Category: Post 8: Tables and Graphs

The data I collected was the number of trees with and without ivy in a 5 m radius. Each block, A through E, had five replicates. It was difficult deciding what sort of statistics to conduct on the data. I wanted to keep it straightforward and focused on the abundance of ivy on trees within the radius and then broaden out to look at each block as a whole. I believe looking at the relative frequency of the presence/absence of ivy will achieve this. Since the blocks vary in size it could potentially be helpful to standardize the frequency somehow between the blocks.

For my research data, I decided to analyze it using the Kruskal-Wallis statistic test. This allows me to compare the median of my response variable to multiple intervals of my predictor variable. I used Minitab 18 to coordinate the data and calculate the medians, means and standard deviations which I then used to for the graph below.

Fig.1. Average squirrel abundance measured at View Royal Park in relation to the number of domestic dogs present.

The average amount of squirrels was calculated within each interval and plotted here with error bars that were calculated using the individual standard deviation. There is an inverse relationship between the number of dogs present and the number of squirrels observed. The statistical test used to analyze this data was the Kruskal-Wallis test. The P-value was found to be P<0.05, therefore, we can reject the null hypothesis and all the means are different.

I found that summarizing my data into a visually representative table or graph was relatively easy. The table I submitted summarizes the number of individuals (abundance) counted within the two habitat areas within Terwillegar Park. My prediction was that bird abundance would increase with increase in tree/forest cover. When I summarized my data into the table it was noticeable that the highest number of individuals was the forested area.  In my paper I plan on creating a visually representative bar graph of the data collected as this will display the difference in abundance easier than in table form.
 

Table 1: Recorded point count data from Terwillegar Park, Edmonton, Alberta.   

The figure below (Figure 1) depicts dabbling duck abundance at my selected point count locations (1 through 8) in Colony Farm Regional Park. My research focused on habitat selection by dabbling ducks within constructed drainage channels that varied in percent cover of emergent vegetation.  I initially wanted to show the average abundance per point count location, however, the total number of ducks observed during my data collection was lower than I would have hoped. Thus, I used the total number of dabbling ducks observed at each point count location. The low numbers also presented some challenges later on when assessing statistical significance, as it would likely have benefited me to have greater sample sizes to work with to truly assess any preference or habitat selection by the ducks I observed. Summarizing the data I did have, however, was fairly straightforward, and is shown below. 

Figure 1. Total number of dabbling ducks counted at point count locations 1 through 8 within drainage channels of varying emergent vegetation percent cover.

During data analysis, I ended up grouping the point count location results into three groups based on emergent vegetation cover: Group 1 = <25% cover, Group 2 = less than or equal to 25% but <50% cover, and Group 3 =  greater than or equal to 50% cover. I divided the groups in this manner to see if there proved to be a significant difference between the varying amounts of emergent vegetation cover and how this may have affected abundance.

Since my initial prediction was that duck abundance would increase with increasing emergent vegetation cover in the drainage channels, my results surprised me. I was not expecting so much variation in my results, with the greatest number of ducks observed within the channel that had only 30% emergent vegetation cover. I also discovered that variations in duck abundance between groups were also not statistically significant.

In future, I would conduct this study later into September when more dabbling ducks are present within the park to generate a larger sample size (seasonal effect). With greater resources, I would have also measured other factors within the drainage channels, including water quality parameters (dissolved oxygen, pH, temperature) and perhaps also water depth. With more time, I also would have conducted sampling events during different seasons over multiple years to see if this affected my results.  It is clear from relevant literature that I have come across during my research project that understanding habitat selection for species can play an important role in habitat management and conservation of protected areas.