Morphological variation suggests that chitinozoans may be fossils of individual microorganisms rather than metazoan eggs

(a). Systematic palaeontology

Incertae sedis group CHITINOZOA Eisenack, 1931

Order PROSOMATIFERA Eisenack, 1972

Family CONOCHITINIDAE Eisenack, 1931, emend. Paris, 1981

Subfamily BELONECHITININAE Paris, 1981

Genus Hercochitina Jansonius, 1964

Hercochitina violana sp. nov. Nõlvak and Liang

(figure 1a–k)

?1985 Hercochitina repsinata (Schallreuter) n. comb; Melchin & Legault, p. 208, pl. 1, fig. 1.

2007 Clathrochitina? sp. n. 1; Goldman et al., p. 267, fig. 13: 1, 4.

Etymology. In reference to the type stratum, the Viola Springs Formation of Oklahoma.

Diagnosis. This species is characterized by its distinct ornamentation, low and multi-rooted longitudinal spines developed in the lower part of the chamber, and jungle-like complex ones on the basal margin. A constriction near the base and a mucron on the base are also evident.

Types. Holotype, sample HWY99-43 (43 m above the base of the formation), figure 1i (repository number NIGPAS19-28); Paratype, sample HWY99-37.8, figure 1a (repository number NIGPAS19-20).

Type locality. Section D in the west side of State Highway 99, 3.3 miles south of Fittstown, Arbuckle Mountains, south-central Oklahoma, USA [22].

Type stratum. The upper part of the Viola Springs Formation, ranging from sample HWY99-37.8 to HWY99-51, Katian, Late Ordovician.

Material. Seventy-three specimens from 50 g of sample HWY99-37.8, 258 specimens from ca 300 g of sample HWY99-43 were measured.

Dimensions. See figure 2 and the electronic supplementary material, S1.

Description. The vesicle size ranges from small to medium and the outline from stout to slender. Chamber subcylindrical to subconical with a short cylindrical neck developing a short flaring collar. The flexure is so broad that it is difficult to identify the boundary between the neck and the chamber. Flanks are straight to slightly convex. Basal margin sharp to blunt. Base is flat with a mucron. Constrictions are inconspicuous or weak in smaller specimens and become stronger and more distinct in larger, always slender, specimens. The constriction diameter is about three-fifths to nine-tenths of the maximum diameter at the base of the vesicle. The vesicle wall is covered by multiple kinds of spines: simple or multi-rooted spines distributed near the collar, low and multi-rooted longitudinal spines developed in the lower part of the chamber around the constriction, much stronger and more complex ones on the basal margin, and simple and multi-rooted spines developed concentrically in the base around the mucron.

Remarks. The new species, H. violana sp. nov., is easily distinguished from other Hercochitina species by its less-developed longitudinally distributed spines in the upper and middle parts of the chamber. The distinct ornamentations more likely represent a transitional form between the typical Belonechitina and Hercochitina, but more closely resemble the ornamentation of Hercochitina by occupying well-developed vertical rows of spines or distinct crests, even if these distinguished features are confined within the lower chamber of the vesicle, which is the main feature of typical Hercochitina.

The species resembles Hercochitina repsinata (Schallreuter) Melchin & Legault [25] both in vesicle size, outline and similar ornamentation. It differs from Belonechitina repsinata Schallreuter, 1981 [26] in developing a remarkable constriction near the base. The holotype of B. repsinata is considered to be a badly preserved Belonechitina robusta (Eisenack, 1959) [27]. Thus, B. repsinata might be an invalid name and the specimens reported by Melchin & Legault [25] could potentially belong to H. violana sp. nov. Additionally, the specimens previously identified as Gen. et sp. n. 1 (Clathrochitina?) [22] are from the same level of the same section as the materials recovered in the present study. They are reassigned to Hercochitina because the spines on the margin are multi-rooted rather than processes and the flexures are inconspicuous.

(b). Morphological variation of Hercochitina violana sp. nov

The morphology of the new species shows variation both in vesicle size and outline (figures 1 and 2). The vesicle length (L in abbr.) varies more than a threefold from 94 to 317 µm. The maximum diameter (D_p) and the diameter of the constriction (D_cons) vary from 54 to 125 µm and 40 to 94 µm, respectively. All the measured parameters are normally distributed and the specimens on the PCA are well grouped, which support the idea that all the specimens belong to a single population.

