The extinction of Chiloguembelina cubensis in the Pacific Ocean : implications for defining the base of the Chattian ( upper Oligocene )

We conducted quantitative biostratigraphic and morphometric studies at two ocean drilling sites in the Pacific Ocean to investigate the extinction of the planktonic foraminifera genus Chiloguembelina, a distinctive biserial genus ranging in the Eocene and Oligocene. The extinction of the last species of the genus, C. cubensis, has historically been used as a correlation event for the base of the Chattian (upper Oligocene; ~ 28 Ma). However, the bioevent has been controversial, as some authors have found a well-defined bio horizon, while others have found a decline in abundance with a continuation of C. cubensis into the upper Oligocene. This has been reflected in the preference of using the Top Common occurrence ( 5% of the foraminiferal assemblage) of the species as a correlation event for the boundary. We assessed the utility of C. cubensis as a reliable biohorizon at Ocean Drilling Program Leg 202 Site 1237 (south-east Pacific Ocean) and Integrated Ocean Drilling Program Expedition 320/321 Site U1334 (eastern equatorial Pacific Ocean). At both sites C. cubensis is a common and consistent component of early Oligocene planktonic foraminiferal assemblages. An abrupt termination of C. cubensis occurs at Site 1237 in magnetic polarity Subchron C10n.1n (27.89 Ma), while at Site U1334 peak abundances of C. cubensis are recorded in the upper Oligocene within Chron 9n with a younger datum for the extinction at ~ 27.15 Ma. Therefore, the biostratigraphic utility of the species as a boundary marker for the base of the Chattian remains unclear. We propose that the phrase “C. cubensis crisis” is the most suitable for describing the extinction or decline in abundance of the species in the mid Oligocene, based upon the findings from these two sites and other age calibrated sites. Morphometric analysis of specimen size at the two sites shows a markedly smaller test size at Site 1237 compared with Site U1334, with specimens from the former being on average 1.37 shorter and 1.45 thinner. Differences in test size and extinction horizon are possibly controlled by ecological preferences, however, our bio chronological compilation of ocean drilling sites does not reveal a latitudinal trend in extinction level.

most suitable being the candidate section for defining the GSSP.A primary correlation event, and a number of secondary event markers, are required to allow correlation between successions, which can include the first or last occurrence of a given fossils species, including planktonic foraminifera.Planktonic fora minifera are microscopic marine zooplankton which have long been used in biostratigraphy (e. g.Cushman 1924, Cushman and Stainforth 1945, Blow 1969, Berggren et al. 1985, 1995, Berggren and Pearson 2005, Wade et al. 2011), as they make excellent zone fossils due to their abundance in oceanic sediment, widespread geographical range, morphologically distinct shells (tests), and rapid evolution (Wade et al. 2011).
Within the Paleogene, 6 of the 9 stages currently have formally defined GSSP's, while the remaining three all have potential GSSP locations and correlation events (Jovane et al. 2010, Agnini et al. 2011, Wade et al. 2012, Coccioni et al. 2008, 2013).One such stage is the Chattian (mid Oligocene), with the candidate GSSP section being the Umbria-Marche basin, northeast Apennines, Italy.The two most suitable boundary markers are the oxygen isotope increase recorded in benthic foraminifera (Oi-2a interval), or the Top Common occurrence (Ͼ 5% of the planktonic foraminiferal assemblage) of the planktonic foraminifera species C. cubensis (Coccioni et al. 2008(Coccioni et al. , 2013)).The latter of these two potential correlation events is the focus of this study, with two ocean drilling sites being analysed primarily for the extinction level of C. cubensis.

Chiloguembelina cubensis
Chiloguembelina is a distinct genus within the Eocene and Oligocene planktonic foraminifera characterised by biserial tests, relatively small size (Ͻ 200 μm) and a microperforate test wall, as opposed to the more commonly found trochospiral, macroperforate planktonic forms of the Cenozoic (Jenkins 1985, Huber et al. 2006).Chiloguembelina cubensis was first described by Palmer (1934) and was initially placed in the genus of Güembelina, before later being assigned to Chiloguembelina by Beckmann (1957).The species has also previously been assigned to the genus of Streptochilus (Poore and Gosnell 1985).Huber et al. (2006) state that C. cubensis likely evolved from Chiloguembelina ototara during the late Eocene, and marks the last species of the genus.A close similarity exists between C. cubensis and C. ototara, with the distinctive troughs and ridges running parallel to long axis of the test, known as striae, on C. cubensis, acting as the main means to differentiate between the two species (Huber et al. 2006).Geographically C. cubensis is widespread and can be considered a cosmopolitan species as it has been recorded in all ocean basins ranging from 65°N to 65°S (Fig. 1).The depth habitat of C. cubensis, as inferred from stable isotope analysis, shows some degree of variability.Poore and Matthews (1984) found that specimens from the Gulf of Mexico and equatorial Atlantic Ocean indicate a shallow, mixed layer habitat, while a higher latitude South Atlantic site (Deep Sea Drilling Project Site 36) showed more intimidate values suggestive of a thermocline dwelling depth.Conversely another South Atlantic site shows values suggestive of a shallow depth habitat (Kelly et al. 2003), while Ocean Drilling Program Site 1218 in the equatorial Pacific Ocean suggested a thermocline dwelling depth habitat (Wade et al. 2007).

