Demystifying Cassiopea species identity in the Florida Keys: Cassiopea xamachana and Cassiopea andromeda coexist in shallow waters

The phylogeny of the Upside-Down Jellyfish (Cassiopea spp.) has been revised multiple times in its history. This is especially true in the Florida Keys, where much of the Cassiopea stock for research and aquarium trade in the United States are collected. In August 2021, we collected 55 Cassiopea medusae at eight shallow water sites throughout the Florida Keys and sequenced COI, 16S, and 28S genes. Mitochondrial genes demonstrate that the shallow waters in Florida are inhabited by both Cassiopea xamachana and a non-native Cassiopea andromeda lineage, identified in multispecies assemblages at least thrice. While C. xamachana were present at all sites, the C. andromeda-mitotype individuals were present at only a minority of sites. While we cannot confirm hybridization or lack thereof between the C. xamanchana and C. andromeda lineages, these previously unknown multispecies assemblages are a likely root cause for the confusing and disputed COI-based species identities of Cassiopea in the Florida Keys. This also serves as a cautionary note to all Cassiopea researchers to barcode their individuals regardless of the location in which they were collected.

Citation: Muffett K, Miglietta MP (2023) Demystifying Cassiopea species identity in the Florida Keys: Cassiopea xamachana and Cassiopea andromeda coexist in shallow waters. PLoS ONE 18(3): e0283441. https://doi.org/10.1371/journal.pone.0283441

Editor: Sergio N. Stampar, Sao Paulo State University Julio de Mesquita Filho Bauru Campus Faculty of Sciences: Universidade Estadual Paulista Julio de Mesquita Filho Faculdade de Ciencias Campus de Bauru, BRAZIL

Received: November 17, 2022; Accepted: March 9, 2023; Published: March 29, 2023

Copyright: © 2023 Muffett, Miglietta. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data and accession numbers are within the manuscript and its Supporting Information files.

Funding: This work was supported by a Texas Sea Grant to author KMM. This work was also supported by Texas A&M University Galveston. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Marine invertebrates have a wealth of cryptic lineages [1, 2]. Although our understanding of the diversity within cryptic marine taxa has grown, precise distributions of species are often challenging to assess with common sampling approaches. Scyphozoan phylogenies have frequently been sampled across large geographic areas but with few individuals at each site [3, 4]. In some pelagic systems, such as Aurelia or Chrysaora, this approach may be sufficient [5, 6]. However, in systems with a strong invasion potential and limited natural dispersion capabilities, like the genus Cassiopea, this shallow or single-site sampling may result in inadequate coverage to identify all species present.

As in other scyphozoan lineages, Cassiopea suffers from poor phylogenetic clarity. This is consequential because C. xamachana, and the genus Cassiopea more broadly, have become an emergent model system for research in symbiosis, behavior, and regeneration [7–11]. Recent work has demonstrated that morphological, symbiosis and ecological differences exist between Cassiopea species [12, 13], however, most non-taxonomic research is done on unspecified Cassiopea [14, 15]. This lack of clarity on the identity of the Cassiopea used in research, is problematic as it may lead to confounded, unreproducible, or less comparable experimental results.

