ECOTROPICOS, Vol. 3 (2): 77-86. 1990Socicdad Vcnczolana de Ecología

PLANT RESPONSE TO HERBIVORY:TWO EXAMPLES FROM THE NEOTROPICS

RESPUESTA DE LAS PLANTAS ALA HERBIVORIADOS EJEMPLOS DEL NEOTRÓPICO

G Wilson Fernandesl.2 and Sérvio P. Ribeiro2

IDepartment of Biological Sciences, Box 5640, Northern Arizona University, Flagstaff, AZ

86011 -U.S.A.2Present Address: Departamento de Biologia Geral, Caixa Postal 2486, Instituto de Ciencias

Biologicas, Universidade Federal de Minas Gerais, 30.161 -Belo Horizonte. MG -BRAZIL

ABSTRACT

The responses of two neotropical plant species to herbivory and frost were studied. Within-plant

comparisolls illdicated that Bricke/lia pinnifolia (Compositae) brallches rcspollded to herbivory by astem gallillg tcphritid specics and frost by producillg largcr alld morc shoots thall ullaffected brallches.

Damagc-rcsponse (mcasured in tcrms of number of shoots alld shoot lcllgth) was sigllificantly higherOll frost affected thall Oll gallcd bra11Ches, which was higher tlla11 Oll ullaffected brallches. Herbivory

by a buprestid illsect Oll tlle tlower stalk of Paepalanthus speciosus (Eriocaulaceae) sigllificalltly il1creased

productioll of compomld um\)els alld il1dividual illflorcsccl1ces. Howcver, fcwer il1dividual il1florcscellces

dcvclopcd to fruit stagc Oll eatcll plallts comparcd to lU1eatel1 plal1ts. No clear illfluel1ce of herbivorcattack was observcd Ol1 host plallt fillal hcight, or Oll overa1l COmpolU1d umbcl diameter. A secol1d

wavc of hcrbivory fo1lowed aftcr plalltS respol1ded to the first ol1e. III tl1Ís secol1d wavc, 72% of tlle

compoul1d umbels were dcstroyed and l1O furthcr plallt respol1Se was observcd. It appears tllat resourcc

mal1ipulation by tlle hcrbivore may bc involved in tl1Ís case. The adaptivc l1atllre of plallt respOl1SCto damages caused by frost alld hcrbivorv is discusscd.

"KEY WORDS: Brickellia pinnifolia, herbi..'ory, Neolropics, PaepalaJI/hus speciosus, planl compcnsalory responses.

RESUMEN

Sc cstudi6 cl cfecto de los herbívoros y Ia fornlaci6n de cscarcha en dos cspccics de plantas

neotropicalcs. La cornparaci6n entre Ias plantas Í1ldic6 que Ias rantas de Brickellia pinnifolia rcspondentanto a Ia forntaci6n de escarcha como a Ia herbivorfa por parte de un tefritfdo quc ind~lce agallascn cl tallo, prod~lcicndo rctofios nlás largos y más numerosos quc cn Ias ramas no afcctadas. La

respuesta al dafio (mcdida cn términos de número y longitud dc retofios) fué significativamente mayoren Ias ramas afectadas por Ia forlnaci6n de escarcha que en Ias que teniall agallas, Ias cuales a su

vez sufricron ll1l dafio mayor que aquellas que no fueron atacadas. EI impacto de un insecto bupréstido

en el tallo de Ias flores de Paepalanthus speciosus (Eriocaulaceae), increment6 significativamente Ia

produci6n de umbclas e Í11florescellCijs i1ldividuales. Sin embargo, en estas plantas muy pocas i1lflorcscencias

individualcs Ilcgaron a fruto cn (omparaci6n con Ias plantas que no sufrieron este ataque. No se

observ6 una clara influencia dei ataque deI herbívoro sobre el peso final de Ia olanta ni sohre f'.1

FERNANDES AND RIBEIRO

diámetro completo de Ia umbela. Una segunda ola de herbívoros Ie sigue a Ia primera \Ula vez que

Ias plantas hatl respondido al ataque. En este segwldo ataque, 72% de Ias umbelas fueron destmfdas

y no se observaron nlás respuestas de Ia planta. AI parecer, en éste ataque existe \Ula nlal1Ípulación

deI recurso por parte de Ios herbfvoros. Se discute Ia naturaleza adaptativa de Ias plantas a datios

ocasionados por Ia fornlación de escarcha y por Ia herbivoría.

