Dynamics of Regional Distribution: the Core and Satellite Species Hypothesis Author(S): Ilkka Hanski Reviewed Work(S): Source: Oikos, Vol

Nordic Society Oikos

Dynamics of Regional Distribution: The Core and Satellite Species Hypothesis Author(s): Ilkka Hanski Reviewed work(s): Source: Oikos, Vol. 38, No. 2 (Mar., 1982), pp. 210-221 Published by: Wiley on behalf of Nordic Society Oikos Stable URL: http://www.jstor.org/stable/3544021 . Accessed: 19/01/2013 18:53

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].

Wiley and Nordic Society Oikos are collaborating with JSTOR to digitize, preserve and extend access to Oikos.

http://www.jstor.org

This content downloaded on Sat, 19 Jan 2013 18:53:43 PM All use subject to JSTOR Terms and Conditions OIKOS 38: 210-221. Copenhagen1982

Dynamicsof regionaldistribution: the core and satellite specieshypothesis

Ilkka Hanski

Hanski, I. 1982. Dynamics of regional distribution:the core and satellite species hypothesis.- Oikos 38: 210-221.

A new concept is introducedto analyse species' regionaldistributions and to relate the patternof distributionsto niche relations. Several sets of data indicate that average local abundance is positivelycorrelated with regional distribution,i.e. the fractionof patchily distributedpopulation sites occupied by the species. This observationis not consistentwith the assumptionsof a model of regionaldistribution introducedby Levins. A correctedmodel is now presented,in whichthe probability of local extinctionis a decreasingfunction of distribution,and a stochasticversion of the new model is analysed. If stochasticvariation in the rates of local extinction and/orcolonization is sufficientlylarge, species tend to fall into two distincttypes, termedthe "core" and the "satellite"species. The formerare regionallycommon and locally abundant,and relativelywell spaced-outin niche space, whileopposite attri- butes characterizesatellite species. This dichotomy,if it exists,provides null hypo- thesesto testtheories about communitystructure, and it mayhelp to constructbetter structuredtheories. Testing the core-satellite hypothesis and itsconnection to the r-K theoryand to Raunkiaer's "law of frequency"are discussed.

I. Hanski,Dept of Zoology, Univ.of Helsinki,P. Rautatiekatu13, SF-OO100 Helsinki 10, Finland.

HpegnaraeTcaq HoBc-m KOoHqeriL;HXen aHas3a permoHamHoro pacrnpreneHM BK- ,pB H ;flH CPaBHeEHm XapaKTepa pacnPEreneH~m C cooTHaIceHeeM HH . HecKam- i(o cepg XgaHmx noKa3anH, EITO BenH'IHa CpHee TI1OKaWMHOA 'iCJeHHOCTh no- JIO2KTeJHO KoppenpyeT C peMOHaWHE&M PaCrqOCTpaHeHfeM, TO eCTh C OTHO- CKTOMlhMM KOIH'IeCTBCM M3atHHO paCrI IfCDKeHX BHXPBbIX CTaV4lH, 3aHabENX gaH- HaiM BRg9t. 3To HajureHae He cOoTBecTByeT rre-mI perlHoHaTbHoro pacrrpae- YieHHH,npancKeHHoVt fleiHcct. 3gecb ripeanaraeTcq HcrpaB.UeHHaHmcGemu'I, B 1(OTOpOOAo6CyxmaeTCq BepOHTHOCT nOKaJhHOrrOI lc'e3HOBeHmH Bfag, RaK ymeHb- KI1aCgdI yHopacKpcTpaHe H cToxacTmecKag BepCHa HOBOg MXenH. Ecnu cToxacTfqecmce Kone6aHf cKppocTek TioKanihHoro Hc'e3HoBeHHH H/lw K1nOHR3aLWH oPpcTaTo'HOrO BendKH, rrpoawmeTcq TeH~eHLUM pa3geneH5 BHX0B Ha xna YemTKoX TrIa, Ha3mBaev "lOCHOBHbml H '"caTemTFmml4. fepmie OATHS1 B CBoeM perliOHe, J1OKWTIEHO MHOrOtIHCneHHU, a CaTeJITHTIe Burl XapaKTepw3y- H0rCSI npoTHBOhincmKi*4 npH3HaKamx. 3Ta HXoT48 ecJI oHa cyAiecTByeT, no3- BOjmeT HcrIoInb3oBaTh HY.1b-rHnOTe3y lNp rIPOBepKH TeOPHH cTpyKTybi cOOgeCTBa H OHa IKeT IICMOYb B Co3gaHwH 6anee COBepMHHOR TeOpMH. O6cyxgla=31 pe- 3YJhTaThi ripoBepKH rmHOTe3bi "OCHOBmx-CaTennHTHtx1x B=OB H ee CBH3H C r-K TeopHekH 6HoTEonwiecKHmcnewtpam PayHmoepa.

Accepted 26 March 1981 ? OIKOS 0030-1299/82/020210-12 $ 02-50/0

2 10 OIKOS 38:2 (1982)

This content downloaded on Sat, 19 Jan 2013 18:53:43 PM All use subject to JSTOR Terms and Conditions itiatedby MacArthur(summarized in his 1972 book). 1. Introduction Whateverview one holds on the importanceof com- It has been popularto defineecology as the studyof petitive and other biotic interactionsin structuring abundanceand distributionof organisms(e.g. An- communities,it is an indisputablefact that communities drewarthaand Birch1954, Krebs 1972), to theextent consist of differentkinds of species: some are widely thatMacArthur and Wilson (1967) statethere is no real distributedwhile othersoccur patchily;there exist lo- distinctionbetween ecology and biogeography.For the cally abundant and locally rare species; and in some purposesof this paper, I defineabundance as the communities species are, at least apparently, well numberof individualsat a local populationsite (for spaced-out in niche space, while in other communities otherdefinitions see Hengeveld1979). Thereare good guildsof similarspecies coexist.One is temptedto pose reasonsto expressabundance as a fractionof the possi- the question: Is it possible to findunifying factors to ble maximumnumbers sustainable at the site (An- simplifythis diversity? drewarthaand Birch1954), but thismay be difficult I suggestsome narrowingdown of this question. It particularlywhen dealing with multispeciescom- willfirst be shownthat dynamics in local abundance and munities.Distribution refers to thenumber of popula- regionaldistribution are interdependent.Incorporating tionsites occupied by the species;this again may be thisobservation into the typeof models of regionaldis- givenas thefraction out of thesuitable ones withinan tributionsuggested and firstanalysed by Levins (1969a, arbitraryor naturalregion. Population sites can be dis- 1970, Levins and Culver 1971) leads to an important creteunits, like true islands; or, like habitat islands, they structuralchange in the basic model. The key question mayhave been delimited more arbitrarily from the rest in the analysisof the revisedmodel is whetherthere is ofthe environment; or theymay be contagious,in which an internalequilibrium point on the distributionscale, case distributionis simply the proportion of totalarea which most species are approaching,or whetherthe occupied.This definition of distribution does not specify species are just heading towards either maximal dis- the typeof spatialpatterns, i.e. the locationsof the tributionand superabundance,or regionalextinction. (occupied)sites in space, which is a relatedbut different question. Theoreticalecology has largelymodelled local abun- 2. Local abundanceand regionaldistribution are dance(e.g. May 1976),while distribution has beenleft, interdependent untilrecently, to biogeographers,with the exception of the largelydescriptive statistical work on animaland Four examples fromdifferent invertebrate taxa are put plantdistributions (Patil et al. 1971, Bartlett1975, forwardto answer the question, are local abundance Tayloret al. 1978, Ord et al. 1980). An exceptionto and regional distributionindependent of each other? thisrule is Levins's(1970, see also 1969a) modelon The answeris no. extinction,which has been followedby a numberof studieson interspecificcompetition (Cohen 1970, Le- vinsand Culver1971, Horn and MacArthur1972, Le- vin1974, Slatkin 1974, Hanski 1981a; see alsoSkellam 1951) andpredation (Vandermeer 1973, Zeigler 1977) C 2- in patchyenvironments, all of which apply Levins's ap- M 0 00 proachto regionalpopulation dynamics and underline C thedifference between local and regionalinteractions. . / The spatialaspect of populationinteractions has re- centlyreceived increasing attention (reviewed by Levin 1976; see also Smith1974, Levin 1977, 1978,Gurney and Nisbet1978a, b, Taylorand Taylor1977, Crowley 1979,Comins and Hassell 1979, Hanski 1981b), and it has becomeclear thatunderstanding of both spatial processes(distribution of thespecies in physicalspace) and resourcepartitioning (distribution of thespecies in nichespace) are essentialcomponents to a satisfactory explanationof theperennial questions: Why are there 5 10 so manyspecies? Why are thereso manyrare species? number of sites occupied (Wiens 1976, Yodzis 1978, Hanski 1979a). Indeed, Fig. 1. Relationshipbetween average local abundance and dis- someecologists (e.g. Simberloff1978) havegone so far tributionin Anasiewicz's (1971) data on bumblebees from as to maintainthat, in manyor mostcases, spatial Lublin, Poland. While calculating average abundance only dynamicsin independentlydeveloping populations those sites were includedfrom which the species was collected explainmost of the (note logarithmicy-axis). Distributionis the numberof sites, "communitypatterns" (see also maximally10, occupied by the species. Each dot in thisfigure Caswell 1976). This contrastswith the approachin- representsone species (the line has been drawnby eye).

