A larger benthic foraminiferal zonation
for the Eocene
of the Caribbean and central American region

Extensive new material (about 100 samples) has been collected from Jamaica over the last 15 years, which we integrate with previously collected material from Jamaica and elsewhere within the Caribbean-American region (Fig. 1 ). In general we have tried to collect free specimens of LBF so that external ornamentation can be recorded, and orientated sections (equatorial and axial) can be cut. This allows us to evaluate whether species can be determined in random orientations in thin sections cut from rock samples.

Typically large samples (a one US gallon or 3.78 litre sized bag) were collected by one of us (SFM, labelled WL) or by two of us (SFM and EÖ, also labelled WL) and broken down by drying and soaking in water, or by repeated freeze-thaw cycles in water. Following breakdown, samples were washed through a 0.5 mm sieve and the LBF were picked. For samples with low yields (or low yields of critical species) further samples were collected and disaggregated. Specimens were sorted into morphospecies based on exterior characteristics and selected specimens were then cut on a glass plate with 600 grade corundum grit to provide orientated equatorial and axial sections. During the cutting process observations were also made of the presence of features (e.g., lateral chamberlets and piles/pillars) that would not be present in the final orientated equatorial thin-section. UWIGM.WL samples are preserved in the collections of the University of the West Indies Geological Museum, Jamaica. EO.WL samples are preserved in the paleontological collections of the Geological Engineering Department of İstanbul Technical University, Turkey. USNM refers to specimens housed in the Smithsonian Institute, Washington - DC. Additional material collected by two of us (ER and NR) is also used in this study, this is prefixed by the letters 'ER' and 'R' respectively. The ER material was collected at various times between the late 1950s and 2005. The NR material was collected in 2017-2018. Some of the ER and NR material includes orientated sections, but most is of unorientated specimens in thin section. A detailed list of samples is provided in the sections below.

Selected samples were also studied for nannofossils. These either represented samples from continuous sections or spot samples associated with particular LBF assemblages. Previous work on nannofossils (e.g., Jiang & Robinson, 1987; Robinson & Jiang, 1997) has been updated for inclusion in this work.

3. Investigated sections in Jamaica

We present brief details of the investigated sections here together with the lithostratigraphic schemes currently in use. Sections are included either because they contain intervals with both LBF and nannofossils, because they yielded free specimens of LBF, or because they are important historically for the descriptions of LBF. We correlate these sections using our LBF zones (American Benthic Zones or ABZ) that are calibrated to the chronostratigraphy using calcareous nannofossils. This paper is concerned with the Eocene which covers zones AE3 to AE16 (Fig. 4 ) and zones for the Paleocene and Oligocene to Miocene will be presented elsewhere. The zones are formally described after the description of the sections. A revision of the lithostratigraphic units will be presented elsewhere, as it is beyond the scope of this paper. Numbers for localities refer to Fig. 3 .

Area 1. Northern Wag Water Belt (parish of St Mary): Ypresian to Priabonian

This area of the northern part of the Wag Water Belt (Fig. 3 ) consists of small-scale tectonic blocks separated by faults. As such, a continuous stratigraphy is not present. The lithostratigraphy follows Burke et al. (1969) and Mann and Burke (1990). LBF have been recovered from the Wag Water Formation, Richmond Formation and the Yellow Limestone (Palmetto Grove and Preston Hill formations) and White Limestone ("Bonny Gate" Formation) groups. The LBF typically occur as transported specimens in turbidites in deep-water shales and marlstone-limestone deposits, which we consider are broadly contemporaneous with the sedimentary rocks they are found in. Chronostratigraphic correlation is based on updated previous nannofossil records (Jiang & Robinson, 1987; Robinson, 1996, 2004) together with additional samples investigated in this work. The samples are listed in Table 2 (with details of the nannofossils in Appendix 1) and a correlation of the formations in this area is shown in Fig. 4 .

A calcareous conglomerate in the Wag Water Formation (at Devon Pen) yielded a LBF assemblage characterised by Tremastegina with orthophragmines (Jiang & Robinson, 1987). This would indicated ABZ3. The Richmond Formation yields LBF in turbidites or shell beds (Robinson & Mitchell, 1999) and indicate zones ABZ3 and ABZ4. Samples from the Richmond Formation around Albany yield ABZ3 foraminifers with nannofossils indicating Zone NP12 (Jiang & Robinson, 1987). Samples from mudstones around Langley yield LBF indicating ABZ4 and nannofossils indicating Zone NP14a (Jiang & Robinson, 1987; Appendix 1). The Wag Water Formation and Richmond Formation, at least in this area, must interfinger; with the Wag Water Formation representing slope deposits that pass into the basinal sandstones and shales of the Richmond Formation (Mitchell, 2021).

The Yellow Limestone Group is represented by the Palmetto Grove and Preston Hill formations (Burke et al., 1969; Robinson & Mitchell, 1999). One sample was collected from the Palmetto Grove Formation (WL3315) and four samples were collected from the Preston Hill Formation (ER152 (= original ER sample R78), ER161, WL4869 and WL4870). No nannofossil data are available for WL3315 (which contains ABZ5 LBF) or ER152/R78 (which contains ABZ7 LBF). Planktic foraminiferal and nannofossil data were published for ER161 (P11, upper NP15, and CP13-lower CP14a; ABZ7-ABZ8 transition) by Robinson (1996). Nannofossil data for samples WL4869 and WL4870, which yield Eulinderina guayabalensis and Planorbitoinella ecuadorensis n. gen. et sp., indicate lower Zone NP15b (Table 2, Appendix 1).

Table 2: Details of LBF and calcareous nannofossil zone for samples from area 1 (northern Wag Water Belt). Column 1 indicates LBF zone assignment based on proximity of samples or fauna. WL GPS positions were measured in the field (ER GPS positions picked from georeferenced original field maps).

* Data from Jiang & Robinson (1987); ⁺ Data from Robinson (1996).

The "Bonny Gate" Formation in this area has also yielded some useful samples. These are referred to the early Bartonian (WL3416: ABZ11) and late Priabonian (ER176, WL4865, ABZ16). Nannofossil data for ER176 (zone NP21) were published by Robinson (2004).

Area 2. North Coast Belt, Rio Sambre Gorge, and area around (parish of St Mary): Ypresian to Lutetian

This is among the most important areas for establishing the succession of faunas from the late Ypresian to the late Lutetian because of the continuous nature of the successions and quality of the exposures present. The lower part of the succession (Guys Hill, Richmond, and lower Palmetto Grove formations) is exposed in the bed and banks of the Rio Sambre, about 0.5 to 2 km downstream from the bridge between Palmetto Grove and Kilancholly, and the upper part (upper Palmetto Grove and Preston Hill formations) is exposed in road cuts between Farm Turn, Kilancholly and Palmetto Grove in St Mary (Fig. 3 ). The section in the Rio Sambre has been the subject of several reports (e.g., Burke et al., 1969; Wise & Constans, 1976; Robinson & Mitchell, 1999). Detailed geological mapping indicates that the section begins with the Pembroke Hall Formation (of Burke et al., 1969), passes through the Guys Hill, Richmond and Palmetto Grove formations and terminates in the Preston Hill Formation. A summary section of the Richmond and Palmetto Grove formations was published in Robinson and Mitchell (1999,p. 38-39).

We have re-measured and recollected this entire section. Three separate sections were logged in the Rio Sambre. Section 1 (18° 17.411" N; 76° 56.039" W) extends from the Richmond Formation (samples WL4054-WL4056: ABZ4) into the lower part of the Palmetto Grove Formation. Section 2 (18° 16.903" N; 76° 55.547" W to 18° 16.991" N; 76° 55.618" W) extends from the Richmond Formation (WL4071-4070, WL4033, WL4035, WL4037, WL4041: ABZ4) into the lower Palmetto Grove Formation (samples WL4049, WL4026: ABZ5). Section 3 (18° 16.930" N; 76° 55.621" W to 18° 16.903" N; 76° 55.547" W) extends from the Richmond Formation (samples WL4057, WL4027, WL4058: ABZ4; sample WL4025: ABZ5) into the lower Palmetto Grove Formation (samples WL4059, WL4024, WL4023: ABZ5). The different sections can be correlated using turbidites (Fig. 5 ) which can be traced between sections. Samples were collected for calcareous nannofossils to chronostratigraphically calibrate the succession. This indicates the presence of zone NP14a and basal NP14b. Additionally upsection reductions in the abundances of small Toweius spp. and Discoaster lodoensis suggest that nearly the entire thickness of Zone NP14a is present (Fig. 5 ). The appearance of Blackites inflatus near the top of the section (Fig. 5 ) indicates the base of the Lutetian. Details of all samples are provided in Fig. 5 and Table 3.

The upper part of the succession is limited to small exposures along roads around Farm Turn, and Kilancholly, where the upper part of the Palmetto Grove Formation and the lower part of the Preston Hill Formation are exposed. Spot samples along these roads have been analysed for their LBFs and zones ABZ6 to ABZ8 are represented (Table 3). These samples have been tied into the chronostratigraphy using calcareous nannofossil identifications from sample splits. Very rich assemblages of LBF are present including dictyoconids, orthophragmines, helicosteginids, nummulitids and Fabiania, amongst others. Table 3 shows the breakdown of sample numbers and zone based on LBF and calcareous nannofossils for samples above the Rio Sambre section. Details of the nannofossils are given in Appendix 1.

Table 3: Details of LBF and calcareous nannofossil zone for samples (above the Rio Sambre section) from area 2 (western St Mary).

Area 3. Lilyfield-Bamboo Area (North Coast Belt, parish of St Ann): Lutetian

This area is historically significant because Cole (1956, 1963) presented information on the LBF from here. Özcan et al. (2019, 2022a) also presented data on the orthophragmines from here. This area (Fig. 3 ) yields good faunas from the Preston Hill and "Chapelton" formations. Above this is a typical shallow-water Clarendon Block stratigraphy which consists of the Troy, "Swanswick" s.l. (see area 4 below), Claremont and Somerset formations; these are hard limestones and were not sampled for LBF because they lack free specimens (they are included on Fig. 4 for completeness). We did not carry out any nannofossil analyses for this area, because this part of the stratigraphy was covered in the area around Farm Turn (Area 2) and the LBF allow a straight forward correlation (Fig. 4 ). Zones ABZ8 and ABZ9 are represented in this area (Fig. 4 ). Details of the samples, LBF zones and nannofossil zone are given in Table 4 and Appendix 1.

Table 4: Details of LBF zone for samples from area 3 (Lilyfield area). (Nannofossils were not studied from this area.)

Area 4. Swanswick and Clarks Town (North Coast Belt/Clarendon Block): Lutetian, Bartonian and Priabonian

This area is significant because it contains the type section of the Swanswick Formation of Hose and Versey (1956 - see also Versey in Zans et al., 1963, where the type section is defined). Unfortunately, the faunas from the type Swanswick Formation do not correlate with the Swanswick Formation as mapped across much of Jamaica (e.g., Versey in Zans et al., 1963; Robinson & Mitchell, 1999; Mitchell, 2013a, 2016). The type Swanswick section contains two different stratigraphic intervals: the lower level consists of yellow-brown impure wackestones (a typical Yellow Limestone lithology) and yields Helicostegina gyralis in association with Linderina floridensis and nummulitids (Wright & Switzer, 1971; Robinson & Mitchell, 1999; herein) and can be assigned to the early Lutetian (ABZ7); in contrast, the upper part of the succession consists of pale cream packstones and yields Heterostegina ocalana and very rare Fabularia verseyi allowing a correlation with ABZ16 (Fig. 4 ); an interval between consists of lime mudstones and is best assigned to the top of the Claremont Formation. Across much of the Clarendon Block, the white grainstones that have been mapped as Swanswick Formation occur between the Troy Formation (dolostones and recrystallized limestones) and the Claremont Formation (predominately lime mudstones: Mitchell, 2004, 2013a), which yield an ABZ15 LBF assemblage. Thus the type Swanswick (here designated as "Swanswick" s.s.) includes only early Lutetian (ABZ7) and late Priabonian (ABZ16) LBF with a probable fault (maybe an unconformity) between, whereas the widely mapped Swanswick Formation (designated here as "Swanswick" s.l.) yields a lower ABZ15 LBF assemblage. It is notable that to the south of Clarkes Town (SW of Swanswick itself) the Preston Hill Formation is exposed and yields late Lutetian LBF (ABZ10, herein) indicating that the Yellow Limestone in the lower part of the type Swanswick Formation is a limestone level developed within the Preston Hill Formation. A revision to the lithostratigraphy is required, and this will be published elsewhere. No nannofossil work was undertaken by us on these successions because of the numerous faults and the generally indurated nature of the limestones. Samples are listed in Table 5.

Table 5: Details of LBF and calcareous nannofossil zone for samples from area 4 (type Swanswick Formation).

Area 5. Sherwood Content (North Coast Belt, parish of Trelawny): Lutetian and Bartonian

Sherwood Content lies in the transition zone between the Clarendon Block and the North Coast Belt (Fig. 3 ) and faunas were described from here by Vaughan (1928) and Cole (1956). The succession consists of three units: the Preston Hill Formation, Chapelton Formation and Healthy Hill Formation. LBF assemblages indicate a succession through the Bartonian (Fig. 4 ) with ABZ11, ABZ13 and ABZ14 represented. No nannofossil work was undertaken in this area because much of the succession here consists of indurated limestones. Samples are indicated in Table 6.

Table 6: Details of LBF and calcareous nannofossil zone for samples from area 5 (North Coast Belt, parish of Trelawny).

Area 6. Spot Valley, Salt Spring and Flower Hill (North Coast Belt): Bartonian

Eocene rocks belonging to the Yellow Limestone and White Limestone groups occur in a structurally complex area to the SE of Montego Bay (Fig. 3 ) in what has been called the Montego Bay Anticline (Meyerhoff & Krieg, 1977). There is a transition here from marlstones of the Preston Hill Formation to the overlying chalks of the "Bonny Gate" Formation (Fig. 4 ), which occurs in ABZ11. The overlying chalks of the "Bonny Gate" Formation have yielded LBF indicating ABZ12 (at Spots Valley) and ABZ14 (south of Flower Hill). Nannofossil samples were taken from the marlstone-chalk sequence in ABZ11 (indicating zone NP16 and CP14a) and from the chalks in ABZ12 at Spots Valley (which were barren). Correlations are shown in Fig. 4 . Samples and LBF and nannofossil zones are listed in Table 7 and Appendix 2.

Table 7: Details of LBF and calcareous nannofossil zone for samples from area 6 (Spot Valley and Flower Hill, North Coast Belt, parish of St James).

Area 7. Yellow Limestone around Albert Town and Litchfield (Clarendon Block): late Lutetian and early Bartonian

The Yellow Limestone around the north-western margin of the Central Inlier (Mitchell, 2013b; Fig. 3 ) consists of four formations: Freemans Hall, Stettin, Litchfield and Chapelton formations (Fig. 4 ) (Robinson, 1996; Robinson & Mitchell, 1999; Mitchell, 2016). Samples from the Stettin Formation yield H. minor in the absence of nummulitids and can be assigned to the earliest Lutetian lower ABZ6 (Fig. 4 ). The overlying Litchfield Formation lacks zonally significant LBF, but the Chapelton Formation contains LBF indicating zones ABZ9 and ABZ11, indicating a hiatus representing ABZ10 (Fig. 4 ). Some nannofossil work on the Chapelton Formation has previously been presented by Robinson (1996) for this area and indicates zone NP16. Samples are detailed in Table 8.

Table 8: Details of LBF and calcareous nannofossil zone for samples from area 7 (Albert Town and Litchfield, Clarendon Block, parish of Trelawny).

* Data from Robinson (1996).

Area 8. Yellow Limestone at Coleyville (Clarendon Block): Bartonian

The area around Coleyville (Fig. 3 ) has a similar lithostratigraphy to that of the area around Albert Town and Litchfield, although the LBF assemblages are different (further from the platform edge). Zonal indicators are largely absent (except for very rare H. minor in the Stettin Formation) until the upper part of the Chapelton Formation. The upper part of the Chapelton Formation at the Forest Reserve yields abundant lepidocyclinids indicating ABZ13. No nannofossil work was undertaken here and correlations are shown in Fig. 4 . Samples are described in Table 9.

Table 9: Details of LBF and calcareous nannofossil zone for samples from area 8 (Coleyville, Clarendon Block, parish of Manchester).

Area 9. Western side of the Clarendon Block, St Elizabeth and St James: Lutetian and Bartonian

Rocks of the Yellow Limestone Group occur in a series of inliers in northern St Elizabeth and southern St James where the rocks of the White Limestone Group have been eroded (Fig. 3 ). The succession has similarities with the succession around the Central Inlier, but the Bartonian has thickened substantially and can be mapped in more detail. The formations present here need revision, but are currently assigned to the Stettin Formation, Helicostegina-beds, Litchfield Formation, Chapelton Formation, Healthy Hill Formation, and 'Ipswich' Formation (Versey, 1957; Atkinson, 1969). The 'Ipswich' Formation can be divided into three distinct mappable units: a lower unit of pale, thickly-bedded wackestone with a platform-type LBF assemblage; a series of brown and locally red packstones and grainstones carrying a lepidocyclinid assemblage (the type Ipswich Formation of Versey, 1957), and an upper unit passing back into wackestones with some lepidocyclinids and platform interior LBF.

ABZ6 is found in the Stettin and basal part of the Helicostegina-beds, zones ABZ7 - ABZ8 in the middle and upper parts of the Helicostegina-beds, ABZ10 in the Chapelton Formation, and zones ABZ13 to ABZ14 are represented in the middle and upper 'Ipswich' Formation (Fig. 4 ). The Yellow Limestone is overlain by a typical lower White Limestone succession which includes the Troy, 'Swanswick' sensu lato, Claremont and Somerset formations (Fig. 4 ). This demonstrates that the Ipswich Formation is demonstrably older than the Swanswick Formation, although the two were mapped as a single unit by the Jamaican Geological Survey (Bateson, 1972). Limited nannofossil data has been provided by Robinson et al. (2018). Samples are listed in Table 10.

Table 10: Details of LBF and calcareous nannofossil zone for samples from area 9 (western Clarendon Block, parishes of St James and St Elizabeth).

