Cryptic mtDNA Diversity of Diopatra cuprea (Onuphidae, Annelida) in the Northwestern Atlantic Ocean

Simple Summary Molecular tools continue to reveal cryptic biodiversity within common and ecologically important species. The decorator worm Diopatra cuprea is an ecosystem engineer of intertidal beds of high-salinity estuaries of the Atlantic and Gulf of Mexico shorelines. Here, we sequenced mitochondrial cytochrome oxidase I (COI) in D. cuprea populations and discover evidence for several deep mitochondrial lineages, suggesting the presence of cryptic diversity. Abstract Marine annelid taxonomy is experiencing a period of rapid revision, with many previously “cosmopolitan” species being split into species with more limited geographic ranges. This is exemplified by the Diopatra genus, which has recently witnessed dozens of new species descriptions rooted in genetic analyses. In the northwestern Atlantic, the name D. cuprea (Bosc 1802) has been applied to populations from Cape Cod through the Gulf of Mexico, Central America, and Brazil. Here, we sequenced mitochondrial cytochrome oxidase I (COI) in D. cuprea populations from the Gulf of Mexico to Massachusetts. We find evidence for several deep mitochondrial lineages, suggesting that cryptic diversity is present in the D. cuprea complex from this coastline.


Introduction
Annelid taxonomy is experiencing a period of rapid revision at every level, from individual species to the entire phylum [1][2][3][4]. Historically, taxonomists of the 19th and early 20th centuries embraced the 'cosmopolitan species concept', believing that marine annelid faunas were dominated by a relatively small number of very widespread species. This view began to crumble in the face of increasingly detailed morphological work beginning in the 1970s, and has now fully collapsed under evidence from modern phylogenetic methods (reviewed by [3]). While we now understand that annelid diversity is vastly greater than was once thought, the work of fully understanding that diversity remains very much a work in progress.
D. cuprea occurs intertidally on the US east coast from Duxbury MA southward (the southern limit is not well established) [21,22]). D. cuprea provides one of the best-described examples of ecosystem engineering in estuarine sediments; the worm builds robust tubes that descend 1 m or more vertically into the sediment, with an above-sediment tube-cap emerging 2-5 cm above the sediment surface. This tube-cap is characteristically decorated with fragments of drift algae, shell, and other debris. The tube stabilizes sediment, alters water flow across the sediment surface, and physically excludes epibenthic predators. Together, these physical effects increase the diversity and abundance of nearby infauna, and of the epifauna living on the tube itself (reviewed by [22]). By attaching to algae, the worm creates an algal canopy in habitats where attached algae would otherwise be rare [23], and this behavior is facilitating the invasive red alga Gracilaria vermiculophylla throughout most of the US east coast [24][25][26][27]. D. cuprea's emergent, decorated tube is typical of the genus, and our model for understanding the genus' ecological role worldwide is largely informed by studies of D. cuprea [22].
The D. cuprea's type locality is in South Carolina, but the name has been applied to Diopatra from Cape Cod through Brazil [10,20]. Diopatra from Brazil have recently been redescribed with four new species, none of which were D. cuprea [20], consistent with a non-cosmopolitan range for this species. The possible existence of cryptic Diopatra diversity in the northwestern Atlantic remains currently unexplored. D. cuprea is known to exhibit a curious latitudinal gradient in its tube-decorating behavior, with worms in Florida decorating far less than worms between Cape Cod and Georgia, even when offered the most commonly utilized algae in controlled conditions [21]. This biogeographic pattern in behavior raises the hypothesis that there may be hidden population-level genetic diversity in this region. Here, we sequenced a portion of the mitochondrial cytochrome oxidase I (COI) gene from D. cuprea populations from the Gulf of Mexico to Massachusetts in order to assess population genetic structure and any cryptic diversity.

