A review of fossil evidence supports a pelagic mode of life (in the water column) of ammonoids, but they may have spent their life close to the seabottom (demersal), planktonically, or nektonically depending upon the ontogenetic stage and taxon. There are good indications for a planktonic mode of life of ammonoid hatchlings, but a broad range of reproductive strategies might have existed (egg-laying, fecundity). Isotope and biogeographical studies indicate that some forms migrated or swam for considerable distances, whereas others may have been primarily transported by oceanic currents during early and/or late ontogeny. Diverse ammonoid habitats are also supported by evidence from predator–prey relationships derived from characteristic injuries and exceptional fossil finds, which indicate chiefly predatory or scavenging lifestyles. Sublethal injuries preserved in some ammonoid shells, as well as rare stomach and coprolite contents, provide evidence of predation by other cephalopods, arthropods and various jawed vertebrates. Various lines of evidence suggest that different groups of ammonoids had quite different ecologies, but shell shape alone can only give upper constraints on ammonoid capabilities, a matter that needs to be considered when interpreting their diversity and evolutionary history.
Ammonoids, a now extinct group of externally shelled cephalopods, had a successful ecological and evolutionary history for over 300 million years, surviving at least three major mass extinction events (House, 1989). These molluscs are palaeontology's model organisms for the study of biostratigraphy, diversity, biogeography and macroevolution due to their wide geographical distribution, high diversity and disparity, and high evolutionary rates (Kennedy & Cobban, 1976; Landman, Tanabe & Davis, 1996; Monnet, De Baets & Klug, 2011). Although preservation of the accretionary shells of ammonoids is nearly ubiquitous, preservation of soft-part anatomy is very rare and often cryptic when present (Klug, Riegraf & Lehmann, 2012). As a result, reconstructions of their appearance are often conjectural and work upon their ecology has been based primarily upon study of the morphologies and facies distribution of their shells (e.g. Ebel, 1992; Westermann, 1996). Classically, ammonoids have been compared with extant nautilids, which is also reflected in historical reconstructions (e.g. Fraas, 1910). However, it is now widely accepted that ammonoids are more closely related to extant coleoids (Engeser, 1996) than to Nautilida; the former thus presents an adequate model for some aspects of ammonoid ecology and appearance (Jacobs & Landman, 1993; Warnke & Keupp, 2005). Nevertheless, coleoids have internalized, and largely reduced shells, leaving Nautilus as the only living analogue to constrain the accretionary growth and function of the ammonoid shell. Here, we review how recent investigations and syntheses, aided by advanced technology and exceptionally well-preserved specimens, can advance earlier models (e.g. Wiedmann, 1973; Batt, 1993; Westermann, 1996) of ammonoid life mode and ecology.
The ammonoid animal in phylogenetic context
Morphological details of ammonoid shells and soft parts reveal their anatomy and confirm the ammonoid's phylogenetic position within the cephalopods. All ammonoid shells have a chambered part (phragmocone) and a final open chamber (body chamber), which houses the soft body of the animal (Fig. 1). Most post-embryonic ammonoid shells are coiled planispirally, but some ‘heteromorph’ taxa have trochospiral, uncoiled or even straight shells. Ammonoid buoyancy was probably controlled not only by shell and soft-tissue growth but also by manipulation of gas and fluid transport through the siphuncle, a tube of ‘connecting rings’ between the phragmocone chambers (Tanabe et al., 2000). Ammonoids had a functional phragmocone similar to the Nautilida, where the mantle secreted the calcium carbonate (aragonite) shell with the septa subdividing the phragmocone into chambers as well as its proteinaceous coating (periostracum; cf. Mutvei, 1964). Additionally, the mantle performed shell repairs evident on ammonoid body chambers (Kröger, 2002). The other living cephalopods, the coleoids (octopods, cuttlefish and squids), are very different from Nautilus, most obviously in the internalization and reduction of the shell. Similar to most extant cephalopods, the mouthparts of ammonoids consist of jaws and a