Exploring abnormal Cambrian-aged trilobites in the Smithsonian collection

Biomineralised trilobite exoskeletons provide a 250 million year record of abnormalities in one of the most diverse arthropod groups in history. One type of abnormality—repaired injuries—have allowed palaeobiologists to document records of Paleozoic predation, accidental damage, and complications in moulting experienced by the group. Although Cambrian trilobite injuries are fairly well documented, the illustration of new injured specimens will produce a more complete understanding of Cambrian prey items. To align with this perspective, nine new abnormal specimens displaying healed injuries from the Smithsonian National Museum of Natural History collection are documented. The injury pattern conforms to the suggestion of lateralised prey defence or predator preference, but it is highlighted that the root cause for such patterns is obscured by the lumping of data across different palaeoecological and environmental conditions. Further studies of Cambrian trilobites with injuries represent a key direction for uncovering evidence for the Cambrian escalation event.


RESULTS
Elliptocephala asaphoides Emmons, 1844, USNM PAL 18350a, Browns Pond Formation (=Schodack Formation) (Cambrian Stage 2, Series 4, age taken from Skovsted & Peel, 2007), eastern New York, USA. Figures 1A, 1B. USNM PAL 18350a is preserved as an external mould with a 'U'-shaped abnormality that truncates pleurae by 8 mm. Abnormality begins at thoracic segment 6, extends into thoracic segment 8, and is 15 mm long. The margin of the abnormality is cicatrised along thoracic segment 6, while thoracic segments 7 and 8 show no evidence for cicatrisation. USNM PAL 443790 is preserved as an external mould and displays a possible bilateral thoracic abnormality. Thoracic segments 3 and 4 on the left side have been truncated into an asymmetric 'V'-shaped abnormality that is 3 mm long and slightly cicatrised (Fig. 2B). The thoracic segments are truncated pleurae by 9 mm. On the right side of the thorax, there is a potential SSI on thoracic segment 6 ( Fig. 2C). However, closer examination of the specimen highlights that there are likely traces of more parts of the pleural spine. In  USNM PAL 729428 is preserved as an external mould with little relief and shows two abnormalities on the right posterior thorax. The more anterior abnormality has a 'U'-shape, is observed on thoracic segments 7-9, and is slightly cicatrised. Thoracic pleurae 7 and 8 are fused together at the abnormality margin. The second abnormality is 'W'-shaped, spans thoracic segments 11-13, and has a cicatrised margin. Both abnormalities truncate the thoracic pleurae by 6 mm.
Olenellus getzi Dunbar, 1925 USNM PAL 729422 is a partial specimen preserved as an external mould and displays two abnormalities on the right side of the thorax. The more anterior abnormality is an SSI on the 6th thoracic segment that shows no evidence of cicatrisation and truncates the pleura by 6 mm. Thoracic segment 7 shows possible evidence of two thoracic pleurae developing from the one thoracic segment. The split between these spines occurs ∼24 mm from the midline of the specimen. As the upper of the two pleurae shows terraced lines indicative of the ventral surface (Lieberman, 1999), it is possible that this specimen instead represents a fragment retained during moulting, or even the chance superimposition of a fragment on a complete specimen. Fragments on the specimen indicate that either of these two scenarios is possible; however, the alignment of the pleura with the segment, and its relative size, support a biological interpretation. Nevadia weeksi Walcott, 1910 USNM PAL 56792a is a partial specimen preserved as an external mould with an abnormality on the posterior right thorax (Figs. 5A, 5B). The abnormality is an SSI on the 16th thoracic segment. The pleura is terminated 2 mm from the thoracic axial lobe, rounded, and shows no evidence of cicatrisation. This abnormality truncates the pleura by 28 mm.
USNM PAL 56792d preserves the posterior section of the exoskeleton and has a bilaterally expressed injury. The plural spine on the 14th thoracic segment on the left side is an SSI that truncates the pleura by 10 mm (Fig. 5D). On the right thoracic side, the 11th and 12th segments show SSIs. The terminus of the 11th thoracic pleura is not rounded and truncated by at least 7 mm (Fig. 5E). The terminus of the 12th thoracic pleura is slightly rounded and truncated by at least 9 mm (Fig. 5E). No abnormalities on this specimen show evidence of cicatrisation. USNM PAL 729419 is preserved as an external mould and displays an abnormal right thorax. Abnormality is an SSI on the 4th thoracic segment that truncates the segment by 8 mm. The margin of the abnormality is rounded and slightly cicatrised. USNM PAL 729421 is preserved as an external mould and displays an abnormality on the left thoracic lobe. The abnormality has a shallow 'W'-shape that begins at the 2nd thoracic segment, ends at ends at the 5th thoracic segment, is 10.5 mm long, and truncates the pleurae by 2 mm. Abnormality margin shows no evidence of cicatrisation. Thoracic pleurae 3 and 4 are fused together at the abnormality margin, while the margins of pleurae 2 and 5 are distorted about the fused section. USNM PAL 729417 is preserved as an external mould and displays an abnormality on the posterior right thorax that extends into the anterior pygidium. The abnormality has a 'U'-shape, begins at the 10th thoracic segment, ends within the first 1 mm of the pygidium, is 4.6 mm long, and truncates the affected thoracic segments by 1.5 mm. The abnormality margin is slightly cicatrised and deforms thoracic segments 10-11.  (Owen, 1985;Conway Morris & Jenkins, 1985;Babcock & Robison, 1989;Babcock, 1993;Babcock, 2003; Babcock, 2007; Robison, Babcock & Gunther, 2015;Bicknell & Paterson, 2018;Bicknell, Paterson & Hopkins, 2019;Pates & Bicknell, 2019). The injuries that affect more than one thoracic segment and are located on thoracic areas that are unlikely to have been damaged by chance (Babcock & Robison, 1989;Babcock, 1993;Pates et al., 2017;Bicknell & Paterson, 2018) represent healed injuries, likely attributable to predation. Those injuries showing exoskeletal cicatrisation reflect an attack that occurred recently within the same intermoult period, as observed in modern arthropods (Ludvigsen, 1977;Bursey, 1977;Owen, 1985;Rudkin, 1985;Halcrow & Smith, 1986). Injuries that occurred during a soft-shelled stage, when individuals were likely more vulnerable to attacks, would likely have wrinkled and deformed the exoskeleton as it would not have been fully mineralised (Conway Morris & Jenkins, 1985;Bicknell & Paterson, 2018). Injuries lacking cicatrisation and showing partial regeneration likely occurred in prior moult stages (Owen, 1985;Pates et al., 2017). The SSIs observed in the studied sample may record attacks or moulting complications. In particular, the injuries on Nevadia weeksi likely reflect complications during moulting as the most elongated thoracic pleurae that would catch or not exit the old exoskeleton cleanly during ecdysis (Šnajdr, 1978;Owen, 1983;Conway Morris & Jenkins, 1985). The studied sample presents possible evidence for injury patterns in an entirely qualitative context. This assessment provides potential support for an interpretation of injuries to trilobites caused by predators showing location specificity for the right side; if all injuries do indeed represent failed attacks (e.g., Babcock & Robison, 1989;Babcock, 1993). Six specimens have right-sided injuries, two specimens have potential bilateral injuries and one specimen has a left-sided injury. Although this sample size is too small to test for statistical significance, these data align with the thesis that either Cambrian predators attacked prey from the right side, Cambrian prey orientated themselves to have the right side attacked, or a combination of both (Babcock & Robison, 1989;Babcock, 1993;Eaton, 2019). Evidence for lateralised injury patterns in trilobite systems was also recently presented in the Silurian-aged Rochester Shale (Bicknell, Paterson & Hopkins,  Bicknell (2019) outlined, studies of injury lateralisation that pool data on injuries from different time periods do not allow researchers to distinguish between potential causes of injuries, as the studied taxa are from different deposits. The true taxon-specific palaeoecological signal is therefore masked by variation in temporal and geographical conditions and it is unlikely that they all were under the same predatory selection pressures (Pates et al., 2017;Bicknell, Paterson & Hopkins, 2019;Pates & Bicknell, 2019). The identification of both Cambrian (Babcock, 1993;Eaton, 2019) and Silurian signals (Bicknell, Paterson & Hopkins, 2019), with the failure to detect a Cambrian signal (Pates et al., 2017;Pates & Bicknell, 2019), and post-Cambrian signal in other cases (Babcock, 1993) demonstrates that the causes of injury lateralisation are best considered on a case-by-case basis. Such an approach provides the best chance of identifying the root causes of an interesting ecological interaction.
No cephalic injuries were reported in this study. This rarity of cephalic injuries has been noted by previous workers (e.g., Owen, 1985;Babcock, 1993;Pratt, 1998). Biological explanations for this pattern could be that predators targeted the thorax and pygidium preferentially, a higher mortality rate of injuries to the head region, and/or trilobites protecting the head region through behavioural actions such as enrolment (Ortega Hernández, Esteve & Butterfield, 2013;Bicknell & Paterson, 2018;Pates & Bicknell, 2019). It is unlikely that it represents sampling bias, as specimens have been collated from a large number of collectors, and trilobite cephala and cranidia provide a wide range of taxonomic, morphometric, and phylogenetic characters and landmarks (e.g., Lieberman, 1999;Webster, 2015). Furthermore, in a bulk sample with no collection bias injured cephala were significantly rarer than thoracic injuries (Pates & Bicknell, 2019).
Analysis of the radiodont oral cone has not provided any definitive evidence to support a durophagous lifestyle for these animals (Whittington & Briggs, 1985;Hou, Bergström & Ahlberg, 1995;Hagadorn, 2009;Hagadorn, Schottenfeld & McGowan, 2010;Daley & Bergström, 2012), despite suggestions that the shape might be suitable for producing 'W'shaped injuries (e.g., Babcock & Robison, 1989;Nedin, 1999). These lines of evidence, combined with the lack of any hard-parts in known radiodont guts (e.g., Daley & Edgecombe, 2014), has led to suggestions that radiodonts may not have fed on hard-shelled taxa at all (with some potential exceptions discussed below).
One final consideration regarding possible predators is the idea that injuries may have been inflicted by shell hammering, as opposed to shell crushing (Pratt, 1998). It has been suggested that raptorial frontal appendages of Yohoia tenuis Walcott, 1912 would have been effective at breaking biomineralised exoskeletons, using similar mechanics to modern-day mantis shrimps (Pratt, 1998;Haug et al., 2012;Bicknell & Paterson, 2018). Analyses of such morphologies with comparisons to extant stomatopods may highlight the effectiveness of such Cambrian shell hammering (Crane et al., 2018).

