Stephanodiscus Ehr. species from Holocene sediments in the Faiyum Depression (Middle Egypt)

Lake Qarun in the Egyptian Faiyum Depression is the shrunken remnant of a much larger lake. To investigate the sedimentary history of the former lake, several continuous lacustrine sediment cores were collected from now terrestrial locations within the basin. The suitability of these sediments for reconstructing past environmental changes was first assessed by describing the gross stratigraphies of three core sequences and secondly by evaluating taxonomic issues concerning sedimentary diatom species within the genus, Stephanodiscus Ehr. The stratigraphic descriptions show basal sections comprised of either coarse sand or weathered Ecocene limestone overlain by lacustrine sediments. Lower sections of the lacustrine sediments are diatom marls, and, in the two lakeside cores, these are laminated. These sections indicate that the lake initially formed by rapid in filling to a considerable depth with freshwater and the diatom-rich marls containing abundant Stephanodiscus and Aulacoseira Thwaites species. A new species and variety of Stephanodiscus, S. neoaegypticus and S. neoaegypticus var. fekrii, are described from the marl section in one core. Another species, S. galileenis Håkansson & Erhlich, was also present and is re-evaluated with reference to the published description. The importance of establishing a sound taxonomic foundation for palaeolimnological studies of Lake Qarun is stressed.


Introduction
Egypt possesses few natural inland lakes but modern day Lake Qarun in the Faiyum Depression is the remnant of a much larger early Holocene lake (Caton-Thompson & Gardner 1934, Hassan 1986). Consequently, exploring sedimentary diatom records in Lake Qarun sediments can potentially help understand the environmental history of the lake. However, issues of sediment formation and dating, diatom preservation and diatom taxonomy all require careful evaluation before palaeolimnological information can be fully exploited (e.g. Flower et al. 2006). Previous work has decribed terrestrial diatomite exposures northeast of modern Lake Qarun (Aleem 1958) and later Przbylowska-Lang (1976) showed benthic diatom species of Synedra Ehrenberg (1830: 60) and Epithemia Kutzing (1844: 33) to be more abundant in upper diatomites from the region. Furthermore, Mueller-Wilmes (1988) found several unknown Stephanodiscus Ehrenberg (1845: 72) species in Lake Qarun sediment as well as Stephanodiscus aegyptiacus Ehrenberg (1854: pl. 33/1, fig. 16). Stephanodiscus species are of particular interest because they no longer grow in the lake today (Fathi & Flower 2003, Mansour 2008) and, being planktonic, are indicative of open lake conditions.
As part of a larger project to investigate the environment history of Lake Qarun (e.g. Keatings et al. 2010), several sediment augur cores where collected from the Lake Qarun basin. This paper firstly summarizes the gross stratigraphy of these cores and, secondly, evaluates some Stephanodiscus species present with in these cores.

Site Description and Field Methods
Modern Lake Qarun (29.487 o N,30.298 o E) lies within the Faiyum Depression and the current surface lake water fluctuates around -43 m below sea level. It is sustained by drainage of Nile water delivered into the Depression through the Bahr Yusef ( Fig.1) Field sampling: In April-May 2008 augur coring (see Flower et al. 2012) was undertaken at six locations and three of these yielded lacustrine sediment sequences. The three cores containing diatom remains were QARU9 and QARU10, collected at the southern margin of Lake Qarun, and GARK1, collected from within the Gharak region in the south-western part of the Faiyum (Fig. 1). FIGURE 1. Lake Qarun situated in northern part of the Faiyum Depression, Egypt. Three terrestrial sediment core locations are shown with QARU9 and 10 on the southern shore of the lake and GARK1 in the Gharak region of south west Faiyum. Note diatomite exposures are indicated to the west and northeast of the modern lake. Summary descriptions: QARU9 (29.43447°N, 30.64213°E) was the longest core at 21.4 m and the upper sections consisted of a sandy layer and silty clays coloured dark brown to grey depending on depth (Fig. 2, left). Only below 1085 cm depth did the bulk sediment become coarsely laminated and below 1730 cm the white laminations became much thinner (at c. 1 mm thick) as the matrix changed to soft olive grey sediment. Below 2030 cm the sediment became silty and at c. 2080 cm coarse grey sand with small pebbles was present. QARU10 (29.40309°N, 30.50830°E) was 10.4 m long (Fig. 2, middle) had a lithology similar to QARU9 but FIGURE 2. Gross sediment stratigraphies of the three cores, QARU9, QARU10 and GARK1 (see text for sediment descriptions). Note the basal sands below finely laminated diatom marl sediment in QARU9 & 10. with shorter sections. At around 960 cm fine laminations appeared but this section was shorter than in QARU9 being only about 60 cm in depth and by 1030 cm brown coarse sands and gravel occurred. GARK1 (29.14840°N, 30.60976°E) was 5.80 m long (Fig. 2, right) and the upper sections comprised mainly silty clay but at 420 cm the sediment changed to a silty clayey grey marl which persisted to 570 cm with some feint irregular laminations. The sequence ended in light grey weathered limestone at 580 cm.
