Phenology, Flowering Synchrony, and Fruit Set of Six Neotropical Shrubs

Author(s): Carol K. Augspurger

Source: Biotropica, Vol. 15, No. 4 (Dec., 1983), pp. 257-267
Published by: The Association for Tropical Biology and Conservation
Stable URL: http://www.jstor.org/stable/2387650 .
Accessed: 08/04/2013 20:21

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.

The Association for Tropical Biology and Conservation is collaborating with JSTOR to digitize, preserve and
extend access to Biotropica.

http://www.jstor.org

This content downloaded from 142.51.1.212 on Mon, 8 Apr 2013 20:21:06 PM

All use subject to JSTOR Terms and Conditions
Phenology, FloweringSynchrony,and FruitSet of Six Neotropical Shrubs1

Carol K. Augspurger
of Illinois,Urbana, Illinois61801 U.S.A.
Departmentof Plant Biology,University

ABSTRACT
Temporalpatternsof flowerproductionand the level of fruitset were determinedfor20 individualseach of six shrub species
(fourfamilies)in a semi-deciduouslowlandforestin Panama. The specieswere:Hybanthus Turnerapanamensis,Rinorea
prunifolius,
sylvatica,Psychotriahorizontalis,
Erythrina var.panamensis,and Pentagoniamacrophylla.
costaricensis Thereweretwo objectivesof
the study: 1) to compareamong species the relationbetweenthe individual'sflowering patternand the population'sflowering
synchrony; and 2) to comparewithinspeciesthe relativeinfluenceof the individual'sand the population'sflowering phenology
on the individual'sfruitset.
The six speciesdifferedin numberof flowersper individual(mean values forspeciesrangedfrom98-2995), how long the
individualproducedflowers(mean valuesrangedfrom3.5-59.0 days),and synchrony of theindividualwithits conspecifics
(mean
values rangedfrom0.48-0.95, wherevalue of 1.0 = perfectsynchrony). Among the six species,populationsynchrony increased
as the mean durationof an individual'sflowering decreased.Populationsynchrony of the firstday, peak (median) day, and the
entireflowering period werehighlycorrelated.
When comparingindividualswithineach species,the individual'sflowernumberwas the best predictorof fruitset. Neither
the individual'slengthof flowerproductionnor its synchrony with conspecificsadded significantlyin explainingthe variancein
fruitset. A regressionincludingthe individual'snumberof flowers, lengthof flowerproduction,and synchrony withconspecifics
as independentvariablesand theproportionfruitset (and its arcsintransformation)as thedependentvariableyieldedno significant
regressions.
The consequencesofthesewidelyvaryingphenologicalpatternsarediscussed.Comparisonsaremade withthetemporalpatterns
observedin othertropicalforests.

OUR UNDERSTANDING OF TROPICAL REPRODUCTIVE synchrony and flowerabundancethroughoutthe popu-

PHENOLOGY is best developed at the communitylevel lation'sflowering period.No quantification ofpopulation
(flowering: Frankieet al. 1974, Putz 1979, Opler et al. synchrony offlowering is availablefortropicalspecies(but
1980a; fruitmaturation and seed dispersal:Frankieet al. see Primack1980 fortemperatemeasures).
1974, Opler et al. 1980a, Foster 1982; germination: Pollinationof an individual'sflowersmay be influ-
Garwood 1983). A fewcomparative studiesare available encedby itsown patternof flowerproductionand/orby
withinone family(Gentry1974) or one genus (Snow its synchrony with conspecifics.Previous experimental
1965, Stiles 1975). Few phenologicalstudieshave fo- studieswith the tropicalshrub, Hybanthusprunifolius,
cused on the populationand the individualscomprising demonstrated that mass-flowering by the individualin-
it (Koelmeyer1959, Bawa 1977, Augspurger1981, Bul- creasedattraction of social bees (Augspurger1980); syn-
lock and Bawa 1981, Bullock et al. 1983). chronywiththepopulationfurther enhancedtheindivid-
The individual'sflowering patternis definedby the ual's seed set (Augspurger1981).
durationof flowering as well as thenumberand temporal My studyof six shrubspeciesaddressedtwo ques-
distribution of flowers.Among tropicalspecies,the flow- tions: 1) How does the populationflowering synchrony
eringphenologyof individualplantsvariescontinuously varyamong species?i.e., do specieswithmass-flowering
betweentwo extremepatterns(Janzen 1971, Heinrich individualshave a higherdegreeofsynchrony thanspecies
and Raven 1972, Gentry1974). At one extremeare withsteady-state individuals?2) Within a species,does
specieswith"mass-flowering" individualsproducinglarge an individual'sfruitset correlatebestwithaspectsof the
numbersof new flowerseach day overa week or less. At individual'sphenologyor thoseof the population?
the oppositeextremeare specieswith "steady-state"in- Shrubswerechosenbecauseoftheease ofquantifying
dividualsproducingsmallnumbersofnew flowers almost daily flowerproductionand determining fruitset. The
dailyformanyweeks. speciesrepresentlargedifferences in floweringseasonand
The population'sflowering patternis definedby its durationof flowering.The followingsix species were
componentindividuals.Variabilityamong individualsin studied:Hybanthus prunifolius(Schult.) Schulze (Viola-
theirresponseto a flowering cue determines thesynchrony ceae), TurnerapanamensisUrban (Turneraceae),Rinorea
of the initiationday and consequently population sylvatica(Seem.) 0. Kuntze (Violaceae), Psychotria
affects hor-
izontalisSw. (Rubiaceae), Erythrina costaricensis
Micheli
var. panamensis(Standl), comb. nov. (Fabaceae), and
I Received6 December 1982, accepted7 April 1983. Pentagoniamacrophylla Benth.(Rubiaceae).

BIOTROPICA 15(4): 257-267 1983 257

This content downloaded from 142.51.1.212 on Mon, 8 Apr 2013 20:21:06 PM

All use subject to JSTOR Terms and Conditions
TABLE 1. Comparisons ofsix shrubspeciesof the individual'sphenology
and pollinationleveland thepopulation'sphenology
and
spatial pattern.Mean values (? 1 standarddeviation)are given;N = 20 individualsperpopulation(exceptN = 10
individualsforfruitset ofErythrina costaricensis
var. panamensis).

