Linings of agglutinated Foraminifera from the Devonian: taxonomic and biostratigraphic implications

The organic linings of agglutinated foraminiferans from the Devonian are documented and described. These linings have been recovered in palynological residues from Australia, France, Pakistan and Siberia and range from the Lochkovian to the Frasnian. Six species are described as new: Hemisphaerammina coolamon, Psammosphaera garraay, Reophanus proavitus, Saccammina mea, Saccammina wingarri and Thurammina mirrka. Three species, with a wide geographical spread and a relatively limited stratigraphic range, may prove to have some utility in intercontinental correlation: Inauris tubulata Conkin & Conkin, Saccammina mea n. sp. and Saccammina wingarri n. sp.


INTRODUCTION
The organic linings of rotalid foraminifera from the Permian to the Recent are well documented (Stancliffe, 1989), but only recently have linings of Palaeozoic agglutinated foraminifera been reported (Winchester-Seeto & Bell, 1994). Continuing study of material from Australia, France, Siberia and Pakistan has shown that Ordovician to Upper Devonian agglutinated foraminiferal organic linings are commonly present in shallow marine limestones, mark and shales. In earlier works reporting organic linings they have been referred to as microforaminifera because of their small size (typically < 200 pm). We believe that such a term is unwarranted as foraminifera are now known to form part of the sub-63 pm fauna (Gooday, 1986a, b;Pawlowski, 1991) and Burnett (1979) refers to foraminifera of the 10-15pm range; thus these smaller forms are part of the entire foraminiferan size range (albeit a size range seldom studied) and require no special name.
Apart from the Allogromiidae, the various foraminiferal genera are defined as either agglutinate or calcareous. Bender (1995) has shown that most agglutinated genera have an inner organic lining and our specimens are obviously congeneric with described agglutinated genera, but do not have any agglutinant covering. This absence of outer wall material may be an environmental response, or just a preservational or procedural artefact. At present we cannot differentiate between these possibilities, so we prefer to use established genera and to ignore the absence of agglutinating wall material. Hohenneger (1990) has suggested that the Allogromiidae and simple Astrorhizidae (in which our specimens are classified) may be more closely related than previously thought and that the amount of agglutinated material present may have little significance and even be environmentally controlled; our studies support this view.
The purpose of this study is to describe and document the inner, organic linings of agglutinated foraminifera, recovered in palynological processing of Devonian material, and to examine their taxonomic and biostratigraphic implications. In particular, we aim to place the species recovered into a more tightly constrained time-frame than most previous studies by referring to the conodont zones from which they were recovered.

METHODS
The foraminiferal linings described in this study were recovered from samples processed in the quest for chitinozoans, thus the criterion for the selection of sections was the prospectivity for acid-resistant fossils. Many of the sections had also previously yielded conodonts, and strong, reliable stratigraphic control is already in place. Samples yielding foraminiferal linings are exclusively from marine strata, dominantly shallow marine limestones and shales.
Processing methods followed those outlined by Paris (198 l), including initial treatment of 50g of crushed rock with 10% HC1 until all the carbonate had been dissolved. This is followed by acid digestion in 50-70% H F for 1-4 days. Nitric acid (concentrated) was used when necessary for surface etching, the dissolution of fluorite salts and the destruction of amorphous organic matter. The residue was then separated through a 53 pm sieve and the coarse fraction was picked with a micropipette. Well-preserved specimens were selected for examination by conventional scanning electron microscopy or with an environmental electroscanner; the advantages of the environmental electroscanner have been outlined by Winchester-Seeto (1993a).

LOCALITY INFORMATION Early Devonian
The majority of foraminiferal linings were recovered from Lower Devonian sequences spanning the Lochkovian-Pragian boundary (i.e. pesavis-sulcatus conodont zones); this is primarily due to the more intensive nature of the chitinozoan and conodont work undertaken on these areas.
The Garra Limestone, near Wellington, central New South Wales (Figs 1 and 2), yielded the most specimens. It is characterized by grey to dark grey, highly fossiliferous limestones from a subtidal, shallow shelf. Studies of conodonts (Wilson, 1989) and chitinozoans (Winchester-Seeto, 1993b) provide a detailed biozonation for sections MUNG, RUN (pesavis Conodont Zone) and GCR (pesavis-sulcatus conodont zones). A section through the Martins Well Limestone Member of the Shield Creek Formation (MW) from the Broken River area of northern Queensland has been dated as spanning the pesavis-sulcatus conodont zones (Benson & Bear in Mawson et al., 1988), although all foraminiferal linings were recovered from the sulcatus Zone. This limestone is a shallow marine bioclastic calcarenite deposited on a broad, stable shelf (for further details see Wyatt & Jell, 1980;Winchester-Seeto, 1993~). The Amphitheatre Group from the Darling Basin, western New South Wales, represented in the Kewell East bore-core (KE DDHl), yielded a moderate diversity of foraminiferal linings. The interbedded grey to dark grey claystones, carbonaceous shales and siltstones represent a marine environment, possibly a transgressive sequence (Bembrick, 1997). Chitinozoan evidence suggests a Late Lochkovian-Early Pragian age (no younger than sulcatus Conodont Zone; Winchester-Seeto, unpubished data).
The Taravale Formation, Buchan Group, from eastern Victoria, yielded a moderate number of Emsian foraminiferal linings. The section along the Gelantipy Road (Gel. Rd.) extends from the dehiscens to the serotinus conodont zones (Mawson, 1987;Winchester-Seeto, 1996). The succession consists of nodular limestones, shales and impure limestones, probably deposited on a broad, gently sloping marine shelf

