Comparative analyses of olfactory systems in terrestrial crabs (Brachyura): evidence for aerial olfaction?

Experimental animals

We analyzed representatives of several different species of brachyurans representing aquatic species (four exclusively marine and one freshwater crab species) as well as four brachyuran species featuring different grades of terrestrial adaptation (Table 1; Figs. 2 and 3). For simplification, the four latter brachyurans are referred as terrestrial brachyurans throughout this text, although all nine species feature terrestrial adaptions at various degrees (Table 1). After shipping, living specimens of Cardisoma armatum, Geosesarma tiomanicum, and Uca tangeri were kept in tanks providing both a water and a land part. Husbandry as well as observation and documentation of these species were conducted between 5 and 14 days until dissection in the laboratory. Before dissection, animals were sexed, and the carapace width as well as the wet weight of each animal was measured. The collection of specimens of Gecarcoidea natalis was permitted by Christmas Island National Park (Australian Government; Department of the Environment; Parks Australia; Permit No.: AU_COM 2010-090-1).

Analysis of antennae and aesthetascs

The first pairs (antennules) and the second pairs (antennae) of post-ocular appendages were cut off prior to brain dissection and were transferred into 70% ethanol (G. natalis and two animals of G. tiomanicum) or in 2% glutaraldehyde in 0.1 M phosphate buffered saline (PBS; further specimens of G. tiomanicum, C. armatum and U. tangeri). Micrographs of these appendages were documented in PBS with a Nikon Eclipse 90i microscope equipped with a digital camera (Nikon DS2-MBWc) and analyzed by the use of the software package NIS-Elements AR. Cuticular auto-fluorescence of the first and second antennae was excited with ultraviolet light (UV) with a wavelength of 340–380 nm eliciting light emissions with a wavelength of 435–485 nm. Aesthetascs of marine specimens such as of Pagurus bernhardus (Linnaeus, 1758), Carcinus maenas, Xantho hydrophilus, X. poressa and Percnon gibbesi were cut off from the lateral flagella with a razor blade and counted on an object slide using UV-excitation as well as bright field illumination.

Histochemistry, immunohistochemistry, and microscopy

The animals were anaesthertized by cooling on ice for 1 h before dissection. Following the protocol by Ott (2008), the dissected brains were fixed in toto for approximately 20 h (room temperature) in 3.7% formaldehyde/zinc-fixative (the ready-to-use formaldehyde/zinc-fixative was obtained via Electron Microscopy Sciences. Cat. No. 15675). For whole-mount preparations, the brains and eyestalk ganglia were dissected, and the retina including all pigments was removed. The whole-mounts were washed three times in HEPES-buffered saline (HBS) for 15 min, subsequently transferred to Dent’s fixative (80% methanol/20% DMSO), and post-fixated for two hours at room temperature. Specimens were then transferred to 100% methanol and stored overnight in the refrigerator and rehydrated stepwise for 10 min each in 90%, 70%, 50%, 30% methanol in 0.1 M Tris–HCl-buffer, and finally in pure 0.1 M Tris–HCl-buffer (pH 7.4). Alternatively, for preparing horizontal brain sections, after anaesthetizing the animals by cooling on ice for 1 h, the brains were dissected and fixed in 4% paraformaldehyde (PFA) in 0.1 M PBS overnight. The dissected brains were washed for 4 h in several changes of PBS and sectioned (80–100 µm sections) horizontally at room temperature using a vibratome (Zeiss Hyrax V590; Carl Zeiss, Oberkochen, Germany). For permeation of cell membranes, both brain whole-mounts as well as brain sections were then preincubated for 90 min in PBS-TX (1% Bovine-Serum-Albumine, 0.3% TritonX-100, 0.05% Na-acide, in 0.1 M PBS; pH 7.4). In contrast to the protocol of Ott (2008), PBS-TX was used instead of PBSd-NGS. Finally, the samples were incubated at 4 °C for 84 h (whole-mounts) or overnight (sections) in the primary antisera. The following sets of reagents were used (compare Krieger et al., 2012):

Set A: rabbit anti-Dip-allatostatin 1 (AST-A; final dilution 1:2,000 in PBS-TX; antibody provided by H Agricola, Friedrich-Schiller Universität Jena, Germany); monoclonal mouse anti-synapsin “SYNORF1” antibody (final dilution 1:30 in PBS-TX; antibody provided by E Buchner, Universität Würzburg, Germany) detected by anti- mouse Cy3 (CyTM3-conjugated AffiniPure Goat Anti-Mouse IgG Antibody; Jackson ImmunoResearch Laboratories Inc., West Grove, PA, USA).

Set B: polyclonal rabbit anti-FMRFamid (in PBS-TX; final dilution 1:2,000; Acris/Immunostar; Cat. No. 20091) detected by anti-rabbit Alexa Flour 488 (IgG Antibody, invitrogen, Molecular Probes); monoclonal mouse anti-synapsin “SYNORF1” antibody (in PBS-TX; final dilution 1:30; antibody provided by E Buchner, Universität Würzburg, Germany) detected by anti- mouse Cy3 (CyTM3-conjugated AffiniPure Goat Anti-Mouse IgG Antibody; Jackson ImmunoResearch Laboratories Inc., West Grove, PA, USA).

Set C: monoclonal mouse anti-synapsin “SYNORF1” antibody (in PBS-TX; final dilution 1:30; antibody provided by E Buchner, Universität Würzburg, Germany) detected by anti- mouse Alexa Flour 488 (IgG Antibody; Invitrogen, Waltham, MA, USA; Molecular Probes, Eugene, OR, USA); counterstain: phallotoxins conjugated to Alexa Fluor 546 (concentration 200 units/ml; Molecular Probes, Eugene, OR, USA) as a high-affinity probe for f-actin.

