During the Devonian Nekton Revolution, ammonoids show a progressive coiling of their shell just like many other pelagic mollusk groups. These now extinct, externally shelled cephalopods derived from bactritoid cephalopods with a straight shell in the Early Devonian. During the Devonian, evolutionary trends toward tighter coiling and a size reduction occurred in ammonoid embryonic shells. In at least three lineages, descendants with a closed umbilicus evolved convergently from forms with an opening in the first whorl (umbilical window). Other lineages having representatives with open umbilici became extinct around important Devonian events whereas only those with more tightly coiled embryonic shells survived. This change was accompanied by an evolutionary trend in shape of the initial chamber, but no clear trend in its size. The fact that several ammonoid lineages independently reduced and closed the umbilical window more or less synchronously indicates that common driving factors were involved. A trend in size decrease of the embryos as well as the concurrent increase in adult size in some lineages likely reflects a fundamental change in reproductive strategies toward a higher fecundity early in the evolutionary history of ammonoids. This might have played an important role in their subsequent success as well as in their demise.
A major macroecological event happened long after the Cambrian Explosion and the Great Ordovician Biodiversification Event, the Devonian Nekton Revolution (Klug et al. 2010). It signifies the rapid occupation of the free water column with nektonic animals, an important component of all post-Silurian and Recent marine ecosystems. This event coincides with an explosive diversification of gnathostomes (jawed fishes), the initial radiation of ammonoids (extinct, externally shelled cephalopods), and the intensification of the radiation of mollusks with planktotrophic larvae or hatchlings (ammonoids, bivalves, gastropods; see also Frýda et al. 2008; Manda and Frýda 2010).
Some early ammonoids have markedly larger embryonic shells than those of younger geological periods, which might be related with differences in fecundity and reproductive strategy (House 1996; Landman et al. 1996). Drushits and Khiami (1970, p. 31) noted a possible trend toward smaller embryonic shells and initial chambers from the Devonian to the Cretaceous, while Birkelund (1981, p. 188) dismissed this because forms with smaller embryonic shells and initial chambers already appeared in the Devonian. These different interpretations might result from a poor stratigraphic resolution, insufficient data, and the poorly studied phylogenetic framework for Devonian ammonoids at that time (compare House 1996; Landman et al. 1996). No uncoiled embryonic shells of ammonoids younger than the Devonian have become known (House 1996; Landman et al. 1996), whereas forms with loosely coiled post-embryonic shells reappeared several times in the Mesozoic (Wiedmann 1973; Dietl 1978; Cecca 1997).
We herein test if there is a trend in size decrease of the umbilical window, of the embryonic shell, and/or of the initial chamber through time and phylogeny in the Devonian. Did this happen independently in several lineages? How were lineages with an open umbilicus and/or a larger embryonic shell affected by Devonian events compared with forms with a closed and/or smaller embryonic shell? Are Devonian embryonic shells larger than that of more derived Carboniferous–Cretaceous ammonoids? We also discuss possible implications and drivers of these coiling trends.
The Ammonoid Embryonic Shell
The embryonic shell (= ammonitella; Drushits and Khiami 1970) of derived ammonoids consists of an initial chamber and a subsequent whorl with a constriction at the aperture (Kulicki 1979; Landman et al. 1996; Fig. 1A). A similar constriction, the nepionic constriction, is also present in the hatchlings of Recent Nautilus (Okubo 1989; Carlson 1991; Uchiyama and Tanabe 1999). The nepionic constriction in most, if not all, ammonoids corresponds with an actual constriction in the shell wall (primary constriction), a thickening of the nacre of the shell (primary varix), and the accompanying trace or pseudoconstriction on the internal mold (varix trace). In ancestral bactritoids, the nacre does not form an abrupt swelling (Doguzhaeva 2002). Abrupt changes in ornamentation, coiling, shell shape, as well as microstructure occur at the nepionic constriction (e.g., Birkelund 1981; Landman et al. 1996). The rather uniform surface of the embryonic shell and large size of the initial chamber speaks for a direct development of ammonoids like in all extant cephalopods (Ward and Bandel 1987; Engeser 1990; Landman et al. 1996).
In many early ammonoids, the nepionic constriction hardly left any trace on the inner surface of the shell tube or internal mold (House 1996). This lack of information is probably caused by a preservational bias (internal molds), but some Devonian ammonoids might simply have lacked this structure such as bactritoids (Doguzhaeva 2002; Kröger and Mapes 2007; Kröger 2008). According to House (1996), the end of the embryonic shell in early ammonoids can be distinguished by the onset of coarse ornamentation.
The ammonoid initial chamber is typically called the protoconch in the literature (e.g., Landman et al. 1996). To avoid confusion with the gastropod protoconch, which is not homologous (Frýda et al. 2008), we will consistently refer to initial chamber in the remainder of the text as suggested by Tanabe et al. (1994a). The initial chamber of ammonoids has a globular to spindle shape, whereas the remainder of the embryonic shell is a straight or slightly coiled shaft in the earliest ammonoids or a tightly coiled whorl in derived ammonoids. Early ammonoids had a small, ovoid initial chamber just like bactritoids and some orthocerids (compare Kröger and Mapes 2007).
