Instantaneous Sampling: An Approach to Estimating Population Size Using Imagery of Unmarked Animals
B.J. Adams, Kanalu Sproat, Randy T. Larsen, Brock R. McMillan
Introduction
- Accurate estimations of population size can help inform wildlife management decisions.
- Traditional approaches to estimating population size can be expensive, risky, and invasive.
- Remote cameras have become a popular method of estimating population size via methods such as Capture-Recapture (CR) which involves placement of anthropogenic marks such as ear tags to re-identify individuals.
- Relying on resightability of marked animals can potentially introduce bias in estimations (ie. animals improperly postured to clearly display marks).
- Our objective is to estimate population size with a developing methodology known as Instantaneous Sampling (IS) and compare it with estimates from a traditional method (CR).
Introduction
- Accurate estimations of population size can help inform wildlife management decisions.
- Traditional approaches to estimating population size can be expensive, risky, and invasive.
- Remote cameras have become a popular method of estimating population size via methods such as Capture-Recapture (CR) which involves placement of anthropogenic marks such as ear tags to re-identify individuals.
- Relying on resightability of marked animals can potentially introduce bias in estimations (ie. animals improperly postured to clearly display marks).
- Our objective is to estimate population size with a developing methodology known as Instantaneous Sampling (IS) and compare it with estimates from a traditional method (CR).
Statistical Concepts
- Instanateous Sampling calculates population density (D̄) within the cameras' viewable area (aij) by taking the mean count (nij) of individuals at location i = 1, 2,... M and occasion j = 1, 2,... J.
- Viewable area (aij) is calculated as a sector of a circle determined by the camera's lens angle (ϴij) and view distance (rij):
Statistical Concepts
- Instanateous Sampling calculates population density (D̄) within the cameras' viewable area (aij) by taking the mean count (nij) of individuals at location i = 1, 2,... M and occasion j = 1, 2,... J.
- Viewable area (aij) is calculated as a sector of a circle determined by the camera's lens angle (ϴij) and view distance (rij):
Methods
- We marked 120 feral sheep (O. gmelini musimon and O. aries) during September 2016 - October 2017 to enable resighting for CR.
- When sheep with ear tags are found in camera imagery, they are considered "recaptured."
Methods
- We marked 120 feral sheep (O. gmelini musimon and O. aries) during September 2016 - October 2017 to enable resighting for CR.
- When sheep with ear tags are found in camera imagery, they are considered "recaptured."
Methods
- We deployed 95 cameras at random locations to detect sheep during June - October 2017.
- Cameras were set to trigger with motion and infrared detection.
- Cameras were also set to trigger every 15 minutes. Images taken in this manner represent "instantaneous" samples of population density.
- Images were analyzed for sheep detections and estimates were derived from data consisting of continuous 4-week periods.
Methods
- We deployed 95 cameras at random locations to detect sheep during June - October 2017.
- Cameras were set to trigger with motion and infrared detection.
- Cameras were also set to trigger every 15 minutes. Images taken in this manner represent "instantaneous" samples of population density.
- Images were analyzed for sheep detections and estimates were derived from data consisting of continuous 4-week periods.
Results
- Using datasets derived from periods of 4 continuous weeks, we generated estimates of population size through IS.
- Estimations fell within the range of the local biologist's best guess between 1600 and 2100 individuals.
- We are currently re-analyzing all detections for instances of photographic recapture in order to implement CR for comparison.
Results
- Using datasets derived from periods of 4 continuous weeks, we generated estimates of population size through IS.
- Estimations fell within the range of the local biologist's best guess between 1600 and 2100 individuals.
- We are currently re-analyzing all detections for instances of photographic recapture in order to implement CR for comparison.
Discussion
- Provided that CR and IS return similar results, IS has potential as a cost-effective and accurate alternative to estimating population size.
- Since IS does not rely on uniquely identifiable individuals to generate estimations of population size, the requirement for captures is eliminated.
- Monitoring techniques involving remote cameras are highly flexible and repeatable.
Discussion
- Provided that CR and IS return similar results, IS has potential as a cost-effective and accurate alternative to estimating population size.
- Since IS does not rely on uniquely identifiable individuals to generate estimations of population size, the requirement for captures is eliminated.
- Monitoring techniques involving remote cameras are highly flexible and repeatable.
