Wednesday, November 28, 2012

Why consistent terminology is important

Some time ago, I received this email from a grad student:
"Do you know the blog zombies ideas in ecology?? I think this is the kind of ideas that could interest you. After reading all these chase papers I just find myself completely lost in the meaning and use of words such as stochastic, random and neutral…. This text kind of help me (it’s a critic of the use of these terms in community ecology) but I’m still lost…do you have any tips on that?? It seems to me that a stochastic process is a process that you can  only predict in probabilistic term. But then it can of confuse me because it seems for me that neutrality is random and randomness is a stochastic process, right??..

And shortly after that a 2nd question from the same student:
"And what to do with that : . All Chase works on beta diversity assumption and works on limiting distance similarity is to throw in the garbage??..."
I took me a while indeed to figure out first the problem, and second a potential solution. I will try to break it down, and I will not use full prose, since I am still trying to work out the best way to present and think about this myself, thus the semi-concept map style of writing below.


  • spaghetti clump of terms in the literature:
    • neutral, drift, random, stochastic, unexplained, error term, probabilistic equivalence, dispersal limitation, noise, chance, spatial processes
    • deterministic, niche-based, selection
  • there is not a lot of confusion about niche-based processes = selection

  • neutral and stochastic pretty similar, while different definitions
    • stochastic - in Chase and Myers 2011: "chance colonization, random extinction and ecological drift", "indistinguishable from random chance alone"
  •  but
    • dispersal does not have to be neutral, while this is implied by the definition of Rosindell et al.
    • neutral and stochastic often used as synonyms in e.g. Chase and Myers
  • So lots of confusion around the first group of terms, hence the spaghetti clump 

  • Fox provides some thoughts on a solution, but I think that could use some more explanation (and he will probably correct me if I am wrong ;-)
  • looking at the four community processes from Vellend 2010, I suggest to identify five, fundamentally orthogonal, questions you need to ask to classify a process (and thus the associated terminology)
    • What is the time frame of the process: ecological versus evolutionary time frame (or is it necessary/useful to include speciation as an important process?)?
    • What is the spatial scale of the process, or the metacommunity context:  within- or between-site (dispersal) interactions? I would refrain from using "local vs. regional" here, because "regional" also has the connotation of "at a larger scale", while the action of these "regional" process are sometimes at the within-site scale. For instance, climatic features such as rain or temperature are regional in nature, but each organism experiences this within a site. So it is a regional process at a within-site scale, in that case (another quagmire of confusing terminology).
    • What is the nature of the species differences: are they niche-based (or under selection sensy Vellend 2010) versus neutral (or probabilistic equivalence of species sensu Rosindell et al.)?
    • What is the stochastic nature of these ecological processes: Do they result in a deterministic or stochastic outcome?
    • How do you detect the signature of these ecological processes? Through randomization procedures, or explained variation, or noise, or ...

Using this scheme, I think you could uniquely identify several terms and patterns in the literature, and solve some of the inconsistencies noted above:

  • drift is a specific type of end product of within-site, neutral, and stochastic, species interactions
  • dispersal can be 
    • neutral: all species similar in dispersal characteristics
    • or stochastic: see limiting dispersal sensu Winegardner et al. 2012 (see e.g. Matias et al. in press) , but with some deterministic patterns (see distance decay or isolation by distance in neutral metacommunities, Jeremy Fox's 2nd zombie idea mentioned above).
    • or deterministic: see efficient or high dispersal sensu Winegardner et al. 2012
  • neutral species interactions can
    • lead to dominance of one species over long time periods, and thus look deterministic
    • be deterministic (see Jeremy Fox's examples in his first zombie idea)
  • selection can be stochastic (see also Jeremy Fox's example in his first zombie idea)


  • It would be easy to fall in the trap outlined by Jeremy Fox, and treat these questions as "ends of a linear continuum". I do think they are ends of a continuum, but this does not have to linear (which is what I think the main danger is that Jeremy Fox wanted to point out).
  • I reserve the right to change my mind on this above classification, based on what I hope to be lots of discussion and exchange of ideas.

No comments:

Post a Comment