As the embryonic shell was completely produced within the egg and the largest part of the soft part was within the body chamber (Ward and Bandel 1987), its diameter is considered to be a good proxy of egg size (House 1985, 1996; Landman et al. 1996). This is corroborated by the similar size of ammonoid hatchlings and eggs (e.g., Etches et al. 2009). In Recent cephalopods, hatchling size not only depends on egg size, but also on developmental temperature and individual hatching conditions (Boletzky 2003). Embryonic shell measurements within and between fossil assemblages can show variations that can be caused by intraspecific variability, phylogenetic variation, ecophenotypic variation, taphonomic biases, taxonomic uncertainty (due to missing post-embryonic shell), and simple errors in measurement due to the small size (Stephen and Stanton 2002). Variation between or within phylogenetic lineages can be quite high (Rouget and Neige 2001; Tanabe et al. 2003), especially within Devonian taxa as demonstrated by Erben (1950, 1962, 1964). A factor of ecophenotypic plasticity of duration of embryogenesis as well as egg and hatchling sizes in Recent cephalopods is temperature (Boyle 1983, 1987; Boyle and Rodhouse 2005). At least during the early evolution of ammonoids, they had a tropical to subtropical distribution (Becker and Kullmann 1996; De Baets et al. 2009). At this time, the climate was warmer than today with a lower latitudinal gradient (Boucot and Gray 1982; Joachimski et al. 2009). Hence, latitudinal temperature differences were probably less important in the Devonian.
The initial chamber is closed off from the subsequent shell by the prismatic (non-nacreous) proseptum. The proseptum is perforated by the siphon and its bulbous beginning within the initial chamber, the caecum (Landman et al. 1996). It is therefore assumed by House (1985) that the embryonic shell could have already functioned as a buoyancy apparatus (neutral buoyancy) with the initial chamber as the first gas-filled chamber and with the caecum for liquid absorption and/or hydrostatic adjustment (House 1985). This could be corroborated for Late Paleozoic Goniatitida and Mesozoic Ammonitida by buoyancy calculations on actual embryonic shells, which consist of an initial chamber and subsequent whorl (body chamber) with a primary varix, but not yet nacreous septa (Shigeta 1993; Tanabe et al. 1995). The correlation between the initial chamber (first phragmocone) and the embryonic shell sizes and volumes in Paleozoic and Mesozoic ammonoids support the hypothesis that the embryonic shell consisted of a gas-filled initial chamber and a succeeding body-filled whorl segment, which could have achieved neutral buoyancy at least at hatching or slightly before or after that (Tanabe and Ohtsuka 1985; Shigeta 1993; Landman et al. 1996).
The pelagic (nektoplanktonic to planktonic) mode of life for hatchlings of more derived ammonoids is not only corroborated by buoyancy calculations within the error range of neutral buoyancy (Shigeta 1993; Tanabe et al. 1995), but also by facies analyses (House 1996; Landman et al. 1996; Mapes and Nützel 2009), by actualistic comparisons (all cephalopods with < 3–4 mm hatchling sizes are planktonic: Bandel and Boletzky 1979; Calow 1987; Nesis 1987; Boyle and Rodhouse 2005; Fig. 10) and also by the wide dispersal of some Mesozoic heteromorphic ammonoids, which were presumably poor swimmers (Ward and Bandel 1987; Landman et al. 1996). Mapes and Nützel (2009) recently suggested that bactritoids (ammonoid ancestors) had planktonic hatchlings as they occur with juvenile planktonic gastropod and ammonoid hatchlings in Carboniferous dysoxic sediments. Hatchlings of orthocerids (bactritoid ancestors) with small initial chambers were probably also pelagic (Kröger et al. 2009). Isolated ammonoid and bactritoid embryonic shells co-occur in early Emsian sediments (Erben 1965; Klug et al. 2008, K. De Baets, pers. obs.). Furthermore, early ammonoids (e.g., Anetoceratinae) are considered to be rather poor swimmers (Klug and Korn 2004), but have a global, subtropical to tropical distribution (De Baets et al. 2009). It is therefore highly likely and the most parsimonious assumption that hatchlings of early ammonoids were already planktonic, but this needs to be further corroborated by (mass) occurrences of Early Devonian ammonoid hatchlings in dysoxic sediments and better preserved embryonic shell material.
The present study is mainly based on newly measured Devonian ammonoids (Appendix S1). The evolutionary relationships between Devonian (sub-) families are based on a cladistic analysis by Korn (2000; slightly updated by Korn and Klug 2002 and De Baets et al. 2009). These relationships are congruent with the fossil record (Fig. 2). Additional measurements on embryonic shells of Carboniferous to Cretaceous ammonoids were compiled from the literature for comparison (see details in Appendix S2).