PALABRAS CLA VE: Brickellia pinnifolia, herbivoría, Neotrópico, Paepalanrhus speciosus, respllestas compensatorias de plantas.

INTRODUCTION

After examining numerous studies on

the beneficial effects of herbivory , Belsky(1986) concluded that there is as yet noconvin.cing evidence that herbivory is

beneficial to plants under natural conditions.However, Paige and Whitham (1987)experimenta1ly demonstrated that undernatural fie1d conditions plants can benefitfrom being eaten. They found that after

the removal of 95% or more of the above

ground biomass of the biennial scarlet gilia,

lpomopsis aggregata, seed production andseedling survival were 2.8 times greater

than that of uneaten controls. Unbrowsed

p]ants produced only single inflorescences,whereas browsed p]ants produced multip]einf]orescences. Daspite Paige and

Whitham's (1987) clear demonstrations of

the beneficial impact of browsingherbivores on p]ants, more studies on theeffects of herbivores on p]ants in both

temperate and tropical regions are needed.

We present preliminary evidence on how

two tropical p]ant species respond to insect

herbi vory .The morphology of Brickelliapinnifolia A. Gray (Cornpositae) is

modified by gal1 formers and frost. Atephritid insect ga]] on stems of B.pinnifolia is very common in Serra doCip6, MG, Brazi]. Gal1ed branches exhibitmodifled morphology b~cause gallformation interferes with normal growth

patterns. Galled branches resembled ' 'witch-

brooms' , , with many shoots growing from

the proximal and distal portions of the gall.Gall formation is frequently cited as a

factor interfering with host plant p1p-morphologyand fitness (e.g., Dennill 1985,

1988; Fernandes 1987 and referencestherein) .Frost also damages terminal shootsand a1ters branch morphology, causingbranches to depart from their normalgrowth pattern, and the death of terminalshoots. Frost affected branches also

resembled "witch-brooms", but only the

terminal portion of the branches wereaffected .

We also censused a second plant

species, the monocot Paepalanthusspedosus (Bong) Kolin (Eriocaulaceae), thatis attacked by an unidentified species of

buprestid beetle. Adults of the beetle chew

the single flower stalk produced by thehost plant; they chew through it until theterminal compound umbel falls off the

plant. Casual field observations suggested

that chewed plants produced more

compound umbels, and individualinflorescences than uncheweli plants (Fig.

1). It appears that P. Spil-'vSUS is an annual

(T .S.M. Grandi, personal communication).In the B. pinnifolia system we asked: 1)

Does herbivory and/or frost influence shootproduction and shoot length? In the P .

speciosus system we asked the followingquestions: 1) Does herbivory influence plant

78

PLANT RESPONSE TO DAMAGE

PRIMARY

UNATTACKED ATTACKED

FIGURE 1. Schematic represelltatioll of P. speciosus plants. OllIy olle ternúllaI COmplU1d lImbeI is

produced by lU1eatell plaIltS. Eatell Í1ldividllalS produce n1aIly more tern1ÍnaI compmld umbeIs. Nevertheless,these are SigllificaIltly smallet thall the ternúllaI compomld lImbeIs of U11eaten plaIltS.

final height? 2) Do eaten plants producemore compound umbels than uneaten

plants? 3) What is the effect of herbivoryon inflorescence quality?

1943, 1945, Joly 1970, Goodland and Ferri1979). We sampled B. pinnifolia in

February 1988, and P. spedosus in April1988. Flower buds in the distal portionof the branches of B. pinnifolia break

dormancy after a vegetative growth phase,thus starting the flowering season. Many

flower heads are produced at the terminal

portion of the branches. Many dormantbuds are also present along the branches.