radula, a minutely toothed, chitinous ribbon used for feeding in molluscs (Nixon, 1996). The ammonoid radula is similar to that of coleoids (in size, shape and the amount of elements per row), but differs from the radula of more basal externally shelled cephalopods and Nautilus (Nixon, 1996; Gabbott, 1999). The close phylogenetic relationship between ammonoids and coleoids (Engeser, 1996; Kröger, Vinther & Fuchs, 2011) is further supported by a transitional series from straight orthoceratid nautiloids, over straight to slightly curved bactritoids (which also gave rise to coleoids) to coiled ammonoids (Erben, 1966; see Fig. 2), which was recently stratigraphically corroborated (Kröger & Mapes, 2007; De Baets et al., 2013). All these closely related extinct groups share similar small, initial chambers of their embryonic shells with the extant coleoid Spirula (Warnke & Keupp, 2005). Developmental, neontological, fossil and genetic studies (Shigeno et al., 2008; Sasaki, Shigeno & Tanabe, 2010; Wani, Kurihara & Ayyasami, 2011; Ogura et al., 2013) indicate that the unique Nautilus features (90 arms, pinhole eyes, large yolk-rich eggs) developed in their lineage separately from the other extant cephalopods (probably after they split from the Orthocerida; Kröger et al., 2011).
The timing of evolutionary divergence of extant nautiloids and the ancestors of coleoids has often been traced back to at least 480 Ma (origin of the Oncocerida; Kröger, Zhang & Isakar, 2009) using the fossil record, but the deep ancestry of extant nautiloids is still debated (Dzik & Korn, 1992; Turek, 2008) and their divergence from extant coleoids has been estimated by the most recent molecular clock analyses to be near to the Silurian/Devonian boundary (∼416 ± 60 Ma; Kröger et al., 2011 and references therein).
Constraints on ammonoid anatomy come from exceptionally preserved specimens and from phylogenetic inference. The rare preservation of soft tissues is often interpreted as a consequence of delays in the pelagic animals' sinking (De Baets et al., 2013) and burial (Wani et al., 2005; Wani, 2007). Still, extraordinarily preserved specimens reveal digestive tracts, gills, questionable eye capsules and details of the buccal apparatus (e.g. Klug et al., 2012). Computer tomography has revealed new morphological details about the ammonoid mouth parts (e.g. Kruta et al., 2011, 2013; Klug & Jerjen, 2012). Ammonoids might have had a lens eye (if the development of the pinhole eye was specific for the nautiloids; Ogura et al., 2013) and the arguable presence of a hood in ammonoids has been deemed unlikely by Keupp (2000) based upon the findings of dorsally tongue-shaped black layers. It is most parsimonious to assume the presence of 10 arms in ammonoids due to the presence of 10 arm buds in the embryonic development of all extant cephalopods (Shigeno et al., 2008). However, both the amount and the appearance of the adult arms might have varied from taxon to taxon, as found in extant coleoids (Westermann, 1996). Based upon the lack of preserved arms in the fossil record and constraints imposed by narrow aperture and mouth parts, Landman, Cobban & Larson (2012) suggested very short arms, whereas Keupp & Riedel (2010) have speculated on very short or delicate arms spanning a mucous net in order to feed on plankton in some derived Jurassic and Cretaceous ammonoids.
Muscle attachment scars are more commonly preserved in fossil ammonoid shells, which reveal how muscles were attached to the shell. A pair of lateral retractor muscles is the most conspicuous structure (e.g. Doguzhaeva & Mutvei, 1996; Richter, 2002), which probably served for jet propulsion and the retraction of the head, the arm crown and the hyponome; sometimes, the portions of the muscle insertion responsible for the hyponome and for the cephalic part can be distinguished. At least some of these muscles were already present at the ammonitella stage; the dorsal muscle scars are commonly long in Triassic and more rounded in Jurassic ammonoids (Keupp, 2000). An additional pair of muscle scars at the umbilical edge is present in Mesozoic ammonoids. A pair of ventrolateral muscles was reported for Aconeceras, which significantly change in shape and position during ontogeny and between species (Doguzhaeva & Mutvei, 1996). Two tongue-shaped ventral muscles were probably connected with the funnel and therefore responsible for the adjustment of the swimming direction (Keupp, 2000).
