Sexual reproduction represents a fundamental aspect of animal biology, but the diversity of reproductive strategies among early Paleozoic metazoans remains obscure. Direct evidence of reproductive strategies comes from exceptionally preserved egg masses in Cambrian and Ordovician euarthropods such as waptiids and trilobites, but anatomical or behavioral adaptations for mating in these taxa are all but unknown. We demonstrate the presence of functionally specialized appendages in the mid-Cambrian (Wuliuan) Burgess Shale trilobite Olenoides serratus. New material of O. serratus preserves significantly modified and reduced endopodites underneath the seventh thoracic and first pygidial tergites. Comparison with extant euarthropods indicates that these specialized limbs are functional analogs to claspers, which are used by sexually mature males to grasp females prior to or during mating. The claspers in O. serratus were most likely used by the male to hold onto the posterior pygidial spines of the female during amplex, similar to the strategy observed in the horseshoe crab Limulus polyphemus. The new appendicular data from O. serratus provide the first direct evidence for trilobite sexual dimorphism based on the non-biomineralized ventral anatomy. Our findings illuminate the reproductive biology of trilobites and indicate that complex mating behaviors observed in extant euarthropods originated during the Cambrian explosion.
Mating and reproduction are critical parts of metazoan life cycles, but information from the fossil record relies on rare cases of specimens preserved in the act of copulation (Joyce et al., 2012), of unfertilized eggs (Caron and Vannier, 2016; Hegna et al., 2017), or of preserved appendages specialized for reproduction (Kamenz et al., 2011). Although euarthropods (animals with jointed limbs) are by far the most abundant bilaterian group in the macrofossil record since the Cambrian Period (Daley et al., 2018), the mating behavior and reproductive biology of early representatives are largely unknown (Hegna et al., 2017). Trilobites dominate the Paleozoic fossil record thanks to their calcitic dorsal exoskeleton (Wilmot and Fallick, 1989), which has produced some insights into their reproductive biology, such as the presence of possible brood pouches on the preglabellar field in several species (Fortey and Hughes, 1998). However, the only available information on their mating behavior relies heavily on using the extant horseshoe crab Limulus polyphemus (Linnaeus, 1758) as an analog to interpret trilobite clusters as representing molting, mating, and spawning events (Speyer, 1985; Karim and Westrop, 2002). The rarity of trilobite fossils, and euarthropod fossils more generally, with preserved ventral anatomy has hampered our understanding of the evolution of reproduction and mating behaviors in deep time.
The Laurentian corynexochid Olenoides serratus (Rominger, 1887) is known from numerous exceptional fossils from the Burgess Shale (British Columbia, Canada) (Wuliuan, Miaolingian; ca. 508 Ma) and preserves the non-biomineralized ventral anatomy including biramous appendages and caudal cerci in substantial detail (Whittington, 1975). We provide strong evidence of appendage differentiation in the trunk region of the iconic Burgess Shale trilobite O. serratus based on new material deposited at the Royal Ontario Museum (Toronto, Ontario, Canada), and demonstrate the presence of modified limbs consistent with a function for mating in adult males.
MATERIAL AND METHODS
Studied specimens are deposited at the Royal Ontario Museum (ROM), the Geological Survey of Canada (GSC, Ottawa), and the invertebrate paleontology collections at the Smithsonian Institution (USNM; Washington, D.C., USA). All specimens were collected from the Burgess Shale. Specimens were photographed under cross-polarized light using a Nikon D850 DSLR camera fitted with a Macro Nikkor 60 mm lens, or Nikon D7500 DSLR camera fitted with a Macro Nikkor 40 mm lens. Figures were produced in Adobe Photoshop CS® and Adobe Illustrator CS®.
To quantify aspect ratios of podomeres (segments) in the endopodites of O. serratus, we selected 13 specimens with preserved appendages showing clear podomere boundaries on at least one appendage (Table S2 in the Supplemental Material1). Length was measured along the axis of the ramus. Width was measured at the midpoint of length, perpendicular to the axis. See Figure S4 for the detailed methodology used for measuring podomeres. For descriptive purposes, we number the appendage podomeres from distal to proximal.
Of the 13 specimens studied, only one (ROMIP 66299) shows evidence of modified endopodites. All other specimens display the conventional limb morphology known for O. serratus.
Specimen ROMIP 66299 from the Walcott Quarry
Specimen ROMIP 66299 from the Walcott Quarry (51°26’18.8”N 116°28’20.1”W) is a completely articulated but partially broken individual in dorsal view, with a length (sagittal [sag.]; measured from tip of the thorax to the pygidium [posterior part of the exoskeleton]) of 51.6 mm (Fig. 1). The specimen represents an adult holaspid bearing seven thoracic tergites (dorsal plates) and five pygidial segments. The cephalon (head shield) is broken diagonally across the glabella (mid-section of the cephalon) from the anterior of the left librigena (components of the cephalic shield) to the axial furrow of the posterior border on the right side. The right pleural lobes of all thoracic and pygidial tergites are missing and broken along the axial furrow. ROMIP 66299 preserves nine appendage pairs; these correspond to a continuous series starting with appendage 7, belonging to the fourth thoracic tergite, running through appendage 15, which belongs to the fourth pygidial segment. The pleural lobes on the left side remain intact and only the 15th appendage pair is exposed on both sides. There is no evidence of preserved appendages underneath the cephalon and anterior thorax.
