Herbivory in Australian forests - a comparison of dry sclerophyll and rain forest canopies

11 Pages·1995·4.5 MB·

by  M D Lowman

—
Herbivory in Australian Forests
A
Comparison ofDry Sclerophyll
and Rain Forest Canopies
MargaretD. Lowman
(CommunicatedbyD.Keith)
—
Lowman,M.D.,HerbivoryinAustralianforests acomparisonofdrysclerophyllandrain
forestcanopies.Proc.Linn.Soc.N.S.W.115:77-87(1995).
Long-term measurementsofinsectgrazing in the canopies ofdifferentAustralian
foresttypeswerecomparedover10years.Foresttypesincludedrainforests(cooltemper-
ate, warm temperate, and subtropical) and dry sclerophyll, all situatedwithin a50 km
radiusofArmidale,NewSouthWales.Similarmethodsofmarkingandmeasuringleavesin
tree canopies were employed, and similar modes ofanalyses were executed after the
leaflifespanwascompleted.Grazinglevelsforaforeststandrangedfromaslowas5-8%
annual foliage consumption to 300% (when three successive flushes were grazed).
Contrarytoearlierliterature,thislongterm,communitylevelstudyshowsthatherbivoryin
forestcanopiesisextremelyvariablebothatintra-andinter-sitespatialscales.
M.D.Lowman. TheMarieSelbyBotanicalGardens,SarasotaFL34236, USA;manuscriptreceived
14June1994,acceptedforpublication15February1995.
—
Thereawaitsa—richharvestforthenaturalistwhoovercomestheobstacles gravitation, ants, thorns,
rottentrunks andmountstothesummitsofjungletrees.
WilliamBeebe, 1917,TropicalWildLife
WhenIfirstcametoSydney Universityasapost-graduatestudent,Iwasverykeentostudythe
rainforestcanopy. Infact, Ihopedtoconstructaswinganddangleprecariouslyupinthefoliageto
surveybutterflypopulations. Havingamuchbetterunderstandingofthestatisticalrequirementsfor
ecologyandthenecessityofreplicationinsamplingdesign,PeterMyerscoughkindlytalkedmeoutof
waitingforelusivepollinatorsin theuppercanopy. Idecidedtostudy leavesinstead, andtheypre-
sented themselves as much betterreplicated unitsforsampling, so importantformyfirst research
endeavor.
Ithenhadtoworkoutsomeprotocolsforworkinginthatecologicalfrontier:theforestcanopy.I
contemplatedhiringmonkeystofetchmyleavesfromtheuppercrowns(oranyothermodeotherthan
danglingprecariouslyfromalimbbymyself), butPeterkindlyinformedmethatlearningtoclimbwas
mybestoption.FrightenedthoughIwas, thatadvicehascarriedmethroughoverfifteenyearsofinten-
siveresearch (subsequently usingmany othermethodsofaccess) on many aspectsofleavesinforest
canopies. Iamgra—tefultoPeterfor—hisfriendship, adviceandcollegiateinteractionsoverthosemany
years, andIhope ashisstudent tocontinuehislegacyofexcellenceinecology.
Introduction
The consumption ofplantmaterial byherbivoresisasubjectofgreateconomic as
wellasecologicalimportance (reviewedin BarbosaandSchultz, 1988;Price etal, 1991).
The mostabundantherbivores in forests are insects, andAustralia is no exception (see
Morrow, 1977; Lowman, 1985). Theimpactofherbivoryonaplantrangesfrom stimula-
tion ofnewgrowth (reviewed in Mattson andAddy, 1973; Lowman, 1982) to negligible
impacts (Lowman, 1984a) toentirestanddieback (LowmanandHeatwole, 1992).