These buds may uecome active after the

dominant buds are damaied or deJtroyed,

MATERIALS AND METHODS

The plants we studied are located inSerra ào Cip6, Minas Gerais State, Brazil,which is characterized by rocky, nutrient-

poor soil and low-nutrient status plants (see

Silveira 1908, Hoene 1927. Hutchinson

ECOTROPICOS Vol. 3 (2). 1990 "n

FERNANDES AND RIBEIRO

measurements, we multiplied the number

of shoots by their lel1gths for eachtreatment in order to have what we called

a '.damage response index". A non-

parametric test to analyze the B. pinnifoliadata was used because of unequr.l variances

among the treatments (Quade test, Conover

1980, Zar 1984).

RESULTS

Frffit al1d gall effects 011 R. pinnifolia

We counted the number of new shoots and

measured the lengtll of galled, frostaffected, and uneaten branches. After a

veget.'lti ve rosette stage, P. speciosusproduces a leafy stalk terminating with onelarge compound umbel (Fig. 1). Thus, a

compound umbel, and several individualinflorescences are produced. Dormant buds

are present along the plant's leafy stalk.There is no study available on the floral

biology of P. speciosus (A.M. Giulietti,personal communication). We measuredplant final height, number of primary

inflorescences, number of flower heads,quality of inflorescences (here measured as

develQped or atrophied) , and inflorescencediameter, as the plant variables affected

by tl1e flower stalk herbivore.Because the number of shoots and

their lengths are not independent

Frost affected and galled branches

produced significantly more shoots and

1onger shoots than unaffected branches of

B. pinnifo1ia (Quade test, p < 0.0001,Fig. 2a). Frost affected (N = 41) and

ga11ed branches (N = 41) produced an

average of 9.37 (SE ..:t 0.62), and 5.56

12 12AÊ3 9:I:f-"Z 6w-lf-O 3O:I:{1)

i..

ffi 9a)~=>Z 6f-OO:I:(I) 3

oGALLFROST FROST GAll

DAMAGEDUNDAMAGED

UNDAMAGEDDAMAGED

FIGURE 2. Brickelia pinnifolia response to frost, alld herbivory by a tcphritid stem galling inscct.A) Shoot nlmJbcr prodllced by frost affected, galled, alld \JIlaffected B. jJinnifolja stenlS. Therc wcrc

statistically significant differenccs i;l thc numbcr of shoots produccd by frost, alld gallcd ~tcms comparcdwith unaffcctcd stcms (Qllade tcst, F280 = 3.15, p < 0.0001). B) Lcilgth of shoots produccd by frost

affcctcd, galled, and unaffcctcd brallclies of B. pinnifolia. Thcre wcre statistically significallt differences

betwccn alI trcatmcnt mcalls (see Table 1 alld tcxt for results).

80

PLANT RESPONSE TO DAMAGE

(SE .:t 0.24) shoots, respectively. On theother hand, unaffected branches producedan average of only 3.15 (SE .:t 0.15)shoots. Branches hit by frost, and by thegall herbivore produced the longest shoots.

Frost affected branches produced shootsaveraging 8.72 cm (SE .:t 0.36 cm), andgalled branches produced shoots averaging

9.62 cm (SE ..::t 0.63); while unaffectedbranches produced shoots averaging 6.5 cm(SE ..::t 0.23, N = 41, Fig. 2b). Multiple

comparisons of treatment means (seeConover 1980: 297) indicated significant

differences between all treatments

(Table 1).

T ABLE 1. Multiplc compariso11S of treatme11t mca11s amo11g effects of damage 011 B. pinnifolia (scc

CO110VCr 1980: 297). The treatmc11ts were CO11Sidcred differe11t from each other if the differc11cc betwee11

thcir sums I s; -s; I cxcccded:

2b (Ai -BJI/2

ISj -Sil > ta12

(b -1)(k -1)

AlI treatments were significantly differcnt among them.

