Integrative studies have revealed cryptic radiations in several Caribbean lineages of heterobranch sea slugs, raising questions about the evolutionary mechanisms that promote speciation within the tropical Western Atlantic. Cyerce Bergh 1871 is a genus comprising 12 named species in the family Caliphyllidae, lacking the photosynthetic ability of other sacoglossans but noted for vibrant colors on the large cerata (dorsal leaf-like appendages) that characterize many species. Two species are widely reported from the Caribbean: C. cristallina (Trinchese, 1881) and C. antillensis Engel, 1927. Here, we present an integrative assessment of diversity in Caribbean Cyerce. Four methods of molecular species delimitation supported seven species in samples from the Caribbean and adjacent subtropical Western Atlantic. Six delimited species formed a monophyletic lineage in phylogenetic analyses but were >9% divergent at the barcoding COI locus and could be differentiated using ecological, reproductive and/or morphological traits. We redescribe C. antillensis Engel, 1927, a senior synonym for the poorly known C. habanensis Ortea & Templado 1989, and describe five new species. Evolutionary shifts in algal host use, penial armature and larval life history may have synergistically promoted the rapid divergence of endemic species with restricted distributions in this radiation, substantially increasing global diversity of the genus.

Specimen collections

Samples comprising 92 specimens were collected from the Caribbean and adjacent subtropical western Atlantic waters. Most specimens were collected by the authors with permission from the State of Florida (FL) under Special Activity Licenses (07SR-1034, SAL-11-1034-SR, SAL-19-1034-SR), or from the host country as needed, including special permission granted to authors by the Autoridad de los Recursos Acuaticos de Panama (ARAP 27-2015 and ANAM SE/A-64-15), and from the Dept. of Fisheries of the Bahamas to Joseph R. Pawlik and participants of research cruises in 2003, 2004, 2007 and 2010. Exceptions were specimens collected from Lauderdale by the Sea, FL (by A. Dimitris); Summerland Key, FL (by Y. Gryzmbowski); and Dominica (by D. Willette). Some Caribbean specimens were identified as C. cf. cristallina based on their distinctive external morphology, but no Mediterranean samples were available for comparison; all other Caribbean samples were provisionally identified as C. antillensis in accordance with recent literature (Valdés et al., 2006; Goodheart et al., 2016).

Specimens were preserved in 95–100% ethanol for molecular analysis. The type material of newly described species was deposited at the Natural History Museum of Los Angeles County (LACM) and the California Academy of Sciences (CASIZ). Non-type vouchers were deposited at the museums listed above as well as at the research collections of Cal State L.A. and the California State Polytechnic University Pomona (CPIC).

Sequences from the National Center for Bioinformatics (NCBI) database were included for 13 specimens representing nine species of nominal Costasiella, Sohgenia and Cyerce spp., based on prior work analyzing relationships within Limapontioidea (Krug et al., 2015, 2018; Filho et al., 2019). Additional specimens of Cyerce were included in phylogenetic analyses to test whether Caribbean samples identified as C. antillensis formed a monophyletic group consistent with an endemic radiation. Samples of two species of the related genus Sohgenia  were also included, given prior studies that recovered Sohgenia  nested within Cyerce (Krug et al., 2015). Molecular studies have consistently recovered Cyerce outside family Caliphyllidae contrary to traditional morphological classification schemes, including phylogenetic studies using a few markers and many taxa (Krug et al., 2015; Christa et al., 2015), and phylogenomic analyses of 1,160 genes for representative taxa (Krug et al., 2022). Given that no sister group has been identified for Cyerce using molecular data, we included an exemplar of Costasiella as the outgroup in all phylogenetic analyses based on phylogenetic results that recovered both genera as early splitting lineages at the base of Limapontioidea (Krug et al., 2022).

ecological and life-history data collection

Specimens of Cyerce spp. were collected from four sites within FL and a further 14 Caribbean sites. Each potential udotacean algal host was separately collected (0.5˗1 kg wet weight) at every sampling site; algal stipes were removed from the substrate, placed in a bag underwater and finally transferred to an aerated bin of seawater, where slugs were removed. Specimens were provisionally identified under a stereomicroscope and where possible, isolated and allowed to lay eggs. To determine if slugs could successfully change host algae, specimens from Tarpon Springs, FL and St Petersburg, FL were collected from Penicillus lamourouxii or Udotea looensis and either maintained on the field host (N=6 slugs per alga) or switched to the alternative host (N=6 slugs per treatment). Slugs were monitored daily for feeding and survival over 4 weeks.

Specimen collections

DNA extraction, amplification and sequencing

Specimen collections

EZNA Mollusc DNA kits (Omega Bio-Tek) or a hot Chelex^® protocol were used to extract DNA from one ceras or ~2 mg foot tissue, following described protocols (Krug et al., 2015; Medrano et al., 2019). Portions of two mitochondrial loci were then amplied by PCR: the cytochrome c oxidase I (COI) gene, using primers LCO1490 and HCO2198 (Folmer et al., 1994) with custom modifications as described (Ellingson & Krug, 2016); and the large ribosomal subunit rRNA (16S) gene, using primers 16Sar-L and 16Sbr-H (Palumbi, 1996). The nuclear Histone H3 gene was also amplified using universal primers (Colgan et al., 1998). Purified products were cycle-sequenced using PCR primers at Retrogen (San Diego, CA) or Source BioScience Sanger Sequencing Laboratory (Santa Fe Springs, CA). Chromatograms were edited in Geneious Pro R8 (Kearse et al., 2012) and alignments performed using default parameters for Multiple Sequence Comparison by Log-Expectation (MUSCLE) (Edgar 2004). Published secondary structure models guided refinements of rRNA gene alignments (Krug et al., 2015). Regions of ambiguous alignment in hypervariable loop regions of rRNA genes were identified using the least stringent criteria in GBLOCKS (Castresana, 2000) and removed prior to analyses.

All data files can be viewed with a standard text editor.

"DNA alignments" are text files in Phylip format used as inputs by the program RaxML, as specified in the README file.

see: https://doi.org/10.1093/bioinformatics/btu033

"Arlequin input files" are input files for the program Arlequin, as specified in the README file.

see: http://cmpg.unibe.ch/software/arlequin35/

"Popart input files" are input files for the program Popart, as specified in the README file.

see: http://popart.otago.ac.nz