The great disparity in ammonoid shell shape has prompted studies for environmental associations to test interpretations correlating differing shell morphologies with different life modes (e.g. Batt, 1993; Westermann, 1996). Ammonoid conch geometry has been used to calculate flow resistance and swimming velocities (Jacobs & Chamberlain, 1996), septal strength and maximal diving depths (Hewitt, 1996), the positions of the centres of gravity and buoyancy (e.g. Hammer & Bucher, 2006), and the orientation of the shell in the water column (Saunders & Shapiro, 1986; Klug & Korn, 2004; see Fig. 2). Such models and calculations can only provide maximum constraints on ammonoid capabilities and are based upon several simplifications and assumptions. Flawed estimates of negative buoyancy and erroneous assumptions about ammonoid growth and soft tissues have led to some extreme interpretations that ammonoids were epibenthic crawlers (e.g. Ebel, 1992). However, it is widely accepted that calculations are in the range of neutral buoyancy (within the limit of estimate) when making reasonable assumptions (Westermann, 1996; Kröger, 2001). Furthermore, novel computer tomographic methods allow for more precise volume determinations of cephalopod shells, which can be used to further refine buoyancy calculations of extinct cephalopods (e.g., Hoffmann et al. 2013). There is also support from modelling results that even heteromorphs with the most aberrant shell morphologies were living in the water column (Okamoto, 1996). Facies distribution and hydrodynamic considerations have led to seemingly detailed hypotheses on ammonoid ecology (Westermann, 1996), with coiled ammonoids being considered as nektonic, demersal and planktonic depending upon their shell shape, orientation and streamlining, whereas most heteromorphic ammonoids have been interpreted as planktonic drifters or vertical migrants. Such models must be more widely tested in time and space with different types of data. The next section will review some recent investigations of associations between ammonoid shell shape and habitat (facies), biogeography and diversification events.
Spatiotemporal associations of ammonoid shell shape and environmental phenomena
Investigations of spatiotemporal and evolutionary trends in ammonoid shell shape generally focus upon bivariate plots of geometric parameters (e.g. Klug & Korn, 2004; Yacobucci, 2004) or the combination of many parameters in ordinated spaces (e.g. Dommergues, Laurin & Meister, 2001; Klug et al., 2005; Saunders et al., 2008; Dera et al., 2010; Korn & Klug, 2012; Brosse et al., 2013). Ordinations detect subtle or complex shape changes within a specific collection of shells (Yacobucci, 2004) or assess the extent of variation within a collection (Brosse et al., 2013), but the results from these studies cannot be directly compared visually or statistically (see McGowan, 2004 for discussion). Particularly relevant to the present discussion are recent approaches that allow direct comparison between studies and focus upon parameters that could influence shell hydrodynamics. Monnet et al. (2012) tested for increasing hydrodynamic streamlining within the long ranging ammonoid family Acrochordiceratidae by comparing involution, whorl compression and ornamentation (ribbing density) in a ternary diagram. Ritterbush & Bottjer (2012) introduced the ‘Westermann's morphospace method’ – juxtaposing umbilical exposure, whorl expansion and overall inflation to compare coiled ammonoid shells from different studies (Fig. 3) to common and hydrodynamically distinctive end-member shapes according to the hypotheses of Westermann (1996). The possible influence of ornamentation and body chamber length (associated with shell orientation; Saunders & Shapiro, 1986; Klug & Korn, 2004) is not directly considered in this approach.
