Codweb: Whole-genome sequencing uncovers extensive reticulations fueling adaptation among Atlantic, Arctic, and Pacific gadids

Abstract

Introgressive hybridization creates networks of genetic relationships across species. Among marine fish of the Gadidae family, Pacific cod and walleye pollock are separate invasions of an Atlantic cod ancestor into the Pacific. Cods are ecological success stories, and their ecologies allow them to support the largest fisheries of the world. The enigmatic walleye pollock differs morphologically, behaviorally, and ecologically from its relatives, representing a niche shift. Here, we apply whole-genome sequencing to Pacific, Arctic, and Atlantic gadids and reveal extensive introgression among them with the ABBA-BABA test and pseudolikelihood phylogenetic network analysis. We propose that walleye pollock resulted from extensive adaptive introgression or homoploid hybrid speciation. The path of evolution of these taxa is more web than a tree. Their ability to invade and expand into new habitats and become ecologically successful may depend on genes acquired through adaptive introgression or hybrid speciation.

Fig. 1. Stylized topology of the classical phylogeny of Pacific, Arctic, and Atlantic gadids on a world map centered on the Arctic.

The Arctic cod and Polar cod clade is an outgroup. Pacific cod and walleye pollock are thought to be separate invasions of an ancestral Atlantic cod into the Pacific. Greenland cod is a reinvasion of Pacific cod into the Arctic and Atlantic at Greenland. Walleye pollock is a sister taxon of Atlantic cod. Walleye pollock differs morphologically [forked tail and missing chin barbel, two traits that define genera within the Gadinae as shared derived characters in a cladistic analysis (16)] and ecologically (semipelagic and schooling) from its closest relatives. Its biology represents a niche shift on the invasion [cf. (44)] of the Pacific. Atlantic cod and the Pacific invaders, Pacific cod and walleye pollock, are ecological success stories. They are dominant players in the ecosystem, and their ecologies translate into the remarkable ability of these native species in their native habitat to support the world’s largest commercial fisheries.

Fig. 2. Mitochondrial and nuclear tree topologies.

Topology of the neighbor-joining tree of average genetic distances of whole-genome mtDNA (A) and of whole nuclear genome (B) among cods. On the basis of whole-genome sequencing with 20 to 30× coverage of the entire genome and a lower (approximately 3×) coverage for Polar cod. The other taxa are Arctic cod, walleye pollock, Pacific cod, Greenland cod, and Atlantic cod from Sable Bank Sab, Trinity Bay Tri, Iceland Ice, and North Sea Nse. The whole-genome coverage translates into an average of 33× coverage for Polar cod mtDNA and up to 528× coverage for the other taxa. The Polar cod mtDNA has 99% identity with the Blue whiting mtDNA included (GenBank accession no. FR751401). Also included for comparison was a Polar cod whole-genome mtDNA from GenBank accession no. AM919429. Distances among taxa for each linkage groups were estimated using ngsdist (54). The whole nuclear genome distance is the average distance over all linkage groups weighted by the length of the linkage group in the number of nucleotides.

Fig. 3. D statistics with two SEs for the whole genome and fit to the classical phylogeny of these taxa.

On the basis of transversions with transitions removed. The green dot is for D not significantly different from zero, and red dotted lines mark the size of deviation of significant values. The dashed vertical line represents no introgression, a D = 0. For each test, the four taxa are arranged in a tree (((H₁, H₂)H₃)H₄), where H₄ is the Arctic cod as the outgroup and every other taxon is rotated into the position of potential introgressor (H₃, second from the right), as well as into the positions of potential recipients of introgression (H₁ and H₂, the two leftmost taxa of each line). A positive D implies that H₂ is closer to the potential introgressor (H₃) than H₁ is. The opposite is true for a negative D. The figure was made with admixturegraph (58).

Fig. 4. Best phylogenetic network and bootstrap support for linkage group LG01.

The inheritance probabilities of the major (γ in red) and minor edges (1 − γ in magenta) represent the proportion of genes that a hybrid inherits from its two parents [using SNaQ method of (32)]. The percentage bootstrap support (blue), from 200 bootstrap replicates with 10 runs per replicate, for the entire reticulation is given at the hybrid node, which represents the bootstrap support for the same hybrid node derived from the same sister clades. The number below an edge (in black) is the length of the edge in coalescent units, and the number above an edge (also in black) is the bootstrap support. The taxon names on the networks are Arctic cod, Polar cod, walleye pollock, Pacific cod, and Greenland cod, and locality names for Atlantic cod are Trinity Bay (Newfoundland), Sable Bank (Nova Scotia), coastal ecotype from North Sea2, coastal ecotype from North Sea8, coastal ecotype from Iceland Coastal3, a Pan I AB fish from Iceland termed Hybrid, Frontal (a composite of frontal ecotype from Iceland), and Coastal4 a composite of coastal ecotypes from Iceland (see Materials and Methods for the description of composites). Gadmor2 is the new alignment (49) of the reference genome. This labeling applies to all networks presented in figs. S2 and S3. LG01 is a linkage group known to harbor two adjacent inversions (23) that are associated with ecotypic variation within Atlantic cod. The major edge tree for LG01 is rooted on Arctic cod (Fig. 5). However, because of root mismatch, the network was rooted on an edge leading to Arctic cod.

Fig. 5. Major edge tree (edges with γ > 0.5) of the network for linkage group LG01 shown in Fig. 4.

The major edge tree is rooted using Arctic cod as outgroup. The major edge tree recovers the conventional phylogeny. The network in Fig. 4 showed a root mismatch and must be rooted on an edge leading to Arctic cod. The numbers below edges are branch lengths in coalescent units, and numbers above branch are bootstrap support of the node. The major edge tree was estimated with SNaQ (32).

Fig. 6. Phylogenetic network for linkage group LG07 with an apparent ancestor to descendant hybridization, the effects of ghosts of hybridization past.

Best phylogenetic network and bootstrap support for linkage groups LG07 phylonetwork with labeling as in Fig. 4. The LG07 linkage group harbors polymorphic inversions (24). The network was rooted on Arctic cod and showed an apparent hybridization of an ancestor to descendant, the effects of a ghost of hybridization past. Also evident is introgression between forms of Atlantic cod, such as from the Northwest Atlantic (Trinity Bay) to the Eastern Atlantic cod.