Maliau dragonfly project

From BiodivBorneo09

Jump to: navigation, search

Fikty Aprilinayati, Andrew Brownjohn, and Mindy Tuan



A long water body in a ditch in Maliau Basin was picked and six dragonfly species observed for intra- and interspecies interactions on a standardised series of actions with graded aggression levels. Dragonflies were given morphonames for ease of observation: Red Wing Belly, Pink Belly, Red Belly, Clubtail, White Belly and Half-Wing were the species present. Three interactions (chase, kick and disturb) were eventually mapped onto an interaction web to visualise the dragonfly community dynamics within the area. Determining a dominance hierarchy was possible by comparing aggressor and recipient interactions and observing the interaction webs. 4 species out of the 6 species present were determined to have specific interactions with just 1 other dragonfly species across all aggression levels: the Red Wing Belly and the White Belly (aggressors) to the Pink Belly and the Half-Wing (recipients) species. Manipulation experiments performed by removing the main aggressor, Red Belly, produced interesting results, where the level of community interactions dropped significantly. However, due to a lack of sufficient number of surveys it is not possible to conclude with any level of confidence that removing the main aggressors from the community impacts the community dynamics at that level all the time. Further studies done with a sufficient number of replicates should be carried out to verify this hypothesis.


Competition for limited resources dominates community interactions between both populations and individuals. This competition can lead to complex behaviors between various similar species all competing for the same habitat location, food resources, and breeding sites. In order to observe the possible behavioral adaptations (especially dominant or aggressive behavior) of species to facilitate resource partitioning, our group selected the self-contained ditch environment along the path from the Maliau Research Center to the Canopy Walkway. The ditch plays home to more than ten dragonfly species, six of which are the most common (see list of dragonflies here). Our team surveyed those six species for half-hour intervals on 4 separate days to gain an understanding as to the nature of their inter- and intraspecific interactions, to find out:

  • Do specific dragonfly species within our ditch area directly compete with one another for the same resources (perching space and territory) and is there a distinct dominance hierarchy?

As a further expansion of our study, we perturbed the system by removing a dominant species and observed the community interactions in its absence, to see if its dominant, combative behavior might play a significant role in the behavior of other species.


In the field

To observe the dragonflies around the ditch, our group selected a single location along the path exposed to sufficient sunlight for dragonfly activity. In a prior survey, we identified the six most common species and standardized our identification of each by close observation. We then split the stretch of ditch split into three equally-sized sub-sites, and each member of our team took a post at one of the three sites for half-hour rounds. Before beginning each half-hour round, a population count of each of the six species for each of the three sites was taken, and all discernable active behavior by members of those species recorded during the time period. Interactions between the dragonflies were standardized before observation, and split into several levels of dominant or aggressive behavior: "chasing", "kicking", "disturbing", and "surveying". Time spent inactive (e.g. perching) was not recorded. At the close of each half-hour, we took five minutes to rotate locations, take new population counts, and then began the process again. In total, we completed 13 surveys over four days, during both morning hours (between 9:30 and 12:00) and afternoon hours (between 12:00 and 16:00). Our survey times were limited by the activity of the dragonflies before 9:30 and after 16:00, due to the low angle of the sun and their dependence on it for activity. The final day of data collection (the last two data surveys) was supplemented by the removal of the most dominant species as determined by total aggressive activities from the three previous days. Our observations after removing the species termed "Red Belly" used the same method as the three previous days.

Data Analysis

  • We hypothesize that there is a pattern to the interactions between each of the dragonfly species, and that by removing a dominant species, the community behavior of the other dragonfly species will change.
  • The null of this hypothesis would result in our not finding any targeted or patterned interactions between specific species, and there will be no dominance hierarchy. Such a lack of dominance hierarchy would lead to no change in community interaction after the removal of the dominant species.

To test this hypothesis, it was necessary to compile each of the three first days of our data sets into a single set organized by species, location, type of behavior, and the recipient of each aggressive action. From this collection, we produced several interaction matrices to map the direction of aggressive behavior between each species. A separate matrix for each species-species interaction: kicking, disturbing, and chasing, was drawn and the results for each compared. The interactions for the final day, during the Red Belly removal, were organized separately, and observed for any patterns in the same manner.


