Edited by Gregory D. Edgecombe, The Natural History Museum, London, United Kingdom, and accepted by the Editorial Board January 29, 2016 (received for review November 14, 2015)
Understanding the evolution of the CNS is fundamental for resolving the phylogenetic relationships within Panarthropoda (Euarthropoda, Tardigrada, Onychophora). The ground pattern of the panarthropod CNS remains elusive, however, as there is uncertainty on which neurological characters can be regarded as ancestral among extant phyla. Here we describe the ventral nerve cord (VNC) in Chengjiangocaris kunmingensis, an early Cambrian euarthropod from South China. The VNC reveals extraordinary detail, including condensed ganglia and regularly spaced nerve roots that correspond topologically to the peripheral nerves of Priapulida and Onychophora. Our findings demonstrate the persistence of ancestral neurological features of Ecdysozoa in early euarthropods and help to reconstruct the VNC ground pattern in Panarthropoda.
Panarthropods are typified by disparate grades of neurological organization reflecting a complex evolutionary history. The fossil record offers a unique opportunity to reconstruct early character evolution of the nervous system via exceptional preservation in extinct representatives. Here we describe the neurological architecture of the ventral nerve cord (VNC) in the upper-stem group euarthropod Chengjiangocaris kunmingensis from the early Cambrian Xiaoshiba Lagerstätte (South China). The VNC of C. kunmingensis comprises a homonymous series of condensed ganglia that extend throughout the body, each associated with a pair of biramous limbs. Submillimetric preservation reveals numerous segmental and intersegmental nerve roots emerging from both sides of the VNC, which correspond topologically to the peripheral nerves of extant Priapulida and Onychophora. The fuxianhuiid VNC indicates that ancestral neurological features of Ecdysozoa persisted into derived members of stem-group Euarthropoda but were later lost in crown-group representatives. These findings illuminate the VNC ground pattern in Panarthropoda and suggest the independent secondary loss of cycloneuralian-like neurological characters in Tardigrada and Euarthropoda.
The nervous system represents a critical source of phylogenetic information and has been used extensively for exploring the evolutionary relationships of extant Panarthropoda (i.e., Onychophora, Tardigrada, Euarthropoda) (1⇓⇓⇓⇓⇓–7). Identification of fossilized nervous tissues has provided a unique perspective on early euarthropod brain neuroanatomy and suggests that broad patterns of extant neurological diversity were already in place by the Cambrian (8⇓⇓–11). The ventral nerve cord (VNC) reflects fundamental aspects of panarthropod body organization that complement the organization of the brain and together illuminate the evolution of the CNS (1⇓–3, 5, 7, 12⇓⇓⇓–16). The early evolutionary history of the panarthropod postcephalic CNS, however, remains obscure due to the exclusive preservation of brains in most available fossils (8, 10, 11). Moreover, the unresolved phylogenetic relationships within Panarthropoda complicate accurate reconstruction of the CNS ground pattern (16⇓⇓⇓⇓⇓–22). In this study, we demonstrate the exceptional preservation of postcephalic neurological features in the early Cambrian fuxianhuiid Chengjiangocaris kunmingensis, an upper stem-group euarthropod (17) from the Xiaoshiba Lagerstätte, South China (23). These fossils clarify the neurological organization of the VNC in early euarthropod ancestors, thereby polarizing the evolution of the panarthropod CNS.
Five individuals of C. kunmingensis display a narrow (maximum width, ∼170 µm) and slightly convex rope-like structure with a metameric pattern that extends medially throughout the body (Figs. 1 and 2, Figs. S1 and S2, and Table S1); its segmental organization and position at the ventral midline identifies this feature as the VNC. This interpretation is supported by comparisons with other preserved components of the internal anatomy. For instance, the VNC can be readily distinguished from the digestive tract of C. kunmingensis, which is expressed as a comparatively larger (maximum width, ∼860 µm) but fully compressed, linear structure running almost the entire length of the animal (23, figure 1 d and e). The VNC extends from at least the five anteriormost reduced trunk tergites (i.e., dorsal exoskeletal plates) to tergite T23 at the posterior end of the trunk (Figs. 1 and 2 and Figs. S1, S2, and S3). Although the VNC would have continued into the head region in life, our material does not preserve any anterior neurological structures that have been previously identified as the dorsal brain or nerves leading to the antennae and specialized postantennal appendages (8). The absence of fossilized brains in these otherwise exceptionally preserved specimens can be attributed to the small sample size and moderate postmortem disarticulation; likewise, other studies addressing neurological structures in Cambrian fossils rarely report the CNS preserved in its entirety (8⇓⇓–11). The VNC also expresses a distinctive dark/light color banding throughout its length, with the dark bands varying from black to reddish-brown between specimens (Figs. 1 and 2 and Figs. S1 and S2C); this variability in color reflects differences in the extent of weathering between individual specimens. Raman spectroscopy indicates the presence of residual organic carbon associated with the dark-colored bands (Fig. S4) but not in the light ones, likely mirroring differences in their original histology and early diagenesis. In addition to the carbonaceous films, the VNC is typified by modest relief relative to adjacent exoskeletal features, attesting to a degree of early diagenetic permineralization before complete degradational collapse (24).