Sample preparations: Sub-samples for preliminary diatom analysis were taken at approximately 40 cm intervals in each core with sampling frequency varying according to changes in stratigraphy. Most samples were placed into Whirl Pac bags for transportation from the sites. The bulk cores were all archived in Egypt for further lithostratigraphic study while the sedimentary diatom sub-samples were transported to the UK for diatom analysis.

Material and Methods
Initial low power microscopical examination of the down core diatom strews indicated that species of Cyclostephanos Round ex Theriot, Håkansson, Kociolek, Round & Stoermer (1987: 346), Stephanodiscus and Aulacoseira Thwaites (1848: 167) and were common in many samples. In QARU9 at least some whole diatom valves were present in all samples between 160 cm and 2030 cm depth, in QARU10 between 260 and 1005 cm depth and in GARK1 between 60 cm and 580 cm. Consequently, all samples were prepared for diatom analysis. After initial freeze drying, diatom analysis was undertaken after sample treatment with hydrogen peroxide (organic matter was <10% of the dry sediment weight in all samples and little oxidation was needed). Hot 50% HCl, was used to remove carbonate minerals. After washing with distilled water, the resulting diatom suspensions were dried on to glass cover slips and mounted in Naphrax. Slide strews were examined at x1000 magnification in a Leitz Labrolux light microscope (LM) with a 1.3 N.A. objective so that diatom species could be assessed and taxonomic issues confronted. Because of the high degree of valve silicification, diatoms were usually viewed without phase contrast optics. Identification difficulties were soon encountered for species within the Stephanodiscus genus and appropriate treated samples from all three cores were used for taxonomic evaluation and photomicrography. LM microphotography used a Leica DM6000 LM with a x100 objective with a Zeiss AxioCam digital camera. Scanning electron microphotography was undertaken in a Jeol JSM-6480LV scanning electron microscope (SEM).

Results
Sediment core GARK1: Although not thinly laminated, the deeper (4.2-5.6 m) sediment in this core contained abundant planktonic diatoms which, in addition to Cyclostephanos dubius (Fricke) Round in Theriot et al. (1987: 346), included a Stephanodiscus species with similarities to S. aegyptiacus. This latter species was reviewed by Håkansson & Locker (1981) and it is noteworthy that the type locality for S. aegyptiacus is the 'See Garag, Fajum, Aegypten'. The unrecognized Stephanodiscus species in GARK1 marl sediment was initially thought to be co-specific with S. aegyptiacus but closer examination in both LM and SEM revealed significance differences.