Erythrina
costari-
Hybanthus Turnera Rinorea Psychotria censisvar. Pentagonia
prunifolius panamensis sylvatica horizontalis panamensis macrophylla
Individuallevel
Phenology
Floweringpattern Mass- Mass- Mass- Intermediate Steady- Steady-
flowering flowering flowering state state
Total numberof flowers 296 (249) 343 (333) 784 (638) 2995 (3054) 434 (254) 98 (90)
Lifespanof flower(days) 1 1 4 1 1 2
Durationof flowerproduction4.65 (.99) 7.70 (1.13) 8.10 (.64) 20.15 (3.34) 44.65 (12.05) 58.75 (17.22)
(days)a
Duration:90% of flower 2.35 (.49) 4.60 (.82) 7.05 (.39) 14.55 (2.50) 29.60 (9.88) 49.40 (16.08)
production(days)a
Synchrony withconspecificsb0.89 (.14) 0.77 (.11) 0.95 (.05) 0.82 (.07) 0.50 (.09) 0.48 (.12)
Pollination
Major visitorsof flowers Social bees Bees & wasps Bees & wasps Bees & wasps Humming- Humming-
Butterflies birds birds
Humming-
birds
Number of fruitsset 253 (229) 91 (69) 107 (91) 451 (578) 52 (43) 36 (28)
% of flowerssettingfruit 82 (13) 31 (16) 14 (8) 13 (9) 15 (10) 33 (28)
Populationlevel
Phenology
Durationof flowerproduction7 12 10 27 80 94
(days)a
Synchrony of firstdayc 0.77 0.92 0.40 2.22 11.82 16.16
Synchrony of peak (median) 0.43 1.13 0.60 2.48 11.45 13.59
dayd
Synchrony of flowering 0.89 0.77 0.95 0.82 0.50 0.48
periode
Spatial density High Medium-high High Medium-high Low Low
a Includesdays withno flowers.
b See Appendix 1A formethodof calculation.
c Calculatedas 1 SD aroundthe mean of the firstflowering day of 20 individuals;highervalues indicatelowersynchrony.
d Calculated as 1 SD around the mean of the peak (median) flowering day of 20 individuals;highervalues indicatelower
synchrony. For the last threespecieswithno clearpeak day (maximumnumberof flowers),the day the median flowerwas open
was used.
e See Appendix 1B formethodof calculation.

Each individual'sfloweringphenologywas quantified

METHODS
The studywas conductedin 1976 in thesemi-deciduous
lowlandforeston BarroColoradoIsland (BCI), Panama.
This well-studiedbiologicalpreserveis describedin detail
elsewhere(Leigh et al. 1982). Croat (1978) provides
descriptionsof the six speciesused in thisstudy.
Beforeflowering began,20 individualsofeach species
wereselected.For fourspecieswithmediumto highden-
sity(Table 1), individualswereselectedwithinone hect-
are;individualsweredumped withineach of thesespecies.
For two specieswithlow density(Table 1), severalhect-
areswererequiredto locate20 individuals.
by countingdaily the numberof newlyopened flowers.
Individualflowerson one individualof each specieswere
taggedto determineflowerlifespan.Summariesof these
observationsprovideddaily and total flowerproduction
and durationof flowering of each individual.
For comparisons amongspeciestwo additionalvalues
were calculated: 1) the day on which each individual's
median flowerwas open; and 2) the numberof days
aroundthatmedianday thatencompassed90 percentof
total flowerproduction.Using the 90 percentlevel of
flowerproductionwas an arbitrary decisionthat elimi-
natedthe tailsof the floweringperiods,i.e., days of very

