Introduction

I worked on this project with a team of 3 others: Noor-e-jehan Umar, Emily Sine, and Nolan Harrington.

Dataset

The United States Department of Agriculture collects quarterly survey data from beekeepers across the country to keep track of bee colony loss rates as well as risk factors. The bee colony datasets were used for TidyTuesday on January 11th, 2022 and contain information about colony numbers as well as colony health stressors from 2015-2021. Surveys were sent to operations which have reported having five or more bee colonies to the USDA.

In one dataset, called colony, each observation contains the year, the month range, and the state where colony numbers were reported. These numbers include number of colonies , maximum number of colonies, lost colonies, percentage lost, colonies added, colonies renovated, and percentage renovated within that time period. In the other dataset, called stressor, each observation contains the year, the month range, the state, the stressor (ie. the stress type experienced by the bees), and the percentage of bee colonies affected by the stressor in a given quarter. The dataset also takes into account that a colony can be affected by more than one stressor in the same quarter.

Question 2: Colony Renovation and Stressors

How do efforts to renovate colonies after 2015 impact colonies experiencing stress?

For this question, we will compare the number of colonies renovated with the number of colonies affected by stressors to determine if renovating colonies has any sort of impact on how many colonies experience stress. A renovated colony is one which has been re-queened or has received a “package” of bees from another colony in an effort to strengthen it (“Honey Bee Colonies”). Our current assumption is that the higher the number of renovated colonies, the lower the number of colonies affected by stressors will be, and vice versa.

We are interested in this question because we want to determine whether or not renovating colonies has helped combat the colony losses caused by health stressors. With more efforts to renovate, it is concerning that bee loss rates are still so high, and it is important to understand how health stressors are or are not changing as a result of renovations so that beekeepers can either ramp up renovation efforts or perhaps focus on other methods of protection. Ultimately, we know that climate change plays a huge role in the demise of bee colonies, so it is possible that renovations alone will not be enough to save these colonies.

Approach

Scatterplot:

For our first graph, we want to see if there were any trends in the number of colonies renovated and the number of colonies stressed. We are interested in whether more renovation leads to less stress. We think a good way to visualize this would be to create scatter plots with the independent variable, colonies renovated, on the x axis, and the dependent variable, colonies stressed, on the y axis. The data points we use are taken across years from 2015-2021, where each point represents the number of stressed hives and number of renovated hives per each state and each year. We will aggregate the data to be for each year rather than each season, because we assume that the impact of renovating hives might take longer than just one month. Furthermore, we will facet by stressor to see whether there are any trends specific to certain stressors, particularly because we saw that colonies had certain stressors that were more prevalent than others, such as Varroa mites.

Beeswarm plot:

Since we noticed that there was more of a significant relationship between the number of colonies renovated and the number of colonies stressed for the Varroa mites stressor as compared to the other ones, we decided to focus on this for our next plot. We will plot the year on the x axis and the percentage of colonies renovated on the y axis, and then color the points in the beeswarm plot by percentage of stress, using a continuous color scale. We feel that this is the best way to visualize the data because we are trying to see if there are any significant trends between percentage of renovation and percentage of colonies affected by Varroa mites since the 2015 policy change which we discussed in our first question (with year being used as a categorical variable because we will visualize trends for each year on the x axis).

Analysis

Discussion

Through the results of our analysis, we found that for most types of stressors, there was little to no relationship between the number of colonies renovated and the number of colonies stressed. This was interesting because we had initially assumed that a higher number of colonies renovated would lead to a lower number of stressors. However, when it came to Varroa Mites (the top stressor), we actually saw a positive correlation between the number of colonies renovated and the stress experienced by the bee colonies, showing that more renovations happen in the same year when more colonies are stressed. This led us to believe that perhaps the colony renovations were not really helping, and there were other factors at play. We knew that Varroa Mites was the most common stressor affecting colonies, which might be the reason that it had the most stress, regardless of how many colonies were being renovated; It is also possible that a larger number of colonies were being renovated DUE to the fact that they were experiencing increasing stress, and that the effect that renovations had were not immediately available with this data.

After looking more closely at the Varroa Mites stressor through the beeswarm plot, we confirmed our suspicions that colony renovation did not seem to be highly correlated with the percentage of stress experienced by colonies. The darker colored points (or more stressed data points) were evenly scattered across the plot, and there were no significant relationships i.e. clusters of colors showing an increase or decrease in the percentage of stress experienced. It is interesting to note that the 2015 White House Executive order for Pollinators increased funding for varroa mite treatment. In the swarm plot, we do start to see more treatment of highly infested colonies following the year 2015, but because of the dataset, we cannot make this claim. Again, this suggested that other factors are affecting the amount of stress experienced by hives.

To see part one of this project, go here.

Data

Mock, Thomas, 2022, “Bee Colonies,” Tidy Tuesday: A weekly data project aimed at the R ecosystem, viewed 16 February 2023, https://github.com/rfordatascience/tidytuesday/tree/master/data/2022/2022-01-11.

References

“Colony Collapse, Climate Change and Public Health Explained.” GW Online Public Health, 19 Sep 2019, https://onlinepublichealth.gwu.edu/resources/colony-collapse-climate-change-public-health/. Accessed 28 Feb 2023.

“Honey Bee Colonies.” USDA, National Agricultural Statistics Service, 2022, https://usda.library.cornell.edu/concern/publications/rn301137d?locale=en. Accessed 28 Feb 2023.

Hunt, Greg & Given, Kristen. Entomology, P. E. “SELECTING A TERMITE CONTROL SERVICE.” Purdue University, https://extension.entm.purdue.edu/publications/E-2/E-2.html. Accessed 3 Mar 2023.

“Legislation.” NY State Senate, https://www.nysenate.gov/legislation/laws/AGM/A15. Accessed 3 Mar 2023.

Milman, Oliver. “US Beekeepers Lost 40% of Honeybee Colonies over Past Year, Survey Finds.” The Guardian, 19 June 2019, https://www.theguardian.com/environment/2019/jun/19/us-beekeepers-lost-40-of-honeybee-colonies-over-past-year-survey-finds. Accessed 28 Feb 2023.

The White House. “National Strategy to Promote the Health of Honey Bees and Other Pollinators.” Pollinator Health Task Force, 2015, https://obamawhitehouse.archives.gov/sites/default/files/microsites/ostp/Pollinator%20Health%20Strategy%202015.pdf. Accessed 2 Mar 2023.