Area 10. Grants Pen, St Andrew (SW Wag Water Belt): Bartonian

The Grants Pen area of St Thomas (Fig. 3 ) is a complex area which has overturned beds and probably significant faulting. The area contains a suite of Yellow Limestone marlstones that are overlain by chalks with at least one chert band. The area is important because it yields LBF. Pseudolepidina trimera occurs in a hard limestone that cannot be dated using calcareous nannofossils; whereas Lepidocyclina ariana (ABZ12) occurs in a Yellow Limestone marlstone which has been chronostratigraphically calibrated using nannofossils to zone NP16 in this work (Appendix 2). Correlations are shown in Fig. 4 , and samples are listed in Table 11.

Table 11: Details of LBF and calcareous nannofossil zone for samples from area 10 (Grants Pen, parish of St Thomas).

Area 11. Highway 2000 around the Golden Vale Junction, parish of St. Ann (Clarendon Block): Priabonian

The road cuts created during the construction of the North-South Leg of Highway 2000 in the parish of St Ann (Fig. 3 ) showed good sections through the Swanswick s.l., Claremont, Somerset, and Walderston Formations. The lower three of these were composed of relatively weakly cemented limestones from which free specimens of LBF could be extracted. These sections were close to the shelf edge in Late Eocene time and contain rich lepidocyclinid assemblages. These were used to understand the succession of lepidocyclinids through the Priabonian, with zones ABZ15 and ABZ16 represented (Fig. 4 ). No nannofossil work was undertaken on these samples. Samples are described in Table 12.

Table 12: Details of LBF and calcareous nannofossil zone for samples from area 11 (road cuts along northern N-S leg of Highway 2000).

Area 12. The central and southern Montpelier-New Market Belt: late Bartonian to Priabonian

The succession in the Montpelier-New Market Belt (Fig. 3 ) covers the interval from the late Lutetian to the Priabonian. Samples were used from the Content-1 well which penetrated the lower part of the succession (Chapelton Formation and overlying marlstones) and from surface outcrops. Surface outcrops can be divided into five Eocene units: Chapelton Formation, marlstones, lower unit of chalks, unit of chalks with cherts, and upper unit of chalks. New road cuts associated with the improvement work on the Mackfield to Ferris Cross main road facilitated the collection of samples from weakly indurated, freshly exposed rocks. A series of nannofossil samples was analysed from here from the late Bartonian into the mid-Priabonian. ABZ10 (samples from the Content Well: Eva, 1980), ABZ14, ABZ15 and ABZ16 are represented in this area (Fig. 4 ). Samples are listed in Table 13 and nannofossil records are included in Appendix 2.

Table 13: Details of LBF and calcareous nannofossil zone for samples from area 12 (Montpelier-New Market Belt).

Area 13. Jerusalem Mountain - Upper Rock Spring Area (parish of Westmoreland)

The Chapelton Formation is exposed around Rock Spring to the north of the Jerusalem Mountain Inlier (Fig. 3 ; Mitchell & Edwards, 2016). The Chapelton Formation here yields abundant free specimens of Helicostegina jeannemairae associated with Operculinoides sp. and can be assigned to ABZ9 (Fig. 4 ). No nannofossil work was carried out here, and the sample is listed in Table 14.

Table 14: Details of LBF and calcareous nannofossil zone for the sample from area 13 (Jerusalem Mountain - Upper Rock Spring, parish of Westmoreland).

4. Zonation scheme

We develop a zonation scheme for the Eocene of the American biogeographic province based on the stratigraphic distribution of selected taxa. The zonation is based on the first appearance datum (FAD) and last appearance datum (LAD) of selected taxa. Planktic foraminiferal zones are indicated by the prefixes 'P' (Berggren et al., 1995) and 'E' (Pearson et al., 2006) and calcareous nannofossil zones by the prefixes 'NP' (Martini, 1971), 'CN' (Okada & Bukry, 1980) and 'CNE' (Agnini et al., 2014). The American Benthic Zones (ABZs) are defined below, and their ages are constrained largely using calcareous nannofossils related to the 'NP' zonation. The zonation scheme for the Eocene is shown in Fig. 6 and consists of 14 zones (ABZ3 to ABZ16). The fourteen Eocene zone are defined below, and the zonation for the Paleocene, Oligocene and early to middle Miocene will be presented in a separate paper.

ABZ3. Tremastegina lopeztrigoi Zone. From the LAD of Ranikothalia catenula (which defines ABZ2) to the FAD of H. wellsi. Small specimens tentatively assigned to T. lopeztrigoi are found in the Nonsuch Formation of Jamaica where they co-occur with Ranikothalia (Robinson & Jiang, 1990; Robinson & Wright, 1993). ABZ3 is recognized by the presence of T. lopeztrigoi, in the absence of either Ranikothalia (ABZ2) or Helicostegina wellsi (ABZ4). The associated LBF assemblage is characterised by long-ranging forms, particularly orthophragmines, such as Orbitoclypeus, Stenocyclina, Pseudophragmina and Asterocyclina (see Özcan et al., 2019, 2022a, for a revision of the orthophragmines). The reference section for the zone is in the Richmond Formation near Albany, northern Wag Water Belt, Jamaica (Fig. 3 ), and is assigned to NP12 (Fig. 6 ; Jiang & Robinson, 1987). Blanco-Bustamentel et al. (1999) suggested that Ranikothalia and Eoconuloides (= Helicostegina herein) wellsi occurred throughout the lower Eocene, but these results have not been proved elsewhere and a reassessment of these successions in Cuba is needed.

ABZ4. Helicostegina wellsi Zone. From the FAD of H. wellsi to the FAD of Nephronummulites macgillavryi. In shelf edge localities, the LBF of this zone are dominated by H. wellsi and T. lopeztrigoi, locally with H. minor, Fallotella and Cushmania. In the deeper shelf succession of the Richmond Formation the zone contains common H. wellsi and T. lopeztrigoi along with a high-diversity LBF assemblage including orthophragmines (Asterocyclina, Stenocyclina, Pseudophragmina and Orbitoclypeus), Fabiania, Cushmania, Fallotella and Eorupertia, and locally H. minor. The reference locality for the zone is in the Rio Sambre section, parish of St Mary, Jamaica (Robinson & Mitchell, 1999; Fig. 5). Our nannofossil determinations indicate that ABZ4 belongs to the lower part of Zone NP14a (Fig. 6 ; Appendix 1). This is consistent with previous age assignments (NP14 and P9: Wise & Constans, 1976; Robinson & Mitchell, 1999). We tentatively place the boundary between ABZ3 and ABZ4 within Zone NP13 as we do not have LBF assemblages indicating Zone NP13. It is unclear if NP13 is absent due to a hiatus, or, more likely, cut out by faulting.

The occurrence of H. wellsi (as Amphistegina elliotti) in Ecuador as recorded by Cushman and Stainforth (1946) is a typical shallow-water occurrence, where H. wellsi is the only species present. H. wellsi is a distinctive species and can easily be recognized in random thin sections. The presence of Tremostegina senni and H. minor in the absence of Operculinoides (Senn, 1940; Caudri, 1972; Jones, 2009) in the Murphy's Beds of Barbados (Upper Scotland Group) suggests assignment to ABZ4 (or possibly lower ABZ6).

ABZ5. Nephronummulites macgillavryi Zone. From the FAD to LAD of N. macgillavryi. The reference section lies in the upper part of the Richmond Formation and lower part of the Palmetto Grove Formation in the Rio Sambre Gorge, parish of St Mary, Jamaica (Figs. 5 - 6). In the lower part of the zone, N. macgillavryi co-occurs with H. wellsi, whereas H. minor occurs throughout the zone. The zone is here correlated with the upper part of the Ypresian (upper Zone NP14a below the entry of the nannofossil Blackites inflatus in Zone NP14b (Figs. 5 - 6). The zone is recognized in Cuba (Rutten, 1935), Jamaica (Robinson & Wright, 1993; herein), and Nicaragua (Andji et al., 2018). Many assemblages are characterised only by N. macgillavryi (as reported by Rutten, 1935; Robinson & Wright, 1993; Andji et al., 2018), and these probably represent shelf edge assemblages lacking the deeper-water orthophragmine component of the fauna. N. macgillavryi is a distinctive species and can easily be recognized in hand specimen and in random thin sections.

ABZ6. Helicostegina minor Zone. From the LAD of N. macgillavryi to the FAD of Operculinoides spp. This is a low diversity interval above the Lower-Middle Eocene boundary, and amongst the Asterogerinoidea only Helicostegina and T. lopeztrigoi are present; longer ranging taxa include orthophragmines (including Stenocyclina advena and Asterocyclina spp.), Fallotella and Cushmania. There is a succession of species of Helicostegina, H. minor occurs in the lower part of the zone and H. gyralis in the upper part of the zone (Fig. 6 ) and thus it can be divided into lower (ABZ6A) and upper (ABZ6B) parts. The reference locality for the zone lies in the upper part of the Palmetto Grove Formation above the Rio Sambre in the parish of St Mary, Jamaica. The zone also occurs in shelf edge assemblages where H. minor occurs in association with inner platform LBF (Verseyella jamaicensis (Cole, 1956), Coleiconus zansi Robinson, 1993, and Fabularia colei Robinson, 1969) in the Stettin Formation of the Central Inlier (Robinson & Wright, 1993; Robinson & Mitchell, 1999). In the Rio Sambre section, the entire zone yields the nannofossil Blackites inflatus indicating Zone NP14b (Fig. 6 ; Appendix 1). Zone NP15a has not been recorded in samples from transects in this area, nor elsewhere in Jamaica, and there appears to be a minor hiatus (representing zone NP15a) between ABZ6 and ABZ7 in Jamaica.

In Cuba, Cole and Bermúdez (1944) recorded an identical assemblage of LBFs from the Peñon Formation, which we can assign to the upper part of ABZ6 based on the presence of H. gyralis in the absence of Operculinoides spp. Torres-Silva et al. (2018) recorded a second assemblage from the Peñon Formation in Cuba in which Operculinoides spp. occurred with H. wellsi (reported as E. wellsi), which we would place in ABZ7.

ABZ7. Operculinoides spp. Zone. From the FAD of Operculinoides ex gr. floridensis and Nummulites ex gr. striatoreticulatus to the FAD of Eolepidina antillea (with R < 5.5). The reference section lies on the road from Farm Turn to Hampstead, parish of St Mary, Jamaica. The zone sees the appearance of numerous taxa including Eulinderina guayabalensis, Planorbitoinella ecuadorensis and Pseudolepidina trimera suggesting an important migration event at this time. The area from where these taxa migrated from is unknown, but could be in the Pacific, and it represents the most significant faunal change in the Eocene of the Americas. Helicostegina gyralis occurs in the lower part of the zone, whereas H. dimorpha occurs in the upper part of the zone. The appearance of Operculinoides ex gr. floridensis (e.g., Op. jennyi Barker) and Nummulites striatoreticulatus (e.g., N. guayabalensis Barker) in large numbers is particularly conspicuous and the genera can easily be recognized in random thin sections. In the Rio Sambre area Operculinoides ex gr. floridensis appears at the base of Zone NP15b, but Zone NP15a is missing. We tentatively put the ABZ6/ABZ7 boundary within Zone NP15a (Fig. 6 ), pending further research. In Mexico, Eu. guayabalensis occurs in association with Operculinoides prenummulitiformis and Nummulites guayabalensis (Barker, 1939, p. 325) indicating ABZ7. The association of H. wellsi and T. senni (cited as Eoconuloides senni) with Operculinoides sp. in Calyx wells 50 and 50A in Trinidad (Caudri, 1996, Fig. 7 ) also suggests assignment to ABZ7.

ABZ8. Eolepidina antillea Zone. From the FAD of Eo. antillea (with R < 5.5) to the FAD of Eo. subplana (with R < 4.5). The reference section lies on the road from Farm Turn to Hampstead, parish of St Mary, Jamaica. At Farm Turn (Rio Sambre area) the transition from ABZ7 to ABZ8 occurs in the upper part of Zone NP15b (Fig. 6 ; Appendix 1). Robinson (1996) reported that a sample from the type locality of the Preston Hill Marl (his locality 2c, ER161) contained Eolepidina with R = 5.5 (n = 14), which is at the transition between Eu. guayabalensis and Eo. antillea (i.e., the ABZ7/ABZ8 boundary) as defined here. This locality has yielded the following plankton correlations: P11, upper NP15, and CP13-lower CP14a (Robinson, 1996) which is consistent with the upper part of Zone NP15c. We therefore place the boundary between ABZ7 and ABZ8 in the upper part of Zone NP15b (Fig. 6 ). Sample USGS 6897 (collected by S.B. Senn) from St. Bartholomew was reported (Robinson, 1996) to have Eo. antillea with R = 5.08 (n = 12) and was very tentatively attributed to Zone P11/P12 (Westercamp & Andreieff, 1983; Robinson, 1996); it is consistent with the age of ABZ8 as determined here.

ABZ9. Eolepidina subplana Zone. From the FAD of Eo. subplana (with R < 4.5) to the FAD of Eolepidina gardnerae (with R = 3.5 or less). The type section lies at Lilyfield in the parish of St Ann, Jamaica. The zone can be recognized in Alabama (Robinson & Jiang, 1995), St. Bartholomew (Robinson, 1996), Jamaica (Robinson, 1996), Mexico (Barker & Grimsdale, 1936), and Margarita Island, Venezuela (Caudri, 1974). A sample from Albert Town (ER1366 from Robinson, 1996, locality 2B) containing Eo. subplana (indicating ABZ9) yielded a few calcareous nannofossils, including Sphenolithus furcatolithoides, Pemma papillatum and Discoaster barbadiensis and was assigned to Zone NP16 (Robinson, 1996). In the Bay Minette Corehole (Baldwin County, Alabama) Eo. subplana with R = 4 occurs in samples at depths of 529.7 m, 543.2 m and 644.4-646.6 m and the nannofossils indicate assignment to Zone NP16 (Robinson & Jiang, 1995). We therefore place the boundary between ABZ8 and ABZ9 at about the NP15/NP16 boundary (Fig. 6 ). Sample ER1485 from St. Bartholomew (Sample 3A: Robinson, 1996) from the "Upper Group" (near Lurin) was reported (Robinson, 1996) to have a R = 3.82 (n = 11) and was tentatively attributed to Zone P12 (Westercamp & Andreieff, 1983; Robinson, 1996), which is consistent with the age of ABZ9 as determined here. On Margarita Island, Venezuela (Caudri, 1974), the following species are present (identifications updated from Caudri, 1974): Eo. subplana (= Lepidocyclina (Polylepidina) antillea of Caudri, 1974), Orbitoina sp., Helicolepidina polygyralis and Operculinoides spp. and can be assigned to ABZ9.

ABZ10. Eolepidina gardnerae Zone. From the FAD of Eolepidina gardnerae (with R < 3.5) to the FAD of Polylepidina chiapasensis (with R > 2.2). The reference section lies at a depth of 3327 feet (= 1037.6 m) in the Content #1 Well, at Content, parish of Westmoreland, Jamaica. Eolepidina from here (as Polylepidina) have been described by Eva (1980) and Robinson (1996) and are also figured here. Robinson (1996) recorded biostratigraphic data from the level that yielded Eo. gardnerae in the Content-1 Well (his locality 2A): core samples from depths of 1010.4 m, 1030.8 m, 1040.6 m and 1070.8 m had rich, but overgrown, nannofossils (including Sphenolithus furcatolithoides, Cribrocentrum reticulatum, Reticulofenestra umbilica, Helicosphaera compacta, Dictyococcites bisectus (small form; = D. scrippsae) and Pemma papillatum: Robinson & Jiang, 1995; Robinson, 1996) that indicate Zone NP16; whereas a sample from 1040.6 m had calcareous nannofossils (including Sphenolithus furcatolithoides, Pemma papillatum and Reticulofenestra umbilica) and was assigned to Zone CP14a (determinations by W.E. Steinkraus in Robinson, 1996). In the Bay Minette Corehole (Baldwin County, Alabama) Eo. gardnerae (ABZ10) with R close to 3 occurs in three samples collected at depths of 515.1 m, 518.8 m and 522.1 m. These samples are assigned to Zone NP16 (Robinson & Jiang, 1995; Robinson, 1996), but the appearance of Cribrocentrum reticulata at 518.8 m (Robinson & Jiang, 1995) means that the uppermost sample that yielded P. gardnerae can be assigned to a level in Zone CP14a. This means that the boundary between zones ABZ9 and ABZ10 must occur within Zone NP16, probably within the upper part of Zone CP13c (Fig. 6 ).

ABZ11. Polylepidina chiapasensis Zone. From the FAD of Polylepidina chiapasensis (with R = 2 - 2.2) to the FAD of Lepidocyclina ariana. The reference section lies on the road from Salt Spring to Flower Hill in the parish of St James, Jamaica. Samples WL4234 and WL4236 from the Preston Hill Formation (Salt Spring to Flower Hill transect) yield Po. chiapasensis and have yielded nannofossils indicating Zone NP16 and CP14a (Appendix 1). We therefore assign the ABZ10 - ABZ11 boundary to an interval within Zone NP16 (within Zone CP14a) (Fig. 6 ).

ABZ12. Lepidocyclina ariana Zone. From the FAD of Lepidocyclina ariana to the FAD of Lepidocyclina ocalana. At Grants Pen in St Thomas (Jamaica), L. ariana (Sample WL4630) occurs with nannofossils indicating upper Zone NP16 (CP14a) whereas a slightly higher sample without LBF (WL4628) yielded nannofossils possibly indicating Zone NP17 (Appendix 1). In the Bay Minette Corehole (Baldwin County, Alabama) L. ariana occurs with nannofossils indicating Zone NP17 (Robinson & Jiang, 1995). We therefore place the ABZ11- ABZ12 boundary in the upper part of Zone NP16 (CP14a). The reference section for the zone lies at Grants Pen in St Thomas, Jamaica.

This zone is widely developed being found in Jamaica, Alabama (Robinson & Jiang, 1995; Robinson, 1997), Florida (Cole, 1944) and Margarita Island, Venezuela (Caudri, 1974). In Margarita Island the LBF assemblage consists of advanced L. ariana, Orbitoina spp., Helicolepidina spiralis trinitatis [= nortoni Vaughan], Helicosteginopsis soldadensis. Caudriella ospinae and Operculinoides spp. (Caudri, 1974, identifications updated); this assemblage occurs with the planktic foraminifer Orbulinoides beckmanni indicating Zone P13 (Caudri, 1974) - this is in complete agreement with the age calibration of ABZ12 herein based on calcareous nannofossils (Fig. 6 ).

