Early Permian Carbonitidae (Ostracoda): ontogeny, affinity, environment and systematics

An assemblage of four Carbonita species was deposited with charophytes, lungfishes and lysorophids in a lenticular mudstone from a Cisuralian freshwater pond deposit from the lowest Permian. Samples contained few adult Carbonita, indicating perhaps a stressed and unstable environment. Two species new to science, C. ovata n. sp. and C. triangulata n. sp., occur together with C. evelinae and C. pungens. Morphological characters of these Carbonita suggest an affinity with the Healdioidea, marine taxa that are probably ancestral to the Carbonitidae. The muscle-scar patterns of Carbonitidae, which comprise closely grouped circular bundles of secondary muscle scars, resemble closely those of Healdioidea and not those of Cypridoidea and Cytheroidea, whose muscle scars are fewer and spaced further apart. The muscle-scar pattern of C. pungens, described here for the first time, is a circular scar with an ascertainable pattern of secondary scars. C. pungens and species of Darwinula are morphologically similar, but study of additional specimens of C. pungens with better-preserved muscle scars is essential to determine their evolutionary affinity.


HISTORY OF CARBONITIDAE
The first known Carbonitidae evolved during the Mississippian, while the youngest are known from the Cisuralian or Guadalupian in the Permian (Cooper, 1942;Kellett, 1943). During this time, Carbonitidae had a wide distribution throughout Europe and North America in continental freshwater deposits (Scott & Summerson, 1943) and other presumably lowsalinity deposits with marine to estuarine affinities (Jones & Kirkby, 1879;Knight, 1928;Anderson, 1970;Tibert & Scott, 1999). Carbonitids are most common in deposits of freshwater swamps, slow streams and lagoons, all of which received freshwater (Scott & Summerson, 1943). The presence of Carbonitidae in marine and estuarine deposits indicates that some species may have tolerated oligohaline conditions or were transported from nearby freshwater habitats (Pollard, 1966).
The classification of genera now grouped into the Family Carbonitidae was debated widely during the twentieth century. Knight (1928) placed Carbonita into the Superfamily Cytheroidea because of the presence of a single species in marine deposits. Scott & Summerson (1943) compared Carbonita fabulina (Jones & Kirkby, 1867) to the modern freshwater species Cypridopsis vidua (Müller, 1776) and placed carbonitids into the Family Cyprididae Baird, 1845, having found no significant differences in the carapaces that would justify their being classified elsewhere. Swain (1976) placed the carbonitids into a separate family, Carbonitidae, which he erected based on their elongate-elliptical outline that is uncommon in species of Cypridopsis Brady, 1867. More recently, Sohn (1985 determined that the muscle scars of Carbonitidae resemble closely those of the marine Healdioidea and not of the Cypridoidea. He erected the Superfamily Carbonitoidea and, again, the Family Carbonitidae. Other researchers have found no conclusive evidence for classification of the carbonitids and placed them into an uncertain suborder and superfamily of the Order Podocopida (Whatley et al., 1993).

ONTOGENY AND SEXUAL DIMORPHISM
Four species of Carbonita occur in the Eskridge samples, including two previously undescribed species: C. evelinae, C. pungens, C. ovata n. sp. and C. triangulata n. sp. All specimens were preserved either as finely recrystallized valves or as steinkerns.  (Zeller, 1968). Carbonita specimens were collected from a green mudstone layer indicated by the asterisk (modified from Watabe & Kaesler, 2004 Graphs of ontogenetic series can show instar and adult stages where specimens are clustered. Stages can be difficult or impossible to distinguish, especially in the fossil record. This blurring of growth stages may result from time-averaging of specimens from different environments or seasons (Whatley & Stephens, 1977). Conspecific ostracodes tend to be smaller in the spring than in the fall (Whatley & Stephens, 1977). Brooks' rule suggests that crustaceans double their volume and increase linear dimensions by approximately the cube root of two (1.26) with each moult (Brooks, 1886). The rule is a general observation of crustacean growth but is not necessarily true of all ostracode ontogenies. Brooks' rule is useful for recognizing instars and adults of ostracodes and may provide insight into heterochrony in the evolution of the Ostracoda or into the timing of such ontogenetic changes as addition of appendages and development of sexual maturity.
