Biological Conservation 78 (1996) 69-96 Copyright © 1996 Elsevier Science Limited Printed in Great Britain. All rights reserved PIi: S0006-3207(96)00019-5 0006-3207/96/$15.00 +.00 ELSEVIER GLOBAL PATTERNS IN THE ESTABLISHMENT AND DISTRIBUTION OF EXOTIC BIRDS

Ted J. Case Department of Biology, 0116 University of California at San Diego, La Jolla, CA 92093, USA

Abstract be influencing habitat distributions of species in both sets I use three separate data bases to examine recipient com- within islands. In both man-made habitats and native munity and site factors that might be influencing the forest habitats, exotic species number and the relative establishment, persistence, and distribution of avian abundance of exotic birds is negatively related to the exotics. All in all, about half the variance between number of native species. After accounting for this local islands~regions in their numbers of successfully and variation, exotic species number is positively related to unsuccessfully introduced species can be accounted for exotic species number for the entire island~region. In by recipient site-specific variables; the most important local surveys the relative abundance of exotic birds com- correlate of success is the number of native species pared to native birds is affected by habitat (non-native extinctions over about the last 3000 years, which reflects habitats have more exotics) and also by the numbers of the degree of human activity and habitat destruction and species of exotics and natives on the island. The relative deterioration through intrusions of exotic predators, her- importance of biotic interactions like competition, appar- bivores, and parasites. Consequently, the number of ent competition through differential disease transmission exotic species gained is close to the number of species or susceptibility, and predation in shaping the abundance lost through extinction. Even after controlling for avian and habitat affinities of exotics and native species can be extinctions, island area correlates positively with intro- difficult to unravel when regional affects are so impor- duced species number. Invasion success does not decline tant. Copyright © 1996 Elsevier Science Limited significantly with the richness of the native avifauna (after controlling for the effects of extinctions and island Keywords: birds, extinction, exotic species, invasibility, area) nor the variety of potential mammalian predators. habitat destruction. The relative proportion of extinct native species across islands~regions is negatively correlated with area and positively correlated with introduced species number and INTRODUCTION the number of endemic species. A strong correlation exists between the number of successes and the number For assorted reasons, some practical and some purely of failures, attesting to the role of persistent acclimatiza- aesthetic, people have purposely introduced various tion societies in increasing species numbers despite high plants and animals from one place to another across .failure rates. The relative success to failure rate increases the globe. What factors determine whether a species with the number of extinct native species. The correlation will become established or not? What site properties between introductions and native extinctions seems to determine whether an ecological system will accept arise because native birds are usually more common, if or repel invaders? Given answers to these questions, not restricted, to native habitats while introduced birds how can we best manage ecological systems to avoid are primary occupants of disturbed and open habitats. As problems? more of an island's area is converted to urban, agricul- One ecological issue affecting answers to these ques- tural and disturbed habitats or altered through the intro- tions is whether natural communities vary in their resis- duction of herbivores and exotic predators, most natives tance to invasions because of intrinsic properties of the lose good living space while most introduced birds, that natural system. Elton (1958) suggested that species-poor frequent open and disturbed areas and have evolved in communities, characteristic of islands or very disturbed predator-rich areas, gain habitat. habitats, were more susceptible to invasion than species- l find little support for the notion that rich avifaunas in rich communities. More recent theoretical explorations themselves repel the establishment of avian invaders at of model communities lend some credence to this the level of whole islands or archipelagoes. However, hypothesis (Case, 1991; Post & Pimm, 1983; Drake, interactions between established exotics and natives may 1983, 1988). The role of habitat disturbance in promoting biologi- Correspondence to: T. J. Case cal invasions is well supported (Rejm~nek, 1989). But e-mail: [email protected] how would we test Elton's hypothesis regarding 69 70 T.J. Case species-richness experimentally? Ideally we would con- failed as well as successful introduction attempts. For struct islands identical in all respects but differing in these locations, mostly places that had acclimatization the richness of the resident fauna. We then would societies in the 1800s, I ask what site and community- introduce the same set of species to each island, con- level factors contribute to establishment success com- trolling for individual numbers released, sex ratio, pared to failure. physiological condition, and other variables that might Finally, since competitive interactions between influence reproductive and survival success. Since this is natives and exotics may potentially be important in generally impossible except at the microcosm scale shaping the proclivity of native species to enter human- (Drake, 1991; Drake et al., 1993; Robinson & modified habitats and exotic species to penetrate native Dickerson, 1984), for real islands we must rely on habitats, I examine the habitat distributions of native interpreting the results of past historical introductions and introduced birds in various habitats and locations where many confounding factors, beyond the nature of in the Pacific region. the recipient community, can and do influence invasion The present study is entirely devoted to identifying success. With introductions there are often no controls factors in the recipient location that potentially infl- and, like all historical work, we must often rely on the uence the establishment, persistence, and spread of sometimes conflicting accounts described by different introduced bird species. It is also important to make actors and observers, no longer living; but the vast comparisons between species: are the same species that number of introductions performed on literally thou- are successful in one place successful in others? What sands of different islands and mainlands worldwide biological attributes of species are associated with estab- offers an opportunity to attempt answers to the ques- lishment success? How similar must climates be in the tions raised in the opening paragraph. old and new homes of species? I reserve these analyses The method of the introduction may influence its for a later paper. likelihood of success, e.g. how many individuals are released, and how they are transported and acclimated METHODS to the new location. Different species have been intro- duced to different places with different climates so Native and introduced birds on islands species-specific autecological factors will also be important. Table 1 contains a list of locations with tallies for their Some species may be better adapted to physical and numbers of native and introduced birds along with climatic features of the new environment and others other geographic and biotic features of the location. may have behavioral and life-history attributes that References are given in Appendix 1. The general refer- enhance population growth when they are rare. We ences at the bottom of Appendix 1 were used for nearly might get spurious associations between invasion resis- all locations. Islands were selected for inclusion in the tance and native biodiversity because of confounding analysis based on two criteria: a complete avifaunal interactions with other variables. For example, the list, and an attempt to balance representation across number of native species on an island generally different regions of the globe (excluding polar regions). increases with its area (Preston, 1962; MacArthur & The analysis is restricted to land and freshwater Wilson, 1967). Yet island area directly can affect intro- species (excluding shorebirds, waders, and strictly sea- duction success by potentially affecting mate-finding going ducks) that maintain breeding populations in the ability. A few individuals of a new exotic species locations (i.e. migrants are excluded). Table 1 is based released into a large area may have difficulty locating on archipelagoes with some minor exceptions (see mates compared to the same number released into a below). Using individual islands as separate data points smaller area. Without experimentally or statistically (rather than lumping islands into archipelagoes) com- partialling out the possible affects of area per se on mits pseudoreplication since the native faunas and introduction success, we might incorrectly conclude introduction histories of nearby islands are typically that species-richness (which covaries with area) was closely linked and birds introduced to one island often explaining the results when, in this scenario, it was colonize nearby islands on their own (Long, 1981; more directly caused by area per se. Moulton & Pimm, 1983). However, a problem arises Here I use three separate data bases to examine with this choice: a single point, e.g. Fiji, includes many recipient community and site factors, including area, islands; each of which may not have the full comple- that might be influencing the establishment, persistence, ment of introduced species present in the entire and distribution of avian introductions. I have assembled archipelago nor the full set of native species. Conse- species lists for a worldwide set of islands and two quently, one can imagine a situation wherein all the mainland locations (continental USA and Australia). I introduced species are confined to a single island within ask what features of the recipient location best account the archipelago that lacks any native species. In this for across-site variation in the number of introduced case the two sets of species never see each other in the species present in the existing avifaunas. archipelago and potentially experience very different The second analysis deals with a smaller set of loca- environments. However, this is rarely the case and I tions for which information exists on the number of have attempted to sort out islands which have unique Introductions and extinctions of birds 71

Table 1. Bird species numbers on various islands of the world and for Australia and the continental US (excludes Alaska) Taxonomy follows Sibley and Monroe (1990). References are in Appendix 1. See also notes to Table 2. An asterisk (*) denotes locations where Holocene bird fossils have not yet been discovered or examined. Missing values in the extinction column indicate an absence of information on both historical and prehistorical extinctions. See text for further details.

Island/mainland No. No. native No. native Area Max. elevation Human Mammalian introduced species species (km 2) (m) settlement predator species extant extant extinct pattern category

Aldabra* 1 20 1 155 100 3 3.0 Andamans/Nicobars* 4 104 0 6462 735 4 4.0 Aruba* 2 32 175 500 4 3.0 Ascension 4 0 1 90 860 3 3.0 Aucklands* 9 11 2 606 610 3 1.5 Azores* 3 20 0 2304 2320 3 3.0 Bahamas 6 101 13 13,939 122 4 4-0 Balearic 1 119 3 5014 1445 4 5.0 Bermuda 7 9 7 54 73 3 3.0 Borneo* 4 386 0 758,870 3810 4 5.0 California Channel 4 56 2 904 753 4 4.0 Canaries 5 47 3 7273 3720 4 3.0 Cape Verde* 5 26 1 4033 2829 3 3.0 Chagos* 7 3 65 20 3 3.0 Chatham 14 14 21 973 213 4 3.0 Christmas (Indian Oc.)* 1 8 0 135 357 3 2.0 Cocos* 1 6 0 47 854 3 3.0 Comoros* 7 52 0 1958 2361 3 4.0 Cook Islands 3 11 10 240 652 4 3.0 Corsica/Sardinia 3 120 1 32,771 2711 4 5.0 Cozumel* 0 75 0 324 24 4 5.0 Curacao* 2 34 0 425 372 4 3.0 Easter 4 0 6 I 16 530 4 2.0 Falklands* 2 28 0 16,053 706 3 4.0 Fiji (mongoose) 8 46 7 15,921 1300 4 4-0 Fiji (non-mongoose) 9 51 3 2375 1200 4 3.0 Galapagos 2 36 1 7855 1524 3 3.0 Great Britain 10 147 5 160,952 1343 4 5.0 Guadaloupe, Mex.* 3 13 6 254 1402 3 1.5 Hawaii (Kauai) 27 18 14 1422 1598 4 3.0 Hawaii (mongoose) 47 27 66 12,136 4206 4 4-0 Henderson 0 4 3 37 33 2 1-0 Jamaica 5 88 6 11,424 2256 4 4-0 Java* 1 327 0 132,400 3676 4 5.0 Juan Fernandez* 2 8 0 164 1500 3 3.0 Kangaroo 15 78 1 3890 190 4 3.0 Kermadec 7 6 1 30 520 2 3.0 Lord Howe 9 6 9 13 853 3 3.0 Madagascar 3 168 13 587,000 2881 4 5-0 Guam (Marianas)* 7 12 7 541 406 4 3.0 Marquesas 5 11 9 910 1260 4 2.0 Mauritius 19 13 14 1865 826 3 4.0 New Caledonia 6 56 14 16,912 1815 4 3.0 New Guinea* 1 513 0 808,000 5000 4 4-0 New Zealand 41 52 36 266,800 3765 4 4-0 Norfolk 12 15 6 40 310 4 2.5 Palau* 4 31 0 440 240 4 3.0 Pemba* 1 67 1 984 85 4 4-0 Puerto Rico 23 82 15 8897 1350 4 4-0 Reunion 19 12 13 2512 3040 3 3.0 Revillagigedo* 3 13 2 233 1113 1 2.0 Rodriguez 7 4 10 109 396 3 4.0 Rottnest* 3 20 2 19 35 3 3-0 Rotuma (Fiji)* 1 16 0 45 300 4 3.0 Saint Helena 8 0 5 125 819 3 3.0 Samoa 3 33 2 3150 1858 4 3-0 Sea of Cortez (Ld.-bridge)* 2 34 0 1713 620 1 1.5 Seychelles (granitic)* 8 16 3 233 913 3 3.0 Societies 12 13 12 1550 2322 4 3.0 Solomons, Central 1 124 2 31,800 3100 4 4.0 (Contmue~ 72 T. J. Case

Table 1. -- eoatd

Island/mainland No. No. native No. native Area Max. elevation Human Mammalian introduced species species (km 2) (m) settlement predator species extant extant extinct pattern category

Sri Lanka* 1 225 1 65,610 2528 4 5.0 Taiwan* 2 132 0 36,125 3998 4 5.0 Tasmania 13 104 1 67,900 1520 4 4.0 Three Kings* 9 10 4 7 300 4 1-0 Tres Marias* 1 34 0 350 613 3 4.0 Trinidad* 2 235 4542 1250 4 5.0 Tristan de Cunha Arch.* 0 6 1 259 2329 3 3.0 Tuamotu* 2 9 0 855 113 4 3.0 Vanuatu* 5 56 0 12,000 1889 4 3-0 Wake* 1 0 1 23 6 4 3.0 Zanzibar* 4 102 0 1658 125 4 5.0 USA (lower 48 states) 13 553 6 7,827,622 4516 4 5.0 Australia 17 466 1 7,680,000 2228 4 5-0

histories compared to others in the same archipelago. species are not counted. Most of the purposeful intro- For example, Guam has had many more introductions ductions were done in the heyday of acclimatization and extinctions than elsewhere in the Marianas societies during the mid 1800s although in Hawaii they archipelago; consequently, Guam is considered alone. continued much longer. A successfully introduced Since I will be examining the role of community and species is one that has established a breeding popula- site factors, I subdivide archipelagoes when they are tion that is still present today; its population must have strongly subdivided by distance or important biological increased numerically and expanded geographically factors that might influence the success of introduced beyond the vicinity of the original introduction sites. species. Fiji and Hawaii are subdivided into those Aboriginal introductions (usually just Gallus gallus) are islands with and without introduced mongoose. inferred based on a lack of subspecific differentiation, The definition of an archipelago usually follows geo- anthrophilic habits of the species, and no appearance in political boundaries, with a few exceptions. Islands are middens or subfossil deposits dated earlier than man's lumped when the islands share endemic species, are appearance on the island group. similar in geological age of formation or isolation, and distance to colonizing sources. Islands whose distance Native species number apart from one another is greater than the distance to This count excludes self-introduced land and freshwater the mainland or other major faunal sources are not species and native species that are now extinct. Here lumped. By these conventions, 'New Zealand' includes and for historical extinctions I follow the taxonomy in only the two major islands, not the many small off- Sibly and Monroe (1990). For island archipelagoes like lying land bridge islands; similarly I do not include in the Canaries, the number of native species is totaled Australia or New Guinea the several off-shore islands. over all islands in the group but no species is counted Corsica and Sardinia, which are similar in size and more than once. were united in the last glaciation, are combined. The Andamans are combined with the nearby Nicobars. On Extinct bird species the other hand, the Mascarenes (Reunion, Rodriguez The total number of land and freshwater bird species and Mauritius) are considered separately because they that became extinct in historical and in prehistorical share relatively few native land bird species or exotic time is tallied. The prehistoric extinctions coincide for predators, and they are about as far apart from one the most part with aboriginal occupation (Cassels, another as they are to other major colonization 1984; Diamond, 1984; Olson & James, 1984; Olson, sources. The distant Rotuma island (politically part of 1989; Steadman, 1989b) and generally are late Fiji) is considered separately from the rest of Fiji. Holocene. Extinctions include true extinctions of a species and extirpations of the species from the Introduced species number island/archipelago in question. If the same species This includes human releases and self-introduced becomes extinct locally on several islands within a species naturally invading the location during the last listed archipelago, it is counted as only a single extinc- hundred years if they survive to the present day; self- tion and then only if no populations survive anywhere introductions usually represent a small proportion of in the listed location (e.g. all islands in the Canaries, all the 'exotic' bird species. Self-introduced species that mongoose-free islands in Fiji minus Rotuma, both replace extirpated native populations of the same major New Zealand islands). Introductions and extinctions of birds 73

