Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Molecular ecology wikipedia , lookup
Theoretical ecology wikipedia , lookup
Ecological fitting wikipedia , lookup
Biodiversity action plan wikipedia , lookup
Introduced species wikipedia , lookup
Occupancy–abundance relationship wikipedia , lookup
Fauna of Africa wikipedia , lookup
Habitat conservation wikipedia , lookup
Latitudinal gradients in species diversity wikipedia , lookup
Distributional Ecology of New Guinea Birds Author(s): Jared M. Diamond Reviewed work(s): Source: Science, New Series, Vol. 179, No. 4075 (Feb. 23, 1973), pp. 759-769 Published by: American Association for the Advancement of Science Stable URL: http://www.jstor.org/stable/1735788 . Accessed: 22/12/2012 20:00 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. . American Association for the Advancement of Science is collaborating with JSTOR to digitize, preserve and extend access to Science. http://www.jstor.org This content downloaded on Sat, 22 Dec 2012 20:00:09 PM All use subject to JSTOR Terms and Conditions Species lliversity on Islands If one were to count all the animal or plant species occurring within an area of I hectare, the result would vary greatly dependng on the location of the census area. The species total would generally be much higher in the tropics than in the temperate zones, higher at the base of a mountainthan at the summit, higher on a large island than on a small island, and higher on an island near a continent than on a remote Recent ecological and biogeographicaltheoriescan island. It is important to understand be tested on the bird communitiesof New Guinea. this variation if, for example, one is establishinga system of national parks to ensure survival of as many native JaredM. Diamond species as possible. Islands such as those of the southwest Pacific lend themselves well as test areas for a quantitativetheory of species diversity As indicated by frequent references understood.New Guinea has served as because each island represents a sepain the syntheses of zoogeography by the birdcolonizationsourcefor the thou- rate experiment, and because most inWallace;(l), of evolution by Mayr (2), sands of islandsof the southwestPacific; sular variation in species diversity can and of ecology by MacArthur (3), the and New Guinea itself behaves as an be predicted from values of only two tropicalisland-continentof New Guinea "islandarchipeSagos' for montane birds, or three readily measuredvariables.We and its birds have played a special role since its mountain ranges are isolated shall see that the diversity of bird in advancing our understandingof ani- from each other by a i'sea" of unin- species on most Pacific islands is in a mal populations. This role developed habitablelowlands.The numberof bird state of dynamic equilibrium that is, partlybecause birdsare the best known, species on these oceanic islands and the diversity is determined by the most easily observed and identifiedani- mountain islands varies with area and island's present physical characteristics mals, and partly because of unique isolation, providing innumerable "ex- and is independentof the island's hisadvantagesof New Guinea itself New periments of nature" whereby the tory. However, on some islands, the Guinea provides a range of habitats niche of a given species can be studied species diversity may also reflect the from tropicalrainforest to glacierswith- as a function of the competing species island's recent history. in distances of less than 16 kilometers, pool. The number of land and freshwater a range of elevations of over 50()0 Within the past decade, new para- bird species coexisting on each tropical meters, and an equatorialposition that digms introsluced by MacArthur and island of the southwest Pacific varies minimizes seasonal migration with its Wilson and their co-workers(3-7) have from 1 for some isolatedkatolls up to associated complications. The rugged revolutionized our understanding of 513 for New Guinea itself. This variatopography, which isolates populations some central questionsof ecology, such tion is due partly to the greater variety in adjacent valleys or on adjacent as: Why do diSerent localities support of habitatspresenton the largerislands. mountains, has promoted speciation very different numbers of animal or However, even within a given habitat within small areas of a single land mass plant species? What determines the type (for example, in tropical lowland by essentiallythe same mechanismsthat distribution of a given species? How rain forest) there are great differences underlie speciation on large continents. do related species manage to coexist? among islands in the number of bird In New Guinea'sexpanses of forest un- These questions are not only of basic species to be found. These differences disturbed by man, niche interrelations scientific interest but are also of prac- are largely predictablefrom an island's retain a simplicity and beauty lost in tical importance in formulating con- area, its distancefrom New Guinea, and altered environments,and distributional servation policies. Furthermore, the its elevation (13). patterns illustrating many intermediate concepts of MacArthurand Wilson are Figure 1 shows the number of bird stages in evolution and in niche dis- proving increasingly helpful in under- species S occurring at sea level on placement are readily identified. The standing human populations. In this islands between 8 and 500 km from number of breeding bird species, 513, article I discuss these questions in the New Guinea, as a function of island is large enough to give rise to the com- light of my studies of New Guinea area A (expressedas squarekilometers). plex interactions characteristicof con- birds, conducted during six expeditions Over a 3-millionfold range of areas the tinental faunas, but not s-o large as to to New Guinea and other islands of the results fit the power functionbe overwhelming.In spite of the physi- southwestPac;fic (8-15). A recent book cal difficuIties of exploration in New discussesin detail many of the examples 5-12.3A022 (l) Guinea, the distributionand taxonomy summarizedhere (15). Many patterns with an average error of 19 percent. of its bird species are by now fairly well observed in New Guineaibirdsare rele- Thus, a tenfold increase in area invant to other groups of animals in other creases species diversity by somewhat The author is professor of physiology at the parts of the worlds especially in the less than a factor of 2. The numbered School of Medicine, University of California, Los Angeles 90024. tropics. deviant points represent islands in var23 FEBRUARY 1973 7ss DistributionalEcology of New GuineaBirds This content downloaded on Sat, 22 Dec 2012 20:00:09 PM All use subject to JSTOR Terms and Conditions As a test of the equilibriumtheory, t represents time. Let us assume conin 1968 and 1969 I resurveyedthe land stant coefficients Ki and Ke, (expressed and freshwater birds of a temperate as year- l ), respectivelyarchi,oelago and of a tropical island (3) E= Ke S(t) whose birds had been surveyed50 years (4) I = Ki [S -S(t)] previously:the nine ChannelIslands off dS/dt = l-E KiS*:southern California (9), and Karkar (Ki + Kc)S(t) (5) Island off northern New Guinea (l 2). is the mainland species pool. On each island I found that between where S: (dS/dt = O), the species equilibrium At 17 and 62 percentof the species present given by is S(.