Developmental and sexual divergence in the olfactory system of the marine insect Clunio marinus

An animal’s fitness strongly depends on successful feeding, avoidance of predators and reproduction. All of these behaviours commonly involve chemosensation. As a consequence, when species’ ecological niches and life histories differ, their chemosensory abilities need to be adapted accordingly. The intertidal insect Clunio marinus (Diptera: Chironomidae) has tuned its olfactory system to two highly divergent niches. The long-lived larvae forage in a marine environment. During the few hours of terrestrial adult life, males have to find the female pupae floating on the water surface, free the cryptic females from their pupal skin, copulate and carry the females to the oviposition sites. In order to explore the possibility for divergent olfactory adaptations within the same species, we investigated the chemosensory system of C. marinus larvae, adult males and adult females at the morphological and molecular level. The larvae have a well-developed olfactory system, but olfactory gene expression only partially overlaps with that of adults, likely reflecting their marine vs. terrestrial lifestyles. The olfactory system of the short-lived adults is simple, displaying no glomeruli in the antennal lobes. There is strong sexual dimorphism, the female olfactory system being particularly reduced in terms of number of antennal annuli and sensilla, olfactory brain centre size and gene expression. We found hints for a pheromone detection system in males, including large trichoid sensilla and expression of specific olfactory receptors and odorant binding proteins. Taken together, this makes C. marinus an excellent model to study within-species evolution and adaptation of chemosensory systems.

a combination of circalunar and circadian clocks, which have made C. marinus a long-standing object of chronobiological research 5,6 .
Finally, C. marinus adults are strikingly sexually dimorphic and display a curious mating behaviour 7,8 . The females have a very cryptic lifestyle: they are wingless and have only small legs, eyes and antennae. They are often unable to shed the pupal skin on their own. Female pupae float on the water surface until they are located by swarming males. Males patrol the water surface much like hovercrafts, with their legs resting on the water and their wings serving merely as propellers. The males free the females from the pupal skin in a stereotyped behaviour involving their large forceps-like hypopygium, the modified last abdominal segment. After a few seconds this directly ends in copulation. In that way males ensure to copulate with virgin females, likely enhancing their reproductive success. After mating, the males carry the females during a short nuptial flight and eventually deposit them on the larval substrates, which are exposed by the low tide and where oviposition takes place. Both sexes die in the rising tide. During the mating procedure, the males depend on the ability to localize the female, which could be triggered by visual, tactile, but also olfactory cues. Males are also involved in choosing the oviposition site. We may expect that the sexual dimorphism also extends to the olfactory system, with males having to perform a greater variety of olfactory tasks.
In order to investigate if the characteristic ecology and life-cycle of C. marinus are mirrored in the olfactory system at the species-level, but also within the species at the developmental and sex level, we used an array of morphological and genomic tools. The combination of scanning electron microscopy (SEM), immunohistochemistry and antennal backfilling, as well as analysis of the recently published C. marinus genome 6 and RNAseq data allowed us to provide the first detailed description of a chironomid's olfactory system, compare the olfactory system of C. marinus larvae, adult males and adult females and estimate the importance of chemoreception for the extreme specialist C. marinus.

Results
Morphology. Periphery: antennae and maxillary palps. In order to see how the olfactory system of C. marinus differs between life stages and between the sexes, we investigated it at several levels. We started in the periphery by examining the main olfactory organs, the antennae and the maxillary palps. Several olfactory sensillum types have been described for dipterans, including basiconic, coeloconic, trichoid sensilla and sensory pits [9][10][11][12] . Especially large trichoid sensilla have been suggested to play a role in sex recognition and pheromone detection 13 . Antennae and maxillary palps of C. marinus larvae, adult males and females ( Fig. 1A-C; red colouration) were screened for the presence of putative olfactory sensilla using scanning electron microscopy (SEM) and confocal laser scanning microscopy exploiting the autofluorescence of the insect cuticle.
