89  Case Study 2 — Peppermint

Now that we have an idea about some of the geometries available to us, we will use a simple case study to extend this idea to a multivariate comparison. This serves as an introduction to the concepts we will explore in detail in the next chapter.

This data-set comes from a doctoral student at the Max Planck Institute for Biogeochemistry in Jena. She was interested in uncovering the effects of possible future climate conditions on sugar production in a plant - namely peppermint. She measured the sugar concentrations of three sugars at a few time points over several weeks in four different plant parts. She grew the plants under normal water conditions or drought and under low or high carbon dioxide concentrations.

In total, she had over 741 data points classified into 48 different time series (consisting of 2 water conditions, 2 \(CO_2\) conditions, 4 plant parts, and 3 sugars). In keeping with the original plot, I have plotted the means and standard errors for each time point. Note that this produces smaller error bars than the \(95\%\ CI\) or the standard deviation, which are more easily understood.

She began by grouping this into 12 plots according to sugar and plant type. This is our starting point and is shown at the top of the page overleaf. Above the plot is a schematic of the grammar of graphics I used to create it - the blue lines depict the mapping of data onto aesthetics and facets. The plots which follow are variations on this theme and are intended to showcase how changing mapping attributes affects how we communicate information. Subtle changes in mapping can have dramatic effects on information perception.

In this case the central question of the research is how will climate change affect sugar content in peppermint?. The plots are an aid to answering this question, they are only an aid that must be interpreted. We are interested in the change in sugar concentration over time under specific growth conditions. Which of the first five plots on the next pages do you think addresses this question the best? i.e. which plot affords the most suitable route to an interpretation?

After testing multiple different mapping types, the final plot, Figure 89.1, shown below, seems to truly depict a clear and easy-to-understand trend for both the researcher and the reader.

Figure 89.1: The final plot of the peppermint data set showing a clear trend over time under specific growth conditions.