Category: Nancy Elliot

Blog Post 6: Data Collection

TRU Open Learning Leave a comment

Hi Class & Professor Elliot,

This past long weekend I finished my field data collection at Coldham Regional Park. I set out early in the morning to beat the summer heat and it took me around 3.5 hours to finish collecting samples.

Coldham Regional Park

Over the past two months I have been regularly thinking about my study design and hypothesis, which I have had to revise several times. From the start, my sample design seemed sound and I did not run into any problems implementing the sampling design on the ground (other than trying to cross the creek!). I took 40 replicate samples over an area of 60m by 50m, divided into four transects, to accurately capture the riparian area either side of Jack Creek.

View of Study Area – Looking East

My prediction was that there would be a greater abundance of large woody vegetation on the eastern side of the creek that has a westerly facing slope. It was interesting during field data collection, I noticed more of a pattern that in the flatter areas on the west side of the creek there was indeed greater abundance of large woody vegetation, however, as the hillside became steeper there was typically less vegetation overall.

View of Jack Creek – Facing South

Another interesting pattern that I wasn’t expecting was the number of spruce trees. I had originally thought there would be greater diversity in the species of trees (e.g. pine, spruce, maple). But during data collection, spruce trees were clearly the most abundant species. Another note was that there were several standing dead spruce trees on the western side of the creek (that gets more sunlight), and healthy, taller spruce trees on the eastern side of the creek.

Spruce Stand – Eastern Side of Jack Creek

Reflecting on my hypothesis, I am interested to see what sort of correlations I can find and other explanations during theoretical research.

Thanks for reading.

-Brittany Lange

Blog Post 4: Sampling Strategies

TRU Open Learning Leave a comment

For this assignment, I was required to use the systematic sampling method, the random sampling method and also the haphazard sampling method in the virtual forest tutorial. The sampling technique with the fastest estimate sampling time ended up being the systematic sampling method with a time of 12 hours and 34 minutes. The next fastest sampling time was the random sampling method at 12 hours and 41 minutes, and the slowest sampling technique ended up being the haphazard method with a time of 12 hours and 49 minutes.

The two most common species were: Red Maple and Witch Hazel

The percentage error of the two most common species:

Systematic:

Red Maple= 9.8% error

Witch Hazel= 12.2% error

Random:

Red Maple= 22.4% error

Witch Hazel= 48.2% error

Haphazard:

Red Maple= 13.1% error

Witch Hazel= 33.3% error

 

The two rarest species were: Sweet Birch and Hawthorn

The percentage error of the two rarest species:

Systematic:

Sweet Birch = 82.7% error

Hawthorn = 14.9% error

Random:

Sweet Birch = 24.4% error

Hawthorn = 62.4% error

Haphazard:

Sweet Birch = 55.7% error

Hawthorn = 71.1% error

 

From the results, on average the trend seems to be that the accuracy increases when there is a greater species abundance, in other words the greater the species abundance the less of a percentage error there was.

Systematic sampling method percentage error (more accurate on average): 29.9%

Random sampling method percentage error: 39.35%

Haphazard method percentage error: 43.3%

So, as you can see the systematic sampling method has the lowest percentage error and therefore, is on average the most accurate.

 

Blog Post 5: Design Reflections

TRU Open Learning 1 Comment

From observations gathered from the Alfred Howe Greenway, Port Moody, BC, in Blog Post 1 and Blog Post 3, when walking along from the south end of the trail (elevation: 118 m) to the north end of the trail (elevation: 50m), there appeared to be a change in pine tree density.

A stratified random sampling strategy was used to measure pine tree density along three points of an elevation gradient using the nearest individual method to select each sampling unit. Along each elevation category (A. 120-110 m, B. 90-80 m, C. 60-50 m) five sampling units were selected by generating a random compass bearing and number of paces using an app. From the randomly generated location, the distance from the nearest pine tree to its neighbour was measured.

Although using the nearest individual method to select the sampling unit was more efficient than formulating a coordinate grid overlaying a map of the area in order to select a specific sample quadrant, a few difficulties were encountered in implementing and interpreting data collected using the nearest individual method, stratified random sampling strategy. When interpreting the results of this particular sampling strategy, the pine tree density of a particular elevation category is measured as the average distance from one pine tree to the next, which would (1) record the upper limit of the pine tree density for that particular area, as there would be a disregard for spaces where there is a significant lack of pine trees. This would particularly effect data in areas where it was observed that pine trees were found in “patches” rather than having a more uniform distribution. Furthermore, when implementing the sampling strategy, due to the steep elevation gradient, (2) difficulties were encountered to generate a random bearing that would generate a sampling location within the desired elevation category. If the same technique is used for a future data collection, perhaps the range for the number of paces should be decreased.

