Composition Of Butterfly Communities In Papua New Guinea Biology Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Composition of butterfly communities (Lepidoptera) along a successional gradient in a lowland rainforest of Papua New Guinea

Abstract: Despite the numerous ecological studies carried out on tropical butterflies, ecological data sets for New Guinea butterflies are rare. A survey was conducted in the Wanang Wildlife Management Area of Papua New Guinea to quantify the species richness, community composition and habitat preference of lowland rainforest butterflies in (1) open area habitats; (2) secondary forest habitats; (3) primary forest habitats; and (4) river/stream habitats. Twenty transects representing these habitats in the study area were chosen: five transects for each habitat, with a length of 300 m for each transect. A total of 3,411 individuals of 176 species were recorded from the various habitats. Results from this study showed differences in the mean number of species per 300 m transect between the sampled habitats. More butterfly species were recorded in the river/stream habitats (mean 65 species) with the primary forest habitats having the least number of species (mean 47 species). Community composition of butterflies feeding in the different habitats was comparatively similar between the secondary forest habitats and primary forest habitats; on the other hand, butterfly communities between the open area habitats and the stream/river habitats were clearly separated despite having the same number of species. Comparing community composition of butterflies recorded along a successional gradient, butterfly communities occurring in the open area habitats were different compared to the secondary forest habitats and primary forest habitats which were mutually relatively similar. Habitat preference was also examined and showed that many species respond strongly to disturbance, showing clear preference to either disturbed or undisturbed forest habitats. These preferences can be used in predicting butterfly community response to disturbance, with species restricted to undisturbed habitats having the most important conservation value.

Keywords: Butterfly ecology, Forest disturbance, Habitats, Hesperioidea, Papilionoidea, Papua New Guinea, Successional gradient.


Butterflies are possibly the best group for assessing and monitoring patterns of terrestrial arthropod diversity (Kremen, 1992; Kremen et al., 1994). Butterfly biology and taxonomy are well known (Gilbert & Singer, 1975; Vane-Wright & Ackery, 1984) and an estimated 90% of species are described (Robbins et al., 1996). Butterflies as a group feed on a wide range of Angiosperms and occasionally other plants or animals (Cottrell, 1984; Ehrlich & Raven, 1964; Singer & Mallet, 1986; Singer et al., 1971) in addition to occurring in many habitats, ranging from disturbed to pristine areas (Brown Jr., 1996; Brown Jr. & Hutchings 1997; Kremen et al., 1993; Thomas, 1991). In fact, the only major regions of the world in which butterflies do not occur are the ice-covered poles. Every other major landmass has a more or less individual fauna (Parsons, 1999).

Butterflies are a large group, with twice as many species as terrestrial birds and about three times the number of mammals, reptiles, dragonflies, mosquitoes, termites, or tiger beetles (Robbins & Opler, 1997). The use of butterflies as indicators in conservation planning has been the focus of authors for several years (Brown Jr., 1991; DeVries et al., 1997; Ehrlich & Murphy, 1987; Kremen et al., 1993; Nelson & Andersen, 1994), and many of the advantages of butterflies as biodiversity indicators are summarized in McGeoch (1998).

Experienced entomologists in the tropics commonly recognised that butterfly compositions, diversity and abundance vary greatly among different forest habitats and along altitudinal gradients. In a northern Vietnam study, Spitzer et al. (1997) pointed out that there were significant differences in butterfly communities and diversity in different successions of montane forests. Furthermore, studies on the diversity of fruit-feeding nymphalid communities in intact and disturbed forests in Ecuador showed that species abundance distributions from intact and disturbed forest areas had significantly different variance (DeVries et al., 1999).

Authors have generally considered disturbance to be one of the most important determinants of ecological community composition (Grime, 1979; Pianka, 1974; Southwood, 1988). Additionally, studies have shown that habitat preferences are closely related to the life history strategies of respective species; as the geographical range is also influenced by the species strategy, the correlations of species geographical range and its habitat preferences are expected and have been documented (Leps & Spitzer, 1990; Novotny, 1991; Spitzer & Leps, 1988). At present, many data sets signifying the influence of disturbance are available for plant communities in temperate regions. In contrast, data from tropical regions are uncommon and quantitative data on insect communities are uncommon also (Spitzer et al., 1993).

