Data.
Throughout the analyses, the experimental units were the insects in the Orders Lepidoptera and Coleoptera that were caught in a light trap at the University of Copenhagen Zoological Museum, Denmark from 1992-2009 (Thomsen et al., 2016). These were the only two Orders provided by the dataset. These units were used to make inferences about the total Lepidopteran and Coleopteran insect populations of Copenhagen.
The predictor variables were monthly precipitation (in millimeters) and monthly average temperature (in degrees Celsius) for the city of Copenhagen. Both predictor variables are continuous numerical data and were collected from ClimateEU software (Marchi et al., 2020). The response variables were monthly and yearly Family counts for both insect Orders. The response variables are discrete numerical data. Table 1 illustrates how the data were organized for this project.
Table 1. Example of the standard data table used for analysis of monthly insect family counts between 1992-2009 and the relationships between insect counts and climate variables.
Categories
Response
Predictors
Exploratory figures bring the data to life!
The figures below show how this project's data looked before any statistical tests were performed:
Notice how seasonal temperatures vary little among years!
Figure 5. Monthly average temperature (in degrees Celsius) of Copenhagen, Denmark from 1992-2009. Only the months when the University of Copenhagen's Zoological Museum was consistently running their Robinson light trap are visualized (April-November). Data sourced from Climate EU (Marchi et al., 2020)
Notice how seasonal precipitation varies quite a bit within years!
Figure 6. Monthly average precipitation (in millimeters) of Copenhagen, Denmark from 1992-2009. Only the months when the University of Copenhagen's Zoological Museum was consistently running their Robinson light trap are visualized (April-November). Data sourced from Climate EU (Marchi et al., 2020)
Notice how July and August have the highest counts, accompanied by large variations!
Figure 7. Total monthly insect counts of ten Coleopteran Families trapped from 1992-2009 in a Robinson Light Trap atop the Zoological Museum of Copenhagen, Denmark. Total counts were calculated only for April-November and only using families with at least one monthly count above 50 individuals are shown. Data sourced from Thomsen et al., 2016.
Notice how July has the highest median count, accompanied by the largest variation!
Figure 8. Total monthly insect counts of thirteen Lepidopteran Families trapped from 1992-2009 in a Robinson Light Trap atop the Zoological Museum of Copenhagen, Denmark. Total counts were calculated only for April-November and only using families with at least one monthly count above 50 individuals are shown. Data sourced from Thomsen et al., 2016.
Notice the strong positive skews at the right ends of the plots!
Figure 9. Normal Q-Q plots for insect counts of two Orders. Counts were calculated only for April-November and only using families with at least one monthly count above 50 individuals. Data sourced from Thomsen et al., 2016. A. Coleopteran normal Q- Q plot, counts are the sum of 10 Families. B. Lepidopteran normal Q-Q plot, counts are the sum of 13 Families.
Notice the zero-inflation leading to the need for non-parametric statistics!
Figure 10. Histograms for insect counts of two Orders, each with 25 breaks. Counts were calculated only for April-November and only using families with at least one monthly count above 50 individuals. Data sourced from Thomsen et al., 2016. A. Coleopteran histogram, counts are the sum of 10 Families. B. Lepidopteran histogram, counts are the sum of 13 Families.