Since some island locations have had much more R. norvegicus but no feral cats or other predators; paleontological attention than others, or have geologi- (3) introduced rats and feral cats with or without native cal structures that better preserve fossils, it is probable rats and feral domestic dogs (but no other predators); that many undocumented prehistoric extinctions await (4) introduced rats, plus feral cats and dogs, plus a discovery (see Pimm et al., 1994 for a clever way to mongoose, and/or a stoat or weasel, or other wild estimate these 'missing' extinctions), but I see no rea- mustelid, viverid, canid, raccoon, or partially carnivo- son why there should be any particular bias in fossil rous (of birds or their eggs) primate; (5) a rich main- discovery with respect to numbers and type of intro- land fauna that includes rats, feral cats, and dogs plus duced species established. Locations without any known several native carnivore species and other potential prehistoric bird fossils are asterisked in Table 1. avian predators such as primates or large insectivores. Missing values in the extinction column are for I test Elton's (1958) hypothesis by determining if a islands where I could not find reliable information negative relationship exists between the number of for both historical and prehistorical extinctions. native species of birds or varieties of mammalian predators and the number of successfully established Island area introductions after partialling out variation that may The total area of all islands in the archipelago or be due to other variables (like island area, or settlement labeled island group is summed. It would also be nice history). Since almost universally the number of native to have a measure of the area still in virgin native habi- species on islands increases with their area and habitat tat. However in practice such measures are inherently diversity, and since a priori we might expect this same subjective since even apparently pristine habitats may trend in introduced species, the numbers of both be altered in subtle ways through human activities and species in both sets might be positively associated introduced mammals, insects, and plants, the creation across islands because they co-vary with island area of edge effects (Diamond & Veitch, 1981; Simberloff, and maximum elevation. To test Elton's hypothesis we 1990; Bolger et al., 1991). Different bird species will need to partial out the expected effects of area and respond to these local and landscape alterations in elevation (habitat diversity) of introduced and native different ways, so a meaningful number that represents species number, so that we can focus on these biotic that habitat remaining which is acceptable for all variables. To do this I first regress introduced species native birds is illusive. As a surrogate of overall number versus island area and maximum elevation and human impact to bird habitats, the number of avian save the residuals, which represent relative introduced extinctions may be a more appropriate 'bio-assay' (see species richness after accounting for these geographic below). covariates. This is repeated for native species richness. A stepwise regression is performed using the residuals Maximum elevation above sea level of introduced species number as the dependent variable This measure serves as a useful surrogate of habitat regressed in a step-wise manner against native species diversity (MacArthur & Wilson, 1967). number residuals, mammal categories, human occupa- tion categories, and the number of extinct native Human occupation h&tory species as independent variables, I do not partial-out Some places may lack avian introductions simply variation due to differences between islands' isolation because people have not been present or interested in distances, because isolation does not significantly affect introducing them. The locations in Table 1 are catego- introduced species number and its affect on native rized as follows: 1, locations with no permanent human species richness, while highly significant, is immaterial settlements or settlements only very recently established to the hypothesis. and still localized; 2, Locations with aboriginal settle- ment but no subsequent lasting European settlement; 3, Establishment success and failure Locations with no aboriginal settlement prior to colo- Table 2 contains a subset of locations for which rela- nization and settlement by Europeans; 4, Locations tively reliable records exist on introduction attempts. with both an aboriginal and European colonial period While the compilation of Long (1981) serves as an of settlement. excellent starting point, I have found a number of instances where introduced species exist or failed Mammalian predators attempts have occurred that are not mentioned in Long A rich native avifauna fauna will often be associated (1981). Also problematic, is that Long's criterion of with a rich fauna of other taxa that might prey upon 'success' is different from mine. Moreover, in a sub- birds. Islands are categorized according to the variety stantial number of cases, the outcome of the intro- of their native and exotic mammalian predators as duction is listed in Long (1981) as uncertain (his 'possi- follows: (0) none; (1) Polynesian rat Rattus exulans or bly successful introductions' category). Therefore the other native rats but no introduced rats (R. rattus or locations included in my Table 2 are those where I R. norvegicus) or other mammalian predators; (1.5) have independently been able to assess successes and feral domestic cats but no rats; (2) Rattus rattus and/or failures. 74 T. J. Case

g~ 0 n3 .o

~g

= -~,.2 ~.~

~o~

~.~ ~j o

>,~ 0 =~ .~ ~)"~=-~ .0 0 ~= 0 ~ 0000000000000000000000 o ~'~_~ •o ~ E::

~1 . ~ ~, .~ 1::~ 0 oo

~ o ~ .~ ~.~-

Z •=~o ~ N

~~ ~ ~ ~ .=

~g {9

~ ~ ~"~.~ ..~ 0 6 ~.o ~ ~.~ ~.~[_ Z

~.=_~-'~ =~~ ~ 0~~ 0 ,~'= = = E"~ L"-~'-

0

',~ ~ 0 ~ ~, .. ~, =~o o.~ ~ o~ -~ N 0 c~ o 0 © ~J Introductions and extinctions of birds 75

The success 'rate' is calculated on a per-species basis: ductions. To determine if competitive interactions the number of successfully introduced species to a loca- between exotics and natives may be influencing habitat tion is divided by the total number of species whose distributions at a finer geographic scale, I surveyed introduction was attempted. In calculating success birds in the following Pacific basin and rim sites from ratios, the problem arises as to what constitutes an 1983 to 1990: Fiji (four islands and 20 sites), Palau 'attempt'. I do not count attempts that involve (two islands, two sites), Hawaii (five islands, 21 sites), releases of only one individual of a species or multiple Tahiti (three sites), Cook Islands (two islands, two individuals but all of the same sex, or migratory sites), Marquesas (two islands, two sites), Vanuatu (two species. An attempt is defined as the release or self- islands, six sites), New Guinea (two sites), New colonization of a land or freshwater species that is not Zealand (two islands, 18 sites), New Caledonia (three native to the location. Reintroductions of native species sites), mainland Australia (19 sites), and Kangaroo are excluded. Multiple releases of the same species are Island off South Australia (one site). This collection of counted only once. To be counted as an attempt, Pacific sites includes those with avifaunas ranging from records must show that at least a pair of individuals few to many introduced bird species and from few to was released into the wild (not simply shipped to a many native bird species. Each site was categorized location). Some exotic birds now occur in the avifaunas into one of four habitats: native forest; exotic forest today because they colonized recently on their own or (usually pines or eucalypt plantations); disturbed native because they are cage-escapees but there is no record of habitats (at least 50% exotic species by areal coverage their importation or release. These species are counted and always secondary growth); and suburban/urban both as attempts and as successes. Since there is an locations (usually city parks, college campuses, or sub- unknown number of failures in this category, the urban housing tracts) with typically much open space success rates in Table 2 are likely to be artificially high. and over 50% exotic vegetation. Table 2 is based on locations with at least 10 docu- Two types of censuses were performed. At all sites mented attempts and where I was able to verify the bird species lists were compiled based on visual and call species' present status either through personal visits or identification. Sites were surveyed over a period of 2-8 through the literature. I have personally conducted h over 1~1 days. The area surveyed was approximately field studies in about half these sites. 2-20 ha. For a subset of locations (n = 52) individual Many failures are immediate; the birds are released birds were counted using the variable circular-plot and disappear soon after. Other exotics maintain thriv- technique (Reynolds et al., 1980). Birds were counted ing and even expanding populations for some time but at between five and 14 stations 100-300 m apart along subsequently become extinct. Should we count the lat- a transect of continuous habitat. Birds were identified ter as failed introductions or as extinctions of success- by sight or by calls for 6 min at each station. Longer ful introductions? Moulton and Pimm (1983) in their time intervals may result in counting the same bird analysis of the Hawaiian situation chose the latter more than once (Reynolds et al., 1980; Scott et al., course and they were aided in their assessment by par- 1986). Because of dramatic differences between islands ticularly good historical records. However, in most in the numbers of bird species and in their habits and places it is difficult to know when a small population habitat structure, these data are not reliable for com- becomes extinct and thus how long it survived before paring absolute densities of birds between locations. perishing. I therefore choose the conservative measure The data are useful, however, as rough estimates of the of counting successes as only those introduced species relative abundances of native vs introduced species at that survive to the present time. Moulton and Pimm different sites and in different habitat types. (1983), Lockwood et al. (1993), Simberloff and Boecklen Additionally, I have compiled each location's native (1991) and others have restricted their analyses to only species number, area, and maximum elevation. These passerines or to passerines plus doves. Exotic game birds values differ from those in Table 1 since now I am and parrots are common introductions but are excluded interested in a specific island's values not archipelago because many parrots are established as cage-bird escapees values. For this purpose mainland Australia was while human hunting pressure could exaggerate the divided into eight geographic regions. failure of game birds. However, such a distinction seems arbitrary, since many finches are also kept as Literature surveys cage birds that end as established exotics (Long, 1981) Bird abundance data in New Zealand and Australia and game birds include examples of both striking were also gathered from studies in the literature (Gibb, successes and failures (Banks, 1981; Pimm, 1991). 1961; McLay, 1974 [a review of five New Zealand studies]; Disney & Stokes, 1976; Friend, 1982; Wall, Relative abundance of introduced and native species in 1983). These studies were performed over much longer native and disturbed habitats time intervals, of a few months to a few years, com- Personal surveys pared to the short-term census data that I collected. The first two data sets are of value in understanding These surveys are analysed in the same way as my own regional patterns in the number and success of intro- surveys to check the robustness of results obtained 76 T. J. Case from my short-term surveys and to examine abundance . I . , . t . L . | . i . i . f . I . data for native and introduced birds. i .8 .6 Statistical analysis Raw variables were examined for normality and trans- .4 formed by the angular transformation (for proportions), .2 square root (maximum elevation), or a log transforma- .gv 0 tion (numbers, counts, and area) if that produced a -.2 • • •; -, • o~ • • • • • Q more normal distribution. Since some variables contain - 4i zero values for some islands/regions, I added the ~n -I ..6 ¸ constant 1-0 to these variables before applying a log Q transform. In stepwise regressions, the variables were "~ -.8 -1.2 -1 -,8 -.6 -.4 -.2 0 .2 .4 .6 .8 stepped first in the forward and then in the backward direction. The backward model begins with all vari- Residuals native species number ables added and successively drops those with insignifi- Fig. 1. A visualization of the negative partial correlation cant partial correlation coefficients with the dependent between log introduced bird species numbers and the log of variable. In the vast majority of cases, direction did not native bird species numbers after partialling out variation in influence the resulting regression model. In the few log area and square root elevation. The regression line is cases where the stepwise models differ in the two direc- shown: Y = 4).298 * X; R2= 0-083; p = 0-013. tions, I report the model that produces a significantly better fit to the data. The residuals around the regres- variance is quite low (about 8.3%). Much more impor- sion were visually examined for homoscedasticity and tant is the number of extinct species, which is the only results are reported only when this assumption is met. independent variable entered into a stepwise regression Statistical analyses were performed using the statistical model, and which correlates positively with the residu- software Statview TM 4.0 from Macintosh. als of introduced species number. The same result emerges for both forward and backward stepwise RESULTS regressions and the regression is highly significant (p <0.0001), explaining 43-4% of the variance (Table Introduced species number 4(a)). Thus nearly half the variation in introduced The correlation matrix for the variables in Table 1 is species richness (controlling for island area and eleva- shown in Table 3. While log area is highly correlated tion) is explained by the number of extinct native with log extant native species number (R 2 = 0-66), the species in the avifauna. A rich native bird fauna per se, correlation between log area and log exotic species however, does not seem to hinder the success and per- number is only marginally significant (Table 3, sistence of exotic birds after controlling for island area, p = 0-07). Hence, there is substantial unexplained varia- elevation and extinct native species. tion in the numbers of introduced species that might be An alternative stepwise multiple regression technique accounted for by other factors. After partialling out the is to ignore any a priori expectations about potential affect of area and elevation on both exotic and affects of island area and elevation on species number. native bird species number, a stepwise regression was Now the log of introduced species number (rather than performed using the residuals of introduced species the residuals of this variable after accounting for area number as the dependent variable and the remaining and elevation) is used as the dependent variable and variables in Table 1 as independent variables. Native area and elevation are included as dependent variables species richness (the residuals after accounting for area in a stepwise regression along with the remaining vari- and maximum elevation) shows a significant negative ables in Table 1. The stepwise regression in the forward partial correlation coefficient (p = 0.013; Fig. 1) with direction first adds extinct native species number as the the introduced species residuals, although the explained most significant independent variable, and it alone

Table 3. Correlation matrix for variables in Table 1

Log extinctions Log introduc. Sqrt elev. Log area Human Mammal cat. Log natives

Log extinctions 1-000 0.664 0.125 0.046 0.171 4)-070 4). 154 Log introduc. 1-000 0-216 0.216 0-188 0.029 0.030 Sqrt elev. 1-000 0- 710 0.187 0.394 0.494 Log area 1.000 0.408 0-673 0.823 Human settlement 1.000 0-473 0-413 Mammal cat. 1.000 0.686 Log natives 1.000 Coefficients greater in absolute value than 0-230 are significant at p -- 0.05. Coefficients greater than 0.302 are significant at p = 0.01. Introductions and extinctions of birds 77

Table 4. Stepwise regression models using the variables in Table 1 to account for variation across locations in the number of exotic or extinct native bird species

(a) Dependent variable: Residuals of log introduced species number after being regressed against log area and sq. root of maximum elevation

Total d.f.: 69 p-value <0.0001 R2:0-444

Variable Coefficient Std Error F to remove

Intercept 0.278 0-051 30.184 Log extinctions 0-532 0.072 52.211

Variables not significant in model: native species number residuals, human settlement, preda- tor category.

(b) Dependent variable: log of introduced species number

Total d.f.: 69 p-value <0-0001 R2: 0.478

Variable Coefficient Std Error F to remove

Intercept 0.285 0.094 9.272 Log area 0.052 0.025 4.473 Log number extinctions 0-531 0.071 55.308

Variables not significant in model: native species number, human settlement, predator category, maximum elevation.

(c) Dependent variable: log of extinct native species number. Table 1 variables plus native species endemism

Total d.f.: 69 p-value <0.0001 R2:0.511

Variable Coefficient Std Error F to remove

Intercept -0.201 0-097 4.570 Sq. root species endemism 0.385 0.122 9.907 Log introductions 0-762 0.109 48.780

Variables not significant in model: extant native species number, human settlement, predator category, maximum elevation, log area.