(, diversity 50 years ago had disappeared,and an approximatelyequal number of species KiS/(Ki + Kc) absent 50 years ago had immigrated. (6) Relaxation to equilibriumfrom an iniThus, the species diversityhad remained tial species diversity S(O) that differs in dynamic equilibrium;Terborgh and St, is described by: from Faaborg (17) obtained similar results for the birds of Mona Island in the [S(t) - S(sq]/[S(O) - Se] = e tXtr (7) West Sndies.While a few of the extinctions and immigrationsin these studies where "relaxationtime" tl. is given by were related to effects of man, most of (8) the changes were of the random kind expected in the abselace of a human The relaxation time is the length of role. As predicted f rom colonization time requiredfor the departureof spetheory (4), most of the extinctions in- cies diversity from equilibrium,IS(t)volved populations that were rare 50 S(ll, to relax to 1/e (or 36.8 percent)of years ago because of such factors as the initial departure,15(°)-Se,ll, where recency of colonization, small island e is the base of naturallogarithms.Actu>N= (12.3) ( I + 0.089 L/ 1000) X size, presence of well-establishedcom- ally, this treatmentis only a crude apreal situation, be(e-D/'"")(A0 22) (2) peting species or small area or margi- proximation of the nal suitabilityof island habitat. Among cause K( and Ki prove to be functions. Similarly, species diversity in other tlaeChannelIslandsthe calculatedmini- Of S(t). However, by applying Eqs. plant and animal groups on other mum turnover rates, expressed as the 3 to 8 to island faunas we can deduce islands increases with island area and percentageof islandspecies immigrating the form of these functions, stipulate decreases with distance from the colo- or becoming extinct per year, range some conditionsfor an improvedmodel, nization source. These two trends con- from 1.2 percent per year for the and come to some conclusions about stitute the fundamental law of island smallest or most barren islands down the distribution of bird species. As an example of an "experimentof biogeography.The exponent of area is to 0.3 percent per year for the largest that permitsestimationof S(0), nature" generally in the range 0.20 to 0.34 island. pair of values of S(t) and t, and one S(.<l, Karfor A values and S the Since (4. l 6). To explain this law, MacArthurand kar fit Eq. l and lie on the general pat- hence calculation of tl., I will describe Wilson (4), and Preston (16), suggested tern of Fig. 1, it is probable that bird the situation on a land-bridge island, that insular species diversity represents peies diversity is near equilibriumon Misol. During the most recent Pleistoa dynamic equilibrium between im- most southwest Pacific islands. How- cene glaciation, when much water was migration and extinction. Islands con- ever, there are some points in Fig. 1 sequesteredin glaciersand sea level was stantly receive immigrantsdue to ran- that deviate conspicuously from the about 100 m below its present stand dom dispersal of individuals from general pattern. These prove to be as- (18), Misol was partof the New Guinea mainlands, and the immigration rate sociated with islands on which natural mainland and must have supported should increase with island area and processes (such as volcanic explosions virtually the full New Guinea lowland proximity to the colonization source. or changing sea level) have displaced avifauna of 325 species. Since the sevOn the other hand, island populations S from its equilibrium value at some ering of the New Guinea-to-Misol risk extinction due to competition and approximatelyk nown recent time. On l-and bridge by rising .sea level about random fluctuationsin population size, such islands S must be gradually re- 10,000 years ago, S(t) on Misol must and extinction rates should be highest turning to equilibriumas a result of a have been relaxing toward the value for the smallest islands with the small- temporaryimbalancebetween immigra- expected for an "oceanic island"of the est populations. Thus, at equilibrium, tion and extinction; this process -may same area, as a result of extinction exwhen the species immigrationrate I on be termed "relaxation."Analysis of the ceeding immigration.Figure 1 (points an island equals the species extinction islands on which species diversity is marked +) shows that Misol and other rate E, there should be more species undergoing relaxation provides further large land-bridgeislands are still superon a large island than on a small island,- insight into immigration, extinction, saturated -that is, their present S(t) values are still considerablyin excess of and more on an island near to other and effects of island history (13). The respective rates of immigration the equilibrium value .for oceanic land than on a remote island. FurthermoreSspecies turn-overrates at equilib- and extinction, I and E (expressed as islands, although considerablyless than rium should decrease with increasing species per year), depend on the in- the initial value of 325. Misol has an stantaneousspecies diversityS(t), where area of 2040 km2 and should have 65 island area and remoten-ess. ious stages of "relaxation,"as will be explained. l n Fig. 2 l ha\e plotted, as a function of island distance D from New Gllillea, the ratio of an island's actual S at sea level to the S value predicted from the island's area and Eq. l. This ratio decreases exponentially with distance, by a factor of 2 for each 2600 km from New Guinea. Thus, the most remote islands of the southwest Pacific (Mangareva and the islands of the Pitcairngroup, 8()()0to 9200 km from New Guinea) have a bird species diversity only 12 percent that of islands of similar size near New Guinea. The mollntains of the higher southwest Pacific islands harbor additional bird species not occurring at sea level. On the average, each 1000 m of elevation L enriches an island's avifauna by a number of montane species equal to 8.9 percent of its avifauna at sea level. Thus, bird species diversityon the New Guinea satellite islands may be summarized by the empirical formula Seq = tl = (Ki + Ke) l SC lENCE, VOL. 179 760 This content downloaded on Sat, 22 Dec 2012 20:00:09 PM All use subject to JSTOR Terms and Conditions lowland species at equilibrium (from Eq. l) but actually has 135. If we take Eq. 7 and substituteS(0) = 325, S(t) = 135, Seq= 65, and t = 10,000 years, we obtain a relaxation time of 7,600 years for the avifauna of Misol. Relaxation times can also be estimated from four other types of "experimentsof nature."These are: the fission of a large island into two smaller islandsbecause of risingsea levels flooding a- low isthmus; contraction in an island'sarea becauseof risingsea levels; gradual extinction, from land-bridge islands, of relict populations of those New Guinea lowland species that never cross water gaps of more than 8 km or even 50 m (such species must have reached the islands at the time of the land bridge, and were isolated without possibility of recolonization after submergence of the land bridge) (l3); and recolonization of volcanic islands such as Krakatau, Long? and Ritter, after cataclysmic eruptions totally destroyed the fauna. From ca]culatedrelaxationtimes obtained by these methods for 19 New Guinea satellite islands, the following conclusions can be drawn(13). I(i)With decreasing island area, extinction rates increase (yielding shorter calculated tr values) because of smaller population sizes. Thus, among land-bridge islands the calculated tr for the'relict populations decreases from 9000 years for an island of area 7800 km2 (Aru), to 6;l00 yearsfor 450 km2(Batanta),and to 2630 yearsifor 145 km2 (Pulu Adi). Species diversities on small land-bridgeislands do not show an excess over equilibrium values (Fig. 1) because relaxationtimes for these islands are much shorter than the 10,000 years that have elapsed since severing of the land bridges. (ii) On supersaturated islands certain species are especially prone to extinction and tend consistentlyto disappearfirst. The resultingreleaseof the remainingspecies from competition tends to increase their population densities and to decrease their risk of extinction.Expressed mathematically,K} on a given island is an increasing function of Sl(t). (iii) The -species that arrive first -at an initia]l.yempty island tend to be'certain species with consistently superior disperlsal abilityn such as those that 1.00-\! characteristically colonize mainland "second-growth"habitats(transientvegetational stages during regrowth of a forest clea'ring).Subsequentimmigrants are drawn from a mainland species pool comprised of progressivelypoorer colonists. Exp-ressedmathematically,K on'a given island is a decreasingfunction of S(t). ;(iY) The probability of extinction is much higher for recently arrived immigrants 'that still have a low population'density, than for established species'that-have saturatedavailable island habitats. Patchiness of Species Distributions . . . . Among the characteristicsof tropical species that'distinguishthem from their high-latitudecounterparts'arethe lesser tendency of tropical'species to disperse, their subjection to greater niche compression' by interspecific competition, and their lower extinction and higher speciation rates. All these characteristics contribute to the striking tropical phenomenon called "patchiness"(3, 7, 15). Whereas the local presence or . . . . * *4 320- *. */ 0.50- / 160- 78 i+t 5 ' + 3+ 0.25._O t) Q 2 1O2s13_/ 80 cn 14@ / *H