Adults: A sexual dimorphism was already apparent in the antennal composition. Female antennae were composed of only 7 annuli (Figs. 1A, 2A), whereas male antennae had 11 annuli (Figs. 1B, 2A). This difference was sustained in sensillum numbers. Males had significantly more small basiconic sensilla (sbs) and small trichoid sensilla (sts) (Fig. 2B). As only one large basiconic sensillum (lbs) was found per annulus, overall the males also had more lbs due to the different number of annuli. For sbs and sts the highest abundance was found on the distal annulus (Fig. 2B, Suppl 1B,C), which in males was much bigger than in females (Figs. 1A,B, 2A). Interestingly, on this annulus the number of sensilla seemed to be not stereotyped: The bigger the proximal annulus, the higher the number of sensilla. Large trichoid sensilla (lts; Fig. 1b1) and mechanosensory chaetic sensilla (cha; Fig. 1A,B) were distinguished based on their different base structures. Cha arise from a round socket that allows movements of the sensilla (Suppl 2A), whereas the wall of lts is continuous with the antennal cuticle (Suppl 2B,C). SEM micrographs could not resolve a porous cuticular wall in C. marinus lts (Suppl 2b). However, lts were completely absent in females (Fig. 2B), suggesting they may play a role in sexual communication, as they do in Drosophila 13 . For all other sensillum types we found pores in the cuticle, supporting an olfactory function (Fig. 1a3,b3. Suppl. 2C,c). In contrast to the antennae, maxillary palps of adults showed no obvious sex differences (Fig. 1A,B,a4,b4). In both sexes the only putative palp olfactory sensilla were basiconic sensilla.
Larvae: The larvae do not show any apparent sexual dimorphism. Porous sensilla were detected on both the antennae and the palps (Fig. 1C). On the antennae these structures belong to the Lauterborn organ (LO) and an additional basiconic sensillum (Fig. 1c1,c3). Three putative olfactory basiconic sensilla were detected on the maxillary palp (Fig. 1c4, the sensillum pair is referred to as "bisensillum" in other chironomids 14 ). Most other sensilla were sensillum styloconicum-like or poreless basiconic sensilla (referred to as "a and b setae" 14 ).
Orco staining: To further investigate a putative olfactory function of antennal sensilla, we used an antibody designed against a conserved motif of the olfactory receptor coreceptor (Orco) 15 . Insect olfactory receptors (ORs) function as heteromultimers composed of a specific ligand binding OR and Orco 16,17 . In contrast to the ligand specific ORs, Orco is highly conserved across insects and is co-expressed in all OR-expressing olfactory sensory neurons (OSNs) 18 . We found Orco-immunoreactivity in OSNs in both male and female antennae (Fig. 3A,B). In adults the antibody labelling is best visible in the dendrites of large basiconic sensilla (Fig. 3a1,B,b1,b3). This is probably due to the fact that the large basiconic sensilla of C. marinus house a higher number of OSNs compared to small basiconic sensilla, as suggested by anti-HRP staining as a general neuronal marker (Fig. 3A,a1). OSN somata displayed weak labelling ( Fig. 3a1,b1,b3). The adult maxillary palps did not show Orco-immunoreactivity. During preparation the palps were left attached on the head and this may have prevented diffusion of the antibodies into the palps. In the larvae dendrites and somata were Orco-immunoreactive (Fig. 3c). In larvae we were able to count the exact number of Orco-immunoreactive OSNs on antennae and palps: There are 17 cells on the antennae with dendrites extending into the Lauterborn organ (Fig. 3c,c1,c2) and two on the palps (Fig. 3c3,c5). The dendrites of the latter two neurons extended into the porous basiconic sensillum that is not belonging to the bisensillum (Fig. 3c4), suggesting the bisensillum does not serve olfactory functions. The female antenna is shorter than the male antenna and consists of fewer antennal segments. Many putative olfactory sensilla can be found on the last antennal segment of antennae of the female (a1) and the male (b1). These include small trichoid sensilla (sts), as well as putative olfactory coeloconic sensilla (cs: a2, b2), large basiconic sensilla (lbs) and small basiconic sensilla (sbs: a3, b3). Large trichoid sensilla (lts) were only found on male antennae (B, b1). Small basiconic sensilla, together with mechanosensory chaetic sensilla (cha) were also found on the maxillary palps of both sexes (a4, b4). Larval antennae are short. The prominent Lauterborn organ (LO) has pores (c1, c2) and might fulfil a chemosensory function. Additionally a basiconic sensillum (bs) can be found (c3). The larval maxilla also bears multiporous basiconic sensilla (c4). Arrowheads in a3, b3, c2 and c3 point towards the pores in the surface of basiconic sensilla. Scale bars: A, B, C: 50 µm; a1, b1, c1, b4: 10 µm; a4, c4: 5 µm; c3: 2 µm; a2, a3, b2, c2: 1 µm; b3: 200 nm.
Brain. Generally, insect OSNs that express the same set of chemoreceptors converge in the same region in the antennal lobe (AL) and form small spherical-like neuropils called glomeruli together with higher order neurons 19 . The number of glomeruli is therefore a good estimate for the number of receptors and for olfactory complexity 20 . We visualized the anatomy of the olfactory brain centres of C. marinus by using an antibody against presynaptic proteins (anti-synapsin) as general markers for neuropil structures 21 .
