Image Credit: Manfred Antranias Zimmer, Pixabay licence, Image Cropped

Invasion of freshwater ecosystems is promoted by network connectivity to hotspots of human activity (2019) Chapman et al., Global Ecology and Biogeography, https://doi.org/10.1111/geb.13051

The Crux

The spread of invasive species throughout freshwater ecosystems is a topic we’ve looked at before on Ecology for the Masses. In a previous paper breakdown we talked about how recreational is heavily responsible for the presence of non-native fish at a European scale.

Our paper this week takes a more local approach. Can we predict the presence of non-native birds, invertebrates and fish by looking at the presence of human activity, and where that human activity is present?

What They Did

The researchers worked with on a database which recorded visits of humans to different freshwater sites for different types of activity and compared this to nearby human indices like the presence of a car park or nearby population density. They then measured the likelihood of human activity at all freshwater sites in England involved in the study, based on those indices.

This was used to produce different measures of connectivity to human activity for freshwater sites. The measures were based on the amount of human activity immediately surrounding the lake, as well as activity upstream and downstream. They also wanted to test whether varying the importance of land further away from the site changed the effect of human activity. These measures were then compared to the overall non-native species richness or birds, crustaceans, fish, molluscs and plants at different locations throughout England.

Did You Know: Machine Learning in Ecology

Machine learning is a practice that has been around for a while. It essentially generates predictive models by detecting patterns in data, then tests those predictive models and adjusts itself based on how well it did. This week’s researchers used machine learning to come up with the likelihood of activity at a freshwater site. Whilst I currently find the technqiue a bit of a black box, it has gained popularity over the last decade or so in ecological circles.

What They Found

The importance of different types of human activity, and where that activity took place, differed between the species groups. For instance, fishing in the immediate vicinity of the site was the best predictor of non-native fish presence, but general recreation downstream was the best predictor for non-native crustaceans.

The importance of accounting for land further away from the site also changed the results. While recreation immediately downstream was a good predictor of non-native plant presence, it improved substantially when it placed more importance on land that was even further downstream. Likewise, whilst fishing was a good indicator of non-native fish presence, it worsened as more land further away from the site was taken into account.

Quagga mussels can cause huge problems in Britain’s freshwater ecosystems (Image Credit: NOAA Great Lakes Environmental Research Laboratory, CC BY-SA 2.0)

Problems?

This study suffers from the same problem that many studies involving this type of data do. Data involving species presence or absence is always going to be unevenly spatially balanced, as certain places are easier to survey and are therefore more likely to have better data. However this study was able to take that into account to a degree, by adding in ‘sampling effort’ as an additional variable into the models which compared human activity to species richness. It turned out to be quite an important variable as well, without which the researchers could have overestimated the importance of other factors.

So What?

Non-native species can be extremely hard to remove from an ecosystem once they’re present. So identifying pathways which help them move into new ecosystems is of vital importance in stopping their spread (read more about this in my recent interview with Helen Roy). If we know human activity is one of the major causes of introduction, and we can identify where that human activity is most intense and most likely to cause introductions, it means we can target educational campaigns and preventative measures in at-risk areas.