Historical perspectives on the biostratigraphic utility of Chiloguembelina cubensis
This review follows the chronostratigraphic framework outlined by Backman et al. (2012), where the Base (B) and Top (T) represents the lowest (first) and highest (last) taxon occurrences respectively.Top common (Tc) is used instead of the Last Common Occurrence (LCO).Where possible, the magnetic polarity Chron where the Top or Tc occurrence of C. cubensis occurs has been given allowing for a better means of global correlation, although some sites lack adequate palaeomagnetic records and in these cases the planktonic foraminiferal biozone in which the Top or Tc occurrence of C. cubensis is used instead.The extinction of C. cubensis has long been used as a bioevent for subdividing the Oligocene, with Berggren (1971), Jenkins and Orr (1972), and Berggren and Amdurer (1973) being the first studies to erect a biozone based upon the event.The bioevent became widely applied following the biochronologies of Berg gren et al. (1985,1995), where the Top or Tc of C. cu bensis is used to subdivide Biozone P21 into subzones P21a and P21b.The zonal nomenclature of the Oli gocene was updated by Berggren and Pearson (2005) to recognise "O" zones, rather than the former "P" zones.Zone O4 (= P21a) and O5 (= P21b) were used but retained Tc of C. cubensis to mark the early/late Oligo cene boundary with magnetochron calibration towards the Top of Chron C10n.1n (Berggren andPearson 2005, Wade et al. 2011).While earlier studies using the bio-event tended to favour the Top occurrence, meaning an abrupt/well-defined extinction, later studies increasingly favoured the use of the Tc occurrence, following the work of Huber (1991) who marked the boundary based upon the "last abundant occurrence" as opposed to the "absolute last occurrence".This was taken further by Leckie et al. (1993), who used LCO to mark the boundary, based upon C. cubensis extending into the upper Oligocene but accounting for less than 5 % of the planktonic fora mi niferal assemblage.Accordingly, this review follows the order of discussing sites which found a well-defined extinction, the sites which found occurrences of C. cubensis in the upper Oligocene, and how reworking may influence the stratigraphic range.Finally, alternatives to the use of C. cubensis in defining the Base of the Chattian are discussed.

Chiloguembelina cubensis
A number of sites at a variety of latitudes have documented the extinction of C. cubensis as a distinct biohorizon.In the western Atlantic an abrupt termination is found on Deep Sea Drilling Project (DSDP) Leg 72 Site 516 (Rio Grande Rise, ~30°S), which is calibrated via palaeomagnetism to within Chron C9 (Berggren et al. 1983).The similar latitude eastern Atlantic Site 522 (Southern Angola Basin, ~26°S) of DSDP Leg 73 shows a distinct biohorizon within Chron C10n (Poore 1984, Poore et al. 1984), which is consistent with the Chron calibration from the higher latitude eastern Atlantic Ocean Drilling Program (ODP) Leg 114 Site 703 (Meteor Rise, ~47°S) (Nocchi et al. 1991, Hailwood andClement 1991).Strontium isotope stratigraphy ( 87 Sr/ 86 Sr) found an age relatively consistent within the mid Oligocene in the eastern Atlantic at DSDP Leg 74 Site 529 (Walvis Ridge, ~28°S) (Hess et al. 1989).The discrepancies in the Chron calibration between the eastern and western Atlantic described above may reflect the poorly defined palaeomagnetics at Site 516.Berggren et al. (1983) suggested the extinction of C. cubensis would be the most useful criterion in boundary definition for the base of the Chattian, which is reflected in Berggren et al. (1985) but chose to use Chron C10n (following Poore 1984, Poore et al. 1984) as opposed to Chron C9.In the South China Sea, an extinction is observed in Site 1148 (~18°N) of ODP Leg 184 (Li et al. 2004), which is also found in ODP Leg 182 Site 1134 (Great Australian Bight, ~33°S) with the two sites having a 10° difference in longitude, although a hiatus is present close to the event in Site 1134 (Li et al. 2003).At ODP Leg 199 Site 1218 (~8°N) the extinction is within Chron C10n.1n (Wade et al. 2007), which was used to provide the calibrated age in Wade et al. (2011).
In the equatorial Atlantic Leg 154 (Ceara Rise, ~4-5°N) the extinction of C. cubensis was shown to be abrupt and synchronous at four different sites (Sites 925,926,928 and 929) (Pearson and Chaisson 1997).This is consistent with findings from the eastern equatorial Pacific namely DSDP Leg 9 Site 77 (~0°N) which again uses 87 Sr/ 86 Sr stratigraphy and finds an age consistent with the mid Oligocene (Hess et al. 1989).
Higher latitude sites within the Southern Ocean favour the Top occurrence bioevent.Stott and Kennett (1990) erected an Antarctic biozonation scheme with the extinction of C. cubensis occurring in the mid Oligocene between AP14a/AP14b, which is equivalent to the O4/O5 (= P21a/P21b) zonal boundary of the tropical planktonic foraminiferal biochronology (Wade et al. 2011), based upon results of ODP Leg 113 Sites 689 and 690 (Maud Rise, ~64-65°S) correlating to the uppermost part of Subchron C10n.1n.However, a site with a similar latitude at the Kerguelen Plateau (ODP Leg 119 Site 744, ~61°S) found the extinction in Chron C11n, although this may be due to an apparent hiatus meaning that Chron C10n was uninterpretable (Huber 1991, Barron et al. 1991).Conversely abrupt extinctions were found in Sites 747 and 749 within Chron C10n, while at Site 748 the extinction occurred within the combined Chron C9/C10 based on lower latitude sites of ODP Leg 120, again from Kerguelen Plateau (~54-58°S) (Berggren 1992).Huber and Quillévéré (2005) revised the Antarctic Paleogene planktonic foraminifera biochronology although the timing of the extinction event and Chron calibration (C10n.1n)from Stott and Kennett (1990) remained unchanged the biozone names adopted were AO2/AO3, with calibrations for the Top C. cubensis being based upon material from ODP Leg 119 Site 738 (Kerguelen Plateau, ~62°S).While the authors used and found the "Last Appearance Datum" (= Top) to mark the lower/upper Oligocene boundary, they preferred the use of Tc due to occurrences in the upper Oligocene outside of the studied sites.

