Ecological memory at millennial time‐scales: the importance of data constraints, species longevity and niche features

Image credit: Blas M. Benito

Abstract

Ecological memory describes how antecedent conditions drive the dynamics of an ecological system. Palaeoecological records are paramount to understand ecological memory at millennial time‐scales, but the concept is widely neglected in the literature, and a formal approach is lacking. Here, we fill such a gap by introducing a quantitative framework for ecological memory in palaeoecology, and assessing how data constraints and taxa traits shape ecological memory patterns. We simulate the population dynamics and pollen abundance of 16 virtual taxa with different life and niche traits as a response to an environmental driver. The data is processed to mimic a realistic sediment deposition and sampled at increasing depth intervals. We quantify ecological memory with Random Forests, and assess how data properties and taxa traits shape ecological memory. We find that life‐span and niche features modulate the relative importance of the antecedent values of the driver and the pollen abundance over periods of 240 yr and longer. Additionally, we find that accumulation rate and decreasing pollen‐sampling resolution inflate the importance of antecedent pollen abundance. Our results suggest that: 1) ecological memory patterns are sensitive to varying accumulation rates. A better understanding on the numerical basis of this effect may enable the assimilation of ecological memory concepts and methods in palaeoecology; 2) incorporating niche theory and models is essential to better understand the nature of ecological memory patterns at millennial time‐scales. 3) Long‐lived generalist taxa are highly decoupled from the environmental signal. This finding has implications on how we interpret the abundance‐environment relationship of real taxa with similar traits, and how we use such knowledge to forecast their distribution or reconstruct past climate.

Publication
Ecography
Blas M. Benito
Blas M. Benito
Data Scientist and Team Lead

Related