Introduction
The geological deep-time record has revealed that drilling predation can be used as a tool to recognize ecological interactions in both fossil and extant communities (e.g., Kelley, Reference Kelley1989; Kowalewski et al., Reference Kowalewski, Dulai and Fürsich1998; Harper, Reference Harper2003). Drilling frequencies have increased over time (e.g., Huntley and Kowalewski, Reference Huntley and Kowalewski2007) and thus represent a useful proxy for a better understanding of the development of ecological interactions through time (e.g., Bengtson and Zhao, Reference Bengtson and Zhao1992; Conway Morris and Bengtson, Reference Conway Morris and Bengtson1994; Kowalewski et al., Reference Kowalewski, Dulai and Fürsich1998; Kowalewski et al., Reference Kowalewski, Hoffmeister, Baumiller and Bambach2005). Early examples of predatory drill holes in echinoderms have been reported from Late Ordovician stylophorans and other Paleozoic echinoderms (Deline, Reference Deline2008 and references therein). There is a rich record of predation on echinoids that has generally been attributed to a number of different organisms (e.g., Nebelsick, Reference Nebelsick1999; Kowalewski and Nebelsick, Reference Kowalewski and Nebelsick2003). Maximum rates of drilling predation on echinoids increase in intensity from the Early Cretaceous to the Recent (Kowalewski and Nebelsick, Reference Kowalewski and Nebelsick2003).
The comparison of drilling predation patterns on the fossil and Recent clypeasteroid echinoid Echinocyamus, which can be studied in large numbers from different habitats, serves to increase the knowledge of predator–prey relationships among echinoids through time. Data on drilling predation exist for several Recent fibulariids including Echinocyamus (Nebelsick and Kowalewski, Reference Nebelsick and Kowalewski1999; Grun et al., Reference Grun, Sievers and Nebelsick2014; Grun and Nebelsick, Reference Grun and Nebelsick2015) and Fibularia (Nebelsick and Kowalewski, Reference Nebelsick and Kowalewski1999). In addition, drillings have been examined for fossil counterparts of Echinocyamus (Ceranka and Złotnik, Reference Ceranka and Złotnik2003; Złotnik and Ceranka, Reference Złotnik and Ceranka2005) and Fibularia (Meadows et al., Reference Meadows, Fordyce and Baumiller2015). Echinocyamus can be abundant in near-to-offshore habitats varying from tropical to cold-water environments (e.g., Mortensen, Reference Mortensen1927, Reference Mortensen1948; Ghiold, Reference Ghiold1982; Schultz, Reference Schultz2006). The test of these echinoids can be well preserved due to internal supports linking the oral and aboral side and thus strengthening the test (e.g., Seilacher, Reference Seilacher1979; Mooi, Reference Mooi1989; Nebelsick, Reference Nebelsick1999, Reference Nebelsick2008; Grun et al., Reference Grun, Sievers and Nebelsick2014).
This study aims to investigate differences in drilling intensity and site selectivity for predation in two time-averaged Miocene Echinocyamus stellatus (Capeder, Reference Capeder1906) (Fig. 1) samples from different paleoenvironmental settings of the Globigerina Limestone Formation in Malta. The tests were analyzed with respect to: (1) test length; (2) drilling frequencies; (3) drill hole length; (4) size selectivity; and (5) site selectivity. The results are compared to previous work on drilling predation on both Recent (Nebelsick and Kowalewski, Reference Nebelsick and Kowalewski1999; Grun et al., Reference Grun, Sievers and Nebelsick2014) and fossil (Ceranka and Złotnik, Reference Ceranka and Złotnik2003; Złotnik and Ceranka, Reference Złotnik and Ceranka2005) Echinocyamus.
Figure 1 SEM microphotographs of nonphosphatized Echinocyamus stellatus from middle Miocene of Malta: (1) aboral view NHMW 2014/0401/0715; (2) oral view (NHMW 2014/0401/0715); (3) test with drill hole (NHMW 2014/0401/0033); (4) the drill hole of (3) in detail. (1–3) Scale bars=500 µm; (4) scale bar=100 µm.
Drill holes, predators, and echinoids
Drilling predation in marine shells can be used as an ecological signal for the interpretation of predator–prey relationships in both modern and fossil environments and has thus been investigated for a number of Recent and fossil organisms such as foraminifera and ostracods (e.g., Reyment, Reference Reyment1966), polychaetes (e.g., Young, Reference Young1969), mollusks (e.g., Kelley, Reference Kelley1988, Reference Kelley2001; Hoffmeister et al., Reference Hoffmeister, Kowalewski, Baumiller and Bambach2004), brachiopods (e.g., Carriker and Yochelson, Reference Carriker and Yochelson1968; Baumiller et al., Reference Baumiller, Leighton and Thompson1999, Reference Baumiller, Bitner and Emig2006; Hoffmeister et al., Reference Hoffmeister, Kowalewski, Baumiller and Bambach2004), crustaceans (e.g., Palmer, Reference Palmer1982), and echinoderms (e.g., Hughes and Hughes, Reference Hughes and Hughes1971, Reference Hughes and Hughes1981; Warén and Crossland, Reference Warén and Crossland1991; Warén et al., Reference Warén, Norris and Templado1994; Baumiller, Reference Baumiller2003). Common drillers have been identified as predatory gastropods including capulids, eulimids, naticids, nudibranchs, muricids, and tonnacids (e.g., Bromley, Reference Bromley1981; Carriker, Reference Carriker1981; Kelley, Reference Kelley1988; Hoffmeister et al., Reference Hoffmeister, Kowalewski, Baumiller and Bambach2004).