Chiloguembelina cubensis
Despite the evidence for an abrupt extinction of  (Stickley et al. 2004).
In the equatorial Pacific a number of sites showed an extended range.As previously mentioned, Leckie et al. (1993) used the Tc occurrence of C. cubensis to mark the base of Biozone P21b on ODP Leg 130 Site 803 (Ontong Java Plateau, ~2°N) in the western Pacific and found the species to be a persistent, but rare component, of the planktonic foraminiferal assemblages up until Chron C7n (Biozone P22).In the eastern equatorial Pacific DSDP Leg 9 Site 77 (~0°N) and Site 79 (~7°N), the authors suggested C. cubensis ranged into the late Oligocene and used the species as a marker across the Leg 9 sites, although no palaeomagnetics are available for these two sites (Jenkins and Orr 1972).
Site U1336 (~7°N) also used the Tc C. cubensis occurrence to mark the base of Biozone O5, although the range charts at this site do not list any occurrences above Biozone O4 (Expedition 320/321 Scientists 2010b), while other sites of the eastern equatorial Pacific (DSDP Leg 34 Sites 320 and 321, 9°S and 12°S) found a range to the early Miocene (Quilty 1976).
Comparable latitude sites in the North and South Atlantic found ranges extending into late Oligocene.Jenkins and Srinivasan (1986) examined the south east Atlantic Site 360 (~35°S) of DSDP Leg 40, in addition to the Pacific sites previously discussed, and found the level to be consistent with the other sites analysed as part of the study.Similarly, at ODP Leg 101 Site 628 (~27°N) the range of C. cubensis is extended slightly beyond the level recorded at Site 803, although the site lacked palaeomagnetic calibration but again C. cubensis was found to be rare in the upper Oligocene so the Tc occurrence was used to mark the base of Biozone P21b (Leckie et al. 1993).In a number of European sections, Tc C. cubensis was used to define the base of Biozone O5.The Fuente Caldera in Southern Spain (~37°N) has a relatively continuous record of C. cubensis with peak abundances occurring in Biozone O2, while the base of Biozone O5 being tentatively placed between the Tc and Top occurrence of C. cubensis (Alegret et al. 2008).In the Northern Adriatic Sea (~45°N) C. cubensis is shown to have a range into the upper Oligocene, although the occurrences in Biozone O5 are generally less than 5 %, therefore the lower/upper Oligocene boundary here is marked by the Tc of C. cubensis (Hernitz-Kučenjak et al. 2014).This is consistent with the similar latitude site in the Umbria Marche Basin (~43°N) in Central Italy, with three sites finding the Tc C. cubensis occurrence within the upper part of Chron C10n.The Tc occurrence of C. cubensis at this locality is the current correlation event for defining the GSSP for the base of the Chattian, with the favoured section being the Monte Cagnero section (Coccioni et al. 2008(Coccioni et al. , 2013)).

Reworking
As C. cubensis is a small planktonic foraminifera species, it is relatively susceptible to sediment reworking (e. g.Poore et al. 1984, Pearson 1995, Wade et al. 2007).Detecting such a process is important in order not to give a false stratigraphic range.For example, a number of authors have directly attributed reworking to younger occurrences of C. cubensis including at DSDP Leg 9 Sites 77 and 79 (Jenkins and Orr 1972), DSDP Leg 73 Site 522 (Poore 1984), ODP Leg 119 Site 744 (Huber 1991) and ODP Leg 126 Site 793 (Herman 1992).Conversely, Hornibrook (1990) discounted reworking of C. cubensis on Chatham Island due to limited outcrops of Oligocene age and the delicate nature of the specimens.Some sites do not discuss reworking and rather directly attribute the occurrence as representing a true stratigraphic range.For example, DSDP Leg 34 suggests a range into the early Miocene, despite only a single specimen of C. cubensis being present within one Miocene aged sample at Site 320, and while Site 321 shows a more continuous stratigraphic range that is restricted to the Oligocene, the age is unclear due to difficulties in biozone correlation (Quilty 1976).
It is worth noting that all of the studies discussed in this section predate Leckie et al. (1993) and Berggren et al. (1995) where Tc became the favourable bioevent.However, due to the limited availability of abundance counts (relative or absolute), it is difficult to assess whether the Tc occurrence of C. cubensis could have been applied at these sites, instead of solely a Top occurrence with all younger occurrences being linked to sediment reworking.

Other potential markers
for the base of the Chattian Van Simaeys et al. (2004) argued that the extinction of C. cubensis is diachronous and latitudinally controlled, so cannot be considered a suitable marker for GSSP assignment.Instead the best marker for the boundary in the southern North Sea basin is a decrease in benthic foraminiferal diversity and bloom in the abundance of the benthic foraminifera Asterigerinoides guerichi.
The authors state that while the latter of these events is correlatable through the North Sea basin, the species is endemic to this region, so based upon the criteria for GSSP definition set out by Remane et al. (1996) this bioevent is not widely correlatable outside of this region, so is an unsuitable bioevent for GSSP definition.
Other markers listed as potential correlation events in the Umbra Marche Basin include the Tc of the calcareous nannofossil species Sphenolithus distentus, the abundance interval of the dinoflagellate genus Svalbardella or the oxygen isotope increase (Oi-2a), with the latter of these having preference due to the potential for global correlation (Coccioni et al. 2008).Oi-2a represents a short lived (70-130 kyr) positive oxygen isotope excursion seen in both planktonic and benthic foraminiferal records, modulated by eccentricity and obliquity cycles, leading to Antarctic ice sheet expan-sion and a decline in global sea levels (Wade and Pälike 2004).

