Missouri Botanical Garden Press

Changes in Plant Community Diversity and Floristic Composition on Environmental and
Geographical Gradients
Author(s): Alwyn H. Gentry
Source: Annals of the Missouri Botanical Garden, Vol. 75, No. 1 (1988), pp. 1-34
Published by: Missouri Botanical Garden Press
Stable URL: http://www.jstor.org/stable/2399464 .
Accessed: 15/07/2013 21:08
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 support@jstor.org.
Missouri Botanical Garden Press is collaborating with JSTOR to digitize, preserve and extend access to
Annals of the Missouri Botanical Garden.
http://www.jstor.org
This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

All use subject to JSTOR Terms and Conditions
Volume75 Annals
Number1 ofthe
1988 Missourl
Botanical
Garden
CHANGES IN PLANT AlwynH. Gentry3
COMMUNITYDIVERSITY AND
FLORISTIC COMPOSITION
ON ENVIRONMENTALAND
GEOGRAPHICALGRADIENTS 2
' This and thefollowingfivepapers comprisetheproceedingsoftheMissouriBotanical Garden's 33rd Annual

SystematicsSymposium-Species Diversity.The symposiumtook place in St. Louis, Missouri on October 10
and 11, 1986.
2 I thanktheNational GeographicSocietyfora seriesofgrantsthatsupportedmuchoftheresearchsummarized
here. Collectionof theMadagascar data set was funded by the WorldWildlifeFund. The coastal Colombianand
Ecuadorian data sets and some oftheAmazonian Peru data weregatheredincidentaltofioristicprojectsfunded
by the National Science Foundation. Additional Peruvian data sets werefunded by USAID (DAN-5542-G-SS-
1086-00) and the Mellon Foundation; study of the Tambopata treeplots was in part funded by a grantfrom
the SmithsonianInstitutionto T Erwin. The data setsfromeasternBrazil and Paraguay weregathered as part
ofan NSF-sponsoredmonographicstudyofBignoniaceae (BSR 83-05040, BSR 86-07113). The Osa Peninsula,
Costa Rica, data were obtained as an OTS class project; the Los Tuxtlas,Mexico, data came from a similar
class projectfor the UniversidadNacional Aut6nomade Mexico; parts of the Colombiandata weregathered as
a project ofthe CursoPos-Grado de Botdnica ofthe UniversidadNacional de Colombia.Amongthe manyfriends
and colleagues who collaborated in gathering the data summarizedhere were R. Neumann, R. Palacios, C.
Crist6bal,and A. Schinini (Argentina);K. Kubitzki,M. Fallen, H. Popppendieck, and W. Lippert (Germany);
J. Miller,D. Faber-Langendoen,E. Zardini, and C. Burnett (U.S.A.); C. Ramirez (Chile); E. Lott (Mexico); D.
Stevens,P. Moreno,and A. Grijalva (Nicaragua); H. Cuadros,E. Renteria,A. Cogollo,M. Monsalve,A. Juncosa,
C. Restrepo,J. Ramos, P. Silverstone,and 0. de Benavides (Colombia); C. Dodson (Ecuador); B. Stein, R. G.
Troth-Ovrebo, and P. Berry (Venezuela); F. Ayala, C. Dtaz, R. Vasquez, N. Jaramillo,D. Smith,R. Tredwell,
K. Young,and D. Alfaro (Peru); A. Peixoto and 0. Peixoto (Brazil); V Vera,J.Ddvalos, and S. Keel (Paraguay);
G. Pilz (Nigeria); D. Thomas (Cameroon); L. Dorr, L. Barnett,and A. Rakotozafy (Madagascar); J. Connell
and J. Tracy (Queensland); J. Tagai (Sarawak); G. McPherson (New Caledonia); and V Kapos (Jamaica).
Additional original data using the same or comparable techniqueswere made available by E. Lott (Mexico),
D. Lorence (Mauritius), and J. Miller and P. White (U.S.A.). I also thankR. Perrozzi, S. McCaslin, G. Fulton,
and especially J. Millerfor computationaland technicalexpertise,E. Zardini for help in thefield and withthe
illustrations,and J. Hall and D. ThomasforprovidingAfricandata. I thankS. Hubbell, T. Givnish,L. Emmons,
P. Ashton,P. Raven, and D. Thomasfor reviewcomments.
3 MissouriBotanical Garden,P.O. Box 299, St. Louis, Missouri63166, and WashingtonUniversity, St. Louis,
U.S.A.
ANN. MISSOURI BOT. CARD. 75: 1-34. 1988.

2 Annals of the
MissouriBotanical Garden
ABSTRACT
Trendsin communitycompositionand diversityof neotropicalforestsas measured by a series of samples of
(1) plants 2 2.5 cm dbh in 0.1 ha, (2) plants over 10 cm dbh in 1-ha plots, and (3) completelocal florulas
are analyzed as a functionof various environmental parameters. These trendsare also compared with those
found in similar data setsfromothercontinents.Altogetherthe basic 0.1-ha data sets are reportedfor 87 sites
in 25 countrieson six continentsand several islands. New data fromten 1-ha treeplots in upper Amazonia are
also compared witheach otherand withsimilar data fromthe literature.Some noteworthy trendsinclude: (1)
Lowland neotropicalplant species richnessis generallyfar moretightlycorrelatedwithprecipitationthan with
edaphic factors. (2). The nearlylinear increase oflowland neotropicalplant species richnesswithprecipitation
reaches an asymptote(communitysaturation?) at about 4,000 mmofannual rainfall. (3) Althoughthespecies
representedin adjacent foresttypeson different substratesmay change dramatically,diversity- tends to change
relativelylittlein upper Amazonia. (4) The species presentat different sites are verydifferentbut thefamilies
representedand theirdiversitiesare highlypredictablefromenvironmental parameters. (5) On an altitudinal
gradient in the tropicalAndes thereis a sharp, essentiallylinear decrease in diversityfromabout 1,500 m to
near the upper limitofforestabove 3,000 m. (6) Thereis no indicationof a "mid-elevationbulge" in diversity,
at least not in the sampled habit groups. (7) Even near timberline,montanetropicalforestsare as diverseas
the most diverse temperateforests. (8) Moist subtropicalforestsare markedlyless diverse than their inner-
tropical equivalents,but dry subtropicalforestsin Mexico are apparentlyricherin species than inner-tropical
dryforests. (9) CentralAfricanforestsare about as species rich as neotropicalforestswithsimilar amountsof
precipitation,butforestsin tropical WestAfricaare relativelydepauperate. (10) Tropical Australasianforests
are no more diverse than equivalent neotropicalforests;the world's highesttree species diversityis in upper
Amazonia, not Southeast Asia. (11) Contraryto accepted opinion, equivalentforestson the threecontinents
are similar in plant species richnessand (with a veryfew notable exceptions) fioristiccompositionbut are
markedlydifferent in structure.The predictabilityof thefioristiccompositionsand diversitiesof tropicalforest
plant communitiesseems strong,albeit circumstantial,evidence that these communitiesare at ecological and
perhaps evolutionaryequilibrium,despite indicationsthat certain aspects of theirdiversityare generated and
maintainedstochastically.
Comparisonsof the species richness(or have reportedthe firstcomparabledata set

otherfacets)of different forestsor different fortropicalforests(Gentry& Dodson, 1987a,
vegetationtypesare oftendifficult because of b). A standardizedsamplingtechniquethat
thedissimilarityoftheavailabledata. In trop- includesonlyplants - 2.5 cm in diameterin
ical Asia thereis a wealthof data fortrees 0.1 ha has also been developedand applied
inlargesampleareas (Ashton,1964, in press; to a seriesoftropicalforests(Gentry,1982b,
Whitmore,1984; Proctoret al., 1983; Kar- 1986b; Lott et al., 1987; Stallingset al., in
tawinataet al., 1981) but fewpublisheddata press; Lorence & Sussman, 1988); the
on nontrees. In the Neotropics there are methodology forobtainingthese O. -ha sam-
several local florulas(Croat, 1978; Dodson ples, each the sum of ten 2 x 50 m belt
& Gentry,1978; Janzen& Liesner, 1980; transects,is discussed in detail elsewhere
Dodson et al., 1985), butuntilrecentlythere (Gentry,1982b, in prep.). The primarydata
have been no tree-plotdata fromhighdiver- set on which this paper is based are these
sityregionsbased on reliableidentifications.0.1-ha samples,whichare now available for
Africahas far more extensivecoverage by 38 lowlandneotropicalsites,11 montaneneo-
regionaland country-wide florasbut no local tropicalsites,and 13 subtropicaland 9 tem-
florulasnor large-plotdata fromhigh-diver- perate-zonesitesintheAmericas.Similardata
sityregions. sets are availablefrom6 sitesin tropicalAf-
Recently,a series of 0.1-ha samples of rica, 3 sitesin tropicalAustralasia,2 sitesin
manyof the world'smostdiverseextra-trop- Europe,and fromseveraltropicalislands:New
ical plantcommunities has been accumulating Caledonia, Madagascar, Mauritius,Jamaica
(e.g., Naveh & Whittaker,1979; Cowling, (Tables 1, 2; Fig. 1). Supplementary data are
1983; Peet & Christensen,1980; Rice & taken fromlocal florulasin the Neotropics
Westoby,1983; Eiten, 1978). Elsewherewe (Dodson& Gentry,1978,1988; Croat,1978;

Volume 75,-Number1 Gentry 3
1988 PlantCommunityDiversity
FIGURE 1. Locations of study sites. Dots = 0.1-ha samples (see Tables 1, 2). Arrows= local fiorulas. For
location of 1-ha treeplots see Gentry,1988.
Janzen& Liesner,1980; Dodsonet al., 1985; so tightlycorrelated(R2 = 0.93) with the

Hammel,pers.comm.(La Selva, Costa Rica)) absolutenumberofspeciesthattheiruse would
and fromtheMakokouregionofnorthwestern add littleto the analysis.Moreover,the wet-
Gabon (Halle, 1964, 1965; Halle & Le forestH' values of 7 to 8 are far above the
Thomas, 1967, 1970; Florence & Hiadik, levels at whichH' has been statistically ana-
1980; iladik & Halle, 1973; Hladik & Gen- lyzed(cf. May, 1975).
try,in prep.). A supplementaldata set is pro-
videdby a seriesof 1-ha treeplotsin various
TEMPERATE-TROPICAL PATTERNS
partsoftheNeotropics(Gentry,1988; Prance
et al., 1976; Campbellet al., 1986; see also Figure2 summarizesthe latitudinaltrends
Gentry,1982b) and Paleotropics(e.g., Ash- in species richness,based on the 74 lowland
ton, 1964, 1977, in press; Gartlan et al., (= < 1,000 m) 0.1-ha sites forwhichcom-
1986). parablesamplesare available.It is wellknown
Here I firstreviewhowthespeciesrichness that tropicalforestsare generallyfar richer
ofplantcommunities changeson fivedifferent than temperateforestsin species (e.g., see
environmental gradients:latitudinal,precipi- Richards,1952; MacArthur,1972). Figure
tational,edaphic, altitudinal,and interconti- 2 indicatesthat for vascular plants species-
nental. Observationson a few noteworthy richtropicalforestsare typicallyan orderof
trendsin foreststructureare also included. magnitudemore diverse. Also apparent in
Second, I analyze some patternsof floristic Figure 2 are several much less well-known
change along the same environmental gra- corollariesto the generallatitudinaldiversity
dients. Finally,I use these analyses to ex- gradient.1) The difference in speciesrichness
aminebriefly the questionof whysome plant betweendifferent tropicalforestsis fargreater
communities have so manymorespeciesthan than the difference betweentemperatezone
others. and species-poortropical forests.Whereas
In all of these analyses I willuse number the temperateforestsamples have 15-25
of species as the simplestand most appro- species and tropicaldry forestones mostly
priatemeasure of diversity,as suggestedby 50-60 species,the samplesof moistand wet
Whittaker(1977). Shannon-WienerH' val- tropicalforestsaverage about 150 species
ues are reportedin Tables 1 and 2, but are and pluvialforestsover 250 species (Gentry,

4 Annals of the
TABLE 1. Site characteristicsfor 0.1-ha samples.
Num-
Alti- Precipi- ber of
Grid tude tation Fami- Numberof
Site Coordinates (i) (mm) lies Species H' Reference
TemperateNorthAmerica
BurlingTract, Virgin- 38055'N 30 1,053 12 21 3.54 Givnishet al.,
ia 77010'W unpubl.
NorthwestBranch, 39002'N 20 1,060 14 20 3.22
Maryland 77002'W
Tyson Reserve,Mis- 38030'N 150 932 12 23 3.26 Zimmerman&
souri(oak woods) 9003 lW Wagner,
1979
Tyson Reserve,Mis- 38030'N 150 932 11 25 3.68 Zimmerman&
souri(chertglade) 90031'W Wagner,
1979
BablerState Park, 38032'N 150 930 13 21 3.61
Missouri 90040'W
CuivreRiverState 39?01'N 140 930 15 26 3.46
Park, Missouri 91000'W
ValleyView Glades, 38015'N 225 930 14 22 3.68
Missouri 90037'W
Indian Cave State 40030'N 320 900 12 23 3.74 Tate, 1969
Park, Nebraska 95043'W
GreatSmokyMoun- 21-30 White, 1983
tainsNational (upper5%)
Park, Tennessee/
N.C.
Europe
West
Siiderhackstedt, 54?N 20 695 10 15 2.19 Walter & Lieth,
Germany 1H1E 1960
AllacherLohe, West 48004'N 530 866 11 20 3.41 Walter & Lieth,
Germany 1 1?'30'E 1960
Temperateand SubtropicalSouth America
Rio Jejui-mi,
Para- 24042'S 150 1,800 31 85 5.40 S. Keel & V.
guay 55030'W Vera, pers.
comm.
Parque El Rey, Ar- 24045'S 1,000 1,500 27 40 4.18 Brownet al.,
gentina 64040'W 1985
Salta, Argentina 24040'S 1,300 712 16 25 3.41 Walter & Lieth,
65030'W 1960
ArroyoRiachuelo, 27030'S 60 1,200 27 47 4.46 Walter & Lieth,
Corrientes,Argen- 58050'W 1960
tina
Altode Mirador,Chile 40014'S 800 4,000 13 16 3.45 Ramirez& Ri-
73018'W veros, 1975
Bosque de San Mar- 39030'S 30 2,316 14 18 3.25 Riveros& Ra-
tin,Chile 73?10'W mirez,1978
PuyehueNational 40043'S 500 3,000 13 16 2.41 Munioz,1980
Park, Chile 72018'W
"Subtropical"CentralAmierica
Chamela,Mexico 19?30'N 50 733 37 92 5.76 Lottet al., 1987
105?03'W

Volume 75, Number1 Gentry 5
1988 Plant CommunityDiversity
TABLE 1. Continued.
Num-
Chamela,Mexico 19030'N 50 733 34 83 5.42 Lott et al., 1987

