The global abundance and size distribution of lakes ponds and impoundme nts J - PDF document

J.J.Cole InstituteofEcosystemStudies,BoxAB,Millbrook,NewYork12545 C.M.Duarte InstitutMediterranid’EstudisAvanc ¸ats(ConsejoSuperiordeInvestigacionesCient ´ficas,UniversitatdelesIllesBelears), MiquelMarques21,Esporles,IslasBaleares,Spain L.J.Tranvik Limnology,DepartmentofEcologyandEvolution,EvolutionaryBiologyCe ntre,Norbyv.20,SE-75236Uppsala,Sweden R.G.Striegl fornia,SantaBarbara,California93106 J.J.Middelburg NetherlandsInstituteofEcology,Korinaweg7,Yerseke,4401NT,Netherl ands Abstract Oneofthemajorimpedimentstotheintegrationoflenticecosystemsintog lobalenvironmentalanalyseshas beenfragmentarydataontheextentandsizedistributionoflakes,ponds, andimpoundments.Weusenewdata sources,enhancedspatialresolution,andnewanalyticalapproachestop rovidenewestimatesoftheglobal ributionshowsthattheglobalextent ofnaturallakesistwiceaslargeaspreviouslyknown(304millionlakes;4 .2millionkm 2 inarea)andis dominatedinareabymillionsofwaterbodiessmallerthan1km 2 .Similaranalysesofimpoundmentsbasedon inventoriesoflarge,engineereddamsshowthatimpoundedwaterscoverap proximately0.26millionkm 2 . processeshavebeenlimitedbecauseknowledgeofthe numberandsizedistributionoflakeshasbeenincomplete (Alsdorfetal.2003;LehnerandDo ¨ll2004).Further, Limnol.Oceanogr., 51(5),2006,2388–2397 E 2006,bytheAmericanSocietyofLimnologyandOceanography,Inc. 2388 pondscanbeconsideredtobe0.001km 2 andthemaximum lakesizethatoftheCaspianSea(378,119km 2 ).The shapeparameter c istheexponentdescribingtherateat whichprobabilitydeclineswithincreasingsize.Thus, c canbeefficientlyestimatedasthenegativeslopeofthe plotofthelogarithmoftheprobabilitythatalake chosenatrandomwillbeofarea( a )greaterthan A as afunctionofthelogarithmof A .Inotherwords, c isequal to  b inEq.3.IfageneralParetodistributionforworld lakescanbediscoveredandshowntohaveinterregional Fig.2.Relationshipbetweenlakesurfaceareaandarealfrequenciesofdi fferentsizedlakes, determinedbydetailedGISanalyses( seesourcesinTable1 )andcomprehensivecounts.Thefilled dotsindicatethefrequenciesoflakesizesfoundbySchuiling(1977)andM eybeck(1995).Other regions’lakefrequenciesareshownbyvarioussymbols. Fig.3.Geographicalanalysisofthepredictedworlddistributionofdens ities(d L ;Eq.2)oflakesbetween1km 2 and10km 2 surface area.PredictionsfollowaworldGISmodelofannualrun-off(Feketeetal. 2005)withageographicalresolutionof0.5 u oflatitudeand longitude.Lakedensitiesareshowninlakesper10 6 km 2 . Lakeabundanceandsizedistribution 2391 generality,thenwecancalculatetheglobalextentofall sizesoflakes. FitoflakeareafrequenciestotheParetodistribution— To testlakesizedistributionsforgeneralfittothePareto distribution,wecollectedinventoriesofalllakeswithin avarietyofgeographicalsettingsrepresentingdivergent topographyandgeology.Datanotderivedfrompublished sources(Table1)weredeterminedfromregionalGIS analysesusingARCView(ESRI).Datawerescrutinized forevidenceofundersamplingatsmalllakesizestoinclude onlyuntruncated,uncensoreddata(Hamiltonetal.1992). Figure4andTable1showstronginterregionalsimilarity intheslopesofthesedistributionalcurves.Because differentregionsholddifferingtotalnumbersoflakes andtheirlargestlakesdifferinsize,curvesarelocatedat differentpointsalongtheabscissa.TheParetodistribution showsasimilarrateofdeclineinabundancewithincreased