434 Journal of Tropical Forest Science 13(3): 434-449 (2001)

IMPACTS OF LAND USE CHANGE ON FOREST COMMUNITIE WESTERE TH SN I N GHAT SOUTHERF SO N INDIA

Ghazala Shahabuddin*

Salim SchoolAli of Ecology, Pondicherry University, Pondicherry 605014,- India

&

Raui fAI

Aurodam, Auroville, Tamil Nadu - 605101, India

Received May 1999______

SHAHABUDDIN, G. & ALI, R. 2001. Impacts of land use change on forest butterfly communitie Westere th sn i n Ghat southerf so n India studiee W . role plantatiodf th e o n agriculture as a refuge lor the of deciduous forest habitat in the Western Ghats of southern India. Abundance, diversity and composition of the butterflie f foresso t habitat were compared with thos f adjoinineo g lime plantations. Butterfly communities were studied using visual censusing techniques along eleven line transects in forest habitat and five line transects in plantation habitat. Observed butterfly densities wer significantlt eno y differen foresn ti t (47 378plantatio8d ± an ) n transects (401 ± 167). Observed species richness was higher in plantations (38 ± 5) compared to forest habitat (34 ± 8) but this difference was not statistically significant. However, ordination analysis revealed that species composition differed between the two types of habitat with forest specialists being replaced by edge and secondary growth specie plantationsn si . Faunal similarity between plantatio foresd nan t transects swa low, with only 33.7%. The study indicated that although butterfly abundance and species richness of plantations were comparable with those of forest, species composition varied significantly between the two types of habitat.

Key words: Butterfl plantatioy- ndeciduou- s fores diversit- t yspecie- s composition- Western Ghats - Palni Hills

SHAHABUDDIN, G.& ALI, R. 2001. Kesan perubahan penggunaan tanah terhadap komuniti kupu-kupu hutan di Ghats Barat di selatan India. Kami mengkaji penman pertanian ladang sebagai perlindungan bagi kupu-kupu daripada habitat hutan daun luruh di Ghats Barat di selatan India. Kelimpahan, kepelbagaian dan komposisi spesies kupu-kupu daripada habitat hutan dibandingkan dengan kupu-kupu daripada ladang limau nipis yang bersebelahan. Komuniti kupu-kupu dikaji menggunakan teknik bancian secara penglihata sepanjani nd g sebelas transek gari habitai d s t hutan nda lima transek garis di habitat ladang. Kepadatan kupu-kupu yang dicerap adalah tidak

* Present address: Chintan Environmental Research and Action Group, 8/404 EastEnd, MayurVihar lExtn., Delhi -110096, India Journal of Tropical Forest Science 13(3): 434-449 (2001) 435

berbeza dengan berert i transeid k hutan (47 3788transe± n da ) k ladang (401 ±167). Kekayaan spesies yang dicerap lebih tingg ) berbandin ladani 5 d i ± 8 g(3 g habitat hutan (34± 8) tetapi dari segi statistik, perbezaan ini tidak bererti. Bagaimanapun, analisis pengordinatan menunjukkan bahawa komposisi spesies berbeza antara dua jenis habitat tersebut, dengan spesies khusus hutan digantikan dengan spesies tepspesien da i s sekunde ladangi d r . Persamaan fauna antara transek ladann gda transek hutan adalah rendah, iaitu hanya 33.7%. Kajian menunjukkan bahawa walaupun kelimpahan kupu-kup kekayaan uda n spesie ladani sd hutan gda n dapat dibandingkan, komposisi spesies berbeza dengan bererti antara kedua-dua jenis habitat itu.

