Bohemicardia hainense (Maurer) revisted: insights into ontogeny and variability of a Devonian hippocardiid rostroconch (Conocardiida: Hippocardiidae)

Re-examination of type material and abundant specimens enable the reinstatement of ‘Conocardium hainense Maurer, 1885’, a species erroneously synonymised with Babinicardia clathrata by Rogalla and Amler. Based on the completely reconstructed morphology, C. hainense can be assigned to the genus Bohemicardia within the Hippocardiidae of the Hippocardioidea. About 600 specimens from a single locality available for study including a wide range of ontogenetic stages allow documentation of allometric growth, with shell width increasing faster than shell height. Protoconchs are not preserved, leaving the development from larval stages to early juveniles unrecorded. The large number of specimens confirms a rather narrow range of morphological variation within the taxon. Based on the preservation of the rarely observed external shell layer, conclusions on shell shape, presence of a hood and external shell ornamentation of the taxon are possible. The material confirms that recognising different shell layer architecture in conocardiid rostroconchs is a prerequisite for specific identification and basic taxonomy.


Introduction
Rostroconchs generally played a minor role in Palaeozoic fossil assemblages and mostly occur in small numbers, if compared, e.g. with brachiopods. Hence, taxonomic concepts are often based on very few specimens, and many species are monotypic. This hampers comprehensive characterisation of intraspecific variability and meaningful morphometry. Moreover, characters vital for identification are often inadequately preserved .
The present study is based on several hundred specimens collected in the early twentieth century from only a few localities of the 'Stringocephalenschichten' (upper Middle Devonian) of the Eifel Hills, the Bergisches Land and Hesse (Germany) (Fig. 1). The specimens had either been assigned to Conocardium hainense Maurer or were simply determined as Conocardium sp. and remained as yet unstudied. The taxon Bohemicardia hainense was established by Maurer (1885: 230) as 'Conocardium hainense' and first revised by Beushausen (1895: 390). Taxonomical errors caused by Rogalla and Amler (2006c) due to presumed missing type material are discussed below.
For rostroconch studies, particularly on ontogeny and variability, abundant material from a single layer is particularly welcome. In this respect, the present contribution provides data not available from most other localities, as there is no previous study on rostroconchs that is based on such a lot of material. This dataset is one of the largest ones of a single rostroconch species and, thus, not only offers data for a comprehensive description of the taxon but also unique insights into intraspecific variability and post-larval growth trends. Our study aims at describing, characterising and interpreting the variability and the ontogeny of a single species of hippocardiid rostroconchs based on a reliable number Schloss-Schönau-Str. 66, 52072 Aachen, Germany of specimens in order to gain basic knowledge about the ontogenetic development of the family or even superfamily. The history of research on the order Conocardiida Neumayr and the superfamilies Conocardioidea Miller and Hippocardioidea Pojeta and Runnegar was compiled and outlined in detail in Amler and Rogalla (2004: 318), and for an overview on the general morphology of conocardiids, we refer to Schröder- Rogalla (2005) as well as Rogalla (2007, 2013).

