Echinolampadoida Kroh & Smith, 2010

Souto, Camilla, Mooi, Rich, Martins, Luciana, Menegola, Carla & Marshall, Charles R., 2019, Homoplasy and extinction: the phylogeny of cassidulid echinoids (Echinodermata), Zoological Journal of the Linnean Society 187, pp. 622-660 : 631-643

Order Echinolampadoida Kroh & Smith, 2010

Family Echinolampadidae Gray, 1851 : Echinolampas depressa : CASIZ 174963; NHMUK 1892.2.25.23; USNM E15144, E15565, E28085, E29737, E32973, E41070 ; UF 1246, 9027 .

DATA COLLECTION, CHARACTERS AND CODING

Data were collected from direct observation of specimens and from the literature. Morphological analyses were performed using a stereo microscope attached to a camera lucida (if available during museum visits). Light application of ethanol with a paintbrush was used to highlight plate boundaries of dry specimens. When authorized by museum curators, fossils were cleaned and polished to reveal plate boundaries and ambulacral pores. Test measurements were taken with digital callipers to the nearest 0.01 mm. Selected drawings were digitized and converted to highresolution images in Adobe Illustrator CS6 using a Wacom Intuos tablet.

Tube foot ossicles, pedicellariae and sphaeridia were removed with thin needles, cleaned and disarticulated using household bleach (~5% sodium hypochlorite solution) for 3–5 min, washed using three changes of distilled water and kept in absolute ethanol. They were then placed on metal stubs with doublesided carbon tape using a dropper, separated from each other using a thin needle, set aside to air-dry and imaged with a Hitachi TM-1000 SEM (Electron Microscope Laboratory, University of California, Berkeley, CA, USA). The plate patterns of the living taxa were visualized using SRμCT images obtained at the ALS-LBNL (beamline 8.3.2), following the protocol described by Souto & Martins (2018).

We selected 98 morphological features based on test shape (11 characters, 25 states), apical system (seven characters, 15 states), aboral ambulacra (20 characters, 54 states), periproct and I5 (19 characters, 54 states), peristome and basicoronal plates (20 characters, 43 states), oral ambulacra and sphaeridia (15 characters, 38 states) and tuberculation and pedicellariae (six characters, 13 states). We did not exclude any character that had high homoplasy indices in previous studies, but characters related to the internal organs were not included, because they are available for only a small subset of the extant species included here. Quantitative character states were defined based on natural, non-overlapping and obvious gaps observed during analysis of a broad spectrum of the taxa involved in the study. Whenever ratios or quantitative measures are presented, these are merely to help delimit parameters for the character states, so that additional taxa can be coded by future workers. Only the largest specimens of each species were measured to reduce biases related to ontogenetic changes. Sixtyfour characters were binary and 34 multi-state, with a total of 98 characters and 242 character states. Four quantitative characters were ordered.

The following list applies to all phylogenetic analyses (see explanation in ‘Phylogenetic analyses’). Ordered characters and characters excluded from analysis 2 (A2; i.e. characters with high percentages of missing data) are indicated.

A) Test shape: aboral view.

1. Test outline: subquadrate ( Fig. 2A) or rounded ( Fig. 2B) (TW> 0.90 TL) [0]; oval (TW 0.75– 0.90 TL; Fig. 2C, D) [1]; elongate (TW <0.75 TL; Fig. 2E) [2]. This character codes for the relationship between the test width and the test length.

2. Test edge in aboral view: uniform, nearly straight edges ( Fig. 2A, E) [0]; curved, greatest in the middle or posteriorly ( Fig. 2B–D) [1]. This character codes for the shape of the ambital plates at interambulacra 1 and 4.

3. Test funnelled posteriorly: no [0]; yes ( Fig. 2D, E) [1]. In tests with a funnelled posterior region, the plates in interambulacra 1 and 4 are nearly straight, and the width of the test decreases rapidly from widest point to the posterior region.

B) Test shape: frontal and posterior view.

4. Shape of the transverse cross-section of the test: triangular ( Fig. 2F, G) [0]; dome-shaped ( Fig. 2H, I) [1]. Triangular tests are slightly inflated, and they increase in height while they diminish in width; domed tests are strongly inflated, and they increase in height while they largely maintain their width.

5. Acute peak at the apical system: absent ( Fig. 2F, I) [0]; present ( Fig. 2G, H) [1]. This character codes for an elevation at the apical system common in Rhyncholampas species and is independent from the transverse cross-section of the test. For instance, Rl. ericsoni has a domed test and an acute peak at the apical system.