VNC in C. kunmingensis. Anterior at top. (A) YKLP 12023 (Holotype), complete specimen preserved in dorsal view with taphonomically dissected head shield (hs) showing internal organization of anterior region. (B) YKLP 12324, complete specimen preserved in dorsolateral view showing preserved VNC extending throughout the almost the entire length of the body. (C) YKLP 12324, magnification of VNC on the anterior trunk region showing the differential preservation of the condensed ganglia (ga) and longitudinal connectives (cn) as dark and light colored bands, respectively, and one-to-one correspondence between the ganglia and the walking legs (wl) (upper box in B). (D) YKLP 12024, magnification of the VNC on the posterior trunk region showing the progressive reduction of size of the ganglia and connectives toward the rear end of the body (lower box in B). Dotted lines highlight the anterior and posterior tergal borders of T15 for comparative purposes with the number of preserved ganglia. ant, antennae; SPA, specialized postantennal appendage; Tn, trunk tergites.
Fine neurological organization of the VNC in C. kunmingensis, YKLP 12026. Anterior is to the left. (A) Completely articulated specimen preserved in laterodorsal orientation with displaced head shield exposing VNC on anterior trunk region. (B) VNC showing the presence of seven sets of condensed ganglia (ga) linked by longitudinal connectives (cn) (box in A). (C) Composite fluorescence photograph of VNC (box in A). (D) Magnification of the VNC (box in B) showing regularly spaced peripheral nerve roots (arrowheads) emerging from the condensed ganglia and connective. (E) Composite fluorescence photograph magnification of the VNC (box in C).
Detail of well-preserved VNC in C. kunmingensis, specimen YKLP 12026. (A) Preserved VNC with seven sets of condensed ganglia. (B) Magnification of four posterior ganglia and their respective connectives. (C) Magnification of three anterior ganglia and their respective connectives, including the fine preservation of regularly spaced peripheral nerve roots (arrowheads) emerging at either side of the VNC. Abbreviations as in Fig. 1.
VNC preservation in C. kunmingensis. (A) YKLP 12322, complete articulated specimen preserved in dorsolateral view with exposed VNC on the posterior trunk region. (B) YKLP 12324a, articulated specimen in lateral view with exposed VNC on posterior trunk region, showing correlation between condensed ganglia and the walking legs. (C) YKLP 12324b, preservation of VNC on anterior trunk region, showing light coloration due to the advanced degree of weathering (compare with Fig. 2). Abbreviations as in Fig. 1. hs, head shield.
Summary of specimens of C. kunmingensis (Cambrian stage 3, Xiaoshiba Lagersttäte) with preservation of neurological structures
Morphological reconstruction of C. kunmingensis. (A) Complete exoskeletal morphology in ventral view showing arrangement of walking legs (wl) and their attachment sites to the body (wlas) relative to the preserved VNC (purple) and tergites (Tn); note that only T1–T5 have a one-to-one correspondence with the walking legs. (B) Overall view of the CNS, including the VNC and dorsal brain (the latter extrapolated from Fuxianhuia protensa; sensu 8); the gap between the VNC and brain reflects lack of paleontological data pertaining this region. (C) Magnification of the CNS; note the one-to-one correlation between the ganglia (ga) and T1–T5, and the presence of up to four ganglia on the remaining trunk tergites. (D) Neurological reconstruction of two condensed ganglia. Other abbreviations as in Figs. 1 and 3.
Raman spectroscopy analysis of VNC in C. kunmingensis. (A) Carbon (1,360 cm−1) and organic carbon (1,604 cm−1) were detected in the dark stripes within the VNC of YKLP 12026. (B) No mineral or organic constituent are detected by Raman spectroscopy in the light areas within the VNC of YKLP 12026. (C) Carbon (1,370 cm−1) and organic carbon (1,603 cm−1) were detected in the dark stripes within the VNC of YKLP 12320. (D) No mineral or organic constituent detected by Raman spectroscopy in the light areas within the VNC of YKLP 12320.
Neurologically, the C. kunmingensis VNC consists of an interconnected series of separate metameric ganglia (dark colored bands) (Figs. 1 and 2 and Figs. S1 and S3). Longitudinal connectives are not morphologically discrete, but their presence is strongly suggested by the alternating light colored bands between the condensed ganglia. The distinct preservation style of these neurological structures likely stems from differences in their original histology; whereas the condensed ganglia are enriched in lipids due to the presence of abundant somata, the connectives consist of neurites and lack cell bodies (1, 3⇓–5). The ganglia have an elongate subelliptic outline (maximum length, ∼600 µm; 3.5:1 length/width) and are roughly three times longer than the connectives (maximum length, ∼206 µm). The ganglia remain separate throughout the entire VNC, with no indication of fusion or specialization. There is a progressive reduction in the size of individual ganglia along the body, with the anteriormost being ∼3 times longer and 1.5 times wider than the posterior ones (Fig. 1 C and D). The preserved proximal portions of the trunk endopods indicate that each ganglion was associated with a single pair of biramous appendages (Fig. 1C and Fig. S2 A and B), which, like the ganglia, become progressively smaller posteriorly (Fig. 1D). As such, each of the five anteriormost reduced tergites (Fig. 1A) correlates with an individual ganglion and leg pair, whereas the comparatively larger tergites T6–T26 overlie up to four ganglia and their corresponding appendages (Fig. 1D and Figs. S2 A and B and S3). This organization reflects the notable dorsoventral segmental mismatch that typifies the fuxianhuiid trunk (23, 25).
