Undergraduate Honours Thesis

The Effectiveness of Video-based Observations on Wildlife Crossings in the Bow Valley (2017)

 

I completed an honours thesis as part of my undergraduate degree at Wilfrid Laurier University. It was based on field work conducted in Alberta during a 10-day fourth-year field studies trip to the Rockies. Below I’ll provide a brief overview of the project. The full thesis is available here:

https://drive.google.com/file/d/1LomFSK1Q2mtd0aP6Tl3drCcHwmLiUSCu/view?usp=sharing

 

Background

Monitoring the usage of wildlife crossing structures is a major topic of research in regard to the mitigation of the effects of human passages on wildlife, especially in the Bow Valley and Banff National Park. Various methods have been used to observe these crossing events, many of which utilize motion-sensitive wildlife cameras. Although studies using these cameras have been conducted, few have made use of the recording of video, opting for the capture of still photographs instead. This thesis explores the effectiveness of video observations in comparison to still images.

The overall objective of this thesis was to explore the usefulness of video observations in capturing wildlife crossings in corridors. Specific objectives included the following:

• To observe and capture the usage of three wildlife corridors in the Bow Valley.
• To create a typology of all the potential variables that can be analyzed within video observations captured in the study, including the movement and behavioural patterns of wildlife.
• To compare the video observations to photographic observations in order to further emphasize the effectiveness of video.

 

Methods

In order to test the viability of video-based wildlife observations within the study area, my group and I set up motion-triggered wildlife cameras on three different wildlife crossing structures (two underpasses along the Trans-Canada Highway, and one overpass over a canal by the Grassi Lakes trail) near Canmore, Alberta.

A map of the three study sites. Larger image available here: https://drive.google.com/file/d/1kxMvgSAhbhkduWXbCc1jxUEAhJskuQZM/view?usp=sharing

Site 1: The Rundle Forebay overpass

Site 2: The Stewart Creek underpass

Site 3: The Dead Man’s Flats underpass

At each of the three study sites, the cameras were placed in a location that would allow for the greatest range of capture possible, while also inflicting minimal damage on the vegetation and natural features of the structure.

A camera is attached to a stick and placed to capture the greatest range possible at one of the study sites

Upon placing the cameras at each study site, they needed to be tested prior to official use in the field. This involved running around in front of the cameras to simulate a wildlife encounter, which was done to test the video capabilities and the range of capture at each location.

The motion-triggered wildlife camera is tested at the Stewart Creek site.

 

Over the course of three days, from August 19-22, 2016,  276 video samples were collected, with only 24 of those samples (representing 21 total crossing events) actually featuring wildlife using the corridors. 8 of these came from Site 1, 11 from Site 2 and 5 from Site 3. The reason there were so many videos even with so little actual crossing events was related to where we set the cameras up at each site. In almost all cases, vegetation located in front of the camera triggered the recording as it blew in the wind, although there were a couple instances of human interference as well. The Rundle Forebay site produced by far the highest number of error videos. The camera here was attached to a tree which blew in the wind and as visible within the video below, vegetation in the lower part of the frame also would have acted to set off the motion detector in the camera.

An extreme example of an error video at the Rundle Forebay site

 

Although there was a high number of error videos that were unusable in the type of analysis that I wanted to perform, many of the videos that did feature wildlife produced quite interesting results.

Perhaps the most interesting crossing captured throughout the study period can be seen in the video below, in which an entire family of bears was caught passing through the Dead Man’s Flats underpass.

A family of bears passes through the Dead Man’s Flats underpass

 

Of the 24 samples containing footage of serving as the main focus for analysis (representing 21 different crossing events in total), 8 came from Site 1, 11 from Site 2 and 5 from Site 3.

 

 

The Typology

These crossing videos provided the basis for the creation of the main typology. The typology contained 31 different variables that can be observed and analyzed within the videos, separated into 6 categories based on the general classifications that can be assigned to each variable.

 

Typology of variables that can be analyzed through video observation.

Larger image available here: https://drive.google.com/file/d/1jgIW2j8tQmmL5uEKuuWI_3k-0V82j9sS/view?usp=sharing

 

The 6 categories can be seen in yellow, and their associated variables in salmon. As visible, these variables were organized with the most general and conditional variables listed first and the more complex and in-depth variables listed at the end, in order to provide the best possible flow and context for all variables. A complete listing of raw data variables can be found at the following link: https://www.dropbox.com/s/ju4qek668xdv3wb/Mackenzie%20Repovs%20Thesis%20Variables.xlsx?dl=0

 

The 6 categories and their respective variables were as follows:

 

1. Conditional Factors

The first category included variables relating to the general characteristics of each video, and can help to provide situational context to an observer. 7 different recorded variables fell within this category, and were for the most part fairly straightforward:

a. Structure Type (overpass or underpass structure)

b. Date

c. Time of Day

d. Weather

e. Number of Animals

f. Species

g. Predator or Prey

 

 

2. Behavioural Traits

This category of the typography contained the behavioural traits exhibited by animals using the the crossing structures. These variables provide the observer with an understanding of the emotion and changes in behaviour of wildlife at the time of crossing.

