Citizen Science and Machine Learning Could Help us Track Climate Change

Climate change is known to be impacting plant species. For example, in many regions spring-like weather is arriving earlier in the year. This impacts the plants and animals in these areas. However, collecting and analyzing the amount of data required to better understand what these impacts are for multiple species and in many different locations is just not possible.

Mora and colleagues (2024) have come up with a possible option. They used data from a plant identification app to generate a database of nearly ten million plant observations in Germany from 2018 to 2021. The app uses an AI interface so users can take a photo of a plant and get an AI-generated ID. But more importantly for the research, it records the location and time the photograph was taken. They then used machine learning to analyze that data in order to quantify the seasonal changes for different plants in different spaces and times.

You might be asking, though, if there is a bias here. Wouldn’t plants be more likely to be photographed at certain times of year (e.g., spring or summer versus winter) and days of the week (e.g., weekends)? The answer is yes, but their algorithms could account for these human factors and still produce results regarding climate-induced shifts. I also wondered about whether the number of observations might decrease over time as users of the app learn how to identify some of the plants themselves. This was not addressed by the researchers but it perhaps isn’t an issue within the timescale considered or is accounted for by new people starting to use the app. Therefore, it might be something to explore in the future.

This is another cool use of AI technologies to support sustainability research. By detecting changes in plants there are lots of implications. One possible implication is how it might impact local pollinators and whether that will then have an impact on agriculture and food security.