High morphological heterogeneity within populations and apparent crypsis across species make Cassiopea difficult to identify [3, 8, 12]. Cassiopea species identification and species boundaries are even more blurred in the Florida Keys, one of the main collection grounds for Cassiopea research in the United States [12]. Cassiopea from Florida are often arbitrarily assigned one of three names: C. xamachana, C. andromeda or C. frondosa. C. xamachana, originally described in Jamaica by Bigelow 1892, theoretically represents the dominant morphotype in the Florida Keys and Caribbean [16]. C. xamachana’s description is distinct from C. andromeda (Forskål, 1775), a Red Sea native and an invasive documented in Brazil, the Mediterranean, and Hawai’i, though many of the characters on which that distinction was made are variable [3, 17, 18]. The separation of C. xamachana and C. andromeda is a point of contention, and the two species have been sometimes considered synonymous. When synonymized, C. xamachana was considered a population of introduced C. andromeda [8, 19]. This synonymization was originally a product of Gohar and Eisawy’s efforts to reduce the many described Cassiopea species into only three species—C. dieuphila (species name not revisited), C. andromeda and C. frondosa—based solely on one morphological character, the rhopaliar number [19]. Rhopaliar number is now recognized as having high intraspecific variability and limited interspecific variability and thus not suitable for species delimitation [3, 13]. Four decades later, genetic sampling of Cassiopea by Holland et al. (2004) across the globe upended this oversimplistic morphological assessment and significantly expanded the known species diversity in the genus but found no evidence of a “C. xamachana” lineage. The “C. xamachana” individuals (n = 4) collected from Bermuda and the Florida Keys by Holland et al. 2004 were not distinct from other global C. andromeda populations [3]. The C. frondosa species, a distinctive deeper water Cassiopea, and a more distant relative also found within the Keys, has remained a stable clade throughout various phylogenies and as such we did not sample these medusae [12] Since Holland et al. (2004) [3], some of the Cassiopea species invalidated by Gohar and Eisawy (1960) were validated and characterized morphologically, however the C. xamachana/C. andromeda distinction was not reevaluated [13], as such we have not met the need for an updated and more in-depth phylogeny of the shallow water Florida Keys Cassiopea.

Here, using a concerted sampling of eight shallow water sites in the Florida Keys, and using mitochondrial and nuclear genes, as well as morphological data, we address the presence and the phylogenetic status of C. xamachana and C. andromeda in the Florida Keys.

In August 2021, 55 Cassiopea were collected by hand in 8 sites along the length of the Florida Keys. All sites were from near-shore water under 2 m depth (S1 Table in S1 File). Each Cassiopea was photographed, diameter measured, and small tissue samples were preserved in ethanol (95%) at room temperature. Six small individuals were fully preserved in 95% EtOH, these samples were later used for limited morphological analysis (S1 Table in S1 File). Medusa density was estimated in area of collection by marking out one square meter with a tape measure and counting medusae within the area. Permitting for these specimens was waived by Florida Fish and Wildlife Conservation Commision.

DNA extractions from the 55 specimens were performed according to a salting out protocol (see full protocol in [20]). Mitochondrial Cytochrome c oxidate subunit I (COI) and 16S ribosomal RNA (16S) were amplified using Cassiopea-specific protocols and primers [13]. Nuclear 28S was amplified using the Cassiopea-specific primers and protocols in Daglio et al. 2017. All products were purified using ExoSap and sanger sequenced at Texas A&M-Corpus Christi Genomics Core (Corpus Christi, Texas) or GeneWiz Azenta (Plainsfield, New Jersey). Fifty-five 514 bp segments of COI, 34 575 bp segments of 16S and 18 822 bp segments of 28S were assembled on Geneious and checked by eye.

Newly produced COI and 16S sequences (55 and 38 sequences respectively) were aligned using Geneious 2022.2.2 (https://www.geneious.com) to COI and 16S sequences from the C. xamachana and C. andromeda published genomes. Provisional sequence identity was assigned through agreement >98% to either the C. xamachana genome [21] or the C. andromeda mitogenome [22]. All sequences were uploaded to Genbank (see S1 Table in S1 File for complete list of accession numbers).

