Island Biogeography: Assembly and Evolution 425 © Lnclosures with lizards remaved 2 Control encloses sth hard © Unenclosed plots vith lizards Figure 14.13 On Bahamian islands lacking, predatory lizards, spiders often exhibit predatory release, reaching densities about ten times as high son islands inhabited by these lizards, Experi- ‘mental removal of lizards from enclosed plots demonstrated that population densities of spiders are strongly influenced by predation. (From, ~ ‘Schoener and Spiller 1987.) Caribbean Anolis in great detail. Although many distributional patterns have ‘been well documented, the underlying causal processes have proved more df= ficult to demonstrate convincingly. Anolis is but one of several important lizard genera on the mainland of trop- ical America, but it is by far the most abundant genus of vertebrates on the Caribbean islands. On the large islands of the Greater Antilles, a few coloniz~ ing species have given rise by endemic speciation and adaptive radiation to a diverse Anolis fauna. Hispaniola, the second largest and ecologically most diverse island, has at least 35 species, which were probably derived from four separate invasions (Williams 1976, 1983). These species occupy a variety of ecological niches. They range from tiny insectivores to large carnivores, with hhead and body lengths ranging from 40 to more than 200 mm. They are mor- phologically, physiologically, and behaviorally specialized for distinetive habi: tats and microenvironments, from sunny sites to deep shade, from open ground and rocks to grasslands and scrub forests to different layers in the complex vegetation of mature tropical and montane forests (Figure 14.14). In contrast, the small islands of the Lesser Antilles each have only one oF {wo generalized species (Roughgarden and Fuentes 1977; Roughgarden et al 1983). When two species coexist on a small island, they differ in body size, prey size, and habitat; but when only one species is present, itis intermediate in size, takes a wide range of prey, and occupies virlually all habitats (Figure 14.15). Clearly, the fundamental niche of an Anolis species that has evolved in the absence of congeners is very broad. Some of the observed niche expansion of Anolis on small islands may represent behaviorally mediated ecological responses to the absence of competing species. Nevertheless, most of the niche shifts in Caribbean species represent evolutionary adaptations to communities containing different numbers and kinds of other species, and these are reflected in morphological, physiological, and behavioral changes (see Rough- garden 1995),426 Chapter 1 Figure 14.14 Diagrammatic representation of the habitats occupied by different Aro species on the northern part ‘of Hispaniola. Their niches differ in elevation, vegetation type, perch height, and post tion on a gradient from sun: light to shade. The species in- dicated here, only a fraction ofthe at least 35 species that inhabit the island, have been produced largely by specia ion and adaptive radiation within the island, (From Montane breadleat shade Moatane open Lowland al stniger forevir Lowland! sora Williams 1983.) Shade Son Jamesby ———*—— Pest Toba a Petit Bateau Pain Union —# Marie Galante . Saha ——*— — 84, Croix St.Kitts A$ * Bacbada —i. St Maarten ‘St. Eustatius —™ * a St Vincent Jaw length (am), ada ma Variation in Anolis densities and niche characteristics between islands and between habitats within islands has also been investigated, especially by Schoener (1968a, 1975), Lister (1976a,b), and Holt (1994). They have shown that when a species occurs on an island or in habitats where there are few other lizards, increases in niche breadth often parallel increases in density (Figure 14.16). Again, itis tempting to attribute these pattems of density compensation and niche expansion to a release from competition with other Anolis species, but this assumption may be unwarranted. The small islands where the lizards show the most spectacular increases in density contain not only fewer Anolis species, but also fewer species of all terrestrial animals. Thus the Anolis popu- lations are potentially released from competition not only with congeners, but also with insectivorous arthropods (¢.., spiders), birds, frogs, and lizards of other genera, and from predation by birds and other lizards, It is also possible that sinall islands support higher standing stocks of insect prey. Holt studied, Figure 14.15 Body sizes of Avolis iaaeds of the Lesser Antilles, Note that all of these islands have either one or two species. When two species co-occut, they tend fo be displaced in size, whereas when only one species i termediate size. (Data fzom Roughgarcen 1974; Roughgarden and Fuentes 