Taxonomic descriptions and evolutionary implications of Middle Eocene

Actinella F.W.Lewis, Oxyneis Round and Nupela Vyverman & Compère are three genera of pennate diatoms known exclusively from freshwater environments that are predominantly dilute and acidic (Round et al. 1990, Sabbe et al. 2001, Siver et al. 2007). Of the three, Oxyneis is the smallest genus and comprises only three species (Round et al. 1990, Metzeltin & Lange-Bertalot 2007), although only two have formally been described within the genus. Oxyneis is an araphid taxon, possessing one rimoportula per valve and a large number of girdle bands, each equipped with a septum (Round et al. 1990, Kingston 2003). In this regard, Oxyneis is closely related to Tabellaria Ehrenb. ex Kütz., a common araphid freshwater diatom. Actinella and Nupela both possess a raphe, although the morphology of this feature is very different, yielding placement of the two genera into separate subclasses (Round et al. 1990). Actinella is characterized by reduced raphe slits, one rimoportula per valve, and frustules that are asymmetric along all three axes. Nupela has a fully formed raphe on one (heterovalvar forms) or both (isovalvar forms) valves and symmetric frustules. Although Nupela was established by Vyverman & Compère in 1991, the majority of species now contained within the genus were originally described as Navicula, a genus that has been split into numerous taxa in recent years (see Metzeltin & Lange-Bertalot 1998, 2007 and references therein). Phylogenies based on gene sequences


INTRODUCTION
Actinella F.W.Lewis, Oxyneis Round and Nupela Vyverman & Compère are three genera of pennate diatoms known exclusively from freshwater environments that are predominantly dilute and acidic (Round et al. 1990, Sabbe et al. 2001, Siver et al. 2007).Of the three, Oxyneis is the smallest genus and comprises only three species (Round et al. 1990, Metzeltin & Lange-Bertalot 2007), although only two have formally been described within the genus.Oxyneis is an araphid taxon, possessing one rimoportula per valve and a large number of girdle bands, each equipped with a septum (Round et al. 1990, Kingston 2003).In this regard, Oxyneis is closely related to Tabellaria Ehrenb.ex Kütz., a common araphid freshwater diatom.Actinella and Nupela both possess a raphe, although the morphology of this feature is very different, yielding placement of the two genera into separate subclasses (Round et al. 1990).Actinella is characterized by reduced raphe slits, one rimoportula per valve, and frustules that are asymmetric along all three axes.Nupela has a fully formed raphe on one (heterovalvar forms) or both (isovalvar forms) valves and symmetric frustules.Although Nupela was established by Vyverman & Compère in 1991, the majority of species now contained within the genus were originally described as Navicula, a genus that has been split into numerous taxa in recent years (see Metzeltin &Lange-Bertalot 1998, 2007 andreferences therein).Phylogenies based on gene sequences I denotes the raphe slit on the head pole.K & L are different focal planes of the same specimen, and J represents a close-up of this same specimen taken with reflected interference contrast optics.The extensive apical pore fields appear dark on this image.M & N represent the same specimen, and P is of the type specimen for O. apporrecta deposited at the ANSP.V-W and AA-AB denote both valves of the type specimen for N. mutabilis deposited at ANSP taken at different magnifications.Note the slight difference in distances between the proximal raphe fissures on both valves.Scale bars: I = 2 µm; J = 5 µm; AB (also for AA) = 5 µm; separate bar = 10 µm for all other images.Except for J, all others taken with differential interference contrast optics.