TREATMENT COMPARISON DIFFERENCE IS, -S,

I s,uJaITecled Sfr~11534.0

I SlUlaffcclcd SgolJ..1I911.0

I Sfr~1 -S.allcd 623.0

~

Hcrbivore Effects 011 P. speciosus only one compound umbel per plmlt (N= 24). In addition, eaten plants produced

significantly more individual inf1orescencesthan uneaten plants (Anova, p < 0.05,

Fig. 3c). Eaten plants produced an averageof 178.0 (SE .:t. 28.3, N= 15) individual

inf1orescences while uneaten plants onlyproduced 128.2 (SE .:t. 14.6, N= 24)

individual inflorescences. Both compoundumbels and individual inf1orescences

produced by eaten plants were significantlysmaller than those produced by uneaten

plants. However, the overall size (total

diameter) of the compound umbel of eaten

Plant final height was not influenced

by the umbel stalk herbivore (Anova, p

> 0.05, Fig. 3a). Eaten plants had an

average height of 113.33 cm (SE .:t 6.33cm, N = 15), and uneaten plants averaged

129.21 cm (SE .:t 8.28, N = 24).

Eaten individl1als produced significantly

more compound umbels than uneaten

individuais (Anova, p < 0.0001, Fig. 1,

and 3b). Eaten individual produced an

average of 5.8 (SE .:t 1.32, N = 15)

compound umbels, while uneaten produced

ECOTROPICOS Vol. 3 (2). 1990 RI

FERNANOES ANO RIBEIRO

140Ê21-:1:~w:1:

1-Z<--1o.

{1)wuzwu{1)wcro..Ju.

~

>-a:<~a:!1-

O'Z

70

6

3

O-L--oATTACK UNATTACKUNATTACKATTACK

50200 Ê2a:w1-W~~o

..Ju.z

ccnw()zwucnwa:o..Ju.~

ozo()wcn

O'Z

25100

.,-L.

Ioo

ATTACK UNATTACKÚNATTACKATTACK

100

2cw0.o..Jw>wc

~u.~

dzo(.)w(/)

50

~

oATTACK UNATTACK

FIGURE 3. Paepalanthus speciosus responsc to a buprcstid bcctlc hcrbivory .A) Final plant hcight

was not influenccd by the inflorcscence stalk herbivore (Anova, FI.37= 1.87, p > 0.05). B) Eaten

individuaIs produced significanlly more primary infloresce11ces thaIl lmeate11 i11dividuals (A11Ova, F 137= 21.59, p < 0.0001). C) Eatcn individuais produced sig1lifica11tly morc seco1ldary ililloresce1lces thá11

\mcate1l i1ldividuals (A1lova, F 1.37 = 2.96, p < 0.05). D) It1floresce1lce size of eate1l iI}dividuals was1lot sig1lificantly differc1lt from lilat of U1leate1l i1ldividuals (A1lova, F t.37 = 0.099, p > 0.05). E)

Sig1lificanlly fcwcr scco11dary inflorcsce11ces devclol)Cd O1l eatc11 plants compared to uneate1l plants(A11Ova, F 1.37 = 9.60, p < 0.002). (scc tcxt for rcsults).

82

PLANT RESPONSE TO DAMAGE

is intriguing. There are at least four

alternative but not mutually exclusivehypotheses that may be responsible for this

plant response pattern.The first hypothesis predicts that apic.1l

buds grow more vigorously than lateralbuds, and plant energy is canalized to the

apical meristems thus enabling more buds

to break dormancy (see Longman 1978,Tomlinson 1978) .In fact, frost affected

only apical meristems whereas the stemgall tephritid destroyed only lateral buds.

The second hypothesis predicts thatfewer buds were damaged by frost thanby the gall tephritid, thus resulting in the

higher numbers and larger shoots foundon frost-affected branches compared with

galled branches. Only terminal buds werehit by frost whereas galling may hadaffected more buds that were perhaps lessactive and less dominant lateral buds, thusreflecting in fewer, and smaller branches .

