Fish Species Diversity in Lakes
Author(s): Clyde D. Barbour and James H. Brown
Source: The American Naturalist, Vol. 108, No. 962 (Jul. - Aug., 1974), pp. 473-489
Published by: The University of Chicago Press for The American Society of Naturalists
Stable URL: http://www.jstor.org/stable/2459679 .
Accessed: 20/06/2011 16:06
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=ucpress. .
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
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].
The University of Chicago Press and The American Society of Naturalists are collaborating with JSTOR to
digitize, preserve and extend access to The American Naturalist.
http://www.jstor.org
The American Naturalist
Vol. 108, No. 962
FISH
CLYDE
July-August 1974
SPECIES DIVERSITY IN LAKES
D. BARBOUR* AND JAMES H. BROWN
Museum of Zoology, Universityof Michigan, Ann Arbor, Michigan 48104, and
Department of Biology, Universityof Utah, Salt Lake City, Utah 84112
For ecologists and evolutionarybiologists interestedin biogeographic
patterns and the relationships among species in natural communities,
islands and insular habitats (e.g., caves, mountaintops,and desert oases)
provide interestingsystemsfor analysis. The numberof species of several
taxa in some isolated habitats representsa dynamic equilibriumbetween
contemporaryrates of colonizationand extinction(see MacArthur 1972).
Diversityof thesebiotas can be accountedfor with littleor no referenceto
historicalevents.For certainothertaxa, however,barriersbetweeninsular
habitats preventcolonizationexcept during periods when they are temporarily abolished and the "islands" are connectedby bridges of suitable
habitat (Brown 1971).
What determinesthe diversity(the number) of fishspecies in a lake or
inland sea? The numberand identityof taxa which have had opportunity
to colonizea lake depend on unique, historicalevents: geologicalconnections
withotherbodies of water inhabitedby fishesand the compositionof those
faunas. Speciation of the colonizingtaxa can also influencethe numberof
species.Ultimately,diversitywill be limitedif the ability of the species to
exist and coexistwithina lake resultsin an equilibriumbetweenthe acquisitionof species by colonizationand speciationand loss by extinction.Our
purposehere is to ask whetheror not the numberof fishspecies in a lake or
sea conformsto the model of island biogeographyproposed by MacArthur
and Wilson (1967) and to assess the relativeimportanceof certainenvironmentalvariables as determinantsof the numberof species present.
METHODS
Statistical analysis.-The results described below were derived by forward stepwisemultipleregressionanalyses of the numberof fishspecies on
a series of independentvariables according to the method outlined by
Draper and Smith (1966). Similar to Hamilton, Barth, and Rubinoff
and logarithmic-to-logarithmic
(1964) we present arithmetic-to-arithmetic
(log 10) analysis of X's on Y as fittedby the computer.The study was
carried out at the Universityof Michigan ComputingCenter utilizingthe
Michigan InteractiveData Analysis System (MIDAS). In this algorithm,
* Present address: Department of Zoology, P.O. Drawer Z, Mississippi State, Mississippi 39762.
473
THE AMERICAN
474
NATURALIST
variables are added to the regressionin order of their partial correlation
afterthe selectionof the mostcorrelatedX with Y, the variable
coefficients;
of those remaining is the
with the highestpartial correlationcoefficient
next to be enteredinto the equation as long as its F statisticfor its added
contributionremainsabove a predeterminedlevel of significance.When F
falls below this value, computationsare terminated.Variables are also deleted fromthe equation if, in the process of adding new Y's, the significance of the F values for their added contributionsfalls below a preset
level. In our analysis,levels of significancefor both inclusion and deletion
of variables were set at a - 1.0, with the result that variables were added
to the regressionas long as the number of degrees of freedomwas not
exceeded.
of determinationand was derived
In tables 1 and 2, r2 is the coefficient
independentvariable with the
of
each
cofficient
by squaring the correlation
of multiple determinationand
numberof fishspecies. R2 is the coefficient
indicates the cumulativeproportionof variabilityin the number of fish
species accounted for by the independentvariables. We stress that the
values of R2 are not independentbut are contingentupon the othervariables whichhave been enteredinto the regressionequation and their order
of entry.
Data.-Values for the numberof fishspecies,latitude,and surface area
of the lakes are froma varietyof sources (Appendix). The numberof fish
species is subject to severalsourcesof error.It may be less than the actual
numberif collectorsfailed to obtain some which were present.This surely
affectsour analysis of the African lakes, as many of them are still poorly
known (G. Fryer and P. H. Greenwood,pers. comm.). Species numbermay
also be depressedif extinctionsowing directlyto human activityoccurred
before collectionswere made. For example, the draining of about 50,000
hectaresof Lake Chapala, Mexico,between1909 and 1912 (Goldman 1951)
undoubtedlycaused local extinctions.The numberof species may be greater
than the actual numberif fisheshave been introducedor if the censuses
include transientswashed in fromstreamsand rivers.We adjusted species
number to account for extinctionsand introductionswhereverpossible.
Finally, the criteriafor recognizingfishspecies differamong ichthyologists,
so that taxonomictreatmentsmay have influencedthe data. The objection
that we have not rigorouslydefineda lacustrinespecies may also be raised.
While it is true that there are species restrictedto eitherlakes or streams,
TABLE
1
COMPARISON OF COEFFICIENTS OF DETERMINATIONS (i-2) AND COEFFICIENTS OF MULTIPLE
DETERMINATION (R2) OF TWO ENVIRONMENTAL FACTORS AFFECTING THE NUMBER OF
FISH SPECIES IN 70 OF THE WORLD 'S LAKES (SEE APPENDIX AND FOOTNOTES
TO TABLE 2)
LOGARITHMIC
LINEAR
VARIABLE
Surface area .........