ABZ13. Lepidocyclina ocalana Zone. From the FAD of L. ocalana to the FAD of Pliolepidina tobleri. The assemblage is characterised by the appearance of species of Lepidocyclina with adauxiliary chambers, in contrast to the assemblage present in ABZ12. The reference locality for the zone is in the "Ipswich" Limestone in the Ipswich area, parish of St Elizabeth, Jamaica. A sample from the Ipswich Limestone (ER3103) yielded a sparse nannoflora, including Chiasmolithus grandis and small Cribrocentrum reticulatum, indicating a range from upper Zone NP16 to Zone NP17 (Robinson et al., 2018). Strontium isotope ratios from this level (Robinson et al., 2018) suggest an age in the range 31.25-38.8 Ma consistent with Zone NP17. We therefore correlate ABZ13 primarily to lower-mid Zone NP17 (Fig. 6 ).

ABZ14. Pliolepidina tobleri Zone. From the FAD of Pl. tobleri to the LAD of Pl. tobleri. The reference locality of this zone lies around the marlstones/chalk boundary at Cold Spring in the Montpelier-New Market Belt, parish of Westmoreland, Jamaica. A sample (WL4444) from the marlstones at Cold Spring, just below a sample (WL4449) with common Pl. tobleri yielded a nannoflora indicating zone NP17 (Appendix 1). We therefore assign Zone ABZ14 to Zone NP17 (Fig. 6 ). Strontium isotope ratios from a sample with Pl. tobleri from Appleton, St Elizabeth, Jamaica (Robinson et al., 2018) gave ages which are too young in stratigraphic terms suggesting diagenetic alteration.

ABZ15. Orbitoina wrighti Zone. From the LAD of Pl. tobleri to the FAD of Heterostegina ocalana. The reference section for the zone lies in the Swanswick Formation s.l. on the N-S Highway 2000 in the parish of St Ann, Jamaica. Nannofossil data from the Montpelier-New Market Belt indicate that ABZ15 ranges from Zone NP18 to the lower part of Zone NP19/20 (sensu Martini, 1971, non Speijer et al., 2020; Fig. 6 ).

ABZ16. Heterostegina ocalana Zone. From the FAD of H. ocalana to the LAD of H. ocalana. The reference locality lies in the White Limestone at Dressikie, parish of St Mary, Jamaica. ABZ16 occurs in the type Swanswick (Robinson & Wright, 1993) and Somerset formations (Robinson, 1974) of the marginal facies of the Clarendon Block, as well as in deep-water turbidites in the Wag Water Belt (Robinson, 2004). Localities in the NW Wagwater Belt (ER176 from deep-water chalks of the White Limestone Group at Dressikie) that have yielded H. ocalana contain a nannoflora assignable to Zone NP21 (Robinson, 2004; Robinson et al., 2018). In Cuba, Torres-Silva et al. (2017) suggested a range of H. ocalana from Zone NP17 to Zone NP19-20/CP15 (but the individual zones were not separated suggesting limited precision) and possibly Zone NP21 (CP16) and P18 (early Rupelian). These data do not seem to relate well to Jamaica and provisionally we suggest that ABZ16 ranges from Zone NP19/20 to lower Zone NP21 (Fig. 6 ). Strontium isotopes from the Claremont Formation (which we would attribute to ABZ16) gave ages of 35.05-35.95 Ma (Robinson et al., 2018), which would indicate a level in lower Zone NP19-20.

Occurrences of what have been regarded as typical late Eocene LBF indicators (e.g., orthophragmines, H. ocalana and Fabularia verseyi) in the Oligocene are likely due either to alteration of Sr-isotope values (giving an age that is too young) or due to reworking of the LBFs into the Oligocene (Robinson, 2004; Molina et al., 2015; Robinson et al., 2018; Torres-Silva et al., 2017). The top of the zone is notable for the disappearance of many typical Eocene taxa including H. ocalana, Fabularia verseyi, Coskinolina cf. douvillei, Pellatispirella matleyi and Butterliniana cubensis.

5. Application to selected sections elsewhere in the American biogeographic province

Having established a zonation for the Eocene strata of Jamaica based on LBF, in this section we show how it can be used in selected areas of the American Bioprovince based on previous studies (that is, our interpretations of previously figured specimens) of LBF. Some areas are difficult to comment on, because of tectonic deformation and associated reworking of faunas. The Cuban arc underwent a collision with the North American Plate (Iturralde-Vinent et al., 2016) in the mid-late Eocene and there is extensive reworking of material; it is therefore difficult to work out successions in Cuba other than for the comments we have provided above. With further work, we beleave that the scheme can be used in Cuba, but will require extensive consideration of reworked material. A considerable amount of work has been undertaken in NE Brazil (e.g., Mello e Sousa et al., 2003), but is largely based on random sections of LBF in thin sections; as such it is difficult to apply the scheme there without a reinvestigation of the material available. Our regional correlation is shown in Fig. 6 .

5.1. Florida. Extensive studies concern the Eocene LBF succession in Florida (e.g., P.L. Applin & E.R. Applin, 1941; Cole, 1941, 1942, 1944, 1945; E.R. Applin & Jordan, 1945; Levin, 1957; Cole & E.R. Applin, 1964; Powell, 2010; Cotton et al., 2018). We have reviewed the occurrences and illustrations in these works and present our reinterpretation of the biostratigraphy in Fig. 7 .

The Oldsmar Formation consists of a lower unit of grey-brown dolostone and an upper unit of white to grey, interbedded layers of lime mudstone, wackestones, packstones and tan dolostones (Miller, 1986). The basal boundary is sharp. In Cedar Keys No. 2 well, Linderina floridensis occurs at 983-1068 ft. (= 300-326 m) and Hel. gyralis and L. floridensis occur at 1301-1308 ft (= 397-399 m) (Cole, 1942, p. 15), but no nummulitids were recorded from this well (although it is possible that nummulitids may occur in more open water environments that Helicostegina). Levin (1957) illustrated typical Hel. gyralis from the Oldsmar Formation. This suggests a correlation with upper ABZ6, or possibly lower ABZ7 (if nummulitids are absent due to facies control). However, the similar embryonic sizes of H. gyralis as given by Levin (1957) with specimens from upper ABZ6 suggests an assignment to upper ABZ6.

The Avon Park Formation of P.L. Applin & E.R. Applin (1944, p. 1686) includes the Lake City Formation of P.L. Applin & E.R. Applin (1944, p. 1693) (Miller, 1986). The formation consists of light-brown limestones and dolostones, and has an unconformable relationship with the underlying Oldsmar Formation (Miller, 1986). The best foraminiferal record from the Avon Park Formation comes from Hilliard Turpentine Company well W-366 (Cole, 1944,p. 33-34) where the following species and levels were indicated (species identifications updated): Eolepidina antillea (R = 5/6, Cole, 1944, Pl. 10, figs. 1-3, 5, 7) at a depth of 1340-1350 ft (= 408-411 m) (ABZ8); Eo. subplana (R = 4, Cole, 1944, Pl. 10, figs. 4, 6; Pl. 12, fig. 3) at a depth of 1300-1305 ft (= 396-398 m) (ABZ9); Lep. ariana (R =  2, Cole, 1944, Pl. 16, figs. 11-13) at a depth of 1065-1070 ft (= 324-326 m) (ABZ12) and Orbitoina cedarkeyensis Cole, 1944, p. 34) at a depth of 1005-1025 ft (= 306-312 m) (ABZ12 or ABZ13?). Above 1005 ft. (= 306 m), there is a change in fauna, with the upper part of the Avon Park Formation being characterised by conical agglutinated foraminifers including Coskinolina, Fallotella and Cushmania (Cole, 1944). Specimens of Cushmania from the Avon Park Formation (e.g., Cole, 1944, Pl. 4, fig. 6) have a relatively large (400 μm) first chamber, and this morphology characterises ABZ13-ABZ15 in Jamaica. Thus the Avon Park Formation probably ranges from ABZ8 to at least ABZ13.

The Ocala Formation consists of white- to cream-coloured, highly fossiliferous wackestones, packstones and grainstones with a basal unconformity with the underlying Avon Park Formation (Miller, 1986). It yields a fauna characterised by Lep. ocalana, Het. ocalana and Eu. chaperi (Cole, 1941, 1942, 1944, 1945; Powell, 2010; Cotton, 2018). This assemblage indicates ABZ16.

5.2. Alabama. Robinson and Jiang (1996) provided biostratigraphic data on calcareous nannofossils and LBFs from the Bay Minette corehole and the section at Little Stave Creek in Alabama (Fig. 7 ). We recognize four depositional units, the upper three of which can be constrained by LBF. The appearance of Operculinoides gravelli indicates either ABZ7 or ABZ8, whereas the presence of Eo. subplana and Eo. gardnerae indicate ABZ9 and ABZ10, respectively. There is a hiatus representing ABZ11 and Lep. ariana indicates ABZ12. There is a further hiatus representing ABZ13-ABZ15 whereas the presence of Lep. montgomeriensis possibly suggests ABZ16. There is a general resemblance to the stratigraphy of Florida with similar hiatuses/unconformities.

5.3. Panama. Cole (1952) described the foraminifers of the Gatuncillo Formation from the Panama Canal Zone, and Woodring (1957) provided additional information on the samples. The samples of Cole (1952) indicate three zones: ABZ14 with Plio. tobleri; ABZ15 without either Plio. tobleri or Heter. ocalana (but some samples with Eu. chaperi); and ABZ16 with Heter. ocalana and Eu. chaperi (Fig. 7 ).

5.4. Mexico (state of Veracruz). The species recorded by Barker and Grimsdale (1936, 1937) and Barker (1939) allow the identification of some zones in the state of Veracruz, Mexico (Fig. 7 ). The association of Hel. gyralis with Pseudolepidina trimera and Operc. jennyi at Jenny Station No. 1573 (11 km southeast of Sabaneta, state of Veracruz, Mexico) indicates lower ABZ7, whereas the presence of Hel. dimorpha indicates upper ABZ7 or ABZ8. The presence of Eu. guayabalensis in association with Operculinoides prenummulitiformis and Nummulites guayabalensis (Barker, 1939, p. 325) also indicates ABZ7. The presence of Eulinderina semiradiata (= Eolepidina antillea) and Eulinderina guayabalensis regularis (= Eu. regularis) in association with Tremastegina would indicate a level around the ABZ7-ABZ8 boundary. Polylepidina chiapasensis var. subplana (= Eo. subplana) and Polylepidina 'chiapasensis' (= Eo. gardnerae) indicate zones ABZ9 and ABZ10, respectively. Frost and Langenheim (1974) figured specimens of Eo. gardnerae as Lepidocyclina (Polylepidina) antillea from central Chiapas, Mexico, which would also indicate ABZ10.

5.5. Margarita Island, Venezuela. Brönnimann (1944), Butterlin (1970) and Caudri (1974) described Eocene LBF assemblages from the Punta Mosquito Formation of Margarita Island. Caudri (1974) figured the fauna in detail from three samples and these forms can be placed in the zonation scheme (Fig. 7 ). Sample B yielded Eo. subplana (= Lepidocyclina (Polylepidina) antillea of Caudri, 1974), Orbitoina sp., Helicolepidina polygyralis and Operculinoides spp. (Caudri, 1974, identifications updated) and can be placed in ABZ9. Sample A yielded L. ariana, Orbitoina spp. and Operculinoides spp. (Caudri, 1974) and can be assigned to ABZ12. Sample P.J.B.145 yielded L. ariana, Orbitoina spp., Helicolepidina spiralis trinitatis [= nortoni Vaughan], Helicosteginopsis soldadensis, Caudriella ospinae and Operculinoides spp. (Caudri, 1974, identifications updated) and can also be assigned to ABZ12.

5.6. Venezuela. Raadshooven (1951) described a series of LBF assemblages from Venezuela. The earliest Eocene assemblage from the Quebrada Grande Limestone (Baralt District, State of Zulia, western Venezuela) consisted of Planorbitoinella nitida (= Lepidocyclina sp. A of Raadshooven) which occurred with Helic. gyralis and Operculinoides cf. oliveri which suggests lower ABZ7 (Fig. 7 ). The San Juan Limestone (Baralt District, State of Zulia, western Venezuela) yields Poly. ecuadorensis in association with Operculinoides spp. (cited as O. jennyi by Raadshooven, 1951); this suggests upper ABZ7 (Fig. 7 ).

5.7. Saint Bartholomew (St. Barths). The Eocene succession in Saint Bartholomew has been described in many works (e.g., Lewis & Robinson, 1976; Westercamp & Andreieff, 1983; Caron et al., 2019; Cornée et al., 2020) and details of the LBFs have been given in works such as Cushman (1919), Cole (1960), Robinson (1996) and Cornée et al. (2020). The succession consists of alternations of shallow-water limestones and volcaniclastic sedimentary rocks which have been intruded by various igneous plutons (Westercamp & Andreieff, 1983; Legendre et al., 2018). The stratigraphy for the Eocene rocks has varied from paper to paper. Lewis and Robinson (1976) placed all the Eocene limestones in the Middle Eocene. Westercamp and Andreieff (1983) recognized six Eocene limestone horizons which they divided into a lower group (lower two limestones) and an upper group (upper four limestones), all of which were assigned to the Upper Lutetian. Caron et al. (2019) named the six limestone horizons from base to top: Flamands (1), Mangin (2), Cayes (3), Gouverneur (4) and Lurin (5-6); they assigned these to the upper Lutetian to Bartonian. In contrast, Cornée et al. (2020) divided the succession into three limestones: a Lower Limestone of Lutetian age, an Intermediate Limestone of late Bartonian/early-late Priabonian age, and an Upper Limestone of Priabonian age. The ages of the limestones as given by Cornée et al. (2020) were largely based on LBF identified in random thin sections to genus level, and judging by their illustrations, there are clearly some misidentifications of species and genera (e.g., Eoconuloides wellsi, Helicostegina, Heterostegina ocalana). Detailed investigations of orientated sections of LBFs from two of the limestones have been given by Cole (1960) and Robinson (1996) and these can be placed within the zonal scheme presented here. The Flamands Limestone (1) of Caron et al. (2019) at Pointe Lézarde yields Eolepidina antillea (it is the type locality: Cole, 1960; Robinson, 1996), which allows assignment to ABZ8 (Fig. 7 ); whereas the Lurin Limestone (5-6) of Caron et al. (2019) yields E. subplana (Robinson, 1996) and can be assigned to ABZ9 (Fig. 7 ). Assuming that the LBFs are not reworked, this would suggest that the limestone sequence on Saint Bartholomew is solely of mid to mid-late Lutetian age (although higher levels could be represented in the upper part of the Lurin Limestone) and a study of orientated sections of LBF from other limestone horizons is needed. Miocene limestones occur above this interval but are not relevant to this work.

5.8. Trinidad. It is difficult to apply the zonation scheme to Trinidad because of the thin successions of Eocene rocks and the extensive reworking of specimens of LBFs. This is clearly seen in Soldado Rock (e.g., Caudri, 1975, 1996) where there are extensive slump deposits and the reworking of older foraminifers into younger layers (Kugler & Caudri, 1975). Ignoring the slump masses, the first appearances of specific foraminifers can be used to suggest some correlations for Soldado Rock. The presence of Operculinoides spp. in lower bed 3 suggests ABZ7 or younger, whereas the absence of typical Lutetian markers suggests that Bed 3 may be as young as Bartonian. The appearance of Pliolepidina tobleri in bed 7 would suggest a correlation with ABZ14. Because of the extensive reworking, we do not show Trinidad on Fig. 7 .

5.9. Barbados. The Scotland District of Barbados also contains Eocene LBFs, but as with Trinidad there is extensive reworking. The Murphys Beds however yield T. senni in association with H. minor without Operculinoides spp. or Nummulites spp. (Brönnimann, 1950). This would suggest an assignment to either ABZ4 or ABZ6. Paleocene Ranikothalia occur throughout (Cizancourt, 1948; Caudri, 1972) indicating extensive reworking from the Paleocene. As a consequence, we do not show Barbados on Fig. 7 .

6. Classification of LBFs and taxonomic notes

Previous classifications of the orbitoidiform foraminifers considered here can be found in Adams (1987), Loeblich and Tappan (1988), Butterlin (1988), and BouDagher-Fadel (2018), with the phylogeny suggested by Barker and Grimsdale (1936) having been generally accepted in all subsequent works. With regard to the Lepidocyclininae, previous workers (e.g., Grimsdale, 1959; Vlerk, 1959; Sirotti, 1983; Butterlin, 1987) have suggested different clades. Our studies of the Asterigerinoidea Orbigny (forms surrounding Amphistegina, Lepidocyclina and their allies) clearly demonstrate that the current classification is polyphyletic, and also that the simple phylogenetic scheme presented by Barker and Grimsdale (1936) is too simplistic, ignores the different forms of test construction in the various groups, and thus should not be the basis for the classification of these foraminifers into families and subfamilies. Here, we introduce a revised classification (based on foraminiferal test morphology and our detailed stratigraphic occurrences) that we think provides a suitable monophyletic scheme and gives an appropriate classification rank to the various clades. We feel that the current classification of taxa into the Helicolepidininae and Lepidocyclininae is not appropriate and prefer to base families on clearly defined phylogenetic series regardless of whether orbitoidiform growth has not or has been achieved. We do not formally describe each genus or species (unless new), but provide synonyms, diagnostic criteria, and discussions that hopefully allow genera and species to be properly identified. We illustrate numerous specimens (some well preserved and some more poorly preserved) of each species so that there is no ambiguity in our specific interpretations. For ease of reference, we also list the bewildering array of synonyms at the generic level that have been previously introduced. The classification scheme adopted for the taxa we use in our zonation is laid out below.

6.1. Terminology

We follow standard guides on terminology for the classification of the foraminifers described here (e.g., Adams, 1987; Hottinger, 2006). We discuss some of these terms in more detail below.