Insectherbivores in forests are difficult to study, due both to theirrelativelysmall
size as well as to their cryptic qualities in a large three-dimensional space (reviewed in
Proc.Linn.Soc.N.s.w., 115,1995
78 HERBIVORY IN AUSTRALIAN FORESTS
Lowman andMoffett, 1993) Whenwalkingthroughaforest,we usuallyfocusourobser-
vationsonanarrowbandofg.reenfoliage,fromabout -2minheight.Thisrepresentsat
most 10% ofthe foliage in mature forests,with the restoften high above ourheads and
consequently beyond our observations. Since the majority ofplant-herbivore relation-
ships occur where the foliage is located, it is obvious that herbivory as a forest process
remainsliterallyoutofreach. Onlyrecentlyhavetechniquesofaccessbeendevelopedto
facilitateresearchinforestcanopies.
It has been previouslyreported thatforests representedvastexpanses ofhomoge-
neous green tissue (e.g. Hairston, Smith and Slobodkin, 1960), but this assumption is
over-simplified.Thelifeofaleaf,whichcomprisesthebuildingblockoftheforestcanopy,
undergoes many complex and critical phases ofgrowth dynamics, including leafemer-
gence,longevity,andphysicalqualitiesthataffectitssusceptibilitytoherbivory,decompo-
sition, andsenescence. In aleafslifespan, itiscriticaltosurvive thevulnerableweeksof
foliar expansionwithoutbeingeaten (Coley, 1983; Lowman, 1985). From a plant's per-
spective,thereexistsanevolutionaryrouletteofrenderingone'sgreenfoliagelesssuscep-
tibletosuccessivegenerationsofdefoliators.Viewedfromthepointofviewofaherbivore,
acomplexworldofdifferentbitesmustberecognized:softversustough,nutritiousversus
non-nutritious, old versus young, apparent versus non-apparent, rare versus common,
and probably other choices that have not yet been detected by biologists (reviewed in
Lowman, 1994).
Recently, the importance of understanding insect pests in forests has led to
increased research on canopydefoliation (e.g., Barbosa and Schultz, 1987, Wong et al,
1991; Lowman and Heatwole, 1992). In Australia, the topic offorest herbivory has fos-
tered lively debates, particularly concerning whether or not higher levels ofherbivory
existascomparedtoothersystems (seeLowman,1987;FoxandMorrow,1983;Landsburg
andOhmart, 1989) Becauseofthevariabilityofmethodsemployedtomeasureherbivory
.
(sensuLowman, 1984),thequestionstillremains:doesAustraliaindeedsustainhigherlev-
elsofherbivoryinitsforestsassomedataindicate,orhavestudiesinthesesystemssimply
beenconductedduringphasesofoutbreak,oraresomemethodsemployedlessaccurate
thanothers?
Historically, most herbivory studies have involved the measurement of levels of
defoliationinforestsatonepointintime.Foliagewastypicallysampledneargroundlevel
in temperate deciduous forests, where annual losses of3 - 10% leafsurface area were
reported (reviewsin Brayand Gorham, 1964; LandsbergandOhmart, 1989). Moststud-
ies, however, could not be extrapolated to evergreen rain forests for three reasons: 1.)
temperate deciduous forests have a comparatively simple phenology with an annual
turnoverofleaves (e.g. Bray, 1961); 2.) measurementsweresometimesmadefromsenes-
cent leaves retrieved from the forest floor (Odum and Ruiz-Reyes, 1970); and 3.) only
destructive discrete sampling was attempted (e.g. Leigh and Smythe, 1978). In short,
defoliationwastreatedasadiscrete,snapshotevent (Diamond, 1986) accountingfornei-
,
thertemporalnorspatialvariability.
