ON THE

Memoirsof the MuseumofVictoria 56(2):623-630 (1997)

ANIMALS ON THEEDGE: THE'CANCELLING-OUT EFFECT

Peter J. Neville and Dennis G. Black

School of Zoology,LaTrobeUniversity, Bundoora,Vic. 3083, Australia

Abstract

Neville,P.J.andBlack,D.G.,1997.Animals onthe edge: the 'cancelling-outeffect'.Memoirs of^theMuseum _ofVictoria 56(2): 623-630.

An examination of ground-frequenting invertebrates of forest edges in Tarra-Bulga National Parkidentifieda 'cancelling-outeffect'ofanimalsattheorderlevel. Somespecies increased in abundancetowards an edge andother speciesdecreasedin abundanceat the sameedge.Suchindividual speciesresponsesaremaskedbyanalysisattheorderlevel.This phenomenonstressestheneedfordocumentationofresponsesatthespecieslevelforamore precise understanding ofthe effects of habitat edgeson distribution patterns of invert- ebrates.

Introduction

EdgeeffectsalongawalkingtrackatTarra-Bulga National Park in Gippsland, Victoria were examined. The greater variety and density of organisms in the boundary zone between eco- systems has been used to recognise edges (Laurence, 1991). It has been documented by Noss(1991)that abiotic factors playanimport- antrolewithin these transitionalzonesbetween edges. Edge effects are described here as the responses of eachspecies to thechangedabiotic conditions found at edges.

Ifspeciesdorespondtoedgestheseresponses are duetochangesin individual microhabitats.

Matlack (1993) detected significant differences in abiotic factors including ^light, temperature, humidity, litter moisture and vapour pressure between the edge interface and the interior of forests. Studies by Bradshaw (1992) indicated that leafdrop, shrub coverand the number of disturbanceadapted^plantsincreased within^the edgezone offorests.

Recent ^{studies of the effects of edges} have revealed that there ^is not a 'typical' edge response oforganisms to edge zones. Sisk and Margules(1993)proposed^sixhypotheticaledge responses whichcan be seen inFigure 1. Figure laillustratesa^habitatgeneralist, wherethereis

no significant change in abundance due ^{to the}

abilityofthisspecies to^utilizeboth^habitatsand the edgezone equally. Figure lbdepicts a habi- tat specialist.These^{speciescan onlyexploit}one habitat and decline rapidlyⁱⁿabundance atthe edge zone of the unsuitable ^habitat. Figure lc represents a habitat generalist edge exploiter.

These species have the ability to exploit both habitats and demonstrate an ^{increase in abun-} dance at the edge zone. Figure Id illustrates a

habitat specialistedgeexploiter,beingfound in

onehabitat, but increasingin abundance atthe edgezonebeforeexhibiting averyrapid decline.

Figure le represents a habitat generalist edge avoider, where these species occurring in both habitatsanddemonstratingdeclinesintheedge regions where they are unable to exploit the environmentsuccessfully. Lastly,Figure Ifisan example of a habitat specialist edge avoider.

Thesespeciesoccurinonlyonehabitatandshow

a decreased abundance as they approach the edge zone. Sisk (1992) was able to recognise thesesixedge responsesinbird communitiesin California.

During the study of edge effects at Tarra- Bulga, what we term a 'cancelling-out effect'

occurswhen analysis iscarried out atthe order level. Thiscancelling-outeffect can be observed when two co-occurring species show opposite edge responses. Our results suggest that at the order level, abundance does not appear to changeat an edge. However, individual species can show dramatic changes in abundance. A number of different edge responses were observed by morphospecies within the four study groups targeted.

Studysite and methods

The study was conducted in Tarra-Bulga National ^Park, 35 km S of Taralgon, approxi- mately 200 km EofMelbourne.The parkison the south-eastern end ofthe Strzelecki Ranges and is well known for its significant stands of cool temperate rainforest and itsprolific popu- lations offerns(Campbell, 1987). It consistsof 1625 hectares with 14 recognised vegetative communities including heavily modified areas (Ashwell, 1991).

623

https://doi.org/10.24199/j.mmv.1997.56.60 28 February 1997

624 P.J. NEVILLEAND D.G. BLACK

a) HABITATGENEHAL1ST b)HABITAT SPECIALIST

1

HABITAT ¹ HABITAT2 HABITAT1 EDGE HABITAT2

c)HABTATGENERALIST ECGEEXPLOITER d) HABITAT SPECIALISTEDGEEXPLOITER

HABITAT1 EDGE HABITAT2 HABfTAT1 EDGE HABITAT2

e) HABITATGENERAL1ST EDGE AVOIDER ^{I) HA8ITATSPECIALIST} EDGEAVOIDER

HABITAT1 EDGE HABITAT2 HABfTAT1 EDGE HABITAT2

Figure 1

.Hypotheticaledgeresponsesforspeciesoccurringnear an abrupt edge betweendifferenthabitat types (Siskand Margules, 1993).