Useful case studies apply these methods to test associations between ammonoid shell shape and environmental characteristics (Tsujita & Westermann, 1998) or evolutionary patterns. Jacobs, Landman & Chamberlain (1994) showed that more streamlined morphs of the Cretaceous genus Scaphites could be more hydrodynamically efficient at high speeds than inflated morphs, and that these streamlined morphs are associated with coarser grained, higher energy inshore habitats of the Western Interior Seaway (cf. Courville & Thierry 1993 for a similar case in a different Cretaceous taxon). Strong intra-species variation in shell morphology without apparent facies association questions a close correlation between shell shape and ecology in several Palaeozoic (e.g. Korn & Klug, 2007; De Baets, Klug & Monnet, 2013) and Mesozoic ammonoids (e.g. Kennedy & Cobban, 1976; Dagys & Weitschat, 1993; Morard & Guex, 2003). However, the situation might be more complex as demonstrated in a study of facies distribution by Wilmsen & Mosavinia (2011), who demonstrated different modes of intra-specific variation within the same Cretaceous ammonoid species (Schloenbachia varians) depending upon the type of environment. More compressed forms are more common in open water and more heavily ornamented, depressed variants in shallow water facies, which is opposite to the facies distribution reported by Jacobs, Landman & Chamberlain (1994) for Scaphites. Wilmsen & Mosavinia (2011) attributed this pattern to selection pressure by predation in addition to hydrodynamics. Oxygenation is another factor classically attributed to influence ammonoid morph distribution (Batt, 1993; Tsujita & Westermann, 1998). However, the model of Batt (1993) does not always uphold as evidenced by more high-resolution stratigraphic studies, which could not find a link between anoxia and the extinction of various morphotypes (e.g. Monnet & Bucher, 2007). Some immigrants that diversified within isolated seaways or basins showed gradual morphologic change, possibly reflecting ecological selection on hydrodynamic properties (Yacobucci, 2004; Klug et al., 2005; see Fig. 2), whereas other groups failed to establish endemic lineages (Fernández-López & Meléndez, 1996).
Detailed studies of ammonoid diversification events also show that selection on shell shape may be more complex than Wiedmann's (1973) model of control by sea level. Early evolutionary lineages of ammonoids show rapid, progressive coiling of the shell, which might have resulted in changes in aperture orientation, locomotion and reproductive strategy (Klug & Korn, 2004; De Baets et al., 2012; see Fig. 2). In addition to the common trend in lineages to produce increasingly tightly coiled shells (Monnet, De Baets & Klug, 2011), heteromorphic post-embryonic shells evolved repeatedly in the Mesozoic, which may be linked to changes in trophic conditions, environmental stress and/or sea-level changes (Cecca, 1997; Keupp, 2000; Guex, 2006). Analyses of taxonomic extinction and rediversification after the end-Permian and end-Triassic mass extinctions (Brayard et al., 2009; Guex et al., 2012) are matched with analyses of selection on morphological disparity (Dera et al., 2010; Brosse et al., 2013) and biogeographic trends (Dommergues et al., 2001; Brayard, Escarguel & Bucher, 2007; Dera et al., 2011; Brayard & Escarguel, 2013). Brayard et al. (2007) discovered that assemblages of ammonoid genera were more similar on opposing coasts of Pangaea than at neighboring sites along the coasts at higher latitude (cf. Brayard & Escarguel, 2013; Brosse et al., 2013). Analysis in ‘Westermann Morphospace’ shows that ammonoid shell forms traditionally interpreted to be more streamlined might have traversed the tropical global ocean (Brosse et al., 2013; see Fig. 3). In contrast, following the end-Triassic mass extinction, earliest Jurassic ammonoid faunas featured chiefly serpenticonic shells (Ritterbush & Bottjer, 2012), which have classically been interpreted as planktonic drifters (Westermann, 1996). However, evidence from shell shape alone is insufficient to corroborate such hypotheses.