The biramous appendages from all previously published and photographed specimens of O. serratus display the same morphology, consisting of a subtriangular protopodite (basal segment of an appendage) (see Bicknell et al.  for revised morphology) connected dorsally to a bipartite, lamellae-bearing exopodite (outer branch of an appendage) (see Hou et al.  for discussion of exopodite connection) and distally to an endopodite (inner branch) composed of seven podomeres (Whittington, 1975; Ramsköld and Edgecombe, 1996; Bicknell et al., 2021; Hou et al., 2021) (Fig. S1A). Although incomplete, appendage pairs 7 to 9 in specimen ROMIP 66299 follow the conventional morphology of biramous appendages (see discussion in the Supplemental Material).
The 10th and 11th appendage pairs in specimen ROMIP 66299 have a substantially different morphology compared to the 7th to 9th, and particularly compared to those typical of O. serratus as described above (Fig. S1A). Endopodite 10 preserves all the podomeres, although the terminal claw is truncated by overlying sediment. Endopodite 11 is less complete, with the claw through podomere 4 obstructed by sediment and overlying appendages. The protopodites of both appendages are subcircular; the proximal margin is obscured by the overlying exoskeleton. The ventral edge of the protopodite is strongly curved and lacks endites. Whereas the protopodite of the 10th and 11th appendages differ mostly in terms of their shape, the corresponding endopodites are dramatically modified in size and morphology relative to conventional O. serratus appendages (Fig. S1). Endopodite 11 closely resembles endopodite 10 but is more flexed ventrally, with an inflection point between the dorsal edge of the protopodite and podomere 7 (Fig. 1). Podomere 7 is square with an inverted W shape, bulging in the middle and with a single endite at the distal margin (Figs. 1 and 2B). Podomere 6 is square with a distal endite slightly shorter than the width of the podomere (Figs. 1 and 2B). Podomeres 5 and 4 closely resemble podomere 6, being nearly square with distally protruding endites (Figs. 1 and 2B). Podomere 3 is rectangular but wider at the distal edge with a thorn-like endite at the margin (Figs. 1 and 2B). Podomere 2 appears slightly rectangular with a longer dorsal margin and thorn-like distal endite on the ventral side (Figs. 1 and 2B). Podomere 1 is truncated by overlying sediment, and only one claw can be seen (Figs. 1C and 1D). Neither endopodites 10 nor 11 show evidence of lateral setae (fine bristles) as observed in conventional appendages (Fig. 1SA). Exopodites for both appendages 10 and 11 are visible but broken. Only the distal lobe of exopodite 10 is visible and appears indented at the terminal margin with no lamellae visible. The proximal and distal lobes of exopodite 11 are visible, with short setae along the distalmost edge of the distal lobe, but no lamellae are associated with the proximal lobe preserved.
Both endopodites 10 and 11 exhibit substantially different aspect ratios in all podomeres (Fig. 2C). Proximally, the difference between reduced and nonreduced appendages is dramatic, highlighted by the fact that endopodite 8 from specimen ROMIP 66299 and endopodite 10 from specimen GSC 34296 both fall near the center of the cluster for all measured podomeres (Fig. 2C). Podomeres 3 and 4 usually develop hourglass or club shapes (Fig. 2B), making them narrow at the midpoint, which could artificially make the reduced endopodites similar despite the obvious differences in morphology.
Significance of Specialized Biramous Appendages
Specimen ROMIP 66299 demonstrates the presence of two distinct sets of biramous appendages in the middle of the trunk (Fig. 1), which is otherwise unknown for O. serratus, or any other trilobite described to date (Table S1). The reduced endopodites of ROMIP 66299 are not only smaller than any other preserved appendages but show a unique morphology of their corresponding protopodite and podomeres. Broken appendages can superficially appear reduced, but given that the endopodites show no cracks or breaks along their length, this is not the case for ROMIP 66299. The changes in protopodite morphology cannot be accounted for by differences in orientation alone. The protopodite does not exhibit a large range of rotation and is usually preserved in an anterior or posterior view (Whittington, 1975; Hou et al., 2021). The presence of ventral endites on appendage 8 indicates the lack of such structures on protopodites 10 and 11 is a legitimate reflection of the morphology. There are different explanations for the significance of the reduced appendages based on the unique organization; namely, that they could be regenerating or are modified for a biological function. However, the interpretation of reparative regeneration does not account for the unique podomere morphology, the similar sizes of endopodites 10 and 11, or the consistent shape of the rounded protopodite (see the discussion in the Supplemental Material).