14* OIKOS 38:2 (1982) 211

This content downloaded on Sat, 19 Jan 2013 18:53:43 PM All use subject to JSTOR Terms and Conditions each community,a positivecorrelation exists between A B average local abundance and distribution(Fig. 2). The thirdexample is fromKarppinen's (1958) study on soil mites (Oribatei) in two foresttypes in North Finland.In bothhabitats, a positivecorrelation between I S. SI abundance and distributionis apparent(Fig. 3). The final example is frommy studies on dung and / 0~~~~~~~~~~'~ 1 carrion beetles in lowland rain forest in Sarawak * D j E F; (Hanski unpubl.). Trapping was carried out with 10 traps for 4 nightsat 12 sites, situatedat least 0.5 km fromeach other in homogeneous virginforest. I have restricted the analysis to the species-rich genus Onthophagus (Scarabaeidae). Once again, a positive

2 4 10 2 4 6 10 2 4 6 80 1 correlationexists between the numberof trappingsites number of sites occupied from which the species was caught and the average Fig.2. Relationshipbetween average local abundance and dis- catch from one site (Fig. 4). I conclude from these tribution,like in Fig. 1, in Kontkanen's(1950) data on examples that a correlationbetween abundance and leafhoppersfrom meadows in East Finland.Each dot in this distributionis the rule in nature. figureis theaverage for several species, which had thesame It is beyondthe scope of thispaper to go into details distribution.Figures A to F referto different"communities", representingwet to drymeadows (from left to right)at early about the causes of this relationship,but it may be (upperrow) and latesummer (lower row). pointedout thatthe level of between-sitemovements is clearlyof crucialimportance. It appears to be common in naturethat emigration takes place muchbefore local carryingcapacity has been reached,perhaps because of Anasiewicz (1971) studiedbumblebees in the parks, reasonsdiscussed by Lidicker (1962) and Grant(1978). squares, lawns, etc. of Lublin in Poland - all good Datum pointsin Figs 1 to 4 resultfrom sampling, but examples of discretehabitat islands in the man-made because both abundance and distributionare under- environment.Average local abundance increasedwith estimated,an increasein sample size should not change thenumber of sitesfrom which the species was recorded the picturequalitatively. There are, of course,truly rare (Fig. 1; only sites in whichthe species was presentare yet widelydistributed species, like the crane Grus grus includedin thecalculation of average local abundance). in Finnishmarshlands (Jirvinen and Sammalisto1976), Kontkanen (1950; see also 1937, 1957) sampled but if communitiesconsisting of reasonably similar leafhoppersfrom meadows in East Finland. He deli- species are studied,true distributionis expected to be mited six "communities"of coexistingspecies, and, in correlatedwith true average abundance. The contrary

3- 10-

00~~~~~~~~~~~ C 0 0~~~~~~~~ C co

* * / con5 X ~~~~~~~~~~St *- 0 0 0 0 C~~ I 0

6 10 15 20 55 30 35 40 45 50 numberof sites occupied 05 10 number of sites occupied Fig.3. Relationshipbetween average local abundance and distribution,like in Fig. 1, in Karppinen's(1958) data on soil Fig.4. Relationshipbetween average local abundance and dis- mites(Oribatei) from two forest types in NorthFinland. Each tribution,like in Fig. 1, inOnthophagus (Scarabaeidae) in the dotin thisfigure is theaverage for several species, which have alluvialforest in Sarawak(Hanski unpubl.). A totalof 12 sites beengrouped into 10 distributionclasses (total number of sites was studiedin homogeneousvirgin forest. Each dot in this was50 inboth forest types). The twokinds of symbols refer to figurerepresents one species (note that y-axis is not thetwo forest types. logarithmic).

212 OIKOS 38:2 (1982)

This content downloaded on Sat, 19 Jan 2013 18:53:43 PM All use subject to JSTOR Terms and Conditions would require spatial variance to decrease by a factor C greater than approximatelythe squared proportional C) decrease in average abundance,which most certainly is not the case at least in mothsand aphids (Taylor et al. 0 1980, see also 1978). Acceptingthat spatial varianceis proportionallyas large in rare as in abundantspecies, local extinctionsare bound to occur (MacArthurand 0 L Wilson 1967), and distributionis, at least in rare a : species, less than the possible maximum.

1 2 3 4 5 8 7 8 9 number of sites occupied 3. Modelsof distribution Fig.5. Relationshipbetween the probability oflocal extinction The paradigm for the dynamicsin local abundance is (per year) and distributionin Simberloffs(1976) data on stillthe logisticequation (Lotka 1925) mangroveisland insects. Maximum number of sites (small mangroveislands) was nine.Each dot representsthe average forseveral species (the line has beendrawn by eye). dN = rN(1-N/K), (1) dt while perhaps the only similarlygeneral model pro- 1-(e+oe)/i. If (5) does not hold, )(p) is a decreasing posed fordistribution is Levins's (1969a) functionbetween 0 and 1. The deterministicequilib- rium,obtained when i2 = 0, is alwaysequal to or greater than the stochasticmode. It = ip(l-p)-ep, (2) should be noted that dt thereare two interpretationsfor D(p) (Kimura 1964). D(p) gives the distributionof p both fora singlespecies where p, a measure of distribution,is the fractionof duringa long period of time, and for a communityof population sites occupied by the species (0 - p ' 1), similarspecies at a givenmoment. and i and e are constantsfor a given species in a given The assumptionthat all the local populationsare the environment.The firstterm in thisequation is the rate same, implicitin model (2), is veryunrealistic. To model of colonizationof emptysites, and the second termis local dynamicsat each populationsite explicitlyis out of the rate of local extinctions.When all suitable sites in question (though see DeAngelis et al. 1979), but the the regionare occupied,p equals 1. The singleinternal relationshipfound in Sect. 2 provides us with an ap- equilibriumof Eq. (2) is stable,f = 1-e/i,and regional proximative,yet qualitativelycorrect, non-constant re- extinctionfollows if e ' i. lationshipbetween p and the "average state" (abun- Levins (1970) subsequentlyanalysed the stochastic dance) in a local population: average local abundance versionof Eq. (2): the extinctionparameter, e, was as- increaseswith increasingp. sumed to be a randomvariable, withmean e and var- Probability of local extinctionincreases with de- iance ie. Assumingno autocorrelation("white noise"), creasing population size (e.g. MacArthurand Wilson the diffusionequation method (Kimura 1974) may be 1967, Christiansenand Fenchel 1977). One would ex- used to analyse the distributionof p, and gives (Levins pect, therefore(see Sect. 2), that e in Eq. (2) is not 1970), constant,but decreases withp. I have found 3 sets of data to test thisprediction. i1(p) = Cp2(0-0)/b-2exp(-2ip/oi), (3) A re-analysisof Simberloff's(1976) resultson ex- tinctionof local (island) populationsof mangroveisland as the limiting(t -m o) distributionof D(p,t). This does insectsshows thate in Eq. (2), a parameterrelated to not depend on the initial value, p(O). Constant C is the probabilityof local extinction,decreases with in- necessary to guarantee that ofktDp)dp = 1. Critical creasingp (Fig. 5). The same resultwas obtainedfrom a pointsof Eq. (3) may be foundfrom the equation, similaranalysis of Kontkanen's(1950) data on leafhop- pers in meadows in East Finland (Fig. 6). 2MSp- d/dp VFOP= 0, (4) The thirdexample is fromBoycott's (1930, see also 1919 and 1936) studyon fresh-watermolluscs in small where M6p and V1Pare the mean and variance of the ponds in the parishof Aldenham in England. Almost a rate of change in the stochasticversion of Eq. (2). This hundredponds were surveyedfor molluscs and plantsin gives the condition, 1915 and 1925. This example is particularlyimportant because the small size of the ponds enabled Boycott ie + ( (5) (1930: 2-3) to make accuratecensuses. The extinctions observed are thusreal. (Simberloff(1976) also triedto for a unimodal distributionD(p) with a peak at p = document all the populations of each island, while

OIKOS 38:2 (1982) 213

This content downloaded on Sat, 19 Jan 2013 18:53:43 PM All use subject to JSTOR Terms and Conditions I shall next explore two simple assumptionsabout the 0.4 colonizationprocess. C (1) If i in Eq. (6) does not depend onp, we obtainthe logistic equation (1). This has been analysed, in the U00. ecological context,by Levins (1969b; see also May 0 .3- X \ 1973 and Leigh 1975). In the deterministiccase, there 0 is one stable equilibriumpoint, either atp = 1, ifi > e', - Q@2- \ \ 0r.1 or at j = 0, ifi < e'. For the stochasticversion, assume .0 thats i - e' is a randomvariable, and thatthere is no 0 autocorrelation.The diffusionequation methodgives,

= 0. 14@p) Cp2(s/s-1)(1p)-2(s/aS+). (7)

1D2 3 D(p) is bimodal ifuS > S. If, on the otherhand, 5 > S, number of sites occupied all populationsapproach maximaldistribution, p = 1. (2) Let us assume that the rate of colonization is Fig. 6. Relationshipbetween the probabilityof local extinction and distributionin Kontkanen's(1950) data on leafhoppersin (s'p+e')p(l-p); the model then becomes, East Finland.Maximum number of sites (meadows) was three. Black dots representthe average for several species and the probabilityof extinctionin five years; open circles give the dp = stp2(1_p), sI > 0. (8) correspondingannual extinctionprobabilities (these resultsin- dicate that "re-colonization"was frequent,which is partlyan artefactof the relatively small sample size, leaving some small Evidently,there is only one stable equilibriumpoint, populationsunnoticed; cf. Kontkanen 1950). P = 1. If s' is a randomvariable, and thereis no autocorrelation,we can again use the diffusionequation Kontkanen (1950) probablymissed manysmall popu- method,which gives, lations.) My re-analysis(Fig. 7) of Boycott's (1930) data closely agrees withthe above results:e is not con- (D(p) = Cexp(-2a'/o'p)p 2(-'/o-2) (1s )-2(s'/?+1) (9) stantbut decreases with. This resultis not quite accu- rate, because more than one extinction-colonization This distributionis bimodal ifs' < oll/3. If the mean is event may have taken place in 10 years (cf. Diamond greaterthan a thirdof the variance,all populationsbe- and May 1977), but the trendis veryclear. come maximallydistributed. At present we may accept the simplesthypothesis A biologicaljustification for assumption (2) about the about the rate of extinction:e'(1-p)p (note thate' = e rate of colonizationis the probablyincreasing number whenp is small). On thisassumption Eq. (2) is replaced of emigrantswith increasing local abundance (e.g. by Dempster 1968, Johnson 1969); presumably,more emigrantsmeans more colonizations. - ip(1-p)-e'p(1-p). (6) dt Addendum C During the preparationof this paper it escaped my 0.6- noticethat there may be certainmathematical problems x 0.5- in the use of the diffusionequation technique in the 0 0.4- analysisof the models in Sect. 3 (Levins's 1970 analysis > is erroneus;see Boorman,S. A. and Levitt,P. R. 1973. 0. OL3- Theor. Pop. Biol. 4: 85-128; and see Roughgarden,J. 90.2- 1979, pp. 384-391. Theoryof populationgenetics and .0 evolutionaryecology: an introduction.MacMillan). A a-0.1 supplementarynumerical analysis of a discrete time versionof Eq. (6) indicates,nonetheless, that the result 5 10 number of sites occupied presentedhere is qualitativelycorrect (Hanski 1982). Fig.7. Relationshipbetween the probability of extinction and distributionin Boycott's(1930) data on fresh-watermolluscs in smallponds in England.This figure gives the "net" extinc- 4. Ecologicalappraisal tionrate in 10 years.While constructing the figure,I have includedthe 34 pondsin Boycott's(1930) classesA and B The present modificationof Levins's model led to a whichdid notdry up duringthe studyperiod. Species have radicallydifferent conclusion fromthe one originally beendivided into three distribution classes, namely 1 or2, 3 to 7, and 10 to 14 pondsoccupied, and the dots in thefigure give drawn by Levins: assumingstochastic variation in the theaverage for each group (altogether there were 16 species). rate of local extinctionand/or colonization, populations

214 OIKOS 38:2 (1982)

This content downloaded on Sat, 19 Jan 2013 18:53:43 PM All use subject to JSTOR Terms and Conditions 100-

0 70- 0.