A subset of 11 representative COI sequences from individuals collected in this study (see Table 1 and S1 Table in S1 File) were aligned with Cassiopea COI GenBank sequences from Holland et al. 2004, Morandini et al. 2016 [17], Daglio et al. 2017, Abboud et al. 2018, Gamero-Mora et al. 2022, Kayal et al. 2013, and Mastigias and Versuriga from Swift et al. 2016 and Sun et al. 2019 (see table) using MAFFT 7 (L-INS-i) [22–25]. The final COI dataset was composed of 89 sequences, 11 of which were produced here, all trimmed to 514bp. The remaining 78 GenBank sequences belonged to the following species: C. andromeda, C. xamachana, C. ornata, C. sp. 1, C. sp. 2, C. sp. 3, C. culionensis, C. mayeri, C. frondosa, and outgroups Mastigias papua and Versuriga anadyomene.

https://doi.org/10.1371/journal.pone.0283441.t001

16S sequences from 10 representative individuals across the Florida Keys (see Table 1) collected in this study were aligned with Cassiopea 16S GenBank sequences from Gamero-Mora 2022 and Daglio et al. 2017 and trimmed to 544 bp. The final 16S dataset was composed of 22 sequences, ten from C. xamachana and C. andromeda from this study (same individuals as COI minus one) and 12 GenBank sequences belonging to C. andromeda, C. xamachana, C. culionensis, C. mayeri, C. frondosa, C. ornata, and outgroup taxa M. papua and V. anadyomene. 16S sequences were aligned using MAFFT (E-INS-i).

Models for all COI codons and 16S dataset were chosen by AICc from MEGA X 11 [26] model tester (COI codon position 1: TN93+I, COI codon position 2: TN93+I, COI codon position 3: TN93+G+I, 16S: GTR+G+I). A dataset combining COI and 16S was run in IQtree 1.6 under an ML framework with support from 1000 aLRT (approximate likelihood ratio test) and 1000 non-parametric bootstraps. Bayesian support was generated in BEAST 1.8 [27] with 10 million steps. All phylogenetic trees were edited with Figtree v1.4.4 (http://tree.bio.ed.ac.uk/software/figtree/).

The 28S dataset consisted of 28 sequences—18 newly produced Cassiopea sequences, 8 Genbank sequences of C. frondosa, C. andromeda, C. ornata, and two sequences belong to outgroups M. papua and V. anadyomene (see S2 Table in S1 File for all accession numbers). All 822 to 846 nucleotide 28S sequences were aligned using MAFFT (G-INS-i). 28S phylogenetic trees were run on IQtree under ML framework with support from aBayes, 1000 SH-aLRT (approximate likelihood ratio test) and 1000 non-parametric bootstraps under the Mega X model tester suggested model (TN93 + G) and edited in Figtree v1.4.

All 55 COI sequences from this study and 29 additional GenBank COI sequences belonging to C. xamachana and C. andromeda were used to generate a COI haplotype network. 38 16S sequences from this study and 7 additional GenBank 16S sequences belonging to C. xamachana and C. andromeda were used to generate a 16S haplotype network. Two additional 16S sequences produced from water samples in the Florida Keys for eDNA purposes (from Ames et al. 2021) were included in the 16S haplotype map as these represented the only Cassiopea 16S data from the Florida Keys outside of the sequenced C. xamachana genome (see Table 1 for a complete list of sequences used). Haplotype networks were built using PopART v1.7 [28].

Cassiopea bell diameter were recorded for each medusa and compared using a one-way analysis of variance of each of these features between sites and between mitotypes performed in R v. 4.0.3.

Of the 55 Cassiopea COI sequences collected in the Florida Keys, 49 had a C. xamachana mitochondrial haplotype and six had a C. andromeda haplotype, as determined by agreement with published C. xamachana and C. andromeda mitogenomes [21, 22]. The COI divergence between the two species was approximately ~7%, as previously reported [13, 22], with no intraspecific divergence within the C. xamachana or C. andromeda collected (Fig 1).

