A continuous water quality monitoring platform on Peter Lake helped raise alarms that the system was approaching an ecological tipping point (Credit: Ryan Batt)
Big shifts in ecosystems–say, from a food web made up of native species to one dominated by invasive–can happen quickly and with relatively little warning. But a new study shows that intensive data provided by environmental sensors could help give managers a heads up for impending tipping points.
“The research is in a fairly early stage,” said Ryan Batt, a graduate student at the University of Wisconsin-Madison’s Center for Limnology and lead author of the study. “But I think in the long term we’d hope that it could contribute to being able to better anticipate these big changes and take action accordingly while the cost to benefit ratio for mitigating environmental impacts is still really low.”
The method for predicting the approach of ecological thresholds–the lines a system crosses before undergoing a shift and settling into a new, perhaps undesirable, state–have mostly been a focus of theoretical and laboratory research. This study, published recently in the Proceedings of the National Academy of Sciences, is an important step towards applying the tools in real ecosystems.
The study also shows the value of high-resolution, long-term data sets generated by increasingly common continuous water quality monitoring systems. The predictions are based on statistical tools that require a lot of data–more than what routine manual monitoring programs can supply.
“That’s the downside to these statistics, they’re very data-hungry,” Batt said. “Weekly sampling won’t cut it. ”
The researchers helped move the method from the lab to real-world systems by conducting an experiment on two small lakes in Northern Wisconsin. The lakes, named Peter and Paul, were once a single hour-glass shaped lake that was divided by an earthen dike in the 1950s, spawning a tradition of whole-lake experiments in which Peter is manipulated and Paul serves as a reference.
This experiment took advantage of the differences in the two lakes’ food webs. While Paul Lake was dominated by largemouth bass, Peter Lake had few bass and was dominated by planktivorous species like minnows, shiners and pumpkin seeds. From 2008 to 2011, the researchers added bass to Peter Lake in an effort to push the food web into a state similar to Paul’s.
Throughout the experiment, water quality conditions on each lake were monitored by a YSI multi-parameter sonde equipped with sensors measuring dissolved oxygen, pH, chlorophyll-a and conductivity. The sensors, suspended from a wooden raft, collected data every five minutes for four years.
As the researchers gradually added bass to Peter Lake, they were pushing the system toward a threshold at which the food web would quickly shift into a stable bass-dominated lake. That event is what the statistical methods are designed to predict, but the exact threshold–for example, the specific number of bass required to make the system switch–is difficult to pin down.
Instead of calculating those thresholds, the statistics instead rely on the concept of ecosystem resilience. That refers to a system’s ability to recover from any sort of shock or disruption. As an ecosystem approaches a threshold, it becomes less resilient. While the threshold is difficult to calculate, scientists can use estimates of resilience to indicate that a tipping point is approaching.
Peter Lake crossed a threshold and quickly shifted into a bass-dominated lake around the end of 2010. More than a year before that, the statistics fueled by the high-frequency sensor measurements of chlorophyll, dissolved oxygen and pH gave an early warning that resilience was low and a tipping point was likely on the way.
Batt said the underlying hope for this type of research is that these indicators of changing resilience could be part of a tool set that someone could use to gain an understanding of how close it could be to a significant change.
“We’ve known for a long time that it’s a big problem for management, because not only are these changes big and difficult to predict, but they’re also extremely hard to reverse,” Batt said. “Once you switch into a state, the system wants to stay there. It just sort of snaps into one spot or the other and it just tries to hold there.”
Perhaps the most important contribution from this study is its proof that environmental sensors can provide both the volume and quality of data necessary to run the statistics. That’s good, because many monitoring programs have already turned to sensors for other purposes.
“A lot of people are using sensors and they’re using them more and more all the time. And it turns out that the sensors give you the kind of data that you need to do this,” Batt said. “Tons of people have these sensor data, and this points out a potentially really cool use for these kind of data.”
Top image: A continuous water quality monitoring platform on Peter Lake helped raise alarms that the system was approaching an ecological tipping point (Credit: Ryan Batt)