1997; Roughgarden et al. 1983Island Biogeography: Assembly and Evolution 427, Anolis on islands off Irinidad and concluded that direct competition with other Anolis or other lizard species could not account for the observed density changes. He suggested that the absence of predatory birds might be the most important of the various factors contributing to increased Anolis density and niche breadth on small islands. Thus, while ecological release and high densities among insular popula tions are often attributed to reduced competition, they are more generally a consequence of the depauperate nature and reduced intensity of interspecific interactions in general. Density overcompensation. One puzzling aspect of insular density com pensation is that the total popullation densities of a few species inhabiting a small island may actually exceed the combined densities of a much greater number of species of the same taxon occupying similar habitats on the main- land. This phenomenon, called density overcompensation or excess density compensation, is fairly well documented, especially for birds on small oceanic islands. In fact, Crowell’s (1962) studies, mentioned above, revealed that just 10 species of small passerine birds on Bermuda together maintained a popula- tion not just equal (o, but about 1.5 times greater than the combined densities, of 20 to 30 species on the mainland, Subsequent studies by MacArthur et al. (1972) on the Pearl Islands south of Panama, by Diamond (1970b, 1975b) on. the Bismarcks and other archipelagos north and east of New Guinea, and by Emlen (1978) on the Bahamas have described similar patterns. Case (19% documented density overcompensation among the lizards on the islands in the Gulf of California compared with the mainland of Baja California and Sonora. Several explanations have been proposed for such overcompensation, and for density compensation as well: 1. Because bird species of lange body size tend to be absent from small islands, the same resources can support substantially larger popula- tions of small birds. Note, however, that Crowell’s (1962) study found that biomass, as well as density, was substantially higher among insu- lar birds than among mainland birds. 2. Bird densities reflect release from competition not only with missing bird species, but also with other taxa, such as mammals and amphib- ians, which use similar foods and other resources, but are even less common on oceanic islands than birds because they are poorer over- water dispersers. 3. Inflated population densities on islands reflect the absence or paucity of predators and parasites, which also tend to be poorly represented ‘on oceanic islands (MacArthur et al. 1972; Grant 1966; Case 1975; George 1987; McLaughlin and Roughgarden 1989), 4, Oceanic islands are more productive, at least in terms of foods and. other resources required by small birds (see Case 1975). (On oceanic islands, renewable food resources are harvested at ral closer to their maximum sustained yields than on mainlands, where intense competition leads to overexploitation and lower productivity of resources. 6. Populations can become more finely adapted to their local environ- ment, and hence attain higher densities, on isolated islands than on continents, where extensive gene flow among populations occupying different habitats tends to prevent specialization for more efficient use of local resources (Emlen 1978, 1975). Males 3 Z| ea o 1 2 3 4 Nuenber of species present Females o 1 2 3 «5 Number of species prevent Figure 14.16 Relationship between the diversity of heights of perches used by males and females of Anolis sangrei and the number of other Anolis species occurring in the same habitat on the same island, Note than when A. sigrel is the only species present, fuses a wide variety of perch heights, but the breadth of this niche dimension con- tracts and A. savgre is exeluded from, arboreal habitats as it encounters an in- creasing number of coexisting con- generic species. (After Lister 19760.)428 Chapter 14 7. The surrounding waters act as a fence, preventing emigration of indi- viduals into marginal habitats (the fence effect; see Krebs et al. 1969; Emlen 1979; MacArthur et al. 1972). As is often the case with the most general patterns in ecology and biogeog- raphy, these patterns may result from a combination of convergent mecha- nisms, rather than just one unique, averriding one. Patterns of combined densities of insular species, including compensation and overcompensation, may be summarized in a graphical model (Figure 14.17A; Wright 1980). We again emphasize that, like their mainland counter- parts, insular communities often are Gleasonian, with species responding, sometimes uniquely to one another