Insect Visitors and Petal Coloration Project

From BioDivBorneo2010

Jump to: navigation, search

Exploring Petal Coloration as a Factor of Insect Visitation Frequency in Maliau Basin, Borneo



A stingless bee visitor

Many factors are recognized to influence insect visitation to plants and insects are a particularly important vehicle for pollen movement in angiosperms. Here, the frequency of insect visitors was analyzed under three different petal color variations in a single Lamiaceae species of Maliau Basin in Sabah, Malaysia, to assess the importance of coloration as an attraction cue for pollinators. The study flowers have a distinct purple coloration on their bottom-most petal thought to potentially be a 'nectar guide'. The manipulation experiment entailed utilizing paint to cover the purple pattern in a subset of flowers and observing differences in the frequency of insect visits of these covered flowers as compared to a natural subset and a control subset of flowers. Data related to environmental factors such as light intensity in the form of cloud cover and shade was also recorded. Other data collected included flower height, flower diameter, time of day, and general insect visitor type.

For the primary color-manipulation section of the experiment, results indicated that both groups of flowers which received experimental manipulations had a lower frequency of visits than the natural group. There was also no significant difference between flowers with a covered color pattern and the flowers with an uncovered pattern. This suggests that the paint medium used to cover the coloration was a deterrent for reasons not related to coloration. It is likely the paint had an unpleasant odor for insect visitors. Data collected indicated that effects of cloud cover and shade on insect visitation frequency were not statistically significant when all observed flowers were considered as a single data set. However, when considered individually, the data indicated that cloud cover had an effect on the number of insect visitors in the subset of flowers with the covered color pattern. Similarly, shade was found to be a relevant factor in both subsets of manipulated flowers.


Flower diameter measure

Animal vectors are important vehicles for pollen movement in many angiosperm species. Organisms such as bees, wasps, ants and birds play a vital role in the fertilization of plants. Due to the intricate relationship between humans and plants throughout most of human existence and certainly since the dawning of widespread agriculture, mechanisms that influence pollinators and thus potentially increase the successful production of fruits and energy-rich plant structures hold an important place in biological study. As in multiple habitats, in Sabah's Maliau Basin, Malaysia, bees play an especially significant role in pollinating the diverse plant flora.There are many characteristics that angiosperm flowers exhibit to attract their specific pollinating organisms. Some cues to attract pollinators include flower shape, coloration, odor, production of nectar and UV patterning [1]. One study done in Lambir Hills National Park in Sarawak, Malaysia, found that flower shape and location (understory versus canopy/gap) are also important factors affecting which species of stingless bees are the primary bee-pollinators for many flower populations in that area [2]. This suggests that bees use visual and light-intensity cues to discriminate when choosing which flowers to visit.

In this study, the possible association between the visual color pattern on the bottom petal of a specific angiosperm species found in Maliau Basin's Study Center and the amount of insect visitors was explored under different light conditions. Multiple insect visitors were observed to interact with and/or enter the flowers in both high-sun and high-shade habitats. In this species of the Lamiaceae family, (more commonly known as the mint family) stingless bees were observed to be a primary pollintor species. These flowers have white petals, are approximately 1.65cm in diameter and 1.9cm in height from the bottom of the ovary to the flower opening.

Flower height measure

These dicots have flowers which exhibit bilateral symmetry. They have five petals: the upper two petals on both sides are completely white, while the bottom 'lip' petal has a bright purple dotted-line leading into the center of the flower. Previous studies have suggested that such coloration often serves as a 'nectar guide' for pollinators [3]. This nectar guide and 'lip' petal is found in a variety of other plant families [4]. Such coloration is an interesting point of study by virtue of such convergent evolution. The Lamiaceae flowers occur in both highly sunlit and shady areas. In a preliminary hour-long observation, stingless bees, stinging bees, ants, and butterflies were also observed visiting this species. The bees, however, exhibit a highly specialized behavior in which they land on the purple 'lip' petal of the flower and enter down into the tube of the flower. They typically remain in the flower for several seconds possibly collecting pollen and/or nectar. The ants can be found all along the plant stalk and do not seem to visit the flower specifically. They also travel from flower to flower very infrequently, and therefore are not likely to be primary pollination vectors.

Here, the role of the intense purple coloration pattern on the frequency of insect visitors to the model plant species is investigated. This study analyzes how the visual factor of petal pattern coloration and environmental factors including light and clouds and might affect frequency of insect visitors. Evidence that these factors play a significant role in plant-pollinator interactions would have valuable implications for further relationship studies in plants and pollinators. Knowledge surrounding specific mechanisms of pollination are important for agriculture, especially with the current increase in human population and interest in both organic farming and bioengineered crops. If specific traits are isolated that appear to increase the frequency of pollinator visits, it may be possible to utilize this data when considering traits important for increasing agricultural yield and ensuring a high level of fertilization on agricultural crops.


