The major aim of this study was to test whether patterns of backfilled neurons can be used as a morphological character complex for the homologisation of cerebral nerves in opisthobranch gastropods. Here we tested whether the criteria for evaluation of homology of the axonal tracing patterns, based on the earlier intra- and interspecific studies [10, 11] can be confirmed by extending the comparison of interspecific innervation patterns. Furthermore we wanted to homologise the cerebral nerves and in turn to postulate primary hypotheses of homology for the CSOs.
Intra- and interspecific variability of tracing patterns
Earlier investigations on Haminoea japonica have shown that the number and the size of the somata within a distinct cluster are correlated with the body size of the individual, whereas the number of clusters projecting into a particular nerve was constant within a species. H. japonica and Acteon tornatilis are relatively small species, which reach sexual maturity when body length approaches 1,5 cm but individuals can reach a body length up to 3 cm (own investigations, unpublished data), Pleurobranchaea meckeli and Archidoris pseudoargus are medium sized (up to 10 cm body length) and Aplysia spp. are large species (up to 30 cm). Besides the body size P. meckeli has the largest CNS due to elongated commissures and connectives rather than ganglion size, whereas the CNS of Aplysia spp. and A. pseudoargus are nearly the same size, regardless of the body length over the range of animal sizes examined.
Throughout our investigation of several taxa of opisthobranch gastropods in the frame of this study and in previous investigations, we could identify uniform tracing patterns via four cerebral nerves which can be attributed to characteristic neuronal cell clusters in the CNS. All investigated opisthobranchs have four major cerebral nerves which innervate the head region (optic nerve excluded). We found that the cellular innervation patterns for the different nerves in the different taxa were highly conserved with regards to the number and location of the clusters. However, in addition to the constant features of the axonal tracing patterns (see Figures 3, 4, 5, 6), we also found variations in these patterns across the investigated taxa. The number of cells within the clusters and also the size of the somata in each cluster vary. The position of the clusters in relation to each other and nervous structures seemed to be the most conserved and thus the most useful character to compare the cellular tracing patterns and to identify homologous clusters across different taxa. However, these features are not completely invariable, but variations are generally relatively minor. This variability might be caused by the fusion of ganglia and/or the strong variation of the morphology of the central nervous system. The lack of pedal clusters for P. meckeli could be more problematic. While our techniques appear to reliably stain central neurons projecting into the nerves, the long connectives between the cerebral and pedal ganglia in this species could exceed the distances over which nickel can travel in our assays.
Under the assumption that the cerebral clusters are homologous we expected to find a correlation for the number of somata within the cerebral clusters between the investigated species. However, due to intra- and interspecific variability we expected the correlation to be low, as the homologisation was not based on the number only, and the absolute number of somata is still correlated to body size .
We found a very high correlation in somata number between Aplysia spp. and Archidoris pseudoargus and a very low correlation between Aplysia spp. and Pleurobranchaea meckeli. Furthermore, the overall correlation is highest within the N2 and N3 somata numbers and lowest in the N1 somata numbers. These differences might indicate a higher similarity in function within the CSOs which are provided by the N2 (tentacles and labial palps) and the N3 (rhinophores) than within the CSOs provided by the N1 (lips) and Nclc (bodywall). The higher correlation between Aplysia spp. and A. pseudoargus might also be caused by a more similar life history as P. meckeli is an active predator of mobile prey whereas Aplysia spp. feeds on algae and A. pseudoargus on sponges, which are both sessile organisms. However, we are very cautious with this hypothesis, which at current is highly speculative and needs more data to be tested.
Homology hypotheses for the cerebral nerves
We postulate homology hypotheses for the cerebral nerves of the investigated species based on the observed conservation of their cellular innervation pattern of the respective nerves. The N1 innervates the lip region. The N2 has two branches (N2a, N2b) in all investigated opisthobranch taxa and is related to the anterior sensory organs. The two branches of the N2 innervate different cephalic areas, predicted to serve different functions . In consequence we distinguish between four types of CSOs: lip – provided by the N1, ASOa – provided by the inner branch of the N2, ASOb – provided by the outer branch of the N2 and PSO – provided by the N3. The Nclc innervates structures of the head region which are involved in locomotion like the body wall or the cephalic shield nevertheless these structures could also perform sensory functions like mechano- or light reception [33–37].