The oldest known freshwater diatoms date to the late Cretaceous and include both radial centric (Ambwani et al. 2003) and pennate forms (Chacón-Baca et al. 2002).Although the identity of some of the fossil forms reported by Chacón-Baca et al. (2002) from 70 My old sediments in Mexico, especially the ones possessing a raphe, has been questioned (Sims et al. 2006), most fossil evidence suggests that the majority of freshwater lineages evolved in the Eocene or later with rapid radiation occurring in the Miocene (Wornardt 1972, Bradbury et al. 1985, Krebs 1994, Sims et al. 2006, Siver & Wolfe 2007).In general, the lack of well preserved and accurately dated fossil remains has resulted in an overall poor understanding of the biostratigraphy and evolution of freshwater diatoms (Barron 1993, Sims et al. 2006), and a lack of geologic mileposts for use in phylogenetic investigations (Sims et al. 2006, Wolfe & Siver 2009) The corrected vertical depth in the core and estimated depth within the lake are also shown.
situated near the Arctic Circle in northern Canada harbour numerous and exquisitely preserved diatom and chrysophyte microfossils (Wolfe et al. 2006, Siver & Wolfe 2005, 2007).
The high quality of fossil preservation, coupled with a wellconstrained Middle Eocene age, make this an exceptional deposit for understanding the early history of freshwater ochrophyte algae.
In this communication, we describe three freshwater diatoms representing the genera Actinella, Oxyneis and Nupela from Giraffe pipe.Other than initial observations of the former genus (Siver & Wolfe 2009), none of these diatoms have previously been recorded from the Paleogene fossil record.Although none of these fossil species have been observed in extant floras, the morphological structures they possess share pronounced affinities with modern congeneric taxa.

MATERIALS AND METHODS
As described in earlier communications, the Giraffe Pipe locality (64º44'N 109º45'W) is a kimberlite diatreme that was infilled by a sequence of lacustrine and paludal sediments, and subsequently covered by Neogene glacial deposits (Siver & Wolfe 2005, Wolfe et al. 2006, Wolfe & Siver 2009).BHP Billiton retreived a 165 m long drilled core (BHP 99-01) from the Giraffe Pipe diatreme in 1999 to assess the kimberlite deposit for potential diamond exploration.The core was taken at 47º and included 68.3 m of lacustrine mudstone sediments overlaid with 44.8 m of peat and terrestrial remains and capped with glacial till.Converting the core to vertical equivalents results in 51.1 m and 32.7 m of lacustrine and peat sediments, respectively.The estimated depth in the Giraffe Pipe waterbody that each sample represents was measured from the point where the lacustrine sediments transition to true peat layers.Based on a 87 Rb/ 87 Sr model from kimberlite phlogopite, emplacement of the kimberlite occurred at 47.8 ± 1.4 My, representing a maximum age for the lacustrine sediments (Creaser et al. 2004).Fission track estimates using tephra beds lying near the termination of the lacustrine phase establish a minimum age for the underlying lacustrine sediments of 40 My (Wolfe & Siver 2009).The portions of the core containing lacustrine sediments are stored in boxes labeled 11-27.Each box has three channels (1-3) and each channel has 1.5 m of core (4.5 m total per box).Samples are recorded as the box # -channel # -cm from the top of the box.Thus, the sample noted as 11-3-90 represents a sample taken from 90 cm along the core length in channel three from box 11.
A total of 300 specimens, 150 of Nupela, 100 of Actinella and fifty of Oxyneis, in 25 samples from the Giraffe Pipe core were examined as part of this investigation (table 1).The samples correspond to corrected vertical depths within the core ranging from 70.09 to 110.67 m, and lakewater depths ranging from 0.51 to 41.09 m.Samples consisted of either untreated chips of mudstone or aliquots from digested slurries.
Approximately 0.5-1.0g of organic mudstone from each lacustrine layer was oxidized with 30% H 2 O 2 under low heat for several hours, centrifuged, rinsed several times with distilled water and stored as a final volume of 10 ml.Aliquots of these slurries were air dried onto heavy duty aluminum foil, attached onto aluminum SEM stubs with Apiezon ® wax and coated with a gold and palladium mixture for two minutes with a Polaron model E5100 sputter coater (Siver 1987).Small chips of mudstone from a few of the lacustrine facies were also mounted directly onto aluminum SEM stubs using carbon tape, coated with the same gold and palladium mixture and grounded with silver paint.All samples were observed with a Leo (Zeiss) 982 SEM.  the exterior surface, the distal and proximal ends of the raphe are slightly rounded, but otherwise unmodified (fig.3A-B).On the interior surface, each raphe slit terminates as a wellformed helictoglossa (fig.3C-F).There is a single rimoportula per valve, usually orientated parallel to the apical axis and positioned at the apex (fig.3A, C-D & F).There are up to four, and most often three, girdle bands per frustule, each opened at the apex and with one row of pores (fig.2F).The specific epithet refers to the discovery site.

