Modeling the pre-collapse community pattern of PETM warming event

Abstract

Abstract We modeled a foraminifera community collapse during Paleocene-Eocene Thermal Maximum with our temperature- and density-dependent model. Besides assessing our model, we investigated the pre-collapse pattern and the warning signals of the collapse with statistical indicators (relative species abundance, hierarchical cluster analysis, principal component analysis, total abundance, species richness, first-order autoregression, relative abundance of pre-collapse dominant species). We found that our model is appropriate for analyzing temperature- and density-dependent data patterns. Our study confirms that the total abundance and the pre-collapse dominant species are consistent temperature- and density-dependent indicators and give large-scale signals (trend changes) at the beginning of major environmental events and at community thresholds, respectively. They are earlier than first-order autoregression, which signals at the collapse zone boundaries. A great deficiency of our temperature- and density-dependent model is that it lacks strong small-scale signals (peaks and troughs). Small-scale signals are important because they can precede the large-scale signals of approaching collapses.