Adults: We were not able to identify a clear glomerular structure of the deutocerebral antennal lobes of C. marinus, neither in females (Fig. 4A) nor in males (Fig. 4B). This is much in contrast to other Diptera. In order to exclude technical limitations due to the size of the brains and the staining technique, we also performed antennal backfilling of male C. marinus (Suppl 3A,C). Again, we could not differentiate distinct glomeruli in the AL (Suppl 3C). So clearly, in C. marinus glomerular input regions are at least not as pronounced as in other insects. Male deutocerebral neuropils appeared bigger than those of the females. The mushroom body calices and lobes, which are brain centres associated with olfaction, are also comparatively small in C. marinus and especially in females (Fig. 4a1,a2,b1,b2). Furthermore, the antennal backfill highlighted strong contralateral connections of the  ORs and GRs belong to the arthropod chemoreceptor superfamily, which are ion channels that in structure look like seven-transmembrane-domain receptors with an inverted topology compared to G-protein coupled receptors 22,23 . To identify members of the three receptor families we made use of the recently published C. marinus genome 6 . Additionally, we generated RNAseq datasets of adult males, adult females and larvae to improve our gene models and to get a first impression of gene expression in the different life stages and sexes.
Olfactory receptors (ORs). While the olfactory receptor coreceptor (Orco) 24 is highly conserved among insects, the ligand specific OR genes exhibit very little sequence similarity between insect orders, and even within the same insect order 18 . This likely reflects lineage-and species-specific chemosensory adaptations. In the C. marinus genome we annotated 66 OR genes (including the Orco), which is a similar number of ORs as in drosophilid flies. Phylogenetic analysis of C. marinus OR candidates along with the set of Drosophila melanogaster and Anopheles gambiae ORs revealed several lineage specific expansions with up to 19 ORs (Fig. 5A). Beside the Orcos of the different species, only AgamOr11, DmelOr56a and CmarOr47 formed a distinct cluster of orthologous receptors (Fig. 5A, dark grey shading). Across the genus Drosophila this particular OR is involved in the geosmin detection system, which senses the presence of harmful microbes 25 . It is possible that this important function is conserved in other dipetrans.
The identification of a similar number of ORs as compared to other dipterans was somewhat discrepant from the comparably simple morphology of the olfactory centres in the C. marinus brain. In order to get an idea of whether or not OR genes are expressed, we generated RNAseq data of heads and bodies of C. marinus adult males, adult females and larvae. In general, gene expression was very low for all receptors, independent of life stage, sex or tissue (Suppl 4A). Despite the limited data, there are several hints suggesting that the sexual dimorphism that was already found at the morphological level is also present at the level of gene expression: Orco expression was detected at about 6-fold higher levels in males compared to females (Suppl 4A). Many ORs do not have detectable expression, but the number of undetectable ORs is much larger in female heads (n = 49) compared to male heads (n = 23) (Suppl 4A). Finally, CmarOr45 is highly expressed specifically in adult male heads (Suppl 4A), which makes it a candidate for being a pheromone receptor.
Our RNAseq data also suggested the expression of various ORs outside the head of C. marinus, possibly being expressed in the legs or genitalia structures of adults or in some sensilla on the larval body. Our results do not rule out that more ORs are expressed at very low levels and an exact quantification will only be possible with additional data.
Gustatory receptors (GRs). GRs belong to the arthropod chemoreceptor superfamily just as the ORs, but in contrast to ORs no universal coreceptor is known. Some GRs are known to bind volatile ligands such as CO 2 , but usually they bind compounds dissolved in aqueous solutions. Additionally, there are certain GR lineages that fulfil a conserved function, such as the sugar, fructose, bitter and CO 2 receptor lineages [26][27][28][29] .
In the C. marinus genome we found 47 GR candidates (Fig. 6). We found direct orthologues for all three insect CO 2 receptors and the bitter receptor DmelGr66a. In contrast, there is no orthologue to the D. melanogaster fructose receptor (DmelGR43a) and only two receptors belonging to the sugar receptor group were identified (CmarGR1 and CmarGR2). Expression of these sugar receptors seemed to be largely restricted to the larval stage (Suppl 4B), correlating with the fact that adults do not feed anymore. Similar to the ORs, GR expression levels were generally extremely low (Suppl 4B). The GR family also showed lineage specific expansions (Fig. 6). The genomic locations of both OR and GR genes indicate that these genes are frequently arranged in tandem arrays of up to eight closely related receptors (Fig. 5B). This supports the birth-and-death model previously suggested for chemosensory gene families 30 . The current data does not allow assessing the fraction of pseudogenes.
