Seed and Seedling Ecology of Neotropical Melastomataceae Author(s): Aaron M. Ellison, Julie Sloan Denslow, Bette A.

Loiselle, Danilo Brens M. Reviewed work(s): Source: Ecology, Vol. 74, No. 6 (Sep., 1993), pp. 1733-1749 Published by: Ecological Society of America Stable URL: http://www.jstor.org/stable/1939932 . Accessed: 03/04/2012 12:07
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.

Ecological Society of America is collaborating with JSTOR to digitize, preserve and extend access to Ecology.

http://www.jstor.org

Ecology, 74(6), 1993, pp. 1733-1749 (?) 1993 by the Ecological Society of America

SEED AND SEEDLING ECOLOGY OF NEOTROPICAL MELASTOMATACEAE'

Departmentof Biological Sciences, Mount Holyoke College, Clapp Laboratory, South Hadley, Massachusetts 01075-1484 USA Departmentof Ecology, Evolution,and Organismal Biology, Tulane University, New Orleans, Louisiana 70118 USA Departmentof Biology, University ofMissouri, 8001 Natural BridgeRoad, St. Louis, Missouri 63121 USA Organizaci6n para Estudios Tropicales,Estaci6n Biol6gica La Selva, Apartado 676, 2050 San Pedro de Montes de Oca, Costa Rica
DANILO BRENtS BETTE A. LOISELLE JULIE SLOAN DENSLOW AARON

M.

ELLISON

M.

in tropicalwet forests Abstract. Models of gap-phase regeneration are based primarily on studies of canopy trees.These paradigms are derived fromstudies of a small number of species whose life historytraitsplace them easily into "pioneer" or "climax" species. Woody plants of the forestunderstory, subcanopy trees and shrubs,are known to affect significantly regeneration dynamicsin treefall gaps, yet we lack fundamentalinformation on life historycharacteristics of understory plants that could permittheirincorporation into gap regeneration models. We investigatedintra-and interspecific variabilityin early forms lifehistory characteristics of understory Melastomataceae species of different growth and adult distributions fromthe lowland tropicalrain forest at La Selva Biological Station, in vitro(22 species),effects on germination Costa Rica. We examined germination patterns of passage throughbirds (5 species), seedling establishmentin the field(3 species), and in different growth lightenvironments (8 species). Established seedlingswere significantly more abundant in gaps than in nearbyunderstorysites. This patternwas generallyapplicable to all melastomes studied,regardlessof the habitatwhere adults were normallymost abundant. Seedlingswere more common in the root pits and on the root mounds of uprooted trees,even thoughtherewas less light available to the seedlings in these micrositesthan elsewhere in the gaps. In a series of trialsin vitro,most seeds germinated germination readilywithin 1-2 wk. Comparisons of rates between seeds collected frombird droppingsand seeds extractedfrom germination fruitsby hand gave varied results.With one exception, seeds of shade-tolerantspecies germinatedmore slowly followingpassage throughbirds, while seeds from high-lightbirds. demanding species germinatedmore rapidlyafterpassing through withingaps Seeds of threecommon species were sown directly into different microsites and understory controlsites. In contrastto the germination trials,emergencepercentages in the fieldrangedfrom4% in understory sites to 21% in gap centers.Survival after1 yr was low (0.1%/0.4%). After1 yr,these seedlingswere small (<2 cm tall) and produced in the shade-house,melastome seedlingsurvival fewleaves. In similarlightenvironments was between 80% and 100%, and growthincreased with increasinglightlevels from2% to understory (corresponding sites)to 20% (small gap centers)offullsun. Withtheexception of two species common to abandoned pastures and river banks, the species used in the withincreases shade-housegrowth did not show significant increasesin growth experiments in lightfrom20% to 40%. Suites ofearlylifehistory traitsdescribedfortropicaltreespecies are not consistentwith the variation in patternseen withinthis one familyof understory
species. seedlingestabKey words: Costa Rica; gap dynamics;light;Melastomataceae; seed germination; lishment; seedlinggrowth; seedlings;seeds.

Manuscriptreceived 13 April 1992; revisedand accepted 16 December 1992; finalversion received 11 January1993.

1734
INTRODUCTION

AARON M. ELLISON ET AL.

Ecology, Vol. 74, No. 6

The ability of buried seeds and suppressed forest seedlings to respond rapidly to increased lightlevels has longbeen recognizedas a crucial following a treefall stage in plant life history, and is a fundamentalcomin tropical forests(e.g., ponent of stand regeneration Schulz 1960, Vdzquez-Ydnes 1974, Ng 1978, Denslow 1980, 1987, Clark and Clark 1987). Seed size, germiamong tropstrategies nation rates,and establishment ical treesare diverse(Ng 1978, Whitmore1983, Foster and Janson 1985, Foster 1986, Swaine and Whitmore 1988, Garwood 1989). "Pioneer" (sensu Swaine and Whitmore 1988), or gap-dependent species, such as Cecropiaspp. (Cecropiaceae),Ochroma lagopusSwartz. Turcz. (Bombacaceae), and Heliocarpusappendiculatus (Tiliaceae), have small seeds withextendeddormancy and rapid germinationand growthin response to increasing light levels (Vdzquez-Ydnes 1974, MorenoCasasola 1976, Holthaijzen and Boerboom 1982, and Dirzo 1988). "Climax" or "nonNfiiiez-Farfan pioneer" species (sensu Swaine and Whitmore 1988), on the otherhand, tend to have largerseeds withlittle followingdispersal dormancy,and rapid germination (Foster 1986). Many tree species fall between these limits,combiningrapid germinationand some shade tolerancewith rapid growthresponses followingcanopy opening(Clark and Clark 1992). Data on seed and seedling ecology of species not encompassed by the pioneer and climax categoriesare have scant,however.In commonusage,thesecategories been applied only to canopy or emergenttrees, and ofplanttypes do notincludethediversity consequently In particular, thatmake up any forest. understory speare ubiquitous in tropcies, shrubsand small treelets, of canopy trees ical forests,can inhibit regeneration (Brokaw 1983, Raich and Gong 1990, Denslow et al. 1991), and yet have been studied rarely relative to propcanopytrees.Detailed studiesofearlylifehistory and seedling suchas germination, establishment, erties, needed to species are therefore growthof understory complementthe relativewealth of comparable informationavailable forcanopy treespecies,and to permit dynamicsin more reliable predictionsof regeneration wet tropicalforests. in characteristics Combinations of earlylifehistory from thoseoutlinedabove speciesmaydiffer understory for rain forestpioneer and climax canopy trees. For example, Vdzquez-Ydnes and Or6zco-Segovia (1984) foundthatalthoughseeds of many understory species exhibit enforceddormancy and require gaps for germination,othershave rapid postdispersalgermination oflight With and no dormancy environments. regardless it regardto germinationand seedling establishment, has been suggested thatunderstory species have a wider ecological amplitude (e.g., Lebr6n 1979) and are less dependent on gaps than are canopy trees (Hartshorn species 1978, Lebr6n 1979, Brokaw 1985). Understory are likely to recruitinto new gaps either from sup-

pressed seedlings or as seeds dispersed into gaps followinggap creation(Garwood 1983, Lawton and Putz 1988, Levey 1988a), but appear to not require gap formationto germinate.This is because understory species oftenproduce fewerseeds, and theirseeds are less abundantin theseed bank thanare seeds ofpioneer trees (VAzquez-YAnes 1974, Uhl and Clark 1983, Whitmore 1983, de Foresta and Prevost 1986, Garwood 1989; but see Alvaraz-Buyllaand Garcia-Barrios 1991). In contrast,survival and growthof seedlings and by gap older plants may be influencedmore strongly Growthof environments than is seedlinggermination. highestin gaps relative species is generally understory to other environments (Denslow et al. 1990), and shrubs flower and fruit more abundantlyin gaps (Croat 1975, of fruits Stiles 1975, Levey 1988a). This concentration in gaps has led Levey (1990) to conclude thatgaps may in tropical rain be "keystonehabitats" forfrugivores forests. Gaps (sensu Brokaw 1982) themselvesare not homogeneous, but encompass substantial variation in propmoisture,soil, and nutrient light,temperature, and Dirzo 1988, Poulerties(Putz 1983, Nutfez-Farffan son and Platt 1989, Petersonet al. 1990). If understory dependenton gaps forsuccessful species are not strictly (Hartshorn 1978, Lebr6n 1979, Brokaw recruitment may be less sensitiveto 1985), then seed germination variation withina gap. That is, microenvironmental variation in we would not expect great intraspecific parts of a and establishmentin different germination treefall gap. Later seedling growth,however, should be influenced more strongly by gap microsite,but data addressingthispredictionare rare.Denslow et al. (1990, and unpublished manuscript) found significantresponses to within-and between-gapvariation in light among rooted cuttings availability,but not nutrients, of several species in the Melastomataceae and the Pipstudies eraceae. As a complementto these long-term of the responses of understory gaps species to treefall (Denslow et al. 1990, Denslow et al., unpublished we examinedin detailtheecologyofseeds manuscript), and seedlingsof a numberof melastomespecies in and We around gaps, and in controlledlightenvironments. the earlylife asked seven specific questions concerning historiesof these species: time courses of newly 1. What are the germination dispersed melastome seeds? birds affect 2. Does passage through germination? 3. How is the frequencyof successfulseedling esin thefieldaffected tablishment by gap microsite? 4. Are melastome seedlingsmore common in gaps or in surrounding forest? vari5. What is thedegreeofintra-and interspecific ation in distribution and abundance patternsof these seedlingswithintreefall gaps?

