My initial field data was measuring the diameters of sagebrush (Artemisia tridentata) along an elevation gradient. I had hypothesized that sagebrush diameters decreased as elevation increased and so far the data does not seem to falsify this. The data I have collected is not surprising. My sampling strategy was systematic sampling as I had to follow the elevation gradient up along certain paths that were suitable for walking on. The only minor difficulty I had was the elevation measuring app I had on my phone did not seem extremely accurate so I had to confirm the elevations on a computer program afterwards. Now that I know the proper elevations for the park it will be easier to continue sampling the rest of my data.
Category: Post 5: Design Reflections
Blog Post 5- Design Reflections
Post 5: Design Reflections
My initial data collection for Module 3 involved taking systematic quadrant samples of native & invasive alien plant communities within two separate locations. Using the multiple transect sampling technique, ten quadrants were laid out along each transect, roughly 5 meters apart. Each quadrant sampled was 1m/square in size, using a homemade PCV pipe observation quadrat device. The first sample location site (Site A) was taken along an of leash dog area; A coastal bluff which was comprised of mostly degraded bare soil & rocky outcrop, also greatly worn due to human & canine traffic. The second sample location site (Site B) was an area that is *supposed to be* off limits to canine & human traffic. I found that this was clearly not the case, as human & canine activity is often observed in these no access areas & unfortunately, I suspect my results were skewed due to this. This location was in the same park as my first site, although the samples were taken approximately 100 feet away from the first site throughout a sloping wetland, lush marsh / meadow. The park contains pubic info boards which describe that it contains SARA plant populations (Federal Species at Risk Act) & to obey the signage which clearly indicates out of bound locations.
I did have some difficulties when implementing sampling strategies for my second sampling site (Site B), in the marshland / meadow transect sample areas. I did not want to tread on a location that was “out of bounds” to the public, so I leaned over the fence (which is only about 3 ft high) & floated my sample device over top of the sample area, recording my observations as I went along the transect. The first sample site was much easier (Site A), as I could actually place my sample device on the ground, making it much easier to observe, write down my observations, & even take pictures when necessary.
The data collected at Site B was quite surprising to me. I expected to find a greater diversity and density of plant species observed, due to the fact that this area was restricted. Unfortunately I found that most people (including a few mountain bikers!) did not observe the areas that contained SARA species & were clearly marked as out of bounds. My low plant counts in Site B reflected this. Site A is a well used, high foot traffic area, so I did expect the plant species & density to be a bit lower than Site B. I was not expecting that my results would show the density & diversity to be as similar in numbers as it turned out to be.
With regard to collecting data, & for the sake of my field research project, I will modify my approach slightly. I will still use the sampling device I made & method used, but I will observe my samples in different areas that what was chosen for Module 3. I will expand my transect line & quadrats out by a few more meters between each sample. I am hopeful that these modifications will improve my research. My explanatory variable – I will observe plant composition communities along a drier & more exposed coastal bluff in one site, vs. plant composition communities observed in a higher elevation site with a different overstory (meadow). I will be looking to observe plant composition community differences between site A (bluff) and site B (meadow).
Blog Post 5: Design Reflections
My initial study design includes the collection of percent covers of moss species along an environmental gradient, i.e. the slope position on a bedrock outcrop. I found that deciding on which sampling strategy to use for this exercise to be somewhat challenging. I chose a strategic sampling strategy in which samples were collected at regular intervals along a gradient. A transect was established along the environmental gradient, starting at the base of the western slope of the rock outcrop, over the crest, and down the eastern slope to the depression. Samples consisted of 18 cm by 18 cm square quadrats because this is the size that my foldable ruler, which was readily available, could produce and it seemed like a reasonable size (i.e. not too big or too small). Quadrats were located along this transect and were spaced every 40 cm.
A challenge that I had when it came to implementing this sampling method was that due to the variation in the micro-topography, there was no single transect line that would include the representative areas in each of the slope positions. I therefore had to deviate from the transect line so that the quadrats located on the western slope were all along the same transect, those along the crest and eastern slope along another, and those in the depression along yet another.
At a glance, the results of this sampling did not show a strong relationship or pattern in the percent covers of the various species. The most notable pattern was that RACCAN occurred only on the crest of the rock outcrop and PYLSPL occurred only in the depression, while DICSCO and PLESCH were present in all of the samples (with the exception of one, which lacked DICSCO). Note that 10 samples were collected instead of just 5 in order to obtain adequate representation of all of the slope positions on this particular rock outcrop while maintaining consistency in sample spacing.
