Category: Nancy Elliot

My study design focuses on measuring leaf length of fallen leaves. While this is admittedly easier than climbing trees to collect leaves, fallen leaves bear little loyalty to their parent tree and could blow wherever the wind takes them. Additionally, there is a risk of sampling bias, as more dried leaves take on a dull red colour compared to some of the freshly fallen leaves that have a more vibrant colour. These leaves tend to stand out, which is great but there is some variation in leaf colour, even among freshly fallen leaves. This means those of a duller colour are more likely to be missed. In response to this, I think I will need to use a small quadrat to focus my leaf collection to be exhaustive within a fixed area. That way, all leaves are evaluated to determine if they are freshly fallen enough to be sampled. (Dried leaves are not used because they cannot sustain the manipulation required for measuring and their reduced weight makes them more likely to be dispersed nearby trees by wind from.)

Initial data collection was successful. Point count locations were selected based on visibility within the drainage channels. Point count locations were also selected based on varying densities of emergent vegetation within the drainage channels in order to provide a representative sample. Point count duration at each point was 5 minutes. Emergent vegetation cover was visually estimated within the visible channel sections. Visibility was greater than 100 metres (m), however, I capped channel section observations to between 60 and 80 metres to increase both the accuracy of waterfowl identification and the overall precision of vegetation estimates.

Prior to my data collection visit on August 17th, I had decided to use a systematic sampling strategy where I overlaid the sitemap with a grid and used a random number generator to select sampling locations. This strategy was unsuccessful due to limited visibility at the randomly selected points. Randomly selected point count locations also did not take into account disturbance of birds within the channels. As the site is diked and exposure to human activity is common, it quickly became important to not only select vantage points that provided adequate visibility but also to select points where waterfowl would be less likely to be disrupted. This modification will improve the accuracy and reliability of my data.

The data collected was generally consistent with my hypothesis in that the total number of waterfowl observed increases with increasing emergent vegetation cover.  I did notice that the total number of waterfowl observed at all point count stations was fairly low. As a result, I will need to ensure that I incorporate additional point count stations and implement enough replicates in order to identify any relevant trends. Initial field data was collected during the hours just before sunset. I intend to collect data just after sunrise in order to determine when waterfowl activity is greatest.

Although the systematic sampling strategy did not pose any major difficulties, I need to consider modifying the strategy slightly to ensure that my replicates are independent of one another.  To ensure independence, I will select channel sections that are at least 100 m apart. This will also reduce the likelihood of counting waterfowl twice. Based on the layout of the survey area, any incoming individuals from nearby point count locations will be highly visible and will not be recorded.

I would also like to be more specific as to what constitutes “emergent vegetation”. Emergent vegetation, for the purposes of this study, will be restricted to yellow pond lily (Nuphar lutea), as it is the most dominant emergent aquatic species within the drainage channels and occurs at varying densities.  This modification will help me to generate a clear and specific research hypothesis and experimental design.

I conducted my research using a simple random sampling technique to sample how many bees were around different types of flowers. To limit my bias, I used a random number generator on my computer to come up with the number of steps to take (between 1-20 steps) and the compass bearings (between 0-360). I used a .5 meter squared quadrate to analyze how many bees were surrounding the flower of interest. I also used a point count sampling technique to observe how many bees were either on the flowers or an inch away from them for a time period of five minutes. I located 6 flower samples using the random sampling technique and measured the amount of bees surrounding them three times.

I had a couple difficulties implementing my sampling strategy. Some of the coordinates from the random number generator lead me to areas with no flowers. For example, the 3 steps and 213 degree compass bearing led me to the middle of the playground. Another difficulty that interfered with my data collection was disruption from children. Some kids playing around the park would approach the flowers I was observing which may have scarred the bees away. The data I collected did not surprise me in any way because it matched my hypothesis. Overall, I believe that my sampling technique works really well for my research. The one modification I will make to improve my research is to make my observations earlier in the morning to limit the disruption from children.