Well-known predators of echinoids are cassid gastropods (e.g., Hughes and Hughes, Reference Hughes and Hughes1971, Reference Hughes and Hughes1981; Nebelsick and Kowalewski, Reference Nebelsick and Kowalewski1999; Złotnik and Ceranka, Reference Złotnik and Ceranka2005; Grun et al., Reference Grun, Sievers and Nebelsick2014). The morphology and size range of holes make it possible to identify potential predators (e.g., Nebelsick and Kowalewski, Reference Nebelsick and Kowalewski1999; Kowalewski and Nebelsick, Reference Kowalewski and Nebelsick2003; Kroh and Nebelsick, Reference Kroh and Nebelsick2006; Złotnik and Ceranka, Reference Złotnik and Ceranka2005; Grun et al., Reference Grun, Sievers and Nebelsick2014). The drill hole diameters produced by cassid gastropods in fibulariids show an extensive range of sizes, for example 0.04 to 1.8 mm in samples from the Miocene of Poland (Złotnik and Ceranka, Reference Złotnik and Ceranka2005) and 0.10 to 4.00 mm from Recent samples (Nebelsick and Kowalewski, Reference Nebelsick and Kowalewski1999; Grun et al., Reference Grun, Sievers and Nebelsick2014). The identification of holes in fossil echinoids can, however, be difficult since the traces may be altered by taphonomic processes and may lack distinctive characteristic features. The drilling of echinoids by cassid gastropods has been studied in detail by Hughes and Hughes (Reference Hughes and Hughes1971, Reference Hughes and Hughes1981), who identified that cassid gastropods use sulfuric acid during drilling to dissolve the calcareous stereom and then remove the etched material by radula movement to expose the underlying stereom to the acid. A circular groove is then cut into the test, and the central disc is pushed into the test cavity. The drill holes produced by cassids are generally circular or elliptical in shape and increase in diameter with the size of the predator (Hughes and Hughes, Reference Hughes and Hughes1971). Drill hole outline can vary from smooth to highly ragged if test surface characters such as ambulacral pores, spine tubercles, and glassy tubercles are intersected (Nebelsick and Kowalewski, Reference Nebelsick and Kowalewski1999; Grun et al., Reference Grun, Sievers and Nebelsick2014). Drill holes in echinoids show concave walls in cross section, caused by differences in stereom density between the plate center and the inner and outer surfaces (Grun et al., Reference Grun, Sievers and Nebelsick2014).
Other known drilling gastropods of echinoids are ectoparasitic eulimids (e.g., Lützen and Nielsen, Reference Lützen and Nielsen1975; Warén and Crossland, Reference Warén and Crossland1991; Warén et al., Reference Warén, Norris and Templado1994). Numerous eulimid gastropods can parasitize a single individual, and each of these can potentially produce several drill holes with a distinct morphology. Such parasitic drill holes show a deeply convex imprint of the eulimid’s snout and a nearly centric complete penetration of the host’s test (Lützen and Nielsen, Reference Lützen and Nielsen1975) and range from 0.1 to 2.3 mm (Warén and Crossland, Reference Warén and Crossland1991).
Geological setting
The Maltese Islands are situated on the Malta-Ragusa Platform and are composed of Cenozoic rocks ranging from late Oligocene (Chattian) to late Miocene (late Tortonian) in age. The succession consists of five major lithostratigraphic units that are well exposed in vertical cliffs around the islands. The strata are generally horizontal, but the stratigraphic succession is locally complicated by synsedimentary tectonics and subsidence (Pedley et al., Reference Pedley, Hughes and Galea2002). Within the succession, two major hiatuses exist that are associated with the deposition of phosphorites.
The studied outcrop is located in northwestern Gozo (Figs. 2, 3), where the strata of the Middle and Upper Globigerina Limestone members (MGLM and UGLM, respectively) are exposed in the Qolla I-Bajda section (36°4'47''N, 14°14'59''E) at Xwejni Bay. This section is directly exposed to the open sea and thus subject to intense weathering. The uppermost few millimeters of the otherwise very durable Globigerina Limestone are relatively soft, presumably due to the constant cycle of seawater soaking and drying. The more resistant parts of the succession of Qolla I-Bajda, namely the terminal hardground of the Middle Globigerina Limestone and associated phosphate conglomerate (Upper Main Phosphate Bed – UMPB), form ledges on which the weathered material accumulates.
Figure 2 Map of Malta and Gozo including geological formations. Sample site (Qolla I-Bajda) is highlighted. Scale bar=5 km. From Kroh (2004: fig. 1).
Figure 3 Field photograph of the Qolla I-Bajda hill near Xwejni, seen from the west. Boundaries of lithostratigraphic units are indicated by white bars, as is the sampling area (hatched in inset). Inset shows the exact position of the site in relation to the surrounding area.