258 Augspurger

This content downloaded from 142.51.1.212 on Mon, 8 Apr 2013 20:21:06 PM

All use subject to JSTOR Terms and Conditions
 low flowerproduction.All subsequentanalyseswerere- Psychotria horizontalis,eightand nine weeks afterflow-
strictedto days encompassing90 percentof flowerpro- ering,respectively; neitherof the lattertwo speciespro-
duction. vided earlyindicationof pollination.
Floweringsynchrony, hereafterreferredto as syn- In Method 2, applied to Turnerapanamensisand
chrony,was quantifiedfromthe perspectiveof both the Erythrinacostaricensis var. panamensis,aborted ovaries
individualand the population.Individualsynchrony is a on the groundwere counteddaily and removed.These
compositemeasureof the amountof overlapof a given specieshad beenexperimentally shownto consistently abort
individual'sflowering days with thoseof all individuals non-pollinatedflowers,but not pollinatedflowers,two
in the definedpopulation;the methodof calculatingin- daysaftereach flower'sfunctional1-daylifespan.By sub-
dividualsynchrony is presentedin Appendix 1A. Popu- tractingnumberof freshly abortedovariesfromnumber
lation synchrony is a compositemeasureof the amount of open flowerstwo days priorto groundcollection,an
of overlapof all flowering days of all individualswith indirectestimatewas obtained of the numberof non-
each other individualin the definedpopulation. The aborted,and assumed pollinated,ovariesremainingon
methodof calculatingpopulationsynchrony is shownin the plant. T. panamensisand E. costaricensis var. pana-
Appendix 1B; populationsynchrony equals the mean of mensisare tall shrubs;thus Method 2 was considered
the values of individualsynchrony as derivedfromthe superiorto a directcount of swollen ovariesfromthe
methodin Appendix1A. The methodof measuringpop- vantagepointof the ground.
ulationsynchrony summarizesthe temporaloverlapbe- Measurements of fruitset occurredeithertwo or five
tweeneach set of two individualsin thepopulationfrom days afterthe flowerwas open in Hybanthus prunifolius,
theperspective of bothindividuals.This measureof pop- Rinoreasylvatica,Turnerapanamensis,and Erythrina cos-
ulationsynchrony is a modification of Primack'smethod taricensis var.panamensis.By thatday, all flowersexper-
(1980), whichsummarizedoverlapfromthe perspective imentallyexcludedfrompollinatorshad aborted,while
of onlyone of the two individuals,viz. the one withthe all experimentally pollinatedovariesremainedon theplant.
shortestflowering period.Neitherof the synchrony mea- Counts made so quicklyafterfloweringwere likelyto
sures takes into accountdifferences among days in the indicateonlywhetheror not pollen was receivedby the
numberofopen flowers; rathertheysimplymeasureover- flower;abortionof experimentally pollinatedovaries,due
lap on all daysan individualhas at leastone open flower. to unknowncauses such as predationor resourcelimits,
Two additionalaspectsof populationsynchrony were occurredonlyafterthisperiod.In Pentagoniamacrophylla
measured.First,synchrony of the firstday of flowering and Psychotria horizontalis,measurements of fruitsetwere
was calculatedas one standarddeviationaroundthemean made weeks afterpollination.Therefore,theirfruitset
of the firstflowering day of 20 individuals.Second,syn- could notbe used to indicatepollinationonly,as selective
chronyof the medianday of flowering was calculatedas abortionand predationof pollinatedovariesmay have
one standarddeviationaroundthe mean of the median occurredpriorto measuringfruitset.
flowering day of 20 individuals.A low standarddeviation The timingand methodof determining fruitset,and
indicatesa highlevel of synchrony. whatfruitsetrepresented, differedamongthesix species.
Percentof flowerssettingfruitand absolutenumber Therefore, interspecificcomparisonsof fruitset wereun-
of fruitsset weredeterminedforeach of 20 individuals warranted; onlyintraspecific comparisons of fruitsetwere
in fivespecies;thesemeasureswerelimitedto ten of 20 made.
individuals in Erythrina costaricensis
var.panamensis.Fruit
setmeasuredthepistillatefunction of thehermaphroditic
flowersand indicatedthat fertilization occurred.This
RESULTS
measurewas the best estimateof the numberof flowers Phenologyvariedwidelyamong the six speciesat both
receiving pollen in sufficient quantityto effectfruitset. the individualand populationlevels (Fig. 1, Table 1).
Two methodswere used to determinefruitset. In The lifespanof a flowervariedfrom1-4 daysamongthe
Method 1, swollenovarieswere counteddirectlyon the species(Table 1). The totalnumberof flowersper indi-
plant.Ovariesof some speciesare knownto swellslightly vidual did not differsignificantly among fiveof the six
withoutpollination.Therefore, experiments werealso car- species(Tables 1 and 2). However,the six speciesvaried
ried out on flowersemasculatedand bagged to prevent significantly in how long the individualproducedthose
pollination.A comparisonof theirovarysize withthatof flowers (Tables 1 and 2). The sixspeciesrepresented three
flowersartificiallypollinatedindicatedthe minimumsize phenologicalpatterns.Three specieshad "mass-flower-
of ovaryrequiredto indicatethat pollinationhad oc- ing" individuals;theirday of medianflowerproduction
curred.This directmethodof determining fruitset was coincidedwiththe day of peak flowerproduction.Two
appliedto individually taggedflowers of Hybanthus prun- specieshad "steady-state"individualswith no marked
ifoliusand Rinoreasylvatica,fivedaysaftereach flower's peak of flowerproduction;individualshad similarnum-
functionallifespan,and to Pentagoniamacrophylla and bersof flowersformanydaysbeforeand aftertheday on
Comparative ofNeotropical
Phenology Shrubs 259

This content downloaded from 142.51.1.212 on Mon, 8 Apr 2013 20:21:06 PM

All use subject to JSTOR Terms and Conditions
TABLE 2. Comparisons of the individual'sphenologicalvariablesamongthe six species.Underlinedvalues indicatemeansnot
significantly
different(P < .05) byStudent-Newman-Kuels MultipleRangetest.Abbreviations refer
tofirstletterofgenus
and speciesofeach species.Speciesare arrangedin orderfromlowest(left)to highestvalues.

Analysisof variance Student-Newman-Keuls

Variable F df Signif. MultipleRange test
Total numberof flowers 14.19 5, 114 P < .0005 Pm Hp Tp Ec Rs Ph
Duration:90% of flowers 110.70 5, 114 P < .0005 Hp Tp Rs Ph Ec Pm
withconspecificsa
Synchrony 79.54 5, 114 P < .0005 Pm Ec . Tp Ph Hp Rs

a
See Appendix 1A formethodof calculation.

whichthemedianflowerwas produced.One specieshad tionanalysisrevealedthatthenumberof flowers, but not

individualswithan intermediate pattern.Finally,the six thedurationofflowering, was significantly
correlated
with
speciesvariedsignificantly in the individual'ssynchrony the numberof fruitsset in fiveof the six species(Table
withconspecifics (Tables 1 and 2; Fig. 1). The six species 4). Thus theseanalysesrevealthatone variablealone,the
fellintothreecategories:thosewitha high,medium,and individual'sflowernumber,was the best predictorof
low degreeof synchrony (Tables 1 and 2). numberof fruitsset in all species,exceptErythrina cos-
Amongthe speciesthe mean flowernumberwas not taricensisvar. panamensis.Variationin the individual's
significantly correlatedwith the mean durationof flow- phenology,and not in the population'sphenology,ex-
eringor populationsynchrony (Table 3). All otherphe- plainedthevariationin the individual'spollination.
nologicalcharacters weresignificantly correlated(Table 3; Identicalanalysesas above wereperformedforeach
Fig. 1). Mean durationof individualflowerproduction species,usingfirst, theproportion of flowerssettingfruit,
significantly correlatedwithpopulationsynchrony of the and second,the arcsintransformation of proportionfruit
first and peak (median)daysofflowering and oftheentire set as the dependentvariable.These analysesyieldedno
flowering period;specieswithshorter individualflowering significant regressions.Furthermore, simple correlation
periodshad a higherdegreeofsynchrony. For speciesthat analyses,in general,showed no significant relationships
initiatedfloweringmore synchronously, peak (median) (Table 4). Only Turnerapanamensisshowed a negative
flowering also tendedto occur more synchronously. Fi- relationbetweenboth theindividual'snumberof flowers
nally,a highdegreeof synchrony forthe entireflowering and its synchrony with conspecifics with the proportion
periodresultedamongspecieswitha highdegreeof syn- of its flowerssettingfruit(Table 4). Also, in threeof six
chronyforfirstand peak (median) days of flowering. species,individualswithlongerflowering periodshad less
To evaluatethe relativeinfluenceof the individual's synchrony withconspecifics(Table 4).
and the population'sphenologyon the individual'spol-
lination,a multipleregression analysiswas performed for
each species.In the firstset of analyses,numberof fruits
DISCUSSION
set was the dependentvariableand flowernumber,90 Clearphenologicalpatternsemergedwhendetailedquan-
percentdurationof flowering, and individualsynchrony titativestudiesat the individualand populationlevels
were the independentvariables.When the population wereundertaken.The six shrubspeciesof thisstudydif-
variable (synchrony) and both the individualvariables feredin threemajorcomponentsof the individual'sphe-
(flowernumber and duration) were all included, the nology:numberof flowers,durationof flowering, and
regression was clearlysignificant forHybanthus prunifol- synchrony withconspecifics.In the discussionI consider
ius, Turnerapanamensis,Rinorea sylvatica,Psychotria possibleconsequencesof and explanationsforsuch pat-
horizontalis, and Pentagoniamacrophylla (P < 0.05). It terns,and wherepossible,make comparisonswith the
was not significant forErythtinacostaricensis var. pana- patternsin othertropicalforests.
mensis;in measuringfruitset in thisspecies,N = 10 in- Historically,researchin phenologyhas been lacking
dividuals,whilein all otherspecies,N = 20 individuals. in two resp&ts: quantification of phenologicalpatterns
Next, whensynchrony was eliminated,theregression and preciseuse of dearlydefineddescriptive terms.Con-
was stillsignificantand therewas no significant reduction sequently,adequate descriptions of temporalpatternsof
in the varianceexplainedby the regression. When dura- flowering arelackingformosttropicalspecies.The timing
tionof flowering was eliminated,the regression was still of the individualand the populationare oftennot dis-
significantand again therewas no significant reductionin criminated. For example,"gregarious"is frequently used
the varianceexplainedby the regression. Simple correla- to imply"synchronous," especiallyiftheflowering period
260 Augspurger