Only 3D-preserved specimens were used in this study to reduce taphonomic biases. In Devonian openly coiled and advolute early ammonoids, measurements can be directly taken on internal molds, whereas in tightly coiled forms, the embryonic shells need to be studied in the plane of symmetry or broken out of specimens as the subsequent whorls envelop the embryonic shell. All our data were collected on internal molds and are therefore directly comparable. Only such specimens were used, which could be confidently assigned at least to genus level. The embryonic shell parameters were measured using a microscope with a micrometer scale (0.01 mm), on photographs taken with a camera attached to a microscope, or with an SEM. Additional measurements derive from the photographs and figures in the literature (enlargement ≥× 5). In rare cases, the data were directly compiled from the systematic descriptions in the literature (Appendix S1). The material figured in this article is housed in the following institutional collections: American Museum of Natural History, New York (AMNH); Steinmann-Institut für Geologie, Mineralogie und Paläontologie (GPIBo); Geologisch-Paläontologisches Institut; Universität Tübingen (GPIT); Czech Geological Survey, Prague (ICh); National Museum, Prague (L); Paleontological Institute, Moscow (PIN); Landessammlung für Naturkunde Rheinland-Pfalz, Mainz (PWL); Museum für Naturkunde, Berlin (MB.C.). Additional repositories and their abbreviations can be found in the Supporting information or references therein.
Initially, we measured initial chamber length (PL) and height (PH; Fig. 3). To compare the initial chamber dimensions of early forms with more derived taxa, we used the initial chamber diameter PD commonly used in the literature (Landman et al. 1996; see Appendix S2). To quantify the shape of the initial chamber, we used the initial chamber ratio PR (= PH/PL).
The cardinal measurements of the embryonic shell are:
1Maximum embryonic shell diameter AD (= ammonitella diameter of Landman et al. 1996): the distance from the ventral adoral end of the nepionic constriction (taken to be the ventral edge of the embryonic shell in internal molds) through the center of the initial chamber to the opposite side of the embryonic shell (Fig. 4).
2Length umbilical window length (UWL) and height umbilical window height (UWH) of the umbilical window: the most readily available morphometric information that can be obtained from both sections and internal molds (Fig. 3). To quantify the shape of the umbilical window, we used the umbilical window ratio UWR (= UWH/UWL). Note that closed umbilici were entered as equidimensional (UWR = 1).
3Initial chamber diameter (PD) and the embryonic shell angle (AA; ammonitella angle of Landman et al. 1996): measured in some specimens, where the edge of the embryonic shell (nepionic constriction) is known; we used it to compare it with literature data (Landman et al. 1996; Fig. 4). In loosely coiled embryonic shell, the shell angle is dependent on both size of the initial chamber/embryonic shell coil and also coiling, that is, the amount of whorls it completes. The angle between the initial chamber aperture and the minimum whorl height at the end of the embryonic shell (Amonitella Whorl Angle, AWA) might be a more appropriate measure for the number of whorls completed and would be independent of size of the embryonic shell shaft/whorl as well as initial chamber size and shape as opposed to the AA. In bactritoids, AWA is zero, whereas the AA is not zero and dependant on the length of the embryonic shell. However, the AWA has not been used in the literature and to facilitate comparisons, we herein used the standard parameter (AA).
The scanned and digitized images of initial chambers, umbilical windows, and adult whorl cross sections were, for a precise outline tracing, aligned and filled with black. Their areas (PA, UWA, and K) were calculated using TpsDig software developed by Rohlf (http://life.bio.sunysb.edu/morph/). The edge of the initial chamber is herein approximated with a direct line connecting two homologous points, which correspond with the dorsal and ventral edge of the aperture of the initial chamber in internal molds and the proseptum in cross section. These edges are easily recognizable on internal molds and in cross sections, whereas the trace of the proseptum cannot always be seen and differs between that seen on the internal mold and in cross section.
Adult Body Chamber Volume Calculation
Adult ammonoids can be recognized by the presence of mature shell modifications at the end of their ontogeny (Bucher et al. 1996). These modifications include septal approximation and changes in coiling, whorl cross section, and ornamentation (Davis et al. 1996; Appendix S4). We herein use the adult body chamber volume as a measure for adult size of ammonoids as it better reflects the volume of the soft body than the shell diameter, especially when comparing forms with very different conch geometries (Bucher et al. 1996; Dommergues et al. 2002). The adult body chamber volume was herein calculated using the formulas suggested by Moseley (1838) as modified by Raup and Chamberlain (1967):
with K is the area of the last aperture (calculated with TpsDig: see above), Ra is the distance from the coiling axis to the centre of gravity, θ is the angular length of the body chamber in radians and
EGG VOLUME AND NUMBER ESTIMATES
Minimum egg volume at hatching was estimated using the formula for the volume of a prolate spheroid with the maximum embryonic shell diameter AD as long axis and the dimension of the embryonic shell orthogonally on it, AD2 (Fig. 4), as short axis:
Note that this is a rough estimate as the soft part volume outside the embryonic shell or the yolk sac (compare Boletzky 1987) is not taken into account because they are unknown.