105003'W
Chamela,Mexico 19030'N 50 733 46 105 5.9 Lott et al., 1987
105003'W
Los TuxtIas,Mexico 18035'N 200 4,953 40 108-109 4.52 Lot-Helgueras,
95008'W 1976
CerroOlumno, Nicara- 12018'N 750 2,000 36 97-98 5.8 -
gua 85024'W
Cerro El Picacho, 13000'N 1,400 2,000 39 65 5.22 -
Nicaragua 85055'W
LowlandNeotropics(12?N to 12?S, ' 1,000 m)
Corcovado,Costa Rica 8030'N 30 3,800 46 132 6.56 Hartshorn,
83035'W 1983
Guanacaste(upland) 10030'N 100 1,600 21+ 53a Hartshorn,
Costa Rica (700 z 8501O'W 1983
m2)
Guanacaste(gallery), 10030'N 50 1,600 33+ 68a Hartshorn,

Costa Rica (800 ; 85010'W 1983
m2)
Curundu,Panama 8059'N 20 1,830 42 90 5.78 Gentry,1982b

79033'W
Madden Forest,Pana- 9066'N 50 2,433 45 126 6.34 Gentry,1982b
ma 79036'W
PipelineRoad, Pana- 901O'N 300 3,000 58 167 6.77 Gentry& Em-
ma 79045'W mons, 1987
Galerazamba,Colom- 10048'N 10 500 21 55 5.05 -
bia 75015'W
Tayrona,Colombia 11020'N 50 1,500 31 65 5.36 -
74002'W
Bosque de la Cueva, 11005'N 360 2,000 36 93 5.5 -
Colombia 73028'W
Tutunendo,Colombia 5046'N 90 9,000 53 258 7.57 Gentry,1986b
76035'W
Bajo Calima,Colombia 3055'N 100 7,470 58 265 7.74 Gentry,1986b
77002'W
Boca de Uchire,Ven- 10009'N 150 1,200 20 66 5.16 Gentry,1982b
ezuela 65025'W
BlohmRanch, Vene- 8034'N 100 1,400 31 68 5.38 Troth,1979
zuela 67035'W
Estaci6nBiol6gicode 8056'N 100 1,312 21+ 59a Gentry,1982b
los Llanos, Vene- 67025'W
zuela (500 m2)
CerroNeblina,Vene- 0050'N 140 3,000 31 97 5.33
zuela (No. 1) 6601?'W
CerroNeblina,Vene- 0050'N 140 3,000 26 83 4.95
zuela (No. 2) 66?11'W
Rio Palenque, Ecua- 0?34'S 200 2,980 50 119 6.15 Dodson & Gen-
dor (No. 1) 79?20'W try,1978

6 Annals of the
MissouriBotanical Garden-
TABLE 1. Continued.
Num-
Rio Palenque, Ecua- 0034'S 200 2,980 43 121 6.18 Dodson & Gen-
dor (No. 2) 79?20'W try,1978
Ecuador
Centinela-, 0034'S 550 3,000 55 140 4.78 Gentry,1986b
79018'W
Jauneche,Ecuador 1016!S 60 1,855 38 96 5.39 Dodson et al.,
79042'W 1985
Capeira,Ecuador 2000'S 50 804 26 60 5.41 Dodson & Gen-
79058'W try,1988.
INPA, Manaus, Brazil 30S 75 1,995 34 101 Gentry,1978
60?W
Mocambo,Belem, 1?30'S 30 2,760 39 131 6.42 Pires & Prance,
Brazil- 47059'W 1977
Linhares,Espirito 19018'S: 50 1,403 53+ ca. 212 7.4 Peixoto& Gen-
Santo, Brazil 40?04'W try,in prep.
Jacarepagua,Rio de 23?05'S 200 1,500 45+ ca. 160
Janeiro,Brazil 43025'W
Tarapoto,Peru 6040'S 500 1,400 38 97 5.96
76020'W
Sucursari,Peru 3015-S 140 3,500 46+ ca. 240' 7.46
72055-'W
Yanamono,Peru (up- 3028'S 140 3,500 50 212 7.49 Gentry& Em-
land)z(No. 1) 72050'W mons, 1987
Yanamono,Peru (up- 3028'S 140 3,500 50 225 7.59 Gentry& Em-
land) (No. 2) 72050'W mons,1987
Yanamono,Peru (ta- 3028'S 130 3,500 51 163 6.67
huampa) 72050'W
Mishana,Peru-(flood- 3047S 130 3,500 58 249 7.63 Gentry& Em-
plain) 73030'W mons, 1987
Mishana,Peru (ta- 3047S 130 3,500 40 168 6.44
huampa) 73030'W
Mishana,Peru (upland 3047S 140 3,500 46 196 7.21 Gentry& Em-
whitesand) 73030'W mons,1987
Bosque von Humboldt, 8050'S 270 2,500 44 154 6.37
Peru 75000'W
Cabeza de Mono, 10020'S 320 3,500(+) 42 147 6.82 Gentry,1988
Peru 75018'W
Cocha Cashu, Peru 1105I'S 400 2,000 49 162 6.78 Gentry& Ter-
71019'W borgh,in
press
Tambopata,Peru (lat- 12050'S 260 2,000 48 149 6.7 Erwin,1985
eriticterrafirme) 69017'W
Tambopata,Peru 12050'S 260 2,000 43 130 6.44 Erwin,1985
(sandyterrafirme) 69017'W
Africa
Makokou,Gabon (No. 0034'N 500 1,755 39 135 6.44 Hladik,-1978
1) 12?52'E
Makokou,Gabon (No. 0?34'N 500 1,755 32 116 6.25 Hladik, 1978
2) 12?52'E

TABLE 1. Continued.
Num-
Site Coordinates (m) (mm) lies Species H' Reference
Omo Forest,Nigeria 7?N 50 1,800 29 73 4.42 Richards,1939

50E
Oban Forest,Nigeria 5010'N 50 4,000 ? (53++)
(200 m2) 8028'E
Mt. Cameroon,Cam- 4?N 230 8,000 37 129 6.31 Richards,1963
eroon 90E
Korup NationalPark, 5?N 50 5,460 43 139 6.34 Gartlanet al.,
Cameroon 8031'E 1986
Belinga,Gabon (500 1?N 750 1,800 26(+) 115 Aubreville,1967
m2) 140E
Perinet,Madagascar 18055'S 950 1,200 52+ ca. 199
48025'E
Australia
Davies RiverState 17005'S 800 2,300 41 115 6.29 Connellet al.,
Park, Queensland 145034'E 1984
Asia
SemengohForest,Sa- 1050'N 20 4,000 47 243 7.39 Walter & Lieth,
rawak 110005'E 1960
Bako NationalPark, 1052'N 30 4,000 39 143 6.5 Ashton,in press
Sarawak 110006'E
TropicalIslands
Rivieredes Pirogues, 222010'S 360 2,200 47 151 6.31
New Caledonia 166050'E
Round Hill,Jamaica 17050'N 40 1,200 31 58 3.96 Kapos, 1982
77015'W (4.47)
Brise Fer, Mauritius 20030'S 600 2,400 26 61 Lorence & Suss-
57030'E man, 1988
aExtrapolated fromnumberof species in sampleof < 1,000 M2.
1986b). 2) The latitudinaldecreasein species 4) Species-poortropicalforestswithsingle-

richnessseems to be asymmetrical about the species dominanceare generallystill much
equator;in theSouthernHemisphereitbegins morediversethanany temperate-zone forest.
near theTropicofCapricorn,butin thenorth 5) South temperateforests,at least in Chile,
it begins well inside the Tropic of Cancer, where data sets are available, have fewer
apparentlynear 12'N latitude.3) Temperate speciesthantemperateforestsin NorthAmer-
zone forestsare verysimilarin species rich- ica, contraryto thepopularperceptionofthe
ness ofwoodyplants,comparedwiththemas- "rich" Valdivianforest;a major reason for
sive differences in species richnessbetween this differenceis that Valdivian forestsdo
temperateand tropicalforestsor betweendif- not have sympatriccongenerslike the up to
ferenttropicalforests.Temperatezoneforests sevenQuercusand fourCaryaspeciestypical
are so massivelydepauperate that even if of 0.1 -ha samplesofeasternNorthAmerican
borealforestswithtwoor threespecies ' 2.5 forests.6) Subtropicaldry forestscan have
cm dbh in 0.1 ha wereincludedin Figure2, morespeciesthando full-tropical dryforests,
theywouldnot significantly change it, even even thoughwetor moistforestsusuallyhave
thoughthe reportedvalues are for some of fewerspecies in the subtropicsthan in the
the reputedlyrichesttemperatezone forests. innertropics.

8 Annals of the
TABLE 2. Site characteristicsfor 0.1-ha samples from upland Neotropics (2 1,000 m, 12"N to 120S).
Parenthesesindicate sites too incompletelysampledfor a meaningfulestimateof numberof species in 0.1 ha.
Median
Grid Altitude Numberof Numberof
Site Coordinates (m) Families Species H'
(Monteverde,Costa Rica (200 m2)) 10048'N 1,550 (33+) (61+)
84050'W
CerroKennedy,Colombia(500 M2) 1105'N 2,600 26 5Oa 4.92
74001'W
(Cuchillode Sap Antonio,Colombia(200 M2)) 10058'N 1,710 (15+) (24+)
73030'W
Finca Zungara,Colombia(600 M2) 3032'N 1,990 37+ 1OOa
76035'W
Farallonesde Cali, Colombia 3o30'N 1,950 55 134-135 6.48
76035'W
Colombia
Finca Mehrenberg, 2016'N 2,290 40 107 4.46
76012'W
La Planada, Colombia 1010'N 1,800 38 116 5.14
77058'W
Pasochoa, Ecuador (400 m2) 0028'S 3,010 21 28a 3.03
78025'W
Venceremos,Peru 5045'S 1,850 46 159 6.65
77040'W
Incahuara,Bolivia 15055'S 1,540 45 130 6.71
67035'W
Sacramento,Bolivia 16018'S 2,450 32 93 4.89
67048'W
aExtrapolated fromnumberof species in sampleof < 1,000 M2.
There are also latitudinaldifferencesin precipitation (Gentry,1982b). However,this

foreststructure. In general,tropicalforests, relationshipis more complexthan originally
far from being open and cathedral-like,are suggested (Gentry, 1982b), and the correla-
denser thantemperate This
forests. difference tionmay not exist at all in the Paleotropics.
is almostentirelydue to small-diameterplants, In tropicalAsia, high rainfallareas such as
lianas, and trees less than 10 cm dbh (also Mt. Cherrapunji, Assam,oftenhave relatively
see Gentry,1982b). Biomass(as extrapolated low plantspecies richness(Ashton,in press).
frombasal area) is roughlyequivalentamong' In tropicalAfrica,two highrainfallsites (>
different tropicalforests(X= 34.9 m2/ha, 5,000 mm per year) in southwestern Cam-
N = 36 (excludingAfrica;X= 70.7 m2/ha, eroon(Korup,Mt. Cameroon)have onlymar-
N = 6)) and northtemperatedeciduousforests ginallymore species in 0.1-ha samples than
(X = 29.6 m2/ha, N = 5) but markedly do samplesfromnortheastern Gabon thatre-
greaterin the Valdivianforests(X= 155.7 ceive < 2,000 mm of annual rainfall.More-
m2/ha,N = 3) as wellas in theirnorthtem- over,a moremonsoonalclimatesitein Nigeria
perate equivalent,the northwestern conifer- (Omo Forest) had many fewerspecies than
ous forests(Waring & Franklin,1979). the Gabon sites despite having similarpre-
cipitationvalues. Thus, it seems likelythat
the generalizationthatdiversity increaseslin-
DIVERSITY VS. PRECIPITATION
earlywithprecipitation (Gentry,1982b) ap-
In the Neotropics,plant species richness pliesonlyinthespecialcase oftheNeotropics,
is stronglycorrelatedwith absolute annual where total annual rainfalland strengthof

300,
200
0
en: :
100 _'
40 30 20 10 0 10 20 30 40 50
OS Latitude ON
FIGURE 2. Species richnessof 1,000 samples of lowland (< 1,000 m) forestas a functionof latitude.
M2
Closed line enclosescontinentalAfricanpoints;dottedline enclosesAsian points;dash-dotline enclosesAustralian

and New Caledonian samples; other tropical and subtropicalpoints all neotropicalexcept anomalously high
diversityMadagascar point at 190S, (circled). Dashed line separates dryforest (bottom)frommoistand wet
forest (top) with three intermediatesites (moistforestphysiognomydespite relativelystrong dry season)
indicated by alternatelines.
the dryseason are stronglycorrelated.A po- mm of annual precipitation (Fig. 3). The re-
tentialtest of the relativeimportanceof dis- lationship is significantly curvilinear (F =
tributionand amountof precipitation comes 4.299, P < 0.05). From 4,000 mm to near
froma single 0.1-ha site in coastal Brazil the wettest place in the world (Tutunendo,
(Linhares),which has the unusual (for the Colombia) there is little or no change in the
Neotropics)conditionof low, evenlydistrib- species richness of neotropical plant com-
uted annualrainfall.Althoughanalysisof the munities as measured by the 0. 1-ha sampling
Linharesdiversity data is notcompleted(Pei- protocol. The regularity of species richness
xoto & Gentry,in prep.) and the site is thus patterns, and especially the apparent lid on
notincludedin Figure3, it is obviousthatits community richness suggested by this as-
estimated212 speciesin 0.1 ha are farmore ymptote,seem strongcircumstantialevidence
than wouldbe expectedfromits 1,400 mm of the kind that zoologists (e.g., MacArthur,
of annual precipitation. 1965, 1969) have construed as representing
While themanyadditional0.1-ha samples niche saturation and communityequilibrium.
now available fromthe lowland Neotropics It is also possible that part of the apparent
generallystrengthen the previouslyreported lid on plant community richness merely re-
relationshipbetweenneotropicalplantspecies flectsthe intrinsiclimitationsof the sampling
richnessand precipitation(Gentry,1982b), technique. Figure 4 compares the accumu-
additionaldata sets at the upper end of the lation of species with sample area for several
precipitationscale stronglyindicatethat the representative sites. In low-diversityforests
relationshipbecomes nonlinear,reaching a the species area curves level off below 500
marked asymptoteat around 4,000-4,500 m2 indicating that most of the species present

10 Annals of the
Missouri Botanical Garden
300 -
250
Q 200 -
0
150
W
E
1 2 3 4 5 6 7 8 9
Annual precipitation (mm x 103)
FIGURE 3. Numbersofspecies in 0.1-ha samples of lowland neotropicalforestvs. annual precipitation.Solid

line = straightline regressionfor siteswith< 5,000 mmannual precipitationwithvisuallyestimatedasymptote.
Dashed line = computergenerated curve: y = 12.37 + 0.0613x - 0.000003598x2. The curve is displaced
slightlyupwardfromthedata points,since questionablemorphospeciesand lostspecimensare treatedas distinct
species by the computerwhile the data points representbest estimatesof species numbers.Data fromTable 1
withsubtropicalsites excluded.
in a givencommunity have been sampled,but thersupportedby preliminary data from1-ha

in species-richvegetationsthe species-area treeplotsin upperAmazonia(Gentry,1988).
curvesshowlittlesignoflevelingoff.To what In these samples onlytrees and large lianas
extenta larger samplingarea mightreveal - 10 cm in diameterwerecensused(Fig. 5).
significantdiversitydifferences betweenthe The twomostspecies-richsites are fromthe
differenthighrainfallsitesremainsunknown. everwet high rainfall (3,000-4,000 mm)
The strongrelationship of species richness Iquitos area of northernAmazonian Peru,
to precipitation
in neotropicalforestsis fur- wherediversityreaches almostridiculousex-
300
Bajo Calima
Tutunendo
250
a 200* /.
Yanamono 2
200
-E
150
z
100 _ _ Tarapoto
50____ _ _
....BlohmRanch
50 *
_ _ - * . * * *Branch
.Northwest
1 2 3 4 5 6 7 8 9 10
Transect number
FIGURE 4. Species-area curvefor 100 m2 high-and low-diversity
subsamplesofrepresentative 0.1-ha samples.

