1968, 1973, 1975, sidered only the breeding avifauna, defined as those 1976: San Clemente 1897. 1907. 1915. 1968. 1973. species of which at least one pair is known, or may 1974: 1975, 1976; Santa Catalina 1905-1969: 19681 reasonably be inferred, to have attempted to nest on 1973-1974, 1975, 1976; Santa Rosa, 1927, 1968, the particular island in the year considered, regardless 1973, 1974, 1976; Santa Cruz 1911-1912, 1968, 1973- of whether or not the nest attempt finally resulted in 1974, 1975, 1976. In the nine cases of years fledged young. Only bird species that do not normally connected by a hyphen, we based such breeding lists rest on salt water were considered (i.e., gulls, ducks, on observations made over several consecutive years and alcids were excluded; shorebirds, waders, and ( p. 545). Table 1 lists, for each island, the number Rallidae were included). of ornithological visits through 1967 which we were The Channel Islands are visited by many non-breed- able to locate as sources of bird records, based on ing migrants and vagrants. Many migrant passerines published papers, museum specimen data, or unpub- sing and even behave territorially for short periods. lished manuscripts or field notes; the year of the Therefore, one of the main practical problems to first visit by an ornithologist; and the number of years which our field observations were devoted was to through 1967 in which at least one visit was made, establish which species were breeding on each island. as well as details of Joness’ visits. Since results of the In most cases that we accept as instances of turnover, great majority of visits to the Channel Islands have proof was obtained by observing active nests or fledg- remained unpublished, a search of only the published lings, and the existence of such evidence is implied literature would grossly underestimate the amount of in all instances in the text when we discuss turnover field work that has been done. without otherwise citing breeding evidence. However, in a minority of cases one or more pair of a species SURVEY STRATEGY were found to be resident on an island during the breeding season at a constant location in appropriate Field time on the islands was allocated with two pri- breeding habitat, and other types of evidence indi- mary purposes in mind: to determine which of the cated breeding, but the nest or fledglings could not species observed on a particular island were breeding be seen. To ignore such cases would introduce prob- and which were not, in most cases by locating active able errors into turnover calculations. In such cases, nests and fledglings, in some cases (see next section) availability of one or more of the following types of by observing other indications of breeding; and to evidence, cited in the text by the code letter in pa- minimize the likelihood of overlooking species that rentheses, was accepted in conjunction with residency might be breeding. Whenever a visit yielded incon- as evidence of breeding: (m)-Adults were repeat- clusive evidence of breeding for a particular species, edly seen carrying nest construction material to a we attempted to return or to have a resident observer place or carrying egg shells or fecal sacs from a place. return to the same site later to resolve the question. ( f )-Adults were repeatedly seen carrying food to a Regarding the problem of breeding populations presumed nest. (s)-Adults of hole-nesting or cliff- being overlooked, two islands, Santa Barbara and nesting species were repeatedly seen entering a pre- Anacapa, are sufficiently small (2.6 and 2.9 km,’ re- sumed nest site in a hole or on a cliff inaccessible to spectively) that it was feasible to examine essentially us for direct observation. These cases involved Acorn the whole island. On the five largest islands (Santa Woodpecker ( Melunerpes formicivorus) and Ameri- Cruz, Santa Catalina, Santa Rosa, San Clemente, San can Kestrel (F&o sparuerius). j)-Juveniles were Nicolas) the availability of roads, jeep transportation, ( observed that were capable of flight but had incom- and resident observers facilitated surveys. On these plete feather growth, were accompanied by adults five islands plus the medium-sized San Miguel (36 and often were being fed by the adults, and were pre- km)’ we used our experience of island habitats and sumed to have been fledged recently on the island of recent breeding populations to design survey strate- observation. (t)-One or more pair of adults ex- gies. Particular attention was paid to specialized habi- hibited clear territorial behavior for a prolonged period tats or to species-rich habitats likely to harbor local in the breeding season. Whether such behavior pro- populations, such as springs, wind-sheltered gullies, streams, marshes, wooded canyons, and pine forest. vides any evidence at all of breeding depends en- Some field time was allocated at night to find noc- tirely on the species, number of pairs involved, and turnal species. A measure of the completeness that length of time involved, as discussed for each case it proved possible to attain is that although much in the text (e.g., see discussion of Huttons’ Vireo additional information was available from other ob- (Vireo huttoni) on p. 535). servers for each island in each survey year as dis- Questions of breeding status were evaluated con- cussed in the preceding section, both Diamond in servatively with respect to turnover calculations if 1968 and Jones in 1973-5 themselves observed almost evidence was inconclusive. That is, if a species was all breeding populations each year except for some represented on an island in the breeding season, but owl populations on the four largest islands, reliably if available evidence did not suffice to make breeding reported to us by other observers. We have not at- tempted to assess turnover in these large-island owl either highly likely or highly unlikely, the species was populations. In practice, it has become clear that assumed to be breeding if this was true in previous and turnover rates are much the highest on the smallest subsequent survey years, and was assumed to be not TURNOVER ON THE CHANNEL ISLANDS 529
6
occasional 1 frequent 3 breeders I breeders “flukes” v I \ in-and-out species 2 l-_------__------T -- regular 1 absentees ------XrTnTiGTGIiticTi~~ Er7iCiZ$aTif%s breeders I, J 0 t
I I I I I I I I I I I 0 1 2345670 9 10
N ( years)
FIGURE 1. Turnover pyramid. Abscissa: number of years, out of 10 survey years, that a particular species bred on a particular island (e.g., N = 10 means that the species bred every year). Ordinate: number of cases of turnover, counting either an immigration or an extinction as one case of turnover, that the species provided on the island within the 10 survey years.
breeding if this was the case in previous and sub- RESULTS sequent survey years. If a year of definite breeding was separated from a year of definite absence by a THE TURNOVER PYRAMID year of presence with uncertain breeding status, the decision affects the calculated time of turnover but Suppose that complete breeding censuses of not the calculated number of cases of turnover. In an island were available for, say, 10 separate such cases we arbitrarily assumed the immigration to years. The results for each species might be precede and the extinction to follow the uncertain depicted in short-hand by numbering the year. Barn Owl (Tyto &a), Saw-whet Owl ( Aegolius years, underlining years of breeding, and de- acadicus), and Burrowing Owl ( Athene cunicularia) are sufficiently difficult to observe, to prove absent, noting immigrations by I and extinctions by or to document as nesting that we conservatively as- E: e.g., lE213E415E617E819E10, or 1234E567- sumed no case of turnover for Barn or Saw-whet owl 18910. The %nn&er characteristics of each and only two particularly well documented cases for Burrowing Owl. In evaluating surveys by earlier ob- species could then be depicted by plotting the servers before our field work began, we considered species as a point in a two-dimensional space, lack of records for a particular species in a particular as illustrated in figure 1. The horizontal axis year as evidence of actual absence only if the intensity of this space is the number of years (N) out of the survey in relation to the size of the island and ease of observation of the species made the lack of of the 10 survey years that the species bred. records significant. For instance, the failure of Joseph The vertical axis is the number of cases of Grinnell to find the conspicuous Peregrine Falcon turnover (#T) which that species provided, (Ealco peregrinus) on a six-day visit to tiny Santa counting either an immigration or an extinc- Barbara in 1897 is considered sufficient evidence for tion as one case of turnover. Points for all absence that we count a subsequent breeding record as an immigration. However, the failure of Green, species are constrained to be within a pyramid Sumner, and Bond to find the Barn Owl on San Miguel whose apex is at N = 5, #T = 9 (correspond- during two visits in 1938 and 1939 is not considered ing, e.g., to lE213E415E617E819ElO). To un- sufficiently strong evidence of absence to warrant counting a nest found in 1968 as proof of an immi- derstand th~step~wal~d form of the pyramid, gration of this nocturnal species. consider that N = 2 years can correspond to 530 H. LEE JONES AND JARED M. DIAMOND
8 San Nice/as Anocapa
11 .e
1 1 - .
1 ? .
164
I I I I I I I I 1 I I I I I I I I I I 0 2 4 6 8 0 2 4 6 8 N (years) N (years)
FIGURE 2. Turnover pyramids for San Nicolas and Anacapa islands, based on the survey years listed in the text. Numbers above the points are the number of species with the indicated ( #T,N) value. The value for the number of absentees (#T = 0 = N) was obtained by subtracting the number of species that have bred on San Nicolas or Anacapa from the 187 species that breed on the southern California mainland.
