Deciphering trophic interactions in a mid-Cambrian assemblage

Abstract

The Cambrian Period (541-485 Mya) represents a major stage in the development of metazoan-dominated assemblages with complex community structure and species interactions. Exceptionally preserved fossil sites have allowed specimen-based identification of putative trophic interactions to which network analyses have been applied. However, network analyses of the fossil record suffer from incomplete and indirect data, time averaging that obscures species coexistence, and biases in preservation. Here, we present a novel high-resolution fossil dataset from the Raymond Quarry (RQ) member of the mid-Cambrian Burgess Shale (7549 specimens, 61 taxa, ~510 Mya) affording new perspectives on these challenging issues. Further, we formulate a new measure of preservation bias that aids identification of those assemblage subsets to which network analyses can be reliably applied. For sections with sufficiently low bias, abundance correlation network analyses predicted longitudinally consistent trophic and competitive interactions. Our correlation network analyses predicted previously postulated trophic interactions with 83.5% accuracy and demonstrated a shift from specialist interaction-dominated assemblages to ones dominated by generalist and competitive interactions. This approach provides a robust, taphonomically corrected framework to explore and predict in detail the existence and ecological character of putative interactions in fossil datasets, offering new windows on ancient food-webs. Significance StatementUnderstanding interactions in paleo-ecosystems has been a difficult task due to biases in collection and preservation of taxa, as well as low time resolution of data. In this work, we use network science tools and a fine scale dataset from the Cambrian period to explore: (i) preservation bias due to ecological/physical characteristics of taxa; (ii) evidence that the magnitude and sign of pairwise abundance correlations between two fossil taxa yields information concerning the ecological character about the interaction. All results in our work derive from using complex system approaches to analyze abundance data, without assuming any prior knowledge about species interactions - thereby providing a novel general framework to assess and explore fossil datasets.