Category: Post 6: Data Collection

I have sampled twelve replicates. There have been a few issues with collecting data, such as implementing the sampling strategy, systematic sampling. When sampling, there was two instances in which I had to turn around and find another spot to continue, unless I wanted to fall down a cliff. Another issue I have run into is that the past two months, there has been quite a bit of precipitation. This has made it difficult to obtain soil samples that are not uncharacteristically wet.

While collecting my data, I noted that the distribution of sagebrush on hill tops is approximately what I expected. However, there was one quadrant in the valleys where there were more sagebrush than expected.

I went out and sampled 5 plots of 1m x 1m each at my site.  I did not have any problems implementing my sample design.  I did notice that the less grass there was the more thistle there was.  The taller the grasses were the less there was of thistle.  And in plots that did not have a lot of grasses there appeared to be significantly more amounts of thistle present.  This goes well with my hypothesis.

Since originally making my hypothesis and making my data collections, I have decided to revise my sampling method slightly: Since I am measuring fallen leaves, there is no way to guarantee the leaves beneath a given tree were grown on that particular plant. Giving more observation to the distribution of leaves on the ground, I shouldn’t be able to make the claim of tree origin with any confidence, so I am going to instead group sampled leaves to a “habitat area.” The aggregation of the data means that I would not be able to look for associations in the data between leaf length and individual tree. However, the integrity of the data is preserved for the purpose of comparing variance within and between and groups (ANOVA), except that there is an increased risk of a Type II error. (Which is better than a Type I error, in my opinion).

I have thus sampled trees within each of two different habitat areas (one near the construction site and roadway; the other farther away) on three separate occasions. In total, 80 leaf length samples were taken from each habitat area over the data collection period.

Throughout this module I needed to get 3 replicate samples completed. I’m collecting a total of 15 replicates (5 from initial data) and either 3 or 4 replicates per module. I haven’t had any problems with implementing my sampling design so far… I will continue to go to my point count stations in the morning when it is nice and cool and I will record the amount of times I see a bird using the area. This no longer takes into account any birds that fly by or and bird calls since they are not technically using the area. I think this will give me more accurate data. So, because of this decision, only part of my data from the first 5 initial observations I recorded is valid and, therefore, I will only be being using half of the data. I will return to all my 5 point count stations in the same order that I did the first time and I’ll continue the process of returning to these stations one at a time over multiple days… This also gives me extra time to complete other parts of this module. So, on Aug 20^th I returned to my first point count station. This station is the station with no shelter, this is the lawn area station. It was 15 degrees Celsius, however, there was a layer of smoke blanketing the area. I only saw 1 bird using the area during the whole 10 minutes… I have a feeling this is due to the smoke. I don’t know how long the smoke will stick around and I’m curious if the smoke might skew my observations; however, I can’t wait forever for it to go away and I will just have to take this into account when I am analysing the data. I then returned to my second station a couple days later on Aug 22^nd . It was 16 degrees and the thick layer of smoke was still glomming above. The second count station is the intermediate station with half shelter and multiple bird feeders within the area.  Even with the thick smoke I saw 3 birds use the area. The next day, Aug 23^rd  ,  I went to visit the third count station. There was no smoke!!! I was a nice morning with clear skies and it was 15 degrees Celsius.  This station is the one right in the middle of the forest with lots of shelter. Here I recorded 3 birds using this station. I will visit the rest of the stations sometime throughout the next couple of days and be finished up by module 8. From the data, I have recorded I am starting to see more of what I was expecting. I did in fact more birds in areas with more shelter than in areas of low to intermediate shelter. I will be very interested when analysing my final data to see if the smoke did in fact have any real affect, but hopefully it will be gone for good now.

I completed my field data collection over the span of two days from 12:00-14:00 at Creekside Park. Throughout the past month, I have been visiting the site and thinking about my research project, notably the hypothesis and study design. My original hypothesis, “the abundance of bees increase around vibrant colourful flowers and decrease surrounding pale white flowers”,  had some subjective terminology that needed clarification. The words ‘vibrant’, ’abundance’ and ‘pale had to either be removed or changed into an objective form. I switched ‘abundance’ with ‘number’, and removed ‘pale’ and ‘vibrant’. I ended up changing my hypothesis to “the number of bees increase around colourful flowers (purple, red, and yellow), and decrease around white flowers”.

I used a random sampling technique to look at  bees around 10 flower samples in the park. I have not experienced many problems with the implementation of my study. The accuracy of my study has improved by implementing my sampling on weekdays, a time frame where their are less people to disturb the bees and flowers.

My field data collection was ultimately collected as a “snapshot” of the site on one day.  I had been planning to collect data on multiple days over a period of weeks, however, it should be unnecessary to collect repeat information in such a short time span.  Increased sampling could also introduce confounding factors, such as counting invertebrates more than once and skewing the data to show greater abundance and diversity when there is none.

I sampled two sites, with each site consisting of three transects.  Because the study area was small, it was possible to sample the entire site with 100% coverage.  Introducing randomization for the sake of randomization in this case would have introduced unnecessary rates of error into my study.  This type of sample design was relatively easy to implement.  My only concern was in counting mobile species more than once.  Luckily, when I did encounter highly mobile species, I was able to keep an eye on them to reduce the error of counting species twice.  In a larger site, it could be beneficial to introduce smaller plot sizes and randomization to avoid counting highly mobile species more than once.

I have noticed a greater diversity of invertebrates in the native plant garden setting, which I expected, but I did not notice a greater abundance of invertebrates, which is something I had expected to observe.  Finding a small colony of ants in the turf lawn setting could account for this discrepancy as their numbers could artificially skew the data.