#T = 1 case (e.g., 12E345678910), 2 cases capa in one or more of nine survey years, 11 (e.g., 1123E45678910),3 cases (e.g., 132345 were established as having participated in turnover. Two species turned over three times 16E78910) or 4 cases (e.g., 1213E4516E?8910), each, two turned over twice each, and seven b¬ #T = 0 or 2 5 cases. - - others turned over once each. Of the 17 spe- Points can be anywhere corresponding to cies that bred on San Nicolas in one or more integral values of N and #T within the pyra- of eight survey years, one species turned over mid, so that species can form a virtual con- three times, two turned over twice, and nine tinuum in their turnover characteristics. How- more turned over once each. Of the remain- ever, it is convenient arbitrarily to divide and ing 17 species on these two islands, for which name different parts of the pyramids’ space as turnover was not proved, some may neverthe- corresponding to different “types” of popula- less have participated in undetected turnover, tions. Points at N = 10, #T = 0 correspond to since proof of nesting in all years was not ob- “regular breeders” that never turn over within tained for most species. However, since evi- the survey period. Points at N = 0, #T = 0 dence of breeding absence is also not com- also correspond to species that never turn over, pelling for these remaining species in any but because they never breed (“absentees”). year, we have conservatively assumed no turn- Points at #T = 1 represent either “terminal over for them, and figure 2 gives minimum extinctions” (species that breed in the first estimates of turnover. years censused, disappear, and do not return) We present turnover findings for the Chan- or else “terminal immigrations” (species that nel Islands by assigning all populations ever are absent in the first years censused, colonize, known to have bred to one of several cate- and remain). Points at N = 1, #T = 2 repre- gories discussed below, proceeding from zero- sent “flukes”: improbable colonists that man- turnover populations to flukes. Since the cate- age to breed one year but immediately dis- gories intergrade, assignments to categories a.ppear. The remaining points (#T 2 2, N are of course somewhat arbitrary. Popula- 2 2) constitute “in-and-out species” that fre- tions of one species may belong to different quently immigrate and become extinct again, categories on different islands. For example, and that may be further designated either as the Red-tailed Hawk (Buteo jamaicensis) is occasional breeders (N i 5) or frequent a regular breeder on Santa Cruz, Santa Cata- breeders (N 2 5). lina, and Santa Rosa, absent on San Nicolas Figure 2 gives two examples of actual turn- and Santa Barbara, a terminal immigrant on over pyramids, based on Anacapa and on San San Miguel, and an in-and-out species on San Nicolas. Of the 23 species that bred on Ana- Clemente and Anacapa. TURNOVER ON THE CHANNEL ISLANDS 531
ZERO TURNOVER: REGULAR BREEDERS ostoma redivivum) . Other species not re- Jones (1975, in prep.) summarized the popu- corded from the islands are strong fliers such lations that are known or assumed to have as the Red-shouldered Hawk (Buteo Zineatus) bred in each survey year, and for which there or average fliers such as the Downy Wood- is no evidence of turnover. Only 12 of the 56 pecker ( Picoides pubescens) , Nuttalls’ species known to have bred on the Channel Woodpecker (P. nuttallii) , Common Screech Islands are not known to have participated in Owl (Otus asio), Plain Titmouse (Pamcs inor- turnover. Some regularly breeding popula- natus), and Brown Towhee (Pipilo fuscus). tions have undergone large fluctuations in Mainland populations of all these species are, breeding abundance that have at times nevertheless, very sedentary. Still other spe- brought them close to extinction. For exam- cies are strong fliers and long-distance mi- ple, on Anacapa, the House Finch (Carpoda- grants overland but, nevertheless, have not GUSmexicanus) was one of the three common- been recorded on the Channel Islands. The est species in 1910 (Willett 1910), was Black-chinned Hummingbird ( Archilochus considered not particularly common by 1963 abxandri), for example, migrates to central and 1964 (Banks 1966)) had declined to about Mexico from coastal southern California, yet six individuals by the 1968 breeding season (J. has never been recorded from the Channel Is- D.s‘ observations), and recovered to about 15 lands. individuals in 1973, 30 in 1974, and 20-30 in In a second category are species that do 1975 and 1976 (L. J.s‘ observations). On the reach islands with suitable habitat but arrive large island of Santa Rosa, the Northern Mock- so rarely or in such low numbers that the ingbird (Mimus polyglottos) was “common” chances of establishing a breeding pair are in 1927 (Pemberton 1928), was widespread low and breeding success has not materialized and numbered about 400 individuals in 1968, to date. For example, in 1927, around the time but had declined to about 30 sparsely dis- that Acorn Woodpeckers invaded and estab- tributed individuals by 1973. Conversely, the lished themselves on Santa Cruz, two indi- Orange-crowned Warbler (Vermivora celata) viduals were seen on Santa Rosa (Pemberton on Santa Rosa was common in 1927, had be- 1928), which offers considerable areas of suit- come confined to Lobo Canyon and numbered able habitat. None has been seen there since, only about 20 individuals in the 1968 breeding so that these few individuals must soon have season, but had again become ubiquitous in disappeared, although an invasion of more all wooded canyons and in chaparral and colonists might have succeeded. From Febru- coastal sage scrub by the 1973 breeding sea- ary 1974 to late April or May 1975 a single son, when the estimated population was 1200. Red-tailed Hawk was resident on San Nico- Thus, even populations with a high time- las, but a second individual did not arrive. averaged abundance on large islands occa- Recause this hawk breeds on San Miguel, sionally crash to dangerously low levels. In which has similar habitat and a smaller area, the long run, it will be the frequency and breeding might well have taken place on San magnitude of these crashes (the temporal co- Nicolas had a mate arrived. Since August efficient of variation), as shown by Leigh 1973 a single Canon Wren (Catherpes mexi- ( 1975), rather than the time-averaged abun- canus) has been resident on Santa Cruz in a dance, that determines the extinction prob- habitat typical of this species. To date, a mate ability of these large populations. for this individual has not arrived. Acorn Woodpeckers, Cation Wrens, and Red-tailed ZERO TURNOVER: ABSENTEES Hawks all differ from the first category of Species that have never bred on some or all absentees in that they do occasionally reach islands can be subdivided into at least five islands, but their colonizations are infrequent. categories. First, some species that breed on A third category of absentees consists of the southern California mainland in habitats represented on the islands have never been species that reach islands but find no suitable recorded from the Channel Islands, even as breeding habitat. In this category are the vagrant individuals. Some of these species are dozens of Sierran and northern species that such weak fliers that they may be physically flood all the islands every year on spring or incapable of reaching even the nearest island fall migration or in the winter but breed on no ( Anacapa, 20 km from the mainland). Such island. Other examples are numerous vagile species may include the Greater Roadrunner species that reach all islands each year and (Gsococc~~x caZifornianus), Wrentit (Cha- breed only on those islands that offer suitable maea fasciata), and California Thrasher ( Tox- habitat (e.g., Common Flicker, Colaptes aura- 532 H. LEE JONES AND JARED M. DIAMOND tus, and Western Flycatcher, Empidonax dif- on Santa Cruz may initially have involved only ficiUs). a single pair, whose offspring returned. Two The fourth category of absentees includes examples support this interpretation. First, the species which often reach islands with suit- island population of Western Flycatchers is able habitat but whose failure to breed is considered an endemic subspecies, of which probably due to the presence of established, all individuals leave each fall to winter in abundant competitors. For example, the wide- Central America and return the following spread abundance of Allens’ Hummingbird spring. Second, banding studies on the main- (Selasphorus sasin) may explain the failure land have shown that Cliff Swallows do return of Annas’ Hummingbird (Calypte anna) and year after year to the same nest site, whereas Costas’ Hummingbird (C. costae) to establish Barn Swallows (Hirundo rustica), which have lasting breeding populations on most islands. successfully colonized all eight islands, are Likewise, among the four species of sparrows much less philopatric ( Speich, pers. comm. ) . that breed on the islands, the breeding absence of some species on islands where they might TERMINAL IMMIGRATIONS be expected may be due to the established breeding presence of other species. The following populations became established The last category of absentees consists of on islands, have continued to breed, ap- long-distance migrants that winter south of parently without interruption, and are suf- the United States, breed in habitats on the ficiently numerous that they are likely to sur- southern California mainland also represented vive some time longer. on the islands, apparently lack close competi- The sole island on which American Kestrels had tors on the islands, and flood the islands dur- been nroven to breed in the early decades of this ing spring migration each year, but neverthe- century was Santa Cruz (nest records in 1906 and less fail to breed. These species present one 1919). Evidence that thev actuallv did not breed is adequate for at least five islands. early visits to Santa of the most puzzling problems in the island Barbara (in 13 years between 1863 and 1918) and avifauna. Examples are the Ash-throated Fly- Santa Catalina (in 21 years between 1861 and 1919) catcher (Myiarchus cinerascens) and Black- failed to record this conspicuous species at all. It was headed Grosbeak (Pheucticus melanocepha- recorded on San Clemente in 1907 but not in 15 other years between 1863 and 1920. San Miguel yielded Zus), for which the arborescent chaparral and only one specimen taken at an unspecified date in woodland of Santa Catalina, Santa Cruz, and 1910 and no observations during visits in 1927, 1938, Santa Rosa provide typical breeding habitat; and 1939. On San Nicolas it was recorded only as a the Warbling Vireo (Vireo gilvus) and North- winter visitor prior to 1975 and did not breed until 1976. The early status on Anacapa and Santa Rosa ern Oriole (Icterus galbula), for which ri- is uncertain, because there was a single March record parian woodland on the same three islands for the former in 1911 without proof of breeding, and seems equally suitable; and the Cliff Swallow because ornithological exploration of the latter before (Petrochelidon pyrrhonota). While one can the first kestrel record in 1927 was incomplete. At present the American Kestrel is breeding on all is- never refute the ad hoc interpretation that the lands, with nests found on five islands and type-s, -t, islands lack some subtle ecological factor im- and -j evidence (see p. 528) available for three is- portant to these species but invisible to orni- lands. thologists, the island habitats seem to us and American Coots (Fulica am~icanu) first appeared to others with whom we have discussed these on Santa Catalina in the winters of 1928 and 1929 (Meadows 1929) following construction of a man- species well within the broad range of habitats made freshwater reservoir, where coots have bred an- occupied on the mainland. Tn addition, the nually since at least 1967. Ash-throated Flycatcher and the Black-headed There have been 2-5 American Oystercatchers Grosbeak refuted this interpretation by start- (Hacmatopus palliatus) on Santa Cruz since at least 1966, where they are apparently interbreeding freely ing in the 1960s’ to breed annually in small with Black Oystercatchers ( H. bachmani) . numbers and in the expected habitat on Except for one winter record on Santa Cruz in 1907, Santa Cruz (p. 533), although they had there were no observations of Killdeers (Char~drius already previously been recorded as spring vociferus1 on anv island until 1927 I Sheldon. unoub- lished field notes). They have been breeding >esi- migrants on Santa Cruz and are still migrat- dents on Santa Catalina, Santa Cruz, and Santa Rosa ing annually through similar habitats on Santa since at least 1968, and possibly since at least 1947 Catalina and Santa Rosa without breeding. on Santa Cruz. None of the groups of observers who visited San We can only guess that these are strongly Miguel prior to 1968 recorded Allens’ Hummingbird. philopatric species which return annually to Since 1968 this species has been a common breeder the same breeding area despite great distances there. The conspicuous Acorn Woodpecker was unrecorded between breeding and wintering sites. The from any island until it invaded Santa Cruz some time establishment of the flycatcher and grosbeak between 1927 and 1930 (Hoffman 1931) and subse- TURNOVER ON THE CHANNEL ISLANDS 533
