Category: Post 5: Design Reflections

Blog Post 5: Design reflections

TRU Open Learning 1 Comment

In module 3, I collected data for my experiment to understand whether the presence of other plant species (carrots, pumpkins, onions, and peas) have an influence on the growth and abundance of bean plants near (in 30cm distance) them. For the small Assignment #1, I chose one garden plot, to collect my data.

I selected the systematic sampling strategy to record data because it would help me to avoid the experimenter bias while choosing the samples. This also seems like the best approach because it would prevent collecting samples that are clustered in the same area, but instead use the samples that are spread out around the garden bed. The first individual bean plant sample was randomly selected, and then the next samples were systematically selected. From the fifth plant north, then the fifth bean plant East. However, the difficulty with this method was: the garden plot was not large enough for the samples to be  spread out perfectly in fives. This is therefore why I plan to use the same approach, systematic sampling technique, but I will record every third plant instead of the fifth. Individual samples will still be spread out, and I will have the opportunity to record even more samples. Also, instead of collecting just the presence of other plants, I will also record what those plants are to be more detailed. This will increase accuracy, and provide more data for the analysis.

The collected data was somewhat surprising because the results were different from my predictions. I predicted that the closer the bean plant would be from these other plants, the more leaves and flowers it would have, which was not reflected in my data. I started to suspect that my hypothesis might be falsified, but I will not know until I collect more data. This opened my mind to think about possible confounding variables.

Some of the confounding variables that could play a role in the abundance of these bean plants could be the type of soil, moisture levels, type of bean plants and planting dates especially between the different garden beds. I will do more observation, and to avoid these possible confounding variables. An approach I plan to take to avoid these confounding variables is to firstly compare the bean plants within the same garden beds, before I could compare the beans in different garden plots who might not share some of these factors.

Finally, more careful observation will increase certainty of more extensive data to be collected on the next trip. In addition, a more detailed recording of data will provide more meaningful data that will lead to a more accurate conclusion.

 

 

Blog Post #5

TRU Open Learning Leave a comment

I decided to study vegetation diversity with increasing distance from the creek in my home town. My hypothesis was that proximity from the creek would effect the variety of plant life growing in the area. I predicted that, as distance from the creek increased, the variety of vegetation would also increase. I predicted the heardier plants like the Cows Parsnip and grass would survive closer to the creek because they are typically able to survive in a variety conditions. They can grow in shade or direct sunlight and in damp areas as well as drier areas. Other species, such as the wild rose, needs to have a bit of shade as well as not be in areas that are too damp or too dry.

I focussed on the portion of the walking trail starting at 15th street and ending near the public library because the whole creek would be too large of an area for me to properly sample. I stratified the area by the creek into five different sections (t1-t5) as shown on my map below. I took eleven samples from each section. I used a random number generator app on my phone to determine how far I would have to walk in each section before placing my 1m^2 quadrat and sampling.

I stuggled with collecting my data for the areas closest to the creek as some areas were very steep and difficult to walk on. I was forced to estimate where I would be placing my quadrat from the top of some of the inclines because I couldn’t actually get down the slope to place it.

I counted how many different species types I found in each section:

T1-8

T2-8

T3-12

t4-8

t5-4

I was surprised to see that the first two sections didn’t have a higher number of species than the third. I wondered if this was due to the mowing and spraying the city does near the walking trail. This also could be due to the plants near the trail being in direct sunlight.

I think that my method of sampling is working for the most part. I would like to think of a more accurate way to smaple the areas nearest the creek but have yet to come up with a solution (if anyone has one, please let me know). Also, I will need to take more samples in each section. Some of the rarer species I listed didn’t get sampled even once. I found myslef walking past some common species every time due to random luck with the number generator. Taking more samples would give me a better idea of the diversity and abundance of the species in each transect.