800
Relatively Poor soil

700 fertile soil
(av. 19.6/ha.)
60'0 0.ianas
*; *F- . : Etrees< 30 cm dbh
3treesat= 30 cm dbh
S
rich soil av.: 106.7/ha
500 poor soil av.: 81/ha
E
0.
All 400-
CoC'
300 S
~~~~0
2
FIGUE6 mer c a Coa CO C C
10 .0c
diameter. E
0
*0~~0
o
a
O
0 0
N Co
E E E M
.0
100
FIGURE 5. Density of trees and large lianas in 1-ha Amazonian plots. Black= lianas; hatched= trees >
o
30 cm dbh; white=trees 10-30 cm dbh. K~~~~~~~~~~~C 1 w
Aseasonal
300
250
Co ~ Dry~ Season
Couva Cobpt
Mn
Subtropical
uln
a
200
~~~~~~~~~~~~~~~~~~~~~~~~~~
0 ~~~~~~~~~~~~~~~~~~~~0
0.
CO
-
0~~~~~~~~~~~~~~~~~~~
1,50
E
E
z
100
50
Yanamono Mishana Manu Cabeza de Tambopata Tambopata INeblina

Mono alluvial Tambopata upland 2

12 Annals of the
tremes(Fig. 6). At Yanamono thereare 300 1977, in press; Gartlanet al., 1986). These
species - 10 cm in diameterout of the 606 authorssuggestthatphosphorus, magnesium,
individualplantsin a hectareplot!The other and potassiumare amongthenutrients whose
1-ha plots in AmazonianPeru are in areas levelsare moststronglycorrelatedwithtrop-
withgenerallygreaterdry season stressand ical plantcommunity diversity.Nevertheless,
less overallprecipitation. Two sites between at least in the Neotropics,soil nutrientsare
100 and 12S latitudehave about 200 species farless important thanbiogeographicfactors
- 10 cm dbh, while several 1-ha plots in or precipitation in determining plant species
different habitattypesat TambopataReserve richness(Gentry,1982b; Stark et al., sub-
in southeastern Madrede Dios (1 2050'S) have mitted ms.). Multipleregression of a seriesof
between 153 and 181 species. Thus tree 31 lowland neotropicalsitesfor which we have
species richnessalso appears to be greatest both soil and species richness data for 0.1-
in aseasonal highrainfallareas, at leastwithin ha samples produced the equation: Species
Amazonia. Richness= 84.48 + 0.025(mean annual
Epiphyte diversity likewise increases in precipitation) - 0.1 00(extractable soil K).
wetterareas. While epiphytes can be well R2= 0.76, N = 31 (Stark et al., submitted
representedin areas withhigh atmospheric Ms.).
humidity but relativelylow rainfall,our data Thus our data indicatethat the nutrient
indicatethat absoluteprecipitation is gener- most closely correlated with neotropical
ally a remarkablygood predictorof epiphyte species richnessis K. The importanceof K
diversity (Gentry& Dodson,1987b). We have agrees with what Ashton (1977, in press)
data sets froma series of local florulasin foundfora largeseriesoftreeplotsin Borneo,
westernEcuador and southernCentralAmer- Gartlanet al. (1986) also foundavailable K
ica; epiphytesvary from9-24 species (2- to be highlyand significantly correlatedwith
4% of the total flora)in dry-forest sites to floristic diversityin a series of sites in Cam-
72-216 species (12-16% of the totalflora) eroon.Our data contrastwiththoseofAshton
in moist-forest sitesto 238-368 species(23- (1977, in press) and Gartlanet al. (1986) in
24% of the total flora) in wet-forestsites thatwe do notfindphosphorusto be strongly
(Gentry& Dodson, 1987a, b). For a series correlatedwithdiversity.This may be due in
of 1,000 m2samplesin whichall plantspecies part to different techniquesof nutrientex-
wereidentified and tabulatedin threewestern traction(ammoniumacetate vs. HCl). It is
Ecuadorian forests,3 epiphytesconstituted also relatedto the factthatthe mostspecies-
2% ofthespeciesin a dryforest,13 epiphytes rich 0.1-ha sample (Bajo Calima, Colombia)
constituted8% ofthespeciesin a moistforest, comes froma peculiarwhiteclay soil with0
and 127 epiphytesconstituted35% of the phosphorusas measuredby our technique.
species in a wet forest(Gentry& Dodson, Whereas Ashton's(in press) data sets in-
1987b). The wetforestat Rio Palenque is so dicate greatestdiversityat intermediate nu-
diversein plantspecies that,even excluding trientvalues, a "humped" nutrient/diversity
tree species, it has more species of herbs curvethatfitsthe modelproposedby Tilman
(includingherbaceousepiphytes)or of shrubs (1982, 1984), I see no indicationin mydata
in 0.1 ha than any nontropicalplant com- of a generaldecrease in diversityon richer
munityin the -world (Gentry & Dodson, soils in the Neotropics.Quite the contrary,
1987a). the most species-richtree plot in the world
at Yanamono, Peru, is on relativelyrichsoil
(Gentry,1988; Stark et al., submittedms.);
DIVERSITY VS. SOIL NUTRIENTS
furthersouth,in an area witha strongdry
There has been much recent interestin season, the 0.1-ha Cocha Cashu sample on
therelationships betweentropicalsoilnutrient unusuallyrich alluvial soil is fartherabove
levelsand plantcommunity richness(Ashton, theprecipitation-diversity regression linethan

200
150 A!---
1 white sand (Mishana & Neblina)
subtropical (Tambopata)
tropical terra firme (Yanamono, ManO, & Cabeza de Mono)
o 1 00,
50
1 2 3 4 5 6 7 8 9 '10
number of individuals
FIGURE 7. Numberof individuals/speciesin 1-ha Amazonian treeplots (plants 2 10 cm diam.).
any othersite (Gentry,1985a). My data do ridge systemtropical-subtropicaldemarca-

fitwell withAshton'salong the low nutrient tionhave fewerspecies than the full-tropical
end of the diversity-soilnutrientgradient, ones on eitherrich or poor soils. Moreover,
wherethereis a generalincrease in species the site withthe most nutrient-poor soil of
richnessfromnutrient-poor to intermediate all, Cerro Neblina,on pure whitesand, has
sites, contraryto the suggestionsof Huston manyfewerspecies thando any of the other
(1979, 1980). sites.Thus the Amazoniantreeplotdata gen-
Anotherway of comparingthe effectsof erallysupporttheidea thatrelativelyrichsoil
on diversity
soil fertility is by comparingoth- correlateswithrelativerichnessintreespecies.
erwiseapproximatelymatched site pairs on Especiallynoteworthy in the contextofthe
fertileand poor soils. The series of six tree relativeimportanceof soil nutrientsand pre-
plotsin AmazonianPeru fallintothreenatural cipitationas determinants of species richness
groupsbased on latitudeand strengthof the is the seriesof 0.1-ha samplesfromdifferent
dryseason (Gentry,1988). Of the two plots substratesin the Iquitos area (Table 3). All
in the everwetIquitos area, one on rich soil of the sites have the high species richness
has (marginally)more species than a nearby (168-212 species) that would be expected
siteon whitesand; on a speciesper individual (Gentry,1982b) in a regionwithhighrainfall
basis the differencewouldbe much stronger and no dry season. While samples fromthe
(Figs. 7, 8). Of two sites fromcentralPeru, forestssubjectivelyjudged likelyto be sub-
the one on richalluvialsoil (ManuiPark) has jected to greaterstress (i.e., seasonally in-
morespeciesthanone on poorsoil(Iscozacin). undated tahuampa or white-sandcampina-
Severalplotsat TambopatasouthoftheHold- rana) have slightly lowerspecies richness,all

14 Annals of the
Missouri Botanical Garden
% of species with numberof individuals tion,when such broader-scalebiogeographic

factorsas latitudeand altitudeare controlled.
This relationship wouldpredictthatthe high-
100_ A
est neotropicala-diversitiesshould be found
in upper Amazonia,where the soils are rel-
80 ' .0 ( ativelyrich,comparedwiththoseof compar-
ablyhighrainfallareas oftheGuayanaShield.
My data for0.1 -ha samplesand for1-ha tree
60 _::::::::: : plotsbothappear to fitthisprediction.More-
over,manyotherkindsof organisms,includ-
ingbirds,reptilesand amphibians,butterflies,
40 0_._ ., and bats, appear to show excactlythe same
patternof greatestdiversityin areas with
relativelyfertilesoils near the base of the
Andes, suggestingthat this relationshipis a
generalbiogeographictrend(Gentry,1988).
It is possiblethat increased productivity on
is O the generallyricher soilsof thisregionmakes
C c n
O
o
possiblefinerniche partitioning and special-
co
(S (U 0 >
E e ; 2 a XC
(U
ofz
E /D Co
izationin otherwise marginal (cf. Em-
habitats
mU
(U Co -a _
(U 0 0 &
mons, 1984; Gentry Emmons,1987).
(U Even thoughthe effectof soil nutrients on
(UE
.0 E E
a-diversitymay be relativelyminor,soil nu-
Y
trientsundoubtedlydo play a major role in

FIGURE8. Percentofspecies withdifferent numbers
of individuals in Amazonian treeplots (plants - 10
contributing to the high overall diversityof
cm diam.). Lower white bar = 1 individual; dotted Amazonian foreststhroughtheir effecton
bar = 2 individuals;hatchedbar = 3 individuals;black A-diversity (e.g., Gentry,1981, 1986a, c).
bar= 10 or more individuals. Note that an average
across all plots of 50% of the species are represented MuchofupperAmazonia,probablymorethan
by single individuals. any other part of the lowland Neotropics,
constitutes a conspicuoushabitatmosaic,with
verydifferent sets of plantspecies occurring
sites are very diverse compared with mois- in adjacent communitieson differentsub-
ture-stressed sites with a strong dry season strates(Salo et al., 1986; Gentry,1986a, c).
or low annual precipitation. Table 3 shows how littleoverlap in species
I conclude that the species richness of neo- thereis betweendifferent, moreorless equally
tropical plant communities generally in-, diverseplant communitieson different sub-
creases with soil fertilityand with precipita- stratesin theIquitosarea. Only3-24 species
TABLE 3. Numberof species shared by 1,000 m2 samples of Iquitos area foresttypes.
Mishana
Yanamono Yanamono Yanamono Mishana Campi- Mishana
No. 1 No. 2 Tahuampa Lowland narana Tahuampa
Yanamono
Terra firmeNo. 1 212 91 20 24 12 14
Terra firmeNo. 2 230 20-21 19 9 8
White-watertahuainpa 16-3 9 5 ca. 19
Mishana
Lowlandnoninundated 249 55 17
Canmpinarana(whitesand) 196 3
Black-watertahuampa 168

Laguna Cocococl-ha
km5
18 spp, shared by plots 3, 4, and 67
co
o poor soil terra firme forest
| | 1 ~~0 o\ld on sandy river terrace/
0 )/?/ < ~~~terafirmet
0 32 spp. shared ~ ~ 27 pp shre
FIGURE 9. Location of 1-ha treeplots in the Tambopata WildlifeReserve,Madre de Dios, Peru. Indicated
species numbers forplots 2 and 5 are approximatesince sampling is notyet complete.Plot 1 data in part based
of Gary Hartshorn (pers. comm.),and the actual numberof species will undoubtedlybe
on field identifications
higheras well. Shared species indicated onlyfor plots (3, 4, 6) completelysampled and identifiedby me.
out of the ca. 200 species sampledfor any of eitherpoor soil plot were shared witha
habitat are shared by a differentadjacent nearbytree plot on rich alluvial soil (coeffi-
habitat. The one exceptionis the Mishana cients of correlationof 10-11%'o) (Fig. 9).
white-sand and floodplainsamples(55 species Incompletely analyzeddata foradditionalplots
overlap),butthesetwovegetationtypeshave in otherforesttypes at Tambopata indicate
similarsubstratesand are not very well dif- thatthey,too,willshowlittleoverlapin species
ferentiated.Whilesomeofthislack ofoverlap withsandysoilor alluvialforests.The unique-
mightbe due to inadequacyof the sampling ly high species richnessof the Tambopata
techniquein such diverseplantcommunities, reserveforsuch well-known groupsas birds
a repeat sample of the same forestat Yan- (Donahue et al., in press)and butterflies (La-
amono gave a much greater,almost 50% mas, 1985) has been suggestedas largelydue
overlapin species; in otherspecies-richmoist to thereserve'shabitatdiversity,
a conclusion
and wet forestssimilarrepeatsamplesof the that clearlyaccords withthe botanical evi-
same vegetationalways give the same ca. dence.
50% overlap in sampled species (Gentry, Thus the high species richnessof woody
1982b), contrasting stronglywiththe < 20% plantsin Amazoniaas comparedwiththe rest
overlapsbetweendifferent communities. Sim- of the Neotropics(Gentry,1982a) is largely
ilarly,fortwo 1-ha tree plotson terrafirme /-diversity due to habitatspecialization.Typ-
foreston poor sandy soil at Tambopata, 83 ically,relatedspecies may fillsimilarniches
species(46% ofthe 181 speciesin plot 1 and in forestson different upper Amazoniansub-
48% ofthe 172 speciesin plot2) wereshared strates(Gentry,1981, 1986c). Dramaticdif-
withthe otherplot, for a coefficient of as- ferencesin specificcomposition,thoughnot
sociationof44%. Only16- 18% ofthespecies intracommunity diversity,accompany spe-

16 Annals of the
Decrease in diversity with altitude

200
180
lowland Amazonian average
160 *Venceremos
moist & wet forest n

average Incahuara
140
Centinela Farallones
La Planada
120 * *Finca Zingara
* Los Tuxtlas
Merenberg
100 * Cerro Olumo ,
en
'-Chamela average
Id \\ All\ ~~~~~*
* Scmn
Sacramento
e 80 ' XII
0
*Cerro El PiCsaco
E
---dry forest average \
Cerro Kennedy
40 * Parque El Rey
Most diverse \
temperate site Salta
Eastern North America average Paaochoa