Tarra-Bulga National Parkinbeingrelatively small could potentially suffer through external forces. A numberof roads border andintersect theNationalParkincreasing theedgeto interior ratio. A number of walking tracks and fire vehicle access tracks also intersect the park.

Theseareas areindanger of invasion byforeign plant species beingcarried in by hikers and/or access vehicles. Most roads and walking tracks havethe potential to divide habitats and allow altered abiotic factors to modify the compo-

sition of edgehabitats. The invasion offoreign plant specieswillenhancethisresponseandmay

cause dramatic structural differences to these habitats. If edge effects are occurring much of the park may be affected.

Two88 mtransect lineswereused as aguide forthesampling oftwohabitats,wetsclerophyll forest and cool temperate rainforest. Each line straddled awalkingtrackandran over40m^into

eachhabitattype.Samplingoccurred every8 m

alongthe transectwithfivepitfalltrapsplacedto the east ofthe transect running parallel to the walkingtrack. Soil and littersampleswere col- lected to the west ofthe transect line at each samplingsite. Three sampling techniques were performed in an attempt to reduce the possi- bility of not sampling all patches within the habitats, and transectswere replicated.

All invertebrates were identified to order usingNaumann(1991),Harvey andYen(1993) and Goode (1987). Four detailed studies were

ANIMALSON THE EDGE ⁶²⁵

carriedoutontheordersAraneae,Hymenoptera and Coleoptera and on the Class Collembola.

The Araneae were identified to family using Davies(1986) andthen tomorphospecies. The CollembolaandColeopterawerebothidentified to family using Naumann (1991) and then to individual morphospecies. Lastly, the Formici- dae were identified to genus using Andersen

(¹991 )andHolldoblerand Wilson(1990)before beingsplitintomorphospecies.Thevoucher^col- lectionhasbeen lodgedintheSchool ofZoology

at La Trobe University. Preliminary statistical analysis hasbeen carried out using chi-squared analysis inorderto identifyedgeresponses.No

statistical results are reported here, but some edge responsesaresignified.Errorbarshave not been included on graphs for the sake of sim- plicity.

Resultsand discussion

Inpresentingthe^results,ahypotheticalexample

is first given to illustrate the cancelling-out effect. Figure 2 shows the abundance ofhypo- thetical species A. It can be observed that this species is a habitat A specialist edge avoider, occurringinlargenumberswithintheinteriorof habitatAand decreasinginabundance towards theedgezone.HypotheticalspeciesBisdepicted inFigure 2B. This^{speciesisa habitat}Bspecialist edge avoider (occurring in high abundances within habitat B and decreasing towards ^the edge zone). At the species level, these two responses^areclear. However,attheorderlevel (Figure2C)itcan be observedthat thereappears tobe relativelyequal ^level ofabundance across

all sites along ^{the transect. It} seems that there

is a single edge response, that of a habitat generalist.

Theactualtrends observedwerenotthis clear.

Figure 3 represents the ^total abundance of 19 collembolan morphospecies ^over one of the transects. Sites Ato Crepresent thewet sclero- phyll forest with site Abeing the most ^interior

siteandCbeing^{thesiteclosest tothe}edgezone.

SitesD ^toF represent the cool temperate rain- forest with D ^{beingthe site closest to the edge}

and F beingthe most ^{internal site.} Thesymbol

WT represents the walking track or edge zone over which the transect straddled. It can be observedfromthis figurethat thereseemstobe

little ifany edge effect present. The only trend observed isthat ofthe habitatgeneralist, where the species present are ableto exploitbothhabi- tats and the edge zone successfully. Values of

A)

LUO z<

QZ

CD<

B)

OUJ

z<

QZ

Z) CD<

-l ^{1 '} T

A B C WT

HABITAT A

25

20-1

15-

10-

5-

0- ^{n r} i

A B C WT D E F

HABITAT B C)

LDO Z<

QZ

Z) CD<

HABITAT A HABITAT B COLLECTION SITES

LEGEND

:

[23 HYPOTHETICAL^SPECIES A HYPOTHETICALSPECIES B

Figure 2. A hypothetical example ofthe'cancelling- out' effect where A and B represent hypothetical speciesandCrepresents thetotalabundanceofhypo- theticalspeciesAandB(WT =walkingtrackoredge zone).

