Plants & People

Dutch Scientists Learn How to Get Better Botanical Data from the Public

Instead of complaining about the limits of volunteer work, some Dutch botanists have found how to improve their understanding of data from citizen scientists.
You can listen to this page as an audio file.

Tracking changes in plant populations works best if you have a lot of eyes in the field. The public have a lot of eyes and, on a nice day, can enjoy a day out wandering through fields. It would make sense to combine the two, and have the public help record plant life. But a problem is that when the public go out for a walk they want to enjoy it, and will often prioritise that over scientific protocols. Instead of complaining about the public, Arco van Strien and colleagues in the Netherlands have tried to develop new statistical tools for surveys, so the public still get to enjoy themselves, but any data they find can still have scientific value. Their results published in Biodiversity and Conservation, show potential.

Golden tulips in the foreground give way to a dramatic cloudy sky. On the horizon is a large windmill, to remind people the survey was in the Netherlands.
Just some of the flora in the Netherlands. Image: Canva.

We tend to think of citizen science as a modern innovation, with the public aiding research over the internet. It’s a lot older than that. In 1902, the National Herbarium of the Netherlands and the Dutch Botanical Society set out to map the flora of the Netherlands. All of it.

They divided the country into grid cells of grid cell of 1.3 × 1.01 km. Volunteers recorded the vascular plants they found until around 1950. It was a lot of data and it’s only recently that’s all been digitised.

One problem in compiling the data was that there was no standard field protocol. That makes it difficult to track changes over time. If one survey differs from another, are you looking at a genuine difference, or were the observers just looking at the same plants in a different way?

Van Strien and colleagues asked volunteers to survey the same cells, so there was at least two surveys to compare. “This within-season replication allows for the application of occupancy models…, which can disentangle detection and occupancy probabilities using records from independent replicated visits. Without such separation between probabilities, higher observer effort over time may result in deceptive positive species trends, whereas it should only affect detection probability. Occupancy models are therefore currently the most powerful tool to adjust for variation in observer effort…” write the authors.

Another factor the botanists take into account is that plant occupancy can vary from year to year anyway. To overcome this, the scientists are aiming to compare occupancy between survey rounds, which take many years. This research paper has come out after the end of the first survey round, running between 2012 and 2019.

The team found that people miss species when they record what they see. The detection probability was 43% on average. For rare species, this can be a lot lower, and in some cases, it wasn’t possible to create an occupancy model. A lot of the methods of improving detection require more people – which isn’t feasible, but there are other options. First, practice helps. 

“The most experienced observers showed higher detection for many species, including species that are often viewed as more difficult, like grasses, tree and shrub species, and species belonging to species-rich genera. This suggests that training observer skills and increasing their experience could raise the detection for many species… Also, cell-phone apps for automatic species identification, such as ObsIdentify, may help to further improve observer skills,” write wan Strien and colleagues. They also suggest that phones could help in other ways.

“[I]t may be helpful to collect extra information on the search activity of the observers. Most observers nowadays already use a smartphone app with GPS to keep track of their search route and the exact time of observations within a grid cell, which enables new opportunities for occupancy modelling… If such information is combined with existing habitat maps, for instance, the time spent per habitat type may be retrieved and used as a covariate for detection probabilities of species associated with particular habitats.”

Despite this room for improvement, overall the method worked well. The survey could detect declines of 10% in the most common plants, and declines of 30% in over half the species. It looks like a viable method to work with people, instead of imagining what could be achieved if people were better botanists.

READ THE ARTICLE

van Strien, A.J., van Zweden, J.S., Sparrius, L.B. and Odé, B. (2022) “Improving citizen science data for long-term monitoring of plant species in the Netherlands,” Biodiversity and Conservation. https://doi.org/10.1007/s10531-022-02457-y

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.

English
%d bloggers like this: