Environmental drivers can shift in effect direction when acting in concert with other factors
Results from a recent paper from our group in Nature Communications
We didn’t see that result coming, but what we got was very clear: an environmental factor (global change factor) can change in its effect direction when applied together with other factors, as reported in our recent paper in Nature Communications.
What we did (the we refers to the joint first-authors Rebecca Rongstock and Huiying Li, as well as co-authors Anika Lehmann and Anja Wulf) is an experiment using a new subtractive design. Instead of piling up more an more factors, which is what we have done before in several studies, now we start from the case of many applied factors (in our case six factors), and always take one out, so to speak. This allows us to gauge what the effect of that one factor was when acting together with the others: just compare what happens when it’s in vs. when it’s out when all the other factors are present. We can then compare that effect to the effect of the factor by itself, so when it is applied alone (in comparison to a control where we did nothing).
What you would expect that for sure the effect size would be different, because other factors are now also active; that would have been hardly shocking. However, what we did find is that some factors changed not only in effect size but also effect direction. That was quite surprising. For example, warming tended to have positive effects on measured soil variables, but the effect of warming was negative when acting together with all the other factors in our factor pool.
That result is interesting on several levels. First of all, we learn something about the mechanism of the effect, because we cannot assume the factor does the same thing when acting alone or in concert. Or maybe it has the same effect but changes how the other factors work. This makes predicting outcomes certainly more complicated.
Perhaps even more importantly, data from single factor effects are not necessarily useful in gauging which factor you should try to eliminate to prevent the worst effects for a soil or ecosystem. From the single effect result, your choice would certainly not have been warming as the ‘bad guy’ to focus on, since it had nominally positive effects on measured soil parameters; warming in fact would have appeared as relatively harmless. However, when in concert with the other factors, warming produced negative effects, so made things overall worse.
This flip in response direction with the addition of this factor is potentially interesting to consider in a restoration context, as we write in the paper: maybe restoration efforts should focus on eliminating or reducing the factor(s) that are worst when acting with others, rather than seeing who’s the bad one when acting by itself. Not sure what to call a factor with that property: not so bad by itself, but makes things worse when applied with other factors — keystone interactor perhaps? (suggestions welcome). Or does it even make sense to come up with a name for that factor rather than for this type of effect? What do you think?
At the very least, this result (and the new experimental approach) makes clear why it is so important to study effects at high dimensionality (meaning number of factors acting at the same time): something very surprising might happen. We could also have expected that the effects of the various factors cancel each other out, for example, or that they add up. That didn’t happen in our case. But until you do the actual experiment, you won’t know.
A further interesting point to make: we definitely need new ideas for experimental designs to tease apart what happens when a lot of factors act at the same time. This is not trivial to do….and there is still a lot to be learned; for example exactly why warming made everything so much worse.
Hope you find the paper interesting; I sure thought this was a cool result. :)
You can also read the press release of our university on this study, if you like.
How do you determine what is 'good' and what is 'bad' (worst) though? I know you talk about negative and positive perhaps more intended as directions, but even that has an undertone of 'preference'.
For example we see increasing soil porosity as 'good' (aeration) and decreasing as 'bad' (compaction) but what if the latter increases nitrogen leaching, therefore it then becomes 'bad' in that context? But to a tap rooted plant, it becomes 'good' because it might access more? We expand to include 3 factors - porosity, nitrogen, root architecture - and multiple directions emerge. So there is no 'right direction' in this sense? I imagine the direction we look at it might also change our interpretation - root architecture, nitrogen, porosity... - plus whether it is additive or subtractive.
Have you considered whether they reverse because our interpretation of 'good' and 'bad' is atomic rather than holistic? As you consider more (or less) factors, is this effect what you are seeing? Or even trying to see it in binary terms is why its surprising? If we assume that systems tend towards organisation rather than chaos or vice versa, how does that change our interpretation of 'good' and 'bad'? And what a system does to 'get better' or 'get worse'?
Perhaps getting into the philosophical but I do think we need to be careful interpreting what 'better' or 'worst' is.
Look forward to reading the paper - Annette