6. Position of ambitus: low ( Fig. 2F–H) [0]; high ( Fig. 2I) [1]. In tests with low ambitus, the widest region is in the oral-most one-third of the test, and the ambitus is at an acute angle; in tests with high ambitus, the widest region is near or at the middle of the test, and the ambitus is at an obtuse angle.

7. Oral posterior I5 plates convex: no [0]; yes [1]. Some cassiduloids have a curvature on the oral posterior I5 plates resulting in an interradial region projected downwards. In addition, there is a depression on the plates at the oral ambulacra 1 and 5, and instead of having an open midline throughout the test, these species have an ‘M-shaped’ bipartite channel ( Fig. 2G). This condition is clearest in Hardouinia but is also present in other genera.

C) Test shape: lateral view.

8. Convexity of aboral region: aboral region flat ( Fig. 2J, K) [0]; slight posterior slope ( Fig. 2L) [1]; steep posterior slope ( Fig. 2M) [2]. Tests with a flat aboral region have a roughly uniform height from the apical system to the periproct. In tests with a posterior slope, the test decreases in height towards the periproct, meaning that the greatest centre of mass is in the anteriormost region of the test; the decrease may be slight or steep.

9. Posterior region of test truncated: no ( Fig. 2K–M) [0]; yes ( Fig. 2J) [1].

D) Test shape: oral view.

10. Concavity of oral region: nearly flat [0]; concavity only near the peristome [1]; concavity starting at test edge [2].

11. Inflation of posterior interambulacra: no [0]; yes [1]. This character codes for the presence of a swollen region in the oral interambulacra 1, 4 and 5, found in Ne. rostellata and Sd. recens .

E) Apical system.

12. Apical system monobasal in adults: no [0]; yes [1]. The apical system in Ap. recens is tetrabasal in juveniles and monobasal in adults, possibly because of the fusion of the genital plates. Given that we do not have information about the ontogenetic changes in most fossil taxa, we chose to code this character for adults only.

13. Length of apical system in relationship to the test length: large (> 8.5% of TL) [0]; medium-sized (6.5–8.5% of TL) [1]; small (<6.5% of TL) [2]. Measurements of the apical system were taken from the anterior margin of ocular plate III to the posteriormost region of the apical system (posterior edge of madreporic plate or of ocular plates I and V). Only mature specimens (i.e. with all gonopores well developed) were measured. For species with strong sexual dimorphism in gonopore size (e.g. Sd. recens ), only males were measured.

14. Number of gonopores: four [0]; three [1].

15. Symmetry among gonopores: symmetric [0]; asymmetric [1]. Gonopores are asymmetric when gonopores 1 and 2 are displaced adorally and aborally, respectively, meaning that the position of the gonopores in the left and right side of the apical system is asymmetric. For species with three gonopores, symmetry was based on the position of the posterior gonopores in relationship to the centre of the apical system.

16. Location of ocular plates: between gonopores [0]; beyond gonopores [1]. In species with strong sexual dimorphism in gonopore size (e.g. Sd. recens ), the ocular plates are located beyond the gonopores in males, and beyond the gonopores or between their adoral edge in females. We coded for the condition in males.

17. Madreporic plate extended posteriorly: no [0]; yes [1]. In some species, the posterior region of the madreporic plate has an acute rather than a curved or flat edge, ending beyond gonopores 1 and 4.

18. Hydropores: abundant, all over madreporic plate [0]; few (≤ 15), confined to a small region in madreporic plate [1].

F) Aboral ambulacra.

Neolampas rostellata does not have developed petals, and their ambulacral system is reduced to single and rudimentary pores, except for the phyllodes, which are well developed. Characters 19–26, 28–30 and 32 code for features of the aboral ambulacra not applicable to this species.

19. Longest petal: I and III roughly the same size [0]; petal I longest [1]; petal III longest [2]. In species with unequal columns of pore-pairs, we measured the longest column.

20. Posterior paired petals short: no [0]; yes [1]. Petals I and V were considered short when their length was <90% the length of the anterior paired petals.

G) Aboral ambulacra: petal III.

21. Petal III, adoral shape ( Fig. 3A): wide (Wa/ Wm> 0.70) [0]; convergent (Wa/Wm = 0.40– 0.70) [1]; tapering (Wa/Wm <0.40) [2]. The ancestral state is a divergent petal, in which the pores at the end of the petal follow the growth in plate width and become more separated; in contrast, in tapering petals, the pores at the end of the petal are often positioned slightly closer to the midline of the petal.