Geological background and stratigraphy
All specimens used for this study come from the Middle Devonian (middle to upper Givetian) of the Eifel Hills, the Bergisches Land and the Lahn Syncline in west-central Germany (Fig. 1).
The sedimentary succession of the upper Middle Devonian (Givetian) in the Eifel region mostly consists of limestones and marlstones. The majority of the studied specimens was collected from heaps of quarried material of the 'Obere Stringocephalenschichten' at the Girzenberg (Girzen Hill) near Sötenich, Sötenich Syncline, Eifel Hills, Germany. Schmidt (1936) and later authors were not able to place precisely the general locality name 'Girzenberg' cited by Beushausen (1895). Thus, the precise stratigraphic level that yielded the rostroconchs is unknown. Quiring (1914) described the marlstones and marly limestones of the Girzenberg as 'Girzenbergschichten', i.e. Girzenberg beds, and Ritz (1931) assigned the locality to the upper Middle Devonian. Schmidt (1936: 315) called this fossiliferous part of the middle to upper Givetian succession 'Girzenberg fauna' without definite stratigraphical position. Today, the rocks quarried at the Girzenberg are assigned to the 'Rodert Schichten' and 'Kerpener Schichten' or Rodert and Kerpen formations, respectively (Ribbert 1985;Struve et al. 2008;STD 2016;Litholex 2021) (Fig. 2).
The original locality of Maurer (1885), now type locality of Bohemicardia hainense, was given as 'Grube Haina(u)' where limestones of Middle Devonian age ('Kalke von Waldgirmes') were quarried. Kegel (1933) placed the limestones from Haina in the lower part of the 'Stringocephalen-Stufe'. Later, the term 'Massenkalk' was introduced for these rocks and is still in use as 'Massenkalk Formation' (Bender 2008;Clausen and Korn 2008;Flick and Nesbor 2021), which, however, is not a formal unit. The precise age of these limestones is still unknown, although these units are placed in the middle to upper Givetian (or even lowermost Frasnian) (Fig. 2).
From a palaeogeographical viewpoint, all localities were situated on the passive continental margin of southeastern Laurussia and adjacent submarine volcanic rises (Grabert 1998;Clausen and Korn 2008). While differing opinions exist on the precise latitude (Golonka 2002;Blakey 2021;Scotese 2021), the study region was certainly situated in the equatorial realm. Due to the rapid sea level rise during the Givetian (Johnson et al. 1985), the palaeogeography of central Europe was characterised by widespread carbonate shelf areas with local reef complexes (Fig. 1). The Eifel region in particular formed a patchwork of different tropical, shallow marine, reefal and lagoonal environments, which today are represented by a great variety of different rock facies (Grabert 1998;Struve et al. 2008). These ecological conditions favoured a flourishing development of rostroconch diversity and abundance (see also Rogalla and Amler 2006a, b, c).

Material
The present data pool consists of about 600 specimens; ca. 120 are embedded in the matrix and ca. 480 are isolated from the sediment. The size of the specimens at hand ranges from 2 to 15 mm (excluding the rostrum; Fig. 3), none of them has the larval shell preserved. Some 100 specimens show the outer shell layer altered to a whitish, soft, chalky matter and variable abrasion stages; in further ca. 120 specimens remains of the outer shell layer are still visible between the ribs of the middle shell layer or preserved only in patches (Figs. 4a,c,7,8,11). In the remaining specimens, only the ornamentation of the middle shell layer is clearly visible (e.g. Figs. 8f-j, 11a-e). Roughly 300 specimens are damaged, 200 of which are variably crushed. In nearly 350 specimens, the base of the rostrum is preserved, rarely more (e.g. Figs. 7a,f,k,8f,11a,f,k). The fragile hood is preserved in very few specimens (Fig. 4a), but the lunulazone is well preserved in most of the specimens. Furthermore, the steinkern formerly filling the orifice is preserved in a few specimens, precisely indicating the former presence of a hood (Figs. 4b,7f).
In addition to the type material, for measurements, description and analyses, 50 well-preserved specimens were selected individually in order to cover different preservation modes and growth stages (Table 1). Originally, all specimens of our study had been determined as Conocardium sp. or Conocardium hainense, respectively. Some specimens labelled as Conocardium clathratum (Archiac and Verneuil, 1842) were used for comparison and turned out to be incorrectly determined.

Methods
The material was studied with conventional methods, i.e. analyses with microscope, measuring tools and basic statistics (Excel). Photographs were taken with the digital microscope VHX-500F (Keyence) and a Leica MC 170 HD, and all measurements by use of a measuring tool provided by the respective microscope software. Based on Amler (1986), Richter and Amler (1994), Gierse (2003) and Schröder- Rogalla (2005), the following measurements were taken (Fig. 3): Length umbo (L U ) refers to the length from the anterior end of the snout to the umbo parallel to the dorsal axis. The length posterior (L P ) indicates the whole length of the rostroconch from the anterior margin to the posterior end of the rostral area (posterior margin) except the length of the rostrum. This is the basic measurement of the length for comparisons. The length of the whole specimen including the length of the rostrum (L R ) is called total length (L total ). Due to mostly incomplete preservation of the rostrum, L R and L total can rarely be measured. In well-preserved specimens, the diameter of the rostrum (D R ) can be measured near the rostral area at the base of the rostrum. Furthermore, the angle α (alpha) can be measured between the vertical axis and the dorsoventral keel or ridge marking the attachment of the hood as indicated by the lunulazone, giving a ratio from the rostral area to the whole specimen. The width (W) indicates the thickness of the specimen. The height (H) is measured from the umbo to the most ventral point of the ventral margin perpendicular to the dorsal axis. The ratio between the total height (H) and the height from the umbo to the orifice (H ori ), i.e. the rostral height, plays a major role for separation the superfamilies. In this study, this relation is used with reservation, as it is adulterated by abrasion and loss of the outer shell layer.  Amler (1986Amler ( , 2016, Richter and Amler (1994), Gierse (2003), Schröder- Rogalla (2005), Amler and Rogalla (2013) and Balik (2018) For details of the general palaeobiology and morphology of conocardiid rostroconchs used for the descriptions, we refer to Amler (1996Amler ( , 2016Amler ( , 2018, Rogalla et al. (2002Rogalla et al. ( , 2003, Schröder- Rogalla (2005), Amler and Rogalla (2013) and Amler et al. (2020).