Summing up, the conservative axonal tracing patterns of the cerebral nerves within the Opisthobranchia allow us to homologise these nerves. In consideration of earlier studies [10, 11] our study supports the hypothesis  that the posterior Hancock’s organ in Cephalaspidea and the rhinophores in Nudipleura (Nudibranchia and Pleurobranchomorpha) and Aplysiomorpha are innervated by homologous nerves. The oral veil (Pleurobranchomorpha), lip organ (Cephalaspidea, Acteonoidea), and the labial tentacles (Aplysiomorpha and Nudibranchia) are also innervated by homologous nerves.
Nevertheless the axonal tracing method has its limitations. The first limitation is the size of the species: the techniques employed here require not only that the CNS be dissected without damage but that the individual nerves be large enough to manipulate (personal note: our limits have been species with a minimum body size of 0.5 cm). The second limitation is the number of specimens required, 5 to 10 undamaged replicates are needed for each nerve. Further restrictions have been discussed in previous studies [10, 11]. A third limitation is that backfilling using NiLys does not permit evaluation of whether some cells might have axons in more than one nerve, thus complicating our analysis. Future work could better employ fluorescent labels which might allow double labelling resulting from backfilling of more than one nerve at a time.
Primary homology hypotheses for the CSOs
As described above we divided the CSOs into the lip and the categories, ASOa and ASOb and PSO (Table 2), innervated by three homologous nerves (N1-N3).
Based on the hypotheses of homology for the cerebral nerves we postulate preliminary homology of the lip, which is innervated by the N1. Morphological and immunocytochemical investigations of the CSOs [38–43] indicated that the lip is supposed to be primarily a contact chemoreceptor organ. It can be assumed that the lip has the same function in all gastropods and represents a homologous structure throughout this taxon.
We also postulate homology of the various ASOs innervated by the N2, including labial tentacles, oral veils, oral tentacles, lip organ and the anterior part of the Hancock’s organ as it was defined by Edlinger . The immunocytochemical and ultrastructural investigations [39, 41–43] indicated that the ASOs are primarily mechano- and contact chemoreceptor organs. As discussed earlier we distinguish between two types/parts of the ASOs provided by the two different branches of the N2. Although homologies of types ASOa on the one hand and ASOb on the other hand are most likely, axonal tracing of the single branches of the N2 is necessary to homologize these parts of the nerve and in consequence to propose a homoloy hypothesis for the respective CSOs, as in some opistobranch species the anatomical differentiation between the ASOa and ASOb is clear but in other species it is not. In the current study due to the lack of specimen we only traced the entire nerve.
The third postulated homology is between the PSOs which are innervated by the N3 and supposed to be primarily olfactory [43, 44]. In our opinion, the posterior Hancock’s organ of Cephalaspidea, as described in Edlinger  and the rhinophores (both innervated by the N3) are homologous structures throughout the Opisthobranchia, as has been stated earlier [7, 8, 17]. Our data confirmes this homology hypothesis as we found similar axonal tracing patterns for the cerebral nerve (see N3 in our definition) which innervates the rhinophores of Nudipleura (Nudibranchia and Pleurobranchomorpha) and Aplysiomorpha, and the posterior Hancock’s organ of the Cephalaspidea . Furthermore similar neurotransmitter contents and similar sensory functions of the rhinophores and the posterior Hancock’s organ have been discussed . Therefore the primary homology hypotheses proposed here are not only supported by the axonal tracing patterns, but also by the specific functions of the CSOs deduced from previous immunocytochemical data.
In the case of the Sacoglossa, where the rhinophores are innervated both by the N2 and N3 [8, pers. obs.] a homology to other rhinophores is questionable and needs further investigation.
The cerebral nerve Nclc does not seem to correspond to a primary sensory organ.