Environmental DNA identifies coastal plant community shift 1,000 years ago in Torrens Island, South Australia

Coastal marine plants, including seagrasses, mangroves, and tidal marshes, are vital for marine life, carbon storage, coastline stabilization, and food webs. However, these habitats are declining due to anthropogenic changes, with the Industrial Revolution marking a turning point for rapid degradation. Understanding historical changes and identifying ecosystem tipping points is crucial for contextualizing present-day changes and achieving successful restoration. Restoration efforts often suffer from the "shifting baseline syndrome," where the perceived natural state is limited to recent observations. Traditional methods like archival observations, isotope analysis, fossil analysis, and pollen analysis provide some historical data but are limited by biases and preservation issues, particularly for seagrasses. The recovery and analysis of eDNA from soil cores offer a promising new tool for reconstructing historical coastal plant communities, particularly when combined with environmental context data.

The existing literature highlights the importance of coastal plant communities and the challenges in understanding their long-term dynamics. Studies have shown the significant ecological and societal value of these ecosystems (Barbier et al., 2011), and the accelerating rate of human impact on the world’s oceans (Halpern et al., 2019). The "shifting baseline syndrome" (Pauly, 1995; McClenachan et al., 2012; Alleway et al., 2023) is a major obstacle to effective conservation and restoration efforts. Traditional palaeoecological methods using pollen, macrofossils (Pedersen et al., 2013; Parducci et al., 2019), and isotopes (López-Sáez et al., 2009; Becker et al., 2020) offer insights into long-term changes, but have limitations in terms of preservation bias and taxonomic resolution. The emerging field of eDNA analysis from sediment cores (Nguyen et al., 2023; Barrenechea Angeles et al., 2023) is showing great potential to overcome some of these limitations. However, methodological challenges remain regarding eDNA degradation and interpretation in soil samples.

This study used four soil cores collected from a grey mangrove-dominated wetland on Torrens Island, South Australia. Radiocarbon (¹⁴C) dating was used to establish a timescale for environmental change. Targeted eDNA capture, focusing on multiple chloroplast gene regions, was employed to reconstruct plant community composition through time, unlike traditional metabarcoding approaches. This targeted method improved the ability to detect coastal plants, even those with degraded DNA. Chemical proxies, including dry bulk density, organic carbon content (%Corg), stable organic carbon isotopes (δ¹³C), and X-ray fluorescence (XRF) elemental analysis (Ca/Fe, Sr, Moly ratio, Br/Cl, Ti/Ca), were used to characterise environmental conditions. Principal Component Analysis (PCA) was applied to reduce dimensionality and identify significant change points in the chemical and eDNA data. Change point analyses were performed on PC1 scores to determine the timing and nature of environmental shifts. Bioinformatic analysis processed eDNA sequences, using PALEOMIX for read processing and trimming, BWA-MEM for alignment, and CD-HIT-EST for clustering to provide taxonomic assignments. Generalized additive models (GAMs) were applied to analyze changes in relative abundances of plant communities with depth and to interpret eDNA signals in context of chemical data. Strict laboratory protocols were followed to minimize contamination during eDNA extraction and processing.

Radiocarbon dating revealed a hiatus in the sediment profile around 1000 Cal Year BP. Pre-hiatus age-depth models were generated for Cores 1, 2, and 4 up to ~1000 Cal Year BP. eDNA analysis revealed a shift in plant community composition around 1000 Cal Year BP. Before this shift, the dominant community comprised subtidal seagrasses, notably *Zostera nigricaulis*, *Ruppia maritima*, *Posidonia australis*, and *Amphibolis antarctica*. After the shift, the community was dominated by the intertidal grey mangrove, *Avicennia marina*. Chemical analysis (PCA of δ¹³C, %Corg, dry bulk density and XRF) supported this shift, showing higher δ¹³C, Sr, and Ca/Fe (indicators of marine environments) before 1000 Cal Year BP, shifting to a higher Ti/Ca (terrestrial influence) and lower Sr, and Ca/Fe afterwards. The higher organic matter in younger sediments supports the establishment of an intertidal environment. The change point analysis confirmed the shift at approximately 63cm (cores 1-4) and 60cm (core 3) aligning with the ~1000 Cal Year BP.

The combined eDNA and chemical data strongly support an ecosystem shift from a subtidal seagrass environment to an intertidal mangrove habitat around 1000 years ago. The presence of mangrove eDNA in older sediments is likely an artifact of modern mangrove root penetration. While mangrove roots stabilize the sediment contributing to coherent stratigraphy, they could introduce a false positive signal of mangrove presence through time. The chemical data—higher δ¹³C, Sr, and Ca/Fe in older sediments— confirm a previous higher sea level supporting the seagrass community. The subsequent decrease in sea level, potentially due to tectonic uplift, or gradual climate-driven sea level fall, exposed the sediments enabling mangrove colonization. This ecosystem shift illustrates a longer timescale of change, significantly impacting our understanding of natural coastal ecosystem resilience. The methodological approach used in this study has potential application in other environments affected by sea-level change or climate warming, allowing for more robust reconstructions of long-term ecosystem dynamics.

This study successfully used a multi-proxy approach combining targeted eDNA capture and soil chemical analyses to reconstruct high-resolution historical changes in a coastal plant community over 4000 years. The findings revealed a significant shift from a subtidal seagrass to an intertidal mangrove ecosystem around 1000 years ago, driven by likely sea-level fall. This demonstrates the power of integrating eDNA analysis with established methodologies for understanding long-term ecosystem dynamics. Future research could apply this approach to other coastal ecosystems and investigate drivers of past vegetation change in more detail.

The study focuses on a single site on Torrens Island, limiting the generalizability of findings. The presence of a hiatus in the sediment profile hindered the generation of complete age-depth models. The eDNA approach may have detected some signals from adjacent habitats due to the sensitivity of the targeted capture approach. While careful laboratory protocols minimized contamination, the possibility of some undetected contamination cannot be entirely excluded. The interpretation of the chemical proxies relies on existing knowledge and assumptions about their relationships to environmental variables. Further research validating these assumptions and improving the interpretation of overlapping signals from multiple sources in sediments may strengthen conclusions.