  1. Does the purple coloration pattern on the bottom petal of the Lamiaceae influence the frequency of insect visitors to the flower? If the purple coloration is covered, will the composition or frequency of insect visitors change as compared to uncovered flowers?
  2. Is there a significant difference in the frequency of insect visitors in different light-level habitats?


Field methods

NAT flower
COV flower
UNC flower

Study areas

21 experimental field sites were located in and around the Study Center of the Maliau Basin Conservation Area. Sampling was carried out from the 30th of June to the 2nd July, 2010.

Experimental design and sampling procedure

A member of the plant family Lamiaceae was chosen as the model organism of this study. This plant is commonly found at Maliau Basin Conservation Area along the road sides and in open areas.

- 3 flower types were created by selectively applying white paint to specific petals. The types were: natural flowers (NAT), covered flowers (COV), and uncovered flowers (UNC). The NAT flowers were not manipulated by any paint effects. The COV flowers had white paint applied to cover the petal with the purple coloration. The UNC flowers had white paint applied to a white petal at the upper right in order to control for unintended effects of painting. A period of at least ten minutes was allowed after paint application for the paint to dry before we began data collection.

-The height and diameter of each flower was also measured, and later turned into a size index. A single experimenter painted all flowers while another experimenter measured all flowers to eliminate differences in techniques.

- For each of the 3 group categories, 4 flowers were chosen using a randomization procedure at each site and each flower were observed for 10 minutes. Time, date, and light habitat were recorded for each flower and during each observation segment all insect visitors were recorded. An 'insect visitors' was defined as any insect that landed on a flower for a full second.

- Each site was at least 5 meters apart in order to avoid pseudo-replication. Sun and shade habitats were selected in order to investigate the effect of light conditions on insect visitors. Cloud cover was also recorded.

- Observations were collected for 3 consecutive days at differing intervals: morning (07.00-10.00 hours) midday (10:00-12:00 hours) and afternoon (12:00-15:00 hours). Each observer recorded data on a rotating schedule of flower types to reduce observer bias.

- There were 25 NAT flowers, 25 COV flowers, and 25 UNC flowers. Thirty-three flowers were recorded in the shade and 42 in the sun.


Data from each site was stored as a spreadsheet, and then imported in the Statistical Analysis Platform R.

A series of general linear models (glm) assuming a Poisson distribution were performed to determine if there were statistically significant relationship between the total number of non-ant visitors. Each test excluded one flower type in order to assess the differences between the other two types. This included:

  • A glm of non-ant visitors as a function of flower type, excluding NAT flowers in order to compare COV and UNC flowers
  • A glm of non-ant visitors as a function of flower type, excluding COV flowers in order to compare NAT and UNC flowers
  • A glm of non-ant visitors as a function of flower type, excluding UNC flowers in order to compare NAT and COV flowers

We ran a second set of glm tests to analyze the effect of cloud cover on the number of non-ant visitors within each flower type, and across all flower types:

  • A glm of non-ant visitors as a function of cloud cover, including all flower types
  • A glm of non-ant visitors as a function of cloud cover, including only NAT flowers
  • A glm of non-ant visitors as a function of cloud cover, including only COV flowers
  • A glm of non-ant visitors as a function of cloud cover, including only UNC flowers

We ran a series of Wilcox tests to assess if shade was a factor which significantly affected the number of non-ant visitors:

  • A Wilcox test of shade and non-ant visitors, including only NAT flowers
  • A Wilcox test of shade and non-ant visitors, including only COV flowers
  • A Wilcox test of shade and non-ant visitors, including only UNC flowers

We created a size-index for flowers by approximating a volume using the equation for a cone: <math>(\frac{1}{3}\,)(pi)(h)(r^2)</math>

Finally, we used another set of glm tests to analyze the effect of flower size on the frequency of non-ant visitation:

  • A glm of non-ant visitors as a function of flower size-index, including all flower types
  • A glm of non-ant visitors as a function of flower size-index, including only NAT flowers
  • A glm of non-ant visitors as a function of flower size-index, including only COV flowers
  • A glm of non-ant visitors as a function of flower size-index, including only UNC flowers

A record of the R commands used to generate derived data from the stored files is given in the Appendix


The raw data for all flowers indicated that the NAT flowers had more visitors than COV or UNC flowers, both in terms of total insect visitors and non-ant insect visitors. Our glm analysis done between the different flower types with respect to the number of non-ant insect visitors indicated that there was a significant difference in visitors between both the NAT and COV flower types (p = 0.0004) and the NAT and UNC types (p = 0.013). However, there was no statistically significant difference between the number of insect visitors between the COV and UNC flower types.