Systematic palaeontology
Repository. Most of the specimens come from the Museum für Naturkunde der Humboldt-Universität zu Berlin, Germany (prefix MB.Ro.). A few specimens originate from the former collection of the Institut für Geologie und Paläontologie der Philipps-Universität Marburg and are now housed at the Senckenberg Forschungsinstitut und Naturmuseum, Frankfurt am Main, Germany (prefix SMF.Mbg.). The syntypes of the Maurer collection are housed at the Naturkundemuseum Darmstadt (prefix HLMD.MR.). The two presumed hypotypes of Beushausen (1895) are from the Beushausen collection of the Goldfuß Museum, Bonn, Germany (IPB beushausen 5, 6). A syntype of Bohemicardia rotunda (Paeckelmann) (SMF.Mbg.3362) and two specimens of Conocardium clathratum belong to the collections of the Senckenberg Forschungsinstitut und Naturmuseum Frankfurt (prefix SMF.).
Diagnosis (emended from Rogalla and Amler 2006c, p. 354). Conch bohemicardiine; snout triangular to nearly nasute in shape, snout rib narrow, snout gape drop-shaped in outline, marginal denticles developed, ventral snout margin diagonal to slightly concave; ventral margin of central body convex to narrowly convex, paralleling dorsal axis; rostral area gently concave, longitudinally ovate, cordate ovate or cordate circular with diagonal rostral keel; surface of conch ornamented with flattened radial costae.
Revised description (Figs. 4,5,6,7,8,10,11). Conch bohemicardiine; body carina not developed; shape of snout triangular to slightly nasute; course of ventral margin of snout running diagonally. Snout gape drop-shaped in outline, marginal denticles weakly developed; snout rib slender. Umbo small, pointed and protruding above dorsal margin; ventral margin of central body convex, running nearly parallel to dorsal axis ('parallel convex' after Rogalla and Amler 2006e); maximum curvature situated in the middle of the central body. Rostral area occupying about 80-85% of total height of the body, its surface covered with flat radial ribs forming a zigzag-shaped rostral keel (Fig. 7n). Juvenile specimens showing heartshaped, longitudinally ovate rostral area increasing in width (laterally) during growth; rostral area of adult specimens circular heart-shaped (cordate) (Fig. 11). Rostral area surrounded by lunulazone (Fig. 7d, i), a row of crescent-shaped growth lines representing attachment line of hood. Ovate orifice located at ventralmost point of lunulazone; in a few specimens orifice preserved as tubular steinkern (Figs. 4b,  7f). Rostrum oriented parallel to dorsal axis and relatively large, occupying about 20-25% of total height (Figs. 7a,f,k,8f,11a,k); in most specimens rostrum broken off or preserved only about 2 mm of its length; rostral clefts present.

Remarks.
A few specimens show a very slightly concave ventral margin of the snout due to variable preservation.