The third hypothesis predicts that the

gall competes for energy with the adjacentbuds thus making them shorter, and

suppressing bud break in some dormantbuds. Insect galls induce dramatic metabolicchanges in their host plants during gall

formation and growth (e.g. Rohfritisch and

Shorthouse 1982, Fernandes 1986 and

references therein) that could be associatedwith the B. pinnifolia's observed response

patterns. Furthermore, galls are nutrientsinks within host plants (see Jankiewicz

et al. 1970, Stinner and Abrahamson1979), so that portion of the nutrientsavailable for plant growth is diverted togall maintenance. On the other hand, infrost damaged branches, all the energy

would be directed to the growth of newshoots because a branche would not have

part of its energy diverted to a different

structure.

plants was not significantly different from

that of uneaten individuaIs [Anova, p >0.33, Fig. 3d; eaten 39.60 cm diameter

(SE ..t 2.34, N= 15), uneaten 36.58 cmdiameter (SE ..t 1.91, N= 24)].

Although significantly more compoundumbels and individllal inflorescences wereproduced on eaten plants compared touneaten plants (Figs. 3b and 3c),significantly fewer flowers on individual

inflorescences developed to fruit stage oneaten plants compared with uneaten plants(Anova, p < 0.002, Fig. 3e). Eighty five

percent of the flowers on individualinflorescences of eaten plants developed

com~letely to the fruit stage (85.44%, SE..t 0.05, N = 15) as opposed to 97%

that developed completely to the next stageon uneaten plants (97.34%, SE ..t 0.01%,N = 24).

Failure to develop was due primarilyto atrophy of the irldividual inflorescences.

Undeveloped (atrophied) individualinflorescences were smaller and blackishcompared to developed ones.

Of the fifteen plal1ts that were attackedby the flower herbivore, 5 (33%) suffered

a second attack which was responsible forthe loss of 72 % o~ the newly produced

compound umbels.

DISCUSSION

Brickellia pinnifolia branches

responded to herbivory by prodllcing ]argerand more shoots than unaffected branches.FlIrthermore, the number of shoots, andshoot ]ength were significantly higher onfrost affected than on gal1ed branches which

were higher than on unaffected branches.The higher number of shoots, 8S wel1 as

]onger shoots, prodllced by l'rost affectedbranches compared with gal1ed branches

83ECOTROPICOS Vol. 3 (2). 1990

FERNANOES ANO RIBEIRO

attack may also have altered the sex ratios

among flowers within individual

inflorescences. Herbivory may delay

flowering such tl1at the newly produced

flowers on eaten plaJltS are not p)llinated

or are more likely to be adversely affectcd

by temperattue or weathcr condition (e.g.

Schemske 1977) .No clear innuel1ce of

herbivory was observed 011 host plal1t fil1al

height, or 011 overall compoul1d umbel

diameter. Innorescel1cc sizc is an important

factor for pollil1ator attraction (c.g. Willson

and Price 1977, Zimmcrman 19BO, Wyatt

1982, Schimid-Hempcl al1d Spciscr 1988,

but see Willsol1 and Rathcke 1974).

Inflorescel1ce display al1d compctitiol1 for

pollil1ators m~y also bc adaptive (c.g.

Campbell 19E5, Campbcll al1d Mottcl1

1985). They may be evcl1 more importal1t

in annuals and il1 habitats sllch as tlle Serra

do Cip6 il1 which there is a marked nowcr

season. Furthermore, the adaptive vallle of

inflorescel1ce display al1d size might be

amplified il1 tropical habitats where plal1ts

are, il1 generar, far away from a

cons.pecific, and thll5 hard to fil1d for thc

pollil1ator species.The sigl1ifical1ce of compcl1satory

production of compound umbcls nl1d

individual innorescel1ces in P. .çp(~Ci().Çll.Ç

is not clear .A secol1d wave of hcrbivory

follows after plal1ts respond to the first

herbivore attack. 111 this sccond wavc of

attack 72% of the compound umbcls wcre

destroyed and no plal1t growth rcsponse

was observed. There arc at Icast two, not

l1ecessarily mutllally cxclllsive, mechal1ismsor hypotheses that may be involvcd hcre.