The terms used for the orientation of the tests of foraminifers are not uniform between different morphologies, and separate schemes are used for trochospiral and orbitoidiform taxa (Hottinger, 2006). In the Helicosteginidae fam. nov. the test is trochospiral, and in the early forms (e.g., Helicostegina wellsi), one side of the test is elevated, and the other side of the test is flat or even concave (Fig. 8 .1). In Helicostegina wellsi, the proloculus is situated close to the flat (or concave) side of the test (Fig. 8 .1), which by definition makes this the 'spiral' or dorsal side of the test (Hottinger, 2006). In H. wellsi the subsidiary chamberlets are formed within the alar prolongations (or replace the septa forming the alar prolongations) on the 'umbilical' or ventral side of the test (Fig. 8 .1). In later species of Helicostegina (e.g., H. gyralis, H. dimorpha, H. jeannemairae sp. nov.) the ventral side of the test can be recognized by the layer of subsidiary chamberlets (Fig. 8 .2-3). In forms with an orbitoidiform test, such as the Lepidocyclinidae, there is an equatorial layer with lateral chamberlets on both sides (Fig. 8 .4). It is only in the earliest forms (Eulinderina guayabalensis) that the proloculus lies closer to one (dorsal) side of the test and there is a hint of low trochospiral coiling (Fig. 8 .5); in later forms the dorsal and ventral sides of the test cannot be distinguished and an equatorial layer is overlain by two sets of lateral chamberlets (Fig. 8 .4). Thus the subsidiary chamberlets in the Helicosteginidae are equivalent to one set of lateral chamberlets as developed in the Lepidocyclinidae. Thus the 'complete flange of equatorial chamberlets' described in H. dimorpha by Adams (1987, p. 294) represents a cut passing from the primary spire into the subsidiary layer of ventral chamberlets. In later species (e.g., H. jeannemairae sp. nov., H. jamaicensis sp. nov.) a true flange of equatorial chamberlets is developed from the primary spire and a subsidiary layer of ventral chamberlets (which have the same morphology as a true equatorial layer) is developed on the ventral side giving the flange a two-layered structure (Fig. 8 .3).

Following Adams (1987, p. 289) we use the terms proloculus and deuteroloculus for the first two chambers in spiral forms (Asterigerinoidea, Nummulitidae) where the second chamber (deuteroloculus) has only one aperture or stolon (R = 3 or more; Fig. 8 .6). We use the term protoconch and deuteroconch in forms where the deuteroconch has two (R = 2) apertures or stolons (Fig. 8 .7). In many of these forms (whether they consist of a proloculus and deuteroloculus or a protoconch and a deuteroconch) the first two chambers frequently have a thicker and darker wall than the succeeding chambers and this represents the embryo (e.g., Adams, 1987; Krüger et al., 1997; Langer & Hottinger, 2000). We use the form of the embryo and subsequent growth as generic and specific characters. In axial sections either the first two chambers (Figs. 8.1-2, 8.5), or one of the first two chambers may be visible (Fig. 8 .3-4).

We use the term chamber for successive growth stages of spiral foraminifers. In some forms, the alar prolongations of the chambers are divided up into smaller chamberlets on the ventral side of the test in trochospiral forms. In some trochospiral forms, the alar prolongations of the chambers on the ventral side of the test interfinger with short ventral chamberlets (Amphistegina, Tremastegina: Fig. 8 .8); the term stellar chamberlets has been used for these short ventral chamberlets (Hottinger, 2006), but may not be appropriate and they may have had a separate origin. We tentatively use the term stellar chamberlet here pending further research on these forms. In other trochospiral forms (e.g., Helicostegina gyralis), the alar prolongations on the ventral side of the test are divided up into a series of subsidiary chamberlets (Fig. 8 .9) connected by a series of ventral stolons. In this case the lumina of the chambers with undivided dorsal alar prolongations can be seen interfingering with the subsidiary chamberlets on the ventral side of the test (H. minor, Fig. 8 .9). In later forms (H. jeannemairae) the ventral layer of subsidiary chamberlets passes into a ventral sheet of chamberlets which mimics the growth in the equatorial layer of orbitoidiform foraminifers (Fig. 8 .3). In contrast, the equatorial layer of orbitoidiform foraminifers has lateral chamberlets on both sides (Fig. 8 .4) of the equatorial layer which are separated by irregular or regular walls and connected by diagonal stolons (Hottinger, 2006).

The equatorial layer in more primitive members of many megalospheric forms of various lineages starts with a proloculus and a deuteroloculus followed by a primary spire of equatorial chambers (Fig. 9 .1). In many forms, the primary spire persists to the edge of the test, but in others the spire terminates in a chamber (R) with two apertures (Fig. 9 .1) from which a spiral and a counterspiral of equatorial chambers encircle either the embryo (of two chambers) or the earlier chambers of the initial spire (following the embryo) and terminate in a closing chamber (Fig. 9 .1). Each chamber in these two spires has two apertures which gives rise to equatorial chamberlets and initiates orbitoidiform growth. In many lineages, the length of the primary spire shows a progressive reduction over time (nepionic acceleration); this may give rise to one PAC (in which case a proloculus and deuteroloculus are still present and R = 3) or two PACs (in which a protoconch and deuteroconch are present and R = 2; Fig. 9 .2). Once an embryo and two PACs are developed, either three of four unequal spires of peri-embryonic chambers encircle the embryo, with two closing chambers; Fig. 9 .2). In more primitive forms the three or four spirals are asymmetrical (Fig. 9 .2), but over time four symmetrical spirals of peri-embryonic chambers encircle the embryo (Fig. 9 .3). Subsequently, some forms develop adauxiliary chambers (Fig. 9 .4) which are connected to the protoconch and/or the deuteroconch by stolons; in these forms short spirals and multiple closing chambers are developed. This progressive evolution of the equatorial layer is repeated in many lineages and this gives rise to homeomorphs which are not phylogenetically related. We use other criteria to separate these different clades.

The size of the embryo has commonly been used in separating species in the Americas (e.g., Cole, 1952; Barker & Grimsdale, 1936; Colletti et al., 2019). However, the size of the embryo has both phylogenetic significance and environmental significance, with variations due to salinity, temperature, and water depth (e.g., Nigam & Rao, 1987; Bryan, 1995; Yu et al., 2016). Given the variation in the size of the embryo in populations of the same species from different environments, the size of the embryo, in itself, is not diagnostic for species identification (we show examples of this in our work here). Notwithstanding this, the size of the embryo helps in distinguishing between different lineages (e.g., Lepidocyclina and Orbitoina) and needs to be taken into consideration (with other characters).

Equatorial chamberlets in an orbitoidiform growth plan are developed from the primary spiral(s) of chambers or from the peri-embryonic equatorial chambers (Fig. 9 .5). Their morphology is dependent on shape (i.e., how the chambers interlock), the radial distance versus the concentric distance of the connecting walls, and the position of the connecting stolons between radially adjacent equatorial chamberlets. The form of equatorial chamberlets in American lepidocyclinids has been investigated by Hohenegger and Torres Silva (2020), but we do not consider this in detail in this work. We illustrate the equatorial chamberlets in the figures of the species that follow.

Lateral chamberlets (seen in axial sections) may form regular stacks (tiers) or alternate. They may be separated by thin walls, thick walls or piles (pillars). The surface of the test may be smooth, have irregular anastomosing patterns representing the walls between the lateral chamberlets, or have piles (pillars or pustules). We illustrate the form of the lateral chamberlets in the figures for each taxon below. In some forms (e.g., Tremastegina, Helicostegina), small pustules may follow the septa on the dorsal and/or ventral side/s of the test. In some forms, the ventral side of the test may be divided into chamberlets or stellar chamberlets (Fig. 8 .8).

Various measurements have been made on the embryos of LBF. This has varied both by geographical region (e.g., Tethys versus the Americas) and by group. For instance, in Cretaceous Orbitoides (Fig. 10 .1) the length and width of the nucleoconch excluding the thickness of the walls has been measured (e.g., Hinte, 1965; Gorsel, 1978). For orthophragmines (Fig. 10 .2) the thickness of the wall has been included in the measurements of the width of the protoconch and the deuteroconch (e.g., Less, 1987). In contrast, for Tethys lepidocyclinids (Fig. 10 .3) the width of the protoconch and deuteroconch (including half the thickness of the walls) has been used (e.g., Vlerk, 1959; Drooger, 1993). Subsequent work has tended to follow a particular scheme for a particular group so as to allow comparisons (e.g., Mitchell, 2006; Less et al., 2018; Özcan et al., 2018, 2021a). For American lepidocyclinids and their allies, different measurements have been made. These are typically the internal widths (Pwi and Dwi) and internal lengths (Pli and Dli) of the protoconch/proloculus and deuteroconch/deuterolocus (i.e., excluding the wall) and the length and width of the embryo (including the wall: Ele, Ewe). Some workers have only given parameters for the embryo, Ele and sometimes Ewe (e.g., Gorter & Vlerk, 1932; Gravel, 1933; Vaughan, 1933; Cole & Ponton, 1934; Ritten & Vermunt, 1932;Rutten, 1935; Eames et al., 1962; Caudri, 1996; Serra-Kiel et al., 2007); some workers have only given lengths (Pli, Dli) or widths (Pwi, Dwi) of the proloculus/protoconch and sometimes deuteroloculus/deuteroconch (e.g., Frost & Langenheim, 1974; Robinson & Jiang, 1995); and some workers both (e.g., Berry, 1930, 1932; Barker, 1932, 1934; Barker & Grimsdale, 1936; Cole & Bermudez, 1944; Cole & Vaughan, 1945; Cole, 1952; Butterlin, 1961; Cole, 1963; Eames et al., 1968; Colletti et al., 2019). In the Tethys the terms height and width have been used, whereas in the Americas the terms length and width have been used. In order to compare our results with those that have been made before in the Americas, we provide details of the width of the protoconch/proloculus (Pwi) and the length of the embryo (Ele) herein. These measurements are shown in Fig. 10 .4. In the Eocene taxa we are dealing with, the proloculus/protoconch and deuteroloculus/deuteroconch are generally of similar sizes; we therefore do not provide parameters for Pli, Dli nor Dwi at the present time. To compare the tightness of coiling in the Helicosteginidae, we also construct coiling diagrams (e.g., Robinson, 1993) using measurements every 90° (Fig. 10 .5).

6.2. Species concepts

Species concepts in American LBF have varied greatly among various specialists using orientated equatorial sections, with some accepting very broad species definitions (e.g., Cole, 1963; Frost & Langenheim, 1974; Robinson & Wright, 1993), while others have narrow species definitions (Cole, 1956, 1958a, 1958b; Robinson & Jiang, 1995, Robinson, 1996, 1997). This is partly due to the material available for study, with random sections in rock samples being much more difficult to identify than isolated specimens from which orientated thin sections can be cut. It may also be due to the reworking of specimens where populations from different stratigraphic levels have been mixed together (e.g., Cole, 1963; Caudri, 1975, 1996). In order to achieve a high-resolution biostratigraphy, multiple isolated specimens are clearly preferable, because this allows detailed measurements of features of the test to be made. We have found that there is generally a uniformity in equatorial sections and 4 or 5 specimens generally allow species to be identified, but (many) more specimens are desirable.

There are also the differences between a typological approach and a statistical approach to be considered. This can be illustrated by considering the evolutionary development within Eolepidina in specimens from populations used in this study. Consider, for instance a typological approach to a set of specimens using the criterion R. A 'population' might include specimens with R = 5, R = 4, and R =  3, which, under a typological approach, might be assigned to Eo. antillea and Eo. subplana, and Eo. gardnerae, respectively. Under a statistical approach, the mean value of R can be calculated, which, with values for R defined for each species) would allow a population to be placed in a single species (so in the previous population if the mean R was 4.2, it would allow the population to be assigned to Eo. subplana). Our samples, albeit often with only a limited number of specimens, calibrated using nannofossils, demonstrate that mean values of R are very useful biostratigraphy, and we define limits for R for different species in this paper.

In the following sections, we describe the various families, genera, and species of LBF that we use in the zonation scheme. We illustrate each species with numerous specimens (some well preserved, others more poorly preserved) so that our species concepts can be used by other workers.

Superfamily ASTERIGERINOIDEA Orbigny, 1839

Family AMPHISTEGINIDAE Cushman, 1927

Remarks. Loeblich and Tappan (1988) included only Amphistegina in the family Amphisteginidae. We place the genus Butterliniana gen. nov. also in this family, but further work may indicate that it belongs elsewhere.

Genus Amphistegina Orbigny, 1826

Type species. Amphistegina quoyii Orbigny, 1826 (by monotypy: Loeblich & Tappan, 1988, p. 610) from the Recent of the island of Rawack, Papua province, Indonesia.

Remarks. Species such as Amphistegina lopeztrigoi Palmer and Nummulites parvula Cushman have previously been placed in Amphistegina; in this work we place these taxa in other/new genera. Caudri's (1975) species grimsdalei (and associated forms), from the Eocene of Trinidad might be an Amphistegina, but this needs more work to confirm and is beyond the scope of this paper.

Genus Butterliniana Mitchell, E. Robinson & Özcan gen. nov.

Type species. Amphistegina cubensis Palmer, 1934, from the Eocene of Cuba.

Diagnosis. Low trochospiral coil, embryo consists of a proloculus and deuteroloculus followed by a spiral arrangement of chambers; stellar chamberlets absent; stolons between chambers without countersepta.

Origin of name. Named after Jacques Butterlin for his work on the foraminifers of the Caribbean region.

Discussion. Carefully prepared thin sections cut perpendicular to the axis on the ventral side (although invariable close to the thick wall of the test and difficult to illustrate) show that the septa divide a short distance from the umbo, but typical stellar chamberlets, as seen in Amphistegina are not present. Butterliniana differs from Amphistegina in lacking stellar chamberlets, and also in having simple stolons between chambers without the development of countersepta. We retain Butterliniana in the Amphisteginidae for the present but recognize that the lack of stellar chamberlets might require placement in a new family in the future. This can only be determined once studies of well-preserved, early forms of the Amphisteginidae have been completed.

Butterliniana cubensis (Palmer, 1934)

(Figs. 11 , 12 )

Previous descriptions. Palmer, 1934, p. 256-257, Figs. 15-16, Pl. 15, fig. 2; Butterlin, 1970, p. 293, Pl. 3, figs. 5-6; Robinson & Wright, 1993, p. 309, Figs. 18.3-5.

Remarks. Cushman (1919) figured four specimens as Nummulites parvula (his Pl. 4, figs. 3-6) from St. Bartholomew which are preserved in the Smithsonian. The Smithsonian catalogue records his Pl. 4, fig. 3 (USNM.MO.328183) as holotype and the other specimens (his Pl. 4, figs. 4-6: USNM.PAL27143, PAL.27143 [number duplicated], and PAL.321801) as paratypes. Cushman (1919, p. 51) clearly stated that his Pl. 4, fig. 3 was the type specimen, and that the other three specimens may not represent the same species; as such these should not be considered part of the type series or as paratypes. The species 'parvula' should therefore only be interpreted based on the holotype. The holotype is a natural, approximately equatorial section in which no indication of the original ornament of the test is visible. It has a diameter of 1.75 mm (as measured from the plate) and shows 5 whorls, with maybe 20 chambers in the final whorl.

Cole (1958a, p. 201, Pl. 25, figs. 17-19; 1969, p. 80, Pl. 16, figs. 1, 3, 5-7, 10-12; Pl. 17, fig. 5) interpreted 'parvula' based on Cushman's (1919) Pl. 4, figs. 3, 6, probably fig. 4, but not fig. 5. He interpreted 'parvula' as a senior synonym of Amphistegina lopeztrigoi Palmer, 1934. Exterior features of A. lopeztrigoi were given by Palmer (1934) and interior features of topotype material were provided by Cole and Gravell (1952). Cole's (1958a) Pl. 25, figs. 17-18, show thick-tested forms in axial section with small chambers which are similar to Cushman's (1919) Pl. 4, fig. 5; whereas his Pl. 25, fig. 19 (and Cole, 1969, Pl. 12) shows an equatorial section with 4¼ whorls, which appears to show forwardly orientated (away from the protoconch) septal necks (countersepta) and 17 chambers in the final whorl. Cole (1958a, 1969) does not provide information on the external ornament of specimens, but he included Cushman's (1919) Pl. 4, fig. 6, in synonymy, and this form clearly lacks pustules and is smooth.

A second interpretation of Cushman's (1919) Nummulites parvula has followed Butterlin (1970), p. 294, Pl. 3, figs. 5-6 (e.g., Robinson & Wright, 1993, p. 309, Figs. 18.3-5; Serra-Kiel et al., 2007, p. 369, Pl. 2, figs. 11-12, 15). This interpretation effectively includes Cushman's (1919), Pl. 4, fig. 5, and Cole (1958a), Pl. 25, figs. 17-18. This form was named Amphistegina cubensis by Palmer (1934), p. 256-257, Figs. 15-16, Pl. 15, fig. 2.

Loeblich and Tappan (1988, p. 611) placed Cushman's Pl. 4, figs. 3, 6 (but not 5), in synonymy with 'Amphistegina' lopeztrigoi Palmer, 1934. Clearly Cushman's, 1919, Pl. 4, fig. 3, is not related to what we name here as Butterliniana, but we are unable to directly assign this specimen to lopeztrigoi because of its very weathered nature.

It is clear that an interpretation of Cushman's (1919) species Nummulites parvula is made difficult because the name-bearing type is so poorly preserved. The holotype might be close to Amphistegina lopeztrigoi as maintained by Cole (1958a, 1969), but would appear to lack ornament (as judged from other illustrated sections that might be attributed to that species) and it is difficult to interpret as that species. Similarly, one of Cushman's (1919) figured specimens (his Pl. 4, fig. 5) is equivalent to the two axial sections illustrated by Cole (1958a, Pl. 25, figs. 17-18) and was named Amphistegina cubensis by Palmer (1934). The name parvula is therefore surrounded by nomenclature issues that make it an ambiguous species of debatable affinity. It seems the best course of action would be to prepare a submission to the ICZN for the suppression of the species parvula in the combination Nummulites parvula Cushman and that Amphistegina lopeztrigoi Palmer and Amphistegina cubensis Palmer should be placed on the list of accepted names. A case will be prepared for this. We use the specific name 'cubensis' for this species here.

Butterliniana cubensis is a distinctive long-ranging species appearing in the late Ypresian and apparently becoming extinct near, or at, the top of the Eocene. Specimens vary from forms which are relatively symmetrical in axial section (e.g., Cushman, 1919, Pl. 4, fig, 5; Butterlin, 1970, Pl. 3, figs. 5-6; Cole, 1958a, Pl. 25, figs. 17-18; Robinson & Wright, 1993, Figs. 18.3-5; Serra-Kiel et al., 2007, Pl. 2, figs. 11-12, 15) to forms that are strongly asymmetrical (e.g., Palmer, 1934, Pl. 15, fig. 2). Asymmetrical forms are present at some levels in the mid and late Eocene in Jamaica, but without an analysis of populations it is not possible to determine if more than one species might be recognizable at this time. The number of chambers in the final whorl also varies significantly; without the study of populations it is not possible to determine if this is due to natural variation or might be stratigraphically significant.