Original material and nomenclature
Conocardium hainense was erected by Maurer (1885, p. 230) based on an unknown number of syntypes, he simply stated that he had several specimens available for study. Two of the syntypes plus two plaster casts of probable further syntypes were available for our revision. Comparison of these specimens with the original figures (Maurer 1885, pl. 9, fig. 21, 21a;here Fig. 5) leads to the conclusion that the illustrations are inaccurate and embellished, but most likely agree. Beushausen (1895, p. 390), who regarded Maurer's specimens as insufficient, revised the taxon based on some(?) original specimens and newly collected material (Fig. 6). Beushausen's descriptions are much more detailed, but he did not re-illustrate the specimens figured by Maurer, nor did he mention the number of specimens available for his study. However, he considered Maurer's specimens as 'distorted and therefore not typical for the shape of the shell' (Beushausen 1895, p. 390, translated from German; see Fig. 5f). Thus, in his opinion the syntypes do not show the correct proportions of the conch.
Most of the material described and illustrated by Beushausen (1895, pl. 29, figs. 1-4) is lost. The morphology of the specimen figured in pl. 29, fig. 1 (coll. Maurer; lost) is clearly different from that of typical C. hainense, and the rostral area extends over the entire conch height; this specimen is therefore excluded from the species concept. Apart from the absence of the first body carina, which is probably abraded, this specimen rather conforms to the description of Bohemicardia clathrata. The specimen illustrated in pl. 29, fig. 2 of Beushausen (1895) from Sötenich (Fig. 6a-d) represents a well-preserved individual of Bo. hainense and perfectly fits the revised species concept. The specimen was kept in the collections of the 'Breslau Museum' (today Muzeum Uniwersytetu Wrocławskiego = Museum of the University of Wrocław, Poland), but is also considered lost (pers. comm. Paweł Raczyński, University of Wroclaw).
Beushausen (1895) also erected the variety C. hainense var. impressa (Fig. 6k-n) and mentioned a depressed rostral area as a characteristic feature. Furthermore, he described and figured one specimen provisionally named Conocardium n. sp. aff. hainense. However, the two specimens from the Beushausen collection that are labelled accordingly are not the illustrated ones and may have been mixed up.
Schröder- Rogalla (2005) and Rogalla and Amler (2006c) erroneously regarded Conocardium hainense as a subjective junior synonym of Babinicardia clathrata (Archiac and Verneuil, 1842), but excluded the variety impressa from the species concept of Ba. clathrata. This error is based on the fact that the type material was considered lost and, thus, not revisited.

Preservation and taxonomy
Most rostroconchs are preserved with both semi-conchs in their original position and contact as the continuous dorsal shell layer(s) prevent(s) a gaping of the 'valves' and disarticulation after death in contrast to bivalves (see Rogalla et al. 2003;Schröder-Rogalla 2005;Amler and Rogalla 2007 for details). Fragile and distal parts of the shell, such as the hood, the most anterior portion of the snout and the rostrum, are often damaged or broken off owing to post-mortem taphonomic processes. The ample material studied, however, allows almost complete reconstruction of the morphology of Bohemicardia hainense including indication of the hood deduced from the specimens embedded in matrix (Fig. 4a) and the presence of the lunulazone (Figs. 3, 4b, 7d, i).
As exemplified by the figured specimens (Figs. 4,7,8,11), for the determination of rostroconchs preservation of the different shell layers is proven fundamentally important. Rostroconchs possess three architectural shell layers, not mirrored by microstructure (Rogalla et al. 2002(Rogalla et al. , 2003Amler and Rogalla 2007). The inner layer is delicate and rarely exposed, and the middle layer is relatively thick, mostly consisting of strong radial ribs (Figs. 5c,e,6e,f,7k,8f,11a,f,k). This is the layer exposed and hitherto described in the majority of rostroconch taxa. The outer layer fills the interspaces of these ribs and smooths the external surface or shows reticulate patterns (Figs. 4a,c,8a). If the outer shell layer and parts of the middle layer are abraded, ill preserved or removed, the outline of the conch is modified significantly, leading to incorrect reconstruction or description of the outline (Amler and Rogalla 2013 and references therein). Many specimens show different modes of preservation, sometimes even in a single specimen (Figs. 7, 8;Amler 1996;Schröder-Rogalla 2005;Rogalla andAmler 2006b, 2007).
In contrast to Schröder- Rogalla (2005) and Rogalla and Amler (2006c, d), it could be proven by the presence of a lunulazone that Bo. hainense possesses a hood and, thus, is a member of the superfamily Hippocardioidea Pojeta and Runnegar and the family Hippocardiidae Pojeta and Runnegar (Rogalla and Amler 2006c). The six subfamilies of the Hippocardiidae differ in the ratio of rostral height to total height. Whereas the Barrandeicardiinae is characterised by a rostral height of only 50% of the total height, the Bohemicardiinae distinguishes a rostral area of about 75-90% of the total height. In all other subfamilies, the rostral area is as high as the entire height of the conch (100%). Consequently, due to the development of the rostral area and the lack of body carinae, Bo. hainense is a member of the Bohemicardiinae, even though the rostral area occupies a greater height than the original subfamily definition requires (for further details on differences within the family Hippocardiidae, see Schröder- Rogalla 2005, p. 133;Rogalla and Amler 2006c, p. 347). Therefore, the subfamily diagnosis was emended here.
Within the Bohemicardiinae, the genera mostly differ in the shape of the snout and the ventral margin. The triangular snout and the convex, curved ventral margin prove an assignment to the genus Bohemicardia Rogalla and Amler, 2006a, b, c, d, e. Consequently, the taxon Bo. hainense is not to be included in Babinicardia clathrata, as erroneously considered by Rogalla and Amler (2006c, d).
The specimens at hand labelled as Conocardium hainense var. impressa Beushausen 1895 (Fig. 6e-j) do not correspond to the description and the illustration of the variety presented by Beushausen. The most important difference from the type subspecies is the development of the rostral area. Beushausen described and illustrated the rostral area as depressed, concave, but both specimens at hand do not show this feature. This seems to have been either an observation error by Beushausen or the specimens are not the figured syntypes. It can be stated, however, due to the fact that the specimens of var. impressa show a different size and, thus, a different growth stage that there is a change in the shape of the rostral area from cordate longitudinally oval to cordate circular, just as in Bo. hainense. Based on these facts and supported by the large number of specimens from Sötenich, the two present specimens of Co. hainense var. impressa are included in the species concept of Bohemicardia hainense.
Furthermore, Beushausen (1895) erected a new species in open nomenclature, Conocardium n. sp. aff. hainense, and figured one specimen. We have two specimens at hand, neither of which is the illustrated one. One specimen is badly preserved, slightly deformed and shows different proportions compared with Bo. hainense, with an elongated snout. The other specimen is better preserved, partly showing the outer shell layer. As a difference from Bo. hainense, Beushausen stated a more elongated shape and more but weaker ribs. Furthermore, both specimens show a distinct and regular commarginal striation crossing the radial ribs, especially visible on the middle shell layer, but also with a fine cover by the outer shell layer. We exclude both specimens from the species concept of B. hainense based on the characters mentioned above and presume affinities with Babinicardia clathrata.