Lithostratigraphy and sedimentary environment
Deposition in the western margin of the Malta-Ragusa Rise represents a major province of Miocene phosphogenesis with a complex origin, transport, and sedimentation of the phosphatized clasts (Pedley and Bennett, Reference Pedley and Bennett1985). The occurrence of large amounts of phosphatized components is generally attributed to marine settings with high organic productivity, low oxygen levels, low sedimentation rates, and low terrigenous input (e.g., Föllmi et al., Reference Föllmi, Gertsch, Renevey, de Kaenel and Stille2008; Tapanila et al., Reference Tapanila, Roberts, Bouaré, Sissoko and O’Leary2008). Following Pedley and Bennett (Reference Pedley and Bennett1985), the phosphatized clasts were deposited during periods of turbulence, interrupted by nondeposition resulting in the formation of hardgrounds as well as the deposition of normal pelagic marine sediments. Rehfeld and Janssen (Reference Rehfeld and Janssen1995) proposed a multiphased development of beds rich in phosphate controlled by sea-level oscillations.
The studied beds follow a prominent horizon in the upper part of the Globigerina Limestone Formation (GLF), which ranges from Aquitanian to Langhian in age. The GLF is predominantly composed of pelagic carbonate limestones deposited offshore. It is subdivided into three members—Lower (LGLM), Middle (MGLM), and Upper Globigerina Limestone (UGLM) members—on the basis of two prominent horizons that can be followed throughout the Maltese Archipelago. The second horizon separates the pale-gray marly limestones of the MGLM from yellowish marly limestones of the UGLM (Pedley, Reference Pedley1992; Föllmi et al., Reference Föllmi, Gertsch, Renevey, de Kaenel and Stille2008).
The basal sediments of the UGLM consist of an ~1 m thick bed containing phosphatic nodules and clasts, phosphatized fossils, and nonphosphatized components within a marly limestone matrix (Pedley and Bennett, Reference Pedley and Bennett1985; Föllmi et al., Reference Föllmi, Gertsch, Renevey, de Kaenel and Stille2008). The sediments overlie an intensely borrowed hardground or omission surface. Various terms have been used to designate these beds (see Bennett, Reference Bennett1980; Pedley et al., Reference Pedley, House and Waugh1978, 1985; Föllmi et al., Reference Föllmi, Gertsch, Renevey, de Kaenel and Stille2008) as well as the preceding hardground surface (Gruszczyński et al., Reference Gruszczyński, Marshall, Goldring, Coleman, Małkowski, Gaździcka, Semil and Gatt2008).
The top of the MGLM shows distinct relief, with a topographic seafloor high in NW and W Gozo (Pedley et al., Reference Pedley, Hughes and Galea2002), where it ends with a distinct terminal hardground with a thalassinoidean burrow system extending up to 1.5 m into the MGLM. In that area, bed thickness of the UMPB is greatest and clast size is largest (Bianucci et al., Reference Bianucci, Gatt, Catanzariti, Sorbi, Bonavia, Curmi and Varola2011). The phosphorite intraclasts and a matrix of planktonic foraminiferal packstones of the overlying UGLM infill this hardground. According to Bianucci et al. (Reference Bianucci, Gatt, Catanzariti, Sorbi, Bonavia, Curmi and Varola2011), the areas where the UMPB is associated with an underlying hardground represent autochthonous phosphatization associated with topographic seafloor highs.
Toward the south and east, clast size and bed thickness decrease. The UMPB Bed is composed of two horizons of phosphorite intraclasts floating in the matrix at Malta and up to five similar horizons at Gozo (Pedley and Bennett, Reference Pedley and Bennett1985; Carbone et al., Reference Carbone, Grasso, Lentini and Pedley1987; Rose et al., Reference Rose, Pratt and Bennett1992). For these occurrences, an allochthonous origin and clast transportation over short distances, presumably from a topographic high in NW Gozo, are assumed.
Different interpretations for the depositional depth estimates assigned to the UMPB can be found in the literature; Carbone et al. (Reference Carbone, Grasso, Lentini and Pedley1987) estimated depositional depths at 25–65 m, while Bennett (Reference Bennett1980) considered the environment to range from deep shelf margin to open shelf sea (see Boggild and Rose, Reference Boggild (née Challis) and Rose1984). According to Challis (Reference Challis1980), the UMPB was deposited at shallower conditions than the MGLM, which is currently thought to have been deposited at depths in excess of 400 m on the basis of the absence of planktonic foraminifera and presence of chert nodules (see Bianucci et al., Reference Bianucci, Gatt, Catanzariti, Sorbi, Bonavia, Curmi and Varola2011). More agreement exists for the paleoenvironment of the UGLM, which is consistently attributed to an upper bathyal setting at 500–600 m (Bellanca et al., Reference Bellanca, Sgarrella, Neri, Russo, Sprovieri, Bonaduce and Rocca2002; Abels et al., Reference Abels, Hilgen, Krijgsman, Kruk, Raffi, Turco and Zachariasse2005) or 500–800 m depth (Bianucci et al., Reference Bianucci, Gatt, Catanzariti, Sorbi, Bonavia, Curmi and Varola2011).