Introduction

Rapid deforestation all over the globe has caused unprecedented loss of bio-diversity in recent decades (Skol Tuckee& r 1993). Muc thif ho s deforestatioe th o t e du s ni expansion of agricultural activities into previously uninhabited forest areas, resulting formatioe inth habitaa f no t mosai whicn ci h forest fragment locatee sar n a n di agricultural lan matrie dus x (Wilcov 1986. al t ee , Rant . 1998)al t ae . currene Inth t situation necessars i t i , evaluato yt role evariouf th e o s typef so agricultura s habitala lan e r forest-dwellinus dfo t g anima d planan l t species (Vandermee Perfect& r o 1997).Type lanf so d use, that provide foragind gan breeding habita foresr tfo t faun corridorr ao animar sfo l movement usee b dn ca , maintaio t n continuity between forest fragments (Salafsky 1993), whic essentias hi l for the maintenance of biodiversity (Taylor et al. 1993). Butterflies are a useful taxo o studt n n thiyi s contex s the a e sensitivt yar e indicator f changeo s n i s vegetation structure, composition and microclimatic conditions that inevitably occur when forests are converted into agricultural land (Kremen 1994, Sparrow . 1994)eal t . Palne th n iI Hills, locate Westere th n di n Ghats mountain syste southerf mo n India, deciduous forest between 500 and 1500 m asl is rapidly being converted into various types of plantations including those of banana, lime and coffee. In this study, we evaluated the use of lime plantations as habitat for forest butterflies in this area through compariso butterflf no y communitie plantationf so s with those of adjoining forest fragments. The major questions addressed in this study were: (1) What was the effect of forest conversion on the abundance and species diversity of local butterfly fauna? (2) Was the butterfly species composition of lime plantations significantly different from tha adjoininf to g natural forest?

Materials and methods

Study area and vegetation characteristics

The Western Ghats are a mountain chain running south-north close to the western

coas f Indio tcoverin d aan altitudinan ga 250o t l stude asl0 rang0m Th .50 y f eo was carried out at Siruvattukadu Kombei (SVK), a valley covering about 80 km 2 in the Palni Hills of the Western Ghats. The Palni Hills are located within the state of Tamil Nadu between 10° 21' to 10° 25' N and 77° 36' and 77° 44' E (Figure 1). 436 Journal of Tropical Forest Science 13(3): 434-449 (2001)

76° E

10 ° N

9°N

STATE BOUNDARY

SCALE : 1.000.001 : 0

Figur e1 Locatio stude th f yno site , Siruvattukadu Kombei, in the Palni Hills of the Western Ghats meae Th n altitud vallee th 750f s eyi o mreceiveK asl SV .averag n a s 100f eo m 0m of rainfall every year, mos whicf to h occurs between Octobe Decemberd ran . The forest vegetation in SVK consists of & mix of moist and dry deciduous associations, with an evergreen understorey layer in many places. Dominant tree specie aree th an s i includ e Miterophera heyneana, Alphonsea sclerocarpa, Celtis wightii, Sapindus emarginata d Diospyrosan melanoxylon e canopth n i wels ya s Murrayaa l paniculata, Tarenna asiatica d Canthiuman parviflora understoreye th n i . Mangifera indica, Terminalia arjuna and Pongamia pinnata dominate the riparian forest. Nectar sources were rare and scattered in this habitat during the period of study. Canopy trees . arjuna,sucT s ha understorey trees including Gardenia obtusad an riparian trees like Asclepias curassavica wer importane eth t source f nectaso r rfo forest butterflies durin stude gth y period. Mixed lime plantations, are one of the dominant agricultural land uses in the area. In this type of land use, 1-4 m tall lime trees (Citrus aurantifolia) were grown in rows, mixed with useful tree species including silk-cotton (Bombax ceiba), jackfruit (Artocarpus heterophylla) and banana (Musa sapientum). Most plantations harboured a high density of herbaceous weeds such as Stachytarpheta indica, Acanthospermum hispidumimd Tridax procumbens which were flowering throughout the period of study. Along with Lantana camara, an invasive shrub of South American origin, the weeds presented important sources of nectar for butterflies during the period of study. Journal of Tropical Forest Science 13(3): 434-449 (2001) 437

In addition to differences in plant species composition, plantation and forest habitat diffe vegetation i r n structure generaln .I , plantations have fewer vegetation strata, more open canopy and greater density of flowers compared with forest habitat (Shahabuddin 1993). Average density of trees (> 20 cm gbh) along the plantation transect compare2 lowm s si 0 , namely10 r thado t pe t2 alon,1. g forest transects, with 4.2 perlO 2 (ShahabuddiOm n 1993). Average densit shrubf yo saplingd san s (all plants > 1 m in height but < 20 cm gbh) is also lower in plantations (0.3 per 100 m2) compare thao dt t measured along forest transect (Shahabuddi) 2 m 0 s 10 (21 r 5pe n 1993).