I think the sampling strategy should be modified to include a broader study area, i.e. additional rock outcrops. This will broaden my scope and allow for the collection of more samples, and will also decrease the impact of any localized patterns that may not be representative of other outcrops in the area. I will try out a couple of different strategies; one option being to collect one random sample from each slope position on several different rock outcrops, and another being to simply repeat the same strategic transect methodology (but perhaps using a spacing of 20 cm instead of 40 cm in order to increase the sample number) on multiple rock outcrops. The sampling method used worked well and I will continue to use this method.
Blog Post 5: Design Reflections
During the initial data collection efforts at Surrey Bend Regional Park, there were no difficulties in implementing my sampling strategy. However, after consultation with the professor we decided that the habitats being examined were not similar enough to compare, and that the anthropogenic influences on them were too different to be able to effectively account for any changes in bird species presence and abundance to only one explanatory variable. As a result, I have decided to change the location of my research project to Burnaby Lake Regional Park and the surrounding area, and focus the study on examining bird species presence and abundance at three sites along an urbanized gradient.
My revised hypothesis is: Bird species presence and abundance is impacted by the percent cover of natural habitat at a site.
My revised predications are:
- Bird species richness will be highest in areas with the highest percent cover of natural habitat.
- Bird species abundance will be highest in areas with the highest percent cover of natural habitat.
My response variable will remain as bird species presence and abundance (continuous variable), and my explanatory variable of percent cover (categorical variable) of natural habitat (forest, wetland, etc…) vs. anthropogenic habitat (roads, buildings, trails) at each site as a whole will also remain the same.
In addition the followings changes will be made in the experimental design in regards to the way data is collected:
- Initial sampling took place mid-day and overall detectability was low. It is standard bird inventory procedure to sample only between dawn and 10:00am to increase detectability. I will now perform all my surveys during this period to maximize the birds being detected by sight and song.
- My new sampling strategy includes two survey sites (point count surveys) in each of the three habitats along the gradient. During the replicate exercise, I expanded the number of point count surveys in one of the sites to five. While it would be ideal to maintain this type of replication it is too time consuming to perform five replicate surveys in each of the three sites. As a result, I will perform two replicates per site, and will conduct surveys at each site on at least three different days.
During the next blog I will go into more detail on the sampling strategy, sample unit, how these changes have impacted the data collection, and any ancillary patterns that appear in the data to support the new hypothesis and predictions.
Post 5: Design Reflections
After walking around Milliken District Park and observing my potential sampling sites, I got a better idea of how I would collect my data and implement my sampling strategy. Prior to doing this, I was a little confused on how I could collect data, but my initial observations helped me make more sense of it. As such, I did not face much difficulty in actually implementing my sampling strategy. I believe that the data that I collected was actually not surprising in any way. I expected some discrepancies, and was somewhat expecting to see some differences in the distribution of the six species, as there are factors such as sun exposure and climate that come into play. As a result, I plan to continue using the same technique in order to collect data. The same technique can also promote consistency in the result obtained. I believe that the time and weather that samples are collected in will have an impact on the results obtained, and thus I will have to take this into consideration when analyzing.
I decided to use the systematic sampling strategy as I felt it would be the best choice in my sampling areas. The first site, Site A, is within the forest, the second site, Site B, is outside of the forest, just before a large pond, and the third site, Site C, is the area on the other side of the pond. Percentage coverage was measured using a 1m2 quadrat, the abundance of the species using a 0.5m2 quadrat, and absence/presence using a 0.25m2 quadrant. These measurements were determined based on the amount of plants, as well as the size of the plants present in the area selected. Each of the three quadrats was placed randomly five times at each site, and data was collected. Abundance was measured using the ACFOR scale (Abundant, Common, Frequent, Occasional, or Rare), which will be explained in more details below. When looking at the variable of presence/absence, an ‘X’ represents the species that were present, and no ‘X’ represents that the species was absent in that region.
Post 5: Design Reflections
After collecting this initial field data, I realize that I might have to expand my site to include beaches along the same coast. Farlow’s seaweed seems to grow very well when it is on its own in a tidepool. However, Farlow’s seaweed doesn’t seem to grow in many tide pools with higher diversity. To collect a large enough sample size, I might need to continue down the coast to visit more tidepools. A larger sample size will give me a more “solid” conclusion.