Materials and Methods
Per site, 10-40 samples were collected by hand, placed into 95% ethanol and returned to the College of Charleston Marine Laboratory in Charleston SC in approximately 2009-2012 (Table 1). Samples were identified using morphological traits previously described [6,28], and extracted using a Nucleospin Tissue and Blood Kit (Maschery-Nagel), with the manufacturer's recommended protocol. A portion of the COI gene was then PCR amplified using the protocol in [29]. These were cleaned with an EXO-SAP-IT protocol and sent for Sanger sequencing with these same primers at a private company. The sequences were trimmed and quality-checked with 4Peaks software v1.8 (nucleobytes.com) and saved as a fasta-formatted file. This yielded a dataset of 539 nucleotides across 153 individuals (GenBank Accession Numbers OQ700009-OQ700161). We used ape 5.0 [30] and a custom code in R v4.2.3 [31] to estimate the genetic distance between individuals, and assigned labels to all 15 unique haplotypes (Hap 1-15). To assess the placement of northwestern Atlantic D. cuprea into a wider phylogeographic context, we used MUSCLE v3.8.31 [32] with default parameters embedded in SeaView 5.0.5 [33] to align our dataset and sequences from published sources ( Table 2). We ran PhyML 20120412 [34] as implemented in SeaView to estimate the phylogenetic relationships with default parameters (GTR + 4 rate classes as a model of evolution; 100 bootstrap replicates). We also generated a Bayesian analysis with MrBayes 3.2.7a [35] with 10,000 generations, a sampling frequency of 100, and a burn-in period of 250 generations. We used R::ape to visualize the Bayesian tree. We also used strataG 1.0 [36] to assess nucleotide diversity, pairwise Weir and Cockerham's F ST and the proportion of unique haplotypes within populations. Finally, we used linear regressions to assess the relationship between latitude and nucleotide diversity, and the latitude and proportion of unique haplotypes.
All the COI haplotypes sequenced from morphologically similar Diopatra cuprea from the east coast of the United States (Gulf of Mexico to Massachussetts) clustered into a single monophyletic clade, although the clade had minimal statistical support (i.e., ML bootstrap and Bayesian posterior probabilities were less than 90%). None of these samples clustered within Clades 1-3 from the work of Hektoen et al., 2022 [12]. Instead, samples were within a lineage labeled as Clade 4-5 of Hektoen et al., 2022 ([12]; Figure 1).
Biology 2023, 12, x FOR PEER REVIEW Figure 1. Cryptic Diopatra cuprea genetic diversity. The Bayesian phylogeny was based on 55 cytochrome c oxidase I. Numbers on edges indicate posterior probability, while asterisks in 90% or greater maximum likelihood bootstrap support (100 replicates). Clade designation numbers after taxa) were defined using the work of [12]. An approximate 5% divergence b haplotypes is indicated by the segment.
Within the northwestern Atlantic D. cuprea samples, we detected five divergen ages that differed between 14.6 and 20.5% (Table 3). We labeled these D. cuprea based on their relationship to each other and their geographic distribution (Figure clade frequencies in Table 1). A1 and A2 are a monophyletic group found mainly in ida; A1 dominated the St. Thersa Beach population in the Gulf of Mexico (81% of types), while A2 dominated the southeastern coasts of Florida (100% of haplotype Fort Pierce and Chicken Island). Clade B dominates the mid-Atlantic states, from eastern Florida (Amelia Island) to Virginia, and rarely occurred in other locations. C was only found in North Carolina and Clade D dominated Massachusetts popula and rarely occurred in mid-Atlantic states.  Clade designations (i.e., numbers after taxa) were defined using the work of [12]. An approximate 5% divergence between haplotypes is indicated by the segment.
