The commissural system in pycnogonid ganglia and other arthropod taxa
Commissural tracts can be used as landmarks that help to characterize identified neurons
with contralateral projections in more detail. The fasciculation of a contralaterally
projecting neurite with a specific commissural tract provides additional information
for homologization with other neurons. In case of the SL-ir neurons in the arthropod
VNC, contralateral projections of supposedly homologous neurons have been repeatedly
assigned either to the anterior or the posterior commissure (e.g. 10], 12], 15], 26]). However, such a simple two-commissure-system per segmental neuromere is not consistently
found throughout arthropods. On the contrary, it is relatively rare in adult animals
(e.g. Branchiopoda 24]; Mystacocarida 49]; Remipedia 26]) and many other taxa exhibit either more complex commissural systems (see 27], 50]–53], this study) or the architecture remains still unsatisfactorily resolved (e.g. in
Myriapoda). Accordingly, a segmental two-commissure-system is not an ubiquitous arthropod
feature and cannot be confidently advocated as plesiomorphic state of the adult VNC.
Similar disparities are encountered during development of the commissural system.
Especially in tetraconates, a distinct two-commissure-system is frequently recognizable
during segmental axonogenesis 54]–56]. But yet again, neither does this apply to all tetraconate representatives (e.g.
16], 57]), nor does it hold for the studied euchelicerates 19], 58] and apparently also not for myriapods 9], 14]. In pycnogonids, there is only a single rather than two embryonic commissures 8], 59] and it is only during post-embryonic development that this primary commissural pathway
is further elaborated into the numerous separate transverse tracts of the adult system.
Hence, the more complex adult commissural systems of pycnogonids, euchelicerates and
supposedly myriapods cannot be convincingly derived from a transient two-commissure
system in development, as may be possible for many tetraconates 4], 60].
As a consequence, when identifying and homologizing contralaterally projecting SL-ir
neurons in intra-specific comparisons, the exact neurite pathway through specific
commissural tracts is beyond doubt very useful. However, whenever dealing with inter-specific
neuron homologization across divergent taxa, correspondences between the investigated
commissural systems should be ascertained first, before a comparison of the exact
course of single contralateral neurite projections makes sense. Strictly speaking,
only contralateral projections through corresponding tracts within the compared commissural
systems can add additional support to inter-specific neuron identity and thus strengthen
a homology hypothesis. In this regard, available data on pycnogonids, euchelicerates
and myriapods remain still unsatisfactory, which hampers arthropod-wide comparisons.
Intra-specific comparison of SL-ir neurons in the pycnogonid walking leg ganglia
In each of the studied pycnogonid species, different SL-ir neuron types have been
reliably identified in the four walking leg ganglia of single individuals as well
as across different specimens. Some of these neurons have been found to be intra-specifically
strictly stereotypic (DMNs, cAMN, iAMNs), whereas others show variations of cell numbers
(ALNs and PLNs) or were not even reliably detected in each species (VMN and PVNs)
(Fig. 9; Table 1).
Fig. 9. Schematic representation of SLI pattern in the investigated pycnogonids and inferred
minimal pattern in the last common ancestor. The left column shows the largely corresponding
SLI pattern of the ventral half of all four walking leg ganglia, the middle column
depicts the pattern characteristic for the dorsal half of walking leg ganglia 2–4,
and the right column for the slightly deviating dorsal half of walking leg ganglion
1 only. All neurons shown in black have been consistently identified in the respective
species. Neurons in gray were only inconsistently identified. In the case of the ALNs
of P. litorale, five neurons are shown in black (minimum number observed) and the additional gray
neurons indicate that higher numbers are frequently observed, but no clearly fixed
maximum cell number was assessable. The lowest row illustrates the reconstructed minimal
pattern of the common ancestor, including only those neuron types and neuron numbers
that have been reliably recovered and therefore homologized across the three investigated
pycnogonid species
At the moment, it remains unresolved whether the observed intra-specific differences
relate to actual variability in the cell number of the respective SL-ir neuron type.
This is because several of the neurons show whenever labeled very low signal intensity
and we cannot fully exclude that non-detection of them may be related to a less successful
immunolabeling. Furthermore, the applied protocol does not allow elucidating whether
some of the neurons in question show significant temporal variations in serotonin
synthesis and content. This might lead to the detection of only a sub-set of SL-ir
neurons of ganglia and individuals (see 61]). Similar uncertainties relating to weakly labeled neurons have been encountered
in previous studies (e.g. 26], 62]). However, regardless of the cause of variation in pycnogonids, it is notable that
the observed variability is – with the exception of the ALNs in P. litorale – restricted to the absence of one or two cells of a specific SL-ir neuron type.