 

Post 5: Design Reflections

TRU Open Learning 1 Comment

I chose the haphazard sampling strategy because the shorelines of Nita lake varies significantly in vegetation abundance as a result of some particularly steep and rocky areas. In order to determine whether elevation from the waterline (flood prevalence) has a relationship with species composition, I needed to sample areas that had sufficient abundance in vegetation and a gradual enough gradient. The results in my first assignment submission are from Site 2, which is one of four sample plots i chose along the shoreline. I am aware that by subjectively selecting sample sites i run the risk of subconsciously tailoring the results to fit my hypothesis, and neglecting other factors aside from elevation that may have an impact on species composition, such as substrate. I analyzed and recorded the substrates in results, and there appeared to be some correlation between substrate type and species composition, particularly in the higher, less flood vulnerable zones.

I had predicted that Alnus rubra would be the overwhelmingly dominant species in the sub-2 meter elevation zones, as this would align with my hypothesis that Alnus rubra will be the dominant tree species of flood prone areas on Nita Lake. My results demonstrated this, with Alnus rubra composing 100% of individuals in the sub 1 meter zone and 95% in the 1-2 meter elevation zone. However, i was surprised to see that this trend in species composition continued past the flood prevalent zones, with Alnus rubra comprising 89% of individuals in the 2-3 meter zone. The variable that stood out to me in this zone was substrate, with Alnus rubra only growing in the areas with deeper soil, in contrast to the two Western red cedars growing in a thin layer of soil over large rock slabs. This made me give more consideration to the impact of substrate, as well as flood disturbance, on the distribution of Alnus rubra, and the colonizing behavior of Alnus rubra in non flood disturbed regions.

This sampling exercise was my first attempt at practicing my elevation calculation methods. I lodged an upright pole (using a level) in the mud at the waters edge, with markings from 0cm (at the waterline) up to 1 meter height on the pole. I had a string attached to the pole at the 1m elevation mark that, with a helper, i ran horizontally across to where it met the rising slope of the shore line, and attached it to the ground with a tent peg. I used a level to make the string horizontal. This gave me the 1 meter elevation mark in my sample plot. I then lodged another pole in the ground at the 1 meter mark and went through the same process to make the 2 meter elevation mark. I did this two more times to make the 3 and 4 meter elevation marks. This was a slow and tricky process to begin with, however after finishing the second mark we became much more efficient at it, and i think it provides sufficient accuracy for my purposes. I also used these horizontal string lines to determine the perimeters of my 10 x 10m plot.

I will continue to use the haphazard sampling strategy as i found it to be successful in recording these results.

Blog Post 5: Design Reflections

TRU Open Learning 1 Comment

I used systematic sampling across each of my three cross-sections to study soil moisture, using a 7inch probe, along the slope of my chosen area. At each interval I’d measure one soil moisture reading, the percent slope using a level and ruler, and the presence/absence of trees in a 2m2 radius.

There were a number of key issues I ran into throughout this process. The process of systematically sampling horizontally across the slope and taking only one reading at each stop did not always render replicates with similar slopes for comparison. In order to address this issue moving forward, I may have to modify my design to include taking more than one percent slope and soil moisture reading within each quadrant to ensure replicates and/or set pre-determined ranges for what constitutes a mild, moderate and severe percent slope for clarity purposes. The second issue I ran into was using my equipment in a consistent manner. Initially I would insert the soil moisture probe to its maximum depth, but ran into issues in later sampling when this was not possible due to soil conditions. To ensure the accuracy of my results in future sampling, I need to improve the consistency from which depth I take my soil moisture readings. I can accomplish this by either taking moisture readings at various depths at each point or marking an insertion limit on the probe itself at some point <7inches to ensure that it’ll be consistently inserted to the same depth. Other general modifications to future sampling will include expanding the size of my quadrants to improve data collection regarding tree sampling, and to gather more detailed information regarding the observed trees in order to better understand the potential impact of soil moisture, as it pertains to slope, on their growth. I will continue to implement systematic sampling; however, it will obviously need to be adapted to account for the larger quadrant sizes. I will also look into whether there is a way to accurately, and more efficiently measure percent slope since using a level and ruler was a tedious, and time-consuming process that will only become more challenging with more sample points.

