Data

My data table included measures of dissolved concentrations of Ca, Cl, K, Mg, Na, NO2, NO3, SO4, and total amounts of NH4, Zn, Ni, and suspended solids for each of the 35 ponds, all measured in mg / L. These continuous predictor variables were included alongside the response variables of taxon rank, or which invertebrates were present, and their respective counts. The water chemistry data was observed and not manipulated as all samples were collected from stormwater ponds in the Edmonton area, and not produced in a lab setting

Figure 4. Box plot depicting invertebrate counts for each invertebrate Order recorded at Millwoods, Orchards, Pylypow, Silverberry, and Windermere ponds respectively. Missing pond values indicate no presence of the respective invertebrates in that pond.

To visualize the abundance of the different invertebrate orders across all ponds I used a box plot with coloured points that correspond to the 5 different ponds (Figure 4). This exploratory graph allowed me to search for errors or outliers in the data, as well as errors that could have occurred during data entry as I identified the individuals. 

Figure 5.  Principle Component Analysis (PCA) plot showing the correlation between water quality variables across 35 ponds. Fertilizer factor includes NO3 and NH4, and Road Salts / Pollutants factor includes Cl, Na, SO4, Ca, and Mg. Ponds sampled for invertebrates are highlighted with coloured circles, the respective colours correlating with the number of invertebrate orders found at each pond.

To visualize the correlation between the different water quality variables across the 35 ponds sampled for water data I used Factor Analysis, a form of Principle Component Analysis (PCA), which allowed me to see possible relationships, as well as possible outliers in respect to the ponds numbered 1 through 35 (Figure 5). The Fertilizer factor includes NO3 and NH4, while the Road Salts / Pollutants factor includes Cl, Na, SO4, Ca and Mg. I added a colour ramp and points to highlight the 5 ponds with invertebrate data, and to show the possible relationships between the number of different orders present at each pond and the water quality variables. A shortfall of this graph would be that it only accounts for two components.

Figure 6. Correlation matrix depicting the correlation between NO3, Cl, and SO4. A dark blue colour indicates a perfect non-linear (negative) relationship, a dark red colour represents a perfect positive linear relationship, and white indicates an intermediate between neither positive nor negative linear interdependency. 

I chose to plot a correlation matrix (Figure 6) against NO3, Cl, and SO4, as these variables served as a representative set of individual pollutants that had long arrows that pointed in different directions in my PCA analysis (Figure 5). A dark blue colour indicates a perfect non-linear (negative) relationship, a dark red colour represents a perfect positive linear relationship, and white indicates an intermediate between neither positive nor negative linear interdependency. 

I created scatter plots for each invertebrate Order, visualizing their abundance at each pond in relation to the ponds respective chloride level. Here I only chose to showcase the scatter plots for Orders Amphipoda (Figure 7) and Ephemeroptera (Figure 8), as Amphipods are a relatively robust Order, and Ephemeroptera are more sensitive (Beermann et al., 2018). I will point out that while both scatter plots show a relationship between invertebrate abundance and chloride concentration, the Ephemeroptera numbers are significantly lower than the number of Amphipoda found at each pond. 

Figure 9. Bar charts representing the number of individuals of Amphipoda, Anomopoda, Coleoptera, Ephemeroptera, Odonata, and Sphaeriidae in each pond respectively. 

I also created bar charts for each invertebrate taxon, visualizing their abundance at each pond concerning the pond's respective chloride level (Figure 9). This was done to provide another visual representation of the data for accessibility. Missing bars indicate that no individuals of that taxon were found in that respective pond.