
Song and Levine (2025) introduce a "covariance criteria" that ties the covariance of gain and loss processes to observed population abundance, providing a model‑structure test for ecological time‑series. Borrowed from queueing theory and later used in biophysics, the method works under stationary conditions without needing exact parameter values. Applied to classic debates—functional response, rapid evolution, and higher‑order interactions—the test rejected most competing models but surprisingly upheld the simple Lotka‑Volterra predator‑prey formulation. The authors also released an R package and highlighted methodological limits to encourage responsible use.

The post argues that truly fundamental ecological research seldom has direct, immediate applications, but a few notable exceptions exist. It highlights trophic cascade studies in lakes as a basis for algal‑bloom management, simple stochastic population‑growth models that shape endangered‑species legislation,...