In the LM valves were large (up to 70 µm in diameter) and weakly concentrically undulate so that frustules possessed only slightly differentiated convex and concave valves. No distinct sub-marginal hyaline silica ring was discernible at the valve face mantle junction (cf. S. aegyptiacus) although an indistinct ring (Fig. 3) is present in larger specimens (> c. 40 µm diameter). Interfascicle ribs (3-4 in 10 µm) were relatively broad and tended to narrow both distally and proximally giving a fine lanceolate appearance (Figs 3, 4). Spines were absent although valves showing marginal dissolution can appear to have spines (Fig. 6). The fascicles are composed of areolae which in LM often appear large and blocky (Fig. 4), especially in well silicified specimens (Fig. 5). At the valve face-mantle junction (VFMJ), the interfascicles occasionally bifurcate (Figs 4 & 5). The fascicles are uniseriate for half to two thirds the length of the fascicle (dependant on valve size) and broaden at the VFMJ to include 3-4 areolae (Figs 4,(6)(7). In smaller (< c. 40 µm diameter) valves (Figs 7-8) the interfascicles appear relatively more lanceolate, depending on the plane of focus  and, compared to larger valves, they more frequently bifurcate at the VFMJ (Fig. 8). In SEM the external valve view confirms only a small degree of concentric undulation with valves being slightly convex ( Fig. 9) or concave (Fig. 10). The latter clearly shows the fine lanceolate nature of the flat interfascicles which narrow distally to accommodate broadening of the fascicles as they become multiseriate. External openings of the valve face areolae (c. 14 in 10 µm) are small (< 1 µm in diameter, unless expanded by dissolution) and most interfascicles have external fultoportulae openings at the VFMJ (Fig. 10). External apertures of rimoportulae ( Fig. 10, arrowed) are sub-marginal on interfascicles. Spines are absent. Internally, rimoportulae (3-6) occur irregularly spaced in the sub-marginal zone (Fig. 11, arrowed), are almost sessile and oriented sub-obliquely (Fig. 12). The areolae become markedly smaller beyond the VFMJ (Fig. 12) and more closely spaced (>30 in 10 µm) compared to valve face areolae (c. 14 in 10 µm). Marginal fultoportulae have three satellite pores and are subtended by short tubes (Fig. 12). The central area fultoportulae form a ring and each possesses two satellite pores (Fig. 13, arrowed). Smaller valves (< c. 20-40 µm in diameter) are also either concave (Fig. 14) or convex (Fig. 15) and valve face areolae appear more coarsely spaced with very small apertures (but metrics are not significantly different from those of larger forms). At least one raised rimoportula aperture occurs submarginally (Fig 14, arrowed). Internally, each valve has one, two or three marginal rimoportulae and a valve with two is indicated in Fig. 16. Marginal fultoportulae have three (sometimes four) satellite pores (Fig. 17). Central area fultoportulae are recessed with two satellite pores (Fig.  18, where well preserved domed cribra are also shown).   Stephanodiscus neoaegypticus var. fekrii Flower var. nov.
Sediment cores QARU9 and QARU10: In the thinly laminated sections of both QARU9 & 10 (see Fig. 2), a third common Stephanodiscus species was present that was not S. neoeagyptiacus but much more similar to S. galileensis Håkansson & Ehrlich (1987: 20). The type locality for this species is sub-surface (fossil) sediment in Lake Kinneret (Israel). They reported a heterovalvic condition where the concave and convex valves differed with regard to fascicle arrangement and to location of fultoportulae. Since some of their material was evidently rather poorly preserved (see their Figs 3, 14 and 16), the species is re-examined here.
In LM, Stephanodiscus galileensis valves show strong concentric undulations according to the plane of focus either on the marginal or central areas (Figs 19 & 20). Valves range from 20-48 µm in diameter. The marginal area (Fig. 19) also displays a characteristic sub-marginal hyaline silica ring and broad interfascicles. Fascicles are c. 5 in 10 µm, proximally uniseriate with large blocky areolae becoming finer and bi-or triseriate in the distal third portion of the valve. Interfascicles are distinct and about 1.5 µm broad in the midradius area. Concave valves show a similarly strong undulation (Figs 21 & 22) with 5-6 hyaline areas (marking fultoportulae, see below) in the mid-radius position on the valve face (Fig 21, arrowed) and these are particularly obvious in well silicified specimens (Fig. 23). Fascicles on these valves (morphotype 'B' of Håkansson & Ehrlich 1987) appear to contain finer areolae and are biseriate for about half the valve distal radius. In this region, interfascicles are often about 1 µm broad but in strongly silicified specimens the width is similar to those in valves with hyaline sub-marginal rings.  In SEM, the external view of convex valves of S. galileensis clearly shows the marked central undulation with irregularly spaced spines subtended from the sub-marginal silica ring (Fig. 24, note the similarity of this specimen with S. aegyptiacus as depicted in Fig. 63, Håkansson & Locker 1981). The fascicles are uniseriate becoming biseriate or triseriate near the valve face-mantle junction (VFMJ) with small areolae (12-13 in 10 µm). External areolar openings are generally < 1 µm in the mid valve face area and, consequently, the fascicles are narrower than the interfascicles. Internally, these convex valves (Fig. 25) show marginal fultoportulae on each interfascicle at the VFMJ, one or two sub-marginal rimoportulae (arrowed) and a ring of central area fultoportulae (Fig. 26) with either two or three satellite pores. The mantle areolae, cribra and fultoportulae are generally as in Håkansson & Ehrlich (1987) except in this material the number of central area processes is lower. The concave valves confirm heterovalvy (Fig. 27) with raised interfascicles each subtended by a spine, distinct external openings of central area fultoportulae (arrowed) and a greater degree of areolation (with distally extended bi-and triseriate fascicles) compared to convex valves. Internally, the concave valves display a ring of sub-marginal fultoportulae (Fig. 28, arrowed) that initiate distally an usually biseriate fascicle (Fig. 29). There are typically three oblique sub-marginal rimoportulae (Figs 28-29). Central area fultoportulae possess two (Fig. 29) or three (Fig. 30) satellite pores and occur in depressions near finely porate domed cribra (Fig. 30).