626 P.J. NEVILLEAND D.G. BLACK

5O

CO

LU _l _lO o

LLo

LUo

-z.<

Q

CQ<

O

1000 A)

150

WETSCLEROPHYLL FOREST

COOL TEMPERATE RAINFOREST

COLLECTION SITES

Figure 3. Total abundance of 19 collembolan mor- phospeciesoverthe transect (WT = walkingtrackor edgezone).

abundance seem to remain relatively constant overthe entiretransect.

However,whenwelookat individualspecies, forexample, aspeciesfromthefamilyBrachys- tomellidaeand a species ofEntomobryidae we can observetwoindividual trends.Brachystom-

ellidaemorphospecies 1 tends todecrease over the transect towardstheedge zone suggesting a habitat generalist edge avoider (Figure 4A).

Entomobryidae morphospecies 1 (Figure 4B) demonstrates the opposite trend in wet sclero- phyll forest,and a mixed response incool tem- peraterainforest.When thesetwospeciesabun- dancesareplotted together (Figure4C)itcan be observedthatthesetwospeciesresponses canno longer be distinguished. The trends in the wet sclerophyllforestarelost anditcan beassumed thatthese species arewetsclerophyllforest gen- eralistswith noresponse totheedgezone. This additive response is very similar to that found

forthetotal collembolan abundance histogram (Figure 3).

Thecancelling-outeffectcanalsobeobserved within the order Coleoptera. Figure 5 presents thetotalabundancedataforthe 116coleopteran morphospeciesateach siteoverthe entire tran-

III III

C WT D E F

B)

150

100'

50-

~~^I

—

WT

C) _150—i 1

hull

E F

WETSCLEROPHYLL FOREST

COOLTEMPERATE RAINFOREST

COLLECTION SITES LEGEND

:

BRACHYSTOMELLIDAEMORPHOSPECIES 1

ENTOMOBRYIDAEMORPHOSPECIES 1

Figure 4. Abundanceof Collembola. A, Brachystom- ellidaemorphospecies I;B,Entomobryidae morphos- pecies1; and C, Brachystomellidae morphospecies 1

and Entomobryidaemorphospecies 1 (WT =walking track oredgezone).

ANIMALSONTHE EDGE 627

A)

C WT D E F

WETSCLEROPHYLL FOREST

COOL TEMPERATE RAINFOREST

COLLECTION ^SITES

Figure 5. Total abundance of 116 coleopteran mor- phospecies over a transect (WT = walking track or edgezone).

sect.Itappearsfromthis figurethatthisorderis

composed ofspecieswhich are habitat general-

ists. There seems to be no clear edge effect.

However, when thesedata are examined at the morphospecies ^{level, edge trends can be} observed. For example, Curculionidae mor- phospecies 1 could be classifiedas awetsclero- phyll habitat specialist edge exploiter (Figure 6A). It can be observed that there are higher abundances of this morphospecies ^{within this} habitatand anincreaseinabundanceattheedge zone.However,Curculionidaemorphospecies² demonstratesa differenttrend (Figure 6B). ^Itis suggested that this morphospecies ^is a habitat generalistedgeavoider,havingdecreased abun- dances ^at the edge zone. When these two mor- phospecies from the same ^family are plotted together, averydifferenttrend can be observed.

Abundances become more^{similarbetweensites} within each habitat. The ^{family level analysis}

leads to a cancelling-outofspecies trends,mask- ing how individual species are responding to Anotherbeetleexample showsthiscancelling- outeffectalittleless clearly. Staphylinidaemor- phospecies ^{1 is} a cool temperate ^rainforest specialist edge exploiter(Figure 7A). Leiodidae

A B C WT D E F

B)

20-

15'

10-

1 ^{I I 1 1} T"

A B C WT D E F

C)

30-

20-

:iUUI

.

B C WT D

WETSCLEROPHYLL FOREST

COOLTEMPERATE RAINFOREST COLLECTION SITES

LEGEND

:

| CURCULIONIDAE MORPHOSPECIES ¹ CURCULIONIDAE MORPHOSPECIES ²

Figure6.AbundanceofColeoptera. A,Curculionidae morphospecies 1;B,Curculionidae morphospecies2;

andC,both morphospecies of Curculionidae(WT =

walkingtrack oredgezone).

628 P.J. NEVILLEAND D.G. BLACK A)

B)

C)

A B C WT D E F

50'

40-

30-

20-

10-

0-

WT

T

D

70

60 50 40

30 20' 10-

r-'-r

I..

