Multisensory integration by polymodal sensory neurons dictates larval settlement in a brainless cnidarian larva

Multisensory integration (MSI) combines information from more than one sensory modality to elicit behaviours distinct from unisensory behaviours. MSI is best understood in animals with complex brains and specialized centres for parsing different modes of sensory information, but dispersive larvae of sessile marine invertebrates utilize multimodal environmental sensory stimuli to base irreversible settlement decisions on, and most lack complex brains. Here, we examined the sensory determinants of settlement in actinula larvae of the hydrozoan Ectopleura crocea (Cnidaria), which possess a diffuse nerve net. A factorial settlement study revealed that photo‐, chemo‐ and mechanosensory cues each influenced the settlement response in a complex and hierarchical manner that was dependent on specific combinations of cues, an indication of MSI. Additionally, sensory gene expression over development peaked with developmental competence to settle, which in actinulae, requires cnidocyte discharge. Transcriptome analyses also highlighted several deep homological links between cnidarian and bilaterian mechano‐, chemo‐, and photosensory pathways. Fluorescent in situ hybridization studies of candidate transcripts suggested cellular partitioning of sensory function among the few cell types that comprise the actinula nervous system, where ubiquitous polymodal sensory neurons expressing putative chemo‐ and photosensitivity interface with mechanoreceptive cnidocytes. We propose a simple multisensory processing circuit, involving polymodal chemo/photosensory neurons and mechanoreceptive cnidocytes, that is sufficient to explain MSI in actinulae settlement. Our study demonstrates that MSI is not exclusive to complex brains, but likely predated and contextualized their evolution.

Classically, MSI studies are conducted at the neuron level where neuronal signals are processed in the brain to make behavioural decisions (Meredith & Stein, 1983;Otto et al., 2013;Stein, 1998;Stein & Stanford, 2008). However, much less research has been conducted on organisms that lack complex nervous systems or brains.
Moreover, zoospores of an Allomyces fungus utilize a multimodal system involving chemo-and phototaxis (Swafford & Oakley, 2018), suggesting the possibility that MSI may have evolved prior to the evolution of complex nervous systems.
Sensory integration is critically important for sessile marine invertebrates that utilize larvae for dispersal and to make irreversible settlement decisions. Because some sensory cues may be better indicators of site quality than others, larvae may place emphasis on select cues, leading to a hierarchy of sensory cues that determine where settlement occurs (Hodin et al., 2018;Kingsford et al., 2002;Müller & Leitz, 2002;Woodson et al., 2007). However, MSI has yet to be demonstrated in marine invertebrate larval settlement, and little is known about the potential for MSI in such organisms that lack complex nervous systems or brains.
The marine hydrozoan Ectopleura crocea is a benthic colonial species with a pan-global distribution in temperate coastal regions.
E. crocea actinulae begin settlement with a larval behaviour called nematocyte-printing, where they use tentacles loaded with cnidocytes to tether to the substrate, presumably in the context of suitable environmental cues (Yamashita et al., 2003). However, little is known about the sensory cues that determine this process, the cell types that receive such information, or the underlying genetic machinery of sensation that coordinates settlement decisions.
Here, we describe integrative studies on the sensory biology of settlement in actinulae of E. crocea. Immunohistochemical studies of larval neural network development indicate that E. crocea larvae first possess a defined nervous system complete with robust tentacular cnidocytes and sensory neurons by the actinula stage of development. Next, to identify the sensory cues involved in settlement, we performed a factorial larval settlement study that investigated the effects of individual and combined cues (e.g. light, chemical (biofilm), and mechanical/surface texture) corresponding to three prominent sensory modalities. We found strong evidence of MSI during larval settlement where the highest rates of settlement occurred in the presence of all three sensory cues and where the effects of cues changed in the presence or absence of other cues, resulting in a sensory cue hierarchy. Developmental transcriptome analyses revealed deep homological links with bilaterian sensory system development and a peak expression of sensory transduction components for each of the three modalities in actinula stage larvae. Lastly, RNA fluorescent in situ hybridization (FISH) studies localize several prominent sensory transcripts including opsin (photosensitivity), PKD2L1 and PKD1L3 (chemosensitivity), and ASIC, Piezo, and TRPA (mechanosensitivity) to sensory neurons and their attendant cnidocytes in settlementcompetent actinulae. Our results demonstrate MSI in the brainless actinula of E. crocea and suggest that this capacity is facilitated by the activities of polymodal sensory neurons with distant ancestry to unimodal primary sensory neurons known from bilaterian animals.