In all three sets, the tissues were incubated in mixture containing the secondary antisera and the nuclear marker HOECHST (33,242; 0.1 µg/ml) for 2.5 days at 4 °C (whole-mounts) or for 4 h at room temperature (sections). Finally, the brain sections were washed for at least 2 h in several changes of PBS at room temperature and mounted in Mowiol^® (Calbiochem) between two coverslips. After secondary antibody incubation, the whole-mounts were dehydrated in changes of ascending glycerol concentrations (1%, 2%, 4% (2 h each), 8%, 15%, 30%, 50%, 60%, 70%, and 80% (1 h each) glycerol diluted in Tris–HCl buffer, with DMSO to 1% final concentration). After the last step of dehydration, the whole-mounts were washed twice for 30 min in 99.6% denatured ethanol. The ethanol was then underlyed by the same volume of methylsalicylate for clearing of the whole-mount brains. After the brains were cleared, the supernatant liquid was removed and the samples were and then mounted in customized chambers (a custom washer from the hardware store was glued between two coverslips as spacer) filled with methylsalicylate and sealed with Mowiol^®. The triple-labeled and sectioned tissues were analyzed using a Nikon eclipse 90i microscope equipped with a digital camera (Nikon DS2-MBWc). The whole-mounts were analyzed by using a confocal laser scanning microscope (clsm; Leica TCS SP5II; Leica, Wetzlar, Germany). The pictures were then processed using the NIS-Elements AR software and Adobe Photoshop CS4. Only global picture enhancement features of Photoshop elements (black to white inversion, brightness, and contrast) were used for all experiments. Three-dimensional (3D) brain reconstructions in addition to volumetric analysis based on optical section series of clsm data were performed using the reconstruction software Amira^® (FEI Visualization Sciences Group, Mérignac, France).

Raw data of brain section series is available from https://www.morphdbase.de under the “media” tab under a combination of the short title “Krieger” and an identifier according to the species and ID of the specimen.

Three-dimensional reconstructions of brains and substructures are based on tomographies of three specimens per species for C. armatum, G. tiomanicum and U. tangeri. For each specimen, surfaces of one DCL including the corresponding olfactory glomeruli and the ipsilateral AcN were generated by manual labeling. Finally, the computed 3D surfaces were slightly smoothed and resulting parameters such as the glomerular number and volume as well as the volume of the whole DCL were analyzed.

Nomenclature

The neuroanatomical nomenclature used in this manuscript is based on Sandeman et al. (1992) and Richter et al. (2010) with some modifications adopted from Harzsch & Hansson (2008)) and Loesel et al. (2013). In favor of a consistent terminology, here we suggest avoiding the term “optic neuropils” (Hanström, 1925; Sombke & Harzsch, 2015) as well as “optic lobes” (Kenyon, 1896). Even if the Greek “optikos” and the Latin term “visus” have the identical meaning, nowadays, “optic” in the field of visional anatomy and physiology refers to the physically refractive components of the eye for the reception of light. To emphasize the perceptive character of these neuropils, we suggest using the term “visual neuropils” which is consistent with, e.g., the visual cortex in mammals, formerly also termed “optic” cortex (Spiller, 1898). All post-retinal components that are related to vision, such as the “optic tract” and the “inner” as well as the “outer optic chiasm” should be consequently renamed, too. Here, we suggest to use “visual tract” (VT) and the “inner” (iCh) as well as the “outer visual chiasm” (oCh) accordingly. However, for all pre-retinal components that are related to vision, the term “optic,” as for example in the dioptric apparatus of the ommatidia, should be maintained. We also discourage the commonly used terms “eyestalk neuropils” (Bliss & Welsh, 1952; Polanska, Yasuda & Harzsch, 2007), “optic ganglia” (Medan et al., 2015), or “eyestalk ganglia” (Harzsch & Dawirs, 1996; Techa & Chung, 2015), usually summarizing the visual neuropils as well as the neuropils of the TM/HN-complex, because these neuropils together can be located more proximal to the central brain and not in the eyestalk in some species, and thus are part of the central brain as exemplified below. Furthermore, the neuropils of the lateral protocerebrum (visual neuropils + TM/HN-complex) do not fulfill the definition of a ganglion (see Richter et al., 2010). The traditional nomenclature of the visual neuropils lamina ganglionaris, medulla interna, and medulla externa has been modified as suggested by Harzsch (2002) to lamina, medulla, and lobula. Because we could not detect any border between the cell body clusters (9) and (11) of olfactory interneurons as described in Sandeman et al. (1992), we collectively refer to them as cluster (9/11) (see Krieger et al. 2010). The term “oesophageal connective” and the corresponding abbreviation OC (British English) are maintained here for simplicity. The olfactory neuropil (ON or OL) is now named the deutocerebral chemosensory lobe (DCL), and the olfactory globular tract (OGT) is now named the projection neuron tract (PNT) according to Loesel et al. (2013). Consequently, the olfactory globular tract neuropil OGTN is now named projection neuron tract neuropil (PNTN). For simplification, the neuroanatomical descriptions are kept restricted to only one hemisphere of the brain and hold true for all specimens studied if not stated otherwise.

The data presented in this study are drawn from different sets of triple-labeling immunofluorescence experiments as laid out above. The localizations of synapsins provides a general labeling of all neuropils in the brain whereas staining of actin is better suited to label neurite bundles and fiber tracts. The two antisera against allatostatin and FMRFamide label specific neuronal subsets and were chosen for a better comparison with other studies that have used the same markers (e.g., Harzsch & Hansson, 2008; Krieger et al., 2010; Krieger et al., 2012). The following abbreviations (color-coded in the figures) identify the markers:

SYN

synapsin-like immunoreactivity (magenta or black)

RFA

RFamid-like immunoreactivity (green or black)

PHA

actin labeling by the use of phalloidin (green or black)

AST

allatostatin-like immunoreactivity (green or black)

NUC

nuclear counterstain with HOECHST-dye H 33258 (cyan or black)

The antennae

In general, the first antennae in brachyuran crustaceans each consists of two branches called the median and the lateral flagellum (Fig. 4). Both flagella are composed of several units, the flagellomeres. Each flagellomere of the lateral flagellum is equipped with one row of the typical unimodal chemosensory sensilla, the aesthetascs, in both marine and terrestrial brachyurans (Fig. 5). A quantification of aesthetasc numbers is provided in Table 2. The shape of the aesthetascs in marine versus terrestrial brachyurans displays marked differences. In the marine species, the aesthetascs are long and slender, whereas in all species featuring a rather terrestrial lifestyle, they are short and blunt (Fig. 5). The second antennae consist of one articulated branch only, composed of multiple antennomeres, and with the low-resolution light microscopic methods used here, we could not detect any striking differences in the sensillar equipment between the marine and terrestrial representatives (Fig. 6).

