New species of the genus Parkiella (Foraminifera) from the Late Cretaceous Central Pacific Ocean: biostratigraphy, biogeography, and the Cretaceous–Palaeogene boundary

Two new species, Parkiella angulocamerata sp. nov. and P. globocamerata sp. nov., are described from the Late Cretaceous (late Maastrichtian) Central Pacific Ocean, DSDP Sites 465 and 171. Examination under the SEM revealed apertural features that were not documented in the original description of Parkiella. An ‘L’-shaped aperture was originally considered as a diagnostic character of Parkiella; we suggest that this is a preservational artefact and question its diagnostic value. The existence of internal toothplates determine both Parkiella described here as members of the family Turrilinidae Cushman 1927. Both species were found to be endemic to the tropical Pacific Ocean. They occurred in sequence throughout the Maastrichtian section at Site 465 with a minimum (one sample) stratigraphic overlap. Parkiella globocamerata is one of the few deep-sea benthic foraminiferal species that indeed became extinct at the Cretaceous/Palaeogene (K/Pg) boundary. However, we document a decrease in both relative abundance and accumulation rate of the species already prior to the K/Pg transition. We propose that the interval between the LO (Last Occurrence) of P. angulocamerata and the LO of P. globocamerata is indicative of uppermost Cretaceous (Micula prinsii Zone) strata in the tropical Pacific.


INTRODUCTION
The benthic foraminiferal fauna from the uppermost Cretaceous and lowermost Palaeogene strata at Site 465 (and from the Late Cretaceous of the Central Pacific Ocean in general) has not yet been described in detailapart from the documentation of the most common taxa by Widmark & Malmgren (1992a, b) and short communication by Widmark & Henriksson (1995). In this paper we describe two new species typical for this fauna, discuss their taxonomic position, biostratigraphic significance, and biogeographic distribution.
During a study on benthic foraminiferal changes across the Cretaceous/Palaeogene (K/Pg) boundary at three DSDP sites, one from the Pacific Ocean (Site 465) and two from the South Atlantic Ocean (Sites 525 and 527), Widmark & Malmgren (1992a) encountered several taxa that had not been previously described in the literature. Among these was a 'buliminid' species from Site 465, tentatively named Buliminella? sp. A. This species, here described under the name of Parkiella glohocamerata sp. nov. and a probable closely related species, Parkiella angulocamerata sp. nov., are the subjects of the present study.
It has been suggested and later demonstrated that deep-sea benthic foraminifera (as a group) were not severely affected by the biotic crisis at the K/Pg boundary (e.g. Douglas & Woodruff, 1981;Emiliani et al., 1981;Berggren, 1984;Hansen et a/., 1987;Thomas, 1990;Nomura, 1991;Kaiho, 1992;Widmark & Malmgren, 1992a) compared to the vast majority of planktonic organism and shallow benthic groups. Some benthic foraminiferal taxa did, however, respond to this event in that they became extinct or declined drastically in abundance across the K/Pg boundary. Bulirninella? sp. A. (= Parkiella globocamerata sp. nov.) was among the relatively few taxa that significantly decreased in relative abundance across the K/Pg boundary (Widmark & Malmgren, 1992a). Here, we present a detailed account on its response to the K/Pg boundary event in terms of both relative abundances and accumulation rates.