20 Temperate average 'a
Valdivian average %
Suderhackfested
0 1 2 3
Altitude in kms.
FIGURE 10. Species richnessof 0. 1-ha samples vs. altitude.Points to rightof dashed line and the calculated
regressionare for Andean sites. Comparativedata ftom otherselected sites to leftof dashed line. Stars are for
individual temperateand subtropicalsites: Sdderhackfestedtis in Germany;Salta and Parque El Rey are in
northwest Argentina;Los Tuxtlas is in Veracruz,Mexico; CerroOlumo and CerroEl Pichaco are in Nicaragua;
Centinela is an isolated ridge west of the Andean Cordillera Occidental in Ecuador. Average species richness
for othersite-seriesindicated by lines spanning appropriate altitudinalrange; Chamela is westernMexican dry
forest.Several of the Andean values are preliminary,being based only on field identifications with herbarium
comparisonof vouchersstill pending or on samples of less than 1,000 m2 (see Table 2).
edaphic conditions, incomplete,the trendof decreasingdiversity

cializationsfor different
oftenrelatedto different soil-nutrientavail- withincreasingaltitudeis clear.Atleastwithin
abilityin differentAmazonianhabitats. the Andes, this inversecorrelationis linear
(Fig. 10), but the relativelylow diversityof
two Central Americanlower montane sites
ALTITUDINAL TRENDS
suggeststhat the extra-Andeandecrease in
Eleven sites in tropical forestsbetween diversity withaltitudemaynotfollowthesame
1,500 and 3,100 m altitude,mostlyin the rules; certainlyCentral American montane
Andes,are includedin Table 2. Althoughthe forestshave very different floristic
composi-
available data set for upland sites is very tionsas well. Althoughtherehas been much

E ~~~~~E
0%o E ~~E E
moO
E
0~ ~~~~~~~ OO ) m'Dt
0 E0
s~~~~~~~~~~~~~~~~1
IJ c
-~~~ 4 0~
`1) OD
01 v z *'E
l
~~~ 0
%.r. 04
o =.
0 ~~~ 0
0 .0
-
0
o ~~~C
C"
0 0
c~~~o~
0
E~~C
0
o E
0o
0T
E 0
~ 0o 0
C4
0C
a
0~~~
E
E.E~ E
~~E ~~
~~~~~~3
O C 0 0
0 E E 0 S-C
C - E 0 0 0
0 ()O
~
0LC
~
0
Ca z 0cCo 0 0 co 0 o 0 .2 0
~~- 0o co C co -
' 0 ca > a. 0o E C0
0
0
0
0) 0
-j
0
> U.W
0CO
0
>->-0 ()
Lowland Mos ~ and ~ we~ cl

oet hooRgo e Uln oet
Lowland Moist and wet Forests Choc6 Region Wet Upland Forests
and Pluvial Forests
FIGURE11. Percent of hemiepiphytes(black portion of bar) in sampled climbersfor 0.1-ha samples of
upland forests (in altitudinalsequence) comparedwithlowland Choc6 area and non-Choc6area samples. Note
apparent peak in hemiepiphytesat 1,800 m.
speculationin the literatureabout a "mid- moststriking is theincreasein sampledhemi-

altitudebulge" in diversity(Janzen, 1973; epiphytesaround 1,800 m (Fig. 11). How-
Janzenet al., 1976; Scott, 1976; Gentry& ever, since increased numbers of hemi-
Dodson, 1987b), thereis no hintof such a epiphytic species (and individuals) are
phenomenonin the data of Figure 10. In- concomitant withdecreasednumbersof free-
stead, there seems to be a constantrate of climbingliana species and individuals,there
decreasingspecies richnessin moistAndean is no netchangein community diversity.Also
forestsfromthe lowlandtropicsto near tree noteworthy is therelativeabundanceofhemi-
line. Unfortunately,no siteshave been sam- epiphyticclimbersin wetlowlandsitesin the
pled fromthe Andeanfoothill regionbetween Chocoarea, a typicalexampleofthetendency
600 and 1,500 m, makingit difficult to judge of the forestsof thisregionto have features
at whataltitudethe decrease in diversitybe- and taxa morecharacteristic ofuplandforests
gins. Clearlythereis no altitudinaleffectup (Fig. 11; Gentry,1986b). At higheraltitudes
to at least 500 m (Cocha Cashu, Peru; see free-climbing lianas take over again, so that
Gentry,1985a). Since samplesfromsites at at 2,500 m and above, hemiepiphytes have
1,700 m wouldbe near theaverage value for completelydroppedout.
lowlandwet-and moist-forest sites(Fig. 10), Even near the tree line above 3,000 m,
we can assume that thereis littleor no de- Andean forestsare more species rich than
crease in diversityup to that altitude. are temperateforests.Our highest-altitude
Althoughthereis no increasein diversity sample, from3,010 m at Pasochoa in the
at middle elevations,there are some note- EcuadorianAndes,has 25 species compared
worthyphysiognomicchanges. One of the withonly21-30 speciesin the richest5% of

18 Annals of the
TABLE 4. habits in local fiorulas (fromGentry& Dodson, 1987b).

Representationof different
Barro
Capeira Santa Rosa Jauneche Colorado
Habit Number % Number %. Number % Number %
Epiphyte(includingstranglers) 8 2 19 3 72 12 216 16
Parasites + saprophytes 4 1 6 1 4 1 12 1
Climbers 112 24 115 18 136 22 258 20
Trees - 10 cm dbh 69 15 142 21 112 19 290 22
Terrestrialherbs,shrubs,treelets 270 58 381 58 280 47 540 41
Total species 463 667 604 1,316
a Data fromB. Hammel(pers. comm.).
some 312 GreatSmokiesMountainssamples is on an unusuallypoor,highlyleached skel-

(White,pers. comm.)and 15-26 (X= 20.5) etal soil (Thomas, pers. comm.).
forthe 13 othertemperate-zone forestslisted West Africanforests,includingNigeria's
in Table 1. Omo Forest in my data set and the Ghana
forestsstudiedbyHall & Swaine(1981), may
be poorer in species for historicalreasons
SOME INTERCONTINENTAL DIVERSITY TRENDS
since there are suggestionsthat most West
At a continentallevel,the Neotropicshave Africanforestsmay have been extensively
many more species of plantsthan do either alteredby Bantupopulationspriorto the first
the Asian or Australasiantropics (Raven, European colonization(Keay, 1953; Jones,
1976; Prance, 1977; Gentry,1982a). Else- 1956). Even thoughmy anomalouslylow di-
where,I have suggestedthat the "excess" versityOmo Forestsite was in a plot of pro-
neotropicalspecies are mostlyin herbaceous, tectedforestconsideredto be climax(though
epiphytic, and shrubtaxa thathave speciated surroundedby a mosaic of otherplots sub-
explosivelyalongthelowerslopeoftheAndes jected to varyingdegreesof degradationhis-
and in southernCentral America. To what torically)(G. Pilz, pers. comm.),a numberof
extent,ifany,does highera-diversity of neo- itsconstituent species,such as Pausinystalia
tropical forestscontribute to the continental macroceras, Spathodea campanulata,
pattern? Markhamia lutea, and Musanga cecro-
While I have relativelyfew comparable pioides, seem morecharacteristic oflate sec-
paleotropicaldata sets, a fewgeneraltrends ondarythan of primaryforest.
seem evident.One surprising indicationfrom Nor is thehighdiversityofCentralAfrican
the available Africandata is thatCentralAf- forestsrestricted to woodyplants.Data com-
ricanforests(X = 127 spp., N = 5) may be parable to a completelocal florulaare avail-
as diversein species ? 2.5 cm dbh as their able forone Africanforestsite at Makokou,
neotropicalequivalents(X 105 spp., N Gabon (Hladik & Halle, 1973; Florence &
9) forsiteswith1,600-2,000 mmof precip- Hladik, 1980; Hladik & Gentry,in prep.).
itation.Even thoughthetwohigh-rainfall sites Comparisonof these data withlocal florulas
in Cameroon do not show the increases in fromthe Neotropicsindicatesthat Makokou
speciesrichnessthatmightbe expectedin the is not onlyas species richas equivalentneo-
Neotropics,they are stillvery diverse,and tropicallocal florulas,butit also has a similar
the drierGabon samplesactuallyhave more habit composition(Table 4; Gentry& Dod-
species than would be expected for similar son, 1987b). Similarly,data from1-ha tree
rainfallvalues in the Neotropics.Moreover, plots indicate that Africanforestsmay be
one ofthehighrainfallsiteswithanomalously almostas richin tree species as comparable
low diversity(Mt. Cameroon)is on the slopes neotropicaland SoutheastAsianforests(Gart-
of an active volcano, and the other(Korup) lan et al., 1986; Thomas,pers. comm.: 138

TABLE 4. Continued. protectedby being an island (Raven & Ax-

elrod, 1974; Axelrod& Raven, 1978).
Rio
Palenque La Selva- Makokou Quite the oppositeof Africa,Asian forests
have been widelythoughtto have more tree
Number % Number % Number %
speciesthanneotropicalforests(e.g., Ashton,
238 23 368 25 66+ 6+ 1977; Whitmore,1984). This conclusionwas
6 1 8 1 9 1
171 16 182 12
based on comparisonof extant neotropical
259 23
165 16 310 21 389 34 data for 1-ha tree plots with similarAsian
475 45 622 42 418 37 data sets. However,the previouslyavailable
1,055 1,490 1,140 neotropicaltreeplotswereall fromareas that
would be anticipatedon biogeographicalor
ecologicalgroundsto have species-poorfor-
spp. in 0.64 ha on transect S, Korup National ests (Gentry,1988). Hectare plots in upper
Park, Cameroon; Gentry, in press). Amazoniaconsistently have moretreespecies
On the other hand, it is noteworthythat than in most Asian forests(Gentry,1988),
my single Madagascar site is richer in species and the most species-rich1-ha plots are in
than any of the continentalAfricansites,which upper Amazonia. Indeed, these plots are so
might be anticipated from the now widely diverse-up to 300 species out of 606 in-
accepted hypothesis that Africa's low conti- dividuals - 10 cm diameterat Yanamono,
nent-wide plant (and bird) species richness Peru-that it is hard to imaginehowa forest
stems largely fromextinctionsassociated with could be much more diverse.
climatic deterioration during the Pleistocene I concludethatplantcommunity diversity,
or late Tertiary, whereas Madagascar was at least of woodyplantsin plots of 1 ha or
15~~~~~~~~~~~~~~~~~~~~~~~~~~~~~C
0~~~~~~~~~~~~~~~~~~~
x ~ ~~ ~ ~~~~~~~20
0 ~~~~~~~~~~~~~~z
o X C
0 > -EL~
Q .0 C
.0~~~~~~~
z
0 0 0
0 .Z 0
FIcSUREI1-2. Basal areas for 1 ha saiples of some different