The unique disarticulation pattern of the head shield in the Xiaoshiba fuxianhuiids (23)—aided by mechanical preparation—reveals submillimetric neurological detail in C. kunmingensis. In specimen YKLP 12026, the exposed VNC extends from the oral region into tergite T6, comprising a total of seven ganglia (Fig. 2 A–C and Fig. S1). Close inspection reveals the presence of delicate nerve roots (maximum length, ∼209 µm; width, 9 µm) emerging at both sides of the VNC. These nerve roots are regularly spaced (∼17–20 µm) and originate from both the ganglia (i.e., segmental) and interganglia connectives (i.e., intersegmental) at acute angles (∼40–60°) relative to the VNC (Fig. 2 C–E and Figs. S1 B and C and S3). The emergent chevron-like pattern most likely resulted from moderate postmortem displacement of the distal portions of the peripheral nerves relative to the VNC. There are at least a dozen paired sets of roots associated with each ganglion and four to six nerve roots associated with the longitudinal connectives. Preferential preservation of the proximal bases of the nerve roots precludes the identification of precise relationships between these structures and the exoskeletal morphology, such as the walking legs. The lack of distal neurological preservation suggests that, similarly to 3D guts in Burgess Shale fossils (24), the VNC and the proximal nerve roots were permineralized during early diagenesis due to the chemically reactive composition of adjacent lipid-rich ganglia (8). Fluorescence microscopy corroborates the anatomical and compositional continuity of the regularly spaced nerve roots and VNC throughout the body (Fig. 2 C and E and Dataset S1).
The Xiaoshiba fossils provide previously unidentified insights on the neurological diversity of Cambrian euarthropods. Within the context of extant representatives, the postcephalic CNS of C. kunmingensis exhibits notable similarities with the thoracic VNC of the notostracan Triops cancriformis, which consists of a homonymous series of separate ganglia linked by short connectives that become progressively smaller toward the posterior end (1, 3). The overall morphology of T. cancriformis further resembles C. kunmingensis—and generally other fuxianhuiids—in the presence of dorsoventral trunk segmental mismatch expressed as appendage polypody (23, 25, 26). A similarly unspecialized VNC is also found in crustaceans typified by a trunk region with undifferentiated limb-bearing segments, such as the remipedes (5, 7) (Fig. S5C). Unlike C. kunmingensis and T. cancriformis, however, the body organization of remipedes does not evince any type of trunk segmental mismatch. Thus, the paleoneurological data available indicate that the fuxianhuiid CNS combines a relatively simple VNC without signs of functional specialization (Figs. 1–3) with a tripartite brain bearing both olfactory and optic lobes (8). Considering the position of fuxianhuiids within upper-stem Euarthropoda (17), this character combination could suggest the evolutionary convergence of either the VNC of Chengjiangocaris with Branchiopoda or the complex brain of Fuxianhuia with Malacostraca. Alternatively, the fuxianhuiid CNS may approximate the complex brain (sans optic lobes) and unspecialized VNC of remipedes (5, 7). If the neurological organization of fuxianhuiids reflects an ancestral condition as suggested by their phylogenetic position, it would support the symplesiomorphic nature of a malacostracan-like brain for Pancrustacea/Tetraconata (6, 8) and also imply an independent increase in VNC complexity among extant crustacean lineages. However, the lack of consensus pertaining to the phylogenetic placement of Branchiopoda, Remipedia, and Malacostraca relative to Hexapoda based on molecular and neuromophological data (5⇓–7, 27) complicates deciding between these hypotheses based on the available paleoneurological information.
Simplified cladogram showing the evolution of the postcephalic CNS in Panarthropoda. Detailed results of the phylogenetic analysis are provided in Fig. S6 and SI Text. The topology supports a single origin for the condensed ganglia (ga) in the VNC in a clade including Tardigrada and Euarthropoda; note that the presence of multiple intersegmental peripheral nerves (ipn) in C. kunmingensis represents an ancestral condition. Given the morphological similarity between peripheral and leg nerve roots, the presence of a single pair of leg nerves (lgn) in C. kunmingensis is hypothetical (dashed lines) and based on the condition observed in crown-group Euarthropoda. †, fossil taxa; ?, uncertain character polarity within total-group Euarthropoda. asn, anterior segmental nerve; cn, longitudinal connectives; co, commissure; dln, dorsolateral longitudinal nerve; ico, interpedal median commissure; irc, incomplete ring commissure; pn, peripheral nerve; psn, posterior segmental nerve; rc, ring commissure. Reconstruction of VNC in Onychophora adapted from ref. 13.