Data collection

Butterfly densities were estimate foresn d i plantatio d an t n habitats usine gth line transect method following Pollar Yated dan s (1993). Five 200-m-long line transects wer plantatioe laith n di whil) n P5 habita eo t eleve I (P t n were established in the forest (Fl to Fl 1). The transects were established in three different plantations wideln i t ybu separated portion same th f eso forest continuum. Transects were laid varieta n i terraif yo n including slope d streamsidesan tha o maximue s tth m rang variatiof eo 7 F capture, s typeo F6 n wa , habitattw f so F5 e eac dn i th e f ho Th . transect5 P d an s were located along streamside whil remainine eth g transects were established along valley slopes and bottom. However, all the transects were locate approximatelt da same yth e elevation, namely aslm . 0 Tabl75 , givee1 n sa overvie locationae th f wo vegetationad an l l attribute eacf so h transect.

Table 1 Attributes of vegetation along study transects in Siruvattukadu Kombei

Source: Shahabuddin (1993). H = high, M = medium and L = low. n.m. = not measured 438 Journal of Tropical Forest Science 13(3): 434-449 (2001)

Before the transect counts started, a reference collection of butterflies was built up. Specimens were identified (Evans (1911), Wynter-Blyth (1957), Ugarte & Rodericks (1960), Satyamurti (1966)). Most butterflies coul identifiee db d during transece th t counts butterfliew fe A . s tha t e identifiecoulb t dno d immediately were capture later dfo r identification. A total of 20 counts was carried out along each of the 16 transects on different days. Thus count0 tota,a 32 f lso were undertaken durin studye gth eacn I . h transect, counts were spaced evenly through the three-month study period (24 March-17 June 1992)a singl r Fo e. count walkee ,w e transecdth t slowl 15-2r fo y 5 minutes. During that time, the vegetation on either side was scanned continuously for speciee butterflie Th distancnumbea d . o st an butterfliem l p 5 al u f sf e ro o s seen during these counts were recorded. Counts were made during sunny weather totae Th l . betweenumbe m p. butterflie3 f a.mn ro 9 d an . givea f so n species, seen count0 2 durine th s l alongal gparticulaa r study transect indicatousen s a s d,wa a r of its population density; an assumption that was proven to be viable by Pollard andYates (1993).

Statistical analysis

Species richness was calculated for each study transect (Magurran 1988). Species richnes totad san l abundance wer parem eco d between fores plantatiod tan n transects usin Kruskal-Wallie gth s test non-parametria , c analysi variancf so e test which account unequar sfo l sample sizes (Soka Rohll& f 1981) abundance Th . f eo each specie compares swa d between fores plantatiod an t n habitats, also usine gth Kruskal-Wallis tests orden I . studo rt effecte yth variouf so s physical variables upon butterfly abundance and species richness, linear regressions were carried out using canopy cover, understorey densit tred yean density (Tabl . Speciee1) s composition of communitie differenn si t habitat compares swa d using detrended correspondence analysis (DCA) techniqua , e which assumes normal response function specief o s s environmentao t l gradients (Jongma . 1995) al e assumptio t ne Th . f normano l response functions of species to environmental gradients is one that is considered biologically robust (Jongma 1995). al t ne . DCA was carrie usint d ou ordinatio e gth n package PC-ORD (Multivariate Analysis of Ecological Data 1997, Version 3). In addition Fortraa , n program called GDIS (Landscape Ecology Laboratory, Duke University uses comparo wa d)t e species compositio typeo tw habitatf so e th f no .

Results and discussion

Butterfly abundance speciesand richness

A total of 8628 butterflies belonging to 94 species in 5 families was sighted during the surveys. In addition, eight species were recorded before the counts began but were not seen subsequently. A classified list of the butterfly species seen during the study is given in Appendix 1. Of the 8628 individuals, 1224 individuals (14%) Journal of 9 Tropical43 Forest Science 13(3): 434-449 (2001)