(d) Dependent variable: the proportion of extinct native species (sq. root transformed). Analysis restricted to those locations with at least one extinct and at least one extant native species

Total d.f.: 45 p-value <0.0001 R2: 0.655

Variable Coefficient Std Error F to remove

Intercept 0-505 0.077 43.088 Log area -0.147 0-022 44-262 Log introductions 0.279 0.064 19.011 Log number of endemics 0-17 0-047 12.931

Variables not significant in model: human settlement, predator category, maximum elevation. accounts for about 44.3% of the variance in introduced repeat the analysis but only include those islands with species. However, log area is also added (Table 4(b)) at least one (or two) extinctions, and at least two intro- with a positive coefficient and the combined two vari- duced species under the assumption that those islands ables account for 48% of the variance in log introduced with very few known extinctions or introductions are species number. This regression is highly significant simply understudied, the stepwise model enters only (p <0.0001) and the same model is reached from both the number of extinct species and this is strongly and forward and backward stepwise procedures. If we positively correlated with numbers of introductions 78 T. J. Case

(d.f. -- 49 and d.f. = 32; both p <0.0001). Finally, if we Discussion, I will examine sources of bias in these data, screen the locations to use only those where subfossil e.g. the likely underestimation of extinctions. Since the birds are known (Table 1), the stepwise model (in both number of extinctions is not significantly correlated the forward and backward directions) again produces with area (both variates log transformed, p >0-7) and the same basic result: the number of extinct species is the number of introductions is only marginally corre- the only variable entered and is strongly and positively lated with area (p -- 0.079, both variates log trans- correlated with numbers of introductions (n -- 37; formed), the correlation between the number of p <0.0007). introductions and extinctions is not simply due to While log transformed species numbers are used in mutual covariation with area. If we include only those the multiple regressions, the precise relationship between islands where there is at least one extinction, under the introductions and extinctions is more easily visualized assumption that those islands with no extinctions are by plotting their untransformed values (Fig. 2). The simply under-sampled, the correlation between log slope here is about 0.74, implying that on average extinctions and log introductions is still highly signifi- about four extinctions have occurred for every three cant (p <0.0001, n = 50). Similarly, including only introduced species gained. Thus species number has on those islands with known subfossil birds, does not sub- average remained roughly steady despite substantial stantially alter either the slope or the significance of the gains and loss of species and varying degrees of habitat correlation (Fig. 2(b)). destruction and deterioration across locations. In the Native extinctions Since extinctions are a reliable predictor of introduc- (a) tions, it is useful to explore the factors contributing to variation in the number of extinct species and rates of 65 extinction across these locations. A stepwise regression .~ 55 was performed using the number of extinct species as u New Zealand /~,~J'f~ o. 45 the dependent variable (log transformed). An addi- ~n • / tional dependent variable was added to those in Table 35 ~g Kauai Mauritiusf~,~ ~, m 1, the level of species endemism, i.e. the number of u "O= 25 endemic species (extant natives including those histori- O • Reunion mr~ cally extinct that are endemic at the species level to the *., 15 c location) divided by the total number of native species. 5 The most significant variable explaining the number of extinct species for Table 1 locations is the number of .... i .... i .... ! .... i .... , .... i .... i ,, S 15 Z5 35 45 55 65 introduced species (Table 4(c)) attesting to the tight and Extinct native species reciprocal relationship between extinctions and intro- ductions. The stepwise model next entered the level of (b) endemism (square root transformation) with a positive i .... i .... , .... i .... i .... i .... i .... I~ affect on extinction; this two-variable model explains 65 ~ I 51% of the variance in the number of extinctions across islands. A backward regression yielded the same final model. I next limited the analysis to only those islands o. 45 s I with at least one extinction and one introduction. This reduces the sample size to 47 but improves homo- scedasticity; the resulting regression model enters only the number of introduced species, which explains 37%

C of the variance (t9 <0.0001). Screening the locations to include only those where some subfossil birds are known (Table 1, asterisked locations eliminated) reduces the sample size to 36 locations, and returns s 15 Z5 35 4S 55 65 endemism to significance with introduced species Extinct native species (p = 0-0004; R 2 = 0-376). Fig. 2. (a) The relationship between the number of exotic and The proportion of extinct native species in the avi- extinct native bird species for the locations in Table 1. fauna (i.e. No. all extinct species / [No. all extinct + line of equality; --, regression line (Introduced = No. extant species]) is a variate of interest since it tells 3-084 + 0.742 * Extinct; R 2 -- 0-699; p < 0.0001; n -- 70). us something about relative extinction rates between Some islands with high numbers of extinct and introduced locations. I eliminate those islands with no extinctions species are labeled. (b) Same as in (a) but only those islands from Table 1 where prehistoric birds have been uncovered and no extant native species, and then apply a square from fossils are included. (Introduced -- 4.135 + 0.704 * root transformation to the proportion of extinct species Extinct; R 2 -- 0.657; p <0.0001; n = 37). in the remaining locations. The proportion of extinct Introductions and extinctions of birds 79 species for this sample of locations is negatively corre- islands, the relationship between percent extinct natives lated with area (Fig. 3(a)). A non-parametric Kendall vs only log area is also tighter (Fig. 3(b); R 2 = 0.407 rank correlation also finds that area and the proportion compared to the set of islands in Fig. 3(a) that of extinct species are inversely correlated across the includes some without prehistoric bird discoveries). entire set of locations (p -- 0.0081). A stepwise regression using the proportion extinct for the reduced Introduction success and failure island set as the dependent variable first enters area One obvious factor that may influence success is simply (with a negative coefficient), then introduced species the number of introduction attempts, since clearly there number (positive coefficient), and finally the number of can only be a few introduced species in the avifauna of endemic species enters with a positive coefficient (per- today if only a few species were introduced. I first cent endemism is not used here since its demominator, examine the relationship between the number of intro- native species number, is the same as the denominator duction successes (i.e. the number of extant breeding of the dependent variable creating the potential for introduced or self-introduced species in the avifauna spurious correlation). This model explains 66% of the today) and the number of failures across locations, explained variance (Table 4(d)). (Note that native then I search for those variables that best explain varia- species number was excluded as a candidate for consid- tion in the relative success rate (defined below). eration since it is in the denominator of the dependent A positive relationship between successes and failures variable). The same set of three independent variables would indicate that areas with high numbers of intro- also are included by stepwise regressions applied to duced species owe their success, in part, simply to more only those islands where some subfossil birds are persistent introduction activities: success will be posi- known (n = 36; R 2 -- 0.610). Similarly, for this set of tively correlated with failures, if successes come through repeated attempts and thus also repeated (a) failures. Alternatively, successes and failures would be negatively correlated if site-specific factors involving 1.1 - -,', ,*l* ,''" "" '" "''" "''" "'',,, characteristics of the biota or consistent differences in .| release methods govern overall success. For example, we would expect communities that intrinsically repel invaders to have both low established numbers of D introductions and high numbers of failures. This • .6 "6 .S pattern would also be expected if some acclimatization societies were consistently more diligent about the introduction or release methods; these places might have low failures and high success relative to other areas. " "" :' " o .I ~L The relationship between introduction success and

0 - - - i - - - i - . - i . . . ~m- % . i . . . I . . . i . . failure for the sites in Table 2 is shown in Fig. 4; it is 0 1 2 3 4 5 6 7 significantly positive (p <0.006; R 2 = 0.32), suggesting log area (sq kin) that generally successes come with repeated failures, (b) but the high degree of scatter also points to other fac-

1.1 , , , I , - -, - - - "- -- '- -- '- - - ' - --'- -- tors that must also be influencing success and failure (and possibly recording errors).

1.~ I I [ I I I I I I

$ • 1.6 "~ .6 • ~ 1.4 • • ~ .S 0 C .4 ._o .3 -- ,,,~ I.O • ~~oo o • • • 2 .'t • ; o i 0.8 ~ o• • 0 - - - i - - - m ...... 0 1 2 3:7" 4 s 6 7 8 0.6 "f • •

k)g area (sq km) 0.4 -o.zs ~.o o3s o.s o3s 1:o 1.1~ fs ~s ~.o 2.zs Fig. 3. The proportion of extinct species in the native avi- fauna plotted (with a square root transformation) as a func- log inla'odueed failures tion of log area. (a) Only islands with at least one extinct and one extant species are included. The regression line Fig. 4. The relationship between the number of successfully is shown: Y = 0-689 - 0-086 * X; R 2 = 0.238; p = 0-0006; introduced exotic birds and the number of failures for n -- 46. (b) Only islands with prehistoric birds described. the locations in Table 2. The regression line is shown (log (Y= 0-943 -0.135 * X; R 2 = 0-407, p <0.0001; n = 37). successes = 0-735 + 0-299 * log failures; R 2 -- 0-322; p = 0-0059). 80 T.J. Case

What other dependent variables in Table 1 best F~ proportion exotic by numbers A. explain the variation in introduction success between E~ proportion native by numbers locations? We desire some measure of success that is 1.2 independent of the number of introduction attempts. 22 20 4 6 Since success rate (column B of Table 2) contains intro- 1.0 duction attempts in its denominator, the two variables O.8 are not statistically independent. I use instead the residuals from the regression in Fig. 4, which represent 0.6 a measure of relative success rate. They are used as the 0.4 dependent variable in a stepwise regression with candi- date independent variables: extant natives, extinct O.Z natives, area, maximum elevation, and mammal preda- 0 tor category, each suitably transformed for normality. The only variable that is Significantly correlated with urban/ native exotic secondary suburban forest forest habitat relative success rate is the number of extinctions (log transformed; Fig. 5). A non-parametric Kendall rank [] proportion exotic by species correlation supports the significance of this correlation B. [] proportion natwe by species (p = 0.039). 1.2 D In summary, the number of successfully introduced 46 32 9 11 species increases with the number of failures and the 1.0 relative success to failure rate increases with the number 0.8 of extinct native species. These results are consistent with Elton's (1958) idea that habitat disturbance increases 0.6 invasion success but does not support his contention that 0.4 species-rich locations, in and of themselves, repel invaders. 0.2 Habitat distributions of exotic and native birds Both on the basis of numerical abundance and species 0 number, human-modified (or non-native) habitats con- urban/ native exotic secondary suburban forest forest habitat tain a greater proportion of exotic birds than do native forest habitats (Fig. 6). Moreover, the local number of Fig. 6. The proportion of individuals (A) and species (B) that exotic species observed in surveys and the regional are exotic or native in the four different habitat types sur- veyed across a range of sites on several Pacific locations. The number of exotic species in the avifauna are highly cor- numbers over the bars are the number of sites in each habitat related in the two most common habitat types (Fig. 7(a)). category. A one-factor ANOVA performed on the angular (There are too few surveys in secondary growth habitats transformation of these proportions shows that habitat and exotic forest habitats to determine their separate significantly affects the proportion of exotic birds in the relationships.) The numerical abundance of exotic birds different habitats. relative to native birds also increases with the overall island's exotic bird species number (Fig. 7(b)). Not too surprisingly the number of native species observed in local surveys is positively correlated with .5 , i , 1 , i ._ , . t , i , t , t . , . I , the number of native species on the island in both .4 -' • I urban/suburban and native forest habitats (p <0.0001, Fig. 8(a)). The proportion of individuals in local surveys that belong to native bird species increases with native .1 species number (Fig. 8(b); p <0.02 for both habitat io types). More germane to Elton's hypothesis, the number of

~ -.Z ~ • 00 • exotic bird species in point surveys significantly declines ~ -.3 with the number of native species present in these sur- -.4. veys in both habitat types (Fig. 9, urban!suburban,

° S ~ p <0-0001; native forest, p <0-001). Since the number .Z 0 ,2 .4 .6 .8 1 1.2 1.4 1.6 1.8 of native species in point censuses is highly correlated with the number of native species in the island avi- log number of extinctions fauna, it is difficult to disentangle the relative effect Fig. 5. The relative success rate for avian introductions plot- that the regional species pool and habitat clearing, ted against the number of avian extinctions for the locations in Table 2. The relative success rate is obtained from the compared to local factors, might be playing in produc- residuals of the plot in Fig. 4. The regression line is shown: ing the negative relationship between natives and Relative success rate -- ~).165 + 0.214 * X; R2 = 0.226. exotics in the point surveys. Introductions and extinctions of birds 81

O urban/suburban • urban/suburban (a) o nrdve forest (a) o n,uve ~r~

1.~ , I . t , i . i . i . i . , . I , I , 1.8 I I I I I I I I

1.6 Q

1.4 .c_ .8 .~ 1.z

0 .6 .~ 1

.4 (J Vt .6 0 .2 -~ .4

O~ C .2 0 0 .J o~ o ._1 • •ell •

*.2 i i J i i i i i -.~' 0 .Z .4 .6 .8 1 1.Z 1.4 1.6 1.8 .$ .75 1 1.ZS 1.5 1.75 2 2.2S Z.S Z.TS Log exotic species on island Log native species on island

(b) (b) 1.8 • i . , . i . J . i . ] . i . i , i • k 1.8 i I , I i I I i ~= 1.6 ~ 1.6 Q liE] D ~ ol c 1.4 .~_ 1.4 .~ ~.z u~ 1.Z

"0.-.R .8

.6 0 .4 •- .4 c .Z ._o e- t:t 0 ._o 0 -.2 t~ 2 ~. -,Z i i i i i r i -.2 0 .2 .4 .6 .8 1 1.2 1.4 1.6 1.8 0 o- .S .7S 1 1.Z5 1.5 1.75 2 Z.Z5 Z.5 Z.7S Log exotic species on island Fig. 7. (a) The relationship between the local number of exotic Log native species on island species and the regional number of exotic species in the avi- Fig. 8. (a) The relationship between the number of native fauna for the two most common habitat t~pes. The regression species in local point surveys and the regional number of lines are shown (urban/suburban habitats: R- = 0.625, p <0-0001, extant native species on the island/archipelago (urban/sub- n = 46; native forest: R 2 = 0-526, p <0.0001, n = 32). (b) The pro- urban habitats: R 2 = 0-536, p <0.0001); native forest: portion of exotic individuals observed (with an angular trans- R 2= 0.515, p < 0-0001). (b) The relationship between the pro- formation) in local surveys plotted as a function of the portion of native individuals in local point surveys and the regional number of exotic species in the avifauna. The regression regional number of extant native species on the island/ lines are shown (urban/suburban: R 2 = 0-399, p = 0.0016, n = 22); archipelago (urban/suburban: R 2 = 0.259, p = 0-0155; native native forest: R 2 = 0-413, p -- 0.0022, n = 20). forest: R 2 = 0.306, p = 0-011).

Are regional effects important even after accounting native species number and exotic species number as for local site-specific variables? To answer this I independent variables in a stepwise multiple regression. regressed the local species number of exotics against Again residuals were saved and averaged for each the local number of native species across all sites (both island/region and then entered as the dependent vari- variables log transformed). I saved the residuals from able in the regional regression analysis. this regression, averaged them across all sites within The results are summarized in Table 5. Regardless of each island/region and then performed a multiple habitat type, exotic species number is negatively related regression using the averaged residuals as the depen- to the number of native species in the points censuses. dent variable and island/region species number for After accounting for this variation, the residuals in exotics and natives as the independent variables (again exotic species number are further positively related to log transformed). In this regression there is thus only a exotic species number for the entire island/region. In both single point per island/region to avoid pseudoreplica- forest habitats and man-modified habitats, the proportion tion. This was repeated separately for non-native and of exotic individuals in surveys is negatively related to native habitats. Finally, this process was repeated the number of native species present in the survey and beginning with the regression of the local proportion of positively related to the number of exotic species. exotic individuals in each census (with an angular Regional species numbers did not significantly correlate transformation) as the dependent variable and the local with the average residuals from these regressions. 82 T. J. Case

significantly more native species in point surveys in • urban/suburban [ Australia than in New Zealand (p <0-0001) and signifi- rn native forest cantly more natives in native forest habitat compared • i , I . | . i , t , | . i , J , i 1.2 to human-modified habitats (p <0.0001) but no signifi- cant interaction term (p >0.9). As for my surveys, the number of exotic species is .~ .8 strongly negatively correlated with the number of

.-~ .6 native species present in point surveys in both habitat types (Fig. 11; p = 0.0002 for native forest and

.u p <0.0001 for modified habitats). For the New Zealand surveys (but not the Australian studies), the authors often supplied estimates of the o 0 _J density of individuals in the study area. The total num- *.2 i - v . i • 7 • L . i • i - ~ - i - ber of exotic birds is positively correlated with the Z 0 .Z .4 .6 .a 1 1.z 1.4 1.6 1.8 number of exotic species present in the survey for both Log native species in survey habitat types (Fig. 12(a)). The number of exotic indivi- Fig. 9. The negative relationship between the number of duals is negatively correlated with the number of native exotic species found in point surveys and the corresponding species in modified habitats but not significantly so in number of native species (log transformed) (urban/suburban: native forest (Fig. 12(b)). A multiple stepwise regression R-' = 0-364, p < 0-0001; native forest: R 2 = 0.194, p = 0.0117). with the log of exotic individuals as the dependent vari- able found that only the number of exotic species Literature surveys for New Zealand and Australia entered into the model for both habitat types. Thus the As for my personal surveys these surveys obtained statistical significance of the negative relationship from the literature also reveal a higher number (and between the number of exotic birds and the number proportion) of exotic species in human-modified habi- of native species in point surveys disappears once tats than in native forest (Fig. 10). New Zealand has between-site variation in exotic species number is substantially more exotic species in its avifauna than accounted for. This suggests that even in fixed habitat Australia and this leads to higher numbers of exotic types exotic species and native species are negatively species being present in point surveys in both native associated across sites, but that those natives that are forest and modified habitats (Fig. 10). A two-factor present in a site are not affecting the numerical abun- ANOVA based on the results shown in Fig. 10 finds a dance of exotics very much. The number of native indi- significant location effect (i.e. more exotic species in viduals is not significantly correlated with either the New Zealand than Australia; p <0.0001) and significant number of native species or the number of exotic species habitat affect (p <0.0001), but the interaction term is in point surveys in either habitat type (not shown). not significant. Similarly, a two-factor ANOVA based Similarly the number of native individuals is not cor- on native species number in the literature surveys finds related with the number of exotic individuals in either

Table 5. A summary of stepwise multiple regression results separating the independent variables into likely regional effects and local effects The sign of the coefficient (positive or negative) is indicated with the statistical significance level by the number of asterisks. * p = 0-05 0.01, ** <0.01 and >0-001, *** <0.001 and > 0.0001, **** <0.0001. The box outlines those cells where local inter- specific competition between natives and exotics might be expected to yield negative signs. The regional variables are regressed against the average residuals (over all sites within a region) from the regression involving only the local variables at each site. NA, not applicable.