Escalation and predation
Escalated evolution reflects selective pressure placed on individuals by predators, parasites, competitors and dangerous prey (Vermeij, 1994;Vermeij, 2013). Such pressures drive the development of adaptive features in prey to avoid, escape, or defend against predators (Vermeij, 1994;Vermeij, 2013;Thompson, 1999;Baumiller & Gahn, 2004). The record of prey escalation includes changes to external shell ornamentation, fluctuation in predation intensity, and prey regeneration frequency (Vermeij, Schindel & Zipser, 1981;Kelley & Hansen, 1996;McShea, 1998;Alexander & Dietl, 2001;Baumiller & Gahn, 2004;Whitenack & Herbert, 2015). Vermeij (1989) suggested that escalation was a major component of evolution during the Cambrian Explosion and that escalated predation pressures drove the variety of defensive features in prey (Vermeij, 1989;Bengtson, 2002;Brett & Walker, 2002;Babcock, 2003;Marshall, 2006;Vendrasco et al., 2011;Wood & Zhuravlev, 2012;Voje et al., 2015). However, there is limited quantitative evidence for this evolutionary explanation (Bicknell & Paterson, 2018). The Cambrian escalation event could potentially be demonstrated by documenting changes in defensive adaptations of Cambrian trilobites. To conduct such a study, specimens of the same species from different stratigraphic levels within the same formation could be examined for injuries and responses to predation. If Cambrian trilobites did experienced escalated evolution, innovation in defensive features, such as increased exoskeletal thickness, or changes to hypertrophied spines (Pates & Bicknell, 2019), would be observed, and their role in response to the predation tested. An increased number of injured specimens at particular levels within the section would indicate that a higher survival rate from attacks (Vendrasco et al., 2011). Trilobites, with their excellent fossil record, high diversity, high disparity, abundance, and long record of predation, therefore represent a suitable system for understanding the Cambrian escalation event.

CONCLUSIONS
The current study of abnormal Cambrian trilobites within the Paleontological collection of the Smithsonian National Museum of Natural History presents nine new examples of injured specimens. These injuries display a range of morphologies that are attributed to failed predation and complicated moulting. The possible predatory groups are discussed, and euarthropods with gnathobases and other forms of robust spines are considered as the most probable predators. It is also highlighted that trilobites represent an ideal study system for documenting quantitative evidence for the Cambrian escalation event and responses of prey items to the first durophages.
• Stephen Pates performed the experiments, analyzed the data, authored or reviewed drafts of the paper, and approved the final draft.

Data Availability
The following information was supplied regarding data availability: The raw data are photographs presented in Figs. 1-8