General arrangement of neuropils in the brachyuran brain

The general morphology of the brachyuran brain, as in other Malacostraca, is composed of three consecutive neuromeres, the proto-, deuto-, and tritocerebrum as extensively reported in previous studies (reviewed in Harzsch, Sandeman & Chaigneau, 2012; Schmidt, in press). In some anomalan species, such as Birgus latro or Petrolisthes lamarckii as well as in the axiid shrimp Callianassa australiensis, the bilaterally paired visual neuropils and the neuropils of the terminal medulla/hemiellipsoid body—complex (TM/HN-complex) are located anteriorly adjacent to the “central” brain as a consequence of elongated axons composing the optic nerve. In all brachyurans studied so far, however, these neuropils are located within the eyestalks, thus being situated at some distance from the central portion of the syncerebrum. Note that in the comparative Fig. 3, for simplicity, only outlines of the central portions of the brains, in the following simply termed the “central brain”—are drawn, without the neuropils of the lateral protocerebrum. In horizontal sections, this central brain appears broader than elongated along the neuraxis (Fig. 3C). The species studied here displayed markedly different carapace widths ranging from 14 mm in G. tiomanicum up to 90 mm in G. natalis. In contrast, the general brain dimensions are rather similar across species as indicated by a range of brain width between 1.4 mm in G. tiomanicum to 2.5 mm in G. natalis and 2.7 mm in C. maenas. Hence, there seems to be only a weak correlation between brain size and body size.

Contrary to most other decapods analyzed so far (see e.g., Sandeman, Scholtz & Sandeman, 1993), a distinct compartmentalization of the brain neuropils is less obvious in brachyurans. For instance, the neuropil boundaries in true crabs are much less distinct than in Anomala (compare Krieger et al., 2012). However, the general organization of the brachyuran brain and arrangement of its subunits can be deduced from anatomical data by tracing nerves as well as interconnecting tracts between the corresponding neuropils as outlined below:

The protocerebral tract (PT) is composed of neurites originating in neuropils of the lateral protocerebrum (lPC). The PT interconnects these neuropils with the proximal part of the brain, the median protocerebrum (mPC). The median protocerebrum is composed of the anterior (AMPN) and the posterior medial protocerebral neuropil (PMPN), which together resemble the shape of a butterfly in horizontal brain sections (compare Figs. 3, 9D, 9E, 11D, 11E, 13C–13G, 15A–15D, 16A1–16C1 and 17). Both neuropils are almost completely fused anterioposteriorly as well as across the midline with their contralateral counterparts into one single neuropil mass in the brachyuran brain, but they appear separated in horizontal sections at the level of the central body (Figs. 11E–11F, 13D, 15B and 15F). Furthermore, the median protocerebrum includes neuropils of the central complex, namely from anterior to posterior: the unpaired protocerebral bridge (PB), the unpaired central body (CB), and the bilaterally paired lateral accessory lobes (Lals).

In all brachyuran species investigated, the neuropils of the deutocerebrum (DC) that extend posteriorly adjacent to the median protocerebrum consist of the unpaired median antenna I neuropil (MAN), the bilaterally paired antenna I neuropils (LANs), the deutocerebral chemosensory lobes (DCLs; formerly referred as olfactory lobes or olfactory neuropils), the accessory lobes (AcNs) and the projection neuron tract neuropils (PNTNs; formerly referred as olfactory globular tract neuropils or OGTNs). Each DCL consists of several to hundreds of barrel-shaped subunits of synaptic neuropil, the olfactory glomeruli (OG) which are arranged in a radial, palisade-like array in the periphery of the lobe. Medially to each DCL, a cluster of somata (9/11) of hundreds of interneurons of varying sizes is present. These neurons extend neurites which enter the DCL via the median foramen (mF), one of three gaps in the palisade-like array of olfactory glomeruli. Furthermore, several hundreds of somata of olfactory projection neurons are grouped in cell cluster (10) posteriorly to each DCL. Their neurites enter each DCL via the posterior foramen (pF), innervate the olfactory glomeruli, and project axons that exit each lobe via its median foramen (mF) in a large bundle that constitutes the projection neuron tract (PNT). The axons of the PNT interconnect each DCL with the ipsilateral as well as contralateral hemiellipsoid body within the lateral protocerebrum by forming a chiasm dorsally of the central body.

The tritocerebrum (TC) posteriorly adjoins the neuropils of the deutocerebrum and is composed of the bilaterally paired antenna II neuropils (AnNs) and further dorsally, of the tegumentary neuropils (TNs).

Lateral protocerebrum: the visual neuropils and the terminal medulla/hemiellipsoid body—complex (TM/HN-complex)

The eyestalks of most malacostracan crustaceans each contain three successive retinotopic neuropils. These three main visual neuropils process visual input and from distal to proximal are termed the lamina, medulla, and lobula. An additional (fourth) neuropil can be found adjacent to the lobula referred to as lobula plate. If present, the lobula plate adheres the lobula. The architecture of these visual neuropils which often are referred to as optic neuropils is best known in crayfish and lobsters (review Harzsch, Sandeman & Chaigneau, 2012) but was also analyzed in a number of marine and amphibious brachyurans including Chasmagnathus granulatus, Hemigrapsus oregonensis (Sztarker, Strausfeld & Tomsic, 2005; Sztarker et al., 2009; Berón de Astrada, Medan & Tomsic, 2011; Berón de Astrada et al., 2013), and Carcinus maenas (Elofsson & Hagberg, 1986; Krieger et al., 2012). Although the visual neuropils are not the focus of the present study, successful eyestalk preparations from C. armatum (Figs. 7A–7D), G. natalis (Figs. 9A–9C), G. tiomanicum (Figs. 11A and 11B), and U. tangeri (Fig. 13A) show that these terrestrial species display well developed visual neuropils that show distinct synapsin-like immunoreactivity (SYN). Distinct clusters of somata become clearly visible distal to each visual neuropil. According to their appearance from distal to proximal and based on nuclear counterstaining, we distinguish cluster (1) associated with the lamina; cluster (2) associated with the medulla; cluster (3) associated with the lobula (Figs. 7A, 9A, 11A and 13A). Their arrangement and layered architecture closely correspond to those of their marine relatives. In the lobula, we could resolve three main layers in all four species (Figs. 7B, 9B, 11A and 13A) suggesting that at the level of resolution we analyzed, the visual neuropils show a high level of similarity.