Species-level comparison
Three species are currently assigned to the genus Bohemicardia Rogalla andAmler 2006c, i.e. Bo. bohemica (Barrande, 1881), Bo. rotunda (Paeckelmann, 1913) and Bo. hainense (Maurer, 1885), which mostly differ in conch proportions. For both Bo. bohemica and Bo. hainense abundant material exists in different preservation and also with almost completely preserved outer shell layer. In Bo. rotunda, however, there is only the lectotype available, without the outer shell layer preserved (SMF.Mbg.3362; Fig. 9), all further syntypes are lost and additional material is not (yet) known.
In comparison, the ventral margin of Bo. hainense is curved, slightly convex, whereas Bo. rotunda shows a tighter convex ventral margin, slightly pointed in the central part. In contrast, Bo. bohemica shows a regular, evenly convex ventral margin. Bo. bohemica, particularly, differs from the other two species in general shape of the central body. The central body is much larger and regularly rounded, the snout is very small, similar to the rostral area and Bo. bohemica has a small rostral field, which is about 75% of the total height only. Bo. bohemica shows a moderately wide variability (see illustrations in Rogalla and Amler 2006c). Specimens of Bo. bohemica are easily distinguishable from Bo. hainense and Bo. rotunda as long as the thick outer shell layer is preserved. In abraded specimens, however, the ventral margin of the central body in Bo. bohemica becomes narrower and resembles specimens of Bo. hainense.
Moreover, the orientation of the rostrum relative to the dorsal axis differs among the species compared. In Bo. hainense, the rostrum and the dorsal axis are almost parallel, whereas in Bo. bohemica rostrum and dorsal axis diverge distinctly dorsally. In Bo. rotunda no details on the position of rostrum can be observed because the rostrum of the lectotype is not preserved (see Rogalla and Amler 2006c for more information).
Based on specimens found at the same locality in the Eifel region, the Wachtberg near Sötenich, Rogalla (2007) erected Rheiacardia lerami Rogalla, 2007. This taxon differs from Bo. hainense in the height of the rostral area (100%), a nasute snout with a slightly sinuate ventral margin and the orientation of the ventral margin of the central body. Additionally, Bieberiana pugnans (Whidborne, 1892) occurs in Middle Devonian strata of central and western Europe. This taxon was described under several names, e.g. Conocardium confusum Beushausen, 1895, Conocardium eifeliense Beushausen, 1895and Conocardium retusum Maurer, 1885, and later revised by Rogalla and Amler (2006c). Bi. pugnans differs from Bo. hainense mainly by the significantly higher rostral area, which accounts for the entire height of the conch. In addition, the snout of Bi. pugnans is straight alaeform and, thus, elongated. Although some specimens figured as Bi. pugnans may be slightly tectonically deformed, its overall appearance is more elongate than Bo. hainense.
Contemporaneously, Babinicardia clathrata (Archiac and Verneuil, 1842) occurs in Middle Devonian strata of the Eifel Hills. It is the only genus and species within the subfamily Babinicardiinae of the family Hippocardiidae, and it is the only species with a body carina, which gives this species a triangular appearance of the central body. This characteristic morphology, which is well recognised both with and without the outer shell layer preserved, differs in a few subtleties from Bohemicardia hainense. Ba. clathrata shows a characteristic, triangular box-shaped central body, delimited from the snout by a body carina, and a comparably straight or very slightly convex ventral margin. The height of the rostral area approximates nearly 100% of the total height, placing the orifice at the posteroventral corner. In well-preserved condition, these differences are easily visible, but if the body carina in Bo. clathrata is abraded, the box-like shape of the central body gets lost and the ventral margin appears more rounded. In contrast, if the ventral margin of the central body in Bo. hainense is abraded, it runs almost parallel to the dorsal axis and the rostral surface occupies almost the entire height of the specimen, mimicking the shape of Ba. clathrata. Thus, careful study of preservation patterns is necessary for proper identification of the specimens.