Overall, the length variation is much greater than the diameter variation (figure 2g). The vesicle outline changes from a stout shape to a slender form with an indistinct but longer neck. The constriction also becomes more distinct as the specimens become larger and slenderer. Taking into consideration the normally distributed parameters, the substantial but continuous variation suggests a longitudinal growth trend rather than intraspecific variation. If the specimens are exhibiting population variation, it is plausible to argue that the predominant length, from ca 150 to 210 µm, potentially represent the mature stage of the species. Individuals longer than 210 µm and shorter than 130 µm show significantly decreased proportionality and may represent the largest grown and juvenile specimens, respectively. Notably, except for some small-chambered taxa of the family Desmochitinidae [22], such as Desmochitina, Calpichitina and Eisenackitina, no specimens smaller than the smallest H. violana sp. nov have been reported from the same sample.

(a). Variation of other chitinozoans

According to the compiled dataset, including 574 valid records (measurement data of the other 19 cases are incomplete; in the electronic supplementary material, S2), the ratio between the longest and shortest specimens within the measured assemblage ranges from 1.03 to 4.43, three-quarters of which range from 1.0 to 2.0 (figure 3a); however, in most species, only a relatively small number of specimens have been measured. When using data with more than 40 specimens measured, the ratio increases: 74% of the 116 valid data range from 1.5 to 2.5 and 19% with an L_max/L_min ratio more than 2.5 (figure 3b). Furthermore, 60% of the data share a ratio around 2.5 and only 15% are less than 2 but still larger than 1.5 when the ratios are based on more than 100 specimen measurements (figure 3c with 20 valid data).

The LOESS regression of 498 valid data (complete records with measured specimen numbers) indicates that variation estimates increase when at least 40 specimens are measured and an expected ratio around 2.7 is estimated (figure 3d). The variability range in the family Desmochitinidae (ratio of ca 2.7) has no evident difference from the Conochitinidae (ca 2.8) and Lagenochitinidae (ca 2.5) in the expected data by LOESS regression (figure 3e–g), even though the size of the dataset within each family varies significantly. These results indicate that Desmochitinidae, characterized by the key structure of operculum and with the only record of cocoon-like preservation [19,28]—several tens of specimens clustered together and covered by an organic-walled sheath—shares a similar morphological variation with other chitinozoans which possess a prosome instead of an operculum.

However, an analysis focusing on the data gathered on Desmochitina specimens, which are characterized by an ovoid vesicle and absence of neck, and sealed with the most typical operculum, indicates a smaller morphological variation. Except for the record of Desmochitina ornensis, some specimens of which occupying a short but distinct neck as shown in the type locality [23], all the other 29 datasets share an L_max/L_min ratio limited to 2 (figure 4a) and two-thirds range from 1.2 to 1.6. The LOESS regression, which might be less reliable owing to the limited data, shows an estimated value around 1.9 and 1.5, respectively, with or without D. ornensis (figure 4b). The variation of Desmochitina minor, within which recorded the only cocoon-like preservation, ranges from 1.1 to 1.9 with an average value of 1.47 based on 11 datasets in which the number of measured specimens varied from 2 to 50. Although the data of Desmochitina are limited for statistical analysis, a plausible trend is that taxa within Desmochitina, especially the D. minor, sharing an L_max/L_min ratio of less than 2 can be estimated, which is smaller than the variation in other taxa occupying a neck.

Except for the group of Desmochitina, on average, the intraspecific chitinozoan vesicle length variation is around 2.7 and can be more than 3 when a sufficient number of specimens are measured. This new analysis of variation is important because the size limit is among the main diagnostic characters used in species definitions and identifications. However, the biological variation in fossil populations has always been difficult to document because of preservation, time-averaging and inability to assess geographical variation. In addition, chitinozoan specimens are commonly compressed and deformed, the diagnostic ornamentations are easily removed or destroyed and have a relatively simple anatomical outline. All of this has contributed to the fact that the range of variability in chitinozoans has often been underestimated, and promoted a general hesitancy for workers to consider a group of specimens with great size variation to belong to a single species. In other words, size variation is potentially even larger within most groups in chitinozoans.