Latitude ..................
q2
B2
r2
B2
.1493+t
.0835-
.1493+
.2476-
.3050+
.0652-
.3050+
.3400-
FISH
SPECIES
475
DIVERSITY
TABLE 2
COMPARISON
OF COEFFICIENTSOF DETERMINATIONS
(r2) ANDCOEFFICIENTSOF MULTIPLE
DETERMINATION
FACTORSAFFECTINGTEE NUMBEROF FISH
(B2) OF ENVIRONMENTAL
SPECIES IN 14 NORTHAMERICANAND14 AFRICANLAKES (SEE LEGENDOF FIG. 2) *
LINEAR
Variable
r2
LOGARITHMIC
B2
Variable
r2
B2
Latitude ........
Surface area .
Shoreline length .
Altitude .3130Transparency
Shore develop.
mentt
Growingseason ..
pH .1409+
Mean annual
rainfall .
Total alkalinity .
Maximum depth .
Mean depth .....
.64372521+
.1672+
.64370.92570+
.95386.96470.97883-
North America
Mean annual
rainfall ......
pH ............
Altitude .........2574Growingseason . .
Drainage area . .
Mean depth ..... .
Mean July
temperature
Total alkalinity
Latitude .........4864Surface area ... . .
Morphoedaphic
index?....
Shoreline length
.5401+ t
.2832+
.4805+
.00550950+
.54007+
.73625+
.91261.94447+
.96243+
.98069-
3121+
.98972.99216+
.99526+
.99691-
.0309.1539+
.999051.00000+
.4936+
.2358+
.03161138.5374+
.98445+
.98660.99771+
5690+
.1452+
.0788+
.1670+
.99899.99952.99996+
.99998-
Africa
Maximum
Surface area .
.4887+
.48879+
.1688.70838depth ......
6947+
.69474+
Conductivity.
Surface area ... . . 5909+
Maximum
.85428+
Latitude ........0051+
4644+
.78408+
.90997+
depth, .....
.94465pH .....
Conductivity... . 19920003+
.78746+
.94828Mean annual
pH ....
Mean annual
rainfall .
0027+
.79462+
rainfall ...
.95050Latitude .....
0806+
0017+
.79620+
* See Statistical Analysis section for computationalprocedures.
f The sign of the correlation coefficientand the partial regression coefficientfor each
independent variable is indicated to the right of the values for r2 and B2, respectively.
t The length of the shoreline of a lake divided by the circumferenceof a circle with
the same area.
? The measured or estimated total dissolved solids (mg/liter) divided by mean depth
(m) .
11See Data section; includes oxygenated depths only of Lakes Tanganyika, Malawi,
and Kivu.
thereare also many that can exist in both places by being sufficiently
generalized or by seeking out similar habitats in both places (Corbet 1961).
Some, such as certain salmonids,may be importantmembersof the lake
communitybut spawn in its tributaries.Because lacustrinespecies are derived fromfluvialancestors,we are dealing withan evolutionarycontinuum
whichwe have not attemptedto subdivide.With the exceptionsnoted,we
have taken the censusesat face value even if they do tend to increase the
numberof species in the lakes. The patterns we describebelow are clear
enough to indicate that such imprecisionmay be tolerated as a firstapproximation.
The values for latitude are for the centerof each lake and are calculated
to the nearest decimal tenthof a degree. All lakes were treated as if they
were in the same hemisphere.Surface areas varied slightlyfromreference
to reference,representingeitherreal changesin the lake levels or errorsof
476
THE AMERICAN
NATURALIST
measurement.The areas of Lakes Chad and Balkhash, widely fluctuating,
shallow bodies of water,are means of maximumand minimumvalues.
Morphometric,
physiographic,chemicaland physicalparameters,most of
the climaticdata, and numbersof species for our samplesof northernNorth
Americanand Africanlakes are fromRyder (1972, tables 1-3), Fryer and
Iles (1972, tables 1, 2), and Greenwood(1964, table 1). The firsttwo references also containmaps showingthe localities of the lakes. These data are
subject to the sources of error discussed by those authors (in particular
Ryder) or their sources and are not reproduced here; we have accepted
them at face value with the exception of the maximumdepths of Lakes
Tanganyika, Malawi, and Kivu. These bodies of water are permanently
stratifiedand, as a result, anaerobic below depths of about 200, 250, and
65 m, respectively(G. Fryer, pers. comm.; Damas 1937). As these values
representfunctionalmaximumdepths,they have been substitutedfor the
geologicalmaximaof 1,470,704, and 465 m. Stratification
may also resultin
loss of oxygenin the deeperwatersof Lake Victoria (Fryer and Iles 1972),
but we disregardthistransitoryphenomenon.
All independentvariables for Africanlakes were enteredinto the regression equationforthosebasins. For the NorthAmericanlakes, two variables,
total dissolved solids and conductivity,were not included in the analysis
because of missingdata. It is verydoubtful,however,that theseparameters
are importantdeterminantsof fishspecies diversityin our sample, as the
formerranges from33 to 150 mg/literand the latter from40 to 596 micromhos.Of the remainingvariables,threewere not added to the regression
because of the limitingnumberof degreesof freedom.These were, for the
linear regression,maximum depth, shoreline development,and transparency.In the nonlinearregression,drainage area, morphoedaphicindex, and
mean Julytemperaturewere omitted.
Some variableswere given as ranges and were averaged for our analysis:
mean July temperature,mean annual rainfall and growingseason (Ryder
1972), and conductivityand pH (Fryer and Iles 1972). Values for mean
annual rainfall for African lakes were estimatedfromJackson (1961) and
should be consideredrough approximations.
RESULTS
Lakes and seas of the world.-Nonlinear multiple regressionof species
numberson area and latitude for a sample of 70 of the world's lakes and
seas accounts for 34%7oof the variation in the dependentvariable; 30.5%
is explainedby area and 3.5% by latitude.The linear model is less effective
and accountsfor only 24.8%oof the variability (table 1). The slope of the
species-areacurve is low (z = 0.15, fig. 1), falling below that usually reportedfor islands (z = 0.20-0.35, MacArthurand Wilson 1967).