O' = cn X _ < 3 ',_ <; O 0.125 , Z /? }10 w l 100 1000 10,000 100,000 l 1,000,000 Area (km2) Distance (km) Fig. 1 (left). Number of resident land and freshwater bird species on New Guinea ssatellite islands, plotted as an funotion of island area' on a' double logarithmic scale. Symbols: *, islands on whch species diversity is presumed to be at equiliblrium (the' remaining, numbered islands are in various s-tages of "relaoxation"after displacement of species diversity from an equilibrium value); , exploded volcanic island!s (1, Long); O, contracted islands (2, Goodenough; 3, Fergusson);' + land-bridge islands (4, Aru;- 5, Waigeu; -6, lapen; 7, S;alawati; 8, Misol; 9, Batanta; 10, Pulu' Adi; 11, 'Ron; 12, {Sohildpadj- d3, island fragments split ofT from a larger -island by flooding of an isthmus ( 13, Batjan- 14, Amboina- 15, New Hanover- 16, Tidore). The straight line was fitted by leiast'mean squ'aresthrough points for all 'is]ands except the'land-bridge islands. No,te that the number of species increases with area; and thait deviations Sforrelaxing islands are more marked for 'llargeislands than for small ' Fi'g. 2 (right). Ordiislands, 'because of high extinction rates and short relaxation times on the latter. [After Diamond (13)] nate (logarithmic scale), number of resident land and freshwater bird 'species (S) on tropical southwest Pacific islands more than 500 km from New Guinea,' divided by number of species expected' on an is]and of equivalent area iess 'than 500 km from New Guinea (calculated from Eq. 1). Abscissa^,island distance from New Guinea. The'graph shows that species diversity decl eases by a factor of 2 per 2600 ktn. [After Diamond ( l3')g 761 23 FEBRUARY 1973 This content downloaded on Sat, 22 Dec 2012 20:00:09 PM All use subject to JSTOR Terms and Conditions 6°S- l IJSS - l s l Fig. 3. Distribution of the Papuan tree creeper, Cliszzactezis 1eucopSzaea, in the mountains of New Guinea. Although mountains and forests with similar tree bark extend uninterruptedfor 1600 km, and although there is no other New Guinea bird in the same family, this bark-feeding species has a distribution gap (unshaded area) of 400 km in the middle of its range (hatched area). [After Diamond (15)1 absence of temperate-zonespecies can generally be predicted from knowledge of their particularhabitat requirements, many tropical species are patchily distribu-tedwith respect to the available habitat-that is, they may be absent at a considerable fraction of the localitles oSering a suitable habitat for them. Temperate-zone biologists, because they are rarely confronted with examples of such patterns, often assume that the evidence for patchiness in the tropics can be dismissed as an artifact of inadequate exploration or insufficiently understood habitat requirements.To prove that a species actually is locally absent rather than just overiooked is certainly more difficult than to prove that it is present. Fortunately, the documentation of patchy distributionsin New Guinea has been facilitated by many New Guinea natives who possess a detailed knowledge of local birds. Some natives were able to name (in their native languages) and accurately describe in advance all bird species that I eventually located in their areas; they could distinguish at a distance obscure sibling species in such taxonomically difficult genera as Sericornis; and could accurately describe other species known to them only from single individuals observed up to 10 years previously (14, 15). When such "walking encyclopedias" of bird lore confirm the permanent local absenceof a species that is regularlyencotlnteredin other areas, one can have corfidence that the species is actually absent and not merely overlooked. It is even more difficult to prove that a locality really does meet all of a bird's 762 habitat requirements, since a skeptic can always claim that some unspecified factor has been overlooked. However, many of the examples of patchy distribution that I describebrieflybelow and in detail elsewhere (15) involve wellstudied species which appear to have distinct and readity defined requirements, and which are ubiquitous in habitats meeting these requirementsin many geographical areas of New Guinea. Since there do seem to be some generalizations emerging about patchiness, it is becoming increasingly unnecessary to invoke unspecified factors as an explanation. There are four main types of patchy distribution. 1) Distributionalgaps in a continuous habitat.The ICentralDividing Range of New Guinea provides an uninterrupted expanse of montane forest for 1600 km. Nevertheless, 18 montane bird species that would otherwise be uniformly distributedhave a distributional gap of severalhundredkilometerssomewhere along the Central Range (Fig. 3 ) . For instance, the finch Lonchura montana occurs commonly in large flocks throughoutthe alpine grasslands of western and eastern New Guinea, where it is the only specialized seedeating bird, yet it is absent in the alpine grasslandof centralNew Guinea,which has similar grass and an otherwise similar avifauna (15). Large distributional gaps in a continuoushabitat also occur in the New Guinea lowlands. The interpretationof these remarkable patterns is discussed below. 2) Very local distributions.The distributionsof many species are patchy to an extremedegree, such that they occur at a few widely separatedlocalities but are absent in similar habitatsover most of New Guinea (15). For instance, the flycatcher Poecilodryas placens is known from six scattered areas, where it inhabitsrain forestswith a well-shaded understoryof small saplings in locally flat terrain up to 1000 m above sea level. The species is absent from hundreds of well-explored iocalities with similarhabitatelsewherein New Guinea. Most of these very local species in New Guinea fall Into one of two categories: distinctive "monotypic genera" with no close relatives (that is, genera consisting only of a single isolated species), or members of large genera consisting of many eccalogicallysimilar species. The highly fragmented distributions of these species suggest that they are slowly becoming extinct, either because they are the last survivors of unsuccessful evolutionary lines (monotypic genera), or because they cannot compete with several ecologically similar relativesin the same genus. 3 ) Complementary checkerboard ranges. Some local absences of species are correlated with the presence of ecologically similar congeners, indicating that one-to-one competitive exclusion occurs in New Guinea mainland habitats as on oceanic islands. For example, the extensive midmontanegrasslands, which are a by-productof human agricultureduringthe last few centuries, have been colonized in irregularcheckerboardfashionby eight Lonchurafinch species native to other habitats. Each midmontane area supports only' one finch species over a considerablelocal range of grass types and heights, altitudes, and rainfall conditions,'but the areas inhabitedby a given species are often scattered hundreds of kilometers apart. Evidence indicates that in each instance the first arrival became established over a local area and was able to exclude potential subsequent colonists of the seven other species, but the identity of each locally successful colonist depended partly on chance (15). In the slightlymore complex situation illustrated in Fig. 4 ("compound checkerboard exclusion"), each local area can supportcoexisting populations of any two species out of three potential colonists, and the identity of the locally missing third species varies irregularly. 