lakesize,regardlessoftheregionoftheearthexamined. Giventhatthesize–frequencydistributionsoflakes followaParetodistributioninmanyregionsdowntovery smalllakesizes(Fig.4),canonicaldataontheabundance oftheworld’slargestlakesshouldenabletheanchoringof Eq.3andthecalculationoftheworldwideabundanceof lakes.Herdendorf(1984)inventoriedtheworld’slargest lakestodocumentthedominanceofthe251largestlakesin theworld’sfreshwatersupply(Fig.4).Heconcludedthat thegreatestimpedimenttothisworkwasthepaucityof accuratemaps.LehnerandDo ¨ll(2004)haverecentlyused GISanalysistodevelopandvalidateagloballake database.Theycombinedanaloganddigitalmapswith databases,registers,andinventoriesoflakestopresentalist of
250,000waterbodies.Consideringonlythe17,357 naturallakes
10km 2 inareaincludedintheiranalysis,we calculateEq.3byleastsquaresregression: N a
A
195,560 A  1
06079
5  ( r 2  0.998; n  17,357; SE b  0.0003).Theexponentof thisrelationship( b , Eq.3 ;– c , Eq.4 )isnearthemiddleof thoseseeninregionalanalyses(Table1)andthetwo canonicallakeareadatasetsdescribeanidenticallakesize distribution(Fig.4).Orthogonalregressionyieldeda99.9 % confidenceintervaloftheestimated b from  1.06245to  1.06056.Theaccuracyandprecisionoftheestimateof b isveryimportantbecausecalculatedlakesizedistribu- tionsandareasareverysensitivetothisparameter.Because theshapeofParetodistributionsissimilaramongdiverse regionsoftheearth(Table1;cf.,LehnerandDo ¨ll2004) andtheparametersofthisdistributionareestimablefrom thecanonicaldatasets,wecanthuscalculatetheglobal extentofpondsandlakes. Thenumberoflakesintheworldcanbecalculated followingtheapproachofVidondoetal.(1997).From Eq.5, c oftheParetodistributionis1.06.Giventhatwe definetherangeoflakesizesas0.001to378,119km 2 (the CaspianSea),integrationofEq.4willindicatethefraction ofalltheworld’slakesandpondsthatiscontainedwithin anyrangeofareas.Agoodestimateof k is0.001km 2 becausethisisthesmallestsizeofpondpractically recognizableinlandscapes.Thefractionofallworldlakes thatarerepresentedbythoseinthecanonicaldataset( f c ) canbecalculatedasfollows: f c
 k c  A  c c max  A  c c min

6  where A c min and A c max aretheminimumandmaximum areasoflakesfoundinthecanonicaldataset(i.e.,10and 378,119)and c isthenegativeoftheexponent( Eq.5 )found
 CoefficientsofEq.3fittedbyleastsquaresregression.GISdatawereder ivedfromoriginalGISanalyzedbytheauthorsof thisstudy.DatasourcesfororiginalGISanalysesarenotedifthedatawer enotthepropertyoftheauthors. Dataset Exponent ( b ,Eq.3; c ,Eq.4) Smallestreliable size(km 2 ) Numberoflakes analyzed( a )Source L’Estrie(Que ´bec,Canada)  0.660.0013,398GIS(Y.T.Prairieunpubl.data) Abitibi(Que ´bec,Canada)  0.670.0011,020GIS(Y.T.Prairieunpubl.data) Finland  0.690.00957,205(RaatikainenandKuusisto1988) EasternLakesSurvey(U.S.A.)  0.760.11,264(Linthurstetal.1986;Landersetal. 