Variant ionotropic receptors (IRs). The third insect chemoreceptor gene family is the IR family, ligand gated ion channels derived from ionotropic glutamate receptors (iGluRs) 31 . D. melanogaster IRs play a role in gustation 32-34 , salt detection 35 , thermosensation 36,37 and hygrosensation 37 . A number of D. melanogaster IRs are involved in olfaction, namely members of the Ir75 clade, Ir41a, Ir92a, Ir84a, and Ir76a 38 . Their main ligands are small water soluble components such as amines and acids, which have an important role in many marine ecosystems 39 . Compared to ORs and GRs, antennal IRs show higher sequence conservation across species 33,40 .
As candidates for olfactory IRs, we found three members of the Ir75 clade and one member of the Ir41a/Ir76a/ Ir92a clade, but no orthologue to Ir84a. Ir8a, Ir25a and Ir76b are considered olfactory IR co-receptors, which similar to the Orco for ORs form heteromultimers with other IRs, but do themselves not mediate specific odorant www.nature.com/scientificreports www.nature.com/scientificreports/ binding 41 . The orthologues of these three IR co-receptors show relatively high expression in C. marinus without obvious sex differences (Suppl 5). This is in line with the finding that coeloconic sensilla, which usually host most IR expressing neurons, did not differ in number between males and females. www.nature.com/scientificreports www.nature.com/scientificreports/ Binding proteins (OBPs and CSPs). In addition to the chemoreceptors, we annotated other gene families involved in olfaction or gustation: odorant binding proteins (OBPs), chemosensory proteins (CSPs) and sensory neuron membrane proteins (SNMPs). The OBP and CSP gene families code for proteins that reversibly bind small ligands, including odorants and tastants 42 . Both are expressed in olfactory and gustatory sensilla 43,44 , but are not restricted to chemosensory tissues [45][46][47][48] . In the C. marinus genome we annotated 41 OBP genes. Based on their cysteine patterns 17 C. marinus OBP candidates were assigned to the Plus-C OBPs, two to the Minus-C OBPs, one to OBP-dimers and three to the antennal binding proteins II (Fig. 8A). In contrast to Plus-C OBPs, C. marinus Minus-C and OBP-dimers likely evolved independently from those of D. melanogaster and A. gambiae (Fig. 8A), as was described for other species 49 . OBPs in general display very low overall sequence conservation 42 and the only OBPs with clear orthology relationships across insects are Obp59a and Obp73a 50,51 . In C. marinus we found a putative orthologue of Obp59a (CmarObp9), but not Obp73a (Fig. 8A). We also found a putative orthologue of the D. melanogaster binding protein LUSH/DmelObp76a, which is CmarObp38 (Fig. 8A). LUSH has been shown to be essential for the detection of 11-cis vaccenyl acetate, one of the D. melanogaster pheromones 52 . In line with that CmarObp38 seems highest expressed in male heads (Fig. 8C). In general, OBPs were much higher expressed www.nature.com/scientificreports www.nature.com/scientificreports/ than receptors. Furthermore, most OBPs showed strong expression differences between larval and adult heads (Fig. 8C), suggesting that marine larvae and terrestrial adults may use different subsets of OBPs to bind waterborne vs. airborne substances. www.nature.com/scientificreports www.nature.com/scientificreports/ With six CSP genes C. marinus has a similar number of CSPs as A. gambiae (8), but less than C. quinquefasciatus (21) or Aedes aegypti (18) 53,54 (Suppl 6A). The conserved cysteine pattern, with 4 cysteines was found in all translated coding regions. CmarCSP3 and CmarCSP4 are much longer than previously described CSP proteins www.nature.com/scientificreports www.nature.com/scientificreports/ (300 and 616 AA), but there is no indication for misannotation and both are supported by transcript sequences. One of the C. marinus CSPs belonged to the 5-helical CSPs, whereas the other five grouped within the 6-helical CSPs (Suppl 6A). CmarCSP1 was strongly expressed in adult heads, whereas CmarCSP2 showed the highest expression in larval heads (Suppl 6B). This may suggest that as for OBPs different CSPs are used in marine larvae vs. terrestrial adults. CmarCSP6 was highly expressed in all tissues (Suppl 6B).

Sensory neuron membrane proteins (SNMPs)
. SNMPs belong to a broad family of receptor proteins homologous to the human CD36 class of scavenger B1 receptor proteins, involved in signalling, transport, and internalization of lipid molecules 55,56 . In insects, SNMPs are characterized as olfactory receptor co-factors, which are required for normal response of OSNs particularly in pheromone detection 57,58 , but are not restricted to pheromone sensitive neurons 59 .