Maximum likelihood COI + 16S tree with posterior probability, aLRT, and bootstrap supports. Individuals in grey were sequenced in this study. All accession numbers are for COI, see 16S accession numbers in Table 1. Asterisk (*) represents under .70 (PP) or 70 (aLRT and bootstrap) support. Species names are assigned in accordance with Gamero-Mora et al. 2022.

https://doi.org/10.1371/journal.pone.0283441.g001

The 16S and COI combined dataset, with sequences from GenBank, and rooted using sequences of M. papua and V. anadyomene, showed C. xamachana and C. andromeda as well supported sister taxa (posterior probability: 1, aLRT: 100%, bootstrap: 100%), and together as sister to a low supported clade that contains C. ornata from the Western Pacific, C. sp 1 from Australia, C. culionensis, C. sp 2 and C. mayeri from the Western Pacific. C. sp. 3 (from Papua New Guinea and Hawai’i) and C. frondosa are at the base of the tree (posterior probability: 1, aLRT: 99.5%, bootstrap: 100%) (Fig 1). Fourteen sampled individuals of undescribed Cassiopea from the Western Pacific (Abboud et al. 2018) remain external to known Cassiopea taxa. Tree topology is in agreement with previous phylogenies [3, 13].

The clade of C. xamachana, as defined by the published genome, is composed by sequences from the Florida Keys and Atlantic Panama, and one isolate from Palau (reported by Abboud et al. 2018). According to the analyzed dataset, except for the single Palau sequence, the C. xamachana mitotype is restricted to the West Atlantic (Fig 2A). COI sequences identified as “C. frondosa” and collected in Panama (Atlantic), also fell within the C. xamachana clade. A third common species, C. andromeda, includes sequences from specimens collected in Hawai’i, Mexico, Bermuda, Brazil, Florida and the Red Sea (Fig 2A & 2B).

(a) Global C. andromeda (black) and C. xamachana (white) distributions from sequences published in relevant scyphozoan or Cassiopea specific phylogenies from 2004–2022 [3, 6, 13, 17, 29]. Locations with both recorded as black-and-white. (b) C. andromeda (black) and C. xamachana (white) isolates from the Florida Keys from this study, pie chart indicates proportion of specimens that were C. xamachana and C. andromeda at each site.

https://doi.org/10.1371/journal.pone.0283441.g002

C. xamachana and C. andromeda had ~3.1% divergence in 16S sequences, lower than that calculated for COI. The six 16S sequences from collected C. andromeda showed no intraspecific diversity. The 32 C. xamachana 16S sequences showed low intraspecific diversity (d = 0.0006) (Fig 3B).

(a) COI haplotype network and (b) 16S haplotype network. One accession number that is consistent with the haplotype is displayed next to each group. Sequence MT709258* is a product of eDNA work and may not represent a genuine haplotype. Grey highlight connects all sequences from each species.

https://doi.org/10.1371/journal.pone.0283441.g003

When considering the COI dataset used for haplotype network building, C. andromeda showed 11 total haplotypes and C. xamachana a single haplotype (Fig 3A). C. xamachana showed three 16S haplotypes, and C. andromeda two, though only one was from specimens collected in this study (Fig 3B). The most common C. xamachana 16S haplotype was present in both the Keys and Panama. The second largest 16S haplotype was present only in upper Keys sites, in conjunction with a single individual from Big Pine Key showing a single nucleotide change. In the 16S haplotype network, there is one intermediate between C. xamachana and C. andromeda individuals, however this sequence was collected as part of an eDNA project and maynot represent a mitochondrial haplotype present in the system [30].

The nuclear 28S sequences (n = 18) from all sampled Cassiopea showed polymorphism, and tree topology was incongruent with mitochondrial genes sequenced (S1 Fig in S1 File). Specifically, sequences belonging to specimens collected in this study identified as C. xamachana and C. andromeda showed no differentiation. Genbank sequences of C. andromeda from Baja California (Genbank Acc. KY611005-7) included gaps not found in the sequences from the Keys and were closely related to two GenBank sequences of “C. frondosa” from Panama (likely C. xamachana) (Genbank Acc. KY611002-3). C. ornata from Palau and true C. frondosa from Key West (GenBank Acc. HM194838 and HM194872) were divergent from the C. xamachana/C. andromeda clade and each other.