quently became a widespread resident breeder. It and-out changes and reasons for the disappearance was first recorded from Santa Catalina in 1955 at the will be discussed later. southeast end (Miller 1955), and by 1968 had spread G. Willett found “quite a few [Bushtit (Psaltriparms over the southeast half of the island as a resident minimus) 1 in the western oak region of Santa Catalina breeder but had not yet reached Isthmus in the north- in the sarincrs of 1904 and 1905” (Howell 1917: west part, and was at Isthmus by 1973. A few indi- loo), th&e being the first years for which consider- viduals reached Santa Rosa in 1927, and one indi- able ornithological information about Santa Catalina vidual reached Anacapa in 1974, without evidence of is available. Subsequent observers have not found this establishing breeding populations. population again despite specific searches, although As mentioned previously, the Ash-throated Fly- the Bushtit is still resident on Santa Cruz. Willett catcher has bred in small numbers on Santa Cruz at did not comment on the breeding status of this species least since 1968. It is still a migrant on all other is- on Santa Catalina. However, his observations prob- lands, as was formerly also true on Santa Cruz. ably pertain to a former breeding population rather Common Starlings (Sturnus vulgaris) were intro- than non-breeding vagrants, because: (a) the Bush- duced by man to the eastern United States from Eu- tit is one of the most sedentary North American bird rope in the 1880s.’ They spread to the California main- species, never having been recorded from any other land by 1944, reached Santa Cruz in 1964, San California island (including Farallones) except Santa Clemente in 1965, and were breeding on all islands Cruz, where it breeds, and Anacapa (one record); except Santa Barbara by 1968 and Santa Barbara by (b) the oak woodlands of Santa Catalina are typical 1972. of the habitat in which the Bushtit breeds on Santa The nearly-endemic island race of Orange-crowned Cruz and the California mainland; (c) quite a few Warbler (V.C. sordida) was not found breeding on were seen in two successive years. San Nicolas prior to 1968, despite visits in 17 years Endemic races of the Bewicks’ Wren (Thryomanes and complete surveys in four years, including 1962-3. bewickii), Rufous-sided Towhee (Pipilo erythroph- By 1968 about eight pairs of sordida were nesting, and thalmus), and Song Sparrow (Melospizu m&o&) this had increased to 20-30 ’ pairs in 1973, and 3040 were formerly common residents of San Clemente but pairs in 1974-6. disappeared due to habitat destruction by goats. The The House Sparrow ( PmSf?T domesticus), another first disappeared by 1968, the second at least by 1976, European exotic introduced to the United States in and the third in 1973. An unmated male Bewicks’ the 19th century, now breeds on San Clemente, Santa Wren that was observed in 1973 and captured and Catalina, and San Nicolas. It colonized these islands photographed in 1974 apparently did not belong to between 1907 and 1915, between 1911 and 1928, and the endemic race and may have been a vagrant from between 1945 and 1961, respectively. another population (Stewart et al. MS). Populations Western Meadowlarks (Sturnella neglecta) were of Song Sparrows and of House Finches disappeared first recorded in 1945 on San Nicolas by Rett ( 1947), from Santa Barbara between 1959 and 1968 following who found several singing in the spring and still pres- habitat damage by rabbits and a fire. ent in the fall. All subsequent visitors have found Hooded Orioles (Icterus cucdutus) may have bred them common, widespread breeders. Colonization regularly on Santa Catalina, nesting having been must have occurred between 1940 and 1945, since proven in 1908, 1911, and 1968 and being probable none was reported on visits in 10 years between 1863 in 1906. They have not bred in 1973-6. and 1940. Like the Ash-throated Flycatcher, the Black-headed IN-AND-OUT POPULATIONS Grosbeak has bred in small numbers on Santa Cruz at least since 1968, but is only a migrant on other is- Most cases of turnover involve populations lands, as was also formerly true on Santa Cruz. that alternatively breed and disappear for Rufous-crowned Sparrows ( Aimophila ruficeps) varying numbers of years. Each of these popu- and Chipping Sparrows ( Spizella passerina) were first recorded on Anacapa in 1940 and have been breeding lations generally comprises only a small num- at least since 1963. though with considerable ber of individuals and remains subject to high fluctuations in abundance. risk of extinction, for one or more of several reasons: (a) the island is very small (cf. nu- TERMINAL EXTINCTIONS merous cases cited below for Anacapa and The following populations bred for many Santa Barbara); (b) the island is large, but years, disappeared, and have not recolonized. the species requires a very large territory (cf. cases of the large hawks and owls cited be- Bald Eagles (Haliaeetus leucocephalus) formerly low); (c) the island is large but the species bred on all eight islands (nests found on seven is- lands, type -j and -t evidence on Santa Barbara). The occupies a specialized habitat (cf. cases of last documented nesting was in 1949 on Santa Rosa American Coot and Red-breasted Nuthatch (egg set in WFVZ), though birds were subsequently [Sitta can&e&s] ) ; ( d) the island is large, seen there until about 1958. Reasons for the dis- and suitable habitat is extensive, but the spe- appearance will be discussed later. cies ’ numbers are severely limited by the pres- Ospreys (Pan&on haliuetus) formerly bred on San Nicolas, Santa Catalina, and San Clemente. The last ence of a competitor (cf. several cases involv- documented nesting was on San Clemente in 1927 ing hummingbirds), On the smaller islands a (egg set in WFVZ). majority of the populations may be in-and-out Peregrine Falcons probably bred formerly on all is- populations: for instance, of the 17 species re- lands except, perhaps, San Nicolas, nesting having been proved on five islands. The last documented corded as having bred in the past 78 years on nesting was on Anacapa in 1939 (Sumner MS). In- San Nicolas, 12 are known to have exhibited 534 H. LEE JONES AND JARED M. DIAMOND
turnover there. In-and-out changes are the Several individuals were present north of Isthmus major source of error in long-term turnover and on Bird Rock in the summer of 1975, but breeding evidence was not obtained. There were no studies, because two surveys at an interval of records of this species for Santa Catalina before more than one year may find a species either 1966 nor from 1967 to 1974, despite a careful search present in both years or else absent in both in the summer of 1974. years but cannot reveal the number of inter- A pair of Mourning Doves (Zen&da macronra) bred on Anacapa in 1970, and may also have done so vening extinctions and immigrations. For in 1968. Only occasional non-breeding individuals instance, an observer who visited San Nicolas were seen in 1963, 1973, 1975, and 1976, and none at only in 1968 and 1974 would have found all was seen before 1958 nor in 1974. Northern Mockingbirds breeding in both years; Because Barn Owls are nocturnal and easily over- one might have assumed no turnover, and re- looked, we have made no assumption of turnover on any island. However, turnover probably has occurred mained unaware of the four cases of turnover at least on Santa Barbara, where there were no rec- that actually intervened (an extinction be- ords before 1968 but up to eight owls present in each tween 1968 and 1969, immigration between survey year from 1968 to 1976. Breeding was proved 1971 and 1972, extinction between 1972 and in 1975. Breeding has also been proved recently on 1973, and immigration between 1973 and San Miguel and Santa Catalina. Similar difficulties arise in documenting turnover 1974). As shown in figure 3, and as is dis- of Burrowing Owls, but evidence is available for Santa cussed in more detail by Diamond and May Barbara and San Nicolas. On Santa Barbara three to (in press), the in-and-out effect causes apparent six pairs were resident, seen virtually daily, and regu- turnover rates calculated from surveys several larly flushed from burrows in 1968, 1972, and 1973. There were one or two pairs in 1974, and two pairs decades apart to be underestimates of the in 1975 and 1976 (type-s and -t breeding evidence). true value by approximately an order of The species was not found on this small island until magnitude. As further yearly surveys of the 1927 despite active searches for it, was scarce or ab- islands have become available, it is becoming sent in 1950 and 1958, and common from 1953 to 1957. On San Nicolas in 1963, Townsend (1968, clear that many populations considered to be amplified by personal communication) found a year- regular breeders or to have undergone termi- round population and a defended burrow (type-s nal immigration or extinction are actually and -t breeding evidence). There are also records of uncertain breeding significance for 1886 and 1945. participating in in-and-out turnover. The However, the sole San Nicolas records for 1968, 1973, known examples follow: 1974, and 1975, despite a total of 36 visits and over 1000 hours of field observations by Jones, Schreiber, The Red-tailed Hawk nested on Anacapa in 1927 Diamond, and others, were of single sightings in Oc- (Pemberton 1928); was not even present in 1910-12; tober 1974 and September and November 1975. Sev- did not nest in 1968 or 1974, when only stray indi- eral individuals were seen between January and May viduals were seen on one or two occasions, nor prob- 1976, but all had left by mid-May. Thus, one ex- ably in 1970; and the status is uncertain for 1939, tinction between 1963 and 1968 may be assumed con- 1973, 1975, and 1976, when apparently territorial, servatively. Breeding has also been documented on vocal pairs were present but a nest could not be lo- Santa Catalina and San Clemente. cated. Several pairs nested on San Clemente in 1907 A pair of Long-eared Owls (Asia otus) nested on (Linton 1908) but not in 1897, probably not in 1915, Santa Catalina in 1909 and probably again in 1910 and not in 1968 nor 1973-6. On San Miguel it did not ( Howell 1917). Since night observations by residents nest in 1938-9 (Sumner MS). One or two pairs were and visitors on Santa Catalina since 1968 have proven resident there in 1968, 1972, 1973, 1975, and 1976, breeding of three owl species and frequently detected and nesting occurred in at least one of these years, two non-breeding caprimulgids, the lack of any rec- but it is unclear exactly when and how often turn- ords for Long-eared Owl since 1910 is considered sig- over occurred. nificant and constitutes evidence of extinction. The Peregrine Falcon bred on Santa Barbara in At least one Annas’ Hummingbird bred, or at- 1908 (Howell 1917) but was not observed in 1897 tempted to breed, on San Clemente in 1915 (type-m during a 6-day stay by Grinnell ( 1897), and could evidence, see p. 546 for details; Howell 1917) but hardly have been overlooked if it had been present not previously or since. It is probably limited by on this tiny island. It bred on San Clemente in 1907 competition from the much commoner resident Al- (Linton 1908) and 1915 (Howell 1917), but was lens’ Hummingbird. (Both species breed on Santa not found in 1897 during 7-day and 11-day stays by Catalina and Santa Cruz, but Annas’ is much less Grinnell ( 1897)) although his camp was less than 3 common). Costas’ Hummingbird probably bred on km from a 1907 nest site. San Miguel in 1968, where Allens’ is common, but Bald Eagles bred on Santa Barbara in 1911 and conclusive evidence is lacking. There are no records 1912. It is possible but not certain that they failed of Costas’ Hummingbird on San Miguel in other years. to breed there in 1897, when Grinnells’ (1897) sole It bred on Santa Barbara in 1911 (Willett 1912) but observations for his g-day visit read, “seen on two was not recorded in 1897, and there has been no indi- occasions flying over the island.” cation of breeding there in the past nine years. A pair of American Coots bred on Santa Cruz in 1922 One pair of Black Phoebes (Sayormis nigricuns) in the marsh at Prisoners Harbor (Dickey and van Rossem 1923 ), but there are no previous or subsequent nested on Anacapa in 1970, but in no other year ex- indications of breeding. cept possibly 1941 (specimen collected in March and A pair of Black Oystercatchers bred on Bird Rock labelled “breeding”). An unfinished nest was found at Santa Catalina in 1966 (Harper 1971) and 1976. on San Clemente in 1907 (Linton 1908), but there TURNOVER ON THE CHANNEL ISLANDS 535