Family HELICOSTEGINIDAE Mitchell, E. Robinson & Özcan fam. nov.

Diagnosis. The family is characterised by involute to secondarily orbitoidiform taxa which have an early stage of trochospiral coiling. There is a progressive development from multiple stolons into a spiral sheet of subsidiary chamberlets on the ventral side in early forms. The development of chamberlets in the peripheral primary spire may lead to the development of an equatorial layer and in some forms a retrovert chamber may develop leading to orbitoidiform growth. In some forms the spiral sheet of equatorial chambers may form a second 'orbitoidiform' layer composed of orbitoidiform chambers on the ventral side of the equatorial layer.

Genera included. Tremastegina Brönnimann, 1950; Helicostegina Barker & Grimsdale, 1936.

Remarks. This family shows the development of advanced forms which either show low trochospiral growth with an equatorial layer or orbitoidiform growth with the ancestral form being Tremastegina. Below, we synonymise Eoconuloides with Helicostegina, and this leaves the genus Boreloides (type species Boreloides cubensis Cole & Bermúdez, 1947) as the only genus in the family Boreloididae Reiss, 1963, a family which we regard as of uncertain affinity at the present time. Further research on Boreloides is needed to determine its affinity.

Table 15 shows the length (including the walls) of the first two chambers (embryo consisting of a proloculus and a deuteroloculus; Ele) for different species belonging to the family. Table 16 shows the width (excluding the walls) of the first chamber (proloculus; Pwi) for different species belonging to the family. Figure 13 tracks the evolutionary development of the family. Figure 14 shows the size of the embryo and protoconch in selected species of Helicostegina. Figure 15 shows a scatter plot of protoconch width versus embryo length, and Fig. 16 shows coiling diagrams for selected species of Helicostegina.

Table 15: Measurements (in μm) of the lengths of the first two chambers (embryo) including the chamber walls (Ele) for species of Tremastegina and Helicostegina from Jamaica (WL numbers) and reported values (¹) and our measurements from figures (²) from other locations.

Table 16: Measurements (in μm) of the width of the first chamber (proloculus) excluding the chamber walls (Pwi) for species of Tremastegina and Helicostegina from Jamaica (WL numbers) and reported values (¹) and our measurements from figures (²) from other locations.

Genus Tremastegina Brönnimann, 1950

Type species. Amphistegina senni Cushman (in Vaughan, 1945) from the Eocene of the Murphys Beds, Upper Scotland Group, Barbados.

Diagnosis. The genus is characterised by 'stellar' chamberlets on the ventral side in combination with forwardly (away from the protoconch) directed septal necks (counter septa) around the stolons between dorsal chambers. The 'stellar' chamberlets are connected to each other by a series of basal stolons, with the most peripheral stolon(s) connecting to the dorsal chamber (Fig. 18 ).

Remarks. Amphistegina differs from Tremastegina in having backwardly (towards the protoconch directed septal necks: O'Herne, 1974), whereas in Tremastegina the septal necks (counter septa) point away from the protoconch (Barker & Grimsdale, 1936; Fig. 13). These septal necks reflect the condition seen in other members of the Helicosteginidae fam. nov., but Tremastegina lacks the spiral sheet of subsidiary chamberlets developed on the ventral side in Helicostegina. Tremastegina is further distinguished from Amphistegina by the presence of the basal stolons, which we regard as the first step in the development of the spiral sheet of chamberlets seen in Helicostegina. We therefore regard Tremastegina as a primitive member of the Helicosteginidae. We provisionally recognize Tremastegina senni (Cushman in Vaughan, 1945) and T. lopeztrigoi (Palmer, 1934) as separate species and discuss this below.

Tremastegina senni (Cushman in Vaughan, 1945)

(Fig. 18 )

Previous descriptions. Cushman in Vaughan, 1945, p. 49, Pl. 19, figs. 1-4; Brönnimann, 1950, p. 257-265, Figs. 1-7.

Recognition. Tremastegina senni is a relatively small species of Tremastegina (typically less than 1 mm in diameter) with a small embryo.

Remarks. Cushman (in Vaughan, 1945, p. 49) stated that the holotype of A. senni came from Senn station S. 34B, but in the plate descriptions (Vaughan, 1945, Pl. 19, figs. 1-4) all the specimens were stated to come from Senn station S. 34C. Brönnimann (1950, p. 255) stated that the holotype did come from Senn station S. 34C, and that the statement by Cushman (in Vaughan, 1945, p. 49) was an error (this was probably communicated to Brönnimann by Senn who sent him material of the species to study from station S. 34C in 1946: Brönnimann, 1950, p. 255). Therefore, Brönnimann's (1950) descriptions refer to topotype material of T. senni.

T. senni was originally compared to T. lopeztrigoi by Senn (1940) who referred to the species as Amphistegina cf. lopeztrigoi following a personal communication from Vaughan (Senn, 1940, p. 1550). T. senni is, however, significantly smaller (about half the size) than T. lopeztrigoi, although this may be partly due to the loss of the final whorl in many specimens (Brönnimann, 1950, p. 259). Cole and Gravell (1952) regarded size of relatively little importance and synonymised 'A'. senni with 'A'. lopeztrigoi. Brönnimann (1950, p. 263) gives the diameter of the first chamber as 34 to 58 μm in 'A.' senni. This compares to the internal size of the first chamber in T. lopeztrigoi of 140-160 μm with walls from 10-30 μm thick (Barker & Grimsdale, 1936, p. 236). The two species may therefore be distinct.

Vaughan (1945) used the presence of 'A'. senni to argue for a middle Eocene age for that part of the Scotland Group in Barbados based on the range of 'A'. lopeztrigoi in Mexico (Barker & Grimsdale, 1936) and Florida (Cole, 1942, 1944). Both Caudri (1972) and Jones (2009) also considered that T. senni was part of the Middle Eocene fauna. However, Tremastegina has a relatively long range first appearing in the late Paleocene (Robinson & Wright, 1993) and extending up to ABZ7 in Jamaica. As noted by Cizancourt (1948) and Brönnimann (1950), the LBFs from the Scotland Group of Barbados have been strongly abraded; Brönnimann (1950, p. 259) stated that most of the specimens of T. senni had either lost their outer whorl or had it significantly worn away showing the interior structure. Furthermore, Ranikothalia occurs in the same deposits as T. senni in Barbados, and the former species is considered reworked from the late Paleocene (Caudri, 1972). Small specimens that could be attributed to T. senni occur in the late Paleocene Nonsuch Limestone of Jamaica (Robinson & Wright, 1993) adding potential support that T. senni may be an older species. There could be a progressive size increase (as well as increase in the size of the proloculus) from the late Paleocene to the mid Lutetian, which may make sequential populations of Tremastegina useful for biostratigraphy. However, until a section is found where the evolution of Tremastegina can be determined, and the variation in the size of the early chambers can be measured, it is impossible to determine if T. senni is a Paleocene, Ypresian or early Lutetian species.

Tremastegina lopeztrigoi (Palmer, 1934)

(Fig. 19 )

Previous descriptions. Palmer (1934), p. 255, Pl. 15, figs. 6, 8; Barker & Grimsdale, 1936, p. 233, Pl. 30, figs. 1-2; Pl. 32, figs. 1-3; Pl. 34, fig. 1; Pl. 38, fig. 3; Robinson & Wright, 1993, p. 311, Figs. 19.6-9; Hottinger, 2006, Figs. 33E-G.

Recognition. Tremastegina lopeztrigoi is a relatively large species of Tremastegina (typically more than 1 mm in diameter) with a large embryo.

Remarks. Cushman (1919) erected Nummulites parvula which Cole (1969) regarded as a senior synonym of A. lopeztrigoi. As discussed above, we regard the holotype of N. parvula as indeterminate and suggest suppression of the specific name. T. lopeztrigoi ranges at least from the mid early Eocene to the mid Lutetian (ABZ7) in Jamaica. It is the nominate taxon for ABZ3, which represents a low-diversity interval for LBFs other than for dictyoconids and orthophragmines.

Genus Helicostegina Barker & Grimsdale, 1936

Synonyms. Helicolepidinoides Tan, 1936 (type species Helicostegina gyralis Barker & Grimsdale, 1936) [junior synonym]; Eoconuloides Cole & Bermúdez, 1944 (type species Eoconuloides wellsi Cole & Bermúdez, 1944) [junior synonym].

Diagnosis. A lenticular to conical helicosteginid with an involute test throughout growth; the primary spire extends to the edge of the test; the alar prolongations on the dorsal side are separated by radial septa; the alar prolongations on the ventral side are represented by strongly spiral septa that merge into the spiral sheet of subsidiary chamberlets in early species, but are not recognizable in later species. The subsidiary chamberlets on the ventral side are separated by arcuate (in the direction of spiral growth) chamberlet walls. The youngest forms show the arcuate septa of the subsidiary chamberlets directed in the direction of growth, whereas in later species they show a radial (orbitoidiform) arrangement.

Remarks. Barker and Grimsdale (1936, p. 234) clearly described the morphology of this genus; their description states "Test multichambered, the earliest chambers coiled in an involute trochoid spire, the chambers of the later coils subdividing ventrally into two or more subsidiary chambers or chamberlets" and "In … H. gyralis the chamberlets are restricted to the ventral region of the test, upon which they form a continuous layer but in … [H. dimorpha] this layer of chamberlets extends peripherally as a more or less undulating compressed flange one cell thick, almost surrounding the early spiral portion which forms eccentric umbones on the dorsal and ventral surfaces of the test." The latter statement is a little misleading (as shown by Barker & Grimsdale, 1936, Pl. 32, figs. 6-7) where a narrow double-layered (in part) flange is developed, with one layer corresponding to the ventral sheet of subsidiary chamberlets and the other corresponding to the alar prolongations on the dorsal side of the test or to the primary spire or equatorial chambers. Certainly in more derived forms (e.g., H. jeannemairae sp. nov.) this would appear to be developed from the primary spire of equatorial chambers.

The relationships between Helicostegina dimorpha and H. gyralis have always been difficult to determine because of the figures of H. dimorpha given by Barker and Grimsdale (1936); yet all specimens of H. dimorpha came from Jenny station No. 1592, and all specimens of H. gyralis came from Jenny station No. 1573. Axial sections of H. dimorpha (Barker & Grimsdale, 1936, Pl. 32, figs. 6-7) show forms in which a thickened wall extends to the margin of the shell, and which contain a few peripheral subsidiary chamberlets in the outer whorl. Yet equatorial sections (Barker & Grimsdale, 1936, Pl. 34, figs. 7, 9) seem to show a relatively short spire and a 'wide flange' of orbitoidiform chambers. However, if straight lines are drawn on the axial sections corresponding to the equatorial plane, the trochospiral coiling of the test means that the equatorial sections will not pass through the spire across the whole of the section, but will also include the peripheral ventral portions of the test with subsidiary chamberlets. This is confirmed from numerous sections of Jamaican specimens, which have identical axial sections, but in relatively symmetrical tests (i.e., very low trochospiral coiling) the complete spire can be seen in equatorial section.

Adams (1987) maintained H. gyralis and H. dimorpha as separate species, but stated (p. 294) that "Cole (1960) considered that H. gyralis should be placed in synonymy with H. dimorpha …", but that "These were not, however, from the same sample or locality, and while it is true that temporal gradation between these species may occur, two individuals seen in median section do not constitute proof, neither does their synonymy warrant the possible loss of important biostratigraphical information." Yet, Adams himself figured specimens from the type locality of H. dimorpha (Jenny station No. 1592) as both H. dimorpha (Adams, 1987, Pl. 1, figs. 1-3) and H. gyralis (Adams, 1987, Pl. 1, figs. 4-8), and a specimen from the type locality of H. gyralis (Jenny station No. 1573) as H. gyralis (Adams, 1987, Pl. 1, fig. 9), thus confusing the issue.

Furthermore, Adams (1987) description of the genus Helicostegina states that the equatorial layer passes into an orbitoidiform layer with a chamber having a retrovert aperture. This is quite a different interpretation to the description given by Barker and Grimsdale (1936) who clearly stated that the lateral chambers on the ventral side of the test are divided into subsidiary chamberlets. We agree with the interpretation of Barker and Grimsdale (1936) and illustrate material herein that demonstrates this without ambiguity. Adams (1987) erroneous interpretation of the structure of H. dimorpha led him to place forms that had been separated as Helicolepidina (e.g., H. spiralis) within Helicostegina. However, the test construction in Helicolepidina is completely different to that of Helicostegina, and the two genera cannot be closely related.

Tan (1936) established the genus Helicolepidinoides for H. gyralis. Tan (1936), however, failed to realise that the difference between H. dimorpha and H. gyralis is due to the equatorial sections cut through a trochospiral test, and this is clearly indicated by the axial sections of H. dimorpha as illustrated by Barker and Grimsdale (1936) as well as by the topotype material illustrated by Adams (1987, Pl. 1, figs. 1-8). Since there is no major constructional difference between H. dimorpha and H. gyralis (other that the extension of the chamberlets into chamber lumina), and particularly since random sections are difficult to identify as either H. gyralis or H. dimorpha, there is no need for the genus Helicolepidinoides as pointed out by Adams (1987). We therefore synonymise Helicolepidinoides with Helicostegina herein.

Eoconuloides wellsi shows a similar test construction to H. gyralis and only differs in a stronger trochospiral coiling giving rise to a conical test; we therefore synonymise Eoconuloides with Helicostegina here. Cole and Bermudez (1944, p. 340) stated that Eoconuloides wellsi lacked the ventral subsidiary chamberlets seen in Helicostegina, yet they are clearly present in the holotype as figured here (Fig. 20.3). We do not regard the conical shape as of generic significance, particularly as microspheric specimens of Helicostegina minor sp. nov. show a weakly conical test. We therefore transfer E. wellsi to the genus Helicostegina, and synonymise Eoconuloides with Helicostegina herein.

The two more derived species of Helicostegina, H. jamaicensis sp. nov. and H. jeannemairae sp. nov., develop orbitoidiform growth in the equatorial layer. While this could be a justification for the establishment of new genera, they are clearly the terminations of two separate lineages. For the present, we feel it is best to leave these two terminal species within Helicostegina, particularly as the trochospiral coiling makes it difficult to distinguish orbitoidiform growth from spiral growth when a spiral layer of subsidiary chamberlets is present. We distinguish species here based on three criteria: the tightness of the coiling, the size of the embryo, and the distribution of subsidiary chamberlets.

Helicostegina wellsi (Cole & Bermúdez, 1944)

(Fig. 20 )

Previous descriptions and figures. Cole & Bermúdez, 1944, p. 341-342, Pl. 1, figs. 4-10; Cole & Gravell, 1952, p. 713, Pl. 92, figs. 1-10; Robinson & Wright, 1993, p. 311, Pl. 18.1-2, Pl. 19.1-5.

Recognition. The species is easily recognized because of its strongly conical form with the ventral side being highly expanded to form a tall pointed conical umbo and the dorsal side being gently convex, flat, or even gently concave. The ventral surface of the test is covered by small papillae, which are difficult to see in poorly preserved specimens. There are about 3½ whorls in megalospheric specimens, and the final whorl has about 20 or 21 chambers. The embryo (the first two chambers - proloculus and deuteroloculus) is relatively small with a length (including the walls) of 100 to 255 μm.

Synonyms. Amphistegina elliotti Cushman & Stainforth, 1946 [junior synonym].

Remarks. Helicostegina wellsi has a test constructed in exactly the same way as H. gyralis and H. minor sp. nov. and they often occur in the same samples; the only difference is that H. wellsi is conical with a flat, weakly concave or very weakly convex dorsal side, whereas H. gyralis is lenticular. H. wellsi is a highly distinctive form which ranges from forms in which the axial height is slightly greater than the diameter to forms in which the axial height is twice that of the diameter. The relative height of the test may have some biostratigraphic significance, but this has yet to be worked out. H. wellsi is easily recognizable, although specimens may be rather small and missed if too coarse a sieve is used in sample preparation. The species can be recognized in the field with a hand lens. Amphistegina elliotti Cushman & Stainforth (1946) has the same morphology (their Pl. 20, fig. 5 shows the alar prolongations on the dorsal side, whereas their Pl. 20, fig. 6 shows the ventral layer of subsidiary chamberlets), and falls within the range of variation of H. wellsi. The ornamentation of H. wellsi consists of small papillae scattered across the whole surface of the test; this contrasts with the central crown of large pustules in H. gyralis (although early examples of H. minor from ABZ4 have an ornament closer to H. wellsi).

We use H. wellsi as the nominate taxon for ABZ4. H. wellsi has its peak abundance in ABZ4 and lower ABZ5 in Jamaica, but also occurs in very small numbers in the upper part of ABZ6 and in zones ABZ8 and ABZ9. The association of H. wellsi with H. gyralis (cited as Eoconuloides senni) and Operculinoides sp. in Calyx wells 50 and 50A in Trinidad (Caudri, 1996, Fig. 7) suggests assignment to ABZ7. Reworked(?) specimens also occur in the Upper Scotland Group in Barbados according to Brönnimann (1950, p. 264).

Kalia and Banerjee (1995) recorded 'Eoconuloides' wellsi from India. They recorded associated planktic foraminifers as indicating zone P12 and P14 (equivalent to ABZ9 to ABZ11), which is a partly younger age range than the range of H. wellsi in the Caribbean as indicated here (ABZ4 to ABZ9). We identify their Pl. 1, figs. 5 and 10 as Asterigerina sp., whereas other specimens are typical Nummulites. We doubt the occurrence of H. wellsi in India.

Helicostegina minor Mitchell, E. Robinson & Özcan, sp. nov.

Figs. 21 , 22 , 23 , 24 , 25 , 26 )

1945 Amphistegina senni Cushman in Vaughan, in part?

1993 Helicostegina gyralis Barker & Grimsdale; Robinson, Pl. 5, figs. 11, 13.

1993 Helicostegina gyralis Barker & Grimsdale; Robinson & Wright, p. 313-314, Fig. 18.6-8.

1993 Eoconuloides lopeztrigoi (Palmer); Robinson & Wright, in part, p. 311, Fig. 19.7 only.

Diagnosis. A species of Helicostegina with a very small proloculus followed by a smaller deuteroloculus and then a primary spiral that extends to the margin of the test; tightly coiled. Alar prolongations radial on the dorsal side and undivided; alar prolongations on the ventral side are spiral with more-or-less continuous walls and subdivided into arcuate subsidiary chamberlets.

Origin name. 'Minor' after its small proloculus.