coul identifiee db d onl familo yt y leve werd an l e therefore excluded froe mth analysis. A total of 34 species was seen 10 times or less across all transects. These species were also excluded from the analysis to avoid erroneous results associated witsamplw hlo e size. Finally, 7263 individuals belongin specie9 5 o gt s were used for the present analysis. The total number of butterflies for each species seen along each transect durincount0 2 e givegs sth i Tabln i . e2 Observed butterfly densities wer significantlt eno y differen foresn i t t (47 3788± ) plantatiod an n transects (40 1167± ) (Kruskal-Wallis test statisti 0.03= ) , c(H df=1 , 0.87)< p . Plantations harboure dgreatea r numbe butterflf ro y species compared foreso t t habita tthit (plantationsbu s forests ) d differenc8 an t ± 5 4 no ± 3 : s 8 3 :ewa statistically 0.21)significan< p , .1 1.57= = f H ,d ( t Overall, the results indicate that butterfly densities and species do not differ significantly between forest and plantation habitats. The effects of habitat disturbances such as fragmentation (Brown & Hutchings 1997), selective logging (Hill et al. 1995) and plantation agriculture (Watt et al. 1997) on butterfly diversity variee ar complexd dan . However t appeari , s that small-scal levew lo l r foreseo t disturbances as seen in coffee plantations, creation of small tree fall gaps and partial forest clearing, often result in increased local butterfly diversity (Janzen 1987, Spitze . 1997al t e r, Watl 1997a t e t , Woo Gillmad& n 1998). Small-scale disturbances encourage light penetratio d creatnan a localle y heterogeneous environment suitable for butterfly activity, even encouraging the growth of certain larval host plants and increasing the local abundance of flowers. However, more severe disturbance suc s large-scalha e loggin r cleago r fellin resuly locagma n i t l loss of diversity even after a period of regeneration (Hill et al. 1995). At SVK, lime plantations were typically small (up to a hectare), located close to the forest and therefore migh consideree b t small-scale b o dt e disturbance t havsno thaed di t significant effects upo t locanne l diversity. In particular, floral diversity could be a major reason for the high diversity of butterflies recorded insid plantationse eth variete Th . herbaceouf yo s weeds that were flowering durin stude gth y period attracted numerous butterflie feedingr sfo , including some from forest habitat contrastn I . , forest habita isolated ha t d dan sparse flowering sources. The nectar sources utilised by butterfly species in the different habitat types were described in detail by Shahabuddin (1997b).

Vegetation structure and butterfly abundance

Linear regressions indicated that observed butterfly abundance along transects was not related to any of the variables related to forest structure such as canopy cove 0.049= 2 0.67r(r = 0.43 1,13)< ,F = ,p f ,d , understorey densit 0.0387= 2 y(r , 0.52= 0.48 1,13< F = ,p f trer ,d )o e densit 0.108= 2 1.58y(r = 0.23 1,13)< ,F = ,p f ,d . However, transect location (streamside or interior forest) was a strong determinant of butterfly abundanc 0.5514= 2 e (r 15.98 = ,0.001)F < ,p . Similarly, transect location had a significant effect upon species richness (r2 = 0.5992, F= 19.44, p< 0.0007) but there was no indication that butterfly species richness was determined by any of measuree th d variables relate vegetatiodo t n structure (canopy cover 0.0160= 2 :r , Journal of Tropical Forest Science 13(3): 434-449 (2001)

Table 2 Ranked scores of butterfly species on first and second DCA axes (in increasing order) Journal of Tropical Forest Science 13(3): 434-449 (2001) 441

F = 0.21, p < 0.65, df = 1,13; understorey density: r2 = 0.0612, F = 0.85, p < 0.37, df=l,13; tree density: r2 = 0.309, F = 5.81, p < 0.03, df = 1,13). Thus the results of regression suggest that the presence of water is perhaps the most important factor affecting local species richness and abundance of butterflies. Earlier analysis also indicated tha higa t h proportion (79%speciee th f o ) s observes recordei K SV n di n d i streamside habitat (Shahabuddin 1997b). High species richnes butterfld an s y abundanc streamsidn ei e habitat coul attributee db neee mud-puddlinr th ddo fo t g among most butterfly species which intensify during the dry season.

