Abstract
Abstract
Background
Endogenous retroviruses (ERVs) are genomic fossils of past retroviral infections and a major, lineage-variable component of mammalian genomes. Bats (Chiroptera), as reservoirs and tolerant hosts of diverse viruses, are of particular interest. Yet comparative surveys of their ERVs inherit two rarely tested assumptions: that each assembly is the species it is labelled as, and that the raw output of de novo detection reflects genome content rather than assembly structure.
Results
We characterised LTR-retroelement content across 29 chromosome- and scaffold-level assemblies(with one contig-level genome that proves central to our final result) spanning 11 of the 21 bat families, using a framework built to test both assumptions. We first verified every assembly's taxonomic identity against three independent lines of evidence (NCBI metadata, Bat1K naming and mitochondrial cytb placement), which removed three misannotated assemblies, and validated the detection pipeline internally (4,292 structurally complete proviruses recovered the canonical 5′-gag-pro-pol-env-3′ domain order) and externally against a benchmark ERV of known distribution. LTR content varied roughly 11-fold among families, from a mean of 1.7% of the assembly in Pteropodidae to 18.9% in Rhinopomatidae. The low content of fruit bats (Pteropodidae) was concordant across five independent genomes and was not explained by assembly contiguity or gene-space completeness. Element abundance and coding integrity varied as two largely independent axes: the families richest in LTR sequence were not those richest in structurally intact proviruses. Among replicated families, fruit bats were distinctive in the proviruses they retained, being markedly enriched for envelope-bearing elements (54.9% versus 21.3% in other families pooled; odds ratio 4.49, 95% CI 3.39–5.93). Finally, an apparently young, expanded ERV signal in one Miniopterus genome was not reproduced by assembly-independent homology quantification or phylogenetic analysis, showing that the raw output of de novo LTR detection can decouple sharply from true genomic content and counterfeit a lineage-specific signal.
Conclusions
Our results provide a taxonomically verified comparative baseline for bat ERVs and, from the same analysis, a general safeguard for de novo transposable-element surveys: confirm apparent lineage-specific signals against assembly-independent evidence before interpreting them as biology.