Diversity of VNC organization in extant Panarthropoda. (A) E. kanangrensis (Peripatopsidae, Onychophora). VNC confocal micrograph stained with a monoclonal antibody directed toward acetylated tubulin (Sigma). Reprinted with permission from ref. 29. (B) M. cf harmsworthi (Eutardigrada, Tardigrada). VNC confocal micrograph stained with combined antityrosinated and antiacetylated α-tubulin immunolabeling. Reprinted with permission from ref. 16. (C) S. tulumensis (Remipedia, Euarthropoda). VNC confocal micrograph labeled for acetylated α-tubulin immunoreactivity (TUB-IR, yellow). Reprinted with permission from ref. 5. (D) S. gregaria (Hexapoda, Euarthropoda). VNC confocal micrograph stained with 8b7 immunocytochemistry (46). Abbreviations as in Figs. 1 and 3.
The neurological organization of the fuxianhuiid VNC leads to more informative comparisons with other panarthropod groups. Apart from Euarthropoda (1⇓⇓⇓–5, 9), the presence of segmental ganglia is also characteristic of the CNS in Tardigrada (18, 28) (Fig. S5B) and has been used to advocate the sister-group relationship between these clades (16, 17, 20). Thus, the presence of segmental ganglia in C. kunmingensis provides a minimum phylogenetic threshold for the evolution of this feature in total-group Euarthropoda (SI Text).
At a broader phylogenetic level, the neurological organization of C. kunmingensis also exhibits marked similarities with the VNC of Onychophora (Fig. 3 and Fig. S5A), namely the presence of numerous regularly spaced peripheral nerve roots (12⇓⇓–15, 29, 30). A major difference, however, is that the onychophoran postcephalic CNS is lateralized and lacks morphologically discrete segmental ganglia despite clearly having a segmented organization (SI Text). Although the proximal preservation of the peripheral nerves makes it uncertain whether the C. kunmingensis VNC possessed complete ring commissures like those in onychophorans (12⇓–14), this feature could potentially add morphological support for the sister-group relationship between Euarthropoda and Onychophora suggested by molecular phylogenies (21, 22). Among crown-group euarthropods, there are only a restricted number of intersegmental peripheral nerves emerging from the longitudinal connectives in the VNC of some pancrustaceans (1, 4, 5) (SI Text, Fig. 3, and Fig. S5 C and D). Alternatively, it is possible that the presence of numerous peripheral nerves in the VNC of C. kunmingensis and Onychophora is symplesiomorphic, as comparable structures are also found in the orthogonal CNS of Priapulida (31) and other protostomes (12, 13, 15). In any event, the postcephalic CNS of C. kunmingensis reveals a unique combination of neurological characters otherwise unknown in any extant group within Panarthropoda (Fig. S3).
To test these different hypotheses and to clarify the polarity of neurological characters in extant and extinct groups, we performed a comprehensive phylogenetic analysis incorporating the available fossil data on the postcephalic CNS of total-group Panarthropoda (SI Text and Dataset S2). The results support the sister-group relationship between Tardigrada and Euarthropoda (Fig. 3 and Figs. S6 and S7), corroborated by several unambiguous synapomorphies of the VNC. These synapomorphies include the presence of condensed segmental ganglia connected by commissures, the anterior shift of the leg nerves following a parasegmental organization, and the presence of a stomatogastric ganglion associated with the tritocerebral segment (SI Text) (14, 16, 18, 19, 28). In this context, the presence of numerous segmental and intersegmental peripheral nerves in C. kunmingensis (Fig. 2 B–E) is recognized as an ancestral condition that persisted in derived members of upper stem-group Euarthropoda (SI Text). Among extant groups, homologs of the intersegmental peripheral nerves are expressed in both Priapulida (31) and Onychophora (12, 13, 15, 29) but are otherwise greatly reduced in number or completely lost in crown-group Euarthropoda and Tardigrada, respectively (1, 5, 16, 28) (Fig. S5); the secondary reduction/loss is recognized as a result of parallel evolution between these lineages. The phylogenetic analysis also indicates that several neurological characters shared between Onychophora and Tardigrada are actually symplesiomorphic and likely reflect the broader ground pattern of Panarthropoda (Fig. 3); these include the median interpedal commissures, an orthogonal-like organization, and paired segmental leg nerves (SI Text). A lateralized VNC is resolved as synapomorphic for Onychophora (SI Text). Taken together, these findings argue against the secondary loss of morphologically discrete segmental ganglia in the VNC of Onychophora (30). By contrast, the evolution of the VNC in crown-group Euarthropoda is the result of secondary simplification relative to the CNS of cycloneuralians and other panarthropods, as informed by the transitional neurological organization of C. kunmingensis (Fig. 3).
Summary of results from phylogenetic analyses. See SI Text and Dataset S2. (A) Strict consensus of 546 most parsimonious trees (MPTs) under equal weights [178 steps; consistency index (CI) = 0.65; retention index (RI) = 0.88]. (B) Strict consensus of 160 MPTs under implied weights (k = 0.1; CI = 0.64; RI = 0.87); this topology is stable when 1 ≤ k ≥ 0.1, with some variation in the number of MPTs. (C) Strict consensus of 146 MPTs under implied weights (k = 3; CI = 0.65; RI = 0.88); this topology is stable when 10 ≤ k ≥ 3, with some variation in the number of MPTs. (D) Strict consensus of 58 MPTs under implied weights (k = 20; CI = 0.65; RI = 0.88); this topology is stable when k ≥ 20, with some variation in number of MPTs. Color coding: stem-group Panarthropoda (dark pink), total-group Onychophora (green), total-group Tardigrada (red), total-group Euarthropoda (blue).