All non-native habitats Native forest habitats Dependent variable Dependent variable

No. exotic species Prop. exotic No. exotic species Prop. exotic in surveys indiv, in surveys in surveys indiv, in surveys

Regional variables (Species source pools No. native species and broad habitat on island alteration) No exotic species pos. ** pos * on island (n = 26) (n = 18) (n = 18) (n = 11)

Local variables (Possible competitve No. native species neg **** neg **** neg * neg *** interactions and in survey habitat variation) No exotic species NA pos **** NA pos ** in survey (n = 66) (n = 32) (n = 32) (n = 20) Introductions and extinctions of birds 83

(a) Literature censuses only [ [ 1 .9 Mean no. .8 exotic .7 species in point .6 surveys .$

.4 .3 oo 0 , .2

.1 .4 .6 .s ~ 1 .z ~.4 ~.6 ~.8 z native forest human-modified log number of native species Fig. I1. The negative relationship between the number of (b) All censuses combined exotic species found in point surveys and the corresponding number of native species for Australian and New Zealand sites gleaned from the literature (modified habitats: R 2 : 0"444, p <0.0001; native forest: R 2 = 0.394, p = 0-0002)•

.7

Mean no. .6 formed variables; 45% for transformed variables). The exotic .s Y second result is that invasion success does not decline species in point .4 significantly with the richness of the native avifauna surveys .3 nor the variety of potential mammalian predators once .2 between-site variation in extinctions and island area are

.1 accounted for.

0 This latter result contrasts with the pattern seen for native forest human-modified reptiles on much the same set of locations analysed Fig. 10. The cell means for the mean number of log exotic here for birds. Case and Bolger (1991) found that species in native forest and all human-modified habitats com- native reptile species number had a highly significant bined for Australian and New Zealand sites. (Error bars are negative partial correlation coefficient with introduced 95% confidence intervals.) (a) Only surveys from the litera- species number. After updating these data for recent ture. (b) Those same literature surveys combined with my discoveries and taxonomic revisions and including own surveys from New Zealand and mainland Australia. information on the numbers of extinct reptile species and human settlement patterns (not considered in the habitat type. We would not necessarily expect negative original paper), I find that island area (with a positive relationships between species abundances of these two coefficient) and native species number (with a negative groups across sites because of gradients in overall coefficient ) are the only significant variables explaining resource productivity, which would impose positive the numbers of introduced reptiles across islands (and covariance. That is, birds of all species will respond USA and Australia) in a stepwise regression model. similarly to differences in avian carrying capacity However, the number of reptilian extinctions is poorly between sites diluting the appearance of any negative known for most of these island locals (see Case et al., correlations that might emerge if we could control for 1992) and the explained variance of the model is only environmental differences. about 15%, much less than that for birds. Native species number may play a larger role in reptiles than in birds in preventing exotic success and persistence DISCUSSION because exotic and native reptile species seem to share Regional patterns habitats more frequently than do exotic and native This review finds two particularly intriguing results. birds (Case & Bolger, 1991; Losos et al., 1993). First, I find a strong correlation between the number of bird species that have successfully been introduced to The rough substitution of extinctions by introductions an island (as well as the relative success rate) and the We should probably not interpret too much into the number of known extinctions of indigenous species in slope of the relationship between species gains and historical and aboriginal times, such that the number of losses, which is about 0-75. While the equilibrium exotic species gained is close to the number of species theory of island biogeography (MacArthur & Wilson, lost through extinction. Indeed the correlation between 1967) posits a close relationship between species colo- introductions and extinctions accounts for 70% of the nizations and extinctions, this expectation is based on variance in introduced species number (for untrans- islands that are receiving natural immigrants and not 84 T.J. Case

(a) when disturbed habitats are saturated with new exotic species, we will have approximately 0.84 S exotic species living in disturbed habitats plus 0-84 S native 3.Z species living in native forest habitats; the total number

.~ Z. of species island-wide is about 68% (= 2 × 0.84 × 100)

~ 2.6- greater than before half the island was deforested. For this scenario, at equilibrium the gain in exotics far out- numbers the loss of natives from extinction. On the = 2 other hand, if exotics and natives showed no differen- tial habitat preferences, then at equilibrium the number of gains and losses are expected to be equal preserving 1 4L~e I the total number of species before and after deforesta- 1.Z -~ tion. With intermediate habitat sharing of exotics and .2 .3 .4 .5 ~6 .7 .8 .9 1 1,1 native species, and with some islands experiencing more log number of exotic species habitat conversion than others, and predators poten- tially reducing carrying capacities for both exotics and (b) natives, and islands probably not close to equilibrium 3.4 .... 'o .... B ...... with respect to their recent and ongoing habitat alter- 3,Z3 ations, the expected relationship between gains and 2.8 losses cannot be deduced. We should therefore take the 2.6 o observed slope of 0.74 as an empirical observation for "~ 2.4 this taxon and group of islands at this time, but o • •.g Z.2 expect no general canonical properties. Since several

'~c 2 • island groups probably have many more extinct species ..~'- 1.a" waiting to be discovered, the slope could decrease. 1.G: Avifaunas with high levels of endemism have experi- 1.4 : • • • enced relatively more extinctions for this broad set of 1.2 .... , .... , .... , .... , .... , .... , .... islands matching the pattern found by McDowall .5 6 .7 .8 .9 1 1.1 1.Z (1969) for New Zealand birds. McDowall found that log number of native species 37% of the endemic New Zealand birds are now extinct Fig. 12. The relationship between the numbers of exotic but only 6% of the native but non-endemic species. individuals in point surveys and the number of exotic species Moreover, he found a tight relationship between the (a) and the number of native species (b) present in the New degree of endemism and the propensity for extinction. Zealand/Australian region. (a) Modified habitats: R 2 = 0-847, Elsewhere (Case, in prep.), I show that there is a p = 0.047, p <0.0001; native forest: R 2 = 0.312, p = 0.047. high correlation across locations in the endemism levels (b) Modified habitats: R 2 = 0.577, p -- 0-0002; native forest: R 2 -- 0-092, p = 0-314. Even though the total number of of extant and extinct species, and endemism levels in exotic bird individuals declines as the number of native land birds are correlated to those of land reptiles. species increases in modified habitats (but not native forest; Endemic reptiles on many islands are also more extinc- (b)) the positive effect of exotic species number on exotic indi- tion- (and extirpation-) prone than non-endemic species viduals is stronger (a) and when both variables are available in a stepwise regression, only exotic species number is entered (Case et al., 1992) and the same pattern appears to be into the model (see text). generally true for land birds (McDowall, 1969; Johnson & Stattersfield, 1990; Steadman, 1991; Adler, 1992). Another contributing factor to the apparent vulnera- simultaneously undergoing major habitat alteration. bility of island endemics could be the lack of recolo- When these conditions are added, the equilibrium nization sources. When a population of non-endemics theory can predict numbers of extinctions very different or regional endemics becomes locally extinct on an from the number of colonizations. Consider the classical island, the island can potentially be recolonized from thought experiment suggested by Preston (1962) of individuals still surviving on other nearby islands. For dividing an island into two equal but now isolated a single-island endemic, however, extirpation and halves. Since the number of species usually scales with extinction are equivalent. area raised to a power of about 0.25; halving area One consequence of the rough substitution of exotics results in only about a 16% loss in species number for natives and particularly endemic natives is that in (i.e. 1 - 0.5°25). If we deforest half an island and if all the calculation of log species-log area relationships for of the native species cannot tolerate this new open the islands in Table 1, the explained variance (72-1%) is habitat, then over time extinctions should eliminate greater when the extant native species number is added about 16% of the native species. Now we introduce to the islands' extinct native species and the total new bird species that can only occupy these new open species number is used as the dependent variable habitats. Call the original number of species S; then compared to when either species set is regressed alone Introductions and extinctions of birds 85

(R 2 = 66.6% for extant natives alone and R 2 = 0.2% for species and Holdaway (1990) attributes the high extinc- extinct natives alone). Similarly, when the extant tion rate on New Zealand before European settlement natives are added to the introduced species component, to the removal of forest (particularly dry forest). log area again explains 72% of the variance. Hawaiian native passerines rarely are seen outside native forest habitats (Richardson & Bowles, 1964; Scott Cause and effect: extinctions and introductions et al., 1986) but the richer avifaunas on Fiji and or Most avian extinctions occurred prior to the majority New Guinea have a complement of species that of species introductions in most locations. For the set frequent clearings, forest edge, and secondary growth of locations in Table 1, there are 139 extinctions since as for most mainland native avifaunas (Pearson, 1977; the approximate year 1600 and 231 prehistorical late Watling, 1983). As more of an island's area is con- Holocene extinctions. At least half of the historical verted to urban, agricultural and disturbed habitats or extinctions also probably occurred before the establish- altered through the introduction of herbivores and ment of introduced species through the efforts of exotic predators, most natives lose good living space acclimatization societies in the mid to late 1800s. This while most introduced birds that frequent open and temporal sequence indicates that most native avian disturbed areas and have evolved in predator-rich areas extinctions are not directly caused by competition with gain habitat. All else being equal, small populations are introduced species, although some could be. more vulnerable than large ones (MacArthur & Wilson, The native extinctions, whether in prehistory or in 1967; Leigh, 1981; Goodman, 1987) and extinctions more recent periods, are for the most part traceable to rates, where measured, generally increase with decreases human activities including hunting, habitat destruction in habitat area (Diamond, 1984; Case & Cody, 1987; and the introduction of exotic predators or large herbi- Soul6 et al., 1998; Bolger et al., 1991). Beginning vores that disrupt habitats (for birds: Simberloff, 1981, in aboriginal times, the conversion of native habitats, 1990; Cassels, 1984; Diamond, 1984; Olson & James, particularly at lower elevations, to disturbed habitats simul- 1984; Ebenhard, 1988; Holdaway, 1989; Olson, 1989; taneously enhanced the success and persistence of intro- Steadman, 1989b; and in reptiles: Case et al., 1992; duced species, while decreasing population sizes and Henderson, 1992; North et al., 1994). The idea that increasing extinction rates of native species. Thus habi- these native extinctions might cause vacant niches for tat conversion and deterioration alone could produce the introduced species is difficult to reconcile with the a correlation between the number of extinct natives fact that there are few clear ecological replacements in and the number of introduced species even without the two species sets. For example, In New Zealand any direct cause or effect between birds in these two giant moas are lost while smaller thrushes and finches groups. are gained; Hawaii experienced a loss of many ducks, Probably adding to the scatter in this relationship geese, raptors, rails and forest nectar feeders but a gain are differences between regions in the degree of Euro- of mostly small seed and insect eaters of open habitats. pean exploitation and relish for introducing exotic bird Worldwide, many endemic rails have become extinct species. Madagascar has lost around 85% of its native but rails or species with ecological niches like rails are habitats (Jolly et al., 1984) and at present has lost 13 not generally gained through introductions. native bird species and at least 13 of the 75 native It must be remembered, however, that direct biotic mammal species (approximately 17°/,, of the native interactions like predation, interference competition, fauna; Jolly et al., 1984) yet it never had an organized and parasitism are easier to see and therefore it is acclimatization society introducing exotic birds and easier to establish proof of their importance. Competi- today it harbors only three naturalized exotic species tion for food is not a direct interaction between two (rock dove, common myna, and house sparrow). At the competitors but is only manifested through the dynamic other extreme are small very isolated islands like consequences on other species: the resources. The level Easter, or the Chagos group that had no or very few of proof needed to demonstrate exploitation competi- native land bird species to begin with. Here the number tion is burdensome (Petren & Case, 1996). Thus, with of extinctions must be zero or small, yet when humans exploitation competition we have the double difficulty broke the isolation barrier by introducing exotic birds of first demonstrating that it is an important force in some of them were successful. Still other deviations present-day communities, and secondly connecting its from the correlation in Fig. 2 may be due to incom- occurrence to patterns at the biogeographic scale. plete fossil and subfossil discovery. While the level of An alternative explanation for the correlation is habitat deterioration in Fiji, Vanuatu, and most of the based on the fact that native birds are usually more Solomons is mild relative to Hawaii or Tahiti, it also common, if not restricted, to native habitats while seems likely that these places may have had more pre- introduced birds are primary occupants of disturbed historic extinctions than are presently known judging and open habitats (Figs 6 and 10, see also Temple, from the ever-expanding number of extinct forms found t981; Moulton & Pimm, 1983; Holdaway, 1990). Temple elsewhere in the Pacific (Olson, 1984, 1989; Olson & (1981) notes that almost all of the remaining indigenous James, 1984; Steadman, 1989b, 1991, 1992; Steadman birds in the Indian Ocean are obligate forest-inhabiting et al., 1990, 1994; Pimm et al., 1994). The absence of 86 T. J. Case described prehistoric extinctions in places like the The number of native extinctions seems to be inte- Seychelles, Comoros, Azores, and Cape Verde Islands grating the amount of native habitat loss, with the is simply because as yet there has been little attempt to absolute area of the island, and the deterioration of search for them (Olson, 1984). Australia is another remaining native habitat through the intrusion of exception with a plethora of introduced species but exotic predators, herbivores, and parasites (van Riper only a single known extinction, the historically extinct et al., 1986; Atkinson, 1989). For example, both paradise parrot. Additional extinct forms are known Hawaii and the USA have lost about 60% of native from the Pleistocene in Australia but none from the forest since human contact, but Hawaii has lost around Holocene, although again one suspects that there may 60% of its land birds while the continental USA has be many extinctions yet to be discovered (Pimm et al., lost about 1%. Here any attempt to explain the differ- 1994). ences in extinction rates to specific biotic interactions Thus the species numbers compiled here should be are complicated by the vast difference in area between thought of as dynamic, subject to change by taxonomic Hawaii and the USA, which alone will have a moderating revisions, new discoveries of extinct and extant forms, affect on extinction rates (Richman et al., 1988). New rediscoveries of species thought to have been extinct, Zealand and Great Britain are more similar in size but and new introductions and invasions of exotics. Since I Britain is a land-bridge island with a richer avifauna first assembled this data set over 10 years ago, many and a diverse set of native predators. New Zealand has revisions and discoveries have required its continual lost about 70% of its native forests since human con- amendment; yet the relationship between introductions tact while Great Britain has lost 93% in about the last and extinctions has increased with these new develop- 5000 years (King, 1984). Since human contact about ments rather than weakened. For example, Easter 40% of the land and freshwater birds of New Zealand Island has four successful introductions but until have become extinct but only 3°/,, of those in Great recently, no known extinctions and no extant native Britain (King, 1984 and Table 1). About one-third land birds. Steadman et al. (1994) recently reported six of the extinct New Zealand birds are various species of subfossil extinct land birds that probably became moas, most very large in body size and hunted for extinct after human contact. In the analysis above, food by native Maoris. King (1984) ascribes additional when I subjected the data to various screens, excluding losses in New Zealand birds to both habitat conversion locations where no prehistoric birds had been described, and introduced ferrets, stoats, weasels, cats, and rats. or eliminating locations with no extinctions or intro- The variety of predators is not a significant correlate of ductions, conclusions were unaffected and probability the number of introductions or their relative success rate levels stayed the same or even increased in some com- for the large set of islands studied here (Table 4 and the parisons in spite of lower sample size. results from Table 2). Of course, my categorization of predation is arbitrary and may not capture the true Other factors influencing introduction success predation risk for most birds in these various locations. Other factors influence both the immediate success and I have experimented with different weighting schemes, persistence of introduced species populations. A strong for example weighting exotic predators higher than correlation exists between the number of successes and native predators, without any substantial affect on the failures across the islands in Table 2, attesting to the conclusion, at least with respect to introduced species role of persistent acclimatization societies in increasing Success. species numbers despite high failure rates. Similarly, Native bird species without an evolutionary exposure even after controlling for avian extinctions, and all to exotic predators appear more vulnerable to exotic other independent variables in Table 1 and 2, island predators than their exotic counterparts (Atkinson, area correlates positively with introduced species number. 1985; Engbring & Fritts, 1988). Atkinson (1985) has While significant, the effect of island area on exotic found that on islands with native rodents or land crabs species number is weak and is not a significant corre- introduced rats have caused fewer bird extinctions than late of relative success rate, possibly because island on islands that were previously predator-free. Presum- area can both help and hurt exotic success at different ably bird species on islands with native predators have stages. In the early stages of the release of exotic birds, evolved effective predator escape behaviors that enable a large area will allow birds to disperse long distances them to evade the introduced rats or perhaps life his- from the release sites, diluting the initial release number tory features which allow populations to persist in spite over space and potentially making it more difficult for of high mortality. Because most introduced species the released birds to find mates. Yet, if the birds suc- come originally from predator-rich continental areas, cessively pass this hurdle and their numbers grow, then they may be less susceptible to introduced predators a wealth of theory suggests that larger areas will allow than endemic predator-naive species. The presence of larger absolute asymptotic population sizes, which in exotic predators (and parasites) might even enhance the turn will increase the likelihood of the population's success of introduced species compared to predator-free long-term persistence (Leigh, 1981; Diamond, 1984; islands by moderating the competitive impact of natives Case & Cody, 1987; Goodman, 1987). on the less susceptible exotics. Introductions and extinctions of birds 87