The most proximal neuropils of the lateral protocerebrum, the terminal medulla (TM; also termed medulla terminalis) and the hemiellipsoid body (HN), which are considered multimodal associative areas (Wolff et al., 2012), are located within the eyestalk and together constitute an almost spherical neuropil mass (TM/HN-complex) in the species studied. They are identifiable in preparations of C. armatum (Figs. 7A–7E), G. natalis (Figs. 9A–9C), G. tiomanicum (Figs. 11A–11C2), and U. tangeri (Figs. 13A and 13B) showing distinct SYN, but were also described in Chasmagnatus granulatus (Berón de Astrada & Tomsic, 2002), Hemigrapsus oregonensis (Sztarker, Strausfeld & Tomsic, 2005), and C. maenas (Krieger et al., 2012). A clear distinction between these two neuropils is difficult because they are tightly adjoined. Therefore, a comparative volumetric analysis was impractical. Nevertheless, our preparations indicate that in Uca, the TM/HN-complex is markedly smaller in diameter compared to all other crabs being analyzed (compare Fig. 13A with Figs. 7A–7E, Figs. 9A–9C and 11A–11B). A compartmentalization of the hemiellipsoid body into one cap and 1–2 core neuropil masses is obvious in G. natalis (Fig. 9C) and in G. tiomanicum (Figs. 11B–11C2), whereas such a subdivision could not be resolved in the other crabs analyzed. According to Sandeman et al. (1992), each of these neuropils is associated with a cluster of neurons, namely cluster (4), which is located closely to the terminal medulla and cluster (5), which is adjacent to the hemiellipsoid body in decapods and contains hundreds of interneurons of minute diameter. However, in the brachyurans studied, a clear separation of these two clusters was impossible. Rather, the TM/HN -complex is surrounded by a confluent cortex of somata which therefore will be referred to as cluster (4/5) here (Figs. 7C, 9C, 11B and 13A–13B).

Median protocerebrum

The median protocerebrum is composed of the closely fused AMPN and PMPN and appears broader than long. In all brachyuran crabs studied, it has a butterfly-shape in horizontal sections (Fig. 3C). The AMPN and the PMPN are identifiable by showing distinct SYN (Figs. 7F, 7H–7J, 9D–9E, 11D–11F, 13C–13E, 15A–15D, 16A1–16C2 and 17), weaker RFA (Figs. 11D–11F, 13D–13E, 13G, 16A1–16C2 and 17) and AST in the periphery (Figs. 9D–9E, 15A–15D and 17). Although allatostaninergic and RFamidergic fibers innervate the whole brain, the V-shaped protocerebral bridge (PB) and especially the cylindrical or cigar-shaped central body (CB) further posteriorly show the densest RFA as well as AST (Figs. 7F–7I, 9E, 11D–11F, 13D, 13G, 15A–15D, 16A2–16C2 and 18) besides distinct SYN (Figs. 7H–7J, 9E, 11E–11F, 13D–13E, 15A–15D, 16A2–16C2 and 17). In sections at the level of the CB, the separation of the AMPN from the PMPN becomes visible (Figs. 7H, 9E, 11F, 11G, 13D, 13F, 15B and 17). Anterior to the PB, the nuclear marker reveals hundreds of somata of varying diameters (5–10 µm in all species studied) that are grouped within the cell cluster (6). This cluster also comprises a subset of few somata with a markedly larger diameter that display distinct RFA (diameters approx. 30 µm in C. armatum, 25 µm in G. tiomanicum, and 20 µm in U. tangeri) and AST (approx. 30 µm in E. sinuatifrons and G. natalis). The neuropils of the median protocerebrum are regularly pierced by blood vessels of the circulatory system (e.g., the cerebral artery (CA); Figs. 9D–9E 11D, 11F, 13C and 13E) and by large tracts of neurites (e.g., the projection neuron tract (PNT); Figs. 7H, 9D, 11D–11G, 13C–13E, 13G, 15B, 15D, 16A1 and 16B1–16B2) and can be inferred from the negative imprint due to the absence of immunoreactivity against the tested antisera. The projection neuron tract consists of neurites of olfactory projection neurons whose cell bodies are located in the somata clusters (10) situated posterior-lateral to each DCL. These neurites connect each DCL to the ipsilateral as well as the contralateral TM/HN-complex within the lateral protocerebrum and constitute a chiasm dorsally to the central body. The cerebral artery (CA) located between median protocerebrum (posterior to the PMPN) and deutocerebrum (anterior to the median antenna I neuropil) is identifiable by the nuclear counterstain of the perivascular cells (Figs. 9D–9E, 11F, 13F and 15A–15B) in horizontal sections. The dorsoventral course of the CA through the central brain could be confirmed in all crabs as it has been shown for C. maenas (Sandeman, 1967) but not its ramifications.

Deutocerebrum with special focus on structures of the primary olfactory pathway

Directly posterior-ventral to the PMPN and the CA, the unpaired median antenna I neuropil (MAN; Figs. 7F–7H, 9D–9E, 11D–11G, 13C–13G and 15B) is present in all brachyuran crabs studied displaying distinct SYN as well as AST and RFA. The border between PMPN and MAN is rather confluent but is identifiable due to the clear position of the CA (compare Figs. 9D and 9E).