GENERALRESULTS
Our studies, so far, indicate that the organic linings of agglutinated foraminifera are fairly widespread both geographically and chronologically. Diverse faunas have been recovered (by palynological processing) from shallow marine environments such as limestones, mark and shales from Ordovician to Late Devonian in age and in localities on three continents (see locality data). Although usually not of great abundance (about 10 per 50g rock sample), they are found in about 50% of samples processed. We have found six of the agglutinate families known from this time span, 12 known genera (plus one indeterminate), of which only two (Hemisphaerammina and Tolypammina) are of attached genera (the others having free tests) and 24 speciessix of established species, seven compared with known species, six new species and five left in open nomenclature because of a lack of specimens. Tubular linings with thin and thick walls are present in many samples from a variety of localities, and may represent broken parts of various genera such as Hyperammina, Rhabdarnmina or Saccorhiza, or may even be Allogromiidae such as Shepherdella, but cannot be further determined and are thus left off faunal lists.
Most of the specimens are highly thermally mature; they are black and many specimens are broken or compressed ~ this may have affected our recovery rates and introduced bias in the types of genera and species preserved. Until comparable work is undertaken in other parts of the world on a wider suite of sediment types, little can be further deduced. The surface of the organic wall may be either smooth or show varying degrees of reticulation, i.e. raised ridges outlining smaller or larger smooth areas. These ridges, we believe, are indications of the outlines of the agglutinate material used in an outer wall, but since lost either by diagenesis or treatment. Those specimens with smooth, unridged surfaces probably had either no agglutinated outer wall or one in which the various grains were sparse and perhaps only very weakly attached. It is well known that some species of foraminiferans show a high degree of grain size selectivity (Heron- Allen, 1915;Petelin, 1970;Bender, 1995;Scott et al., 1998). Thus these species would show fairly uniform reticulations on the outer surface of the organic lining. That the surface of the lining may, however, show differing sizes of reticulations is also consistent with the results of Allen et al. (1998), who have found that several agglutinate species show fractal (i.e. self-similar) grain distribution in the test wall. The study of these aspects of foraminiferan test structure is just in its infancy and how, or if, they may be applied to fossil assemblages lies in the future.
Thin sections of the sampled limestones and shales have proved of little use; the 'normal' foraminiferans are very rare and seldom found in thin section; and those that have been seen have no organic layer present, most likely due to the diagenetic changes in the often partly recrystallized sediments. Both Hedley (1962) and Bender (1995) have commented on the rapid shrinkage and decay of the inner organic lining upon death of a foraminiferan.

Taxonomic considerations
Many species show a variety of 'holes' in the surface of the lining; these are dominantly apertures or pores, but may also be due to breakages caused by diagenesis or in extraction procedures. The occurrence of the main aperture(s) (of the order of 10pm) for species such as Saccammina and Thurammina is easily determined either by size and/or position on a neck or protuberance. Close examination of the test wall, however, shows the presence, quite often, of smaller openings (Plate 1, fig 6); these holes, usually 1-5pm in diameter, are termed pores. The smooth walled species do not have such pores present. Within any one species the pores appear to be of relatively constant size and numbers, but differences occur between species, and is carried to an extreme case in Gen. et sp. indet. (Plate 1, figs 12 and 13) in which the test is heavily perforated. We are not aware of any previous mention of such pores in the test wall of allogromiids. It is also possible that the pores are due to some form of chemical degradation of the test during diagenesis or in the processing of the sediments, or due to boring by parasites or predators.
The various genera can be subdivided into morphological groups (i.e. into species) based on characters such as shape, surface structures, wall thickness, number of apertures, all of which have previously been accepted as specific characters for normal-sized foraminiferans. Figure 3 lists the features used in