and to environmental conditions. That is, while the overall pattern exhibited by a particular insular community may be density compensation, some of its species are likely to exhibit atypically low populations, while densities of others are compensatorily high, For example, Crowell (1962) found that common crows, house sparrows, and starlings were extremely rere in all but farmland habitats on Bermuda, while catbird, cardi- nal, and white-cyed vireo populations were 3 to 30 times higher in comparison to the same habitats on the mainland. Similarly, in a study of Caribbean (A) Comunity level (6) Species level 1 Overcompensation 4 Density compensation © Density stan Spaces Density 4 . Ibsen / aE / stasis “ gk Wemsity E Joven ° 2 Species chnoss Species richness (Grislnd are} icrisland ses) Figure 14.17 _ (A) Insular population densities (number of individuals per unit of area) may exhibit one of three qualitatively different responses to variations in species richness (or island area). If population densities are not influenced by com petition, predation, and other interactions among species, then the densities of all species combined should increase in proportion to the total number of sity stasis). On the other hand, if interspecific interactions do limt insular popula- tions, they may exhibit density compensation or avercompensation on species-poor islands (ie, populations of one or a few insular inhabitants may equal or exceed those of a much richer assemblage of species) (B) Population densities of individual species also may exhibit three qualitatively different responses to lfferences in species richness (or island area) among islands. Density enhancement (sensi jaenike 41978): on very small, species-poor islands, population levels of a species may fall below those ofits conspecifics on the mairlland, but increase as island area increases and environmental conditions become more suitable. Density inflation: population levels may continue to increase, with increasing area, above their mainland levels, until interspecific interactions begin to regulate population densities ofthe focal species. Density stasis: at intermediate to high levels of species richness, many. species may regulate one another's densities such that few, if any, exhibit any con- sistent trend of population density with species richness or island area. (A after ‘Weight 1980.)Island Biogeography: Assembly and Evolution 429 insects, Janzen (1973) found that generalists, especially the Homoptera, tended, to exhibit density compensation on islands, while other insects did not. Aspecies-based alternative to Wright's model (Figure 14.17B) introduces an additional density pattern termed density enhancement (see Jaenike 1978) Population densities of some, and perhaps most, species on very small islands, may indeed be much lower than their typical densities of the mainland, but, increase with island area. Larger islands not only support more species, some of which may be mutualists or commensals, but also tend to provide less harsh, more stable, and otherwise more favorable habitats for particular species, Alternatively, population densities of at least some species may remain rwlatively stable over a broad range of species richness—a pattern that Williamson (1981) termed density stasis. In fact, the same species may exhibit, all three patterns—density enhancement on tiny islands, density inflation on islands of intermediate size, and density stasis on larger, more species-rich islands (see Figure 14.17B, species A). Pechaps density inflation also varies sys- tematically among species: highest for supertramps and lowest for those species restricted to the richest, and largest, islands. We admit that these are largely untested predictions, but the pertinent data seem to be available, at least for some insular communities. In summary, studies on the ecological responses of insular populations have been, and will certainly continue to be, fertile ground for research on the forces influencing community structure. The comparative approach enables identifi- cation of some intriguing patterns, and variation among islands provides the opportunity for natural experiments, while their isolation, small size, and sim- plicity often enable manipulative experiments in @ controlled, but realistic, arena. Evolutionary Trends on Islands Flightlessness and Reduced Dispersal Ability on Islands ‘As with mariners shipwrecked near a coast, it would have been better for the good swimmers if they had been able to swim stil further, whereas it would ave been better for the bad swimmers if they had not been able to swim a al and had stuck to the wreck, (Darwin 1859, p. 177) With this one metaphor, Darwin explained a truly fascinating