Type specimens. Holotype: UWIGM.WL3667-5; Paratypes: UWIGM.WL3667-8, UWIGM.WL3667-7, UWIGM.WL3667-6, UWIGM.WL3667-4, UWIGM.WL3667-9; all from sample WL3667, Stettin Formation (ABZ6), Stettin, Trelawny, Jamaica.

Description. Microspheric forms have a diameter of up to 2.0 mm, megalospheric forms have a diameter of up to 1.6 mm. The ornament consists of a central crown of pustules with small pustules along the septa. Microspheric forms have a very small proloculus, with about five whorls; there are 33 chambers in the final whorl. Megalospheric specimens have the first two chambers (proloculus and deuteroloculus) with a length of 114-181 um (including the walls), with means of 129 um (sample WL4052) and 150 um (sample WL3667). The internal width of the proloculus (Pwi) ranges from 60-89 μm (Table 16). The first two chambers are followed by 3½ to 4½ whorls, with 27 to 31 chambers in the final whorl. Chambers on the dorsal side undivided with simple arcuate septa; dorsal chambers visible from ventral side (where they form the marginal series of chambers). Septa on the ventral side of the test can be traced through the subsidiary chamberlets and are strongly spirally arranged. Alar prolongations of chambers on the ventral side divided into subsidiary chamberlets, with arcuate walls, and extending up to the spire of the previous whorl. Axial sections range from relatively symmetrical to asymmetrical forms. Septa on the dorsal side are only occasionally visible in orientated axial sections, so that most of the alar prolongations on the dorsal side of the test are 'open.' In axial section, the walls of the ventral chamberlets are closely spaced and fill the alar prolongations on the ventral side of the test.

Remarks. H. minor can be distinguished from T. lopeztrigoi by the presence of subsidiary chamberlets in or replacing the alar prolongations on the ventral side, which extend up to the keel of the primary spire of the previous whorl. The well-marked septa forming the alar prolongations are retained on the ventral side such that a spiral pattern is quite obvious. This is lost in later species where a true spiral sheet of subsidiary chamberlets is developed which have a more typical orbitoidiform arrangement of the chamberlets. H. minor differs from H. wellsi by the presence of large tubercles on the surface of the test that are concentrated at the centre (crown) and the fine series of papillae that mark the courses of the septa across the flanks of the test on the dorsal side.

Helicostegina minor ranges from ABZ4 to ABZ6. It is replaced in the upper part of ABZ6 by H. gyralis. The species is recorded from Jamaica (Robinson & Wright, 1993, herein) and Barbados (Fig. 26.3).

Helicostegina cf. minor Mitchell, E. Robinson & Özcan, sp. nov.

(Fig. 27 )

Remarks. Four specimens from middle ABZ7 show an equatorial section with a small proloculus and small deuteroloculus combined with a spiral wall that extends to the margin of the test. We do not have an axial section so cannot determine if they have an incipient double layer, and we attribute these specimens to Helicostegina cf. minor sp. nov. Further material will need to be collected to determine whether this species should be retained in H. minor sp. nov. or placed in a new species transitional between H. minor sp. nov. and H. jamaicaensis sp. nov.

Helicostegina gyralis Barker & Grimsdale, 1936

Figs. 28 , 29 , 30 , 31 , 32 , 33 )

Previous descriptions and figures. Barker & Grimsdale, 1936, p. 236-237, Pl. 30, figs. 3-5; Pl. 32, figs. 4-5; Pl. 34, figs. 2-6; Pl. 37, fig. 6; Cole & Gravell, 1952, p. 713, Pl. 92, figs. 11-21.

Recognition. Helicostegina gyralis is a lenticular species, but shows a low trochospiral coil such that in many specimens it is difficult to cut true equatorial sections, and the sections often pass from the primary spire in the centre of the test into part of the spiral sheet of subsidiary chamberlets. Barker and Grimsdale (1936, p. 236) stated that the first two chambers have a length of 185-240 μm and there are about 25-30 chambers in the 'final whorl'. Measurements of populations from Jamaica are given in Tables 15 and 16. The dorsal surface of the test is ornamented with coarse piles (pillars) on the central area of the test with finer papillae aligned along the septa separating the alar prolongations/chambers. Axial sections of most specimens lack chamberlets at the periphery of the test, although a few specimens (e.g., Barker & Grimsdale, 1936, Pl. 32, fig. 4: reproduced in Fig. 28.4 here) show a few chamberlets at the periphery, but not the clear double row of chamberlets seen in H. dimorpha.

Remarks. H. gyralis can be distinguished from T. lopeztrigoi by the presence of a spiral sheet of subsidiary chamberlets on the ventral side. These can be seen during section preparation even if the resulting equatorial section apparently shows none (which is the case in relatively symmetrical specimens). H. gyralis differs from H. minor in having a larger embryo and in being less tightly coiled. H. gyralis differs from H. wellsi by its shape, and by the presence of large tubercles on the surface of the test that are concentrated at the test's centre and the fine series of papillae that mark the courses of the septa across the flanks of the test on the dorsal side.

The two populations of H. gyralis (upper ABZ6 and lower ABZ7) from Jamaica plot in different areas on Coiling Parameter vs. Embryo Length diagrams (Fig. 17). Since we have not measured the Coiling Parameter on specimens of the type population of H. gyralis from Mexico, and since the coiling parameter is highly dependent on reproduction of figures at the correct scale in publications (unless a scale bar is present), we do not separate the Jamaican populations at the present time. Pwi of specimens from upper ABZ6 is smaller than for lower ABZ7 (Table 16, Fig. 14), and the specimen of H. gyralis from Florida figured by Levin (1957) falls within the range of the upper ABZ6 population from Jamaica; this may be of stratigraphic significance and populations from Florida need to be measured to determine this.

H. gyralis ranges through the middle of ABZ6 to the lower part of ABZ7 in Jamaica. The type locality of H. gyralis is Jenny Station No. 1573 (11 km southeast of Sabaneta, state of Veracruz, Mexico: Barker & Grimsdale, 1936); this locality has also yielded Pseudolepidina trimera (Barker & Grimsdale, 1937) and Operculinoides jennyi (Barker, 1939) and can therefore be assigned to lower ABZ7. H. gyralis is widely distributed across the American region being reported in Mexico (Barker & Grimsdale, 1936), Florida (Cole, 1942; Levin, 1957), probably Venezuela (Raadshooven, 1951), Cuba (Cole & Bermudez, 1944; Cole & Gravel, 1952) and Jamaica (herein). Previous records from the Stettin Formation of Jamaica (e.g., Robinson, 1993; Robinson & Wright, 1993) are here placed in Helicostegina minor sp. nov.

Helicostegina dimorpha Barker & Grimsdale, 1936

(Figs. 34 , 35 , 36 , 37 )

Previous descriptions and figures. Barker & Grimsdale. 1936, p. 235, Pl. 30, fig. 6; Pl. 32, figs. 6-7; Pl. 33, fig. 9; Pl. 34, figs. 7, 9; Pl. 37, fig. 3.

Recognition. According to Barker and Grimsdale (1936), the spire of H. dimorpha has a length of 18 to 21 chambers, inclusive of the proloculus. However, because of the trochospiral coiling it is probable that this is only the part of the spire that is cut in the section and in reality the spire is longer extending to the margin of the test. This is confirmed by Adams' (1987) Pl. 1, fig. 2, a topotype, which shows the spiral wall extending to the periphery of the test. Barker and Grimsdale (1936, p. 236) reported that the first two chambers (proloculus and deuteroloculus) had a length of 240 μm. There is no orbitoidiform layer and the apparent presence of one is due to equatorial sections cutting the spiral sheet of subsidiary chamberlets because of the trochospiral coiling of the test. Axial sections show the presence of subsidiary chamberlets in the peripheral area of the dorsal chambers, although the primary septa of the spire continue to the margin of the test.

Remarks. The type material of Barker and Grimsdale (1936) is confusing because the two equatorial sections pass from the primary spire into the spiral sheet of subsidiary chamberlets and this makes it look like the primary spire passes into an orbitoidiform layer. This is clearly not the case in the two axial sections that are presented (Barker & Grimsdale, 1936, Pl. 32, figs. 6-7, reproduced here in Fig. 35.3-4). Adams (1987) illustrated topotype material that shows that the primary spire extends to the margin of the test, yet even he talked about an orbitoidal layer. H. dimorpha occurs in upper ABZ7 (samples WL2098 and WL3309/WL3309A) and lower ABZ8 (sample WL2100) in Jamaica.

Helicostegina dimorpha has only been found in the middle and upper parts of ABZ7 and lower ABZ8 in Jamaica, where it is locally common. It is distinguished from H. gyralis by the acquisition of subsidiary chamberlets in the periphery of the chamber which, with the equatorial chamberlets, produces an incipient double layer.

Helicostegina jamaicensis Mitchell, E. Robinson & Özcan, sp. nov.

(Figs. 38 , 39 , 40 )

Diagnosis. A species of Helicostegina with a small embryo (proloculus and deuteroloculus), a thickened spiral wall which terminates before the margin of the test, and with two layers (equatorial, and ventral sheet of subsidiary chamberlets) in the flange. Low lateral chambers present during adult growth.

Type specimens. Holotype: EO.WL2047-51; Paratypes: EO.WL2047-21, EO.WL2047-22, EO.WL2047-23, EO.WL2047-26, EO.WL2047-4, EO.WL2047-41, EO.WL2047-53, EO.WL2047-57, EO.WL2047-57, UWIGM.WL3532C-07, UWIGM.WL3532C-08. Samples WL2047 and WL3532C (both from the same level), Preston Hill Formation, Lilyfield, St Ann, Jamaica.

Description. Only megalospheric specimens seen. Megalospheric specimens are up to 1.5 mm in diameter and 0.8 mm thick. The surface is covered with regularly arranged piles (pillars). The equatorial layer consists of the embryo and a primary spire of 17-20 chambers (including the proloculus and deuteroloculus) occupying 1½ to 1¾ whorls. The length of the embryo (Ele) ranges from 135 to 190 μm (including the walls). The primary spire has a thickened wall which does not extend to the margin of the test. The last chamber in the primary spire has two apertures and gives rise to orbitoidiform growth of the equatorial layer. In axial sections the alar prolongations of the spiral chambers on the dorsal side are not divided into chamberlets. The ventral side of the test is divided into subsidiary chamberlets, and this extends as a layer parallel to the equatorial layer giving the test a double flanged appearance. 'Lateral' chamberlets (i.e., additional layers of dorsal and ventral chamberlets) are added in the adult stage, but do not extend onto the flange which just has a thickened wall.

Remarks. The species is distinguished from H. gyralis and H. dimorpha by its small embryo, and in the spiral wall not reaching the margin of the test. It is distinguished from H. minor sp. nov. and H. cf. minor sp. nov. by the fact that the spiral wall does not reach the edge of the test whereas in the latter forms the spiral wall extends to the margin of the test; thus orbitoidiform growth is neither achieved in H. minor sp. nov. nor in H. cf. minor sp. nov., but is in H. jamaicensis sp. nov.

Distribution. Only recorded from ABZ8, upper Lutetian, along the deep-water facies of the Preston Hill Formation, Lilyfield, St Ann, northern Jamaica.

Helicostegina jeannemairae E. Robinson, Mitchell & Özcan, sp. nov.

(Figs. 41 , 42 , 43 , 44 , 45 , 46 , 47 )

Diagnosis. Early growth stage as in Heterostegina dimorpha with alar prolongations open on the dorsal side and merging into subsidiary chamberlets on the ventral side. Adult test shows purely annular growth (alar prolongations absent in adult test) with an orbitoidiform (equatorial) layer and a ventral layer of subsidiary chamberlets giving a two-layered structure in axial section.

Type specimens. Holotype: UWIGM.WL3641-13 (Figs. 41 .3, 43 .6). Paratypes: UWIGM.WL3641-2 to 6; UWIGM.UWLGM3641-8 to 9; UWIGM.WL3641-12; UWIGM.WL3641-14, WL3641-16 to 21; UWIGM.WL3641-23 to 25. All from Chapelton Formation, Yellow Limestone (sample WL3641), Upper Rock Spring, parish of Westmoreland, Jamaica. There are several hundred additional paratypes in sample WL3641 that have not been numbered or cut.

Description. Tests are flat lenticular with swollen umbos and a diameter of up to 6.4 mm. The ornament consists of small scattered piles (pillars) across the whole test. The population is dominated by megalospheric specimens with very rare microspheric specimens. Microspheric specimens are only slightly larger (6.4 mm) than the megalospheric specimens (largest specimen is 5.7 mm).

Microspheric specimens have a very small proloculus that is followed by a primary spire of about 40 chambers. The spire is terminated by a retrovert chamber with two solons and this gives rise to the orbitoidiform layer.

The megalospheric specimens have an embryo consisting of two chambers with a thickened chamber wall. The length of the embryo ranges from 278 to 400 μm (mean 340, SD 46.4, n = 10), with the proloculus being slightly larger than the deuteroloculus. The primary spire consists of one and a half whorls, is involute, and contains 15 to 23 chambers including the embryo (mean 19.1, st. dev. 2.88, n = 10). The primary spire contains subsidiary chamberlets on the ventral side, whereas the dorsal chambers extend onto the acute keel of the spire (this is developed as in Helicostegina dimorpha). The primary spire has a thickened wall which terminates at the point where orbitoidiform growth commences. The primary spire gives rise to two spirals (primary and secondary) through a chamber with two stolons (one forward directed and one retrovertly directed), and these spires coil around the periphery of the test and end in a closing chamber. Each chamber in the secondary spires gives rise to additional chamberlets which initiates orbitoidiform growth of the equatorial layer. The orbitoidiform chamberlets are low arcuate. The orbitoidiform layer occupies about four-fifths of the diameter of the test in large specimens. The ventral side of the test is composed of a single sheet of subsidiary chamberlets that develop from the sheet of secondary chamberlets on the ventral side of the primary spire. The chamberlets in the ventral subsidiary sheet have a typical orbitoidal form with two stolons as in the equatorial layer. Once orbitoidiform growth is initiated, there are no more alar prolongations and the test wall is compact without lateral chambers which results in the formation of a wide double flange. This indicates that growth of the adult test was facilitated by annular addition of orbitoidiform chamberlets to both the equatorial layer and to the ventral sheet of subsidiary chamberlets.

Remarks. The species is named after the late Jeanne Mair Robinson who, with her children, discovered the first specimens of this species at Point, St James (Jamaica), although the type locality is situated in Westmoreland Jamaica where free specimens are abundant. The population from Point (Fig. 41 ) has more whorls in the spire than specimens from the type population and many have a one-biop morphology (Fig. 41 .2-5, 7) in axial section. We regard these specimens as transitional to H. dimorpha, but because of the wide flange we place them in H. jeannemairae sp. nov. The species is common in some areas of western Jamaica (St. James and Westmoreland), but has not been found in the marlstones (Preston Hill Formation) or limestones (Chapelton Formation) along the North Coast Belt of Jamaica where H. jamaicensis sp. nov. is common. It may be that H. jeannemairae sp. nov. developed as a shallow-water species and H. jamaicensis sp. nov. as a deeper-water species.

Distribution. Helicostegina jeannemairae sp. nov. has only been found in sections in western Jamaica that lack zonally diagnostic forms. In St James it occurs in the Yellow Limestone above levels yielding H. cf. dimorpha and Eolepidina (probably lower ABZ8) and below levels yielding Eo. gardnerae (ABZ10), so an attribution to ABZ9 seems likely. At Rock Spring (Westmoreland) it occurs with only rare specimens of Operculinoides sp.

Family HELICOLEPIDINIDAE Tan, 1936

Diagnosis. The family is characterised by forms showing lenticular growth with the spiral and umbilical sides generally not determinable in axial sections. The equatorial layer consists of a primary spiral of chambers with equatorial chamberlets developing on the outside of the primary spiral. The alar prolongations on the ventral (spiral) side of the test are not subdivided into subsidiary chamberlets as in the Helicosteginidae, and lateral chamberlets are present on each side of the equatorial layer.

Genera included. Helicosteginopsis Caudri, 1975; and Helicolepidina Tobler, 1922.

Remarks. The family is separated from the Helicosteginidae by the lack of a subsidiary sheet of chamberlets on the ventral (spiral) side of the test and the presence of lateral chamberlets on either side of the equatorial layer. We do not describe these forms as they are rare in our samples, but do provide our best estimates of ages.

Genus Helicosteginopsis Caudri, 1975

Type species. Helicostegina soldadensis Grimsdale in Vaughan & Cole, 1941.

Helicosteginopsis soldadensis (Grimsdale in Vaughan & Cole, 1941)

(Fig. 48 .1, 3-4)

Previous descriptions. Grimsdale in Vaughan & Cole (1941); Caudri (1975, 1996); Adams (1987).

Age. The material from Trinidad comes from levels that have been assigned to the upper Eocene (Vaughan & Cole, 1941; Caudri, 1975, 1996), but contains reworked material from Paleocene and middle Eocene levels such that the age of the species cannot be determined from Trinidad. In Margarita Island (Caudri, 1974), Venezuela, H. soldadensis occurs in sample P.J.B.145 with L. ariana, Orbitoina spp. and Helicolepidina spiralis trinitatis [= nortoni Vaughan] and can be assigned to ABZ12. In Panama, Cole (1941) recorded H. soldadensis with Eu. chaperi and H. ocalana indicating ABZ16. We provisionally give a range estimate of zones ABZ12 to ABZ16.

Genus Helicolepidina Tobler, 1922

Type species. Lepidocyclina (Helicolepidina) spiralis Tobler, 1922.

Synonyms. Helicocyclina Tan, 1936 (type species: Helicolepidina paucispira Barker & Grimsdale, 1936).

Helicolepidina sp. A

(Fig. 48 .2)

Remarks. Barker and Grimsdale (1936, Pl. 35, fig. 4) figured a specimen that they ascribed to Eulinderina guayabalensis var. regularis Barker & Grimsdale, but differs in having some equatorial chamberlets developing on the outside of the primary spire (Fig. 49.4). This form does not have a thick spiral wall. We ascribe this specimen (from the ABZ7- ABZ8 boundary zone) to Helicolepidina sp. A.

