Estimating animal density in three dimensions using capture‐frequency data from remote detectors
Abstract Remote detectors are being used increasingly often to study aquatic and aerial species, which move significantly differently from terrestrial species. Camera‐trapping studies have used frameworks based on animal movement patterns to show that a species’ detection frequency, along with movem...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2021-03-01
|
| Series: | Remote Sensing in Ecology and Conservation |
| Subjects: | |
| Online Access: | https://doi.org/10.1002/rse2.159 |
| _version_ | 1818615356624732160 |
|---|---|
| author | Juan S. Vargas Soto Rowshyra A. Castañeda Nicholas E. Mandrak Péter K. Molnár |
| author_facet | Juan S. Vargas Soto Rowshyra A. Castañeda Nicholas E. Mandrak Péter K. Molnár |
| author_sort | Juan S. Vargas Soto |
| collection | DOAJ |
| description | Abstract Remote detectors are being used increasingly often to study aquatic and aerial species, which move significantly differently from terrestrial species. Camera‐trapping studies have used frameworks based on animal movement patterns to show that a species’ detection frequency, along with movement speed and detector specifications, can be used to estimate absolute population density. This approach, however, has not yet been adapted to cases where movement is three‐dimensional. Here we adapt one such framework to three‐dimensional movement, to characterize the relationship between population density, animal speed, characteristics of a remote sensor's detection zone and detection frequency. The derivation involves defining the detection zone mathematically and calculating the mean area of the profile it presents to approaching individuals. We developed two variants of the model—one assuming random movement of all individuals, and one allowing for different probabilities for each approach direction (e.g. that animals more often swim/fly horizontally than vertically). We used computer simulations to evaluate model performance for a wide range of animal and detector densities. Simulations show that in ideal conditions the method approximates true density well, and that estimates become increasingly accurate using more detectors, or sampling for longer. Moreover the method is robust to violations of assumptions, accuracy is decreased only in extreme cases where all detectors are facing the same way. We provide equations for estimating population density from detection frequency and outline how to estimate the necessary parameters. We discuss how environmental variables and species‐specific characteristics affect parameter estimates and how to account for these differences in density estimations. Our method can be applied to common remote detection methods (cameras and acoustic detectors), which are currently being used to study a diversity of species and environments. Therefore, our work may significantly expand the number and diversity of species for which density can be estimated. |
| first_indexed | 2024-12-16T16:32:37Z |
| format | Article |
| id | doaj.art-71576394af62463bb298c5201b2571a0 |
| institution | Directory Open Access Journal |
| issn | 2056-3485 |
| language | English |
| last_indexed | 2024-12-16T16:32:37Z |
| publishDate | 2021-03-01 |
| publisher | Wiley |
| record_format | Article |
| series | Remote Sensing in Ecology and Conservation |
| spelling | doaj.art-71576394af62463bb298c5201b2571a02022-12-21T22:24:33ZengWileyRemote Sensing in Ecology and Conservation2056-34852021-03-0171364910.1002/rse2.159Estimating animal density in three dimensions using capture‐frequency data from remote detectorsJuan S. Vargas Soto0Rowshyra A. Castañeda1Nicholas E. Mandrak2Péter K. Molnár3Department of Ecology and Evolutionary Biology University of Toronto 25 Willcocks Street Toronto Ontario Canada M5S 3B2Department of Ecology and Evolutionary Biology University of Toronto 25 Willcocks Street Toronto Ontario Canada M5S 3B2Department of Ecology and Evolutionary Biology University of Toronto 25 Willcocks Street Toronto Ontario Canada M5S 3B2Department of Ecology and Evolutionary Biology University of Toronto 25 Willcocks Street Toronto Ontario Canada M5S 3B2Abstract Remote detectors are being used increasingly often to study aquatic and aerial species, which move significantly differently from terrestrial species. Camera‐trapping studies have used frameworks based on animal movement patterns to show that a species’ detection frequency, along with movement speed and detector specifications, can be used to estimate absolute population density. This approach, however, has not yet been adapted to cases where movement is three‐dimensional. Here we adapt one such framework to three‐dimensional movement, to characterize the relationship between population density, animal speed, characteristics of a remote sensor's detection zone and detection frequency. The derivation involves defining the detection zone mathematically and calculating the mean area of the profile it presents to approaching individuals. We developed two variants of the model—one assuming random movement of all individuals, and one allowing for different probabilities for each approach direction (e.g. that animals more often swim/fly horizontally than vertically). We used computer simulations to evaluate model performance for a wide range of animal and detector densities. Simulations show that in ideal conditions the method approximates true density well, and that estimates become increasingly accurate using more detectors, or sampling for longer. Moreover the method is robust to violations of assumptions, accuracy is decreased only in extreme cases where all detectors are facing the same way. We provide equations for estimating population density from detection frequency and outline how to estimate the necessary parameters. We discuss how environmental variables and species‐specific characteristics affect parameter estimates and how to account for these differences in density estimations. Our method can be applied to common remote detection methods (cameras and acoustic detectors), which are currently being used to study a diversity of species and environments. Therefore, our work may significantly expand the number and diversity of species for which density can be estimated.https://doi.org/10.1002/rse2.1593D movementideal gas modelpopulation densitypopulation surveysrandom encounter modelremote detectors |
| spellingShingle | Juan S. Vargas Soto Rowshyra A. Castañeda Nicholas E. Mandrak Péter K. Molnár Estimating animal density in three dimensions using capture‐frequency data from remote detectors Remote Sensing in Ecology and Conservation 3D movement ideal gas model population density population surveys random encounter model remote detectors |
| title | Estimating animal density in three dimensions using capture‐frequency data from remote detectors |
| title_full | Estimating animal density in three dimensions using capture‐frequency data from remote detectors |
| title_fullStr | Estimating animal density in three dimensions using capture‐frequency data from remote detectors |
| title_full_unstemmed | Estimating animal density in three dimensions using capture‐frequency data from remote detectors |
| title_short | Estimating animal density in three dimensions using capture‐frequency data from remote detectors |
| title_sort | estimating animal density in three dimensions using capture frequency data from remote detectors |
| topic | 3D movement ideal gas model population density population surveys random encounter model remote detectors |
| url | https://doi.org/10.1002/rse2.159 |
| work_keys_str_mv | AT juansvargassoto estimatinganimaldensityinthreedimensionsusingcapturefrequencydatafromremotedetectors AT rowshyraacastaneda estimatinganimaldensityinthreedimensionsusingcapturefrequencydatafromremotedetectors AT nicholasemandrak estimatinganimaldensityinthreedimensionsusingcapturefrequencydatafromremotedetectors AT peterkmolnar estimatinganimaldensityinthreedimensionsusingcapturefrequencydatafromremotedetectors |