The results were surprising to me in that they were not aligned with my prediction. I predicted soil moisture would be highest at the bottom of the slope, and what I found during this preliminary research was that it was highest at the midpoint. I have some ideas about why this might be, but evidently, I will have to wait until I’ve collected more data to comment on the findings with any reliability.

Blog Post 5 – Design Reflections

TRU Open Learning Leave a comment

The initial collection of my data began with deciding how large I wanted to set my transect and the degree of information that I want to convey. I used a random sampling strategy so picking an area to sample was not difficult. At first, I wanted to make five-meter transects running perpendicularly into an ecotone, and then split the five-meter section into 1m2 quadrats. Upon walking the transect I quickly found out that this would be too small, and adjusted by stretching my transect into a fifteen-meter transect.

My sampling strategy is to use cover class and average height within a m2 transect to help express the density and health of Himalayan blackberry within an ecotone. At first, I had trouble with being too methodical in taking average heights of plants. I would record four individual heights within a quadrat and then average the four numbers. I quickly learned that this was going to take too much time. I decided to note the three highest sprouts or patches, and then take an average height of these. For cover class I used six different categories ranked one through six (0-5%;5-25%; 25-50%; 50-75%; 75-95%; 95-100%). Moreover, when recording my data would also try to identify each plant’s species within the differing ecotone zone, but upon reviewing my notes I thought that this may have been outside of the scope of my research.

I also made a physical transect line via flaggers tape with one-meter quadrats sectioned off. I was able to string this along two pathways in my backyard, but have a feeling this will be hard to replicate in the field with growing blackberry vines. For this reason, I decided to replicate 1 m via a large stride, and then each stride I walked would be equal to one transect.

Blog Post 5 – Design Reflections

TRU Open Learning Leave a comment

I did have some difficulty with implementing my sampling strategy in the field. I had previously selected several locations on a map of the area found from google maps that when I was onsite, I would measure and 2×2 foot square and assess the area for mosses as well as record the temperature. On site, I used my cell phone with the google maps app to find the locations I had previously selected. Even though I had visited the site before, I did not have extensive knowledge of the terrain and soon realized that some of the locations I had selected were on private property or very difficult to get to in order to survey them. For the data I did collect, I took pictures of the specimen with my cell phone – that automatically geotagged them – and used iNaturalist to identify them before completely filling in my field data sheet. I was also overwhelmed by the number of species present that I was not comfortable identifying, so I think that I should choose a few types to collect data on instead of trying to document everything I see. I definitely need to modify my approach since I did not get nearly enough data points to make any inferences due to the poorly selected locations on inaccessible terrain, and my lack of confidence identifying species. This data was collected a few months ago now, and due to the recent health environment I have left the city I had started the study in, and have decided that I will be continuing and using data collected by observers on iNaturalist. This way, I will still be able to collect species and location data without being present in Victoria, B.C.I will also be able to use exact coordinates of the locations I surveyed which may be helpful when it comes to displaying the data later on in the process of this project. 

Post 5: Design Reflections

TRU Open Learning Leave a comment

I had a lot of trouble implementing my initial sampling strategy. Originally, I intended to systematically observe forb density and richness (0.5m by 0.5m quadrats) along transects extending down the slope from the uplands, through the riparian zone, and ending at the South Saskatchewan River. However, the varying steepness of the study site would not allow me to descend straight down the hill in many locations. This forced me to navigate to the subsequent plots from various angles and paths (in the interest of preserving the integrity of the transects). Knowing the amount of time that this would take when scaling my replications up to statistically valid levels, I opted to change my sampling method to a haphazard one while out in the field. I would select an appropriate location (which was, obviously, subjective) and lay the quadrat down before examining the forbs too closely at that location (in an attempt to mitigate some of my bias).