Study sites
Ocean 8.394Ј W) and contains a complete Oligocene between cores 11H to 26X.Like Site 1237, the site is characterised by pelagic nannofossil ooze, although the sedi-mentation rate is considerably higher at an average of 24 m/my in the lower Oligocene, decreasing to 14 m/ my in the upper Oligocene (Expedition 320/321 Scientists 2010a).
When the two sites are backtracked it can be seen that Site 1237 was located in an open ocean setting compared to the current day locality (Fig. 2) at a palaeolatitude of approximately 19°S in the mid Oligocene within an oligotrophic gyre (Shipboard Scientific Party 2003).Conversely, Site U1334 was located in an area of complex oceanography within the equatorial upwelling zone, meaning productivity was likely higher relative to other areas of the Oligocene ocean (Expedition 320/321 Scientists 2010a).The backtrack of Site U1334 (Fig. 2), indicates a palaeolatitude of approximately 2°N in the mid Oligocene (Pälike et al. 2012).

Materials
From Hole 1237B, we examined 36 samples from core sections 30H-1 to 34H-5, equivalent to 272.35 to 316.37 metres below sea floor (mbsf) or 305.96 to 357.48 metres composite depth (mcd), covering ~24.2 Ma to ~32 Ma.To accurately quantify the biostratigraphic changes and, ultimately the extinction level of C. cubensis, the samples between 31H-3 to 32H-4 (284.9-296.4mbsf) were studied every 50 cm and increased to 3 cm resolution between 31H-7 to 32H-1.The samples above 31H-3 and below 32H-4 are at a 3 m resolution.
From Hole U1334A, 38 samples were studied, corresponding to 141.64 to 198.14 metres core depth below seafloor (m CSF) and equivalent to 162.39 to 239.38 m CCSF-A (metres composite core depth below sea floor) from Westerhold et al. (2012).The core sections covered 16H-1 to 21H-7, spanning Chron C9n to Chron C11n.2n, equivalent to ~26.7 Ma to ~30.5 Ma.The data was plotted relative to the depth scale of Westerhold et al. (2012), using the available composite shipboard magnetostratigraphy for all three drilled holes at Site U1334 (Expedition 320/321 Scientists 2010a).

Abundance counts
Samples were wet sieved through a 63 μm mesh and the 63-150 μm size fraction was examined using a binocular light microscope.For the relative abundance counts a total of 300 planktonic foraminifera specimens were counted with the number of biserial planktonic forms (C.cubensis) recorded relative to the number of nonbiserial planktonic forms.The dominance of radiolarians within the Hole U1334A samples meant relative abundance counts were not possible for the samples in which planktonic foraminifera constituted a low percentage of the entire microfossil assemblage.In order to account for this, absolute abundances were applied for all samples from Hole U1334A and took into account the number of biserial planktonic forms for a single picking tray.In each case a similar amount of sediment was spread across the tray in order to eliminate bias.

Scanning Electron Microscope imaging
Twenty-three and twenty-five specimens were imaged using the scanning electron microscope from Holes 1237B and U1334A respectively.These specimens were mounted on a 10 mm SEM stub, before being coated in gold-palladium using a Gatan ion beam coater.They were imaged using a JEOL JSM-7401F scanning electron microscope at University College London.SEM imaging allowed for clearer distinction between the biserial species, which relied primarily on the taxonomic concepts presented in Huber et al. (2006).

Morphometric analysis
For Hole 1237B and Hole U1334A, morphometric analysis was conducted on 59 and 69 specimens respectively.The selected specimens were mounted on a picking tray in lateral view and measured using an Olympus S2X7 microscope and Image-Pro Premier Computer Software.For each specimen the maximum test length was measured automatically, while the maximum test width was measured manually.

Magnetostratigraphic record and updated age-depth model for Site 1237
As  Flower and Chisholm (2006).This allowed for the definition of biozones, with the Top or Base of key index planktonic foraminifera fossils providing a core level for the bioevents, which in turn acted as a means for calibrating the magnetic polarity chrons.The calibration ages for the planktonic foraminifera came from Wade et al. (2011) and the ages for the Chron boundaries are from Pälike et al. (2006).The age-depth model allowed a reassessment of the sedimentation rate for the Oligocene and lower Miocene to be made.