forest types.Lige-average leasa area
foresttype;, bar-= 1 d-.
0 fr
B sa

20 Annals of the
% of species
10 20 30 40 70
a 1
50 60 so 90 100
Yanamono ao | > | X | > | C | X |, | (D | 0= | > ( 46 other families
upl.1 |
Tambopata ? | |oI| -.I2[=1? rm
IKKI^'30 otherfamilies
Tambopata upl. 2 r | t Z > | > ] co I t | mI m
DI| | | | C 34 other families
~~~~ I I~~~~~~~~~~~~~~~~~IM
Tambopata alluv. r I 0 | , | | | | 31 other families
Cocha Cashu r. o '|X|@|m|r|<|P 0 37 other families

Tambopata~~~~~~~~~~~
.p.
Cabeza de Mono co I o '

mI m ni- oI5]oI
D |=|> I
m 28 otherfamilies
(a~~~~~~~~~~~~~~~~~~~~ InmiIieI
Mishana r >c | > | ' | , | ? | ' | < |- | = |< |v | 31 other families
Koroaalup.
D
rO . O
?|t|<| D|?)|D12te aiis
Neblina r m t. | | | | D) | -> | > | 21fother families

I I ~ ~ ~ ~~~~
ci,'~~~MI
Anulha.u | 0 | | CD
WI other families
(pers. comm.);
(200 the
Makokou
are) the total for the- 10 cm (db s e f 280.1-a potsliest in
e
plotsa p Mono
de as o r (D n Fml 0 cili 3e oerfamiies
KuaelaBaongtaeae
to Xig)
Kornmon mel eta
= Meliaceae,
fia sap = Sapotaceae
(KrpadMkku)n (the second,
otes if Asi
present,=Sapindaceae),myrI=
(Kal Beaog fAmiules.
= Melastomataceae,
mel = Meliaceae, sap = Sapotca (h second i prset = Sapndcee) mr
Myristiaceae, yt Mytaceae.Note Legumnosae, he domiant
tht Leguminos, fa Iy in lll2eotopical
'
myt
Myristicandae, = Myrtaceae.Notethat families fandAficans
4 other
the foests o all hree

cntinets aremostlycompoed of pecie beloning t th sameD
few w1oodyfamilies.
(Yanamonoti
juto
Xinu) Cpentrseal
Afica (Kteocrupandeakoinou)anSotheast
Asia.lont (Kuala
Bxelalfong Andulao).ps
pleforsts
e p al otherwe nontined Farmly codeoaet ofirt
pettersbelof thet familiaeex
smfew t
less,has a similarrangeofvariationaccording Perinet),and a few otherislands(e.g., New

to local environmental conditionsin all three Caledonia: 20 palms - 10 cm dbh/ha at
of the world's main tropical regions; what Rivieredes Pirogues). While stem densities
happens at larger spatial scales remainsan of trees - 10 cm dbh may be similarfrom
open question. continentto continent (Dawkins,1959), trop-
Althoughtropicalforesta-diversitymay ical Africanforeststend to have more large
be similaron different itsstructure
continents, treesand higherbasal areas (and presumably
is not. For example,lowlandneotropicalfor- biomasses) (70.7 m2/ha vs. 34.9 m2/ha) than
ests have fewerlianas than Africanforests do neotropicalor Australasianforests(Fig.
and morelianas thanAsian forests(Emmons 12). On the other hand, Asian dipterocarp
& Gentry,1983). Large palms as a major forestsmay have uniquelyhigh densitiesof
and characteristic canopyelementoflowland small polelike trees. Such structuraldiffer-
terra firmeforestseem largelyrestrictedto ences, onlybeginningto be discovered,may
the Neotropics(Gentry& Emmons, 1987), be criticalto forestorganisms.For example,
Madagascar (20 palms - 10 cm dbh/ha at theintercontinental differenceinliana density

AV. FOR 20 LOWLANDNEOTROPICALMOIST ANDWET SITES

c I o
zZ
O
]
5?|
B
S
z
F
g ^ | O | S| | | |i~in |E Q~i| g53|R~~ta~vfor89
0 F r Ps
fams.
|
with-2 8PP.av
AV. FOR 4 SITES ON WHITESANDS
Q { 1 r ^
Av. S for428 tfams.with'2spp.av.
;t t iS
3; z
4|gc||| |r 1 >| ||
||:|l ~POF -e2sppa
for
35 fams. with AV. FOR 2 LOWLANDPLUVIALSITES
Ac
]5 Ioo ' | o 2 tt |et i i |w| !|2| l |n||; Ac. 38 fams- with AV. FOR AFRICA (5 SITES W/O OMO)
rl~ ~ ~ ~~ o r1 E|?I J, I
~ ~~~~~~~~~~~~V z]Dl>| ,lB|gt?IaIH ~o11
0 U
I-
Avi for 27 fams. with AV. OF 2 BORNEO SITES
| l 3 |g
J I <2 spp. av.I
c[ | | Wl > 1H g|> 1?X|SF1>l Sc 21fams. with1 spp.DAVIES RIVER ST. PK., AUSTRALIA
| *, o | ,l FS| | 4F | > { ? |W T?F ? | | v || |s FS~ S F 16 fams. with RIVIERDESPIROGUES,

NEWCALEDONIA
ha 1 55 5 lapieah
FIGURE 14. Comparisonof average numberofspecies per familyfor 0.1-ha samplesfromsubsetsof different
lowland neotropicalforesttypes withequivalentpaleotropical data. From top to bottomthe columnsrepresent:
1) average for 20 lowland neotropicalmoistand wet sites; 2) average for 4 neotropicalsites on whitesand;
3) averagefor 2 pluvial-forestsites in Choc6; 4) continuationof 3; 5) averagefor 5 centralAfricansites (i.e.,
excludingOmo); 6) averagefor 2 Bornean sites; 7) continuationof 6; 8) Davies River State Park, Queensland,
Australia; 9) Riviere des Pirogues, New Caledonia. Shortestcolumnsegmentsare two species tall.
may have been the-criticalfactorselecting rica,on therichvolcanicsoilofthe Mt. Cam-

locomotoradaptations among
for different eroon plot, several families, especially
canopy vertebrateson the three continents Rubiaceae,Apocynaceae,and Euphorbiaceae
(Emmons& Gentry,1983). have morespeciesthanlegumes,butthisfor-
est, on the lowerslopes of an active volcano,
may not be strictlyprimary.
FLORISTICS
The otherfamiliesthatcontributemostto
Neotropicalplantcommunities are put to- species richnessof different plant communi-
getherin decidedlynonrandomways. Thus ties are also predictable.In the Neotropics
community-level frequencyof different seed the same 11 families Leguminosae,Lau-
dispersaland pollinationsyndromesis gen- raceae, Annonaceae, Rubiaceae, Moraceae,
erallypredictablefromenvironmental param- Myristicaceae,Sapotaceae, Meliaceae, Pal-
eters (Gentry,1982b, 1983). Similarly,the mae, Euphorbiaceae, and Bignoniaceae-
floristiccompositionof differentplant com- contribute abouthalf(38%-73%; X = 52%)
munitiesis remarkablyconsistent,at least at of the species richnessto 0.1-ha samples of
the familiallevel. Legumes are virtuallyal- any lowland forest.At least eight of these
ways the dominantfamilyin neotropicaland families are always among the ten most
Africanlowland primaryforests.The only species-richfamiliesin any lowlandneotrop-
neotropicalexceptionsare on extremelyrich ical moist or wet forest(Fig. 14; Gentry,
soils where Moraceae become very diverse 1987b). Similarly,in 0.1-ha samplesof low-
and are occasionallyas species-richas Le- land neotropicaldry forests,Bignoniaceae,
guminosaein 0.1-ha plots (Gentry,1986b, the preeminent liana family,is alwayssecond
c). Of the 43 continentalneotropicallowland only to Leguminosae in its contributionto
0.1-ha samples between23.50N and S lati- species richness(Fig. 15).
tudes, 39 had Leguminosae as the most Somewhatsurprisingly, the dominantfam-
species-richfamily.The dominance of le- iliesin neotropicalforestsalso tendto be the
gumesin the Neotropicsand Africais equally mostspecioseon othercontinents. Rubiaceae,
apparentwhenonlytrees - 10 cm dbh are Annonaceae,and Euphorbiaceaeare always
considered(Fig. 13). Indeed, legumes con- among the ten most species-richfamiliesin
tributealmostexactlyas muchto thediversity Africaand Asia, just as theyare in the Neo-
of neotropicaland Africanforestsas dipter- tropics.The restof the 11 mostspecies-rich
ocarps do in SoutheastAsia. Similarly,in Af- neotropicalfamilies(Lauraceae, Moraceae,

22 Annalsofthe
MissouriBotanicalGarden
I
~~ 1~~~~ o14other
Capeiral c I I | | I- families
Boca de Uchire| I| c I I ; h .InI WHlI Rt
Llanosi I| | c | |I families t 0m)
Blohm Ranch I | >' I 1 1 20 other families
Tayrona| I| | I| I 20 other families
Guanacaste(upQ c I w IW 14 other 1
Guanacaste(gal)I r | I cI > | II o In kki|13 mother .
Tarapoto E 0
FIGURE 15. Numberof species per familyfor 0.1-ha samples of lowland neotropicaldryforests.For three
sites withsample areas of < 1,000 m2-Llanos, Guanacaste (upland) and Guanacaste (gallery), with500 n2,
700 i2, and 800 m2 of sample area, respectively-the actual values are inside the solid outlines with the
anticipated number.of species in 1,000 m2 indicated by the dottedoutline.Shortestcolumnsegmentsare two
species tall.
Sapotaceae, Palmae, Myristicaceae, Meli- are only occasionallyrepresentedby one or

aceae, and Bignoniaceae)are all represented twospecies in the Neotropicsand are absent
in at least somesamplesfrombothAfricaand frommy Asian samples. Apocynaceae and
Asia and, except for Bignoniaceaeand Pal- Sapindaceae almostalwaysturnup in samples
mae, are among the ten most species-rich fromany continentbut are generallyamong
familiesin at least one Africanor Asian sam- the ten most species-richfamiliesin Africa
ple. Thus, withthe exceptionof the substi- (alwaysin the case of Apocynaceae)but only
tutionof DipterocarpaceaeforLeguminosae rarely elsewhere. Disproportionately repre-
as the most species-richwoody familyin sentedin Asia, besidesDipterocarpaceae,are
SoutheastAsian forests,pantropicalfamilial Myrtaceae(always among the most species-
compositionof lowlandforestsis remarkably rich familiesvs. almost always presentbut
similar. onlyrarelyamongthemostspecies-richfam-
Otherminordifferences includeEbenaceae ilies in the Neotropicsand representedby a
(almostalwayspresentin Africaand Asia and singlespeciesin a singlesampleon continental
among the ten most species-richfamiliesin Africa).Othernoteworthy anomaliesinclude
abouthalfthe samplesfromthosecontinents 9 speciesofProteaceae,7 ofElaeocarpaceae,
but onlyoccasionallyrepresentedin the neo- and 6 ofMonimiaceaeintheQueenslandsam-
tropicalsamples,neverby morethana single ple (thesethreefamiliesranking3rd,5th,and
species), Olacaceae (usually representedon 6th in diversityafterLauraceae, Myrtaceae,
all continentsbut generallyamong the ten and Rubiaceae), 7 Araliaceae species and 5
mostspecies-richfamiliesin Africa,neverso of Cunoniaceaein the New Caledoniasample
in Asia or the Neotropics),and Sterculiaceae (ranking5thand 8th,respectively, in familial
(alwaysamongthetenmostspecies-richfam- diversity), and 8 and 3 species,respectively,
iliesin Africa;representedby 1-3 species in of Xanthophyllum(Polygalaceae) at Semen-
almostall neotropicaland Asian samples,al- goh and Bako, Borneo.
thoughamongthe ten mostspecies-richfam- Put anotherway, all of the paleotropical
ilies onlyin Cocha Cashu, Peru). Dichapeta- forestssampledwere constitutedalmosten-
laceae are almostalwaysamongthe ten most tirelyof the same plantfamiliesencountered
species-richfamiliesin Africansamples but in equivalentsamples of neotropicalforests.

Although13 familiesnot representedin the An average of 30% (withextremesof 25%

Neotropicswereincludedin the paleotropical at Korup to 34% at Belinga) of the genera
samples,and althougheach Africanand Asian at the six continentalAfricansites are neo-
sampleincluded1-3 familiesnotrepresented tropicalgenera,nearlyall also includedin the
in the Neotropics,withtwo exceptions,the neotropicalsamples.When completelocal flo-
sum contribution of all of these to species ras are compared,genericconcordancebe-
richnessof the Asian and Africanforestsis tweentropicalAfricaand the Neotropicsre-
negligible.The two exceptionsare Diptero- mainsequallyhigh.Thus 30% of the genera
carpaceae in tropicalAsia and Pandanaceae representedat MakokouGabon also occur in
in Madagascar (3 spp.), Queensland(2 spp.), the Neotropics.Both sets of figureswouldbe
and New Caledonia(4 spp.). Excludingthese much higherif such tenuouslydifferentiated
twofamilies,an average of 2 species(and ca. genera as Pycnanthus and Virola (Myristi-
3 individuals)per sample was contributed to caceae) or Macrolobium and its segregates
paleotropicalcommunity by families (Leguminosae)were consideredto be conge-
diversity
not included in the equivalent neotropical neric.
samples.At thislevel New Caledoniawas the GenericoverlapbetweentropicalAsia and
mostdistinctive, withone specieseach ofBal- the Neotropicsis less, averaging23%, and
anopaceae, Epacridaceae, Oncothecaceae, betweenAustralasiaand the Neotropicsin-
and Pittosporaceae,plus 4 of Pandanaceae. termediate(25% neotropicalgenera in the
The Madagascar sample included,besides 3 Queensland sample, 26% in the New Cale-
Pandanaceae,a speciesofSarcolaenaceaeand donia one). These relationships mightbe pre-
twoofPittosporaceae,theQueenslandsample dictablefromCretaceousand Tertiaryplate
a speciesof Balanopaceae and 2 of Pandana- tectonichistoryand thetimetableofGondwa-
ceae (plus one ofthe sometimesCunoniaceae nan breakup. In this light,it is especially
segregateDavidsoniaceae).In Africa,Ancis- interesting thatabout36% ofthegenerasam-
trocladaceaewas representedby one individ- pled at Perinet,Madagascar, are sharedwith
ual at one site, Medusandraceaeby one in- the Neotropics,the highestvalue forany pa-
dividualat one site, and Scytopetalaceaeby leotropicalsite.
twospeciesat one site.Onlyin thelattercase There are also consistentand predictable
didan endemicfamilycontribute significantly floristicchanges along environmentalgra-
to a site's diversity,with Ouabangia alata dients,at least in the Neotropics.On poorer
the 5thmostcommonspecies(13 individuals) soilsfamilies likeBurseraceae,Lauraceae, and
at Korup and Rhaptopetalumcf. coriaceum Sapotaceae become moreprevalent,whereas
representedby threeindividualsat the same on the richestsoils palms and Moraceae are
site. It is perhaps worthnotingthat several disproportionately speciose.
ofthe endemicfamiliesincludedin the above In neotropicalareas witha strongdrysea-
totalare somewhatdubioussegregates-Pan- son, floristiccompositionis likewisepredict-
daceae (from Euphorbiaceae), Irvingiaceae able. Leguminosae are always the most
(from Simaroubaceae), and Ixonanthaceae species-richfamily,withBignoniaceae,rep-
(from Linaceae). Lowland tropical forests resentedmostlyby wind-dispersed lianas, al-
throughoutthe world are overwhelminglyways second (Fig. 15).
made up of the same plantfamilies,withthe On an altitudinalgradientin the Andes,
exceptionof the Dipterocarpaceae for Le- Lauraceae consistently replace Leguminosae
guminosaesubstitution in SoutheastAsia. as themostspecies-rich familyat intermediate
Even at thegenericlevel,thereare striking elevations(Fig. 16). Otherfamiliesthatcon-
floristic similarities
betweenthe compositions tributeto the diversityof middle elevation
of lowlandtropicalforestson different con- forestsare Rubiaceae, Melastomataceae,Eu-
tinents.The generic similarityis especially phorbiaceae,Moraceae,Guttiferae, treeferns,
markedbetweenAfricaand South America. (hemiepiphytic) Araceae, and Palmae. Fam-

24 Annals of the
CO ilies like Bignoniaceae,Sapotaceae, Myristi-

caceae, Meliaceae, Sapindaceae, Bursera-
w > ceae, and Chrysobalanaceaeare especially
noteworthy as absent or much more poorly
LCJ
V 2 represented thanin lowlandforests.At higher
elevations(> 2,000 m), Melastomataceae,
o Compositae,Rubiaceae, and tree fernsbe-
ERIC
come moreprevalent,althoughof these only
U C o V
Compositaeincrease in absolute numberof
species. At even higheraltitudes,Aquifolia-
MY I L
ceae, Myrtaceae,and Theaceae become rel-
F
O
CHR _C02
ativelymoreimportant, whilenear timberline
CC Compositaeand Ericaceae predominate.
2 IIYRT
E
04
O
0~~~~ 0O > Curiously,the site at 1,000 m altitudeat
0~~~~ CC Perinet,Madagascar, had virtuallyan iden-
J
tical familial compositionto the middle-
111111P ARAC C
elevationneotropicalsite; in additionto Lau-
raceae being the most speciose family,
SAco _O V
< 0
GUUTT
GUTT > r Rubiaceae, Euphorbiaceae, Moraceae, and
Guttiferaefollowedin species richness;the
FLAF CYCLW
onlysubstantial differences are a transposition
oftherolesofMelastomataceae(morespecies
MYRT 0 in the Andes) and Myrtaceae (more species
at Perinet),the presence of several species
ofMonimiaceaeand Oleaceae in Madagascar,
CO
and the frequencyof hemiepiphytic Araceae
in the Neotropics(Fig. 17). A Queensland,
FER14~~~~~~~1
EUP11
Australia,site from850 m was also rather
similarin familialcompositionto the Andean
PAL11 ET 0g
K , b 00m50osaper middle-elevation sites,again withLauraceae

MORRA 2u
neoropca mos-
RUB
an
NT ~
e-oetsmpe;2vrg dominating,closely followedby Rubiaceae,
thoughwithgreaterprevalenceofsuchsouth-
FIGURE F6NmerN ofspce e aiyfr01 ern familiesas Proteaceae, Elaeocarpaceae,
ALM
and Myrtaceae.Such strikingly repeatedpat-
LELU MEL AQT D
ternsin partsoftheworldso widelyseparated
BIGN >~~~~~~~~I todaycan hardlybe due to chance.
MORA 0LA E
LAU Many of the major latitudinalchanges in
floristiccomposition are wellknown,withfam-
FIGURE16. NumberofQUI
spce e famlycor01
Lehe URrg .roKney C ic ugr) ) Cer ilies such as Fagaceae and Juglandaceaere-
leftto right coumns are 1) aveag fo 0loln
neotropical
mist- and wetfores sapls;2 aerg placing the tropicaltaxa in North America
for sitsRbewee 1,0-2,00m(CroEihao
Incahuar, Venceemos, FrralnsB
eCl);3Iv (Fig. 18). Perhaps less emphasizedare how
remarkably similarin familialcompositiondif-
ferenteastern NorthAmericanforestsare.
Kennedy,
ColombiaU(,0R m 0 mUofsapl are) Whilespecies,and to some extentgenera,do
change fromplace to place, froma world
5) Pasochoa, Ecuador (3,010 m, 200 m2 of sample

area). Site datafromTable 2. Shortestcolumnsegments
are two species tall.

I I II"dI
~~~~~~~~~~' ~r8fas.wt
I| I ~
SoI EI r |IS I
<a ~ 1E ft=,|E|e I~
|I~i IC IIM acmies. witahc | PERINET, MADAGASCAR
Av. for 62 fams. with<2 spp. av. | AV. FOR 4 NEOTROPICAL SITES 1000-2000 m.
FIGURE 17. Number ofspeciesperfamilyfor0.1-ha sampleat Perinet,Madagascar (950 m) (top two

