Considerations and Cons

Wildlife Photo Identification Technology saves time and increases accuracy to help researchers better gather data on their animal populations of focus; however, there are flaws in each technology that must be considered before utilization.

Photograph Quality: Photo Identification software can usually identify patterns and match poor quality pictures with more accuracy than human eyes, but even software can be useless when faced against a photo taken during a storm, of an animal running by, or of an animal partially hidden.

Animal Variability: Many animals have physical features and coat patterns that change seasonally or over their lifetime. Wildlife Photo Identification assumes that animals are unchanging, and therefore may not recognize the same animal over time. This could cause the assumption of more individuals within a species then are actually present. It is possible to photo identify animals with little variability within their species, but the rate of identification increases as species variability decreases (Morrison et al., 2011).

Human Error: With some projects generating massive numbers of photographs, researchers are beginning to branch out to the public for help sorting through pictures and identifying animals. Even trained researchers can make mistakes in identifying animals. When the untrained public is permitted to assist, it increases the chance of misidentification and decreases confidence in the results.

Technology Error: Some photo identification technology cannot detect 3-D aspects of animals. If a software recognizes the left side, it may not recognize the right side. Same goes for front/back and dorsal/ventral orientations (Hiby et al., 2012; McClintock et al., 2013). Similarly, animals photographed in front of or next to each other may be identified as one unique individual.

Incompatibility: It is essential that technology used for field research functions as it should. Traveling to fix and item or shipping an item into the field is costly and time consuming. When choosing the best technology, the environment it is exposed to must be considered. Can it hold up to the weather? Does all of the technology sync properly with each other? Is internet required for the technology, and if so is internet reliably accessible in the field?

Sources:

Hiby, L., Paterson, W. D., Redman, P., Watkins, J., Twiss, S. D., & Pomeroy, P. (2012). Analysis of photo-id data allowing for missed matches and individuals identified from opposite sides. Methods in Ecology and Evolution, 4(3), 252-259. doi:10.1111/2041-210x.12008

Mcclintock, B. T., Conn, P. B., Alonso, R. S., & Crooks, K. R. (2013). Integrated modeling of bilateral photo-identification data in mark–recapture analyses. Ecology, 94(7), 1464-1471. doi:10.1890/12-1613.1

Morrison, T. A., Yoshizaki, J., Nichols, J. D., & Bolger, D. T. (2011). Estimating survival in photographic capture-recapture studies: overcoming misidentification error. Methods in Ecology and Evolution, 2(5), 454-463. doi:10.1111/j.2041-210x.2011.00106.x

The Future of WPI

The reach of Wildlife Photo Identification goes beyond identifying species or individual animals. It can be used to map territories, determine biodiversity, and discover population trends. This can be invaluable when faced with wicked threats such as climate change.

HP Earth Insights has already stepped up to this monumental task through the creation of the Wildlife Picture Index, an interactive website allowing users to view cumulative statistics based on wildlife pictures from around the world (HP Earth Insights, 2013).

From large scale to small scale, this technology can make a difference. The most common individual identifiers for wildlife are implants (microships, transmitters), attachments (leg/neck bands, ear tags), and mutilation (scale/ear clippings, digit removal). Each of these techniques cause considerable stress with the requirement of catching and physically manipulating the animal in some way, often more than once. They are also by no means foolproof: mutilations heal, attachments fall off, and microchips malfunction.

Wildlife Photo Identification allows researchers to identify animals without ever needing to touch them. Even with current methodology this is possible. Upcoming technology, such as IBEIS holds promise for a future where stressful identification methods might not be necessary at all. In the future, I hope to see greater speed, accuracy, mobility, and tech compatibility, as well as immediate open-source data uploading to track global patterns of wildlife conservation.

With increased technological capabilities we can begin to branch beyond the animals themselves and into their habitat and symbiotic relationships. Kühl and Burghardt (2013) give some thoughtful example of what this might look like in their article about animal biometrics.