Ammonoid habitats through ontogeny
Like in other molluscs, accretionary shells of extant and fossil cephalopods record stable isotopic signatures of δ13C and δ18O throughout the animal's lifespan. Only pristine aragonite preservation allows measurement of these isotopic signals along the shell spiral, from embryonic to adult stages. Recent studies (cf. Lukeneder et al., 2008, 2010; see Fig. 4) include shells of modern Nautilus, the common Atlantic cuttlefish (Sepia), the internally shelled deep sea squid Spirula as well as several Jurassic to Cretaceous ammonoids (Baculites; Fatherree, Harries & Quinn, 1998; Cadoceras, Hypacanthoplites, Nowakites, Perisphinctes; Lécuyer & Bucher, 2006) and fossil nautiloids (Cenozoic Aturia; Schlögl et al., 2011). Dynamic transitions in an isotope record through ontogeny may have been influenced by transitions in the animal's environment (sea water temperature, salinity or habitat depth) and metabolic processes (diet, reproduction).
Division of δ18O isotopic records of modern and fossil cephalopod shells into embryo, juvenile, mid-age and adult stages allows interpretation of transitions in habitat temperature and indirectly depth (see Fig. 4; Lukeneder et al., 2010). Various strategies are documented in the stable isotope record of the shells of derived Jurassic and Cretaceous ammonoids, some of which can be compared with modern cephalopods. Cadoceras demonstrates an isotopic record similar to modern Nautilus and the cuttlefish Sepia, suggesting that it would have been a planktonic migrant travelling from warm shallow to cool deep waters, then returning to the shallows to spawn. In contrast, the ammonoid Hypacanthoplites rose from cold, deep habitats to warmer, shallower habitats in adulthood, which resulted in a record partially similar to the modern deep sea squid Spirula. Habitat temperature changes of 6–10°C (Fig. 4) would imply a depth change of 50–700 m today, and potentially more in unstratified Mesozoic seas (Lukeneder et al., 2010), although these values may be overestimates because the chambers of ammonoid shells have been estimated to only withstand hydrostatic pressure of depths up to 400–500 m in most ammonoids (Hewitt, 1996). Alternatively, changes in temperature might also be related to horizontal migrations or seasonal changes in water temperature (Lécuyer & Bucher, 2006). As in modern cephalopods, it seems reasonable to assume that ammonoids initiated migrations due to changes in diet or later in the mating/spawning phase (Lukeneder et al., 2008, 2010). Such migrations through ontogeny would explain the separation of juvenile and adult ontogenetic stages of several taxa in the fossil record. Comparison of oxygen isotope data of ammonoids with exceptionally well-preserved co-occurring plankton and benthos demonstrates that at least some Cretaceous ammonoids might have lived and fed close to the seafloor (Moriya et al., 2003). Recent improvements in the precision and affordability of isotopic studies should allow the generation of more data to further test these apparent trends and their association with shell shape and environmental variables.
The accretionary shells of ammonoids record the life history from embryo to adult, which makes them important subjects of ontogenetic studies (Davis et al., 1996; Klug, 2001; Gerber, Eble & Neige, 2008; Lukeneder et al., 2010; De Baets et al., 2012). The end of the embryonic and adult stages is marked by constrictions and changes in shape, shell thickness and ornamentation (see Davis et al., 1996 and Landman et al., 1996 for a review). They were probably fecund producers of planktonic young.