Evolution of the nervous system in Panarthropoda under an alternative phylogenetic hypothesis grouping Onychophora and Euarthropoda as sister taxa. Note that that these scenarios for the evolution of the panarthropod nervous system are less parsimonious compared with the topology obtained in the present study (Fig. S6). (A) Evolutionary scenario favoring substantial convergent evolution of the VNC in Tardigrada and Euarthropoda. (B) Evolutionary scenario favoring the secondary simplification of the neurological organization in Onychophora, derived from ancestors with a ganglionated VNC (30). †, fossil taxa; ?, uncertain neurological character polarity within total-group Euarthropoda.
The integration of paleoneurological data with the present understanding of the developmental biology of extant groups illuminates major issues pertaining the evolution of the nervous system in Panarthropoda. Our results indicate that the ground pattern of the panarthropod postcephalic CNS incorporates a paired, but not lateralized, VNC with median interpedal commissures, an orthogonal organization with complete ring-commissures, paired leg nerves, and intersegmental peripheral nerves (SI Text); in this context, an orthogonal VNC with complete ring commissures is a symplesiomorphy inherited from a cycloneuralian-like ancestor. The presence of condensed segmental ganglia, a lateralized organization, anteriorly displaced leg nerves following the parasegments, and a stomatogastric ganglion (SI Text) (Fig. 3), represent derived features that were acquired later in the evolutionary history of the different panarthropod phyla.
All of the described material is deposited at the Key Laboratory for Paleobiology, Yunnan University, Kunming, China (YKLP).
Fossils were photographed with a Nikon D3X fitted with a Nikon AF-S Micro Nikkor 105-mm lens. For close-up images, a LEICA M205-C stereomicroscope fitted with a Leica DFC 500 digital camera was used under directional illumination provided by a LEICA LED5000 MCITM. Geochemical analyses were performed with an inVia Raman microscope (Renishaw). Fluorescence photography was performed as described in ref. 32.
The data matrix includes 50 taxa and 95 characters (Dataset S2 and SI Text). The analysis was run in TNT (33) under New Technology Search, using Driven Search with Sectorial Search, Ratchet, Drift, and Tree fusing options activated in standard settings (34, 35). The analysis was set to find the minimum tree length 100 times and to collapse trees after each search. All characters were treated as unordered. For an initial analysis, all characters were treated as equally weighted (Fig. S6A); subsequent repetitions with variable concavity values (k) were used to explore the effect of different degrees of homoplasy penalization to test the robustness of the dataset (Fig. S6 B–D) (36).
The dataset used for the phylogenetic analysis has been updated from that presented by Yang et al. (36), including the formulation of new neurological characters to resolve large-scale relationships within Panarthropoda. The crown-group euarthropods Limulus polyphemus (Chelicerata, Xiphosura) and Triops cancriformis (Mandibulata, Notostraca) were included as their neurological organization has been extensively studied (1⇓–3, 37) and to appropriately polarize the characters in the analysis. Additional extant taxa were included based on recent studies describing aspects of the brain and VNC organization; these include the eutardigrades Macrobiotus cf harmsworthi and Hypsibius dujardini (16, 18), the heterotardigrades Echiniscus testudo, Actinarctus doryphorus, and Batillipes pennaki (28, 38), the peripatopsid Metaperipatus blainvillei, and the peripatid Epiperipatus biolleyi (12, 13, 15). Fossil and extant taxa are scored according to a single model of head segmental organization that is informed by developmental studies on extant Onychophora, Tardigrada, and Euarthropoda (36).
Characters 1–80 largely follow those of Yang et al. (36); only those with minor modifications are outlined here. Characters 81–86 from Yang et al. (36) have been substantially revised to integrate the paleoneurological data available from C. kunmingensis and better reflect the diversity of VNC morphology in Panarthropoda; these changes are described below.
Characters modified from Yang et al. (36).
(0) lobopodous ambulatory limb
(1) lobopodous sensorial limb
(2) lobopodous limb with sclerotized jaw
(3) arthropodized antenniform with distinct podomeres
(4) arthropodized short great-appendage/chelicerae
(−) inapplicable: paired appendages (character 1) absent.
This character is scored for as present for the deutocerebral chelicerae of Limulus (2, 37) under the proposed evolutionary transition from the short-great appendages of megacheirans into the first appendage of crown-group Chelicerata (9).
(0) undifferentiated lobopodous limb
(1) specialized papillae
(2) arthropodized biramous walking leg with distinct podomeres
(3) arthropodized uniramous specialized postantennal appendage
(4) arthropodized biramous second antennae
(−) inapplicable: paired appendages (character 1) absent.