Habitat distributions and local diversity Diamond and Veitch (1981) concluded from their While I find little support for the notion that rich avi- study of the exotic and native birds in New Zealand that faunas in themselves repel the establishment of avian extinctions of native birds were not caused by competition invaders at the level of whole islands or archipelagoes, from exotics but rather from habitat destruction, hunting, interactions between established exotics and natives and introduced predators. Forest alteration by browsing may be influencing habitat distributions of species in and logging as well as a lack of competition from native both sets within islands. One possible explanation for birds may have then allowed the exotics to penetrate the striking inverse relationship between introduced into forest habitats. It was impossible for them to assess species number and relative abundance and native the relative importance of these two factors. The conclu- species number and relative abundance is that local sions of Mountainspring and Scott (1985), and Scott et competition between species in these two sets limits al. (1986) for the Hawaiian situation are not dissimilar. their distribution and abundance. Alternatively sites To distinguish between these two hypotheses to may vary in subtle habitat features at the local and at explain the differential penetrance of exotics into forest the landscape scale; exotics and natives may be track- habitats, i.e. diffuse competition between natives and ing these habitat features differentially such that sites exotics, on the one hand, and differential responses of that are favorable for exotics are unfavorable for natives and exotics to habitat and landscape alterations natives. on the other, controlled experiments are needed with Since most native birds in the Tropical Pacific are careful attention to matching habitat features at both primarily birds of closed canopy while most introduced the local and landscape scale. The two hypotheses are species are birds of more open habitats, the bird species not mutually exclusive and both may be operating to sets and their abundances might be expected to vary varying degrees in different locations. inversely across a range of habitats from open to closed. While all sites are categorized according to Native species in non-native habitats floral and vegetation characters, urban/suburban sites Why do these various locations differ so strikingly in still differ in the proportion of native vs exotic plant- the numbers of native species that colonize urban/sub- ings around houses and other buildings. Green (1984) urban habitats? In species-rich areas like North America found that the total amount of native vegetation at a or Australia, with much habitat diversity, the native site was the most influential factor governing the number avifauna contains species adapted to the variety of of both native birds (with a positive affect) and exotic available habitats. Urban/suburban habitats contain birds (with a negative affect) in suburban habitats native plant species as well as exotic introductions and around Melbourne. Similarly, native forests probably the avifauna is similarly heterogeneous. A small book differ in subtle ways by the actions of exotic mammals devoted to the common birds of urban Melbourne lists and insects. Probably even more important than local 35 species of native passerines and eight exotics that habitat differences, forest sites will vary in important the lay person could commonly observe around the city landscape-level features (like patterns of fragmentation (Reid et al., 1971). The ability of native species to occupy and edge effects) that influence their access to exotics urban/suburban habitats where they coexist with exotics and the long-term persistence of exotic and native sub- is typical for other continental avifaunas (Califomia: Vale populations within a larger metapopulation. Some for- & Vale, 1976; South Africa: Brooke et al., 1986; New- est sites are simply subsections of continuous native man, 1993; West Africa: Winterbottom, 1933; Malay forests while others are smaller forest fragments sur- penninsula: McClure, 1961; Ward, 1968; India: Beehler rounded by disturbed habitats that harbor more exotic et al., 1987) and those on very large islands (New birds (Askins et al., 1987). Such near-by reservoirs of Guinea: Bell 1986; Cuba: Pozas & Balat, 1981). exotics might spill over to these forest fragments Because of their remoteness and small size, islands (Smallwood, 1994). Similarly, small landscape patches typically have fewer native species than continents of urban/suburban habitats imbedded in relatively (MacArthur & Wilson, 1967). Smaller land masses in undisturbed forest might be expected to harbor fewer tropical areas also have less habitat diversity for two exotic and more native birds than sites located in reasons, area and maximum elevation are positivley expansive human-modified open habitat. correlated (see Table 3). Islands with taller peaks gener- In this matter, local species diversity is responding to ally have greater topographic diversity. Also tall mas- local factors in the form of competitive interactions sive mountains create strong windward vs leeward and/or habitat variation, but also to regional and his- regional differences in climate on an island. Leeward torical processes affecting the diversity of the source habitats are generally more arid and more open. Con- pool and the overall pattern of landscape alteration sequently, small islands seem to contain proportionally (Askins et al., 1987; Ricklefs & Schluter, 1993; Small- fewer bird species that are pre-adapted to the generally wood, 1994). Table 5 and Fig. 10 indicate that point open habitats associated with urbanization and agricul- (or alpha) diversity, whether in non-native habitats or ture but it is these species that are most likely colonists native forest, is responding to both regional and local of urban/suburban habitats (Bell, 1986). Bell found species variables. that only about 3.2%-4.8% (depending on the city) of 88 T.J. Case

New Guinea rainforest bird species have colonized in native and non-native habitats can be quantified urban areas of Papua New Guinea. While a much using the dissimilarity measure z of Preston (1962): greater proportion of native savannah and forest edge xl/: + yl/: = 1 (1) species also occupy urban areas, Additionally, lowland habitats, where most urban centers are concentrated, where x is the fraction of total species number that is have been particularly decimated on many islands (Per- found in native habitats and y the fraction found in netta & Watling, 1979; Olson & James, 1982; Moulton non-native habitats. If z = 1, its limiting value at one & Pimm, 1983). Consequently, compared to higher ele- end of its range, the faunas are completely dissimilar. vation forest birds, few native lowland species are left As z approaches zero, at the other extreme, there is no in lowland disturbed habitats to have a potential share dissimilarity. Because no closed form analytic solution for in lowland towns and cities with introduced species. In z exists, Preston (1962) provides a table for its solution. short, smaller, flatter, islands with greater lowland The values of z increase from Fiji to Tahiti. This habitat destruction and urbanization have fewer potential quantifies the observation that habitat sharing of native native species to colonize urban centers. and exotic birds is generally greater in areas with larger This pattern comes into sharper focus by a contrast native avifaunas that have not suffered as extensive a between Fiji, Hawaii, and the Societies. Although simi- loss of lowland habitats and their indigenous bird species. lar in overall land mass, Kauai and Tahiti have experi- Bird species numbers in each habitat type are proba- enced more extinctions and habitat destruction than bly not yet in equilibrium for their area and degree of has Viti Levu in Fiji. Today on Viti Levu twelve native habitat conversion. This latter result was also seen by species (representing 44% of the total native species of Bell (1986) who compared historical bird surveys to passerines and doves) commonly make use of disturbed present-day results in New Guinea urban areas. Birds modified habitats (Table 6). In Kauai and Tahiti native were gradually adjusting to new habitat formation and passerines and doves are essentially restricted to upland species occurrences did not appear to have reached an native forest. The overall dissimilarity of the avifaunas equilibrium.

Table 6. Habitat distributions of passerines and eolumbiformes relative to native and non-native habitats on three Pacific islands The number of species (and % of total species in the row heading) is given for each cell. The value of = is Preston's measure of the dissimilarity of two faunas, here applied to birds of native and non-native habitats (see text).

(a) Fiji -- Viti Levu (from Watling (1982) and personal surveys)

Extant native species = 27 Introd. species = 8

Species in only Species in native and Species only in native habitats non-native habitats non-native habitats

Native species 14/52% 12/44% 1/4% Exotic species 0/0% 3/43% 5/57%

z = 0.44.

(b) Hawaii -- Kauai (from Scott et al. (1986); Richardson & Bowles (1964) and personal surveys)

Extant native species = 13 Introd. species = 17

Species in only Species in native and Species only in native habitats non-native habitats non-native habitats

Native species 13t100% 010% 0t0% Exotic species 0/0% 9•53% 8/47%

= 0.612.

(c) Societies -- Tahiti (from Thibault & Rives (1975) and personal surveys)

Extant native species = 5 Introd. species = 9

Species in only Species in native and Species only in native habitats non-native habitats non-native habitats

Native species 4/80% 1/20% 0/0% Exotic species 0/0% 2/22% 7/78%

z= 0.72. Introductions and ext&ctions of birds 89

Is there an asymmetry in the interactions between on islands are generally staying similar through gains exotics and natives? If present-day interspecific compe- of exotics balancing losses of natives, worldwide the tition is contributing to a redistribution of natives and total number of bird species has declined and avifaunas exotic species across habitats, then the results of Table are losing and continue to lose their regional distinctions. 5 suggest that exotics may be hindering native expan- In local surveys the relative abundance of exotic sion into non-native habitats more than natives are birds compared to native birds is affected by habitat preventing exotics from penetrating native habitats. (non-native habitats have more exotics) and also by the The negative associations between natives and exotics numbers of species of exotics and natives on the island. are stronger in non-native habitats compared to native The island (or regional) species pool of exotic and forest, but sample sizes are also larger giving more native species, along with the island's overall degree of power to the comparision. Moreover, this conclusion is habitat alteration, affects the species diversity at the not supported by the results from New Zealand/ local scale and, in turn, the relative abundance of Australian surveys where the abundance of native birds exotic birds and native birds in native and non-native has been more rigorously measured. Here native bird habitats. The relative importance of biotic interactions abundance is not significantly affected by either the (like competition, apparent competition through differ- number of exotic species or their summed numerical ential disease transmission or susceptibility, and preda- density in either native or non-native habitats. Yet, the tion) in shaping the abundance and habitat affinities of abundance and diversity of exotics in Australia and exotics and native species, can be difficult to unravel New Zealand declines sharply with the number of when regional affects are so important. Future studies native bird species, at least in non-native habitats will gain more mechanistic insights by focusing on a (Fig. 12(b)). This last effect disappears in a stepwise few exotic and native species and comparing their habi- regression, since native species number is not a signifi- tat distributions across a number of different archipela- cant variable accounting for variation in exotic abun- goes, similar in climate and habitat but differing in dance or diversity once the local number of exotic native species composition and the level of habitat species is entered into the equation. In sum these destruction and conversion. results suggest that habitat variation and the regional source pools of exotics and natives provide additional ACKNOWLEDGEMENTS explanatory power in accounting for local diversity and abundance patterns. Still caution is urged; more This work was supported by the National Science detailed species-specific studies are needed that compared Foundation (grant DEB-9220621 and BSR-9107739). I the same exotic species in places that vary in their num- thank Mike Gilpin, Mike Moulton, Stuart Pimm, bers and types of competitors and predators. Trevor Price and David Steadman for useful insights.