We collected 55 samples in 8 localities along the Florida Keys (S1 Table in S1 File). With 49/55 individuals, C. xamachana was more frequently found in samples (Fig 2B). C. xamachana was found in all sites and C. andromeda in three of eight sampling sites. In the three sites that hosted the two species, C. xamachana was more abundant than C. andromeda (proportion of C. andromeda: 3/10 on Cudjoe Key, 1/10 on Key West, 2/5 on Key Largo). Two of the five monospecific C. xamachana sites were shallow lagoons (< 1m depth). The other two were low coverage tidal oceanic sites with low density and large individuals. The three sites of cohabitation were densely populated sites (>10 medusae/m2) with calm water but direct oceanic exposure. Both the Key Largo site and the Key West site were at marinas.

C. andromeda have been found on both the Atlantic and Gulf sides of Key Largo, as well as on the Gulf side of Cudjoe Key and Key West.

At each location, C. andromeda and C. xamachana presented the same color type and similar morphology (see Supplementary Materials), with no apparent character that could distinguish between the two species. Overall, individuals with the C. andromeda mitotype had somewhat smaller diameters (mean = 4.8 cm) than C. xamachana (mean = 7.8 cm) but not to a significant degree (ANOVA, F (2,47) = 3.69, p = 0.061). Two of the sites where C. andromeda were found (Cudjoe Key and Key Largo Ocean Bay Marina) had primarily small individuals, and site was the most important factor in size determination (ANOVA, F (1,47) = 5.08, p = 0.029). As all specimens were preserved in ethanol as opposed to formalin, no in-depth comparative morphological analyses were performed. A general inspection showed that C. andromeda (n = 2) and C. xamachana (n = 3), when devoid of symbionts, had different bell markings, however, with such a limited dataset, no conclusions could be drawn (S2 Fig in S1 File). All specimen photographs are included in supplementary materials.

With regard to coloration, Cassiopea collected from this work were blue, white, purple, pink, brown, and green. This is a well-known phenomenon in Cassiopea, and the exact dynamics of color are poorly understood. Color profiles were generally consistent within sites but were highly variable between sites (Fig 4). There was no consistent difference in color bell, oral arm or paddle coloration between species within each site at which the mitotypes were cooccurring.

“CX” for Cassiopea xamachana and “CA” for Cassiopea andromeda.

https://doi.org/10.1371/journal.pone.0283441.g004

C. xamachana, has been genetically undersampled in the Florida Keys relative to its use in research and laboratory work [11, 14, 21, 31]. A belief that C. xamachana was a C. andromeda subpopulation likely contributed to this lack of focus on dense geographic sampling [3, 6]. Moreover, in the last 20 years C. xamachana COI isolates have occasionally been misidentified as C. frondosa, a very distantly related Cassiopea, further adding to the taxonomic confusion. Our data support the notion that C. xamachana and C. andromeda mitotypes are now both found in the Florida Keys. Although mitochondrial DNA suggests these two species are reciprocally monophyletic clades, additional nuclear genes are necessary to confirm their monophyly. The Cassiopea collected here were not distinguishable in the field and they inhabited the same shallow water, sometimes coexisting side by side in the same location. 89% of our samples matched C. xamachana. C. xamachana was also sampled in 100% of sampling sites (8/8), while C. andromeda was sampled in 38% (3/8). We thus show that in our sampling effort, the C. xamachana mitotype is more abundant both in terms of number of jellyfish and locations where it is found.

Our results also confirm some findings of recent eDNA analyses conducted in the same area that recorded 16S residues of both species in the Florida Keys [30]. While we find the C. xamachana and C. andromeda mitotypes in our data, we fail to find the intermediate 16S signature found by Ames et al. 2021, a sequence that may have been an interspecific chimera. Despite having more representatives in this study, there was no diversity within the C. xamachana COI sequences and little diversity in their 16S profiles. This may represent a higher degree of continuity across the Florida Keys and Panama than expected. Further study of the exact boundaries that impact Cassiopea genetic populations is needed. While C. xamachana has been found in Brazil and Palau [6, 32], it does not yet have the non-native range demonstrated by the C. andromeda clade.