State 4 is introduced to reflect the appendicular morphology of the notostracan Triops (3, 26, 39).
(0) lanceolate dorsal blades
(1) simple oval paddle with marginal spines
(2) bipartite shaft with lamellar setae
(3) numerous podomeres, each bearing a single setae
(4) book gills
(−) inapplicable: trunk exites (character 53) absent.
States 3 and 4 are introduced to reflect the exite morphology of the notostracan Triops (26, 39) and the xiphosuran Limulus (40), respectively (41, character 10).
(0) one
(1) two
(2) three
(3) four
(4) six
(5) seven
(−) inapplicable: terminal claws (character 64) present.
State 4 is introduced to reflect the presence of six toe-like claws in the heterotardigrade Batillipes pennaki (28).
(0) one
(1) two
(2) three
(−) inapplicable: dorsal condensed brain (character 82) absent.
See character 83 in Yang et al. (36) and also introductory statements above. Limulus and Triops are scored as having three brain neuromeres based on ample neurological data on these model organisms (2, 3, 37).
(0) protocerebral innervation
(1) deutocerebral innervation
(2) tritocerebral innervation
(3) innervation from multiple neuromeres
(−) inapplicable: condensed dorsal brain (character 82) absent.
See character 84 in Yang et al. (36). State 2 has been included to reflect the innervation pattern of Triops, in which the stomatogastric nervous system is associated with the tritocerebral neuromere (3).
(0) unpaired
(1) paired
The general organization of the ventral nerve cord in Ecdysozoa can be broadly classified based on whether it is paired or unpaired. Within the scope of the present phylogenetic analysis, an unpaired nerve cord is characteristic of Priapulida (31), whereas a paired nerve cord is found in all members of Panarthropoda (1, 3, 12, 13, 16, 18, 19, 28, 29, 38, 42, 43) (Fig. S5). In fossil representatives, a paired ventral nerve cord has been identified in Alalcomenaeus sp. (9) and is confirmed for C. kunmingensis in this study (Figs. 1 and 2). A paired VNC is scored as present for L. unguispinus based on the presence of two descending tracts on the anterior trunk region (10). Although Hou et al. (44) have reported the presence of a VNC in the Chengjiang lobopodian Paucipodia inermis, it is not possible to discern whether the organization of this feature is paired or unpaired, and thus it is scored as uncertain pending a formal description of the putative nervous system in this taxon.
(0) absent
(1) present
Tardigrada and Euarthropoda have a rope ladder-like VNC with morphologically discrete condensed ganglia (1⇓⇓–4, 16, 18, 28, 37, 43) (Fig. S5 B–D), in contrast to the ladder-like VNC of Onychophora (12, 13, 16, 29, 38) (Fig. S5A). The presence of median commissures has been recently demonstrated in Tardigrada (16, 19, 28) and thus suggests that the possession of ganglia and transverse commissures represent fundamentally linked neurological features (45). Priapulida have an unpaired nerve cord associated with a net-like system of neural connectives (31), and thus this character is scored as absent. The present data confirm the presence of condensed ganglia in C. kunmingensis (23). The early Cambrian megacheiran Alalcomenaeus sp. also possesses a VNC with condensed ganglia (9) Condensed ganglia are scored as uncertain in Lyrarapax (10), as the preservation of this taxon does not allow the organization of the VNC to be resolved. The lobopodian P. inermis has been described as possessing condensed ganglia (45). Given that the veracity of these observations has yet to be confirmed by additional study of the material, this character is treated as uncertain in the main analysis (Fig. S6). However, we explored the potential implications of putative ganglia in P. inermis by running a second set of analyses that scored this character as present. Under equal weights and implied weights with k ≥ 20, the analyses produced identical topologies as the original tests (Fig. S6 A and D). However, under implied weights with parameters 20 < k ≥ 0.1, P. inermis was recovered within stem-group Tactopoda (i.e., basal to total-group Tardigrada and total-group Euarthropoda) instead of within stem-group Onychophora. The latter topology ultimately supports a single origin for the condensed ganglia in Panarthropoda, as informed by our original analyses, and carries the same general implications for the evolution of CNS in this clade (Fig. 3 and Fig. S6).
(0) absent
(1) present
(−) inapplicable; VNC unpaired (character 85).
This character reflects the organization of the ladder-like paired VNC of Onychophora (12 ,13, 15, 16, 29, 38) (Fig. S5A). This character is scored as uncertain in Lyrarapax (10), as the preservation of the postcephalic nervous system does not allow the organization of the VNC to be resolved.
(0) absent
(1) present
(−) inapplicable; VNC unpaired (character 85).
Median interpedal commissures are conspicuous in the ladder-like VNC of Onychophora (12, 13, 16, 29, 38) (Fig. S5A). Recently, Mayer et al. (16) described in detail the presence of transverse interpedal median commissures that fall outside of the condensed ganglia in the nerve cord of Eutardigrada (Fig. S5B), suggesting that this represents a plesiomorphic condition reminiscent of the possibly orthogonal organization hypothesized for an ancestral panarthropod. Schulze and Schmidt-Rhaesa (42) and Schulze et al. (28) have reported similar nonganglion associated commisures in the Heterotardigrada, indicating that this represents a widespread condition within Tardigrada. The VNC of Euarthropoda lacks signs of interpedal commissures (1, 2, 16, 37, 46) (Fig. S5 C and D); this character is scored as absent for Alalcomenaeus sp. (9) and scored as uncertain in Lyrarapax (10) and C. kunmingensis (23 and this study), as it is not possible to resolve the preservation of interpedal commissures in the available material.