Variability and post-larval ontogeny
The large number of specimens of Bohemicardia hainense allows to extend the species description with some details on growth stages from post-larval juvenile to adult specimens. Furthermore, based on the 50 well-preserved  (Paeckelmann, 1913). a-d Original illustrations of Paeckelmann (1913, pl. 6, fig. 1, 1a-c). e-h lectotype SMF.Mbg.3362 from the Dorp Limestone (Knappertsbusch quarry) of Beek near Wuppertal-Elberfeld, Germany, in right lateral (e), dorsal (f), ventral (g) and posterior (h) view. Scale bar 1 mm specimens selected individually for measurements (Table 1), we received enough data on the variability of Bo. hainense concerning the basic dimensions of the conch (Fig. 10). This has been impossible so far for most other rostroconch taxa as usually there is not enough well-preserved material available. In this study, the largest specimen has a total length of about 15 mm and the smallest 2.5 mm, but not all of them were included in the measurements due to incomplete preservation. To estimate the range of variation and for comparison of the morphological development, three dimensions were determined in all specimens measured: length (L P ), height (H) and width (W). Although really basic the three ratios, L P to H, L P to W and H to W, clearly show very low dispersion of values (Fig. 10). The mean value x for L P to H is 1.22, for L P to W 1.36 and for H to W 1.11. The standard deviation in all three relations is 0.09, and the standard error of mean is 0.01. These data confirm that the taxon is relatively well circumscribed and characterised.
During post-larval ontogeny, the length of the specimens grows faster in comparison to the height and width. The width is small at first and equalises with the height values in the course of growth. The increase in width and height at different rates causes a change, particularly, in the shape of the rostral area. The smallest specimen shows a narrow and high rostral area (cordate longitudinally oval), whereas the larger specimens show a much rounder (cordate circular) rostral face (Fig. 11) being the result of increasing width relative to the height.
The number of radial ribs of the middle shell layer in the central body of the conch does not increase when the shell grows. Instead, both ribs and interspaces become broader ventrally. With increasing conch width, the number of ribs covering the rostral area increases slightly. A third morphological change concerns the diameter of the rostrum. In small specimens, the size of the rostrum is relatively larger than in bigger individuals. Its relative diameter decreases from 2/5 of the entire conch height to 1/4 (Fig. 11a, k).
The ontogenetic growth from the larval to the early juvenile stage is not preserved in the studied material. Deduced from earlier observations (Schröder-Rogalla 2005), rather differing relations of the measured dimensions are expected compared with our analysis (Fig. 10). Growth of the twolobed, pseudo-bivalved larval shell in anterior and lateral, but slowly in posterior direction terminates at about 1 mm length (Schröder-Rogalla 2005) and the early juvenile shell is formed below the protoconch. Thus, specimens of 1-2 mm in length are required to report on this growth stage.