Ammonoids produced a large number of offspring (compared with extant Nautilus), which is supported by their evolutionary history as well as size class distributions and high abundance of early ontogenetic stages in fossil assemblages (Landman et al., 1996). During their early evolutionary history, ammonoids show a fast, progressive coiling, a size reduction of their embryonic shell and a synchronous trend towards larger adult size in some lineages (Klug & Korn, 2004; see Fig. 2), which probably resulted in the increase of fecundity (De Baets et al., 2012). Higher fecundity and mortality of derived ammonoids is supported by small hatchling size and local occurrences of large numbers of preserved embryonic shells in the Late Devonian to Mesozoic (Landman et al., 1996; Tomašových & Schlögl, 2008; De Baets et al., 2012). Ammonoid hatchlings might have been an abundant food source as indicated by the remains of smaller ammonoids in the stomachs of larger ones (Keupp, 2000; see Fig. 5b) and by the abundance of embryonic shells in a coprolite of a yet unidentified predator (Tanabe, Kulicki & Landman, 2008). A planktonic mode of life for hatchlings is supported by studies on their shells: buoyancy calculations within the error range of neutral buoyancy (Shigeta, 1993; Tanabe, Shigeta & Mapes, 1995), their common occurrence in dysoxic sediments, where they are associated with planktonic gastropods and little to no benthos (Mapes & Nützel, 2009), and their small size compared to the adults, similar to modern cephalopods that disperse as plankton (De Baets et al., 2012). The wide distribution of various loosely coiled or not-planispirally coiled heteromorphic ammonoids, which are interpreted to be poor swimmers, also supports the hypothesis of a wide transportation by oceanic currents during the planktonic stage (Ward & Bandel, 1987; De Baets et al., 2012). The different geographical ranges of ammonoids with different shell morphologies and similar embryonic shell sizes, however, have been interpreted to indicate that at least some ammonoids could also have dispersed considerable distances later in their ontogeny (Brayard & Escarguel, 2013), similar to some modern coleoids.
Whether or not ammonoids had a vast variety of reproductive strategies just like modern coleoids is poorly known, but various strategies also known from their extant relatives have been suggested for different ammonoids including egg-laying on the seafloor or on algae (Westermann, 1996), floating egg masses (Mapes & Nützel, 2009), broadcast spawning (Walton, Korn & Klug, 2010) or brood care up to hatching (Ward & Bandel, 1987). Etches, Clarke & Callomon (2009) have suggested benthic egg deposition of some Jurassic ammonoids like in modern Sepia; direct evidence for their ammonoid origin (like ammonoid remains inside the eggs) is, however, still missing. The smallest embryonic shells and the most extreme cases of dimorphism (e.g. largest differences in size – up to 15:1 – and morphology) occur in some Jurassic ammonoids (such as Phlycticeras and Oecoptychius; Schweigert & Dietze, 1998). Such examples are not rare and suggest that some derived ammonoids not only had a very high fecundity, but possibly also performed brood care until hatching, analogous with some extant octopods (e.g. Argonauta). It is important to note, however, that ammonoid shells could not have been used as egg cases like in the modern pelagic octopus Argonauta, which produces a brood chamber that superficially resembles an ammonoid shell (Hewitt & Westermann, 2003). Differential investment into reproduction between males and females is interpreted in species with sexual dimorphism, which has been reviewed by Davis et al. (1996). The smaller forms (microconchs) have been interpreted to be the males, whereas the larger forms (macroconchs) might be the females because of the larger space for gonads (and thus eggs) as well as occasional umbilical bulges (Landman et al., 2012). Not all ammonoids show a pronounced dimorphism (Davis et al., 1996), and in some, the males might even have been slightly larger than females (as in extant Nautilus: Saunders & Landman, 1987). The large variability in ammonoid adult size and shape as well as differences in the extent of sexual dimorphism suggests that large differences in reproductive strategies between different groups of ammonoids existed (De Baets et al., 2012).