Biostratigraphic implications
There have been few attempts to use Palaeozoic agglutinate foraminiferans for biostratigraphic correlation. This is due to a variety of factors, including the small number of studies globally, poor stratigraphic control of the material studied, and the problems involved in differentiating species when the forms are simple.
Of the 24 species identified in this study, eight, so far, are only known from one region of Australia; a further nine species, however, occur in areas of Australia separated by hundreds of kilometres (e.g. Sorosphaera sp. cf. S. confusa  There are a number of very long ranging species or ones with disjunct, long ranges: Webinellinoidea similis, Hyperammina devoniana, Amphitremoidea sp. cf. A . citriniforma, Lagenamminu ovata and Psammosphaera cava (Fig. 4). This list highlights the problems associated with determining species in organisms with a very simple morphology, and may limit the utility of some foraminifera species for biostratigraphy.

SYSTEMATIC DESCRIPTIONS
All figured and type specimens are lodged with the Australian Museum, Sydney, Australia, and are labelled with numbers prefixed with AMF.
Taxonomic conventions used in this study follow Loeblich & Tappan (1988  Type species. Psammosphaera fusca Schulze, 1875 Remarks. The differentiation of species within Psammosphaera has been based on the size and/or the coarseness of the test wall (Moreman, 1930;Dunn, 1942;Mound, 1968), although varying wall thickness, test size and the grain size used were not considered to be reliable indicators for specific diagnoses of such simple organisms by Browne & Schott (1963) or McClellan (1966). We consider that the smoothness or otherwise of the organic lining surface can be used as a diagnostic feature as an indicator of the original wall texture. Loeblich & Tappan (1988: 28) state that Psammosphaera has no inner lining, but Bender (1995) found that P. fusca had an inner organic lining; the presence of reticulate ridges on some individuals suggests that originally these specimens had an agglutinated outer test. A number of specimens of each of the species described here show equatorial splitting into two equal halves which, although, looking like Hemisphaerammina, may be differentiated by their 32 very much thinner wall and rough-edged sutural boundary.
Psammosphaera cava Moreman, 1930  Remarks. Psammosphaera cava is a very simple foraminiferan; the shape may vary from globular to slightly ovate, but this is not a preservational effect. The smooth surface suggests that either the living animal did not have any agglutinate coating, or that any grains were not strongly attached, or that the test wall was made of small grains, as described by Moreman (1 930).
2). Remarks. This species is easily separated from Psammosphaera cava by its reticulate surface (i.e. numerous raised ribs outlining various sized small areas). These raised ribs possibly represent the boundaries of the various sized sand grains that may have once covered the test and have now been lost by some cause (preservational or procedural) not yet understood. The size ranges of P. cava and Psammosphaera garraay are the same. Numerous specimens of P. garraay show a partial equatorial 'tear'. This may be actual splitting or only an attachment scar similar to that seen in the Recent P. fusca Schulze.
(Plate 1, fig. 7 Remarks. The species shows a variable number of grain impressions on the wall, but never any attached grains, and these impressions show a range in size. This species is not as elongate as A . eisenacki (Bell 1996, Conkin & Conkin, 1964 or A . kielcensis Malec (1992: p. 280). Amphitremoida citroniforma has previously been recorded from the Ordovician (Llanvirnian) of northwest Germany (Riegraf & Niemeyer, 1996) and from the lower Silurian of Illinois (Dunn, 1942); our species is only compared with A . citroniforma because of disjunct ranges.  (Bell, 1996). The organic wall is finely reticulate, which agrees with the small, uniform grains used in the test of the normal agglutinated specimens.
(Plate 2, figs 3-5) sulcatus conodont zones. Description. Test free; a flattened, rounded chamber (broken), followed by a short neck; aperture rounded at the end of a neck; wall of body chamber is coarsely reticulate, with larger and smaller defined areas, but the neck is relatively smooth. Dimensions. Diameter of neck, 42-50 pm. Remarks. Our specimens have broken body chambers, and it is not clear what the original shape would have been. Apart from size, Lagenammina sp. is close to L. talenti Bell 1996, but shows a more constricted neck. It is also similar to L. silnica Malec 1992 in having a short neck, but most of the body is missing in our specimens and so cannot be accurately compared. The difference between L . talenti and L . silnica may only reflect preservational differences.
Thurammina sp. cf. T. arcuata Moreman, 1930 (Plate 3, fig. 8 (1963) extended the concept of the species to include specimens with more apertures and suggested that, with enough specimens, an ontogenetic sequence would show an array of apertural projections.