Besides the deutocerebral chemosensory lobe (DCL) and the accessory lobe (AcN), other neuropils of the lateral deutocerebrum can be found within the confluent mass of the central brain composed of parts of the proto-, deuto- as well as the tritocerebrum (compare Fig. 17) such as the lateral antenna I neuropil (LAN; Figs. 7F, 9D, 11D, 13D–13E and 15B) and in a few preparations, the projection neuron tract neuropil (PNTN; Figs. 15B and 16A1). However, a complete, in-depth reconstruction of their definite outlines remains challenging. In all species investigated, the AcN and, in particular the DCL are the most delimited structures within the otherwise confluent brachyuran brain. The DCL is composed of several dozens (60–80 in G. sesarma, U. tangeri and C. armatum up to almost 200 in G. natalis—see Table 2 for further information) of barrel-like to conical cardridges, termed olfactory glomeruli (OG), of varying sizes (see Fig. 18). From a limited number of investigated specimens of U. tangeri, it appeared that in two males analyzed the number of olfactory glomeruli exceeded that of one female by a factor of ca. 2 (36 OG in ♂versus 76–80 OG in ♀), whereas the males featured approximately a third of the average female glomerular volume (see Table 2) resulting in an almost equal volume of the entire DCL in both sexes. In all marine brachyuran species studied, the numbers as well as the average volumes of olfactory glomeruli markedly exceed those of the co-studied terrestrial brachyurans (Table 2 and Fig. 18), though the general brain dimensions are somewhat similar (see Figs. 3 and 17). In brain sections of aquatic representatives of Brachyura (in the four exclusively marine; and to some degree, in the freshwater species E. sinuatifrons), the olfactory glomeruli are larger and more elongated compared to those of the terrestrial species studied here. A clear regionalization of each olfactory glomerulus into a cap, a subcap, and a base region (from the periphery of the DCL to its center) appears more pronounced in aquatic brachyurans than in the terrestrial species (Fig. 18). The cap and base regions show stronger SYN (Figs. 8A, 8D1, 14E, 15E, 16A1, 16B1, 17 and 18) than the subcap region in these species. In a subset of experiments, the subcap region shows distinct RFA, but RFA is weaker in the base region and is absent in the cap region (i.e., in P. gibbesi, X. hydrophilus, X. poressa, U. tangeri; Figs. 18A and 18B), whereas the subcap region shows the most distinct AST and becomes absent towards the base region in each OG (in E. sinuatifrons, Figs. 18A and 18B; G. natalis, not shown; and in C. maenas, see Krieger et al., 2012). Anteriomedial to the median foramen of each DCL, the accessory lobe (AcN) becomes visible, consisting of dozens of microglomeruli that show distinct SYN but widely lack RFA as well as AST. The diameter of the almost spherical AcN ranges from 50 µm (in U. tangeri, X. hydrophilus, X. poressa, and P. gibbesi; Figs. 14A, 14D–14E and 16A3–16C3) up to 100 µm (75 µm in C. armatum, 100 µm in G. natalis as well as G. tiomanicum; Figs. 7F, 8A, 8E, 9D, 10C, 12A, 12D and 12F). Further medial and between the PMPN and the AcN, a somata cluster of hundreds of interneurons (ca. 5–8 µm in diameter) appears. This cell cluster (9/11) is clearly revealed by the nuclear counterstaining (Figs. 8C, 9E, 10A, 10E and 15B) and contains subpopulations of several to dozens of allatostatinergic (Figs. 9E and 15B) and RF-amidergic interneurons (Figs. 7G, 12A–12C, 13D–13E, 16A1, 16A3 and 16B1) that are markedly larger in diameter (from 12 µm in U. tangeri; and 16 µm in X. hydrophilus; up to 30 µm in G. tiomanicum; i.e., see Fig. 12C). Neurites of cell cluster (9/11) enter each DCL via the median foramen (mF; Figs. 7F, 8C, 10B, 12B, 14A–14C, 15B, 16A1, 16B1 and 17). Lateroposterior to each DCL, a group of hundreds to thousands of olfactory projection neurons house their somata within cell cluster (10). These neurites of projection neurons enter the DCL via the posterior foramen (pF; Figs. 8B–8C, 9D, 10D, 12C, 14B–14C and 15B), connect with the olfactory glomeruli, exit the DCL via the median foramen (Figs. 8C, 12B, 14C, 15B, 16A1, 16B1 and 17), and finally project to the ipsilateral as well as the contralateral TM/HN-complex by forming a chiasm at the dorsal level of the central body (not shown). The entirety of neurites of projection neurons constitute the projection neuron tract (PNT) whose somata are housed within cell cluster (10). According to its position, the projection neuron tract neuropil (PNTN) becomes visible medial to the mF in a few preparations showing distinct SYN (Figs. 15B, 16A1 and 17).

Tritocerebrum

The tritocerebral antenna II neuropil (AnN) and further dorsally the tegumentary neuropil (TN) compose the posteriormost parts of the central brain, being located anterolaterally to the esophagus. An identification of the neuropil borders is difficult due to their confluent connection to the deutocerebrum. The AnN that receives chemosensory as well as mechanosensory input from the second antenna is identifiable in a few preparations by tracing back the course of the antenna II nerve (A_IINv; Figs. 9D and 9E). Since we were unable to trace back the course of the presumably thin tegumentary nerve (TNv), the precise position and shape of the tegumentary neuropil remains uncertain.

Visual ecology and the protocerebrum

Terrestrial brachyurans have been prime examples to study visual ecology in crustaceans (reviews by Zeil & Hemmi, 2006; Zeil & Hemmi, 2014; Hemmi & Tomsic, 2012). Visual orientation has been very well studied in members of the genus Uca but poorly in any of the other ocypodid species (Zeil & Hemmi, 2014 and references therein). Field experiments for individuals of U. tangeri have shown that they can recognize predators at greater distance, triggering an escape behavior. In addition, the animals react to their own mirror image and can visually distinguish the gender of their conspecifics (Altevogt, 1957; Altevogt, 1959; Von Hagen, 1962; Korte, 1965; Land & Layne, 1995; Zeil & Al-Mutairi, 1996). Representatives of the genus Uca can also distinguish colors (Korte, 1965; Hyatt, 1975; Detto, 2007), which is an important factor for social interactions (Detto et al., 2006; Detto, 2007). Ultraviolet light, for example, is reflected by the claw of Uca-males which attracts females (Detto & Backwell, 2009). Aspects of homing and path integration were also thoroughly analyzed in members of the genus Uca (e.g., Hemmi & Zeil, 2003; Layne, Barnes & Duncan, 2003a; Layne, Barnes & Duncan, 2003b; Walls & Layne, 2009). Clearly, vision plays an essential role in the ecology of Uca.