In recentyears, thecomplextemporalandspatialpatternsofleafgrowthdynamics
inforestcanopieshavecausedecologiststoexpandtheirsamplingdesigns.Forexample,
the traditional methods ofmeasuringherbivorybydestructive samplingofsmallquanti-
tiesofleaveshavebeenexpanded (reviewedinLowman, 1984b;LandsburgandOhmart,
1989) Whereasearliermeasurementsofforestherbivorywereconductedovershorttime
.
spans,wererestrictedtounderstoryfoliage,andinvolvedverylittlereplicationwithinand
between crowns, more recent studies have incorporated larger sampling regimes. And,
when herbivorywasmonitoredoverlongerperiods (> 1 yr) andincludedwiderrangesof
leafcohorts (includingdifferentageclasses,species,andheights),higherlevelsofgrazing
were reported (Coley, 1983; Lowman, 1985; Lowman and Heatwole, 1992). Long-term
measurementshavealsoillustrated thehighvariabilityofherbivory,bothtemporallyand
spatially,withinastand (e.g.Coley, 1983;Lowman, 1985;BrownandEwel, 1987; 1988).
Proc.Linn. Soc.n.s.w.,i15, 1995
M.D.LOWMAN 79
In this study, I compare aspects ofthe spatial and temporal heterogeneity ofher-
bivorybetween adjacent rain forest and dry schlerophyll tree crownswithin New South
Wales, Australia, (but the comparisons between Australia and other continents are still
open todebate). Ialsoemphasize themethodologicalchallengesassociatedwith studies
ofherbivoryasacanopyprocess (becauseobviouslythereliabilityofmethodshasanenor-
mousimpactontheaccuracyoftheresults) Whereasbiologistshavesuccessfullycounted
.
andmeasuredtheabundanceofherbivorousmolluscsonatwo-dimensionalsystemsuch
asintertidalrockyshores (e.g., Underwood and Denley, 1984) the heightandstructural
complexity of forest canopies make it more difficult to count and measure grazing
impactsthere.
Methods
Atleasttwo representative standsofeach ofsix differenttypes ofAustralian forest
wereselectedforfieldwork.Thesewereclassifiedaswetforest:cooltemperate,warmtem-
perate,andsubtropicalrainforests;anddryforest:healthysclerophyllforest,ruralstands
ofeucalypts that typify the outback environment, and dieback stands ofeucalypts that
have recently come to dominate the agricultural landscapes throughoutAustralia (see
Lowman, 1982;andLowmanandHeatwole, 1992forfurthersitedescriptions).Allofthe
forestswere situatedwithin 50 km ofArmidale, NewSouth Wales, at approximately 30°
20'S.
At least five tree species from wet and from dry forests were selected for field
measurements.Bothcommonandrarespecieswereselectedineachforesttype,sincethe
overallaimwastoexamineherbivoryatthecommunitylevel.Forobviouslogisticreasons,
itwasnotpossibletomeasure thecanopiesofalltreespecies;butinsomecases,adjacent
trees to the studysampleswere also measured. Trees thatwere studied in greatestdetail
included:
—
rain forests Ceratopetalum apetalum D. Don (Cunoniaceae), Doryphora sassafras
Endl. (Monimiaceae), Dendrocnideexcelsa (Wedd.) Chew (Urticaceae), Nothofagus moorei
F.Muell. (Fagaceae),an—d Toonaciliata(F.Muell) Harms (Meliaceae);
and dry forests Eucalyptus blakelyi Maiden, E. viminalis Labill, E. melliodora
A. Cunn. ex Schauer, E. caliginosa Blakely and McKie, and E. nova-anglica Deane and
Maiden.
Leafgrowth and herbivorywas measured monthlyfor5 years inwetforests (1979-
1984) and dry forests (1983-1988). Because ofthe longevity ofthe evergreen leaves of
somespecies (e.g.D. sassafrasleaveslivedbetween2-12+yrs,Lowman 1992),morethan
fiveyearsoffieldmeasurementsweremadeonsomeoftherainforesttrees.Leafcohorts
weremarkedinthecanopyrepresentingdifferentlightregimes,heights,species,individ-
ual crowns, and sites. In total, over 10,000 leaves were monitored over the duration of
theirlifespans.Isolatedeventsinthelifeofaleafwerequantified,includingdateofemer-
gence,lengthofsurvival,proportionofleaf-arealossestoherbivores,dateofsenescence,
andrateofdecay.Onlyherbivoryisreportedhere,althoughotherinformationwasneces-
sarytocalculateannuallevelsofgrazing.
