It was at a lunchtime seminar of our research group that Cindy De Jonge introduced a new concept to me: using variation in cell membrane lipids (affectionately called brGDGT lipids by those who love them) as a thermometer for the past: these molecules remain super stable in soils and sediments (for millions of years!), and their structure can be related to environmental conditions when the organism lived. What makes them most interesting as a ‘paleothermometer’ is that at the global scale, their distribution in soils changes with mean annual air temperature (and soil pH).

However, there was a lot of remaining noise, so I learned, and Cindy was on a mission to find out what else was happening. As I was working hard on the role of soil temperature – rather than air temperature – as driver of ecological patterns, we figured there was a beautiful match. What if we would go to our elevational gradients in northern Scandinavia, where we had in-situ measurements of soil temperatures across a 1200 meter gradient, and took soil samples there? We could relate the distribution of the membrane lipids in the soil to the in-situ measured temperature, and see if we would get better relationships than with coarse-grained air temperature. Makes sense, right? Indeed, one would expect that bacterial lipids relate more strongly to the temperatures in the soil, which are especially in the snow-rich northern Scandes, largely disconnected from what macroclimate data would tell you.

So we set to work, in what could be seen as an example of ‘rapid’ scientific progress. We obtained a little grant to do the labwork, attracted a motivated master student (Robin Halffman, my first master student actually to publish his master work as a paper, what an incredible achievement!) and went to the mountains to get us some soil.

After a lot of labwork, data processing and even more thinking, we finally got our story – out now in Organic Geochemistry!: things were complicated (who would have expected that!?). Luckily, there was Cindy to see the forest for the trees in an impressively complex real-world ecology detective case: indeed, brGDGTs did NOT relate better to in-situ soil than to free-air temperature, despite our smart hypothesis. Instead, it turns out the soil chemistry (e.g. its pH) is a much better predictor, and that this soil chemistry seems to act through the bacterial community: different soil chemistry –> different bacterial community –> different membrane lipids.

So what do we learn from this exercise? A few important things:

1. Despite the nice correlation of brGDGTs with temperature at the global scale, the local scale is clearly more complex. Indeed, local drivers of bacterial community distribution can be very different from temperature, with especially soil chemistry very important. This does not jeopardize the potential of the brGDGTs as paleothermometer, however, but should warn us to use it – at the moment – only at a coarse spatial resolution.
2. It looks like (at least to my understanding) that the structure of membrane lipids is less related to environmental conditions itself, than to the bacteria making them. A better understanding of soil bacterial biogeography is thus urgently needed to take this further.
3. My favourite conclusion: microclimate data is tremenduously important across so many branches of ecology, especially in the soil. Nevertheless, we should remain critical: it will not improve the story in all cases.We used one year of in-situ measured temperature data (versus decadal averages of free-air temperature). We know, however, that these membrane lipids are super stable in the soil, so it is actually unlikely that they relate strongly to annual fluctuations in local temperature. Better correlations might be obtained when using long-term soil temperature proxies (as now FINALLY are becoming available, see this preprint). Plans for future tests, for sure!