60- 0 h 50 2 50 1 E C) U)40-

0 30- 1 2 3 4 5 6 7 8 9 number of sites occupied 0 - 20 Fig. 8. Frequencydistribution of species' distributionsin Sim- berloff's(1976) data on mangroveisland insects.Maximum numberof sites was nine. 10-

0 1 2 3 4 5 7 8 1 11 12 number of sites occupied tendtowards either one or the otherof the (determinis- Fig.10. Frequencydistribution ofspecies' distributions inLin- tic) boundaryequilibria, P = 0 and P = 1, while in Le- kola's (1916) data on anthropochorousplants in Russian vins's model p would hover around a stable internal Karelia,U.S.S.R. The sitesin question are isolated old houses equilibrium,0 U ' 1. Whichresult more correctlyre- and smallvillages, numbering 12 (sites8 to 19 in Linkola flectsreality? 1916). We recall that D(p) may be interpretedeither as the distributionof p values in one species in a longperiod of time,or as the distributionof p values in manysimilar 12 species were found from6 islands, and 32 species species at one momentof time (cf. Kimura 1964 for fromall 9 islands.The null hypothesisthat the number analogous interpretationsin population genetics). To of species is equal in these two classes is rejected: test the latter qualitatively,we require that all the x2 = 9.09, P < 0.01. species mayestablish local populationsat thesame sites, Onthophagus species in the lowland rain forestin and that interspecificinfluences on model parameters Sarawak (cf. Fig. 4) also show a clear dichotomyinto are density- and frequency-independent.My model two sets of species (Fig. 9). I shall use the term"core" thenpredicts bimodality of p's, peaks close to unityand species forthe locallyabundant and regionallycommon zero, while Levins's model predictsunimodality with species, and the term"satellite" species forlocally and the peak not veryclose to unityor zero. regionallyrare species. In the case of mangroveisland Simberloff's(1976) data (cf. Fig. 5) supportthe pre- insects(Fig. 8), the same termscan be used, although sentmodel; the distributionof the numberof mangrove "intermediate"species are now frequent. islands occupied by differentspecies of insectsappears Another data set to test this predictionis due to a bimodal (Fig. 8). To testthis formally, we observe that study by Linkola (1916) on the occurrence of an- thropochorousvascular plants near houses and villages, isolated by naturalforest, in Russian Karelia (thenFin- land, studyarea ca. 10000 kM2). There was a clear size 0) 7~ effect,large villages having more species than small ones (Linkola 1916), and colonization of isolated .6 houses and small villageswas perhaps not random,be- t5 cause some species were lacking systematicallyfrom them (thoughsome would do so by chance only). For 0 4- thesereasons, I have restrictedthe analysisto 12 similar sites in Linkola's (1916) material.The frequencydis- tributionof occurrencesat the 12 sitesis clearlybimodal C (Fig. 10), which stronglysupports the present model El I (see Hanski 1982 for a full analysisof Linkola's material). 5 10 It sufficesto mentionhere that Kontkanen's(1950) number of sites occupied resultson leafhoppersand Anasiewicz's (1971) results Fig. 9. Frequency distributionof species' distributionsin on bumblebees also supportthe presentmodel. A full Onthophagusin tropical lowland forestin Sarawak (Hanski analysisof thesetwo studies will be presentedelsewhere unpubl.). Maximumnumber of sites was 12. (Hanski unpubl.).

OIKOS 38:2 (1982) 215

This content downloaded on Sat, 19 Jan 2013 18:53:43 PM All use subject to JSTOR Terms and Conditions abstraction,but in manygroups of closelyrelated 5. Moreabout testing the core-satellite hypothesis specieshabitat selection may be sufficientlysimilar. An The core-satellitespecies hypothesis is a simplenull extensionof thecore-satellite hypothesis into analyses hypothesisto explainregional rarity. There are twoim- of nicherelations (in Sect. 8) requires,in anycase, a portantpremises in themodel: (1) regionalpopulation distinctionbetween habitat and niche(Whittaker et al. dynamicsis important,and (2) stochasticityin regional 1973). populationdynamics is important.Stochastic variation Thereis thedanger that counter-evidence (unimodal in theparameters of regional dynamics (extinction, col- distributionof p's) is dismissedon the basis thatthe onization)may be due to eitherdemographic or en- specieswere not, after all, similarenough. This would vironmentalstochasticity. be wrong;the question about habitat selection must be As the examplesgiven in the previoussection resolvedbefore the test is performed.In anycase it is showed,testing the main prediction of the model is sim- safestto includeonly similar population sites (habitat ple: Is thedistribution ofspecies' regional distributions patches) in the analysis, which, to someextent, removes bimodal?If the distribution is clearly bimodal, there are thisproblem. These restrictionsdo notmean that this groundsfor a dichotomy,and for the use ofthe concept theorycould not work on anyspecies. Careful selection in a strongsense. Otherwise, one is leftwith the option ofthe species and of the sampling sites is requiredonly of labellingthe opposite ends of a continuum,like the becauseof problemsin testingthe hypothesis. r-K speciesdistinction (MacArthur and Wilson1967) To repeat,the core-satellitehypothesis should be hasbeen replaced by the r-K species continuum (Pianka testedonly with sets of specieswhich may establish 1970, 1972, 1974, Southwood 1977). populationsat thesame sites; or, if such data are availa- Othermodels than the presentone maypredict a ble,with records for single species in thelong course of bimodaldistribution of regionaldistributions, and to time.I believethat the former requirement was fulfilled morerigorously test this model necessitates intensive in the above examples.A counter-exampleis the dis- studieson therates of local extinction and colonization. tributionalecology of water-striders(Gerris). Vep- If data are availableon therate of changein distribu- salainenhas shown,in a seriesof papers (see especially tion,validity of Eq. (6) maybe testeddirectly. Alterna- 1973,1974a, b, 1978,Jarvinen and Vepsalainen 1976), tively,one may tryto documentchanges in species' thatGerris species, nine of which occur in Finland, show statusfrom the core to thesatellite class, or viceversa, significantecological and morphologicaldifferences in betweenregions or in time.Such changes are predicted theiradaptations to livingin differentkinds of lakes, to occureven if the pattern of environmentalstochas- ponds,streams, etc., includingwing dimorphismin ticityremains stationary. An alternativemodel, which manyspecies. In thiscase habitatselection is clearly we maycall an "adaptation"hypothesis, states that core differentin differentspecies, and the core-satellite speciesare betteradapted to theenvironment than are hypothesisshould not be used for the whole set of satellitespecies, and does not predict changes from core species,though it couldbe used foreach speciessepa- tosatellite class, or vice versa. Note that also the present rately.In thelatter context, one couldtalk about core modelallows for interspecific differences in s and 2 and satellitepopulations, and comparisonsshould be (Sections3 and 8). madebetween regions or times. L. R. Taylor'swork (1974, 1978,Taylor and Taylor Assumingthe reality of core and satellite species, one 1977, Tayloret al. 1978, 1980, Taylorand Woiwod mayrephrase Hutchinson's (1959) questionand ask: 1980) on insectabundance and distributionhas dem- Why,in a givencommunity, are theren core and m onstratedthe ever-changingpatterns of regionaldis- satellitespecies? Constraintson core and satellite tributions,anticipated by Andrewarthaand Birch speciesdiversity are entirelydifferent, which warrants (1954), and Taylor'swork has shownthe interdepen- two questions(n and m) instead of one (n+m). dencebetween abundance and distribution.His results Nevertheless,satellite species may become, besides re- implythat local extinctionsand colonizationsare fre- gionallyextinct, also core species, and a corespecies can quentphenomena (see also Den Boer 1977, Ehrlich moveto regionalextinction only through a stageas a 1965,Ehrlich et al. 1980), and thatspatial population satellitespecies. Therefore the numbers of species in the dynamicsis importantin all species. twokinds are notindependent of each other. The gravestdifficulty in testingthe core-satellite A specificprediction may be derivedfor the most specieshypothesis in a multispeciescontext is habitat universaltrend in speciesdiversity, namely that the selection.How doesone identifythe "population sites" numberof species tends to increasewith area, whether suitablefor the species?The reasonfor insisting on theregion in questionis an islandor partof themain- "similarspecies" in testing the multispecies prediction is land.Because satellite species survive as a setof small just this;if the speciesare so similarthat they have populations,their regional existence should hinge on similarhabitat selection, there is no problem.Then the thesize of the region (e.g. Hanski1981a). Therefore, as numberof potentially inhabitable sites is thesame and regions- e.g. islands- become smaller,the proportion thedenominator in calculatingthep's is thesame for all ofsatellite species should decline. If Diamond's(1975, species.Identical species are, of course, an unattainable see also 1971, 1973,and others) series of speciesfrom