MATERIAL AND METHODS
The studied material was obtained from an 84m long section including the Maastrichtian, the K/Pg boundary, and the lowermost Palaeocene at the Pacific DSDP Site 465 (Hole 465A, southern Hess Rise; Fig. 1, Table 1); a lithostratigraphic column of the section investigated is given in Fig. 2. The upper part of the section (Cores 3-8) consists of foraminifernannofossil chalk and ooze and nannofossil ooze; the K/Pg boundary occupies a 'mixed zone of Danian and Maastrichtian sediments', approximately 30 cm thick ( Fig. 2A) (Thiede et a/., 1981a). Below Core 3, the section is divided by a thick cherty drilling breccia represented by four short cores (4, 6, 7, 8; Core 5 is missing) between 67.5 and 105.5 mbsf (Thiede et al., 1981a). The lower part of the sequence (Cores 9-12) consists also of foraminifer-nannofossil ooze and nannofossil ooze and chalk with occasionally higher chert contents (Fig. 2B) Maastrichtian at Site 465 was studied using the fragmentation of planktonic foraminifera following the method described in Malmgren (1987). Usually more than 300 specimens per sample were counted to determine the proportion of fragments and its precision (95% confidence intervals). Two intervals of strong dissolution ( > 45% of fragments of planktonic foraminifera) were identified (Fig. 3A). The first interval is represented by one pyrite containing sample in the 'mixed zone' above the K/Pg boundary. The second interval (identified in the four lowermost samples from Core 3) coincides with the upper range limit of the cherty drilling breccia. Remaining samples are relatively well preserved and the fragmentation of planktonic foraminifera indicates low to moderate calcium carbonate dissolution during their deposition. Altogether 31 samples from Hole 465A were analysed with a closer sampling towards the K/Pg boundary in order to obtain a higher resolution at this transition (  (Berggren et al., 1985;Henriksson, 1993) (Table 3). The accumulation rate (number of specimens/ kyr/cm2) was used to visualize the changes in abundance of P. globocameratu sp. nov. across the K/Pg boundary. It approximates a measure of population density and does not depend on fluctuations in abundances of other species within the rest of the benthic foraminiferal fauna. The number of specimens/kyr/cm2 was calculated as D*N*S, where D is the density of the sediment (set to 1.925 g/cm3 in this study), N is the number of specimens per gram sediment and S is the sedimentation rate in cm/kyr as given in Table 3.   Table 2. Data for DSDP samples used in the present study. 'Weight' is total dry sample weight, 'N' is the number of Parkiella spp. nov. specimens encountered in each sample, and 'B' is the total number of benthic foraminifera in samples. 'Zone' refers to planktonic foraminifera zones as identified by Boersma (1981). Sample 20 was barren of foraminifera, and sample 23 was not included in the study due to poor preservation. K/Pg denotes the Cretaceous-Palaeogene boundary. Zone abbreviations: Geu = "Globigerina' eugubina; Mixed = 'mixed zone' of Madstrichtian and Danian sediments, Abm = Abuthomphalus mayaroensis; Cco = Giobotruncana contusa; Gga = Globotruncana gunsseri;

SYSTEMATIC DESCRIPTIONS
The genus Parkiella was recently established by Seiglie et al. (1993) on the type species Buliminella gabonica de Klasz and RCrat, 1962. It includes a group of morphologically distinct species described from Upper Cretaceous shelf deposits from West Africa (Cameroon and Gabon) by de Klasz & Rerat (1962) and de Klasz et al. (1963). The diagnosis of Parkiella is based on its 'low trochospiral coil' and 'complex 'L'-shaped aperture' (Seiglie et ul., 1993). These characteristics together with the generally lower number of chambers/whorl separate Parkiellu from Buliminella Cushman 1927 and Praebulimina Hofker 1957 according to Seiglie et al. (1993)  Parkiella on the basis of apertural characteristics .  found it problematic, however, to place Parkiella in a higher systematic category, since the specimens in their well samples were completely replaced by pyrite or completely pyrite infilled, which made it impossible to observe any internal apertural features. Nevertheless, based on the 'mophological characteristics' alone, Seiglie et al. (1993) placed Parkiella in the Turrilinidae Cushman 1927. The main argument for this allocation was that a taxon with such morphological characteristics would possess a toothplate and, therefore, should be placed in Turrilinidae . A close examination under the SEM elucidated some new apertural features in our Parkiella species. Well preserved apertures were found to be equipped with a delicate apertural flap (Fig. 4G), which covers an analogue of the periapertural depression as in Bulimina (see Verhallen, 1986). In addition, an apertural lip ( Fig. 4B-C, G-H) is present in most specimens. The apertural flap is, however, often broken off as shown in Fig.  4B-C, H-I. In even worse preserved specimens the aperture is excavated along the basal suture of the last chamber into an 'L'shaped opening (Fig. 4C, H-I); sometimes even the apertural lip is destroyed by postmortal processes (Fig. 41). These 'L'-shaped openings observed in both species described here is, hence, a preservational artefact. Although we did not have the possibility to examine the type material of Parkiella, the validity of this character as diagnostic of this genus has to be questioned. In order t o examine the internal apertural features of our between such apertural features and 'turrilinid test-morpholo-Parkiellu spp. nov., the last chambers in specimens of each gies' as suggested by Seiglie et al. (1993). species were removed (Fig. 4A, D). Both species exhibit a short The finely perforate wall structure of P . globocamerata sp.
