Category: Post 6: Data Collection

Blog Post 6: Data Collection

TRU Open Learning Leave a comment

On November 10, 2019 between 10:45 AM and 4:00 PM I collected my field data at Kalamalka Lake Provincial Park along Cosens Bay Trail, Vernon BC. I sampled my three sites, Site 1 Eastern Area, Site 2 Riparian Area and Site 3 Upland Area. On the day of my field collection the temperature was approximately 12 degrees Celsius and throughout the day the weather varied from sunny, to some sun with cloudy conditions.

I sampled Site 1 Eastern Area first between 11:00 AM and 12.35 PM. I sampled Site 3 Upland Area second between 1:15 PM and 2:15 PM. I sampled Site 2 Riparian Area third between 2:25 PM and 3:20 PM. At each site I sampled 10 quadrats, my quadrat was 1.5 m by 1.5 m that I built using PVC pipe and PVC elbows. The area of each quadrat represents 2.25 m2. I sampled a total of 30 quadrats. Within each quadrat I recorded the time, weather conditions, number of stems of snowberry, snowberry cover class (1-6), upland slope (%), slope aspect, soil moisture, light, pH and recorded the other vegetation observed within each quadrat. The number of stems of snowberry within each quadrat will be processed to represent the density of snowberry. The cover class of snowberry represents the percentage cover of snowberry within each quadrat. I used a clinometer for the upland slope (%), a compass to record the slope aspect and a soil moisture meter to record the soil moisture, light and pH within each quadrat.

During my field collection, I had to adjust the area I sampled at Site 2 Riparian Area due to the dense vegetation and presence of poison ivy. The riparian area was long enough that I was able to move my sampling location south and still had enough space to implement my sampling design. Before I went into the field, I printed off field collection sheets, field maps and grid coordinates to make it more efficient to locate each quadrat in the field which I found very helpful. I timed my field collection when it was not raining, and before the first snowfall. One issue I found during my field collection was that most of the snowberry had lost it’s leaves and berries, so I had to spend more time at each quadrat identifying the snowberry stems. Ideally, I would have collected my field data before the leaves had fallen to make identification more efficient.

On review of my data collection, I found that snowberry was more abundant in Site 1 Eastern Area, where I expected snowberry to be more abundant in Site 2 Riparian Area. I also found the soil moisture in Site 1 Eastern Area was lower than Site 3 Upland Area where there was no snowberry observed. This observation was not what I predicted, I predicted that dry soils would not support snowberry distribution. My hypothesis, that snowberry distribution is determined by slope gradient percentage is supported by my field collection. I predicted that snowberry will be present in areas where slope is less than 20%. My field collection supports this, however the slope gradient percentage that I recorded ranged from 3% to 19% and then 35% to 47% where snowberry was present between 3% to 19% and not present between 35% to 47%. Ideally one of my sites would have a slope gradient percentage somewhere between 19% and 35% which would support or falsify my hypothesis.

I also observed that the soil depth in Site 1 Eastern Area and Site 2 Riparian Area was deeper than the soil in Site 2 Upland Area. I observed this when I inserted the soil moisture meter within each quadrat. I had not included soil depth as a variable that may contribute to snowberry distribution, therefore I will be considering this as a variable moving forward.

I may consider adding additional sites along Cosens Bay Trail to strengthen my data set.

Blog Post #6 – Data Collection

TRU Open Learning Leave a comment

For my initial data collection, I only looked at three trees, systematically sampled on the North side of the park, as a whole, but as my observations went on, I decided to look at each tree in five different sections from top to bottom. For each of these sections, I noted down that percentage of leave that haven’t changed color yet, and then once I have all five sections, I took the average percentage of the total leaves that haven’t changed color. Each observation was initially taken three days apart, but after some revision, I decided it would be best to take observations every one or two days as the humidity level and leaf color changes more that I anticipated. This way, I could potentially get more accurate results.

I didn’t have too much trouble implementing my design, but there definitely had to be some changes like taking observations for frequently and using a hygrometer to better measure the humidity level.

So far, my observations do support my hypothesis that when humidity level decreases, than the leaf color change increases. But there are some observations that don’t follow exactly.

Blog Post 6: Data Collection

TRU Open Learning Leave a comment

During my initial observations I determined the size of my entire study area (Figure 1 and 2). The size of is approximately 28m by 30m, which I then divided into four quadrants 14m by 15m. As discussed in my previous post after trial and error it was determined individually counting the species to research diversity and population density of invasive pond weeds would be best represented in a range. By utilizing a chart, I have visited the pond a handful of times, once a week and made note of the population densities for various plant species, based on their location surrounding the pond (N,E,W,S) and the ratio of interacting factors.