216 OIKOS 38:2 (1982)

This content downloaded on Sat, 19 Jan 2013 18:53:43 PM All use subject to JSTOR Terms and Conditions D-trampsto High-Sspecies is matchedwith the conti- which has been much used especially in plant ecology nuumfrom core to satellitespecies, the proportion of untilthe 1960's (e.g. Boosting1956, Hanson and Chur- satellitespecies indeed decreases with decreasing island chill 1961, Mueller-Domboisand Ellenberg 1974), and size. Diamond (1975: 358), however,considers his whichis of special relevancehere. To see this,divided D-trampsto be "r-selected",and High-Sspecies to be into 5 segmentsof equal length(-0.2, 0.21-0.4, etc.), "K-selected",which is incontrast with the present con- and denote by A to E thenumbers of species fallinginto ceptualization- core species are certainlynot the5 classes. Raunkiaer's"law of frequency"states that "r-selected". A > B > C > F < E. Quite unexpectedly,the simple theorysuggested in Sect. 3 predictedRaunkiaer's "law of frequency". Nonetheless,with papers by Gleason (1920, 1929) 6. The core-satellitehypothesis and r-Ktheory and Romell (1930), criticismof the "law of frequency" There existsa commonbasis for the core-satellite started to accumulate (Gause 1936a, Preston 1948, hypothesisand theby now well established r-K species Williams 1950), culminatingin the above-mentioned theory(MacArthur and Wilson1967, Gadgil and Bos- "execution" by McIntosh (1962). It had been shown sert 1970, Pianka 1970, 1972; but see Wilburet al. that the frequencydistribution of species' frequencies 1974, Southwood1977, Christiansenand Fenchel depends,in Williams's(1950) words,on "the numberof 1977, Schaffer1979). Both hypothesesstem from the quadrats,the size of the quadrats,and on the Index of same model - the logisticequation - which has been Diversityof the population." This criticismis justified. appliedat thelevel of local abundance in the r-K model, In view of the connectionto the core-satellitehypo- and at the level of regionaldistribution in the core- thesis, one significantdifference between the "laws" satellitemodel. should be pointedout (see also Hanski 1982). Nonetheless,the r-K species concept is usedin a de- Frequencyis the fractionof (usually small) samples, terministicfashion to predictproperties of species from typicallyquadrats, in whichthe species occurs,all'sam- theproperties of theirenvironment (e.g. Pianka1970, ples having been taken fromthe same homogeneous 1974, Southwood1977, Vepsalainen1978), whilethe community.Distribution, as it was definedin the intro- core-satellitedistinction is caused in the model by ductionand used in themodels (Sect. 3), is a measureof stochasticvariation in spatialpopulation dynamics. Un- occurrence on the between-sitescale. Although the like the satellitespecies, r-species are thoughtto be "true" populationlevel may be difficultto specify(for frequentlylocally abundant in comparisonwith an extreme example see Brussard and Ehrlich K-species,but this does not follow from the mathemati- 1970a, b), the distinctionbetween distribution and fre- cal model(logistic equation). quency is an importantone wheneverregional popula- Althoughthe two concepts are fundamentallydiffer- tion dynamicsare important,i.e. whenevermany local ent, core species are relatedto K-species,and, less populationsare studied.It has been pointedout thatthe obviously,satellite species are relatedto r-species. highest(E) of Raunkiaer's frequencyclasses is more inclusivethan the lower ones, because the frequency classes include unequal densityclasses (Gleason 1929, 7. A historicalperspective Ashby 1935, McIntosh 1962). But unlikebetween density(abundance) and frequencyclasses in homogeneous It is commonin ecology that authors - ortheir readers - communities,there is no simple statisticalrelationship find"new" ideas precededby earlierworkers (Hutch- betweendistribution and local abundance,the correla- inson1978, McIntosh 1980), and nowadayspreferable tions in Figs 1 to 4 (Sect. 2) being due to ecological byDarwin. This may be viewedas a markof soundness processes (notwithstandingproblems of sampling;Sect. in the argument- or is McIntosh(1962) correctin 2). In fact,the purposeof usingthe "law of frequency" claimingthat "certain ideas seemto be invulnerableto was to determinethe homogeneityof a standof vegeta- attackand persist although subjected to multipleexecu- tion (or a communityof animals; see e.g. Kontkanen tions"?The core-satellitehypothesis is not an exception 1950); bimodality(D < E) was namelyexpected only in tothe rule. The ironyhere is thatthe idea McIntosh was homogeneous stands, which is an interestingcon- executingin 1962was nothing else but bimodality of the vergenceto my independentlythought requirement of distributionofspatial occurrences - the very prediction similarhabitat selection in the species to be analysed fromthe models in Sect.3. (Sect. 5). G. F. Gause (1936a: 323, see also 1936b) wrote: "The mostimportant structural property of biocoenosis is the existenceof definitequantitative relations be- 8. tweenthe abundantspecies and therarer ones." One Concludingremarks: visiting Hutchinson's niche suchrelation, which Gause (1936a) discussedat length, space is Raunkiaer's"law of frequency"(Raunkiaer 1913, Afterthese observations and theorizing,the readermay 1918, 1934; a pioneeringwork by Jaccardin 1902), ask: What is gained by calling regionallycommon and

OIKOS 38:2 (1982) 217

This content downloaded on Sat, 19 Jan 2013 18:53:43 PM All use subject to JSTOR Terms and Conditions locally abundantspecies core species, and rare species satellitespecies? My answeris twofold.If the frequencydistribution of species' regionaldistributions is indeed bimodal,this is p interestingfor its own sake, because it appears not to be the null hypothesisfor many ecologists, who ratherex- pect the kind of unimodalitypredicted by Levins's model. Secondly, and more importantly,if such a dichotomyexists in many natural communities,this should help us to provide a functionalexplanation for patternsof abundance and distribution.To take an example, if the core-satellitehypothesis is upheld, one may proceed by restrictingthe application of the equilibriumtheory (MacArthur 1972, May 1973, 1976) to the core species, and employingappropriate non- equilibriummodels for the satellite species. Caswell (1978) presumablyhad a similaridea in mindwhen he, after discussingthe virtues of equilibriumand non- equilibriummodels in ecology,conjectured: "Perhaps a communityconsists of a core of dominantspecies, which interactstrongly enough among themselvesto arriveat equilibrium,surrounded by a largerset of non-equilibriumspecies playingtheir roles againstthe background of the equilibriumspecies." 0 In the introductionI referredto a thirdstructural x propertyof communitiesbesides abundance relations Fig. 11. A schematicrepresentation of the hypothesis,ex- (Engen 1978) and spatial distributions(Simberloff plainedin Sect.8, thatcore species are betterspaced-out in 1978): distributionof nichespace thana randomsub-set of the same size of all species, or strictlyspeaking their species,and better spaced-out than satellite species. x andy are "niches", in Hutchinson's(1957) niche space. The pe- twoniche dimensions, andp denotesdistribution, as in the rest rennial question is how well spaced-out niches are in ofthe paper, which varies from 0 to 1. Opencircles represent niche space. Intuitionsays and theory(e.g. MacArthur nichepositions, and the black dots give the position of species 1972, Lawlor and Maynard Smith in the3-dimensional space. It shouldbe recalledthat the ar- 1976) predictsthat gumentis stochastic(see text),and thesituation depicted in interspecificcompetition causes betterspacing-out, and thisfigure is a staticpicture of a dynamicprocess. ultimatelyand ideallyleads to a uniformdistribution of nichesin niche space. In view of the controversyabout the importance of competition in structuringcom- Recall thatthe model is: munities (Paine 1966, Harper 1969, Janzen 1970, Dayton 1971, Connell 1975, 1978, Caswell 1976, i = 1, ... .,n Glasser 1979), thisis an importantquestion. The prob- dt =siPi(_ -i), (Eq. 6) lem is that,in practice,other factorsbesides competi- tioncome intothe play,making any "test" difficult.It is where n is the numberof species, and si is a random not surprising,therefore, that this kind of argumenthas variable withmean 9i and variance a'. The probability led to widely varyingconclusions (MacArthur 1972, that species i is a core species at a given time is an Schoener 1974, Sale 1974, Inger and Colwell 1977, increasing functionof islo'. How does interspecific Southwood 1978, Stronget al. 1979, Pianka et al. 1979, competitioninfluence this ratio? Competitionshould Hanski 1979a, Lawlor 1980). The difficultyis in the increase the rate of extinction,it should decrease the formulationof a proper null hypothesis(see especially rate of colonization- hence competitionwill decrease s Lawlor 1980; the null hypothesisis not necessarilya - and it will probablyincrease oa. Consequently,g/oa random distributionof niches in niche space, Hanski will decrease, and the probabilityof species i staying/ 1979a, Grantand Abbott 1980), and in themultitude of becoming a satellite species increases. The closer the factorspotentially - and in practice- causingchanges in competitor(s),the strongerthe effect.Naturally, if there niche position. are two close competitors,both of which are core This is wherethe core-satellitehypothesis may prove species, this model only predictsthat one of them is useful. The followingheuristic argument shows that likelyto become a satellitespecies. The stochasticna- there exists, after all, at least one unequivocal null tureof the singlespecies model is preservedin the mul- hypothesis:if interspecificcompetition is importantin tispeciescontext. structuringcommunities, core species should be better We may visualize species in a space constructedof spaced-out in niche space than satellitespecies. Hutchinson's(1957) niche space and of one extraaxis,