Species distribution

Of the 60 species, 28 (47%) showed no significant difference in abundance between fores plantatiod tan n transects observes wa t I . d tha (25%5 t1 ) species were significantly more abundant in plantation habitat while 17 (29%) were more abundant in forest habitat (Appendix 1). Species such as Kallima horsfieldi, Papilio polymnestor, patnia, Cyrestis thyodamas and Parantica aglea were more abundant along forest transects othee th n r O hand. , certain species suc s ha Junonia hierta, Danaus chrysippus and Chilades laius preferred open agricultural areas and were generally restricte themo dt . However, nearly halrecordee fth d species, including Pachliopta hector and Eurema hecabe, did not show significant variation in abundance between plantatio foresd nan t habitats. The restriction of butterfly species to certain habitats could be explained by combinatioa r o e on thref no e important factors locae :th l abundanc theif eo r larval host plants (Gaonkar, pers. comm.), adult feeding sites and their preference for a certain leve canopf lo y shad whico et h the physiologicalle yar y adapted. Highest densitie f moso s t butterfly species were generally seen where these conditions overlapped. For example, the abundance of Papilio demoleus in plantation habitat was possibl boto t attractio s e hit ydu limr nfo e shrub whicn s(o ovipositedt hi d an ) its tendency to fly in open areas. Similarly, Talicada nyseus was only seen close to its larval food plant, Kalanchoe spp., found growing in a few spots on the sandy banks of the stream. Exact reasons for the habitat preferences of individual species can foune b onlt detailedou y y b d stud theif yo r biologica movemend lan t patterns.

Species composition

DCA revealed a clear separation of plantation and forest transects along the first axis (Figure 2). The first DCA axis accounted for 42.3% of the variation in butterfly community composition, indicating that fores plantatiod an t n transects differed substantially from each other. An examination of first axis DCA scores showed that species reportedly restricted to forest habitat, such as Cupha erymanthis, Dophla evelina and Mycalesis patnia (Larsen 1987c), were replaced by species that were characteristic to more open areas, such as Chilades putli, Lampides boeticus, Papilio demoleus and junonia hierta (Larsen 1987a, b, c; Table 3). The second axi Figurn si accountee2 additionan a r dfo l 16.9 variatio%f o specien i s composition among the study transects. The clustering of transects F5,F6, F7 and 442 Journal of Tropical Forest Science 13(3): 434-449 (2001)

60 O 40 20 \ o

0 20 40 60 80 100 120 140 160 180 200 220 Axis 1

• Forest (interior) O Forest (riparian) A Plantation (interior) A Plantation (riparian)

Figure 2 Results of cletrended correspondence analysis (DCA) of the 16 study transects buse butterfln do y species composition

P5 along the second DCA axis indicated that the variation in species composition along the second axis may be related to the presence of flowing water as these were transecte th s located alon streamsidee gth . Tabl confirme2 s that several speciee sar much more common along streamside forest transects than non-stream forest transects. These species include Graphium doson, Neptis hylas, Jamides bochusd an Jamides celeno. Community similarity analysis using GDIS indicated tha meae tth n within group similarity among forest transects was 52% while that among plantation transects was 55.4% meae Th . n between group similarit specien yi s composition (usinl gal transects mucs wa ) h lower with 33.7%. Randomisation tests showed that within group similarities were significantly greater than between group similarities (p = 0), thereby indicating that there was a significant difference in overall species composition between forest and plantation habitats.

Plantation agriculture and biodiversity

Plantation agricultur bees agriculturae eha th nf o considere e on le lanb o dt d uses that are relatively hospitable to forest faunaand useful as buffer zone vegetation, providing both primary and secondary habitats for species. r exampleFo , shade coffee plantation Indin i s a (Shahabuddin 1997a), shifting cultivation plots in the Amazonian rain forest (Dufour 1990) and forest gardens in Indonesia (Salafsky 1993) have been recorded to harbour a significant proportion e nativoth f e vertebrate faun f surroundino a g forests. However n equaa , l number of studies have recorded impoverished faunas in plantations in comparison with natural forests, including traditional agroforests in Indonesia (Thiollay 1995) Journal of Tropical Forest Science 13(3): 434-449 (2001) 443