internal toothplate, which connects the 'periapertural depres-nov. is demonstrated in Fig. 4F; each pore seems to be sions' of successive chambers (Fig. 4A, D-E). The presence of a surrounded by six radially arranged calcite crystals. No distinct toothplate in our Parkiellu spp. nov. determines them t o be dimorphism in terms of micro-and macrospheres could be placed in the family Turrilinidae a n d supports the relationship observed in either of the species described herein.

BIOSTRATIGRAPHIC DISTRIBUTION AND THE K/PG BOUNDARY
Parkiella angulocamerata sp. nov. and P. globocamerata sp. nov. occurred in sequence throughout the studied section at Site 465 (Table 2) with almost no overlap in their respective vertical ranges. Parkiella angulocamerata was found in the lower part, whereas P. globocamerata was encountered in the upper part of the sequence. The lower range limit of P. angulocamerata sp. nov. could not be satisfactorily determined due to insufficient sampling and recovery below Core 12. Sample no. 30 represents the lowermost level, at which the species was found; no specimens of the species were encountered below that level, either in the lowermost part of Core 12 (sample no. 31), or in the chert drilling breccia from Core 14 during a preliminary survey of sample 465A-l4cc, 0-2 cm. The uppermost sample containing P. angulocamerata sp. nov. (sample no. 21) also contains the lowermost specimens of P. globocamerata sp. nov., which indicates that both species may have lived at the same time.
Parkiella globocamerata sp. nov. was found in samples from the interval between the upper limit (sample no. 21) of the 38.6m thick chert and the 'mixed zone' at the K/Pg boundary (sample no. 3). The lower range limit of P. globocamerata sp. nov. lies between the bottom of Core 3 and the top of Core 10. No specimens of Parkiella were found in this 60m thick interval; Cores 4-7 are characterized by high chert content resulting in poor foraminiferal preservation and Cores 8-9 by exceptionally low dissolution (Fig. 3A) resulting in generally lower abundance of benthic foraminifera (Table 2). Relative abundances of both species were determined in all samples, except samples no. 20 and no. 23, which were too poorly preserved to yield reliable foraminiferal counts. Parkiella angulocamerata sp. nov. fluctuated in relative abundances between 1.4 and 2.2% in Cores 12-10, declined to 0% through Cores 9-7, and increased to 0.8% in Core 3, sample no. 21 (Fig.  3B). The proportion of P. globocamerata sp. nov. steadily increased in the lower part of Core 3 from 1.0 to 2.5%. It suddenly declined to 0% between 63.91 and 63.50 mbsf, and above this interval it fluctuated between 0.5 and 1.3%, with a single excursion to 1.9% in sample no. 8. Figure 5 illustrates the decline across the K/Pg boundary of P. globocamerata sp. nov. in terms of accumulation rates (number of specimens/kyr/cm2) plotted against the depth in hole; note the lag in extinction just above the boundary due to the 'mixed zone of Maastrichtian and Danian sediments' of Thiede et al. (1981a). Accumulation rates for P. globocamerata sp. nov. declined from a mean Maastrichtian value of about 3.3 specimens/kyr/cm2 to about 0.3 in the 'mixed zone', and finally to zero above the 'mixed zone' in the lowermost Palaeocene (Fig. 5); note that the decrease in relative abundance above 63.5 mbsf can also be observed in terms of accumulation rates. Employment of accumulation rates, a proxy for population density, enables us to conclude that the population of P. globocamerata sp. nov. indeed declined already below the K/Pg boundary, i.e. before it became extinct, and that the simultaneous decrease in relative abundances of this species is not simply the result of a drastic increase in absolute abundance of other species within the benthic foraminiferal fauna.