CONCLUSIONS AND FUTURE DIRECTIONS REFERENCES All in all, about half the variance between islands in Adler, G. H. (1992). Endemism in birds of tropical Pacific their numbers of successfully and unsuccessfully intro- islands. Evolut. Ecol., 6, 29f~306. duced species can be accounted for by recipient site- Ashmole, N. P. (1963). The extinct avifauna of St. Helena specific variables; the most important correlate of Island. Ibis', 1036, 390-408. success is the number of native species extinctions, Atkinson, I. A. E. (1985). The spread of commensal species of Rattus to oceanic islands and their effects on island avi- which reflects the degree of human activity and habitat faunas. In Conservation of island birds: case studies for destruction and deterioration through intrusions of the management of threatened island species, ed. P. J. exotic predators, herbivores, and parasites. Presumably Moors. International Council for Bird Preservation, Cam- some of the remaining variation can be explained by bridge, pp. 35-81. individual differences between the sets of species intro- Atkinson, I. A. E. (1989). Introduced animals and extinc- tions. In Conservation for the twenty-first century, ed. D. duced from place to place, idiosyncrasies in the individ- Western & M. Pearl. Oxford University Press, Oxford, pp. ual methods of introduction for different species, other 54-75. site specific factors not included in this analysis, and Bacallado, J. J. (1976). Notas sobre la distribucion y evolu- truly stochastic factors affecting the establishment and cion de la avifauna Canaria. In Biogeography and ecology persistence of exotic species. All of these factors have in the Canary Islands, ed. G. Kunkel. W. Junk, The Hague, pp. 413-32. been shown to affect introduction success in birds and Baez, M. (1992). Zoogeography and evolution of the avi- other vertebrates (Hengeveld, 1989; Pimm, 1991). fauna of the Canary Islands. In Papers in avian paleontol- Many of the same species that have been introduced ogy, ed. K. E. Campbell. Natural History Museum of Los successfully to one place have also been successfully Angeles, Los Angeles, CA, pp. 425-31. introduced to others around the globe (Simberloff & Balmford, R. (1981). Early introductions of birds to Victoria. Vict. Nat., 98, 96-105. Boecklen, 1991), while the losses of island birds from Balouet, J. C. & Olson, S. L. (1989). Fossil birds from Late extirpation seem to have disproportionately fallen on Quaternary deposits in New Caledonia. Smithson. Contrib. endemic species. Consequently, while species numbers ZooL, 469, 1-38. 90 T. J. Case

Banks, R. C. (1976). Wildlife importation into the United species in shaping the abundance and distribution of island States, 1900-1972. USDL Fish. Wildl. Serv., Spec. Scient. reptiles. Evolut. Ecol., 5, 272-90. Rep., Wildl., No. 200. Case, T. J. & Cody, M. L. (1987). Island biogeographic theo- Banks, R. C. (1981). Summary of foreign game bird libera- ries: tests on islands in the Sea of Cortez. Amer. Sci., 75, tions, 1969-1978. USDL Bur. Sport Fish. Wildl. Spec. Sci- 402-11. ent. Rep., Wildl., No. 239. Case, T. J., Bolger, D, T. and Richman, A. (1992). Reptilian Bannerman, D. A. (1963). Birds of the Atlantic Islands. Oliver extinctions: The last ten thousand years. In Conservation & Boyd, London. biology, ed. P. L. & S. K. J. Fiedler. Chapman & Hall, Bannerman, D. A. & Bannerman, W. M. (1983). The birds" of New York, pp. 91-125. the Balearic. Croom Helm, London. Cassels, R. (1984). The role of prehistoric man in the faunal Beehler, B. M., Krishna Raju, K. S. R. & Ali, S. (1987). extinctions of New Zealand and other Pacific islands. In Avian use of man-disturbed forest habitats in the Eastern Quaternary extinctions, ed. P. S. Martin & R. G. Klein. Ghats, India. Ibis, 129, 197-211. University of Arizona Press, Tucson, AZ, pp. 708-67. Beehler, B. M., Pratt, T. K. & Zimmerman, D. A. (1986). Caum, E. L. (1933). The exotic birds of Hawaii. Bernice P. Birds of New Guinea. Princeton University Press, Princeton, Bishop Mus. Occ. Pap., 10, 1 55. NJ. Cawkell, E. M. & Hamilton, J. F, (1961). The birds of the Bell, H. L. (1986). Occupation of urban habitats by birds in Falkland Islands. Ibis, 103, 1-27. New Guinea. Proc. West. Foundn Vert. Zool., 3, 1-48. Cheke, A. S. (1987). An ecological history of the Mascarene Benson, C. W. (1950). A contribution to the ornithology of Islands with particular reference to extinctions and intro- St Helena and other notes from a sea voyage. Ibis, 92, 75-83. ductions of land vertebrates. In Studies of Mascarene Island Benson, C. W. (1960). The birds of the Comoro Islands: birds, ed. A. W. Diamond & A. S. Cheke. Cambridge Uni- result of the British Ornithologists' Union expedition of versity Press, Cambridge, pp. 5 89. 1958. Ibis, 103h, 5-106. Clunie, F. (1984). Birds' of the Fiji bush. Fiji Museum, Suva, Benson, C. W. & Penny, M. J. (1971). The land birds of Fiji. Aldabra. Phil Trans. R. Soc., Lond., 260, 417-527. Cody, M. L. (1983). The land birds. In lslandbiogeography in Berger, A. J. (1972). Hawaiian birdlife. University of Hawaii the Sea of Cortez, ed. T. J. Case & M. L. Cody. University Press, Honolulu. of California Press, Berkeley, CA, pp. 210-45. Blabar, S. J. M. (1990). A checklist and notes on the current Crowell, K. L. (1962). Reduced competition among the birds status of the birds of New Georgia, West Province, of Bermuda. Ecology, 43, 75-88. Solomon Islands. Emu, 90, 205-14. Crusz, H. (1984). Parasites of the endemic and relict verte- Bohl, W. H. & Bump, G. (1970) Summary of foreign game brates: a biogeographical review. In Ecology and biogeogra- bird liberations 1960--1968. USDI Fish. WildL Serv., Spec. phy in Sri Lanka, ed. C. H. Fernando. W. Junk, The Scient. Rep., WildL, No. 130. Hague, pp. 321-52. Bolger, D. T., Alberts, A. C. & Soulr, M. E. (1991). Occur- Curtiss, A. (1938). A short zoology of Tahiti. Guide Printing rence patterns of bird species in habitat fragment: sam- Co., Brooklyn, N.Y. piing, extinction, and nested species subsets. Amer. Nat., Delacour, J. (1966). Guide des oiseaux de la Nouvelle-CalO- 137, 155. donie et de ses dependances. Delachaux and Niestle, Bond, J. (1980). Birds of the West Indies. Houghton Mifflin, Neuchatel. Boston, MA. Diamond, A. W. (1984). Biogeography of Seychelles land birds. Bostic, D. L. (1975). A natural history guide to the Pacific of In Biogeography and ecology of the Seychelles Islands, ed. North Central Baja California and adjacent islands. Biology D. R. Stoddard. W. Junk, The Hague, pp. 487-504. Education Expeditions, Vista, CA. Diamond, A. W. (1985). The conservation of landbirds on Bourne, W. R. P. (1957). Breeding birds of Bermuda. Ibis, 99, islands in the tropical Indian Ocean. ICBP Tech. Publs., 3, 94-105. 85-100. Bourne, W. R. P. (1966). Further notes on the birds of the Diamond, J. (1985). Rats as agents of extermination. Nature, Cape Verde Islands. Ibis, 108, 425-9. Lond., 318, 602-3. Bregulla, H. L. (1992). Birds of Vanuatu. Nelson, Oswestry. Diamond, J. (1991). A new species of rail from the Solomon Brattstrom, B. H. (1990). Biogeography of the Islas-Revillagi Islands and convergent evolution of insular flightlessness. Mexico. J. Biogeogr., 17, 177-83. Auk, 108, 461-70. Brooke, R. K., Lloyd, P. H. & de Villiers, A. L. (1986). Alien Diamond, J. M. (1984). Historic extinctions: a rosetta stone translocated terrestrial vertebrates in South Africa. In The for understanding prehistoric extinctions. In Quaternary ecology and management of biological invasions in southern extinctions, ed. P. S. Martin & R. G. Klein. University of Africa, ed. I. A. W. Macdonald, F. J. Kruger & A. A. Ferrar. Arizona Press, Tucson, AZ, pp. 824-62. Oxford University Press, Cape Town, pp. 63-74. Diamond, J. M. & Jones, H. L. (1980). Breeding land birds Bruner, P. L. (1972). Field guide to the birds" of French Polyne- of the Channel Islands. In The California islands, ed. D. M. sia. Bernice P. Bishop Museum, Honolulu. Power. Santa Barbara Museum of Natural History, Santa Bullock, D. (1977). Round Island--a tale of destruction. Barbara, CA. Oryx, 14, 51-8 Diamond, J. M. & Marshall, A. G. (1976). Origin of the New Bump, G. (1963). History and analysis of tetraonid introduc- Hebridean avifauna. Emu, 76, 187-200. tions into North America. J. Wildl. Manage., 27, 855-67. Diamond, J. M. & Veitch, C. R. (1981). Extinctions and Cain, A. J. & Galbraith, I. C. J. (1956). Field notes on birds introductions in the New Zealand avifauna: cause and of the eastern Solomon islands. Ibis, 99, 262-95. effect. Science, N. Y., 211,499-501. Carr, G. S. (1980). Historic and prehistoric avian records Disney, D. H. J. & Stokes, A. (1976). Birds in pine and native from Easter island. Pacif. Sci., 34, 19. forests. Emu, 76, 133-8. Case, T. J. (1991). Invasion resistance, species build-up and Drake, J. A. (1983). Invasability in Lotka-Volterra interac- community collapse in model competition communities. In tion webs. In Current trends in food web theory report on a Metapopulation dynamics, ed. M. E. Gilpin & I. Hanski, food web workshop, ed. D. L. DeAngelis, W. M. Post & pp. 239-66. G. Sugihara. National Technical Information Service, Oak Case, T. J. & Bolger, D. T. (1991). The role of introduced Ridge, TN, pp. 83-90. Introductions and extinctions of birds 91

Drake, J. A. (1988). Models of community assembly and the Greenway, J. L. (1967). Extinct and vanishing birds of the structure of ecological landscapes. In Proceedings of the world. Dover Press, New York. International Conference in Mathematical Ecology, ed. T. Guild, E. (1938). Tahitian aviculture: acclimatization of Hallam, L. Gross & S. Levin. World Press, Trieste. foreign birds. Avic. Mag., 5, 8-11. Drake, J. A. (1991). Community assembly mechanics and the Guild, E. (1940). Western bluebirds in Tahiti. Avic. Mag., 5, structure of an experimental species ensemble. Amer. Nat., 284-5. 137, 1-26. Hadden, D. (1981). Birds of the North Solomons. Wau Ecol- Drake~ J. A., Flum, T. E., Witteman, G. J., Voskuil, T., ogy Institute, Wau, Papua New Guinea. Hoylman, A. M., Creson, C., Kenny, D. A., Huxel, G. R., Hannecart, F. & Letocart, Y. (1983). Oiseaux de Nouvelle Larue, C. S. & Duncan, J. R. (1993). The construction and Caledonie et des Loyautes. assembly of an ecological landscape. J. Animal EcoL, 62, Harris, M. P. (1973). The Galapagos avifauna. Condor, 75, 117-30. 265-78. Druett, J. (1983). Exotic intruders. The introduction of plants Hart-Davis, D. (1972). Ascension--the story of a South and animals into New Zealand. Heinemann, Auckland. Atlantic island. Constable, London. du Pont, J. E. (1976). South Pacific birds. Delaware Mus. Harting, J. E. (1873). Rare or little known Limicolae. Ibis, 3, Nat. Hist. Monogr., Ser. 3. 260-9. Duffey, E. (1964). The terrestrial ecology of Ascension Island. Haydock, E. I. (1954). A survey of the birds of St Helena J. Appl. EcoL, 1, 219-51. island. Ostrich, 25, 62-75. Ebenhard, T. (1988). Introduced birds and mammals and Heilprin, A. (1889). The Bermuda Islands. Academy of Natu- their ecological effects. Swed. Wildl. Res. Viltrevy, 13, ral Sciences, Philadelphia. 1-107. Heinzel, H., Fitter, R. & Parslow, J. (1972). The birds of Edgar, A. T. (1965). The Kermadecs expedition, 15-17 Britain and Europe. Collins, London. November 1964. Notornis, 12, 1-43. Henderson, R. W. (1992). Consequences of predator intro- Elliott, H. F. (1957). A contribution to the ornithology of the ductions and habitat destruction on amphibians and reptiles Tristan da Cunha group. Ibis, 99, 545-86. in the post-Columbus West Indies. Caribb. J. Sci., 28, 1-10. Elton, C. S. (1958). The ecology of invasions by animals and Hengeveld, R. (1989). Dynamics of biological invasions. Chap- plants. Chapman and Hall, London. man & Hall, London. Engbring, J. (1988). FieM guide to the birds of Palau, Conser- Henry, G, M. (1955). A guide to the birds of Ceylon. 2nd edn. vation office Palau, Koror, Palau. Oxford University Press, London. Engbring, J. & Fritts, T. H. (1988). Demise of an insular avi- Hertlein, L. G. (1963). Contribution to the biogeography of fauna: the brown tree snake on Guam. Trans. West. Sect. Cocos Island including a bibliography. Proc. Calif. Acad. Wildl. Sot., 24, 31-7. Sci., 32, 219-89. Falla, R. A., Sibson, R. B. & Turbott, E. G. (1983). The new Hindwood, K. A. (1940). The birds of Lord Howe Island. guide to the birds of New Zealand and outlying islands. Emu, 40, 1-86. Collins, London. Holdaway, R. N. (1990). Changes in the diversity of New Ffrench, R. (1980). A guide to the birds of Trinidad and Zealand forest birds. N.Z.J. Zool., 17, 309-21. Tobago. Harrowood Books, Newtown Square, PA. Holdaway, R. N. (1989). New Zealand's pre-human avifauna Fleming, C. A. (1939). Birds of the Chatham Islands. Emu, and its vulnerability. N.Z.J. Ecol., 12, 11-25. 38, 380~13, 492-509. Holdgate, M. W. (1960). The fauna of the mid-Atlantic Forbes, H. O. (1893). A list of the birds inhabiting the islands--a discussion on the biology of the southern cold Chatham Islands. Ibis, 1893, 521-46. temperate zone. Proc. R. Soc. Lond., Ser. B, 152, 550~7. Forbes-Watson, A. D. (1969). Notes on bireds observed in Holloway, J. D. (1977). The lepidoptera of Norfold Island, the Comoros on behalf of the Smithsonian Institution. their biogeography and ecology. W. Junk, The Hague. Atoll Res. Bull., 128, 1-23. Holmes, D. A. & Nash, S. V. (1989). The birds of Java and Ford, H. (1979). Birds. In Natural history of Kangaroo Island, Bali. Oxford University Press, Oxford. ed. M. J. Tyler, C. R. Twidale & J. R. Ring. Royal Society Holyoak, D. T. (1974). Undescribed land birds from the of South Australia, Adelaide, pp. 103-14. Cook Islands. Pacif Ocean. Bull. Brit. Ornithol. Club, 94, Friend, G. R. (1982). Bird populations in exotic pine planta- 145-50. tions and indigenous eucalypt forests in Gippsland, Victoria. Horton, D. R. (1984). Red kangaroos: last of the Australian Emu, 82, 80-91. megafauna. In Quaternary extinctions, ed. P. S. Martin & Gibb, J. A. (1961). Ecology of the birds of Kaingaroa forest. R. G. Klein. University of Arizona Press, Tucson, AZ, pp. Proc. N.Z. Ecol. Soc., 8, 29-38. 639-80. Gill, F. B. (1967). Birds of Rodriquez Island. Ibis, 109, Hummelinck, P. W. (1940). Studies on the fauna of Curacao, 383-90. Aruba, Bonaire, and the Venezuelan Islands, Vol. 1. den Boer, Goodman, D. (1987). The demography of chance extinction. Utrecht. In Viable populations for conservation, ed. M. E. Soulr. Jehl, J. R. & Everett, W. T. (1985). History and status of the Cambridge University Press, Cambridge, New York, pp. avifauna of Isla Guadalupe, Mexico. Trans. San Diego Soe. ! 1-34. Nat. Hist., 20, 313-36. Goodman, S. M. (1994). Description of a new species of Jehl, J. R. & Parkes, K. C. (1982). The status of the avifauna subfossil eagle from Madagascar: Stephanoaetus (Aves: of the Revillagigedo Islands, Mexico. Wilson Bull., 94, Falconiformes) from the deposits of Ampasambazimba. 1-19. Proc. Biol. Soc. Wash., 107, 421-8. Jenkins, J. M. (1983). The native forest birds of Guam. Gorman, M. L. (1975). Habitats of the land birds of Viti Ornithol. Monogr., 31, 1~1. Levu, Fiji Islands. Ibis, 117, 152-61. Johnson, S. W., Millie, W. R. & Moffet, G. (1970). Notes on Grant, P. R. & Cowan, I. M. (1964). A review of the avi- the birds of Easter Island. Ibis, 112, 532-8. fauna of the Tres Marias Islands, Nayarit, Mexico. Condor, Johnson, T. H. & Stattersfield, A. J. (1990). A global review 66, 221-8. of island endemic birds. Ibis, 132, 167-80. Green, R. J. (1984). Native and exotic birds in a suburban Jolly, A., Oberlr, P. & Albignac, R. (1984). Madagascar (key habitat. Aust. Wildl. Res., 11, 181-90. environments). Pergamon, Oxford. 92 T. J. Case