The fauna of the UMPB has been intensively studied and includes both phosphatized and nonphosphatized components. The former comprises vertebrate bones, mollusks including pteropods and cephalopods, corals, serpulids, barnacles, terebratulid brachiopods, bored pebbles, and echinoderms. Nonphosphatized biogenic components include bivalves, barnacles, and echinoids (Pedley and Bennett, Reference Pedley and Bennett1985). Phosphatized skeletons can be well preserved including thin-shelled pteropods, which are preserved as phosphatic internal molds (Rehfeld and Janssen, Reference Rehfeld and Janssen1995; Janssen, Reference Janssen2012).
Biostratigraphy
The terminal hardground of the MGLM is associated with a hiatus of unknown duration, and hence dates given for the UMPB in the literature vary from late Burdigalian to Langhian. The history of chronostratigraphic position of the UGLM is discussed in detail in Bianucci et al. (Reference Bianucci, Gatt, Catanzariti, Sorbi, Bonavia, Curmi and Varola2011) and Foresi et al. (Reference Foresi, Verducci, Baldassini, Lirer, Mazzei, Salvatorini, Ferraro and Da Prato2011). The base of the UGLM has been traditionally assigned to the Langhian on the basis of calcareous nannoplankton and planktonic foraminiferans (nannozone CN4, NN5; Giannelli and Salvatorini, Reference Giannelli and Salvatorini1972; Mazzei, Reference Mazzei1985). Föllmi et al. (Reference Föllmi, Gertsch, Renevey, de Kaenel and Stille2008, fig. 11) place the UMPB at the Burdigalian/Langhian boundary. Foresi et al. (Reference Foresi, Verducci, Baldassini, Lirer, Mazzei, Salvatorini, Ferraro and Da Prato2011) assigned the base of the UGLM (and thus the UMPB) to the Burdigalian. Difficulties in giving a chronostratigraphic age to the UMPB bed are exacerbated by the fact that: (1) the base of the Langhian has not been formally defined; (2) there is a significant time gap represented by the hardground at the base of the UMPB (Föllmi et al., Reference Föllmi, Gertsch, Renevey, de Kaenel and Stille2008); (3) the base of the UMPG across the Maltese Archipelago has a diachronous nature (Foresi et al., 2001); and (4) these marker beds can be followed by one or more omission surfaces (up to five in all) making exact correlation difficult. In Gozo, five individual horizons constitute the UMPB bed, while only one or two are present in the rest of the Maltese Archipelago (see also Föllmi et al., Reference Föllmi, Gertsch, Renevey, de Kaenel and Stille2008).
Studied fauna
Malta is known for its rich echinoid fauna, which has been studied by numerous authors (e.g., Wright, Reference Wright1855, Reference Wright1864; Gregory, Reference Gregory1891; Stefanini, Reference Stefanini1908; Lambert, Reference Lambert1909; Cottreau, Reference Cottreau1914; Zammit-Maempel, Reference Zammit-Maempel1969). Challis (Reference Challis1979, Reference Challis1980), in a study of the paleoecology and taxonomy of Maltese echinoids, recognized 13 echinoid biofacies with Echinocyamus showing a wide distribution ranging from lagoonal to offshore facies. The UMPB at the base of the UGLM contains a macrofauna including the echinoid genera Echinocyamus, Pericosmus, Lovenia, Spatangus, Schizaster, Echinolampas, Brissopsis, Psammechinus, and Sardocidaris. Echinocyamus is particularly common in the basal part of the member, directly on top of the underlying hardground.
The minute clypeasteroids from the GLF were first assigned to the species Fibularia melitensis by Lambert (Reference Lambert1909). Cottreau (Reference Cottreau1914), in his revision of Mediterranean Neogene echinoids, synonymized Fibularia melitensis with Echinocyamus stellatus (Capeder, Reference Capeder1906), a species first reported from coeval strata of the northern coast of Sardinia. Subsequent authors including Rose (Reference Rose1974, Reference Rose1975) and Challis (Reference Challis1980), working on the Maltese echinoid fauna, followed Cottreau (Reference Cottreau1914). Since more than 30 fibulariid species names appear in the literature of the Mediterranean Miocene, and since 18 of these species names were established by Capeder (Reference Capeder1906) alone, it is likely that the diversity of European Neogene fibulariids is overestimated. According to the descriptions of Capeder (Reference Capeder1906), Echinocyamus stellatus seems to be the best match for the Maltese specimens.
Sedimentary environment
The investigated echinoids are preserved in two distinct modes, as phosphatized and as nonphosphatized specimens. Phosphatized tests originate from a period of low sedimentation rate and accumulated at the top of the terminal hardground of the MGLM at a topographic high in NW Gozo. The nonphosphatized specimens, by contrast, derive from a slightly later period when sedimentation rate was higher and phosphatization was absent. Bioturbation is most likely responsible for sediment mixing and time-averaging of the two samples. Age difference between the two samples is poorly constrained because the exact duration of the hiatus linked with phosphatization is unknown. The phosphatized bioclastic sediments are closely associated with the underlying hardground on top of a topographic high on the paleoseafloor, which is interpreted as the site where phosphatization occurred (Bianucci et al., Reference Bianucci, Gatt, Catanzariti, Sorbi, Bonavia, Curmi and Varola2011). By contrast, there is no evidence of transport for the nonphosphatized material that can be considered of autochthonous origin.
