DNA virus–host patterns in lake and marine environments over the last glacial cycle

TL;DR

This study uses ancient DNA from lake and marine sediments to explore how virus-host relationships have changed over thousands of years, revealing patterns in virus communities and their interactions with hosts.

Contribution

The study demonstrates the use of sedimentary ancient DNA to reconstruct ancient viral assemblages and their host interactions over glacial cycles.

Findings

Lake and marine virus communities differ, with lakes dominated by Caudoviricetes and marine sites including Algavirales.

Virus-host co-variation is stable in marine systems, especially for phytoplankton-infecting viruses.

Bacteriophages and hosts show antagonistic patterns, possibly linked to environmental shifts affecting viral cycles.

Abstract

Viruses are integral to population dynamics, biogeochemical cycling, and host evolution, making them essential for ecosystem function. We explore long-term virus–host interactions mainly within microbial ecosystems in lake and marine environments across the late Pleistocene and Holocene. Sedimentary ancient DNA (sedaDNA) from five Siberian lakes and three Subarctic/Antarctic marine cores were analysed to infer past DNA virus taxa from metagenomic sequences. Viruses accounted for 357 161 reads (0.089% of total mapped reads), distributed across 2084 unique viral taxa. Virus communities differ between lakes and marine sites, with lakes dominated by Caudoviricetes and marine environments featuring Caudoviricetes and Algavirales. Each time series shows compositional changes from the Pleistocene to the Holocene, supporting sedaDNA as a tool to reconstruct time-resolved ancient viral assemblages. Among the most abundant viruses, we identified 83 virus–host pairs documented in published literature, spanning bacterial, archaeal, and eukaryotic hosts, and assessed their associations based on co-occurrence correlations. Over millennia, virus–host co-variations are particularly stable in marine systems, especially for phytoplankton-infecting viruses. However, in the Bering Sea, we find a lack of virus–host correlation, likely because an Arctic Pelagibacter strain expanded after the Bering Strait opened, potentially due to absent viral infection, although database limitations prevent clear interpretation. Antagonistic patterns also appear between bacteriophages and hosts, possibly linked to shifts between lytic and lysogenic cycles in response to environmental changes. This study demonstrates that sedaDNA time-series can reveal ancient viral community structures and long-term ecological patterns, highlighting the value of ancient viromes in understanding ecosystem-specific responses to environmental change.

Figures