In this study emphasis is placed on PNG butterflies. While the biology and taxonomy of butterflies in PNG is well documented (Forbes, 1977; Parsons, 1983, 1984a, 1984b, 1986, 1991, 1999; Sands, 1979, 1981; Szent-Ivany & Carver, 1967), little is known about their ecology. Moreover, the majority of ecological studies of tropical butterflies have mainly been confined to tropical Africa, South East Asia and the Neotropics thus leaving gaps in the data sets for PNG.

Since ecological studies of PNG butterflies are very few (e.g. Bowman et al., 1990), this study aimed at quantifying species richness, community composition and habitat preferences of butterflies feeding in different habitats and also along a successional gradient to fill some of the gaps. The study examined the following questions: (1) Are there any differences in species richness between the main habitat types in the study area? (2) How similar are the butterfly communities between the main habitat types in the study area? (3) How similar are the butterfly communities along a successional gradient in the study area? (4) What is the relationship between the degree of disturbance intensity and habitat preference of butterfly species in the study area?

The information gathered from this study is fundamental in providing ecological data for the conservation of PNG butterfly communities, based on a comprehensive checklist of species and statistics of their ecological determinants.


Study site

This study took place at the Wanang Wildlife Management Area (WWMA) (05°15'S, 145°16'E) in the Madang Province of Papua New Guinea (Fig. 2). The study site is situated within a floristically mixed lowland evergreen rainforest and average temperature (26.7 °C annual mean) and rainfall (3,760 mm annually) are typical characteristics of tropical regions that experience distinct wet and dry seasons (Whitmore, 1998). The WWMA covers approximately 10,018 ha of protected unlogged forest but in some areas, traditional practices of shifting cultivation (Lea, 1976; Shearman et al., 2008) have created pockets of secondary forests within the protected area. The WWMA is also surrounded by large areas of production forest (Ramu Block 1), most of which will be "selectively" logged once operations start.

Habitat and transect descriptions

I selected 4 habitat types (open habitats, secondary forest habitats, primary forest habitats and river/stream habitats in primary forest) that represented the main habitats in the WWMA to investigate the species richness, community composition, and habitat preference of butterflies (Fig. 3). In each habitat five 300 m long transects were established along existing paths and trails as well as watercourses. The first three habitats form a typical successional series.

Open area habitats (transects 1-5) comprised of completely cleared forest and an old unused road; all of which had virtually no canopy cover. Most of the plants in these areas were mainly shrubs and grasses reaching to about a meter in height. Secondary forest habitats (transects 6-10) comprised of 6 to 9 year old abandoned gardens following shifting cultivation which had some amount of canopy cover ranging from 15%-40%. Most of the plants in the secondary forest habitats were in the family Euphorbiaceae, Moraceae, and Ulmaceae. Primary forest habitats (transect 11-15) comprised of trees having a reasonably closed canopy cover 70% and greater along with three distinct vegetative layers typical of tropical primary forests. See Damas (2008) for full description of floristic composition of the study area. River/stream habitats following primary forest (transects 16-20) varied in terms of canopy cover but strictly followed watercourses. Plants along this habitat were somewhat mixed: secondary plant species colonised the edges of the watercourses while primary plant species dominated the inland regions of the left and right riverbanks.

Sampling period and sampling times

The monthly surveys lasted from March to September 2008 with 10 days in each month thus cumulating to 70 days in total. During the survey 280 sets of data were collected. Butterflies were sampled along transects using methods similar to those described for butterflies in temperate regions by Pollard (1977).

Each day four transects were sampled when weather permitted. The first two transects were sampled during the morning half of the day usually between 0900-1100 hours local time. This was then followed by the sampling of the other two transects during the afternoon half of the day usually between 1300-1500 hours local time. Therefore after the first five days of sampling (day 1-5), all twenty transects would have been sampled. Consequently, to get a better idea of the species richness and composition for each transect per habitat type, it was necessary to reverse the order of sampling for the next five days (day 6-10); meaning what was sampled in the morning during the first five days (day 1-5), was sampled in the afternoon. Moreover, what was sampled in the afternoon during the first five days (day 1-5), was sampled in the morning.

