Recently, we invited you to ask us anything. Today’s question comes from Chrissy Hernandez, who asks (paraphrasing): where do ecological systems fall on the continuum from random to deterministic?
Jeremy’s answer:
Guest poster Karen Abbott’s answer is better than mine will be.
I don’t think ecological systems can be placed on a unidimensional continuum from “deterministic” to “stochastic”. The interplay of deterministic and stochastic forces is too rich and complicated for that. It’s not just that there’s black and white, with shades of grey in between. It’s more like, there are different sorts of black, and different sorts of white, and in between is a rainbow of colors that you can get by combining different sorts of black and white in different proportions. Think for instance of phenomena like quasi-cycles and noise-induced chaos, that arise when you combine the right sorts of deterministic and stochastic ingredients in the right ways. One classic statement of this point of view is Bjørnstad et al. 2001.
I do think it’s possible to place ecological systems on a continuum from “less to more intrinsically predictable, given information about past system states.” Pennekamp et al. (2019) is a great paper on this. I’m toying with some ideas for building on their work, and I’m sure I’m far from the only one. I like their approach of using permutation entropy to measure the intrinsic predictability of ecological systems, because it turns questions about intrinsic predictability from unanswerable armchair philosophy questions into answerable empirical questions. The hypothetical question of whether ecological systems would be perfectly predictable, if only we had perfect information about all the subatomic particles in the world, doesn’t strike me as a useful question to ask. Because there’s obviously no way we’re ever going to have nearly as much information as that. In contrast, asking “How intrinsically predictable are the future states of ecological systems, given the information we actually have about their past states?” strikes me as a question that’s very useful to ask. You can answer that question, and you can imagine building on the answer in various ways.
Brian’s answer:
Whether a system is deterministic or stochastic is not a property of the system, it is a property of how we study the system. Saying something is stochastic is just a statement of willful ignorance about the detailed mechanisms and processes. To be clear, I don’t mean that in a bad way – willful ignorance saves a lot of time and money, and I choose willful ignorance all the time in macroecology.
Think about the idea of drift in population abundance that many people equate with a stochastic process. That’s really just saying I don’t want to get into the details about why those three rabbits all died this month (cold snap, virus, rare combination of bad luck). But I could if I wanted to and had enough data. That last sentence is the line between what is stochastic and what is deterministic.
This is a little different than Chrissy’s question, but close enough for me not to be able to resist a hot take! For this reason, I think it is terrible that the main outcome from Hubbell & Bell’s neutral theory has been this descent into the claim that systems are on a continuum form stochastic to deterministic. Systems aren’t anywhere on that spectrum. Again it’s our methods and foci of interest that are on that spectrum. And what a typical outcome, sadly, for ecology as a field to take a big bold idea and devolve into “everybody is right, some systems are more one and some are more the other”. That is not an advance of knowledge in my opinion. But that is what ecology does over and over! (Jeremy adds: I think it was really, really unfortunate that, for a while there, many ecologists started using “neutral” and “stochastic” as synonyms. Not only are they not synonyms, they have nothing to do with one another!)