September 1993

MELASTOMES

IN A NEOTROPICAL

RAIN FOREST

1735

of intact 6. How do lightconditions,representative forest,gap edges, and gap centers,affectearly seedlinggrowth? variation in these life 7. What is the interspecific historyparameters,and how does this variation and adult distribution? relateto plant life-form
STUDY SITE

All fieldwork was conducted at the La Selva BiologicalStation(10028' N, 8359' W) oftheOrganization forTropical Studies. This is a reserveof premontane (sensu Holdridge et al. 1971) in the wet tropicalforest Atlanticlowlands of Costa Rica (see Hartshorn 1983 fora more detailed description).Average annual rainfall is -4000 mm, with a less rainyseason fromJanApril, when average monthlyrainfallis uary through <200 mm. The soils at La Selva are derived from volcanic parentmaterial(Sancho and Mata 1987, Sollins et al., in press) and are high in available nitrogen (Vitousek but low in phosphorus and other nutrients and Denslow 1986, Denslow et al. 1987, Sancho and active Mata 1987). Availabilityof photosynthetically radiation (PAR) under intact canopy at La Selva is (Chazdon and Fetchbetween 1 and 2% offullsunlight er 1984). The estimated canopy turnovertime (time in a singlelocation) at La Selva is 118 betweentreefalls + 27 yr (Hartshorn 1978). Most gaps at La Selva are small (88% < 300 M2; Sanfordet al. 1986) with incident PAR levels of -9-20% of full sunlightat their centers (Chazdon and Fetcher 1984, Denslow et al. 1990). Available PAR in large gaps (400 m2-600 M2) (Chazdon and Fetcher1984, is - 20-3/5% offullsunlight Denslow et al. 1990).
STUDY SPECIES

melastomes) are The Melastomataceae (henceforth, one of the most species-richtropical plant families, comprising - 190 generaand 4800 species (F. Almeda, a personal communication).Melastomes are primarily of the species neotropicalgroup;more than two-thirds occur in Centraland South America. All common perenniallife-forms, i.e., herbs,small shrubs,subcanopy are represented and canopytrees, lianas,and epiphytes, withinthe melastomes,althoughthe majorityof speat La shrubs. In the understory cies are understory Selva, themelastomesare themostspeciose shrubfamily (Hammel 1990). In total,22 melastome species were used in the labestaboratoryand fieldexperimentsin germination, lishment, and seedlinggrowth(Appendix 1). Seeds of all of thesespecies are verysmall (100-1000 .g; A. M. data). The majorityof these speEllison, unpublished cies (16 of22) wereshrubs,5 weretreelets or subcanopy trees,one was an herb(Appendix 1). Adult individuals of each of these species are found most commonlyin habitats rangingfrom abandoned pastures to undis-

turbed forest,as well as in swamps and along river banks (Appendix 1; J. S. Denslow and A. M. Ellison, We assignedthespecies' light manuscript). unpublished listed in Appendix 1 based on observed requirements patterns(J. S. Denslow patternsof adult distribution and A. M. Ellison, unpublishedmanuscript)and independent observations of seedling distributionpatterns (A. M. Ellison and D. Brenes M., unpublished data). Miconia nervosahas been considered a "highlight"species in otherstudiesat La Selva (e.g.,Denslow et al. 1990); we consider it a "low-light"species here withrespect because ithas a muchbroaderdistribution to ambient lightconditions than the other high-light species in Appendix 1. All of the species listed in Appendix 1 were used to studytime courses of seed germination in vitro. Subsets of these species were used of dispersalby examiningthe effect in the experiments experimentson fieldestablishbirds on germination, ment,and studies of early seedlinggrowth.For these we selected species representative latterexperiments, and adult habitats,and for life-forms of the different numbersof seeds or seedlings. whichwe had sufficient was biOur selectionof species forthe experiments was the ased in two ways. Availabilityof freshfruits Of the -77 melastome species at constraint. primary La Selva (F. Almeda, personal communication),we on only the 22 species listed in Apencounteredfruits pendix 1. Second, we wished to compare our data with in treeresultsobtained fromstudiesofrootedcuttings fallgaps at La Selva (Denslow et al. 1990, and unpubThese studies have focused on six lished manuscript). M. barbinervis, common melastomes:Miconia affinis, M. nervosa,and M. simM. gracilis,M. multispicata, plex. These species were included in our experiments whenever possible. M. gracilis did not germinatereliably in vitro or in the field,and was not used in the M. multispiseedling growthstudies. M. barbinervis, cata, and M. simplex seeds were not available in sufstudies in ficient quantityforuse in the establishment the field.
PATTERNS OF GERMINATION: GERMINATION IN VITRO

STUDIES

Methods We examined the germinationrates of freshmelaFive stomeseeds in vitro(question 1,see Introduction). of each of 22 common melastome species to 20 fruits multipleplantsat variouslocations werecollectedfrom within La Selva. Ripe berries of these species were crushed,and the seeds separated fromthe pulp in a cup of water. The water and pulp were decanted and the seeds wereair dried (20'C) for72 h and thenstored dryat 20'C untilused (1-3 mo). Nepsera aquatica produces dry,dehiscentcapsules,and theseeds wereeasily shaken out of the capsules. Althoughwe do not know of storingthese seeds before the physiologicaleffects use, the rates and total numbers of stored seeds that

1736
1 .0

AARON M. ELLISON ET AL.

Ecology, Vol. 74, No. 6

0.8 -0.6 ? 4
04

Species of disturbed areas:

0 N. aquatic * Co. subcrustulata v

Y~~~~~~VV
|o
/
V'? -I V

0.2 0.0 1.0 z

li

//

v M. barbinervis

M. fomgfolia C c.dentata

O V *+

B 0.8F

wv

* -v

*High-light species: * L. mexicana M. affinis O L. dichotoma

0.6

/forest

04 -v
02-

1.0

0.8 0.6 0.4 0.2

0.0

JA_

/ I I 7/1 ' *
f111 c;'fi' < 0
X /

7.8 -Vt-V-VV

forest species:
V

Low-light

; * _= C. discolor C. japurensis

I /L

/ 7 /

0 C. crenulata
?
0

A M. simplex L,, M. rtervosa

v L. consimilis
*0. milcrantha

0. macrophylla * M. multispicata

M. cerntroclesma

10

20 DAYS

30

40

FIG. 1. Germinationpercentageover time of melastome seeds in petridishes.

fromseeds planted markedly did not differ germinated after collection(forthose fivespecies also immediately withbirds describedbelow, testedin the experiments and one additional species [Miconia multispicata]examined by Gonzalez J. 1991). One hundred seeds of each species were placed on paper in each of two replicate 10 cm saturatedfilter diameterpetriplates (except 50 seeds/dishof M. barand C. subcrustulata).Petri plates were then binervis lights fluorescent placed 20 cm beneath full-spectrum ( 250 ,mol.m-2 s-') on a 12:12 light:dark photocycle. Temperaturewas not regulated;ambient dayat La Selva was _250C throughout time temperature did not appear and plate temperature the experiment, by the lights.Every 48 h, to be increased significantly 0.5 mL of distilledwaterwas added to the petridishes to preventdesiccation. Every 2-3 d for 29 d we exwith visible amined the seeds, counted all germinants radicles,and removed any seeds that had become infectedwith fungi(in all cases < 5%). It appeared that had ceased after1 mo, but we maintained germination

the plates foran additional 15 d to ensurethatno late germinants were missed. Results Seeds of most species germinated within10-12 d of sowing (Fig. 1), and littleto no change in cumulative was observed after3 wk. Gergermination frequency minationrates of shade-tolerant species were variable (Fig. 1). Clidemia discolor,C. japurensis,Leandra consimilis,Miconia nervosa,and M. simplex,germinated rapidly, with cumulative frequencies quickly approaching100%. Clidemia crenulata,Miconia centroand 0. desma, M. multispicata,Ossaea macrophylla, micranthagerminatedmore slowlyand with variable success. Clidemia densiflora,Miconia dorsiloba, M. gracilis,and M. grayumiidid not germinate at all (Fig. 1) althoughfew (<5%) of the seeds were infectedby fungi.The remainingspecies, common to more disturbed areas (3 species) and abandoned pastures (5 with the species), all germinatedwith high frequency,

September 1993

MELASTOMES

IN A NEOTROPICAL

RAIN FOREST

1737

and Miconia barbiexceptionsof Leandra dichotomy nervis (Fig. 1), ofwhich < 20% ofthe seeds germinated. in germinathe differences To examine statistically tion ratesbetweenspecies common to abandoned passpecies of the forest,and shade-toltures,high-light erantspecies (Fig. 1), we divided each of these species groups into two sub-groups:those with time to 50% germination< 10 d; and those with time to 50% gerdifferences mination > 10 d. There were no significant timesamong thethreegroupsofspecies in germination
(X272 =1.30;

P > .50, G test).

EFFECTS OF

PATTERNS OF GERMINATION:

BIRDS ON SEED GERMINATION

Methods Many melastome fruitsare eaten by birds, which also disperse theirseeds (Denslow et al. 1986, Levey 1988a, b, 1990, Denslow and Gomez-Diaz 1990, Loibirds has been selle and Blake 1990). Passage through to enhanceseed germination, but experimental thought (Izhaki and Safriel studieshave producedmixedresults offruit theeffects also investigated 1990). We therefore rates(quesconsumptionby birdson seed germination tion 2, see Introduction). We collectedadditional fruits MiConostegia subcrustulata, of Clidemia densiflora, five M. simplex,and Ossaea macrophylla, conia affinis, species that had exhibited disparate responses in the
germination experiments in vitro.

resentsa successive day, with time proceeding from the bottom leftcornerto the top rightcornerof each plot. We used a least squares linear regressionmodel to fita line throughthese points. If the slope of this different from 1, then we line was not significantly in cumulative concluded that therewas no difference at any givbetweentreatments germination frequency deviations froma slope of 1 inen time. Significant frequency of dicated that the cumulative germination consumed by birds increased more seeds fromfruits rapidly(slope > 1) or less rapidly(slope << 1) thanthat of controlseeds. If the data were betterfitby a polynomial (quadratic or cubic) regressionmodel than a that there were also diflinear one, then we inferred ferencesin the time of onset of germinationbetween control seeds and seeds frombird droppings.All regressions were forced throughthe origin,because at forboth time = 0, cumulative germination frequency
treatments = 0.