columns)compared
withaverageforfourmid-elevation sites(1,000-2,000m) (bottom
neotropical twocolumns).
Notetheremarkable
similarity
offamilialcomposition.
Perinetdata basedonlyonfieldidentifications
pending
herbarium ofvouchers.
comparison Shortestcolumnsegmentsare twospeciestall.
perspectivethe overall floristiccomposition position are highly predictable from

ofmostoftheseforestsis as similaras is their environmental and geographicalfactors,with
diversity.The contrastingly austral compo- maximumplant communitydiversityoccur-
sitionoftheValdivianflorais also wellknown. ringin full-tropical lowlandareas withrichto
There are also floristicsimilaritiesbetween intermediately infertilesoils and highannual
the austral and northtemperateones. For precipitationand/or littledry-seasonstress.
example, gymnospermsand Fagaceae be- Such patternsare oftentakenas evidenceof
come moreprevalentin bothnorthtemperate niche saturationand communityequilibrium
and south temperateareas. One interesting (MacArthur,1965, 1969; Cody, 1975; see
and previously unremarked floristic
difference also Whittaker,1977).
betweentheValdivianforestsand theirnorth- Much ofthecontroversy aboutequilibrium
ern equivalentsis thatthe formerlack sym- vs. nonequilibrium communitieshas focused
patriccongeners.The difference in diversity on the role of niche specificityvs. stochastic
betweeneasternNorthAmericanand Valdi- generationor maintenanceof diversity(e.g.,
vian forests(as well as betweenthe North Hubbell, 1984; Hubbell & Foster, 1986;
Americanforestsand mytwoEuropeansam- Ashton,1969; Connell,1978). My data sug-
ples) is almostentirelyaccountedforby this gestthateven thoughtropicalforestscontain
lack ofsympatric speciesin generalike Quer- many different plant species, they are far
cus and Carya. Why Chilean Nothofagus fromrandomassemblages.Can thesedata and
species,unliketheirnortherncousins,should conclusionsbe reconciledwiththe very dif-
be almostentirelyallopatricis unclear, but ferentones of Hubbell (1979; Hubbell &
the effectof thispatternon the diversityof Foster, 1986, 1987)? Below I willfocus on
the southtemperateforestsis obvious. several pointsthat may be relevantto this
debate.
From a somewhatdifferent perspective,
DISCUSSION
some authors(e.g., Federov, 1966) have ar-
To thispointI have attemptedto present gued that the exceedinglyhigh diversityof
a seriesofobservationsofchangesin diversity tropicalforestsis too great to be accounted
and floristic
composition on variousgradients. forby nichespecificity; therefore,some kind
I now focusbrieflyon some theoreticalgen- ofnonselectiveor stochasticmechanismmust
eralizationsthat would seem to derive from be invoked.However,it seems to me thatit
these data. is stochasticallymost unlikelythat the ex-
The overall message is that plant com- tremespecies richnessof forestslike that at
munitiesare put togetherin decidedlynon- Yanamono, Peru, with 300 species out of
random ways. Diversityand floristiccom- 606 individualsin a hectare, would result

26 Annals of the
fromrandomprocesses,unlessthereis a po-
tentialsampleuniverseofmanythousandsof
tree species. Forty-eight species are repre-
0
sentedin the first50 individualssampledat
Yanamono,and the 65 individualsin the first
I.:
2 Yanamono 0.1-ha subplot constitute 58
z
0) species.Such highlevelsofdiversity, farfrom
o o indicatingstochasticity, would seem to indi-
0
0
cate verystrongecologicalpressuresresulting
N x
o s inphenomenally lowdensitiesoftheindividual
< a,
species (and highcommunitydiversity).
MALP 1
The strikingregularitiesin the patterns
GUTI O
0
discussedabove clearlyindicatethatat some
LECY
levels both communitycompositionand di-
MYRT
FLAC
versityare highlypredictable.How this re-
CHRY lates to communityequilibrium remains
BURS clouded,however,in part because of defini-
APOC VIOL
tional problems.Hubbell & Foster (1986)
SAPI VERB definedan equilibriumcommunityas one in
TILT
POLY which a particularcombinationof species
MELI HIPP
FLAC maintainsitselfagainstoutsideperturbations,
IYRII CELA whereasthepredictablediversities ofdifferent
EUPH
ANNU
URTI
tropicalforestswithsimilarenvironments but
SAPS
different assemblagesof species is moreakin
PALT
SOLA
to the "equilibrium"theoryof island bio-
SAPO
MYRT geography(MacArthur & Wilson, 1967),
consideredby Hubbell as a nonequilibrium
COMP
. co
RUBI ARAC .
U)
0
theorybecause of the taxonomicrandomness
RUBI
involved.ManyAmazonianforestsare clearly
richerin treespeciesthanequivalentCentral
LAUR SAPT O
W Americanforests(Gentry,1988). They also

PIPE - ~ ~ have muchgreaterhabitatdifferentiation and
PALF co0
3-diversity (Gentry,1986a). Thus some of
the higherdiversity of the Amazonianforests
(wtfret8 N 00
m a E- r0 l
4 a
maybe due to the"mass effect"phenomenon
of Shmida & Wilson(1985), withaccidental
jj
ANN
AMUR R
immigrantsadapted to other environments
contributing to the a-diversityof
significantly
colunRA
LESS
BIGN
:1
LEGS
PTJSRES
vraefr2
LGS FG
oln
FG
netoial an individual Amazonianforest.Arguingalong
MEET
similarlines fromthe nonequilibrium
LIG
mois- ad wt-fres saple;) Lo Tutas, 0eic
view-
MOala 4gnia(rfoet 24?0S1,0 ,30m;5
L4gniAU (mis foet a44', SBAS l.;4 MEE point, Hubbell & Foster (1986) suggested
(28?15'-3 9?2 '); 6)Sd erakteoGemn
that biogeographicalpatternplays a major
FIGURE18. Numberofspecies per familyfor some role in tropicalforesta-diversity:if the re-
0. 1-ha samples at different
latitudes.Fromlelft'
to right,
gional diversityis greater, as it certainlyis in
columns are: 1) average for 20 lowland neotropical
moist-and wet-forest samples; 2) Los Tuxtlas,Mexico Amazonia, more species, on the average,
(wetforest,18035'N, 200 m alt.); 3) Parque El Rey,
Argentina(moistforest,24045'S, 1,000 m alt.); 4)
Salta, Argentina (dry forest,24040'S, 1,300 in); 5)
average for seven temperate North American sites (54"N); 7) average for three Valdivian, Chile, sites
(28015 'N-3902 'N); 6) Suderhackstedt, Germany (39?30'S-40043'S).

Volume75, Number1 Gentry 27
shouldoccur in individualforestsdue purely subcanopytreeLeonia glycicarpa,whichhas

to phenomenaassociatedwithpatchdynamics 6-19 individualsperhectare,and thecanopy
and local immigrations. But there are also tree Symphonia globuliferawith 1-4 indi-
problemswithsuch interpretations. That the vidualsper hectare,again excludingthe sec-
familiesand generarepresentedin these dif- ond growthplot.
ferentsamplesare so predictablestrongly sug- At theoppositeextremeare the 21 habitat
geststhat at least some kindof familial-spe- specialistslistedin Table 5, locallycommon,
cificnichesmay be involved.Moreover,the but occurringin only a singlehabitat:good
apparentpartitioning of the species of each soil,poor soil,swamp,or secondgrowth.The
familyintodifferent ofspeciesspecialized extremecase is Lueheopsis hoehnei,the ab-
sets
for different substratesin Amazonia seems solutedominantin the swampplot with265
strongcircumstantial supportforselectionist trees,but completelyabsent elsewhere;that
interpretations (Gentry,1985b). thisis onlythesecondrecordofsuch a locally
Data fromseveral 1-ha tree plots in the commonspecies fromPeru is instructiveas
Tambopata Reserve, Madre de Dios, Peru, to the state of AmazonianPeruvian floristic
can be used to documentthe effectof sub- knowledge.Anotherexampleworthmention-
stratespecificityon speciescomposition. Data ing is Sparrea schippii, previouslyunre-
are available fromtwo completelyidentified portedfromPeru (or indeedfromAmazonia),
1-hasamplesfromsimilarpoor-soilterrafirme whichis the fourthmost commonspecies in
forestseparated by about 1.5 km, a com- the alluvialplot.The largenumberof species
pletelyidentified plotin matureforeston rich thatare completelyfaithful to a singlehabitat
alluvial soil, and fromas yet incompletely (presumablyalso includingmany additional
identifiedplotsin youngriversidesecondary less commonspecies not listedin Table 5) is
forest,in swamp forest,and in forestin a a good example of the importanceof niche
transitionalarea betweenthe rich floodplain specificity in maintainingoverallAmazonian
and poor-soiluplands(see Fig. 8). As is usually species diversity.Each community is richin
the case in species-richtropicalforests,most largepartbecause ithas manyspeciesunique-
of the species sampledwere representedby ly adapted to a specificsubstrate.
one or two individualson a single plot and Perhaps more interestingfromthe view-
are inadequatelysampled to draw any con- pointof ecologicaltheoryare the othertwo
clusionsabouthabitatspecificity. Table 5 lists distributionalcategoriesindicatedin Table 5.
the speciesthatoccur in all threecompletely The firstare species thatare commonin one
sampledplotsplusall speciesthatare common habitatbut also have a fewindividualsin one
(i.e., > 10 individuals)in at least one of the or moreof the otherhabitats.Some of these
plotsplus a fewotherselectedspecies. mayrepresentcases of"mass effect"(Shmida
At one extremeare 13 species thatoccur v& Wilson,1985), withan occasionalindivid-
both in the two poor-soilplots and in the ual survivingbut not reproducingoutsideits
alluvialplot.These mightbe classed normalecologicalrange. The second pattern
fertile-soil
as ecologicallyinsensitive;none of themoc- is speciesofthepoorsoilforestthatare com-
cursin the secondaryforest,six ofthem(and monin one ofthetwosampleplotsbutabsent
possiblymore) in the swamp plot, and most fromthe other.These may be examples of
(perhapsall) ofthemon theintermediate plot. "ecological equivalents"(Shmida & Wilson,
Alloftheseare essentially uniformly dispersed 1985), where due to some accident of dis-
withsimilarnumbersof individualsin each persalorestablishment, a givenspeciesoccurs
hectare.Bertholettiaexcelsa, whichhas 1- at one sitebut not at anotherwhereit would
2 large emergenttreesper hectarethrough- be equallywelladapted.The ecologicalequiv-
out the Tambopata Reserve (except in sec- alenthypothesisseems especiallygermaneto
ondaryforest),is a good exampleof thispat- Cordia, whereCordia mexiana and C. pan-
tern. Other good examples include the amensis occur in one upland plot while C.

28 Annals of the
TABLE 5. Differencesin occurrenceand abundance in different 1-ha treeplots of some commonTambopata

species (all species occurringin all threecompleteplots or with 10 or moreindividualsin any one plot plus a
few others). Plot 1 is relativelyfertileterrafirme;plot 2 is swampforest;plots 3 and 6 are on poor sandy,
upland terrafirme;plot 4 is on rich alluvial soil; and plot 5 is in young riversidesecondaryforest.
Plot Number
1 2 3 4 5 6
Ecologicallyinsensitive,? uniformly
dispersed
Bertholettiaexcelsa 2 (1) 2 1 - 2
Eschweileracoriacea 6? ? 4 2 - 9
Glycidendronamazonica 1 (1) 1 2 - 1
Leonia glycicarpa 19 (1) 12 12 - 6
Lindackeria paludosa 4 (1) 10 1 - 4
Minquartia guianensis 1 - 3 1 - 2
Nectandra cissiflora ? ? 1 1 - 4
Neea divaricata X ? 2 5 - 4
Ocotea rubrinervis ? ? 7 7 - 7
Oenocarpus mapora 1 (1) 4 3 (4) - 3
Symphoniaglobulifera 1 (3) 3 3 - 4
Tapirira guianensis 1 - - 1 - 2
Clarisia racemosa 1 3 2 - 2
Ecologicallysensitivebut widespread
Amaioua corymbosa 1 (1?) 11 1 _ 6
Euterpe precatoria 17 4 2 2 - 4
Iriartea deltoides 106 (15) 7 86 - -
Iryantherajuruensis 5 (3) 19 5 - 22
Iryantheralaevis 17 (2) 13 2 - 31
Mabea - (1) 1? 11 - -
Pourouma minor 18 (6) 17 1 - 44
Pseudolmedia laevis 14 (5) 4 7 - 1
Siparuna decipiens 13 3 8 1 - 9
Socratea exorrhiza 10 5 1 38 7+ 2
Tetragastrisaltissima 1 (2+) 7 1 - 22
Ecologicallyrestricted
Cecropia membranacea - - - - 46+
Ficus insipida - - - 13
Citharexylumpoeppigii - - - - 9+
Sparrea schippii - - - 17
Rinorea viridifolia X - - 27
Astrocaryummurumuru - - 16
Myroxylonbalsamum 1 ? - 6 - -
Cordia lomatoloba - 2-
2
Cordia nodosa X - - 1
Mauritiaflexuosa - 1 - - - -
Lueheopsis hoehnei - 265 - - - -
Pithecellobiumlatifolium - 12 - -1
Rouchera punctata 1 (2) 19 - 15
Virola sebifera 1 (1) 18 - 15
Ouratea - 14 - - 14
Euceraea nitida - (1) 6 - _ 8
Cedrelinga cateniformis - - 3 - - 5
Cordia mexiana - - 10 - - -
Cordia panamensis - ? 4
Cordia toqueve - ? - - - 4
Cordia ucayaliensis - - - - - 7