(0) absent
(1) present
The nerve cord of Priapulida (4, 31) and Onychophora (12, 13, 15, 16) displays a typical orthogonal organization, consisting of several ring-like commissures and peripheral nerves that intersect with additional dorsal and lateral longitudinal nerve strands, forming a distinctive net-like pattern. The nervous system of Tardigrada also displays dorsal and lateral longitudinal nerves that are intersected by transverse peripheral nerves (16, 18, 38, 43), although these organisms lack complete ring commissures (see character 90). Schulze and Schmidt-Rhaesa (28) reported loop-like neurites that extend dorsally in E. testudo; this organization is similar to that observed in eutardigrades (16), and thus this character is scored as present. However, an orthogonal-like organization has not been resolved in the heterotardigrades B. pennaki and A. doryphorus (28); pending additional data, this character is scored as uncertain as it is unknown whether this absence of legitimate or whether this aspect of the morphology has not been resolved due to differences in specimen fixation and/or ontogeny (41, p. 51). Although the presence of peripheral nerves (see character 93) throughout the VNC of C. kunmingensis may suggest the presence of an orthogon-like organization, as expressed in Onychophora, this character is conservatively scored as uncertain due to the lack of distal preservation of the nerve roots.
(0) absent
(1) present
(−) inapplicable; VNC organization is not orthogonal (Character 90).
This character distinguishes the complete ring commissures of Priapulida and Onychophora (12, 13, 15, 31) from the incomplete peripheral commissures that emerge from the condensed ganglia in the nerve cord of some tardigrades (16, 18, 42). This character is scored as uncertain for C. kunmingensis given the incomplete distal preservation of the peripheral nerves (character 93).
(0) absent
(1) present
(−) inapplicable; paired appendages absent (character 1).
The leg nerves emerging from the condensed ganglia in the VNC in Tardigrada and Euarthropoda evince an organization that reflects the position of the parasegmental boundaries in these organisms, resulting in an anterior displacement relative to the actual position of the trunk appendages (16, 28, 42). By contrast, neuronal tracing indicates that this anterior displacement is not expressed in Onychophora (16). Although the orientation of the peripheral nerves in the ganglia of C. kunmingensis could potentially suggest the anterior displacement of the leg nerves relative to the appendages in fuxianhuiids, this character is conservatively scored as uncertain as it is not possible to confirm this condition given the lack of distal preservation of the nerve roots.
(0) absent
(1) present
(−) inapplicable; paired appendages absent (character 1).
This character describes the condition observed in onychophorans and tardigrades, in which each leg is innervated by two nerves (12, 13, 16, 18, 28, 42, 45); by contrast, a single nerve innervates each leg in Euarthropoda. This character is scored as uncertain for C. kunmingensis as the incomplete preservation of the nerves in the VNC does not allow resolving their precise relationship with the leg innervation.
(0) absent (Tardigrada, crown-group Euarthropoda)
(1) present (Priapulida, Onychophora, Chengjiangocris)
Regularly spaced peripheral nerves are characteristic of the orthogonal nervous system of Priapulida (31) and Onychophora (see “interpedal nerves” in refs. 12, 13, 15, and 16) (Fig. S5A). Although the VNC of some crown-group euarthropods displays the so-called intersegmental nerves outside of the segmented ganglia (Fig. S5 C and D), this character is scored as absent as these nerves do not occur regularly throughout the entire length of the corresponding connectives (1, 4, 5, 7, 37, 47). Intersegmental peripheral nerves are entirely absent in Tardigrada (16, 28) (Fig. S5B). The exceptional preservation in C. kunmingensis demonstrates the persistence of segmental and intersegmental peripheral nerves throughout the length of the VNC in representatives of upper stem-group Euarthropoda (Figs. 1 and 2 and Fig. S1).
(0) absent
(1) present
A stomatogastric ganglion associated with the segment bearing the (tritocerebral) second leg pair has been recently described for various species within Eutardigrada, including M. cf harmsworthi and H. dujardini (16, 18, 19); this structure has been interpreted as potentially homologous with the stomatograstic ganglion of Euarthropoda [2, 3, 46; see refs. in Mayer et al. (19)], and hypothesized as a potential synapomorphy of Tardigrada and Euarthropoda. However, neurological studies of Heterotardigrada have been unable to find evidence for the presence of the stomatogastric ganglion (28, 42); this character is scored as uncertain for Heterotardigrada pending the input of additional neurological data. This character is scored as uncertain for C. kunmingensis and all other fossil taxa.
(0) absent
(1) present
(−) inapplicable; dorsal sclerotized integument (character 33) present.
Cirri are spine-like cuticular projections that represent a distinguishing feature of the anterior region in Heterotardigrada (28).