Results Consumption by birds affectedthe rate and frebut not preciselyas we quency of seed germination, had hypothesized (Fig. 2). Consistent with our hyfed to pothesis,Clidemia densifloraseeds fromfruits earlierand more rapidly birdsgerminated significantly than corresponding controls(best fitcurve was cubic, > 1). However,thetotalpercentage withall coefficients of C. densifloraseeds that did germinatewas much lower than any otherspecies in this experiment.Also in difference as hypothesized, therewas no significant seeds that rateofConostegiasubcrustulata germination had passed throughbirds relative to controls(linear, with slope 1). Consonant with our predictions,Miconia simplex birds began germiseeds that had not passed through soonerthanseeds ofthesespeciesthathad passed nating cumulativegerbirds. Contrary to prediction, through mination frequency of seeds frombird droppingsincontrols creased more rapidlythan the corresponding coeffithird-order polynomial-term (cubic regression, cient > 1). Also consistentwith our hypothesis,seeds of M. affinis frombird droppingsbegan germinating earlier than control seeds. However, cumulative germinationfrequency of controlseeds increasedat a sighad comnificantly more rapid rate once germination menced (quadratic regression, polynomial-term > 1). The germinationbehavior of Ossaea coefficient macrophyllawas similar in all respectsto that of M. simplex.

and In those experiments, Conostegia subcrustulata M. simplex had germinatedreadilyand rapidly,with total cumulative germinationapproaching 100%. We thatpassagethrough birdswould therefore hypothesized germination even inhibit or might notenhance further, and 0. macrophylla of these two species. M. affinis in thein vitro more slowlyand less reliably germinated trials,and forthese two species we hypothesizedthat passage throughbirds could marginallyenhance gerat all did not germinate mination.Clidemia densiflora in the in vitro trials,and we predictedthat bird consumption would dramatically enhance germination seeds. and rate of C. densiflora frequency Fruitsof each species were fedto Red-capped Manakins (Pipra mentalis)caughtin mistnets(Loiselle and Blake 1991). Seeds were then separated from fresh droppingsand placed into petri dishes on moistened filter paper. For controls,seeds were handled as in the in vitro germination experiments. Individual petri dishes contained eitherthe seeds fromone dropping (conor the seeds fromone intactfruit (1-7 replicates), trols,2-3 replicates).Germinationwas scoreddaily for 6 wk, with the exception of C. densiflorawhich was GERMINATION AND ESTABLISHMENT IN THE FIELD followedfor 16 wk. Methods We plotted the mean cumulative proportion(over Our results from the above studies indicated that all replicatesof each species) of seeds thatgerminated afterconsumptionby birds on the abscissa, and the germinationpotential of melastome seeds was relainvestigatedthe corresponmean cumulativeproportionof controlseeds thatger- tivelyhigh. We therefore minated on the ordinate. Each point of the plot rep- dence of these resultswith seed germinationand es-

1738 1.00

AARON M. ELLISON ET AL.

Ecology, Vol. 74, No. 6

,'

Clidemia

densiflora

,'

Conostegia subcrustulata

<

z 0.75
0.50

,~~~~~~~,

E-

1.00

rF

"
m~107'

1X
1 oI. /

***e icniap~,,

=
CI)

0.25

4~~~~~~~~~~~~~~~~~0
0.000

0.!
Micomia

00.

1.00

29t

Ossaeea

macrophylla

0 10 0

/~~~~~4

0.25/

o
0.500

015~~~~~~~~~~~~~~~~~~~~~~~~~~~1

0.00

0.25

0.50

0.75

1.00

0.00

0.25

0.50

0.75

1.00

CUMULATIVE
FIG. 2.

FRACTION

OF CONTROL SEEDS

THAT GERMINATED

birds relativeto paired time of melastome seeds passed through Comparison of cumulative germination through numbers indicate particulardays afterthe startof the germination controls.Each point represents a single day; forclarity, ). The ) along with its 95% confidenceinterval(. trials. The best-fit regressionline throughthe points is plotted ( of birds. . : Methods forinterpretation effects See Patternsofgermination: dashed line (slope = 1) is provided forreference. and statisticaldiscussion.

tablishmentin the field,and examined the effects of different on seedlingemergencein lightenvironments gaps (question 3, see Introduction). Seeds of Miconia affinis, M. gracilis,and M. nervosa wereextracted from as in thein vitrogermination fruits experiments,but instead of being stored dry, were planted within 2 d of collection directlyinto recent gaps (< 1 yrold) createdbythefallofsingletrees.Three "replicate" gaps were used in this experiment, all on the same soil type (Jaguarconsociation; Sancho and Mata 1987) on ridgetops. Three light treatments (plots) wereestablishedwithineach gap. In the centerof each gap, vegetation<2 m tall was cleared froma 5 x 5 m plot to createthe high-light In the mid-light treatment. treatment, seeds were plantedin a 5 x 5 m plot within the gap where the low vegetationwas leftintact.The low-lighttreatment seeds were planted into 5 x 5 m plots in the adjacent forestunderstory. In each plot,

we established three 100 x 75 cm quadrats. In each and planted 25 seeds of quadrat we removed the litter each species 2 mm beneaththesoil. Seeds wereplanted 3 cm apart in three rows separated by 20 cm. In total, we planted 675 seeds of each species: 75 seeds (threereplicatequadrats) of each species in each light in each site. Seeds were sown in September treatment 1988, and quadrats were sampled once every2 wk for 1 yr.At each samplingdate, all seedlingsthatemerged (cotyledons visible above the soil) were noted and markedwitha small plastic stake adjacent to the seedling. Height (? 1 mm) and number of leaves were recorded. Plots were weeded once every2 wk. Annual incident direct and indirectsunlightat 50 cm above each quadrat was estimated from hemisphericalcanopy photographs (Rich 1988, 1989) taken witha 7.5 mm fisheye Minolta lens. Photographswere taken 1 wk afterthe seeds were sown. Negatives were

September 1993
TABLE

MELASTOMES

IN A NEOTROPICAL

RAIN FOREST

1739

sits.

1. Seedling emergenceand survival of Miconia affinis, M. gracilis,and M. nervosasown into gaps and understory Withingaps Vegetationremoved to 2 m Emerging Low vegetationintact Emerging No. 6 14 5 % 8 19 7 Number surviving after1 yr 0 0 0 Number surviving after1 yr 0 1 3 In forest understory Vegetationintact Emerging No. 7 6 3 % 9 8 4 Number surviving after1 yr 1 0 0

Species M. affinis M. gracilis M. nervosa

No. 9 21 19

% 12 28 25

digitizedand analyzed forestimatedannual directsite tions 4 and 5, see Introduction). We definedseedlings factor(DSF) (directbeam radiation) and indirectsite as plants < 15 cm tall, lackingwoody stems,and not factor (ISF) (reflected skylight) as a proportionof total sproutsattachedto largerplants. incident sunlightusing softwaredeveloped by Rich In areas considered primaryor secondaryforest we (1989, Rich et al. 1993). The global site factor(GSF; in 10 naturally sampled seedlingdistribution occurring equivalent to the total site factorof Anderson 1964) single-tree treefall gaps (<300 M2) that we estimated is a weighted sum of DSF and ISF (0.45 DSF + rangedin age from6 to 30 mo since treefall (D. Brenes 0.55 ISF; Rich et al. 1993). These measures of light M. and A. M. Ellison, personal observations), and in availabilityare strongly correlatedwithavailable pho- 10 paired understory sites. Understorysites were lotosynthetically active radiation (PAR) and/orphoto- cated between 30 m and 50 m of the gap edge (sensu synthetic photonflux density (PPFD) as measuredwith Brokaw 1982). These understory sitesincludeda fallen quantum sensors (Rich et al. 1993). trunk,approximatelythe same lengthas that of the paired gap's "gap-maker," that was surroundedand Results overtoppedby low (_ 2 m) vegetation.Such sitescould old gaps or understory in each treatment were sim- represent trees that had fallen The lightenvironments ilar in each of the threesites into which Miconia seeds withoutdisturbingthe forestcanopy, and contained were sown. GSF was 0.10 ? 0.009 (mean + 1 SE, n = micrositessuch as root pits, root mounds, and fallen 3) in thehigh-light treatment, 0.05 ? 0.029 in themid- boles that in gaps were known to affect seedling dislight treatment, and 0.006 ? 0.003 in the low-light tribution(e.g., Putz 1983). Site locations and charactreatment.Emergence of seedlings in the three light teristicsof gaps and paired understory sites are given was low relative to the number of seeds in Appendix 2. treatments In each gap and paired understory sown (Table 1). Cumulative emergenceafter1 yrof all site,we counted treatments (sec- melastomeseedlingsin therootpit,on therootmound, species was highestin the high-light and in three 1 m wide ond-growth vegetation removed) and lowestin the un- on top of the fallenbole itself, derstory (Table 1), and therewere no between-species transectsparallel to the bole at 1 m, 5 m, and 10 m in emergencewithrespectto the threelight distancefromthe fallenbole. Each transect was as long differences (X24 = 6.85, .5 > P > .1). Overall, total as the lengthof the fallenbole measured fromits base environments main branch.Our samplingconventionasemergence(pooled over the three species) increased to the first sumed the fallenbole to be in the gap center;transects with increasinglight(X22 = 19.39, F < .001). In total,only fiveseedlingsremainedalive after1 yr ? 1 m fromthebole therefore werenear thegap center, (Table 1). Resultsare pooled forthethreesitesin Table while transects? 10 m fromthe bole were nearerthe 1. All were <20 mm tall after1 yr of growthand the gap edge. We also used the convention that transects few leaves were all < 1 cm2 in area. Many seedlings labelled + 1, + 5, and + 10 m were to the rightof the appeared to have been browsed,but we have no quan- bole (looking up the trunk),and -1, -5, and -10 m forthese seedlings. transects wereto theleft. titativedata on herbivory based Seedlingswereidentified on comparisonswith seedlingvouchers (Denslow and Gomez Diaz 1990). Seedlingsthatwerenot identifiable PATTERNS OF DISTRIBUTION AND ABUNDANCE OF from voucherswerecollectedand grownin shade housESTABLISHED SEEDINGS es at La Selva untillarge enough to identify. Methods In each gap and controlsite, hemisphericalcanopy We sampled the naturaldistribution and abundance photographs weretakenonce, in April 1990, to provide of melastome seedlingsin gaps and in the undisturbed a measureoflight weretaken environment. Photographs at La Selva to determinewhetherthe composi- 50 cm above groundlevel in the centerof the root pit, forest tion and abundance of melastome species differed be- on the bole at its center,and in the centerof the + 1, tweengaps and understory or among microsites + 5, and + 10 m transects.Only one photograph(on forest, (ques- the bole) was taken in each understory withingaps and understory lightenvironments site,as canopy

1740
TABLE

AARON M. ELLISON ET AL.

Ecology, Vol. 74, No. 6

2. sites.

Total number(with% of total) of Melastomataceae seedlingsfoundin naturaltreefall gaps and understory control In gaps Species No. 0 5 7 10 1 21 4 41 0 0 15 168 34 0 4 42 2 0 14 368 % 0.0 1.4 1.9 2.7 0.3 5.7 1.1 11.1 0.0 0.0 4.1 45.7 9.2 0.0 1.1 11.4 0.5 0.0 3.8 In understory No. 1 0 0 4 0 2 0 5 1 1 1 3 1 1 0 0 0 1 0 21 % 4.8 0.0 0.0 1.6 0.0 8.0 0.0 23.8 4.8 4.8 4.8 14.3 4.8 4.8 0.0 0.0 0.0 4.8 0.0

Adelobotrys adscendens (SW.) Triana Clidemia discolor(Triana) Cogn. Clidemia ephiphytica (Triana) Cogn. var. trichocalyx (Blake) Wurdack Conostegia setosa Triana Henrietteatuberculosa(Donn. Sm.) L. 0. Wms. Leandra dichotoma(D. Don) Cogn. Leandra granatensisGleason Miconia affinis DC. Miconia appendiculata Triana Miconia centrodesmaNaud. Miconia gracilisTriana Miconia multispicataNaud. Miconia nervosa(J. E. Smith) Triana Miconia punctata (Desr.) D. Don ex DC. Miconia serrulata(DC.) Naud. Miconia simplex Triana Miconia stevensianaAlmeda Ossaea macrophylla (Benth.) Cogn. Triolena hirsute(Benth.) Triana Total

cover was not assumed to vary as widely in the understory.Photographswere analyzed as described in experiment (see Germination and thefield germination in thefield:Methods). establishment Results Overall, melastome seedlings were 20-fold more abundant in gaps than in correspondingunderstory sites (Table 2). Of those species consideredhigh-lightdemanding (Appendix 1), only Miconia affinisand Leandra dichotomaoccurredin the gaps thatwe samspecies listed in Appendix pled. The other high-light 1 are moretypicalofhighly disturbed areas (abandoned and are not common in small pasturesand riverbanks), In total,368 seedlings werefound theforest. gaps within [X ? 1 in the 10 gap sites (24.9 ? 9.89 seedlings/M2 SE] over all gap sites),whileonly21 totalseedlingswere sites (1.1 ? 0.27 found in the 10 paired understory
seedlings/M2 over all control sites) (t9 = 2.32, P
=

paired t test). Of the 19 species encountered,14 were The more common in gaps than in the understory. otherfivespecieseach wererepresented by onlya single individual.Amongthe seven gaps in whichmelastome seedlingsoccurred,the number of seedlingsincreased with gap age (r2 = 0.69, F1 6 = 13.15, P significantly
=

.045,

In these sites,we examined in more detail the seedof the more common species lingdistribution patterns whose abundance permitteddetailed statisticalanalM. gracilis, ysis: Leandra dichotoma,Miconia affinis, M. multispicata, M. nervosa,and M. simplex.All spemorecommon cies exceptM. gracilisweresignificantly in treefallgaps than in the forestunderstory (Fig. 3, in seedling abunTable 3). To examine differences

.01).

dances among gap microsites, we analyzed within-gap abundance patternsusing G testson two-waycontingencytables, where the six rows were location within gap (pit, mound, bole, and transectsat + 1 m, + 5 m, and ? 10 m) and the columns were log2 seedlings/M2. Degrees of freedomforeach G test are given by (row 1) x (column 1), where (row 1) = 5, and (column 1) varies between 1 and 3, dependingon the actual number ofnon-zeroseedlingcategories.Mean (? 1 SE) number of seedlingsin each within-gapsamplinglocation are shown in Fig. 3. All six of these species were more abundant in root pits and mounds than elsewhere within gaps. These differences were significant in the cases of M. affinis (X2 = 23.15, df= 10, P = .01), M. multispicata(X2 = 31.38, df = 15, P = .008), and M. nervosa(X2 = 15.12, df= 5, P = .009). The otherspecies showed a tendency towards greaterabundance in root pits and on root mounds (M. gracilis: X2 = 17.98, df= 10, P = .06; L. dichotoma:X2 = 10. 1, df = 5, P = .07; M. simplex: X2 = 8.08, df= 5, P= .15). Overall, seedings were not distributed randomly withingaps. Location withingap (i.e., in the root pit, on therootmound, on thebole, or at varying distances fromthe bole) had a significant effect on seedlingdensity (Table 4, Fig. 3). Tukey's a posteriorimultiple range test indicated that significantly more (P < .05) seedings were found in root pits or on root mounds than in any other location. Seedling densities in root pits and root mounds, however,did not differ signififrom cantly each other(P > .05, Tukey's multiplerange test). Similarly,comparisons between seedling densities in the otherlocations withingaps and understory controlsdid not reveal significant differences (P > .05).

September 1993 4

MELASTOMES
|

IN A NEOTROPICAL dichotoma

RAIN FOREST Miconia multispicata

1741

Leandra
=In

20F
10

=iIn gaps

2-

understory

Miconia

affinis

Miconia

nervosa

~40

Miconia

gracilis

20

16 Fn

Miconia

simplex

P
FIG. 3.

| 1mjj5m|| 1Omj |j~j1m

LOCATION
sites. Locations Seedling densityof six most common melastome species in the 10 gaps and paired understory are: P, root pit; M, on root mound; B, on root bole; I 1ml, in transects+ 1 m frombole; 15m 1, in transects+ 5 m from bole; I 1Om1,in transects? 10 m frombole. Data are means and 1 SE.

ANOVA also indicatedthatvariationin seedlingabundance amongwithin-gap sitesdiffered betweenyounger and older alluvial soil types(interaction termin Table 4). In gaps on younger alluvial soils, seedlingswereless abundant in root pits or root mounds relativeto other siteswithinthesegaps. In gaps on weatheredalluvium (Arboleda consociation), seedlingswere always much more abundant in root pits and on root mounds. In all gap sites, lightlevel (estimated as GSF) was highest atop thebole, and declined steeplytowardsthe gap edge (Fig. 5). Light levels atop fallen trunksin sites were lower than light levels at gap understory edges (Fig. 5). Because of shading effects by the root mound, lightlevels in the root pit were comparable to those found between 1 and 5 m away fromthe fallen bole (Fig. 5). Thereweresignificant differences between locations withingaps (P < .001, ANOVA; Fig. 5), as well as significant differences betweengaps in all three lightmeasures (P = .001, ANOVA; data not shown). Location withina gap was a stronger determinant of seedlingabundance than was variation in lightavailabilitywithingaps (Fig. 5). Average seedling density was two orders of magnitudegreaterwithinroot pits than in any otherlocation withingaps, althoughlight levels within root pits were intermediaterelative to otherpositions withinthe gaps (Fig. 5). ANOVA indicated that overall, only location within gap had a

significant effect on seedlingabundance (Table 4); the additionofany measureoflight availability, global site factor,direct site factor,indirect site factor,or unweightedopenness,as a covariate in the ANOVA did not improve the fitof the model.
SEEDLING GROWTH IN CONTROLLED LIGHT ENVIRONMENTS

Methods Althoughprevious fieldand shade-house studiesindicated that all melastome cuttingsresponded positivelyto some increase in lightavailability (Denslow
TABLE 3.