218 OIKOS 38:2 (1982)

This content downloaded on Sat, 19 Jan 2013 18:53:43 PM All use subject to JSTOR Terms and Conditions givingthe extentof spatial distribution,p, whichcorre- Caswell, H. 1976. Communitystructure: a neutral model lates, as we saw in Sect. 2, with abundance (Fig. 11). analysis.- Ecol. Monogr. 46: 327-354. - 1978. Predator-mediatedcoexistence: a non-equilibrium The greaterthe densityof other species in the neigh- model. - Am. Nat. 112: 127-154. bourhood of any species in niche space, the smallerthe Christiansen,F. B. and Fenchel, T. M. 1977. Theories of probabilitythat this species is, at a given time,a core populationsin biologicalcommunities. - Springer,Berlin. species. Hence, core species are not expected to be a Cohen, J. E. 1970. A Markov contingency-tablemodel for replicatedLotka-Volterra systems near equilibrium.- Am. random sub-setof all the species withrespect to niche Nat. 104: 547-560. position;we expectcore species to comprisesuch a sub- Coins, H. N. and Hassell, M. P. 1979. The dynamicsof set withinwhich species are betterspaced-out from each optimallyforaging predators and parasitoids.- J. Anim. otherthan species are withina trulyrandom sub-set. It Ecol. 48: 335-353. J. H. 1975. Some mechanisms structure followsthat core species are also betterspaced-out than Connell, producing in natural communities.- In: Cody, M. L. and Diamond, satellitespecies. J. M. (eds.), Ecology and evolution of communities,The A within-communityanalysis, such as suggestedhere, Belknap Press of Harvard Univ. Press, Cambridge. provides a concrete point of reference to test null - 1978. Diversityin tropicalrain forestsand coral reefs.- Science 199: 1302-1310. hypothesesabout communitystructure. Still, the pre- Crowley, P. H. 1979. Predator-mediatedcoexistence: an sent model cannot be but one step towardsbetter un- equilibriuminterpretation. - J. Theor. Biol. 80: 129-144. derstandingof communitystructure and its evolution. Dayton, P. K. 1971. Competition,disturbance, and community For instance,one could argue thateven if core species organization:the provisionand subsequent utilizationof space in a rockyintertidal community. - Ecol. Monogr.41: are betterspaced-out than satellite species in a com- 351-389. munity,this is perhaps not due to competitionbut to DeAngelis, D. L., Travis,C. C. and Post, W. M. 1979. Persis- predation. Other studies are necessary to establish tance and stabilityof seed-dispersedspecies in a patchy which explanationis correct.Theoretical work is also environment.- Theor. Pop. Biol. 16: 107-125. J. needed to clarifythe expectationsin multispeciescom- Dempster, P, 1968. Intra-specificcompetition and dispersal: as exemplifiedby a psyllidand its anthocoridpredator. - munities.The value of the presentapproach can onlybe In: Southwood,T. R. E. (ed.), Insectabundance. Symp. R. judged afterseveral ecologists have tried it indepen- Ent. Soc. London, No. 4, Blackwell Scient.Publ., Oxford. dentlyon theirown data. Tests on dungbeetles (Hanski Den Boer, P. J. 1977. Dispersal power and survival.Carabids in a cultivated countryside.- Misc. papers 14, Land. 1980a) and bumblebees (Hanski unpubl.) have given Wageningen,The Netherlands. encouragingresults (see also Hanski 1979b, 1980b). Diamond, J. M. 1971. Comparisonof faunalequilibrium turn- over rates on a tropicaland a temperateislands. - Proc. Natl Acad. Sci. USA 68: 2742-2745. Acknowledgements- I wish to thank several colleagues for - 1973. Distributionalecology of New Guinea birds.- Sci- theiruseful comments on thedifferent versions of themanu- ence 179: 759-769. script:R. Alatalo,J. Cartes, E. Connor,B. Don, Y. Haila,0. - 1975. Assemblyof species communities.- In: Cody, M. L. Jarvinen(twice), Y. Mdkinen,E. Ranta,U. Safriel,D. Sim- and Diamond, J. M. (eds.), Ecology and evolution of berloff,and K. Vepsalainen.0. Prys-Jonesdrew my attention communities,The Belknap Press of Harvard Univ. Press, to papersby Gause. Cambridge. - and May, R. M. 1977. Species turnoverrates on islands: dependence on census interval.- Science 197: 266-276. Engen, S. 1978. Stochasticabundance models. - Chapman and References Hall, London. Anasiewicz, A. 1971. Observations on the bumble-bees in Ehrlich,P. R. 1965. The population biology of the butterfly Lublin. - Ekol. Polska 19: 401-417. Euphydryaseditha II. The structureof the JasperRidge Andrewartha,H. G. and Birch,L. C. 1954. The distribution colony.- Evolution 19: 327-336. and abundance of animals. - Chicago Univ. Press, - Murphy,D. D., Singer,M. C., Sherwood, C. B., White, Chicago. R. R. and Brown, I. L. 1980. Extinction,reduction, sta- Ashby, E. 1935. The quantitativeanalysis of vegetation.- bilityand increase: the responsesof checkerspotbutterfly Ann. Botany 49: 779-802. (Euphydryas) populations to the California drought.- Bartlett,M. S. 1975. The statisticalanalysis of spatialpatterns. Oecologia (Berl.) 46: 101-105. - Monogr.Apple. Prob. Statistics,Chapman and Hall, Lon- Gadgil, M. and Bossert,W. 1970. Life historyconsequences of don. naturalselection. - Am. Nat. 104: 1-24. Boycott,A. E. 1919. The freshwatermollusca of the parishof Gause, G. F. 1936a. The principlesof biocoenology.- Quart. Aldenham: an introductionto the studyof theiroecologi- Rev. Biol. 11: 320-336. cal relationships.- Trans. Hertf. Nat. Hist. Soc. 17: - 1936b. Some basic problemsin biocoenology.(In Russian, 153-200. summaryin English) - Zool. Zh. 15: 363-381. - 1930. A re-surveyof the fresh-watermollusca of the parish Glasser, J. W. 1979. The role of predation in shaping and of Aldenham afterten years withspecial referenceto the maintainingthe structureof communities.- Am. Nat. 113: effectof drought.- Trans. Hertf.Nat. Hist. Soc. 19: 1-25. 63 1-641. - 1936. The habitatsof fresh-waterMollusca in Britain.- J. Gleason, H. A. 1920. Some applications of the quadrat Anim. Ecol. 5: 116-180. method.- Bull. TorreyBot. Club 47: 21-33. Brussard,P. F. and Ehrlich,P. R. 1970a. Adult behaviourand - 1929. The significanceof Raunkiaer's law of frequency.- population structurein Erebia epipsodea (Lepidoptera: Ecology 10: 406-408. Satyridae).- Ecology 51: 880-885. Grant,P. R. 1978. Dispersal in relationto carryingcapacity. - - and Ehrlich, P. R. 1970b. The population structureof Proc. Natl Acad. Sci. USA 75: 2854-2858. Erebia epipsodea (Lepidoptera: Satyridae).- Ecology 51: - and Abbott, 1. 1980. Interspecificcompetition, island 119-129. biogeography and null hypotheses. - Evolution 34: 332-341.