4) Distributionalislands on a mainland. Comparisonof mountainsseveral kilometersapart within the same range invariably reveals faunal differences This content downloaded on Sat, 22 Dec 2012 20:00:09 PM All use subject to JSTOR Terms and Conditions SCIENCE, VOL. 179 that cannot be explained by differences in habitat (15). For example, species characteristicof lowerlmontane forest at 800 to 1100 m were compared on four peaks of the North Coastal Range. In order of decreasing.area and summit elevation, the mountains and their numberof lower montanespecies were: Menawa (1890 m high), 45 lower montane species at 800 to 1100 m; Nibo (1560 m), 36 species; Somoro (1420 m), 34 species; and Turu (1140 m), 26 species. The four mountainshad structurally similar forest at 800 to 1100 m, but the smaller mountains had less area of such forest and, correlated with this, fewer bi'rdspecies characteristic-of this forest. Similar findings have been described for Andean birds and North American mammals ( 19 ) . These patterns are reminiscent of the fundamental species: area relation of island biogeography(see page 760). Evidently, dispersal rates of birds betwee.n'New Guinea mountainsseparatedfrom each other by valleys a few kilometers wide are so low that the peaks behave as islands. Significanceof Patchiness While much remains bafRing about patchy distributions in the tropics, such distributionsappear to be caused by the synergistic'effects of two characteristics.of tropical species comparedto temperate-zone species. These are 'the lower dispersalrates of tropical species, which prolong the existence of distributional gaps as temporary or nonequilibriumphenomena;and the greater pressure from interspecificcompetition, which stabilizes gaps' as indefinitely miaintainedor equiiibrium phenomena. That the mean dispersal distances of many New Guinea bird species are less than'several kilometers is indicated (i) by the absences of montane species on mountains separatedby several kilometers from populations of the same species on other mountains;(ii) by the numerous species absent on oceanic islands a few kilometers or even only a few meters from the New Guinea mainland (13); and (iii) by the presence of distinct subspecies or even semispecies on mountains separatedby valleys several kilometers wide or on islands separated by straits several kilometers wide. Once a species has become locally extinct for whateverreason, immigration from populations in immediately contiguousareas is so slow that distribu23 FEBRUARY 1973 Fig. 4. Compound checkerboard exclusion: distributions of three Melidecteshoneyochromelas;B, M. belfordi;R, M. sufocrissalissuperspecies. Most eaters. O, MeZidectes t two species with mutually exclusive altimountainous areas of New Guinea SUppO1 tudinal ranges. At each locality depicted on the map of New Guinea, the letters above and below indicate the species present at higher and lower altitudes, respectively. The identity of the locally missing thind species is subject to irregular geographical variation. tional gaps in a continuous habitat may persist for long periods of time, albeit as nonequilibrium phenomena. For example, the finch Lonchura montana (discussedabove3may have disappeared from the alpine grassland of central New Guinea severalmillenniaago, when the area of this habitat was reduced by encroachmentof forest; and this finch may dispersetoo slowly to have refilled the whole gap since then. In the more seasonal temperate latitudes, by contrast,the annualnorth-southmigrations, and the postbreeding wanderings of many nonmigratoryspecies, flood suitable habitats with potential colonists of most species each iyear. Mortality due to climatic fluctuations in temperate Iatitudesplaces a premium on dispersai ability to recolonize vacated territories, whereas the greater stability of the tropics selects against dispersal. Effects of interspecific competition on local distributionare easiest to recognize in the one-to-one situation of simple checkerboard exclusion, where species A, for example, occurs only in the absence of species B and vice versa, or in compound checkerboard exclusion, where species A occurs only in the absence of either species B or species C. The eompetitors in these situations are usually, t-houghnot always, close relatives within the same genus. It is more diicult to determinethe connections in so-calied diffuse competition,where the absence of a species is due to the combined efdfectsof many species, each somewhatdistantlyrelatedto the absent species and potentially overlapping it ecologically only in part (3, 10). A decline in species B following an invasion of its close competitorA may permit B's other close competitor C to increase, depressing the population of C's competitor D and ultimately affecting species ecologically far removed from A. Correlated with the greater diversity of species in the tropics than at higher latitudes is the fact that tropical species have more closely packed niches than their temperate counterparts, such that the local survival of a given species may be criticallydependent on the mix of competitorsas well as on the suitabilityof the habitat (3, pp. 231 ff3. Thus, competition may stabilize some distributionalgaps indefinitely. These considerations help us to understand the differing numbers of species on differentcontinents.Low dispersal may allow localized populations of disappearingspecies to linger, relatively undisturbedby influx of competitors, for long periods of time before final extinction. Conversely,low dispersal favors high rates of speciation, since the first stage of speciation depends on the effectiveness of geographical barriers or distributionalgaps (Fig. 8). Because the number of species at equilibrium on a continent depends on a balance between speciation rates and extinction rates, tropical habitatsshould be expected to have a greater number of species at equilibriumthan habitats at higher latitudes (Fig. 5). Furthermore, with increasing continental area, extinction rates should decrease (because of larger population sizes and more local refuges), and speciationrates should increase (because of more popu763 This content downloaded on Sat, 22 Dec 2012 20:00:09 PM All use subject to JSTOR Terms and Conditions _ . _ _ _ _ b Fig. 5. Speciation rates (solid curves) and E1 extinction rates (dashed curves3 on a JZ sa single land mass, as a inction of the l/ number of species (abscissa). The ordiU n?te is the tota1'number of species pro*ov ir duced through speciat-ionor lost through extinction per 10S000 years. The intersection of the curves (N1,^1V2) determines the number of species at equilibrium. The S2 speciation cu'rve'is approximately linear, o o / / / because the probability'of speciation of a given species'is appro:ximatelyindependent a.O of the presence' of other species but extinction rates increase sharply as more O / f species are packed,-together, ()n a large AS-_ */ * tropical land massn high' speciation 3nates N2 N:I: (curve Sl) and 1QW e,xtinctionrates (EI) Number of species yield' a high nvimberof species at equilibrium (N1). On a similar-'sizedtemperate lancl mass or a smaller tropical Iand mass, lower speciation rates (S2?=and higher exThe best understoodsegregating r mechatinction rates' (E2) yield fewer species at nism among New Guinea birdsdepends equilibrium ( N2 ) . their altitudinal range (Fig. 6). With increasingelevation C. mglrinabecomes increasinglyabundantuntil it-suddenly disappears at 1643 m an altitude not far above its altitude of masimum abundance.At this elevation C. rohsttz suddenly appears near its masimum abundance and becomes progressively less common toward Mt. KarinluiSs summit. Prolonged observationsat the transition altitude showed that the tWQ species ,are interspecifically territorial at 1643 mnand no individual of either species was ever found transgressingthe range of the other. There is no change in vegetationat this altitude the nearest i'ecotone' (a border 'between forest types) lies 338 m higher. In all, the New Guinea avifauna contains about on altitude and involves an inlteresting 45 pairs, 13 trios and 3 quartets of behaviolnal response. relatedspecies which ]replaceeach other . ,. Altitudinxxl segregation. O]n New altitudinally with similar abruptness. lations isolated over greater distances). Thus, one should also expect more Guinea mountains not distur rbed by For a given species pair the transition species at equilibriumin a large center man, forest extends uninterrupt :ed from altitude shows minor local variation of speciation, such as the Amazon sea level to timberline at around 3800 correlatedwith local conditionsof rainBasin, than in a small center, such as m. Many bird species especialElythose fall, exposure, and slope but no-sysNew Guinea (Fig. 5). witk no closely related species s in the tematicgeographicalvariationoverNew areanexhibit gradual changes iin abun- Guinea'. Altitudinal sequences with dance with altitude. However in many sharp transitions also occur among Ecological Segregationof Species other instances one finds seque ences of birds in the Peruvian Andes (21) and two, three or even four closel f rela-ted on all large mountainous islands of A central problem of ecology con- species replacing each other, abruptly Indonesia eind the southwest Pacific. cerns community- organization: How with altitude (15)* For examptleSCrStric,ttransitionsmay be vi,olatedby are resourcesdivided among the species teroseelis murina and CratterosceZis young birds: when one finds an indiof an ecological community and what robusta,two abundantand ver y similar vidual outside the normal altitudinal mutual arrangements permit the c o- warblersthat glean for insects ]near the range of its species it generally proves existence of closely relatedspecies (20)? ground dif3er ecologically m ainly in to be a juvenile or an immaturebird. Or X, a . §e /< . . . . . 