This content downloaded from 142.51.1.212 on Mon, 8 Apr 2013 20:21:06 PM

All use subject to JSTOR Terms and Conditions
 Hybanthus Turnera Rinorea Psycho/rna
prun/fo//us panamensis syl votica horizon/a/is
20 - ___

15 -

5>- -4 <-*>>- _

20 -<

$ << ~~~-4.> - _ _ _ _ _ _

1~~L L L 1 1 1 1<*11 <I*III

00 10 10 0 10 0 70 20 30
15- . Erey/hr/na
cos/ar/censis
20 -< .

01 - -4-

5 -_>

o 10 20 30 40 50 60 70 80
t~4-
5 _____ <___(.>
z 0 N

<-4 * > -

Y ~ ~~ ~<N.M~ ~ ~ ~A >B_R

,I I I , I , , , , I I

0 10 20 30 40 50 60 70 80 9

DnAY NUMBE

FIGURE 1. Timingof flowering of individualsarrangedin rankorderof the day the medianflowerwas open (N= 20 foreach
ofsix species).Each horizontallinerepresents by an individual.A brokenlineindicatesdiscontinuous
thetotaldurationof flowering
flowerproduction.The bracketsenclosethe period aroundthe median flowerday duringwhich 90%oof an individual'sflowers
wereopen. The heavydot indicatesthe day the medianflowerwas open; thisday corresponds to thepeak day (maximumnumber
of flowers)formass-flowering individualsof the firstthreespecies.

PhenoloayofNeotropical
Comparative Shrubs 261

This content downloaded from 142.51.1.212 on Mon, 8 Apr 2013 20:21:06 PM

All use subject to JSTOR Terms and Conditions
TABLE 3. Pearsoncorrelation ofthesix species.Analysesare at thepopulationlevel;N = six
characters
forphenological
coefficients
species.Correlation werecalculatedfromTable 1 usingmean valuesfor numberofflowersand durationof
coefficients
flowerproductionunderindividualleveland valuesforsynchrony
underpopulationlevel.

Total Duration: Population Population Population

numberof 90% of synchrony of synchrony of synchrony of
flowers flowers firstday peak day flowerperiod
Total numberof flowers
Duration:90% of flowers -.198
Populationsynchronyof firstdaya -.334 .976***
Populationsynchronyof peak daya -.302 .964** .995***
Populationsynchronyof flowering periodb .328 -.891* - 949**c -.962**c
a Calculatedas 1 SD aroundthe mean.
b See Appendix 1B formethodof calculation.
c The correlationis negativebecause of differences
in the methodsused to calculatesynchronyvalues. For synchronyof firstand
peak days,a low standarddeviationindicatesa high degreeof synchrony.In contrast,forsynchronyof the entireflowering
period,
high values indicatehigh synchrony.
* P < .05, ** P < .005, *** P < .0005.

is short(Holttum1935, LongmanandJenik1974). With in thetropics.Quantitativedata to testthatgeneralization

thatdefinition,it is unclearwhetherthe shortflowering
period refersto the individualand/or the population.
Only when populationsynchrony is perfectare the du-
rationsof the individual and the population periods
equivalent.
Baker(1959) suggestedfromqualitativeobservations
thatasynchronous floweringof thepopulationis common
have beenlacking.Frankieet a/. (1974) listed13 percent
of treespeciesin thewet lowlandrainforest of Costa Rica
as "unsynchronized" in theirpopulationflowering, but
"unsynchronized" was not definedin that study. The
quantitativedata fromthe presentstudyindicatedthat
the shrubspeciesfell along a continuumof population
synchrony.All weresynchronous to a degree;theysimply

TABLE 4. Pearsoncorrelation forvariousphenological

coefficients characters fruitsetand numberoffruitsset. Analyses
and percent
are at the individuallevelfor each species;N = 20 individuals,exceptN = 10 individualsforfruitset of Erythrina
costaricensisvar. panamensis.

Erythrina
costari-
censis
Hybanthus Turnera Rinorea Psychotria var. Pentagonia
Variables prunifolius panamensis sylvatica horizontalis panamensis macrophylla
Total numberof flowers-
% fruitset .207 -.454* -.054 .263 -.128 -.149
Duration:90% of flowers-
% fruitset .290 .207 .043 .194 .135 .197
Synchrony withconspecificsa-
% fruitset -.283 -.472* .096 .146 .368 .313
Total numberof flowers-
totalnumberof fruit .991*** .708** .652** .875*** .513 .510*
Duration:90% of flowers-
totalnumberof fruit .344 -.268 .167 .241 .401 .295
Synchrony withconspecificsa-
totalnumberof fruit -.393 .396 .089 .081 -.152 .240
Total numberof flowers-
duration:90% of flowers .325 -.209 .335 .440 .398 .247
Total numberof flowers-
synchrony withconspecificsa -.376 .383 -.210 -.115 -.072 .031
Duration:90% of flowers-
synchrony with conspecificsa -.994*** -.629** -.108 -.732** -.090 .134
a
See Appendix 1A formethodof calculatingindividualsynchrony.
* P < .05, ** P < .005, *** P < .0005.