OIKOS 38:2 (1982) 219

This content downloaded on Sat, 19 Jan 2013 18:53:43 PM All use subject to JSTOR Terms and Conditions Gurney, W. S. C. and Nisbet, R. M. 1978a. Predator-prey - 1957. On the delimitationof communitiesin researchon fluctuationsin patchyenvironments. - J. Anim. Ecol. 47: animal biocoenotics.- Cold SpringHarbour Symp.Quan. 85-102. Biol. 22: 373-378. - and Nisbet, R. M. 1978b. Single-speciespopulation fluc- Krebs, C. J. 1972. Ecology: the experimentalanalysis of dis- tuations in patchy environments.- Am. Nat. 112: tributionand abundance. - Harper & Row, New York. 1075-1090. Lawlor, L. R. 1980. Structureand stabilityin naturaland ran- Hanski, I. 1979a. The communityof coprophagousbeetles. - domly constructedcompetitive communities. - Am. Nat. D. Phil. thesis,Univ. of Oxford,Oxford (unpubl.). 116: 394-408. - 1979b. Structureof communities.(In Finnish,summary in - and MaynardSmith, J. 1976. The coevolutionand stability English.) - Luonnon Tutkija 83: 132-137. of competingspecies. - Am. Nat. 111: 79-99. - 1980a. The communityof coprophagousbeetles (Coleop- Leigh, E. G. Jr.1975. Populationfluctuations, community sta- tera, Scarabaeidae and Hydrophilidae) in northern bilityand environmentalvariability. - In: Cody, M. L. and Europe. - Ann. Ent. Fenn. 46: 57-74. Diamond, J. M. (eds.), Ecology and evolution of com- - 1980b. Structureof insectcommunities: the core and the munities,The Belknap Press of Harvard Univ. Press, satellitespecies. - In: Abstractsof the XVI Int. Congress Cambridge. of Entomology,Kyoto. Levin, S. A. 1974. Dispersion and populationinteractions. - - 1981a. Exploitative competition in transient habitat Am. Nat. 108: 207-228. patches.- In: Chapman,D. G., Gallucci, V. and Williams, - 1976. Population dynamicmodels in heterogeneousenvi- F. M. (eds.), Quantitativepopulation dynamics. Statistical ronment.- Ann. Rev. Ecol. Syst. 7: 287-310. ecology, Vol. 13, Intern. Co-op. Publ. House, Fairland, - 1977. Spatial patterningand the structureof ecological Maryland. communities.- Some mathematicalquestions in biology, - 1981b. Coexistenceof competitorsin patchyenvironment VII, Vol. 8. Am. Math. Soc., Providence,R. I. withand withoutpredation. - Oikos 37: 306-312. - 1978. Patternformation in ecological communities.- In: - 1982. Distributionalecology of anthropochorousplants in Steele, J. H. (ed.), Spatial pattern in plankton com- villages surroundedby forest.- Ann. Bot. Fenn. 19 (in munities.Plenum, New York. press). Levins, R. 1969a. Some demographicand genetic consequ- Hanson, H. C. and Churchill,E. D. 1961. The plant com- ences of environmentalheterogeneity for biological con- munity.- Reinhold,New York. trol.- Bull. Ent. Soc. Am. 15: 237-240. Harper,J. L. 1969. The role of predationin vegetationdiver- - 1969b. The effectof randomvariation of differenttypes of sity.- BrookhavenSymp. Biol. 22: 48-62. population growth. - Proc. Natl Acad. Sci. U.S. 62: Hengeveld, R. 1979. On the use of abundance and species- 1061-1065. abundancecurves. - In: Ord, J. K., Patil,G. P. and Taillie, - 1970. Extinction.- In: Gerstenhaber,M. (ed.), Some C. (eds.), Statistical distributionsin ecological work. mathematicalproblems in biology.Lectures on mathema- Statisticalecology, Vol. 4, Intern. Co-op. Publ. House, ticsin the lifesciences 2. Am. Math. Soc., Providence,R. I. Fairland,Maryland. - and Culver,D. 1971. Regional coexistenceof species and Horn, H. S. and MacArthur,R. H. 1972. Competitionamong competitionbetween rare species. - Proc. Natl Acad. Sci. fugitivespecies in a harlequinenvironment. - Ecology 53: U.S. 68: 1246-1248. 749-752. Lidicker,W. Z. Jr.1962. Emigrationas a possible mechanism Hutchinson,G. E. 1957. Concludingremarks. - Cold Spring permittingthe regulationof populationdensity below car- Harbour Symp. Quant. Biol. 22: 415-427. ryingcapacity. - Am. Nat. 96: 29-33. - 1959. Homage to Santa Rosalia or Whyare thereso many Linkola, K. 1916. Studienuber den Einflussder Kulturauf die kindsof animals?- Am. Nat. 93: 145-159. Flora in den Gegenden nordlichvom Ladogasee. I. - Acta - 1978. An introductionto populationecology. - Yale Univ. Soc. Fauna et Flora Fenn. 45. Press., New Haven, Conn. Lotka, A. J. 1925. Elements of physicalbiology. - Williams Inger,R. F. and Colwell, R. K. 1977. Organizationof contigu- and Wilkins,Baltimore. ous communitiesof amphibiansand reptilesin Thailand. - MacArthur,R. H. 1972. Geographical ecology. - Harper & Ecol. Monogr. 47: 229-253. Row, New York. Jaccard,P. 1902. Lois de distributionflorale dans la zone al- - and Wilson, E. 0. 1967. The theoryof island biogeog- pine. - Bull. Soc. Vand. Sci. Nat. (Lausanne), 38. raphy.- PrincetonUniv. Press, Princeton. Janzen,D. H. 1970. Herbivoresand the numberof treespecies May, R. M. 1973. Stabilityand complexityin model ecosys- in tropicalforests. - Am. Nat. 104: 501-528. tems.- PrincetonUniv. Press, Princeton. Jarvinen,0. and Sammalisto,L. 1976. Regional trendsin the - (ed.) 1976. Theoretical ecology. Principlesand applica- avifaunaof Finnishpeatland bogs. - Ann. Zool. Fenn. 13: tions.- Blackwell Sci. Publ., Oxford. 31-43. McIntosh, R. P. 1962. Raunkiaer's "law of frequency".- - and Vepsaldinen,K. 1976. Wing dimorphismas an adap- Ecology 43: 533-535. tive strategyin water-striders(Gerris). - Hereditas 84: - 1980. The background and some currentproblems in 61-68. theoreticalecology. - Synthese43: 195-255. Johnson,C. G. 1969. Migrationand dispersal of insectsby Mueller-Dombois, D. and Ellenberg, H. 1974. Aims and flight.- Methuen,London. methodsof vegetationecology. - JohnWiley, New York. Karppinen, E. 1958. Untersuchungenuber die Oribatiden Oosting, H. J. 1956. The study of plant communities.- (Acari) der Waldboden von Hylocomium-Myrtillus-Typin Freeman,San Francisco. Nord-Finnland.- Ann. Ent. Fenn. 24: 149-168. Ord, J. K., Patil, G. P. and Taillie, C. (eds.) 1980. Statistical Kimura,M. 1964. Diffusionmodels in populationgenetics. - J. distributionsin ecological work.- Statisticalecology, Vol. Appl. Prob. 1: 177-232. 4. Intern.Co-op. Publ. House, Fairland,Maryland. Kontkanen,P. 1937. QuantitativeUntersuchungen uber die Paine, R. T. 1966. Food web complexityand species diversity. Insektenfaunader Feldschichtauf einigen Wiesen in Nord- - Am. Nat. 100: 65-75. Karelien. - Ann. Zool. Soc. 3, No. 4. Patil, G. P., Pielou, E. C. and Waters,W. E. (eds.) 1971. Spa- - 1950. Quantitativeand seasonal studieson the leafhopper tial patterns and statistical distributions.- Statistical faunaof the fieldstratum on open areas in NorthKarelia. - ecology,Vol. 1. Penn. State Univ. Press, Pennsylvania. Ann. Zool. Soc. 13, No. 8. Pianka, E. R. 1970. On r- and K-selection.- Am. Nat. 104: 592-597.