Constraints on predator–prey relationships
The diets of ammonoids can be reconstructed from crop and stomach contents, coprolites, regurgitates (‘Speiballen’) and bite marks. Ammonoids are usually interpreted as predatory [including scavenging (Tanabe et al., 2013) and cannibalism (Keupp, 2012)] like most extant and fossil cephalopods (Nixon, 1996). One of the only exceptions known so far is the deep sea vampire squid (Vampyroteuthis), which, despite its daunting name, is a detritivore (Hoving & Robison, 2012). The diet of Palaeozoic and Triassic ammonoids is poorly understood, whereas the assumed loss of the catch/bite function of jaws of post-Triassic aptychi-bearing ammonites (‘Aptychophora’) and the morphology of the radula of some of these (Engeser & Keupp, 2002; Kruta et al., 2011) has been used to suggest that they mostly fed on zooplankton. This is largely supported by the rare ammonoid crop/stomach contents as recently summarized in Kruta et al. (2011) and Keupp (2012), although small benthic prey items are also present. Prey items reported so far include Foraminifera, benthic Malacostraca (eryonids), Ostracoda, Isopoda, stalkless crinoids (see Fig. 5a), pseudoplanktic Bivalvia, planktonic Gastropoda and small ammonites (see Fig. 5b). Great caution has to be applied when interpreting such rare findings because post-mortem transported remains, scavengers or gregarious inhabitants (Kruta et al., 2011; Klompmaker & Fraaije, 2012) might be confused with actual prey items. An exploitation of lower levels of the food chain (Kennedy & Cobban, 1976) would also be in line with the rather sluggish swimming capabilities inferred for externally shelled cephalopods when compared with coleoids and fishes (Jacobs & Chamberlain, 1996) and the fast recovery of ammonoid diversity following extinction events (e.g. Brayard et al., 2009). Their position in the food chain could potentially be tested by nitrogen isotope composition of amino acids in their shells, as has recently been performed for Spirula (Ohkouchi et al., 2012).
Synchrotron technology has increased the level of details known about the ammonoid mouth parts, which can help constrain their capacity to process prey (Kruta et al., 2011). Many pre-Jurassic ammonoids had non-mineralized jaws similar to modern coleoid beaks, and a set of homodont or heterodont radular teeth (Klug & Jerjen, 2012). In the Jurassic and Cretaceous, many ammonoids had a mineralized lower jaw (aptychus) and an array of more complex radular teeth (Engeser & Keupp, 2002; Kruta et al., 2011; Tanabe et al., 2013; see Fig. 5c). In some, the buccal architecture is more similar to rare pelagic coleoids than to most other modern cephalopods (Kruta et al., 2011). The size and shape of both teeth and jaws vary substantially, and this may prove to covary with phylogeny, shell shape or both (Kruta et al., 2011, 2013; Klug et al., 2012; Tanabe et al., 2013). At least some lower jaws within two major ammonite lineages, the phylloceratids and lytoceratids, bore similarities to modern and fossil nautilids. Tanabe et al. (2013) proposed that these ammonoids convergently evolved a scavenging predatory feeding mode similar to modern Nautilus, but this awaits further corroboration from crop or stomach remains. Modern Nautilus feeds on small fish, crustaceans, nematodes, echinoids and tentacles of other nautilids (as summarized in Tanabe et al., 2013).