The first hypothesis is OI1C of mcristcm

1imit.'ltion il1 which axillary bllds localcd

in the co:npoul1d umbcl stalk had alrcady

developcd into sccondary il1norcsccl1ccs

prior to the first wave of attack. ll1e

The fourth hypothesis predicts thatfrost response is more prcdictable than

gall response and that plants have evolvedto respond to frost damage with longerand more numerous shoots. Frost is a

phenomenon of high predictability in thehigh elevations of the Serra do Cip6 thataffects alI B. pinnifolia's unhardened buds,whereas galling is phenomenon dependenton highly variable factors such as host

plant defense, host plant quality, herbivoreabundance, and herbivore host findingbehavior, for instance. Frost would thenbe more predictable both in space and time

than galling. However, more work is calledto u~ravel the mechanisms and processesbehind the trends fol1l1d.

Each stem of B. pinnifolia termil1atesin many flower heads. We specl1late thatmore bral1chil1g wiII Iead to more flowersal1d frui ts being prodl1ced. We do not

kl1ow, however, how mal1y viable seedsthe new bral1ches prodl1ce, al1d whetherthe replacemel1t of damaged shoots affects

timil1g of seec dispersaI al1d seedlil1g

establishment. Another importantul1al1swered question is how tllis extra

expel1ditl1re of reSOl1r<;es affects survivorshipand Iife time seed ~rodl1ction.

Herbivory by a buprestid beetle onthe flower stalk of P. speciosus

significal1tly increased prodl1ction of

compound l1mbels and individualinflorescences. However, fe\Ver flowers inthe il1dividl1al il1florescel1ces developed intofrl1its on eaten plants compared withul1eatel1 plal1ts. The "atrophy" al1dcol1seql1ent frl1it faiIl1re observed on eatenil1dividl1al inflorescences migh( be the resultof herbivory. Flowers il1 thís family are

gel1eralIy unisexual al1d male al1d femaleflowers occur il1 the same Ilower heads(Dahlgreel1 et al. 1985) al1d herbivore

84

PLANT RESPONSE TO DAMAGE

In summary, plants may respond to

abiotic or frost damage, and herbivory inmany different ways. Many historic and

ecological factors may influence theseresponses, such as the type of damage,

plant phenology, plant nutritional status,plant phylogenetic constraints, as well asthe kind of interactions plants have with

their herbivores (see Maschinsky &Whitham 1989, Whitham et al. 1990).

Genera]izations about whether a p]ant's

compensatory responses to herbivory, aswell as physical damage, are adaptive forthe plant can be made only after the who]espectrum of responses of both plant and

herbivore are fully explored and morestudies are performed in both tropical and

temperate systems.

second hypothesis is one of resourcelimitation in which the plant could notrespond to the second attack because therewere no resources left. These p]ants livein nutrient-poor soil habitats. Thus, p]ants

may be constrained in their response to

herbivory after the "limited" energybacktlp supply is used. The differentia]response of nutrient-stressed and hea]thyp]ants to herbivory remains to be ful]y

eva]uated. In simi]af systems, Paige andWhitham (1987), and Hendrix (1979, 1984)did not observe a second wave of attackby the herbivore on their sttldy p]ants. Asecond wave of attack by the herbivore

wou]d jeop~rdized the compensatory growth

response of the host p]ant. The argumentthat p]ant overcompensation is adaptive tothe plant (see Paige and Whitham 1987)was only partiaI]y supportcd in this study.

An herbivore's view of these findings

indicates that the P. speciosus flower

herbivore may be manipu]ating itsresources. The plant response to the firstattack wou]d cause many more resourcesto be produced which wou]d then be usedby the herbivore. If the herbivore ais

specialist on this p]ant specjes and/or this

plant fami]y, this strateg.Y would beadaplive for the herbivore. In fact, weobserved similar damage on other c]ose]yrelated Paepalanthus species. A resource

manipulation strategy wou]d be importantto the herbivore for adjusting pheno]ogica]lyto P. speciosus if p]ant popu]ations present

assynchronous flowering seasons in thea]titudina] gradient of Serra do Cip6, andl

.or Paepalanthus species present different

tlowering phenologies. Nevcrthe]ess, morestudies are ca]]ed to unraveI the

mechanisms and processes invo]ved in this

system.

ACKNOWLEnGMENTS

We are grateful to the following

persons for their comments on earlier draftsof this manuscript and/or assistance: S.