In C. marinus we found three SNMP genes, two belonging to the SNMP1 and one to the SNMP2 cluster (Suppl 6C). SNMP2 was found in two isoforms. CmarSNMP1-1 was highest expressed in male heads, but was also expressed in female heads and male body, containing legs and wings (Suppl 6D). CmarSNMP1-2 was expressed in both male and female heads, but barely in other tissues. CmarSNMP2 had a broad expression. Generally, our RNAseq data supported a role of SNMPs outside the head (Suppl 6D).

Discussion
The life cycle of C. marinus is very special in several aspects: The larvae are truly "submarine", which is very rare among insects, as most insects associated with marine habitats are living in bridging habitats like salt marshes 60 . This implies that the transition from larvae to adults is also a transition from marine to terrestrial habitats, on top of the transition from a feeding to a reproductive stage. In C. marinus the latter transition is particularly pronounced, as adults are extremely short-lived, and dedicate the few hours they live exclusively to reproduction. Finally, the adults are strongly sexually dimorphic, which likely facilitates successful reproduction. Because of these large differences between the tasks and ecological requirements of the life stages and sexes, we can expect that studying the olfactory system of C. marinus may reveal general patterns and mechanisms in the evolutionary specialisation of olfactory systems.
Living in a marine environment. Adaptions to the marine habitat might be represented by expansions of certain gene families. As species-specific expansions in ORs and GRs are common, reflecting their birth-and-death mode of evolution, it is difficult to infer their ecological significance without functional testing. Still, the expansion of the IR7 gene family to 47 members in C. marinus is extraordinary. In other dipterans, members of this clade are expressed in larval and adult gustatory organs 33,61 . Thus, C. marinus IR7 receptors are the prime candidates for gustatory adaptation to the marine habitat.
In C. marinus, larval life lasts from 6 weeks to 1 year, depending on latitude. Adult life is reduced to a few hours. This divergence is reflected in the fact that adult olfactory centres are very simple compared to other dipterans, whereas the larval olfactory system seems normally developed. Larval antennae and palps are equipped with all common larval olfactory sense organs and show a relatively high number of Orco-positive neurons. In A. gambiae each larval OSN expresses the Orco and one specific OR, which are 12 in total 62 . In C. marinus we found 17 Orco-positive neurons in the antennae and two in the maxillary palps. We did not find a glomerular structure in the larval deutocerebrum, in line with findings in other chironomids 63 . However, in some midges synaptic markers, such as the synapsin antibody we employed, may not necessarily display glomerular structures even if present 64 .
Adults of C. marinus do not fly. They either hover on the water surface or crawl on the exposed algae and rocks, maintaining contact with water at all times. Thus, with the transition from marine larval life to adult life on the water surface, airborne chemosensory stimuli add to waterborne stimuli, but do not replace them. In line with that, there is a large overlap of expressed receptors between larvae and adults, suggesting that some olfactory tasks may be similar. For example, adults need to identify suitable algae for oviposition, and these algae are also the places where the larvae live and feed. In contrast, some CSPs and many OBPs seem to be expressed at different levels in larvae compared to adults, suggesting that adaptations to the detection of airborne vs. waterborne stimuli may rather happen at the level of odorant/tastant carriers, not necessarily at the level of receptors. This seems reasonable, as many substances may be transmitted both by water and air and can be detected with the same receptor, but need to be taken up from different media.
Sexual dimorphism in the olfactory system. The difference in behavioural tasks performed by males and females is reflected at all levels of the olfactory system. Starting from the peripheral sensory organs, females are characterized by shorter antennae, with a smaller number of most sensillum types, as well as the complete absence of large trichoid sensilla. The reduction in sensillum types is reflected in lower Orco expression in female heads. Finally, olfactory brain centres are smaller in females.
Males probably perform more complex olfactory tasks, as they need to locate the females and navigate to the oviposition sites during the nuptial flight. The presence of large trichoid sensilla, highly expressed SNMPs and LUSH, three important elements of pheromone detection systems, suggest the presence of pheromones in C. marinus. As females are basically immobile and often cannot shed the pupal skin without male assistance, the most likely scenario is that females emit a pheromone to make sure that they are located by males. Signals could be directly emitted by the females or could be water-insoluble compounds evaporating from the pupal skin, when the pupae reach the water surface. Future investigations on the presence of a C. marinus pheromone will help to understand, if the sex-specific differences in the olfactory system are only due to the different behavioural tasks or also due to the presence of sex-specific pheromones. (2020) 10:2125 | https://doi.org/10.1038/s41598-020-59063-7 www.nature.com/scientificreports www.nature.com/scientificreports/ The very passive role of females in the mating behaviour makes female choice unlikely. But once females are deposited on the algal substrates they seem to choose the exact oviposition site. This choice may involve the detection of algal odours or of odours emitted by eggs of conspecifics or larval faeces, as was described in other nematocera 65 . ORs expressed in the female abdomen may play a role in egg laying behaviour.