(b). Variation in fossil and extant invertebrate eggs

Fossils of aquatic invertebrate eggs are generally rare [29–35] and little is known about their size variation owing to limited numbers of specimens and often poor preservation. Almost all reported fossil eggs of aquatic invertebrates have a simple rounded or ovoid shape without any plug or lid-covered opening. They are different from chitinozoans both in outline and main features. The reported ratio of L_max/L_min is generally below 2 [34,36,37], which is smaller than the estimated size variation of chitinozoans when enough specimens were measured, with the exception of Desmochitina. Overall, the fossil egg-size variation is poorly known and has thus little significance in the analogous study owing to the limited number of case studies as well as limited number of eggs in each individual case.

However, the size variation of eggs is well studied among a wide phylogenetic range of extant animals [38–42]. Therefore, we compiled a dataset of 50 CV estimates, representing six phyla, 10 classes, 20 orders and 45 species of extant aquatic (mostly marine) metazoan species (electronic supplementary material, S3.2), to make a comprehensive view of egg-size variation. The distribution of these values is normal (Shapiro–Wilk W-test; W = 0.916863, p < 0.0018; figure 5) with a mean of 6.59, a standard error of 0.590, a median of 5.35, and a range of 0.7–16.67. The upper and lower 99% confidence intervals are 8.23 and 5.00, respectively.

However, the CV of size variation on H. violana is 20.68, with an L_max/L_min ratio of 3.4. A one-tailed, one-sample t-test [43] indicated that the CV of H. violana is significantly different from the above distribution (t = 3.344, p < 0.001). This allows us to conclude that extant aquatic invertebrates in our broad phylogenetic survey exhibit smaller variation in egg size than exists in chitinozoans. The t-test rejected the hypothesis that the CV of H. violana reported here is part of the same distribution with a high level of probability. In other words, data on the egg-size variation of both fossils and extant invertebrates are incompatible with the hypothesis that chitinozoans are eggs.

(c). Morphological implications on the biological affinity

Apart from the distinct size variation of chitinozoans, the predominant egg hypothesis does not explain well the complex structure of prosome, which is the primary character for subdividing chitinozoans at the order level. For example, systematic studies of prosome morphologies have documented diverse forms and structures among chitinozoan species [44,45]. Another detailed SEM image of a prosome [14] shows a delicate and complex structure consisting of several transverse septa connected by a central vertical structure. It is remarkable to see that there is a thin membrane-like structure around the septa and a wide membrane covering the lowermost thick septum, the rica. This structure is much more complex than is known in any eggs. It is more likely a key functional structure for the chitinozoan individual, even though the specific function remains hitherto unknown. Also, the strong processes and complicated spines developed in some chitinozoans, together with the rapid evolutionary trend, especially in their ornamentation [19,25,46], are atypical of metazoan eggs.

In the history of exploring the biological affinity of chitinozoans, three main stages can be summarized: (i) relationship with various groups of Protozoa mainly based on the vesicle outline from the 1930s to 1980s; (ii) eggs of unknown metazoans, ignited by the discovery of the cocoon-like clusters [28] during the 1960s to 1980s; and (iii) widely adopted as eggs or egg capsules of an unknown group of soft-bodied metazoans since the 1990s. Detailed information on different hypotheses can be found in the literature [14,19,23,47]. The success of the egg hypothesis is mainly owing to the hermetically sealed vesicle which restricts communication between the inner chamber and the outer environment, and the cocoon-like clustered specimens. However, among the 57 recognized chitinozoan genera [48,49], the cocoon-like clusters have only been observed in a single genus—Desmochitina [19,28]. Furthermore, the cocoon-like preservation could also make sense to be interpreted as a reproduction stage, for instance, a multiple fission stage of an individual organism.

In addition to the cocoon-like clusters in Desmochitina, there are linear or coiled catenary forms, which are common in Desmochitina, Bursachitina, Armoricochitina and Linochitina [17,50]. Those forms have also been compared to the eggs of polychaetes and molluscs [16,23,51]. However, the difference is that the eggs in similar catenaries lack a covered opening sealed by a prosome or operculum, which is the diagnostic character of chitinozoans. Besides, the various types of attachments seem to be not suitable to be evaluated for the biological affinity significance because many invertebrates display a similar mode of egg laying [21].