22. Petal III, shape of columns of respiratory podia: both straight ( Fig. 3B, C) [0]; inner straight and outer bowed ( Fig. 3D) [1]; both bowed ( Fig. 3E, F) [2]. This character codes for the change in width of the plates in the petal: some have constant width throughout; others increase and then decrease in width.

23. Petal III, width of poriferous zone in relationship to interporiferous zone ( Fig. 3A): very wide (Wr> Wm) [0]; wide (Wm ≥ Wr and Wm <2Wr) [1]; narrow (Wm> 2Wr) [2]. This character codes for the relationship between the region of the petal responsible for gas exchange (poriferous) and the region with more primary spine coverage (interporiferous). In some species, the region for gas exchange is very reduced and the area with primary spines is large; in others, the region for gas exchange takes up most of the petal area.

24. Petal III, length of a and b columns of respiratory podia: equal or differ by one pore-pair ( Fig. 3A) [0]; differ by more than one pore-pair [1].

H) Aboral ambulacra: petals II and IV.

25. Shape of anterior paired petals: straight ( Fig. 3B) [0]; V-shaped ( Fig. 3C) [1]; oval ( Fig. 3D) [2]; tulip-shaped ( Fig. 3E) [3]; leaf-shaped ( Fig. 3F) [4]. These states are usually distinguished by the width of the ambulacral plates throughout the petal length and by the position of the inner pores. In straight petals, the ambulacral plates have roughly the same width, and the columns of respiratory podia are straight and parallel. In the other petal shapes, the ambulacral plates increase in width towards the middle of the petal. This increase may be continuous throughout the petal, while the inner column is straight but not parallel, resulting in V-shaped petals; or the ambulacral plates decrease in width from the middle to the end of the petal, resulting in a bowed outer column of respiratory podia. In the presence of bowed outer columns, straight and parallel inner columns result in oval petals, bowed and open (i.e. broadly separated) inner columns result in tulip-shaped petals, and bowed and tapering (i.e. end of petal is nearly closed) inner columns result in leaf-shaped petals.

26. Length of a and b columns of respiratory podia, paired anterior petals: equal or differ by one pore-pair ( Fig. 3B–D, F) [0]; differ by two to four pore-pairs ( Fig. 3E) [1]; differ by five or more pore-pairs [2].

I) Aboral ambulacra: petals I and V.

27. Shape of the ambulacra at paired posterior petals: uniform ( Fig. 3H) [0]; bowed ( Fig. 3G, I) [1]. The shape was evaluated based on the difference between the widest region of the petal and the width of the plates at the end of the petal. In bowed petals, the width of the plates increases up to the middle of the petal and then decreases by> 25% towards the end.

28. Width of a and b columns of respiratory podia (Wr in Fig. 3A) at paired posterior petals: same width [0]; width of inner column <80% width of outer column [1]. The a and b columns of respiratory podia are usually symmetric, but in a few cases, columns 1b and 5a are wider than 1a and 5b.

29. Length of a and b columns of respiratory podia at paired posterior petals: equal or differ by one pore-pair [0]; unequal and differ by more than one pore-pair [1]. Most cassiduloids have columns of pore-pairs of the same size, but in a few species the number of pore-pairs may vary significantly between individuals of the same species (i.e. Rl. pacifica , whose columns may differ by three to seven pore-pairs). This condition appears early in the life of the echinoid because of different timing in the development of both columns and, therefore, should not be influenced by the size of the specimen.

J) Aboral ambulacra: general characteristics of petals.

30. Shape of outer (i.e. adradial) respiratory podia in paired petals: slit-like [0]; elongated [1]; rounded [2]. In all species, the outer pores are rounded internally, but in some cases these pores expand as they approach the surface of the test, becoming elongated (width is maintained) or slit-like.

31. Density of primary tubercles in interporiferous zone: high ( Fig. 4A) [0]; low ( Fig. 4B, C) [1]. The density of tubercles is high when there is no space among them, and low when the mean distance among primary tubercles is greater than the diameter of a single tubercle so that additional tubercles could be accommodated among them.

32. Tuberculation of poriferous zone: miliary tubercles only [0]; miliary and one or two sparse primary tubercles [1]; miliary and three to five often reduced primary tubercles [2]; six or more reduced primary tubercles [3]. Some species have small primary tubercles in the poriferous zones; although reduced in size, these are still larger than the miliary tubercles.

33. Last inner pore of paired petals on occluded plate: no [0]; yes [1].

K) Aboral ambulacra: plates beyond posterior paired petals.