Actinella giraffensis
Location in core: Actinella giraffensis has been found in eight sections of the core, all from core boxes 14-16 (table 1).The samples containing this species range in corrected vertical depth from only 81.63-86.63m, representing estimated lake depths of between 12.05-17.05m. -Type: holo-: here designated as the circled specimen on microscope slide marked "Giraffe Pipe, 15-2-40, LM2", deposited at the Academy of Natural Sciences of Philadelphia, PA, USA, as ANSP GC 64856, and illustrated in figure 1D-E  × 1.8-2.1 µm (center) × 4.5-5.5 µm (inflation).The narrow portion of the middle of the valve is approximately the same width as the rostrate apices, and about one-third that of each inflated portion (fig.4C).The mantle is of the same width, except around the apices and sometimes at the inflations where it often widens slightly (fig.4F).Striae are uniseriate, parallel to slightly radiate, linear to slightly curved, often unevenly and alternately spaced and consist of very small and round areolae (fig.4A-D).Striae continue over the rounded valve margins and terminate half way down the mantle (fig.4F).On our specimens the areolae lacked coverings.The axial area is thin and narrow within the constricted center of the valve and along the rostrate apices, but is broadly lanceolate within the expanded portions of the valve (fig.4C-D).Extensive and well-developed apical pore fields are found on both apices, each consisting of numerous rows of small pores (figs 1J, 4B & D).A single rimoportula is found within a stria near the margin close to the point where the valve is extended to form the rostrate apex (fig.4B-D).The rimoportula is oriented perpendicular to the apical axis.Copulae are open, have a welldeveloped septum that extends approximately the length of the rostrate apex, and a single row of pores along the margin (fig.4E-F).The closed end of each copula is expanded and contains three to four short rows of pores.Frustules with up to 10 copulae were observed.Valve ridges were not observed.The specific epithet refers to the extended rostrate apices.
Location in core: Oxyneis apporrecta has been found in four sections of the core, all from core boxes 14-16 (table 1).The samples containing this species range in corrected vertical depth from only 80.71-87.29 m, representing estimated lake depths of between 11.13-17.71m. -Type: holo-: here designated as the circled specimen on microscope slide marked "Giraffe Pipe, 15-1-131, LM2", deposited at the Academy of Natural Sciences of Philadelphia, PA, USA, as ANSP GC 64857, and illustrated in figure 1P On some valves, the axial and central areas can be very broad, and the distinction between the two less apparent (fig.5D).A fully developed raphe is found on each valve of the frustule, however, the distances between the proximal fissures are usually not equal making the frustule slightly heteromorphic (fig.1AA-AB).The raphe is linear to slightly curved with distal raphe fissures that deflect strongly in the same direction down onto the mantle, where they become recurvate (figs 5A & E,  6A-B).The proximal raphe fissures are straight and terminate as small pores on the external surface (figs 5A, C & E-F, 6C).Internally, the distal raphe ends terminate as small, well formed, helictoglossa (fig.6F), and the proximal raphe ends form small recurved "T-shaped" structures that are deflected moreso to the secondary side of the valve (fig.6D).Externally, the areolae are circular to elliptical and covered with hymenes that are flush with the external valve surface (fig.6A-C & E).Hymenes between adjoining areolae are often fused together (fig.6A-B & E).The diameter of the areola decreases significantly between the external and internal valve surfaces, and areolae open internally as small circular to elliptical pores that lack coverings (fig.6D & F).The striae continue onto the mantle, usually as a single row of elongated areolae.The specific epithet refers to the variable nature of the valve outline.