Within the northwestern Atlantic D. cuprea samples, we detected five divergent lineages that differed between 14.6 and 20.5% (Table 3). We labeled these D. cuprea clades based on their relationship to each other and their geographic distribution (Figure 2; see clade frequencies in Table 1). A1 and A2 are a monophyletic group found mainly in Florida; A1 dominated the St. Thersa Beach population in the Gulf of Mexico (81% of haplotypes), while A2 dominated the southeastern coasts of Florida (100% of haplotypes from Fort Pierce and Chicken Island). Clade B dominates the mid-Atlantic states, from northeastern Florida (Amelia Island) to Virginia, and rarely occurred in other locations. Clade C was only found in North Carolina and Clade D dominated Massachusetts populations, and rarely occurred in mid-Atlantic states.  The overall FST across all populations was 0.728 (p < 0.001) and the overall FST across the four regions (FL-Gulf; FL-Atlantic; mid-Atlantic; MA) was 0.826 (p < 0.001). Tables 4  and 5 indicate the pairwise FST among the populations and regions, respectively.

Discussion
In morphological terms, all the individuals collected from the Gulf of Mexico through Massachusetts are consistent with the currently described Diopatra cuprea (Bosc 1802). While we found no statistical evidence for a monophyletic clade of D. cuprea in the northwestern Atlantic, all five COI clades were clustered away from the recently defined Clades 1, 2 and 3 in the work of Hektoen et al., 2022 [12]. This paper utilized two mitochondrial (COI and 16s rDNA) and one nuclear locus (28S rDNA) to assess phylogenetic relationships; it is likely that the application of these two other loci would allow greater resolution of the D. cuprea lineages and their exact placement within the wider phylogeny presented in the work of Hektoen et al., 2022 ([12]; see [11]). We also note that the type locality of D. cuprea was South Carolina, which is strongly dominated by Clade B [19].
The phylogeography of the widespread Diopatra cuprea mirrors that of several nearshore and estuarine species in North America. The monophyletic split between the A1 and A2 versus the others (B, C and D) likely reflects the Gulf of Mexico vs. eastern seaboard, which now have a secondary contact zone in eastern Florida near Cape Canaveral [40][41][42], and was driven by the separation of these water bodies during the Pliocene or Pleistocene. The split between Massachusetts Clade D from the mid-Atlantic clades (B and C) likely mirrors historical separation during the Pleistocene [43] and has a secondary contact zone near the New Jersey/New York border [44]. There are few species whose geographic range spans the Gulf of Mexico through New England; of these, one that displays similar phylogenetic breaks is the monocot Spartina alterniflora [45], which lives in close proximity to the high-salinity marshes that also constitute D. cuprea habitats.
The populations were not reciprocally monophyletic. Each population was dominated by one or two COI clades, but most populations harbored 1-2 other clades at low frequency. As examples, the Gulf of Mexico population has a low frequency of the mid-Atlantic B clade, while a North Carolina population has a low frequency of the Gulf of Mexico A1 clade and Massachusetts D clade. These may reflect low-frequency dispersal events between the regions, incomplete lineage sorting, or a mix of both.
Low-frequency dispersal events could occur naturally or via human-assisted transport. Unlike Diopatra in Europe, D. cuprea is not harvested for bait in the US (most likely because it is more difficult to obtain than D. neapolitana [22]). Transport as a bait worm is, therefore, unlikely. Diopatra's larval period is quite short, making transport in ballast unlikely [46]. If human-assisted transport plays any role in these distributions, it is most likely via the transport of newly settled juveniles in mud associated with bivalve aquaculture. This is also the mechanism proposed for the D. biscayensis disjunct distribution in France [46]. Galaska et al. [46] found that individual D. biscayensis in the disjunct population were disproportionately found near the ropes used for mussel culture. None of our sampling sites were (to our knowledge) especially close to commercial aquaculture sites, but bivalve aquaculture certainly does occur throughout the eastern United States. We can tease apart the relative importance of lineage sorting and more recent dispersal by assessing the phylogeographic patterns in the nuclear genome (e.g., [38]). Moreover, future sampling efforts could also evaluate whether rare clades are associated with aquaculture sites.
Overall, our results provide preliminary evidence that cryptic diversity is indeed present in D. cuprea populations of the northwestern Atlantic. This is consistent with the explorations of other Diopatra populations in other parts of the world.