We estimated the number of eggs in ammonoid females using the approximation that 8% of the adult body chamber volume is filled with gonads (compare Klug 2001b, 2007; Korn and Klug 2007). This is a simplification as the gonad to body chamber-volume-ratio is unknown in fossil cephalopods and rarely measured in Recent cephalopods (mostly, the gonad to total weight-ratio = gonado-somatic index is given: see Boyle 1983; Boyle and Rodhouse 1985). Things are further complicated by unknown packing of the eggs in the gonads and allometric relationships between the gonado-somatic index and size. The maximum gonado-somatic index for Nautilus is about 8% (Tanabe and Tsukahara 1987).
Even if the gonad volume (∼8%) would apply to ammonoids, this may result in underestimates as most eggs and embryos of Recent cephalopods grow after egg deposition (Boyle 1983; Boletzky 1987; Boyle and Rodhouse 2005 and references within).
EMBRYONIC SHELL COILING AND UMBILICAL WINDOW CLOSURE IN PHYLOGENY
All early Emsian ammonoids with a known embryonic shell had an umbilical window (Fig. 5; see also Appendix S1). In early Emsian taxa, where the extent of the embryonic shell is known, the umbilical window is partially surrounded by the embryonic shell coil and partially by the post-embryonic whorl (Fig. 1). The most basal taxa such as Metabactrites, Ivoites, and Erbenoceras have an egg-shaped initial chamber, an almost straight embryonic shell shaft, and loosely coiled post-embryonic whorls; the umbilical window is very large and not yet enclosed (the initial chamber does not touch the subsequent whorl). Early representatives of other slightly more derived lineages such as Gyroceratites heinrichi (Mimoceratidae), Irdanites kaufmanni/Convoluticeras erbeni (Paleogoniatitinae), and Teicherticeras primigenitum (Teicherticeratinae) also have an open umbilical window. In more derived taxa of these lineages, the initial chamber touches the subsequent whorl (enclosed umbilical window), although they occasionally still have a secondary “opening” between the embryonic shell and the subsequent whorl due to the less-curved embryonic shell (Fig. 3). All known embryonic shells of the Mimosphinctidae, Mimoceratidae, and Teicherticeratidae have an embryonic shell coil, which completes less than one-half whorl (Fig. 8; compare Erben 1964; Bogoslovsky 1969; House 1996). In more derived taxa such as Mimagoniatites and Archanarcestes (Mimagoniatitidae) as well as Anarcestes and Sellanarcestes (Anarcestidae), the embryonic shell coil is about one-half whorl, but the umbilical window is still partially surrounded by the post-embryonic whorls. In more derived taxa, the umbilical window becomes fully enclosed by the embryonic shell coil and eventually disappears.
Late Emsian taxa with an open umbilical window are not known, they all have enclosed umbilical windows (Fig. 5). Two late Emsian genera (Anarcestes and Mimagoniatites) containing several species with an enclosed umbilical window demonstrate that at least two families with an opening in the first whorl crossed the Emsian/Eifelian boundary, but disappeared at the Choteč Event (Fig. 5; compare Klug 2002; Monnet et al. 2011b). This means that all Eifelian taxa with an (enclosed) umbilical window (Fig. 5) are boundary crossers and also the youngest taxa with an umbilical window (Fig. 2). Reports of “Anarcestes” above the Choteč Event (Becker and House 1994) are questionable and an open umbilicus still needs to be demonstrated in these forms. The first taxa without an umbilical window already appeared in the latest Emsian (Fig. 5). The umbilical window is independently closed in the Agoniatitidae and Latanarcestidae of the suborder Agoniatitina (from Achguigites and Latanarcestes/Mimanarcestes with an open umbilicus to Paraphyllites and Chlupacites without an umbilical window) and Anarcestidae of the suborder Anarcestina (from Praewerneroceras with an open umbilicus to Werneroceras without an umbilical window) more or less at the same time in the late Emsian before the Choteč Event (Fig. 2). The taxa of these families with an (enclosed) umbilical window disappear long before the Choteč Event in the latest Emsian. Only representatives without an opening within the first whorl of two of these lineages (Agoniatitidae and Werneroceratidae) survived the Choteč Event and give rise to all subsequently known ammonoid families.
Forms with an umbilical window and an openly coiled embryonic shell disappeared more rapidly than among gastropods (about 20 Ma as opposed to ∼250 Ma in gastropods after they start coiling: Nützel and Frýda 2003). The trend toward increased coiling of the embryonic shell continued after the closure of the umbilical window as derived ammonoids became even more involutely coiled (Bogoslovsky 1969; Landman et al. 1996). This is hard to quantify as in the literature mostly measurements or figures of median sections (through the plane of symmetry) are available (no information on whorl overlap).