are no breeding-season records for other years despite 1975, and August 1976). There are no earlier records thorough searches of all available habitat in 1968 and for Anacapa, although this same canyon was checked 1973-g including the area where Linton found the frequently. Huttons’ Vireo is generally sedentary and 1907 nest. not migratory, so that these birds are unlikely to be On Anacapa, the Horned Lark (Eremophila al- winter visitors or migrants. Because the National Park pestris) was one of the three commonest land birds Service forbids landing on West Anacapa between in June 1910; breeding was documented in 1927 and mid-March and early August in order to protect the 1932 (egg sets in WFVZ); several were seen in 1934, Brown Pelican colony, we have not witnessed breeding but no other details are available; none was found in in this case. We infer a breeding pair from singing, April 1939 or May 1940; a breeding male was col- territorial behavior, and prolonged residency of this lected in 1941: none was observed in 1962. 1963. or sedentary species. 1968, and only one uncertain record in 1964; several One pair of Orange-crowned Warblers bred on territorial pairs were flushed underfoot from apparent Santa Barbara in 1975, and probably in 1939 (Hunt nest sites in April and May 1970; and none bred in and Hunt 1974), but not in 1968, 1972, 1973, 1974, 1973-6. Thus, one must assume at minimum an ex- or 1976; nor did visits before 1939 or in 1953 through tinction after 1932, an immigration between 1968 and 1958 provide any indication of breeding. 1970, and another extinction between 1970 and 1973. Pemberton (1928) found a colony of House Spar- One pair of Barn Swallows nested on San Nicolas rows in 1927 on Santa Rosa. The population persisted in 1975 and 1976, but the species had not been re- at least until 1951 but has more recently been ab- corded even as a migrant until September 1973. On sent (at least since 1968). There have been no records Santa Barbara either one or two pairs bred in 1975 on Santa Cruz since a sighting in 1915 (Dawson (type-m and -t evidence) and probably bred in 1912, 1915). but not in 1897, 1968, 1972, 1973, 1974 or 1976. On A colony of up to 40 pairs of Red winged Blackbirds San Clemente a few have been present locally in each (Age&us phoeniceus) bred on Santa Cruz from at summer and bred in at least some summers since least 1968 until 1975 in willows and cattails at Prison- 1968, and possibly bred in 1915, but were not seen ctrs Harbor, where one individual was also collected in other years before 1915. in 1948. None of the numerous earlier observers on One pair of Common Ravens (Corvus corux) nested Santa Cruz found this species, although many of on Anacapa in 1911 (Burt 1911) and possibly in 1939; these observers used Prisoners Harbor as their base. records for other early years are lacking; none was It did not breed in 1976. present in the 1963, 1964, 1968, 1973 or 1976 breed- Lesser Goldfinches ( Curduelis psaltria) bred on ing seasons; and one or two individuals were resi- Santa Rosa in 1927 (Pemberton 1928), 1974, and dent during the 1970, 1974 and 1975 seasons, but probably 1976, but not in 1968 or 1973. breeding was doubtful. Ravens bred on San Miguel in 1939 (Sumner MS) and probably in earlier years, “FLUKES” but none has been seen in recent years. On Santa Barbara ravens were recorded by at least five ob- In the following three cases a species which servers between the 1860s’ and 1939, bred on at least a priori must have been considered unlikely to some occasions, were not seen by Wright and Snyder colonize a certain island nevertheless did (1913) in 1912, and did not breed in 1968 or 1972-6. Red-breasted Nuthatches and possibly Coopers’ breed there one year, disappeared the follow- Hawks (Accipiter cooperii) nested in the-pine forest ing year, and has not bred since. State bird of Santa Cruz in 1911 (Howell 1917). The nuthatch lists provide numerous examples of the main- was present and apparently breedingin’ the summers of 1971 and 1976, absent in other recent summers. land analogues of these island flukes. One pair of Northern Mockingbirds bred on San Lesser Goldfinches breed in woodland and ar- Nicolas in 1968, 1972, 1974, 1975 and 1976, but not borescent chaparral on Santa Cruz, Santa Catalina, in 1963 or any earlier survey year, nor in 1969, 1970, and (intermittently) on Santa Rosa, and frequently 1971, or 1973. reach other islands without breeding. Because An American Robin nest was found on Santa Cruz San Miguel lacks such habitats, is relatively in 1935 (E. Harrison, pers. comm.; Badger, unpubl. barren, and has practically no vegetation over 2 m field notes in WFVZ ). high, San Miguel would have been considered un- At least one pair of Swainsons’ Thrushes (C&arms suitable for breeding of this species. Nevertheless, one ustulutus) nested on Santa Catalina in 1976 and pos- pair nested in May 1975 in open Bucchuris-grassland sibly also in 1963 at Eagles’ Nest, which supports habitat. the lushest riparian vegetation on the island. In Brewers’ Blackbird (Euphugus cyanocepha2us) has 1968, 1973, and 1975 we and other observers searched often reached all the islands as a spring or fall mi- the Eagles’ Nest area frequently but did not find this grant. The sole breeding on any island is for San thrush in the breeding season. Nicolas in 1963 and possibly 1962, when a group of One pair of Loggerhead Shrikes (La&us Zudovi- at least two pair nested (Townsend 1968). &anus) nested on Santa Barbara in 1906 (WFVZ). White-crowned Sparrows (Zonotrichiu Zeucophrys) A pair nested on Anacapa in 1899 (Willett 1912), winter abundantly on all the islands. There are no 1973 and 1976, and one or two pairs probably in records of breeding nor indeed of summer residency 1968, but none in 1906612, 1934, 1939, 1963, 1964, except for San Nicolas in 1963, when Townsend 1974, or 1975. One pair bred on San Miguel in ( 1968, amplified by personal communication) ob- 1973, probably also in 1968 and 1975, and possibly served small numbers resident through the summer in 1976, but there are no records for earlier years. in the east-central part of the island, including fledg- A territorial pair of Huttons’ Vireos has been found lings being fed by adults. The White-crowned Spar- in a wooded canyon of West Anacapa on all visits to row breeds in similar habitats on the mainland coast, this canyon for the past four years (March, August, but this record represents a temporary southward and November 1973, March 1974, March and August range extension of about 120 km. 536 H. LEE JONES AND JARED M. DIAMOND
DISCUSSION or a decrease, depending on the forests’ ex- tent. As will be seen from the following exam- EFFECTS OF MAN ON TURNOVER ples, mans’ long-term effect on the Channel Much of Johnsons’ (1972) and Lynchs’ and Islands avifauna has been disproportionately Johnsons’ (1973, 1974) criticism of published to eliminate high-turnover populations and to turnover studies focuses on effects of man. create low-turnover populations. The types of These authors argued that man, by modifying human effects are as follows: habitats and eliminating bird populations, has 1. Direct introduction of species to islands. inflated turnover rates well beyond the values Man has successfully introduced California that would otherwise have been observed. Quail (Lophortyx californicus) and other game The actual problem is more complex: man species to several islands: one species each to has acted both to increase and to ’ decrease three islands, two species each to two islands. turnover, in several direct and indirect ways, These introductions have no direct effect on some immediately obvious and some less calculated turnover because we do not count obvious. Mans’ most direct effects have been them as immigrations. They might have an in- the introduction of populations to islands, and direct effect by competing with native popu- the elimination of native populations by shoot- lations and potential immigrants. The indirect ing and other means. Less direct effects have effect of introduced species must be enormous been mediated by mans’ destroying, creating, on Hawaii but is slight on the Channel Islands, or modifying habitats; by directly or indirectly where introductions have been few, relatively increasing or decreasing the mainland pool unsuccessful, and confined to one bird guild. of potential colonists; and, perhaps most im- The native population or potential immigrant portant in the long run, by shifting the bal- most likely to be affected by competition from ance, through these various effects, between the introductions is the California Quail itself, high-turnover and low-turnover populations but it is a weak flier of which no vagrants on an island. As a model for understanding have even reached an island, and its sole ap- how these different effects may either in- parently native island population (on Santa crease or decrease turnover, consider the small Catalina) may have been introduced by Amer- area of pine forest on Santa Cruz, where the indians (Johnson 1972). Red-breasted Nuthatch breeds intermittently 2. Direct elimination of species from is- but turns over because it cannot build up a lands. Populations can be eliminated directly sufficiently large population to survive for by shooting, egg collecting, poisoning, destruc- long. If man destroyed such a forest, or cre- tion of nests, use of pesticides, and release of ated a forest on an island initially lacking one, predators, as opposed to indirectly by changes the immediate effect would be to inflate turn- in habitat. The only clear examples among over by one case, the nuthatchs’ terminal ex- land and fresh-water bird populations of the tinction or terminal immigration, respectively. Channel Islands involve the terminal extinc- In the long run the effect on the turnover rate tions of the Bald Eagle, Peregrine Falcon, and of the island avifauna would variously be: if Osprey from all islands on which they formerly a small forest were destroyed, to decrease bred as well as from the southern California turnover (by eliminating a rapidly turning mainland. These extinctions are surely due over population); if a small forest were al- somehow to man, but the exact causes are in- lowed to expand in area, to decrease turn- adequately understood. While the widespread over (by increasing the nuthatchs’ population use of insecticides after 1947 was probably the and reducing its risk of extinction); if a small final blow to the eagle and falcon, both seem or large forest were created on an island ini- to have been declining in numbers before tially without one, similarly to increase or de- then. Ospreys had become generally rare in crease turnover, respectively. (Recall that a California by 1944 ( Grinnell and Miller 1944 ) , faunas’ turnover rate T, expressed as percent- last bred definitely on the Channel Islands in age of the breeding species turning over per 1927, and probably lingered on the islands year, is the number of cases of turnover per only until the early 1930s,’ long before intro- year divided by the average number of species duction of insecticides. Various islands now S breeding in a single year: see eq. 1, p. 539. have feral cats and rats, and Banks (1966) Addition of species that do not turn over de- speculated about their relevance to temporary creases T by increasing S.) If man extermi- declines (not extinctions) in Rock Wrens (Sal- nated nuthatches or introduced a nuthatch pinctes obsoletus) and House Finches on Ana- species on the mainland, the long-term effects capa, but there is no evidence to support on turnover could again be either an increase this speculation. TURNOVER ON THE CHANNEL ISLANDS 537
3. Additions to the mainland species pool. This prediction is confirmed by the numerous The Common Starling, House Sparrow, Rock in-and-out cycles that have been observed for Dove (Columha U&a), and Spotted Dove large hawks and owls on the Channel Islands. (Streptopelia chinensis), which were re- In-and-out cycles were recorded for Peregrine leased in North America by man and now Falcon, Red-tailed Hawk, and possibly Bald breed in California, have colonized 8, 4, 1, and Eagle on one or more islands even before 1915, no Channel Islands, respectively. An indirect and the absence of annual surveys at that time addition to the southern California mainland makes it likely that many other in-and-out pool is the Brown-headed Cowbird (Moloth- cycles went undetected. The effect on turn- TUSater), which was virtually absent from the over of eliminating the three large raptors has Pacific slope of southern California in 1900 therefore been, in the long run, to remove one but had become common by 1930 in associa- of the most rapidly turning-over groups in the tion with agriculture. To our knowledge it island avifauna, and to reduce turnover by all has never bred on a Channel Island. The lat- the in-and-out cycles that the raptors would ter three species make no contribution to our have undergone if they had not been elimi- turnover calculations (because the Rock Dove nated. may have colonized its sole island of breed- 5. Terminal extinctions or immigrations ing, Santa Catalina, before the first bird sur- due to habitat changes. Some terminal ex- vey), while the Common Starling has con- tinctions or immigrations may be direct conse- tributed eight terminal immigrations, the quences respectively of mans’ destroying or House Sparrow three terminal immigrations creating habitats. As background for identi- and (on Santa Rosa) one in-and-out cycle. fying such cases on the Channel Islands, re- The 11 populations that have contributed no call that each Channel Island was hardly in further turnover since their initial immigra- pristine condition at the time of the first tion are all sufficiently well established now thorough bird survey (some year between that they seem also unlikely to contribute to 1897 and 1939, depending on the island). Be- turnover in the near future. Thus, the immedi- fore Europeans arrived, the islands had been ate effect of these mainland pool additions inhabited for at least 12,000 years by Amer- was to generate cases of turnover, but the indians (Orr, in Philbrick 1967), who may long-term effect thereafter has been to reduce have eliminated the Channel Islands popula- avifaunal turnover rates for each island, by tion of the extinct flightless duck (Chendytes diluting the island avifauna with zero-turnover Zawi; Morejohn 1975). Grazing animals had species. been introduced to all islands by Europeans 4. Eliminations from the mainland species at least by the first half of the 19th century, pool. Three members of the island avifauna possibly earlier on most islands. Records of have been virtually eliminated from the south- the number of sheep or goats maintained, and ern California mainland pool since the first accounts of early pioneers, show that grazing thorough island bird surveys in 1897: Bald on many islands was heaviest in the mid-19th