1988) WesternEurope  0.775751(Schuiling1977) WesternLakesSurvey(U.S.A.)  0.780.01752(Clowetal.2003) Amazonbasin(SouthAmerica)  0.790.14,482(Sippeletal.1992;Hamiltonetal.1992) Adirondacks(U.S.A.)  0.790.012,125GIS(J.J.Coleunpubl.data) Medianofregionalestimates  0.79 World’slargelakes  0.83–251(Herdendorf1984) Florida(U.S.A.)  0.880.055,346(Shaferetal.1986) Meanofregionalestimates  0.89 Laurentides(Que ´bec,Canada)  0.900.001562GIS(Y.T.Prairieunpubl.data) World’slargestlakes  1.061017,357(LehnerandDo ¨ll2004) Oklahoma(U.S.A.)  1.190.1444GIS(OklahomaCenterforGeospatial Information2004) Orinocobasin(SouthAmerica)  1.220.1956(HamiltonandLewis1990;Hamiltonet al.1992) NorthDakota(U.S.A.)  1.3340.0017,239GIS(NorthDakotaStateWater Commission2003) 2392 Downingetal. forthecanonicaldataset.SolvingEq.6indicatesthatthe canonicallakedatasetcontainedafractionoftheworld’s lakesequalto5.712 3 10  5 .Divisionofthenumberof canonicallakesusedtocalculateEq.5bythisfraction estimatesthatthereareintheneighborhoodof304million pondsandlakes(  0.001km 2 )intheworld.Thistotal numberofworldlakesisdefinedasN t . Further,becauseEq.3fitsconsistentlyoverawiderange oflakeareas(Fig.4andTable1)andbecauseEq.5 anchorsthiscanonicalfrequencydistributiontotheknown sizesoftheworld’slargestlakes,thenumberofworldlakes canbeapproximatedoveranysizerange.If A min and A max aretheminimumandmaximumsizesoflakesinagiven sizerange,thenumberoflakesoverasizerangecanbe calculated: N A max  A min
 N t k c A  c max  A  c min

7  (seeTable2).Because c isgreaterthanunity,Eq.7shows thattherearemanysmalllakesandfewlargelakes. Likewise,theaveragesizeandtotalareacoveredbylakes inagivensizerangecanbecalculatedusingthePareto distribution(Vidondoetal.1997).Aftersimplification,the averageareaoflakesoverasizerangeof A min to A max can becalculated:
A A A min  A max
c   A max A c min  A c max A min c  1
 A c max  A c min

8  Thetotallandareacoveredbylakesofanysizerangecan becalculatedastheproductofEqs.7and8. Analysesofcanonicallakedataandtheexponentsof ParetodistributionsoflakesizesderivedbyregionalGIS revealsomesurprises.Contrarytopreviouspredictions (Schuiling1977;Wetzel1990;Meybeck1995;Kalff2001), smalllakes,notlargeones,appeartorepresentthemost lacustrinearea.Althoughlakes  10,000km 2 inindividual lakeareaconstitutenearly1 3 10 6 km 2 ,theselakesmakeup onlyabout25 % oftheworldlakearea.Together,thetwo smallestsizecategoriesoflakesinTable2comprisemore areathanthethreelargestsizecategories.Whenconvertedto d L units(numberper10 6 km 2 ofEarth’ssurface;Table2), extensionofcanonicallakedatatosmalllakesusingthe ParetodistributiontracksRobertWetzel’sconcept(Wetzel 1990)ofthelikelyabundanceofsmalllakes(seeFig.1). Undercountingsmalllakeshasledtosignificantunder- estimatesoftheworldlakeandpondarea.Worldlakesand pondsaccountforroughly4.2 3 10 6 km 2 ofthelandarea oftheearth.Thismorethandoublesmostquantitative historicestimates(Kalff2001;Wetzel2001;Shiklomanov andRodda2003).Naturallakesandponds  0.001km 2 compriseroughly2.8 % ofthenon-oceaniclandarea;not 1.3–1.8 % aspreviouslysupposed. Reservoirsandimpoundments— Theforegoinganalysis madeeveryattempttoexcludeconsiderationofanthropo- genicimpoundmentsofwater.Artificialwaterbodiescan, however,beofgreatimportanceinmanyprocesses(Dean andGorham1998;St.Louisetal.2000)andshouldbe includedinglobalanalyses.Thesizedistributionsof naturallakesandimpoundmentsarebothextensionsof Fig.4.PlotsofdataontheaxesimpliedbyEq.3.StatisticalfitsofEq.3to thesedataare showninTable1.Dataareonlyplottedthroughouttherangeoflakesizesth atcouldbe reasonablyexpectedtobecomprehensivelycensusedusingtheresolution ofGIScoverages available(seeTable1).Theblacklinesrepresentcanonical(complete)c ensusesofworldlakes (Herdendorf1984;LehnerandDo ¨ll2004). Lakeabundanceandsizedistribution 2393 