It does not relate to neuron clusters of generally higher cell numbers – a feature
considered as being typical for euchelicerate taxa (see below).
Inter-specific comparison of SL-ir neurons and the SLI pattern of the last common
ancestor
Beyond intra-specific correspondences, striking inter-specific similarities in terms
of soma position and neurite projection patterns are assessable for most of the detected
SL-ir neuron types. This holds in particular for the DMNs, cAMN, iAMNs and PLNs, but
to some extent also for the ALNs (Fig. 9; Table 1). These correspondences strongly support the homology of the neuron types across
the three species.
Interestingly, the inter-specific similarities do even extend to some pattern variations
between different WLGs. The most eye-catching instance is the consistent lack of a
second DMN in WLG1 as compared to WLG2-4. Likewise, WLG1 of C. japonicus and Meridionale sp. shows a corresponding pattern of only one instead of two dorsal ALNs, as well
as only three instead of four iAMNs (Fig. 9; Tab. 1).
The discovery of considerable similarities in the SLI pattern of all three species
leads to the question of the ancestral pattern of Pycnogonida. A handful of phylogenetic
analyses performed mainly during the last decade have sought to resolve the internal
phylogeny of extant pycnogonid taxa 63]–67]. Yet, whereas some hypotheses based on earlier ideas (e.g. 68], 69]) could be clearly rejected, an uncontested stable pycnogonid phylogeny has still
not emerged. For this reason, we here refrain from proposing any of the three species-specific
SLI patterns to represent a more ancestral or derived state within crown-group pycnogonids.
However, despite the lack of a reliable internal phylogeny, a comparison between the
investigated species can nonetheless yield a provisional minimal SLI pattern for their last common ancestor, which may have been the stem species
of crown-group pycnogonids (depending on which of the phylogenetic hypotheses is favored;
see Fig. 10).
Fig. 10. Phylogenetic position of the last common ancestor of the three investigated pycnogonid
species. Simplified phylograms of the two most comprehensive phylogenetic analyses
of internal pycnogonid relationships (For simplicity, Nymphon floridanum, which renders Nymphonidae polyphyletic in Arango and Wheeler (2007), has been omitted).
Depending on which hypothesis is favored, the last common ancestor (green arrows) of the three investigated species is the stem species of crown-group pycnogonids
(Arango and Wheeler 2007, left side) or of all extant pycnogonids to the exclusion of Austrodecidae (Arabi et al. 2010,
right side). The groups highlighted in green include the three species investigated in this
study
If an identified SL-ir neuron type is present in the walking leg ganglia of all three
species, it seems reasonable to assume that this neuron has already been present in
the walking ganglion of their last common ancestor. Accordingly, we propose that DMNs,
cAMN, iAMNs, ALNs, and PLNs have been part of the ancestral repertoire. In cases of
intra- and/or inter-specific variations of neuron numbers (iAMNs, ALNs, PLNs), we
suggest the lowest consistently observed number to be provisionally put into the ancestral
pattern (Fig. 9). A very conspicuous instance of such variation is found in the ALNs, which appear
in clusters of variable cell numbers in P. litorale but only as single neurons or neuron pairs in Meridionale sp. and C. japonicus. In line with our minimal approach, we place in this case only a single rather dorsally
positioned ALN into the ancestral pattern (Fig. 9).
As the pattern of SL-ir neurons in WLG1 consistently deviates in some aspects from
the one in WLG2-4 (see above), the reconstructed ancestral pattern of WLG1 is also
slightly different (Fig. 9).
SL-ir neurons in Euchelicerata versus Pycnogonida
Recent phylogenetic analyses tend to favor Pycnogonida as the sister group to Euchelicerata
[e.g. 70]–73] , see also 42], 74]). In euchelicerates, the SLI pattern in the VNC has been studied only in few representatives,
including spiders (Cupiennius salei75]), opilionids (Rilaena triangularis52]), scorpions (Pandinus imperator, Androctonus australis25], 39]) and xiphosurans (Limulus polyphemus25]). In these four taxa, each ventral neuromere gives rise to either one or two clusters
of SL-ir neurons with often variable cell numbers, ranging species-specifically from
minimally four (R. triangularis) to more than 60 (P. imperator, A. australis). It has been proposed that the euchelicerate stem species featured one anterior
and one posterior cluster of SL-ir neurons in each hemiganglion, at least some of
the neurons sending contralateral projections into the other ganglion half 10], 26], 40].
