Anomaly detection uncovers emerging novelty in coral assemblages of the Great Barrier Reef

Abstract

Earth's biosphere is being altered by human actions, with all ecosystems, especially coral reefs, rapidly transforming in response to climate change and local stressors. As these changes manifest, ecology needs tools to measure change from past states, termed "ecological novelty", within complex multidimensional data such as taxa-abundance time series. Here we use an anomaly detection method new to ecology, Local Outlier Factor (LOF), to measure ecological novelty via neighborhood density in compositional space. We estimated novelty using LOF within 106 coral time series, spanning 29 years, from a long-term monitoring program on the Great Barrier Reef (GBR). We tested whether reef assemblages were becoming more novel over time and whether novelty correlated with common ecological indicators: coral cover, diversity and assemblage turnover. We found novelty correlated with low coral cover, low diversity and high turnover. Even after accounting for these correlations, we found reefs became more novel over time, with novelty across the GBR increasing by 15% between 1998 and 2021. We conclude that GBR coral assemblages are systematically shifting away from past compositional states, a detectable trend on a decadal scale that is increasing across all GBR regions. Our results indicate that anomaly detection methods can contribute to both ecological research and management by providing a new way to quantify ecological novelty under climate change and anthropogenic pressures. Novelty can be estimated from compositional data without needing to establish reference baselines, providing a unique, non-equilibrium perspective that can advance ecological research aims and incorporate into conservation and restoration decision-making frameworks across different taxa and ecosystems. For reef assemblages, the widespread and consistent increase in novelty we identified echo warnings of a global ecological crisis. While a novel species assemblage does not necessarily translate into altered reef functionality or loss of resilience, our findings point to a new for reef management to account for the increasing gap between past reef states and emerging states of the future.