Results of Wilcoxon's matched-pairs signed-ranks tests contrasting seedling densities in and out of gaps for the six most common species. Two-tailed P .002 .019 .0001 .0008 .116 .028

Light requirement* High High Low Low Low Low

Species Leandra dichotoma Miconia affinis Miconia multispicata Miconia nervosa Miconia gracilis Miconia simplex

Z-score 3.06 2.35 4.84 3.35 1.57 2.20

* Lightrequirement: Low, some degreeof shade tolerance; High, high-light-demanding.

1742
20

AARON M. ELLISON ET AL.

TABLE

Ecology,Vol. 74, No. 6

5. Survivorshipfromfinaltransplant to harvest(10 wk) of melastome seedlingsin threedifferent lightlevels.

|IIm In gaps
10 _

In understory

requirement*

Light

level(0/ of full Light

Species 2% sunlight) 20% 40% Survivorship(%)

cn
-4

PI

MOUND PIT

0. 2 0A

0.2

High Conostegia subcrustulata 100 100 100 High Nepsera aquatic 100 60 100 High Miconia longifolia 100 95 85 High Miconia barbinervis 100 100 100 High Miconia affinis 90 100 80 Low Miconia multispicata 80 95 100 Low Miconia nervosa 95 90 75 Low Miconia simplex 65 70 30 * Lightrequirement: Low, some degreeof shade tolerance; High, high-light-demanding.

soil is likelyto have mycorrhizaepresent.Sixtyseedlingsof each species (except 20 each of M. barbinervis and C. subcrustulata) were assigned randomlyto three 0 -10 -5 -1 +1 +5 +10 light treatments:2%, 20%, and 40% of full sun (20 DISTANCE FROM BOLE (m) These lightlevels were similarto seedlings/treatment). FIG. 4. Density of melastome seedlings away from the those measured in the rain forestunderstory, small gapmakerbole and in tip-uppits and mounds within10 gaps (<300 M2) treefall gaps, and large (>400 M2) treefall and paired understory sites. gaps (Chazdon and Fetcher1984, Denslow et al. 1990). Every week we measured total stem lengthas the et al. 1990, and unpublished manuscript), responsepatheight of the plant fromsoil level to base of top-most ternsof melastomeseedlingswere undocumented.We on themain stemplus thelength leaf ofany lateral pair examined seedling growthof eight species of melastome under threedifferent controlledlightlevels in branches. After10 wk, all survivingplants were harshade-housesat La Selva (question 6, see Introduction). vestedand separatedinto leaves, stems,and roots.ToNewly germinatedseedlingsof Conostegia subcrustu- tal leaf area of each seedlingwas measured witha LIM. barbinervis, M. longifolia, M. COR leaf area meter. To assess changes in biomass lata, Miconia affinis, we multispicata,M. nervosa, M. simplex and Nepsera allocation patternsover the threelighttreatments, dried 40 and separately (50WC, h) weighed (?0.0001 g) aquatica weretransplanted into 50 x 25 x 6 cm plastic leaves, stems, and roots. flatsof heat-sterilized alluvial sand fromthe banks of in 20% of the Rio Puerto Viejo. After8 wk growing Results full sunlight, the seedings developed one pair of true in shade-houseswas generally Seedlingsurvivorship leaves and were transplanted into flatsof sterilesand > 80% in all lightlevels (Table 5), except forM. simand alluvial soil (50:50). Clear plastic sheetingwas placed 1 m above the flatsto preventrain drops from 20 uprooting seedlings. The plastic uniformlyreduced
ambient light by

Twelve weeks later (two pairs of true leaves, -10 mm tall),each seedlingwas transplanted intothecenter of a 300 mL, 6 cm diameterplastic cup filledwith a 50:50 sand: soil mixture.This mixturewas not sterilized. All of these species are mycorrhizal (J. S. Denslow and A. M. Ellison, unpublishedmanuscript), and althoughwe did not deliberatelyinfectthe soil, fresh
4. ResultsofANOVA testing effects of location within gap and soil typeon seedlingdensity. Source ss 96.95 5.96 28.84 59.30 191.05 df 5 3 14 65 87 MS 19.39 1.99 2.06 0.91 2.21 F 21.25 2.18 2.26 P <.001 .099 .014

5 5%.

15
CQ

0.5
0.4

Root pit

0.3

+5m +10m +m

F~~~~~11
0.1-

0.2 _

TABLE

I
Understory 0.00
I I I

Bole

0.0

Location (L) Soil type(S) L x S Residual Total

0.05

0.10

0.15

GLOBAL SITE FACTOR FIG. 5. Relationship betweenseedlingdensityand global site factor(GSF) at fivelocations withingaps and in the adValues shown are means ? 1 SE. jacent forest understory.

September 1993
TABLE

MELASTOMES

IN A NEOTROPICAL

RAIN FOREST

1743

lightconditions(n = sample size). Intraspecific 6. Mass (g) at finalharvestof melastomeseedlingsin thethreedifferent between final (P < .05, Tukey's HSD test for multiple comparisons among means) across light treatments differences letters. superscripted biomasses are indicated by differing Light level (% of fullsunlight) 20% n
X I1 SE

Light require- Lifement* form*

2% Species
X ? ISE

40%
X ? 1 SE

n 6 17 20

7 S Conostegia subcrustulata High 20 10.009 + 0.00 a S Miconia longifolia High 20 0.013 ? 0.002a H Nepsera aquatic High 0.004 ? 0.0004a 6 S Miconia barbinerx'is High 18 0.008 ? 0.00 a T Miconia affinis High 16 0.013 ? 0.002' T Miconia multispicata Low 19 0.004 ? 0.0002a S Miconia nervosa Low 0.003 ? 0.0002a 13 S Miconia simplex Low * See Appendix 1 fordefinition and life-form. of lightrequirement 0.017 ? 0.009a understory, species of the forest plex, a shade-tolerant in which survivorshipdeclined dramaticallywith increasinglightlevel. Survivorshipof the other species did not varywith lightlevel (Table 5). heavier at harvestin All species were substantially relative to the mass of their the 20% lighttreatment (Table 6). grownin the 2% lighttreatment conspecifics Only the biomass of Nepsera aquatic, a high-light bespecies common to riverbanks,increased further tween 20% and 40% sunlight(Table 6). Harvest bioat 20% light(Tamass of Miconia nervosawas greater in growthrate among species and ble 6). Differences treatmentparallel those in final biomass (data not shown). Leaf area and specificleaf mass were oftenhigherat thanat the2% levels typical 20% or 40% offullsunlight of forestunderstory, although the patternswere not
TABLE

0.811 + 0.185b 0.105 ? 0.020b

0.083 ? 0.029a 0.061 ? 0.041a

7 19 7 20
12

0.031 ? 0.008a
0.082 ? 0.014b 0.089 ? 0.023b 0.008 ? 0.002b

19 18 14

0.889 ? 0.115b 0.109 ? 0.038b 0.325 + 0.105b 0.055 ? O.0ab 0.029 ? 0.010ab

0.080 ? 0.020b 0.028 ? 0.005a 0.008 ? 0.00 lb

16 20 15
6

or shade-tolconsistentwithin high-light-demanding erant groups (Table 7). Relative allocation to shoots among lighttreatconsistently and rootsdid not differ ments(Table 7).
DISCUSSION

patternsof germiEarly life historycharacteristics, difand early seedlinggrowth, nation, establishment, feredamong the melastomes that we studied. Their seeds' responses to passage throughbirds, likelihood of establishment,and seedling growthresponses to and adult light conditionsvariedwithlife-form varying did habitat.In addition, suites of these characteristics not co-occur in the same combinationsthathave been found for pioneer or climax canopy tree species. In variation closing,we discuss patternsof interspecific in these characters that we observed among mela-

lightconditions (% of at finalharvestof melastome seedlingsin the threedifferent 7. Morphological characteristics differences (P < .05, Tukey's HSD test for multiple comparisons among means) across light full sunlight).Intraspecific letters.Sample sizes as in Table 6. between finalbiomasses are indicated by differing superscripted treatments Specificleaf mass (g/cm2) 20% 40% 2%
1.3a
0.19 2.6a 2b 0.12 2.6a 2.7c 0.16
4.5b

requirement*

Species

Statistic
X
SE

Leaf area (cm2)

2% 6.8a
3.89 1.8a

20%
127.6b
16.68 23.6b

40%
130b 12.89
13 .9b

Root: shoot ratio (g/g) 40% 2% 20%

0. 9a
0.15 1.3a 0.25
0.3b

High Conostegia subcrustulata

High High High High Low Low Miconia Nepsera Miconia Miconia Miconia Miconia longifolia aquatic barbinervis affnis multispicata nervosa

1.1a
0.10
0.4ab

X
SE

0.31
7.0a

4.42
27.3ab

4.27
60.6b

0.11 0.8a 0.09

2.1a

0.09
.1
a

0.46
2.
lb

0.5 a 0.03
0.4a

0.03
0.7b

0.05
0.5a

X
SE

1.34
0.9a

8.71
14.5a

15.98
8.5a

0.16
.9a

0.25
4
Oa

0.04
1a

0.11
0.9a

0.10
0.4a

X
SE

0.25 1.7a 0.24

3.5a

9.43
7.3b

0
5.9a

0.58
2.4a

0.23
2.7ab

0
3.7b

0.51
0.7a

0.23
0.4a

0
0.5a

X
SE

1.94
20.8a

1.99
13.1l

0.3
1.9a

0.40
2.2ab

0.57
4.3b

0.15
0.7a

0.06
0.7a

0.11
0.7a

X
SE

0.82 1.la 0.12

4.32
21.4b

3.08
7.7ab

0.15
1.6a

0.09
2.6b

0.98
3.3b

0.08
0.8a

0.11 0.7 a 0.17 0.8a 0.1

0.06
0.4a

X
SE

5.56

1.96

0.16 4.5a 1.04

0.11 16.8a 5.32

0.41 13.4a 3.84

0.23 0.7a 0.10

0.04

Low
*

Miconia simplex

X
SE

0.4a
0.11

1.2a
0.38

0.3a
0.12

0.6a
0.10

Low, some degree of shade tolerance;High, high-light-demanding. Lightrequirement:

1744

AARON M. ELLISON ET AL.