Source:

HP Earth Insights. (2013). Wildlife Picture Index. Retrieved from http://wpi.teamnetwork.org/wpi/dashboard

Kühl, H. S., & Burghardt, T. (2013). Animal biometrics: quantifying and detecting phenotypic appearance. Trends in Ecology & Evolution, 28(7), 432-441. doi:10.1016/j.tree.2013.02.013

Technology

Wildlife Identification begins with an image. Depending on the situation, this image may be created through a variety of methods including professional photography equipment, a cellphone, or a camera trap. The image may be collected directly from field research or supplied by the public. It is vital at this stage that the images be tagged with any relevant corresponding information such as location. Once images are collected, they must be analyzed to determine what species and/or individuals are present within that image. Once again, there are choices on how to best decipher what is contained in photographs. Here I will go into some details on the technology available to help researchers sort through their images with greater speed and accuracy.

Wild.ID: This software was originally created under the name desk TEAM for the Tropical Ecology Assessment and Monitoring (TEAM) Network to better identify the creatures caught on their camera traps. Wild.ID is a continuation of deskTEAM, modified to have greater efficiency and to be accessible to conservationists around the globe. Once downloaded, Wild.ID functions offline and allows researchers to connect to camera traps, analyze images, and identify animals. This imaging still requires researchers to go through and label pictures, but the offline capabilities, camera trap compatibility, and easy to use formatting of this software increases speed and organization to the process of identifying animals (TEAM Network, 2015).

Wild-ID: The similarly named Wild-ID was created by Dartmouth College to quickly identify unique individual animals by coat patterns. Still images of coat patterns are uploaded into the program and marked. Every future image uploaded of an animal with the same pattern is instantly identified as that individual. While it is limiting on species that can be studied, this method has been shown to be incredibly accurate for species with distinct markings (Bolger et al., 2012; Bolger and Farid, 2015).

IBEIS: The Image-Based Ecological Information System (IBEIS) is a soon to be released pattern identification software, much like Wild-ID. It is a continuation of HotSpotter software, which was shown to be slower yet more accurate than Wild-ID (Crall, 2013). Created in a collaboration of the University of Illinois-Chicago, Princeton University,  Rensselaer Polytechnic Institute, and Wild Me, IBEIS promises to deliver unlimited photo identifications, primate facial recognition, and environmental context analysis (Costelloe, 2015).

Instant Wild: When the image data is too massive for researchers to sort through, bring it to the citizens! That is exactly what the Zoological Society of London did with Instant Wild. Specialized Instant Detect camera traps transport pictures immediately into an online database accessed through phone applications around the world. Members of the public may then help researchers in sorting through and labeling images (Zoological Society of London, n.d.).

Wild Track: Sometimes it proves too difficult to take a picture of an animal and other methods of identification must be used. Wild Track has created a solution through footprint identification technology (FIT). Researchers simply upload pictures of animal footprints into the software which then identifies the species with a 95%-98% accuracy (Wild Track, 2016).

Sources:

Bolger, D., & Farid, H. (2015, September 9). New Dartmouth Software Tracks Wildlife with Photos, Not Tranquilizers. Retrieved March 06, 2017, from http://www.dartmouth.edu/press-releases/wildlifetrackinging02315.html

Bolger, D., Morrison, T., Vance, B., Lee, D., & Farid, H. (2012). A computer-assisted system for photographic mark-recapture analysis. Methods in Ecology and Evolution, 3(5), 813-822. doi:10.1111/j.2041-210x.2012.00212.x

Costelloe, M. (2015). Great Zebra and Giraffe Count is a huge success. Retrieved from http://ibeis.org/wordpress/

Crall, J., Stewart, C., Berger-Wolf, T., Rubenstein, D., & Sundaresan, S. (2013). HotSpotter – Patterned Species Instance Recognition.

TEAM Network. (2015, October 6). Wild.ID Flyer. Retrieved from http://www.teamnetwork.org/files/page/Wild.ID_InformationFlyer.pdf

Wild Track. (2016, December 5). Non-Invasive Wildlife Monitoring. Retrieved from http://www.wildtrack.org/

Zoological Society of London. (n.d.). EDGE of Existence. Retrieved from http://www.edgeofexistence.org/instantwild/

Why Identify? Pros

In order to conserve a species, data needs to be collected to ensure understanding of its current condition and to set obtainable goals for its preservation. How many individuals are within the population? How large and what shape is the range where these individuals live? How do the individuals relate and interact with each other? To reach maximum wildlife conservation success, individuals must be identified.