Materials and methods
Materials
Fossil tests of the clypeasteroid echinoid Echinocyamus stellatus (Fig. 1) were picked from two bulk sediment samples of the UMPB at the base of the UGLM exposed in the Qolla I-Bajda section at Xwejni Bay, northern coast of Gozo, Malta. The material was collected from the lowermost part of this member, 0.5 to 1.5 m above the weathered terminal hardground of the MGLM (hatched area in Fig. 3).
Material preparation
Bulk sediment samples were dry sieved in the field with an effective mesh size of 1.5 mm. Sieved residuals were processed in the lab by wet sieving through standard sieve sets. Specimens with complete ambitus were cleaned in hydrogen peroxide and an ultrasonic bath. For SEM analysis, tests were additionally cleaned in Rewoqat (Lierl, Reference Lierl1992) to remove fine particles and photographed with a Jeol JSM 6610-LV scanning electron microscope.
Specimens were separated into phosphatized and nonphosphatized specimens. Phosphatized specimens are easily recognizable by their typically brown color and waxy, worn appearance, while the nonphosphatized echinoids are light gray to white and often retain a well-preserved test surface.
Measurements
Tests are measured using an ocular scale. All deviations of the mean or median are listed as standard deviations. The test length is the longitudinal axis, which represents the maximum distance between the anterior and posterior ambitus; test width is the maximum distance between the lateral test margins, perpendicular to the longitudinal axis. Test length and width were compared using Pearson’s correlation coefficient to detect possible allometry. Differences of test length between phosphatized and nonphosphatized specimens were analyzed using a Mann-Whitney U test.
Drill holes
Drill holes are examined for the: (1) outline; (2) length and width; (3) position of drill hole center; and (4) section profile. The drill hole length is the longest distance between drill hole margins, and drill hole width is the widest distance between margins perpendicular to the drill hole length. Spearman’s rho was used in order to test for correlation of drill hole length and width. Potential differences of the drill hole length between specimen types were examined using a Mann-Whitney-U test.
The drilling frequency is defined as the number of drilled tests, either single or multiple drilled, divided by the total specimen number. Pearson’s chi-square was used to test for differences in drilling frequencies between sample types.
Size and site selectivity
Spearman’s rho was used to test for size selectivity between test length and drill hole length. Site-dependent drilling preference was explored using four binomial tests comparing drilling frequencies between the (1) oral and aboral side; (2) petal area and ambital disc aboral side; (3) ambulacral and interambulacral fields; and (4) anterior and posterior (Fig. 4).
Figure 4 Model of Echinocyamus stellatus from the Miocene of Malta: (1–4) aboral test side; (5–8) oral test side. Compared areas are (1, 5) oral vs. aboral site; (2, 6) petal area vs. ambital area; (3, 7) ambulacral fields vs. interambulacral fields; (4, 8) anterior vs. posterior. Compared areas are highlighted in dark and light gray. White areas are not included in the comparison.
The oral side is that part of the test that is visible in perpendicular view from above the peristome. In the same way, the aboral side is that part that is visible in perpendicular view from above the apical disc. Since the test is almost planar, all parts of the tests are covered. The petal area is the expansive petalodium (Zachos, Reference Zachos2015), which is the sum of ambulacral and interambulacral fields enclosed by the ends of the petals. The ambital area is the area between the petal area and the ambitus (outline of the test). Ambulacral fields correspond to the ambulacra; interambulacral fields correspond to the interambulacra. The anterior of the test is the oral and aboral area of the test that is located anterior of the sutures between interambulacrum 1 and ambulacrum II and ambulacrum VI and interambulacrum 4. The posterior of the test is the oral and aboral area of the test that is located posterior of this line. The apical disc and the peristomal area are excluded in this comparison.
The application of a binomial test on unequally dimensioned areas requires an adjustment of the compared area proportions. Thus, the ratios of the compared areas were calculated by dividing the area of each zone by the total area of all of the examined zones (Eq. 1):
$${\rm A}_{{\rm r}} \,{\equals}\,{{{\rm A}_{{\rm i}} } \over {{\rm A}_{{{\rm i{\plus}j}}} }}$$
where Ar=area ratio; Ai=surface area I; Aj=surface area j; Ai+j=total are of all involved zones.
Repositories and institutional abbreviations
Specimens were exported under the Malta National Museum of Natural History’s permit no. T/00/1 and are stored at the Department of Geology and Palaeontology at the Natural History Museum Vienna (NHMW) under the repository numbers NHMW 2014/0400/0001 to NHMW 2014/0401/0724.
Results
A total of 1,053 tests were collected, 259 were phosphatized (24.6%) and 794 (75.4%) were nonphosphatized specimens. In total, 136 drill holes are distributed on 117 individuals (overall drilling frequency=11.1%): 101 tests feature single drill holes; 14 feature two; 1 features three; and 1 features four drill holes. Only a single drill hole could not be measured or have its position assigned due to taphonomic alteration. Another drill hole could not be measured due to abrasion, although its position was determined. From the 101 tests with single drill holes, 44 are phosphatized and 57 are nonphosphatized. Both the phosphatized and nonphosphatized samples feature seven double-drilled tests each, while specimens with three and four drill holes are phosphatized.
