A genome-wide investigation of the effect of farming and human-mediated introduction on the ubiquitous seaweed *Undaria pinnatifida*

The study investigates how human activities—specifically domestication (cultivation) and biological introductions—have shaped genome architecture and diversity within a single species, Undaria pinnatifida. Domestication involves human-driven selection (intentional and unintentional) that can imprint characteristic genomic signatures, while biological introductions expose populations to novel environments and demographic processes (founder effects, admixture) that also leave genomic footprints. U. pinnatifida is uniquely suited for studying both processes concurrently: it has been cultivated at scale in its native Northeast Asia since the 1950s and has been independently introduced across multiple continents over the last ~50 years. The authors aim to generate a high-quality reference genome and use whole-genome resequencing of individuals from natural, cultivated, and introduced populations to disentangle the relative contributions of demography and selection to observed genomic differences.

The paper situates its work within growing evidence of human-driven evolutionary change affecting genomes of domesticated and introduced species. Prior genome-wide studies have illuminated domestication processes (e.g., in rice and sorghum) and, to a lesser extent, invasion genomics. However, no prior study has simultaneously examined independent domestication and introduction events within the same species. U. pinnatifida offers a rare case: centuries of exploitation followed by cultivation (Japan, Korea, China) and recent global spread to at least 14 countries. Previous work suggested distinct introduction histories between regions (e.g., France via aquaculture, New Zealand via shipping), temporal stability in some introduced populations, and high phenotypic plasticity in brown algae, all motivating a genome-wide, comparative approach across origins.

- Reference genome: DNA from a cultivated sporophyte (Wando, Korea; harvested Nov 2015) sequenced using PacBio long reads (~100×) and polished with Illumina paired-end reads (~32×). Assembly: 3,876 contigs; total 634 Mb; contig N50 406 kb. A genetic map anchored and ordered 72.7% (461 Mb; 1,325 contigs) into 30 linkage groups (chromosomes). Repeats masked using RepeatModeler/RepeatMasker; annotation combined homology- and transcript-based pipelines (8 cDNA libraries; proteins from seven species); 20,716 protein-coding genes predicted. - Comparative genomics: Synteny analyses with the Chinese U. pinnatifida assembly and Ectocarpus siliculosus; orthology across 19 stramenopile taxa; maximum-likelihood phylogeny and Dollo parsimony for gene family evolution. - Population sampling and resequencing: 41 individuals from 9 populations in three categories: natural (Goseong, Korea; Tongyeong, Korea), cultivated (Wando, Korea; harvested 2015 and 2017), and introduced (France: Thau, Roscoff; New Zealand: Lyall Bay, Wellington Harbour sampled in 1987 and 2017). Generated 853.77 Gb of cleaned paired-end data (avg 20.69 Gb/individual), mapped to the reference (avg depth 30.67×; avg coverage 94.77%). - Variant discovery: GATK pipeline; initial 25,414,685 variants; filtered to 7,253,541 high-quality variants (6,123,124 SNPs; 1,130,417 indels). Variant distribution assessed (majority intergenic; 3.07% exonic). - Population genomics analyses: PCA, maximum-likelihood phylogeny, and admixture (R package LEA) to infer structure; estimated genetic diversity (heterozygosity), linkage disequilibrium (r² and decay), and runs of homozygosity (ROH). One cultivated individual (Kr_Wando2015_4) showing strong admixture/introgression was excluded from downstream comparative analyses. - Selection scans: Calculated, in non-overlapping windows (50 kb), reduction of diversity, delta Tajima’s D, and FST; combined using DCMS (decorrelated composite of multiple signals) with window-level P values to identify putative selected regions. Enrichment analyses and exploratory transcriptomics for genes in selected windows.