Suter, T.G. Whitham, C. Ghering, H.R.Pimenta, J. Romero-Napoles, J.Maschinski, and two anonymous reviewers.We thank Giulietti for identifying the

plants. L.A. Renn6 and U.G. Castro

(ICB -Universidade Federal de MinasGerais) provided the field facilities whichare greatly acknowledged. A Sigma Xi

grants-in-Aid and a scholarship of theConselho Nacional de Pesquisas (CNPq)(process N 200.747 /84-3-Z0) to GWF isalso acknowledged.

REFERENCES

Bclsky, A.J. 1986. Docs hcrbivory bC11Cfit pla11tS?A rcvicw of thc cvidc11ce. America11 Natllralist

J27: 870-892.

CampbcJI, D.R. 1985. PoJJi11a(or shari11g a11d scedsct of Ste//aria pubera: compctitio11 for

poJli11atio11. EcoJogy 66: 544-553.

ECOTROPICOS Vol. 3 (2). 1990 85

FERNANOES ANO RIBEIRO

CampbcIl, D.R. al1d A.F. Mottcl1. 1985. The;

mechal1ism of competitiol1 for pollil1ati0l1 bctwcel1two forest herbs. Ecology 66: 554-563.

Col1over, W .J .1980. Practical 11011 paramctrics.Secol1d editiol1. Wiley, Ncw York.

Dahlgrccl1, R.M.T., H.T. Clifford al1d P.F. Yeo.

1985. The fan1ilics of the Mol1ocotyledol1s:strllctllrc, cvollltiol1, al1d taxol1omy.

SprÍllgcr- Verlag, Berlil1.Del1l1ill, G.B. 1985. Thc effcct of the gall wasp

Trichilogaster acacialongifoliae (Hymcl1optcra:Pteromalidac) 011 reprodllctivc potcl1tial andvegctative growth 011 the wccd Acacia longifolia.AgriclIltllrc. Agr. El1virol1mcl1t 14: 53-61.

Del1l1i11, G.B. 1988. Why a gall formcr cal1 bc a

good biocol1trol agel1t: thc gaIl wasp Trichlogasteracaciaelongifolia al1d thc wccd Acacia longifolia.

Ecological El1tomology 13: 1-9.Fcmal1dcs, G.W. 1986. Col1trôle poplllaciol1al dc

galhas de il1setos. Séric Mol1ografias cm Cil1cia

c. TccI1010gia., Brazilial1 Embassy/SECTEC,Washil1gtol1, D.C., U.S.A.

Fcmal1dcs, G .W .1987. 1I1sect galls: their ecol10n1icimportal1cc and col1trol. Rcvista Brasilcira deEl1tomologia 31: 379-398.

Goodlalld, R. alld M.G. Fcrri. 1979. Ecologia doccrrado. Edllsp alld Itatiaia, Belo Horizol1te,

Brazil.

Helldrix, S.D. 1979. Compel1satoly rcprodllctiol1Íll a bicll11ial hcrb followillg illscct defloratiol1.

Occologia 42: 107-118.Hclldrix, S.D. 1984. Variatiol1 il1 sccd wcight at1d

its cffcct 011 gcrmÍllatiol1 il1 Pa.l.tinaca sativa L.

(Umbcliferac). Amcricall Jollmal of Botally 71:

795-802.Hocl1c, F.C. 1927. Aspcctos C flora das scrras de

Mil1as Gcrais. Rcvi~a Ccrcs 3: 85-93.Hlitchil1soll, G.E. 1943. The biochcmistry of

alllmillllm al1d of ccrtaill rclated clcmcl1ts.

QlIartcrly Rcview of Biology 18: 1-29,128-153, 242-262, 331-363.

HlitchillSOIl, G.E. 1945. AllIn1imIm il1 SoilS, plallts

alld allimals. Soil Scicl1cc 60: 29-40.Jallkiewicz, L.S. , H. Flich al1d M. Al1toszcwski.