Ten complete valves of C. evelinae were collected, representing three distinct instars ( Fig. 3; Pl. 1, figs 1-3). Bless & Pollard (1973) constructed a growth series of specimens of C. evelinae from the Westphalian of Great Britain and the Netherlands. Their data demonstrate two clusters of data points with mean height and length of approximately 0.37 mm and 0.87 mm and approximately 0.49 mm and 1.02 mm, respectively. Bless & Pollard (1973) inferred these two clusters to be the A-1 instar and the adult. If so, then the three instars in the Eskridge samples are probably the A-1 to A-3 instars, suggesting that no adults were collected. The largest specimens here have a mean height of 0.39 mm, similar to that of Bless & Pollard's A-1. While the mean height of instar A-1 of their specimens is similar to those in the Eskridge samples, the mean length differs: Eskridge specimens average 0.61 mm in length compared with 0.87 mm for Bless & Pollard's (1973). Their specimens range in height: length ratio from 0.36 to 0.52, while the Eskridge specimens range from 0.56 to 0.82. Such differences could stem from the very small sample size of the current study, be the result of environmental differences or be due to taxonomic differences. Specifically, the C. evelinae here may be, instead, C. humilis Jones & Kirkby, 1879, as this species has also a punctate surface ornamentation and a height: length ratio of 0.60 to 0.70 (Pollard, 1966). C. humilis, however, has rounded, not elongated punctae as is characteristic of C. evelinae and its anterior end is pointed more acutely than that of C. evelinae. The dearth of specimens of C. evelinae in the current samples makes it difficult to determine whether these specimens of C. evelinae follow Brooks' rule ( Fig. 3).
The specimens of C. pungens are a single valve of an adult and 17 A-1 instars ( Fig. 4; Pl. 1, figs 4-6). The adult specimen is similar in size to Bless & Pollard's (1973) adults, which have a mean height and length of 0.28 mm and 0.60 mm, respectively (Fig. 4). Specimens measured and illustrated by Anderson (1970) are 0.30 mm by 0.68 mm and 0.26 mm by 0.56 mm, both comparable in size to the adult specimen here (Fig. 4).
The samples yielded specimens of C. ovata n. sp. from adults to the A-3 instar ( Fig. 5a; Pl. 1, figs 7-8). Adults and A-3 instars are easily identifiable, with gaps separating them from other instars. Instars A-1 and A-2, however, have a broad distribution, with no appreciable gap between them. The dotted line in Figure 5a, separating the A-1 and A-2 instars, is based on Brooks' rule. This divided the collection into 16 adults, 38 A-1 instars, 39 A-2 instars and 8 A-3 instars. Adult height: length ratios range from 0.52 to 0.68. The width of adult valves was also measured to show variation in the posterior end and to assess sexual dimorphism. Widths of adults seem to be distributed bimodally, with four adults having a width less than 0.18 mm and nine adults having a width greater (Fig. 6a), suggestive of sexual dimorphism, although further investigation with a larger sample size is needed.
The samples yielded 53 specimens of C. triangulata n. sp.: 13 adults, 19 A-1 instars, 29 A-2 instars, and 2 A-3 instars ( Fig. 5b; Pl. 1, figs 13-14). Adult and all instars are easily identifiable, with gaps separating them from one another. Adult height: length ratios range from 0.58 to 0.69. The widths of adults may be distributed bimodally, with seven adults having a width less than 0.19 mm and six adults having a greater width (Fig. 6b), again suggestive of sexual dimorphism.