The extentofreplication ofleaveswithin acrownwasdetermined bypilotstudies
using leaf size to indicate environmentally different regions in the canopy (Lowman,
1985) Forexample, because C. apetalumleavesvariedsignificantlyinsizewith respectto
.
light levels, canopy heights, individual trees and sites, leaves within each ofthese cate-
gorieswere monitored. In contrast, D. excelsahad a homogeneous canopy, so all leaves
withineachtreewerepooledasonepopulation.Ingeneral,between200-1000leaveswere
measured to calculate herbivory for a species. Further information on the numbers of
replicates and regions ofcrown sampled for both wet and dry forest are reported else-
where (Lowman, 1992;LowmanandHeatwole, 1992,respectively).
ProcLinn.Soc.N.S.W.,115,1995
80 HERBIVORY IN AUSTRALIAN FORESTS
COO
<O<o!
co < z4J
DC
CO 1W- ™c
D u
<
Z c3
E
CO e
1-
E
^ 111
h<-
O(O <LrU
Q.
5
Proc.Linn. Soc.n.s.w., 115, 1995
M.D.LOWMAN 8]
a.
o
-a
-A n:
o
O) .£ fc
3 TOXd!.>fer>f>
LUUJ LULU
II II II II
LfUtLULULULU .Si)
Proc.Linn.Soc.n.s.w.,115,1995
82 HERBIVORYIN AUSTRALIAN FORESTS
Profile diagrams were constructed throughout several forest sites for each forest
type,usingahypsometerandstandardforestrytechniques (seeLowman, 1982).Idealized
forestdiagramswere constructed from these measurements, and used here to map the
herbivorywithineachforestcommunity.
Results
Averages ofleafsurface area loss ofall the leaves were calculated and mapped to
illustrate the herbivory for each forest community (Figs. 1,2). Herbivory in individual
canopies ranges from negligible (e.g. < 3% for Toona ciliata) to over 300% of annual
foliageproductionindrysclerophylltreeswheresomeeucalyptsre-foliatedthreesucces-
sive times after defoliation (see also Lowman and Heatwole, 1992). Herbivory levels
varied significantly both between species and between forest types, with the dieback
standsexhibitingthe highestgrazinglevels, some tosuchextremes thatcrown mortality
wasalsoobserved.
Herbivory in rain forests was quite different from neighbouring dry sclerophyll
canopies. In rain forests, there were greater differenceswith vertical stratification from
top to bottom of the canopy. For example, Ceratopetalum apetalum had 9.4% leaf area
grazedintheuppercanopy,ascomparedto35%intheunderstoryofthewarmtemperate
rainforest,almosta4-folddifference.Incontrast,eucalypttreeshadmorehomogeneous
herbivory throughout the crown of each individual; but more wide-ranging levels of
grazingbetweenspeciesandsites. Sometreesindrysclerophyllwoodlandslostaslittleas
8%leafareaperyear (e.g.Eucalyptusblakelyi),whereasE. nova-anglicainruralpastureslost
as much as 300% in a given year (i.e. scarab beetles ate the entire crown three times
successively, followed by re-leafing) Although the dietary qualities ofeucalypt foliage
.
have been studied elsewhere (see Landsburg, 1990; Fox and Morrow, 1983), it is none-
thelessphenomenalthatlevelsofgrazingvarysoenormouslyamongneighbouringtrees.
In order ofincreasing levels ofannual grazing, Australian temperate forestswere
ranked as follows: healthy dry sclerophyll woodlands (13%), subtropical rain forests
(14%), warm temperate rain forests (22%), cool temperate rain forests (27%), healthy
standsofsclerophyll treesinruralpastures (35%), anddiebacksclerophyll treesinrural
pastures (89%).