2500 - 2000 Fig. 6. Altitudinal ranges of the warblers CraterosceZisrobmsta (*) and C. murina (0) on the'west ridge of Mt. KarimuiS New Guinea. On the left, eacRhmark represents one individual heard, seen, or collected at the given altitude''(the paucity of records at 650 to 1050 m results from my. having spent little time at this altitude). The right-hand side gives 'the relat;ve abundance in ' the whole avlfa.una-that is the percentage of bird individuals of all species estimated as being C. robust or C. m7lrin. The two species replace 'each other abruptly at 1643 m, ard each species reaches'its masimum abundance near tEs altitude. Many other .species show equally sharp'.transitions although the altitude of masimum abundance frequently differs frc>n the altitude 'at which the species transibion occurs .! - *; :: ::#S C. robusta C. murina - *. g g 1 SOO a 3 / ._ C5 o O [000 r oW 500- - L__, o 764 ]__ 1 t l I { _ 4 6 8 10 12 S0totai of bgrdindividuals 2 SCIENCE, VOI. 179 This content downloaded on Sat, 22 Dec 2012 20:00:09 PM All use subject to JSTOR Terms and Conditions Among New Guinea birds generally, there is a characteristicDaltitudinal dependence of population structure. Typically, immature birds are found in a fringe at the bottom of the altitudinalrange; somewhat higher, one finds immature birds plus nonbreeding adults, with females usually appearing at lower altitudes than males; next comes the optimal part of the species' range, with breeding adults of both sexes; and, finally, another fringe of immature birds but few adults appears at the upper altitudinal limit of some but not all species. This population structure is manifested to an extreme degree by some birds of paradise, whose displaying adult males may becompressed into the top 180 m of the altitudinal range, with adult females and especially immature birds being found up to 1000 m below the lowest adult male. UItimately the altitudinal sequences are related to differential adaptations that are continuous functions of altitude, one species being preferentially adapted to higher altitudes, cooler temperatures,and more montanevegetation than the other species. But the very sharpnessof the transitionsimplies that competition superimposes a special behavior pattern on these differential adaptations,since the transitionsdo not coincide with sharp changes in temperature or vegetation. That is, each species must be capable of surviving over a wider altitudinal range than it actually inhabitsand must be excluding related species from the range in which it is competitivelysuperiorto them. To confirm this interpretation, the local poptllation of one member of a sequence would have to be removed in order to ascertainwhether the adjacent species would expand their altitudinal ranges. This test of competition arises in New Guinea under two types of naturally occurring conditions. The first test of niche expansion associated with relief of competition occurs on small or isolated mountains or islands (8, 10, 15). Because of the dependence of species diversity on area and isolation and because of the random element in colonization, one or another member of an altitudinal sequence may be missing on such a "mountainisland."As shown in Fig. 7, under these circumstancesa low-altitude species may expand into the range of a missing high-altituderelative, a highaltitude species may expand into the range of a missing low-altituderelative, and a high-altitudespecies and a low23 FEBRUARY 1973 altitude species may expand simultaneously into the range of a missing middle-altituderelative. Reconstructionof the process of speciation provides the other test for competition and clearly demonstrates the origin of altitudinal sequences ( 15) . Speciation occurs when an initially continuous population breaks up into geographical isolates, the isolates diverge, reestablish geographical contact and perfect their reproductive isolating mechanisms, and finally reinvade each other's geographical ranges if there are sufficient ecological differences to permit coexistence (2). While the slowness of evolution rarely permits following a given species through this process with time, the distribution patterns of diSerent New Guinea montane species represent "snapshots" of seven different stages in a continuous speciation process (Fig. 8): 1) A single montane species extends from the western to the eastern end of New Guinea, oiccupying the same altitudinal range at all longitudes. 2) The local population in one area dies out, so that the east-west distribution becomes discontinuous. 3) The eastern and western populations, now isolated, diverge sufficiently that they would probably not interbreed if contact were established;hence, they are assumed to be distinct species. They may also develop slightly diSerent altitudinal ranges. 4) Both populations reexpand geographically until their geographiclal ranges abut but--donot overlap. There is no or little interbreeding, proving that the populations are in fact distinct species. S) Each species begins to expand geographically into the range of the other, so that there is geographical overlap ("sympatric" distribution) for a short distance. Within the zone of sympatry the two species segregate altitudinally,each being confined to the altitudinal range in which it is competitively superior to the other. The narrower altitudinal range inside than outside the zone of sympatry is a clear demonstration of niche compression due to competition. 6 ) Expansion continues, and the western species reaches the eastern end of New Guinea and overrunsthe entire geographicalrange of its eastern sibling species. The eastern species has given up the upper (or lower) part of its altitudinal range throughout its whole geographical range7 and there is no altitudinal overlap. 7) The eastern species continues to expand until it has reached the western end of New Guinea. The two species are now sympatric over the entire length of New Guinea, with mutually exclusive altitudinal ranges. Evolution Mt. 4000Mt. may then continue in the direction of Wilhelmina Mt. Michael either stage 8a, 8b, or 8c. Saruwaged 8a) Stages 1 through 7 may be repeated one or two more times to yield ,.X..'.'...i..'V..X.'T--T' series of three or four closely related ..'.-'.-."-.'-'ly.'S-iT-'..i: .T..T,i1..Xi:0f,00i; species sympatric over the whole of 3000-i.-0,-.-X-00-..,;i.X--0ft;: New Guinea but occupying mutuallyexti.'---iX'TS'-0.Xt-itA.,X,-t clusive altitudinalranges. ,.,i,,,i,,,,,,000t,t,,,,f,.; ..;;.;f X00.; TS,,;fE4i,!00X 8b) Two species that have become ,...t,--0f."t.,f-'T,',;d,'T,,lL; 3 sympatric with mutually exclusive 0'-'."i't-'TiX'T-i' iTE.-40;-0T'tT-'.iT.X.';.'tl.it altitudinal ranges may diverge in other zr: .0T,,-T,T,';ttEt;0'S-'t;00tl-07 2000niche parameters besides altitudinal -,000t,-;i..0f-,X,0X-0tT;t .t,,000;-Xti.-E,;i...-,;X..t.,. preference (for example7 they may Tit-'t;-S't"t'.'.;.-t;-.X:-;X.:00 iS.Xt; :ti ,0T ,:fE.t ..l, ly TlL,0 develop differentdiets or foraging tech-00t-'T-'T iT-'tt.ti".t;000.-i; niques), so that partial altitudinaloverlap as well as complete geographical 1000 / overlap becomes possible. 8c) Each species becomes genetically of molded to its compressed altitudinal Fig. 7. Altitudinal ranges of the honey- range to a degree such that the range eaters PtiZop)ora perst)iata (diagonally now reflects innate survival ability hatched bar) and P. guisei (solid bar) on three New Guinea mountains. On Mt. rather than competitive compression, a Michael, where both species are present, gap between the ranges of the two their altitudinal ranges are mutually ex- species develops, and neither species clusive. On Mt. Wilhelmina, where P. expands altitudinallyon removal of the guisei is absent, and on Mt. Saruwaged, other. where P. pe)striata is absent, the remainThis process seems to be the principal ing species takes over most or all of the mechanism by w'nichthe rich diversity altitudinal range of its missing relative. 