34. Expansion of the posterior ambulacra beyond petals ( Fig. 3A): uniform ( Fig. 3G) [0]; slight expansion ( Fig. 3H) [1]; strong expansion ( Fig. 3I) [2]. Expansion (Wx) was estimated in relationship to the width of the plates at the end of the posterior petals (We): slight expansion is an increase by ≤ 100%, and strong expansion is an increase of> 100%.

35. Orientation of posterior ambulacra beyond petals: curved anteriorly ( Fig. 3H) [0]; straight expansion following ambulacra ( Fig. 3G, I) [1].

36. Amount of expansion (Wx) in relationship to greatest width of posterior petals (Wp) ( Fig. 3A): petal> 5% wider ( Fig. 3G) [0]; same width ( Fig. 3H) [1]; expansion> 5% wider ( Fig. 3I) [2]. This ratio apparently changes with ontogeny (i.e. expansion grows faster with ontogeny); therefore, only the larger specimens of each species were analysed.

37. Shape of ambulacral plates beyond posterior petals: rectangular, wider than long [0]; squared or slightly longer than wide [1]. This character codes for the first four plates after the end of the petal. Rectangular plates are present when the ambulacrum is wide and the pores beyond petals are close to one another.

38. Placement of pores beyond posterior petals: near or at adradial suture ( Fig. 4A) [0]; between adradial suture and the middle of the plate ( Fig. 4B) [1]; running thorough the midline of the plate ( Fig. 4C) [2]. For this character, we analysed the placement of the pores from the ambital plates. This character codes for the position of the pore on the outside of the test (i.e. external view). Given that the pores may not follow a straight path across the stereom (i.e. through the test wall), their position from the inside of the test side may differ.

L) Periproct and interambulacrum 5.

39. Periproct position: aboral [0]; marginal [1]; oral [2].

40. Presence of a prominent aboral hood over periproct: no ( Fig. 5A, C) [0]; yes ( Fig. 5B) [1]. A hood is formed when the aboral plates framing the periproct curve and extend, forming a hood that covers the periproct opening from above.

41. Shape of lateral plates framing periproct: bent inwards ( Fig. 5C, E) [0]; straight ( Fig. 5A, B, D) [1]. Initially, the periproct of the cassiduloids was placed in a groove formed by the bending of the lateral plates framing it. But in many groups, the lateral plates do not bend and are narrower as a result.

42. Lateral plates framing periproct supported internally by "buttresses": no [0]; yes ( Fig. 6A) [1]. The plates framing the periproct may be supported internally by an additional layer of stereom that connects them.

43. Periproct with subanal shelf: no ( Fig. 5A, C) [0]; yes ( Fig. 5B) [1]. A subanal shelf is formed by the inward and horizontal elongation of the adoral plates framing the periproct.

44. Periproct orientation: longitudinal (width <length) [0]; equant (width = length) [1]; transverse (width> length) [2]. The lateral plates framing transverse periprocts are usually shorter and narrower than the lateral plates framing longitudinal periprocts.

45. Periproct tear-shaped: no [0]; yes [1]. In a tear-shaped periproct, the width increases from the aboral to the oral region.

46. Plates on periproctal membrane: one row of medium-sized plates and many small plates ( Fig. 6B) [0]; two rows of medium-sized plates and few small plates ( Fig. 6C) [1]; one row with three large plates ( Fig. 6D) [2]. Not included in A2.

47. Anus placement in peristomial membrane: in the centre ( Fig. 6B) [0]; on the aboral edge ( Fig. 6C, D) [1]. Not included in A2.

48. Shape of interambulacral plates beneath periproct: concave, forming a groove [0]; convex [1]. This character is inapplicable to the taxa with a periproct near or at the oral surface.

49. Minimum number of plates on I5, between the basicoronal plate and the base of the periproct (plate identity in parenthesis and referenced in Fig. 6E): ≥ 11 (5.a.11 onwards) [0]; ten (5.a.10) [1]; nine (5.a.9) [2]; eight (5.a.8) [3]; seven (5.a.7) [4]; six (5.a.6) [5]; five (5.a.5) [6]. This character codes for the position of the periproct with respect to specific plates within I5. The number of plates may undergo a slight variation within a species (usually by only one plate), but it does not vary with the size of the specimen. This number is also not related to the general size of the species; for instance, Rl. mexicana (TL = 70 mm) and Eu. australiae (TL = 26 mm) have the same number of plates. Ordered 0–1–2–3–4–5–6.

50. Minimum number of plates framing the periproct ( Fig. 6F): ten or more [0]; nine [1]; eight [2]; seven [3]; six [4]; five [5]; four [6]. This character represents the sum of the number of plates framing the entire periproct (i.e. on both sides) and is not necessarily correlated with the size of the periproct, given that the length of the plates may vary across taxa. Ordered 0–1–2–3–4–5–6.