In the following, we focus on size changes of the umbilical window through phylogeny and time. In the boxplots, we see that length, height, and area of the umbilical window are reduced through the time-ordered phylogenetic sequence within the Agoniatitina and Anarcestina (Fig. 6A–D, I–K). The largest umbilical windows are known from the Anetoceratinae (Fig. 6E–H). If we remove them from the boxplots, we also see the subsequent reduction in umbilical window size more clearly (Fig. 6I–K). All sampled Agoniatitidae have a closed umbilical window. The trend is also visible when following the centroid of monophyletic groups through the Agoniatitina and Anarcestina (Fig. 7, top left).
No clear trend is evident in the umbilical window ratio (Fig. 6). A reduction of the umbilical window is not only observed between orders (from Agoniatitina to Gephuroceratina, from Agoniatitida via Anarcestina to Pharciceratina: Fig. 2) or between monophyletic groups within orders. An independent reduction of the umbilical window is seen between subfamilies (e.g., Anetoceratinae to Mimosphinctidae in Fig. 7, bottom left) and genera within a subfamily (e.g., the Anetoceratinae, Fig. 7, bottom left). A trend could even be evidenced between species of a single genus (e.g., the anagenetic Gyroceratites lineage, Fig. 7, bottom right). A reduction of the umbilical window size through time is also evident when grouping all measurements according to stratigraphic range (Fig. 7, top right). Forms, which are known from both the early and the late Emsian fall within the umbilical window size range of early Emsian forms and forms known from the late Emsian and Eifelian within the size range of late Emsian forms (Fig. 7, top right).
CHANGES IN INITIAL CHAMBER SIZE AND SHAPE
No clear trend in initial chamber size (length or height) or area could be distinguished in the Early to Middle Devonian ammonoids (Fig. 6A–D). Some Early (Mimosphinctes) and Middle Devonian ammonoids (Agoniatites) show the largest initial chamber length and height known in the entire history of ammonoids (compare Appendices S1 and S2; Fig. S1). In the case of the initial chamber area, this might also be partially related with the simplification of connecting the lower and upper edge of the initial chamber with a straight line (thus abstracting the shape of the proseptum). This is unavoidable due to our somewhat heterogeneous dataset (cross sections and internal molds) and the sometimes poor preservation of the proseptum.
There is a clear trend in initial chamber length to height ratio (Fig. 6, uppermost row). This change is related with the shape change of the initial chamber from a symmetrical, elongated initial chamber (U-shaped in cross section) in Devonian bactritoids and the earliest ammonoids to spindle-shaped (spiral in cross section) in derived ammonoids (Fig. 1). The increased coiling of the initial chamber (from ovoid-uncoiled over globular-coiled to spindle-shaped-coiled) is also reflected in the angles between lirae (ornamentation) on the initial chamber (Klofak et al. 1999, 2007; Klofak and Landman 2010). The shape change of the initial chamber contributed to closing the umbilical window in addition to the increased coiling of the first whorl (including both embryonic shell and the first part of the post-embryonic shell). Interestingly, our data indicate that first the post-embryonic shells coiled, then the embryonic shell, and then also the initial chamber started coiling in early ammonoids (compare Figs. 5, 6, and S1).
CHANGES IN EMBRYONIC SHELL SIZE
In the Devonian ammonoids, a richer dataset is now also available for embryonic shells (Appendix S1; compare Erben 1964; Bogoslovsky 1969; Landman et al. 1996; Klofak et al. 1999, 2007; Klofak and Landman 2010). Our data show that the largest embryonic shells (up to 4 mm) are known from Devonian ammonoids, namely in plesiomorphic Early Devonian ammonoids (e.g., Mimosphinctidae). The embryonic shell sizes of Erbenoceras and Mimosphinctes represent the two outliers among the dimensions of Early Devonian embryonic shells (Fig. 9). There is a trend toward smaller embryonic shells (<1.5 mm) in several lineages during the Early and Middle Devonian (Figs. 8–10). This is largely related with the increased coiling of the embryonic shell as reflected by changes in the embryonic shell angle (Fig. 9, lowermost row). There is a subsequent steady increase in embryonic shell angle from the Early Devonian to the Carboniferous followed by a slight decrease toward the Cretaceous, but values of the Early Devonian have never been reached thereafter. An exception to the trend toward smaller embryonic shells is the lineage leading to Agoniatites, which has the largest known initial chamber of all ammonoids (compare House 1985, 1996). By contrast, the main trend in nautiloid evolution might have been the increase in size of the embryonic shell (Dzik 1981, p. 162, 1984), possibly also in the lineage leading up to Recent Allonautilus and Nautilus (Mapes et al. 2007).
