How to read a scientific paper (2/2)
By Adrienne Funderburg, research program specialist | Read part one here
In late 2019, the Lilly Center published a paper entitled “Blue-Green Algae (Cyanobacteria) Patterns and Predictions in 12 Lakes in Kosciusko County, Indiana” in a scientific journal. While written for lakes and streams researchers, this paper is brimming with info relevant to everyone who loves the lakes, professional scientist or not! This post is the second half of summary you can use to guide you through the paper.
Here’s what we found after we dug through piles of data: While our lakes experienced very similar atmospheric conditions as one another, their nutrient levels, water clarity, algae/cyanobacteria populations, and microcystin levels varied a lot. Tables 2, 3, and 4 in the paper display these results by lake, and the paragraphs surrounding them describe the patterns we observed in these numbers. But Table 4 boasts some of the most perplexing content of them all. Take a look at Big Barbee and Winona Lake. These lakes had the lowest summer average for microcystin, but as shown in the next column, also had the highest cyanobacteria (BG algae) counts of all the lakes. Dewart and Wawasee experienced just the opposite: high microcystin levels and smaller cyanobacteria populations. Water clarity doesn’t shed any light on this relationship; some lakes had clearer water but more toxin, others the opposite, and some lakes with similar levels of toxin had vastly different clarity.
We also found some temporal trends common to our lakes, such as decreases in water quality approaching July each year, and slight differences in the summers themselves (in temperature, precipitation, wind, and other external factors) influenced the lakes equivalently.
The most successful of our models for predicting toxins and cyanobacteria populations were those that took the most variables into account, giving correct results 77-80% of the time. Table 8 displays each model type and its percentage of correct and incorrect results. The models calibrated to prioritize public safety gave false negatives (5-7% of the time) and false positives (“unsafe” calls when actually safe) 15-16% of the time. And, as anticipated, models optimized for accuracy over safety were correct slightly more often.
So what does all of this mean for us as researchers? Or more importantly, what does it matter for you as a lake enthusiast?
First, the similar spatial patterns and dissimilar clarity, nutrient, toxin, and cyanobacteria levels emphasize the uniqueness of each lake. Even though our county lakes are geographically close, they need to be addressed with strategies that take into account their own water quality and local strengths and challenges.
Second, these results lead us into more questions on cyanobacteria and microcystin levels. Are different species of cyanobacteria influencing microcystin occurrence in our lakes? Are zebra mussels, which eat green algae and leave behind cyanobacteria, a major factor in our lakes’ algae populations and toxin levels? These two questions in particular have guided our sampling and data analysis already as of last year and will continue to do so as we get closer to answering them and finding new questions again.
Third, we were able to confidently say that predictive efforts, while helping us learn more about cyanobacteria and microcystin in our lakes and beyond, were not reliable enough to rely on for public health and safety. Only actual measurements of microcystin, if brought in-house and analyzed weekly, would be sufficiently quick and accurate to keep lake recreationists informed. Thanks to funding from the K21 Health Foundation, we purchased that equipment and began our own weekly microcystin testing in-house last summer and plan to make results available on our website this upcoming summer as we test. Keep your eyes open for that, the results of our 2019 zebra mussel survey, other new Lilly Center research by following us on Facebook and signing up for our e-newsletter as you enjoy our beautiful lakes!