Location in core: Nupela mutabilis has been found in thirteen sections of the core, spanning a wide range of depths (table 1).Samples containing this species range in corrected vertical depth from only 70.09-110.67m, representing a wide range of estimated lake depths from 0.51-41.09m. -Type: holo-: here designated as the circled specimen on microscope slide marked "Giraffe Pipe, 11-3-90, LM2", deposited at the Academy of Natural Sciences of Philadelphia, PA, USA, as ANSP GC 64858, and illustrated in fig.1V-W

DISCUSSION
Oxyneis is a small genus containing only two named taxa, O. binalis (Ehrenb.)Round in Round et al. 1990, and Oxyneis binalis var.elliptica (Flower) Kingston (Kingston 2000).A third taxon recently described from Florida peat has been placed within Tabellaria as T. binalis var.brevistriata Metzel-tin & Lange-Bert.This diatom is nearly identical to O. binalis, with striae restricted to the margins (Metzeltin & Lange-Bertalot 2007).Oxyneis apporrecta represents a fourth taxon that has not been observed in modern collections and must be considered extinct.In this regard, it is noteworthy that a small elliptical taxon virtually identical to O. binalis var.elliptica is also present in Giraffe Pipe sediments (fig.7), however, we have not observed specimens belonging to O. binalis.Reduction of the rostrate apices on O. apporrecta specimens would essentially yield a taxon similar to O. binalis.
As originally described by Round (in Round et al. 1990), Oxyneis differs from the closely related genus Tabellaria in spine complexity, position of the rimoportula, details of the girdle bands and valve shape.Some, but not all, species of Tabellaria form a row of short spines along the valve margin.In many diatom genera spines are used to join frustules into filaments.This is not the case in Tabellaria where commonly observed zig-zag, linear and stellate arrangements of cells are accomplished by mucilage secretions through the apical pore fields.Round (in Round et al. 1990) noted that the marginal spines on valves of Tabellaria binalis (Ehrenb.)Grunow (in Van Heurck 1881) were significantly more complex than those on other Tabellaria species, often fused together, and used to connect frustules valve to valve to form straight chains.The more complex spine development was a primary character used by Round to establish Oxyneis.
Interestingly, Round (in Round et al. 1990) stated that the elliptical valve morphotype of O. binalis (= Oxyneis binalis var.elliptica) formed zig-zag colonies, presumably through mucilage secretions of the pore fields and not using marginal ridges.In this regard, O. binalis var.elliptica was perhaps more similar to Tabellaria.Siver & Hamilton (in press) present specimens of both O. binalis and Oxyneis binalis var.elliptica with massive marginal ridges, structures readily distinguished from individual spines and not reported for Tabellaria.However, Siver & Hamilton (in press) also note that in both of these extant Oxyneis taxa ridges were primarily restricted to terminal valves within a filament.In fact, most isolated valves lacked ridges and were presumably held together, valve-to-valve, by mucilage secretions from the extensive pore fields.Perhaps ridge formation in Oxyneis serves as a means of separating valves from one another, much like separating spines in Aulacoseira, and are not used for linking cells together.Thus, spine and ridge formation is not a universal feature found on all frustules of modern Oxyneis taxa.In fact, Metzeltin & Lange-Bertalot (2007) do not recognize Oxyneis, in part, because they believe differences in spine and ridge structure should be used to differentiate at the species, not genus, level.Further, Kingston (2003) does not even mention spine or ridge structure in his description of Oxyneis.We have yet to observe marginal spines or ridges on any Oxyneis apporrecta valves, and we must conclude at this time that these structures are not formed by this organism.We further presume that valves of Oxyneis apporrecta were held together through mucilage secretions from the extensive apical pore fields that extend over the mantle and well onto the valve face.