and coffee plantations in Guatemala (Greenberg et al. 1997). It appears that the faunal similarity of between human modified habitat and natural forest depends primarily upon the corresponding degree of similarity in vegetation structure, plant species compositio diversityd nan r exampleFo . maintenance th , diversa f eo e tree composition, presence of canopy shade and a multi-layered vegetation structure in coffee plantations improve suitabilite dth plantatiof yo n agricultur nativr efo e fauna (Perfecto et al. 1996). However, the few studies that had extended such investigations to communities showed that effects of habitat conversion thesn o e animalmore b y e sdramatima c than those observe mammalr dfo d san birds (Didham 1997, Wat1997). al greatet e te .Th r sensitivit insectf yo changeo st s in habitat variables can be attributed to their narrower specialisation on plant specie dependencd san varieta n eo differenf yo t micro-habitat thein si r life cycle. are also more vulnerable to changes in microclimate than vertebrates theio t e r poikilothermidu c natur thereford ean more ear e affecte changey db n si forest structure including canopy cover and understorey density (Greatorex-Davies 1993. eal t , Thomas 1994) e presenTh . t study confirmed previous observations thaconversioe th t foresf no plantatioo t n agricultur resuly edrastima n ti c alterations in the composition of arthropod communities, although net diversity may not be significantly reduced.

Sampling design

Certain problems with sampling design need to be considered while interpreting result e studye th exampler th f so Fo . questionabls i t ,i e whethe considen ca e rw e rth forest transects as true controls for comparison with plantation habitat. The forest degrades wa habitaK somn dSV i n i te place exposed san moderato dt e levelf so human disturbance, importantly, those caused by firewood and fodder extraction. In several places, herbs and shrubs that are indicative of disturbance in moist deciduous forest habitat were seen exampler Fo . , Lantana camara d Scutiaan spp. (from drier scrub habitats) were found scattered throughou foreste th t . Such sampling problems are unavoidable when studying the effects of human disturbance pervasive th o t e e du influenc humaf eo n activit arey an a n y(Freesi e 1997). Seasonal variation in butterfly movement patterns also needs to be considered when interpreting the results of the present study. Studies in seasonally dry areas indicate that movement pattern f butterflieso changy seasonaa ma s n eo l basis (Braby 1995)seasony r exampledr Fo e . ,th severan i , l specie e knowar s o nt aggregate in sheltered riparian areas to avoid dessication (De Vries 1987). Dry season aggregation may affect the results of the present study; more individuals are likely to be recorded inside forests than would be during other times of the t seasonwe e year,th reproductivn I . e need f butterflieo s restricy ma s t their distributio areao nt s with relatively higher densitie theif so r host plant othed san r plant whicn so h they oviposit. Temporal trend floran i s l densit distributiod yan n are also likely to alter butterfly diversity and abundance, given the scarcity of nectar source foresn si te hig habitath hd degrean t mobilitf eo mosf yo t species while seeking nectar. Long-term studies covering an entire annual cycle are required completelo t y investigate habitat selection among butterfly species. 444 Journal of Tropical Forest Science 13(3): 434-449 (2001)

Conclusion

Despit limitationss eit r stud,ou y shows that butterfly communitie mixef so d lime plantations differ considerably from thos deciduouf eo s forests extensioe Th . f no such agricultural practice Palne th n isi Hillhavy sema deleterious effects upon the local biodiversity of insects. We, therefore, recommend that other types of agricultural land use be investigated with regard to their conservation value so that more diversity-friendly land use can be propagated in this part of the Western Ghats.

Acknowledgements e stud s carrieTh wa yt witou dh financia d infrastructuraan l l support from Development Alternatives Palne Delhth w d i Hill,Ne ian s Conservation Council, Kodaikanal through the Sustainable Development Programme. We are grateful to B. Ramesh,N.R . Parthasarth Rajendrad yan helr nfo p with plant identification. We also thank H. Gaonkar and K. Kunte for help with updating the butterfly nomenclature.