C WT D

WETSCLEROPHYLL FOREST

COOLTEMPERATE RAINFOREST COLLECTION SITES LEGEND

:

STAPHYLINIDAEMORPHOSPECIES LEIDODIDAEMORPHOSPECIES 1

morphospecies 1 could be classified as a cool temperate rainforest specialist showing dra- maticdecreases in wetsclerophyll forest and at theedge ofthe rainforest (Figure 7B). However, when these two species from different families are plotted together (Figure 7C) it can be observed that the differences in abundance betweenadjacentsitesdecreases.Onceagain,by simply adding the abundances for two species together, the individual trends observed at the specieslevel areslightly cancelledout.

Although this cancelling-out effect has been observed withintheCollembolaandColeoptera, the Formicidae remain variable in abundance overthe transect(Figure 8). The cancelling-out effect does not seem to apply for ants in this study. Anumber offactors may explainthis. It

hasbeen demonstrated bynumerousresearchers that particulargroups ofspecies prefer particu- lar environmental conditions. The most abun- dantant atthesesiteswasaspeciesofNotorious whicharegenerallyknown topreferwettersites,

being cool-adaptedspeciesandareoftennoctur- nal foragers (Andersen, 1986). They are often competingdirectlyagainstIridomyrmexspecies and ^it is believed that this has caused the noc- turnal habit to evolve. The second most

LU<

QO

rro

u.

LLo

LUo

-z.<

Q

-z.

CD<

o

h- 700

600-

A B C WT D E F

WETSCLEROPHYLL FOREST

COOL TEMPERATE RAINFOREST

Figure7.AbundanceofColeoptera. A, Staphylinidae morphospecies 1;B, Leiodidae morphospecies 1;and C,bothStaphylinidaemorphospecies1 andLeiodidae morphospecies 1 (WT = walking track or edge zone).

COLLECTION SITES

Figure 8. Total abundance of 12 morphospecies of Formicidae (Hymenoptera) over a transect (WT =

walkingtrack or edgezone).

ANIMALSONTHE EDGE 629

commonant specieswasa species ofMonomo-

rium. whichare knownto preferopen habitats.

Due to the unspecialised behaviour of many

speciesofMonomoriumthey are very successful opportunists (Andersen, 1984). The thirdmost

commonspeciesbelongsto thegenus Iridomyr- mex, offunctionalgroupone, orthe dominant group. These species are abundant, aggressive and are known to monopolise resources (Andersen, 1984). Theyalso prefersunny,open conditions. Competitive interactions between antshasled tothe developmentoftightly struc- turedcommunitiesinthe aridzone(Greenslade,

1979)andmaybepartly responsible for thevari- ation observed overthetransect. Theability of ants to forageoverlargedistanceswillalso playa role inthe variationofabundanceoverthe tran- sect. Due ^{to these} high levels of mobility, the association between the organisms and a par- ticularmicrohabitatwillbe reduced. Therefore, particularantspecieswillbefoundinmoresites

along the transectdue to theirability to forage overlarge areas.Forantssocialinteractionsmay

alsoplay a largerole inthe variable distribution observed alongthe transects.

One other order of arthropods were exam- ined, the Araneae. As for the Collembola and Coleoptera, spiders demonstrate similar abun- dances acrossall sites alongthe transectsat the orderlevel(Figure9).Once^{again, theonlytrend} observed attheorderlevelisthatofthe habitat generalist.

However, whenexaminingthe spiderdataat the morphospecies^levelanumberoftrendscan be observed. Figure 10Adepicts theabundance of Cycloctenidaemorphospecies ¹. Thisspecies

isdeclining towardsthe edge inwetsclerophyll habitatanddecliningtowardstheinteriorofthe cool temperate rainforest. The abundance of Cycloctenidaemorphospecies²can be observed in Figure 10B. This morphospecies ^represents an ^{increase within the} middle sites ofthe wet sclerophyll habitat and ^{increases towards the} interiorofthe cooltemperaterainforest. Figure IOC includes the abundance of both morphos- pecies of Cycloctenidae. The trend within ^this family changes to that ofthe habitat generalist for the cool temperate rainforest.

Other factors play a role inthe cancelling-out effect. Sampling bias, and the difficulty in pro- viding a sampling ^{regime that will account for} patchiness within the environment ^{is also} important.Three sampling^techniqueswereper- formedduringthisstudyinan attempttoreduce the possibilityof notsamplingallpatcheswithin the habitats, and ^{transects were} replicated.