The position of the rimoportula, being close to the ends of the cell as opposed to near the central region, was another character used by Round (Round et al. 1990) to differentiate Oxyneis from Tabellaria.Our review of published records for Oxyneis clearly indicates a more apical position for the rimoportula, and Oxyneis apporrecta is no exception.On valves of Tabellaria, rimoportulae are almost always positioned near the center of the valve or close to the central inflation (Round et al. 1990, Krammer & Lange-Bertalot 1991).A notable exception is Tabellaria ventricosa Kütz., a species possessing short valves with a wide central inflation and a single rimoportula near one apex (Krammer & Lange-Bertalot 1991, Metzeltin & Lange-Bertalot 2007).This fact also caused Metzeltin & Lange-Bertalot (2007) to question the validity of the genus Oxyneis.
Features of the girdle bands and overall valve shape also differ between species of Oxyneis and Tabellaria.Frustules of Oxyneis species have open girdle bands that possess a septum on the end opposite of the opening, and a row of pores that usually lines the entire structure.In contrast, Tabellaria species have two distinct types of girdle bands, including complete bands with one or two septa and open bands that often lack septa altogether.The septa on Oxyneis girdle bands are short and less developed compared to those of Tabellaria where the septum often extends to the central inflation (Patrick & Reimer 1966, Round et al. 1990, Kingston 2003).Lastly, Tabellaria girdle bands lack the characteristic row of areolae found on Oxyneis.
All species of Tabellaria have valves with terminal and central inflations, whereas species of Oxyneis lack central inflations and, in fact, can be highly constricted in the central region.Inflated regions of the valve on Oxyneis specimens are either in an apical position (e.g.O. binalis) or along the valve (e.g.Oxyneis apporrecta), but not in the center of the valve.The degree of development of the septa on Oxyneis and Tabellaria may simply be related to the position of inflations along the valve.In conclusion, despite the controversy over whether Oxyneis should remain a separate genus (Round et al. 1990) or be merged with Tabellaria (Metzeltin & Lange-Bertalot 2007), it is clear that both taxa are very closely related.Since most recent authors have maintained separation of these two genera, we are electing to describe the taxon from Giraffe Pipe sediments as Oxyneis because of valve shape, position of the rimoportula and similarities of the girdle bands with modern taxa.
When the genus Nupela was originally described, Vyverman & Compère (1991) noted it to be isovalvar with both valves possessing a fully formed raphe and, more importantly, to have uniquely structured areolae.Since the original description, over 38 species have been described or transferred into Nupela (Monnier et al. 2003, Potapova et al. 2003, Siver et al. 2007), many of which are heterovalvar with respect to the degree of development of the raphe (Lange-Bertalot 1993, Metzeltin & Lange-Bertalot 1998, Potapova et al. 2003).In this regard, the slight heterovalvar condition found on N. mutabilis frustules due to differences in the lengths of the raphes fits the situation found in many species within the genus.The unique structure of the areolae on all Nupela species, including those on N. mutabilis, is a very stable and reliable character of the genus (Potapova et al. 2003, Siver et al. 2007).In addition to the structure of the areolae, N. mutabilis possesses all of the characters identifed by Siver et al. (2007) as shared among Nupela species, including external hymenes covering one or more areolae, distal raphe fissures deflected to the same side of the valve, straight external proximal raphe fissures ending as small pores, a single row of elongated areolae on the mantle and small helictoglossae.All samples containing N. mutabilis consistently had specimens with valve shapes ranging from linear to linear-lanceolate and possessing a relatively wide range of L:W values.Such variability in valve shape is not uncommon for Nupela taxa (Siver et al. 2007).