Morphology vs. molecular biology. At first sight, it seems contradictory that the antennal lobe (AL) of C. marinus shows no glomeruli, which would suggest a very simple olfactory system, but at the same time a normal number of sensilla and olfactory receptors were found. But interestingly, aglomerular ALs have been described for other insects that spent most of their life as aquatic larvae and have only a short adult life-span, e.g. the chironomid Chironomus dolichotomus, Ephemeroptera, Odonata, Plecoptera and some Trichoptera 63,66 . In contrast, dipterans with an aquatic development but long adult lifespan possess a glomerular antennal lobe (e.g. A. gambiae, A. aegypti). Possibly, the typical glomerular organization is only necessary when a long adult life requires signal modulation and integration in the course of learning. C. marinus adults live only for a few hours and spend this time exclusively for reproduction. Their reproductive behaviours are highly stereotyped 8 , making it possible that the respective olfactory tasks consist mostly of hard-wired innate responses that do not require learning. This may also be reflected in the small size of the mushroom bodies, especially in females. Furthermore, a lack of a glomerular AL is often correlated with a pronounced reduction or even loss of the mushroom body calyx 67 , as we also found in C. marinus.
Finally, a relatively high number of OR genes seems to be expressed in tissues other than the head, especially in the larval abdomen. This may be explained by the presence of olfactory sensilla in the abdomen or by a possible co-option of ORs into non-olfactory functions. Both scenarios would be worth further investigation.

conclusions
Our detailed characterization of the C. marinus olfactory system underlines that not only species, but also different life stages and sexes of the same species, are equipped with sensory systems that are tuned to their specific needs: In C. marinus, the long-lived larvae show a normally developed olfactory system, whereas in the short-lived adults the olfactory system is much simpler. Differences in gene expression between the life stages may reflect their marine vs. terrestrial life-style. The olfactory systems of males and females are strikingly sexually dimorphic, much in line with the highly specific tasks they have to fulfil during the stereotyped mating behaviour. Taken together, this suggests that C. marinus is an excellent model for further studies on the evolution and adaptation of olfactory systems.

Methods
Animal rearing. All experiments were performed on individuals from the Jean laboratory strain of Clunio marinus, which was also used to establish the C. marinus reference genome. The laboratory stock was bred according to Neumann 2 . Briefly, animals were kept in 20 × 20 × 6 cm plastic containers with sand and natural seawater diluted to 15‰ with desalted water. They were fed diatoms (Phaeodactylum tricornutum, strain UTEX 646) in early larval stages and nettle powder in later stages. Temperature in the climate chambers was set to 20 °C and the light dark cycle (LD) was 16:8 hours. All animals were subject to an artificial moonlight cycle with an incandescent flashlight bulb (about 1 Lux) switched on all night for four successive nights every 30 days. Because of the synchronized development, adults were always available for sampling at the same time of the artificial moonlight cycle (days 12 to 15) and of the LD cycles (zeitgeber time 19 to 22). Larvae were sampled at different times of the artificial moonlight cycles, depending on the experimental requirements.
Scanning electron microscopy (SeM). For investigation of peripheral olfactory structures we used female, male and larval C. marinus preserved in 70% ethanol. Tissue was dehydrated in an ascending ethanol series (80%, 90% 96%, 3 × 100% ethanol, 10 min each), critical point dried (BAL-TEC CPD 030, Bal-Tec Union Ltd., Liechtenstein), mounted on aluminium stubs with adhesive tape and sputter coated with gold on a BAL-TEC SCD005 (Bal-Tec, Liechtenstein). Specimens were examined in a LEO 1530 Gemini scanning electron microscope (Zeiss, Germany) set at 10 kV and 11 to 16 mm working distance.