In addition to this, when collecting samples, the seaweed could not be completely taken off the rock. A small portion of its base remained strongly attached. The seaweed that was collected also contained other organisms. I separated the larger ones that would make a significant difference in the weighing but the smaller ones remain attached.
The major problem I encountered was that my scale only works in whole numbers so my precision is limited. I ended up having to weigh multiple samples at once in order to make the samples weighable. The samples less than one gram in weight were not detected by my scale.
The data that I’ve collected so far is not surprising. It does follow my initial hypothesis.
Blog Post 5 – Design Reflections
My first visit to get my five samples did not go well. I knew going in that it was going to be tricky, and it was. Sampling in the deeper water was difficult, even with a long aquatic net. The samples were dirty and difficult to count. Even with the extended reach, I knew I was disturbing the water too much and mostly likely scaring insects away. It was a lot of fun though, haha!
A new plan came to mind, and my second visit went much better. I decided to run a transect line 3’ from the shore, marked every 4’. Along the shore the plant density varies from no plants to very dense, and not in a predictable pattern. I cut the bottom off a five-gallon bucket and placed that down at the transect marker. Because I wasn’t too far from shore I was able to do it without disturbing the water too much, and it kept the insects contained. The bucket is white and water quite clear, but it was still too difficult to count without taking a water sample. I lightly disturbed the water with the net so the insects would swim up out of the plants. I tried to do it in a circular pattern for a three second count (so each is roughly uniform), and then removed the sample with a canvas net. I then placed it into a white bucket so it’s easy to see the insects. I realize I am not getting all the insects in that sample space, but hoping for a representative sample. I found the best way to count the insects was to focus on each insect species separately. The plants were easy to count after the insect sample was out. This plan came about with a lot of trial and error that day. I feel like this is the best I can do with the resources I have. The one thing I will do differently is rather than taking the samples in a line, I will do transect #10 then #5, #9 then #4 etc., to reduce scaring neighbouring insects away from the next sample area. Also, I can’t return any insects to the pond until I am finished so I don’t double count!
My predictor variable is plant density, and my response variable is number of aquatic insects. My initial data does show more insects in the denser plant areas. I noticed one type of insect only appeared in the samples with plants. It isn’t the best representative sample as the majority of them had no plants, and one landed on top of a big rock. But the rock sample had more insects than the two samples with no plants! I also recorded the number of different insect species within each sample:
Blog Post 5. Design Reflections
The first day I went counting is not included in the dataset as the count was made at 6 p.m. In order to avoid intense heat during the day I decided that it will make sense to do the counts later in the afternoon, but only a few species were found. I found a forum for beekeepers which stated that flights appear from 9 a.m. to 2 p.m. and then training flights of the young can be observed from 3 p.m. to 5 p.m. First two days I used total number of individuals at the moment of observation counts. But for results to be more statistically significant and to avoid errors, further observation included hour counts at the site divided in 5 sets 20 min each which allows me to find the mean value for further analysis.
The data collected so far is consistent with widely accepted hypothesis that urbanization gradient effects abundance and richness of pollinators community, but studies which compare urban and suburban areas rarely include remote suburban areas like McArthur island park and it appears that both values are higher in this park.
Blog Post 5: Design Reflections
The initial sample collection went well and I had no difficulties implementing the sample strategy. The data I collected was very surprising as two of the four trees had buds and those trees were at opposite ends of the courtyard. I expected to collect similar data from trees at similar point in the courtyard, where that was not the case. Tree four had buds which is at the highest point in the courtyard and tree three had buds which is at the lowest point. I plan to continue sampling the trees based on the same approach I used in my initial data collection. Sample collection was easy and I feel like it was effective use of my time.
Post # 5: Design Reflections
- There were some difficulties in implementing my sampling strategies. It is sometimes difficult to measure out transects if people are nearby, it was also not ideal to count individual bees as when many accumulated, it was hard to keep track. It is also hard to set a control as I would physically have to restrict a bush from any bee pollination and legally I am not allowed to do that. I think in modifying my approach I am also going to relocate my experiment to somewhere less busy where I can spend a significant amount of time without being disturbed. So in saying this, I am going to take my study into my backyard where I have an abundance of vegetation and animals. I am now going to study the effects of watering on fruit trees. My hypothesis is: Fruit trees with more watering of the soil will increase better fruit production.