Specimens of Actinella giraffensis can be reliably assigned to this genus because of the heteropolar nature of valves, the clavate shape of frustules in girdle view and the position and nature of the raphe slits.The location of the raphe slits on the valve face margin, as opposed to on the mantle, also separates it from the closely related genus Eunotia.Within the genus, A. giraffensis does not match any extant species, but certainly belongs with species possessing simple, as opposed to complex, head poles.As noted in a previous communication (Siver & Wolfe 2009), A. giraffensis is most closely aligned with Actinella parva Vanhoutte & Sabbe, a species only known from the highlands of Tasmania (Sabbe et al. 2001).It is not yet known if Actinella species possessing simpler head poles and lacking marginal spines are basal or derived taxa within the genus, but it is noteworthy that the few species from the fossil record have less complex valve shapes compared to many tropical species known today from South America (Kociolek et al. 2001, Metzeltin & Lange-Bertalot 2007) and Africa (Kociolek & Rhode 1998, Sabbe et al. 2001), as well as most Australian endemics (Sabbe et al. 2001).
As already discussed for Eunotia (Siver & Wolfe 2007) and the thalassiosiroid genera Cyclotella, Discotella and Puncticulata (Wolfe & Siver 2009), our findings significantly extend the fossil record for Actinella.As noted by Siver & Wolfe (2009), in addition to the presence of Actinella in Giraffe Pipe sediments, three other fossil records exist for this genus.Actinella penzhica Lupikina & Dolmatova was described from Oligocene to late Eocene deposits on the Kamchatka Peninsula (Lupikina & Dolmatova 1982), A. miocenica J.Y.Li from Miocene sediments in northeastern China (Li 1988), and A. pliocenica from Pliocene remains in France (Héribaud 1902).Interestingly, none of the fossil remains of Actinella were found in presently tropical areas and none from the Southern Hemisphere, regions of the world where most Actinella species are found today.The presence of Actinella in Giraffe Pipe with a minimum age of 40 My significantly extends the stratigraphic range of this freshwater genus.
To our knowledge, Giraffe Pipe sediments also represent the first fossil records for the araphid genus Oxyneis and the raphe-bearing genus Nupela.Although fossils of Oxyneis are previously lacking from the fossil record, specimens of the closely related genus Tabellaria are known from late Eocene-early Oligocene (Lupikina & Dolmatova 1975) and Miocene (Wornardt 1972, Bradbury et al. 1985) sediments.In addition, Tetracyclus Ralfs (Ralfs 1843), another genus within the family Tabellariaceae Kütz.(Kützing 1844), also dates to at least the late Eocene-early Oligocene (Lupikina & Dolmatova 1975).Thus, our discovery of Oxyneis in the Middle Eocene is consistent with these previous finds, and Pl. Ecol. Evol. 143 (3), 2010 collectively these records support an Eocene or earlier evolution for the Tabellariaceae.
As reviewed by Sims et al. (2006), the diatom raphe had evolved by the Paleocene and it is likely that this structure is even older, perhaps originating in the Cretaceous.Lohman & Andrews (1968) reported 34 freshwater diatom taxa, including twenty raphid forms, from Late Eocene sediments in Wyoming, U.S.A., supporting the view that most modern raphid lineages had evolved by this time (Sims et al. 2006).Thus, the presence of Nupela mutabilis in the Giraffe Pipe core is consistent with the known fossil record in regards to evolution of the raphe.On the basis of SSU rRNA sequences and preliminary molecular clock calculations, Sorhannus (2007) predicted that early representatives of modern araphid diatoms were present by the Early Cretaceous (~ 100 My), whereas raphid taxa had evolved by the Lake Cretaceous (~ 75 My).In this regard, it is of interest that details of the raphe found on Nupela mutabilis, as well as other Eocene pennate diatoms (Lohman & Andrews 1968), are strikingly similar to those found on modern congeneric forms, suggesting that the fossil record of raphe origins converges with molecular estimates (Sims et al. 2006, Sorhannus 2007).In fact, Bradbury et al. (1985) noted that lacustrine diatoms from the Miocene, Oligocene and late Eocene had structures such as areolae, rimoportulae and raphes that made them "morphologically modern", a theme also echoed by Lohman & Andrews (1968).Our findings from Giraffe Pipe sediments support these previous works and extend them to include additional structures such as the fultoportulae (Wolfe & Siver 2009).