Clearly, this putatively ancestral euchelicerate pattern deviates significantly from
our observations in pycnogonids. Only in P. litorale, a numerically variable, cluster-like arrangement is assessable for the ALNs (with
apparently ipsilateral neurites). The remaining array of SL-ir neurons, however, is
very similar to the other two investigated pycnogonids, featuring several stereotypic
and individually identifiable cells that occur either singly or in pairs.
This pattern disparity between pycnogonids and euchelicerates results in a problematic
situation for the reconstruction of the ancestral state in the chelicerate lineage.
To resolve this, an out-group comparison might help to achieve character polarization
(see below). Nonetheless, however, also critical re-evaluation of the hitherto advocated
euchelicerate pattern would be desirable. For instance, the proposed plesiomorphic
‘two-cluster pattern’ with contralateral neurite projections 10], 25] has in fact only been found in the opisthosomal neuromeres of second-stage larvae
of L. polyphemus25] (the adult pattern remains still unknown). By contrast, neither the more anterior
walking leg-bearing prosomal neuromeres of the same species, nor any ventral neuromere
of the other studied adult euchelicerates show a corresponding pattern. Instead, there
is only a single cluster per hemiganglion. Accordingly, the reconstructed pattern
is solely based on the interpretation of the xiphosuran larval opisthosomal neuromeres
as representing the plesiomorphic euchelicerate condition. Deviations from this are
interpreted as apomorphic simplifications 25], but a proper phylogenetic analysis is still missing.
Likewise, if taking our pycnogonid data into account, an extrapolation of the variability
of SL-ir neuron numbers and the lack of individually identifiable neurons to the entire
Chelicerata can be questioned. Even more so, as at least for one euchelicerate representative
constant numbers of stereotypic SL-ir neurons are actually reported. In the harvestman
R. triangularis, exactly four neurons are described per prosomal walking leg hemi-neuromere, all
of which with ventro-medial somata and ipsilaterally projecting neurites 52]. Hence, if clusters of variable cell numbers are advocated as being plesiomorphic
for Chelicerata, two independent evolution events within the pycnogonid and opilionid
lineages would have resulted in individually identifiable SL-ir neurons with largely
constant numbers.
SL-ir neurons in Tetraconata and Myriapoda versus Pycnogonida
Several studies have focused on the comparison and homologization of SL-ir neurons
in the VNC of different tetraconate taxa (see, e.g. 10], 24], 36]). As a result, a relatively simple pattern of one anterior and one posterior pair
of SL-ir neurons has been suggested for the tetraconate stem species (e.g. 10]). These neurons were proposed to have been unipolar, with ipsilaterally as well as
contralaterally extending projections (type B sensu Harzsch 25]). In the meantime, studies on additional crustacean groups have demonstrated the
presence of additional SL-ir neurons in remipedes and cephalocarids, several of them
being found in medial position 26], 62]. As discussed in detail by Stegner and colleagues 62], these findings might necessitate a re-evaluation and extension of the ancestral
SLI pattern in Tetraconata.
One key argument for the need of a re-evaluation of the ancestral tetraconate pattern
is the situation in Myriapoda, the putative sister group to Tetraconata (=Mandibulata
hypothesis; see 42], 73]). The only available study of the SLI pattern in the myriapod VNC revealed a segmental
array of nine to twelve neurons (centipedes and millipedes, respectively), the somata
of which being arranged singly, in pairs or in groups of four in characteristic positions
within the ganglionic cortex 25]. However, apart from one distinct pair of postero-lateral neurons in centipedes that
project contralaterally, the course of their neurites could so far not be reliably
traced.
Given the SLI pattern disparity between pycnogonids and euchelicerates, an out-group
comparison especially to myriapods is one of the necessary next steps when trying
to establish character polarization for the chelicerate lineage. Notwithstanding persisting
limitations regarding neurite morphology, some noteworthy overall correspondences
in the general pattern type can be assessed between pycnogonids and myriapods. These
correspondences include (1) a segmental set of SL-ir neuronal somata that are arrayed
predominantly singly or in pairs within the cortex, (2) the stereotypic positions
of these somata with different lateral and medial as well as anterior to posterior
positions, and (3) their similar numbers per hemi-ganglion. Due to these features,
several SL-ir neurons can be reliably recognized at the single cell level in different
ganglia and across different specimens. This stands in clear contrast to the reconstructed
euchelicerate pattern with neuron clusters of variable cell numbers.