Thurammina sp. cf. T. subsphaerica Moreman, 1930 (Plate 3, fig. 5  Remarks. Most of the specimens are broken, distorted and compressed, suggesting that the organic lining is very thin in this species. This species has been compared to T. subsphaerica because the papillae are rounded as in T. subsphaerica and there are simple apertures on each papilla, but the presence of small apertures between the papillae, ranging down to 0.5 p, has not been observed before. Thurammina subsphaerica has been recorded from the Silurian of Illinois by Dunn (1942).
Thurammina sp. (Plate 4, figs 2 and 3) Material. One specimen from MW (sample 25.4). Distribution. Martins Well Limestone, Broken River, Queensland, sulcatus Conodont Zone. Description. Test free; an organic foraminiferal lining; 'blocky' surface, possibly due to distortion; few apertures of varying size, not raised from the surface. Dimensions. Diameter of chamber, 69 pm; diameter of apertures, 0.5-1.5pm. Remarks. As this is the only specimen showing this surface, it is unclear as to whether it is not purely a preservational feature, (e.g. the impressions of pyrite framboids).  (Conkin & Conkin, 1970 ). These workers showed that, contrary to the initial description of W . similis (Stewart & Lampe, 1947), there is a small aperture present which could be described as '. . . a single subcentrally located aperture which resembles a pin prick and looks as if a pin had been forced from the exterior into the interior of the test. An apertural protuberance is present on the interior of the shell.' In our case the internal edge of the aperture is fairly recurved; in normal sized agglutinated test foraminiferans from this locality the external appearance of the aperture is a very small arcuate opening. W . similis ranges from the Middle Devonian to Lower Carboniferous in the USA (Conkin & Conkin, 1981, with synonymy), and in Poland occurs in the Lower-Middle Devonian (Malec, 1992 Gutschick, 1962: fig. 4, AMF102674, KE DDHl 448.51, x300; fig. 5, AMF102672, MSh 2, x400; fig. 6, AMF102675, MSh 2, x400; fig. 7, AMF102673, Pt. Hibbs 68669, x400; fig. 8, AMF102676, MW 39.9, (x300). fig. 9, Reophanus proavitus sp. nov., holotype, AMF102677, GCR 105, x200. fig. 10. Tolypammina tantalu Bell, 1996, AMF102678, GCR 53.7, x 120. Figs 11, 12. Hyperammina devoniana Crespin, 1961: fig. 11, AMF102679, MW 39.9, x300; fig. 12 for this species. In the megalospheric forms the ratio of prolocular diameter to length of specimen is one to three or four, whereas in the microspheric form this ratio is one to seven or eight. The microspheric form shows a gradual increase in test diameter from the proloculum, whereas in the megalospheric form the proloculum is slightly constricted from the tubular chamber, which gradually increases in diameter towards the apertural end. Our major difference to H . sappingtonensis is that the microspheric form is very much shorter than the megalospheric form.
H. sappingtonensis is recorded from the Upper Devonian of Louisiana (Conkin & Conkin, 1964) and the Lower Carboniferous (Kinderhookian) of Missouri and Illinois . Conkin & Conkin (1964) have suggested that H . sappingtonensis is almost certainly a junior synonym of H. kahlleitensis Blumenstegal 1969 from the Upper Devonian of Germany, but that doubt exists as to the size range given for the German specimens.
Family Ammodiscidae Reuss, 1962Genus Tolypammina Rhumbler, 1895 Type species. Hyperamminu vagans Brady, 1879 Tolypammina tantula Bell, 1996 (Plate 4, fig. 10 Material. Three specimens from Gel. Rd. (sample 16T/65) Distribution. Tardvale Formation, eastern Victoria, serotinus Conodont Zone. Description. Test free; an organic foraminiferal lining; globular; no apparent large aperture, but the otherwise smooth wall is perforated with many small pores, rounded to angular, placed randomly over the surface. Dimensions. Diameter, 67-73 pm; diameter of apertures, 0.5 pm. Remarks. Although these specimens may belong in Psammosphaera, the many perforations have not been seen in any other member of that genus. A number of the larger perforations seem to have a slightly raised smooth ridge about them. It is, of course, possible that these perforations are the result of chemical reactions during processing.

SUMMARY
The present study of Devonian (Lochkovian to Frasnian) organic foraminiferal linings recognizes 24 species, of which six are new. Organic foraminiferal linings show potential for inter-and intra-continental biostratigraphic correlations as some species have a wide geographical, but stratigraphically limited, range.
This study is the first to integrate taxonomic and biostratigraphic information from foraminiferal linings with that of intact agglutinate foraminiferans from Palaeozoic sequences. This recognition of the significance of the linings opens up a new source of foraminiferal fossils from which to draw information. Thus we now have the potential to not only examine the full size range of foraminiferans and to gain a more complete understanding of the range of variation within these organisms, but also to appreciate the influence of environmental factors on their