- Genome assembly and content: Undaria pinnatifida Kr2015 genome is 634 Mb (largest reported for brown algae), with 30 chromosomes. 72.7% anchored to pseudochromosomes; 20,716 predicted genes (78.25% supported by transcript data). Repeats comprise 52.1% (330.3 Mb), including ≥19.14% transposable elements (~121 Mb). Repeat insertions are broadly homogeneous across chromosomes and associated with reduced gene density relative to Ectocarpus siliculosus; heterochromatin/euchromatin domains are not clearly segregated. Synteny with E. siliculosus is largely conserved despite deep divergence (128.9–220.2 Mya), with evidence for limited splitting/fusion events explaining chromosome number differences (28 vs 30). - Population structure: Whole-genome variants separate individuals by geography (PCA, phylogeny, admixture), with introduced populations forming two distinct clusters (France vs New Zealand). Temporal samples from Wellington Harbour (1987 vs 2017) are genetically similar, indicating stability over ~30–60 generations. - Genomic landscape by origin: • Natural populations: high genetic diversity (mean ≈ 0.0044), rapid LD decay (half-maximum at 3.95 kb), substantial ROH but comparatively shorter total lengths (mean total ROH ≈ 80.9 Mb; average ROH length ≈ 1.13 Mb). • Cultivated (Korea): unexpectedly similar to natural populations—high diversity (mean ≈ 0.0040), rapid LD decay (3.14 kb), and lowest homozygosity (low ROH coverage; average ROH length ≈ 0.96 Mb). Farming practices (large indoor fertilization pools mixing many sporangia and sources) promote outcrossing, maintain diversity, and reduce selfing. • Introduced: reduced diversity and elevated LD/ROH consistent with founder effects/selfing. - France: diversity mean ≈ 0.0015; LD half-maximum ≈ 27.33 kb; high homozygosity (mean total ROH ≈ 338.5 Mb; average ROH ≈ 1.79 Mb). Pattern consistent with sequential founder events (aquaculture vector: Thau to Brittany/Roscoff). - New Zealand: diversity mean ≈ 0.0022; LD half-maximum ≈ 10.47 kb; mean total ROH ≈ 201.2 Mb; average ROH ≈ 1.08 Mb. Closer to native-range profiles than France, consistent with repeated introductions via shipping and local spread (e.g., leisure boating). Temporal decrease in ROH length over 30 years suggests repeated introductions/admixture. - Selection analyses: • Cultivated vs natural: 224 windows under putative selection (P < 0.025 DCMS), encompassing 508 genes; enriched processes include glycolipid biosynthesis and cytokinesis, with some genes in carbohydrate pathways (alginate, mannitol, sulfate fucan). Developmental gene families with many copies (e.g., Cupin-like, C2H2 zinc-finger, imm-upregulated) were not overrepresented, suggesting phenotypic differences may be polygenic and regulatory. Exploratory transcriptomics hints at differential expression of selected genes. • New Zealand (Wellington 1987 vs 2017): no clear enrichment of specific biological functions; signals likely dominated by neutral demographic effects given founder events and limited time. Some selected genes relate to stress, homeostasis, and membrane functions.

The genomic differentiation among natural, cultivated, and introduced U. pinnatifida populations reflects combined demographic and selective processes driven by human activities. Introduced populations exhibit signatures of founder events (reduced diversity, slower LD decay, longer ROH), with regional differences matching known vectors and histories (stronger bottlenecks in France via aquaculture; more repeated introductions in New Zealand via shipping). In contrast, large-scale Korean cultivation maintains high diversity and low LD by promoting outcrossing through mixed sporangia fertilization in large pools, counteracting the species’ natural tendency for selfing. Selection scans in cultivated populations highlight metabolic and regulatory pathways potentially linked to cultivated traits (e.g., yield-related carbohydrate biosynthesis), though effects appear modest and dispersed, consistent with polygenic control and regulatory changes rather than strong sweeps at a few loci. Over shorter timescales in New Zealand, demographic history and similar environments likely mask or limit detectable adaptive divergence. Altogether, the findings show how domestication and various introduction pathways reshape genome architecture and diversity in predictable yet context-dependent ways.

A high-quality chromosome-level reference genome for U. pinnatifida and resequencing of 41 individuals across natural, cultivated, and introduced populations reveal clear genomic signatures of human influence. The work supports distinct introduction scenarios for France and New Zealand and shows that large-scale cultivation in Korea maintains high genetic diversity and low linkage disequilibrium, with potential implications for conservation (farmed stocks as reservoirs of evolutionary potential). The catalogue of putative selected regions provides hypotheses about traits under cultivation. Future research should integrate targeted sampling with phenotype and environment data, perform QTL mapping between breeding lines and natural individuals to resolve domestication genetics, conduct GWAS across native and introduced ranges to identify variants underlying phenotypic divergence, and assess connectivity between cultivated and natural populations to evaluate the role of escaped cultivars in generating genetic novelty.

- Demographic history (bottlenecks, reduced Ne) can produce genomic patterns that mimic selective sweeps, complicating inference; DCMS-detected regions should be interpreted cautiously. - Functional interpretations of selected windows risk overinterpretation without complementary phenotypic or expression evidence; the study’s exploratory transcriptomics is preliminary. - The time span in New Zealand (30–60 generations) may be insufficient for strong selection signals to emerge from standing variation; similar environments between Korea and New Zealand could further limit detectable divergence. - One cultivated individual showing admixture with natural populations was excluded, and sample sizes per population are modest, which may limit power to detect fine-scale substructure or selection. - Some assembly discrepancies relative to other genomes could reflect recombination or assembly artefacts, introducing uncertainty in synteny-based inferences.