Drill hole morphology
The drill holes are circular to subcircular (Fig. 1.3, 1.4) with outlines ranging from smooth to ragged depending on the position of the hole on the test. Drill holes crossing ambulacral pores tend to be more ragged in outline than those located on pore-free areas. Drill hole margins crossing tubercles or glassy tubercles are also more ragged due to the higher local stereom densities of these structures. In cross section, drill hole walls are slightly concave as the diameter of inner and outer surfaces is narrower than that of the intermediate area of stereom. This concave section can, however, be affected by abrasion of the margins. Healed or incomplete drill holes as well as radula scratching traces or attachment scars around or within the drill holes are absent.
Test size
Tests vary from 1.90 to 5.75 mm (mean=3.44±0.65 mm) in length and from 1.50 to 4.90 mm (mean=2.88±0.57 mm) in width for the phosphatized specimens. Nonphosphatized individuals range from 1.55 to 5.15 mm (mean=2.81±0.62 mm) in length and 1.20 to 4.50 mm (mean=2.41±0.56 mm) in width (Fig. 5). The Pearson’s correlation coefficient indicates that test lengths and test widths are strongly correlated (r=0.98, p<0.001, N=1,053; Fig. 6). Phosphatized individuals are significantly larger than nonphosphatized individuals (t-test=-14.13, p<0.001, N=1,053; Fig. 5).
Figure 5 Echinocyamus stellatus from the Miocene of Malta. Test lengths compared among phosphatized and nonphosphatized individuals. N=number of involved individuals; min.=minimum test length; max.=maximum test length; mean=mean test length; SD=standard deviation; frequency=absolute numbers.
Figure 6 Echinocyamus stellatus from the Miocene of Malta. Pearson’s correlation between test length and test width. N=number of involved individuals; r=Pearson’s correlation coefficient; p=p-value of the statistical test.
Drill hole size
Drill holes in phosphatized individuals vary from 0.25 to 4.00 mm (median=1.05±0.73 mm; Fig. 7) in length and from 0.25 to 2.90 mm (median=0.93±0.64 mm) in width. Drill holes in nonphosphatized individuals vary in length from 0.35 to 2.10 mm (median=0.90±0.44 mm) and in width from 0.25 to 1.70 mm (median=0.75±0.39 mm). Spearman’s correlation coefficient indicates that drill hole length and width are strongly correlated (rho=0.93, p<0.001, N=134, Fig. 8). Drill holes in tests from phosphatized and nonphosphatized specimens are similar in length (Mann-Whitney U=1,899.50, p=0.129, N=134, Fig. 7).
Figure 7 Echinocyamus stellatus from the Miocene of Malta. Drill hole lengths compared among phosphatized and nonphosphatized individuals. N=number of involved individuals; min.=minimum drill hole length; max.=maximum drill hole length; mean=mean drill hole length; SD=standard deviation; frequency=absolute numbers.
Figure 8 Echinocyamus stellatus from the Miocene of Malta. Spearman’s correlation between drill hole length and drill hole width. N=number of involved individuals; rho=Spearman’s correlation coefficient; p=p-value of the statistical test.
Drilling frequencies
The phosphatized specimens show a drilling frequency of 20.5% (53 individuals), while nonphosphatized specimens show a frequency of 8.1% (64 individuals; Fig. 9). This difference is statistically significant (chi²=29.03, p<0.001, N=1,052, p<0.001, N=1,052).
Figure 9 Echinocyamus stellatus from the Miocene of Malta. Drilling frequencies compared between phosphatized and nonphosphatized individuals. N=number of involved individuals.
Size and site selectivity
The analyses indicate that a significant correlation between test length and drill hole length is present for phosphatized specimens (rho=0.48, p<0.001, N=64, Fig. 10), while no significant correlation has been detected in the nonphosphatized sample (rho=0.19, p=0.180, N=70, Fig. 10).
Figure 10 Echinocyamus stellatus from the Miocene of Malta. Spearman’s correlation between test length and drill hole length among phosphatized and nonphosphatized individuals. N=number of involved individuals; rho=Spearman’s correlation coefficient; p=p-value of the statistical test.
The binomial tests indicate that drillers of both phosphatized and nonphosphatized samples drill predominantly into the aboral test side (Table 1, Fig. 4). Predators drill equally into the anterior and posterior areas in both of the two samples (Table 1, Fig. 4). The petal area of the phosphatized samples is more frequently drilled than the ambital area, as the ambulacral fields are also more frequently drilled than the interambulacral fields (Table 1, Fig. 4). By contrast, nonphosphatized individuals show no evidence for selectivity of petal area or the ambulacral fields (Table 1).
Table 1 Drilling site comparison on tests of Echinocyamus stellatus. An ‘X’ in ‘site selectivity’ indicates the exact binomial test (exact p°) is significant, based on a significance level of 0.05.