Helicolepidina spiralis Tobler, 1922, and Helicolepidina nortoni Vaughan, 1936

(Fig. 49 .1)

This group of species has been discussed by Raadshooven (1951) who recognized six forms (morphotypes) with some stratigraphic significance; his forms C and D correspond to H. spiralis Tobler and his forms E and F, to H. nortoni (= H. spiralis trinitatis Brönnimann). Caudri (1975) recorded H. nortoni (her H. spiralis trinitatis) from ABZ12 on Margarita Island, Venezuela. Cole (1952) recorded H. nortoni (as H. spiralis) from Panama in samples lacking both Pl. tobleri and H. ocalana (ABZ15) and in samples with H. ocalana (ABZ16). Thus we infer a range of zones ABZ12 to ABZ16 to H. nortoni. We assign tentative ages of ABZ7 to ABZ8 for H. sp. 1 (forms A and B of Raadshooven, 1951) and ABZ8 and ABZ10 for H. spiralis, based on Raadshooven (1951); further work is needed to confirm the ranges.

Helicolepidina polygyralis Barker, 1932

(Fig. 49 .3-4)

Remarks. Barker (1932, 1934) placed megalospheric forms with diameters of 1.6 - 2.4 mm and 2.5 - 3 whorls in this species; it therefore has more whorls than other species of Helicolepidina. Caudri (1974) recorded H. polygyralis from ABZ9 on Margarita Island, Venezuela.

Helicolepidina paucispira Barker & Grimsdale, 1936

(Fig. 49 .2)

Remarks. Barker and Grimsdale referred H. paucispira to the upper Eocene. Cole (1945, p. 17) reported this species from water wells at Tallahassee in the Upper Eocene of Florida which can be assigned to either upper ABZ15 or ABZ16 due to the presence of Eu. chaperi.

Family PSEUDOLEPIDINIDAE Mitchell, E. Robinson & Özcan fam. nov.

Diagnosis. The family is characterised by the development of a double equatorial layer together with lateral chambers on both the dorsal (umbilical) and ventral (spiral) sides. The double equatorial layer develops from an initial single equatorial layer.

Remarks. The biloculine embryo is joined by a further chamber on the ventral side that we consider to be part of the embryo. There is/are one or two PACs developed in addition to this third embryonic chamber. Pseudolepidina developed a lepidocyclinid orbitoidiform morphology before the Lepidocyclinidae and therefore cannot be derived from that stock. We also see it difficult to generate Pseudolepidina from the Helicosteginidae or the Helicolepidinae, again because of its advanced equatorial plan. The subfamily appears to be a separate radiation, presumably from an unknown rotalid ancestor.

Genera included. Pseudolepidina Barker & Grimsdale, 1937 (type species: Pseudolepidina trimera Barker & Grimsdale, 1937); Triplalepidina Vaughan & Cole, 1938 (type species: Triplalepidina veracruziana Vaughan & Cole, 1938). We do not consider Triplalepidina in this work as we do not have specimens of this genus.

Genus Pseudolepidina Barker & Grimsdale, 1937

Type species. Pseudolepidina trimera Barker & Grimsdale, 1937 from the Lutetian of Jenny station 1573, 11 km SE of Sabaneta, state of Veracruz, Mexico.

Diagnosis. A genus of Pseudolepidinidae developing a double equatorial layer, but lacking a compact shell layer between the two parts of the equatorial layer; with a 'third PAC' developed; and well-developed lateral chamberlets.

Remarks. Pseudolepidina is distinguished from Triplalepidina in lacking a compact shell layer between the two parts of the double equatorial layer.

Pseudolepidina trimera Barker & Grimsdale, 1937

(Figs. 50 , 51 , 52 , 53 , 54 , 55 )

Previous descriptions and figures. Barker & Grimsdale, 1937, p. 172, Pl. V. figs. 1-3; Pl. VIII, figs. 1-5; Adams, 1987, p. 299, Pl. 3, figs. 11-16.

Recognition. The double equatorial layer as seen in axial sections is diagnostic, particularly when the third PAC is visible in such sections. Oblique sections of Lepidocyclina or Orbitoina may give the impression of showing two equatorial layers and can be confused with Pseudolepidina, so the presence of the additional PAC is critical to identifications of specimens in random orientations in limestone thin sections. There is considerable variation in the embryo and PACs with specimens having one or two PACs when cut in equatorial section; the position of the 'third PAC' (which we regard as part of the embryo) seems variable and it may also be seen in some equatorial sections.

Remarks. The type series of Barker and Grimsdale (1937) came from Jenny station 1573, 11 km SE of Sabaneta, state of Veracruz, Mexico, where it was associated with H. gyralis and Operculinoides jennyi (Barker & Grimsdale, 1936, 1937; Barker, 1939) and can be assigned to ABZ7. Adams (1987, Pl. 3, figs. 15-16, rephotographed in Fig. 54.8 here) figured a specimen from Goldschmidt station 837, close to the village of Yecuatla, state of Veracruz, Mexico, where it was associated with Eulinderina regularis and Eolepidina antillea (reported as Eulinderina semiradiata: Barker & Grimsdale, 1936); this can be assigned to the ABZ7- ABZ8 transition interval. Robinson (2000) recorded Ps. trimera from the type Swanswick Formation and from a level in the Chapelton Formation of Jamaica. We have specimens from the type Swanswick Formation at Swanswick House and this can be assigned to ABZ7. The occurrence in the Chapelton Formation needs confirmation, and we would assign this to the early Bartonian (maybe ABZ12 or ABZ13) - we do not show this occurrence on the stratigraphic chart (Fig. 6 ) since it needs confirmation.

Family LEPIDOCYCLINIDAE Scheffen, 1932

Diagnosis. A family developed from a spiral (eulinderine) ancestor without alar prolongations. We recognize two subfamilies (Lepidocyclininae and Orbitoininae subfam. nov.) which each show a progressive reduction in the primary spire from eulinderine to polylepidine to lepidocycline to nephrolepidine or eulepidine. All but the most primitive forms have 'lateral' chamberlets developed.

Remarks. The Lepidocyclinidae show clear and separate radiations of genera that are obvious from the detailed work we have undertaken in Jamaica. This has been suggested previously for the Eocene lepidocyclines (e.g., Grimsdale, 1959; Vlerk, 1959; Sirotti, 1983; Butterlin, 1987), but age relationships between different forms have previously been poorly constrained. From the samples collected from Jamaica we can see two clear radiations, which we place in separate subfamilies: the Lepidocyclininae and Orbitoininae. The largest problem is with homeomorphy, as each radiation gives rise to advanced lepidocycline morphologies which are difficult to distinguish from each other on morphological features (particularly in random thin sections). This classification has the advantage of retaining all lepidocycline-like forms in a single family, but separating them into their separate radiations. Although the orbitoidiform plan has been used to define families or subfamilies previously (e.g., Loeblich & Tappan, 1964, 1988; Adams, 1987; Boudagher-Fadel, 2018), we believe that individual phylogenetic radiations should not be arbitrarily split into subfamilies based on the acquisition of an orbitoidiform growth plan. If such a course was taken, at least four separate subfamilies would be required in the Lepidocyclinidae. We therefore give subfamily status to the two individual evolving clades that show a progressive change from cyclic to orbitoidiform growth.

Each radiation shows a progressive change up section. Early forms have an embryo consisting of a proloculus and a deuteroloculus followed by a short spire of large, peri-embryonic chambers (eulinderine forms) that gives rise to a chamber (R) with a retrovert aperture from which two spires (primary and secondary) of relatively large peri-embryonic chambers emanate. Up section, the position or R progressively decreases until the development of first one PAC (unilepidine) and then two PACs (polylepidine). With one PAC two spirals of peri-embryonic chambers extend around the embryo and meet at a closing chamber. With two PACs three or four, asymmetrical spirals of peri-embryonic chambers extend around the embryo with two closing chambers (Eva, 1980). A true orbitoidiform growth plan (lepidocycline) is achieved when the four spires emanating from the PACs become bilaterally symmetrical about the embryonic (protoconch-deuteroconch) axis. Subsequent development sees additional stolons connecting adauxilary chambers to the protoconch and/or deuteroconch, and a reduction in the size (and consequent increase in number) of the peri-embryonic chambers. Further development sees an increase in the width of the deuteroconch which then partially envelops the protoconch (nephrolepidine to eulepidine forms).

Equatorial chambers and chamberlets can be divided into several types. The first two chambers represent the embryo. In the earliest forms there is only a single stolon connecting the proloculus to the deuteroloculus. Subsequent chambers in the primary spire of early forms have two stolons: a single stolon connecting back to the embryo, or earlier chambers and a single stolon connecting to subsequent chambers. Once an R chamber, one PAC, or two PACs are developed (or for that matter adauxilary chambers), these chambers and any subsequently peri-embryonic chambers have three stolons, one connecting to the chamber that the chamber arises from and two connecting to subsequent chambers/chamberlets (note, that the presence of adauxiliary chambers adds to the numbers of stolons connecting to the protoconch and/or deuteroconch). The closing chamber(s) mark(s) a change to four or six stolons and all subsequent chamberlets after this have either four stolons or six stolons. We define equatorial chambers as those representing the embryo and the peri-embryonic chambers (including any closing chambers) and equatorial chamberlets as those that arise from the peri-embryonic and closing chambers.

There are three broad arrangements of equatorial chamberlets when viewed in equatorial sections. The first has a predominantly radial form and occurs in Eolepidina and Orb. rdouvillei (similar patterns are seen in Pseudolepidina). The second only shows arcs of equatorial chamberlets with neither a prominent radial nor a prominent concentric arrangement of chamberlets and occurs in Lep. ariana. The third is characterised by a series of concentric or annular rings of equatorial chamberlets and occurs in Lep. ocalana, Eulepidina chaperi and Eulepidina undosa. These arrangements result from the way that chamberlets are arranged, from the chamberlet shapes and from the distribution of stolons (four versus six) connecting chamberlets. The form of these arrangements is not of generic significance, since they repeatedly occur at similar stages in the development of different lineages.

Subfamily LEPIDOCYCLININAE Scheffen, 1932

The breakdown of the Lepidocyclininae into genera has been demonstrated by Robinson & Jiang (1995) and Robinson (1996, 1997) and we follow their scheme here, with modifications, recognizing eulinderine (R > 7, no lateral chambers), eolepidine (R > 3, with lateral chambers), unilepidine (R = 3), polylepidine (R = 2), lepidocycline (R = 2) and eulepidine morphologies (Fig. 56 ). Different stages of development are given generic status and have useful biostratigraphic value. The described genera are somewhat arbitrary and are retained here for historical reasons rather than because they are fully merited. We discuss the use of generic names under each genus below. The proloculus/protoconch and embryo sizes in the Lepidocyclininae gets larger through the Eocene, although there are periods of stasis (Figs. 57 , 58 , 59 ), and the embryo is generally larger than the embryo size of co-occurring populations of the Orbitoininae (other than for Pliolepidina).

Frost and Langenheim (1974) synonymised many species in this group (although their figured populations can often be assigned to individual species as recognized here) and their scheme clearly obscures the biostratigraphic value demonstrated by Robinson and Jiang (1995) and Robinson (1996, 1997). We do not follow Frost and Langenheim's (1974) synonyms here, but develop our own classification of these forms. Many studies are based on random sections in thin sections of rock samples (e.g., Robinson & Wright, 1993; Serra-Kiel et al., 2007); yet many random sections are difficult to place in the evolutionary scheme as understood here and, if possible, workers should try to obtain free specimens to produce orientated sections for better biostratigraphic resolution. An alternative is to cut large numbers of random sections hoping to get some more-or-less equatorial sections by chance. Yet some genera (e.g., Eulinderina) and species (e.g., Lepidocyclina macdonaldi) are very distinctive in random sections, and provide some useful biostratigraphic resolution.

Robinson (1996, 1997) introduced mean R values (R), being the mean number of chambers (including the protoloculus and deuteroloculus) to the first retrovert chamber in populations of eulinderine and polylepidine forms. He showed that R values fell within narrow limits within populations, and many additional populations analysed from Jamaica in this work show a similar narrow limit for R in any one sample. We therefore define limits for mean R size for identifying species here (Table 17). We also provide measurements for the embryo (combined length of the proloculus and deuterolocus or protoconch and deuteroconch, including the thickness of the walls; Ele, Table 18) and width of the proloculus/protoconch excluding walls (Pwi, Table 19) for different species as given in the literature and in measured specimens/populations for Jamaica. The transition from typical four-spired lepidocyclines (L. ariana) to lepidocyclines with adauxilary chambers (L. macdonaldi, L. ocalana), to nephrolepidine-like forms (Eu. chaperi) is not straight forward and needs to be investigated in more detail. Other recent papers (e.g., Molina et al., 2015, Pl. 1.5; Hohenegger & Torres Silva, 2020, Fig. 2) confuse isolepidine forms (L. ocalana) with adauxilary chambers for nephrolepidine forms (Eu. chaperi).

Numerous generic names have been proposed for the Lepidocyclininae, many of which have been synonymised. The names retained here break the lineage(s) up into a series of genera, and the generic names have a long historical usage. Our placement of species into different genera is based on our revised definitions of the genera as indicated below.

Genera included. Eulinderina Barker & Grimsdale, 1936; Eolepidina Tan, 1939; Polylepidina Vaughan, 1924; Lepidocyclina Gümbel, 1868; Eulepidina Douvillé, 1911.

Table 17: Criteria for distinguishing species of Eulinderina, Eolepidina and Polylepidina based on mean R values.

Table 18: Measurements of mean R and the length of the nucleoconch (or first two chambers) including the walls for specimens/species of Lepidocyclininae from Jamaica and elsewhere. Measurements include nucleoconch (chamber) walls.

* specimens measured from illustrations in publications (which is dependent on accuracy of plate reproduction when only magnifications were given). † specimens regarded as reworked. B & G, 1936 = Barker & Grimsdale, 1936; G & H, 1935 = Gravell & Hanna, 1935.

Table 19: Measurements of mean R and the width of the proloculus/protoconch (Pwi) excluding the walls for specimens/species of Lepidocyclininae from Jamaica and elsewhere. Measurements include nucleoconch (chamber) walls. † specimens regarded as reworked.

Genus Eulinderina Barker & Grimsdale, 1936

Diagnosis. Lepidocyclininid with a primary spire comprising six or more chambers before the first retrovert chamber (i.e., the chamber that gives rise to two separate spires); lateral chamberlets absent.

Remarks. We formally define limits of R for the two species of Eulinderina here for use for biostratigraphic purposes (Table 17). The two taxa assigned to Eulinderina, Eu. guayabalensis (Nuttall, 1930) (R = 7 - 9) and Eu. regularis Barker & Grimsdale, 1936 (R = 5.5 - 7), appear to be rare outside of Mexico and are not currently incorporated into the zonation. It is likely that the Lepidocyclininae initially evolved in Mexico (close to where Jamaica was situated at the time: Pindell, 2009; Mitchell, 2020) during the Lutetian and subsequently spread across the American LBF bioprovince.

Eulinderina guayabalensis (Nuttall, 1930)

(Figs. 60 , 61 .1-5)

Previous descriptions. Barker & Grimsdale, 1936, p. 238, Pl. 32, figs. 8-9; Pl. 34, figs. 8, 10-11.

Recognition. Eu. guayabalensis is recognized by its long primary spire before the development of a retrovert aperture, and also by its lack of lateral chamberlets and a laminate shell seen in axial section.

Remarks. Eu. guayabalensis appears to be rare outside of Mexico. We have several specimens from the Preston Hill Formation (samples ER152 and WL4869) from Grants Town (St Mary, Jamaica) with R values of 7 and 9 (mean = 8.5, n = 4). Barker and Grimsdale (1936) give the length of the embryo of Eu. guayabalensis as 160-230 μm, which is notably smaller than specimens of this species from Jamaica that have an embryo with a length of 212-329 μm. The age of the material described by Barker and Grimsdale (1936) is difficult to assign to our biostratigraphic zonation because they did not list associated species (particularly nummulitids); it could be that their material was from ABZ6 rather than ABZ7, but this will have to await studies of topotype material. Samples ER152 and WL4869, both from the Preston Hill Formation at Grants Town, yield Eu. guayabalensis in association with Operculinoides and orthophragmines (Robinson & Mitchell, 1999, p. 35; herein) and are assigned to ABZ7 and NP15b. For the present, we assume that Eu. guayabalensis is likely restricted to ABZ7.

Eulinderina regularis Barker & Grimsdale, 1936

(Fig. 61 .6-9)

Previous descriptions. Barker & Grimsdale, 1936, p. 238, Pl. 30, figs. 7, 9; Pl. 32, figs. 10-11; Pl. 35, figs. 4-5; Pl. 37, fig. 5.

Recognition. Eu. regularis is recognized by its shorter primary spire before the development of a retrovert aperture, and also by its lack of lateral chamberlets and a compact shell seen in axial section.

Remarks. Barker and Grimsdale (1936) recorded Eu. regularis (their Eu. guayabalensis var. regularis) from the same locality as Eolepidina semiradiata (= Eolepidina antillea); this suggests that this locality (assuming the material was from the same sample, rather than locality) contains an assemblage of specimens at the transition between Eulinderina and Eolepidina. We have not recorded Eu. regularis from Jamaica. In Mexico, its occurrence with Eolepidina antillea (reported as Eu. semiradiata) and Ps. trimera (Barker & Grimsdale, 1936; Adams, 1987) indicates it occurs around the ABZ7 to ABZ8 transition.

Genus Eolepidina Tan, 1939

Type species. Eulinderina semiradiata Barker & Grimsdale, 1936 (junior synonym of Lepidocyclina antillea Cushman, 1919) from the Lutetian, close to the village of Yecuatla, state of Veracruz, Mexico (Barker & Grimsdale, 1936).

Diagnosis. A Lepidocyclininae with a short primary spire of equatorial chambers (from 3 to 6), none or one PAC, and vacuole-like lateral chambers.

Discussion. Eolepidina is distinguished from Eulinderina by the presence of lateral chamberlets. The type locality of Eo. semiradiata (= Eo. antillea) appears to show specimens in transition between Eulinderina and Eolepidina. Because the presence or absence of lateral chamberlets is easily determined in random sections (as seen in thin sections of rock samples), the separation of the two genera is certainly warranted for biostratigraphic purposes.

We recognize three species of Eolepidina: Eo. antillea, Eo. subplana and Eo. gardnerae. The last species has one PAC and has been placed in Polylepidina by some workers (e.g., Cole, 1929). We define Polylepidina as containing two PACs with three or four asymmetric spires of peri-embryonic chambers, which leads us to place the species gardnerae in the genus Eolepidina.

Eolepidina antillea (Cushman, 1919)

(Figs. 62 , 63 , 64 , 65 , 66 )

Previous descriptions and figures. Cole, 1960, Pl. 11, figs. 6, 13; Pl. 12, figs. 1-8.