Oligocene magnetostratigraphy for Site 1237
To accurately assess the timing of the extinction of C. cubensis at Site 1237, a re-examination of the magento-and biostratigraphy was required.The shipboard palaeomagnetic record for the interval from 290 to 319 mbsf was not interpreted during the expedition, so The extinction of Chiloguembelina cubensis in the Pacific Ocean 317 we have reassessed the magnetostratigraphy with new and revised planktonic foraminifera biostratigraphy.A near complete Oligocene and basal Miocene sequence has been inferred from the available palaeomagnetic and biostratigraphic data (Figs.3, 4; Table 1).The Chron interpretations were based primarily on the declination data as this showed a clearer degree of variance between normal and reversed polarities, with the inclination data acting as a secondary means to evaluate Chron assignment (Fig. 3).
Our upper Oligocene magnetostratigraphic interpretation is broadly similar to the findings of the Shipboard Scientific Party ( 2003) and Flower and Chisholm (2006).The record spans Subchron C10n.1n at 290.85 mbsf to Chron C6Cn.2n at the Oligocene/ Miocene boundary in core 29H-4 (Fig. 3).The base of Biozone O5 is determined by the Top occurrence of C. cubensis between samples 31H-7, 75-77 cm and 32H-1, 35-37 cm, at 290.84 Ȁ 0.02 mbsf, within Subchron C10n.2011), except C. cubensis which is from this study.Oligocene ages are calibrated to the astronomical timescale of Pälike et al. (2006), while Miocene ages are calibrated to the astronomical timescale of Lourens et al. (2004).Primary marker species are in bold.299.05 mbsf (32H-6) the magnetostratigraphy has not been interpreted.This contains the Subchron C10n.2n and subsequent short-term reversal (C10n.1r),but there are no clear changes in the inclination or declination data and so no interpretation can be made.
One drawback with the biostratigraphy from Shipboard Scientific Party ( 2003), is the 9 m resolution of the core catcher samples creating a large degree of error in the depth of the Top or Base of the index fossils from this site.However, in order to reduce this error both biostratigraphic and magnetostratigraphic datums have been used to calibrate the depths but this still remains problematic for the Base of ʻGlobigerinaʼ angulisuturalis, which is the marker for the base of the Zone O4 within Chron C11n.1n (Wade et al. 2011).The base of ʻGlobigerinaʼ angulisuturalis at Hole 1237B could not be resolved and therefore Zones O3 and O4 are undifferentiated.A higher resolution study on the Oligocene assemblages of planktonic fora minifera at this site would provide a better means of calibration in resolving the polarity Chrons and planktonic foraminiferal biozones.

Hole 1237B
Planktonic foraminifera are the dominant microfossil component throughout all the samples at Site 1237, allowing for a precise means to characterise the abundance changes in C. cubensis (Fig. 5; Table 2).The three basal samples, between 310.35 to 316.37 mbsf (34H-1 to 34H-5), marks the period of peak abundance in biserial forms which ranges from 8.67-14 % of the total assemblage.The low abundance seen in core section 33H-5 (306.88 mbsf) can be attributed to the anomalousness of the sample, which falls in a band of The extinction of Chiloguembelina cubensis in the Pacific Ocean 319

M1b
Depth (mbsf)   2) in constrained within a 3 cm interval, in which the base of Zone O5 is placed at 290.84 Ȁ 0.02 mbsf.

Hole U1334A
Site U1334 was extremely rich in radiolarians, so the abundance of planktonic foraminifera in each sample was controlled by the dominance of radiolarians mean- The extinction of Chiloguembelina cubensis in the Pacific Ocean 321 ing the absolute and relative abundance records of C. cubensis were erratic (Fig. 6; Table 3).However, the two records mirror one another relatively well with only a few discrepancies existing, providing confidence in our data.The basal samples, between core sections 21H-2 to 21H-7, equivalent to 231.8 to 239.38 m CCSF-A have relatively low abundances before a period of marked increase between cores 20H-4 to 21H-1 (224.60 to 230.28 m CCSF-A).Between 224.60 m CCSF-A and up until the inferred base of Zone O5 at 187.74 m CCSF-A there are abrupt fluctuations in the abundance of C. cubensis in which the peaks range from 29-54 specimens, which are separated by one to two samples with low abundances.
Between 17H-3 to 18H-3 (177.02-187.74m CCSF-A), C. cubensis is still present but in extremely low abundance (Յ 1 %), suggesting that the Tc occurrence of the species can be used to define the O4/O5 biozone boundary at 187.74 m CCSF-A.However, the two samples at 174.06 and 175.47 m CCSF-A (17H-1 and 17H-2 respectively), within Chron C9n, have the high-est occurrences of biserial planktonic forms in Hole U1334A.SEM analysis on 10 specimens from these two samples (Plate 4c-4l) confirms that these biserial forms are C. cubensis, so questions the use of Tc occurrence as a boundary marker at this site.A single, broken specimen was found within core section 16H-5 (168.36 m CCSF-A) but was inferred as being reworked due to the poor preservation state and isolated occurrence of the specimen (Plate 4m).

Morphometric analysis
The measured specimens from Site U1334 are larger in width and length compared to those from Site 1237 (Fig. 7; Supplementary Tables S1 & S2; Supplementary Fig. S3).The specimens from Site 1237 tend to show less variability in test size with length and width ranging between 110-160 μm and 75-100 μm respectively.Comparatively specimens from Site U1334 have a larger spread in test size with the majority of specimens having a test length of 170-240 μm and test width between 100-160 μm.Fig. 6.Absolute and relative abundance changes for biserial planktonic foraminifera at Site U1334 between 162-239 metres composite core depth below sea floor (m CCSF-A) of Westerhold et al. (2012).The palaeomagnetic and biozone data come from the initial shipboard report (Expedition 320/321 Scientists 2010a), except for the Base of Zone O6 which is based on Wade et al. (2016).All data has been calibrated to the astronomical chronology of Pälike et al. (2006).The arrow designates the Top and Top Common (Tc) occurrence of C. cubensis at this site.