Embryonic shell diameter values of derived Mesozoic ammonoids mostly fall between 0.5 and 1.5 mm (Fig. 9; compare Landman et al. 1996); although there are some exceptions (see Appendix S2). Several of the Mesozoic ammonoids (e.g., Jurassic Oppeliidae: Palframan 1966, 1967; Neige 1997 or Cretaceous Scaphites: Landman 1985), which possess the smallest embryonic shells, also show clear evidence for dimorphism (Davis et al. 1996). The latter points toward the importance of embryonic shell size in reproductive strategy. The lower limits of embryonic shell diameter (a proxy for egg and hatchling size) and initial chamber size (∼0.5 and 0.3 mm, respectively) might point to certain constructional and physiological constraints (cf. Vance 1973; Boletzky 1993, 1997). Yolky eggs of a few hundred microns are considered to be a minimum size from which an actively jetting cephalopod can be formed (Boletzky 1993).
CHANGES IN ADULT SIZE
Some of the more derived Devonian ammonoids have markedly larger adult sizes (body chamber volumes) than their early relatives (e.g., Erbenoceras, Mimosphinctes, and Gyroceratites). This is evident when comparing diameter and body chamber volume for some taxa where size at maturity is well established due to adult modifications (Fig. 11; compare Chlupáč and Turek 1983; Klug 2001a; Korn and Klug 2002, 2003; De Baets et al. in press; Appendix S4). Ammonoids could increase the volume of the body chamber by increasing the diameter, body chamber length, involution, or lateral width (Guex 2003). Trends toward increased diameter, coiling (involution), body chamber length and lateral width have been demonstrated for various ammonoid lineages during the Devonian (Fig. 11; Korn and Klug 2003; Klug and Korn 2004; Monnet et al. 2011b). Our data also illustrate the importance of using body chamber volume as a size measure instead of diameter: for example, Mimagoniatites fecundus has a slightly smaller diameter than Erbenoceras advolvens but a three times larger body chamber volume (Fig. 11).
Several theories have been put forward about the implications and causes of progressive coiling of early ammonoids and shelled mollusks in general (House 1985, 1996; Vermeij 1987; Nützel and Frýda 2003; Kröger 2005; Klug 2007; Frýda et al. 2008; Klug et al. 2010). They involve changes in reproductive strategies, mode of life, and predatory pressure. In many cases (e.g., ammonoids and gastropods; compare House 1996; Fryda and Nützel 2003; Fryda et al. 2008), it is unclear if the loss of openly coiled inner whorls could have occurred by the extinction of lineages with forms having an umbilical window or by the gradual closure of the window, through evolution of ontogeny. We will now briefly discuss how progressive coiling of the embryonic shell might have changed the reproductive strategy in early ammonoids and possible drivers of the progressive coiling of the embryonic shell of early ammonoids.
CHANGES IN REPRODUCTIVE STRATEGY
In Recent cephalopods, there is a clear trade-off between egg size/volume and fecundity or number of offspring (Fig. 10; compare Calow 1987; Boyle and Rodhouse 2005). The progressive coiling and reduction in size of the embryonic shell within the early evolution of ammonoids might therefore have partially involved a change in reproductive strategy as suggested by House (1996), especially in forms with larger adult sizes. This is not only supported by actualistic comparisons, but also by our egg number estimates. Early ammonoids (e.g., Mimosphinctidae) had markedly larger embryonic shells and therefore a lower fecundity (35–135) than more derived ammonoids (e.g., Mimagoniatitidae, Agoniatitidae, and Gephuroceratidae) with estimates between 3400 and 220,000 eggs, which falls in the range of pelagic coleoids (Fig. 10).
However, the trade-off between the amount of eggs per female and egg size is not a perfect one (Boletzky 1987, 1997; Calow 1987; Boyle and Rodhouse 2005). The size of the adult or more precisely the egg to adult size ratio also needs to be considered. In the loosely coiled E. advolvens with a similar adult shell diameter as in M. fecundus, the body chamber volume is considerably smaller, thus offering less space for gonads and eggs. The importance of the adult body size is also indicated by Gyroceratites, which has a rather small embryonic shell (∼1.5 mm) and adult body size yielding a lower egg number estimate (135) than Erbenoceras (Mimosphinctidae). The embryonic shell size remains an important factor influencing fecundity, however, even in forms with large body chamber volumes such as Agoniatites. This taxon has a seven times higher body chamber volume, but also a larger embryonic shell (2.3 mm) than Mimagoniatites, which results in only 1.2 times higher fecundity estimates compared with the mimagoniatitid. More derived ammonoids with larger body chamber volumes and smaller embryonic shells in the Late Devonian might have had markedly larger fecundities (egg estimates up to 220,000 for Manticoceras). This is further corroborated by indications for mass spawning events (Korn and Klug 2007; Walton et al. 2010) and pronounced dimorphism (Makowski 1962, 1991) in Late Devonian derived ammonoids, including Manticoceras.
Bactritoids are the immediate ancestors of ammonoids. Their embryonic shells are large and they had narrow body chambers (Mapes 1979; Doguzhaeva 2002; Mapes and Nützel 2009), which probably indicate a lower number of offspring per female than in ammonoids. Their reproductive strategy might have been between that of nautiloids and ammonoids (Mapes and Nützel 2009). Even if some bactritoids and ammonoids produced floating egg masses as suggested by Mapes and Nützel (2009), the shape and size of the gonads (and fecundity) must have been constrained by the volume and form of the body chamber.