1970. Prclimillary Shldics 011 thc trallslocatiol1of 14C-Iabcllcd assimilatcs al1d 32pO4 towards thcgall cvokcd by Cynips (Diploleps) quercus-folii

L. 011 oak Icaves. Marccllia 36: 163-172.

Joly, A.B. 1970. COI1l1cca a vcgetacao brasileira.Edllsp, São Palllo.

Lollgmall, K.A. 1978. COIltr01 of shoot extcl1siol1alld domlal1cy: extcmal al1d illtcmal factors, p465-495. Jn P.B. Tomlillsol1 al1d M. H.

Zimmcmlalll1 (cds.). Tropical trces as livillg

systcms. Cambridgc Ul1ivcrsity Prcss, Lolldol1.

Maschil1ski. J. at1d T.G. Whitham. 1989. TlIecol1til1uum of plant respol1ses to hcrbivory: tlle'

il1fluel1Ce of plal1t associatiol1. l1utriel1t availability,al1d tin1il1g. Americal1 Naturalist 134: 1-19.

Paige, K.N.. al1d T.G. Whitham. 1987.Ovcrcompel1satiol1 il1 rcspol1Se to mammalianhcrbivory: thc advalltagc of being catel1. Americall

Naturalist 129: 407-416.Rohfritsch, O. atld J.D. Shorthouse. 1982. Il1scct

galls. p 131- 152. ln: G. Kahal and J.S. Schcll

(cds.). Molccular biology of plal1t tul!lors.Acadcn1ic, New York.

SchenlSke. S.W. 1977. Floweril1g phenology and

sccd sct il1 Caytonia virginica (Portulacaccae).Bullctil1 of thc Torrey Botanical CllIb 104:

254-263.Schin1id-Hcmpcl, P. al1d B. Speiscr. 1988. Effccts

of il1f1orescel1cc size Ol1 pollil1atjol1 il1 Epilobiuln

angustifolium. Oikos 53:98-104.Silvcira, A.A. 1908. Flora e serras n1incjras.

Inprcl1sa Ofjcial, Bclo Horizol1te, Brazil.Stjnl1er. B.R. al1d W.G. Abrahamsol1. 1979.

El1ergctics of thc Solidago canadensis -stem

gall il1sect -parasitoid guild il1teractiol1. Ecology60: 918-926.

Tomlj\1sol1, P .B. 1978. Bral1chjl1g al1d axjsdjffercl1tiatiol1 il1 tropjcal trees. p. 187-207. ln:

P.B. Tol11ljl1sol1 al1d M.H. Zimmcrnlal1n (cds.).

Tropjcal trces as livil1g systems. Cambridge

Ul1ivcrsjty Press. Lol1dol1.Whjtham, T.G.. J. Maschjnskj. K.C. Larsol1 al1d

K.N. Paqige. 1990. Plal1t rcsl)()l1ses to herbivory:thc col1til1uum from l1egativc to positivc andUIldcrlyil1g physiological mechal1isms. ln: P. W.

Price, T.M. Lcwil1sohl1, G.W. Fcmandes al1d

W. W .Bcl1sol1 (cds). Plant-al1imal jl1tcractjons:

Evollltiol1ary ccology of tropjcal al1d tcmpcratcrcgjol1s. Wilcy. New York (jl1 press).

Willsol1, M.F. al1d P.W. Price. 1977. Thc cvolutiol1of il1f1orcscC11ce sjze iI1 Asclepias (Asclcpiadaccac).

Evolutiol1 31: 495-511.

Wjllsol1, M.F. and B.J. Rathcke. 1974. Adaptivedcsigl1 of thc floral display il1 Asclepias syriacaL. Amcrical1 Naturalist 92: 47-57.

Wyatt, R. 1982. II1f1orcsccl1ce archjtccturc: howf1owcr mll11bcr. arratlgcmCl1t, atld phcnology cffcctpollil1atjol1 and frujt set. Amerjcal1 Jollmal of

Botal1y 69: 585-594.Zar. J.H. 1984. Bjostatistjcal al1alyses. Second

edjtjol1. Prcl1tjcc-Hall, New Jersey.Zimmcrmal1, M. 1980. Rcproductiol1 in

Polemonium: competitiol1 for pollillators. Eeology

61: 497-501.

86