In all species examined here, the neuropils of the lateral protocerebrum are located within the eyestalks in some distance to the central brain (compare Sandeman et al., 1992; Sandeman, Scholtz & Sandeman, 1993). In this study, the three visual neuropils (lamina, medulla, and lobula) could be identified in all individuals of the different species, and their location and anatomy matches that of other described brachyuran species (Tsvileneva, Titova & Kvashina, 1985; Sandeman et al., 1992; Sandeman, Scholtz & Sandeman, 1993; Sztarker, Strausfeld & Tomsic, 2005; Sztarker et al., 2009; Krieger et al., 2012; Berón de Astrada et al., 2013). However, the small lobula plate, the fourth visual neuropil, could not be found in any of the analyzed species, most likely because of technical difficulties but was previously identified in other brachyuran species such as Chasmagnathus granulatus, Hemigrapsus oregonensis (Sztarker, Strausfeld & Tomsic, 2005; Sztarker et al., 2009) and C. maenas (Krieger et al., 2012). The terminal medulla (or medulla terminalis) and the hemiellipsoid body are considered to function as secondary higher-order neuropils (Wolff et al., 2012; Wolff & Strausfeld, 2015). They integrate different modalities such as visual and olfactory information that were already preprocessed in the primary sensory brain centers (visual neuropils and deutocerebral chemosensory lobes; reviewed in Schmidt, in press). Furthermore, the TM/HN-complex receives input from the ventral nerve cord (VNC) and other regions of the central brain. The terminal medulla and especially the hemiellipsoid body are also referred to as centers of learning and memory that functionally correspond to the mushroom bodies in hexapods. For the brachyurans studied here, there were only few species-specific differences visible in our preparations. We conclude that all terrestrial brachyurans examined here have well developed visual neuropils. Thus, they possess a neuronal substrate for a sophisticated analysis of the compound eye input. Therefore, as in their marine counterparts, visual cues most likely play important roles in the terrestrial brachyurans’ behaviors such as food search, mating, and orientation.

Chemical senses: the peripheral olfactory pathway

It is well established that marine crustaceans use chemical cues to locate mates, signal dominance, recognize individual conspecifics, find favored foods and appropriate habitats, and assess threats such as the presence of predators (reviews e.g., Derby et al., 2001; Grasso & Basil, 2002; Derby & Sorensen, 2008; Thiel & Breithaupt, 2011; Wyatt, 2011; Derby & Weissburg, 2014). Malacostracan crustaceans that live in aquatic habitats use several systems for detecting chemicals, and these are distributed over their body surface, walking appendages, and mouthparts. We will focus our discussion on those sensilla concentrated on the two pairs of antennae (reviews e.g., Hallberg, Johansson & Elofsson, 1992; Hallberg & Skog, 2011; Schmidt & Mellon, 2011). The first antennal pair (the antennules) is equipped with specialized olfactory sensilla (aesthetascs) in addition to bimodal chemo- and mechanosensilla, functioning as contact-chemoreceptors, whereas the second pair of antennae is only equipped with the latter. The tips of the first antennae (more specifically the lateral flagellum) bear a tuft region with arrays of aesthetascs that house branched dendrites of olfactory sensory neurons (reviews by Hallberg, Johansson & Elofsson, 1992; Hallberg & Hansson, 1999; Mellon Jr, 2007; Hallberg & Skog, 2011; Schmidt & Mellon, 2011; Derby & Weissburg, 2014). There are multiple studies on the ultrastructure of these aesthetascs (e.g., Ghiradella, Case & Cronshaw, 1968a; Ghiradella, Case & Cronshaw, 1968b; Snow, 1973; Wasserthal & Seibt, 1976; Tierney, Thompson & Dunham, 1986; Spencer & Linberg, 1986; Grünert & Ache, 1988; Gleeson, McDowell & Aldrich, 1996), but unfortunately none of these studies includes any of the species analyzed in the present work.

We observed that all four terrestrial brachyurans studied here have shorter antennae in relation to their body sizes, and feature markedly fewer and shorter aesthetascs compared to their marine relatives (Table 2; Figs. 3A, 3B and 4–5). These findings suggest that possessing short and hidden first antennae equipped with few, short, and blunt aesthetascs seems to be a shared feature and most likely a specific adaptation in all terrestrial brachyurans. We suggest that this feature may be an adaptation to minimize water loss across the cuticle. Furthermore, a typical marine (brachyuran) array of long and slender aesthetascs will likely collapse out of water and most likely will be non-functional on land. Studies on other terrestrial crustacean taxa such as representatives of the Isopoda and Anomala support the idea that all terrestrial crustaceans share a size reduction of antennal sensilla including the aesthetascs (compare Hansson et al., 2011). The aesthetascs of terrestrial hermit crabs of the taxon Coenobitidae, for example, display striking differences to those of marine hermit crabs in that they appear short and blunt (compare Ghiradella, Case & Cronshaw, 1968a; Stensmyr et al., 2005). In robber crabs, Birgus latro, the largest known land arthropods, they are confined to the ventral side of the primary flagella and are flanked by presumably bimodal contact-chemoreceptive sensilla. A preliminary analysis using classical histology and transmission electron microscopy (TEM) revealed that, in contrast to marine crustaceans, the aesthetascs of Coenobitidae have an asymmetric profile, with the protected side lined with a thick cuticle (Tuchina et al., 2015). The exposed side is covered with a thinner cuticle, a feature that most likely is necessary to enable the passage of odors (Stensmyr et al., 2005). These and other morphological features were interpreted as mechanisms to minimize water evaporation while maintaining the ability to detect volatile odorants in gaseous phase (Stensmyr et al., 2005). Furthermore, antennal olfaction at least in coenobitids is assumed to depend on activity of the asthetasc-associated epidermal glands discharging their secretion to the base of related aesthetascs. By the aid of the mucous secretion covering the entire thinner cuticle, aesthetascs are provided with a moist, sticky layer essential for binding, sampling, and finally perceiving (after transcuticular passage) volatile odors (Tuchina et al., 2014).