Discussion
The measurementofherbivoryin evergreen forestcanopiesmaybe more compli-
catedthanpredictedbeforecanopyaccesswasareality,becausethecyclesofleafturnover
arenotalwaysseasonallydistinct (e.g.,Lowman, 1992).Theexistenceofmanycohortsor
leafpopulationswithinonecrown,requiresamorecomplexsamplingdesigntoascertain
both annual defoliation and cumulative herbivory over a leafs life span. In Australian
evergreen forests where leaf longevity was also extremely variable, the canopy was
composed ofacomplex mosaic ofdifferentaged leaves,with differentsusceptibilities to
herbivores. Leaflife spans ranged from as short as 4-6 months (e.g., Dendrocnideexcelsa,
Urticaceae) (Lowman, 1992) up to 25 years (e.g. Araucaria sp., Aracaceae) (Molisch,
1928). The average age ofan Australian subtropical rain forestcanopyleafrangedfrom
2-4years (sun) to4-12years (shade) (Lowman, 1992).
Over this ten year period, herbivory was measured using long-term monitoring
techniques and repeated visits to measure leaves and their associated phenological
changes (see Lowman, 1984b). This long-term samplingyielded grazinglevels thatwere
2-3 times higher than those reported in short term studies ofotherevergreen forests
(cf. Leigh and Smythe, 1978). It also revealed an enormous difference in grazing sus-
ProcLinn. Soc.n.s.w., i15, 1995
. . .
M.D.LOWMAN 83
ceptibilitybetweendifferentspeciesandwithindifferentleafcohortsononetreecrown.
Sowhatdotheserelativelyhighlevelsofinsectgrazingmeanintermsofthedynam-
ics ofthe forest canopy community? First, the variability in levels ofgrazing are higher
than previouslyassumed, even in adjacentforests. And second, the tolerance oftrees to
levelsofgrazingappearsmuchhigherthanpreviouslythought,andexhibitsaplasticityin
susceptibilityto defoliation thatmaybeveryimportantto subsequentmanagementand
regenerationofforeststands.
The tolerance ofthe dry sclerophyll forest canopies to outbreaks is illustrative of
their strong response to stress, probably a consequence ofmany thousands ofyears of
adaptationtophysical (aswellasbiological) limitations.Conversely,intherainforest,the
environmental 'stresses' may be more subtle within one crown, such as the changes in
microclimateasoneprogressesfromgroundtouppercanopythroughthecomplexlayers
offoliage. Comparative studies ofinsects in these two habitats will provide further in-
formation on their trophic structures, especially relative proportions of herbivores
(Kitchingetal, 1993andunpublisheddata)
The heterogeneity of defoliation is a consequence of a leafs environment and
phenology, with different leaf cohorts exhibiting different susceptibilities to grazing
(sensuWhittam, 1981).Fromtheselong-termstudiesofherbivoryinforestcanopies,lam
nowabletoisolate"hotspots"inthecanopy,wheregrazingwillbepredictablyhigher(Fig.