Eagle, Peregrine Falcon, and Osprey, as dis- century. Thus, transformation of large areas cussed above. Elimination of these large rap- from chaparral and woodland to grassland or tors from the species pool had the immediate sparse shrubs had been accomplished long effect of inflating turnover through terminal before the first bird surveys. The problem is extinctions, but the effect on turnover in the therefore to recognize changes in habitat be- long run has been opposite. These raptors, tween the surveys that could underlie changes like other predatory species or large species, in bird populations. Comparison of the islands were present on the islands in much lower as we see them now with photographs and number than species at lower trophic levels descriptions at the times of the first thorough or species of smaller size. On smaller islands bird surveys suggests the following changes, there were never more than a few breeding as summarized from a detailed account (Jones pairs, often only one pair, of each large raptor 1975 and in prep.). San Nicolas and San Mi- species. Even on an island as large as San Cle- guel, which were already rather barren at the mente (145 km”) Linton found only two pairs first survey, remain relatively barren despite of Peregrine Falcon in 1907. As we will show, termination of grazing within the last three extinction rates increase steeply with decreas- decades. Santa Cruz, Santa Rosa, and Santa ing population size (figure 6). Hence, one Catalina, the three largest and most heavily would expect the large raptors to have had wooded islands, retain the same mix of habi- exceptionally high turnover rates, as discussed tats as on first survey and still have grazing, by Hunt and Hunt ( 1974) and Jones ( 1975). though now more carefully managed than be- 538 H. LEE JONES AND JARED M. DIAMOND fore. San Clemente, already stripped of most seen on Anacapa and Santa Barbara and have of its chaparral before the first survey, has ex- intermittently been summer residents of perienced further reduction of understory doubtful breeding status on Anacapa. (3) thickets in the few ravine woodlands, due to Some terminal immigrants use both natural goats. Between 1918 and 1959 farming, rab- and man-made nest sites, and have had ac- bits, and a fire destroyed most thickets on cess to the latter since long before the first Santa Barbara, the island most changed be- bird survey. For instance, nests of Barn Swal- tween surveys. On Anacapa termination of lows on the islands are usually in natural sea grazing in the 1930s’ has permitted scrub to caves, though sometimes on buildings or piers. expand at the expense of grassland and ice Barn Owls on the islands usually nest on cliffs, plant (Gasoul crystallinum), but extensive though perhaps sometimes on buildings. No areas of the latter habitats remain. change of conditions between earlier and later Five terminal extinctions and one terminal surveys is involved. (4) Some terminal im- immigration are obviously the result of habitat migrants use both natural and man-made alteration by man, as discussed previously in habitats. For example, the Killdeer, which the text: extinctions of the House Finch and colonized three islands, usually requires prox- Song Sparrow on Santa Barbara, and of the imity of fresh water for breeding. On Santa Bewicks’ Wren, Rufous-sided Towhee, and Rosa and Santa Cruz it breeds only in natural Song Sparrow on San Clemente, due to habi- habitats (along streams); on Santa Catalina, tat damage; and immigration of the American in natural habitats (along streams, along the Coot to Santa Catalina, due to habitat cre- coast, and on Bird Rock) and also near sev- ation. These appear to us the only cases in eral man-made reservoirs. which man-induced habitat alteration has Assessment. We accept 119 cases of turn- played a decisive role in terminal immigration over as having occurred since the first reason- or extinctions, but several other types of cases ably complete bird surveys in 1897, counting warrant discussion: (1) Banks (1966), in each immigration or extinction as one case. comparing the breeding avifauna he found on Of these, about 33-39 cases, or 27-33% of the Anacapa in 19634 with that found by earlier total, can be attributed to the effects of man, observers, attributed the decline and (as he depending on how one assessesthe three Ana- believed) extinction of the Horned Lark, and capa cases and the three raven cases discussed immigration of the Rufous-crowned and the above. The man-related cases include 9-11 of Chipping sparrows, to reduction in grassland the 37 terminal immigrations, and most (23-26 and expansion of scrub. This interpretation out of 33) of the terminal extinctions, but only seems at most only partly correct. Anacapa 1 or 2 of the 43 in-and-out changes and none still has approximately 40 ha of habitat simi- of the six fluke changes. However, these fig- lar to that in which Horned Larks are abun- ures greatly exaggerate the contribution of dant on San Nicolas and Santa Barbara, and terminal immigrations and extinctions to total Horned Larks recolonized and bred on Ana- turnover, because comparison of surveys up capa since Banks’ visit. Our censuses of 1968- to 79 years apart detects all terminal extinc- 1976 reveal large year-to-year fluctuations in tions and immigrations but overlooks about numbers of the two sparrows, both of which 90% of the in-and-out changes (fig. 3). A better assessment can be made by consider- nearly disappeared in 1976 despite continued ing only those cases of turnover since 1968, expansion of scrub. On an island as small as because the shorter intervals between surveys Anacapa, even the most abundant species (one to at most five years) mean that fewer has such a small population that turnover due in-and-out changes escaped detection. Of the to year-to-year population fluctuations unre- 29 cases detected between 1968 and 1976, only lated to habitat succession is high. (2) Man four (immigration of the Common Starling may have contributed in several ways to ex- to Santa Barbara, extinction of the Bewicks’ tinctions of the Common Raven on the three Wren, Rufous-sided Towhee and Song Spar- smaller islands (Anacapa, Santa Barbara and row on San Clemente) involve effects of man. San Miguel). However, interpretation of The conclusion that turnover measured at these extinctions remains unclear, because ra- short intervals is less affected by man than vens continue to breed abundantly on the re- turnover at long intervals is hardly surprising. maining five islands in the presence of man; However, the real problem with long-time- breeding on Anacapa and Santa Barbara may base studies is not that they inflate turnover formerly have been intermittent rather than rates by a factor of about 1.5 due to effects of annual; and in recent years ravens have been man, but that they underestimate turnover TURNOVER ON THE CHANNEL ISLANDS 539
IO breeding species that are replaced by other breeding species per year. Unresolved ques- 1. . tions of breeding presence or absence were evaluated conservatively with respect to con- clusions of turnover, as discussed on pp. 528- 529. Thus, the resulting turnover rates are minimum estimates for two reasons: con- servative evaluation of species lists for the survey years, and neglect of in-and-out changes between survey years. Figure 3 plots these apparent values of T against At on a double logarithmic scale for Anacapa, using all pair-wise combinations of . survey years and fitting a straight line through the points. Apparent T values decrease by an 0. I J, / I order of magnitude as the survey interval At I IO 100 increases from 1 year to several decades. From At (years) the fitted line of figure 3, the “true” value of FIGURE 3. Turnover rates were calculated for Ana- T for Anacapa-i.e., that measured at one-year capa Island by eq. 1, from comparison of breeding intervals-is 3.7%/year. The apparent value lists in the survey years listed in the text and taken in measured at the longest survey interval (At = all pair-wise combinations. Ordinate: turnover rate T (%/year); abscissa: interval between surveys At 67 years) is only 0.4l%/year. The reason for (years). Scales are logarithmic. The point with an the decrease in T with increasing At is that arrow denotes a pair of surveys at an interval At = 1 more cases of undetected in-and-out changes that exhibited zero turnover. Note that T decreases are completed within longer intervals. Thus, with At, because of the in-and-out effect. the in-and-out effect causes surveys performed at an interval of several decades to under- rates by a factor of about 10 due to over- estimate turnover by approximately an order looked in-and-out changes (next section). An- of magnitude. Diamond and May (in press) nual censuses minimize this problem. Mans’ provide examples from other island avifaunas, overall effect on the species pool has been to and show how the decline in apparent turn- remove three rapidly turning-over species and over rate and also in its statistical fluctuations to inject two non-turning-over “regular breed- with time, as illustrated in figure 3, is related ers.” Because of this transformation of the mathematically to the immigration and extinc- pool, the turnover rates revealed by annual tion probabilities of each breeding species. censuses within the past decade may be lower Figure 4 depicts the number of breeding than those that would have been observed species S on each island in each survey year. early in this century. The number written over the line connecting each pair of surveys is the percent turnover TURNOVER VALUES (not turnover rate), defined as 100 (I + E)/ ( S1 + Sp) and meaning the percent of the avi- For each island and for each pair of years in fauna that turned over between surveys. Sev- which a reasonably complete survey was avail- eral points can be noted from figure 4: able, we calculated relative rates of avifaunal On no island is S perfectly constant (ordi- turnover, T, based on the 119 cases of turn- nate values of figure 4). However, the fluctu- over discussed on pp. 532-535. T was cal- ations in S are relatively small. The sole culated as instance in which the change in S between T=lOO(I+E)/At(&+Sp) (es. I) surveys exceeds 25% of Ss’ initial value is be- tween 1910 and 1968 on Santa Barbara, where where S1 and Sa are the number of breeding species in the earlier and later survey year S declined by 36% due to the fire-related ex- respectively, I and E are respectively the num- cess of extinctions. S fluctuates relatively more ber of immigrations and extinctions based only on the smaller, more species-poor islands on comparison of the lists for the two years ( Anacapa, Santa Barbara, San Miguel, San (i.e., ignoring any known in-and-out changes Nicolas: average S = 9-17 species, A = 2.6- in the intervening years), and At is the num- 58 km2) than on the larger, more species-rich ber of years between the two surveys. The islands (Santa Cruz, Santa Rosa, Santa Cata- factor 100 in eq. 1 means that the units of T lina, San Clemente: average S = 23-38 spe- are %/year; i.e., the percentage of the islands’ cies, A = 145-249 km2). Excluding the 1910- 540 H. LEE JONES AND JARED M. DIAMOND
Anacapa .
Santa Barbara IO IO-/ 44 I3 0 ‘San Nicolas
I I I I I I I I I I I I I I 1 1900 1910 1920 1930 1940 1950 1960 1970 Year
FIGURE 4. Fluctuations in number of breeding species S for each island (ordinate). The abscissa is the survey year. The number written over the line connecting each pair of the points is the percent turnover be- tween krveys (not turnover rate).
1968 interval on Santa Barbara, the average Barbara (net loss of 4 species due to fire) magnitude of change in S (irrespective of and San Clemente (net loss of 7 species since sign) between surveys is 11% on the small, 1915, partly due to habitat damage by goats). species-poor islands, 4% on the large, species- Even for these two islands, in-and-out turn- rich islands. We interpret this effect as the ex- over has been superimposed on the net de- pected increase in statistical variability, as cline, as several extinctions were offset or population sizes and species numbers decrease reversed by immigrations. with decreasing area. These increased tem- Average values of apparent turnover rates, poral fluctuations in S on small islands are the calculated by averaging values obtained from counterpart of the increased spatial fluctua- all pairs of successive censuses on each island, tions noted previously; if one compares S on are OS-l.O%/year for the four larger islands, islands of similar area censused in the same 1.2-3.6%/year for the four smaller islands. Val- year, the coefficient of variation increases ues calculated only from the one-year survey with decreasing area (Diamond 1974, Dia- intervals of the period 1973-1976 are higher mond and Mayr 1976). (0.9-l.g%/year and 1.6-5.6%/year for the There is considerable turnover in species larger and smaller islands, respectively) and even when S itself is the same or nearly the are more meaningful, because errors produced same in two successive surveys. The only in- by the in-and-out effect at long survey inter- stances in which there is no turnover between vals are avoided. Our best estimates, the one- surveys are in some short intervals of l-5 year values estimated by the curve-fitting pro- years. Three pairs of successive censuses re- cedure of figure 3, are similar: 0.5-1.4%/year veal turnover despite no change in S: San and 0.84.9%/year for the larger and smaller Nicolas, 1897-1945, turnover 13%, S constant islands, respectively. For almost all islands, at 8; San Nicolas, 1963-1968, turnover 30%, these values (based on more surveys than Dia- S constant at 10; Santa Cruz, 1975-1976, turn- mond [1969] used in his turnover calculations) over 3%, S constant at 38. On Santa Catalina are considerably higher than his values (0.3- between 1908 and 1968, S was nearly con- 1.2%/year), his underestimates being greatest stant (33-34), but turnover was 16%. for the islands with the highest turnover rates. The relative constancy of S in the face of The reason for Diamonds’ underestimations is turnover is in accord with the MacArthur- that his survey interval was 51 years, so that Wilson hypothesis that species number on an the in-and-out effect lowered his apparent T island may track an equilibrium value. Non- values below the true ones, as illustrated by equilibria1 trends in S appear only for Santa figure 3. TURNOVER ON THE CHANNEL ISLANDS 541