Butterfly recording and identification

Walking at an even pace, all butterflies (Hesperioidea and Papilioidea) seen were recorded within an imaginary belt of 5 m on both sides of a transect and 7 m above the ground; the time needed for completing each transect walk was approximately 20-30 min. Species that could not be identified in the field (Lycaenidae and Hesperiidae) were collected with a hand-held net, then identified in the hand and released or killed and taken as voucher specimens to allow for better identification later in the laboratory. The identification and nomenclature of butterflies followed Parsons (1999). All dead specimens were kept at the New Guinea Binatang Research Center laboratory for referencing purposes.

Data analysis

The analysis was performed using complete samples from each transect. The completeness of the survey was assessed by constructing a species accumulation curve to show the rate of new species discovery during the survey. The number of total species expected in the study area was estimated using the Chao 1 richness estimator (Chao, 1984; Colwell & Coddington, 1994) and executed with the EstimateS program (Colwell, 2006). Mean species richness between the habitats was tested using one-way ANOVA with the Tukey test and executed with the SigmaStat statistical software (SPSS, 1997).

Data on community composition were analysed by means of CCA ordination (ter Braak, 1986) and performed using CANOCO version 4.51 (ter Braak & Smilauer, 1998). Canonical correspondence analysis (CCA) is a multivariate method for relating the community species composition to underlying environmental variables. The CCA is a method based on the assumption of unimodal response of species abundance to environmental gradient and uses weighted averaging for calculation of ordination scores (Leps & Smilauer, 2003). In all compositional analyses, the recorded species were taken as species data and different habitats as environmental data.


Species richness

During this study 3,411 individuals representing 176 species from 90 genera in the WWMA were recorded. Many species of butterflies were rare in the samples: 28 species were represented by only a single individual, whereas only 23 species had sample size N > 10 individuals, and 48 species N > 20 individuals. Nymphalid butterflies represented 41% (72 species) of the total surveyed fauna, followed by lycaenids at 34%, (59 species), hesperiids at 14% (24 species), and papilionids 7% (12 species), while pierid butterflies were the least represented in the study area with only 5% (9 species). The species sampled and their abundance in the 20 transects is presented in Appendix 1.

All 20 transects from the study site provided a sufficient estimate of the local species richness as the species accumulation curve appeared to have reached an asymptote (Fig. 4). The observed species richness of 176 was approaching the estimated total species richness of 196 estimated by the Chao 1 richness estimator.

Species richness in particular habitats ranged from 45 to 81 species (mean 65) in the river/stream transects, from 53 to 66 species (mean 60) in the open area transects, from 49 to 66 species (mean 59) in the secondary forest transects, and from 41 to 64 species (mean 47) in the primary forest transects. Mean butterfly species richness differed significantly among the habitats (ANOVA with Tukey test, P < 0.05). Species richness was highest in river/stream habitats and declined in order open area habitats, secondary forest habitats and finally primary forest habitats (Fig. 5). Results of pair-wise comparisons of species richness between habitats showed that the only significant difference was between the primary forest habitats, characterised by the lowest richness, and river/stream habitats, characterised by the highest richness, while other pair-wise differences between the habitats were not significant.

Community composition and habitat preference

The importance of having different habitats in any given area is essential for maintaining biodiversity. Theoretically, different habitats have different communities of vertebrates and invertebrates. The CCA performed on species occupying the different habitats revealed a clear separation between the community in open area habitats and river/stream habitats whereas the community between secondary forest habitats and primary forest habitats was comparatively similar (Fig. 6). In this analysis each habitat was coded as one variable.