Remarks. Eo. antillea has an ornament consisting of scattered pustules, and we define that R = 5.5 - 4.5. The secondary spire of chambers developed from the first retrovert chamber is often short with small chambers and contrasts with the primary spire developed from the first retrovert chamber which is long with larger chambers. We follow the revision of Eo. antillea as proposed by Cole (1960) and Robinson (1996). We synonymise Eulinderina semiradiata Barker & Grimsdale, 1936 with Eo. antillea here. Eo. antillea occurs in the Preston Hill Formation in the North Coast Belt and northern Wag Water Belt in Jamaica, where it is associated with Operculinoides and small Nummulites. It is the nominate taxon for ABZ8. At the type locality in St Bartholomew (recollected as samples ER1524-ER1526), Eo. antillea occurs with Fallotella cf. cookie.

Some specimens show a thick spiral wall which extends up to the chamber before the chamber with the first retrovert aperture (e.g., Fig. 66.2). Other specimens (e.g., Fig. 66.3-4) lack a thickened spiral wall, but occur with specimens which have a thickened spiral wall. Further work is needed to determine if these are separate morphospecies, or simply variants of a single morphospecies.

Eolepidina subplana (Barker & Grimsdale, 1936)

(Figs. 67 , 68 , 69 , 70 )

Previous descriptions. Barker & Grimsdale, 1936, p. 241, Pl. 31, figs. 1-2; Pl. 32, figs. 12-13; Pl. 33, fig. 8; Pl. 35, fig. 6; Pl. 37, fig. 2.

Recognition. Eo. subplana is recognized by its low, lenticular test with lateral chambers and a primary spire consisting of typically 4 chambers. We place populations with R = 4.5 - 3.5 in Eo. subplana.

Remarks. The ornament of Eo. subplana is variable. Some forms have an ornament consisting of regular pustules that extend across the entire surface, whereas other forms (Chapelton Formation at Lilyfield) have an ornament that consists of a central crown of pustules with radiating furrows in the marginal zone. These ornament types are also seen in Eo. antillea. Eo. subplana is the nominate species for ABZ9. The species has been recorded from Mexico (Veracruz), Jamaica, Florida, and St Bartholomew.

Eolepidina gardnerae (Cole, 1929)

(Figs. 71 , 72 , 73 , 74 )

Previous descriptions and figures. Cole, 1929, p. 60-62, Pl. 1, figs. 1-6; Pl. 2, figs. 1-2.

Recognition. Eo. gardnerae has one PAC and two subequal spires. This is clearly visible in material from the type series (Cole, 1929, Pl. 1, fig. 5; Pl. 2, fig. 2). We define R as ranging from 2.2-3.5. If a second PAC is present (e.g., Figs. 26.3, 26.6) it either produces no spires or very short ones.

Remarks. We place the species gardnerae in the genus Eolepidina as we regard the development of two PACs as being of generic value. Eo. gardnerae is distinctive and easily distinguished from P. chiapasensis, based on the number of PACs, when several specimens are available for study. Although there is undoubtedly a graduation from Eo. gardnerae to P. chiapasensis to Lepidocyclina ariana, the separation of Eo. gardnerae from P. chiapasensis is useful biostratigraphically. Eo. gardnerae is the nominate species for ABZ10. The species has been recorded from Florida/Alabama (as Polylepidina: Robinson & Jiang, 1995), Jamaica and Veracruz (as Polylepidina chiapasensis by Barker & Grimsdale, 1936, Pl. 35, figs. 7-8) and Chiapas (as Polylepidina antillea by Frost & Langenheim, 1974, Pls. 30-31), Mexico.

Genus Polylepidina Vaughan, 1924

Type species. Polylepidina chiapasensis Vaughan from the Middle Eocene of Chiapas, Mexico.

Diagnosis. A lepidocyclininid with two PACs and three or four asymmetrical primary spires of periequatorial chambers. Lateral chamberlets developed on both sides, with low numbers (usually 3 to 5, but up to 6 or 7) of tiers and with vacuole-like chamberlets.

Discussion. We follow Robinson (1997) in regarding the presence of two PACs and asymmetrical primary spires as the diagnostic criterion for discrimination at the generic level.

Polylepidina chiapasensis Vaughan, 1924

(Figs. 75 , 76 , 77 , 78 , 79 , 80 )

Previous descriptions and figures. Vaughan, 1924, p. 807-808, Fig. 4; 1929a, p. 288-289, Fig. 3.

Recognition. Po. chiapasensis is flat discoidal and has numerous small to large pustules extending across most, or all, of the surface, yet sometimes it has a marginal band with radial striations formed from very-fine granules (as in Eo. semiradiata (Barker & Grimsdale, 1936) = Eo. antillea (Cushman, 1919)). Here, we define R as ranging from 2.0 - 2.2, and typically three or four asymmetrical peri-embryonic spirals develop from the two PACs.

Remarks. P. chiapasensis is externally similar to Lepidocyclina ariana and can only be distinguished in equatorial sections (P. chiapasensis is characterised by R = 2.0 - 2.2 with three or four asymmetrical peri-embryonic spirals, whereas L. ariana has R = 2 with two sets of two symmetrical peri-embryonic spirals). In most specimens, the three or four peri-embryonic spirals are markedly differently developed, with one spiral usually being dominant and having larger chambers. In a few specimens (e.g., Fig. 80.3-4) the spirals have chambers of equal sizes and on first glance the spirals appear symmetrical; yet a closer inspection reveals the spirals are asymmetric as the closing chambers (c) lie closer to one PAC (P) that the other.

Polylepidina chiapasensis is the nominal species for ABZ11, and its presence indicates a level in the lower Bartonian. The species is widespread in Jamaica in the Chapelton Formation at Albert Town and Litchfield, and in the Preston Hill Formation at Salt Spring.

Genus Lepidocyclina Gümbel, 1868

Type species. Nummulites mantelli Morton, 1833 from the Marianna limestone (lower Oligocene), near Claiborne, Alabama, USA (Morton, 1833; Vaughan, 1927).

Synonyms. Isolepidina Douvillé, 1915 (type species Nummulites mantelli Morton, 1833) [objective synonym]; Amphilepidina Douvillé, 1922 (type species Orbitoides sumatrensis Brady, 1875) [junior synonym]; Polyorbitoina Geyn & Vlerk, 1935 (type species Lepidocyclina (Polylepidina) proteiformis Vaughan, 1924) [junior synonym]; Multilepidina Hanzawa, 1932 (non Silvestri, 1937) (type species Lepidocyclina (Multilepidina) irregularis Hanzawa, 1932) [junior synonym].

Remarks. We restrict Lepidocyclina here to forms which have two PACs and an embryo with the protoconch and deuteroconch divided by a more-or-less straight to gently curved wall. The more 'primitive' forms lack adauxiliary chambers and include Lep. ariana Cole & Ponton. The more 'advanced' forms have adauxiliary chambers and include Lep. ocalana Cushman and Lep. macdonaldi Cushman. We do not think that separating these forms into separate subgenera (or genera) is useful as we are probably dealing with several distinct lineages that show a progression from forms lacking adauxiliary chambers to forms bearing adauxiliary chambers. The relationships with other 'simple' lepidocyclinids, such as Lep. canellei Lemoine & R. Douvillé from the Oligocene, are also unclear at the present time and we provisionally include all these forms in Lepidocyclina pending further work.

Lepidocyclina ariana Cole & Ponton, 1934

(Figs. 81 , 82 , 83 , 84 )

Previous description and figures. Cole & Ponton, 1934, p. 142-143, Pl. 2, figs. 1-5; Cole, 1944, p. 61-62, Pl. 1, fig. 14; Pl. 13, fig. 3; Pl. 14, figs. 1-7; Pl. 16, figs. 11-13; Pl. 17, fig. 13.

Remarks. In equatorial view L. ariana is characterised by a typical four-spiral peri-embryonic chamber plan developed from two PACs; it lacks adauxilary chamberlets and has semi-circular equatorial chamberlets. In axial section, it possesses relatively few lateral chamberlets that have thick walls with small irregular vacuoles.

Occurrence. In Jamaica, L. ariana occurs during a strongly regressive interval, linked either to a major sea-level fall, or to local tectonics, and has only been found in deposits within basins; equivalent deposits on the platforms are either lacking, or represented by platform interior facies without lepidocyclinids. In Florida and Alabama the species is common at some localities (e.g., Cole, 1944, 1945; Robinson & Jiang, 1995; Robinson, 1997). The species, with its distinctive lateral chamberlets is restricted to ABZ12.

Lepidocyclina macdonaldi Cushman, 1919

(Figs. 85 , 86 )

Previous description and figures. Cushman, 1919, p. 94, Pl. 40, figs. 1-6; Cushman, 1920, p. 77, Pl. 34, figs. 1-3; Cole, 1952, p. 16, Pl. 7, figs. 1-19; Pl. 8, figs. 1-4; Pl. 14, fig. 11.

Remarks. Cole's (1952) material from Panama includes several species which show differences in the form of the lateral chambers. In 1956, Cole (p. 221) stated that three species of Lepidocyclina, namely Lep. ariana, Lep. proteiformis and Lep. macdonaldi could be separated on the basis of their lateral chamberlets: the first (Lep. ariana) had relatively thick but straight floors and roofs of the lateral chamberlets, and the chamberlet cavities were rather low; the second (Lep. proteiformis) had relatively thin and straight lateral floors and the cavities of the lateral chamberlets were open; the third (Lep. macdonaldi) had much thicker and curved floors and roofs of the lateral chamberlets and the cavities of the chamberlets were more appressed. Lep. ariana as figured here is relatively distinct, because of its basic four-spiral arrangement of peri-embryonic chambers and its thick walls to the lateral chamberlets that have thin spaced vacuole-like chambers. Lep. macdonaldi is also relatively distinct, in the form of its lateral chamberlets (thin) and thick curved roofs and floors; it is very distinctive in thin section. Cole (1956, Pl. 27, figs. 4-5) shows equatorial sections of Lep. macdonaldi from Jamaica, which appear to show the dark lines (cavities) above the equatorial layer, with the thick curved floors showing a bright area inside the cavities. Yet, the embryo is relatively similar to Lep. proteiformis and forms such as Lep. ocalana. The equatorial section of Lep. proteiformis illustrated by Cole (1956, Pl. 27, fig. 6) has an embryonic length of about 675 μm (as measured from his illustration), and therefore falls well outside the range of forms such as Lep. ariana and within the range of forms such as Lep. montgomeriensis and Lep. ocalana, although it has a simple four-spiral plan of peri-embryonic chambers. In fact, the lateral chamberlets of forms attributed to Lep. proteiformis from Jamaica and Mexico by Cole (1956) have lateral chambers similar to Lep. ocalana. We therefore tentatively attribute Lep. proteiformis of Cole, 1956 to the Lep. ocalana group.

Cole (1956, Pl. 27, figs. 3-5) illustrated three equatorial sections he referred to Lep. macdonaldi; these had embryo lengths of 520, 720 and 620 μm, respectively (as determined by measurement of the figures with the scale given). Four equatorial sections from sample WL4449 (ABZ14) from Cold Spring have an embryo length ranging from 488 μm to 766 μm (mean 660 μm, standard deviation 124.1 μm) and show the development of one or two adauxilary chambers. Both Cole's (1956) illustrations and our sectioned specimens have larger embryos than typical of Lep. ariana, and fall within the range of Lep. ocalana. It seems that Lep. macdonaldi and Lep. ocalana are parallel evolving lineages in the mid to late Eocene.

Our material shows the very distinctive features of the lateral chamberlets when seen in axial sections. The lateral chamberlets are strongly arcuate, offset in an en-echelon fashion, and do not show regular vertical tiers. Scattered piles (pillars) are also seen in the axial sections. This is consistent with material from the type locality in Mexico. Note that our specimens do not show an annular arrangement of equatorial chamberlets. This species resembles the arrangement seen in Lep. ariana in this respect rather than Lep. ocalana.

Range. Specimens of Lep. macdonaldi are common in ABZ14 (sample UWIGM.WL4449) at Cold Spring and Sherwood Content (samples UWIGM.R1068 and UWIGM.R1072) and sparingly so in the type Swanswick section at Swanswick House (ABZ16). In Panama, judging by a reinterpretation of the specimens illustrated by Cole (1952), the species ranges from zones ABZ14 to ABZ16.

Lepidocyclina ocalana Cushman, 1920

(Figs. 87 , 88 , 89 )

Previous description and figures. Cushman, 1920, p. 70-71, Pl. 27, figs. 1-4; Pl. 28, figs. 1-3; Cole, 1941, p. 41-43, Pl. 13, figs. 1-7; Pl. 16, figs. 1-4, 6-10, 15. Molina et al. (2015, Pl. 1.5) and Hohenegger and Torres Silva (2020, Fig. 2) illustrated specimens of Lep. ocalana erroneously under the name Lepidocyclina chaperi.

Recognition. Lep. ocalana in equatorial section combines a relatively large embryo, with a straight boundary between the protoconch and deuteroconch, with adauxilary chambers on the protoconch and deuteroconch, a relatively large number of peri-embryonic chambers, and a generally annular arrangement of equatorial chamberlets. At any level, there is a gradation from forms with four peri-embryonic spires to forms with adauxilary chambers; we place all these forms in Lep. ocalana and consider them natural variants of a single morphospecies. In axial section Lep. ocalana has a small number of layers of lateral chamberlets (up to about 7), which have open vacuoles, which are about as wide as the floors/roofs. Piles (pillars) are well developed.

Remarks. The species differs markedly from Lep. macdonaldi in axial section, with the later having arc-shaped, thin lateral chamberlets. Axial sections of the two species cannot be confused, but equatorial sections can. Equatorial sections sometimes show one larger and one smaller embryonic chamber (e.g., Fig. 87.4-5), but this may be due to the embryo being tilted at an angle to the equatorial plane (e.g., Fig. 87.1).

Range. In Florida Lep. ocalana occurs in the Ocala Limestone with Eulepidina chaperi and Heterostegina ocalana (Cole, 1941, 1943). In Panama, this species occurs with Pliolepidina tobleri, Eulepidina chaperi and Heterostegina ocalana (Cole, 1952). We have specimens in Jamaica from the middle unit of the Ipswich Formation (although these have a smaller embryo; Fig. 87), the Swanswick Formation sensu lato, the Claremont Formation and the Somerset Formation. The species therefore ranges through ABZ13 to ABZ16, and there is an increase in the size of the embryo passing from ABZ13 to ABZ14/ABZ15.

Genus Eulepidina Douvillé, 1911, p. 59

Type species. Orbitoides dilatata Michelotti, 1861 from the early Miocene of northern Italy (Michelotti, 1861).

Synonyms. Trybliolepidina ven der Vlerk, 1928 (type species Orbitoides (Lepidocyclina) ephippioides Jones & Chapman, 1900) [junior synonym].

Remarks. We follow Robinson (2004) in placing Lepidocyclina chaperi as the earliest species of Eulepidina.

Eulepidina chaperi (Lemoine & R. Douvillé, 1904)

(Figs. 90 , 91 )

Previous descriptions and figures. Cole, 1952, p. 23-27, Pl. 8, figs. 5-8; Pl. 9, figs. 3-19; Pl. 10, figs. 1-10; Pl. 11, figs. 1-8; Pl. 12, figs. 1-15; Pl. 20, figs. 8-10; Pl. 23, figs. 8, 11-12.

Recognition. The species is easily distinguished in equatorial section by the deuteroconch partially enveloping the protoconch, by the small and numerous peri-embryonic chambers and by the annular arrangement of the equatorial chamberlets. Specimens range from lenticular to weakly selliform (saddle-shaped). The species is distinguished from Oligocene species (such as, Eu. undosa and Eu. favosa) by its relatively similar diameters of the protoconch and deuteroconch, whereas in the Oligocene species the deuteroconch has a much greater diameter than the protoconch. The length of the embryo of Eu. chaperi is notably larger than that of co-occurring specimens of Lep. ocalana. Eu. chaperi occurs in upper ABZ15, ABZ16, and in the earliest Oligocene (ABZ17).

Subfamily ORBITOININAE Mitchell, E. Robinson & Özcan subfam. nov.

Diagnosis. The subfamily shows a parallel development to the Lepidocyclininae through eolepidine, unilepidine, polylepidine and lepidocycline forms to nephrolepidine forms. Early members of the subfamily have a relatively large number of well-developed lateral chamberlets in contrast to the early members of the Lepidocyclininae, which have none. The earliest member of the subfamily is Planorbitoinella gen. nov., which has well-developed lateral chamberlets. It is likely that the Orbitoinidae initially evolved in north-eastern South America during the Lutetian and subsequently spread across the American LBF bioprovince.

Genera included. Planorbitoinella gen. nov., Polyorbitoinella gen. nov., Orbitoina Schenck & Frizzell, 1936, Pliolepidina Douvillé, 1915, Nephrolepidina Douvillé, 1911.

Remarks. The evolutionary development of this subfamily is parallel to that of the Lepidocyclininae and gives rise to homeomorphs. Details of the Orbitoininae are give in Tables 20-21 and Figs. 92 , 93 , 94 , 95 . Planorbitoinella gen. nov. gives rise to Polyorbitoinella gen. nov., which in turn gives rise to Orbitoina, and Orbitoina gives rise to Pliolepidina. Nephrolepidina (not shown) with its small embryo, probably evolved from Orbitoina, but this requires further work. Here we describe the forms that have so far been found in Jamaica and comment on selected forms from the American LBF bioprovince. Furthermore, as we get higher in the Eocene, the distinction between the Lepidocyclininae and the Orbitoininae becomes more difficult because of convergent morphologies, and assignment of specimens to a genus becomes more problematic. Table 20 provides measurements of the length of the embryo, including the walls (Ele) for different species belonging to the Orbitoininae. Table 21 provides measurements of the width of the proloculus/protoconch, excluding the walls (Pwi) for different species belonging to the Orbitoininae. Variations of the length of the embryo (Ele) and width of the proloculus/protoconch are shown in Fig. 92 and a plot of Ele versus Pwi is shown in Fig. 93 .

Table 20: Measurements of the length of the embryo (Ele) for species of Orbitoininae

*Specimens measured from illustrations in publications (which is dependent on accuracy of plate reproduction when only magnifications were given). ⁺ Specimens considered reworked.

Table 21: Width of Proloculus/Protoconch (Pwi) for species of Orbitininae from Jamaica