TABLE 5. Continued.
Plot Number
1 2 3 4 5 6
not explainableby ecology
Major densitydifferences
Bixa arborea - 15 -
Hevea guianensis 1 15 -
Pseudolmedia laevigata 1 23 - - 4
Ocotea (domatia) ? 1 - - 10
Arrabidaea tuberculata - - - - - 4
X = presentin Hartshornplotbut numberof individualsnot known.
? = presenceor absence not knowndue to incompleteidentifications
(plot 1) or samplingnot yet completed
(plot 2).
()= species occurringin plot 2 onlyin corneron higherground.
toqueve and C. ucayaliensis occur in the generallynot abundantat different siteswith

other.Both of the lattersituationsrepresent similarecology,theyare usuallypresent. If
the patternthoughtto be prevalenton Barro we take the three rich-soiltree plots as an
Colorado Island (Hubbell & Foster, 1986), example, all the abundantTambopata allu-
withhighdiversity of a givencommunity due vial-plotspeciesare presentin the Yanamono
in largepartto nonequilibrium fluctuations in plot,and all theabundantYanamonospecies,
its species. except Otoba glycicarpa and Carapa gui-
At a differentlevel,mydata on plantcom- anensis,are presentat Tambopata.All ofthe
munitycomposition also seem muchless pre- abundantCocha Cashu species are at Yana-
dictableand much more in accordance with mono,and all butfourofthe abundantYana-
the nondeterministic, nonequilibriumview- monospeciesare at Cocha Cashu. If we com-
point.In the nine 1-ha tree plots that have pare the sandy-soilplots fromMishana and
been analyzed,therewas not a singlerepe- Cabeza de Mono, there is only one species
titionof a most-dominant species. Although abundantat both sites (Hevea guianensis),
one (or a few)species is always much more but all the commonCabeza de Mono species
common, there is a different"dominant" are presentat Mishana,and mostofthe com-
speciesin each plot.Even iftheseveralmost- monMishanaspeciesare at Cabeza de Mono.
dominantspecies in each plotare compared, In contrast,onlyfourspecies thatare abun-
thereis littleoverlap.Consideringonlythose dant at any poor-soilsite are presentat all
species with10 or moreindividualsper hect- in any rich-soilsite,all at Yanamono, which
are in at least one tree plot generatesa list has an intermediate level of soil fertility.
The
of 54 species documentedto be relatively relevantpoint is that althoughthe species
commonlocally somewherein upper Ama- presentat a sitemay be predictable,the fre-
zonia.Butofthese,onlyfivespeciesare abun- quency of a particularspecies in different
danton twodifferent plots.The sharedabun- forestsseems entirelyunpredictableand is
dantspeciesincludeAstrocaryum murumura, likelydeterminedstochastically.This is sim-
commonon all three rich-soilplots (Yana- ilar to the concept that Shmida & Wilson
mono,Cocha Cashu,Tambopataalluvial),and (1985) have termed "ecological equivalen-
Hevea guianensis, commonon three poor- cy," i.e., the coexistenceof species withef-
soil plots.The othersharedabundantspecies fectivelyidenticalniche and habitatrequire-
are Otoba parvifioraon twoofthethreerich- mentsforlargelystochasticreasons.It is also
soil sites (Yanamono and Cocha Cashu) and the patternthatwouldbe predictedby Hub-
Iriartea deltoidea on a differentpair of rich- bell's (1979; Hubbell & Foster, 1986) com-
soil sites (Cocha Cashu and Tambopata al- munitydrifttheory.Indeed, Grubb (1986)
luvial).But even thoughthe same speciesare generalizedthatthe relativelysparse or rare

30 Annals of the
species that of necessity constitute the bulk plantcommunity thata biogeographer might
of the species of species-rich communities define.The numerousmicrohabitat specialists
should interact so infrequentlywith each oth- suggestedby a casual glance at a series of
er that niche differentiationbecomes largely the Hubbell-Foster BarroColoradoIslanddis-
irrelevant. tributionmaps become nonspecialistsif the
The same conclusion arises from the 0.1- relativelyfewindividuals thatoccurawayfrom
ha samples. There are almost always a few a favoredhabitatare emphasized.I suspect
very common species in any sample. One of thatdifferences in taxonomicfocusmay also
these is usually much more common than all relateto theinterpretational Hub-
differences.
the others; at 11 sites the most common bell focusedentirelyon a particularcombi-
species was between two and seven times more nationof species in addressingthe question
common than the second most common of communityequilibrium.My data suggest
species. Yet the only repeat of a "most dom- thatwhilethe species thatmake up different
inant" species among 25 moist- and wet-for- communitiesmay be very inconstantfrom
est sites is Catoblastus velutinus, shared be- place to place, at the same timethe different
tween Rio Palenque and immediatelyadjacent families(and perhapsgenera)thatcontribute
Centinela in western Ecuador. For 12 dry- to community floristic are verycon-
diversity
forest sites there was not a single repeat of sistent.Perhaps the familyis the basic unit
a 'most dominant" species. on whichselectionforlowpopulationdensities
Even if all 213 species that are dominant (and thusindirectly forhighspecies richness)
or subdominant in any of these samples (i.e., occurs. For example, many seed predators
among the most common 5-10 species) are and leaf-eating insectsare host-specificat the
considered, only 38 are repeated in two or generic or familial,as well as the specific,
more differentsamples; ten of these repeated level (Janzen,1975, 1980, 1984). If family-
common species (13 if Rio Palenque and Cen- specificpredatorsand/orfamily-specific com-
tinela are considered part of the same site) petitionare added to the scenarioof dynamic
are in repeat samples of the same forest. forestswithsomenichedifferentiation, an ex-
Thus, only 24 species are abundant at more planationpleasing to selectionistsand non-
than one site. One species, Socratea exor- selectionistsalike could beginto take shape.
rhiza, is abundant at four sites, and three As indicatedin Table 1, the familialdiversity
species jessenia bataua, Arrabidaea ox- of tropicalmoistand wet forests,unlikethe
ycarpa, and Arrabidaea pubescens -are speciesrichness,is bothhighand remarkably
abundant at three sites. Ten of the 24 species constant.It is certainlywithinthe realm of
abundant at more than one site are shared possibilitythat this is due to ca. 50 family-
between different dry forests, twelve are specificnichesin a givenforest,whereasthe
shared between differentmoist forests (typi- differing species richnessof different forests
cally between Central America and Amazon- could be largelystochasticallygeneratedby
ia), and one (Mansoa verrucifera) is abundant factorsrelatingto higherturnoverinthemore
in one dry-forestand in one moist-forestsite. species-richsiteson bettersoilsand withhigher
A major part of the debate on whether productilvities.
tropical-forestecosystems are at equilibrium A different type of reconciliation,espe-
or nonequilibriummay be a by-productof the ciallyof the differences betweenupper Am-
scale of a particular study or the focus of a azonian and CentralAmericanspecies rich-
particular author. The "rare" species that ness, theircauses, and the equilibrium status
"random walk" througha 50-ha plot on Bar- of the forestsinvolved,mightcome froma
ro Colorado Island-are -mostlycommon under- differentapproach to the data. Hubbell &
storyor second growthspecies that would be Foster (1986) emphasizedthat niche differ--
regarded as permanent an-dcontinuous memn entiation amongBarroColoradoIslandspecies
bers of the more comprehensive moist-forest seems to consist mostlyof separationinto

abouta dozengeneralizedmulti-species guilds intodifferent communities maybe completely

based on degreeof shade toleranceand pref- random.
erenceforedaphicor topographicmicrosites.
Indeed they suggestthat lack of niche dif-
ferentiation mightmake possiblethe co-oc- LITERATURE CITED
currenceofmanypotentialcompetitors which ASHTON, P. 1964. Ecologicalstudiesin the mixeddip-
are forcedto sharethesame generalized niche. terocarpforestsofBruneiState.OxfordForest.Mem.
Butbyand largethelowlandforestsofCentral 25: 1-75.
* 1969. Speciationamongtropicalforesttrees:
America are composed of the same species somedeductionsin thelightofrecentevidence.Biol.
that in Amazoniawould be regardedas the J. Linn. Soc. 1: 155-196.
mostwidespreadand weediestspeciesoftheir * 1977. A contribution of rainforest research
to evolutionary theory.Ann.MissouriBot. Gard.64:
respectivefamiliesor genera(Gentry,1982a, 694-705.
1986c). A relativelydepauperate Central * 1984. Biosystematics oftropicalforestplants:
Americanforestmade up of species adapted a problemof rare species. Pp. 497-518 in W. F.
Grant(editor),PlantBiosystematics. AcademicPress,
forweedygeneralizedstrategiesmighthave Toronto.
littlein commonwithan upper Amazonian . Ecological studies in the mixed dipterocarp
forestcomposedmostlyofnarrow-habitat spe- forestof northwestBorneo III. Patternsof species
richness(in press).
cialists. Given a few millionmore years of AUBREVILLE, A. 1967. La for&tprimairedes Montagnes
evolution,perhapsCentralAmericanforests de Belinga.Biol. Gabon. 3: 96-112.
mightseem much closer to ecological equi- AXELROD, D. & P. RAVEN. 1978. Late Cretaceousand
Tertiaryvegetationhistoryof Africa.Pp.77-130 in
libriumthan theydo today. M. Werger (editor),Biogeographyand Ecology of
It is no accidentthat many tropicalbiol- SouthernAfrica.W. Junk,The Hague.
ogistswho have consideredthe question of BROWN, A. D., S. C. CHALUKIAN & L. M. MALMIERCA.
1985. Estudiofloristico-estructural de un sectorde
whytropicalforestsare so richinplantspecies selva semideciduadel noresteArgentina1. Compo-
have greatlymodifiedtheiroriginalviews,no sici6nfloristica, densidad,y diversidad.Darwiniana
matteron whichside of the equilibrium/non- 26: 27-41.
CAMPBELL, D. G., D. C. DALY, G. T. PRANCE & U. N.
equilibrium questiontheybegan(Ashton,1969 MACIEL. 1986. Quantitativeecologicalinventory
vs. 1984; Hubbell,1979 vs. Hubbell& Foster, of terrafirmeand varzea tropicalforeston the Rio
in press;Gentry,1982b vs. 1982a and Gen- Xingu,BrazilianAmazon.Brittonia38: 369-393.
CODY, M. L. 1975. Towards a theoryof continental
try& Dodson, 1987b). We stillknowso little species diversities:bird distributionsover Mediter-
abouttropicalforests thatgeneralizationselude raneanhabitatgradients.Pp. 214-257 in M. Cody
us. Almostcertainlythere are elementsof & J. Diamond (editors),Ecology and Evolutionof
Communities.Belknap,Harvard, Cambridge,Mas-
truthon both sides of the question. Quite sachusetts.
possiblydifferent forestsand thedifferenttaxa CONNELL, J. H. 1978. Diversityin tropicalrainforests
that make them up will oftenprove to be and coral reefs.Science 199: 1302-1310.
, J. G. TRACEY & L. J. WEBB. 1984. Compen-
doing thingsquite differently. Althoughwe satoryrecruitment, growth,and mortality as factors
have notyetreachedanythinglike a consen- maintaining rainforesttreediversity.Ecol. Monogr.
sus on how different factors,or even which 54: 141-164.
COWLING, R. M. 1983. Diversityrelationsin Cape
factors,interactto determinediversity,it shrublands and othervegetationsin theSoutheastern
seems abundantlyclear fromthe data pre- Cape, South Africa.Vegetatio54: 103-127.
sented here that there are discernibleand CROAT, T. 1978. Flora of BarroColoradoIsland. Stan-
fordUniv. Press, Stanford,California.
surelydeterministic patternsin the species DAWKINS, H. C. 1959. The volumeincrementofnatural
richnessofdifferent plantcommunities. I con- tropicalhigh-forest and limitations on its improve-
clude thatwhichfamilies,howmanyspecies, ments.EmpireForest.Rev. 38: 175-180.
DODSON, C. & A. H. GENTRY. 1978. Flora of the Rio
and possiblywhatindividualspecies make up Palenque Science Center.Selbyana 4: 1-628.
a tropicalplant communityare to a large & . 1988. Flora de Capeira. Banco
extentdeterministic and predictablefromsim- Centraldel Ecuador, Quito.
, -- & F. M. VALVERDE. 1985. La Flora
ple environmental parameters;how common de Jauneche.Banco Centraldel Ecuador, Quito.
thespeciesare and howtheyare puttogether DONAHUE, P. K., T. PARKER III, B. SORRIE & D. SCOTT.

32 Annals of the
BirdsoftheTambopataNatureReserve,Dept. Madre & C. DODSON. 1987a. Contribution of non-

de Dios, Peru (in press). treesto species richnessof tropicalrain forest.Bio-
EITEN, G. 1978. Delimitation of the cerradoconcept. tropica19: 149-156.
Vegetatio36: 169-178. & . 1987b. Diversityand phytogeog-
EMMONS, L. H. 1984. Geographicvariation in densities raphyof neotropicalvascular epiphytes.Ann. Mis-
and diversitiesof non-flying mammalsin Amazonia. souriBot. Gard. 74: 205-233.
Biotropica16: 210-222. & L. H. EMMONS. 1987. Geographicalvari-
& A. H. GENTRY. 1983. Tropicalforeststruc- ation in fertility and compositionof the understory
ture and the distribution of glidingand prehensile- of neotropicalforests.Biotropica19: 216-227.
tailedvertebrates.Amer.Natur. 121: 513-524. & J. TERBORGH. 1988. Comparison and dy-
ERWIN, T. L. 1985. TambopataReservedZone, Madre namicsofthe Cocha Cashu maturefloodplain forest.
de Dios, Peru: historyand descriptionofthereserve. In: A. Gentry(editor),FourNeotropicalForests.Yale
RevistaPeruana Entomol.27: 1-8. Univ. Press, New Haven, Connecticut(in press).
FEDEROV, A. A. 1966. The structureof the tropical GiVNISH, T. J.,J.W. TERBORGH& D. M. WALLER.Plant
rain forestand speciaton in the humidtropics.J. form,temporalcommunitystructure,and species
Ecol. 54: 1-11. richnessin forestherbs of the Virginiapiedmont
FLORENCE, J. & A. HLADIK. 1980. Cataloguedes pha- (unpubl.).
nerogameset des pteridophytes du Nord-estdu Ga- GRUBB,P. J. 1986. Problemsposed by sparse and
bon(Sixiemeliste).Adansonia,ser. 2, 20: 235-253. patchilydistributed speciesin species-rich
plantcom-
GARTLAN, J. S., D. M. NEWBERY, D. W. THOMAS & P. munities.Pp. 207-225 in J. Diamond & T. Case
G. WATERMAN. 1986. The influence oftopography (editors),Community Ecology.Harper & Row, New
and soilphosphoruson the vegetationof Korup For- York.
est Reserve,Cameroun.Vegetatio65: 131-148. HALL,J.B. & M. D. SWAINE. 1981. Distribution and
GENTRY, A. H. 1978. Diversidadee regenaracaoda EcologyofVascularPlantsin a TropicalRain Forest,
capoeira do INPA, com referenciaespecial as Big- ForestVegetationin Ghana. W. Junk,The Hague.
noniaceae. Acta Amazonica8: 67-70. HALLE,N. 1964. Premiereliste de Phanerogameset
. 1981. Distributional patternsand an addi- de Pteridophytes des environsde Makokou, Kem-
tional species of the Passifiora vitifoliacomplex: boma et Belinga.Biol. Gabon. 1: 41-46.
Amazonianspecies diversitydue to edaphicallydif- - 1965. Seconde liste des Phanerogameset
ferentiated communities.P1. Syst. Evol. 137: 95- Pteridophytes du N.-E. Gabon(Makokou,Belingaet
105. Mekambo).Biol. Gabon. 1: 337-344.
. 1982a. Neotropicalfloristicdiversity:phy- & A. LE THOMAS. 1967. Troisiemeliste des
togeographical connections betwenCentraland South Phan6rogamesdu N.-E. Gabon. Biol. Gabon. 3(2):
America,Pleistoceneclimaticfluctuations, or an ac- 113-120.
cidentof the Andean orogeny?Ann. MissouriBot. & . 1970. Quatriemeliste des Pha-
Gard. 69: 557-593. nerogameset Pteridophytes du N.-E. Gabon (Bassin
1982b. Patternsof neotropicalplantspecies de l'Ivindo).Biol. Gabon. 6: 131-138.
diversity.Evol. Biol. 15: 1-84. HARTSHORN,G. S. 1983. Plants (Introduction).Pp.
1983. Dispersalecologyand diversity in neo- 118-183 in D. Janzen(editor),Costa Rican Natural
tropicalforestcommunities.Sonderbd. Naturwiss. History.Univ. Chicago Press, Chicago.
Verh. Hamburg7: 303-314. HLADIK,A. 1978. Phenologyofleafproductionin rain
- 1985a. Algunosresultadospreliminaresde forestofGabon:distribution and composition offood
estudiosbotanicosen el Parque Nacional del Manu. forfolivores.Pp. 51-71 in G. Montgomery (editor),
Pp. 2/1-2/24 in M. Rios (editor),ReporteManui. Ecology of ArborealFolivores.SmithosonianInst.
Centrode Datos para la Conservaci6n,La Molina, Press, Washington,D.C.
Peru. & N. HALLE. 1973. Cataloguedes Phan6ro-
. 1985b. Plant distribution and diversitypat- games du N.-E. du Gabon (Cinquiemeliste).Adan-
ternsin AmazonianPeru. Natl. Geogr. Soc. Res. sonia, s6r. 2, 13: 527-544.
Rep. 10: 245-252. HUBBELL, S. P. 1979. Tree dispersion, abundance,and
- 1986a. Endemismin tropicalvs. temperate diversity in a tropicaldryforest.Science 203: 1299-
plantcommunities. Pp. 153-181 in M. Soul6 (edi- 1309.
tor), ConservationBiology.Sinauer Press, Sunder- . 1984. Methodologiesfor the study of the
land, Massachusetts. originand maintenanceof tree diversityin tropical
. 1986b. Species richnessand floristiccom- rainforest.Pp. 8-13 in G. Maury-Lechon, M. Had-
positionofChoc6regionplantcommunities. Caldasia ley& T. Younes(editors),The Significance ofSpecies
15: 71-91. Diversityin TropicalForestEcosystems.BiologyIn-
- 1986c. Summariode patronesfitogeograficos ternational(IUBS). Special Issue 6.
neotropicalesy sus implicacionespara el desarrollo & R. FOSTER. 1986. Biology,chance and
en la Amazonia.RevistaAcad. Colomb.Ci. Exact. historyand the structureof tropicalrain foresttree
16: 101-116. communities. Pp. 314-329 in J.Diamond& T. Case
. 1988. Tree species richnessof upper Ama- (editors),Community Ecology.Harper & Row, New
zonianforests.Proc. Natl. Acad. U.S.A. 85: 156- York.
159. & . 1987. La estructuraespecial en
. The transectmethod.In: A. Peixoto(editor), gran escala de un bosque neotropical.RevistaBiol.
ManualforPhytosociology and FloristicsoftheMata Trop. 35 (Suppl. 1): 7-22.
Atlantica.CNPq, Brasilia(in prep.). & . Structure,dynamics,and equilib-