theearth’shypsometry(Strahler1952),whichisfunda- mentallyinfluencedbytheerosionalforceofthewater supply.Aslandscapesarecoveredwithimpoundedwater, however,smalldepressionsareaggregatedintolargerlakes. ExpressedintermsofaneffectontheParetodistribution, thiscouldlead c (Eq.4)tobelarger. Becausetheearthclearlyhasmoresmalldepressions thanlargeones(bothdryandwet),itislikelythataPareto distributioncouldfitforimpoundmentsaswellasfor naturallakes.Infact,Eq.2suggeststhattherearewater- richregionsoftheearththatareunderservedbynatural lakesowingtotheirtilted,erosional,river-dominated topographycausedbyveryhighrun-off(Fig.3).As humansinstallimpoundmentsinmoreofthedepressions thatwillholdwater,impoundedwaterscouldcovereven moreareathannaturallakes,whilefollowingsimilarsize distributions.Ifmorelargeimpoundmentsarebuiltthan smallones,however,therecouldbedifferencesinthe exponentsoftherelationshipsofthetypeshowninEq.3 fornaturalandimpoundedwaterbodies. Theareaimpoundedbylargedamsisincreasing worldwide.Ithasbeenestimatedthatthevolumeofwater inimpoundmentsincreasedbyanorderofmagnitude betweenthe1950sandthepresent(Shiklomanovand Rodda2003).Asanexample,Fig.5showsthetrendinthe areacoveredbyimpoundmentsintheUnitedStates.The semi-logtrendisapproximatelylinearfrom1700to present,butdeceleratedaround1960.Theannualaverage rateofincreaseinimpoundedareaduringthistermwas about4 % .Therateofincreasesince1960hasslowedto about1 % peryearperhapsbecauseoftheincreasingrarity ofvacantland. TheInternationalCommissiononLargeDams(ICOLD; www.icold-cigb.org)tracksdataondamsaroundtheworld thatareofsafety,engineering,orresourceconcern.These dataarepurposefullybiasedtowardlargedams,most notablythose
15mheight.Thedataarethuslikelyto providethebestestimateofimpoundmentswiththelargest impoundedareasandprogressivelylessexhaustivecover- ageofsmallerimpoundments.Restrictingananalysisof Eq.3tothe41largestimpoundmentsfromtheInguri impoundment(13,500km 2 )downto1,000km 2 yields N a
A
2,922,123 A  1
4919
9  ( r 2  0.97; n  41; SE b  0.0435).Thestronglynegative exponentofthisequationindicatesthatthesmallestofthe largeimpoundmentscomprisemoresurfaceareathanthe
 Thenumbers,averagesizes,andareasofworldlakescalculatedfromEqs.7 and8.Thetotalareaoflakesinthesizerange iscalculatedastheproductofcalculationsfromEqs.7and8.Valuesarein clusiveoflowerboundsbutexclusiveofupperbounds. A min (km 2 ) A max (km 2 )Numberoflakes Averagelake area(km 2 ) Totalareaof lakes(km 2 ) d L (lakesper 10 6 km 2 )Source 0.0010.01277,400,0000.0025692,6001,849,333Eqs.7,8 0.010.124,120,0000.025602,100160,800Eqs.7,8 0.112,097,0000.25523,40013,980Eqs.7,8 110182,3002.50455,1001,215Eqs.7,8 1010015,90524.7392,362106Lehnerand Do ¨ll(2004) 1001,0001,330248329,8169Lehnerand Do ¨ll(2004) 1,00010,0001052,456257,8560.7Lehnerand Do ¨ll(2004) 10,000100,0001637,978607,6500.1Lehnerand Do ¨ll(2004)
100,0001378,119378,1190.007Lehnerand Do ¨ll(2004) Alllakes304,000,0000.0124,200,000 Fig.5.RateofchangeinimpoundedareaintheUnited Statesforallwaterimpoundmentswithdamslistedaspotential hazardsorlowhazarddamsthatareeithertallerthan8m, impoundingatleast18,500m 3 ,ortallerthan2m,impoundingat least61,675m 3 ofwater(USACOE1999).Alldatawereignored wherethedateofdamconstructionwasunknown(ca.12 % of impoundedarea)ornaturallakes(e.g.,LakeSuperior)werelisted asimpoundments.Thedashedlineshowsasemi-logregression ( r 2  0.95). 2394 Downingetal. largestofthem.Theintentionalbiasofthisdatasettoward largedamsprogressivelyincreasestheexponentofthis equationassmallerimpoundmentsareincluded.The averagerelationship,consideringalloftheICOLD impoundmentsdownto1km 2 ,is N a