3). These 'hotspots' represent foliage with greatest susceptibility to herbivores, such as
new leaf flushes, colonizing species that are characterized by soft tissue, lower shade
regions ofthe canopywhere insects aggregate to feed in the absence ofpredators, and
canopyregionsthatattractmoreinsectsduetothepresenceofflowers,epiphytesorvines
(e.g.,Lowman, 1992;Lowman,MoffettandRinker, 1993;Lowman,unpublished).These
regionsaredifferentbetweenrainforestanddrysclerophyllcanopies. In the dryforests,
wherethephysicalenvironmentthroughoutthecanopyislessstratified,grazingwasmore
homogeneous throughout the canopy ofan individual tree, but entire crowns ofsome
specieswere grazing hotspots (e.g. E. nova-anglica). In contrast, the rain forestcanopies
exhibitlessmagnitude ofinter-speciesvariation, butobviousgrazingpreferenceswithin
individualcrowns (e.g.youngleavesinthemid-canopy)
For example, Nothofagus mooreihad approximately eight cohorts ofleaves present
within one tree crown at one point in time, each with varying levels ofsusceptibility to
insect attack. Young leaves that emerged during spring (Oct. - Nov.) were the most
preferredbycommonhost-specificbeetlelarvae thatemergedsynchronouslywithflush-
ing; whereas old leaves (> 1 yr) from summer flushes and from the previous yearwere
highlyresistant to grazing. In addition, herbivoryvaried significantlybetween branches
andindividualcrowns,butnotwithlightregimeorheight (SelmanandLowman, 1983)
More large-scale comparisons between forest communities are needed to better
understand the impact ofherbivory as an ecological process. For example, the annual
levelsofdefoliation inAustralian tree speciesrangedfrom aslowas 2-3% in subtropical
rain forests to as high as 300% in nearby dry sclerophyll (Eucalyptus) stands (Lowman,
1992; Lowman and Heatwole, 1992). Does this imply that one forest is healthier than
another?Are differentmechanismsregulatinginsectdefoliatorsandsubsequentfoliage
responses between two forests? Are the trophic structures ofherbivores and predators
intrinsicallydifferent?
Theprospectofincreasedecologicalcomparisonsbetweenandwithinforestsisan
incentivetodevelopbetterprotocolsforfieldsamplingofeventssuchasgrazing.Thepro-
cessofherbivoryhasimportantconsequencesinforestecosystems,botheconomicallyin
termsofpestmanagementandecologicallyin termsofmaintenanceofspeciesdiversity.
Forexample,whatspeciesareappropriate to sample? Is theregreatervariation withinor
betweenforests?Andhowdowe tackle thesequestionswithstatisticalandbiologicalaccu-
racy?Andperhapsmostimportantlyinthecurrenturgencyofforestconservationissues,
can we apply such community level measurements to improve the management and
ProcLinn. Soc.n.s.w.,115,1995
84 HERBIVORYIN AUSTRALIAN FORESTS
Si
.12 o
fcfrl £P3 Cu
C H-
COLU
u T3
z hj
.Sx;
111 u ao
7() X<iu V
< U 3
H
w X<u W.
n * E
o
a.u
CNJ y
1 >
•si
n °
c MS2 2o3
T—
SH313W Nl30NV1SIQ
Proc.Linn. Soc.n.s.w., i15, 1995
—
M.D.LOWMAN 85
I/5
Ww ^2kJV^'k/ ' coO)
X ^ I ScI -
H
Z
f| 1%3v2^-,^C<^0<-^ Urouj ZN ^X'\^nv^^IlIrV^^\L7.^__/ — e0)n
* sl^/vx' \ c
CQ
/^S» pa
u
w
I?
I
2
a.
©
e 4JX!
U03
bDC
c
.2?
O, Q "2 -
§ o
—
© Q.
M r%r^\JJ~W"^\ mCm Ew
PQ C/0
V
>
* a
H
M -J
z O)
^
M * !2
« /V// ^%s^ <au.
W
in .be
*" ^j,
Proc.Linn.Soc.n.s.w., us,1995
, , .
86 HERBIVORY IN AUSTRALIAN FORESTS
restoration policies in forests thathave been altered byhuman activities orothersevere
stresses?
FutureDirectionsforStudy
I pursued mystudies ofrain forestherbivory at SydneyUniversity, where I shared
office space with the graduate students ofTonyUnderwood, whose concepts ofexperi-
mentaldesignon rockyintertidalorganismshavegreatlyimprovedscientificmethodsin
thatecosystem (e.g.Underwood, 1988).How,Iwondered,couldonequantifyandsample
with similar statistical rigour in the canopy? Obviously, the forest canopy has several
obviousdifferencesfromtherockyintertidal,namelythat:
1. itisextremelythree-dimensionalwithheightsofupto50-60m (vs.two-dimensional
ontherockyshore)
2. ithasorganismsrangingahundred-foldinsizee.g.,seedlingsvsadulttrees,thripsvs
sloths (incontrasttoamorehomogeneousrangeintheintertidal)
3. ithasanairsubstrate (vswater) thatisdifficultforhumanmobility.