The genera Oxyneis and Actinella are largely restricted to very acidic habitats (Round et al. 1990), and although Nupela is more widespread, this genus is also most often associated with acidic waterbodies (Siver et al. 2007 and references therein).Flower (1989) reported both Oxyneis binalis and O. binalis var.elliptica (under Tabellaria) as being confined to lakes usually with pH < 5, and Camburn & Charles (2000) reported an average weighted mean pH for O. binalis (including specimens of O. binalis var.elliptica) of 5.1 for lakes in the Adirondack Mountains of New York.In a forthcoming publication, Siver & Hamilton (in press) observed both Oxyneis binalis and O. binalis var.elliptica as common elements of the diatom floras of very acidic and often humic-stained waterbodies along the Atlantic Coastal Plain, U.S.A. Kingston (2003) also noted Oxyneis as typical of very acidic lakes, including peat bog ponds, and Patrick & Reimer (1966) and Siver & Hamilton (in press) found this genus typical of waterbodies low in nutrients and dissolved salts.
Like Oxyneis, Actinella taxa are primarily recorded from very acidic environments, often high in dissolved humic content (Round et al. 1990, Kociolek & Rhode 1998, Sabbe et al. 2001, Kociolek et al. 2001, Metzeltin & Lange-Bertalot 2007).The acidic nature of Actinella holds true for tropical taxa from South America (Kociolek et al. 2001, Metzeltin & Lange-Bertalot 2007) and Africa (Kociolek & Rhode 1998), Australasian endemics (Sabbe et al. 2001) and the widely distributed and common species A. punctata F.W.Lewis.To date, both A. giraffensis and O. apporrecta have only been uncovered in samples from core boxes 14, 15 and 16.Coupled with the fact that many of these samples also contain sig-nificant numbers of Eunotia specimens (Siver & Wolfe 2007) and several chrysophyte species aligned with modern acidophilic taxa (P.A. Siver, Connecticut College, U.S.A., unpubl.res.), we conclude that the waterbody that existed between approximately 11 and 18 m was very acidic.
Interestingly, although Nupela mutabilis is found widely throughout the core, ranging from boxes 23 to 11, it is largely absent from the section of the core containing A. giraffensis and O. apporrecta, suggesting that the former taxon is less abundant in very acidic waters.Although much more work needs to be done before a complete reconstruction of the paleoenvironment represented by the Giraffe Pipe waterbody can be appreciated, our preliminary observations support the concept that the initial lake was more alkaline and acidified over time.

CONCLUSION
In conclusion, the diatoms reported here provide valuable mileposts that refine the existing fossil record and provide new temporal constraints for molecular phylogenetic investigations.By the Middle Eocene (≥ 40 My), the orders Tabellariales, Eunotiales and Naviculales were already established in North American freshwater habitats, unambiguously displaying the range of morphological characters that define each of these orders.The search for primitive forms that inform more directly the evolution of the raphe system in pennate diatoms and the initial colonization of nonmarine habitats must therefore address older sedimentary sequences.

Figure 1 -
Figure 1 -Light micrographs of Actinella giraffensis (A-I), Oxyneis apporrecta (J-R) and Nupela mutabilis (S-AB).D & E represent slightly different focal planes of the type specimen for A. giraffensis deposited at ANSP, and H is a girdle view of this species.The arrow on fig.I denotes the raphe slit on the head pole.K & L are different focal planes of the same specimen, and J represents a close-up of this same specimen taken with reflected interference contrast optics.The extensive apical pore fields appear dark on this image.M & N represent the same specimen, and P is of the type specimen for O. apporrecta deposited at the ANSP.V-W and AA-AB denote both valves of the type specimen for N. mutabilis deposited at ANSP taken at different magnifications.Note the slight difference in distances between the proximal raphe fissures on both valves.Scale bars: I = 2 µm; J = 5 µm; AB (also for AA) = 5 µm; separate bar = 10 µm for all other images.Except for J, all others taken with differential interference contrast optics.