765 - .. ....- - a) ._ 0.. .=L This content downloaded on Sat, 22 Dec 2012 20:00:09 PM All use subject to JSTOR Terms and Conditions - this overlap band the neighbor of a given individualwould as likely belong to the other species as to the same species, whereas in the distributionsof Fig. 6 each individual is surrounded by individuals czf the same species. In fact, altitudinal segregation of congeneric species on mountainousPacific Stage Stage islands with fewer than 66 species 30003000always involves broad overlap bands, C) 1 and sharp transitionsoccur only on the 2000 more species-richislands. This contrast z suggests two selective pressuresunderlying sharp transitions.First, in species1 000 {ooopoor areas the niche of a given spe cies is nearly as fbroad as permitted Oo] loy its intrinsic adaptationsand is little compressedby interspecificcompetitione 30003000 these conditionsthe success oiTa 7 Under 2 dispersingjuvenile in colonizing a new 2000 territory will depend mainly on visible 2000features of the environment.In species1000rich areas,'however,the task-of identi1000fying a territory that offers a high probability of breeding success for a Ogiven species becomes much more -3000difficult, since success now depends 30008a critically on the presence and abun3 dance of many other competingspecies. a 2000This in effect would requirea dispersing s 1 M 3 juvenile to perform a faunal survey in 111111 10001000addition to merely judging the appear: ance of the habitat. Second, a selective OOpressure may arise from the difficulty of finding mates in a species-richcom3000- munity where many species are rare. 30008b Thus, I postulate that, on species-rich 4 2000islands but not on species-poorislands, 2000dispersingyoung birds are programmed 1000or they learn by early genetically, 10P0experience, to seek out habitats where their species is already established, OOrather than habitatswhere they will be surrounded by individuals of other n 30Q0300018c species. The presence of individualsof 1 5 their own species becomes the only 20002000reliable indicator that the habitat is suitable for them. By waiting in the juvenile fringe or "kindergarten"at 1 1000the lower or upper limit of the f | ' 1 O I altitudinal range, the juveniles are in o 132 138 144 150 a ringsideposition from which they can 132 138 444 150 eventually seize an optimal territory vacated by an adult. Longitude (°E) h Otherspatialsegregatingmechanisms Fig. 8. Stages in the evoltltionary transformation of one species into two species with further ecological segregating ,I Two mutually exclusive altitudinal ranges. See text for details. In each diagram altitudina] range is plotted as a function of longitude for some species or pairs of species in thee mechanisms, vertical stratificationand t1 habitatsegregation,resemblealtitudinal mountains of New Guinea. Thus, in stages 1 through 4 there is neither geographica] nor altitudinal overlap, in stages 5 through 8a and 8c there is geographical but noh segregation in that closely related altitudinal overlap, and in stage 8b both geographical and altitudinal overlap. Eackd species become sorted out spatially. such distributionalpattern representsa "snapshot"of a diSerent stage in speciation and permits reconstruction of the whole process. Examples of species in each stage are:: Bird species are more stratified 1, Ifrita kowaldi; 2, Climactesis leucop11aea;3, Melidectes nouXuysi and M. princepsi; vertically in the New Guinea forest poifosoma and P. 1zattamensis;6 4, ..Parotialawesi and P. carolae; 5, Pac11ycep1lalopsis I' than in forests with fewer bird species Ael blyo)nis macg}egoriae and A . subala}is; 7, Ducula rufigaste} and D. c11alsonota but similarvegetationalstructure.Thus, 8a, Eupetes caerulescens, E. castanonotus, and E. Ieucostictus; 8b, Neopsittacus mus standard bird nets 2 m in height and schenbroekii and N. pullicauda; 8c, MeZiphagaanaloga and M. orienfalis. of bird species in the mountains of l!Jew Guinea evolved out oiTthe lowlands (15). What is the significance of sharp altitudinal transistions? If there were o special behavioral mechanism pro- ducing such transitions, one would expect two competing,territorialspecies with differentaltitudinaladaptationsto replace each other in an overlap Iband which would consist of a mosaic of territories of the two species (2Z). ln fW - - ._ _ , S -t dooo-sts5ssIIsIazIIIIIIIIIIIIIID SCIENCE, VOL. 179 766 This content downloaded on Sat, 22 Dec 2012 20:00:09 PM All use subject to JSTOR Terms and Conditions resting on the ground catch only about half of the forest species locally present in New Guinea montane rain forest but all species in the species-poorrain forest of New Zealand. Some of the species in New Guinea forage only in the canopy; others forage regulalrlyat 4 to 15 m above- the ground but never descend to 2 m and hence are never caught in nets; and others forage only on the ground or up to a height of 2 m. As in the case of altitudinal ranges, vertical foraging ranges expand on species-poor mountains or islands where vertically abutting competitors are absent (10) . Some congeners become sorted by occupying different habitat types, often to the mutual exclusion of each other. For example, whereas there is only one species of barn owl (genus Tyto) in most parts of the world, New Guinea has three species: Tyto capensis in grassland, T. alba in partly wooded areas, and T. tenebricosa in forest. Among species sorting by habitat, spa-tial expansion is a frequent response to the absence of competing congeners on islands (10, 11). Expansion of New Guinea second-growth species into forest on islands is especially common, since New Guinea forest species are frequently absent on islands because of their poor dispersal ability concentratedtowardthe main branches, the smaller birds toward the periphery (Fig. 9). Among congeners sorting by size in New Guinea, the ratio between the weights of the larger bird and the smaller bird is on the average 1.90; it is never less than 1.33 and never more than 2.73. Species with similar habits and with a weight ratio less than 1.33 are too similar to coexist locally (that is, to share territories)and must segregate spatially.For instance, the cuckooshrikes Coracina tenuirostris and C. papuensis segregate by habitat on New Guinea, where their average weights are 73 grams and 74 grams, respectively, but they often occur together in the same tree on New Britain, where their respective weights are 61 g and 101 g. New Guinea has no locally coexisting pairs of species with similar habits and with ia weight ratio exceeding 2.73, presumably because a 163 123 _< /t Fruit (mm) / <L<_ A 7 414 -L &, _ jt/ & 20 Nonspatial SegregatingMechallisms Spatial overlap of closely related species is possible if they separate on the basis of time, diet, or foraging techniques. Infrequently, closely related bird species segregate by occupying the same space at different times of day or of the year. The kingfisherMelidora macrorhinais nocturnal,whereas other kingfishersare diurnal. The south New Guinea savanna near Merauke is alternately occupied by two marsh hawks, Circus approximansin the dry season and C. spilonotus in the wet season. Differences in body size provide the commonest means by which closely related species can take the same type of food in the same space at the same time (15). Larger birds can take larger food items than can smaIler birds, but smallerlDirdscan perch on more slender branches than can larger birds. One can frequently see a bird foraging out along a branch up to the point where the branch begins to bend under its weight. In a tree occupied by birds of many species the larger birds are often 802 slAf-.,Lz'> t40 Fig. 9. Schematic representation of niche reIations among the eight species of Ptilislopus and Ducula fruit pigeons in New Guinea lowland rain forest. On the right is a fruit of a certain diameter (in millimeters), and on the left are pigeons of different weights (in grams) arranged along a branch. Each pigeon weighs approximately 1.5 times the next pigeon. Each fruit tree attracts up to four consecutive members of this size sequen-ce. Trees with increasingly large fruits attract increasingly large pigeons. In a given tree the smaller pigeons are preferentially distributed on the smaller, more peripheral branches. The pigeons having the weights indicated are: 49 g, Ptilinopus nanus; 