By visual inspection and analysis of the data collected after three days of surveys, we found that while the Red Wing Belly species was the most abundant, the Red Belly species was by far the most aggressive towards a large number of dragonflies. Using this information, we removed the Red Belly individuals from the community of dragonflies at the ditch on the fourth day of surveys and noticed that there was a big drop in the number of interactions between species once they were removed. Simply put, the level of interactions that go on in each day dropped significantly when they were removed.

Analysis of our data using R statistics (diagram.package) produced interaction webs (see Figures 1-3), where it is visibly noticeable that four dragonfly species had specific species interactions throughout the three levels of aggression. These comprise the Red Wing Belly and the White Belly (aggressors) interacting with the Pink Belly and Half-Wing dragonflies (recipients) respectively.

Fig. 1: Interaction web of interspecies dragonfly chases above a threshold of 8
Fig. 2: Interaction web of interspecies dragonfly disturbing actions above a threshold of 3
Fig. 3: Interaction web of interspecies dragonfly kicks above a threshold of 2


Data Interpretation and Argument

Based on the data we have obtained, we have determined that there appears to be a relationship between species pairs: the "Half-Wing" and "White Belly" and the "Red Belly" and "Red Wing Belly". The disturb, kick, and chase data all overlap on these particular interactions, and the overlap suggests that this is a nonrandom pattern. Our ability to collect data during the four days that we were allotted suffered from the weather; most of our time was spent surveying during hours that were overcast, while dragonflies are most active when it is sunny. However, not all is lost, as all four days of survey were overcast and thus provided a consistency in weather.

The data collected on the final day, in particular, is difficult to interpret, for we can interpret the lack of activity within the community either to foul weather or the lack of an aggressive species. By comparing the first three days of sampling to the final day, it is possible to argue that the Red Belly species has an important role to play in the activity of the dragonfly community as a whole, but it would be necessary to complete at least one more full day of sampling to verify that removing the Red Belly had an effect, rather than inclement weather. Our lack of statistical analysis to determine whether the patterns of behavior seen over the course of these four days of observation were repeatable resulted from an insufficient pool of data and a lack of time to complete the complex calculations necessary to determine such a pattern. However, based on the agreement between each of the interaction matrices, it appears very likely that the interactions between those four species are not entirely random, and are actually targeted. An extended period of observation and subsequent statistical analysis would likely support this claim.

The reason for the species-species targeting that we observed is difficult to establish. Based on non-rigorous observation, it is possible that the targeted interactions result from a closer similarity in perch choice between Red Bellies and Red Wing/Bellies (low twigs) compared to the choices of Half-Wings and White Bellies (high blades of grass). However, this is a general observation, and there were instances in which Red Bellies and Half-Wings were seen perching on the same twig, and so on.

Extension upon our findings

Additional samplings of the dragonfly community, both with the Red Belly and without it, would help us verify that our findings are accurate. If it were possible, a statistical test to verify the non-random nature of the interaction data (comparing the times a species interacted with a specific species versus all of the others) that we obtained would be necessary in order to consider what we have observed to be an actual phenomenon.

It is worth noting that our team attempted, during the third day of observation, to remove the most populous species of the dragonflies, the "Red Wing Belly", and determine whether this had an effect on the community dynamics. Our motivation stemmed from the fact that intraspecies chasing between Red Wing Bellies tends to "cascade", causing multiple chases (both inter- and intraspecies cascades) in a small period of time. Such high activity levels appeared to have a direct effect on individuals of other nearby species, and to test this, we tried eliminating the population for several observation periods. We succeeded for the first half-hour period, but during our second observation period, the species began to recolonize the area from other nearby regions and became impossible to remove completely. As such, we were forced to continue with non-removal sampling. During the first sampling period in which we removed Red Wing Bellies, we noticed a significant decrease in dragonfly activity, but this can be attributed to the weather. It would be highly interesting, if a group had a more efficient method of dragonfly removal, to see if the community dynamic was in fact altered as we hypothesized, but it was beyond our ability to sample in this particular case.