A
20,107 A  0
8647
10  ( r 2  0.97; n  9,604; SE b  0.0154).Integrationofthis equationcertainlyresultsinanunderestimateofthearea coveredbyimpoundments(cf.,Eqs.9and10)becauseit ignoresmanyimpoundmentsformedbysmalldams. CalculationsfollowingEqs.6–8(Table3)show,however, thatthereareatleast0.5millionimpoundments  0.01km 2 intheworld,andtheycover
0.25millionkm 2 oftheearth’slandsurface.Thisisasmallerareathan estimatesbasedonextrapolation(DeanandGorham1998; St.Louisetal.2000;ShiklomanovandRodda2003)but nearlyidenticaltoGIS-basedestimates(LehnerandDo ¨ll 2004).Largeimpoundmentdatasets(e.g.,Smithetal. 2002)suggestthatsmallimpoundmentscoverlessareathan largeones(Table3). Theprecedinganalysisincludesonlythoseimpound- mentswithlarge,engineereddamsandignoressmall impoundmentscreatedusingsmall-scaletechnologies.The areacoveredbysmallimpoundmentshaslargelybeen amatterofspeculation(St.Louisetal.2000).Farmand agriculturalpondsareagrowingandgloballyuninventor- iedresource.Theyareconstructedassourcesofwaterfor livestock,sourcesofirrigationwater,fishcultureponds, recreationalactivities,sedimentationponds,andwater qualitycontrolstructures. Figure6showstheareaofwaterimpoundedby agriculturalpondsinseveralpoliticalunits.Thereis climaticregularityinthefractionoffarmlandthatis convertedtopondstructures.Underdryconditions,farm pondsarerare,owingtothedifficultyofcollectionand conservationofsufficientstandingwater.Uptoabout 1,600mmofannualprecipitation,farmpondsarean increasingfractionoftheagriculturallandscape.Inmoist climatessuchasGreatBritain,Tennessee,andMississippi, farmpondsmakeup3–4 % ofagriculturalland. Thestatistical–climaterelationshipshowninFig.6was usedwithdataonareaunderfarmingpractice,pondsize, andestimatesofannualaverageprecipitation(FAO2003) toestimatetheglobalareacoveredbyfarmpond impoundments.Thisareasumsto76,830km 2 worldwide. Theaccuracyofpredictionsoffarmpondareawereverified usingpublisheddataonfarmlandareas(USDA2004)and normalprecipitationaverages(NOAA2000)fortheUnited States.Thismethodestimatestheareaoffarmpondsinthe contiguousUnitedStatestobe21,600km 2 ,remarkably closetothe21,000km 2 estimatedbyGIS(Smithetal. 2002).Thepredictedworldareaoffarmpondsismorethan sixtimestheareapredictedbyextrapolationofthelarge damsdatabaseandnearlydoublethetotalareacoveredby impoundmentsbetween100km 2 and1,000km 2 inarea (Table3).Suchsmallimpoundmentsaregrowingin importanceatannualratesofincreasefrom0.7 % inGreat Britain,to1–2 % intheagriculturalpartsoftheUnited
  Thenumbers,averagesizes,andareasofworldwaterimpoundmentscalcula tedfromEqs.7,8,and10.Thetotalareaof impoundmentsinthesizerangeiscalculatedastheproductofcalculation sfromEqs.7and8.Valuesareinclusiveoflowerboundsbut exclusiveofupperbounds. A min (km 2 ) A max (km 2 ) Numberof impoundments Averageimpoundment area(km 2 ) Totalareaof impoundments(km 2 ) d L (impoundments per10 6 km 2 ) 0.010.1444,8000.02712,0402,965 0.1160,7400.27116,430405 1108,2952.7122,44055.3 101001,13327.130,6407.55 1001,00015727141,8501.05 1,00010,000212,70657,1400.14 10,000100,000327,06078,0300.02 Allimpoundments515,1490.502258,570 