The Bee-nanza of Data Collection in Field Research

Let’s go shopping! Our list: a black and yellow striped sweater, a pair of sunglasses, and honey. If you had to purchase these items yourself, you would probably want to minimize the distance you travel between stores and between shelves within a store. As individuals, we like to maximize efficiency and the way to be most efficient is to minimize travel time. However, if you and several thousand of your closest friends all went shopping together, you might decide to pick a larger store, even if the shelves were a little farther apart as a result. When in a group, suddenly having enough to go around becomes much more important.

A honey bee “shopping” for pollen at a California Buckwheat “store”. You can see the pollen she has collected already hanging off her back legs like saddlebags. (Image Credit: Keenan Gilson)

Similarly, when “shopping” for pollen, individual honey bees minimize the distance they must travel between plants and between flowers on a plant. We expect them to prefer plants that are close together and have high inflorescence density, meaning that the flowers of each plant are also close together. The inflorescence density of a plant can be measured as the area of flowers on the plant compared to the total area of the plant. However, honey bees live in large colonies with several thousand “shoppers”. As such, to maximize the pollen collected by the colony as a whole, honey bees should choose big plants with a large blooming area, meaning that there are a lot of flowers overall but not necessarily close together. The blooming area of a plant can be measured as the total area of the plant with flowers.

Blooming area is the region surrounded by the red line. Inflorescence density is the region denoted by black and white lines compared to the area of the whole plant.

Both individual and colony reward factors impact honey bee foraging patterns, a fancy way of saying that both individual and group dynamics influence where honey bees “shop” for pollen and nectar. The same way that stores track which demographics to appeal to with targeted campaigns and ads, understanding which factors most strongly determine the plants bees choose to forage can help us to improve agricultural field design. As honey bees pollinate most of the world’s crops - including peaches and cherries, two summer time favorites of the Bee Lab - this is a key importance to the global crop industry. Personally, I was also interested in the question of whether bees are more concerned about how quickly they can individually collect food or how quickly the colony as a whole collects food. Determining whether honey bees tend to prefer plants with high inflorescence density or with large blooming areas can help us to expand our understanding of eusocial insects’ social dynamics. This summer, to answer this question, the Harvey Mudd Bee Lab conducted an experiment on how local inflorescence density and blooming area affect honey bee visitation rates. We performed two manipulations on sets of California Buckwheat that altered either local inflorescence density or blooming area and observed honey bee visitation rate beforehand and afterwards. As any biological researcher knows, the power of analysis is dependent upon the quality and quantity of data. So where did the data set come from? We collected it at the field station across the street from campus of course! Choosing Plants Each set of California Buckwheat consisted of three plants - two plants to undergo the density or area manipulations and one control. As such, we would like for these plants to be a similar as possible such that differences in honey bee visitation rates can be attributed to our experimental manipulations as opposed to climatic conditions or differences in the initial number of flowers on each plant. Similarly, if you want to see which store people prefer buying honey from, you would not want factors such as distance or price to impact consumer choices.

Kate (HMC ‘22) and Tom (HMC ‘22) collecting GPS data on a set of California Buckwheat at the Bernard Field Station.

However, plants are all different and it was difficult to find three exactly similar plants no matter how hard we tried. Hence, we decided to prioritize choosing plants of roughly similar size within close proximity to control for external variables such as honey bee preference based on salience and climatic conditions. Secondarily, we selected sets of plant in similar stages of bloom and with comparable initial inflorescence densities to minimize the extent to which honey bee preference based on movement and pollen collection efficiency. The plants were then randomly assigned treatments. One plant served as the control and was not altered. The other two plants underwent manipulations that reduced the total number of inflorescences on the plant by roughly fifty percent. In one manipulation, roughly half of the inflorescences were removed from around the edges of the plant. As only those inflorescences towards the center of the plant were left, this reduced blooming area leaving while local inflorescence density unchanged. In the other manipulation, approximately half of the inflorescences were removed evenly from the plant. As the inflorescences had been thinned out, this decreased local inflorescence density while keeping the blooming area constant. 