Evidence of predation on ammonoids is more difficult to match to specific predator taxa. Apart from ammonoid remains within other ammonoids, ammonoid jaws have been found in plesiosaur stomachs (Sato & Tanabe, 1998), but more frequently jaws and shell fragments are found in coprolites of jawed fishes and as yet unidentified predators (Mehl, 1978; Keupp, 2012). Successful or attempted predation can also be interpreted from fatal and repaired damage marks on ammonoid shells. It is usually difficult to assign such shell injuries to a particular predator (Kröger, 2000; Klug, 2007; Klompmaker, Waljaard & Fraaije, 2009), except in rare cases where predator and prey are found in association (Vullo, 2011), or the predator can be precisely deduced from shape and position of the injuries (Martill, 1990; Mapes & Chaffin, 2003; Keupp, 2006; Richter, 2009). Injuries attributed to pelagic faunas include damage by sharks (in Carboniferous ammonoids; Mapes & Chaffin, 2003), jawed fishes (Martill, 1990; Keupp, 2012), controversial mosasaur bite marks (on Cretaceous ammonoids; e.g. Tsujita & Westermann, 2001) and triangular bite marks typical for coleoids or nautilids (Keupp, 2012). In addition to establishing trophic specialization, associations with benthic or pelagic predators and prey can support or reject interpretations of ammonoid habitat. Small aperture pits attributed to stomatopod damage on shells of platycones (Cleoniceras; see Fig. 5d) and planorbicones (Desmoceras) would be consistent with their interpretation as fitting a demersal life mode, but long slits attributed to scissor-bearing eryonid crustaceans on cadicones (Ptychites) and serpenticones (Dactylioceras, Subolenekites, Pleuroceras, Douvilleiceras; cf. Fig. 5e) would not match a planktonic interpretation for these shapes (cf. Fig. 3). Other biotic interactions can also constrain information on ammonoid ecology (Keupp, 2012). Epizoa living on the shell during the lifetime of the ammonoid can give information about their orientation and position in the water column (e.g. Keupp et al. 1999). For example, Hauschke, Schöllmann & Keupp (2011) used the distribution of a barnacle encrusted in vivo on a straight-shelled ammonoid to suggest a more horizontal orientation during swimming, which has also been independently inferred from shell modelling (Westermann, 2013). It is clear from the studies presented herein that interpretations of ammonoid ecology for certain taxa and time frames cannot necessarily be generalized to other time frames or forms with similar shell shapes. More data are necessary before such models can be used to investigate local or global changes in diversity.
Ammonoid ecology and their extinction
Ammonoids (as well as belemnites with equally small embryonic shells) went extinct after the Cretaceous Mass Extinction, whereas nautilids with larger embryonic shells survived (Wani et al., 2011). The ammonoid extinction at the end of the Cretaceous has often been attributed to their mode of life and reproductive strategy (e.g. Ward, 1996; Tanabe, 2011), but the exact mechanisms are still debated. The planktonic feeding habit of adult aptychus-bearing ammonoids (Kruta et al., 2011), and the planktonic habit of hatchlings (Tanabe, 2011), have been used to explain ammonoid extinction as a consequence of the collapse of plankton at the end of the Cretaceous. However, only calcifying plankton went through a major crisis during this event (Alegret, Thomas & Lohmann, 2012; Hönisch et al., 2012). Other authors have tried to explain the demise of ammonoids by the impact of oceanic acidification, particularly on ammonoid hatchlings (Alegret et al., 2012; Arkhipkin & Laptikhovsky, 2012), although the evidence for a rapid acidification of oceanic surface waters during the end-Cretaceous Mass Extinction is not yet unequivocal (Hönisch et al., 2012). Interestingly, ammonoids went through extreme bottlenecks during other mass extinctions (leaving only a few taxa), but rapidly recovered taxonomic diversity in the aftermath of these extinction events (House, 1989; Guex, 2006; Brayard et al., 2009; Korn & Klug, 2012).
The tools available to determine ammonoid life mode have recently increased in variety, precision and accessibility. Perhaps, exponential progress can now be made as these tools are used in combination to challenge the emerging picture of ammonoid ecology. All evidence suggests that ammonoids spent most of their life in the water column, but it most likely differed between taxa and through ontogeny within a species, ranging between demersal (nektobenthic), nektonic and/or planktonic. Due to their non-benthic ecology, their habitat and mode of life can only be constrained by combining information from isotope and facies analyses, studies on function, morphology and intra-specific variability of their hard parts and direct or indirect evidence for biotic interactions. The potential influence of a wide range of ecological preferences needs to be considered in studies on extinctions, biostratigraphy and macroevolution of ammonoids.
We would like to thank Helmut Keupp (FU Berlin) for supplying pictures and Christian Klug (University of Zürich) for proofreading a previous draft and supplying figures. We greatly appreciate the thoughtful and constructive reviews by two anonymous reviewers. David Button (University of Bristol) and Laura Wilson (The University of New South Wales, Sydney) proofread a previous version of the paper.