Fig.6.Relationshipbetweenthesurfaceareaoffarmponds andtheannualaverageprecipitationinseveralpoliticalunits. DatasourcesforfarmpondnumbersaregiveninWebAppendix 1.Thelineisaleastsquaresregression( r 2  0.80, n  13)where theareaoffarmpondsexpressedasapercentageoftheareaof farmland(FP)riseswithannualaverageprecipitation(P;mm)as FP  0.019e 0.0036P . Lakeabundanceandsizedistribution 2395 States,toindryagriculturalregionsofIndia(seereferencesinWebAppendix1,http://www.aslo.org/lo/toc/vol_51/issue_5/2388a1.pdf).ConsistentregionalhypsometrypermitscalculationofthesizedistributionandareacoveredbylakesbyanchoringaParetodistributionfunctiontoacanonicalsizedistribu-tionoftheearth’slargestlakes.Naturallakesandpondsareestimatedtocoverabout4.2millionkmoftheearth’ssurface(Table2),whereasimpoundmentscover260,000km(Table3),andfarmpondscoverabout77,000km.Thesedata,takentogether,indicatethatlakes,ponds,andimpoundmentscoveroftheearth’ssurface.Thisismorethantwiceasmuchasindicatedbypreviousinventoriesbecausesmalllakeshavebeenunder-censused.Onaglobalscale,ratesofmaterialprocessing(e.g.,carbon,nitrogen,water,sediment,nutrients)byaquaticecosystemsarelikelytobeatleasttwiceasimportantashadbeenpreviouslysupposed.Sincethenumericalandarealcoverofsmallwaterbodiesismuchgreaterthanwaspreviouslyassumed,processesthataremostactiveinsmalllakesandpondsmayassumeglobalsignificance.Onalocalscale,previousanalyseshadindicatedthatsmallaquaticsystemswerespatiallyunimportant,yetsmallwaterbodiesdominatetheglobalareacoveredbycontinentalwaters.Becausestudiesofsmallaquaticsystemshavebeenunderemphasized,futureworkshouldemphasizetheglobalroleandcontributionofsmallwaterbodies.,D.,D.LC.V.2003.Theneedforglobal,satellite-basedobservationsofterrestrialsurfacewaters.EOS269,275–276.,D.W.,.2003.Changesinthechemistryoflakesandprecipitationinhigh-elevationnationalparksinthewesternUnitedStates,1985–1999.WaterResour.Res.1171,[doi:10.1029/2002WR001533].,J.J.,N.F.C.2001.Carbonincatchments:Connectingterrestrialcarbonlosseswithaquaticmetabolism.Mar.Freshwat.Res.,W.E.,E.G.1998.Magnitudeandsignificanceofcarbonburialinlakes,reservoirs,andpeatlands.Geology,M.,N.HB.P.1993.Statisticaldistributions.2nded.JohnWileyandSons.FAO.2003.Aquastatdatabase.FoodandAgricultureOrganiza-tionoftheUnitedNations.,B.M.,C.J.VW.G.2005.UNH/GRDCcompositerunofffields.V1.0.UniversityofNewHampshireandGlobalRunoffDataCentre.,L.2004.Lakes:Formandfunction.TheBlackburn,W.1922.DieSeenderErde.Peterm.Mitteilungen,1–169.[InGerman.]AMILTON,S.K.,ANDW.M.LEWIS.1990.PhysicalcharacteristicsofthefringingfloodplainoftheOrinocoRiver,Venezuela.Interciencia491–500.[InSpanish.]———,J.M.M,M.F.GW.M.L1992.Estimationofthefractaldimensionofterrainfromlakesizedistributions,p.145–164.G.E.PettsandP.A.Carling[eds.],Lowlandfloodplainrivers:Ageomorphologicalper-spective.Wiley.ERDENDORF,C.E.1984.Inventoryofthemorphopmetricandlimnologiccharacteristicsofthelargelakesoftheworld,technicalbulletin.TheOhioStateUniv.SeaGrantIPCC.2001.Thecarboncycleandatmosphericcarbondioxide,p.183–237.IPCC,ClimateChange2001.CambridgeUniv.,J.2001.Limnology:Inlandwaterecosystems.Prentice,D.H.,W.S.O,R.A.LD.A..1988.Easternlakesurvey:Regionalestimatesoflakechemistry.Environ.Sci.Tech.,B.,P.D.2004.Developmentandvalidationofaglobaldatabaseoflakes,reservoirsandwetlands.J.Hydrol.,R.A.,D.H.L,J.M.E,D.F.B,W.S.O,E.P.MR.E.C.1986.Populationdescriptionsandphysico