The data collected (although having a sample size of less than 10 measurements for each elevation category) represented a linear decrease in the average pine tree density along the elevation gradient. Although this result supports an initial hypothesis of pine tree density decreasing along the elevation gradient, the difficulties encountered (mentioned above) in interpreting the data brings the accuracy of the result to question (ex. perhaps there might be an exponential decline).

It would be favourable to continue to build future research of the Alfred Howe Greenway around pine tree data collection, as they seem to be found along the entire trail, in contrast to other plant species that only appear at one point of the trail or are seasonal. Perhaps a modified approach of a point-centered quarter method will be used for measuring pine tree species density in order to attain more accurate pine tree density measurements. By using the point-centered method, the distance of each pine tree (with the average interpreted as the pine tree density for a particular elevation category) would be recorded as a measurement of the nearest individual from each quarter to the centre point of the quadrant. As a result, recordings of solely the upper density limits will be avoided.

For the second portion of this blog post I commented on kmcara’s Blog Post 3: Ongoing Observations .

EDIT: From researching for archives about the Alfred Howe Greenway on the Port Moody, BC, Government website, it was found that the north point of the trail encompasses an area formerly used as a landfill site from the 1950’s to 1982, in addition to being used for green waste up until 2002. (City of Port Moody, 2018; Payne, 2015)

This offers a clear explanation for the significantly contrasting abundance of pine trees (and overall species diversity) between the north point (formerly a landfill) and south point (historically forested area) of the trail.

As a result of this newly found information I will perhaps shift my area of study to observe any current impact the former landfill has on the surrounding ecosystem.

Blog Post 4: Sampling Strategies

TRU Open Learning Leave a comment

Three sampling strategies, inclusive of systematic, random, and haphazard, were used in the virtual sampling tutorial. The technique with the fastest estimated sampling time was systematic sampling (12 hours and 31 minutes).

Eastern hemlock and sweet birch were the two most common species in this tutorial. Systematic sampling yielded the lowest percent error for eastern hemlock (1.98%), however, percent error for sweet birch was lowest when the haphazard sampling technique was applied (17%).

Eastern Hemlock: Systematic = 1.98% error, Random = 32.3% error, Haphazard = 10.8% error

Sweet Birch: Systematic = 36.17% error, Random = 46.21% error, Haphazard = 17% error

Overall, when comparing percent error results for the two most common species, haphazard sampling had the lowest overall average percent error (13.9%), compared to systematic sampling (19.1%) and random sampling (39.3%).

The random sampling strategy proved to be the most accurate technique for the two rarest species: striped maple and white pine. Percent error for striped maple using the random sampling technique was still quite high, despite having the lowest error of all the techniques applied.

Striped maple: Systematic = 76% error, Random = 50.9% error, Haphazard = 114% error

White pine: Systematic = 100% error, Random = 2.4% error, Haphazard = 100% error

Random sampling, on average, was the most accurate technique when used to sample the rarest species (26.7% error), as compared to systematic sampling (88% error) and haphazard sampling (107% error).

Overall, greater species abundance led to greater accuracy.

Post 5: Design Reflections

TRU Open Learning 1 Comment

For my study, I hypothesize that Sagebrush (Artemisia tridentate) will be more abundant on hill tops versus valleys. I did not have any difficulties with my sampling strategy. The data that I collected was surprising to me; the south side of the hill tops had little to no mature sagebrush, only juveniles that were less that 20 cm tall. I was expecting a more even distribution with more mature sagebrush on the south side. I did not yet gather data for the valleys. For next time, I will continue to use the same sampling strategy but I will change my approach slightly.  In the future, I will use rope or something similar in order to show the exact edges of the quadrant. In order to avoid bias, I will use a predetermined amount of space between each quadrant and measure accurately. I will also obtain soil samples in order to determine the soil moisture content.

Blog Post 5: Design Reflections

TRU Open Learning 1 Comment

Hello Class & Professor Elliot,

During the collection of initial field data in Module 3, I found the most difficult part was trying to design a sampling unit that would accurately represent the area I was trying to study. Once I had devised a plan to span an environmental gradient on both sides of Jack Creek, I found it was relatively easy to put together a sampling method. The difficulties in implementing the sample unit happened more on the ground when specific points I wanted to measure either had nothing to sample or weren’t easy to access on foot. The data I collected was surprising (in the context). I tested soil moisture to see what would happen and it was uniform, even as I got closer to the creek. I only took measurements on one side of the creek for my initial data so I am interested to see the differences on the eastern side. I plan to use the same technique for my larger data collection, however, I will need to modify what I am sampling as discussed below.

Previously, I was counting all vegetation present and I did not take diameter breast height (DBH) measurements for large woody vegetation. To date, I have only used desktop review to analyze the gradient or metres above sea-level (MASL). For my larger field data collection, I will use the compass and elevation reader on my smartphone to collect this data at each replicate point. I think calculating DBH and aspect will be most important to my study. By understanding the amount of aspect this will potentially show any underlying processes in microclimates.

