Orthologous genes are commonly found together on the same chromosome over vast evolutionary distances. This extensive physical gene linkage, known as macrosynteny, can be seen between bilaterian phyla as divergent as Chordata, Echinodermata, Mollusca, and Nemertea and likely reflects the importance of genome organization to gene regulatory landscapes. Here, we report a unique pattern of genome evolution in Bryozoa, an understudied phylum of colonial invertebrates. Using comparative genomics, including phylogenetic reconstruction and orthologous gene mapping, we reconstruct the chromosomal evolutionary history of five bryozoans. We infer the ancestral bryozoan genome organization and identify multiple ancient chromosome fusions followed by gene mixing, leading to the near-complete loss of bilaterian linkage groups. A second wave of rearrangements, including chromosome fission, occurred independently in two bryozoan classes, further shuffling bryozoan genomes. We also discover at least five derived chromosomal fusion events shared between bryozoans and brachiopods, supporting the traditional yet highly debated Lophophorata hypothesis. Finally, we show that chromosome fusion and fission processes led to the separation of bryozoan Hox clusters. Our findings demonstrate that the canonical bilaterian genome structure has been lost across an entire phylum, reveal that linkage group fission can occur very frequently in specific lineages, and provide a powerful source of phylogenetic information.

Chromosome-scale assemblies of Bugulina stolonifera (GenBank accession no. GCA_935421135), Cryptosula pallasiana (GCA_945261195), Membranipora membranacea (GCA_914767715), and Watersipora subatra (GCA_963576615) were sequenced as part of the Darwin Tree of Life project and retrieved from the NCBI Datasets. A chromosome-scale assembly of Cristatella mucedo was obtained from the NCBI Gene Expression Omnibus (GSE169088). None of the bryozoan assemblies used have publicly available gene models, so gene prediction was performed in-house. Repetitive elements were first annotated with RepeatModeler (v2.0.4) and masked using RepeatMasker (v4.1.5) in sensitive mode. Gene prediction and annotation were performed using the BRAKER3 pipeline (v3.0.3), incorporating hints from RNA sequencing (RNA-seq) data as aligned reads in bam format.