Autofluorescence for sensillum count. The sensillum numbers of 8 male and 8 female antennae were counted using the high cuticular autofluorescence at 488 nm excitation. Heads or single antennae of ethanol fixed material were washed several times in 0.1 M phosphate-buffered saline (PBS, pH 7.4, Sigma Aldrich, USA) and mounted in 50% glycerol in PBS between two cover slides. For visualization samples were scanned using confocal laser scanning microscopy (LSM 880, Zeiss, Germany). Cuticle was exited using the 488 nm line of the microscope's argon laser. All pictures were taken using a 40 × W objective (C-Apochromat 40x/1.20 W Korr M27). Scanning resolution was set to 1024 × 1024 pixel. Because of the small size of the antennae it was possible in most of the cases to scan through the complete antennae and visualize both front and back side. In all other cases the object was inverted and rescanned. By using landmarks such as big sensilla it was possible to combine both data sets for counting. www.nature.com/scientificreports www.nature.com/scientificreports/ goat-anti mouse antibody linked to Alexa Fluor 546 at a dilution of 1:200 (Invitrogen, USA) for 3 days at 4 °C. The incubation was followed by washing steps as described above. For visualization, samples were mounted in 50% glycerol on a microscope slide and scanned using confocal laser scanning microscopy (LSM 510 Meta or LSM 880, Zeiss, Germany). Alexa 546 was exited using the Helium Argon 543 laser. Signals were detected by a spectral detector (555-681 nm). All pictures were taken using a 40 × W objective (C-Apochromat 40x/1.2 W UV-VIS-NIR or C-Apochromat 40x/1.20 W Korr M27). Scanning resolution was set to 1024 × 1024 pixel. 3D reconstructions were prepared with Amira 4.1.1 (Mercury Systems, Germany). olfactory receptor coreceptor (orco) immunohistochemistry. Antennae and whole heads of fixed male, female and larval C. marinus were treated as described in the part brain morphology. The primary antibody against Orco (originally termed anti-R2, kindly provided by Prof. Jürgen Krieger, University of Halle-Wittenberg, Germany) was raised against a 17 amino acid peptide (NH2-NQSNSHPLFTESDARYH-COOH) that is fully conserved in the Bombyx mori, Heliothis virescens and Manduca sexta Orcos 15,68 . This peptide is also highly conserved in the C. marinus Orco, with only two conservative amino acid replacements (NH2-NQSNSHPLFTESNSRYH-COOH). A BLAST search of M. sexta Orco against all C. marinus proteins shows that the two species' Orcos have 65% amino acid identity, while the second best blast hit (CmarOR1) only has 27% identity. In a BLAST search of the 17 amino acid peptide against all C. marinus proteins only Orco had a significant BLAST hit. We therefore do not expect the antibody to cross react with antigens other than Orco. The anti-Orco antibody was applied at a 1:500 dilution in 2% NGS-PBST and incubated for 4 days at 4 °C. Detection of the Orco antibody was performed by incubating in a goat-anti rabbit antibody linked to Alexa Fluor 488 at a dilution of 1:200 (Invitrogen, USA) for 3 days at 4 °C. In addition, we added goat anti-horseradish peroxidase (anti-HRP) antibodies conjugated to Cy3 (Jackson Immuno Research, USA) at a dilution of 1:50 to visualize neuronal tissue and Hoechst33342 (ThermoFisher Scientific, Germany) as nuclear marker (1:2000). For visualization, samples were mounted in 50% glycerol on a microscope slide and scanned using confocal laser scanning microscopy (LSM 880, Zeiss, Germany). Alexa Fluor 488 was exited using the 488 nm line of the microscope's argon laser, while a Helium Argon 543 laser was used to activate Cy3. Signals were detected by a spectral detector (quasar: 490-553 nm and 555-681 nm). Hoechst was excited by a laser diode at 405 nm. Pictures were taken using a 40 × W (C-Apochromat 40x/1.20 W Korr M27) or 63 × W objective (NCI PlanNeofluar 63x/1.3 Dic Imm Korr). Scanning resolution was set to 1024 × 1024 pixel.

Antennal nerve backfill.
For several living males one antenna was cut carefully without destroying the cuticle of other body parts. The animals were placed in 1xPBS (without detergent) and carefully centrifuged until the animals were submerged. The animals were incubated in a micro ruby solution (Tetramethylrhodamin and biotin; ThermoFisher Scientific, Germany) for two hours at room temperature and immediately fixed in 4% PFA in 0. Annotation of olfactory genes. Annotation of olfactory genes was based on the CLUMA1.0 reference genome. For all examined gene families (ORs, GRs, IRs, OBPs, CSPs, SNMPs) the respective set of D. melanogaster genes served as a BLAST query to retrieve C. marinus homologs from the CLUMA1.0 gene models (e-value cut-off e −2 ). Then putative C. marinus homologs were blasted against NCBI's nr database and only retained if the best hit in another species was a member of the gene family under investigation. In a second step, all identified C. marinus homologs served as BLAST queries against the CLUMA1.0 gene models to obtain additional homologs, which were also checked for being members of the respective gene family as described above. Finally, all gene models were manually curated and improved by comparison to the available transcript data and protein sequences from other insects, as were used for automated annotation of the C. marinus reference genome 6 . ORs, GRs and IRs were further checked for the presence of a 7tm domain, OBPs and CSP for their characteristic cysteine residue patterns. A few gene models are partial. The resulting AA sequences are provided in the supplement and the corrected gene models will be published with the second release of the C. marinus reference genome (manuscript in preparation).