However, in terrestrial Anomala, contrary to terrestrial Brachyura, the first antennae are extensively enlarged and the number of aesthetascs markedly increased as compared to marine representatives (Table 2), and there is evidence that Coenobitidae may have evolved good terrestrial olfactory abilities (Greenaway, 2003). In fact, behavioral studies have suggested that these animals are very effective in detecting food from a distance and in responding to volatile odors (Rittschof & Sutherland, 1986; Vannini & Ferretti, 1997; Stensmyr et al., 2005). These omnivorous crabs are attracted by volatiles emitted by many different sources such as seawater, wellwater, distilled water (Vannini & Ferretti, 1997), crushed conspecifics or snails (Thacker, 1994), fruits, seeds, flowers (Rittschof & Sutherland, 1986; Thacker, 1996; Thacker, 1998), and finally even horse faeces and human urine (Rittschof & Sutherland, 1986). By conducting a two-choice bioassay with Coenobita clypeatus using an arena with a centrally placed shelter with two pit-falls on each side, Krång et al. (2012) found that the animals were strongly attracted to natural odors from banana and apple. Furthermore, wind-tunnel experiments with C. clypeatus suggest that these animals display a behavior that may be described as odor-gated anemotaxis (C Mißbach, J Krieger, S Harzsch, BS Hansson, 2015, unpublished results). Furthermore, electrophysiological studies using electroantennograms in B. latro confirmed that the aesthetascs respond to volatile substances (Stensmyr et al., 2005). In aquatic crustaceans, antennular flicking enhances odorant capture by shedding the boundary layer (Koehl, 2011; Reidenbach & Koehl, 2011; Mellon Jr & Reidenbach, 2012). Coenobitidae also show flicking behavior similar to that seen in their marine relatives, thus maximizing odor sampling (Stensmyr et al., 2005). Mellon Jr & Reidenbach (2012) suggested that considering the higher kinematic viscosity of air versus water and the resulting lower Reynolds numbers, the aesthetascs of Coenobitidae nevertheless operate in a range where boundary layer shedding could be effectively achieved by antennular flicking. Taken together, these behavioral and morphological observations suggest that terrestrial Anomala evolved aerial olfaction and actively use their first pair of antennae to detect volatile odors.

In contrast to this highly sophisticated olfaction-related behavior of Coenobitidae, our limited observations in the laboratory of the terrestrial brachyurans C. armatum, G. tiomanicum, and U. tangeri suggest that their first pair of antennae extended and that flicking behavior occurred only if animals were immersed in water but the antennae were not exposed to the air. This holds also true for the second pair of antennae, except for Uca tangeri. In Gecarcoidea natalis, we did not observe that the first as well as the second pair of antennae were exposed in their terrestrial habitat as observed in three field trips to Christmas Island (J Krieger, MM Drew, S Harzsch, BS Hansson, 2012, unpublished obs.). These animals enter the water only during the spawning season (Orchard, 2012). As laid out above, crabs orient very well on land, and many studies have suggested vision to be the dominating sense in terrestrial Brachyura. Our morphological results and preliminary behavioral observations suggest that, contrary to Anomala, the detection of volatile substances plays only a minor role in the sensory ecology of Brachyura while on land. If it holds true that the first antennae in brachyurans are only functional in an aquatic environment, we may expect to see this reflected in the organization of primary processing areas within the brain. With respect to the critical cost-benefit ratio of maintaining the highly energy-demanding nervous tissue, providing processing capacities for poorly used sensory modalities may be too costly, so that these brain areas become reduced during evolution.

Chemical senses: the central olfactory pathway

The chemosensory neurons associated with the aesthetascs versus the bimodal non-aesthetasc sensillae (contact-chemoreceptors) on the first antennae of malacostracan crustaceans innervate distinct regions in the brain (see review of Schmidt & Mellon, 2011; Derby & Weissburg, 2014). The axons of the olfactory sensory neurons (OSNs) associated with the aesthetascs target the deutocerebral chemosensory lobes (in our previous studies termed olfactory lobes), whereas the axons associated with non-aesthetasc sensilla innervate the lateral antenna 1 neuropil (LAN; for other crustacean chemosensory systems see Schmidt & Mellon, 2011). For all species studied in this paper, the deutocerebral chemosensory lobes of the deutocerebrum, the accessory neuropils, the lateral antenna I neuropil and the median antenna I neuropil were well identifiable. Their structure and arrangement corresponds to that described in other Brachyura (Sandeman et al., 1992; Krieger et al., 2012). Also, the projection neuron tract and the cerebral artery could be depicted as characteristic landmarks. In all species investigated, the deutocerebral chemosensory lobes (DCL) share the typical malacostracan organization, featuring a radial array of barrel- to wedge-shaped olfactory glomeruli that form the thick synaptic layer of the lobe, with their apices pointing inwards (compare Schachtner, Schmidt & Homberg, 2005; Schmidt & Mellon, 2011). Although the DCLs of the species studied here have a similar overall organization, the relative size of the DCL to the central brain displays the most striking difference between aquatic and terrestrial brachyurans. While in all aquatic brachyurans studied, the DCLs are comparably large, they are conspicuously much smaller within terrestrial brachyurans, an observation also made in the land crab Chiromantes haematocheir (Honma et al., 1996). This relation also applies to the number and size (length especially) of olfactory glomeruli which are higher in all aquatic brachyuran species studied (see Table 2). These morphological aspects seem to be strongly correlated with the reduction of aesthetasc number and size as discussed above and therefore may represent another adaptation to terrestrialization (compare Figs. 17 and 18). However, a linear correlation between the number of aesthetascs and number of olfactory glomeruli could not be identified here, which is in accordance with the varying convergence ratios (aesthetascs/glomeruli) reported by Beltz et al. (2003). In summary, morphometric quantifications of neuronal structures have indeed to be considered as rough estimates to infer sensory processing performance of a species, and the species-dependent lifestyles play of course a large role for the evaluation of olfactory capacity.