NorthAmericanlakes.-Both linear and nonlinearmultipleregressionof
species numbersagainst 12 independentvariables accounts for essentially
all the lacustrinevariation (table 2) for 14 northernNorthAmericanLakes
FISH
SPECIES
477
DIVERSITY
3~ ~
~~~~~~0
014
100~
26
~~~~~~~~~~~~~~~~0
5'
U
67
6;@52
Uo
@63
cc
062
0566@629
64
44
M
190
1.0
Oh1
10
100
0.54
4
047
*39
%58 @6
0 66
J40
32
@
8v
0 *12
21
0
4i
59
*
*
28
2
* 6 *469
@3
]6*
*033.*35
049 4
_
41
2
1,000
AREA OF LAKE (KM2)
10,000
100,000
FIG. 1.-The relationship of area to number of fish species for 70 of the
world's lakes and seas. The basins are numbered the same as in the Appendix.
(fig. 2). In the formermodel,mean annual rainfall,pH, and altitude account for 91% while latitude and surface area account for about 92%c of
the variabilityin species numbersin the analysis of transformeddata. The
remainingvariables by themselvesare inconsequentialas determinantsof
diversity.The slope of the species-areacurve (z = 0.16, fig. 2) is almost
as low as it is for the sample of worldwidelakes.
African lakes.-Linear and nonlinearmultipleregressionof the number
of species against six independentvariables for 14 African lakes (fig. 2)
accounts for about 95%oand 80%oof the variation,respectively(table 2).
Maximumdepth and surfacearea are the most importantfactorsaffecting
the species numbersin the linear analysis and togetheraccount for about
85%oof the variation.Surface area and conductivityexplain about 71%oin
the nonlinearanalysis. The slope of the species-areacurve (z = 0.35, fig.2)
falls in the upper range of the values for islands (MacArthur and Wilson
1967).
012
013
014
1
*
100
10
100
1,000
AREA Or LAKE
(1122)
4
1,0
FIG. 2.-Area-species curves for 14 northernNorth American and 14 African
lakes. North America (triangles): Opeongo, 1; Caynga, 2; Seneca, 3; :Kootenay, 4; Keller, 5; Big Trout, 6; La Ronge, 7; Ontario, 8; Erie, 9; Great
Slave, 10; Great Bear, 11; Michigan, 12; H~uron, 13; Superior, 14. Africa
2; :Kivu, 3; Tana, 4; Rukwa, 5; Edward, 6;
(circles): Nabugabo, 1; C:hilwva,
Bangweulu, 7; Mweru, 8; Albert, 9; Rudolf, 10; Chad, 11; Malawi, 12; Tanganyika, 13; Victoria, 14.
478
THE AMERICAN
NATURALIST
DISCUSSION
Lakes are insular habitats. It is temptingto assume that they acquire
and lose species in the same manner as oceanic islands or some isolated
habitats-so that species diversityrepresentsan equilibriumbetweenrates
of contemporarycolonization and extinction (MacArthur and Wilson
1967; Simberloffand Wilson 1969; Diamond 1969; Vuilleumier 1970;
Culver 1970). However, fishescannot disperse across terrestrialbarriers.
They colonize lakes only when suitable bridges of aquatic habitat connect
the lakes with otherbodies of water containingfishes.Lakes normallyare
colonized fromrivers and streams,but colonistsconceivablycould immigrate throughdirect connectionswith anotherlake or with the ocean. For
colonizationis zero or nearlyso. Many
mostlakes,the rate of contemporary
lie in interiorbasins where they are completelysurroundedby terrestrial
barriers.Othersare drained by river systemsand must already have been
colonizedby all of the available fluvialfishescapable of exploitinglacustrine
habitats, so there is no significantimmigrationat present. Fish species
diversityin lakes should not representan equilibriumbetweencolonization
and extinctionin the sense of MacArthurand Wilson.
The diversityof fishesin lakes depends on the effectsof two kinds of
events.First, theremustbe a colonizationepisode in whichthe lake is connectedby a water bridgeto a source of fishspecies. This connectionis normallya riveror a streamwhichacts as a filter,permittingonlyselectedtaxa
to immigrateinto the lake. Lacustrine and marine species must be able to
pass throughthe fluvialhabitats,and fiuvialspecies must at least be able
to spend most of their lives in lacustrinehabitats (even if they reproduce
in tributarystreams) if theyare to colonizethe lake. The periods of colonization are likelyto be briefand to followimmediatelythe geologicalevents
which establishaquatic connectionsbetweena lake and other waters containing fishes.After the colonization episode has provided the original
fauna, diversitycan be alteredby interactionsamong species, and between
species and the lacustrine environment.Extinctionsowing to small population size, and oftencompoundedby the effectsof competition,predation,
or changesin the aquatic habitat,may reduce the diversityof the fauna.
time and environmentaldiversity,additional
However,if thereis sufficient
colonization, intralacustrinespeciation, or speciation in satellite lakes
(Greenwood1964) may increasespecies diversity.
The unique and importantrole of historyin biogeographyprobably is
nowherebetterdemonstratedthan in the distributionof freshwaterfishes.
in faunal studiesand by the students
This has been repeatedlydemonstrated
of particulargroups of fishes(e.g., Darlington 1957; Smith 1966; Pflieger
1971; Jenkins,Lachner, and Schwartz 1972). Ancient drainage patterns
can oftenbe worked out on the basis of taxonomicrelationshipsof fishes.
However, much variation in species diversitybetween islands or habitat
islands can be accountedforon thebasis of ecologicalvariablessuch as island
size, climate, and habitat diversity.Are there also general patterns of
FISH
SPECIES
DIVERSITY
479
lacustrinefishspecies diversitythat can be explained in termsof ecological
characteristicsof lakes? To what extentcan we detectsuch overall patterns
despitethe pervasiveinfluenceof historyon freshwaterfishdistribution?
We have attemptedto answer these questions using multiple stepwise
regressionto analyze the relationshipbetweenthe number of fishspecies
and two environmentalvariables for 70 lakes of the world. We also performedmoredetailed analyseson two morelocalized (North Americanand
African) groups of lakes for whichadditional kinds of environmentaldata
were available. Multipleregressionis an a posteriorimethod,and its results
should be interpretedwith caution, particularlyin inferringcausal relaIn general,our resultsprovide interestingpreliminaryinformationships.tion on the factorsinfluencinglacustrinefishspecies diversity.