Palaeoecology and possible life habits
The palaeoecological distribution of conocardiid rostroconchs seems to have been controlled by various environmental factors such as nutrient supply, sediment grain size and stability, salinity, oxygen content, and water temperature. The wide range of basic morphologies in rostroconchs as well as palaeobiological constraints derived from the basic molluscan bauplan lead to the assumption that several lifestyles were realised for the occupation of different niches, e.g. reef margin, soupy mud or unstable sand (Richter and Amler 1995;Rogalla and Amler 2005a, b;Amler and Rogalla 2013;Amler 2016). Thus, the morphology of many Conocardiida reflects their life habits, such as mobile epibenthic crawling, immobile semi-endobenthic sediment sticking, and the loss of the hood in advanced conocardiids indicates a rather mobile, almost endobenthic life position (Runnegar 1978;Pojeta 1987;Richter and Amler 1995;Amler and Richter 1997;Schröder-Rogalla 2005;Amler and Rogalla 2013). The related trophic perspective, particularly an alternative interpretation of the life habit of hippocardioid rostroconchs, was discussed elsewhere (Schröder-Rogalla 2005; Rogalla and Amler 2005a, b;Amler and Rogalla 2013;Amler 2016). Accordingly, hippocardiids most probably obtained a specific life position relative to the sediment surface. With respect to mobility, stable position within  Table 1) the substrate and efficient orientation of the posterior surface of the hood, only one life position is able to meet each of the demands properly, a rather horizontal orientation of the rostral area parallel with the surface of the substrate (Amler 2016).
The morphology of Bohemicardia hainense supports these assumptions except for the large conch size required for photosymbiosis and the relatively large diameter of the rostrum. The short, compact conch and the relatively large diameter of the posterior rostral area plus the hood (compared with height and length) were hardly suitable for rapid or even moderate locomotion, even if the grain size of the sediment offered sufficient pore space for grain movement (Amler 2016). The small size of the taxon, the relatively small rostral area surrounded by a narrow hood if compared with Devonian taxa from North America [e.g. Hoareicardia cunea (Conrad, 1840)] and Mississippian taxa [e.g. Hippocardia hibernica (Sowerby, 1812)], as well as the relatively wide diameter of the rostrum, indicate an early evolutionary stage of the mixed feeding habit.

Conclusions
• The study of a relatively large number of specimens and the re-examination of preserved type material enabled the reinstatement of 'Conocardium hainense Maurer, 1885'. • The completely reconstructed and redescribed morphology of C. hainense allows the assignment to the genus Bohemicardia within the Hippocardiidae of the Hippocardioidea. • The large number of specimens shows a rather narrow range of morphological variation within the taxon. • The abundant material available, including a wide size range of specimens, allows the description of the ontogenetic development. A change of shell dimensions during growth, particularly a distinct increase in width versus height is documented. Thus, the shape of the rostral area changes from narrow and high (cordate longitudinally oval) to roundish (cordate circular) in larger specimens. • Protoconchs are not preserved leaving the development from larval stages to early juveniles unrecorded. • Based on ample preservation of the rarely observed external shell layer information on the proper shell shape, presence of a hood and external shell ornamentation of Bo. hainense and on hippocardiids in general are possible. • The studied material confirms the difficulties in recognising different shell layer architecture in conocardiid rostroconchs, which is a prerequisite for specific determination. • Bohemicardia hainense appears to have been an abundant taxon in late Middle Devonian shallow shelf environments at the southern margin of Laurussia. • From the stratigraphic viewpoint, an evolutionary development from Bohemicardia bohemica (late Lower Devonian) to Bo. hainense in the Middle Devonian and Bo. rotunda in the early Upper Devonian may be plausible. Unfortunately, several strata between the recorded occurrences are void of rostroconchs and the precise age of the latter species is unknown, leaving the postulated lineage unproven. • The morphology of Bohemicardia hainense supports the assumption of an early evolutionary stage of the mixed feeding habit in hippocardiids for this taxon.