Although we have analyzed admittedly only few specimens (two males and one female), our findings nevertheless hint at a sexual dimorphism of the DCL of U. tangeri. In addition to numerous reports of sexual dimorphism of insect brains (e.g., Koontz & Schneider, 1987; Homberg, Christensen & Hildebrand, 1989; Rospars & Hildebrand, 2000; Jundi et al., 2009; Streinzer et al., 2013; Montgomery & Ott, 2015) especially of the primary olfactory system, such a sexual dimorphism in crustaceans is well described from the DCLs in Euphausiacea and Mysidacea (Johansson & Hallberg, 1992). Furthermore, Loesel (2004) suggested a sexual dimorphism in central body architecture in the genus Uca. However, further investigation of sexual dimorphic features within the brain of crustaceans is crucial to understand the general principles in crustacean communication and their underlying structures. Their pronounced sex-specific external morphology regarding courtship behavior (e.g., the conspiciuous heterochely and eye stalk extensions in males) indicates that representatives of the genus Uca can serve as favorable study organisms to explore such aspects.

It has been well documented in aquatic malacostracans including crayfish, clawed and clawless lobsters, marine brachyurans, and hermit crabs (Schachtner, Schmidt & Homberg, 2005; Schmidt & Mellon, 2011; Krieger et al., 2012; Polanska et al., 2012) that the olfactory glomeruli are regionalized along their long axis to provide an outer cap, a subcap, and a base region. The subcap region of decapod olfactory glomeruli displays another level of subdivision when viewed in cross-sections and is separated into a central rod, a core region, and an outer ring. These patterns of subdivision of decapod olfactory glomeruli have been suggested to mirror a functional subdivision (Schmidt & Ache, 1997). Such a regionalization was not very obvious in the terrestrial brachyuran glomeruli which we analyzed. In conclusion, it seems obvious that the reduced sensory input to the deutocerebral sensory lobe in terrestrial brachyurans decreases the processing demands in the system, which in turn is reflected in the small size of olfactory glomeruli in addition to the lowered structural and functional complexity therein. These findings are also supported by the behavioral observations described above and support the idea that, while on land, olfaction is subordinate to vision in brachyurans. Along these lines, neuroanatomical studies of the olfactory system in marine versus terrestrial isopod crustaceans also suggested that in the terrestrial animals the deutocerebral chemosensory system has lost some of its importance during the evolutionary transition from water to land (Harzsch et al., 2011; Kenning & Harzsch, 2013).

Contrarywise, neuroanatomical studies analyzing the central olfactory pathway in terrestrial Anomala including Coenobita clypeatus (Harzsch & Hansson, 2008; Polanska et al., 2012; Wolff et al., 2012), and Birgus latro (Krieger et al., 2010) in comparison to several marine anomalan taxa of the subgroup Paguroidea (Krieger et al., 2012) suggested that in both terrestrial species, the primary olfactory centers targeted by antenna 1 aesthetasc afferents strongly dominate the brain and display conspicuous side lobes that are not present in the marine representatives, suggesting that a significant elaboration of brain areas involved in olfactory processing has taken place. The DCLs are markedly enlarged, and the number of olfactory glomeruli is increased compared to other marine anomalans studied (Table 2).

The tritocerebrum: antenna II neuropil and flow detection

In arthropods, the detection of flow is essential for tracking odor sources but also for anemotaxis, and in crustaceans, antenna 2 most likely plays a major role in detecting flow. In many malacostracan crustaceans, the second pair of antennae bear mostly mechanosensory sensilla as well as bimodal chemo- and mechanosensory sensilla (Schmidt & Mellon, 2011) presumably working as contact-chemoreceptors. This pair of appendages is associated with the tritocerebral neuromere, and its afferents target the bilaterally paired antenna 2 neuropils (AnN) that extend posterolaterally to either side of the esophageal foramen (Figs. 9D–9E, 11D, 13E and 15B). In some representatives of Decapoda, this neuropil is transversely divided into segment-like synaptic fields, suggesting a somato- or spatiotopic representation of the mechanoreceptors along the length of the second antenna (reviewed in Krieger et al., 2012). In the marine anomalan Pagurus bernhardus, this neuropil is elongate and the transverse segmentation is very obvious (Krieger et al., 2012), enforcing the idea that the sensory array of antenna 2 may be mapped along its length. In terrestrial Anomala, the antenna 2 neuropils of C. clypeatus and B. latro are rather inconspicuous, as is a transverse segmentation (Harzsch & Hansson, 2008; Krieger et al., 2010). In isopods, which are considered the most successful terrestrialized crustaceans, a transverse segmentation of the prominent AnN, as has been shown for hermit crabs, is clearly identifiable in marine but indistinct in terrestrial isopods (Harzsch et al., 2011). Since in terrestrial isopods, the first pair of antennae is highly reduced in size and the associated DCL seems to be absent, the idea arose that the pronounced second pair of antennae and its associated antenna 2 neuropils together may function as the major sensory organ (Harzsch et al., 2011; Kenning & Harzsch, 2013). Contrarywise, in both marine (Krieger et al., 2012) as well as terrestrial brachyurans, the antenna 2 neuropil is part of a large neuropil mass that is composed of both deuto- and tritocerebral portions, thus making the antenna 2 neuropil hardly identifiable. A transverse segmentation has not been detected so far, neither in marine nor in terrestrial brachyurans. From these data and especially from behavioral observations in Brachyura, one could argue that flow detection on land would have to be realized by other body parts rather than in the second pair of antennae. Undoubtedly, further analyses are required to clarify the functional relevance of the second pair of antennae and to check for structural as well as functional differences that may represent adaptations for detecting flow in water versus in air.