-chemicalrelationships,p.136.CharacteristicsoflakesoftheeasternUnitedStates.V.1.U.S.EnvironmentalProtectionAgency.,M.1995.Globaldistributionoflakes,p.1–35.Lerman,D.M.ImbodenandJ.R.Gat[eds.],Physicsandchemistryoflakes.Springer-Verlag.,E.1929.Thescopeandchiefproblemsofregionallimnology.InternationalRevuedergesamtenHydrobiologieNOAA.2000.State,regionalandnationalmonthlyprecipitationweightedbyarea;1971–2000(andpreviousnormalsperiods),HistoricalClimatologySeries,p.17.NationalOceanicandAtmosphericAdministration,NationalEnvironmentalSatel-lite,Data,andInformationService,NationalClimaticData.2003.NorthDakotastate-widearealhydrologicfeatures[Internet].Bismarck(ND):NorthDakotaStateWaterCommission;[accessed2004March].Availablefromhttp://www.state.nd.us/gis/ENTERFOR2004.Inlandwaterresources:AllOklahomalakes[Internet].Stillwater(OK):OklahomaStateUniversity/StrategicConsultingIn-ternational;[accessed2004March].Availablefromhttp://,M.L.,Y.T.P.2004.Respirationinlakes.A.delGiorgioandP.J.L.Williams[eds.],Respirationinaquaticsystems.OxfordUniv.Press.,V.1897.Coursd’e´conomiepolitique.Lausanne.AATIKAINEN,M.,ANDE.KUUSISTO.1988.Suomenja¨rvien¨ra¨japinta-ala.TerraCHUILING,R.D.1977.Sourceandcompositionoflakesediments,p.12–18.H.L.Golterman[ed.],Interactionbetweensedimentsandfreshwater.ProceedingsofaninternationalsymposiumheldatAmsterdam,theNetherlands,September6–10,1976.Dr.W.JunkB.V.,S.A.1956.EvolutionofdrainagesystemsandslopesinbadlandsatPerthAmboy,NewJersey.Bull.Geol.Soc.Amer.,M.D.,R.E.D,J.P.HW.C.H1986.GazetteerofFloridalakes.WaterResourcesResearchCenterofFlorida.HIKLOMANOV,I.A.,J.C.RODDA.2003.Worldwaterresourcesatthebeginningofthetwenty-firstcentury.CambridgeUniv.Press.IPPEL,S.J.,S.K.HAMILTONANDJ.M.MELACK.1992.InundationareaandmorphometryoflakesontheAmazonRiverfloodplain,Brazil.Arch.Hydrobiol.385–400.[In2396Downingetal. S MITH ,S.V.,W.H.R ENWICK ,J.D.B ARTLEY , AND R.W. B UDDEMEIER .2002.Distributionandsignificanceofsmall, artificialwaterbodiesacrosstheUnitedStateslandscape.Sci. TotalEnviron. 299: 21–36. S OBEK ,S.,G.A LGESTEN ,A.-K.B ERGSTRO ¨ M ,M.J ANSSON , AND L.J. T RANVIK .2003.Thecatchmentandclimateregulationof pCO 2 inboreallakes.GlobalChangeBiol 9: 630–641. St.L OUIS ,V.L.,C.A.K ELLY ,E.D UCHEMIN ,J.W.R UDD , AND D. M.R OSENBERG .2000.Reservoirsurfacesassourcesof greenhousegasestotheatmosphere:Aglobalestimate. BioScience 50: 766–775. S TRAHLER ,A.N.1952.Hypsometric(area-altitude)analysisof erosionaltopography.Bull.Geol.Soc.Amer. 63: 1117– 1142. T HIENEMANN ,A.1925.DieBinnengewa ¨sserMitteleuropas:Eine LimnologischeEinfu ¨hrung.E.Schweizerbart’scheVerlags- buchhandlung.[InGerman.] USCOE.1999.Nationalinventoryofdams.UnitedStatesArmy CorpsofEngineers.[Availableonlineathttp://crunch.tec. army.mil/nid/webpages/nid.html]. USDA.2004.2002Censusofagriculture;summaryandstate data,GeographicAreaSeries.UnitedStatesDepartmentof Agriculture,NationalAgriculturalStatisticsService. V IDONDO ,B.,Y.T.P RAIRIE ,J.M.B LANCO , AND C.M.D UARTE . 1997.Someaspectsoftheanalysisofsizespectrainaquatic ecology.Limnol.Oceanogr. 42: 184–192. W ETZEL ,R.W.1990.Land-waterinterfaces:Metabolicand limnologicalregulators.Int.Verein.Theor.Limnol.Verh. 24: 6–24.[InGerman.] ———.2001.Limnology:lakeandriverecosystems.Academic Press. Received:10July2005 Accepted:5March2006 Amended:6April2006 Lakeabundanceandsizedistribution 2397