How to read a scientific paper (1/2)
By Adrienne Funderburg, research program specialist | Read part two 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! The following is a summary you can use to guide you through the paper or read on it’s own to understand our recent cyanobacteria and microcystin research.
If you work in manufacturing, product development, or data collection, you’re likely familiar with the phrase “quality control.” This is the process of double-checking the product for consistency and validity. Academic research papers are run through a quality control process too, called “peer-review,” in which multiple professional researchers in the area of study edit and critique a paper submitted for review. They aren’t just looking for correct their/there/they’re’s and complete sentences, but for proper scientific method, references to previous published work in the area, and logical conclusions based on trustworthy data. Research papers that get the green light from their reviewers and the journal editor are compiled and made accessible to other researchers, and then it can serve as a reference and contribute to the body of knowledge on the subject. Our paper was published in Proceedings of the Indiana Academy of Science.
We focused this research on cyanobacteria and microcystin. (If you’re wondering what those are and why they matter, click here to get oriented!) We know that cyanobacteria and their toxins are a global and complex issue. Cyanobacteria populations and microcystin levels tend to be highest in mid-summer, which is peak season for lake recreation and algae/cyano growth. However, researchers have not yet found a consistent relationship between cyanobacteria growth and their toxin production. Scientists are performing experiments and studies to answer questions such as, “How do cyanobacteria decide when to produce toxin?” and “How can we best predict toxin production?” The Lilly Center investigated our own preliminary questions with a 2009 pilot study and a subsequent 2010-13 study on our Kosciusko County lakes.
Countless more questions on these topics are being asked around the world and across the U.S., but we developed this 2015-17 study focused on one globally significant problem and one locally relevant question. We aimed to address the problem of uncertainty in public health when on the lakes by testing a way to rapidly assess, or predict, toxin levels. While hunting for a solution to that problem, we asked the question, “What are cyanobacteria and microcystin conditions like in lakes across Kosciusko County recently?” The words “across” and “recently” inform our analysis of the data; we looked for patterns in cyanobacteria and microcystin over space and over time to gain understanding of their influences.
To answer our question and address the issue we previously identified, we studied local recreational hotspots (Kosciusko County’s 12 all-sport lakes; Table 1 on page 4) over the peak recreational and cyano/toxin season (summertime) over three years (2015, ‘16 and ‘17). Each of our 12 lakes was sampled weekly over the summer, and we collected data on atmospheric conditions, water quality, nutrient levels, algae/cyanobacterial pigment levels as comparison variables to our key data points: algae/cyanobacteria populations and microcystin levels.
With data in-hand, we began data analysis: the translation from numbers into understanding. To answer our question about cyano and toxin conditions and patterns, the analysis process consisted of creating tables, calculating averages and percent differences, and hunting for notable spikes and dips in values over the weeks and years, and between individual and groups of lakes. These make relationships between parameters, or a lack of them, easier to spot.
As for a quick assessment of toxin levels, mathematical models (math formulas used to describe the real world) were created out of individual lake variables, pairs of variables, and larger groups of variables to see which lake and models did best at predicting microcystin levels and cyanobacteria populations. We double-checked the accuracy of the models’ predictions with our real toxin and cyano measurements from each sampling day, and calculated their accuracy. We assessed the models in two ways: by their accuracy, and by their safety, the second of which is important when working with public health concerns like microcystin. The safer model trades some accuracy for caution, generating fewer false negatives, or reports of “It’s safe!” when in fact toxin levels may be high.