Prediction in ecology

Some digestions after reading

Houlahan, J. E., McKinney, S. T., Anderson, T. M., & McGill, B. J. (2017). The priority of prediction in ecological understanding. Oikos, 126(1), 1–7.

  • Prediction is not (yet) a central focus in ecology
  • If prediction were central in ecology, model selection would be done very differently
  • “Parsimony holds a hallowed place in model selection”
  • When data are limited, complex models are more likely to capture idiosyncrasies of the data rather than true underlying processes
  • However, this problem is entirely due to limitations the data impose on our ability to detect the underlying process, rather than on any inherent value of simple models
  • Ecological models can be improved by
    😊 identifying important variables
    🤔 elucidating functional relationships
    😩 improving parameter estimates
  • Without prediction, it is difficult to know if we understand more today than we did yesterday
  • Without temporal transferability, our understanding is ephemeral and temporary
  • Also hard to generalise
  • Many ecological hypotheses are quantitative
  • Many ecological studies are still null-hypothesis testing
  • Many truly continuous variables are still used categorically
  • Popper (paraphrased): “…demonstrating understanding is a slow and iterative process that relies on (in)correct and risky predictions.”
  • Then why aren’t we doing more predictions?
    • Funding sources and employers look for novelty, not confirmatory/reproduced science
    • Validating published models is unappealing
    • Pure science without immediate real-world implementations not worth getting super-accurate prediction