References

BRABY . 1995F . M , . Reproductive seasonalit tropicayn i l satyrine butterflies: season strategiey dr e th . r sfo Ecological Entomology : 5-1720 . BROWN, K. S. & HUTCHINGS, R. W. 1997. Disturbance, fragmentation and the dynamics of diversity in Amazonian forest butterflies . 91-11Pp . Laurance0n i Bierregaard& . F . ,W . (Eds.,R ) Tropical Forest Remnants: Ecology, Management and Conservation of Fragmented Communities. University of Chicago Press, Chicago. DEVRIES, P.J. 1987 Butterfliese .Th of Costa Rica Theird an Natural History. Princeton University Press, Princeton. 327 pp. DIDHAM . 1997K . overvieR n , .A invertebratf wo e response foreso st t fragmentation . 303-32Pp . 0n i Watt, A. D., Stork, N. E. & Hunter, M. D. (Eds.) Forests and Insects. Chapman & Hall, London. DUFOUR, D. L. 1990. Use of tropical rain forests by native Amazonians. Bioscience40(9): 652-659. FREESE,C . 1997 H . "Use Losr .o Th et debate" I eIt l-48inFreese,Cp .P . (Ed..H ) Harvesting Wild Species. Implications for Biodiversity Conservation. John Hopkins University Press, Baltimore. EVANS, W. H. 1911. A list of butterflies of the Palni Hills. Journal of the Bombay Natural History Society 20: 380-391. GREATOREX-DAVIES.J MARRS , SPARKS& . 1993, HALL. H N. . L H. . e influenc. . R , Th M .T , f shadeo n eo butterflies in rides of coniferised lowland woods in southern England and implications for conservation management. Biological Conservation 63: 31—41. GREENBERG, R., BICHIER, P., ANGON, A. C. & REITSMA, R. 1997. Bird populations in shade and sun coffee plantation Centran si l Guatemala. Conservation Biology 11(2): 448-459. HILL, J. K., HAMER, K C., LACE, L. A. & BANHAM, W. M. T. 1995. Effects of selective logging on tropical forest butterflies in Bum, Indonesia. Journal of Applied Ecology 32: 754—760. JANZEN, D. H. 1987. diversity of a Costa Rican dry forest: why keep it and how? Biological Journal of the Linnean Society 30: 343-356. JONGMAN,R . G.,TEH . R BRAAK,C.J.F VA& . N TONGEREN . (Eds.R . F . ) O ,1995 . Data Analysis Communityn i and Landscape Ecology. Cambridge University Press, Cambridge. pp 9 .29 KREMEN, C. 1994. Biological inventory using target taxa: a case study of the butterflies of Madagascar. Ecological Applications 4(3): 407-422. LARSEN, T. B. 1987a. The butterflies of the Nilgiri mountains of southern India (: Rhopalocera). Journal Bombaye oth f Natural History Society (1)4 8 : 26-54. Journal of Tropical Forest Science 13(3): 434-449 (2001) 445

