Thread by @eva_gril

Hurray! My 1st first-author paper is now published - open-access - in @MethodsEcolEvol 🥳doi.org/10.1111/2041-2…
"Slope and equilibrium: A parsimonious and flexible approach to model #microclimate" ⛱️🌡️
Let me tell you what this new method is all about...

As ecologists, we know that temperature matters to organisms... And especially #microclimate temperature -locally experienced by biodiversity- rather than just the available coarse-grained temperature, which is based on weather station data (macroclimate, resolution 1 to 9 km)

So, we want to predict #microclimate from macroclimate, using vegetation / topography drivers.
The usual approach is to model offsets (the difference between T°micro and T°macro), and it works, but with some drawbacks... We argue for an alternative method to model microclimate!

Our approach relies on two simple parameters: the slope captures the linear relationship between #microclimate and macroclimate, while the equilibrium is the point at which micro = macroclimate!
A slope < 1 means the microclimate temperature is buffered compared to macroclimate

Basically, we use the linear relationship "Microclimate = a x Macroclimate + b"
The slope [a] has already been used in several microclimate studies. But to obtain microclimate, we also need the intercept parameter [b]... That's where the novelty is.

The equilibrium is better than an intercept because it has a biophysical meaning, so it can be modelled. During wet cloudy days with little radiative fluxes, a “mild” temperature is expected in and out of the habitat: T°micro = T°macro.
➡️The equilibrium follows mean temperature

Once we have the slope and equilibrium, we have the intercept! As they are mathematically related: intercept = equilibrium x (1 - slope)
We can then fully reconstruct microclimatic patterns from relevant local drivers and from macroclimate data readily available worldwide 🗺️

Although applicable to other habitats, we demonstrate the relevance of our method by focusing on forest understoreys and open grasslands across 13 contrasted sites, using a year of hourly temperature from sensors set at 1m (microclimate) and paired weather stations (macroclimate)

From these paired hourly temperatures, we applied linear regressions and thus extracted the slope and monthly equilibrium for each sensor, separating the leaf-on/leaf-off period because some of our forests are deciduous.
Here is an example of the process for a given site:

Note that we could have chosen different time scales for our regressions, such as weeks, but the monthly scale is common in microclimate studies.
We then used linear mixed-effects models to investigate the main determinants of the slope and equilibrium, our response variables!

The slope was chiefly determined by well-known microclimate drivers, stand structure predictors interacting with the period: stand type (conifer vs broadleaf); shade-casting ability; stand age; dominant height; stem density; and cover of the upper and lower shrub layer

In contrast, forest structure had no explanatory power on the equilibrium. As expected, we found the equilibrium to be positively related to mean macroclimate temperature

In brief, this general linkage between microclimate and macroclimate temperatures can be applied to any location or time if we know the mean macroclimate temperature (equilibrium) and the buffering or amplifying capacity of the habitat (slope).
Very nice perspectives ahead... 🌄

If you are interested in using this method, or if you still find this "equilibrium" parameter a bit fuzzy, you can read the whole paper at doi.org/10.1111/2041-2…
Tell me what you think!

The data and code are freely available on Figshare: doi.org/10.6084/m9.fig…
There is also a full RMarkdown report in .html, which means you can look at the code + outputs directly on your web browser, without even opening R!

MANY thanks to my fabulous co-authors, who came up with the idea, designed the experiment and/or helped write the paper: @f_spicher, @Linegreis, Michael B. Ashcroft, Sylvain Pincebourde, Sylvie Durrieu, Manuel Nicolas, @BRichard_Ben, @GuillaumeDecocq, @RonanMarrec, @EkoLogIt ✨

Also, deep thanks to all involved with the project such as the @ONF_Officiel agents from the #RENECOFOR network, @Emi_Gallet and coworkers from my lab @UMR_EDYSAN, our funders @CNRS @AgenceRecherche, and two dedicated reviewers including @RebeccaASenior 🙏