262 Augspurger

This content downloaded from 142.51.1.212 on Mon, 8 Apr 2013 20:21:06 PM

All use subject to JSTOR Terms and Conditions
variedin theextentof theirsynchrony. This resultimplies (Augspurger1980). Hence,attraction and pollinationare
that the dichotomy"synchronous"versus "asynchro- not necessarily equivalentforthesespecies.
nous" may not alwaysbe appropriate. The abundantresourceused to attractpollinatorsto
The mean numberof flowersper individualdid not mass-flowering individualshas the potentialto produce
differsignificantlyamong fiveof the six species;a high low inter-plant movement.This would resultin low out-
variationin flowernumberoccurredamong individuals crossing, particularlyforself-compatible species(e.g., Hy-
withineach species,however.Specieswithmass-flowering banthusprunifolius,see Augspurger1980), or in low
and steady-stateindividualshad roughlyequivalentnum- percentagefruitset for self-incompatible species (e.g.,
bers of flowers.Therefore,the major difference in their Turnerapanamensisis heterostylous and self-incompati-
phenologicalpatternwas in thedurationof flowering and ble; personalobservation).In Hybanthus prunifolius de-
flowerproductionper day. tailedobservationsindicatedthatbees visitedonlyabout
Mean durationof the individual'sflowering period 5 percentof availableflowerson an individualand then
rangedamongthesix speciesfrom4.6-58.8 days(2-49 moved, frequentlyto other conspecifics(Augspurger
days for90% of flowerproduction).Shortflowering pe- 1980). Inter-plantmovementbetween mass-flowering
riods occurredin the dry season, long ones in the wet treeshas also been observed(Frankieet al. 1976). This
season.The onlyextensivecommunity studyof the phe- movementmaybe promotedbyseveralfactors:aggressive
nologyof tropicalshrubsand treeletsoccurredin dryand interactionsamong pollinators(Frankie et al. 1976,
wet forestsof Costa Rica (Opler et al. 1980a). In that Rausherand Fowler1979), presenceof toxiccompounds
study,individualsin treespeciesin the dryforestpro- in nectar(Rhoades and Bergdahl 1981), or variability
ducedflowers forshorterperiods(mean= 7.9 weeks)than among individualsin timing,amount,and composition
did individualsin treeletand shrubspecies(mean = 11.7 of nectarsecretions (Frankieand Haber 1983). Bees con-
weeks).The bloomingperiodvariedby habitat,however; tinuallymonitorchangingresourcelevels; detectionof
shrubsin the dryhill foresthad an individualaverageof thesechangesapparently promptslargebee pollinatorsto
4.7 weeks,while thosein drysecondaryforestflowered make inter-tree movements,therebyeffecting cross-polli-
for 19.8 weeks. Opler et al. (1980b) noted that early nation.
successionalspeciesgenerallyhave moreextendedflower Mass-floweringmayalso lowerflower predation(Bawa
periodsthan those in maturecommunities.Continuous 1983) or, in later reproductivestages, seed predation
and extendedflowerproductionby treeletand shrubin- (Augspurger1981). The shortflowering periodmay im-
dividualswas greaterin wet forest(28% of the species) pose some riskif it occursduringunfavorableweather
and riparianforest(30%) than in dryhill forest(17%) conditionsforpollinationor forsubsequentreproductive
(Opler et al. 1980a). In a moreaseasonalforestin Ma- stages.The shortflowering by individualsof Hybanthus
laysia,Putz (1979) notedthat66 percentof speciesflow- prunifolius was followedin one yearby an extremedry
eredneitherforlong periodsnor at regularintervals;the periodwhenmanyoftherapidlydevelopedfruitsaborted
durationof flowering was highlyvariablewithina tree due to drought(Augspurger1978).
species. In contrastto mass-flowering, otherspecieshad in-
Bawa (1983) outlinedpossibleselectiveforcesaffect- dividuals with steady-statefloweringproducinga few
ing the lengthof the individual'sflowering period. Of flowersregularlyformanydays. They werevisitedby a
particularimportanceare: regulationof pollen flow,for- low diversity of hummingbird species,predominantly non-
aging behaviorof pollinators,rate of fruitdevelopment territorial foragers,althoughsome territoriality was ob-
in responseto resourceavailability,and habitat(as ob- servedat largeindividualsof Erythrina costaricensisvar.
servedabove). The lengthmay also be affectedby the panamensis.This group of pollinatorsare knownto reg-
availabilityof an environmental cue to synchronize an ularlyvisitwidelyspacedindividualplants(Stiles 1975).
individual'sflowering withconspecifics (see below). The individual'slow flowerproductionper day promotes
The shortperiod of mass-flowering individualsmay cross-pollination and perhapsincreasesthenumberof dif-
be a responseto competition forpollinators:theirsuper- ferentindividualswith whicha plant mates. The latter
abundantresourcesare thoughtto attractmanytypesof dependson theregularity of thepollinator's foragingpath.
opportunistic pollinatorswithdensity-dependent foraging Extendedflowering lessensthe risk of flowering during
behaviors(Janzen1967, Heinrichand Raven 1972, Gen- sporadicbad weather(Bawa 1983). It may also allow
try1974, Frankieet al. 1976). My observations showed the regulationof the relativenumberof flowersversus
that many different foraging speciesvisited each of the fruitsin responseto changingavailabilityof resources
threespecieswith mass-flowering individuals.However, (Lloyd 1980).
a morethoroughstudyof Hybanthus prunifolius showed The six shrubspeciesrangedin the populationsyn-
that most of the 22 speciesof visitorswere ineffectual chronyfrom.48-.95 on a scale of 0-1. The population
pollinatorsand thatone social bee effected mostfruitset synchrony of the firstday, peak (median) day, and the