Beyond that, however, homologization of single SL-ir neuron types between pycnogonids
and myriapods faces the same uncertainties that have been previously encountered in
a similar attempt made for cephalocarids and myriapods (see 62]). Preliminary working hypotheses can be discussed, but well-founded homology assumptions
remain more often than not elusive. For instance, in all investigated myriapods at
least one posterior pair of neurons is present (neuron type ‘e’ sensu Harzsch 25]), which – where known – sends out contralateral projections. Obviously, the two distinctive
posterior DMNs in the WLG2-4 of pycnogonids with their contralateral projections appear
to be promising candidates for homologization. However, in myriapods, the somata of
these characteristic posterior neurons are positioned laterally, whereas the pycnogonid
DMNs are found far medially. While this difference of soma position need not necessarily
imply non-homology, the situation is complicated by pycnogonid walking leg ganglia
featuring at the same time the (mostly) paired PLNs with postero-lateral soma position.
Yet, as far as we were able to trace their primary neurite, it runs ipsilaterally
rather than contralaterally. Accordingly, we are faced with two potential pycnogonid
homologues to the myriapod postero-lateral neurons ‘e’, each of them showing only
one of two applied indices for homology. To complicate matters even further, myriapods
feature an additional neuron type ‘d’ 25], with medial soma that may be shifted posteriorly, but the neurite course of which
remains unknown. Apart from the already discussed DMNs, also the less reliably detected
VMNs or PVNs of pycnogonids might be considered as potential homologue to this ‘d’
neuron of myriapods.
In our opinion, this amply demonstrates that homologization attempts at the level
of SL-ir neuron types is anything but straightforward outside of the tetraconates,
and even in the latter taxon more challenging than previously assumed 26], 62]. The here presented pycnogonid data are one of the first steps for a broader data
basis in chelicerates (see also 39]). Beyond that, additional investigations on myriapods are still needed to estimate
more confidently the true potential of SL-ir neuron patterns in the arthropod VNC
within a phylogenetic-evolutionary framework.
Individually identifiable SL-ir neurons – an ancestral arthropod or panarthropod feature?
Based on evidences from different areas of the CNS, it has been suggested that individually
identifiable neurons might represent an apomorphic feature that has evolved only in
the mandibulate lineage (e.g. 10], 40]). This view is clearly challenged by the SLI pattern of Pycnogonida. Not only are
most of the SL-ir neurons of pycnogonids individually identifiable. Additionally,
there are also several overall correspondences between the pattern type of pycnogonids
and myriapods. Hence, it appears rather likely that the VNC of the arthropod stem
species has already possessed a more stereotypic segmental array of SL-ir neurons.
If this should hold true, a pattern of SL-ir neuron clusters with variable cell numbers
would have to be considered as being apomorphic either for the entire euchelicerate
lineage or at least for some of the euchelicerate taxa.
The original idea of the more variable cluster pattern being plesiomorphic for arthropods
found some support from studies on Onychophora 28], 29], which are generally accepted as close arthropod relatives 42]. In these animals, numerous SL-ir neurons are evenly distributed along the two ventral
medullary cords, with no indications of segmental arrays, let alone stereotypic individually
identifiable cells. Notably, however, it is not beyond doubt whether the gross anatomy
of the onychophoran VNC represents a good approximation for the plesiomorphic condition
of arthropods, or is rather a highly derived characteristic of the onychophoran lineage
itself 30], 76]. In the meantime, two recent studies on Tardigrada – being as well potential close
relatives of arthropods – have provided data on the SL-ir neurons in the VNC 30], 31]. Interestingly, both studies point out the presence of few SL-ir neurons in each
ventral ganglion, their somata being found in stereotypic positions, although exact
locations are unfortunately not described. These new findings may indicate an even
earlier appearance of individually identifiable SL-ir neurons already in the panarthropod
lineage 30]. To a certain extent, this view depends on whether the ganglionated VNC of tardigrades
and arthropods is considered to be homologous 30], or rather interpreted as having evolved two times independently.