Back

Label invariance: a guiding principle for ecological models

Gibbs, T.; Mazzarisi, O.; Fant, L.; An, R.; Grilli, J.; Barabas, G.; Song, C.

2025-07-19 ecology
10.1101/2025.07.15.664968 bioRxiv
Show abstract

Ecological models, though diverse in form, are strengthened when they obey guiding principles. We formalize and advocate for a foundational principle we call "label invariance", which says that a models dynamics must remain the same when identical individuals are arbitrarily grouped into distinct sub-populations. This principle is a necessary consequence of trait continuity--the observation that ecological interactions change continuously as organisms become more similar. Violation of label invariance often implies a hidden, intrinsic niche differentiation between species, which may obscure the mechanisms of biodiversity maintenance. We provide a general framework for constructing both deterministic and stochastic models that follow label invariance. We further demonstrate its utility as a complementary, non-statistical tool for empirical model selection. In sum, label invariance provides an important test for evaluating existing ecological models and a guide for developing new ones, promoting clarity in model assumptions from the outset.

Matching journals

The top 3 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.