| Field collection of E. crocea colonies
Larvae were obtained by collecting adult E. crocea colonies at the UNH Coastal Marine Lab (CML) pier in New Castle, NH in June and July 2020. Colonies were cultured and maintained in unfiltered seawater with aeration at a 12/12 L:D cycle at 18°C as per Mackie (1966).
Colonies were undisturbed overnight to allow spawning and actinula release. Larvae were identified and collected the following day for settlement experiments, molecular analyses or staining.

| Immunohistochemistry and confocal microscopy
Immunohistochemistry was performed on four developmental stages: star embryos, pre-actinulae, actinula larvae and juvenile polyps.
Following the primary antibody, samples were washed five times with 5-min incubations in PBST and blocked as before. Samples were incubated with the secondary antibody, Alexa Fluor 546-conjugated anti-

| Experimental design of the larval settlement study and materials
The larval settlement study was designed as a 7 × 2 × 2 splitplot Randomized Complete Block Design (RCBD) with 10 blocks ( Figure 2a). The blocking variable was the experimental day that F I G U R E 1 Nervous system development of Ectopleura crocea actinulae larvae. (a, b, d-g) Immunohistochemistry (IHC) staining of four developmental stages of E. crocea larvae where red staining corresponds to acetylated alpha-tubulin immunoreactivity in neural cells, green staining corresponds to contractile F-actin in muscles and stereocilia, and blue staining corresponds to DNA in nuclei. (a) Stage 1, star embryo. (b) Stage 2, preactinula. (c) Light micrograph of an actinula larva. The arrows point to the basal protrusion (bp), tentacles (te) and the developing oral tentacles (ot). (d) Aboral end of Stage 3-4 actinula larva. Arrows point to the basal protrusion (bp) and ganglion neurons (gn) in the tentacles. (e) Oral end of actinula larva at stage 3-4. The arrow is pointing to the developing oral tentacles (ot). (f) High magnification of a stage 3-4 actinula tentacle. An abundance of cnidocytes (cn) are found at the tips of the tentacles. Additionally, ganglion neurons (gn) extend down the tentacles and connect to the actinular nerve net. (g) A metamorphosed juvenile polyp. (h) A light micrograph of an adult E. crocea polyp. The arrows point to the oral tentacles (ot), the aboral tentacles (te) and the gonophores (go). F I G U R E 2 Settlement responses of actinula larvae to three sensory cues. (a) Cartoon depiction of the larval settlement experimental design: 7 × 2 × 2 Split-plot RCBD with 10 replicates. The block variable is an experimental day and a single replicate, depicted as a grey rectangle. All four combinations of presence/absence of biofilm and surface texture (depicted as a-d treatments) were exposed to all seven light conditions in a single block (replicate). (b) The individual effect of light cues on settlement. Larval settlement is significantly higher (p = .0027) in the presence of a light cue compared to the absence of a light cue (darkness). (c) The individual effect of chemical cues (biofilm) on settlement. Larval settlement is significantly higher (p = <2e-16) in the presence of a chemical cue compared to the absence of one. (d) The individual effect of mechanical cues (surface texture) on settlement. Larval settlement is significantly lower (p = <2e-16) in the presence of a mechanical cue compared to the absence of one. (e) The effect of the seven light conditions on settlement. Larval settlement was significantly lower in darkness (p = .00272) and in red wavelengths of light (p = .01137), and significantly higher in green wavelengths of light (p = .00167). (f) The effect of chemical cues (biofilm) across the seven light conditions. There were no significant differences in settlement in any wavelength of light (blue p = .51; green p = .60; red p = .88) in the presence of a chemical cue. (g) The effect of mechanical cues (surface texture) across the seven light conditions. There were no significant differences in settlement in any wavelength of light in the presence of a mechanical cue (blue p = .99; green p = .79; red p = .79). (h) The effect of chemical and mechanical cues combined across the seven light conditions. There was a significant increase in settlement in the presence of green wavelengths of light (p = .02) and significantly lower settlement in red wavelengths of light (p = .003). (i) The interaction of chemical cues and light cues. There was no significant interaction between these two cues (p = .33). (j) The interaction between mechanical cues and light cues. There was no significant interaction between these two cues (p = .89). (k) The interaction between chemical and mechanical cues. There was a significant interaction between these two cues (p = <2e-16) with an additive effect in the presence of both cues. (l) The three-way interaction of chemical, mechanical, and light cues. There is a significant three-way interaction (p = .02) where higher settlement rates occur in the presence of all three cues compared to other combinations.
replicates were performed on, where larvae in one block were from colonies collected at the same time the day before the experiment.
The blocking variable accounted for variation between colonies and days across the study, where each block contained all 28 treatments. An experimental unit was a single petri dish that contained 10 actinula larvae. We used the area under the curve (AUC) as our response variable, which was calculated with the settlement percentage. The use of AUC as our response variable allowed repeated measures to be collapsed, providing a more robust analysis while still providing information on how the settlement rate changed over time (over seven-time points). Additionally, we considered three levels of larval metamorphosis during quantification (larval stage, settled/attached larvae, and metamorphosed juvenile polyp). The three treatment factors were the sensory cues of interest which included seven levels of light conditions including the absence of light, two levels of chemosensory treatments (presence/absence of biofilm), and two levels of mechanosensory treatments (presence/absence of surface texture). Replicates were performed in experimental chambers described below. The first randomization of the split-plot design was applied at the chamber level and entailed the assignment of the seven light conditions described in Table 1. The second randomization included the assignment of chemical and mechanical cues, which were applied at the petri-dish level in a factorial presence/ absence structure (Table 2).  Figure S1A). An additional piece of acrylic (30.48 cm × 30.48 cm) with the same thickness was placed in the centre of the box to create two separate chambers, Figure S1B. Heatless LED lights (Super bright LEDs, part #: WRLFA-RGB6W-60) were placed in the cutouts, which allowed control of four different light intensity settings for the wavelengths red (630 nm), green (520 nm) and blue (460 nm).

