Mapping Species Distribution with Citizen Science
Occupancy models for citizen-science data (2018) Altwegg & Nichols, Advances in Modelling Demographic Processes, 10, p. 8-21
Species distributions maps are great. I remember rifling through animal encyclopedias as a kid, checking out the distributions of my favourite animals, just assuming that people knew exactly where to find all these organisms. But the reality is that figuring out exactly where species live is extremely difficult.
It’s made easier, however, by the use of citizen (or community) science. This occurs when volunteers involve themselves in projects in which they observe and report the presence or absence of a species in a given area, which is then used to determine a species’ distribution. This data is obviously incredibly useful to any ecologist, but it comes with some drawbacks. This paper attempts to summarise those drawbacks and outline ways to work around them.
What They Did
Unlike many of the papers we summarise on Ecology for the Masses, this paper didn’t have a strictly experimental phase. It is instead a review of the use of citizen science in ‘occupancy models’. Occupancy models essentially try to give the likelihood of a species occurring in an area, and thus are a major contributor to mapping species distributions. The paper lists a range of issues with incorporating citizen science into these models, 3 of which we’ll focus on: variation in observer capability, the choice of where to sample, and the problem of species movement.
Did You Know
Citizen science data undergoes many fluctuations over the course of a year. Sightings of seasonal species often peak at the beginning and end of seasons, as community excitement prompts excitement. But they also peak during weekends and holidays, as we’re more likely to hit the great outdoors then. When using citizen science, it can also be useful to take into account things like bad weather, as observers will generally be less eager and attentive during thunderstorms.
What They Found
This is a big one. Many occupancy models combine the probability of a species being in a given area with the probability of an observer seeing that species. But what if your observer skill varies? Variation among observer skill can often lead to very low estimates of species occurrence. This can be helped by encouraging multiple observers to visit sites and giving abundance data as opposed to presence/absence data, which can help quantify variation between 2 observers. Providing information on observer techniques can also provide insight.
Where to sample?
Citizen scientists are likely to visit easier to reach, more interesting spots. This can lead to higher probability in more these areas, which can result in inaccurate distribution maps. Study designs therefore need to encourage people to visit harder to reach places or less visually intriguing areas.
The closure problem
A lot of models assume that the probability of detecting species remains the same in one area, ie. that the area is ‘closed’. But this isn’t always the case. Species become harder to detect at certain parts of the year, whether due to physical (eg. fur colour darkens in summer) or behavioural (eg. stays in burrows to avoid cold) changes. Some species may even migrate away from entire areas. So studies that incorporate citizen science need to be temporally specific, and acknowledge species biology.
I love practical, applied examples in my textbooks, so I was a little surprised that they weren’t used more in this study. Showing the results of varying observer skill or levels of closure of an environment on a hypothetical distribution map based on citizen science would have been a great visualisation of the problems it can present.
This report is a valuable guide for ecologists who use citizen science data. In calling for more rigorous specifications and more documentation of observation procedures, it could lead to much more accurate species distribution maps in the future.