riumstatusof old-growth foreston Barro Colorado * 1969. Patternsofcommunities in thetropics.

Island. In: A. Gentry(editor),Four NeotropicalFor- Biol. J. Linn. Soc. 1: 19-30.
ests. Yale Univ. Press, New Haven, Connecticut(in * 1972. GeographicalEcology:patternsin the
press). distribution of species. Harper & Row, New York.
HUSTON, M. 1979. A general hypothesisof species & E. 0. WILSON. 1967. The Theoryof Island
diversity.Amer.Naturalist113: 81-101. Biogeography. PrincetonUniv.Press,Princeton, New
- 1980. Soil nutrients and treespeciesrichness Jersey.
in Costa Rican forests.J. Biogeogr.7: 147-157. MAY, R. 1975. Patternsof species abundance and
JANZEN, D. H. 1973. Sweep samplesof tropicalfoliage diversity.Pp. 81-120 in M. Cody & J. Diamond
insects:effectsof seasons, vegetationtypes,eleva- (editors),Ecology and Evolutionof Communities.
tion,timeof day, and insularity. Ecology54: 687- Belknap,Cambridge,Massachusetts.
708. MufOz, S., M. 1980. Flora del Parque Nacional Pu-
- 1975. Interactionsof seeds and theirinsect yehue. EditorialUniversitaria, Santiago,Chile.
predators/parasitoids on a tropicaldeciduousforest. NAVEH, Z. & R. H. WHITTAKER. 1979. Structuraland
Pp. 1547186 in P. Price(editor),Evolutionary Strat- floristicdiversityof shrublandsand woodlandsin
egies of ParasiticInsectsand Mites. PlenumPress, northern Israel and otherMediterranean areas. Veg-
New York. etatio41: 171-190.
- 1980. Specificity of seed-attackingbeetlesin PEET, R. K. & N. L. CHRISTENSEN. 1980. Hardwood
a Costa Rican deciduousforest.J. Ecol. 68: 929- forestvegetationof the NorthCarolina Piedmont.
952. Ver6ff.Geobot.Inst.ETH Stiftung RubelZurich69:
- 1984. Two ways to be a tropicalbig moth: 14-39.
Santa Rosa saturniidsand sphingids.OxfordSurv. PEIXOTO, A. & A. H. GENTRY. Diversidadee composicao
Evol. Biol. 1: 85-140. da matade tabuleiroda ReservaFlorestalde Linhares
& R. LIESNER.. 1980. Annotatedchecklistof (EspiritoSanto, Brasil).Rev. Bras. Bot. (in prep.).
lowlandGuanacasteProvincejCosta Rica, exclusive PIRES, J. M. & G. T. PRANCE. 1977. The Amazon
of grasses and nonvascularcryptograms. Brenesia forest:a naturalheritageto be preserved.Pp. 158-
18: 15-90. 193 in G. Prance & T. Elias (editors),ExtinctionIs
, D. M. ATAROFF, M. FARINAS, S. REYES, N. Forever.New York BotanicalGarden,New York.
RINCON, A. SOLER, P. SORIANO & M. VERA. 1976. PRANCE, G. T. 1977. Floristic inventory of the tropics:
Changesin the arthropodcommunity along an ele- wheredo we stand? Ann. MissouriBot. Gard. 64:
vationaltransectintheVenezuelanAndes.Biotropica 659-684.
8: 193-203. , W. A. RODRIGUEZ & M. F. DA SILVA. 1976.
JONES, E. W. 1956. Ecologicalstudieson therainforest Inventarioflorestalde um hectarede mata de terra
of southernNigeriaIV (Part 2). The plateau forest firmekm. 30 da estradaManaus-Itacoatiara.Acta
of the OkomuForestReserve.J.-Ecol.44:83-117. Amazonica6: 9-35.
KAPOS, V. 1982. An EcologicalInvestigation ofSclero- PROCTOR, J.,J.ANDERSON, P. CHAI & H. VALLACK. 1983.
phyllyin Two TropicalForests.Ph.D. Dissertation. Ecological studiesin four contrastinglowlandrain
WashingtonUniversity, St. Louis, Missouri. forestsin GunungMulu NationalPark, Sarawak. I.
KARTAWINATA, K., R. ABDULHADI & T. PARTOMIHARDJO.
Forestenvironment, structureand floristics.
J. Ecol.
1981. Composition and structureof a lowlanddip- 71: 237-260.
terocarpforestat Wanariset,East Kalimantan.Ma- RAMIREZG., C. & M. RIVEROS G. 1975. Los alerzales
layan Forester44: 397-406. de CordilleraPelada: floray fitosociologia.
Medio
Ambiente1: 3-13.
KEAY, R. W. J. 1953. An Outlineof NigerianVege-
tation,2nd edition.Ministryof Information, Lagos. RAVEN,P. H. 1976. Ethics and attitudes.Pp. 155-
179 in J. Simmonset al. (editors),Conservationof
LAMAS, G. 1985. Los Papilionoidea(Lepidoptera)de la
ThreatenedPlants. Plenum,New York, London.
Zona Reservadade Tambopata,Madrede Dios,Peru. & D. AXELROD. 1974. Angiospermbiogeog-
I. Papilionidae,Pieridaey Nymphalidae(En Parte).
raphyand past continentalmovements.Ann. Mis-
RevistaPeruana Entomol.27: 59-73. souriBot. Gard. 61: 539-673.
LORENCE, D. & R. SUSSMAN. 1988. Diversity,density, RICE, B. & M. WESTOBY. 1983. Plant species richness
and invasionin a Mauritianwet forest.In: Modern at the 0.1 hectare scale in Australianvegetation
Systematicsin AfricanBotany: Proceedingsof the comparedto othercontinents.Vegetatio52: 129-
AETFAT EleventhPlenaryMeeting,June 10-14, 140.
1985. Monogr.Syst. Bot. MissouriBot. Gard. 25: RICHARDS, P. W. 1939. Ecologicalstudieson the rain
187-204. forestof southernNigeria.I. The structureand flo-
LOT-HELGUERAS, A. 1976. La estaci6nde biologiatrop- risticcomposition of the primaryforest.J. Ecol. 27:
ical Los Tuxtlas:pasado, presentey futuro.Pp. 31- 1-61.
70 in A. G6mez-Pompa,C. Vazquez-Yanes,S. del . 1952. The TropicalRain Forest.Cambridge
Amo Rodriguez& A. Butanda C. (editors),Rege- Univ. Press, Cambridge.
neraci6nde Selvas. CECSA, Mexico. 1963. Ecologicalnoteson West Africanveg-
LOTT, E. J., S. H. BULLOCK & J. A. SOLIS-MAGALLANES. etationIII. The uplandforestsof Cameroonsmoun-
1987. Floristicdiversity and structureof a tropical tain. J. Ecol. 51: 529-554.
deciduousforestof coastal Jalisco.Biotropica19: RIVEROS G., M. & C. RAMIREZ G. 1978. Fitocenosis
228-235. epifitasde la asociaci6n Lapagerio-Aextoxiconetum
MAcARTHUR, R. 1965. Patternsof species diversity. en el Fundo San Martin(Valdivia-Chile).Act. Ci.
Biol. Rev. 40: 510-533. Venez. 29: 163-169.

34 Annals of the
SALO, J., R. KALLIOLA, I. HAKKINEN, Y. MAKINEN, P. * 1984. Plantdominancealongan experimental

NIEMELA,M. PUHAKKA& P. COLEY. 1986. River nutrientgradient.Ecology65: 1445-1453.
dynamicsand thediversityofAmazonlowlandforest. TROTH, R. G. 1979. Vegetational typeson a ranchin
Nature322: 254-258. the centralLlanos of Venezuela. Pp. 17-30 in J.
SCOTT, N. J., JR. 1976. The abundanceand diversity Eisenberg(editor),VertebrateEcologyin the North-
oftheherpetofauna oftropicalforestlitter.Biotropica ern Neotropics.SmithsonianInst. Press, Washing-
8: 41-58. ton,D.C.
SHMIDA, A. & M. WILSON. 1985. Biological determi- WALTER, H. & H. LIETH. 1960. Klimadiagramm-Wel-
nantsof species diversity.
J. Biogeogr.12: 1-20. tatlas.Fischer-Verlag,Jena.
STALLINGS, J. R., L. WEST, W. HAHN & I. GAMARRA. WARING, R. H. & J. F. FRANKLIN. 1979. Evergreen
Primatesand theirrelationshipto habitatin the Par- coniferousforestsof the PacificNorthwest.Science
aguayanChaco. In: K. Redford& J. Eisenberg(ed- 204: 1380 1386.
itors),Mammalsof the Americas,Essays in Honor WHITE, P. 1983. EasternAsian-easternNorthAmer-
of Ralph M. Wetzel. In press. ican floristicrelations:the plant communitylevel.
STARK, N., A. H. GENTRY & H. ZUURING. Prediction of Ann. MissouriBot. Gard. 70: 734-747.
tropicalplantspecies diversity
fromsoil and precip- WHITMORE, T. C. 1984. TropicalRain Forestsof the
itation.Biotropica(submitted). Far East. Clarendon,Oxford.
TATE,J.,JR. 1969. Pileatedwoodpeckerand otherbirds WHITTAKER, R. H. 1977. Evolutionofspeciesdiversity
at IndianCave StatePark. NebraskaBirdRev. 37(4): in land communities. Evol. Biol. 10: 1-87.
57-60. ZIMMERMAN, M. & W. L. WAGNER. 1979. A description
TILMAN, G. D. 1982. ResourceCompetition and Com- ofthe woodyvegetationofoak-hickoryforestin the
munityStructure.PrincetonUniv. Press,Princeton, northernOzark highlands.Bull. Torrey Bot. Club
New Jersey. 106: 117-122.


Missouri Botanical Garden Press

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Missouri Botanical Garden Press

Uploaded by

Copyright:

Available Formats

Changes in Plant Community Diversity and Floristic Composition on Environmental and

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

' This and thefollowingfivepapers comprisetheproceedingsoftheMissouriBotanical Garden's 33rd Annual

ANN. MISSOURI BOT. CARD. 75: 1-34. 1988.

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

Comparisonsof the species richness(or have reportedthe firstcomparabledata set

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

Janzen& Liesner,1980; Dodsonet al., 1985; so tightlycorrelated(R2 = 0.93) with the

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

TABLE 1. Site characteristicsfor 0.1-ha samples.

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

Chamela,Mexico 19030'N 50 733 34 83 5.42 Lott et al., 1987

Guanacaste(gallery), 10030'N 50 1,600 33+ 68a Hartshorn,

Curundu,Panama 8059'N 20 1,830 42 90 5.78 Gentry,1982b

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

Omo Forest,Nigeria 7?N 50 1,800 29 73 4.42 Richards,1939

1986b). 2) The latitudinaldecreasein species 4) Species-poortropicalforestswithsingle-

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

There are also latitudinaldifferencesin precipitation (Gentry,1982b). However,this

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

Closed line enclosescontinentalAfricanpoints;dottedline enclosesAsian points;dash-dotline enclosesAustralian

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

Annual precipitation (mm x 103)

FIGURE 3. Numbersofspecies in 0.1-ha samples of lowland neotropicalforestvs. annual precipitation.Solid

in a givencommunity have been sampled,but thersupportedby preliminary data from1-ha

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

Relatively Poor soil

Yanamono Mishana Manu Cabeza de Tambopata Tambopata INeblina

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

any othersite (Gentry,1985a). My data do ridge systemtropical-subtropicaldemarca-

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

% of species with numberof individuals tion,when such broader-scalebiogeographic

trientsundoubtedlydo play a major role in

TABLE 3. Numberof species shared by 1,000 m2 samples of Iquitos area foresttypes.

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

o poor soil terra firme forest

| | 1 ~~0 o\ld on sandy river terrace/

0 )/?/ < ~~~terafirmet

0 32 spp. shared ~ ~ 27 pp shre

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

Decrease in diversity with altitude

moist & wet forest n

Eastern North America average Paaochoa

edaphic conditions, incomplete,the trendof decreasingdiversity

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

Lowland Mos ~ and ~ we~ cl

speculationin the literatureabout a "mid- moststriking is theincreasein sampledhemi-

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

TABLE 4. habits in local fiorulas (fromGentry& Dodson, 1987b).

some 312 GreatSmokiesMountainssamples is on an unusuallypoor,highlyleached skel-

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

TABLE 4. Continued. protectedby being an island (Raven & Ax-

FIcSUREI1-2. Basal areas for 1 ha saiples of some different

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

Cocha Cashu r. o '|X|@|m|r|<|P 0 37 other families

Cabeza de Mono co I o '

Neblina r m t. | | | | D) | -> | > | 21fother families

the foests o all hree

less,has a similarrangeofvariationaccording Perinet),and a few otherislands(e.g., New

This content downloaded from 205.133.226.104 on Mon, 15 Jul 2013 21:08:47 PM

AV. FOR 20 LOWLANDNEOTROPICALMOIST ANDWET SITES