76 g, P. pulc11ellus;123 g, P. supesbus; 163 g, P. orl1atus; 245 g, P. pe11atus; 414 g, Ducula sulSga*ter;592 g, D. zoeae; 802 g, D. pinon. medium-sized bird of relative weight +/2.73 = 1.65 can coexist successfully with both the large species and with the small species. Thus one finds a sequence of three or more species rather than just two species of such different sizes. For example, the eight fruit pigeons of the genera Ptilinopus and Ducula coexisting in the lowland forests of New Guinea form a graded size sequence over a 16-fold range in weight (Fig. 9). May and MacArthur (6) predicted on theoretical grounds that sprecies segregatingalong a single niche dimension in a Ructuatingenvironment must maintain a certain minimum niche difference. This minimum spacing seems in fact to have been reached in nature by those bird species that segregate according to size. Thus, on Pacific islands with 30 to 50 bird species, the ratio between the weights of pairs of birds segregatingby size is approximately 4, but this ratio has already been compressed to 2 on islands with 100 species. On New Guinea (513 species) the average value of this ratio is compressedno further, and the extra species are accommodated by expanding the size sequences to smaller or larger birds or else by finer subdivision of space or foraging techniques. Similarly sized species that take the same food may overlap spatiallyif they harvest the food in different ways. For instance,small insectivorousbirds differ in tactics according to whether insects are caught in midair by sallying, are pounced on and plucked off surfaces, are gleaned off surfaces, pried out of bark, taken from flowers, or are extracted from epiphytes and accumulations of dead leaves. Species with a given type of strategy further differ in the ratio between traveling time and stationary time, in the frequency of movements, and in their average rate of travel (23). Thus, the montane Rycatchers Pachycephala modesta and Poecilodryas albonotata differ in that the former remains perched for an average of 2 seconds, the latter for an average of 30 seconds between moves; and in that the former travels 1 m, and the latter 12 m, per move. Pachycephala modesta could be described as a quick and cursory searcherz Poecilodryas albonotata as a slow and seI>ctive searcher. Finally, relatedspecies may segregate by different diets. For example, the whistler Pachycephala leucostigma eats mainly fruit while other whistlers eat . maln y lnsects. 23 FEBRUARY 1973 767 This content downloaded on Sat, 22 Dec 2012 20:00:09 PM All use subject to JSTOR Terms and Conditions Some General Features of Competition increase linearly or even more rapidly with species diversity,so that an island Despite the abundant distributional with few species also has a low density evidence for competition between New of individuals (11). However, on the Guinea bird>;species, one rarely sees Pearl Islands oS Panama, total populaa member of one territorial species tion densities of birds are even higher fightinga memberof anotherterritorial than on the Panama mainland l(25). species. Once territoriesare established, Part of the difference between these fighting simply does not pay: the win- two sets of results may depend on ner as well as the loser may be injured, whether the island colonists are as or both combatants may attract the well adapted to the available habitat attention of a predator. Even among as are the mainland species they remigrant North American thrushes that place. Also, the low populationdensities must reestablish their territories each on the New Guinea satellite islands, spring, interspecificchases and aggres- and on old isolated islands such as sive behavior disappear within a week Madagascar (26) and New Zealand, after arrival on the breeding grounds may reflectgenetic deteriorationin their (24). In tropicalrain forest, where many isolated populations, because of small bird species are permanently resident gene pools, reduced intraspecific and and are relativelylong-lived, and where interspecific competition, and (frevisibility inside the forest is poor, quently)short populationsurvivaltimes neighboring individuals may confine to extinction (1.1). their aggressive behavior for years to songs and calls a.nd simply learn to Conclusions avoid each others foraging space. Species that colonize species-poor The concepts by which MacArthur islands, where they are freed of competition from close relatives, often and Wilson have transformed the broaden the spatial parameters of the science of ecology in the past decade, niche by immedia.telyoccupyinga wider and the results of ecological studies range of habitats, altitudes, or vertical such as mine on New Guinea bird foraging positions. However, the col- communities, have implications for onists rarely expand their diets or conservation policies. For example, range of foraging techniquesuntil after primary tropical rain forest, the most relatively long periods of time on an species-rich and ecologically complex evolutionary scale (11). This combina.- habitat on earth, has for millions of tion of spatial elasticity with tactical years served as the ultimate evoluand dietary conservatismreRectspartly tionary source of the world's dominant the degree of genetic programming plant and animal groups. Throughout underlying the stereotyped foraging the tropics today, the rain forests are strategies of birds, and partly the being destroyed at a rate such that economics of feeding. Given a certain little will be left in a few decades. set of inflexible tactics, the economics When the rain forests have been reof energy yield and energy expenditure duced to isolated tracts separated by in foraging dictate the diet but leave open country, the distributionof oblito the individualthe decision about the gate rain forest species will come to space in which the tactics can be applied resemble bird distributions on New Guinea land-bridgeislands after severprofitably(3, chap. 3). How does the combined population ing of the land bridges.The smaller the density of all bird species on an island tract, the more rapidly will forest compare with the combined population species tend to disappear and be redensity on a mainland?That is, istands placed by the widespread secondhave fewer species than mainla.nds,but growth species that least need protecthe types of competitive release we tion (13). This ominous- process is have examined mean that the island illustrated by Barro Colorado Island, colonists frequentlyhave broaderniches a former hill in Panama that became and higher populationdensitiesthan on an island when construction of the the ma.inland.How well does competi- Panama Canal flooded surrounding tive release compensate on islands for valleys to create Gatun Lake. In the the population densities of missing succeeding 60 years several forest bird mainland species? Studies on this species have already disappearedfrom problem of density compensationhave Barro Colorado and been unable to provided conSictingresults. In the New recolonize across the short intervening Guinea area, total population densities water gap from the forest on the nearby in similar habitats on different islands shore of G;atunLake. 768 The consequences of the speciesarea relation (Fig. 1) should be taken into considerationduring the planning of tropical rain forest parks (13). In a geographical area that is relatively homogeneouswith regardto the fauna, one large park would be preferable to an equivalent area in the form of several smaller parks. Continuousnonforest strips through the park (for example, wide highway swaths) would convert one rain forest "island" into two half-size islands and should be avoided. If other considerationsrequire that an area be divided into several small parks, connecting them by forest corridors might significantly improve their conservation function at little further cost in land withdrawn from development. Modern ecological studies may also be relevant to the understiandingof human populations. For instance, during a long period of human evolution there appear to have been not one but two coexistent hominid lines in Africa, the AustralopithecusrobustusA . boisei l("Zin janthropus') line, which became extinct, and the A ustralopithecusafricanus-A. habilisline, which led to Homo sapiens (27). The need to maintain niche differences between these lines must have provided one of the most important selective pressures on the ancestors of modern man in the late Pliocene and early Pleistocene. Thus, any attemptto understandhuman evolution must confront the problemof what these ecological segregating mechanismswere. To what extent were contemporaneous species of the two lines separated by habitat, by diet, by size difference, or by foraging technique, and were their local spatial distributions broadly overlapping or else sharpenedby behavioralinteractionsas in the case of the Craterosceliswarblers of Fig. 6? To take another example, there are striking parallels between the present distributionsof human populations and of bird populations on the islands of Vitiaz and Dampier straits between New !Guineaand New Britain. Some of these islands were sterilized by cataclysmic volcanic explosions within the last several centuries. The birds that recolonizedthese islandshave been characterizedas coastal and smallisland specialists of high reproductive potential, high dispersal powers, and low competitive ability, unlike the geographically closer, competitively superior, slowly dispersing, and breeding birds of mainlandNew Guinea.