Variation in size and height: length ratios of adults of both C. ovata n. sp. and C. triangulata n. sp. could result from sexual  (1970) and Bless & Pollard (1973) showing groupings of adults and instars A-1 to A-3.  (1970) and Bless & Pollard (1973) showing groupings of adults and instar A-1.
dimorphism, which has been postulated for C. humilis (Jones & Kirkby, 1879) (Pollard, 1966;Bless & Pollard, 1973 and C. inflata (Jones & Kirkby, 1879) (Anderson, 1970), based on differences in the swelling of the posterior end. Bless & Pollard (1973 found also noticeable differences in height: length ratios, degree of arching of the dorsum and relative positions of maximum height of C. humilis among ostracodes from separate localities. Bless & Pollard (1975) recognized that the variation among localities could result also from conditions other than sexual dimorphism: parthenogenetic reproduction in an unstable environment or a combination of syngamic reproduction during unstable environmental times and parthenogenetic reproduction during stable environmental times.
Height: length ratios of adults of both C. ovata n. sp. and C. triangulata n. sp. are quite variable, but their distribution is not bimodal and the amount of variation is similar to that of the A-1 and A-2 instars. The widths of the valves are distributed bimodally (Fig. 6). This may indicate that the variation in adult specimens is sexual dimorphism, or it may stem from life in an unstable environment and time-averaging of the fossils. It is unfeasible to speculate on sexual dimorphism and the nature of the variability in C. ovata n. sp. and C. triangulata n. sp. until further information is available from study of larger collections that include other localities.

EVOLUTIONARY AFFINITY
Muscle-scar pattern, overlap of the hinges and free margins, ornamentation and shapes of species of Carbonita, along with their environment of deposition, have been used to establish their evolutionary affinities to other known ostracodes. Evolutionary affinities with the Superfamilies Cytheroidea (Knight, 1928), Cypridoidea (Scott & Summerson, 1943;Swain, 1961Swain, , 1976 and Healdioidea (Sohn, 1985) have been suggested. Species of the Superfamily Darwinuloidea and the Late Palaeozoic genera Whipplella Holland, 1934, andGutschickia Scott, 1944a, also have superficial similarities to some species of Carbonitidae (see Table 1). Although these taxa have not been suggested as ancestral to the carbonitids, Anderson (1970) and Bless & Pollard (1973) suggested that the genera Whipplella and Gutschickia are synonymous with Carbonita.  The value of muscle scars in the systematics of these Ostracoda has been debated. Scott (1944b) argued that muscle scars, when present, vary insignificantly within genera, whereas Gramm et al. (1972) found healdiid's muscle scars to be monotypic at the family level. Kristan-Tollmann (1977) judged that because fossil ostracodes have few characteristics, the musclescar pattern is of primary importance in taxonomic determination. Muscle-scar patterns of Bairdia McCoy, 1844, have changed little since the Carboniferous (Sylvester-Bradley, 1950). Alternatively, Smith (1964) found considerable variation in muscle scars among specimens of Cypridopsis vidua and Chlamydotheca arcuata Sars, 1901, and judged that they should not be used in classification.
Muscle scars of a few specimens of Carbonita pungens (Pl. 1, fig. 5) and C. triangulata n. sp. (Pl. 1, fig. 14) were observed and used to evaluate tentatively the evolutionary affinities. The muscle-scar pattern has an almost circular outline containing smaller secondary muscle scars within it (Figs 7a, c). Poor preservation of the muscle scar made it difficult to determine the number and shape of the secondary muscle scars, but they appear to be rounded to elongated (Figs 7b, d). Anderson's (1970) illustration of the muscle-scar pattern of Carbonita shows peripheral scars surrounding a few central scars (Fig. 8a). Sohn (1985) suggested that Carbonita's muscle-scar pattern resembles those of the Healdioidea. Pre-Carboniferous healdiid muscle-scar patterns are known only from a single Early Devonian species (Gramm, 1987). This species has a circular musclescar pattern with convex rows of 7-9 secondary muscle scars (Fig. 8b). Carboniferous and Permian healdiids have a circular muscle scar with many secondary muscle scars (Gramm et al., 1972;Kristan-Tollmann, 1977) (Fig. 8c). The secondary muscle scars can be divided into two groups: a central group of 6-8 secondary scars that vary in arrangement and a peripheral group that varies in arrangement and also in number (Gramm et al., 1972). The muscle scar of healdiids, as is true of Carbonita and many other kinds of ostracodes, is located slightly anterior of the median (Scott, 1944b).
