Category: Post 6: Data Collection

I have made several changes to my data collection strategy since last time. While I am still using a haphazard sampling method with the help of Google maps, and point counts for estimating Black-billed magpie abundance in relation to human presence in Edmonton’s Hermitage park, I have moved to a dependent double-observer approach. Also, I have reduced the time frame spent at each point count to 20 minutes. There was an additional one-minute settling time given at each point, and five minutes were given to walk from one point to another. Additionally, I have chosen to assess human presence solely by recording the number of pedestrians observed during given time frame. Doing so resulted, I believe, in less bias than recording various predefined “human traces”.  Moreover, this time, I chose to record separately: 1. No. of birds seen within the first 5 minutes, 2. No. of birds seen flying over. Doing so will give the reader an idea of whether the presence of observers might have attracted more birds after the first five minutes. Birds seen flying over are generally noted separately in current similar literature, because they cannot be recorded in standard density calculations (Gregory et al. 2004).

To increase the accuracy of my results, I have increased the number of point counts from five to nine which I sampled on two different days (March 24th and March 28th 2018).

As I will explain further in my final report, my literature review as well as previous comments on my past assignments has led me to make those changes. My intention was to minimize bias, standardize my sampling as much as possible, and increase the use of randomization/replication in my sampling strategy. Using a dependent double-observer approach made it much easier to implement my sampling design. Not only did it reduce observer bias, less coordination was needed than it was using different observers for each point count. The only problem encountered was that most of the time, the walking time took a bit less than five minutes, so the real “settling time” was more than a full minute at most points. It was hard to keep track of the exact time spent at each location for that reason.

Gregory, R. D., D. W. Gibbons, and P. F. Donalds. 2004. Bird census and survey techniques. Pages 17–55 in R. E. Green, W. J. Sutherland, and I. Newton. Bird ecology and conservation: a handbook of techniques. Oxford University Press, New York, New York, USA.

I have made several revisions to my sampling in the field. While the overall strategy will stay the same, haphazard with subjective location selection, I will be decreasing the size of my quadrats from 100cm x 100cm to 30cm x 30cm.This reduction will allow for quicker observation of the damaged area and give an overall clearer picture. In addition to the reduction of my quadrats I will now be looking at the damage to the Kentucky Bluegrass species instead of the Ryegrass species. With the turn from winter to spring, Kentucky Bluegrass has shown itself to be the dominant species in the grassy area. Also, I will now be taking 10 replicate samples, 5 grazing and 5 non-grazing, in order to satisfy the rule of ten. In satisfying the rule of ten I hope to achieve a more accurate representation of the ecological processes that are occurring. There were no major difficulties in implementing the original sampling design, but I believe that these changes will allow for a more streamline process. There is one ancillary pattern that may have an effect on the original hypothesis. This pattern is the increasing presence of the number of geese in the grazing areas with increasingly warm weather. It appears that on warm and sunny days the number of geese in the grazing area of the park will increase by anywhere from 2 to 8 geese. While this does not change the fact that the geese are damaging the Kentucky Bluegrass in the area, it is worth observing the damage done on days where geese numbers are higher versus the damage done on days where geese numbers are at 2 or 3.

From the three tree categories (sheltered, partially sheltered, and exposed) 10 trees were randomly selected and examined for the presence of living moss. Data collection was on March 14 and started at 4:00 and finished at 5:30 when all 30 locations were examined, the weather was cloudy and slightly rainy that day. Currently the living moss present on the three different tree categories shows no significant different. Right now the presence of living moss seems to be fairly constant across the different categories. This goes against my hypothesis that there will be an increase of living moss on more sheltered trees. More research is needed to further determine the relation of living moss on different levels of exposed trees.

Data Collection

I collected data on the species of moss present and their abundance on varying slope positions on rock outcrops. For the purposes of data collection, the slope positions of the rock outcrops were divided into discrete categories for data collection and included the west slope, crest, east slope, and depression between outcrops. The faces of the slopes were distinguished from the crest and the depressions by an increase in slope to greater than 20%.

A total of 4 rock outcrops were sampled, providing 4 replicate study plots. Note that these were the only rock outcrops in the area that included all slope positions; others nearby were embedded in the slope and consisted only of a west face and crest. These are expected to be independent of each other, but it can be noted that they will not be independent of other external factors affecting all of the replicates, including overstorey and understorey cover and proximity to forest edge. The percent cover and species of overstorey and understorey vegetation were recorded for each replicate, as well as the distance to the cut forest edge.

A randomized strategic sampling strategy was used to select the sample locations along the environmental gradient of the slope positions on the rock outcrops. The number of samples collected for each slope position on each rock outcrop was determined based on the relative size of the given slope. At least 2 samples were collected per slope position, while those greater than 5 m2 in area had 3 samples and those greater than 10 m2 in area had 4 samples. The locations of the sample within each slope position was randomly assigned in order to capture the randomness of the moss cover.

The sample unit is an 18 x 18 cm quadrat because this is the size which was readily available and was a reasonable size of similar scale to plots used in the literature reviewed. The quadrat was placed on the ground parallel to the slope. All species of moss occurring within each quadrat sampled were identified and the cover determined for each. The nomenclature used was from the field guide Plants of Coastal British Columbia by Pojar and MacKinnon (1994). Cover was estimated using a cover classification scale based on the percent cover of each species, as used in the Daubenmire method (Daubenmire 1959).

Reflections
I did not have any problems implementing my design, but I had hoped that I would have found more suitable rock outcrops in order to have more replicate plots. Patterns that I have noticed right away are that the most dominant species occur in all slope positions (PYLSP, PLESCH, DICSCO), while some of the less common species of moss occur only in one slop position. These include RACCAN, which only occurs on the crest, HYPSUB and HETPRO, which only occur on the eastern slopes, and KINORE and RHYLOR, which only occur in the depressions. I also noticed that the abundance of PYLSPL increases across the environmental gradient (from western slope to depression). This difference in the species of moss occurring in the different slope positions is what I had predicted and is encouraging as it indicates that a significant trend may occur.

Previously I had stated my hypotheses to be:
1. There is a significant difference in the percent cover of moss species between plots.
2. There is a correlation between any significant differences in percent cover and slope position.
My prediction was that there will be a significant difference in the mean percent cover of mosses of different species between plots in different slope positions.

Based on feedback and the results of field sampling, I would like to change these to be:
1. Different species of moss will grow on different slope positions on rock outcrops.
2. The density of cover of moss species growing on rock outcrops will be different on different slope positions.
My new prediction is that there will be differences in the mean percent cover of several different species of mosses between sample locations on different slope positions on rock outcrops.

My field data collection activities consisted of sampling three sites for the presence of seven different bird species over five consecutive days. Point counts were taken at the same time (10:30am) for each of the consecutive sampling days. Species abundance was determined from the resulting data.

Overall, my sampling strategy was quite easy to implement. I was able to borrow a set of binoculars from the Thompson Rivers University science faculty. This aided in my point counts as I was better suited to see and identify the birds present. The most challenging thing for me was identifying the bird species. For the most part I was able to identify the species present, however, there were other times when I had to use a species identification sheet. This could lead to inaccurate data if incorrect identifications occurred.

Through the sample collections, I was able to notice that there were trends in which bird species were present at the various sites. It became obvious early on in the study that certain bird species favoured one of the three sites. Each site had a different dominant bird species. This demonstrates how differing landscapes alters bird species abundance. This follows my prediction in that the different sites will have different bird species that are associated.