The reduced endopodites of specimen ROMIP 66299 represent the first instance of substantial biramous appendage specialization in trilobites that is not directly linked with feeding ecology. Some Cambrian artiopods and extant euarthropods have modified appendages specialized for food processing, with robust protopodites where the inner margin is elongate and studded with stout gnathobasic spines (Bicknell et al., 2018, 2021; Holmes et al., 2020). These transversely elongated protopodites distribute the mechanical strain sustained from food mastication more effectively than shorter morphologies (Bicknell et al., 2021). Olenoides serratus has subtriangular protopodites with short medial margins and long ventral spines, making it unable to crush biomineralized shells (Bicknell et al., 2021). In species that have protopodites specialized for durophagy (eating behavior for consumption of hard-shelled or exoskeleton-bearing organisms), such as Redlichia rex, those appendages are located close to the posterior-facing mouth opening (Holmes et al., 2020; Bicknell et al., 2021). The reduced endopodites of O. serratus are poorly suited for durophagous predation or mastication given their position in the body, the absence of gnathobasic endites, and the round outline of the protopodite. Given the low likelihood that the differentiated appendages 10 and 11 were used for feeding based on their appearance and position within the body, we propose that a reproductive purpose is the best available explanation when comparing the functional morphology of ROMIP 66299 and extant euarthropods.
Adult male euarthropods have repeatedly evolved appendages to clasp onto females in both Chelicerata and Mandibulata. Male horseshoe crabs have specialized pedipalps, which they use to hold onto the opisthosomal spines of females (Botton et al., 1996; Gerhart, 2007; Bicknell et al., 2018) (Figs. 3C and 3D). Euarthropod claspers generally feature an enlarged and bulbous fixed “finger” articulated with a more-gracile moveable finger that closes and grasps onto the female (Figs. 3A, 3C, and 3E). In contrast to modern claspers with unsegmented moveable fingers (Figs. 3C–3F), the functional analogue of the moveable finger in O. serratus would consist of the reduced endopodite composed of multiple podomeres (Fig. 3A). The reduced endopodites are preserved in more ventrally flexed positions than other appendages, showing a sharp V-shaped outline along the ventral edge that could allow males to grasp onto the pleural spines of the female, similar to mating in horseshoe crabs (Botton et al., 1996; Gerhart, 2007). Assuming a similar size between males and females, males would be able to reach and clasp the fourth and fifth pygidial pleural spines of the female during amplex (Fig. 4A), consistent with a Limulus-like mating behavior (Fig. 4B).
Proposed instances of sexual dimorphism in trilobites include differences in size (Hu, 1971) and discrete exoskeletal features (e.g., the brood pouches of Fortey and Hughes ). The interpretation of specimen ROMIP 66299 as an adult male indicates that sexual dimorphism in O. serratus was expressed only in the appendages (see the discussion in the Supplemental Material) and thus can only be confirmed by the identification of claspers. By contrast, immature males and females would share a similar appendicular morphology with conventional endopodites (Fig. S1A). If ventral sexual dimorphism is expressed in other trilobites, it might be nearly impossible to identify sex based only on the exoskeleton (see the discussion in the Supplemental Material).
Insights into Trilobite Reproductive Biology
Horseshoe crabs are commonly used as analogs for trilobites preserved as clustered or overlapping specimens, which have been considered as representing possible spawning events (Karim and Westrop, 2002; Paterson et al., 2007). The clasper interpretation of the reduced endopodites strengthens this comparison. Male trilobites would have vied for the optimal position near the female to have the best chances of fertilizing the eggs upon release (Fig. 4). Mate protection is known in both horseshoe crabs and branchiopods (e.g., fairy shrimp and water fleas) with claspers (Brockmann, 1990; Botton et al., 1996; Gerhart, 2007), given that fertilization is external and the male must be in the correct position to successfully mate. The new fossil evidence of O. serratus represents the earliest record of differentiated biramous appendages used for reproduction and demonstrates a previously unknown degree of limb specialization within trilobites for a non-feeding function. Our findings reveal sexual dimorphism in trilobites, with specimen ROMIP 66299 informing the adult male morphology for this species. The discovery of appendicular claspers in trilobites illuminates the reproductive strategies of a dominant metazoan clade of the Paleozoic Era and shows that the existence of complex behaviors and anatomical adaptations for mating dates back to at least the mid-Cambrian.
We thank Jean-Bernard Caron (Royal Ontario Museum, Toronto, Canada), Michelle Coyne (Geological Survey of Canada, Ottawa, Canada), and Mark Florence and Douglas Erwin (Smithsonian Institution, Washington D.C., USA) for facilitating access to specimens; Holly Sullivan for the artistic reconstruction; and Jared Richards for assistance with data visualization. We thank John Paterson, Greg Edgecombe, and an anonymous reviewer for their comments that helped improve our work. This work is published by a grant from the Wetmore Colles Fund (Harvard University).