The management of free-roaming cat populations is important because, unsupervised, cats can prey on wild animals and spread diseases to other species. Research shows that spay and neuter programs can reduce the number of cats placed and killed in shelters. However, it remains unclear what proportion of cats in an area need to be sterilized to see a population reduction, and how long it takes to see that decline.
Currently, the gold standard for estimating population size is the mark-recapture technique. For example, a small number of cats are identified and “marked” by taking their picture. Later, another group is marked, and scientists record and analyze the number of cats photographed in both sightings. Motion-activated camera traps are usually used for this research, but such technology is expensive and often produces low-quality photos that are difficult to identify.
In this study, researchers aimed to validate the use of smartphone photos as a way to identify free-roaming cats. Specifically, they wanted to understand whether cat advocates can successfully match smartphone photos of cats compared to university science students (who are usually trusted to analyze camera trap research).
In other words, the researchers wanted to know if untrained public citizens can correctly act as “citizen scientists” to identify cats in photos used for mark-recapture analysis. Evidence suggests that tapping into crowdsourced free labor does not cause a significant decrease in the rigor of the scientific process.
Using an iPhone XS, a researcher collected cat photos over two months across Washington and Oregon, during periods when free-roaming cats are active. Photos were taken by walking along a street and photographing any observed cats, which is how volunteers would be instructed to collect data. No filters or zoom lenses were used, and afterward, the photos were not enhanced.
Afterward, 151 cat advocates and 17 life science students were shown two cat photos and asked whether the same cat appeared in both photos. Researchers assessed common features of the photos that were most, and least, identifiable, as well as the demographic characteristics of respondents.
It appears that cats are generally identifiable from smartphone photos, as advocates’ matching attempts were correct 98% of the time compared with students’ 97.5%. Students made 27% more errors than cat advocates, suggesting that advocates can be relied upon for studying free-roaming cats. There were two types of error: false positive (matching photos that were not the same cat) and false negative (ignoring a match). Both groups showed a false positive trend, which, in practice, can underestimate the true population size.
Secondarily, the researchers considered which characteristics make a volunteer better at cat identification. They found that the strongest-performing volunteers had a companion cat, a bachelor’s degree or higher, and had no previous cat volunteer experience. It’s possible that volunteer experience led to overconfidence and less caution in selecting matches. Regardless, the authors point out that recruiting previous volunteers as citizen scientists may be counterproductive.
The researchers also accounted for the factors that make a cat easier to identify in a smartphone photo. The results showed that non-black cats were more identifiable. Color outranked other aspects like face visibility and the presence of a collar. However, this does not mean that black cats cannot be identified. Instead, when building a photo log to estimate population size, it’s important to consider numerous features of each photo, including (but not limited to) color.
This study reveals that professional experience does not guarantee expert-level identification of individual animals; enthusiastic non-professionals may be just as accurate at identifying animal photos, perhaps even more so. When planning a study of free-roaming cats, animal advocates may consider using smartphone photos and recruiting citizen science volunteers to reduce cost, engage the public, and potentially improve the quality of their data.
The post Do Cat Advocates Make Good Citizen Scientists? appeared first on Faunalytics.
Currently, the gold standard for estimating population size is the mark-recapture technique. For example, a small number of cats are identified and “marked” by taking their picture. Later, another group is marked, and scientists record and analyze the number of cats photographed in both sightings. Motion-activated camera traps are usually used for this research, but such technology is expensive and often produces low-quality photos that are difficult to identify.
In this study, researchers aimed to validate the use of smartphone photos as a way to identify free-roaming cats. Specifically, they wanted to understand whether cat advocates can successfully match smartphone photos of cats compared to university science students (who are usually trusted to analyze camera trap research).
In other words, the researchers wanted to know if untrained public citizens can correctly act as “citizen scientists” to identify cats in photos used for mark-recapture analysis. Evidence suggests that tapping into crowdsourced free labor does not cause a significant decrease in the rigor of the scientific process.
Using an iPhone XS, a researcher collected cat photos over two months across Washington and Oregon, during periods when free-roaming cats are active. Photos were taken by walking along a street and photographing any observed cats, which is how volunteers would be instructed to collect data. No filters or zoom lenses were used, and afterward, the photos were not enhanced.
Afterward, 151 cat advocates and 17 life science students were shown two cat photos and asked whether the same cat appeared in both photos. Researchers assessed common features of the photos that were most, and least, identifiable, as well as the demographic characteristics of respondents.
It appears that cats are generally identifiable from smartphone photos, as advocates’ matching attempts were correct 98% of the time compared with students’ 97.5%. Students made 27% more errors than cat advocates, suggesting that advocates can be relied upon for studying free-roaming cats. There were two types of error: false positive (matching photos that were not the same cat) and false negative (ignoring a match). Both groups showed a false positive trend, which, in practice, can underestimate the true population size.
Secondarily, the researchers considered which characteristics make a volunteer better at cat identification. They found that the strongest-performing volunteers had a companion cat, a bachelor’s degree or higher, and had no previous cat volunteer experience. It’s possible that volunteer experience led to overconfidence and less caution in selecting matches. Regardless, the authors point out that recruiting previous volunteers as citizen scientists may be counterproductive.
The researchers also accounted for the factors that make a cat easier to identify in a smartphone photo. The results showed that non-black cats were more identifiable. Color outranked other aspects like face visibility and the presence of a collar. However, this does not mean that black cats cannot be identified. Instead, when building a photo log to estimate population size, it’s important to consider numerous features of each photo, including (but not limited to) color.
This study reveals that professional experience does not guarantee expert-level identification of individual animals; enthusiastic non-professionals may be just as accurate at identifying animal photos, perhaps even more so. When planning a study of free-roaming cats, animal advocates may consider using smartphone photos and recruiting citizen science volunteers to reduce cost, engage the public, and potentially improve the quality of their data.
The post Do Cat Advocates Make Good Citizen Scientists? appeared first on Faunalytics.