Size distributions:-To help substantiate the two varieties of S. neoeagypticus, diameters of 70 random valves (S. neoaegypticus sensu lato) were assigned to size classes and the distribution plotted (Fig. 31a). These data show a bimodal size distribution with two modal size classes, 30-40 μm and 50-60 μm; the larger class was always composed of S. neoaegypticus valves and the smaller class was mainly S. aegypticus var. fekrii. For S. galileensis, diameters of 60 random valves were assigned to size classes and the distribution plotted (Fig. 31b). There were approximately equal numbers of convex and concave valves resulting in a unimodal size class distribution with 35% of valves in the 30-35 μm size class.  Table 1 where valve size, extent of uniseriate areolae serve best to discriminate the latter in routine LM examination. Another small form, similar to S. neoagypticus var. fekrii but with distally wider (4/5 areolae) fascicles, was recognised in GARK1 and tentatively named as cf. asteriodes but this taxon requires further study. Comparable data for S. aegyptiacus from Håkansson & Locker 1981 are included in Table 1 to demonstrate differences between this species and the two newly described S. neoaegypticus taxa. The most discriminating features of the latter are a larger valve diameter range and differences in valve face areolar structure. Strongly concentric valve undulations a distinct marginal hyaline ring and marginal spines are all lacking in S. neoaegypticus sensu lato. The morphological characteristics of S. galileensis obtained from the Qarun core material compare favourably with the published data for this species (Håkansson & Ehrlich 1987) shown in Table 2. The original description and the existence of heterovalvy is generally substantiated and only very minor differences were found in this study. The latter extends slightly the original description by increasing the valve size range and including more variation in the number of rimoportulae and in the arrangement of satellite pores around fultoportulae.  Håkansson & Locker (1981) are also given for comparison; na indicates data not available, ? indicates inferred value.

Discussion
The presence of diatom-rich marl sediment in the basal sections of all three cores provide evidence of a lake that was formerly much more extensive than today. The planktonic diatoms in these sections were dominated by Aulacoseira and Stephanodiscus taxa and, while A. granulata (Ehrenberg) Simonsen (1979: 58) is common in the Nile River today (Talling et al. 2009), the latter is not. Sediment records show that species of the genus Stephanodiscus were formerly widespread and often very abundant in large East African lakes during the Plio-Pleistocene and Holocene (Gasse 1980(Gasse , 1987. However, few species present in the Faiyum core sections appeared to be recognisable as previously described African endemics. Those Stephanodiscus species present in the Faiyum sediments appear to have developed within the lake rather than from populations brought northwards into the lake via Nile floods. In the thinly laminated marl sections of cores QARU9 and 10, Stephanodiscus galileensis was fairly common; it has not been previously reported for the Faiyum but its locality corresponds to the fossil type material from Lake Kinneret (Håkansson & Erhlich 1987). The species seems to have a very restricted Middle Eastern distribution although Krammer & Lange-Bertalot (2004) note that it is a planktonic form in eutrophic African freshwaters. Tuji et al. (2003) comment on the similarities of S. akanensis Tuji, Kawashima, Julius & Stoermer (2003: 2) (so far a Japanese endemic species) to S. galileensis. Mueller-Wilmes (1988) recorded 11 unidentified Stephanodiscus species in Lake Qarun sediment cores and S. galileensis probably occurred Marginal fultoportulae interfascicle termini, raised interfascicle termini, raised interfascicle termini, below spines Satellite pores 3 at margin, 2 centrally 3/4 at margin, 2 centrally 3 at margin, 2 centrally Rimoportulae 3-6, sub-marginal usually 2, sub-marginal at least one, marginal within this unknown group. Other unidentified Stephanodiscus taxa could correspond to different dissolution states of this species where erosion of both the marginal and the broad interfascicle areas can alter markedly valve appearance in LM. The marked heterovalvy in S. galileensis described by Håkansson & Erhlich (1987) is supported in this investigation as are the characteristics of both the convex and concave valves. Failure to recognise this heterovalvic condition can also cause species identification problems in routine counting since the concave valve is distinctly more finely striated and has a higher density of fascicle areolae (cf. Table 2). As noted by Håkansson & Erhlich (1987), the distinctive hyaline marginal ring is characteristic of the convex valve and, while not definitive of the species, it is a useful initial identification feature. The descriptions of both frustular valves of this species are extended slightly by reporting details ultra-structure features (cf. Table 2) associated with the valve processes and the valve size range is extended to smaller forms (to 20 μm valve diameter).