Alignments with other dipteran members of the respective gene families were carried out using the MAFFT plugin in Geneious Pro 5.0.4 (E-INS-I parameter set; MAFFT 7.453 for the OR tree) 69,70 . Dendrograms were calculated using maximum likelihood analysis with FastTree2 71,72 and displayed and edited with FigTree (http://tree. bio.ed.ac.uk/software/figtree). Candidates were named with the abbreviation for the gene family and ascending numbers with the exception of receptors were a clear homology could be assigned.
RnA isolation. RNA was extracted from larval, male and female heads and abdomen. Each of the total of six samples consisted of at least 20 pooled individuals. Total RNA was isolated from sample material homogenized in TriReagent (Molecular Research Centre, Cincinnati, OH, USA) using ball bearings and a TissueLyzer (Qiagen, Hilden, Germany) and subsequently purified using the DirectZol kit (Zymo Research) following the manufacturers' guidelines. The integrity of the RNA was verified using an Agilent 2100 Bioanalyzer and a RNA 6000 Nano Kit (Agilent Technologies, Palo Alto, CA). The quantity of RNA was determined using a Nanodrop ND-1000 spectrophotometer. Since the goal of the experiment was not to identify differences between individuals, but (2020) 10:2125 | https://doi.org/10.1038/s41598-020-59063-7 www.nature.com/scientificreports www.nature.com/scientificreports/ rather to obtain a robust tissue-specific gene expression map, we have opted for a pooling strategy, combining large numbers of individuals per sample to massively reduce biological variability and the impact of individual outliers on observed gene expression differences.
RnA-Seq and expression analysis. RNA-Seq was carried out for 6 different RNA samples using poly(A) + enriched RNA fragmented to an average of 150 nucleotides. Sequencing was carried out by the Max Planck Genome Centre Cologne (MPGCC) on an Illumina HiSeq 2500 Genome Analyzer platform using single read (1 × 100 bp) technology. This yielded approximately 25 million reads for each of the 6 samples. Quality control measures, including the filtering of high-quality reads based on fastq file scores, the removal of reads containing primer/adapter sequences, and trimming of the read length, were carried out using CLC Genomics Workbench v9.1 (http://www.clcbio.com). The same software was used for de novo transcriptome assembly, combining all 6 RNA-Seq samples, and selecting the presumed optimal consensus transcriptome as previously described 73 . The final de novo reference transcriptome assembly of C. marinus contained 27,239 contigs (minimum contig size = 200 bp) with a maximum contig length of 17,430 bp. The transcriptome was annotated using BLAST, Gene Ontology and InterProScan searches implemented in BLAST2GO PRO v4.1 (www.blast2go.de) 74 and used to verify some of the olfactory genes based on the CLUMA1.0 reference genome.
Digital gene expression analysis was carried out using CLC Genomics workbench to generate BAM mapping files, QSeq (DNAStar Inc.) to remap the Illumina reads from all 6 samples onto the CLUMA1.0 reference transcripts (with manually curated and corrected olfactory gene sequences), and finally by counting the mapped sequence reads to estimate the expression levels, using previously described parameters for read mapping and normalization (for a detailed description of the procedure see 73,75 ).
The short read data have been deposited in the EBI short read archive (SRA) with the following sample accession numbers: ERS2954175-ERS2954186. The complete study can also be accessed directly using the following URL: http://www.ebi.ac.uk/ena/data/view/ PRJEB30068. ethics approval. There are no regulations on work with C. marinus and thus no ethics approval was required.
Nevertheless, we have minimized the use of animals to the numbers that were absolutely required.

Data availability
As NCBI or ENA do not accept revised gene models, all protein sequences of olfactory genes as used in and revised for this study are given as Supplement 7. They will be included in the next release of the C. marinus reference genome. The short read data have been deposited in the EBI short read archive (SRA) under accession numbers ERS2954175-ERS2954186, project number PRJEB30068.