Discussion
Drill hole origin and morphology
The trace fossils documented here can be assigned to the ichnotaxon Oichnus simplex (Bromley, Reference Bromley1981) (see Wisshak et al., Reference Wisshak, Kroh, Bertling, Knaust, Nielson, Jagt, Neumann and Nielsen2015 for a discussion of purported synonymy between Oichnus and Sedilichnus). Cassids are known drilling predators of echinoids (e.g., Hughes and Hughes, Reference Hughes and Hughes1971, Reference Hughes and Hughes1981) and are interpreted as the cause of drill holes in both Recent (Nebelsick and Kowalewski, Reference Nebelsick and Kowalewski1999; Grun et al., Reference Grun, Sievers and Nebelsick2014) and fossil (Złotnik and Ceranka, Reference Złotnik and Ceranka2005) fibulariids. According to previous morphological descriptions of the drill holes (Hughes and Hughes, Reference Hughes and Hughes1971, Reference Hughes and Hughes1981; Nebelsick and Kowalewski, Reference Nebelsick and Kowalewski1999; Grun et al., Reference Grun, Sievers and Nebelsick2014; Meadows et al., Reference Meadows, Fordyce and Baumiller2015), with respect to shape and size, cassid gastropods (Tonnacea) are the most likely drillers of the Maltese Echinocyamus. It is highly unlikely that eulimids are the producers of drill holes in the studied Maltese samples as Lützen and Nielsen (Reference Lützen and Nielsen1975) reported much smaller drill hole diameters for eulimid drillings. SEM micrographs of eulimid drill holes published by Warén and Crossland (Reference Warén and Crossland1991) are obliquely cylindrical or conical as opposed to the drill holes in the Maltese echinoids, which are perpendicular to the test surface. In addition, attachment scars on the test surface attributed to parasitic eulimids are missing.
Echinocyamus stellatus from middle Miocene sediments of Malta do not show allometric growth during ontogeny. These results are similar to those found in Recent Echinocyamus samples from the Red Sea (Nebelsick and Kowalewski, Reference Nebelsick and Kowalewski1999) and the Mediterranean Sea (Grun et al., Reference Grun, Sievers and Nebelsick2014), as well as in fossil examples from Poland (Złotnik and Ceranka, Reference Złotnik and Ceranka2005). The studied echinoids from the two preservation modes show clear differences in average test length, suggesting that they indeed belong to separate samples (Fig. 5). There may be several reasons for such separation, including primary differences in the test size or taphonomic bias due to transport mechanisms during winnowing.
By contrast, drill hole length does not vary significantly between the samples. The drill hole diameter can be correlated to the size of the radula and thus to the size of the gastropod predators (Hughes and Hughes, Reference Hughes and Hughes1971, Reference Hughes and Hughes1981). The similar drill hole lengths in both echinoid samples suggest similar predators of similar size range. The drill holes in Echinocyamus stellatus are between 0.25 to 4.00 mm in length, reflecting the lower end of drill hole diameters reported for Recent cassid gastropods by Hughes and Hughes (Reference Hughes and Hughes1981) on regular adult sea urchins. This is similar to the size range observed for drill holes in modern fibulariids (Nebelsick and Kowalewski, Reference Nebelsick and Kowalewski1999). The variation from small to larger drill holes in Echinocyamus stellatus and the comparison to drill hole lengths produced by adult cassids species suggest that Echinocyamus stellatus served as a food source for both juvenile and adult cassid gastropods.
There is a trend of increasing drill hole length with increasing test length in the phosphatized sample. It cannot be rejected that predators selected the prey by test size, but it is likely that relatively large drill holes weaken the test integrity more than smaller drill holes. Bioturbation, transport, and extended exposure on the seafloor during phosphogenesis may have resulted in a taphonomic filter against the preservation of larger drill holes on smaller tests. By contrast, the more autochthonous, rapidly buried, nonphosphatized material does not show such a trend even though they feature drill holes of the same length range as the phosphatized material. The absence of size selectivity was also reported for drilled fibulariid samples from the Miocene of Poland (Złotnik and Ceranka, Reference Złotnik and Ceranka2005), Oligocene of New Zealand (Meadows et al., Reference Meadows, Fordyce and Baumiller2015), Recent Red Sea (Nebelsick and Kowalewski, Reference Nebelsick and Kowalewski1999), and Recent Mediterranean Sea (Grun et al., Reference Grun, Sievers and Nebelsick2014).
Drilling frequencies
The drilling frequencies of phosphatized and nonphosphatized samples are clearly different. This mirrors the results from Recent studies in the Red Sea where frequencies differed highly among facies from around 60.6% to 80.2% for Echinocyamus crispus (Mazzetti, Reference Mazzetti1893) and 22.0% to 83.3% for Fibularia ovulum (Lamarck, Reference Lamarck1816) (Nebelsick and Kowalewski, Reference Nebelsick and Kowalewski1999). Grun et al. (Reference Grun, Sievers and Nebelsick2014) also reported highly variable drill hole frequencies in samples taken around a single island in the Mediterranean Sea ranging from absent to 21.7%. It is difficult to ascertain the reason for the different rates of drillings in the Recent material (Grun et al., Reference Grun, Sievers and Nebelsick2014), and even more complicated for the fossil material examined in the present study. The higher frequencies shown by phosphatized (20.5%) over the nonphosphatized (8.1%) samples may be due to a number of ecological parameters both biotic (population densities of both predators and prey) and abiotic (such as sedimentation rates) in their contemporary environment.