Ecology,Vol. 74, No. 6

stores, and speculate on the relationshipof this vari(question ation to plantlifeformand adult distribution 7, see Introduction).

rrios 1991), and much less than those observed for large-seeded species (e.g., Augspurger 1984, Howe like othof melastomeseedlings, 1990). Establishment er small-seeded species, was low in forestunderstory Germination environments.In field experiments,emergence and environments All the melastomes examined produce very small survival were highestin the high-light gaps forall species planted.Naturallyestabtrials in vitro indicated some of treefall seeds. The germination degreeof induced dormancyamong at least 18 of the lished melastome seedlings also were more common 22 melastomes,whose adult habitatsrangefromopen in gaps than in the forestunderstory. Seedlingsof seven melastome species, Clidemia disWe base this conclusion understory. pasturesto forest M. gracilis, on these species' responses to germinationfollowing color,Leandra dichotoma,Miconia affinis, were all M. simplex, M. and nervosa, M. multispicata, dry storage. manual extractionand prolonged cool, gaps than in nearby However, we have little basis for speculation about foundmore commonlyin treefall matureindividuals of two Reproductively under understory. requirements dormancycapacityor germination field conditions. Among canopy trees, only pioneer of these species, C. discolorand M. gracilis,were not to gaps, and in factare common in the unof disturbedhabitats have been restricted species characteristic at La Selva (J. S. Denslow and A. M. Ellison, found to have extended dormancy (Vaizquez-Ydnez derstory The otherfivespecies appear manuscript). 1974, Moreno-Casasola 1976, Holthaijzen and Boer- unpublished (Levand flowering boom 1982, Vdzquez-Ydnezand Or6zco-Segovia 1984, to be dependenton gaps forfruiting uncommonly were found that 1990). Species 1 988a, ey four Dirzo 1988). The and Foster 1986, Nfifiez-Farfdn species that did not germinatein the in vitro trials, in our search forestablished seedlingsare also found and rarelyas matureadults at La Selva (J. S. Denslow and M. gracilis, Miconia dorsiloba, Clidemia densiflora, species common in the A. M. Ellison, unpublished manuscript). However, are all understory M. grayumii, at La Selva. On seedlings of two of these species, Conostegia setosa, dense shade of the undisturbedforest the basis of trials in vitro,we cannot draw any con- Henriettea tuberculosa,were more common in gaps while the resttended to occur than in the understory, clusions about dormancyin these fourspecies. in the understory. We had hypothesized that passage throughbirds more frequently illusField contrastsbetween gaps and understory ofspecies that enhancegermination would significantly had responded poorly in the in vitro trials. This hy- tratedapparentgap-dependenceof many melastomes this resultis evidence of pothesis was supported for Clidemia densiflora,the forsuccessfulestablishment; Within on seedlingestablishment. only species testedof the fourthathad not germinated a stronglighteffect (1975) gaps, seedling distributionswere not related to estiin thein vitrotrials. Mayerand Poljakoff-Mayber birdsand defecationof mated lightavailability;seedlingswere most abundant thatpassage through suggested seeds generallyshould enhance germination in root pits and on root mounds althoughlightavailpulp-free lower in root pits than in gap of fleshy fruited species. However, otherstudies have abilitywas significantly measuresof lightwithquantitative However, centers. should be suggestedthat germinationof such species to seedwerenot relatedsignificantly more variable (Izhaki and Safriel 1990 and references in gap microsites ger- ling distributionwithin gaps, and higherlight levels and thatless predictableor asynchronous therein), (e.g., Mur- have been oftencited as one of the main reasons that minationshould be favoredevolutionarily after tree seedlings and saplings grow betterin gaps (e.g., responses ray 1988). We examined germination passage throughonly one bird, Pipra mentalis.Other Frankie et al. 1974, Denslow 1980, Putz 1983, Augsbirds feed on melastome fruits (Denslow et al. 1986, purger1984, Fetcheret al. 1987). Putz (1983) found Levey 1988b, 1990), and results of identical experi- thatpioneer treeseedlingscolonized pits and mounds these birds could yield resultsdifferent at BarroColorado Island, Panama. He attributed mentswithdifferent found to improvedconditionsforgermination Pipra, like most frugivorous findings fromours. In particular, theexposureof minincluding gizzard thatcould abrade seed withinthesemicrosites, species,lacks a grinding withothergermination eral soil. Tiny roots of lilliputianseedlingsare able to coats. Additional experiments the soil unimpededby theprespenetrate conditions or bird species could be used to examine immediately criticallyour observations that germinationpatterns ence of litteror a dense root mat. However, nutrient or adult habitat among availability(extractableP, N mineralization)is lower are variable withinlife-form in the root throwzones of gaps at La Selva (Vitousek these melastomes. and Denslow 1986). Establishment species eshave describeddifferential Otherworkers responsesto within-gapmicrosites(Brattablishment (on theorderof 1 x 10-3 Establishment probabilities and Dirzo 1988, Petersonet M. gracilis, and M. nervosa across all ton 1976, NUiez-Farffin forM. affinis, habitats) were low but similar to those observed for al. 1990). Both the higherlightconditions in treefall at gaps and the presence of exposed mineral soil appear othersmall-seededspecies,e.g., Cecropia obtusifolia Los Tuxtlas, Mexico (Alvarez-Buyllaand Garcia-Ba- to enhance seedling establishmentin high-light-de-

September 1993

MELASTOMES

IN A NEOTROPICAL

RAIN FOREST

1745

melastomespecies. Workmandingand shade-tolerant Peterson et al. (1990) found ing in temperateforests, formscolonized difgrowth thatspecies withdifferent small-seedwithin gaps. In particular, microsites ferent species were more common in pits ed, wind-dispersed speanimal-dispersed and mounds,whilelarge-seeded, cies and species that relymore heavily on vegetative weremore common in less disturbedmireproduction crosites. to thispattern seed rain may contribute Differential as well. Althoughno quantitativedata are available, seed rain may be higher in gaps than in the forest as well Fruitproductionin shade-tolerant understory. species is higher in gaps as high-light-demanding (Denslow et al. 1986, Levey 1988a, 1990). Increased birds and higherrates of fruit activityof frugivorous removal (Levey 1988a, b) also may produce higher rates of seed rain. Levey (1988a, 1990) also has sugto gap gestedthatseeds may be disperseddifferentially edges wherebirds perch followingfeeding.Since root dispersal defineone edge ofa gap, differential pitsoften to the higherdensityof seedlings may also contribute in root throwzones.

any suite of responses common to species fromparticular habitats. For example, Conostegia subcrustulata, foundonlyin abandoned pasturesand along roadsides, and the shade-tolerantMiconia simplex had similar biomass allocation patternsto leaves, stems, and roots at all lightlevels. However, total leaf area in and specificleaf mass of these two species differed had Conostegia subcrustulata each lightenvironment. greaterleaf area but lower specific leaf significantly mass than M. simplex in each of the threelightenvironments.The othersix species, adults of which were fromabandoned pasturesto foundin habitatsranging (Appendix 1) showed more similar forestunderstory biomass allocation responsesto each otherthan to either C. subcrustulataor M. simplex in each lightenvironment.As with germinationand establishment in of earlyseedlinggrowth patterns therefore, patterns, in adult distridifferences these species did not reflect butions.
CONCLUDING REMARKS

Among La Selva melastomes,variationin seed gerand seedlingdistrimination,seedlingestablishment, bution was not consistentamong species fromsimilar in different lightenvironments Seeding growth habitats.It may not be possible to generalizebroadly categoriesderived fromstudies of cause fromregeneration levels have been citedas a primary Higherlight of more rapid growthof treeseedlingsand saplingsin canopy trees (e.g., pioneer vs. climax: Swaine and gaps (e.g., Frankie et al. 1974, Denslow 1980, Putz Whitmore 1988, Whitmore 1989) and based on corestablishmentand growthchar1984, Fetcheret al. 1987). Although related germination, 1983, Augspurger plant families(cf. the patternsof light intensityand wavelengthcom- acteristicsof species fromdifferent between shade-house and natural gap Martinez-Ramos et al. 1989). Our data froma single position differ imconstraints our shade-house resultsare consistent plant familysuggestthat phylogenetic environments, traitsand thatthe some lifehistory influence with field experimentson some of the same species portantly oftenused to describelifehis(Denslow et al. 1990). Growthresponsesvarybetween suites of characteristics may be decoupled where phylogenetic the two extremesclassically described for photosyn- torystrategies For example,thespeciose genera are strong. ratesof high-light-demanding constraints theticresponses:growth levels,where- Miconia, Clidemia, Conostegia, and Ossaea all prospecies continueto increaseat high-light berriescontaining largenumbersof ofshade tolerant species plateaus at lower duce small,watery as thegrowth lightlevels. Among the species we studied,both high- very small seeds, traits often ascribed to pioneer or (N. aquatic, Co. subcrustulata)and colonizing species. Species in these genera, however, light-demanding sitesheaviofhabitatsincluding shade-tolerant (Mi.simplex)exhibitedtheexpectedpat- occurin a wide variety terns.Betweenthese extremes,however,we foundno ly disturbedby human activityand intacttropicalrain In additionthisstudyhas shownthat understory. consistentpatternsin seedlinggrowthrates to distin- forest ratesof these species and seedlinggrowth adult habitat dis- germination guish among species with different a priori tributions. Species rankingsbased on mass at harvest are similarand thateven species characterized were similar at 2% and 20% of full sunlight.These as highlyshade tolerantappear to require gaps and and establishment. typical disturbedsoil forgermination the lightenvironments growthpatternsreflect However, none of the species that we characterized mosaic: deeply shaded forestunof a gap-understory were foundas seedlingsin eiChazdon as high-light-demanding derstory (lightlevels at 1-3% of fullsunlight; in the rain forest environments by occasional small gaps thergap or understory and Fetcher1984) punctuated (lightlevels at 10-20% of fullsunlight;Denslow et al. at La Selva. In a separate studyof seed rain into four in this heterogeneousenvi- gaps in this forest(Denslow and Gomez Diaz 1990), 1990). Species persisting ronment might be expected to exhibit considerable 122 seeds of 20 melastomespecies werecollectedover shade tolerance as well as some plasticityof growth 1 yr. None of the species characterizedhere as highoccurred in those collections (J. S. response in the presence of the enhanced lightlevels light-demanding Denslow and A. E. Gomez-Diaz, unpublisheddata). associated withcanopy disturbance. seed rain may thusaccount forthe Similarly,comparisons of allocation patterns be- Lack of an effective did not reveal absence of seedings of these species in gaps, although tween species within light treatments