Light intensity was measured using an LI-COR Biosciences LI-1000
DataLogger.
To obtain chemical cues, we allowed petri-dishes to generate biofilms by placing dishes in mesh bags attached to the UNH CML pier for 1 week at a depth of 1.5 m (Lee et al., 2008(Lee et al., , 2014. The following week, Petri dishes containing natural biofilms were transported to the laboratory in seawater collected at site and were used in settlement studies immediately (Corcoll et al., 2017). To obtain mechanical stimuli, the inner surface of 100 mm × 15 mm plastic petri dishes (Thermo Scientific) were roughened prior to biofilm generation with 36-grit ceramic alumina sandpaper (Lowes; Model #: 9150-052) in a circular motion on the outer part of the dish, then with three, non-overlapping lateral motions to ensure full coverage. Dishes were rinsed in DI water and then sea water immediately prior to use.

| Larval settlement
Actinula stage larvae were collected and placed in petri dishes where each dish contained 10 larvae. Dish treatments corresponded to the predetermined randomized dish conditions for the 28 experimental treatments ( Figure 2a). Larvae were identified under a microscope, following the work of Yamashita et al. (2003), where we sought out actinula with stiff tentacles, small and circular bodies and short aboral poles.
One block (replicate) of all 28 treatment conditions began once larvae were collected at the actinula stage. Metamorphic stages were recorded at 0, 2, 4, 6, 8, 12 and 24 h. Larval quantification was assessed on a presence/absence metamorphosis scale with three levels: larvae that were still in the actinula phase and had not metamorphosed or settled; larvae that had settled (attached to the substrate but had not completed metamorphosis), and larvae that had completed metamorphosis into a juvenile polyp. From this information, we then calculated the AUC using the settlement percentages at each time point, which combined the number of settled and metamorphosed larvae (# of Settled + # of Metamorphosed/total # of larvae in dish). The following equation was used to calculate the AUC (Mukherjee et al., 2010;Shaner, 1977): where y i is the proportion of settled and metamorphosed larvae at the i th time point; t i is the timepoint in hours where larvae were observed, and n is the total number of observations per petri-dish in a replicate.
Normally distributed data were compared statistically by a split-plot RCBD three-way analysis of variance (ANOVA) and with orthogonal contrasts in the R environment (R Core Team, 2022).