The other Stephanodiscus species of note found to date in the Faiyum sediment cores is described as a new species Stephanodiscus neoaegypticus. This species occurred most commonly in GARK1, from a region that would have been partially isolated from the main lake in the early Holocene. Although the 'See Garak' area of the Faiyum is the type locality for Stephanodiscus aegyptiacus Ehr., this species appeared absent in Gharak sediment where only a few S. galileensis and the newly described species were common. Ehrenberg (1854) noted that S. aegyptiacus is an extinct form that also occurs in Atlantic Passat winds but the latter does not substantiate its occurrence as North African. Associated diatoms in his drawing (in Håkansson & Locker 1981) do however indicate a sedimentary source within the Faiyum. The only extant lake (Birket Abu Talib) in the Gharak is small and shallow and receives agricultural drainage so is most unlikely to support S. aegyptiacus today. While its occurrence in the Faiyum remains to be established, the latter species seems unreported in recent lakes except for several sites in Florida (Whitmore 1989).
Despite the genus name, Stephanodiscus neoaegypticus is distinctive for its total lack of spines in all specimens examined in this study. While marginal spines are characteristic of the genus (e.g. Krammer & Lange-Bertalot 1991) may other characters of S. neoaegypticus (fascicule structure, arrangement of valve face and mantle areolae, the external rimoportula and the criba) are all typical of Stephanodiscus. The new species has few characters in common with S. galileensis (cf . Tables 1 & 2). Similarly, it is clearly separated from S. aegyptiacus Ehr., a species for which heterovalvy is not confirmed (Håkansson & Erhlich 1987). There are, however, issues with the description of S. aegyptiacus, the convex valve figured in SEM from the type material ( Fig. 63 in Håkansson & Locker 1981) has a marginal hyaline ring, raised interfascicular ribs and fascicular areolar structure exactly similar to S. galileensis. The internal valve image (Fig. 65 in Håkansson & Locker 1981) of S. aegyptiacus shows a central area ring of fultoportulae and sub-marginal rimoportulae in the same positions as in S. galileensis. Further, an external valve of S. galileensis, Fig. 87 in Håkansson (2002), and the external valve of S. galileensis, Fig. 63 in Håkansson & Locker (981) seem identical. The diagnosis of S. aegyptiacus is also virtually the same as that for the convex valve of S. galileensis excepting a smaller size limit for the former. As noted by Håkansson & Locker (1981), other workers have expressed doubt in the validity of S. aegyptiacus (e.g. VanLandingham 1978) and while the species concept is valid, conspecificity with S. galileensis is suspected. If heterovalvy can be demonstrated in the original material of S. aegyptiacus, then S. galileensis may well be synonimous.
Validating and establishing diatom species in ancient lake sediments is often made difficult by poor preservation, but establishing a secure taxonomic foundation is essential to assure integrity for biostratigraphies within and between lakes. Furthermore, a secure taxonomy is needed for both extant and fossil species so that species distributions through time and space can be determined. Provenance of individual species as well as information about the diversity of either contemporary communities or fossil assemblages is needed to help understand how diatom diversity in ancient lakes developed and sometimes declined. Questions remain about the origin of the diatom species recorded in Lake Qarun sediments but it is already clear that the Stephanodiscus flora of the early Holocene deep lake arose and declined with some rapidity (cf. Flower et al. 2012). Until a secure chronology is developed for a continuous Holocene sequence for Lake Qarun, the precise timing of the major environmental changes in this once great lake remains elusive.