1746

AARON M. ELLISON ET AL.

Ecology, Vol. 74, No. 6

our data indicate thatsuch seedlingsshould be able to survive initially following germination. Low adult growthrates and perhaps survival in light environmentscharacteristic ofthisforest (Denslow et al. 1990) apparently account forboth the observeddistributions of adults and the lack of seed sources forgap colonization. Species lacking an appreciable level of shade toleranceas adults may be unable to surviveextended periodsoftimein theforest understory betweencanopy opening events. The combined effects of inadequate of adult survivalin lowseed rain and low probability adult lightenvironments may thus produce restricted distributions althoughseed and seedlingtraitsare similar to those of more widelydistributed species.
ACKNOWLEDGMENTS

We thank Joel Alvarado C., Guillermo Chaverri S., and Elizabeth Farnsworthfor assistance in the field.The OrgaLe6n, nizationfor Tropical Studies,and especiallyJoseArturo and David and Deborah Clark, provided logistical support and access to a marvelous environmentin which to work. and FrankAlmeda, Carol Augspurger, Elizabeth Farnsworth, Doug Levey provided valuable comments on early versions of this paper. This work was supported by NSF BSR 8605106, and is contribution No. 15 fromthe La Selva Treefall Gaps Project.
LITERATURE CITED

Anderson,M. C. 1964. Studiesofthewoodland light climate I. The photographic computationoflight condition.Journal of Ecology 52:27-41. Alvarez-Buylla,E. R., and R. Garcia-Barrios. 1991. Seed and forestdynamics: a theoreticalframework and an example fromthe neotropics.American Naturalist137:133154. Augspurger, C. K. 1984. Lightrequirements of neotropical treeseedlings:a comparativestudyof growth and survival. Journalof Ecology 72:777-795. Bratton, S. P. 1976. Resource division in an understory herb community:responses to temporal and microtopographic gradients.American Naturalist110:679-693. Brokaw,N. V. L. 1982. The definition of a treefall gap and its effect on measures of forestdynamics. Biotropica 14: 158-160. 1983. Groundlayerdominance and apparent inhibition of tree regeneration by Aechmea magdalenae (Bromeliaceae) in a tropical forest.Tropical Ecology 24:194200. 1985. Treefalls,regrowth and communitystructure in tropicalforests.Pages 53-69 in S. T. A. Pickettand P. S. White,editors.The ecology of natural disturbanceand patch dynamics.Academic Press, Orlando, Florida, USA. Chazdon, R. L., and N. Fetcher. 1984. Photosynthetic light environmentsin a lowland tropical rain forestin Costa Rica. Journalof Ecology 72:553-564. Clark, D. A., and D. B. Clark. 1987. Andlisis de la regeneraci6nde drbolesdel dosel en bosque muyh(imedotropical: aspectos teoreticos y practices. Revista de Biologia Tropical 35 (Supplement 1):41-54. Clark, D. A., and D. B. Clark. 1992. Life history diversity of canopy and emergenttreesin a neotropicalrain forest. Ecological Monographs 62:315-344. Croat, T. B. 1975. Phenologicalbehavior of habit and habitatclasses on BarroColorado Island (Panama Canal Zone). Biotropica 7:270-277.

de Foresta, H., and M.-F. Prevost. 1986. Vegetationpionniere et grainesdu sol en fort Guyanaise. Biotropica 18: 279-286. among tropicalrainDenslow, J. S. 1980. Gap partitioning forest trees.Biotropica 12 (Supplement):47-55. gaps and tree species di1987. Tropical rainforest 18:431Annual Review ofEcologyand Systematics versity. 452. Denslow, J. S., and A. E. Gomez Diaz. 1990. Seed rain to Canadian Journal gaps in a neotropicalrain forest. treefall of Forest Research 20:642-648. Denslow, J.S., T. Moermond,and D. J.Levey. 1986. Spatial treesand shrubs. displayin understory componentsof fruit Pages 37-44 in A. Estrada and T. H. Fleming, editors. Frugivoresand seed dispersal. Dr. W. Junk,The Hague, The Netherlands. Denslow, J. S., E. Newell, and A. M. Ellison. 1991. The and on thegrowth effect of understory palms and cyclanths survival of Inga seedlings.Biotropica 23:225-234. Denslow, J.S., J.C. Schultz,P. M. Vitousek,and B. R. Strain. gap 1990. Growthresponses of tropical shrubsto treefall environments. Ecology 71:165-179. Denslow, J. S., P. M. Vitousek, and J. C. Schultz. 1987. in a tropicalrainforest soil. limitation Bioassays ofnutrient Oecologia (Berlin) 74:370-376. Fetcher,N., S. F. Oberbauer, G. Rojas, and B. R. Strain. y el 1987. Efectosdel regimende luz sobre la fotosintesis crecimientoen plantulas de arboles de un bosque Iluvioso tropical de Costa Rica. Revista de Biologia Tropical 35 (Supplement 1):97-1 10. Foster,S. A. 1986. On the adaptive value of largeseeds for Botanical trees:a reviewand synthesis. tropicalmoistforest Review 52:260-299. Foster, S. A., and C. H. Janson. 1985. The relationship conditionsin tropical betweenseed size and establishment woody plants. Ecology 66:773-780. Frankie,G. W., H. G. Baker,and P. A. Opler. 1974. Comparative phenological studies of trees in tropical wet and ofEcology in thelowlands ofCosta Rica. Journal dryforests 62:881-919. in a seasonal tropGarwood, N. C. 1983. Seed germination ical forest in Panama: a community study. Ecological Monographs 53:159-181. . 1989. Tropical soil seed banks: a review.Pages 149and R. L. Simpson,editors. 209 in M. A. Leck, V. T. Parker, Ecology of soil seed banks. Academic Press, San Diego, California,USA. Gonzalez J., E. 1991. Recolecci6n y germinaci6nde semillas de 26 especies arb6reas del bosque humedo tropical. Revista de Biologia Tropical 39:47-51. among famofdiversity Hammel, B. 1990. The distribution ilies, genera,and habitattypesin the La Selva flora.Pages 75-84 in A. H. Gentry, editor.Four neotropicalrainforests. Yale UniversityPress, New Haven, Connecticut,USA. dynamG. S. 1978. Tree fallsand tropicalforest Hartshorn, ics. Pages 617-638 in P. B. Tomlinson and M. H. Zimmermann,editors. Tropical treesas living systems.CambridgeUniversityPress, Cambridge,England. 1983. Plants. Introduction. Pages 118-183 in D. H. Janzen,editor. Costa Rican natural history.Universityof Chicago Press, Chicago, Illinois, USA. Holdridge,L. R., W. C. Gremke,W. H. Hatheway,T. Liang, in tropicallife and J.Tosi, Jr. 197 1. Forestenvironments zones: a pilot study.Pergamon Press, Oxford,England. Holthaijzen, A. M. A., and J. H. A. Boerboom. 1982. The BioCecropia seedbankin the Surinamlowland rain forest. tropica 14:62-68. Howe, H. F. 1990. Survival and growthof juvenile Virola of herbivoryand canopy surinamensisin Panama: effects closure. Journalof Tropical Ecology 6:259-280.

