Lambir butterfly project

From BiodivBorneo09

Jump to: navigation, search

An Analysis of Patterns of Coloration and Size Differences in Forest and Forest Edge Populations of Lepidoptera

Abigail Faith Schoenberg and Kristina Prus



Lepidoptera habitat partitioning between the Montane Dipterocarpe Forests of Lambir Hills National Park protected by the canopy and the unprotected outskirts of the forest (those meeting grass or open land) were studied to determine whether there were specific morphologies significantly associated with each habitat. Butterfly traps and transect walks were used to sample the Lepidoptera variation between the two habitats, at 3 different sites. Coloration, i.e. light or dark, and size, i.e large (>5cm in wingspan) or small, were noted for each specimen in the study. Ninety-nine Lepidoptera were surveyed over a 6-day period. The study found a significant association between coloration and habitats, such that lightly colored butterflies existed at higher frequency on the edge of the forest than those inside the forest. We also found that Lepidoptera are more highly abundant on the edge of the forest than on the inside. The study found no significant association between location of the specimen and size of wingspan was reported. This data can be used for future studies that require finding locations of specific Lepidoptera.


We investigated the system of adult Lepidoptera habitation and their patterns of location. Our interest in this system stemmed from the geological history of the region that caused the diversity we see. Over past 2 million years, there have been 4 Ice Ages during which Borneo was connected to the Malay Peninsula, Sumatra, Java, and Palwan in the single mass called Sundaland; during these Ice Ages, butterflies from the different islands interacted and mixed, and when the last Ice Age ended about 12,000 years ago, Borneo was left with a diverse pool of butterfly species (Otsuka 2001). We were interested in exploring patterns among the resulting diversity through resource partitioning; Lepidoptera are known to have relatively small microhabitats, due to their limited pollination abilities (Hill et al. 2004). Therefore, we expect the Lepidoptera to have specific home ranges that are based on morphological variations.

Furthermore, based on observations at Lambir Hills National Park, the forest edge is a very different habitat than the inside of the forest. The edge, or the part of hte forest bordered directly by short grasses and houses, is a very bright environment receiving much sunlight during the day and having an open sky exposed to birds and other predators. The inside of the forest is a very different environment, not receiving much sunlight and having a covered sky, protecting the inhabitants from aerial predators and reducing agility of flying organisms due to the high density of trees. These differences can result in morphological variations based on the habitat where Lepidoptera live. Variation in coloration can account for these different habitats, since a darkly colored Lepidoptera on the forest edge would stand out in the sunlight, just as a brightly colored Lepidoptera would stand out in the dark forest. The density of trees is also much more in the forest than on the edge, so a smaller Lepidoptera may have greater maneuvering abilities in flight inside the forest.


What are the differences in morphology of Lepidoptera in different niches at Lambir Hills National Park? We predict that Lepidoptera occupying niches inside the forest will be darker and smaller than Lepidoptera occupying niches on the forest edge. Finding such a relationship between morphology and habitat partitioning could aid further understanding of specific species' life histories, as well as finding other patterns in the forest such as trends in butterfly pollination events of various plants.


We recorded Lepidoptera data at 3 different sites: at the CTFS plot, at the trail near Hilltop Lodge, and at the trail on the way to the canopy crane. We used two different methods to record the Lepidoptera in the field: Butterfly traps and transect walks.

Two separate traps were set up at each site: one on the edge of the forest (where the canopy ends and open land begins), and one a fifteen-minute walk into the forest, at least 10m off the trail. We placed two rotten bananas under each trap as bait. Traps were set up in the morning and left to sit for 24 hours before collection. We then collected the caught Lepidoptera and noted color, size, and number, along with the site number and whether the trap was on the edge or inside the forest.

Transect walks at each sight were 400m long – 200m in one direction and 200m back along the same path. A transect walk was done along the edge of the forest near each site, as well as on the trails within the forest at each sight. The 400m were walked over a span of 20 minutes, carefully timed so that our measured 50m marker was passed every 2.5 minutes. This gave us a sufficiently slow walk so that we could remain attentive to the task of searching for Lepidoptera; it also helped us prevent pseudo replication by granting each subject time to relocate. We classified and recorded the size and color of every individual butterfly or moth we saw within 5m to the left or right of the trail; we both watched for Lepidoptera, and the role of recording the data was switched 10 minutes into each walk.