This included:

a. Reason/Motive for Crossing

b. Focus

c. Displayed Behaviour:

The displayed behaviour of an animal using a corridor was determined by closely examining the general characteristics of movement and facial expressions of an animal, and inferring their overall mood and general behavioural patterns. The following behaviours were observed across the population of video observations: alertness (6 videos), calmness (2 videos), curiosity (3 videos), caution (1 video), fear (1 video), excitement (2 videos), wandering (5 videos), pursuit (2 videos), chasing (1 video) and walking (1 video). The most common behaviour observed was alertness, with 6 videos containing an overall displayed behaviour of this nature by the animals in observation. An example of an animal exhibiting cautious behaviour can be seen below.

An animal exhibits cautious behaviour

 

d. Facial Expressions

e. Tenseness

f. Confidence

 

 

 

3. Bodily Movements/Functions

Variables in this category were less focused on behavioural aspects of crossing than the previous category, and instead were more focused on the natural bodily functions of the animals or their physical responses to various disturbance factors. They provide a different perspective on behavioural patterns than those in category 2.

a. Number of Head Nods

b. Number of Rapid Head Movements

Rapid head movements are apparent when an animal makes a large shift in focus. Often this is accompanied by a large, stiff movement of the head and neck area in a particular direction. In the case of this study, these head movements indicated some sort of disturbance to the animals in their crossing. Animals exhibiting these movements were often alarmed by something throughout the duration of their crossing, however in many cases it is unclear exactly what provoked them.

Check out the video below and pay attention to how many times the animal exhibits a rapid head movement!

Animal exhibits multiple rapid head movements

 

c.  Number of Blinks

d. Number of Ear Twitches

e. Number of Tail Movements

 

 

 

4. Path/Crossing Related Variables

The variables in this category related to the paths and travel of wildlife within the corridor, and are the most relevant variables to the movement patterns of an animal as it uses a corridor. These include variables corresponding to distance, speed, direction, distractions and stops.

a. Distance Traveled

b. Number of Stops

c. Number of Times Distracted from Path

d. Number of Directional Changes:

If an animal changes direction while crossing a structure, that may indicate a diminished sense of comfortability while using that structure. There is a possibility that the animal may be frightened or alarmed by something in the corridor, or that they just get a generally unsafe impression from that corridor. In the video below, you can see that this elk was clearly disturbed by something which caused it to turn around and run back in the direction it came.

Elk switches direction in travel

 

e. Number of Trips/Stumbles

f. Number of Rapid Changes in Speed

g. Number of Ground Sniffs

h. Number of Glances at Traffic

 

 

5. Interaction-Dependent Variables

The following variables were specifically related to the observations featuring multiple animals crossing at once, and have to do with the interactions that may take place between animals throughout their crossing. Observations without multiple animals were not considered.

a. Number of Glances at Another Animal:

The number of times that an animal looked or glanced over at another animal within the same crossing event was recorded, and is visible in the video below, as a deer observes the behaviour of the individual in front of it.

A deer pays keeps a close eye on its companion

 

b. Number of Speed Adjustments

 

 

6. External/Technical-Based Variables

The variables in this final category of the typology were related to external disturbances to to the crossings, as well as the aspects pertaining to the quality of the video and positioning of the camera itself.

a. Awareness of Study Camera

b. Number of Outside Disturbances

c. Interferences to Video Quality

 

 

 

Discussion

The results of this study act to highlight the importance of video-based observations, specifically acting to promote their usage over photographic observations. Of the 31 variables examined within the typology, approximately 19 of those classify as video-specific. What this means is that 19 of the variables in the typology are only observable within observations that in video format; they cannot be properly examined, or examined at all, through still-image observations. The majority of these video-specific variables were based on more quantitative analyzation, including any variable based on a count of a particular behavioural or movement pattern.

An example of a comparison between the capabilities of video observations to photographic observations with regard to their effectiveness at capturing the details of wildlife crossing events can be seen below.

 

Photographic observation of a deer crossing event

Video observation of the same deer crossing event

 

If one were to look at both of these observations in order to analyze the number of ear twitches exhibited by an animal, only the video clip would suffice. The photograph, while providing a nice snapshot of the animal upon crossing and a general understanding of the situation, is simply unable to capture such a specific variable. The intricateness of the movement of the ears, at least in the case of this crossing event, is almost impossible to observe through photographic observations, unless the photos were taken continuously upon a triggering of their motion sensors.

While the video provides an actual sample of the motion that the animal was exhibiting and allows an easy interpretation of its behaviour, the photograph carries much more uncertainty. Upon analyzation of the image with the figure, it is apparent that the behaviour and movement patterns of the animal are not totally clear. In some cases, depending on the exact moment when a photo is snapped and what the animal is exhibiting in its motion at the time, photographs can provide a sufficient basis for understanding such variables as confidence, however there is no guarantee that a photo captured will be able to properly communicate that kind of understanding to a researcher. Video observations in this instance provide a much more reliable form of data that can be analyzed at a much greater level than photographic observations.

 

Conclusion

The results of this study provide a tool for those in the field of wildlife corridor monitoring to identify the many possibilities of video-recording technology in comparison to photographic-based research. Due to the fact that so little has been done to compare these two methods of observation, the typology created may be beneficial to others in the field who may be considering the usage of motion-triggered video cameras. It is hoped that what has been done through this study can serve as a stepping stone for future studies that are similar in nature.