(10, 11, 13). It remains to be seen whether This content downloaded on Sat, 22 Dec 2012 20:00:09 PM All use subject to JSTOR Terms and Conditions SCIENCE, VOL. 179 15. the people of the Vitiaz-Dampier islands, the Polynesians, and other humanpopulationsthat colonize insular or unstableh;abitatsalso have distinctive poptllationecologies. 6. 7. 8. References and Notes 1. A. R. Wallace, The Geographical Distribution of Animals (Macmillan, London, 1876). 2. E. Mayr, Systematics and the Origin oJ ,Species (Columbia Ulliv. Press, New York, 1942); Arlimal Species and Eltolution (Harvard Univ. Press Cambridge, Mass., 1963). 3. R. H. MacArthur, Geographical Ecology (Harper & Row, New York, 1972). 4, - -- and E. O. Wilson, The T}teory ot [sland Biogeogra phy (Princeton Univ. Press, Prunceton, N.J., 1967). 5. Is Elfolution 17, 373 (1963); R. H. MacArthur and R. Levins, Proc. Nat, Acad. Sci. U.,S.A. 51, 1207 (1964); R. -H. MacArthur, Biol. Rev. Cambridge 40, 510 (1965)-- , H. Recherfi M. Cody, Amer. Natur 100, 319 (1966?; R. H. MacArthtlr and E. Pianka, H* 9. 10. 11. 12. 13. 14. - Avifauna of the Eastern Highlands ibid., p. 603; E. O. Wilson, Evolution 13, 122 (1959); Amer. Natur. 95, 169 (1961); and D. S. Simberloff, Ecology 50, 267 (1969). R. M. May and R. H. MacArthur, Proc. Nat. Acad. Sci. U.S.A. 69, 1109 (1972). E. O. Wilson, Psyche 65, 26 (1958). J. M. Diamond, A mer. Mus. Nol . No. 2284 (1967); Explorers J. 46, 210(1968); and J. W. Terborgh, Auk 85, 62 (1968); J. M. Diamond, Amer. Mtes. Nos!. No. 2362 (1969); J. Terborgh and J. M. Diamond, Wilson Bull. 82, 29 (1970). J. M Diamond, Proc. Nat. Acad. Sci. U.S.A. 64, 57 (1969). -, ibid. 67, 529 (1970). , ibid., p. 1715. -, ibid. 68, 2742 (1971). , ibid. 69, 3199 (1972). , Science 151, 1102 (1966). ot New Guinea (Nuttall Ornithological Club Cambridge, Mass., 1972). 16. F. W. Preston, Ecology 43, 185 (1962). 17. J. Terborgh and J. Faaborg, Atlk, in press. 18. R. F. Fl}nt, Glacial and Pleistocene Geolog> (Wiley, New York, 1957). 19. F. Vuilleumier, Amer, Natxr. 104, 373 (1970); J H. 13rown ihid. 105, 467 (1971). e umanlzlng the Earth Rene J. Dubos How gray and drab, unappealingand which were most Rourishingin antiqunsignificant,our planet would be with- uity are now among the poorest in the out the radiance of life. If it wereCPnot world. Some of their most famous cities covered with living organisms the sur- have been abandoned;lands which were face of the earth would resemble that once fertile are now barren deserts. of the moon. Its colorful and diversified Disease, warfare,and civil s;trifehave appearance is largely the creation of certainly played important roles in the microbes, plants, and animals which collapse of ancient civilizations;but the endlessly transform its inanimate rocks primarycause was probablythe damage and gases into an immense variety of caused to the quality of the soil and to organic substances. Man augments still water supplies by poor ecological pracfurther this diversification by altering tices. Similarly today, the environment the physical characiterIstics of the land, is being spoiled in many parts of the changing the distribution of living world by agricultural misuse or overthings, and adding human order and use, by industrial poisoning, and of fantasy to the ecologtcal determinism course by wars. of nature. The primary purpose of the recent Many of man's interventionsinto na- United Nations Conference on the Huture have, of course, been catastrophic. man Environment, held in Stockholm History is replete with ecological dis- in June 1972, was to formulate gIobal asters caused by agriculturaland indus- approaches to the correction and pretrial mismanagement. The countries vention of the environmental defects resultingfrom man's mismanagementof The author is a professor emeritus of the the earth. I shall not discuss the techniRockefeller University, New York 10021. This article is the text of the B. Y. Morrison MemoriaI cal aspectsof-these problems,but rather lecture, sponsored by the Agricultural Research Service of the U.S. Department of Agriculture. shall try to look beyond them and preThe lecture was delivered at the annual meeting sent facts suggesting that man can acof the AAXS, 29 December 1972, in Washington D.C. Reprints may be obtained from Dr. Robert tually improve on nature. In my opinNelson, Public lnformation Office, Agrlcultural ion, the human use of natural resources Research Service, U.S. Department of Agriculture Washington, D.C. 20250. and of technology is compatible with 2(). D. Lack, Ecol*geial Isolation in Birds (Harvard Univ. Press, Cambridge,Mass., 1971). 21. J. W. Terborgh,Ecology 52, 23 (1971). 22. M. L. Cody, ibid. 51, 455 (1970). 23. ; --I s Anzer. Ncxtltr. 102>t07 (19683. 24. D. H. MorsefiWilson Bull. 83, 57 (1971). 25. R. H. MacArthur, J. M. Diamond, J. R. Karr, Ecology 53, 330 (1972). 26. J. R. Karr, personalcommunication. 27. D. Pilbeam, The A scent of Man (Macmillan, New Yorks 1972); L. S. B. Leakey, Nalure 209, 1279 (1966), F. C. Howell, ibid. 223, 1234 (1969); W. Schaffer, Amer. Natllr. 102, 559 ( l 968). Z8. l thank R. H. MacArthurand J. W. Terborgh, for stimulatingdiscussions;the National Geographic Society, Explorers Club, American Philosophical Society, Chapman Fund, and Sanford Trust of the American Museum of NaturalHistory, and Alpha Helix New Guinea Programof the National Science Foundation, for support; M. Cody, A. Grinnell, S. Kaufman-Diamond,S. Krasne, and G. Szabo, for criticismof the manuscript;D. Amador,for permission to use facilities of the American Museum of Nattlral History; and more residents of New Guinea than can be mentioned hy name. tor makinvficldworkpossible. ecological health, and can indeed bring out potentialitiesof the earth which remain unexpressed in the state of wilderness. The disastrousecological consequences of many past and present human activities point to the need for greater knowledge and respect of natural laws. This view is succinctly expressed by Barry Commoner in his fourth law of ecology: "Nature knows best." I shall first discuss the limitations of this law. When left undisturbed, all environments tend toward an equilibriumstate called the climax or mature state by ecologists. Under equilibrium conditions, the wastes of nature are constantly being recycled in the ecosystem, which becomes thereby more or less self-perpetuating. In a natural forest, for example, acorns fall to the ground and are eaten by squirrels,which in turn may be eaten by foxes or other predators; the dead leaves and branches the excrements of animaIs, are utilized by microbes, which return their constituents to the soil in the form of humus and mineral nutrients. More vegetation grows out of the recycled materials, thus assuring the maintenance c)f the ecosystem. When applied to such equilibrated systems,the phrase "Natureknows best' is justified, but is in fact little more than a tautology. As used in this phrasesthe word nature simply denotes a state of affairs spontaneouslybrought about by evolutionary adaptation resulting from feedbacks which generate a coherent system. There are no problems in undisturbed nature; there are only solutions, precisely because the equilibrium state is an adaptive state. 23 FEBRUARY 1973 - This content downloaded on Sat, 22 Dec 2012 20:00:09 PM All use subject to JSTOR Terms and Conditions 769