SEM imaging
The majority of specimens imaged from Site 1237 (Plates 1-2) and Site U1334 (Plates 3-4) possessed striae and so represented C. cubensis.A number of specimens imaged from both sites lacked striae and were tentatively identified as C. ototara (Plate 1c; 3a; 3k; 4a), however these specimens are recrystallized and comparatively poorly preserved so it is difficult to accurately assign these specimens to either C. cubensis or C. ototara.
Anomalous specimens were identified in both Site 1237 (Plate 1i) and U1334 (Plate 4e).While both specimens show the distinctive striae associated with C. cubensis, the test morphology is more variable.The imaged specimen from Site 1237 (Plate 1i) has no clear final and penultimate chamber, with the arrangement of these chambers more closely resembling those of triserial forms.This specimen was recovered from a section in which the lithology changed from nannofossil ooze to partially lithified mudstone and may there- The extinction of Chiloguembelina cubensis in the Pacific Ocean 323 fore be more diagenetically altered than the other specimens hosted in nannofossil ooze.The specimen from Site U1334 (Plate 4e) has more ovate chambers in aperture view with an extremely wide low arched central aperture and more pronounced striae.This specimen does resemble the Mesozoic planktonic form Heterohelix globulosa, however the crust beneath Site U1334 extends only to the Eocene (~38 Ma) (Expedition 320/321 Scientists 2010a) and no evidence of reworking or contamination is present within these samples.This therefore likely represents an end member of the C. cubensis morphospecies as no other specimens bear such an extreme morphology with all the other imaged specimens from this depth being consistent with the C. cubensis species concept.
On the whole the specimens from Site U1334 were better preserved then those from Site 1237, although nearly all of the specimens had some form of diagenetic alteration.The best preserved specimens from Site 1237 were found in core section 32H-1 (291.85 mbsf) within Subchron C10n.1n, while the best preservation in Site U1334 was found in the upper Oligocene specimens in Chron C9n in core sections 17H-1 (174.06 m CCSF-A) and 17H-2 (175.47 m CCSF-A).

Revised age-depth model for Site 1237
The revised age-depth model from the magnetostratigraphic and foraminifera datums shows four main changes in sedimentation rate (Fig. 8), with two periods of slower sedimentation between 306.8-309 mbsf and 277.85-292.8mbsf with average rates of 2.5 m/ my and 3.6 m/my respectively, which are separated by a period of faster sedimentation (8.4 m/my).In the uppermost Oligocene and extending into the lower Miocene (257.31-277.85mbsf) the rate becomes faster with an average of 9.1 m/my.The Shipboard Scientific Party ( 2003) recorded an average rate between 8-9 m/my for the Oligocene, and while this is consistent with the data between 292.8-306.8mbsf, the rest of the Oligocene is generally markedly slower, although it is worth noting that the initial report only accounted for the biostratigraphic datums.
While the majority of data points closely agree with the revised age model, the Top Chron C10r (299.05 mbsf), the Top Tenuitella gemma (276.88 mbsf), Tc Trilobatus primordius (287.60 mbsf) and Base of Globoquadrina dehiscens (257.64 mbsf) plot slightly further from the idealised sedimentation line.However, the two former planktonic foraminifera species represent sec-  ondary marker species, so may not be as well constrained as those primary markers.In the case of G. dehiscens the degree of depth error associated with the planktonic foraminifera bioevents is likely to be responsible.The interval around Top Chron C10r is a period where the inclination and declination became unclear which led to Chron 10n.2n not being interpreted (Fig. 3).

Chiloguembelina cubensis
The primary aim from the analysis of Site 1237 and Site U1334 was to determine the biostratigraphic utility of C. cubensis.The results from Hole 1237B strongly support C. cubensis as a distinct bioevent.The species is observed at 290.85 mbsf (32H-1) and is absent in the next upcore sample at 290.82 mbsf (31H-7), allowing for an extremely precise extinction level due to the increased sampling resolution (290.835Ȁ 0.015 mbsf; Tables 2, 4, Fig. 5).This level is equivalent to upper Subchron C10n.1n, which when directly correlated to various time scales (Table 4) gives an age of 28.285 Ma (Cande andKent 1995), 28.188 Ma (Luterbacher et al. 2004) and 27.888 Ma (Pälike et al. 2006).These findings are consistent with other Oligocene mid latitude sites (20-40°S) from ODP expeditions, namely The Great Australian Bight, Leg 182 Sites 1130 and 1134 (Li et al. 2003), which find an abrupt extinction of The age inferred from Site 1237 is ~100 kyr younger than other sites using the species as a boundary marker, which have an age of ~28 Ma when astronomically calibrated (e. g.Berggren 1992, Wade et al. 2007).This younger age is likely a reflection on the high sampling resolution (3 cm) through the extinction interval.
The results from Hole U1334A are contrasting to those from Hole 1237B.A peak abundance of C. cubensis occurs in the upper Oligocene within Chron C9n, with an astronomical age for the Top occurrence being ~27.15Ma (as per Pälike et al. 2006).This raises questions regarding the biostratigraphic utility of C. cubensis as a reliable marker, at least at Site U1334.Although Tc occurrence of the species has become the widely accepted boundary maker (Berggren et al. 1995, Berggren and Pearson 2005, Coccioni et al. 2008, Wade et al. 2011), this is on the basis of rarer (Ͻ 5%) and sporadic occurrences of C. cubensis in the upper Oligocene (e. g.Huber 1991, Leckie et al. 1993).While this is consistent with the low abundance observed through Chron C9r, the rapidity of the peak abundance and inferred extinction at 172.01 Ȁ 2.05 m CCSF-A within the basal region of Chron C9n (~27.15Ma) presents a previously unobserved phenomenon within C. cubensis.The species is susceptible to reworking (e. g.Poore et al. 1984, Pearson 1995, Wade et al. 2007) and while such a process can never be fully discounted, it is unlikely that this is the reason for this peak in abundance.The planktonic foramini feral assemblages observed are consistent with upper Oligocene taxa, with no evidence of other reworked taxa, and SEM analysis shows that planktonic fora minifera at this level are well preserved.In addition, reworking is not identified through the core 17H interval in other microfossil groups, such as the calcareous nannofossils (Bown and Dunkley Jones 2012) and radiolarians (Expedition 320/321 Scientists 2010a).The lack of reworking in any of the microfossils groups therefore suggests this peak is a true occurrence of C. cubensis.Our new data from Site U1334 contrasts to nearby Site 1218 where Wade et al. (2007) found an extinction in Subchron C10n.1n with no higher recorded occurrences in younger sediments.However, Site 1218 is extremely siliceous rich through this interval and the foraminiferal preservation is inferior to Site U1334 (Wade et al. 2016).In addition, abundance counts were not employed in the planktonic foraminifera biostratigraphy, so these reasons likely account for this discrepancy between the two sites.
Van Simaeys et al. ( 2004) questioned the utility of C. cubensis as a reliable boundary marker on the basis that the extinction event is diachronous and latitudinally controlled.By considering the sites which possess a magnetostraigraphic record through the mid Oligocene and where C. cubensis is present (Fig. 9), it can be seen there is no clear latitudinal control on the extinction level of the species.
The results from Site 1237 and the global comparison dataset (Fig. 9), confirm that there is a large scale global decline in the population of C. cubensis in the mid Oligocene.Previous authors have suggested an abundance threshold of 5% to define the Tc C. cubensis, the peak abundance within Chron C9n at Site U1334 indicates that the Tc occurrence of 5% may not be the best criterion.In addition, the relative abundance datasets presented in this study (Figs. 4 and 5), records from the southern Spain (Alegret et al. 2008) and the northern Adriatic Sea (Hernitz Kucejak et al. 2014) show abundances below 5 % through some periods of the early Oligocene.Therefore, applying the Tc criterion for C. cubensis may prove difficult due to fact the record seems to fluctuate somewhat erratically.By taking into consideration these factors, we suggest that the new phrase "Chiloguembelina cubensis crisis" best describes the event in the mid Oligocene.