Figure 2 -
Figure 2 -Scanning electron micrographs of Actinella giraffensis depicting external (A, C), internal (B, D) and girdle (E-F) views.Note the short raphe slits of variable lengths situated on the valve margin and terminating as small helictoglossa (B, D).Frustules are clavate in girdle view (E) and possess open girdle bands (arrow, F).Scale bars are 2 µm (F), 5 µm (A-D) and 10 µm (E).
Siver, Wolfe & Edlund, sp.nov.Description: Specimens of Actinella giraffensis have a slight dorsi-ventral design, acutely rounded and unmodified foot and head poles, and are clavate in girdle view (figs 1A-I & 2A-F).Valves are small, narrow, club-shaped and range in size from 18-40 µm × 1.8-3 µm.The striae are parallel to the transapical axis except at the apices where they become radially aligned around the poles (fig.3A-F).The striae become more closely spaced at the apices, continue onto the mantle and range in density from 22-27 in 10 µm.The areolae are small, circular, evenly spaced pores that lack coverings on both the exterior (fig.3A-B) and interior (fig.3C-F) surfaces.The mantle is shallow, ranging in depth from 0.8-1.3µm at the head pole to 0.5-0.8µm at the foot pole, yielding the slender clavate shape (figs 1H & 2E-F).The raphe slits are straight, restricted to the ends of the valve, of unequal lengths and range in length from 1.5-5 µm (figs 1I & 3A-F).Each raphe slit lies on the valve face along the ventral margin.On

Figure 3 -
Figure 3 -Scanning electron micrographs of Actinella giraffensis revealing the external (A, B) and internal (C-F) morphology of head (A, C, E) and foot (B, D, F) poles.Note the short raphe slits of variable lengths positioned on the valve margin (A-F) and terminating as small helictoglossa (C-F).The rimportulae are parallel with the apical valve axis and can be found on either the head (A, C) or foot (D, F) pole.Scale bars are 2 µm.

Figure 4 -
Figure 4 -Scanning electron micrographs of Oxyneis apporrecta showing external (A, B, F) and internal (C-E) surfaces of valves (A-D), girdle (F) and girdle bands (E).Note the biundulate shape of the valve (A, C), rostrate apices (A-E), marginal and more apical position of the rimoportulae (B-D), and the extensive apical pore fields (A-D, F).Girdle bands are open and possess a septum that extends to the valve inflation (E).Scale bars are 2 µm (B, D, F) and 5 µm (A, C, E).

Figure 5 -
Figure 5 -Scanning electron micrographs of external (A, C, E-F) and internal (B, D) views of Nupela mutabilis Note the linear to linearlanceolate outline of valves, rostrate apices, wide range in the L:W ratio, deflected distal raphe fissures (A, C, E), lanceolate axial area and circular to elliptical central area (A-C, E, F).Scale bars are 2 µm (C), 5 µm (A, B, D, E) and 10 µm (F).

Figure 6 -
Figure 6 -Additional scanning electron micrographs of external (A-C, E) and internal (D, F) views of Nupela mutabilis.Note the highly deflected distal raphe fissures (A-B), the small and straight proximal raphe fissures (C), T-shaped internal fissures (D) and helictoglossa (F).Hymen coverings are eroded from some valves (A, C, E) and present on others (B).The often coalesed nature of the hymenes is shown in A and E. Scale bars are 500 nm (E), 1 µm (A) and 2 µm (B-D, F).

Figure 7 -
Figure 7 -Scanning electron micrographs of a frustule (A) and girdle bands (B, C) of a diatom from Giraffe Pipe sediments that is largely identical to the modern taxon Oxyneis binalis var.elliptica (Flower) Kingston (Kingston 2000).Scale bar is 2 µm.