If we firstconsiderall 70 lakes of the world for which we have data,
surfacearea accountsfor about 30%oof the variabilityin fishspecies diversity. This figureis quite large despite the very wide geographicrange and
in age, geographicisolation, ecology,and fishtaxa among
the differences
thelakes. However,extremesof someof thesefactorsdo accountforextreme
deviationsfromthe regressionline in figure1. For example,lakes in northernmostNorth America and Asia such as Great Bear (17; numbersrefer
to fig.1 and Appendix) and Taimyr (49) are cold and unproductive.The
Aral Sea (35) and Lake Balkhash (37) lie in interiorbasins of Asia; the
latter is partly saline and extremelyshallow,while the formerhas had a
complexhistorywhichapparentlyincluded periods of high salinity (Zenkevitch1963). Issyk Kul (42) and Lake Titicaca (33) lie in isolated interior
basins at highaltitudes.Lakes Chapala (29), Paitzcuaro(30), and Zirahuen
(31) lie at fairly high elevations on the faunisticallydepauperate Mesa
Central of Mexico. All the above lakes contain relativelyfew fishspecies
for their size. In contrast,lakes of relativelyhigh diversityinclude small
temperatelakes such as Walnut (68) and Waccamaw (67), whichare more
productive and lie in faunisticallyricher areas. The largest number of
lacustrinefishesare found in the lakes of tropical Africa such as Malawi
(7), Tanganyika (13), and Victoria (14).
It is interestingthat latitude accounts for only a small fraction (3.5%o)
of the totalvariabilityin fishspecies diversitywhen viewed on a worldwide
scale. This may be due in part to the fact that the aquatic environment
tends to be climaticallybufferedby the physical propertiesof water. Its
major cause, however,probablylies in the many physical barriers,some of
which were mentionedabove, which preclude the easy movementof fishes
fromdrainage to drainage. We would expect a much greatereffectof latitude on the diversityof terrestrialorganismson islands or habitat islands.
The species-arearelationshipis logarithmicand can be describedby the
exponentialequation S = CAZ- where S and A are numberof species and
area, respectively,and C and z are fittedconstants.For the lakes of the
world,the species-areacurvehas a low slope (z = 0.15) comparedwiththat
forotherorganismsinhabitingislands or isolated habitats (MacArthurand
Wilson 1967; Vuilleumier 1970; Brown 1971). This is interestingbecause
480
THE AMERICAN
NATURALIST
normallyisolated sample areas such as islands are thoughtof as having
steeperspecies-areacurvesthan nonisolatedsample areas withincontinuous
habitats (Preston 1962a, 1962b). Fishes in lakes are obviouslymoreisolated
than many terrestrialorganismson islands or habitat islands. We suggest
two explanations,one ecological and the otherhistorical,for the low slope
of the species-areacurve. First, lakes may be more homogeneousthan isolated terrestrialhabitats,so that the diversityof habitats and resources
increasesmore slowlywith area in lakes than it does on land. This would
obviouslyaffectthe abilityof lakes to supportcoexistingfishspecies. Alternatively,historicaleventsmay tend to preventlarge lakes fromacquiring
as manyspecies as theycan supportecologically.The filteringnature of the
fluvialconnectionscoupled with the slow biogeographicmovementof fishes
resultsin a relativelysmall numberof taxa available to colonize a lake;
immigrantspecies may fillsmall lakes to capacity but not large ones. As a
result,speciationis requiredto filllarger basins,and theresimplymay not
have been sufficient
time since the initial colonizationepisode to reach an
equilibriumbetweenspeciation and extinction.These alternativeexplanations are reexaminedbelow,whenwe discuss diversityin the African lakes.
The lakes in the northernUnited States and Canada representa geographicallylocalized subsampleof temperateand subarcticlakes for which
we wereable to obtain data on waterquality and climate.The analysis indicates that the numberof fishspeciesin theselakes is related to surfacearea,
latitude, and climaticvariables correlatedwith latitude (mean July temperature,growingseason, and mean annual rainfall). These variables account for about 90% of the variabilityin species diversitywhen log-transformeddata are used. When the linear model is used, thereare statistically
significantcorrelations(at the .05 level) betweenspecies diversityand both
pH and altitude. It is unlikelythat these correlationsare biologicallysignificantbecause of the small absolutevariationin the independentvariables.
The slope of the species-areacurve (z = 0.16) is almostthe same as for the
total sample of lakes throughoutthe world. Two phenomenaprobably account for the strongcorrelationbetweenlatitude (or related climaticvariables) and species diversity.First, there is probably a direct effectof
climateon species diversityof fishesin aquatic habitatsjust as thereis for
terrestrialorganisms.Comparedwiththe moresouthernlakes, the northern
ones tend to be cold and unproductive,and the growthand reproductionof
aquatic organismsare restrictedto the midsummermonths.Fish must have
specialized adaptationsin order to toleratesuch climaticextremes.For the
reasons given below,there has probablynot been sufficient
time for many
taxa to acquire such adaptations.Second, the historicalpatternof colonization of theselakes is relatedto latitude; the morenorthernlakes have been
formedmost recentlyby the retreat of the Wisconsin ice sheet. A few
species apparentlyhave colonizedtheselakes by movingeastwardalong the
marginsof retreatingglaciersfromunglaciatedAlaska. However,the major
sourceof colonistshas beenthe diversefaunas of theriversof the Mississippi
Basin and the Atlantic Coast (McPhail and Lindsey 1970). Thus, latitude
FISH
SPECIES
DIVERSITY
481
tends to be correlatedwith distance fromthe source of colonistsand the
severityof barriersposed by the filterbridges.