Shrubs
ofNeotropical
Phenology
Comparative 263

This content downloaded from 142.51.1.212 on Mon, 8 Apr 2013 20:21:06 PM

All use subject to JSTOR Terms and Conditions
 entireflowering periodwerehighlycorrelated.This sug- flowering occurredin responseto drought.Corner(1952)
gests that variabilityamong individualsin theirinitial and Holttum(1953) suggestedit was associatedwitha
responseto the flowering cue subsequentlyaffectssyn- decreasein temperature accompanying storms.
chronythroughoutthe flowering period. In threetem- Fourth,populationsynchrony was also enhancedby
perateshrubssynchrony varied from.34-.74 (Primack theproductionof well-developedpreformed flowerbuds
1980) (but note differences betweenthisstudyand Pri- thatgo dormantand await a flowering cue (Rutgersand
mack'sin methodof quantifying synchrony; see methods Went 1915, Seifriz1923). The time fromcue to flow-
and Appendix1). Putz (1979) notedthatabout 50 per- eringwas quiteshort,thusleadingto lowvariability among
centofMalaysiantreespecieslackedintra-population syn- individualsin developmentalratesand high population
chrony.Opler et al. (1980a) observedthat in the wet synchrony. Hybanthus prunifoliusand Rinoreasylvatica
forestof Costa Rica, populationasynchrony was more formedpreformed flowerbuds; Turnerapanamensisap-
commonin shrubsand treelets(33% of species)thanin pearedto do so facultatively. The bud of Rinoreasylva-
trees(13%). In thewet forest,in general,synchrony was tica mostcloselyapproachedthe size of its open flower;
notalwaysmarked.In thedryforest, slightly moreofthe thatspecieshad the highestpopulationsynchrony of the
"brief" (less than 2 weeks per individual)specieswere six species.
synchronous (85%) than the extended(greaterthan 2 The two specieswithlow populationsynchrony, Er-
weeks) species(79%). Frankieet al. (1974) foundthat ythrinacostaricensis var.panamensisand Pentagoniamac-
speciesof treesweremoresynchronous in the drythanin rophylla,sharedcommonfeaturesthat are oppositethe
the wet forest. above threespecies.Theirindividualsexhibiteda steady-
In the currentstudy,highvalues of populationsyn- stateflowering phenologyin responseto an unknowncue
chronywere associatedconsistently with fourvariables. in the wet season. They did not formwell-developed,
First,populationsynchrony increasedamong species as dormantflowerbuds and theyoccurredin quitelow den-
themeanindividualflowering durationdecreased.Species sity.Similarasynchrony of flowering,as well as leaf fall
withmass-flowering individualshad nearlytwiceas high and shoot emergence,was observedin populationsof
synchrony values as specieswithsteady-state individuals. Erythrina peoppigianain Costa Rica (Borchert1980); an
This quantitative conclusionconfirmed manyearlierqual- increasein its flowering synchrony occurredwithincreas-
itativeobservations. Specieswithmass-flowering individ- ing drought.
uals were commonlyobservedto have high population Finally, the sixth species, Psychotriahorizontalis,
synchrony (Muller 1882, Schimper1903, Spruce 1908, sharedfeatureswithbothgroupsof species.This species
Coster 1926, Kerling 1941, Janzen 1967, Opler et al. may flowerin the late dryor earlywet season or both
1976). followingheavy rains. It facultatively made preformed
Second,highpopulationsynchrony was associatedwith flowerbuds in responseto weak rainsin the dryseason.
high population density.Thus, Hybanthusprunifolius, The population'sflowering synchrony was fairly high,but
Turnerapanamensis,and Rinoreasylvaticahad a very its daily flowerproductionwas relativelylow and was
high, but ephemeralflowerdensityin the forestas a spreadoveran intermediate lengthof time.
whole. This flowerdensityaccrued fromthe flowering Severalhypotheses existto explainselectionfora high
patternof both the individualand the population,and degreeof synchrony withconspecifics: 1) increasein the
fromthe spatialdistribution of thepopulation. potentialforcross-pollination, a requirement if the indi-
Third,thespecieswithhighephemeralflowerdensity vidual is self-incompatible (Holttum 1953, 1954; Baker
also shareda commonand unambiguousflowering cue. 1959; Bawa 1983); increasein theefficiency of energetics
The specieswithmass-flowering individualsall flowered of pollinators(Heinrichand Raven 1972); attraction of
in the dryseason,a few days aftera heavyrain caused pollinators(Corner 1952, Beattie et al. 1973, Gentry
sudden and large changesin soil moisture.The specific 1974, Opler et al. 1976, Bawa 1983); and escape from
cue requirements have been identifiedexperimentally for flowerand later seed predators(Beattie et al. 1973).
Hybanthus prunifolius.That shrubrequiresfirst a drought Experimental testsof thesehypotheses are limitedto the
of 1-2 months,afterwhichit flowers rapidlyaftersudden last two. In Hybanthusprunifolius,both pollinatorat-
changesin soil moistureassociatedwith rainfall(Aug- tractionand seed predatoravoidancewereenhancedin a
spurger1982). Similarly,aseasonal showersduringthe naturallysynchronous populationrelativeto an experi-
dryseason in Costa Rica have been shown to correlate mentally-induced asynchronous population(Augspurger
withthe onsetof anthesisby severalspecieswithshort, 1981).
highlysynchronized flowering (Opler et al. 1976). Poore Selectionforsynchrony has apparently beenlessstrong
(1968) and Medway (1972) also observedthe impor- withinspecieswithindividualswith steady-state flower-
tanceof externalcues in affecting synchrony in the more ing. For long-flowering individuals,being out of phase
aseasonal tropicalforestsof the Old World; gregarious by a fewdays withconspecifics, especiallythoseat some