The Superfamilies Cytheroidea and Cypridoidea both contain freshwater species and have been suggested as having evolutionary relationships to carbonitids (Knight, 1928;Scott & Summerson, 1943;Swain, 1961Swain, , 1976. The muscle scars of both superfamilies differ from carbonitids in being not arranged  circularly. Cyprid muscle-scar patterns contain a cluster of scars that is not radially or linearly arranged (Swain, 1961) (Fig. 8d), while cytherids have typically a nearly vertical row of four scars with a few anterior scars (Howe & Sylvester-Bradley, 1961) (Fig.  8e). Scars of both these superfamilies are less complex than the scars of healdiids and carbonitids. Freshwater species of Carbonitidae, Whipplella, Gutschickia and Darwinuloidea all evolved during the Late Palaeozoic and have broadly similar muscle-scar patterns. The muscle-scar pattern of Whipplella comprises radially arranged, elongated secondary muscle scars (Fig. 8f) and resembles that of Suchonella Spizharsky, 1937(Sohn, 1977. The muscle-scar pattern of Gutschickia is an almost circular outline containing smaller secondary muscle scars (Sohn, 1977) (Fig. 8g). The muscle-scar pattern of Darwinula Brady & Robertson, 1885, consists of up to 12 palmate scars that are unequal in shape (Sohn, 1976) (Fig. 8h).
The muscle-scar pattern of carbonitids resembles closely those of Healdioidea and not those of Cytheroidea and Cypridoidea, thus suggesting that healdiids are closer evolutionary ancestors of carbonitids. The similarities in muscle-scar pattern of all the Late Palaeozoic freshwater genera, even those that have not been regarded as Carbonitidae, suggest also a close evolutionary affinity.
Another feature of Late Palaeozoic freshwater ostracodes to be used in classification is the valve overlap. Holland (1934) used valve overlap to distinguish species of Whipplella from Carbonita. Differences of valve overlap, however, may not be a good taxonomic indicator. Reversals are known to occur in Cytheroidea, Cypridoidea and Bairdioidea Sars, 1888, both among genera and within species (R. C. Whatley, pers. comm., 2004) and in Carbonita (Bless & Pollard, 1973).
Valves of carbonitids have generally the right valve overlapping the left valve along the venter (Jones & Kirkby, 1879;Anderson, 1970;Bless & Pollard, 1973), with the exception of most specimens of C. pungens, which have the opposite pattern (Bless & Pollard, 1973). Darwinulids (Swain, 1961) and Gutschickia (Scott, 1944a) have the same overlap pattern as carbonitids, while in healdiids (Shaver, 1961) and Whipplella (Holland, 1934) the left valve overlaps the right. Cyprids and cytherids have both patterns of overlap. In the dorsum along the hinge, the left valve of Carbonita overlaps the right valve (Kellett, 1935;Anderson, 1970;Bless & Pollard, 1973). The same pattern of dorsal overlap occurs in Gutschickia (Scott, 1944a), but Whipplella has the reverse overlap pattern, right valve over left valve (Holland, 1934).