51. Presence of primary tubercles in posterior region of I5 basicoronal plate and adoral region of plates 5.2: absent or few [0]; abundant (more than five tubercles) [1]. Some species have tubercles near the peristome regardless of the presence (and width) of a naked zone. The tubercles may be sparse along the phyllodes or randomly distributed (variable within a species), or abundant along the phyllodes and in the middle of the plate.

52. Naked zone running along oral I5: absent ( Fig. 7A) [0]; reduced ( Fig. 7B) [1]; developed ( Fig. 7C–E) [2]. A reduced naked zone has only a small reduction in tubercle density, and it does not reach the posterior edge of the test.

53. Width of I5 naked zone in relationship to the test width: narrow (<10% TW; Fig. 7C, D) [0]; wide (≥ 12% TW; Fig. 7E) [1]. This character was coded based on the broadest region of the naked zone, usually in the middle.

54. I5 granular: no [0]; yes [1]. Although the naked zone is free of primary spines, miliary spines are still present. In some species, there is an increased density of miliary tubercles, giving a granular appearance.

55. Pits on I5: absent [0]; finely pitted [1]; deeply pitted [2]. The distribution of pits in the naked zone was very variable. Therefore, we coded only for their size rather than their distribution.

56. Pits on aboral edge of interambulacral basicoronal plates: absent [0]; present ( Fig. 8E) [1]. Pits may be large and deep (as in some Eurhodia ) or small and shallow (as in Au. longianus ).

57. Naked zone running along oral ambulacrum III: absent ( Fig. 7A, B) [0]; narrow ( Fig. 7C) [1]; wide ( Fig. 7D, E) [2]. The width of the naked zone in ambulacrum III was estimated based on the naked zone in I5. Naked zone III is usually larger, but in some species it is narrower than the naked zone in I5. If the naked zone was equally narrow or equally wide throughout, species were coded as having a narrow or wide naked zone III, respectively.

M) Peristome and basicoronal plates.

58. Peristome orientation: transverse (width> 1.1 times length) [0]; equant (width = 0.9– 1.1 times length) [1]; longitudinal (width <0.9 times length) [2].

59. Shape of peristome: (sub)pentagonal ( Fig. 7G–J) [0]; oval ( Fig. 7A–F) [1]. The peristome in some cassiduloids (i.e. C. infidus ) develops from a circular to a pentagonal shape, passing through a subpentagonal stage when juvenile.

60. Peristome position: near the centre or slightly anterior [0]; very anterior [1]. Peristomes were considered very anterior when their posterior edge was <42% of the TL from the anterior ambitus.

61. Accretion of stereom on interambulacral basicoronal plates: absent or low ( Fig. 8A, C) [0]; high ( Fig. 8B, D) [1]. In some species, there is accretion of a thick stereom layer on the basicoronal plates, forming solid bourrelets.

62. Deep depression on interambulacral basicoronal plates: absent ( Fig. 8B, D) [0]; present ( Fig. 8A, C) [1].

63. Bourrelet 5 bulged anteriorly: no ( Fig. 7F, H–J) [0]; yes ( Fig. 7G) [1]. In some species, the posterior region of the peristome is strongly convex adorally; in others, this anterior projection is weak or absent, resulting in a nearly flat posterior edge.

64. Bourrelets pointed: no [0]; yes ( Fig. 8E) [1]. Developed bourrelets are usually smoothly bulged, but sometimes they project outwards (towards the sediment), forming a pointed tip.

65. Bourrelets tooth-like: no [0]; yes ( Fig. 8F) [1]. In tooth-like bourrelets, the sides are straight instead of rounded, and the aboral region of the bourrelet is wider than the adoral region.

66. Bourrelet 5 poorly developed: no [0]; yes [1]. This character codes for the development of bourrelet 5 in relationship to bourrelets 2 and 3. Despite being undeveloped, bourrelet 5 may still be slightly bulged, pointed or tooth-like.

67. Basicoronal plates 1 and 4 narrower than basicoronal plate 5: no ( Fig. 7I, J) [0]; yes ( Fig. 7G) [1].

68. Oral surface of I5 basicoronal plate longer than wide: no, wider or equant ( Fig. 7F, G) [0]; yes ( Fig. 7I, J) [1].

69. Aboral edge of I5 basicoronal plate expands beyond aboral edge of ambulacrum V basicoronal plate: no ( Fig. 7G) [0]; yes ( Fig. 7I, J) [1].