Controls on the abundance of Chiloguembelina cubensis at Sites 1237 and U1334
At Site U1334 there is a clear link between the abundance of radiolarians and planktonic foraminifera, as high radiolarian abundance is coincidental with low planktonic foraminiferal abundance and vice versa.A similar link between abundances of these two microfossil groups has been recorded in the Oligocene and Eocene in the equatorial Atlantic Ocean at Ceara Rise (Diester-Haass and Zachos 2003).The abundance of C. cubensis was found to decrease following oxygen isotope (Oi) events in the Adriatic Sea (Hernitz-Kučenjak et al. 2014).However, we find little correlation between the high resolution stable isotope record at nearby Site 1218 (Wade and Pälike 2004) and our abundance data (Supplementary Fig. S1 and S2), suggesting that abundance is not controlled by palaeoceanographic factors.However, the onset of peak abundance in C. cubensis in Chron C9n is near synchronous with a positive shift in the planktonic δ 18 O record (Supplementary Fig. S1) which was attributed to a migration of D. venezuelana to a deeper depth habitat (Wade and Pälike 2004).
The causal mechanism for the peak abundance of C. cubensis at Site U1334 is unclear, reworking has been discounted based on previously discussed evidence (Section 5.2).Boersma and Premoli Silva (1989) found that Cenozoic biserial planktonic foraminifera had the most depleted δ 13 C values and suggested these forms occupied the oxygen minimum zone.We therefore speculate that a localised, temporary expansion of the oxygen minimum zone may have led to this peak in abundance in C9n.The enhanced abundance is coincidental with a period of gigantism in Paragloborotalia opima (Wade et al. 2016) and the migration to deeper depths observed in Dentoglobigerina venezuelana at Site U1334 (Matsui et al. 2016) and Site 1218 (Wade and Pälike 2004; Supplementary Fig. S1), which Matsui et al. (2016) suggested was coincidental with a deepening of the thermocline.The synchronicity of the peak in C. cubensis, gigantism in P. opima and depth migration of D. vene zuelana suggests that these events may have been driven by complex changes in oceano -graphy, at least on a localised scale through Chron C9n at Site U1334.
At Site 1237 the abundance of biserial planktonic foraminifera is greatest in Zone O2.This is consistent with records for the mid-latitude sections of Spain and the Adriatic Sea (Alegret et al. 2008, Hernitz-Kučenjak et al. 2014) at least for the basal part of Zone O2.This peak in abundance is likely due to the co-current range of both C. cubensis and C. ototara through the more basal region of Zone O2, while in the latter part of Zone O2 only C. cubensis is occurring so explains why the relative abundance of biserial forms begins to decline and remains relatively low through the rest of the record at Site 1237.
served in the most optimal conditions (Schmidt et al. 2006), while the largest test sizes for an overall assemblage are found within subtropical oligotrophic gyres (Kucera 2007) with size being found to decline in high productivity, upwelling areas (Schmidt et al. 2004a).While limitations exist, if the life cycle is viewed in terms of the r/K selection theory it should follow that r-selectors reproduce quickly and grow to a smaller size due to a variable or unpredictable climate (e. g. upwelling environments), while K-selectors are associated with the opposite, this being a larger size and slower reproduction associated with predictable and near constant climatic conditions (e. g. oligotrophic gyres) (Pianka 1971).In the fossil record there is little difference in the overall test size of the planktonic foraminiferal assemblages in the mid-high and low latitudes prior to ~42 Ma, but the latitudinal offset becomes more pronounced from the Eocene-Oligocene boundary (~34 Ma) onwards and remains the case until the modern day (Schmidt et al. 2004b).Therefore, by taking these factors into consideration it is possible to deduce what the key control is on the test size of C. cubensis test size at least between these two sites.When compared it can clearly be seen that the specimens from Site U1334 are longer and wider than specimens in Site 1237 (Fig. 7), this therefore suggests the latitudinal offset, and hence temperature, is the most important control on test size at least in terms of the studied sites.