WETSCLEROPHYLL FOREST

COOL TEMPERATE RAINFOREST

COLLECTION SITES

Figure 9. Total abundance of 31 morphospecies of Araneae overa transect(WT= walkingtrack oredge zone).

Generalist trends mayappearatthefamilyand order levels through the combination ofinad- equate sampling procedures, chance variation andreal lifedifferences inindividual responses.

Without statistical analysisit is not possible to sayiftheedge responses observedarereal.How-

ever,preliminarytests using chi-squaredanaly- sis indicate significant trendsforsome species.

These trends are lost ifanalysis is carried out above specieslevel.

Ithas beendemonstratedthata cancelling-out effectoccursatboththefamilyandorderlevelof analysis.Toobservetrendsinecologicalsystems examiningsuchfactorsasedgeeffectsitisessen- tial that classification of organisms must be completedto the species ormorphospecies^level.

Only when this has been done ^{will real} edge responses be identified. The cancelling-out effect discussed here may have wider implica- tionsforecological studies involvinganalysisof colonisation trends of organisms after more widespread disturbances such asfire.

Acknowledgements

ThankstotheMuseumofVictoriaforproviding partial funding for the project through an honours grant. The authors also thank the

630 PJ. NEVILLEANDD.G. BLACK A)

B)

C)

A B C WT D E F

A B C WT D E F

A B C WT D E F

WETSCLEROPHYLL FOREST

COOLTEMPERATE RAINFOREST COLLECTION SITES

LEGEND

:

CYCLOCTENIDAE MORPHOSPECIES1

CYCLOCTENIDAE MORPHOSPECIES2

Figure 10. Abundance ofAraneae. A, Cycloctenidae morphospecies 1;B,Cycloctenidae morphospecies2;

and C, both Cycloctenidae morphospecies 1 and 2

(WT = walkingtrack oredgezone).

anonymous reviewer for helpful comments on the manuscript.

References

Andersen A.N., 1984. Community organization of antsinthe Victorian Mallee. Victorian Naturalist

101(6): 248-251.

Andersen, A.N., 1986. Patterns of ant community

organization in mesic southeastern Australia.

AustralianJournalof^{Ecology 11: 87-97.}

Andersen, A.N., 1991. 'TheantsofsouthernAustralia

— a guide to the Bassian fauna? CSIRO:

Melbourne.

Ashwell, D.A., 1991. A vegetation survey ofTarra- Bulga National Park, Victoria. Arthur Rylah

InstituteforEnvironmentalResearch, Technical ReportSeries 105.

Bradshaw, F.J., 1992. Quantifying edge effect and patch size for multiple-usesilviculture —adis- cussion paper. ForestEcology and Management

48: 249-264.

Campbell, R., 1987.NationalParks ofVictoria,South Australia and Tasmania. Reader's Digest Services: SurreyHills.

Davies, V.T., 1986. Australian spiders (Araneae) —

collection, preservation and identification.

Queensland Museum Booklet 14, Brisbane.

Goode, J., 1987. Insects of^Australia. Angus and Robertson: North Ryde.

Harvey, M.S. and Yen,A.L.,1993.Wormstowasps:an

illustratedguide toAustralia's terrestrial invert- ebrates. OxfordUniversity Press: Melbourne.

Holldobler, B. and Wilson, E.O., 1990. The ants.

Harvard University Press: Cambridge.

Greenslade, P.J.M., 1979. A guide to ants of South Australia. SouthAustralian Museum:Adelaide.

Laurence, W.F., 1991. Edge effectsin tropical forest fragments: applicationsofamodelforthe design of nature reserves. Biological Conservation 57:

205-219.

Matlack, G.R., 1993. Microenvironment variation withinand amongforestedgesitesinthe eastern United States. Biological Conservation 66: 185- 194.

Naumann, I.D., 1991. The insects ofAustralia — a textbook for students and research workers.

MelbourneUniversity Press: Melbourne.

Noss,R.F., 1991. Effectsof edgeandinternal patchi- nessonavianhabitat useinan old-growthFlorida

hammock. Natural Areas Journal11(1): 34-46.

Sisk,T.D., 1992. Distributionof^birdsandbutterflies in heterogeneous landscapes. Ph.D. dissertation, Stanford University: Stanford.

Sisk, T.D. and Margules, C.R., 1993. Habitat edges and restoration: methods for quantifying edge effects and predicting the results of restoration efforts.Pp. 57-68in:Saunders,D.A.,Hobbs,R.J.

andEhrlich, P.R. (eds),Natureconservation3—

the reconstruction of fragmented ecosvstems.

Surrey Beattyand Sons: Perth.