Figure 1: Study Area


Figure 2: Study Area Coordinates

The following photos were taken October 20, 2019 at 5:00 pm on a sunny fall evening. The weather was approximately 10 degrees celsius and somewhat windy.

Blog Post 6: Data Collection at Whispering Woods

TRU Open Learning Leave a comment

On October 2nd between 15:47 and 16:27 I collected field data at Whispering Woods park, Calgary AB. The weather was a sunny 7°C with a light breeze. I took 10 replicates of P. tremuloides (aspen tree) from the bottom of the hill, and 10 from the top of the hill. I used a simple random sampling design to find these replicates. A random number generator was used to locate initial tree, then random number generator was used for subsequent replicates by counting the number of trees based on the random number. I walked from West to East and then back again when I reached the last tree on each side. This is a modification from my previous systematic sampling design. I believe is this new method is an improvement, as it allows me to take 10 replicates instead of 9 and uses a more random process.

This was the second day of snow melting since a large snowfall over the weekend, thus I anticipated having difficulties using my soil moisture probe. I predicted the forest floor to be too wet that the probe wouldn’t be able to record an accurate moisture level, or that the snow would skew all my data. However, it appeared that any leftover snow was sitting on the thick layer of native grasses, such that I only had to brush this snow away to isolate the untouched soil underneath. Thus, soil readings were no trouble. Other than this, I had no problems implementing my sampling design.

So far, the patterns observed have been mostly in favour of my predictions. At the bottom of the hill, the mean soil moisture has been higher at the base of the trees, the mean pH has been more neutral, and the mean percentage of yellow leaves, and leaves lost has been lower. Upon reflection, this appears to suggest that the soil conditions are more moist at the bottom of the hill, which serves as a proxy for tree health.

I look forward to collecting more data over the next few weeks to track these patterns further.

Madeleine Browne

BLOG POST 6

TRU Open Learning Leave a comment

I did eight replicates in total, all at the Bee Garden outside of the Ken Lepin building. I have had some issues with the spread-out area, with there being a fair amount of plants to keep my eye on and trying to ensure I do not miss any bees. I have noticed some slight patterns, such as there being a fair amount of bees the higher the temperature and with a lower temperature and rain there seems to be fewer bees.

Blog Post 6 Data Collection

TRU Open Learning Leave a comment

My field data collection began when I revised my plots to include 2 separate plots of wet and dry soil. I created a field data table similar to the activity in Module 3. I set up the table to include the 6 replicates I have chosen; Hydrocotyle Heteromeria, Trifolium repens, Glechoma hederacea, Bellis perennis, Poa pratensis and Elymus repens.

My design is a Logistical Regression experiment as I am determining a categorical predictor variable. The predictor variable for the hypothesis is soil moisture. I have determined areas which contain high levels of soil moisture and areas of less soil moisture using a ‘soil moisture meter.’ According to Gotelli and Ellison (2004), am hoping to determine the “effects of X on Variable Y.” My experiment will help me determine if the effects of moisture variable ‘X’ limits the abundance of Hydrocotyle plant ‘Y’.

I have not had any trouble implementing the Logistical Regression sampling design, on the systematically placed transects. My data table has a categorical predictor of ‘absence or presence” of the replicates in each quadrate in the two sample plots. The only issue that I had not accounted for was the fact that it was so time consuming. Looking at 16 quadrates in two 5x5m sections to determine each species took me hours.

 

To decrease the chances that the experimental data results may not be a representation of the actual patterns occurring, I will have a large scale area to sample, greater than 1m2 (Gotelli 2004, Englund 2003).

 

I am performing a natural experiment in which the two plots are in natural settings and have not been manipulated. I am performing a snapshot survey of the plots (in the month of September) instead of a trajectory experiment which would be done over time and years (Gotelli 2004). Snapshots work well because the replicates are more likely to be independent of one another as compared to the trajectory experiments (Gotelli 2004).

 

I will be using 6 replicates which are the 6 most common species found in my lawn experiment. Using the “Rule of 10” I have systematically set up 4 transects in a North/South and East/West direction to give me 16 study plots in each of my 2 designated “high moisture content” and “low moisture content” areas. The quadrate size is 17x17cm, which will ensure that the samples are far enough apart to be independent. Both of the plots are homogeneous in climatic conditions. I don’t not need a control group as I am not manipulating the experiment, I am surveying the natural landscape.

 

Grain: Smallest unit of study = the absence or presence of the replicate in the 17x17cm quadrate

Extent: Total area encompassed by all sampling units = 2.7m2 of sampling area

Citation

Gotelli and Ellison. 2004. A primer of Ecological statistics. Chapter 6; Designing a Successful Field Study. Web. Accessed TRU.