Conclusions
Based on our quantitative biostratigraphic analyses from the Pacific Ocean, the biostratigraphic value of C. cubensis as a correlation event for the base of the Chattian remains unclear.Site 1237 strongly supports the utility of the species as a reliable bioevent, whereas the results from Site U1334 show peak abundances in the late Oligocene (C9n).While the Tc occurrence of C. cubensis has become the favourable bioevent for the boundary marker, this is on the basis that the upper Oligocene occurrences of the species are rarer and more sporadic, which is not supported by our data from Site U1334.Our compilation from ocean drilling and land sections shows that there is no latitudinal control on the extinction level of C. cubensis, however we do find differences in test size between the two studied sites, which we suggest is controlled by temperature.We pen the phrase "Chiloguembelina cubensis crisis" to describe the extinction or reduction in the abundance in C. cubensis in the mid Oligocene.Table S1 Measured test length and width for specimens from Hole 1237B, as well as the inferred aspect ratio (width/length) for each specimen.In total 59 specimens were measured from 12 core sections.

Fig. 1 .
Fig. 1.Global occurrences of Chiloguembelina cubensis from the Deep Sea Drilling Program (pluses), Ocean Drilling Program (unfilled circles), Integrated Ocean Drilling Program (crosses) and other sections (stars) plotted on a bathymetric map extracted from the NOAA server at a 10 m resolution and plotted using the MarMap R package (version 0.9.2).
Drilling Program Leg 202, Site 1237 was drilled in the southeast Pacific Ocean (16°0.421ЈS, 76°22.680Ј W) in 3217 m water depth.Hole 1237B recovered a 55.2 m near complete Oligocene, characterised primarily by pelagic sedimentation of nannofossil ooze with a sedimentation rate between 8-10 m/my (Shipboard Scientific Party 2003), although a revised agedepth model is presented in this study.Integrated Ocean Drilling Program Expedition 320 Hole U1334A (4790 m water depth) is located in the eastern equatorial Pacific Ocean (7°59.998ЈN, 131°5

Fig. 2 .
Fig. 2. Cumulative average sea-surface chlorophyll concentrations (mg m 3 ) of the eastern Pacific Ocean for the period September 1997 to November 2010 derived from the SeaWiFS satellite (http://oceancolor.gsfc.nasa.gov/).The unfilled circle and diamond points show the backtracked positions for Site U1334 (Pälike et al. 2012) and Site 1237 (Shipboard Scientific Party 2002) respectively across the Eocene -Oligocene (E/O), Rupelian-Chattian (R/C), Oligocene -Miocene (O/M) and at 10 Ma.The filled points show the present day positions of Site U1334 and Site 1237.

Fig. 3 .
Fig. 3. Palaeomagnetic interpretation for Site 1237 based upon the available inclination and declination data from the Shipboard palaeomagnetics survey (http://www-odp.tamu.edu/database/)and biostratigraphy (Shipboard Scientific Party 2003, Flower and Crisholm 2006 and this study).The ages of the biostratigraphic datums come from Wade et al. (2011), except C. cubensis which is from this study.Oligocene ages are calibrated to the astronomical timescale ofPälike et al. (2006), while Miocene ages are calibrated to the astronomical timescale ofLourens et al. (2004).Primary marker species are in bold.

Fig. 7 .
Fig. 7. Comparison of the length and width of measured tests of Chiloguembelina cubensis at Hole 1237B and Hole U1334A.

Fig. S2 .
Fig. S2.Absolute and relative abundances of biserial planktonic foraminifera from Hole U1334A plotted against the δ 13 C benthic and planktonic foraminifera records, and oxygen isotope (Oi) events intervals at ODP Leg 199 Site 1218(Wade and  Pälike 2004).
large proportion of the sites which find late Oligocene occurrences of C. cubensis are in the south west Pacific and Austral Realm.
C. cubensis, a number of other study sites have shown an extended range of C. cubensis into the late Oligocene.A part of this study, we constructed the lower Oligocene palaeomagnetic record at Site 1237 as this data was not interpreted by the Shipboard Scientific Party(2003).The available data was downloaded from the Ocean Drilling Program database (http://www-odp. tamu.edu/database/) with both the inclination and declination data being considered.The record was analysed between 255.01 to 319.19 mbsf and consisted of 3946 data points.In order to better identify normal and reverse polarities a five period moving average was fitted to both datasets.

Table 1
Lourens et al. (2004)used in the construction of the magnetostratigraphic record for Hole 1237B (Fig.4).The Fig 4 tie points relate to the numbered crossed in Fig.4.Biostratigraphic datum depths are from Shipboard Scientific Party (2003), except Top Tenuitella gemma and Chiloguembelina cubensis (this study) and Paragloborotalia opima(Flower and Crisholm 2006).Biostratigraphic datum calibrations are fromWade et al. (2011), except C. cubensis (this study, see Table4for calibration details).Oligocene ages are calibrated to the astronomical timescale ofPälike et al. (2006), while Miocene ages are calibrated to the astronomical timescale ofLourens et al. (2004).Planktonic foraminifera bioevents are to two decimal places in the Neogene and one decimal place in the Palaeogene.

Table 2
Relative abundance data for Hole 1237B showing samples, depths, ages and the inferred Chron and biozone for each sample.

Table 3
Relative (Rel.Abun.) and absolute abundance (Abso.Abun.) data for Hole U1334A showing samples, depths, ages and the inferred Chron and biozone for each sample.
Pälike et al. (2006)2004)embelina cubensis in the Pacific Ocean 329 Revised age depth model for Site 1237, taking into account datums presented in Table1.Table4Depth and age of Top C. cubensis in Hole 1237B.Calibrated ages to the palaeomagnetic chronologies ofCande and Kent (1995)andLuterbacher et al. (2004), and the astronomical time scale ofPälike et al. (2006).
The extinction of Chiloguembelina cubensis in the Pacific Ocean 331