Whether C. pungens belongs to the genus Carbonita or Darwinula has been debated. The general left over right valve overlap is the reverse of most Carbonita and Darwinula species, and it lacks also a dorsal overlap at the hinge, characteristic of Carbonita. The muscle-scar pattern of C. pungens has not been observed previously. Cooper (1946) reclassified C. pungens as Darwinula pungens after reviewing Pennsylvanian specimens from Illinois. His reclassification was based on the maximum valve overlap being around the ends, not around the ventral margin, and lack of a dorsal overlap at the hinge -both characteristics of carbonitids. His specimens' heights and lengths are larger than those of other C. pungens. Anderson (1970) retained C. pungens in the genus Carbonita but regarded Cooper's D. pungens to be not synonymous with C. pungens because the right valve overlapped the left valve along all free margins. Pollard (1966) and Bless & Pollard (1973) disagreed with Anderson and regarded Cooper's specimens of D. pungens as Carbonita pungens, citing that the variance was the result of the structural simplicity of a small, thinly shelled species. Other Late Palaeozoic Darwinula species have been recognized from West Virginia (Sohn, 1976(Sohn, , 1985 and Virginia (Sohn, 1985) and have a typical darwinulid muscle-scar pattern. Some Darwinula species from Virginia are larger than C. pungens. Without being able to observe the muscle-scar pattern of C. pungens, its difference, if any, from Darwinula could not be assessed.  (1970, fig. 1g). (b) Healdia? obtusa (Abushik, 1971), Lower Devonian, redrawn from Gramm (1987, fig. 1). (c) Bythocyproidea sp., Mississippian, redrawn from Gramm et al. (1972, fig. 1). (d) Cypridopsis vidua (Müller, 1776), Holocene, redrawn from Moore (1961, fig. 21). (e) Cythere lutea (Brady & Norman, 1889), Holocene, redrawn from Moore (1961, fig. 185). (f) Whipplella sp., Permian, redrawn from Sohn (1977, fig. 1). (g) Gutschickia sp., Pennsylvanian, redrawn from Sohn (1977, fig. 1). (h) Darwinula stevensoni (Brady & Robertson, 1870), Holocene, redrawn from Sohn (1976, pl. 3).

PALAEOENVIRONMENTAL INDICATIONS
The presence of Carbonita species, charophytes [Stomochara moreyi (Peck, 1934)], lungfishes (Gnathorhiza sp.) and lysorophid amphibians (Brachydectes elongates Wellstead, 1991) and the absence of marine fossils in the green mudstone lens in the Speiser Shale indicate a freshwater environment, not a tidally influenced marine mudflat, as suggested previously (Schultze, 1985). Laterally adjacent to the mudstone lens is a red palaeosol (Hembree et al., 2004) that suggests it developed on a floodplain or coastal plain near a lake or pond (Picard & High, 1972). The mudstone lens has been interpreted as a pond deposit that filled a topographic low, possibly the result of eolian deflation or a shallow temporary stream cut (Hembree et al., 2004). The pond's source of freshwater is likely to have been from rainwater or overland flow as a result of local flooding (Hembree et al., 2004).
The distribution of carbonitids is variable both laterally and vertically within the Eskridge pond. There is no evidence of pooling of specimens in the lysorophid burrows or of the accumulation of specimens at desiccation surfaces. The variable distribution is probably the result of abiotic factors of wind or water, predation, a combination of both, or poor preservation.
Many of the specimens are nested, with small instars inside larger instars or adults. Often only two specimens are nested, but in some instances three or four specimens are nested. Commonly the same species are nested with each other, but occasionally two species are nested together. Lane (1964) noticed similar nesting of carbonitid valves in other Lower Permian rock units in Kansas. The nesting could be the result of water or wind influences or such biotic factors as stacking within the digestive tract of a predator.
The Eskridge samples produced few adult specimens and no specimens of instars before A-3. This could be the result of poor preservation or of their being too rare to occur in the samples. The dearth of adult specimens could also be the result of an unstable environment. The small size and isolation of the pond would have made it sensitive to local fluctuations in seasonal climate (Picard & High, 1972), producing an unstable environment for the carbonitids. Hembree et al. (2004) observed three subaerial exposure surfaces with desiccation features and lysorophid aestivation burrows within the pond lens, suggesting that these surfaces formed during drying episodes of the pond during dry seasons. Drying of the pond would have affected water quality and chemistry, including ionic strength, temperature, reduction-oxidation potential and pH. Some modern freshwater ostracods species can be sensitive to temperature changes (Echols et al., 1975;Delorme, 1990Delorme, , 1991 and specific conductivity and redox potential of the solution (J. L. Castle, pers. comm., 2004). A stressful and unstable environment may have caused the Eskridge carbonitids to have high juvenile mortality rates, dwarfism, or, less likely, early sexual maturity (Bless & Pollard, 1975).