Character classifications are as follows: “bright” Lepidoptera included any specimen with a strikingly visible amount of blue, red, orange, yellow, and white; “dark” Lepidoptera included specimens that were predominantly brown, black, or dark grey; “small” Lepidoptera were those with a horizontal forewing wingspan of 5cm or less; “large” Lepidoptera were those with a horizontal forewing wingspan greater than 5cm.

To determine the significance of the associations , we used Chi Squared tests and Fischer tests on the specimen data to see if there was a statistically significant difference between the coloration and size of the observed Lepidoptera and in which habitat they were observed. We then used a General Linear Model of the binomial family to determine if the proportions of Lepidoptera of in terms of coloration and size in each area were significant.

Finally, we used a General Linear Model of the poisson family to analyze the count data to confirm whether the difference in abundance between forest and edge suggested by the plots in the figures was, in fact, significant.


Figure 1

As anticipated, our results showed that coloration had a significant association with the location of the Lepidoptera; the Chi-Squared test returned a p-value of .05173, and the Fisher test returned a p-value of .04821. The General Linear Model in the binomial family also confirmed that the proportion of light and dark butterflies in the forests and on the edges was, in fact, significant (Figure 1). As proportions show, Lepidoptera with a light coloration exist in a significantly higher proportion on edges than within the forest, confirming our hypothesis that lightly colored butterflies would exist in higher abundance on forest edges.

Contrary to our hypothesis that larger butterflies would be found on the forest edge, no tests suggested that size was significantly associated with location; the Chi-Squared test revealed a p-value of 0.3168, and the Fisher Test returned a p-value of 0.2389. Additionally, the General Linear Model of the binomial family did not suggest significance of the proportions in Figure 2; the p-value was 0.2.

Finally, the General Linear Model of the poisson family did reveal significance in the difference of abundance between the edge and the interior of the forest. Lepidoptera were more abundant on the edges by a p-value of 3.54e-9.


Our study confirmed the hypothesis that brightly colored Lepidoptera were more often found on the edge of the forest and that dark colored Lepidoptera were most commonly seen inside the forest. We imagine that our original rationale explains why brightly colored Lepidoptera are significantly associated with location in a higher proportion on edges than within the forest. Lepidoptera within the forest need dark coloration for camouflage purposes, while Lepidoptera on the edges need bright colors to warn potential predators or blend in with the generally more brightly coloration on the edge of the forest due to the sunlight on bright green grasses.

Our idea that Lepidoptera were more likely to be smaller sized on the inside of the forest and larger on the edge was not confirmed; size distribution was more or less equivalent in both locations. We have also considered reasons for which our hypothesis concerning size and location was proven false. Perhaps size is not significantly associated with location because the ability to move nimbly is unassociated with size, or because the ability to move nimbly does not protect from predators as well as appropriate coloration would. Further study is needed to see if these reasons are indeed true. Another study suggested that butterflies have eyespots on areas of the wings that are easily broken so that predators would grab those areas of the wing (Hill et al. 2004). If that is true, then it would make sense that size does not matter as much as the patterns and colors on the wing, which is in line with what our study found.

Finally, we appreciated the unexpected discovery that abundance of Lepidoptera is much higher on edges than in the forest. We hypothesize that this is because of the lack of regular flowering cycles in tropical rainforests; Lepidoptera are not guaranteed sources of nectar or rotting fruit within the forest, but there are chances of finding unnatural nectar sources on forest edges (flowers planted by humans, etc.).

There are a number of confounding factors that might explain our data. One of the most important is that we were unable to include the butterflies of the canopy in our study. Therefore, we only counted part of the forest community and all of the forest edge community. Additionally the use of butterfly traps may have skewed data, as they are prone to catching only butterflies of the family Sateridae. Therefore, in future studies, it may be best to only include transect walks (both on the forest floor and perhaps on canopy walkways) in the study. Merging the data from the two collections was the best way to give them equal weight, but it is a major confounding factor that, in future studies, could be easily avoided. Additionally, we regret that we were unable to identify the majority of the Lepidoptera that we caught; therefore, we have not included an appendix with a list of such species.


(See summary of argument)


  • Eastwood, Rod. Personal Interviews. 14-17 June 2009.
  • Hill, R. & Vaca, J. Differential wing strength in Pierella butterflies (Nymphalidae, Satyrinae) supports the deflection hypothesis. Biotropica, 2004, 36, 362-370.
  • Otsuka, Kazuhisa. Butterflies of Borneo and South East Asia. Kota Kinabalu: Iwase Bookshop Sdn. Bhd., 2001.