Jones, H. L. & Diamond, J. M. (1976). Short-time-base stud- Milberg, P. & Tyrberg, T. (1993). Na'fve birds and noble ies of turnover in breeding bird populations on the Califor- savages--a review of man-caused prehistoric extinctions of nia Channel Islands. Condor, 78, 526-49. island birds. Ecography, 16, 22%50. King, C. (1984). Immigrant killers: introduced predators and Mittermeier, R. A. (1972). Jamaica's endangered species. the conservation of birds in New Zealand. Oxford University Oryx, 11, 258-62. Press, Auckland. Moreau, R. E. (1966). The bird faunas of Africa and its King, W. B. (1981). Endangered birds of the world." ICBP bird islands. Academic Press, New York. red data book. Smithsonian Institution Press, Washington, Moreau, R. E. & Pakenham, R. H. W. (1941). The land verte- DC. brates of Pemba, Zanzibar, and Mafia: a zoogeographical Knox, G. A. (1969). The New Zealand off-shore islands. study. Proc. Zool. Soc. Lond. Ser. A, 110, 97-128. Micronesia, 5, 465-91. Moulton, M. P. (1993). The all-or-none pattern in introduced Kramer, P. (1984). Man and other introduced animals. Hawaiian passeriformes: the role of competition sustained. Biol. J. Linn. Soc., 21,253-8. Amer. Nat., 141, 105-19. Lack, D. (1976). Island biology. University of California Moulton, M. P. & Pimm, S. L. (1983). The introduced Press, Berkeley, CA. Hawaiian avifauna: biogeographic evidence for competi- Langrand, O. (1990). Guide to the birds of Madagascar. Yale tion. Amer. Nat., 121,669 90. University Press, New Haven, CT. Mountainspring, S. & Scott, J. M. (1985). Interspecific com- Leigh, E. G. (1981). Average lifetime of a population in a petition among Hawaiian forest birds. Ecol. Monogr., 55, varying environment. J. Theoret. Biol., 90, 213-39. 219-39. Lever, C. (1977). The naturalized mammals' of the British Isles. Munro, G. G. (1944). Birds' of Hawaii. Tongg, Honolulu. Hutchinson, London. Nelson, E. W. (1889). Natural biology of the Tres Marias Lever, C. (1985). Naturalized mammals of the world. Long- Islands: general account of the islands with reports on man, London. mammals and birds. N. Amer. Fauna, 14, 7-62. Lever, C. (1987). Naturalized birds' of the world. Longman, Newman, K. (1993). Birds of South Africa, 4th edn. Southern London. Book Publishers, Johannesburg. Lionnet, G. (1984). Extinct birds of the Seychelles. In Bio- Newsome, A. E. & Noble, I. R. (1986). Ecological and physi- geography and ecology of the Seychelles Islands', ed. D. R. ological characters of invading species. In Ecology c?] Stoddart. W. Junk, The Hague, pp. 505-11. biological invasions, ed. R. H. Groves & J. J. Burton. Cam- Lobin, W. (ed.) (1984). Fauna und Flora der Kapverdischen bridge University Press, Cambridge, pp. 1-20. Inseln. Senckenbergische Naturforschende Gesselschaft, North, S. G., Bullock, D. J. & Dulloo, M. E. (1994). Changes Frankfurt. in the vegetation and reptile populations on Round Island, Lobin, W. (1986). Beitrag zur Fauna und Flora der Kapverdischen Mauritius, following eradication of rabbits. Biol. Conserv., Inseln. Senckenbergische Naturforschende Gesselschaft, 67,21 8. Frankfurt. Olson, S. L. (1975). Paleoornithology of St. Helena Island, Lockwood, J. L. & Moulton, M. P. (1994). Ecomorphological South Atlantic Ocean. Smithson. Contrib. Paleobiol., 23, 49. pattern in Bermuda birds: the influence of competition and Olson, S. L. (1984). Fossil birds of the islands of the world. implications for nature preserves. Evolut. Ecol., 8, 53-60. Proc. lOlst Amer. Ornithol. Soc. Mtg. Lockwood, J. L., Moulton, M. P. & Anderson, S. K. (1993). Olson, S. L. (1989). Extinction on islands: man as a catastrophe. Morphological assortment and the assembly of communi- In Conservation for the twenty-first century, ed. D. Western ties of introduced passerines on oceanic islands: Tahiti vs & M. Pearl. Oxford University Press, New York, pp. 50-3. Oahu. Amer. Nat., 141, 398-408. Olson, S. L. & Hilgartner, W. B. (1982). Fossil and subfossil Long, J. (1981). Introduced birds of the world. David & birds from the Bahamas. Smithson. Contrib. Paleobiol,, 48, Charles, London. 22-60. Losos, J. B., Marks, J. C. & Schoener, T. W. (1993). Habitat Olson, S. L. & James, H. F. (1984). The role of Polynesians use and ecological interactions of an introduced and a in the extinction of the avifauna of the Hawaiian islands. native species of Anolis lizard on Grand Cayman, with a In Quaternary extinctions, ed. P. S. Martin & R. G. Klein. review of the outcomes of anole introductions. Oecologia, University of Arizona Press, Tucson, AZ, pp. 768 80. Berl., 95, 525-32. Olson, S. L. & James, H. G. (1982). Prodromus of the fossil Loustau-Lalanne, P. (1962). Birds of the Chagos archipelago, avifauna of the Hawaiian Islands. Smithson. Contrib. Zool., Indian Ocean. Ibis, 104, 67-73. 365, 1-59. MacArthur, R. H. & Wilson, E. O. (1967). The theory of Olson, S. L., Pregill, G. K. & Hilgartner, W. B. (1990). Stud- island biogeography. Princeton University Press, Princeton, NJ. ies on fossil and extant vertebrates from San Salvador Marsh, L. & Pope, E. (1967). Norfolk and Lord Howe (Watling's) Island, Bahamas. Smithson. Contrib. Zool., 508, Islands. Aust. Nat. Hist., 15, 392-400. 1-15. Mayr, E. (1965). The nature of colonizations in birds. In The Owen, R. P. (1977a). A checklist of the birds of Micronesia. genetics of colonizing species, ed. H. G. Baker & G. L. Micronesia, 13, 65-81. Stebbins. Academic Press, New York, pp. 30-45. Owen, R. P. (1977b). Terrestrial vertebrate fauna of the Palau McClure, H. E. (1961). Garden birds in Kuala Lumpur, Islands. Report from the Office of the Chief Conservation- Malaya. Malay Nat. J., 15, 111-13. ist, Trust Terr. Pacif., pp. 1-15. McDowall, R. M. (1969). Extinction and endemism in New Paynter, R. A. (1955). The ornitho-geography of the Yucatan Zealand birds. Tuatara, 17, 1-12. Peninsula. Bull. Yale Peabody Mus. Nat. Hist., 9, 1 347. McKean, J. L. & Hindwood, K. A. (1965). Additional notes Peake, J. F. (1971). The evolution of terrestrial faunas in the of the birds of Lord Howe Island. Emu, 64, 7%97. western Indian Ocean. Phil, Trans. R. Soc. Lond., B, 260, McLay, C. L. (1974). The species diversity of New Zealand 581-610. forest birds: some possible consequences of the modifica- Pearson, D. L. (1977). A pantropical comparison of bird tion of beech forests. N.Z.J. Zool., 1, 179-96. community structure on six lowland forest sites. Condor, Medway, L. & Marshall, A. G. (1975). Terrestrial vertebrates 79, 232M4. of the New Hebrides: origin and distribution. Trans. R.. Penny, M. (1974). The birds of the Seychelles and the outlying Soc. Lond. B, 272, 423-65. islands. Collins, London. Introductions and extinctions of birds 93

Pernetta, J. C. & Watling, D. (1979). The introduced and Rejm~nek, M. (1989). Invasibility of plant communities. In native terrestrial vertebrates of Fiji Pacific. Science, N.Y., Biological invasions." a global perspective, ed. J. A. Drake, 32, 223~14. H. A. Mooney, F. di Castri, R. H. Groves, F. J. Kruger, Petren, K. and Case, T. J. (1996). An experimental demon- M. Rejm~inek & M. Williamson. John Wiley, New York, stration of exploitation competition in an ongoing inva- pp. 369-88. sion. Ecology (in press). Reynolds, R. T., Scott, J. M. & Nussbaum, R. A.. (1980). A Philibosian, R. & Yntema, J. A. (1977). Annotated checklist variable circular-plot method for estimating bird numbers. of the birds', mammals, reptiles and amphibians of the Virgin Condor, 82, 309-13. Islands and Puerto Rico. Information Services, St. Croix, Richardson, F. & Bowles, J. (1964). A survey of the birds of US Virgin Islands. Kauai, Hawaii. Bernice P. Bishop Mus. Bull., 227, 1-51. Phillips, J. C. (1928). Wild birds introduced or transplanted Richman, A. D., Case, T. J. & Schwaner, T. D. (1988). Natu- in North America. US Dep. Agric. Tech. Bull., 61, 63. ral and unnatural extinction rates of reptiles on islands. Phillips, W. W. A. (1953). A revised checklis't of the birds of Amer. Nat., 131,611 30. Ceylon. Natural History Museum of Ceylon. Ricklefs, R. E. & Schluter, D. (1993). Species diversity: Pickering, R. (1981a). An annotated checklist of the birds of regional and historical influences. In Species diversity in Vanuatu. Misc. Pubis" Vanuatu Nat. Sci. Soc., 1, 1 8. ecological communities." historical and geographical perspec- Pickering, R. (1981b). The birds of Vanuatu, 1. The intro- tives, ed. R. E. Ricklefs & D. Schluter. University of duced species. Naika, 2, 10-12. Chicago Press, Chicago, IL, pp. 350-63. Pimm, S. L. (1991). The balance of nature? Ecological issues in Ridpath, M. G. & Moreau, R. E. (1966). The birds of Tas- the conservation of species and communities. University of mania: ecology and evolution. Ibis, 108, 348-93. Chicago Press, Chicago, IL. Ripley, S. D. & Beehler, B. M. (1989). Ornithogeographic Pimm, S. L., Moulton, M. P. & Justice, L. J. (1994). Bird affinities of the Andaman and Nicobar Islands. J. Bio- extinctions in the central Pacific. Phil. Trans. R. Soc. Lond., geogr., 16, 323-32. B, 344, 27-33. Robertson, C. J. R. (1985). Complete book of New Zealand Pittingill, O. S. (1973). Passerine birds of the Falkland birds'. Readers Digest Services, Sydney. Islands: their behavior and ecology. Living Bird, 12, Robinson, J. V. & Dickerson Jr., J. E. (1984). Testing the 95 136. invulnerability of laboratory island communities to inva- Pizzey, G. & Doyle, R. (1980). A field guide to the birds of sion. Oecologia, Berl., 61, 169 74. Australia. Princeton University Press, Princeton, NJ. Rolls, E. C. (1984). They all ran wild. Angus & Robertson, Post, W. M. & Pimm, S. L. (1983). Community assembly and London. food web stability. Math. Biosci., 64, 169 92. Ryan, C. S. (1906). President's address: on European and Power, D. M. (1972). Numbers of bird species on the Califor- other birds liberated in Victoria. Emu, 5, 110-19. nia Islands. Evolution, 26, 451-63. Saunders, D. A. & Rebeira, D. C. P. (1983). The birds of Pozas, G. & Balat, F. (1981). Quantitative and qualitative Rottnest Island. J. R. Soc. West. Aust., 66, 47-52. composition of the bird community in a garden suburb of Schodde, R., Fullagar, P. & Hermes, N. (1985). A review of Havana (Cuba). Folia Zool., 30, 155-66. Norfolk Island birds: past and present. Aust. Natn. Parks Pratt, H. D., Bruner, P. L. & Berrett, D. G. (1987). The birds" and Wildl. Serv., Spec. Publs., No. 8. of Hawaii and the tropical Pacific. Princeton University Schwartz, C. W. & Schwartz, E. R. (1949). The game birds in Press, Princeton, NJ. Hawaii. Board of Commissioners of Agriculture and Pregill, G. K. (1981). An appraisal of the vicariance hypothesis Forestry, Honolulu. of Caribbean biogeography and its application to West Scott, J. M., Mountainspring, S., Ramsey, F. L. & Kepler, Indian terrestial vertebrates. Syst. Zool., 30, 147-55. C. B. (1986). Forest bird communities of the Hawaiian Pregill, G. K. (1982). Fossil amphibians and reptiles from New Islands'." their dynamics, ecology, and conservation. Cooper Providence Island, Bahamas. Smithson. Contrib. Paleobiol., Ornithological Society, Las Cruces, New Mexico. 48, 8-21. Scott, S. L. (1987). Field guide to the birds of North America, Pregill, G. D., Crombie, R. I., Steadman, D. W., Gordon, 2nd edn. National Geographic Society, Washington, DC. L. K., Davis, F. W. & Hilgartner, W. B. (1991). Living and Severinghaus, S. R. & Blackshaw, K. T. (1976). A new guide late Holocene fossil vertebrates, and the vegetation of the to the birds" of Taiwan. Mei Ya, Taipei. Cockpit Country, Jamaica. Atoll. Res. Bull., 353, 1-19. Sibley, C. G. & Monroe, B. L. (1990). Distribution and taxon- Preston, F. W. (1962). The canonical distribution of com- omy of birds of the world. Yale University Press, New monness and rarity, Part I. Ecology, 43, 410-32. Haven, CT. Raffaele, H. A. (1983). A guide to the birds of Puerto Rico and Simberloff, D. (1981). Community effects of introduced the Virgin Islands. Fondo Educativo Interamicano, San species. In Biotic crises in ecological and evolutionary Juan, Puerto Rico. time, ed. M. H. Nitecki. Academic Press, New York, pp. Raffaele, H. A. (1989). The ecology of native and introduced 53-81. granivorous birds in Puerto Rico. In Biogeography of the Simberloff, D. (1990). Community effects of biological intro- West Indies, ed. C. A. Woods. Sandhill Crane Press, ductions and their implications for restoration. In Eco- Gainesville, FL, pp. 541-66. logical restoration of New Zealand islands, ed. D. R. Rand, A. L. (1936). The distribution and habits of Madagas- Towns, C. H. Daugherty & I. A. E. Atkinson. Department car birds. Bull. Amer. Mus. Nat. Hist., 72, 143499. of Conservation, Wellington, pp. 128-36. Rand, A. L. & Gilliard, E. T. (1968). Handbook of New Simberloff, D. (1992). Extinction, survival, and effects of birds Guinea birds'. Natural History Press, Garden City, New introduced to the Mascarenes. Acta Oecologica, 13, 663-78. York. Simberloff, D. & Boecklen, W. (1991). Patterns of extinction Recher, H. F. (1977). Colonization and extinction: the birds in the introduced Hawaiian avifauna: a reexamination of of Lord Howe Island. In Lord Howe Island, ed. N. Smith, the role of competition. Amer. Nat., 138, 300-27. Australian Museum, Sydney. Skottsberg, C. (1956). The natural history of Juan Fernandez Reid, A. J., Shaw, N. J. & Wheeler, W. R. (1971). Birds of and Eastern Island. Almqvist & Wicksells, Uppsala. Victoria urban areas, 2nd edn. Gould League of Victoria, Slud, P. (1967). The birds of Cocos Island (Costa Rica). Bull. Melbourne. Amer. Mus. Nat. Hist., 134, 261-96. 94 T. J. Case