Population: Wildlife Photo Identification is an efficient way to discover how many species live in an area, the density of each population, and track individual animals throughout their lives. A good example of this was the use of Wild-ID to measure the population of two species of newts (Mettouris, Megremis, & Giokas, 2016). This is a necessary starting point for studying biodiversity, ecology, and vulnerability statuses.

Range: Properly geotagged images from a variety of locations have the potential of unveiling valuable information about species and individual territories, as well as migratory trends.

Relation: When tracking individual animals over time, data can begin to be collected  on how the individuals are related. A male and female photographed together during breeding season, may be assumed as mates. Lineages can be tracked as the mothers of each generation are photographed with their young.

Behavior: Not only can Wildlife Photo Identification reveal behavioral information caught on camera, it can also capture images of animals in which physical capture is unlikely. Even an animal as illusive as the snow leopard can be tracked using digital photography identification (Jackson, Roe, & Hunter, 2006). The element of danger and risk to the animal and researcher also decreases when cameras take place of physical capture.

Alternatives: Traditional identification methods involve capturing and marking the animal in some way, such as a leg band or cut in the ear lobe. More often than not, the identification method chosen requires the researchers to recapture the animals in order to see their previous marking. Markings may also heal, fall off, or malfunction. The process of marking and recapturing is time consuming and increases stress in the animal, causing a higher physical risk to the researcher. Wildlife Photo Identification technology allows researchers to mark animals without the need for capture at all (McClintock, 2013).

Time and Cost: The technology for Wildlife Photo Identification involves some initial time to install, learn, and configure. It also requires a method to secure photos, upload photos, and use the software, adding additional cost. (The software itself is often free or cost effective.) When compared to the time and cost of physically tracking, capturing, and marking animals, Wildlife Photo Identification is by far the best way to go.

Sources:

Jackson, R. M., Roe, J. D., Wangchuk, R., & Hunter, D. O. (2006). Estimating Snow Leopard Population Abundance Using Photography and Capture–Recapture Techniques. Wildlife Society Bulletin, 34(3), 772-781. doi:10.2193/0091-7648(2006)34[772:eslpau]2.0.co;2

McClintock, B. T., Conn, P. B., Alonso, R. S., & Crooks, K. R. (2013). Integrated modeling of bilateral photo-identification data in mark–recapture analyses. Ecology, 94(7), 1464-1471. doi:10.1890/12-1613.1

Mettouris, O., Megremis, G., & Giokas, S. (2016). A newt does not change its spots: using pattern mapping for the identification of individuals in large populations of newt species. Ecological Research, 31(3), 483-489. doi:10.1007/s11284-016-1346-y

Birds

Gaglio, D., Cook, T. R., Connan, M., Ryan, P. G., & Sherley, R. B. (2016). Dietary studies in birds: testing a non-invasive method using digital photography in seabirds. Methods in Ecology and Evolution, 8(2), 214-222. doi:10.1111/2041-210x.12643

• Used digital photography to identify the prey of greater crested tern (Thalasseus bergii) to show that photograph technology can reach beyond simply identifying individuals of a species, but also their diets.

Reptiles

Chassagneux, A., Jean, C., Bourjea, J., & Ciccione, S. (2013). Unraveling Behavioral Patterns of Foraging Hawksbill and Green Turtles Using Photo-Identification. Marine Turtle Newsletter; Swansea, 137, 1-5.

• Researchers identified individual hawksbill turtles (Eretmochelys
  imbricata) and green turtles (Chelonia mydas) with assistance from a computer-assisted process called the TORSOOI database.

Knox, C., Cree, A., & Seddon, P. (2013). Accurate identification of individual geckos (Naultinus gemmeus) through dorsal pattern differentiation. New Zealand Ecological Society, 37(1), 60-66.

• Photographed the dorsal side of jeweled geckos (Naultinus gemmeus). Gave new pictures of randomly selected geckos to volunteers and asked them to identify the individual geckos within a picture database. Volunteers had 100% success rate.
  

Mammals

Bolger, D., Morrison, T., Vance, B., Lee, D., & Farid, H. (2012). A computer-assisted system for photographic mark-recapture analysis. Methods in Ecology and Evolution, 3(5), 813-822. doi:10.1111/j.2041-210x.2012.00212.x

• A detailed study on the development and use of Wild-ID on Masai giraffe (Giraffa camelopardalis tippelskirchi).