Something that I found surprising in my data was how a close examination of the forbs at each quadrat revealed how low in abundance they could be. I found myself, often times, looking at shrubs and saplings (of which I am excluding from my study). When this occurred: I would choose a location to sample based on it containing high abundance of broad-leaf foliage, begin examining the species, learn that they were mostly shrubs (such as Alemanchier alnifolia, or Rosa acicularis), then have to move on to the next quadrat without having any data related to my study of forb density. Having this preliminary data is useful because it does indicate that, when moving forward with my formal data collection for the study, I will need to ensure I have a high level of replications in order to capture the forb diversity in the area.

I do not intend on continuing the sampling strategies I implemented for module 3. I am planning on moving towards a simple random approach to laying down quadrats throughout the region. While transects are a good approach to this site (in theory), I believe that they are too difficult to implement in the study area. In addition, I would like to ensure that I am controlling my own bias and ensuring that the statistical analyses I would like to use are not compromised. Therefore, I, having had some more time to think about it, will be randomizing the coordinates for my replication locations. I acknowledge that randomizing individual quadrats will have the same navigational challenges as transects. However, I also believe that generating coordinates to unnavigable quadrats, needing to discard those points, and generate new coordinates is more favorable than breaking the integrity of systematic placements of quadrats in a transect (or severely restricting the locations that I can chose to generate unbroken transects).

Blog Post-5

TRU Open Learning 1 Comment

At the very beginning it was difficult for me to implement randomization in collecting data, however other than that there was no difficulty in collecting the data as I was using random sampling strategy. The data collected wasn’t surprising as it seems to be supporting the hypothesis. In order to modify the approach, I kept on collecting more data to get better results to support the hypothesis and make the conclusion stronger. The only difficulty for me to collect the data was the weather as birds don’t get out of their place in every weather and the area where I am living the weather changes really quick.

 

Blog Post 5: Design Reflections

TRU Open Learning 1 Comment

For module three I needed to collect plant biodiversity data up a slope which began at the creek that ran through my site. The goal was to see how plant diversity changed as you moved away from the water source. My approach to doing this was systematic and involved taking a transect line surveying 1×1 metre quadrants in succession up the slope. I left a meter spacing between quadrant surveyed. My method was pretty straight forward and I didn’t really have any difficulty since I could choose exactly where I wanted the transect to be within the previously observed site. I do foresee an issue with access however if I take the same approach with the larger project, which will likely involve setting the transect at a particular distance from the last, if the chosen distance ends up being somewhere which takes me into thick brush. Other than that possibility, the strategy seemed to work well to gather the data I wanted, which coincided with my previous observations and prediction. The man made gravel path at the top of the slope however seems to have it’s own affect on biodiversity and will need to be controlled for in my final project. One way I’m thinking of doing this is by also surveying sites on the opposite side of the stream as well where a gravel path does not exist, and looking for patterns in the differences. This way I can determine how much, if any, effect the path is having on my predictions.

Blog Post 5

TRU Open Learning Leave a comment

My initial experimental design and strategy took some significant re-thinking. After initial data gathering session in my study area, I realized I was unable to gather accurate enough data that I could then compile in a useful way. I therefore decided to switch from surveying for mule deer prints using transects, to using a plot system.

The problem with the transect system I originally planned to use was that I would have been unable to accurately monitor any individual trail created by mule deer through the snow. That is to say, keeping track of what I defined primary trails and secondary trails in my report would have been very difficult. Also, I believe transects would have provided better randomization, but made it difficult to accurately count individual deer tracks.

So, I switched from transects to five plots, four  of which had a anthropogenic snowshoe trail that ran the length or width of the plot. I surveyed for mule deer presence in the plot by counting individual tracks. I counted tracks on the established anthropogenic trail, on well used, previously created deer trails (primary trails), and on newly created trails (secondary trails).