The patternof fishspecies diversityin the 14 lakes of tropical central
fromthat in the NorthAmericanlakes. In
and east Africa is very different
the linear model,maximumdepth and surfacearea account for about 85%o
of the variability in species numbers,with latitude explaining an additional 5%o.In the nonlinearmodel,surface area and conductivityaccount
for about 71% of the variability,withmaximumdepth explainingan additional 8%. Of particularinterestis the relativelysteep slope (z = 0.35) of
the species-areacurve. This figureis similar to the highestvalues obtained
forterrestrialorganismsinhabitingoceanic islands (MacArthurand Wilson
1967), and it approximatesthe value (z = 0.43) obtainedby Brown (1971)
for the mammalson isolated mountaintopswherethereis no contemporary
colonization.The steep slope for the African lakes contrastswith the much
shallowerslopes (z = 0.15-0.16) for the lakes of the world and the North
Americanlakes. Comparedwithmany otherlacustrinesystems,some of the
African lakes are relativelyold and the center of a great deal of rapid
speciation,particularlyin the family Cichlidae (Greenwood 1964; Fryer
and Iles 1972). They are also situated in the centerof a rich faunal area.
Apparentlytime and other conditions (such as the formationof satellite
lakes, the heterogeneousdistributionof substratetypes,and the biological
characteristicsof cichlidswhichappear to preadapt themfor rapid speciation) have been adequate for speciationto fillthe larger of theselakes and
at least approach an equilibriumwith extinction.
This suggeststhat the low species-areacurvesfor otherlacustrinesystems
are the resultof a small numberof initial colonistsand inadequate timeand
opportunitiesfor speciation to fill the large lakes to equilibria. This is
supportedby the fact that only in the African lakes does depth contribute
to determiningthe numberof fishspecies. The fluvialspecies
significantly
which normallycolonize lakes initially tend to occupy habitats such as
shallowwaters,inletsof tributarystreams,and surge zones along the shore
whichare similarto thosetheypreviouslyexploitedin the streams (Corbet
1961). One would expect that it would require a period of speciation and
adaptive radiation to produce fishes of sufficientecological diversityto
exploit all habitats and resources of large lakes potentiallysuitable for
fishes.Apparentlywhen this stage is reached the diversityof habitats is
relatedto depth as well as surfacearea.
Conductivity,an estimationof the degree of mineralizationof water,
varies in the African lakes from25-26 micromhosfor Lake Nabugabo to
1,630-12,000micromhosfor Lake Chilwa. The lakes also vary in their concentrationsof individual anions; Lake Nabugabo is very low in Ca, Lakes
Rukwa and Rudolf are rich in Na, and Lake Kivu is rich in Mg. These data
suggest that osmoregulatoryor other physiologicaldifficultiesmay have
preventedthe colonizationof certainlakes or caused extinctionsas minerals
slowlyaccumulatedin the basins withtime.However,bear in mind that the
volcanic and tectonicactivitywhich resulted directlyor indirectlyin the
482
THE AMERICAN
NATURALIST
increasedmineral contentof these waters also created geographicbarriers
to fishmovements.Until these factorscan be separated,the significanceof
physiologyas a determinantof fish species diversityin this region will
remainunclear.
Finally, we emphasizethat our subsamplesof African and North American lakes were determinedby the availabilityof physical and limnological
data. Had many more lakes fromboth regions been included, we predict
that both curves would have shown pronouncedtoes to the left of their
point of intersection.This is the result of therebeing a relativelygreater
numberof fishspecies available in both regionsto colonizesmall lakes from
streamsand rivers.Further,the veryhighnumbersof speciespresentin the
smaller North American lakes (fig. 1) suggest that the smaller lakes of
the temperateand the tropics will show comparablelevels of fish species
diversity.To the right of the intersectionthe tropical curve, for reasons
discussedabove,will rise muchmoresteeplyas space becomesless of a limiting factor. This interpretationis consistentwith the findingsof Patrick
(1966) for tropical and temperaterivers.
It is becomingapparent that thereare several distinctpatternsof species
diversityfor organismsdistributedon islands or insular habitats.Since all
islands lose species by the same process (extinction),the various patterns
are determinedby how species are acquired. We distinguishfour patterns
of insularspeciesdiversityas follows:
1. Many insular biotas representequilibria between colonization and
extinction(MacArthur and Wilson 1967; Diamond 1969; Simberloffand
Wilson 1969; Culver 1970; Vuilleumier1970). This occurs wheneverthere
is a significantand continual immigrationof potential colonists. Colonization-extinctionequilibria tend to result in species-area curves with z
values in the range 0.20-0.35 (MacArthur and Wilson 1967). The remaining patternsoccurwhencolonizationis not continualbut confinedto periods
when the islands were connectedby bridges of suitable habitat to an area
containingsuitable colonists.
2. Some biotas should representequilibria between speciation and extinction,provided that the area of the island in question is large enough
to allow speciationto occur. This takes place whenhabitatbridgesadmitted
a relativelysmall numberof colonistssufficiently
long ago for speciationto
reach an equilibriumwith extinction.In this case the slope of the speciesarea curve should be steeper than for colonization-extinction
equilibrium
biotasbecause of theslow rate of speciation(relativeto colonization)and the
highrate of extinctionon small islands.We expectfishfaunas of the tropical
African lakes to be closer to such an equilibriumthan any of the other
lacustrine assemblages we have studied because of the great amount of
speciation which has occurred within the Cichlidae. On the other hand,
the smaller lakes of our sample are probablytoo small to allow extensive
multiplicationof species; theirfaunas are largelythe resultof colonization.
It is encouraging,however,that the z value for the African faunas is 0.35,
at the upper extremeof the range for colonization-extinction
equilibrium
FISH
SPECIES
DIVERSITY
483
biotas. The two precedingpatternsrepresentequilibrial situationson differenttime scales. The two remaining patterns representnonequilibrial
situationsbecause the habitatbridgeshave been relativelyrecent.
3. Islands have less than equilibrial numbers of species when small
numbersof colonizerscrossedthe habitatbridgestoo recentlyforsignificant
subsequent speciation. In this case, large islands tend to be particularly
impoverishedand the slope of the species-area curve (at least initially)
will be less than that for a colonization-extinction
equilibriumbiota. The
fishfaunas of many large, geologicallyrecent lakes apparently represent
such a subequilibrialpattern; for samples including such lakes we obtain
z values in the range 0.15-0.16.
4. Islands have more than equilibrial numbersof species when a large
numberof species (often an entire"mainland" biota) crossedthe habitat
bridgestoo recentlyfor subsequentextinctionsto reduce the biotas to equilibrium.The absenceof colonizationand thehighrate of extinctionon small
islands combineto produce species-areacurveswith steeperslopes than for
biotas characterizedby colonization-extinction
equilibria. Brown (1971)
showedthat the boreal mammalfauna of isolated mountainsin the western
United States representedsuch a supraequilibrial fauna; he obtained a
species-areacurvewith a z value of 0.43.
SUMMARY
Stepwisemultipleregressionwas used to obtain preliminaryinsightsinto
the environmentalparameterswhich influencethe number of fish species
occurringin lakes. Results are summarizedas follows:
1. For a sample of 70 lakes and inland seas fromthroughoutthe world,
surfacearea and latitude account for about one-thirdof the variabilityin
fishspeciesdiversity.
2. For a subsample of 14 North American lakes, latitude and surface
area accountforabout 90% of the variabilityin species numbers.The large
effectof latitudecan be explained in termsof climaticseverityand isolation
fromsourcesof colonization.
of 14 African lakes, surface area, depth, and con3. For a subsammple
ductivitywere primaryvariables affectingspecies diversity.
4. For 70 lakes of the world and 14 North Americanlakes, the slope of
the species-areacurve was low (z- 0.15-0.16). Apparentlythis reflectsthe
fact that,whereasa relativelylarge numberof species are available to colonize small lakes, larger ones are impoverished.Most of the latterare recent
enough that there has not been time for endemic speciation to reach an
equilibriumwithextinction.We expectthe large lakes of tropical Africa to
be the closestto such an equilibrium,and it is encouragingthat the slope
of the species-areacurve for the sample includingthese bodies of water is
quite steep (z = 0.35).
5.. On the basis of our data and analyses of species distributionin other
insular habitats,four patternsof species diversityare distinguished.
484
THE AMERICAN
NATURALIST
ACKNOWLEDGMENTS
Numerouspeople aided this study with criticism,discussion,suggestions,
and in some cases unpublisheddata. The followingwere particularlyhelpful: J. Addicot, R. M. Bailey, W. Y. Brockelman,W. R. Bussing, F. P.
Cichocki,G. Fryer, D. M. Gillespie, W. A. Gosline,P. H. Greenwood,J.
Kethley,K. IKurawaka,J. D. McPhail, R. R. Miller, N. C. Negus, R. A.
Ryder,W. B. Scott, D. J. Stewart,and S. H. Smith. We especially thank
H. Wilbur for help in a preliminaryanalysis of the data. Computertime
was made possibleby fundsmade available to R. R. Miller throughthe Universityof MichiganMuseumof Zoologyfor that purpose. Finally we thank
G. Fryer and T. D. Iles; the original publisher,Oliver and Boyd; and the
Americanpublisher,T. F. H. Publications,for permissionto analyze data
appearingin theirbook,The CichlidFishesof theGreatLakes of Africa:
TheirBiologyand Evolution.
FISH
SPECIES
485
DIVERSITY
APPENDIX
NUMBER OF FISH SPECIES, SURFACE AREA, AND LATITUDE FOR 70 LAKES AND SEAS
OF THE WORLD
Number
of Species
Africa:
1. Albert ....................
2. Bangweulu .68
3. Chad .93
4. Chilwa .13
5. Edward .53
6. Kivu .17
7. Malawi .245
8. Mweru .88
9. Nabugabo .24
10. Rudolf .37
11. Rukwa .22
12. Tana .18
13. Tanganyika .214
14. Victoria .177
Canada:
15. Athabasca .21
16. Big Trout (Ontario) .
17. Great Bear .12
18. Great Slave .26
19. Keller .13
20. Kootenay .19
21. La Ronge .19
22. Opeongo ......22
46
24
Great Britain:
23. Loch Lomond .15
24. Windermere.9
Surface
Area (kM2)
5,346.0
2,072.0
17,500.0
673.0
2,150.0
2,370.0
28,490.0
4,413.0
28.5
9,065.0
3,302.0
3,626.0
32,893.0
69,484.0
7,154.0
616.0
31,153.0
27,195.0
406.0
399.0
1,425.0
60.0
Latitude
1.70 N
11.10 S
13.00 N
15.30 5
0.50 S
2.00 S
12.00 S
9.00 S
0.60 S
3.50 N
8.00 S
12.00 N
6.00 S
1.00 S
59.20 N
53.80 N
66.00 N
61.40 N
64.00 N
49.50 N
55.00 N
45.70 N
71.0
15.0
54.30 N
Guatemala:
25. Pet6n .23
26. Yzabal .48
98.0
684.0
17.00 N
15.50 N
Italy:
27. Maggiore .21
212.0
46.00 N
Japan:
28. Biwa .46
676.0
35.20 N
Mexico:
29. Chapala .14
30. Pftzcuaro .7
31. Zirahuen .5
1,080.0
111.0
8.0
20.20 N
19.60 N
19.40 N
Nicaragua-Costa Rica:
32. Nicaragua .40
8,264.0
11.50 N
Peru-Bolivia:
33. Titicaca .18
9,065.0
16.00 S
Philippine Islands:
34. Lanao .20
357.0
7.90 N
64,500.0
31,500.0
18,500.0
1,125.0
423,488.0
436,000.0
165.0
6,206.0
45.00 N
54.00 N
46.00 N
60.20 N
43.00 N
42.00 N
51.20 N
42.00 N
Soviet Union:
35. Aral Sea .17
36. Baikal .50
37. Balkhash .5
38. Beloe (Vologda region)
39. Black Sea .156
40. Caspian Sea .74
41. Gusinoe .13
42. Issyk Kul .................
22
11
56.10 N
THE AMERICAN
486
APPENDIX
Number
of Species
43.
44.
45.
46.
47.
48.
49.
50.
Ladoga ...................
Leprindo .14
Onega ....................