The decrease in embryonic shell size and the increase of adult size (body chamber volume) resulted in higher fecundities in some derived ammonoid lineages and augmented the upper limit of fecundity in ammonoids. However, ammonoids have a broad range in shell shape and size, also in the Devonian (Korn and Klug 2003), so that a wide variation in reproductive strategies and number of offspring is plausible as in modern coleoids (Landman et al. 1996). This is further corroborated by varying degrees of dimorphism and different adult modifications in various groups through time (Davis et al. 1996). Among Devonian ammonoids (Fig. 11), egg numbers per female might have varied between 35 and 220,000 at egg sizes between 1 and 4 mm. Nevertheless, even smaller forms such as Prolobites (diameter < 40 mm) might have had a higher fecundity than Nautilus (Walton et al. 2010).
DRIVING FACTORS IN DISSAPPEARENCE OF LOOSELY COILED EMBRYONIC SHELLS
In ammonoids, a loosely coiled embryonic shell and an umbilical window both represent the plesiomorphic condition as they derived from bactritoid ancestors with more or less straight conical shells (Korn 2000; Klug and Korn 2004). Trends toward a smaller umbilical window and more closely coiled embryonic shells are documented throughout the Agoniatitida, between and within suborder, (sub-) families, as well as within genera. The umbilical window is independently lost in at least three lineages, while other lineages with (larger) umbilical windows became selectively extinct around some Devonian bio-events, while lineages with a closed umbilicus survived these events (Fig. 2). All Middle Devonian ammonoid taxa derived from an ancestor with a closed umbilical window, which can be traced back to two families. This indicates that both driven (i.e., iterative evolution) and sorting trends due to differential extinction of clades played a role in the disappearance of the umbilical window in ammonoids. Both mechanisms have also been suggested to explain trends toward smaller, more tightly coiled gastropod juvenile shells (Nützel and Frýda 2003; Wagner and Erwin 2006; Frýda et al. 2008). The simultaneous increase in coiling of the inner whorl and disappearance of the umbilical window in several ammonoid lineages speaks for a common, selective factor, or other external factors operating on them (Nützel and Frýda 2003; Frýda et al. 2008; Monnet et al. 2011a).
The trend toward increased coiling of the embryonic shell and closure of the umbilicus might be a response to increased predatory pressure as has been suggested for various ontogenetic stages in mollusks groups occupying different habitats (Vermeij 1987; Nützel and Frýda 2003; Kröger 2005). This fits quite well with the presence and diversification of various predatory groups in the Devonian, which were capable of preying on ammonoid hatchlings, which include phyllocarids, conodonts, gnathostomes, ammonoids, and other cephalopods (Brett and Walker 2002; Kröger 2005; Berkyová et al. 2007; Klug et al. 2008). This would be in line with the evidence of predation on Recent pelagic cephalopod and gastropod juveniles by gnathostomes, cephalopods, arthropods, and various other pelagic invertebrates and their young (Nixon 1987; Vecchione 1987; Hickman 2001; Ibáñez and Keyl 2010). Shells with a closed umbilical window have several advantages over a shell with an open umbilicus or umbilical window: loosely coiled shells are less sturdy and mechanically weaker compared with coiled shells with a closed umbilicus and can be more easily crushed by predators and the animal takes up less space with the same volume (Vermeij 1987; Brett and Walker 2002; Nützel and Frýda 2003; Wagner and Erwin 2006; Klug 2007).
As the ammonoid shell is used as a buoyancy apparatus starting already in the hatchlings (or even earlier), the changes in coiling affected the orientation of the aperture (from a downward to upward position), maneuverability, and streamlining in the water column just like in postembryonic stages (Westermann 1996; Westermann and Tsujita 1998; Klug and Korn 2004; Kröger 2005). This might have had certain advantages in escaping predators, but also in food gathering. Young ammonoid hatchlings already had a hyponome and were capable of active locomotion by jetting water from their hyponome as indicated by a bend in the growth lines (hyponomic sinus) in Devonian ammonoids directly after hatching (Erben 1964; House 1965). The smaller, more coiled hatchlings without an umbilical window (∼1 mm) might have a less unstable locomotion (cf. Kröger 2005) and their more spherical shape might have reduced the surface area and friction at their small size (Jacobs and Chamberlain 1996; Klug 2007). However, shape might play a subordinate role in embryonic shells of ammonoids as smaller forms have less power relative to drag than in larger forms, which would have made the effect on the absolute swimming speed minimal. Ammonoid hatchlings were probably unable to swim against currents (Jacobs 1992; Jacobs and Chamberlain 1996).