| Library preparation, sequencing and read processing
Colonies of E. crocea were collected at the UNH CML pier in May and June 2019. We collected six replicates of the six developmental stages for sequencing: embryos, preactinula, actinula, settling actinula, settled actinula and metamorphosed juvenile polyps (Yamashita, 2003), where each stage had a total of 125 larvae collected. Samples were stored in RNAlater (Thermo Fisher Scientific, AM7021) in −20°C until total RNA was extracted using the PureLink The seven levels of the light condition factor with their corresponding measurements.

Light conditions in study
Wavelength ( River Protocol (ORP) was used for assembly, which performed read trimming, read normalization, read error correction and assembly using a multi-kmer/multi-assembler approach, merging those assemblies into one final high-quality assembly (MacManes, 2018).
The ORP also produced quality metrics from TransRate and BUSCO.

| Gene expression
We used Salmon (Patro et al., 2017) to quantify transcripts and EdgeR (Chen et al., 2020) to identify differentially expressed transcripts in pairwise comparisons of the six developmental stages.

| RNA fluorescent in situ hybridization
We identified the highest expressed transcripts of selected sensory genes from the three sensory gene sets. We then designed RNA-probes for our target sequences using the Stellaris RNA FISH platform (Biosearch Technologies) with the custom probe design service following their recommendations for probe design.
Samples were stained according to the Stellaris FISH protocol (https://www.biose archt ech.com/suppo rt/resou rces/stell aris-proto cols) with alterations to the protocol. Samples were fixed overnight at 4°C in 4% paraformaldehyde (PF; Sigma, P6148) in PBS. The following day, the samples were washed five times with 5-min incubations in PBST. Samples were then incubated in Prot K (1 μg/μL) for 10 min at room temperature. Immediately following the Prot K incubation, the solution was removed, and the samples were incubated in glycine (4 μg/μL) for 10-min at room temperature. The samples were then washed twice with 5-min incubations in PBST.
The samples were fixed in 4% PF for 30-min at room temperature

| Larval nervous systems reach full development by the actinula stage of development
The actinula stage of E. crocea possesses morphological and cellular features such as the basal protrusion and tentacles replete with cnidocytes, which are associated with larval settlement (Yamashita et al., 2003). However, the structure of the nervous system throughout larval development has not been described.
We examined larvae at four developmental stages, including the metamorphosed juvenile polyp stage, using immunohistochemistry (IHC) and confocal microscopy ( Figure 1). The state of the nervous system at the earliest stage, the star embryo (Figure 1a), appears granular and undifferentiated, which we interpret as a contiguous assemblage of neural progenitor cells (Leclère et al., 2012;Rentzsch et al., 2017). The larval nervous system becomes increasingly differentiated as development proceeds from the preactinula (Figure 1b), where we see the migration of neural progenitors from the endoderm to the ectoderm (Leclère et al., 2012), to the actinula stage (Figure 1c-f), whereupon the tentacles are loaded with neurons and cnidocytes ( Figure 1f).
Additionally, the basal protrusion, the structure that contacts the substrate during settlement, contains a concentrated ring of neural cells (Figure 1d). These data are consistent with the actinula stage being the competent stage for settlement and suggest that actinulae have the capacity to integrate sensory information using the tentacles and the basal protrusion.