Pestovo .17
Sea of Azov .17
.21
Seliger
Taimyr .13
Teletskoe .14
United States:
51. Black (North Carolina)
52. Canandaigua .37
53. Cayuga .60
.113
54. Erie
55. Huron .99
56. Jones (North Carolina)
57. Keuka
58. Michigan .114
59. Ontario .112
60. Otisco .17
61. Owasco .10
62. Salters (North Carolina)
63. Seneca .39
64. Singletary (North Carolina) .
65. Skaneateles .14
66. Superior .67
67. Waccamaw .36
68. Walnut (Michigan) .30
69. White (North Carolina)
Yugoslavia-Romania:
70. Ohrid .17
48
28
10
13
30
14
14
19
NATURALIST
(Continued)
Surface
Area (kM2)
Latitude
18,400.0
24.0
10,340.0
2.0
38,000.0
221.0
4,650.0
231.0
61.00 N
56.50 N
61.50 N
58.30 N
46.00 N
57.20 N
74.50 N
51.60 N
5.2
41.0
171.0
25,719.0
59,596.0
0.8
44.0
58,016.0
19,477.0
10.0
85.0
1.3
174.0
2.6
54.0
82,414.0
36.0
0.8
5.2
34.70
42.80
42.70
42.20
44.50
34.70
42.50
44.00
43.50
347.0
N
N
N
N
N
N
N
N
N
42.80 N
42.80
34.70
42.60
34.60
42.80
47.50
34.30
42.60
34.60
N
N
N
N
N
N
N
N
N
41.00 N
NOTE.-Data in this Appendix are from the following sources. Africa: United States
Board on Geographic Names (1955a); Greenwood (1964); Fryer and Iles (1972).
Canada: Canadian Board of Geographic Names (1953, 1958); Canadian Permanent
Committeeof Geographic Names (1969); Department of Mines and Technical Surveys,
Geographical Branch (1962); McPhail and Lindsey (1970); Ryder (1972). Great
Britain: Le Cren, Kipling, and McCormack (1972); Maitland (1972). Guatemala:
United States Army Map Service (1940a, 1940b; surface areas by planimetry); R. R.
Miller (personal communication). Italy: United States Board on Geographic Names
(1956b); Grimaldi (1972). Japan: United States Board on Geographic Names (1955b);
Horie (1969); K. Kurawaka (personal communication). Mexico: De Buen (1943, 1944);
United States Board on Geographic Names (1956a); C. D. Barbour (unpublished).
Nicaragua-Costa Rica: United States Board on Geographic Names (1956c); W. R.
Bussing (personal communication). Peru-Bolivia: Eigenmann and Allen (1942);
Tehernavin (1944); Brooks (1950). Philippine Islands: Myers (1960); Frey (1969).
Soviet Union: Kozhov (1963); Zenkevitch (1963); Zhadin and Gerd (1963); United
States Board on Geographic Names (1970). United States: Hankinson (1908); Greeley
(1927); Frey (1949); Louder (1962); Ryder (1972). Yugoslavia-Romania: Stankovic
(1960). The primary source for surface areas was Gresswell and Huxley (1965). Many
of the latitudes were calculated from Bartholomew (1955-1959) or appropriate topographic maps.
FISH
SPECIES
DIVERSITY
487
LITERATURECITED
Bartholomew, J. 1955-1959. The Times atlas of the world. Mid-Centuryed. Vols. 1-6.
Times, London.
Brooks, J. L. 1950. Speciation in ancient lakes. Quart. Rev. B3iol.25:30-176.
Brown, J. H. 1971. Mammals on mountaintops: non-equilibriuminsular biogeography.
Amer. Natur. 105:467-478.
Canadian Board of Geographic Names. 1953. Gazetteer of Canada: British Columbia. Gov.
Printing Office,Ottawa. 641 pp.
1958. Gazetteer of Canada: Northwest Territories and Yukon. Gov. Printing
Office,Ottawa. 89 pp.
Canadian Permanent Committeeof Geographic Names. 1969. Gazetteer of Canada: Saskatchewan. Gov. Printing Office,Ottawa. 173 pp.
Corbet, P. S. 1961. The food of non-cichlid fishes in the Lake Victoria basin, with
remarks on their evolution and adaptation to lacustrine conditions. Proc. Zool.
Soc. London 136:1-101.
Culver, D. C. 1970. Analysis of simple cave communities.I. Caves as islands. Evolution
29:463-474.
Damnas,H. 1937. Recherches bydrobiologiques dans les Lacs Kivu, Edouard, et Ndalaga.
Institut des Pares nationaux de Congo belge. Exploration du Pare national
Albert. No. 2. Fase. 1. Mission H. Damas, 1935-36. 128 pp.
Darlington, P. J. 1957. Zoogeography: the geographical distribution of animals. Wiley,
New York. 675 pp.
De Buen, F. 1943. Los lagos Michoacanos. I. Caracteres generates. El Lago de Zirahuen.
Rev. Soc. Mexicano Hist. Nat. 4:211-232.
1944. Los lagos Michoacanos. II. Patzeuaro. Rev. Soc. Mexicano Hist. Nat. 5:99125.
Department of Mines and Technical Surveys, Geographical Branch. 1962. Gazetteer of
Canada: Ontario. Gov. Printing Office,Ottawa. 614 pp.
Diamond, J. M. 1969. Avifaunal equilibria and species turnover rates on the Channel
Islands of California. Proc. Nat. Acad. Sci. 64:57-63.
Draper, N. R., and H. Smith. 1966. Applied regressionanalysis. Wiley, New York. 407 pp.
Eigenmann, C. H., and W. R. Allen. 1942. The fishes of western South America. University of Kentucky, Lexington. 494 pp.
Frey, D. G. 1949. Morphometryand hydrographyof some natural lakes of the North
Carolina coastal plain: the bay lake as a morphometrictype. J. Elisha Mitchell
Sci. Soc. 65:1-37.
1969. A limnological reconnaissance of Lake Lanao. Verh. Int. Verein Limnol.
17:1090-1102.
Fryer, G., and T. D. Iles. 1972. The cichlid fishes of the great lakes of Africa: their
biology and evolution. Oliver & Boyd, Edinburgh. 641 pp.
Goldman, E. A. 1951. Biological investigations in Mexico. Smithsonian Misc. Collection,
vol. 115.