A few other assemblages of carbonitids have been described from the Cisuralian of Kansas (Lane, 1964;Peterson & Kaesler, 1980). Lane (1964) found Carbonita associated commonly with darwinulids, fish teeth and scales, and Geisina Johnson, 1936, a brackish-water ostracode genus. They occurred also less commonly with Gutschickia and with charophytes. Lane (1964) interpreted the assemblages with darwinulids, charophytes and fish teeth and scales as occurring in freshwater deposits. The assemblages with Geisina were interpreted as being from freshwater to slightly brackish-water environments. He observed also that the assemblages of carbonitids were associated more commonly with green shales than red ones, suggesting that they were deposited in environments characterized by reduction of mud in a semi-permanent pond of freshwater or brackish water. Peterson & Kaesler (1980) collected an assemblage of six species of Carbonita from Upper Pennsylvanian deposits. They inferred the lagoon and tidal-mudflat sediments in which the assemblage was deposited to have been influenced strongly by influx of freshwater. One species, C. inflata, was abundant in the freshwater deposits but occurred commonly also in nearshore environments, indicating either tolerance of saline conditions and the ability to live in a variety of environments or that their carapaces were transported into these environments. Reexamination of specimens they identified as C. inflata show them to be conspecific with C. ovata n. sp.

CONCLUSIONS
The collection of four species of Carbonita from a Cisuralian pond deposit provides new and supporting information to the solution of problems associated with ontogeny, systematics, evolutionary affinities and palaeoenvironment indicators.
1. The assemblage of carbonitids, lungfish, lysorophids and charophytes in a mudstone lens indicates a freshwater pond environment, which agrees with the environmental interpretations of Hembree et al. (2004). 2. The dearth of adult specimens of Carbonita suggests a stressful and unstable environment. Stressful environmental conditions may have favoured either dwarfism or a high juvenile mortality rate. 3. The circular muscle-scar pattern of Carbonita with secondary muscle scars and its location anterior of the median suggest a close affinity to the healdiids, as Sohn (1985) proposed. 4. Differences in muscle-scar pattern and geological ranges indicate no close affinity to either cytherids or cyprids as proposed by Knight (1928) and Scott & Summerson (1943), respectively. 5. The circular muscle-scar patterns of C. pungens are the first to be described for the species, but the arrangement of the secondary muscles could not be ascertained. The carapace of C. pungens is similar in shape to that of the darwinulids, but additional study of specimens with better-preserved muscle scars is essential to determining their affinity. 6. The posterior widths of adult specimens of C. ovata n. sp. and C. triangulata n. sp. indicate a bimodal distribution, suggestive of sexual dimorphism in these species.  Latreille, 1806 Order Podocopida Müller, 1894Suborder Metacopina Sylvester-Bradley, 1961 Superfamily Healdioidea Harlton, 1933 Family Carbonitidae Swain, 1976Genus Carbonita Strand, 1928 Type species. Carbonia agnes Jones, 1870; designated subsequently by Bassler & Kellett, 1934, p. 237.

Remarks.
The largest specimens from the Speiser Shale are smaller than Bless & Pollard's (1973) adults, perhaps for reasons discussed earlier.
Remarks. The left valve over right valve overlap is the reverse of typical Carbonita. The elongated carapace shape and valve overlap resemble closely those of the Permian Darwinula hollandi Scott, 1944a. Scott (1944a Distribution. Upper Pennsylvanian to Cisuralian of Kansas.

Remarks.
The carapace of C. ovata n. sp. is shaped similarly to C. inflata (Jones & Kirkby, 1879) but lacks the punctae ornamentation and the specimens from the current study are smaller. Widths of the posterior portions of adult specimens are distributed bimodally, suggesting sexual dimorphism. All specimens identified by Peterson & Kaesler (1980)

Distribution. Cisuralian of Kansas.
Remarks. The carapace is shaped similarly to C. subangulata (Jones & Kirkby, 1879). C. subangulata, however, is known only from Mississippian rocks and the specimens from the current study are appreciably smaller than C. subangulata described in the literature (Anderson, 1970), suggesting that the two are not conspecific. Widths of the posterior portions of adult specimens are distributed bimodally, indicating sexual dimorphism.