70. Adoral edge of I5 basicoronal plate more than twice as wide as adjacent ambulacrum V basicoronal plate: yes ( Fig. 7G, I) [0]; no ( Fig. 7J) [1].

71. Size of basicoronal plates I and V along the perradial suture: short [0]; medium-sized [1]; enlarged [2]. The size was estimated based on the orientation and size of the second ambulacral plate. When the basicoronal is short, the second plate is diagonal to the midline of the phyllode ( Fig. 7H). When it is medium-sized, the second plate is perpendicular to the midline of the phyllode ( Fig. 7G), and when it is enlarged, the second plate is a demiplate ( Fig. 7J).

72. Shape of ambulacral basicoronal plates: flush or wall-like [0]; bent [1]. The shape of the plates apparently influences where the first pores are located: flush basicoronal plates are slightly curved, and the first pores sometimes start deep inside the peristome ( Fig. 8G); wall-like basicoronal plates are straight, and the first pores are placed at the adoral region of the plate, often facing the inside of the peristome ( Fig. 8H); and in bent plates, a high proportion of the plate is on the adoral side, and the first pores are located close to the phyllodes, facing outwards ( Fig. 8I).

73. Adoral region of ambulacral basicoronal plate depressed: no [0]; yes ( Fig. 8F) [1]. Depressed plates are often enlarged aborally, and their lowest region is usually lower than the peristomial opening.

74. Ambulacral basicoronal plates with more than two pores: no [0]; yes [1].

75. First pore in ambulacral basicoronal plate modified into a buccal pore: no ( Fig. 9D) [0]; reduced ( Figs. 7H, 9E) [1]; distinct ( Figs. 7G, I, J, 9A–C, F) [2].

76. Distance between first and second ambulacral pores: near ( Fig. 9A–D, F) [0]; far ( Fig. 9E) [1]. When the pores are far from each other, there is a large and noticeable gap between them.

77. Placement of second ambulacral pore: in the middle of the plate, often aborally ( Fig. 7G) [0]; aborally and near the adradial suture ( Fig. 7H) [1]. One plate on each pair of ambulacral basicoronal plates in cassiduloids has at least two pores. In species with only two pores, the second pore is always placed aborally, but in species with more than two pores, the placement of the pores along the anterior–posterior axis will vary according to the number of pores present.

N) Oral ambulacra: phyllode III.

78. Shape of outer column of anterior phyllode: straight ( Fig. 9A, D) [0]; barrel-shaped ( Fig. 9B) [1]; triangular ( Fig. 9C, E, F) [2]. Straight phyllodes have parallel columns of pores; barrel-shaped phyllodes have their greatest width in the middle; and triangular phyllodes have their greatest width adorally.

O) Oral ambulacra: phyllodes II and IV.

79. Arrangement of columns of paired anterior phyllodes: one column (inner column absent; Fig. 9A, E) [0]; one column and scattered pores ( Fig. 9C, F) [1]; two complete columns (inner column throughout phyllode; Fig. 9B, D) [2]. In phyllodes in which the inner column is complete, the outer column is usually composed only of demiplates.

80. Shape of outer column of paired anterior phyllodes: rows of three ( Fig. 9D) [0]; straight or barrel-shaped ( Fig. 9A–C, F) [1]; tapering ( Fig. 9E) [2]. In tapering phyllodes, the first phyllopores are widely separated while the last phyllopores are very close to each other.

81. Maximal number of primary plates in paired anterior phyllodes: ≥ 11 [0]; eight to ten [1]; five to seven [2]; up to four [3]. Character states were chosen based on intraspecific variability. For example, some species had specimens with five to seven or eight to ten pores but never outside of these ranges. Ordered 0–1–2–3.

P) Oral ambulacra: phyllodes I and V.

82. Size of posterior phyllodes: long, last phyllopore aboral to second interambulacral plate [0]; short, last phyllopore adoral to third interambulacral plate [1].

83. Maximal number of primary plates in posterior phyllodes: ≥ 12 [0]; eight to 11 [1]; four to seven [2]. Character states were chosen based on intraspecific variability. Also, in phyllodes with up to seven pores the pores are spaced out, but in phyllodes with ≥ 12 pores the pores are close together and the phyllode is highly developed. Ordered 0–1–2.

84. Arrangement of outer phyllopores in external view: pores in a uniform column ( Fig. 9A–E) [0]; pores scattered ( Fig. 9F) [1]. Phyllopores are usually placed near the adradial suture, but in some species the phyllopores are also found in the middle of the plate or near the perradial suture. These pores were considered as part of the outer column because they are not homologous with pores in occluded plates.