Greeley, J. R. 1927. Fishes of the Oswego watershed. Pages 95-107 in A biological
survey of the Oswego river system. Vol. 4. State of New York, Conservation
Department, Supplement to the SeventeenthAnnual Report.
Greenwood,P. H. 1964. Explosive speculationin African lakes. Proc. Roy. Inst. 40:256269.
Gresswell,R. K., and A. Huxley, eds. 1965. Standard encyclopedia of the world's rivers
and lakes. Putnam, New York. 384 pp.
Grimaldi, E. 1972. Lago Maggiore: effects of exploitation and introduction on the
salmonid community.J. Fisheries Res. Board Canada 29:777-785.
Hamilton, T. H., R. H. Barth, and I. Rubinoff. 1964. The environmentalcontrol of
insular variation in bird species abundance. Proc. Nat. Acad. Sci. 52:132-140.
Hankinson, T. L. 1908. A biological survey of Walnut Lake, Michigan. Pages 155-288
488
THE AMERICAN NATURALIST
in Reportof the Biological Surveyof the State of Michigan,State Board of
GeologicalSurvey.
Horie,S. 1969.Late Pleistocenelimnetichistoryof JapaneseancientLakes Biwa, Yogo,
Suwa, and Kizaki. Mitt.Int. VereinLimnol.17:436-445.
for TechnicalCoJackson,S. P., ed. 1961. Climatologicalatlas of Africa.Commission
operationin Africa South of the Sahara, JointProject No. 1, Lagos. 55 pp.
Jenkins,R. E., E. A. Lachner,and F. J. Schwarz.1972. Fishes of the centralAppalachian drainages,theirdescriptionand dispersal.Pages 43-117 in P. C. bolt,
historyof the biota of the southernAppalachians.III.
ed. The distributional
Inst. Monogr.4.
VirginiaPolytechnic
Vertebrates.
Kozhov,M. M. 1963.Lake Baikal and its life.Monogr.Biol. 11:1-344.
effectsof exploita1972.Windermere:
Le Cren,E. D., C. Kipling,and J. C. McCormack.
J. FisheriesRes. Board
on the salmonidcommunity.
tion and eutrophication
Canada 29:819-832.
Louder,D. E. 1962. An annotatedchecklist of the NorthCarolina bay lakes fishes.
J. Elisha MitchellSci. Soc. 78:68-73.
R. H. 1972. Geographicalecology.Harper & Row, New York. 269 pp.
MacArthur,
Monogr.
R. H., and E. 0. Wilson.1967. The theoryof island biogeography.
MacArthur,
Pop. Biol. 1:1-203.
Canada
fishesof northwestern
McPhail,J. D., and C. C. Lindsey.1970. The freshwater
and Alaska. FisheriesRes. Board Canada Bull. 173.
J. FishMaitland,P. S. 1972. Loch Lomond: man's effecton the salmonidcommunity.
eriesRes. Board Canada 29:849-860.
Myers,G. S. 1960. The endemicfishfauna of Lake Lanao, and the evolutionof higher
taxonomiccategories.Evolution14:323-333.
Patrick,R. 1966. The CatherwoodFoundationPeruvian Amazon expedition.Mongr.
Acad. Nat. Sci. Philadelphia14.
studyof Missourifishes.Univ. Kansas Pub., Mus.
Pflieger,W. L. 1971. A distributional
Natur. Hist. 20:225-570.
and rarity.I. Ecology
of commonness
Preston,F. W. 1962a. The canonicaldistribution
43:185-215.
and rarity.II. Ecology43:410of commonness
. 1962b.The canonicaldistribution
432.
Ryder,R. A. 1972. The limnologyand fishesof oligotrophicglacial lakes in North
America(about 1800 A.D.). J. FisheriesRes. Board Canada 29:617-628.
of islands: the
D. G., and E. 0. Wilson.1969. Experimentalzoogeography
Simberloff,
colonizationof emptyislands.Ecology50:278-296.
Smith,G. R. 1966. Distributionand evolutionof the NorthAmericancatostomidfishes
Misc. Pub. Mus. Zool. Univ.
of the subgenusPantosteus,genus Catostomius.
Michigan129.
Stankovic,S. 1960. The Balkan Lake Ohridand its livingworld.Monogr.Biol. 9:1-357.
V. V. 1944. A revisionof the subfamilyOrestiinae.I and II. Proc. Zool.
Tchernavin,
Soc. London114:140-233.
UnitedStates ArmyMap Service.1940a. CentralAmerica,Series E501, 1:250,000sheets
PrintingOffice,
NE 16S IV, NE 16S 1, NE 15S VI, ed. 1-AMS. Government
D.C.
Washington,
1940b.CentralAmerica,Series E501, 1:250,000,sheetsND 16N 1. Government
D.C.
Washington,
PrintingOffice,
UnitedStates Board on GeographicNames.1955a. GazetteerNo. 1, BritishEast Africa.
D.C. 601 pp.
Washington,
PrintingOffice,
Government
D.C.
Washington,
PrintingOffice,
. 1955b. GazetteerNo. 12, Japan. Government
731 pp.
D.C.
Washington,
PrintingOffice,
. 1956a. GazetteerNo. 15, Mexico.Government
750 pp.
FISH SPECIES DIVERSITY
489
PrintingOffice,
* 1956b.GazetteerNo. 23, Italy and associatedareas. Government
D.C. 369 pp.
Washington,
Washington,
PrintingOffice,
* 1956c. GazetteerNo. 25, Nicaragua. Government
D.C. 49 pp.
PrintingOffice,
1970. GazetteerNo. 42, U.S.S.R. Vols. 1-7. 2d ed. Government
D.C.
Washington,
Vuilleumier,F. 1970. Insular biogeographyin continentalregions.I. The northern
Andes of South America.Amer.Natur. 104:373-388.
New York.955 pp.
L. 1963.Biologyof theseas of theU.S.S.R. Interscience,
Zenkevitch,
Zhadin,V I., and S. V. Gerd.1963. Fauna and floraof the rivers,lakes and reservoirs
of the U.S.S.R. Israel Programfor ScientificTranslation,Jerusalem.626 pp.