Englund, G. and S.D. Cooper. 2003. Scale effects and extrapolation in ecological experiments. Advances in Ecological Research 33: 161-213. Accessed TRU.

Post 6: Data Collection

TRU Open Learning Leave a comment

I did not make any changes to my stratified random sampling method from Module 3, however, I decided to increase the number of replicates from five to 10 for each stratum, as I observed variation within each stratum that could be reduced with a greater number of samples. I continued to use a distance-based sampling method with the random number generator to create two numbers, an angle and a distance in cm to determine the location of the sample unit to measure. I did not encounter any other problems since revising the data collection technique mentioned in Blog Post 5. A potential ancillary pattern I noticed was that the stratum with a high level of exposure to sunlight, which appeared to consist of the tallest plants, was also the furthest away from the creek, with the highest elevation. The moderate exposure strata was second closest, and the low exposure strata was adjacent to the creek. This prompted me to consider the effect of soil moisture and elevation as potential confounding factors affecting plant growth.

 

Edit:

I collected my data again following my revised hypothesis and study design outlined in Blog Post 5. I implemented the stratified random sampling method using quadrats to measure density of the Canada goldenrod. While this method was more difficult that simply measuring the height of individual plants, I believe it will provide more conclusive results of the effect of sunlight exposure on plant success. This sampling method was more time consuming, but included a larger sample size. I sampled 30 replicates in total, 10 from each strata. Potential ancillary patterns are similar to those observed with height measurements, since density and height appear to be correlated.

Blog Post 6 – Data Collection

TRU Open Learning Leave a comment

In total I sampled 75 replicates (25 replicates from each sampling region). I created one 60 inches by 60 inches plot in each sampling region and then proceeded to divide that plot into 25 12 inches by 12 inches quadrants. The only problem I encountered in implementing my sampling design was the same as listed in blog post 5. My sampling design was quite time consuming, but I chose to continue with my design because it was the most accurate. In my hypothesis I predicted that the less shade (and therefore more sunlight) the clovers have access to the higher the abundance/frequency would be. The patterns that I noticed did support my hypothesis. When revisiting my study location, I did notice that the grass had been cut and that the grass was cut much shorter in the larger open areas of grass than the grass around the trees (likely because of the difficulty of maneuvering a lawn mower around the trees). This did make me consider that the clovers in sampling region “shade” and “partial-shade” may experience more competition than the clovers in “no-shade” because of the longer grass. I am considering that this may be a contributing factor to the lower abundance in those areas. However because my data collection was only 7 days apart and the data was very similar I do not think it had too much of an effect.

Blog Post 6: Data Collection

TRU Open Learning Leave a comment

Field data collection activities: I got my Dad to record the values while I measured them. We located each of my replicates, measured width of branches at base to look for any limited growth patterns between the crowded and spaced replicates. We sampled buds with a 0.25m2 quadrat.

 

How many replicates: 30 in total, 10 per site

 

Problems implementing sample design: the quadrat: although I still believe was the best way to collect bud abundance, may have a moderate percent error. It was hard to count all the buds present in the quadrat because it is only 2 dimensional and there were some buds further behind those on the surface that still fell within the quadrat. I realize that the quadrat is more precise when placed flat on the ground, but this method was the only thing I could think of that would be somewhat accurate.

 

Ancillary patterns that caused hypothesis reflection: realized that all replicates are within the same soil conditions, very close proximity, so that wouldn’t be a major contributing factor to differences in growth, however, those in areas of high density might experience more competition for those soil resources.

Post 6: Data Collection

TRU Open Learning Leave a comment

I sampled 30 replicates (10 from each site). The problems I experienced implementing my sampling design were the same as in Blog Post 5, where the uneven ground of Site 2 was difficult to maneuver as I was pacing out my steps. The thick vegetation around the rocky outcrop of Site 3 was also a challenge, but ultimately did not stop me from collecting any samples. Some patterns that I have noticed include the complete absence of Western redcedar trees from Site 1, and the domination of Site 2 by the Western redcedars. In my hypotheses, I predicted that the lower elevation and higher moisture content of Site 2 would promote the growth of the Western Red Cedar, which has been supported so far. Because Site 1 is so sandy and sand doesn’t retain as much moisture as loam or silt, the complete absence of Western redcedars is not very surprising. Elevation and aspect could also play a part in their distribution, as Site 1 is at a higher elevation than Site 2 but equal to that of Site 3. Site 3 receives more sun (aspect) than Site 1 however, which may explain the species distribution patterns I have noticed.