Smallwood, K. S. (1994). Site invasibility by exotic birds and Temple, S. (1981). Applied island biogeography and the con- mammals. Biol. Conserv., 69, 251-9. servation of endangered island birds in the Indian Ocean. Smith, K. D. (1965). On the birds of Morocco. Ibis, 107, Biol. Conserv., 20, 147-61. 493-526. Thibault, J. C. & Rives, C. (1975). Oiseaux de Tahiti. Editions Smythies, B. (1968). Birds of Borneo, 2nd edn. Oliver & Boyd, du Pacifique, Papeete, Tahiti. London. Thomson, G. M. (1922). The naturalisation of animals and plants Soper, M. F. (1968). The Kermadec Islands. Animals, 11, in New Zealand. Cambridge University Press., Cambridge. 112-19. Tikader, B. K. & Das, A. K. (1985). Glimpses of animal life of Soul6, M. E., Bolger, D. T., Alberts, A. E., Wright, J. H., Andaman and Nicobar Islands. Zoological Survey of India, Sorice, M. & Hill, S. (1988). Reconstructed dynamics of Calcutta. rapid extinctions of chaparral-requiring birds in urban Turbott, E. G. (1961). The interactions of native and intro- habitat islands. Conserv. Biol., 2, 75-92. duced birds in New Zealand. Proc. N.Z. Ecol. Soc., 8, Stager, K. E. (1957). The avifauna of the Tres Marias 62-6. Islands, Mexico. Auk, 74, 413-32. Turbott, E. G. (1963). Three Kings Islands, New Zealand. A Steadman, D. W. (1986). Holocene vertebrate fossils from study in modification and regeneration. In Pacific Basin Isla Floreana, Galapagos. Smithson. Contrib. Zool., No. 413. biogeography, ed. J. L. Gressitt. Bishop Museum Press, Steadman, D. W. (1988). A new species of Porphyrio (Aves: Honolulu, pp. 485-598. Rallidae) from archaeological sites in the Marquesas Turbott, E. G. & Buddle, G. A. (1948). Birds of the Three Islands. Proc. Biol. Soc. Wash., 101, 162-70. Kings Islands. Rec. Auckland Inst. Mus., 3, 319-36. Steadman, D. W. (1989a). Extinction of birds in Eastern Turner, J. S.,Smithers, C. N. & Hoogland, R. D. (1968). The Polynesia: a review of the record and comparisons with conservation of Norfolk Island. Aust. Conserv. Foundn. other Pacific island groups. J. Archaeol. Sci., 16, 177-205. Spec. Publs., No. 1. Steadman, D. W. (1989b). A new species of starling (Sturnidae, Vale, T. R. & Vale, G. R. (1976). Suburban bird populations Aplonis) from an archaeological site on Huahine, Society in west-central California. J. Biogeogr., 3, 157 65. Islands. Notornis, 36, 1614. van Riper, C., III, van Riper, S. G., Goff, M. L. & Laird, M. Steadman, D. W. (1991). Extinct and extirpated birds from (1986). The epizootiology and ecological significance of Aitutaki and Atiu, Southern Cook Islands. Pacif Sci., 45, malaria on the birds of Hawaii. Ecol. Monogr., 56, 327-44. 325-47. Vesey-Fitzgerald, D. (1940). The birds of the Seychelles. Ibis', Steadman, D. W. (1992). Extinct and extirpated birds from 4, 480-9. Rota, Mariana Islands. Micronesia, 25, 71-84. Voous, K. H. (1957). The birds of Aruba, Curacao and Steadman, D. W. (1993). Bird bones from the To'aga site: Bonaire. Stud. Fauna Curacao, Aruba, Bonaire & Other prehistoric loss of seabirds and megapodes. In The To'aga Carib. Isl., No. 7. site: three millennia of Polynesian occupation in the Manu'a Voous, K. H. (1983). Birds of the Netherlands Antilles, 2nd islands, American Samoa, ed. P. V. Kirch & T. L. Hunt. edn. De Walburg Pers, Utrecht. University of California Archaeological Research Facility, Wall, L. E. (1983). Birds in Tasmanian pine plantations. Berkeley, CA, pp. 217-28. Proc. R. Soc. Tasmania, 117, 125-33. Steadman, D. W. & Kirch, P. V. (1990). Prehistoric extinc- Walters, M. (1988). Probable validity of Rallus nigra Miller, tion of birds on Mangaia, Cook Islands, Polynesia. Proc. an extinct species from Tahiti. Notornis, 35, 265-9. Natl. Acad. Sci., USA, 87, 9605-9. Ward, P. (1968). Origin of the avifauna opf urban and subur- Steadman, D. W. & Martin, P. S. (1984). Extinction of birds ban Singapore. Ibis, 110, 238-55. in the late Pleistocene of North America. In Quaternary Watling, D. (1982). Birds of FO'L Tonga, and Samoa. Mill- extinctions: a prehistoric revolution, ed. P. S. Martin & wood Press, Wellington. R. G. Klein. University of Arizona Press, Tucson, AZ, pp. Watson, G. E., Zusi, R. L. & Storer, R. W. (1963). Prelimi- 466-77. nary fieM guide to the birds of the Indian Ocean. US Steadman, D. W. & Olson, S. L. (1985). Bird remains from National Museum, Smithsonian, Washington, DC. an archaeological site on Henderson Island, South Pacific: Wetmore, A. (1960). Pleistocene birds in Bermuda. Smithso- man-caused extinctions on an 'uninhabited' island. Proc. nian Misc. Collec., 140, 1-11. Natn. Acad. Sci., 82, 6191-5. Williams, G. R. (1973). The natural history of New Zealand. Steadman, D. W. & Zarriello, M. C. (1987). Two new species A. H. and A. W. Reed, Wellington. of parrots (Aves: Psittacidae) from archeological sites in Williams, W. D. (1974). Biogeography and ecology in Tasma- the Marquesas Islands. Proc. Biol. Soc. Wash., 100, 518-28. nia. W. Junk, The Hague. Steadman, D. W., Pahlavan, D. S. & Kirch, P. V. (1990). Winterbottom, J. M. (1933). Bird population studies: a pre- Extinction, biogeography, and human exploitation of birds liminary analysis of the Gold Coast avifauna. J. Anita. on Tikopia and Anuta, Polynesian outliers in the Solomon Ecol., 2, 52 97. Islands. Bishop Mus. Occ. Pap., 30, 118-53. Wodzicki, K. & Wright, S. (1984). Introduced birds and Steadman, D. W., Casanova, P. V. & Ferrando, C. C. (1994). mammals in New Zealand and their effect on the environ- Stratigraphy, chronology, and cultural context of an early ment. Tuatara, 27, 77-104. faunal assemblage from Easter Island. Asian Perspect., 33, Wolff, T. (1967). The fauna of Rennell and Bellona, Solomon 79-96. Islands. Phil. Trans. R. Soc. Lond. B., 255, 321-43. Stoddart, D. R. (1971). Settlement, developoment and con- Wood, C. A. & Wetmore, A. (1925). A collection of birds servation of Aldabra. Phil. Trans. R. Soc. Lond., B, 260, from the Fiji Islands. Ibis, 55, 814-55. 611-28. Wood, C. A. & Wetmore, A. (1926). A collection of birds Stonehouse, B. (1962). Ascension Island and the British from the Fiji Islands, Part III. Ibis, 56, 91-136. Ornithologists Union Centenary Expedition 1967-69. Ibis, Woods, R. W. (1975). The birds of the Falkland Islands. 103b, 107-23. Anthony Nelson, Oswestry. Introductions and extinctions of birds 95

APPENDIX 1. References to avifaunas, extinct species and avian introductions for land and freshwater birds by location

Island/mainland References

Aldabra Watson et al. (1963); Duffey (1964); Moreau (1966); Benson (1967); Benson & Penny (1971); Hart Davis (1972); Penny (1974); Diamond (1984) Andamans/Nicobars Tikader & Das (1985); Ripley & Beehler (1989) Aruba Hummelinck (1940); Voous (1957, 1983); Pregill (1981); Olson (1984) Ascension Stonehouse (1962); Watson et al. (1963); Duffey (1964); Greenway (1967); Hart-Davis (1972) Auckland Greenway (1967); Knox (1969); Cassels (1989) Australia Ryan (1906); Pizzey & Doyle (1980); Balmford (1981); Horton (1984); Rolls (1984) Azores Bannerman (1963) Bahamas Buden & Schwartz (1968); Bond (1980); Olson & Hilgartner (1982); Pregill (1982); Olson et al. (1990); Milberg & Tyrberg (1993) Balearic Ridpath & Moreau (1966); Bannerman & Bannerman (1983) Bermuda Bourne (1957); Wetmore (1960); Crowell (1962); Pregill (1981); Olson (1984); Heilprin (1889); Lockwood & Moulton (1994) Borneo Smythies (1968) California Channel Power (1972); Jones & Diamond (1976); Diamond & Jones (1980); Milberg & Tyrberg (1993) Canaries Bannerman (1963); Smith (1965); Bacallado (1976); Baez (1992); Milberg & Tyrberg (1993) Cape Verde Bourne (1966); Moreau (1966); Lobin (1984, 1986) Chagos Loustau-Lalanne (1962); Watson et al. (1963); Stoddart (1971); Diamond (1984) Chatham Forbes (1893); Fleming (1939); Greenway (1967); Knox (1969); Cassels (1984) Christmas Watson et al. (1963) Cocos Hertlein (1963); Slud (1967) Comoros Benson (1960); Watson et al. (1963); Forbes-Watson (1969); Diamond (1984) Cooks Holyoak (1974); Steadman (1986, 1991); Pratt et al. (1987); Steadman & Kirch (1990); Milberg & Tyrberg (1993) Corsica-Sardinia Heinzel et al. (1972); Milberg & Tyrberg (1993) Cozumel Paynter (1955) Curacao Hummelinck (1940); Voous (1957, 1983); Olson (1984) Easter Stottsberg (1956); Johnson et al. (1970); Carr (1980); Milberg & Tyrberg (1993); Steadman et al. (1994) Falklands Cawkell & Hamilton (1961); Pittingill (1973); Woods (1975) Fiji Wood & Wetmore (1925,1926); Gorman (1975); Pernetta & Watling (1979); Watling (1982); Clunie (1984); Pratt et al. (1987); Steadman (1989b) Galapagos Harris (1973); Olson (1984); Kramer (1984); Steadman (1986) Great Britain Ridpath & Moreau (1966); Heinzel et al. (1972); Lever (1977) Guadalupe Mex. Bostic (1975); Jehl & Everett (1985) Hawaii Caum (1933); Munro (1944); Schwartz & Schwartz (1949); Berger (1972); Moulton & Pimm (1983); Olson & James (1982, 1984); Pratt et al. (1987); Simberloff & Boecklen (1991); Moulton (1993) Henderson Fosberg et al. (1983); Steadman & Olson (1985); Pratt et al. (1987); Milberg & Tyrberg (1993) Jamaica Mittermeier (1972); Lack (1976); Bond (1980); Pregill (1981); Pregill et al. (1991); Milberg & Tyrberg (1993) Java Holmes & Nash (1989) Juan Fernandez Skottsberg (1956); Moreau (1966) Kangaroo Edgar (1965); Greenway (1967); Soper (1968); Ford (1979); Pizzey & Doyle (1980); Cassals (1984); Newsome & Noble (1986) Kermadec Edgar (1965); Greenway (1967); Soper (1968); Knox (1969) Lord Howe Hindwood (1940); McKean & Hindwood (1965); Greenway (1967); Marsh & Pope (1967); Recher (1974) Madagascar Rand (1936); Watson et al. (1963); Diamond (1984); Langrand (1990); Milberg & Tyrberg (1993); Goodman (1994) Marianas (Guam) Owen (1977a,b); Jenkins (1983); Pratt et al. (1987); Engbring & Fritts (1988); Steadman (1992) Marquesas Bruner (1972); Pratt et al. (1987); Steadman & ZarrieUo (1987); Steadman (1988); Milberg & Tyrberg (1993) 96 T. J. Case

Mauritius Watson et al. (1963); Greenway (1967); Bullock (1977); Diamond (1984); Cheke (1987); Simberloff (1992) New Caledonia Delacour (1966); Hannecart & Letocart (1983); Balouet & Olson (1989) New Guinea Rand & Gilliard (1968); Diamond & Marshall (1976); Beehler et al. (1986); Bell (1986) New Zealand Thomson (1922); Turbott (1961); Williams (1973); Diamond & Veitch (1981); Druett (1983); Falla et al. (1983); Cassels (1984); Wodzicki & Wright (1984); Robertson (1985) Norfolk Marsh & Pope (1967); Turner et al. (1968); Holloway (1977); Schodde et al. (1985) Palau Owen (1977a,b); Pratt et al. (1987); Engbring (1988) Pemba Moreau & Pakenham (1941); Moreau (1966) Puerto Rico Bond (1961); Lack (1976); Philibosian & Yntema (1977); Schwartz (1978); Pregill (1981); Raffaele (1983, 1989); Milberg & Tyrberg (1993) Reunion see Rodriguez and Mauritius Revillagigedo Jehl & Parkes (1982); Brattstrom (1990) Rodriguez Watson et al. (1963); Moreau (1966); Gill (1967); Diamond (1985); Cheke (1987); Simberloff (1992) Rottnest Pizzey & Doyle (1980); Saunders & de Rebeira (1983) Saint Helena Harting (1873); Benson (1950); Haydock (1954); Ashmole (1963); Greenway (1967); Olson (1975, 1984) Samoa Watling (1982); Pratt et al. (1987); Milberg & Tyrberg (1993); Steadman (1993) Sea of Cortez Cody (1983) Seychelles Vesey-Fitzgerald (1940); Watson et al. (1963); Peake (1971): Penny (1974); Diamond (1984, 1985); Lionnet (1984) Societies (Tahiti) Curtiss (1938); Guild (1938, 1940); Delacour (1966); Bruner (1972); Thibault & Rives (1975); Pratt et al. (1987); Waiters (1988); Steadman (1989a); Lockwood et al. (1993); Milberg & Tyrberg (1993) Solomons Wolff (1967); Diamond & Marshall (1976); Hadden (1981); Blabar (1990); Diamond (1991); Milberg & Tyrberg (1993) Sri Lanka Phillips (1953); Henry (1971); Ridpath & Moreau (1966); Crusz (1984) Taiwan Severinghaus & Blackshaw (1976) Tasmania Ridpath & Moreau (1966); Williams (1974); Pizzey & Doyle (1980) Three Kings Turbott & Buddle (1948); Turbott (1963); Knox (1969) Tres Marias Stager (1957), Grant & Cowan (1964); Nelson (1989) Trinidad Ffrench (1980) Tristan de Cunha Elliott (1957); Holdgate (1960) Tuamotus Bruner (1972); DuPont (1976); Pratt et al. (1987) United States Phillips (1928); Bump (1963); Mayr (1965); Bohl & Bump (1970); Scott (1987); Banks (1976, 1981); Steadman & Martin (1984) Vanuatu Cain & Galbraith (1957); Medway & Marshall (1975); Marshall (1976); Pickering (1981a,b); Diamond & Bregulla (1992) Wake Greenway (1967); Owen (1977a,b); Pratt et al. (1987) Zanzibar Moreau & Pakenham (1941); Watson et al. (1963); Moreau (1966) General King (1981); Long (1981); Atkinson (1985, 1989); Lever (1985, 1987); Johnson & Stattersfield (1990); Sibley & Monroe (1990)