Data were also obtained regarding environmental factors that may affect frequency of insect visitation. When analyzed with respect to the frequency of non-ant visitors in all flower types, cloud cover did not appear to have a significant affect on number of insect visitors. However, when analyzed in each flower type individually, cloud cover was shown to have a significant negative effect on non-ant visitation in only the COV flower type (p = 0.053). Data for flowers in the shaded vs. unshaded habitats indicated that shade had a significant negative effect on insect visitors in both the COV (p = 0.001) and UNC flower types (p = 0.003) but not the NAT type.

We also found that flower size was not a significant factor affecting the frequency of non-ant visitation in any of the flower types, either combined or separately.


The raw data results suggest that the paint utilized in this experimental manipulation had a significant negative effect on insect visitation in the plant of study. As seen in the data, the total number of insect visitors for the NAT flower type (56) was almost 1.5 times the total number of insect visitors in either the COV (38) or UNC (35) flower types. The manipulation paint was intended to have no effect on insect visitation other than to cover the bottom petal color pattern. However, the paint did have a slight odor which was detectable to the experimenters up to five minutes after application. Because of this, the experimental procedure included ten minutes wait time for the paint to dry, but this may not have been enough to eliminate detection by the insects. Thus the paint itself likely significantly affected the results obtained. A side aspect to note is the behavior of the stingless bees observed when approaching a COV flower. Often times, the stingless bees would approach a COV flower, hover around it uncertainly, and then either land on the wrong petal, or fly away or to another flower type. While we were unable to quantify and analyze this exact behavior, it is relevant to the study in that it suggests the lack of purple coloration may have been a source of confusion to the insect visitors. Further studies may be able to investigate this phenomenon and better qualify the causes of such behavior.

Despite the unforeseen consequence of the paint, the lack of a significant difference between the COV and UNC flower types was largely unexpected. Previously conducted studies have indicated that such color patterns on the petals of angiosperms are often important for pollinator attraction and flower choice. Thus results were expected to indicate a significant decrease in the frequency of insect pollinators to the COV flower type as compared to both UNC and NAT. In theory, the NAT and UNC would not have a significantly different frequency of insect visitors. The results obtained, however, suggest that factors beyond visual cues, such as scent, shape, habitat or UV-light likely play a large role in the attraction of insect visitors to this particular angiosperm. In particular, there may be a slight odor emitted by nectar production in the experimental flower species that is masked or altered by the presence of the paint. Alternatively, there may be an entirely different UV-light pattern undetected by humans that guides or attracts insects. The flowers are also quite small, so it is possible the paint altered the thickness of the petal and deterred insects in that regard.

The results obtained from the analysis of the cloud cover data were also slightly surprising. It is unclear why clouds might have a larger effect on flowers that have their purple pattern covered. One possibility is that the lower light level created by clouds made the manipulated pure-white flowers relatively more difficult to see under those circumstances without a bright purple target. This may have increased the likelihood that insect visitors would pass by the COV flowers more often than either the NAT or UNC.

For the sites in which the flowers were in the shade, the data indicated that there was a significant decrease in the frequency of insect visitors for the COV and UNC flower types, but not the NAT flower type. There were less insect visitors overall to both the COV and UNC flowers as compared to the NAT flowers, which makes this result particularly interesting because borderline effects are more often recognized in larger data sets than smaller data sets. A more extensive data set has a larger amount of variance, so the fact that shade seemed to have an effect only in the two smaller data sets is unexpected.

Beyond the repellent effect of the paint application, there are other factors that may have altered the results. Sources of error include difficulty in determining a true insect visitor, observer bias, replication in insect visitation, differences in paint application between flowers, handling effects of flowers, presence of observers and possible relevant unrecorded factors. While the effect of the paint generated inconclusive results in regards to our original question, the results are nonetheless still relevant to pollination factors and further studies are potentially useful. Further studies could investigate previously mentioned other cues that may attract insect visitors to the Lamiaceae flowers, and it should be well-noted that chemicals in the immediate surroundings of plants play have a significant effect on insect interactions.


  1. Lauren Ruane, Personal Communication
  2. Nagamitsu. T. et al (1999). Preference in flower visits and partitioning in pollen diets of stingless bees in an Asian tropical rain forest. The Society of Population Ecology and Springer – Verlag Tokyo, vol. 41: 195-202.
  3. Johnson, SD. and Dafni. (1998). Response of bee-flies to the shape and pattern of model slowers: Implication for floral evolution in a Mediterranean herb . Functional Ecology, vol. 12: 289-297.
  4. Manning, A. (1956). The effect of nectar guides. Behavior, vol. 9: 113-139.


Downloadable PDF:

Our "R" Analysis