85. Phyllodes tapering: no [0]; yes ( Fig. 9E) [1]. In tapering phyllodes, the last phyllopores are very close to each other.

86. Occluded plates in posterior phyllodes: absent or rare ( Fig. 9G) [0]; few ( Fig. 9H) [1]; many ( Fig. 9I) [2]. The presence and number of occluded plates is usually conserved within a species, but in some cases we found one or two specimens with one occluded plate even though occluded plates were absent for the species. These occurrences were considered rare. The abundance of occluded plates was also assessed by taking the ratio between the number of occluded plates and the number of primary plates into account. In phyllodes coded as having many occluded plates, at least one-third of the plates are occluded.

87. Presence of primary tubercles on phyllodes: absent [0]; present aborally ( Fig. 9D) [1].

Q) Oral ambulacra: sphaeridia.

88. Location of sphaeridial pits in posterior phyllodes: near buccal pores only [0]; throughout phyllodes [1]. This character was coded as unknown for three species ( K. malayana , K. florescens and Rp. marmini ) because they have very short phyllodes, making it challenging to assess whether the sphaeridial pits are restricted to a small region near the peristome or whether they would be widespread if the phyllodes were larger.

89. Sphaeridia placement: in open pits [0]; concealed by a thin layer of stereom [1].

90. Sphaeridial pits greatly reduced: no [0]; yes [1].

91. Number of sphaeridial pits: seven or more [0]; up to six [1].

R) Oral ambulacra: plates beyond phyllodes.

92. Shape of ambulacral plates beyond phyllodes: transverse [0]; square [1]; longitudinal [2]. This character codes for the ambulacral plates in the oral region only.

S) Overall test tuberculation.

93. Tubercle size: aboral tubercles ≥ 60% as large as oral tubercles [0]; aboral tubercles <60% diameter of oral tubercles [1].

94. Oral tubercles with bosses displaced from centre: no [0]; yes [1]. Species with enlarged areoles have larger spines on the oral region of the test that aid in locomotion.

T) Pedicellariae.

95. Teeth on blade of ophicephalous pedicellariae: teeth form an open-U blade and run down on the edges of the neck ( Fig. 10A) [0]; teeth form a semi-oval blade and run down in the middle of the neck ( Fig. 10B) [1]; teeth form an oval blade and are absent in the neck ( Fig. 10C) [2]. Not included in A2.

96. Size of teeth on distal region of ophicephalous pedicellariae: coarse ( Fig. 10B, C) [0]; fine ( Fig. 10A) [1]. Not included in A2.

97. Tridentate pedicellariae blade in relationship to base: long and narrow ( Fig. 10D) [0]; short and broad ( Fig. 10E) [1]. Not included in A2.

98. Teeth on base of tridentate pedicellariae ( Souto & Martins, 2018: table 2): absent [0]; present [1]. Not included in A2.

A data matrix (Supporting Information, Appendix S1) was constructed in Mesquite v.3.51 ( Maddison & Maddison, 2018). Phylogenetically uninformative characters were not included; polymorphic characters were retained. Inapplicable characters were coded as ‘–’, and missing data were coded as ‘?’. For some characters, we were able to exclude a subset of the character states for a particular taxon, but we were unsure of the remaining character states. These partial uncertainties were included within curly brackets and not coded as missing data. We used the command ‘mstaxa = variable’ to differentiate partial uncertainty and polymorphism. Missing data often result in a high number of equally parsimonious solutions and reduced resolution. Partial uncertainty should ameliorate these effects.

After coding the characters, we estimated the completeness of all fossil taxa ( Table 2). Rowe (1988) defined completeness as the percentage of missing data (owing to non-preservation and inapplicability) in relationship to the total number of characters in the matrix. In our estimation of completeness, only the characters with missing data owing to non-preservation were considered. We think that inapplicability should not affect the estimation of completeness, because if fossil preservation allowed for the detection of inapplicability of characters, then the preservation for that character should be considered to be good.

PHYLOGENETIC ANALYSES

Four cladistic analyses were conducted using the software PAUP* v.4.0a163 ( Swofford, 2003) using the parsimony optimality criterion. In all of these, heuristic searches were performed using stepwise random addition sequences with 1000 replicates (start = stepwise, addseq = random, randomize = addseq, nreps = 100) followed by tree bisection–reconnection branch swapping (swap = tbr, multrees = yes). Five trees were held at each step of stepwise addition (hold = 5). Branches without unambiguous optimizations were collapsed (pset collapse = minbrlen). Trees with the best score were retained (filter best = yes, permdel = yes). Finally, strict consensus and 50% majority-rule consensus trees were generated. Clade support was determined with bootstrap resampling ( Felsenstein, 1985) using the full ‘heuristic’ option with 1000 heuristic replicates, and character changes were optimized using the ‘accelerated transformation’ (ACCTRAN) option. Batch files with commands are available in the Supporting Information (Appendix S2) and in Morphobank ( O’Leary & Kaufman, 2012) project P3287.