LARSEN, T. B. 1987b. The butterflies of the Nilgiri mountains of southern India (Lepidoptera: Rhopalocera). Journal of the Bombay Natural History Society 84(2): 291-316. LARSEN, T. B. 1987c. The butterflies of the Nilgiri mountains of southern India (Lepidoptera: Rhopalocera). Journal Bombaye oth f Natural History Society 84(3): 560-584. MAGURRAN, A. E. 1988. Ecological Diversity and its Measurement. Princeton University Press, Princeton. 179pp. PERFECTO, I., RICE, R. A., GREENBERG, R. & VAN DER VOORT, M. E. 1996. Shade coffee: a disappearing refug biodiversityr efo . Bioscience : 598-60846 . POLLARD, E. & YATES, T. J. 1993. Monitoring Butterflies for Ecology and Conservation. Chapman & Hall, London. 274 pp. RANTA, P., BLOM, T., NIEMELA,J.,JOENSUU, E. & SITONEN, M. 1998. The fragmented Atlantic rain forest of Brazil: size, shap distributiod ean foresf no t fragments. Biodiversity Conservationd an 385: 7 - 403. SALAFSKY . 1993N , . Mammalia buffea f o e r nzonus e agroforestry system bordering Gunung Palung National Park, West Kalimantan, Indonesia. Conservation Biology 928-933: 7 . SATYAMURTI, S. T. 1966. Descriptive Catalogue of the Butterflies in the Collection of the Madras Government Museum. Government of Madras, India. 241 pp. SHAHABUDDIN . 1993G , . Butterflie indicators sa habitaf so t typ foresd ean t conditio Siruvattukadn i u Kombei, Lower Palni Hills, Western Ghats. M.Sc. thesis, Pondicherry University, . Indiapp 5 .8 SHAHABUDDIN, G. 1997a. Preliminary observations on the role of coffee plantations as avifaunal refuges Palne th n i Wester e Hillth f so n Ghats. Journal of the Bombay Natural History Society : 10-2194 . SHAHABUDDIN, G. 1997b. Habitat and nectar resource utilisation by butterflies found in Siruvattukadu Kombei, Palni Hills, Western Ghats. Journal Bombaye oth f Natural History Society : 423-42894 . SKOLE, D. & TUCKER, C. 1993. Tropical deforestation and habitat fragmentation in the Amazon: satellite data from 197 19888o t . Science 260: 1905-1910. SOKAL, R. R. & ROHLF, F.J. 1981. Biometry. Freeman, New York. 859 pp. SPARROW , EHRLICH MURPHY, SISK& D. R. . . . 1994 R D . T , .. . H , P D ,, Technique d guidelinesan r sfo monitoring neotropical butterflies. Conservation Biology 8(3): 800-809. SPITZER JAROS,J.K., , , HAVELKA & LEPS J. , J. 1997, . Effect small-scalof s e disturbanc butterflon e y communitie Indo-Chinesn a f so e montane rain forest. Biological Conservation 9-15: 80 . TAYLOR , FAHRIGD. . P , ,HENEIN L. , MERRIAM& . K , . 1993G , . Connectivit vitaa s yi l elemen f landscapto e structure. Oikos 68: 571-573. THIOLLAY.J. 1995. The role of traditional agroforests in the conservation of rain forest bird diversity Sumatran i . Conservation Biology 9(2): 335-353. smaly THOMASl Wh . 1994 cold-bloodeA . . ,J d insects pose different conservation problem birdo st n si modern landscapes. Ibis 137: S112-S119. UGARTE RODERICKS& . ,E . 1960,L . Butterflie Palne th f so i Hills complementar,a y list.Journal of theBombay Natural History Society 57: 270-277. VANDERMEER, J. & PERFECTO, I. 1997. The agroecosystem: a need for the conservation biologist's lens. Conservation Biology 11: 591-592. WATT, A. D., STORK, N. E., EGGLETON, P., STIVASTAVA, D., BOLTON, B., LARSEN, T. B., BRENDELL, M. J. D. & BIGNELL. 273-28 Pp . 1997E Hunter.. Stork, & WattD 6, n . . i (Eds. D. E D . . . ,A , N ,M ) Forestsd an Insects. Chapma Halln& , London. WILCOVE, D. S., McLELLA, C. H. & DOBSON, A. P. 1986. Habitat fragmentation in the temperate zone. Pp. 237-256 in Soule, M. E. (Ed.) Conservation Biology: Science of Scarcity and Diversity. Sinauer Associates, Massachusetts. WOOD GILLMANe effect& . Th . B 1998 , disturbancf P o s. M , foresn eo t butterflies usin methodo gtw s of sampling in Trinidad. Biodiversity and Conservation 7: 597-616. WYNTER-BLYTH, M. A. 1957. Butterflies of the Indian Region. Bombay Natural History Society, Bombay. 52. 3pp Appendix 1 Abundance of butterfly species along the study transects

* = Seen outside of count hours, P = more abundant in plantation habitat, F = more abundant in forest habitat. 0 = No difference between forest and plantation habitats, - = not analysed because of insufficient samples.

(Continued) Appendix 1 - continued

Total number of butterfly Habitat Taxon Common name Forest habitat Plantation habitat preference 5 P 4 P 3 P 2 P I P l Fl 0 F1 Fl 9 F F2 8 F F3 7 F F4 6 F F5

* = Seen outside of count hours, P = more abundant in plantation habitat, F = more abundant in forest habitat. differenco N = 0 e between fores plantatiod an t n habitats t analyseno = - , d becaus insufficienf eo t samples. (Continued) Appendi continue- x1 d

See*= n outsid f couneo t hours mor= P , e abundan plantation ti n habitat mor= F , e abundan foresn i t t habitat. 0 = No difference between forest and plantation habitats, - = not analysed because of insufficient samples. (Continued) Appendix 1 - continued

See= * n outsid counf eo t hours mor= P , e abundan plantation i t n habitat mor= F , e abundan foresn ti t habitat. differenco N = 0 e between fores plantatiod tan n habitats t analyseno = ,- d becaus f insufficieneo t samples.