While there has been no demonstrated divergence in behavior or tolerance to environmental factor between C. xamachana and C. andromeda, there has also been no study formally comparing them. Our data show that wild-caught Cassiopea, even from single locations present clear hazards for comparative analysis if not properly identified [4]. This brings into foreground the inadvisability of treating results from Floridian samples and other locations as representatives of one clade without genetic evidence. In addition to genetic study, the Keys population would benefit from careful morphological and ecological analysis within mixed assemblage sites to parse whether these cooccurring populations have distinct diagnostic morphometrics or ecological features.

The mitochondrial markers analyzed in this work present evidence of historical genetic separation between C. xamachana and C. andromeda, the nuclear marker (28S), however, does not. As Cassiopea has very few published 28S sequences, some of which certainly suffer from the same issues of misidentification as the COI isolates, firm conclusions cannot be drawn as to the usefulness of 28S for species delimitation. Given the low mitochondrial divergence relative to a 10% benchmark [1], the 28S results may indicate introgression and hybridization between C. andromeda and C. xamachana. In-depth study of a larger array of nuclear markers is needed to parse the hybridization potential or degree of C. xamachana and C. andromeda within the Florida Keys.

As only 11% of individuals found in the Keys in this sampling effort were C. andromeda, only relatively dense sampling (either large collections from multiple locations or environmental sequencing) could identify the mix. In Brazil, multiple Cassiopea invasions have resulted in both C. xamachana and C. andromeda populations, but these were found in separate sampling efforts [17, 29, 32]. Locations with Cassiopea, especially those with a paucity of sequences, may present the sort of assemblages already identified here in the Florida Keys, Hawai’i, Brazil, the Philippines and Palau, and may require multiple rounds of sampling to parse [3, 6, 13]. Some Pacific Cassiopea populations remain unidentified (see the sequences without species identity in Fig 1), further hampering our understanding of invasion history within the genus. Greater sampling numbers are needed to further characterize species distributions within the Keys and elsewhere.

Finally, in constructing the phylogeny, we found instances of Cassiopea misidentification in GenBank. Three COI sequences identified as Cassiopea frondosa were instead C. xamachana. Additionally, five sequences identified as C. xamachana were instead C. andromeda, despite their originating text correctly identifying their species affinity [3]. This accounts for five of sixteen total identified C. andromeda sequences. This indicates a general problem with GenBank sequences and is a result of the taxonomic confusion that has surrounded Cassiopea species.

Cassiopea in the Florida Keys has long been defined as two species, the deeper-water, distinctive C. frondosa and the shallow-water C. xamachana [12]. In 1960 and again in 2004, C. xamachana was relegated to a junior synonym of C. andromeda [3, 19]. Using a phylogenetic approach, we show that C. andromeda and C. xamachana mitochondrial genotypes are both found in sympatry in the Florida Keys, showing no obvious morphological differences. We show that it is difficult to determine the population history of Cassiopea collected in shallow water in the Keys without proper molecular barcoding. This is relevant because a wealth of research has been performed with various Cassiopea without a proper assessment of the species it was conducted on. We also found evidence that Cassiopea research has suffered from frequent species misidentification. This paper calls for deeper sampling of jellyfish assemblages within Cassiopea and other highly cryptic scyphozoan genera. It also indicated that a proper species identification that involves molecular barcoding is essential for any work on Cassiopea, especially from Florida. This even more crucial as Cassiopea continues to successfully establish itself as an emerging model system for physiological studies and as a proxy for investigations on zooxanthellae-Cnidaria interaction. Caution should be exercised in generalizing result from published studies that assumed Cassiopea identity without explicitly investigating species identification with molecular tools.

https://doi.org/10.1371/journal.pone.0283441.s001

(PDF)

We thank the Florida Fish and Wildlife Conservation Commission for their help and guidance.