Nucleolites scutatus , Ap. recens and Ec. depressa are the most distantly related taxa; hence, they were used as outgroups in PAUP to root the tree. All characters were treated as equally weighted, and continuous characters not derived from ratios were ordered (additional analyses with unordered and reweighted characters were also performed). Analysis 1 (A1) included all ingroup (45) and outgroup taxa (21), all 98 characters, and coded for partial uncertainty. To analyse the effect of missing data on the resulting topology, characters coded for <20% of the species (N = 6) were excluded from analysis 2 (A2). In analysis 3 (A3), partial uncertainties (N = 34) were converted into missing data (‘?’). Analysis 4 (A4) aimed to examine the influence of fossil taxa on the tree topology; hence, it included only extant taxa (six ingroup taxa and six outgroups).

USING STRATIGRAPHY TO CHOOSE THE BEST TREE

Temporal data have been applied in parsimony-based phylogenetic reconstruction in two different ways: a priori, as discrete characters used to build phylogenies [e.g. stratocladistic methods ( Bodenbender & Fisher, 2001)]; and a posteriori, as a separate dataset to test phylogenetic hypotheses (e.g. Day et al., 2016). Although many agree that temporal data should be used in association with phylogeny ( Gauthier et al., 1988; Huelsenbeck, 1994; Fox et al., 1999), stratocladistics has been severely criticized, especially because the concept of homology does not apply to time, but also because of the way that time is binned into stratigraphic intervals (Smith, 2000; Sumrall & Brochu, 2003). Fisher (2008) reviewed the main concerns raised by critics and provided a discussion addressing them. However, most software does not support temporal data, making the implementation and testing of this method challenging.

Here, we used temporal data a posteriori to calculate four stratigraphic congruence metrics [the gap excess ratio (GER; Wills, 1999; Wills et al., 2008), the modified Manhattan stratigraphic measure (MSM*; Siddall, 1998; Pol & Norell, 2001), the relative consistency index (RCI; Benton & Storrs, 1994) and the stratigraphic consistency index (SCI; Huelsenbeck, 1994)] for each MPT and determine which MPT best fits stratigraphy. Different metrics of stratigraphic congruence and their refinements have been proposed, all assessing whether the FAD of a taxon corresponds to its placement in the phylogeny and/or the length of the ghost lineages.

Tests were performed using the DatePhylo and StratPhyloCongruence functions of the ‘strap’ R package ( Bell & Lloyd, 2015). Input files consisted of the MPTs and a list with the FAD and LAD of each taxon. We adopted a conservative approach and included uncertain ages in the temporal range of species (see Table 2). Analyses were performed using the ‘basic’ dating method ( Smith, 1994), which sets the root length at 0 Myr. Polytomies were treated as hard (hard = TRUE), and outgroups and topologies were fixed. Given that the temporal data come from stratigraphic intervals rather than absolute ages, we treated FADs and LADs as uncertain, and two values were randomly drawn from within the interval (randomly.sample.ages = TRUE, samp.perm = 1000). Estimated P-values were then calculated for these metrics from 1000 randomly generated trees (rand. perm = 1000). The R script to estimate the stratigraphic congruence metrics is provided in the Supporting Information (Appendix S2) and in Morphobank ( O’Leary & Kaufman, 2012) project P3287.

TREE CALIBRATION

Stratigraphic ranges of species ( Table 2) were obtained from the literature and museum records; absolute dates were not available. Five additional extinct taxa [ Paralampas platisterna ( Smith & Jeffery, 2000) , Rhyncholampas cookei Sanchez-Roig, 1952 , Rhyncholampas fontis ( Cooke, 1942) , Rhyncholampas smithi Srivastava et al., 2008 and Glossaster? apianus ( Besaire & Lambert, 1930) ] were used to calibrate the phylogeny. For each cassidulid genus, we targeted the oldest species and species occurring in different geographical areas. However, we included only the five species whose literature data allowed for a reliable phylogenetic placement. These species were manually added a posteriori to the best tree using assignable synapomorphies (e.g. node dating; Table 4), which was then calibrated using the ‘basic’ method and plotted against the geological time scale of the International Commission on Stratigraphy ( Cohen et al., 2013; updated).