| Differences in the sensory environment impact larval settlement
Next, we investigated the sensory information actinula integrate during the settlement decision. We performed a factorial settlement study (Figure 2) assessing the impact of photosensory, chemosensory and mechanosensory cues on larval settlement, which we analysed using a three-way analysis of variance (ANOVA). Our  Figure S2, Table S2).
First, we examined the impacts of the three individual sensory cues on settlement. We found that each sensory cue significantly impacted settlement (Figure 2b-d; Table S1). Specifically, the rate

| Multisensory integration and a sensory cue hierarchy determine settlement in actinulae
Multisensory integration is detected by examining statistical interactions between different sensory conditions (Stein et al., 2009;Stevenson et al., 2014). Therefore, we examined the two-way and three-way interactions between sensory cues as they relate to settlement rate using a three-way ANOVA and contrasts based on our experimental approach (Figure 2a; Table S4). First, we assessed the two-way interactions beginning with the interaction between light cues and chemical cues and found no significant interaction across experiments (p = .33; 70 experiments total; Figure 2f,i). Similarly, no significant interaction (p = .89) was observed between light cues and mechanical cues (Figure 2g,j). Conversely, the combination of chemical and mechanical cues displayed a significant positive interaction (p = <2e-16; Figure 2k; Table S4). Alone, chemosensory cues increased larval settlement. But when combined with a mechanical cue, settlement rates were significantly enhanced revealing an additive effect when both cues are present, reversing the effect of the mechanical cue. Recall that when the chemical cue was absent, the mechanical cue had a significant negative influence on settlement rate (p = <2e-16; Figure 2k).
Next, we examined the three-way interaction between photo-, chemo-, and mechanosensory cues and found a significant

| Sensory gene expression peaks in competent larvae and diminishes during and after metamorphosis
To assess genetic correlates of the observed sensory response, we  Table S5). It is noteworthy that many of the transcripts that show differential expression in development for either the photo-or mechanosensory gene sets have shared functions in both.
Similar analyses of the sensory perception of chemical stimulus gene set indicate a markedly different trend as few developmentally differentially expressed transcripts were recovered for this set ( Figure 4b). However, they do indicate strong stage 4 (competent actinula) differential expression of homologues of both PKD2L1 and PKD1L3, which dimerize and facilitate sour taste perception in mammals (Fain, 2020;Ishimaru et al., 2006).

| RNA fluorescent in situ hybridization reveals evidence for polymodal sensory cells
Our studies of show some co-expression with TRPA, but most TRPA expression is observed in cnidocytes to the exclusion of PKD2L1, which is F I G U R E 3 Sensory cue hierarchy of larval settlement in actinula larvae. The highest rates in settlement occur in the presence of all three cues, where significantly higher settlement rates occurred in the presence of green light, and significantly lower settlement rates occurred in the presence of red light. The second tier consists of chemical cues which significantly increases larval settlement rates, but not to the same magnitude as having all three cues present. Light did not interact with chemical cues, leading to the lack of a light hierarchy in this second tier. The third tier is the light cues only treatment, where the highest rates of settlement occur in green light and the lowest rates occur in red light. Lastly, is the Mechanical cue tier which significantly decreased settlement rates in the absence of other cues. Additionally, the scale of colour indicates the strength of a cue (darker colours indicate stronger influence).
expressed primarily in sensory neurons (Figure 5d-f). Our results were confirmed by quantifying colocalization using Manders' Overlap ( Figure S4) and control experiments were conducted to validate expression ( Figure S5). Additionally, we examined zsections and 3D projections of confocal data to corroborate colocalization ( Figures S7-S9, Video S1).

F I G U R E 4
Gene expression over developmental time in actinula larvae for three sensory gene sets. Heatmaps of significantly differentially expressed genes over development for the following three gene sets: (a) sensory perception of light stimulus; (b) sensory perception of chemical stimulus; and (c) sensory perception of mechanical stimulus. High expression is signified by yellow and low expression is light blue. The grey panels to the right contain abbreviated descriptions of gene functions from The UniProt Consortium (2021). The transcripts highlighted in red are the transcripts used to make RNA Fluorescent in situ hybridization probes. Gene trees of the orthogroups of probes are given in Figure S6. Developmental stages: 1 star embryo, 2 preactinula, 3 pre-competent actinula, 4 actinula in nematocyteprinting stage, 5 settled (attached) actinula and 6 metamorphosed juvenile polyp.  Hobmayer et al., 1990;Plachetzki et al., 2012;Westfall, 2004;Westfall & Kinnamon, 1978). The cnidocytes that facilitate nematocyte printing are located in the tentacles, and a ring of neurons is present at the basal protrusion located on the aboral end of actinulae (Figure 1d). Studies of other cnidarian larvae (planula) have shown that the aboral region is a site for sensory integration (Matsushima et al., 2010;Schwoerer-Böhning et al., 1990;Seipp et al., 2007;Tran & Hadfield, 2013;Vandermeulen, 1974). This provides support to the hypothesis that the basal protrusion of actinulae may also be involved in sensory integration, but further research is needed. Furthermore, we propose that the sensory interplay between cnidocytes and their adjacent sensory neurons facilitate sensory integration during hydrozoan larval settlement.