Two manipulations were performed on each set of California Buckwheat. The “Density” manipulation altered local inflorescence density while not impacting blooming area. The “Area” manipulation altered blooming area while not impacting local inflorescence density.

While it took a bit of practice, we eventually got the hang of both manipulations and enjoyed the chance to chat while working together as the need for simultaneous observation meant that much of our time at the field station was spent quietly counting bees on our own. Counting Inflorescences and Flowers

One of the most frustrating parts of data collection was counting the number of inflorescences and flowers on the selected plants. The number of inflorescences allows us to approximate local inflorescence density. Moreover, counting the number of open flowers per inflorescence gives insight into the blooming stage of the plant as well as a metric for the attractiveness of the inflorescence within the selected quadrat region to honey bees. To make it easier on ourselves, we chose a 25 cm by 25 cm quadrat region for observation. In addition to allowing for more precise honey bee visitation observation measurements and shorter observation times, this meant we only needed to count the number of inflorescences and flowers within the small selected square as opposed to on the entire plant. Moreover, we found that the number of open flowers on a few clusters could be used to accurately estimate the number of open flowers per inflorescence, saving us time and aggravation. 

Try to count the number of inflorescences within this quadrat and number of open flowers on this cluster!

We found twenty three inflorescences within the quadrat region above and counted thirty two flowers (not all of which are visible from this angle). We then estimated approximately one hundred and forty-five flowers for the entire inflorescence, meaning over three thousand open flowers within the entire quadrat region. The number of honey bees that visited this region during a fifteen minute period was counted before manipulations were performed, thirty minutes after manipulations were performed, and twenty four hours after manipulations were performed. This allows us to analyze both the short and long term effects of the two manipulations to local inflorescence density and blooming area compared to a baseline visitation rate provided by the control plant as well as those measurements taken prior to manipulation. To Bee or Not to Bee? Observing bees is actually quite fun, as they have a tendency to bumble around in an endearing manner and are rather fuzzy up close. But, for accuracy, it is important to be able to tell what is a honey bee and what is a mimic. 

These are all honey bees (Apis mellifera)! You can tell from the wing shape and stripe pattern.

Honey bees, like those above, have a barrel-shaped body that appears fuzzy and two pairs of wings. Mimics, like those below, generally have thinner bodies (wasp), a single set of wings (hoverfly), or no wings at all (longhorn beetle). Another useful way to distinguish between honey bees and their more fraudulent brethren is by their flight patten. Bees typically bustle from flower to flower in a slow yet direct manner that appear orderly. Wasps' and hoverflies' flight style typically involves scanning behavior and more back and forth movement. Meanwhile, longhorn beetles cannot fly and so crawl up inflorescence stems to reach flowers.

These are not bees! They are a wasp and a beetle, respectively.

One would think that after spending nearly three weeks painstakingly creating a procedure for an experiment, everything might go according to plan. And one would be right. Putting in time early on to think through the experiment design and future analysis definitely paid off. We knew exactly what data we wanted and how to collect it before stepping out into the hundred degree weather. This is not to say that everything went down without a hiccup - forgotten cables for flying the drone as well as hot temperatures and windy afternoons, less than ideal conditions for western honey bees which prefer more moderate temperatures and low wind, complicated our ventures to the field station. But overall we had a lovely time collecting field data and the data we collected was definitely worth it. Further Reading:

Donaldson-Matasci, Matina. (2013). Honeybees and Monocultures: Nothing to Dance About. Scientific American, Guest Blog. Kearns, Carol A., and David W. Inouye. (1993). Techniques for Pollination Biologists. University Press of Colorado, Colorado. Sih, A., and Baltus, M. (1987). Patch Size, Pollinator Behavior, and Pollinator Limitation in Catnip. Ecology, 68: 1679-1690. Image Credit: All unattributed photos by Macallan Penberthy.