The frequencies of drilling predation (8.1% and 20.5%) in Echinocyamus stellatus from the Maltese Globigerina Limestone are similar to those reported from the Miocene of Poland. Złotnik and Ceranka (Reference Złotnik and Ceranka2005) show drilling frequencies of 3.8%, 10.9%, and 15.2% for Echinocyamus linearis (Capeder, Reference Capeder1906), Echinocyamus pusillus (Müller Reference Müller1776), and Echinocyamus pseudopusillus (Cotteau, Reference Cotteau1895), respectively. These results are similar to the Oligocene of New Zealand, where Meadows et al. (Reference Meadows, Fordyce and Baumiller2015) reported a drilling frequency for Fibularia sp. of 7.0%. Nebelsick and Kowalewski (Reference Nebelsick and Kowalewski1999), however, revealed clearly higher frequencies from the Recent Red Sea (60.6% to 80.2% for Echinocyamus crispus and 22.0% to 83.3% for Fibularia ovulum).
A possible taphonomic filter against large drill holes in small tests may result in an underrepresentation of drilling frequencies. Such a filter should, however, have a rather small effect on the overall drilling frequency since the test length distribution of drilled specimens roughly follows the normal distribution of nondrilled specimens (Fig. 4).
Site selectivity
The selectivity for the aboral test side in the Maltese Echinocyamus stellatus has also been reported for several Echinocyamus species from both fossil and Recent environments (Nebelsick and Kowalewski, Reference Nebelsick and Kowalewski1999; Złotnik and Ceranka, Reference Złotnik and Ceranka2005; Grun et al., Reference Grun, Sievers and Nebelsick2014) as well as for Fibularia (Nebelsick and Kowalewski, Reference Nebelsick and Kowalewski1999; Meadows et al., Reference Meadows, Fordyce and Baumiller2015). The drilling preference to the aboral side, especially to the petalodium, may reflect an optimization strategy of the predator reducing the amount of energy (e.g., for producing less acid or by removing less stereom due to pore occurrences) and thus may reduce the time for drilling, leading to potentially less disturbance and more drillings over time. The reduction in drilling time for optimization is reasonable since: (1) the predator attacks the buried prey from the aboral side and thus reduces handling time; (2) the aboral side of the test is thinner than the oral side and thus less energy is needed to drill into the test (Złotnik and Ceranka, Reference Złotnik and Ceranka2005); (3) the presence of ambulacral pores on the aboral side of the test means that less material needs to be removed; (4) targeted internal organs are located underneath the petalodium; and (5) in Echinocyamus, internal supports are present beyond the petal area.
The preference for the petal area seems to be characteristic in Echinocyamus prey (Nebelsick and Kowalewski, Reference Nebelsick and Kowalewski1999; Złotnik and Ceranka, Reference Złotnik and Ceranka2005; Grun et al., Reference Grun, Sievers and Nebelsick2014). The fact that there is no significant preference for drilling into the petal and the ambital areas of the nonphosphatized individuals may be due to the generally smaller size of these echinoids. Smaller prey size with constant drill hole diameters results in less specific site selectivity as opposed to the phosphatized examples. There is also evidence for selectivity by predators of the larger phosphatized individuals for the ambulacral areas, which contain relatively large ambulacral pores, rather than the interambulacral areas. Again, the smaller, nonphosphatized individuals show no such selectivity.
Conclusions
The two distinct samples of phosphatized and nonphosphatized Echinocyamus stellatus from the Miocene Maltese Island Gozo can be statistically distinguished with respect to test length and drill hole frequency, but they cannot be distinguished on the basis of drill hole length alone. Drilling patterns are quite similar, while minor differences in site selectivity may be size specific.
The holes in the Miocene Maltese echinoid tests are interpreted as Oichnus simplex (Bromley, Reference Bromley1981) trace fossils. The drill holes are circular to subcircular in outline. Drill hole margins are smooth or ragged depending on the microstructures of the drilling area. The drill hole wall is concave in section and perpendicular to the test surface.
Comparisons of the morphological characteristic and sizes of the drill holes from Malta with descriptions of other drill holes in Echinocyamus species indicate that the drillers of the Miocene Echinocyamus stellatus are most likely cassid gastropods.
Drilling frequencies vary from 8.1% in nonphosphatized echinoids to 20.5% in phosphatized echinoids. Variations in drilling frequencies may be due to variations of predator densities or a number of environmental differences and mirror the high variation of reported drilling frequencies within various Recent depositional environments.
The fact that predators drill more frequently into the aboral test side may be due to handling effects and the time and energy saved by drilling the thinner, highly porous areas of the test. The position of internal organs and internal supports may also be important factors.
Drilling predation traces can be used for analyzing direct predation patterns on well-preserved fossil echinoid tests. The drill holes can also provide indirect evidence for the existence of predators and allow for the analyses of specific drill hole morphologies, sizes, and drill site preferences.
Acknowledgments
We thank G. Zammit-Maempel (formerly Malta National Museum of Natural History) for the collection and export permit. Thanks to the Natural History Museum Vienna, Austria, for providing lodging during a research stay. Thanks also to T. Baumiller (Department of Earth and Environmental Sciences, University of Michigan, USA), G. Dietl (Department of Earth and Atmospheric Sciences, Cornell University, USA) and S. Zamora (Instituto Geológico y Minero de España, Zaragoza, Spain) for reviewing, editing, and constructive suggestions.