We know of two other published analyses due to a 24 - Population not turning over 60 ---4 Immigrant populations “““x Extinct populations 20 16 / \ LARGE ISLANDS % E 30- 20- IO- O- I / I I IO 100 1000 lO,OOO Population SW (#Pars) FIGURE 6. Incidence of extinction as a function of SMALL ISLANDS population size, calculated from the data of figure 5. For each breeding population size class (abscissa), the ordinate gives as a percentage the number of pop- ulations in that class that became extinct since the first surveys, divided by the total number of popula- tions in that class. Some colonists who reach the islands fail to breed, either because of the absence of suit- I I I I I I IO 100 1000 10,000 able habitat ( step 2a), or else despite the pres- Population Size (x Pairs) sence of suitable habitat (step 2b) but cor- FIGURE 5. Population size and turnover on the related with the presence of competing Channel Islands. On each island the average breeding species, or else correlated perhaps with strong population of each species was estimated as falling philopatry. Other colonists reach an island into one of nine size classes (l-3 pairs, 4-10 pairs, already occupied by a breeding population of 1130 pairs, etc). Populations were then grouped as conspecifics, into which they are absorbed to whether they immigrated, became extinct, or never turned over since the first surveys. The number of (step 2~). Still other colonists find an unoc- populations (ordinate) in each size class (abscissa) cupied island with suitable habitat and oom- was summed for the four larger islands (above) and mence to breed (step 2d). for the four smaller islands (below ) . The new breeding populations founded by those colonists reaching step 2d survive for conclusion does not mean that island popula- varying periods of time before eventual ex- tions are of two sorts: small ones that are in tinction (steps 3-5). Once colonists com- dynamic equilibrium, and large ones that are mence to breed, the new population may grad- not. Instead, figures 5 and 6 simply mean that ually increase in numbers until it reaches the turnover rates decrease continuously with pop- carrying capacity K, the number of individuals ulation size. of the species that the island can support at equilibrium. Since probability of extinction SYNTHESIS decreases with increasing population size (fig. In this concluding section we present a model 6), many or most new breeding populations summarizing our observations on community must fluctuate out of existence soon after first dynamics. We find it convenient to delimit breeding, before the saturating population K arbitrarily five stages in colonization: dis- has been reached (step 3a; many of the in-and- persal, commencement of breeding, saturation out-extinctions on pp. 533-535 provide exam- of carrying capacity, survival for only a short ples). For most of the immigrants that sur- time, and survival for long times. vive step 3a, K is so low, often just a single As illustrated in figure 7, the process begins pair, that even populations which reach K with the dispersal of colonists to islands from (step 3b) soon disappear (step 4a; see p. 533 the mainland or from other islands (steps la for the numerous alternative reasons for low and lb). Some species have never reached the K, such as small island, large territory, special- islands at all (e.g., Wrentit, Red-shouldered ized habitat, or limitations by competitors, and Hawk), while other species reach every is- pp. 533-535 for numerous examples among in- land every year in numbers. and-out populations). A few immigrants do TURNOVER ON THE CHANNEL ISLANDS 543 FIGURE 7. Model of stages in colonization and turnover. See text for discussion, find an empty island with high K, succeed in breeding populations have appeared on Ana- achieving K, and then survive for long times capa at a rate of 0.67/year over the past de- (steps 4b, 5). Th ese are the immigrants who cade (step 2d), but about half of these fail to might begin to diverge genetically and even- survive more than one breeding season (steps tually might form endemic subspecies. In all, 3a, 4a). Only two species have Anacapa popu- more than half of the man-unrelated immi- lations that exceed 100 individuals and thus grant populations known to have been are likely to survive for a long time (step 4b). founded on the Channel Islands since the first We emphasize that the process illustrated in bird surveys have disappeared again. Most of figure 7 is going on continually, and that a the remainder have achieved only a tenuous “snapshot” of an island at any moment reveals existence and number less than 50 individuals, populations of different ages and at different generally much fewer. Only about six natural stages of colonization. Thus, one survey of immigrant populations (American Kestrel on an island avifauna is like viewing “a single Santa Catalina and San Clemente, Acorn frame of a movie film rather than viewing the Woodpecker on Santa Cruz and Santa Cata- movie itself” (Levins and Heatwole 1973: lina, Orange-crowned Warbler and Western 1056). Populations turn over at various rates, Meadowlark on San Nicolas) now exceed 50 ranging from rapid turnover of the uncommon individuals. These are the sole natural immi- species or recent arrivals that account for most grants who succeeded in achieving K on a turnover, to very slow turnover of abundant large island providing extensive areas of suit- populations. Not only do populations of dif- able habitat. ferent species on a given island differ in turn- Consider the avifauna of Anacapa to il- over frequency, but also the relative propor- lustrate how the mainland species pool be- tions of populations with different turnover comes attenuated from step 1 to step 5. About frequencies differ according to island size. On 345 land-bird species have been recorded from a very small island most populations are small the southern California mainland, of which ca. and relatively short-lived. The larger the is- 187 breed. Of the total pool, 170 have been land, the more the distribution of population recorded from Anacapa, of which ca. 107 are ages becomes skewed to include some large, recorded every year, the others being recorded long-lasting populations (fig. 5). Thus, the less often than annually (step 1). Of these proportion of endemic forms increases with is- IO7 species, ca. 73 do not find suitable breed- land area (Mayr 1965). ing habitat on Anacapa (step 2a), while ca. In short, the practical message of our short- 34 do (steps 2b-2d). Of these 34 species, on time-base turnover studies is that turnover cal- the average 17 were already breeding on Ana- culations based on censuses spaced at long in- capa in the preceding year (step 2~). New tervals reveal only the tip of an iceberg. TO 544 H. LEE JONES AND JARED M. DIAMOND obtain even approximately accurate estimates ACKNOWLEDGMENTS of turnover rates in bird communities requires For access to and general assistance on the islands we annual surveys. thank the personnel of Channel Islands National Mon- ument (San Miguel, Anacapa, Santa Barbara), Pacific SUMMARY Missile Test Center at Point Mugu (San Nicolas, San Miguel), Naval Underseas Center (San Clemente), Al Estimates of fauna1 turnover rates based on Vail and Bill Wallace CSanta Rosa), Carey Stanton, surveys many years apart are likely to be un- Henry Duffield, and the University of ~&&forma Marine Institute at Santa Barbara (Santa Cruz), and derestimates because of the “in-and-out ef- Doug Propst and personnel at the University of South- fect” (a population immigrating and then go- ern California Marine Biological Laboratories (Santa ing extinct, or vice versa, between surveys). Catalina). We also thank the many other people, some Hence surveys of breeding land and fresh- of whom are mentioned in the text, who generously water bird species on the eight California supplied us with their field notes or assisted us directly on the islands. Thomas R. Howell, Kimball Channel Islands have been carried out an- Garrett, and Lloyd F. Kiff made valuable suggestions nually since 1973, for comparison with earlier on the manuscript. surveys. An introductory section summarizes available sources of information on Channel APPENDIX Islands birds, our survey strategies, and our criteria for breeding presence or absence. Johnson ( 1972) and Lynch and Johnson ( 1973, 1974) As depicted graphically by a “turnover pyra- criticized Diamonds’ ( 1969) study of turnover on the Channel Islands, and Lynch and Johnson ( 1974) also mid,” populations of different species on the criticized studies of turnover on Mona Island in the same island fall into categories that turn over Caribbean by Terborgh and Faaborg ( 1973) and on at very different rates: zero-turnover popula- Karkar Island in the Southwest Pacific by Diamond ( 1971). Lvnch and Tohnson claimed that thev re- tions, consisting of either regular breeders or ~ , else of absentees (members of the mainland appraised the data and critically examined the records that served as the basis for Diamonds’ assessments of species pool that never breed on the island breeding presence or absence on the Channel Islands. for any of five reasons); terminal immigrants, Their reassessment was that most instances of turn- that colonized once in the survey interval and over Diamond accepted are either invalid or unproved, persisted; terminal extinctions, species that because of inadequate evidence for breeding presence or absence in one or both survey years compared; disappeared in the survey interval and did not or, if proved, are due to direct or indirect effects of recolonize; in-and-out populations, that im- man. Johnson ( 1972) assumed natural turnover to be migrated and became extinct repeatedly; and sufficiently low that one can reconstruct directions flukes, populations that once colonized unex- and approximate times of island colonizations several pectedly and soon disappeared. Numerous thousand years ago from present-day distributions, even for such vagile species as the Grange-crowned examples of Channel Islands populations in Warbler and the Horned Lark. Lvnch and Tohnson each category are given. concluded that the turnover estimates by Diamond Man has acted both to increase and to de- and by Terborgh and Faaborg appear severely in- crease turnover in several direct and indirect flated, sometimes by as much as an order of magni- tude. They predicted ( 1974 : 383 ) , “If equilibrium ways. The in-and-out effect causes turnover turnover rates are indeed as high as has been claimed values based on comparison of surveys several in the studies cited, then changes in the species com- decades apart to underestimate the true values position of resident insular avifaunas should be de- by about an order of magnitude. Conservative tectable within, say, 5-10 years.” estimates of turnover rates for the Channel Is- The present study confirms this prediction: changes are detectable on most Channel Islands in one or two lands, based on one-year survey intervals, years. Diamonds’ previous estimates have proved to range from O.S%/year for the larger islands to be underestimates rather than overestimates. It re- 56%/year for the smaller islands. Relative mains to point out examples of factual errors in the and absolute turnover rates decrease with is- critiques by Lynch and Johnson. We list, first, errors of a general nature; next, examples of invalid state- land area. Despite this turnover, species num- ments concerning specific populations; and finally, ber on most islands has remained approxi- difficulties underlying Johnsons’ reconstructions of mately constant because the immigration rate historical colonizations. approximately equals the extinction rate. Extinction rates decrease steeply with popu- GENERAL CONSIDERATIONS lation size. Relatively few immigrant popula- ( 1) Among their guidelines for proving existence of tions succeed in achieving large numbers. turnover, Lynch and Johnson (1974: 372, column 2, On the basis of these observations, a model lines 38-50, and p. 376, col. 2, lines 26-39) recom- of the stages in colonization is proposed. Criti- mended that documentation of breeding populations should preferably be supported by collection of voucher cisms directed by Lynch and Johnson at previ- specimens; asserted that “data on gonad condition [of ous turnover studies are considered in an ap- the specimens] are important when one must decide on pendix. the likelihood of breeding versus migrant or casual TURNOVER ON THE CHANNEL ISLANDS 545 status”; and considered that, because Diamond did any one of them is almost sure to add one or more no collecting in 1968, “the possibility of misidentifi- new migrants or winter visitants to the list” (end of cation of at least a few species cannot be dismissed.” Howells’ paragraph: Howell 1917: 5, lines 25-26). While collection of specimens is often justified in other Howells’ belief that a visit was likely to add new types of studies, the reasons why specimens should migrants or winter visitors to the lists for most is- not be collected in turnover studies on the Channel Is- lands is justified by our experience: we are still add- lands are compelling. First, many of the populations ing new migrants or winter visitors on almost every turning over consist of one or a few pairs, so that visit. Howells’ implicit belief that a visit was not collecting is likely to eliminate the populations and likely to add new breeding species also seems alter the outcome- of the natural experiment. Second, justified, for two reasons. First, having reexamined no breeding bird snecies of the Channel Islands uresent virtually all the original papers, specimen collections, significant problems of field identification except for and egg collections consulted by Howell, we find female hummingbirds. Finally, as has often been that the records of breeding species obtained by dif- pointed out, enlarged gonads fail to prove whether the ferent collectors in the same or successive years on the collected individual was breeding on that island or same island largely duplicate each other, implying was on migration (e.g., Lofts and Murton 1973). To relatively thorough coverage of most islands. Second, establish turnover, it is necessary to abstain from col- except for Santa Rosa and San Miguel, the breeding lecting a suspected colonist, observe it repeatedly and species totals extracted from Howell for the other six determine whether it nests, and then examine whether islands in the years up to 1917 are close to those ob- it also nests in the following year. served since 1968 (1917 total the same as 1968 for (2) Lynch and Johnson considered it necessary to one island, slightly higher for three, slightly lower for demonstrate that “the mere presence of a bird species two). This suggests that early breeding coverage of on a given island, even if several individuals are seen these six islands was not at all inadequate, and we during the peak breeding season, does not constitute pointed out (p. 526) that the motivation of early col- strong evidence, much less proof,‘ of resident status” lectors resulted in far more complete coverage of (1974: 372. col. 2. lines 24-28: emuhasis of not and breeding species than of migrants and winter visitors. quotes around proof‘ are by Lynch and Johnson). For Santa Rosa and to a lesser extent San Miguel, the Their wording, italicization, and use of quotes im- visits of early ornithologists were too few or brief for plied that Diamond had claimed mere presence to be breeding coverage to be complete, and Howells’ strong evidence or proof of residence, and that this breeding totals are significantly below modem ones, claim required refutation. Hence they summarized as already discussed by Diamond (1969). Naturally, records of migrants and vagrants on the Farallon Is- whether lack of early records for a particu- lands off northern California to support “our [i.e., lar breeding species on a particular island indicates their1 contention that the presence of nonbreeding absence or simply lack of observation depends entirely land birds on continental islands is the rule rather than on the species ’ conspicuousness, the islands’ size, and the exception” (p. 373, col. 2, lines 16-19, also de- the survey intensities, and we have assessed each case scribed as “our thesis” in col. 1, line 7). In fact, of individually (see p. 529). However, the intensity of course, the presence of numerous migrants and va- coverage up to 1917 provides no support for dismiss- grants on the California coastal islands has been well ing turnover estimates based on pre-1917 surveys as known since the late 19th century, and caused Dia- severely inflated, as Lynch and Johnson claimed. mond in 1968 and Jones since 1973 to devote most (4) According to Lynch and Johnson (1974: 379, field effort to determining breeding vs. migrant status. col. 2, lines l-6), “The MacArthur-Wilson (1967) A by-product of our turnover studies has been the equilibrium model predicts that, other things being addition of 1012species to the list of migrants recorded equal, large islands near a source of potential colonists from the Channel Islands. In addition, the presenta- should have (1) more soecies. and (2) a lower rate of _ I /I tion of Farallon records bv Lvnch and Tohnson ( 1974: species turnover than smaller, more remote islands.” 373, their Table 1 ), as *monthly totals for all non- Half of this paraphrase is correct. It is true that large breeding land birds combined, gives a misleading islands are predicted to have a lower turnover rate, picture of the nature of the problem. For the Chan- because of lower extinction rates. However, Mac- nel Islands as for the Farallones, the great majority Arthur and Wilson (1967) predicted near islands to of nonbreeding individuals in the June and July totals have a higher, not lower, turnover rate than more re- are recorded in the first half of June and latter half mote islands, because of higher immigration rates. of July. Many of the late-June and early-July records (5) Diamond ( 1969) calculated turnover by com- are of vagrants from eastern North America whose paring his 1968 surveys with the avifauna recorded nonbreeding status is scarcely in doubt. We do not by Howell ( 1917), whose compilation was based automatically assume any land bird observed in the mainly on field studies between 1897 and 1915. Lynch latter half of June and first half of July to be breed- and Johnson (1974: 375, col. 2, lines 36-37) claimed ing, but we do observe few species on the islands in that the result of this procedure is “to overestimate the this period that are not clearly breeding, and fewer turnover rate by some unknown fraction.” Actually, still whose status is ambiguous. the effect on turnover calculations of combining con- (3) Lynch and Johnson (1974: 376, col. 1, lines secutive years into a single survey, for comparison 14-16) quoted Howell (1917: 5, lines 23-25) to with a later survey completed within one year, is two- support their contention (p. 382, col. 2, lines 49-50) fold: to tend to underestimate immigrations, but to that Howells’ summaries do not provide a reliable overestimate extinctions. The two errors are thus op- data base for computing equilibrium turnover rates: posite and tend to cancel. In the present paper the “ As‘ a matter of fact, there has been comparatively turnover rates we calculated mostly utilize surveys little work done upon these islands by anybody . . . ’ ” completed within one year, with 9 additional surveys (italics and termination of the quotation are by Lynch based on combined observations for several consecu- and Johnson). Their termination of the quotation re- tive years. versed Howells’ meaning, for the remainder of his (6) In their first critique, Lynch and Johnson sentence read, “. . . and a visit of several weeks to ( 1973) claimed, “We have reappraised the data ana- 546 H. LEE JONES AND JARED M. DIAMOND lyzed by Diamond (Proc. Nat. Acad. Sci. U.S. 1969, from the vicinity of the partys’ skinning table for sev- 64: 57-63) and offered as support for his thesis of eral days in March 1915. Finally, dozens of visits and historically high rates of avifaunal turnover in the hundreds of man-hours of observation on San Cle- Channel Islands of California. Critical examination mente since 1967 have yielded only one record of a of records that supposedly document individual in- transient Annas’ Hummingbird in five breeding sea- stances of extinction and immigration leads us to the sons. Thus, the species is surely not breeding now, following conclusions . . .” This claim is false, ancl and the observations by Howell, Dickey, and Huey Lynch and Johnson did not repeat it in their second constitute a typical example of in-and-out turnover. (1974) paper. Of the nine islands that Diamond (2 ) Lvnch and lohnson ( 1974: 378, col. 1) ques- surveyed in 1968, Lynch and Johnson examined none tioned e&in&ions of the Red-breasted Nuthatch-and‘ of Diamonds’ records at all from his surveys for six Coopers’ Hawk on Santa Cruz: “While it is indeed islands, and for the other three islands examined only possible that these species were once resident, preliminary, briefly annotated species lists that Dia- neither has ever been shown to nest on this or any mond precirculated to several people to elicit infor- other of the Channel Islands . . . .” For the Red- mation and that did not attempt to provide detailed breasted Nuthatch, Howell (1917: 99; see also documentation of records. Howell and van Rossem 1911: 210) stated, “On May 1 119111 I watched an individual excavating a nest STATEMENTS CONCERNING site in a dead stub . . . .” For the Coooers’ Hawk the SPECIFIC POPULATIONS evidence is ambiguous, but Lynch and Johnson un- derstated the case: “. . . I believe it to be resident Of the numerous populations for which Lynch and in small numbers . . . . On April 25, 1911, I saw a Johnson disputed Diamonds’ assessment of turnover, pair at the lower edge of the pines which acted very many have been discussed in the text of this paper, much as if they had a nest near by” (Howell 1917: and others will be discussed elsewhere. Since space 54; see also Howell and van Rossem 1911: 209). does not permit us here to rediscuss every case in the (3) Lynch and Johnson (1974: 378, col. 1) ques- light of Lynchs’ and Johnsons’ comments, we consider tioned extinction of the Black Phoebe on San Cle- seven cases to illustrate the types of errors involved mente after a nesting record in 1908, on the grounds in their arguments: that absence in subsequent years was not established ( 1) Lynch and Johnson ( 1974: 377, col. 2) re- (code E2 in their Table 3). On our p. 535 we de- jected Diamonds’ conclusion of an extinction of Annas’ scribe the recent searches for this species that provide Hummingbird on San Clemente between 1917 and adequate documentation of absence. 1968, on the grounds that breeding was not adequately (4) Lvnch and Tohnson C1974: 378. col. 2) stated demonstrated in 1917 (code El in Table 3 of Lynch that Barn‘ Swallows have never been shown to breed and Johnson) and that absence of a breeding popula- on Los Coronados. Howell (1917: 87) stated, “A tion was not adequately demonstrated in later years. limited number occur on the Coronados during the Here it is useful to quote Lynch and Johnson at spring and summer, breeding in the sea caves where length. “The inclusion of the Annas’ Hummingbird their nests must be frequently dampened by the . . on the list of extinctions for San Clemente Island spray.” (Diamond 1971) appears to be yet another instance (5) Regarding extinction of the Bushtit on Santa of pseudoturnover. Many of the early ornithologists Catalina, Lynch and Johnson ( 1974: 378, ~01s. 1 and who worked on the Channel Islands had visited San 2 ). concluded. I “. . . the earlv sight records bv Wil- II r v Clemente prior to Howells’ summary. Among these lett (fide Howell 1917) may well have involved a were Howell himself, Breninger, Howard, Linton, vagrant group which was blown to Catalina from Dickey, and Huey. Some of these biologists spent the mainland.” On p. 533 we state why this inter- periods of weeks on the island during the height of the pretation is implausible. breeding season, but the only Annas’ Hummingbird ( 6 )Lynch and Johnson ( 1974: 380,381) discussed ever reported by any of these workers was the indi- in detail the published surveys of San Nicolas in vidual seen by Howell in March 1917 [sic]. Even 1897 and 1962-3, in the mistaken belief that “from though this individual appeared to be gathering nest- the time of Grinnells’ visit in the late 19th century ing material (Howell 1917: 63), the fact that the until 2 Mav 1962-l Tanuarv 1964 (Townsend 1968) species had never been noted previously, even as a no general survey of the avifauna of San Nicolas Is- vagrant, and has not been reported subsequently to land was attempted.” Lynch and Johnson discarded all breed on San Clemente implies that any possible six apparent cases of turnover suggested by compari- breeding population never exceeded a very few pairs. son of these two publications and concluded that ap- Therefore, Diamonds’ failure to find this species dur- parent natural turnover between 1897 and 19624 was ing a few days ’ field work on this large (56 square zero. In fact, the results of one general survey based on miles) island cannot be accorded much weight.” In three visits by two observers totalling 35 days in 1945 these sentences Lynch and Johnson listed six visitors were published in this journal (Rett 1947). Another to the island prior to Howells’ summary, implied that general survey, by L. Miller in 1938, is described in none of these investigators but Howell saw Annas’ an unpublished manuscript by Miller in the UCLA Hummingbird, and implied that there was some un- library. San Nicolas has the best documented and certainty what the individual seen by Howell was one of the two highest turnover rates of any Channel doing. First, Breninger and Howard commented only Island (12 cases of turnover since 1963 alone). briefly on a few of the species they found, thus it is Lynchs’ and Johnsons’ erroneous conclusion of zero not known if they saw this species. Second, Dickey turnover partly reflects the inevitable underestimate and Huey visited the island with Howell as a single that the in-and-out effect causes with widely spaced party. Third, Howells’ account of the partys’ ob- survey years, but also reflects more specific errors. servations does not state that the hummingbird “ap- Thus, “The Bald Eagle and Osprey, now extinct on the peared” to be gathering