The intensity of habitat disturbance is an important determinant of species composition. It was analysed using CCA analysis where the first axis captured differences in community composition of butterflies in response to disturbance, while the second axis corresponds to differences caused by other factors - (highest in open areas, intermediate in secondary forest and lowest in primary forest habitats). The analysis revealed important effect of disturbance on community composition of butterflies as the first axis explained 20.6% of the total variability in butterfly composition (P < 0.01, Monte Carlo test). The open area habitats were particularly well separated from the secondary forest and primary forest habitats which were comparatively similar (Fig. 7). The first axis coordinate obtained for butterfly species in this analysis also characterises the distribution of each species with respect to disturbance, as low scores correspond to the tolerance for high disturbance in open areas, while high scores indicate the preference for undisturbed primary forest. This analysis thus permitted ranking butterfly species according to their sensitivity to disturbance (Fig. 8).


Species richness

The habitat-heterogeneity hypothesis - states that large areas hold more kinds of habitats and therefore support more species (Williams, 1964). This study sampled the main habitats that were in the study area and a total 176 species out of an estimated 196 species were recorded for the WWMA (Fig. 4). This is approximately 21% of the identified species in PNG (Parsons, 1999). Particularly numerous were the nymphalid butterflies (72 species) recorded during the study (Appendix 1). Presently 223 species in 55 genera of nymphalids are listed from NG of which 201 species in 53 genera occur in PNG (Parsons, 1999); therefore this result is expected.

To explain such richness of butterfly fauna, the flora of the study area has to be considered. The WWMA accounts for more than 200 species of woody plants (Damas, 2008). Most adult nymphalid butterflies are known to gain all their nutritional requirements by feeding on the juices of rotting fruits or plant sap (DeVries & Walla, 2001; DeVries et al., 1997, 1999). During this study, most of the plants were fruiting which probably explain why more species of nymphalids were recorded. Similar results of species richness were also recorded in a lowland dipterocarp rainforest of Borneo (Willott et al., 2000) and Sierra Leone (Sundufua & Dumbuyab, 2008) using similar methods. The Sierra Leone study reported 195 species of butterflies of which a significant proportion were nymphalids. Similarly, the Borneo study recorded 185 species of which nymphalids comprised a significant proportion also. The result of the present study implies that the WWMA is exceptionally rich in butterfly fauna most probably owing to the mosaic of heterogeneous habitats that are in the study area. The 10,018 ha WWMA is large enough to accommodate many different habitats and thus the result illustrates this.

A significant difference was also detected in the mean numbers of species between the main habitats in the study area (Fig. 5). More species of butterflies were recorded in the river/stream habitats compared to the other three habitats which were sampled in the study area. The reason may probably lie in the feeding and courtship behaviour of tropical lowland butterflies and their close association with riverine habitats. Waterways act as natural pathways for butterflies and the males of many species can be observed patrolling up and down in search of females (Parsons, 1999). Furthermore, riverine habitats are used by the majority of butterflies mainly to imbibe mineralised water from the damp sand of the creek sides and nectar from flowers along the banks (Parsons, 1991). This was very much evident during the study. The secondary vegetation of the creek margins at the edge of the watercourse was evidently diverse in plant species in various stages of regrowth; ideal for females to lay their eggs on tender young foliage (Parsons, 1999).

Community composition

Butterfly communities in the main habitats surveyed displayed similarities between the secondary forest habitats and the primary forest habitats (Fig. 6). The two habitats had a total of 107 (secondary forest habitats) and 100 (primary forest habitats) species of butterflies correspondingly. Between the two habitats, they shared 77 species. A similarity result between these two habitat types might be expected due to secondary forest succession. Each plant species is most abundant at a particular successional age (its successional optimum). Both species traits and microclimate change in a predictable way during succession, therefore it might be expected that species with similar successional optima share more herbivores than species differing in their successional optima (Leps et al., 2001). In the study area, shifting cultivation is common. After about three years of intensive gardening, the garden sites are abandoned where secondary forest succession begins. The transects in the secondary forest habitats utilised these abandoned garden sites; which were about 6 to 9 year old secondary forests at the time of this study. Equally in the primary forest, some natural forest gaps were present within close proximity of the primary forest transects also having roughly the similar ages of secondary regrowth as to the secondary forest habitats. This probably explains why secondary forest habitats and primary forest habitats had comparatively similar butterfly communities.