September 1993

MELASTOMES

IN A NEOTROPICAL

RAIN FOREST

1747

Izhaki, I., and U. N. Safriel. 1990. The effect of some Mediterranean scrublandfrugivores upon germination patterns. Journalof Ecology 78:56-65. Lawton, R. O., and F. E. Putz. 1988. Natural disturbance and gap-phaseregeneration in a wind-exposed tropical cloud forest. Ecology 69:764-777. Lebr6n, M. L. 1979. An autecological studyof Palicourea riparia Bentham as related to rain forestdisturbance in Puerto Rico. Oecologia (Berlin) 42:31-46. Levey, D. J. 1988a. Tropical wet foresttreefallgaps and distributions of understory birds and plants. Ecology 69: 1076-1089. 1988b. Spatial and temporalvariationin Costa Rican fruit and fruit-eating birdabundance. Ecological Monographs58:251-269. 1990. Habitat-dependentfruiting behaviour of an and tropicaltreefall understorey tree, Miconia centrodesma, in Costa Rica. Jourhabitatsforfrugivores gaps as keystone nal of Tropical Ecology 6:409-420. Loiselle, B. A., and J. G. Blake. 1990. Diets of understory fruit-eating birds in Costa Rica. Studies in Avian Biology 13:91-103. Loiselle, B. A., and J. G. Blake. 1991. Temporal variation in birds and fruits along an elevational gradientin Costa Rica. Ecology 72:180-193. Martinez-Ramos, M., E. Alvarez-Buylla,and J. Sdrukhan. 1989. Tree demographyand gap dynamics in a tropical rain forest. Ecology 70:553-558. Mayer,A., and A. Poljakoff-Mayber.1975. The germination of seeds. Pergamon Press, Oxford,England. Moreno-Casasola, P. 1976. Viabilidad de semillas de drboles tropicales y templados: una revision bibliogrifica. Pages 471-526 in A. G6mez-Pompa, C. Vazquez-Yanes, S. del Amo Rodriguez, and A. Botanda Cervera, editors. Investigacionessobre la regeneraci6n de selvas altas en Veracruz, Mexico. CompafniaEditorial Continental,Mexico City, Mexico. Murray,K. G. 1988. Avian seed dispersal of three neotropicalgap-dependentplants. Ecological Monographs 58: 271-298. Ng, F. S. P. 1978. Strategiesof establishmentin Malayan forest trees.Pages 129-162 in P. B. Tomlinson and M. H. Zimmermann, editors. Tropical trees as living systems. CambridgeUniversityPress, Cambridge,England. Nuinez-Farfan, J., and R. Dirzo. 1988. Within-gapspatial in a Mexican tropheterogeneity and seedlingperformance ical forest. Oikos 51:274-284. Peterson,C. J.,W. P. Carson, B. C. McCarthy,and S. T. A. Pickett. 1990. Micrositevariationand soil dynamicswithin newlycreatedtreefall pits and mounds. Oikos 58:39-46. Poulson, T. L., and W. J. Platt. 1989. Gap light regimes influence canopy treediversity.Ecology 70:553-555. Putz, F. E. 1983. Treefall pits and mounds, buried seeds, and the importanceof soil disturbanceto pioneer treeson Barro Colorado Island, Panama. Ecology 64:1069-1074.

Raich, J. W., and W. K. Gong. 1990. Effectsof canopy in a Malaysian dipteroopeningson treeseed germination carp forest. Journalof Tropical Ecology 6:203-218. Rich, P. M. 1988. Video image analysis of hemispherical Pages 84-95 in P. W. Mausel, editor. canopy photography. First special workshop on videography.Terre Haute, Indiana, USA. . 1989. A manual for analysis of hemispherical Los Alamos Technical Manual LAcanopy photography. 11733-M. Rich, P. M., D. B. Clark, D. A. Clark, and S. F. Oberbauer. in a tropical studyof lightenvironments 1993. Long-term wet forestusing quantum sensors and hemisphericalphoand Forest Meteorology,in press. Agricultural tography. Sancho, F., and R. Mata. 1987. Estudio detallado de suelos. Estaci6n Biologica La Selva, Organizaci6n para Estudios Tropicales, San Jose, Costa Rica. Sanford,R. L., Jr.,H. E. Braker,and G. S. Hartshorn. 1986. Canopy openingsin a primaryneotropicallowland forest. Journalof Tropical Ecology 2:277-282. in northSchulz, J. P. 1960. Ecological studieson rainforest ern Suriname. Verhandelingen der Koninklijke NederNatuurkunAfdeling landse Akademie van Wetenschappen de Tweede Reeks 53:1-267. Sollins, P., F. Sancho, R. L. Sanford,and G. G. Parker. In press. Soils of La Selva. In L. McDade, K. Bawa, H. Heseditors.La Selva: ecologyand penheide,and G. Hartshorn, Universityof natural historyof a neotropical rainforest. Chicago Press, Chicago, Illinois, USA. and humphenology, Stiles,F. G. 1975. Ecology,flowering mingbirdpollination of some Costa Rican Heliconia species. Ecology 56:285-301. Swaine, M. D., and T. C. Whitmore. 1988. On thedefinition of ecological species groups in tropical rain forests.Vegetatio 75:81-86. Uhl, C., and K. Clark. 1983. Seed ecology of selected Amazon basin successionalspecies. BotanicalGazette 144:419425. Vdzquez-Ydnes, C. 1974. Studies on the germinationof seeds of Ochroma lagopus Swartz. Turrialba 24:176-179. Vdzquez-Ydnes, C., and A. Or6zco-Segovia. 1984. Ecoin thetropicalhumid forests ofseed germination physiology of the world: a review. Pages 37-50 in E. Medina, H. A. Mooney, and C. Vdzquez-Ydnes, editors. Physiological ecology of plants of the wet tropics. Dr. W. Junk,The Hague, The Netherlands. Vitousek, P. M., and J. S. Denslow. 1986. Nitrogenand gaps of a lowland tropphosphorousavailabilityin treefall ical forest. Journalof Ecology 74:1167-1178. Whitmore,T. C. 1983. Secondary succession fromseed in Abstracts44:767-779. tropicalrain forests.Forestry 1989. Canopy gaps and the two major groups of forest trees.Ecology 70:536-538.

1748

AARON M. ELLISON ET AL. APPENDIX 1

Ecology, Vol. 74, No. 6

Melastomataceae used in germination and seedlinggrowth experiments. All collectionlocations refer to La Selva Biological Station. All collectiondates are in 1988. Species Life-form* Habitatt Light requirement

Clidemia crenulataGleason? S L P/S C. densiflora (Standl.) Gleason? S P L C. dentata D. Don.? S H S/O C. discolor(Triana) Cogn.? S S L C. japurensis DC. var. heterobasis (DC.) Wurdack? S R/Sw L 0 Conostegia subcrustulata (Beurl.) Triana H S Leandra consimilisGleason? S P/S L L. dichotomy (D. Don.) Cogn.? S H S/O L. mexicana (Naud.) Cogn.? H S S/O Miconia affinis DC. T H S/O M. barbinervis (Benth.) Triana 0 S H P M. centrodesma Naud.? S/T L M. dorsiloba Gleason? P T L P M. gracilisTriana? S L P M. grayumiiAlmeda? S L 0 M. longifolia(Aubl.) DC. H S M. multispicata Naud. T S L M. nervosa(J.E. Smith) Triana S P/S L P M. simplex Triana S L Nepsera aquatic (Aubl.) Naud. R/Sw H H P/S Ossaea macrophylla(Benth.) Cogn.? S L 0. micrantha(Sw.) Macf. ex Cogn.? T P/S L * Life-forms: H, herb;S, shrub(multipletrunks, much secondary littlesecondarywood on main stem); T, tree(singletrunk, wood on main stem). t Habitat wherecollected: P, primaryforest; S, secondaryforest;0, abandoned pasture;R, riverbanks; Sw, swamps. t Lightrequirement: L, some degree of shade tolerance;H, high-light-demanding. ? Used in germination studyonly.

September 1993

MELASTOMES

IN A NEOTROPICAL

RAIN FOREST

1749

APPENDIX 2 Description of sites sampled forMelastomataceae seedlings.The location given is the nearesttrail,the distance along that trail,and the direction(degrees E of magneticN) fromthe trail. The orientationof the fallenbole is given as degrees E of magneticN. Location
______

Site 1 Gap Understory 2 3 4 5 Gap Understory Gap Understory Gap Understory Gap Understory 6 7 8 9 10 Gap Understory Gap Understory Gap Understory Gap Understory Gap Understory

Bole species Cespedesia macrophyllaSeem. (Ochnaceae) Pentaclethramacroloba (Willd.) 0. Ktze (Mimosaceae) P. macroloba P. macroloba unidentified P. macroloba P. macroloba P. macroloba PterocarpusrohriiVahl (Papilionaceae) panamensis (Pittier)Record Dipteryx (Papilionaceae) Pentaclethramacroloba P. macroloba Hymenolobiummesoamericanum Lima (Papilionaceae) P. macroloba P. macroloba P. macroloba Dussia macroprophyllata (Donn. Sm.) Harms (Papilionacae) P. macroloba Inga sp. (Mimosaceae) P. macroloba

______

Trail (m) 0E

______

O rienta-

Approxiage (mo) 24
mate

tion (0E) 2250 2600 3000 3250 1400 2000 2290

3300 1550 1540

consociation* Arboleda Arboleda

Soil

CCC 575, 1600 CCC 550, 1400 CCC 650, 3050 CCC 620, 3250 CCC 685, 1400 CCC 720, 1400 SSO 150, 800 SSO 150, 800 SURA 250, 2400 SURA 200, 1540 CEN 640, 1300 CEN 650, 3400 CEN 690, 2480 CEN 700, 600 SSO 750, 1400 SSO 760, 1400 CCL 200, 2400 CCL 270, 1300 CCL 650, 2920 CCL 725, 1680

...

12

Arboleda Arboleda Arboleda Arboleda Arboleda Arboleda La Selva La Selva

30 18 7

3050 1610 2010

1390

3 27 24 21
... ...

Chanchera Chanchera Chanchera Chanchera Arboleda Arboleda Holdridge Holdridge Holdridge Holdridge

2300 2200 1700 1650 2510 3100

10

* Soil taxonomyfollowsSancho and Mata (1987) and Sollins et al. (in press). The Arboleda consociation is a weathered alluvial soil, the Chanchera and La Selva consociations are alluvial soils fromthe middle trace of the Rio Sarapiqui, and the Holdridge consociation is an alluvial soil fromthe middle terraceof the Rio Puerto Viejo.