The deep-sea benthic foraminiferal taxa that became extinct at the K/Pg boundary were mainly endobenthic and they are assumed to have been generally more dependent on relatively high food-fluxes from the euphotic zone than epibenthic taxa (e.g. Thomas, 1990;Widmark & Malmgren, 1992a). The extinction of P. globocamerata sp. nov. suggests that the species could have had an endobenthic mode of life and needed the relatively high food-fluxes that were provided by the primary producers in the eutrophic zone. When a large portion of these primary producers became extinct at the end of the Cretaceous, there was a devastating decline in food availability at deep seafloor and some eutrophic taxa, such as P. globocamerata sp. nov., could obviously not sustain the new, rougher conditions that emerged at the beginning of the Cenozoic. The decrease in both relative abundance and accumulation rate of the species observed about 1 m below the K/Pg boundary at Site 465 may represent a prelude to the environmental disturbances that caused the biotic crisis at this transition.
The upper range limits of both species are well defined and documented at Site 465. Both limits lie within Core 3, where sediment recovery and stratigraphic control are excellent. The restricted vertical distribution of both species makes them biostratigraphically useful and we propose that the interval between the LO (Last Occurrence) of P. angulocamerata sp. nov. and LO of P. globocamerata sp. nov. may indicate uppermost Cretaceous (upper Maastrichtian) in the tropical Pacific. This interval is roughly comparable to the calcareous nannofossil Micula prinsii Zone at Site 465 (see Fig. 6). Widmark (1995) conducted a biogeographic survey encompassing terminal Cretaceous (Abathomphalus mayaroensis Zone) DSDPjODP material from all over the world (  fig. 5), which also is identical to P. globocamerata sp. nov. The only information provided on the vertical distribution of Buliminella sp. ( = P . globocamerata sp. nov.) was its occurrence in Maastrichtian-?early Palaeocene strata at Site 171 (Douglas, 1973).

BIOGEOGRAPHIC DISTRIBUTION AND ITS EVOLU-TIONARY IMPLICATIONS
In the two samples available from Site 171 we encountered P. globocamerata sp. nov. in the upper sample only, whereas P. angulocamerata sp. nov. was found in both samples; this corresponds to the stratigraphic succession we observed at Site 465. The relative abundances of Parkiella in the samples from nov. and the Maastrichtian planktonic foraminifer (Boersma,198 I), and nannofossil (Cepek, 1981;Henriksson, 1993) zones recognized at Hole 465A. Last Occurrences (LOs) of both Parkiella species are well defined within Core 3, which enables us to propose that the interval between the LO of P . globocamerata sp. nov. and LO of P . angulocamerata sp. nov. is indicative of upper Maastrichtian strata in the Central Pacific. This interval is slightly longer than the nannofossil Micula prinsii Zone identified at Site 465 by Henriksson (1993). Cores 4 7 were omitted in the figure; vertical scale changes below Core 3.
Site 171 are, however, significantly higher (6.6 and 8.9%; see Table 2) than at Site 465. This may indicate that the palaeoenvironmental conditions at Site 171 were closer to the ecological optimum of the two species. Both Sites 171 and 465 were located within the tropical Pacific sector of the Tethyan realm at the end of the Cretaceous (Fig. I). Consequently, we can assume that during at least the Late Maastrichtian both P . globocamerata sp. nov. and P . angulocamerata sp. nov. were endemic to the Pacific realm. Other representatives of the genus Parkiella, including Parkiella gabonica, P. hrevispira and P . mamelligera, have only been reported from shelf deposits from a restricted area in Gabon and Cameroon, West Africa (de Klasz & Rerat, 1962;de Klasz et al., 1963;. Figure 7 illustrates the stratigraphic ranges of species referrable to the genus Parkiella. Two scenarios of the phylogenetic relationship between the geographically separated West African and Central Pacific stocks are possible: (I) a common ancestor of both stocks occurred (at the latest) in the Turonian; (2) the Central Pacific stock descended from P . mamelligera, the only representative of the 'older' West African stock that possessed spines (Fig. 7). A third alternative might be   Widmark (1995). Localities (sites) where Parkiella was encountered are highlighted in bold. *Data on pakdeolatitudes for the sites investigated were obtained (or estimated) from various DSDP and ODP reports. that our new species were not at all related to the West African stock, and represented an independent lineage within the turrilinids (Fig. 7). In the latter case, P . globocarnerata sp. nov. and P . angulocamerata sp. nov. should not be included in the genus Parkiella, but be considered as a separate deep-water genus morphologically parallel to shallow-water Parkiella.