The results of the cladistic analysis provide strong support for the sister-group relationship of Tardigrada and Euarthropoda (i.e., Tactopoda), to the exclusion of Onychophora (20, 36); this topology is retrieved in the equally weighted analysis (Fig. S6A) and throughout a broad range of homology penalization (Fig. S6 B–D). The stability of Tactopoda allows reconstructing the evolutionary history of the VNC within Panarthropoda in detail (Fig. 3) and indicates that mapping the neurological data (characters 82–94) to match the sister-group relationship between Onychophora and Euarthropoda (21, 22) results in a less parsimonious reconstruction of character transformations (Fig. S7A). More specifically, the latter scenario would require the convergent evolution of several neurological features in Tardigrada and Euarthropoda, namely the presence of condensed ganglia in the nerve cord, the anterior shift of the leg nerves following a parasegmental organization, and the presence of a stomatogastric ganglion (Characters 86, 91, and 94). An alternative scenario in which all these features are symplesiomorphic for Panarthropoda would favor the substantial secondary simplification of the VNC in Onychophora (30) (Fig. S7B); however, the lack of clear developmental or neurological data supporting the loss of all these complex features in the CNS of Onychophora argues against this evolutionary scenario (12, 13, 15).
The variability observed between the different analyses indicates that phylogenetically basal lobopodians within the stem lineages of Onychophora and Euarthropoda represent rogue taxa. Aysheaia pedunculata, Onychodictyon ferox, Jianshanopodia decora, and Megadictyon haikouensis are variably recovered as either the most basal members of stem-group Euarthropoda (Fig. S6 B and C) or in a less resolved position within stem-group Panarthropoda (Fig. S6 A and D); only at high concavity values is O. ferox resolved as basal within stem-group Onychophora (Fig. S6D). This variability reflects that of previous analyses (20, 36) and indicates that additional material is necessary to better constrain the morphology and phylogenetic affinities of these lobopodians. By contrast, taxa that occupy a more crown-ward phylogenetic position evince a substantial degree of stability within each of the total-groups, even though it is not possible to resolve the internal relationships within Luolishaniidae (stem-group Onychophora) and Radiodonta (stem-group Euarthropoda) with the present dataset. Within stem-group Onychophora, the basal lobopodians Paucipodia inermis, Diania cactiformis, and Onychodictyon gracilis evince some plasticity in their position. It is likely that the formal description of the putative VNC in P. inermis (44) will help to elucidate its precise affinities and further inform the early evolution of the nervous system in early onychophoran ancestors. The internal topology of total-group Tardigrada is stable in all of the analyses and differs from previous studies in that O. ferox was not recovered within this group under any degree of homology penalization. The analyses further allowed to recover a monophyletic Heterotardigrada but were unable to resolve the phylogenetic position of the Orsten Siberian tardigrade (48) relative to the crown-group, despite the overall eutardigrade like appearance of this fossil, including the presence of claws with multiple branches. Finally, the topology within total-group Euarthropoda is largely stable, save for the fluctuating positions of A. pedunculata, O. ferox, J. decora, and M. haikouensis discussed before. The only additional variation occurred in the position of the gilled lobopodian Pambdelurion whittingtoni; whereas in most analysis this taxon was recovered as crown-ward relative to Kerygmachela and stem-ward relative to a clade including Radiodonta and Deuteropoda, at low concavity values (Fig. S6B), P. whittingtoni was resolved as sister-taxon to Radiodonta. The relationships between other members of stem-group Euarthropoda are consistent with previous analyses (17, 20, 49), including the recent finding that Opabinia regalis may occupy a more crown-ward position relative to Radiodonta than previously considered (36).
We thank K.-S. Du and J.-F. He for assistance with fossil collection. B. J. Eriksson (University of Vienna), G. Mayer (University of Leipzig), and G. Bicker (University of Veterinary Medicine Hannover) generously contributed photographic material for Fig. S5. This work was supported by National Natural Science Foundation of China (NSFC) Grants 41472022 and U1402232 (to J.Y. and X.-g.Z.), a research fellowship at Emmanuel College and a Herchel Smith fellowship (both University of Cambridge; to J.O.-H.), and a Ludwig Maximilians Universität München excellent Junior Researcher fund and NSFC Grant 41528202 (to Y.L.).
↵1Present address: Department of Zoology, University of Cambridge, Cambridge CB2 3EJ, United Kingdom.
Author contributions: J.Y., N.J.B., and X.-g.Z. designed research; J.Y., J.O.-H., N.J.B., Y.L., J.-b.H., T.L., and X.-g.Z. performed research; Y.L. and G.S.B. contributed new reagents/analytic tools; J.O.-H. analyzed data; J.O.-H. and N.J.B. wrote the paper; J.Y. collected and prepared all the fossil material; J.O.-H. performed light photography; N.J.B. and X.-g.Z. discussed and approved the manuscript; Y.L. and G.S.B. performed immunohistochemistry and living animal microscopy; and J.-b.H. and T.L. collected fossil material and performed photography.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission. G.D.E. is a guest editor invited by the Editorial Board.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1522434113/-/DCSupplemental.
