While I was working on young cod for my honours thesis, I joined a project testing whether the stress hormones wildlife biologists use to judge a population’s health actually reflect it. It was my first meta-analysis.
Led by Dr. Levi Newediuk, with me as second author. Published in Conservation Physiology, 2023.
Stress hormones are one of the easiest things to measure in a wild animal, from a blood sample, a feather or even droppings. So conservation biologists often read high levels as a sign that a population is in trouble. That only works if hormones really track health. Checking it meant pulling together results from across the literature, and this project was where I first learned how to do that.
Helped sift 519 search results down to the studies that measured both stress hormones and health in the same animals.
Helped write the R code that combines 109 results from 41 species and tests whether hormones and health move together.
Helped write the manuscript. Published in Conservation Physiology in 2023, with me as second author.
A population’s stress hormones didn’t reliably tell us how healthy it was.
On average, stress pushed hormones up and health down, as expected. But study by study, the two often didn’t move together, even in populations of the same species. Meanwhile, more than half of recent conservation studies we checked (18 of 35) had judged a population’s health from its hormones without testing that link.
72 of 107 results landed where you’d expect…
…but 35 didn’t.
To test whether hormones and health actually move together, we fit a model that compares the two directly across studies while allowing for how precise each study was. Its best estimate of the link sits almost exactly at zero, and the likely range stretches from a strong link in the expected direction to a strong link in the opposite one. In plain terms, the data can’t say whether more stress hormone means worse health, better health or neither.
The likely range covers almost every possible link…
…and only 48% of it falls on the expected side, about a coin flip.
In the paper we set out why a hormone reading can mislead. First, why high stress hormones are usually a bad sign. Then three cases where reading them that way gets it wrong.
In the paper: