Kassandrina, SOUTO & MARTINS, 2018

GENUS KASSANDRINA SOUTO & MARTINS, 2018 View in CoL

Emended diagnosis: Medium-sized eurhodiids with straight-edged, oval test; oral region concave. Petals short; anterior petals oval; posterior petals of uniform width. Poriferous zone very wide when compared with interporiferous zone, with sparse primary tubercles. Ambulacra beyond posterior petals expanded; pores placed in middle of plate, along suture. Periproct on aboral surface, narrow, longitudinal. About seven to nine interambulacral plates between basicoronal plate 5 and base of periproct. Naked zone with large or small pits. Peristome transverse, pentagonal. Bourrelets pointed. Phyllodes short; up to five phyllopores per half; occluded plates scattered. Tridentate pedicellariae long, thin.

Includedspeciesbasedonpresentanalysis: K.malayana (type species) and K. florescens .

HOMOPLASY AND CHARACTER EVOLUTION

The low CI and moderate RI ( Fig. 15; Supporting Information, Appendix S6) indicate that most characters are homoplastic, which helps to explain the low bootstrap values. A similar result was obtained by Smith (2001) and Kroh & Smith (2010) in their phylogeny of the post-Palaeozoic echinoids, which suggested that the evolutionary history of the cassiduloids involved multiple shuffling of character states (shuffling here does not refer to lateral gene transfer, but to the constant character state changes as a result of homoplasy) rather than the evolution of novel traits ( Smith, 2001; present paper). Kier (1962), Suter (1994a) and Saucède & Néraudeau (2006) also attributed the high level of homoplasy, and consequently low phylogenetic resolution, to parallel evolution in the cassiduloids. In fact, parallelism and reversals are frequent among irregular echinoids that evolved to live in similar environments (e.g. Kier, 1974; Smith, 2001; Saucède et al., 2003).

The evolution of the apical system from four to one genital plate (i.e. tetrabasal vs. monobasal) has been poorly studied. It is unclear whether some genital plates reduced in size until they disappeared, leaving a single enlarged plate, whether the genital plates fused to form a solid plate, or whether there was some combination of these processes. In the cassiduloid clade described here, the apical system changed from monobasal to tetrabasal in some faujasiids, and then apparently reverted to monobasal in F. apicalis .

Other major transitions concern the peristome and periproct. In the reconstructed cassiduloid phylogeny, the orientation of the peristome changed from transverse to longitudinal and vice versa, the periproct position changed multiple times from marginal to aboral and once to oral (all within clade E), and the orientation of the periproct changed multiple times from longitudinal to transverse. These transitions are affected by the rate and orientation of plate growth and the rate of plate addition. Different lineages could be affected differently. For example, a transverse periproct is not necessarily framed by fewer plates than a longitudinal periproct, and the number of interambulacral plates from the peristome to the periproct is not necessarily higher if the periproct is aboral rather than marginal, although the number of plates tends to be lower in species whose periproct is oral ( Figs. 16, 17).

Souto & Martins (2018) showed that the bourrelets in cassiduloids may be formed by the accretion of stereom onto the basicoronal plates or by an internal depression on the basicoronal plates that projects the bourrelets outwards. Our morphological analyses indicate that these conditions can also co-occur. For example, Rl. pacifica has a slight depression in the interambulacral basicoronal plates and a thick accretion of stereom, whereas K. florescens has a deep depression in the interambulacral basicoronal plates and a slight accretion of stereom. Given that slight depressions are difficult to detect in fossils and unbroken extant species, we did not code for it. Whether both conditions co-occur or not, usually only one is responsible for the formation of the bourrelets. Usually, tooth-like and pointed bourrelets in faujasiids are formed by a deep depression in the plates, whereas the bulged and pointed bourrelets in cassidulids are formed by a strong accretion of stereom.

Micro-computed tomography has provided insights about the different ways in which bourrelets are built ( Souto & Martins, 2018), but there is still much to learn about other cassiduloid novelties, such as modifications of the naked zone and apical system. These novelties are usually coded for presence vs. absence or tetrabasal vs. monobasal, respectively, but without an examination of their ontogeny and deeper homologies we are likely to be missing important parts of the story that can lead to more nuanced coding schemes.

USING FOSSILS TO RECONSTRUCT PHYLOGENIES

The inclusion of fossil species provided better resolution of phylogenetic relationships in the cassidulids, allowed for the delimitation of supraspecific taxa and detected taxonomic inconsistencies that have not been assessed before. For example, K. malayana was classified in the cassidulids ( Mooi, 1990b; Suter, 1994b), and many have considered Rl. pacifica congeneric with C. caribaearum (e.g. Agassiz, 1869; Mortensen, 1948a), but the analyses performed here show that Rhyncholampas and Cassidulus have been separated for ≥ 60 Myr. Characters responsible for this nesting include: scattered arrangement of phyllopores in posterior phyllodes and convex shape of bourrelet 5, which is shared with C. mitis and C. briareus , petals with a narrow poriferous zone, a high number of sphaeridia and naked zone with reduced pits, which is shared only with C. mitis .

Smith (2001) and Kroh & Smith (2010) also recovered different results from phylogenies with and without fossils ( Fig. 1C, E). As a result of unique combinations of character states that have often been erased in Recent species, fossils generally improve phylogenetic resolution ( Huelsenbeck, 1994). However, this improvement will depend on tradeoffs between the completeness and temporal position of the fossils. For example, young fossils with low completeness may worsen the phylogenetic resolution ( Huelsenbeck, 1991). In the phylogeny reconstructed here, completeness was relatively high (74–100%), but one of the species with lower completeness ( Rl. riveroi , 75–77% complete) and dating back to the Late Oligocene resulted in a trichotomy, whereas a taxon from the Late Cretaceous with similar completeness ( G. welschi , 74–75% complete) had a better resolution.

Fossils may not be as important when recovering the relationships of very closely related extant taxa or of taxa whose diversity trajectories are skewed towards the recent. However, because the diversity trajectory of cassiduoids is bottom heavy and its few extant species are relicts, descending from ancient lineages separated by tens of millions of years, adding fossils is necessary to recover the morphological information lost since those lineages split. Phylogenetic studies that include fossil invertebrates are still uncommon for numerous reasons, including the incompleteness and lack of knowledge of the fossil record, the physical separation of biological and palaeontological collections, and the different research traditions between these two disciplines. Also, although the resolution of the present analysis has not been diminished by the amount of missing data, their negative effect in phylogenetic resolution usually prevents the inclusion of fossil taxa in evolutionary studies ( Donoghue et al., 1989). As observed here, the effect of missing data can be reduced if more characters are added ( Wiens, 2003; Prevosti & Chemisquy, 2010) even if they are highly homoplastic. Our results strongly recommend the inclusion of fossil taxa whenever possible.