Comparisons were also made between open area habitats and river/stream habitats. A total of 118 species of butterflies were recorded respectively for both habitats. Even though both habitats shared 88 species, the CCA revealed dissimilarities between the two habitats (Fig. 6). This may be associated with habitat characteristics of the habitats in question. More individuals were recorded in the river/stream habitats (966 individuals) than in open area habitats (934 individuals). According to Parsons (1999), individuals tend to congregate along riverine habitats more often than open area habitats mainly for imbibing mineralised water among other activities such as patrolling for mates and seeking host-plants. These resources clearly lacked in open area habitats where it was much drier and relatively homogenous in terms of host-plants.

Community composition along a successional gradient showed differences between the open area habitats and the secondary forest and primary forest habitats. Butterfly communities in open area habitats were clearly separated from the more similar secondary forest and primary forest communities (Fig. 7). More species of butterflies were recorded in the open area habitats than in the secondary forest habitats and primary forest habitats. The open area habitats comprised approximately 67% of the total number of species recorded in the study area, whereas the secondary forest habitats and the primary forest habitats included only 61% and 59% correspondingly. The number of shared species along the successional gradient observed 78 species between the open area habitats and secondary forest habitats; 77 species between the secondary forest habitats and the primary forest habitats; and only 72 species between the primary forest habitats and the open area habitats.

Interestingly enough, the study recorded a decline in species richness with increasing canopy cover. This result may be due to the more opportunistic ubiquitous species (Spitzer et al., 1993). Approximately 48% of the species in the open area habitats were ubiquitous species - e.g. Notocrypta renardi, Papilio ulysses, Eurema hecabe, Pithecops dionisius, and Parthenos aspila. This species have a wider geographic range which gives them the advantage of scouring for resources effectively which probably influenced the results. Similar findings were also documented in Vietnam (Spitzer et al., 1997, 1993; Vu & Yuan, 2003).

Habitat preference

Habitat preferences of butterfly species in the study area were also examined. Sixty-three species having 10 individuals and greater were analysed (Fig. 8). The results showed that species having similar resource requirements in essence coexist. Probably the two most important resources to a butterfly are: (1) host-plant; and (2) solar-radiation (sunlight).

A total of 35 species had CCA scores of less than zero, of which 13 had strong preferences for more disturbed habitats. Species which prefer more disturbed habitats either have wider geographic ranges or are more specific to those habitats that have the host-plant in close proximity. For example, the nymphalid species Ypthima arctoa (CCA score -2.6) was only encountered in the open area habitats in the presence of its host-plant; Poaceae: Imperata cylindrical (Parsons, 1999). The same can also be said for Junonia erigone (CCA score -2.6), which was encountered almost every time in well lit habitats. The other species were primarily ubiquitous of nature. Of the 63 species analysed, 15 species preferred less disturbed habitats and had CCA scores >1. Only 2 species had CCA scores >2. These were Elymnias cybele and Prothoe australis. According to Parsons (1999), E. cybele and P. australis prefer well shaded forest understorey. This means that both species are more habitat specific, thus having the smallest geographic range and therefore having the most important conservation value.

Results obtained by Spitzer et al. (1993) and Sundufua & Dumbuyab (2008) concluded that butterflies confined to the climax forest tend to have a small range of distribution of endemitic type, whereas species associated with disturbed habitats are much more widely distributed. This patterns needs to be verified also for NG as it has clear conservation implications - species dependent on undisturbed forests are at the same time also more restricted geographically, making them sensitive to human disturbance.


This study builds on the ecological knowledge of butterflies particularly for PNG. It examined the species richness, community composition and habitat preference of butterflies feeding in the main habitat types and along a successional gradient in the WWMA. The results showed differences in the mean species richness and community composition of butterflies observed during the study. Much of the variability in the observations was due to feeding strategies, reproduction and habitat requirements and preferences of the butterfly communities. Although these findings may not appear to be new, one important aspect of this study is that it documents the butterfly communities for a much long time frame in order to truly understand the community compared to previous butterfly work in PNG (e.g. Bowman et al., 1990). They are also the first detailed results coming from a large tropical island fauna and from the island of New Guinea in particular. The study is also timely in the sense that the surrounding forests will be logged and information gathered maybe used as baseline data for future comparisons to better understand the effects of forest disturbance on insect communities in the tropics.