For the second part of this blog post, I have decided to comment on M. Myles recent post on their recent field observations, as our study designs are similar.

Blog Post 3: Ongoing Field Observations

TRU Open Learning 1 Comment

The organism(s) I plan to study for my field research project include waterfowl and their allied species. During my subsequent field visits, I have observed different species of waterfowl (e.g., wood duck, mallard) utilizing the smaller waterways of the park where emergent vegetation is present.  Underlying processes that may cause this pattern include the use of the emergent vegetation by waterfowl for foraging purposes and for protection from predators.

Although there are multiple environmental gradients within the park, I have only observed waterfowl within specific aquatic habitats, inclusive of the drainage ditches and one pond/marsh area. This could be due to habitat preferences of individual species, life stage, foraging potential and presence of predators. Waterfowl may occur within the old field habitat adjacent to the drainage ditches and marsh area, however, abundance and height of grasses within the park at this time of year greatly reduce visibility. As such, I will only be using the visible waterways as potential study areas.

Based on these observations, my initial hypothesis is that waterfowl prefer to use aquatic habitats where emergent vegetation cover is present.  I predict that relative abundance of waterfowl will increase where emergent vegetation is present and decrease in areas where emergent vegetation is absent. A potential explanatory variable is percent cover of emergent vegetation (continuous). A potential response variable is waterfowl abundance (categorical).

Field Notes Blog Post 3

Blog Post 4: Sampling Strategies

TRU Open Learning 1 Comment

Three sampling methods were used in gathering data from the Snyder-Middleswarth Natural Area in the virtual forest tutorial: systematic sampling, random sampling, and haphazard sampling.

The systematic sampling method had the fastest estimated sampling time of 12 hours and 34 minutes. In contrast, the random sampling method had an estimated sampling time of 12 hours and 44 minutes. The haphazard sampling method had the longest estimated sampling time of 12 hours and 59 minutes.

The percentage error of the two most common and two rarest species for each sampling method are summarised below:

As observed from Table 1.4, on average, when the percentage errors of the two most common with the two rarest species in each sampling method are compared, it can be observed that accuracy increases with greater species abundance.

Furthermore, the systematic sampling method was more accurate on average (21.1%) than the random sampling method (30.78%) or the haphazard method (37.1%).

This result was surprising, having previously assumed that a method employing more randomisation in sample selection would have a smaller percentage error in its data collection. Perhaps because the species density distribution did not vary considerably along the y-coordinate plane, the samples collected using the systematic sampling method were (on average) more representative of actual data than the random sampling method (which had selected areas at random to sample, regardless of the evident topographic gradient).

It would be interesting to observe whether the relative accuracies of the sampling strategies would change if a much larger sample size was tested. Additionally, if would be interesting to observe whether a stratified random sampling method would have greater accuracy than the systematic sampling method for the Snyder-Middleswarth Natural Area.

 

Blog Post 2: Sources of Scientific Information

TRU Open Learning Leave a comment

The source of ecological information I have chosen for this blog entry is a journal article titled:

“Assessing the effect of emergent vegetation in a surface-flow constructed wetland on eutrophication reversion and biodiversity enhancement”

This journal article is an example of academic, peer-reviewed research material for the following reasons:

Academic Source:

  • Written by experts in the field

  • Includes in-text citations

  • Contains a bibliography

Peer-reviewed Material:

  • Includes a revision submission date and acceptance date

  • Credit is given to anonymous reviewers in the Acknowledgements section of the article.

Research Article:

  • Methods and Results sections within the article indicate that original research was conducted.

Blog post 9

TRU Open Learning Leave a comment

The design of my field research was relatively simple, which led to some paranoia about missing a key concept because of the simplicity.  However, given the intended use of the data and the relatively small sample site, I am confident that the design was appropriate for the question.

One feature of the field research that turned out to be unnecessary was the detailed locational data that I collected.  For each sample collected, I noted the exact location where it was found along the transect and in the plot.  This information proved to be of little use later on.  Collecting this data was probably the most time consuming aspect of the data collection, so designing a similar study in the future that eliminates that component would be ideal.

I was also interested to see how much human error or bias can be introduced into study designs and I have read other studies with a critical eye to this point.  It is difficult to design a study which does not introduce some kind of bias, whether it is related to financial constraints, areas of interest, knowledge, or any number of other potential biases and errors that are lurking.  This illustrates the importance of having many studies of a topic showing an observed effect before we can say with much confidence that there is a genuine observation being made that reflects the hypothesis.  I can now appreciate just how difficult it is to design a study in a way that reduces error and bias.  And even when that is done well, it is still important to note that there are some biases and errors which cannot be completely eliminated in such complex systems with the various constraints to doing research with limited time and funding.