Thelogisticsofcountingandmanipulatingherbivoresin theforestcanopymaybe
morecomplicatedthanonanintertidalrockplatform,buttheadvantagesofimplement-
ingasoundsamplingprotocolareenormous.
Differentcomponentsofaforestcanopymustbe quantified to measure a specific
canopy process. In the case ofherbivory, all foliage components plus active herbivores
require measurement. Initialobservations,usingropesoraplatform,areidealfordeter-
miningtheorganismsinvolvedinfoliagegrazing.Itshouldbeemphasizedthatnocturnal
surveysarealsoimportantforevaluatingherbivore activity. Samplingprotocolsareillus-
tratedatdifferentspatialscales,rangingfromecosystemtositetoindividualtree (Fig.4)
All seven spatial scales are important for a thorough ecological understanding ofher-
bivoryasacanopyprocess,althoughdifferentstudiesmayprefertoapproachresearchat
thelevelofspeciesorofecosystem.
Ashabitatdestructioncontinuestoreducetheworld'sforests,canopieswillbecome
reducedboth in areaand in diversityofspecies. Itispredicted thatmanycanopyorgan-
ismshave alreadydisappearedbefore theywereeverscientificallydescribed, andmostof
themarepresumedtobe insects (Erwin, 1982, 1991;Wilson, 1992) includingmanyher-
,
bivores. Understandingthemaintenanceofspeciesdiversityintropicalhabitatsisstillan
urgent priority (Connell, 1978). The complex interactions between canopyfoliage and
defoliators is an arenafor ecological change as a consequence ofhuman activities. The
concept ofa forest pest usually implies a foliage-feeding insect and such outbreaks are
often the result of human perturbation (e.g., gypsy moth, reviewed by Elkington and
Liebold, 1990).AnotherexampleisthedeathofmillionsofeucalypttreesinAustralia,the
resultofacompleximpactofhumanactivitiesin theruralregionsresultinginoutbreaks
of a scarab beetle (Lowman and Heatwole, 1992). Although pest outbreaks are still
regardedasrelativelyrareeventsinforests,itisobviousthatthenaturalprocessesregulat-
ing canopy foliage and their defoliators require further study to fully understand the
implicationsofimbalancesthatresultfromhumanimpacts.
References
—
Barbosa,P.and—SCHULTZ.J.C.1987. InsectOutbreaks.AcademicPress,NewYork.
Bray.J.R.1961. Primaryconsu—mptioninthreeforestcanopies.Ecology45,165-7.
Bray,J.R.andGorham,E.1964. —Litterproductioninforestsoftheworld.Adv.EcolRes.2:101-157.
Brown,B.J.andEwel,J.J. 1987. Herbivoryincomplexandsimpletropicalsuccessionalecosystems.Ecology
68,108-116. —
Brown,B.J.andEwel,J.J.1988. Responsestodefoliationofspecies-richandmonospecifictropicalplantcom-
munities.Oe—cologialo,12-19.
Coley, P. D. 1983. Herbivoryanddefensivecharacteristicsoftree speciesin alowland tropical forest.Ecol.
Monogr.53:20—9-33.
Connell,J.H.1978.—Diversityintropicalrainforestsandcoralreefs.Science199,1302-1310.
Diamond, J. 1986. Overview: laboratory experiments, field experiments and natural experiments. In:
CommunityEcology (J.DiamondandT.J.Case,eds.),pp. HarperandRow,NewYork.
Proc.Linn. Soc.n.s.w.,115, 1995