264 Augspurger

This content downloaded from 142.51.1.212 on Mon, 8 Apr 2013 20:21:06 PM

All use subject to JSTOR Terms and Conditions
distancein theserarespecies,may have littleconsequence ical patternsof tropicaltreesis not the resultof biotic
forcross-pollination and fruitset,relativeto a mass-flow- factorsalone, but is also influencedby treearchitecture
eringindividual.In the lattergroup,beingasynchronous and vegetative, bud, and flowerdevelopment.
to the same degreewould have a major effect,gready Withina species,variancein onlyone aspectof the
loweringthepopulationdensityat whichit floweredand individual'sphenology,flowernumber,was usefulin ex-
the numberof nearbyindividualswith which it could plainingthe varianceamong individualsin the number
mate.Extremely highdensitiesof flowersat one timecan of fruitsset. Durationof flowering and synchrony with
not be achievedby a populationof steady-state individ- conspecificswerenotsignificant variables.The proportion
uals. In thesespecies,the individualby itselfis not using of fruitsetby theindividualwas notsignificantly affected
high densityto attracta density-dependent pollinator. byanyofthethreevariables.Theseresultsdo notindicate
Furthermore, littledensityeffect is galned fromoverlap- thatdurationand synchrony per se are of no importance
ping in timepreciselywiththe population. to the individual.Rather,theyillustratethat,at the ex-
It is impossibleto know todaythe orderin whicha istinglevel of variationamong individualsin the popu-
species'characteristicsbecamelinkedin thepast. Selection lation,in conjunctionwith all othervariablesaffecting
on bud formation, season of flowering, typeof flowering fruitset,theyare not predictiveof levelsof fruitset. To
cue, lengthof flowering, patternof flowerproduction, testdirectlythe effectof synchrony on the individual's
and synchrony with conspecifics is likelyto be dosely fruitset requiresexperimental manipulationof the indi-
intertwined. The differences in synchrony amongthespecies vidual'sflowering timeawayfromthatofconspecifics. At
may be due simplyto selectionon bud formation, the such extremesin variationin timing,the importanceof
flowering cue, and season of flowering. These factorsre- synchrony may be more readilydetected,as was dem-
late directlyto the mechanisman individualuses to time onstratedin the earlierstudiesof Hybanthus prunifolius
itsflowering; indirecdy theyaffect itssynchrony withcon- (Augspurger198 1).
specifics.Alternatively, selectiondirectlyon the individ-
ual's synchrony, due to pollinatorattractionand move-
ACKNOWLEDGMENTS
mentand the importanceof cross-pollination, may vary
among the speciesand explainthe observedvariationin I thankP. Amman,T. Beliz,N. Garwood,C. Gyllenhaal-Davis,
populationsynchrony. These evolutionary factorsdirectly D. Janos,and L. McHargue forfieldassistance,and K. Bawa,
G. Frankie,K. Hogan, D. Janzen,M. Melampy,and D. Ra-
regulatepopulationsynchrony affecting binowitz
by differentially for constructivecommentson earlierdraftsof the
the reproductivefitnessof its componentindividuals. manuscript.The researchwas fundedby a SmithsonianInsti-
Borchert(1983) arguedthatthe evolutionof phenolog- tutionPredoctoralFellowship.

LITERATURE CITED
AUGSPURGER, C. K. 1978. Reproductiveconsequencesof flowering synchronyin Hybanthusprunifolius (Violaceae) and other
shrubspeciesof Panama. Ph.D. Thesis. University of Michigan,Ann Arbor.
1980. Mass-flowering of a tropicalshrub (Hybanthusprunifolius): influenceon pollinatorattractionand movement.
Evolution34: 475-488.
1981. Reproductivesynchrony of tropicalplants:experimental effectsof pollinatorsand seed predatorson Hybantbus
prunifolius (Violaceae). Ecology62: 775-788.
1982. A cue forsynchronous flowering.In E. G. Leigh,Jr.,A. S. Rand, and D. M. Windsor (Eds.). The ecologyof
a neotropicalforest:Seasonalrhythms and longer-term changes,pp. 133-150. SmithsonianInstitution Press,Washington,
D.C.
BAKER, H. G. 1959. Reproductive methodsas factorsin speciationin flowering plants.Cold SpringsHarborSymp.Quant. Biol.
24: 177-199.
BAWA, K. S. 1977. The reproductive biologyof Cupania guatemalensis Radlk. (Sapindaceae). Evolution31: 52-63.
1983. Patternsof floweringin tropicalplants. In C. E. Jones and R. J. Little(Eds.). Handbook of experimental
pollinationbiology,pp. 394-410. Van NostrandReinhold,New York.
BEATTIE, A. J., D. E. BREEDLOVE, AND P. R. EHRLICH. 1973. The ecologyof the pollinatorsand predatorsof Fraseraspeciosa.
Ecology54: 81-91.
BORCHERT, R. 1980. Phenologyand ecophysiology poeppigiana0. F. Cook. Ecology61: 1065-1074.
of tropicaltrees:Erythrina
1983. Phenologyand controlof flowering in tropicaltrees.Biotropica15: 81-89.
BULLOCK, S. H., AND K. S. BAWA. 1981. Sexual dimorphismand the annual flowering patternin Jacaratia dolichaula (D.
Smith) Woodson (Caricaceae)in a Costa Rican rainforest.Ecology62: 1494-1504.
, J. H. BEACH, AND K. S. BAWA. 1983. Episodic flowering and sexual dimorphismin Guarea rhopalocarpaRadlk.
(Meliaceae) in a Costa Rican rainforest.Ecology64: 851-862.
CORNER, E. J. H. 1952. Wayside treesof Malaya, Vol. 1, Second Ed. Government PrintingOffice,Singapore.
COSTER, C. 1926. PeriodischeBliiteerscheinunger in den Tropen.Annalesdu JardinBotaniquede Buitenzorg35: 125-162.
CROAT, T. B. 1978. Flora of BarroColorado Island. Stanford UniversityPress,Stanford,California.943 pp.
ComparativePhenologyof NeotropicalShrubs 265

