CLYTIIDAE, Cockerell, 1911

FAMILY CLYTIIDAE

When all species of Clytia are compared, the PCA shows that most of the variation (PC1) is related to the presence of erect colonies, and the number, length, diameter and perisarc thickness of the internodes (NIS, LIS, DIS, PIS) separate Clytia linearis (Thorneley, 1900) and some specimens of C. elsaeoswaldae Stechow, 1914 , C. cf. gracilis ( Sars, 1850) sp.1 and C. cf. hemisphaerica (Linnaeus, 1767) sp.1 from the remaining Clytiidae ( Fig. 5A View Figure 5 ). However, when data for species of C. cf. gracilis and measurements related to internodes are excluded from the analysis, further morphological patterns among species with erect colonies become evident ( Fig. 5C, D View Figure 5 ). Clytia linearis is distinguished by its longer hydrothecae and cusps (LH, HCMax, HCmin; Fig. 5C, D View Figure 5 ), although the range of variation of cusp height overlaps with those of other species ( Fig. 6A, B View Figure 6 ). Likewise, C. elsaeoswaldae is separated by the larger hydrothecal diameter (DHMa, DHMe, DHB, DBC; Fig. 5A, C, D View Figure 5 ), but this character is more informative when compared to species of C. cf. gracilis and C. cf. hemisphaerica , with which it shows less overlap ( Fig. 6C View Figure 6 ). Further comparisons show that C. elsaeoswaldae has a thicker diaphragm on average than C. linearis , as well as specimens of C. cf. gracilis and C. cf. hemisphaerica ( Fig. 6D View Figure 6 ). However, morphological variation is high and certainly attenuates these differences, leading to large overlaps among species.

The second direction accounting for most variation (PC2; Fig. 5A, B View Figure 5 ) is related to perisarc thickness (PHMa, PHMe, PHB, PPMe) and length:diameter ratio of the hydrotheca (HRatio). It sets apart Clytia sp.2 and Clytia noliformis (McCrady, 1859) because of their thicker perisarc, and Clytia sp.1 , C. cf. gracilis sp.5 and C. paulensis because of their more cylindrical hydrothecae ( Figs 5A View Figure 5 , 6E, F View Figure 6 ). Although evident when directly compared among these species, differences in HRatio are not evident in all PCAs, probably because of the slight variation shown by the remaining species of Clytia ( Fig. 6F View Figure 6 ).

Specimens of C. cf. gracilis , although not clearly individualized, can be set apart from each other when compared as a group: C. cf. gracilis sp.B, C. cf. gracilis sp.1 and sp.2 have larger hydrothecae and pedicels (LH, DHMa, DHMe, DHB, DP) with higher and more numerous cusps (NC, HCMax, HCMin), while C. cf. gracilis sp.3 and sp.4 have, in general, lower values for those characters ( Fig. 5E, F View Figure 5 ). If measurements

1 Specimens identified as Clytia sp. from He et al. (2015) clustered with specimens of Clytia cf. gracilis sp. 3 in the phylogeny of Cunha et al. (2017), and should be referred to that species. However, since we were not able to study the morphology of these specimens, they were not considered in the proposed re-identifications.

2 Specimens identified as Clytia gulangensis from He et al. (2015) clustered with specimens of Clytia cf. gracilis sp. 5 in the phylogeny of Cunha et al. (2017) (see discussion). Since we were not able to study the morphology of these specimens, they were not considered in the proposed re-identifications.

3 Specimens identified as Clytia gracilis sp.A from Lindner et al. (2011) clustered with specimens of Clytia cf. hemisphaerica sp. 1 in the phylogeny of Cunha et al. (2017), and should be referred to that species. Specimens identified as Clytia xiamenensis from Zhou et al. (2013) also clustered with Clytia cf. hemisphaerica sp.1, but these results are only based on 16S sequences (see Cunha et al., 2017), and should be confirmed. Since we were not able to study the morphology of these specimens, they were not considered in the proposed re-identifications.

related to erect colonies are excluded from the analysis (LIS, PIS, NIS, DIS), C. cf. gracilis sp.1 and C. cf. gracilis sp.B can be further separated from C. cf. gracilis sp.2 by the length (LH) and length:diameter ratio of the hydrotheca (HRatio; Fig. 5F View Figure 5 ), although these differences are too small to be informative and delimit lineages. Specimens of C. cf. gracilis sp.5 spread along the four quadrants of the graph because of their high variation in the characters examined ( Fig. 5E, F View Figure 5 ). Additional comparisons with literature descriptions show that morphological variation is pronounced in the presumably typical C. gracilis , and the lineages analysed here could fit one or more descriptions (Supporting Information, Table S4).

Specimens of C. cf. hemisphaerica are not separated by any of the morphological measurements, showing intermediate values for most of the characters evaluated ( Fig. 5A–D View Figure 5 ; Supporting Information, Fig. S2 View Figure 2 ). Characters that are important to differentiate other species of Clytia are uninformative for lineages of C. cf. hemisphaerica , especially because of their wide range of variation and extensive overlap. This variability is also seen when descriptions from the literature are compared (Supporting Information, Table S5; Supporting Information, Fig. S3 View Figure 3 ).

Additional PCAs, including characters from the gonotheca, show less conspicuous patterns of differentiation among species (Supporting Information, Fig. S2 View Figure 2 ). Clytia hummelincki (Leloup, 1935) has been shown to not be part of Clytiidae in previous phylogenetic analysis ( Cunha et al., 2017) and, therefore, was not included in the PCAs with this family.