paradox of insu Jar biotas: many insular forms, especially those on the most isolated oceanic islands, have little or no ability to disperse to other islands. As many have remarked, there is no greater anomaly in nature than a bird that cannot fly (Darwin 1859). Yet, flightless birds and insects are relatively common on many ‘oceanic islands, as are other animals and plants with litte ability or propensity for dispersal. And herein lies the paradox: how could these relatively seden- tary forms have colonized such remote ecosystems? During Darwin’s day, the extensionists may have used these abservations to bolster their argument that ancient landbridges once connected even the most isolated archipelagoes with the continents. As we saw in Chapter 2, nothing “vexed” Darwin more than these post hoe scenarios of the earth's dynamism, Instead, he offered an explanation embodied in the above quote. There was no doubt, at least in Darwin's mind, that these peculiar insular forms were the products of evolution, and that their ancestors had arrived via long-distance dispersal Unfortunately, Darwin attributed some of these patterns to what was termed the “law of disuse.” For example, he believed “that the nearly wingless430. Chapter 14 condition of several birds, species which now inhabit several oceanic islands, tenanted by no beast of prey, has been caused by disuse.” He acknowledged, however, that the fightlessness of many insular forms, such as 200 of the 550 described beetles of Madeira Island, was “wholly, or mainly due to natural selection.” The key point is that the selective forces operating on a potential immigrant may be entirely different from those operating. on its insular descendants. In Darwin's metaphor, selection during immigration favors “good swimmers,” but once they (or their descendants) have reached the “wreck” (island), selection then favors “bad swimmers” —those less likely to swim off or be carried away by winds and lost at sea. Birds. Flightlessness ond reduced dispersal ability have evolved repeatedly. ina variety of birds and insects (see Carlquist 1974). Derived flightlessness is reported in at least cight orders of birds (Struthioniformes, Anseriformes, Psittaciformes, Strigiformes, Columbiformes, Gruiformes, Ciconfiformes, and. Passeriformes) and in most oceanic archipelagoes. On New Zealand, some 25 to 35% of the terrestrial and freshwater binds are—or were, in the case of extir- pated forms—fightless. Similarly, 24% (20 species) of Hawaii's endemic birds were flightless. The long list of flightless birds includes a cormorant of the Galépagos (Phalacroconax harrisi) and many flightless giants, including the dodo (Ruphus cucullatus) of Mauritius, the solitaires (R. solitarius and Pezophaps solitariaort) of Réunion and Rodriguez, the elephant birds (Aepyomithidae) of Madagascar, and the moas (Paclyornis spp.) of New Zealand (see review by McNab 1994a). Flightlessness is especially common among the rails, having, evolved independently within at least 11 groups of rails (Diamond 1991). In fact, Olson (1973) has speculated that flightlessness in rails may evolve over a time span of decades to centuries, rather than millennia. While this may seem astounding, paleontological evidence suggests that, before human coloniza tion, most Pacific islands were inhabited by a flightless rail species (see Stesd- man 1986 and 1989; Steadman and Olson 1985) Currently, the most widely accepted explanation for the evolution of fight- lessness in birds involves selective pressures associated with the absence of predators and limited resources on islands (McNab 1994; Diamond 1991). ‘Under reduced pressure from predators, and for that matter, from competitors, as well, insular populations would be likely to undergo ecological release Evolutionary responses could include changes that would conserve energy, such as reduiced size overall, ora reduction in metabolically expensive tissues, such as the otherwise large flight muscles. On the mainland, such selection to conserve energy is of course countered by the selective advantage of being able to escape ground- atively simple model shown in Figure 14.22B successfully accounts for the a : Gone ul, renjsinslagely untested. How much ofthe residual variation in LLagomorphs 6 figure 14.22A can be accounted for by factors in addition to the body size of Rodents 6 the ancestral spedies? The model predicts, for example, thal the relative body Camivores B sizes of insular mammals should increase with island area and isolation, and. Artiodactyls 5 that they should Be higher for herbivores than for carnivores. Consistent with Same Fos ioe —~SCSC~*~*~*~*S*SCSCSCSCSRe Former prediction, the body sizes of tri-colored squirrels and fruit bats of