220 OIKOS 38:2 (1982)

This content downloaded on Sat, 19 Jan 2013 18:53:43 PM All use subject to JSTOR Terms and Conditions - 1972. r and K selectionor b and d selection?- Am. Nat. - 1978. Bates, Williams,Hutchinson - a varietyof diver- 106: 581-588. sities.In: Mound, L. A. and Waloff,N. (eds.), Diversityof - 1974. Evolutionaryecology. - Harper & Row, New York. insect faunas. Symp. R. Ent. Soc. London, No. 9, - Huey, R. B. and Lawlor, L. R. 1979. Niche segregationin Blackwell Sci. Publ., Oxford. desertlizards. - In: Horn, D. J.,Mitchell, R. D. and Stairs, - and Taylor, R. A. J. 1977. Aggregation,migration and G. R (eds), Analysis of ecological systems.Ohio State populationmechanics. - Nature,Lond. 265: 415-421. Univ. Press, Columbus. - and Woiwod, I. P. 1980. Temporal stabilityas a density- Preston,F. W. 1948. The commonnessand rarityof species. - dependent species characteristics.- J. Anim. Ecol. 49: Ecology 29: 254-283. 209-224. Raunkiaer,C. 1913. FormationsstatistiskeUnders0gelser paa - Woiwod, I. P. and Perry,J. N. 1978. The density-depen- Skagens Odde. - Bot. Tidskr.Kobenhavn 33: 197-228. dence of spatialbehaviour and the rarityof randomness.- - 1918: Recherchesstatistiques sur les formationsvegetables. J. Anim. Ecol. 47: 383-406. - Biol. Medd. 1: 1-80. - Woiwod, I. P. and Perry,J. N. 1980. Variance and the - 1934. The life formsof plants and statisticalplant geo- large scale spatial stabilityof aphids,moths and birds.- J. graphy.- Clarendon Press, Oxford. Anim. Ecol. 49: 831-854. Romell, L. G. 1930. Commentson Raunkiaer's and similar Vandermeer,J. H. 1973. On the regionalstabilization of lo- methodsof vegetationanalysis and the "law of frequency". cally unstablepredator-prey relationship. - J. Theor. Biol. - Ecology 11: 589-596. 41: 161-170. Sale, P. F. 1974. Overlap in resource use, and interspecific Vepshliinen,K. 1973. The distributionand habitatsof Gerris competition.- Oecologia (Berl.) 17: 245-256. Fabr. species (Heteroptera,Gerridae) in Finland.- Ann. Schaeffer,W. M. 1979. The theoryof life-historyevolution Zool. Fenn. 10: 419-444. and its applicationto Atlanticsalmon. - Symp.Zool. Soc. - 1974a. The wing lengths, reproductivestrategies and London 44: 307-326. habitatsof HungarianGerris Fabr. species (Heteroptera, Schoener, T. W. 1974. Resource partitioningin ecological Gerridae). - Ann. Acad. Sci. Fenn. A 202. communities.- Science 185: 27-39. - 1974b. The life cycles and wing lengthsof FinnishGerris Simberloff,D. 1976. Experimentalzoogeography of islands: Fabr. species (Heteroptera,Gerridae). - Acta Zool. Fenn. effectsof island size. - Ecology 57: 629-648. 141. - 1978. Colonization of islands by insects:immigration, ex- - 1978. Wing dimorphismand diapause in Gerris- determi- tinction,and diversity.- In: Mound, L. A. and Waloff,N. nation and adaptive significance.- In: Dingle, H. (ed.), (eds.), Diversityof insectfaunas. Symp. R. Ent. Soc. Lon- The evolution of migration and diapause in insects. don, No. 9, Blackwell Sci. Publ., Oxford. Springer,New York. Skellam,J. G. 1951. Random dispersalin theoreticalpopula- Whittaker,R. H., Levin, S. A. and Root, R. B. 1973. Niche, tions.- Biometrika38: 196-218. habitat,and ecotope. - Am. Nat. 107: 321-338. Slatkin, M. 1974. Competition and regional coexistence. - Wiens, J. A. 1976. Population responses to patchyenviron- Ecology 55: 128-134. ments.- Ann. Rev. Ecol. Syst. 7: 81-120. Smith, A. T. 1974. The distributionand dispersal of pika: Wilbur, H. M., Tinkle, D. W. and Collins, J. P. 1974. En- consequences of insular population structure.- Ecology vironmentalcertainty, trophic level, and resource availa- 55: 1112-1119. bilityin life historyevolution. - Am. Nat. 108: 805-817. Southwood,T. R. E. 1977. Habitat,the templetfor ecological Williams,C. B. 1950. The applicationof the logarithmicseries strategies.- J. Anim. Ecol. 46: 337-365. to the frequencyof occurrenceof plantspecies in quadrats. - 1978. The componentsof diversity.- In: Mound, L. A. - J. Ecol. 38: 107-138. and Waloff,N. (eds.), Diversityof insectfaunas. Symp. R. Yodzis, P. 1978. Competitionfor space and the structureof Ent. Soc. London, No 9, Blackwell Sci. Publ., Oxford. ecological communities.- Lecture notes in Biomath., Strong,D. R., Szyska,L. A. and Simberloff,D. S. 1979. Tests Springer,Berlin. of community-widecharacter displacementagainst null Zeigler, B. P. 1977. Persistenceand patchinessof predator- hypotheses.- Evolution 33: 897-913. prey systemsinduced by discrete event population ex- Taylor,L. R. 1974. Monitoringchange in the distributionand change mechanisms.- J. theor.Biol. 67: 677-686. abundance of insects.- Report RothamstedExp. Station for 1973, Part 2.

OIKOS 38:2 (1982) 221

This content downloaded on Sat, 19 Jan 2013 18:53:43 PM All use subject to JSTOR Terms and Conditions