Evolutionary effects of predatory pressure on young ammonoids are hard to test in Devonian ammonoids at the moment as direct evidence for predation on ammonoid hatchlings like in dacryoconarids or planktonic gastropods (Hickman 1999, 2001; Berkyová et al. 2007) is lacking. The absence of direct evidence in the Devonian could be caused by (1) the scarcity of hatchlings with a preserved shell, (2) difficulties to detect injuries on smaller specimens, and (3) the more or less complete destruction of the shell by the predator (Slotta et al. 2011). Furthermore, no difference could be detected in sublethal injuries in post-embryonic conchs in similar preservation between loosely coiled ammonoids and bactritoids from the same localities in the early Emsian (Klug 2007). Evidence for predation is also scarce for post-Devonian ammonoid juveniles. It consists of ammonoid stomach contents (Nixon 1996) and coprolites (Tanabe et al. 2008) in addition to sublethal injuries as in Nautilus (Ward and Bandel 1987).
Similar coiling trends or patterns could also be generated by architectural or developmental constraints (Morita 1993, 2003; Wagner and Erwin 2006; Futuyma 2009), although this mechanism alone would not explain the synchronicity of these events in various ammonoid lineages and other mollusk groups during the Devonian. Such constraints probably must have played a role, but it is at the moment impossible to test to what extent and how they controlled the evolutionary trend toward tightly coiled embryonic shells because the soft parts are hardly known and little is known about the morphogenesis of mollusks (e.g., Urdy et al. 2010a,b).
The simultaneous disappearance of several lineages with larger and/or more openly coiled embryonic shells during Devonian events speaks for a similar external factor affecting them related with environmental changes. These events are linked with regression/transgression couplets, anoxia (black shales), and algal blooms, which indicate changes in nutrient availability (House 2002; Berkyová et al. 2009). The anoxia, which is commonly invoked to explain changes in diversity of benthic organism around these times, probably did not affect the young stages of ammonoids as no indication exists for benthic egg masses and the hatchlings were most likely pelagic (probably planktonic). As taxa with less coiled and/or larger embryonic shells preferentially became extinct at the Daleje, Choteč, and Kačák Events, not only the ammonoid hatchling size and coiling, but also fecundity might have played an important role during these events. So far, only the extinction of all ammonoids (small embryos: Landman et al. 1996) at the end of the Cretaceous and survival of nautilids (large embryos: Wani et al. 2011) have been related with differences in reproductive strategies of these two groups (Ward and Bandel 1987; Kennedy 1989; Tanabe 2011).
In the course of the transition from bactritoids to ammonoids as well as the early evolution of ammonoids, their embryonic shells changed in coiling and size. This happened independently, convergently, and simultaneously in several lineages and even in other mollusk groups. Such a synchronicity can be considered as strong evidence for common driving factors, which could have been related with various ecological factors. Changes in reproductive strategy were one of the effects, which is supported by the decrease in embryo (and thus egg) volume and the increase in body chamber volume in some lineages. This embryo size decrease is achieved first by an increase in coiling and then by a reduction in initial chamber size as well as initial chamber shape. To which degree, predatory pressure and/or competition for planktonic food sources of planktonic juveniles and larvae played a role is currently not possible to test, but both might have played a role as indicated by diversification of various predators in the Devonian and changing planktonic communities. A more tightly coiled shell changed the orientation of hatchlings in the water column as well as improved their chances to survive predation, their maneuverability, and their mobility, although they still were too small to swim effectively against currents. Loosely coiled embryonic shells with an umbilical window got lost during progressive coiling of the embryonic shell within lineages and preferential extinction of lineages with exclusively loosely coiled embryonic whorls. This indicates that both iterative evolution and selective extinction (sorting) were involved in the disappearance of loosely coiled inner whorls. The preferential extinction during events related with important ecological changes suggests the possible importance of embryonic shell coiling, hatchling size, and/or reproductive strategies for survival during such events.
Associate Editor: R. Dudley
This study was funded by the Swiss National Science Foundation (Project numbers 200021–113956/1 and 200020–25029). The Moroccan colleagues of the Ministère de l’Energie et des Mines (Rabat and Midelt) and the Uzbek colleagues from the State Committee on Geology and Mineral Resources of the Republic of Uzbekistan (Tashkent) and the Kitab State Geological Reserve benignly provided permits for the field work and for the export of samples. We greatly appreciate the patient help of N. A. Meshankina, F. A. Salimova (Tashkent), U. D. Rakhmonov (Kitab) during our field work in the Kitab State Geological Reserve (Uzbekistan). S. Klofak (New York) made some SEM photographs of Gyroceratites specimens available to us. V. Turek (Prague) kindly produced a photograph of the inner whorls of one of the Prague specimens. S. Urdy (Zürich) made us aware of the MATLAB function Binofit. C. Monnet (Lille) introduced R to KDB. V. Ebbighausen (Odenthal) provided pictures of specimens in his possession. K. Bylund (Spanish Fork) kindly proofread the manuscript for us. We are greatful to G. Vermeij (Davis), K. Tanabe (Tokyo), the associate editor R. Dudley, and an anonymous reviewer for valuable comments and suggestions on previous versions of this manuscript.