This content downloaded from 142.51.1.212 on Mon, 8 Apr 2013 20:21:06 PM

All use subject to JSTOR Terms and Conditions
FOSTER, R. B. 1982. The seasonalrhythmof fruitfallon BarroColorado Island. In E. G. Leigh,Jr.,A. S. Rand, and D. M.
Windsor(Eds.). The ecologyof a neotropicalforest:seasonalrhythms and longer-term changes,pp. 151-172. Smithsonian
InstitutionPress,Washington,D.C.
FRANKIE, G. W., AND W. A. HABER. 1983. Why bees move among mass-flowering neotropicaltrees.In C. E. Jonesand R. J.
Little(Eds.). Handbook of experimental pollinationbiology,pp. 361-372. Van NostrandReinhold,New York.
, H. G. BAKER, AND P. A. OPLER. 1974. Comparativephenologicalstudiesof treesin tropicalwet and dryforestsin the
lowlandsof Costa Rica. J. Ecol. 62: 881-919.
, P. A. OPLER, AND K. S. BAWA. 1976. Foragingbehaviorof solitarybees: implicationsforoutcrossing of a neotropical
foresttreespecies.J. Ecol. 64: 1049-1057.
GARWOOD, N. C. 1983. Seed germination in a seasonaltropicalforestin Panama: a communitystudy.Ecol. Monogr.53: 159-
181.
GENTRY, A. H. 1974. Floweringphenologyand diversity in tropicalBignoniaceae.Biotropica6: 64-68.
HEINRICH, B., AND P. H. RAVEN. 1972. Energeticsand pollinationecology.Science 176: 597-602.
HOLTTUM,R. E. 1935. The flowering of Tembusu trees(Tagraea fragrans)in Singapore1928-35. Gardens' Bulletin.Straits
Settlements. Singapore8: 73-78.
1953. Evolutionary trendsin an equatorialclimate.Soc. Exp. Biol. Symp. 7: 159-173.
1954. Plantlifein Malaya. Longman,London.
JANZEN, D. H. 1967. Synchronization of sexual reproduction of treeswithinthe dryseason in CentralAmerica.Evolution21:
620-637.
1971. Euglossinebees as long-distance pollinatorsof tropicalplants.Science 171: 203-205.
KERLING, L. C. P. 1941. The gregariousflowering of Zephyranthes roseaLindl. Annalesdu JardinBotanique de Buitenzorg51:
1-42.
KOELMEYER, K. 0. 1959. The periodicityof leaf changeand flowering in the principalforestcommunitiesof Ceylon.Ceylon
Forester4: 157-189.
LEIGH, E. G., JR., A. S. RAND, AND D. M. WINDSOR. 1982. The ecologyof a neotropicalforest:Seasonal rhythmsand longer-
termchanges.SmithsonianInstitutionPress,Washington,D.C. 946 pp.
LLOYD, D. G. 1980. Sexual strategiesin plants. II. An hypothesisof serial adjustmentof maternalinvestmentduringone
reproductive session.New Phytol.86: 69-80.
LONGMAN, K. A., AND J. JENIK. 1974. Tropicalforestand its environment. Longman,London. 196 pp.
MEDWAY, L. 1972 Phenologyof a tropicalrain forestin Malaya. Biol. J. Linn. Soc. 4: 117-146.
MULLER, F. 1882. Bemerkungen zu Hildebrand'sAbhandlungiiberdie Lebensdauerund Vegetationsweise der Pflanzen.Engler's
Jahrbiicher. Bd. II.
OPLER, P. A., G. W. FRANKIE, AND H. G. BAKER. 1976. Rainfallas a factorin the release,timing,and synchronization of
anthesisby tropicaltreesand shrubs.J. Biogeogr.3: 231-236.
) AND . 1980a. Comparativephenologicalstudiesof treeletand shrubspeciesin tropicalwet and dry
forestsin the lowlandsof Costa Rica. J. Ecol. 68: 167-188.
, H. G. BAKER, AND G. W. FRANKIE. 1980b. Plantreproductive characteristics
duringsecondarysuccessionin neotropical
lowlandforestecosystems.Biotropica12 (Supplement):40-46.
POORE, M. D. D. 1968. Studies in Malaysianrainforest. I. The foreston Triassicsedimentsin JenglesForestPreserve.J. Ecol.
56: 143-196.
PRIMACK, R. B. 1980. Phenologicalvariationwithinnaturalpopulations:Floweringin New Zealand montaneshrubs.J. Ecol.
68: 849-862.
PUTZ, F. E. 1979. Aseasonality in Malaysiantreephenology.Malay. For. 42: 1-28.
RAUSHER, M. D., AND N. L. FOWLER. 1979. Intersexual aggressionand nectardefensein Chauliognathus distinguendus(Coleoptera:
Cantharidae).Biotropica11: 96-100.
RHOADES, D. F., AND J. C. BERGDAHL. 1981. Adaptivesignificance of toxicnectar.Amer. Nat. 117: 798-803.
RUTGERS, A. L., AND F. A. WENT. 1915. PeriodischeErscheinungen bei den Bliitendes Dendrobium crumenatum Lindl. Annales
JardinBotanique de Buitenzorg29: 129-160.
SCHIMPER, A. F. W. 1903. Plant-geography upon a physiologicalbasis. ClarendonPress,Oxford.839 pp.
SEIFRIZ, W. 1923. The gregariousflowering of the orchidDendrobium crumenatum. Amer.J. Bot. 10: 32-37.
SNOW, D. W. 1965. A possibleselectivefactorin the evolutionof fruiting seasonsin tropicalforest.Oikos 15: 274-281.
SPRUCE, R. 1908. Notes of a Botaniston the Amazon and Andes. Vol. 1. In A. R. Wallace (Ed.), JohnsonRepr.,London. 518
PP.
STILES, F. G. 1975. Ecology,flowering phenology,and hummingbird pollinationof some Costa Rican Heliconiaspecies.Ecology
56: 285-301.

(modifiedfromPrimack1980).
APPENDIX 1. Methodsof CalculatingSynchrony
A. Synchrony of a given individualwithits conspecifics:
X,, the index of synchrony forindividuali, is definedas:

(n-1)(fi)
,=
synchronously,
where, e = numberof daysboth individualsi and j are flowering j#i
f = numberof daysindividuali is flowering;
n = numberof individualsin population.
266 Augspurger

This content downloaded from 142.51.1.212 on Mon, 8 Apr 2013 20:21:06 PM

All use subject to JSTOR Terms and Conditions
 When X = 1.0, perfectsynchrony occurs,i.e., all flowering
days of individuali overlapwithall flowering
days of each other
individual,j # i, in the population.
When X = 0.0, no synchrony occurs,i.e., no overlapoccursamong any of the flowering days of individuali and any other
individual,j =#i, in the population.

B. Synchrony of the population:

is definedas:
Z, the index of populationsynchrony,
n

n ,=

whereX, is synchrony
of individuali withits conspecifics
frompartA (above).

ComparativePhenologyof NeotropicalShrubs 267

This content downloaded from 142.51.1.212 on Mon, 8 Apr 2013 20:21:06 PM

All use subject to JSTOR Terms and Conditions