nesting material in March island, were present and possibly resident but good “1917,” but rather that it did collect bits of cotton evidence for breeding is lacking” (Lynch and John- TURNOVER ON THE CHANNEL ISLANDS 547 son 1974: 380, col. 1). In fact, an Osprey egg was RECONSTRUCTION OF COLONIZATION taken on San Nicolas in 1901 by B. Trask and is in BY ENDEMIC RACES THOUSANDS the MVZ collection; a set of three eggs was taken there OF YEARS AGO by C. B. Linton on 15 April 1909 ( WFVZ ); and Wil- lett (1912, 1933) recorded it as having bred commonly. Lynch and Johnson (1973) claimed that the number Rett ( 1947) and others (egg set in WFVZ) found of endemic bird subspecies occurring on the Channel eagles nesting. “Two species (meadowlark and [Brew- Islands argues against high turnover rates. Even if ers]’ blackbird) are native colonists from the mainland turnover of these endemic races was in fact slow, this that well may have arrived to breed in direct response would not indicate slow turnover of the avifauna as to habitat alteration by man and his livestock ’ (Lynch a whole, since turnover rates vary conspicuously and Johnson 1974: 380, col. 2). As discussed on among species. In addition, inter-island movement of our p. 533, Western Meadowlarks colonized some time all but two of the 13 species with endemic subspecies between 1940 and 1945, while Brewers’ Blackbirds on the Channel Islands, and of 9 of the 18 endemic did not colonize successfully and disappeared after subspecies themselves, has been detected in this cen- breeding in 1962 and 1963. In 1897 man and his tury, in most cases even within the past decade. Five livestock had already been on San Nicolas for at least of these endemic subspecies (Selasphorus sasin seden- half a century and had converted most of the island tarius, Eremophila ulpestris insularis, Lanius ludoui- into typical habitat for these two bird species. It ciunus anthonyi, Vermivoru celuta sordida, and Aimo- is implausible that removal of the livestock in the phila ruficeps obscura) have established new breeding 1930s’ was the cause of the colonizations, because populations on previously unoccupied islands. John- on all seven other Channel Islands during this entire son (1972) attempted to use present distributional period both meadowlarks and livestock coexisted patterns of the endemic subspecies to reconstruct their abundantly; and the meadowlark still coexists abun- ages (in geological time) and island-to-island coloni- dantly with livestock today on the four islands that zation routes. For instance, he suggested that the continue to support livestock. Lynchs’ and Johnsons’ Rufous-crowned Sparrow first colonized Anacapa in claim (1974: 380, col. 2) that Townsend noted no the Xerothermic or post-Xerothermic period after the evidence of breeding in the Burrowing Owl and that end of the Pleistocene, spread from there to Santa his observations of the easily identified White-crowned Cruz. then died out on Anacaoa ( Tohnson 1972: 301. Sparrow require confirmation contradicts Townsend 310); and that the Western Flycatcher, Horned Lark: ( 1968, amplified by personal communication ). Orange-crowned Warbler, and Allens’ Hummingbird (7) Regarding Diamonds’ study of the New Guinea first colonized the northern islands in the cool-moist satellite island of Karkar in 1969 for comparison with period between the end of the Pleistocene and begin- Meeks’ 1914 studv. Lvnch and Tohnson (1974: ning of the Xerothermic, the former two species pos- 3813) suggested chat Diamond could- “easily; ’ have sibly arriving before the latter two, and spread fairly overlooked the five apparently extinct species (two quickly thereafter to the southern islands (Johnson rails and three pigeons ), and that all 11 of the ap- 1972: 301, 309, 312). In fact, on-going turnover parent 1969 immigrants might “quite possibly” have and inter-island movements of these and other en- been present in 1914 despite Meeks’ failure to obtain demic bird subsoecies of the Channel Islands have them. These assessments should surprise workers fa- probably distorted their original distributions com- miliar with the species involved and with field meth- pletely beyond recognition and obliterated a11but their ods in the New Guinea region. On Karkar, as else- most recent histories. These movements make at- where in the New Guinea region, many resident people tempted reconstructions of their island distributions are walking encyclopedias of natural history with thousands of years ago too speculative to be useful. detailed knowledge of virtually all resident bird species and of many rare straggIers (Mayr 1932, Gil- LITERATURE CITED liard and LeCroy I961 and many other -papers, Dia- mond 1966 and 1972. Bulmer 1974 and manv other BANKS, R. C. 1966. Terrestrial vertebrates of Ana- papers ) ; their help was used extensively by Diamonds’ capa Island. Trans. San Diego Sot. Nat. Hist. 14: party (consisting of Diamond plus four New Guinea 175-188. residents and six Karkar residents ). The pigeons found BROWN, J. H., AND KODRIC-BROWN, A. Arthropods on by Meek but not by Diamond (Grey Pigeon Duculu thistles : small-scale equilibrium insular biogeog- pistrinaria, Piiion Imperial Pigeon Duculu ninon. Nico- raphy. Submitted to Ecology. bar Pigeon C~hnas~nic~b&u) are three of the best BULME~I, R. N. H. 1974. Folk biology in the New known and most conspicuous common species of the Guinea Highlands. Sot. Sci. Inform. 13:9-28. New Guinea region. The loud calls of the former two BURT, H. C. 1911. An early spring trip to Anacapa species carry for a mile or more. The two rails found Island. Condor 13: 164-167. by Meek but not by Diamond are, of course, much DAWSON, W. L. 1915. Supposed new records for shyer but are distinctive species well-known to natives Santa Cruz Island. Condor 17:203-204. elsewhere and are sometimes common. Karkar natives DIAMOND, J. M. 1966. Zoological classification sys- stated that these two rails were absent but correctly tem of a primitive people. Science 151:1102- predicted that Diamond would instead find the equally 1104. DIAMOND, J. M. 1969. Avifaunal equilibrium and shy rail Amaurornis olivuceus, a new immigrant, com- species turnover rates on the Channel Islands of mon on Karkar. While a few of the 11 apparent im- California. Proc. Natl. Acad. Sci. U.S.A. 64: migrants could have been overlooked by Meek, this 57-63. is unlikely for such conspicuous, common, and easily DIA~IOND, J. M. 1971. Comparison of fauna1 equi- collected species as Willie Wagtail (Rhipiduru Zeu- librium turnover rates on a tropical island and a cophrys), Cicada Bird ( Corucinu tenuirostris), and temperate island. Proc. Natl. Acad. Sci. U.S.A. Brahminy Kite (Huliustur in&s). For some of these 6812742-2745. immigrants, Karkar residents were able to give details DIAMOND, J. M. 1972. Biogeographic kinetics: Es- of when and where the species had first appeared. timation of relaxation times for avifaunas of south- 548 H. LEE JONES AND JARED M. DIAMOND west Pacific islands. Proc. Natl. Acad. Sci. U.S.A. LOFTS, B., AND R. K. MURTON. 1973. Reproduction 69:3199-3203. in birds, p. l-107. In D. S. Farner and J. R. King DIAMOND, J. M. 1974. Colonization of exploded [eds.], Avian biology. Vol. 3. Academic Press, volcanic islands by birds: the supertramp strat- New York. egy. Science 184:803-807. LYNCH, J. F., AND N. K. JOHNSON. 1973. Avifaunal DIAMOND, J. M. 1975. The island dilemma: lessons of turnover rates in the California Channel Islands: modern biogeographic studies for the design of a reanalysis. Cooper Ornithol. Club, 44th Ann. natural reserves. Biol. Conserv. 7: 129-146. mtg., Abstr. DIAMOND, J. M. 1976. Relaxation and differential ex- LYNCH, J. F., AND N. K. JOHNX)N. 1974. Turnover tinction on land-bridge islands: applications to and equilibria in insular avifaunas, with special natural preserves. Proc. 16th Intern. Omith. reference to the California Channel Islands. Con- Cong., 616-628. dor 76:370384. DIAMOND, J. M., AND R. M. MAY. Species turnover MACARTHUR, R. H. AND E. 0. WILSON. 1963. An rates on islands: dependence on census interval. equilibrium theory of insular zoogeography. Evo- Science, in press. lution 17:373-387. DIAMOND, J. M., AND E. MAYR. 19768. Species-area MACARTHUR, R. H., AND E. 0. WILSON. 1967. The relation for birds of Solomon Archipelago. Proc. theory of island biogeography. Princeton Univ. Natl. Acad. Sci. U.S. 73:262-266. Press, Princeton. DICKEY, D. R., AND A. J. VAN ROSSEM. 1923. Addi- MACUIRE, B., JR. 1963. The passive dispersal of tional notes from the coastal islands of southern small aquatic organisms and their colonization of California. Condor 25: 126-129. isolated bodies of water. Ecol. Monogr. 33:161- GILLIARD, E. T., AND M. LECROY. 1961. Birds of 185. the Victor Emanuel and Hindenburg Mountains, MAYR, E. 1932. A tenderfoot explorer in New New Guinea. Bull. Am. Mus. Nat. Hist. 123: Guinea. Nat. Hist. 32:83-97. l-86. MAYR, E. 1965. Avifauna: turnover on islands. GRINNELL, J. 1897. Report on the birds recorded Science 150: 1587-1588. during a visit to the islands of Santa Barbara, San MEADOWS, D. C. 1929. Notes on the avifauna of Nicolas and San Clemente, in the spring of 1897. Santa Catalina Island. Condor 31:129-130. Pasadena Acad. Sci. 1: l-25. MILLER, A. H. 1955. Acorn Woodpecker on Santa GRINNELL, J., AND A. H. MILLER. 1944. The dis- Catalina Island, California. Condor 57:373. tribution of the birds of California. Pacific Coast MOREJOHN, V. 1975. Evidence of survival to re- Avifauna 27: l-608. cent times of the extinct flightless duck, Chen- HARPER, C. A. 1971. Breeding biology of a small dytes lawi. Pacific Seabird Group, Ann. mtg., colonv of Western Gulls iLarus occidentalis Abstr. wymani) in California. Condor 73:337-341. PATRICK, R. 1967. The effect of invasion rate, spe- HOFFMAN, R. 1931. Saw-whet Owl and California cies pool, and size of area on the structure of the Woodpecker on Santa Cruz Island. Condor 33: diatom community. Proc. Natl. Acad. Sci. U.S.A. 171. 58: 1335-1342. HOWELL, A. B. 1917. Birds of the islands off the PEMBERTON, J. R. 1928. Additions to the known coast of southern California. Pacific Coast Avi- avifauna of the Santa Barbara Islands. Condor fauna 12: 1-127. 30: 144-148. HOWELL, A. B., AND A. J. VAN ROSSEM. 1911. Fur- PHILBRICK, R. N. [ed.]. 1967. Proceedings of the ther notes from Santa Cruz Island. Condor 13: symposium on the biology of the California Is- 208-210. lands. Santa Barbara Botanic Garden, Santa Bar- HUNT, G. L., JR., AND M. W. HUNT. 1974. Trophic bara. levels and turnover rates: the avifauna of Santa POMER,’ D. M. 1972. Numbers of bird species on Barbara Island, California. Condor 76:363-369. the California Islands. Evolution 26:451463. JOHNSON, N. K. 1972. Origin and differentiation of RETT, E. Z. 1947. A report on the birds of San the avifauna of the Channel Islands, California. Nicolas Island. Condor 49: 165-168. Condor 74:295315. SCHOENER, A. 1974. Experimental zoogeography: JONES, H. L. 1975. Studies of avian turnover, dis- colonization of marine mini-islands. Am. Nat. persal and colonization of the California Channel 108:715-737. Islands. Ph.D. dissert., Univ. California, Los An- SIMBERLOFF, D. S., AND E. 0. WILSON. 1969. Ex- geles. perimental zoogeography of islands. The coloni- JONES, H. L. In prep. Birds of the California Chan- zation of empty islands. Ecology 50:278-296. nel Islands. SULLIVAN, A. L., AND M. L. SHAFFER. 1975. Bio- LEIGH, E. G., JR. 1975. Population fluctuations, eeograuhv of the Menazoo. Science 189:13-17. community stability, and environmental vari- TERB;R&, _ J.. 1974a. Preservation of natural di- ability, p. 51-73. In M. L. Cody and J.M. Dia- versity: the problem of extinction prone species. mond Ieds.1, Ecology and evolution of communi- Bioscience 24~715722. ties. Harvard Univ. Press, Cambridge. TERUORGH, J. 1974b. Fauna1 equilibria and the de- LEVINS, R. 1975. Evolution in communities near sign of wildlife preserves, p. 369380. In F. Gol- equilibrium, p. 16-50. In. M. L. Cody and J. M. ley and E. Medina [eds.], Tropical ecological sys- Diamond Ieds.1, Ecology and evolution of com- tems : trends in terrestrial and aquatic research. munities. Harvard Univ. Press, Cambridge. Springer, New York. LEVINS, R., AND H. HEATWOLE. 1973. Biogeogra- TERBOIXH, J., ANI) J. FAABO~~. 1973. Turnover and phy of the Puerto Rican bank: Introduction of ecological release in the avifauna of Mona Island, species onto Palominitos Island. Ecology 54: Puerto Rico. Auk 90 :759-779. 1056-1064. TOWNSEND, W. C. 1968. Birds observed on San LINTON, C. B. 1908. Notes from San Clemente Is- Nicolas Island, California. Condor 70:266-268. land. Condor 10:82-86. WHITCOMB, R. F., J. F. LYNCH, P. A. OPLEH AND C. TURNOVER ON THE CHANNEL ISLANDS 549 S. ROBBINS. 1976. Island biogeography and con- WILSON, E. 0. 1975. Sociobiology. Harvard Univ. servation: strategy and limitations. Science 193: Press, Cambridge, MA. 1027-1029. WILSON, E. O., AND E. 0. WILLIS. 1975. Applied WILLETT, G. 1910. A summer trip to the northern biogeography, p. 522-534. In M. L. Cody and J. Santa Barbara Islands. Condor 12: 170-174. M. Diamond feds.], Ecology and evolution of WILLETT, G. 1912. Birds of the Pacific slope of communities. Harvard Univ. Press, Cambridge. southern California. Pacific Coast Avifauna 7: W~GIIT, II. W., AND G. K. SNYDER. 1913. Birds 1-122. observed in the summer of 1912 among the Santa WILUTT, G. 1933. A revised list of the birds of Barbara Islands. Condor 15:86-92. southwestern California. Pacific Coast Avifauna No. 21. Department of Physiology, University of California WILLIAMSON, G. B. Equilibrium turnover rates for Medical Center, Los Angeles, CA 90024. Accepted islands. Am. Nat., in press. for publication 7 July 1976.