Güthlin, D., Storch, I., & Küchenhoff, H. (2013). Is it possible to individually identify red foxes from photographs? Wildlife Society Bulletin, 38(1), 205-210. doi:10.1002/wsb.377

• Tested camera trap photo identification by the public (no software) of red foxes (Vulpes vulpes) and found that over 50% of images were not correctly identified.

Hohnen, R., Ashby, J., Tuft, K., & Mcgregor, H. (2013). Individual identification of northern quolls (Dasyurus hallucatus) using remote cameras. Australian Mammalogy, 35(2), 131. doi:10.1071/am12015

• This study tested the effectiveness of camera trap photographs to identify northern quolls (Dasyurus hallucatus), and makes some suggestions on how to modify settings to take better pictures of target animals.

Sirén, A., Pekins, P., Abdu, P., & Ducey, M. (2016). Identification and Density Estimation of American Martens (Martes americana) Using a Novel Camera-Trap Method. Diversity, 8(1), 3. doi:10.3390/d8010003

• Recommendations and comparisons of camera trap methods to get the most identifiable pictures possible of American Martens (Martes americana).

Amphibians

Bendik, N. F., Morrison, T. A., Gluesenkamp, A. G., Sanders, M. S., & O’Donnell, L. J. (2013). Computer-Assisted Photo Identification Outperforms Visible Implant Elastomers in an Endangered Salamander, Eurycea tonkawae. PLoS ONE, 8(3). doi:10.1371/journal.pone.0059424

• Tested reliablitiy of results between Wild-ID high resolution images, Wild-ID low resolution images, and Visible Implant Elastomers (VIE) on Jollyville Plateau salamanders (Eurycea tonkawae). Found that, as long as the images are high resolution and the animals are variable enough, Wild-ID is a comparible method to VIE.

Mettouris, O., Megremis, G., & Giokas, S. (2016). A newt does not change its spots: using pattern mapping for the identification of individuals in large populations of newt species. Ecological Research, 31(3), 483-489. doi:10.1007/s11284-016-1346-y

• Used Wild-ID to successfully identify individuals within two newt species populations, alpine newts (Ichthyosaura alpestris) and smooth newts (Lissotriton vulgaris).

Sannolo, M., Gatti, F., Mangiacotti, M., Scali, S., & Sacchi, R. (2016). Photo-identification in amphibian studies: a test of I3S Pattern. Acta Herpetologica, 11(1), 63-68. doi:10.13128/Acta_Herpetol-17198.

• A test of the I3S Pattern Software on Triturus carnifex to see if the technology is applicable to amphibians.

Marine Animals

 

Gore, M. A., Frey, P. H., Ormond, R. F., Allan, H., & Gilkes, G. (2016). Use of Photo-Identification and Mark-Recapture Methodology to Assess Basking Shark (Cetorhinus maximus) Populations. Plos One, 11(3). doi:10.1371/journal.pone.0150160

• This study examined the ways in which wildlife photo identification was and was not beneficial in tracking individual basking sharks (Cetorhinus maximus).

Florida Fish & Wildlife. (n.d.). Manatee Photo-identification Program. Retrieved from http://myfwc.com/research/manatee/research/photo-identification/program/

• A website by the Florida Fish and Wildlife Service describing their use of photo identification on manatees.

Marineland. (2017). Dolphin Photo Identification. Retrieved from http://www.marineland.net/home/gacfs/research/dolphin-photo-identification

• The Marineland website about their photograph identification program on dolphins.

Town, C., Marshall, A., & Sethasathien, N. (2013). Manta Matcher: automated photographic identification of manta rays using keypoint features. Ecology and Evolution, 3(7), 1902-1914. doi:10.1002/ece3.587

• Researchers successfully used a Scale-Invariant Feature Transform (SIFT) algorithm to identify individual manta rays (Manta alfredi and Manta birostris).

Urian, K., Gorgone, A., Read, A., Balmer, B., Wells, R. S., Berggren, P., Hammond, P. S. (2014). Recommendations for photo-identification methods used in capture-recapture models with cetaceans. Marine Mammal Science, 31(1), 298-321. doi:10.1111/mms.12141

• Suggests ways to reduce bias in photograph selection when preforming wildlife photo identifications on cetaceans.