| A sensory cue hierarchy and MSI during the larval settlement decision
Our larval settlement study identified a sensory cue hierarchy, where the highest rates of settlement occurred in the presence of chemical, mechanical, and light cues, with green light being the most permissive to settlement (Figure 3). This is not surprising as marine invertebrate larvae are known to integrate information from different modalities (Birch et al., 2023;Crisp, 1974;Ettinger-Epstein et al., 2008;Hadfield, 2011;Hadfield & Paul, 2001;Hodin et al., 2018;Holst & Jarms, 2007;Morello & Yund, 2016;Müller & Leitz, 2002;Pawlik, 1992;Say & Degnan, 2020;Whalan et al., 2015;Woodson et al., 2007). This type of sensory integration, and the likelihood that some cues are more important than others, was the basis for the proposal that a hierarchy of sensory cues dictates larval settlement in a species-specific manner (Ettinger-Epstein et al., 2008;Hodin et al., 2018;Kingsford et al., 2002;Woodson et al., 2007).
We also found a However, this differs from current evidence for anthozoan planulae where some species' planulae prefer to settle in red wavelengths of light (Foster & Gilmour, 2016;Mason et al., 2011Mason & Cohen, 2012), while others settle across different wavelengths, suggesting species-specific preferences and a mechanism of niche differentiation (Mundy & Babcock, 1998;Strader et al., 2015). In the case of E. crocea larval settlement, green light, which penetrates only to shallow depths in the water column, may serve as a depth meter ensuring that colonies of E. crocea settle at shallow depths where prey items like plankton are abundant. We note that our finding of the preference of E. crocea larvae to settle in areas illuminated in green light does not in itself indicate the capacity for discrimination between wavelengths of light in E. crocea larvae. It is more likely that actinulae larvae are effectively "colour blind" but show a higher sensitivity to green light due to a limited photoreceptor palate with sensitivity in that range. Indeed, we uncovered only three closely related opsin transcripts from all E. crocea larval stages ( Figure S6).

| The cellular basis for MSI in actinulae
Temporal patterns of gene expression also illuminate the sensory determinants of the settlement decision in E. crocea. While each of the three gene sets used for GSEA interrogation contained >200 genes, only a subset of those genes have orthologs in the E. crocea transcriptome and are differentially expressed between stages ( Figure 4). The mechanosensory gene set includes the largest number of active genes while the chemosensory gene set contains the least number of genes. This disparity is partly due to the relative degree of conservation in gene function between cnidarian genomes and those of model organisms from which annotations are based, and partly due to genes that are differentially expressed in E. crocea larval development. In addition, the mechanosensory and photosensory gene sets share a number of genes in common due to pleiotropy. It is somewhat surprising that so few chemosensory genes were recovered by our screen given that chemosensitivity is the predominant sensory cue in larval settlement. However, we did identify orthologs of both PKD1L3 and PKD2L1, which have been implicated as key components of the sour taste (pH) transduction pathway in mammals (Fain, 2020;Ishimaru et al., 2006) and have previously been implicated in cnidarian chemosensitivity (McLaughlin, 2017). Furthermore, biofilms vary in acidity (Dexter & Chandrasekaran, 2000), which could allow for the assessment of different settlement sites based on their chemosensory properties.
These analyses identify two pulses of sensory gene expression: one occurring early (stage 2, preactinula) and consisting largely of regulatory and structural factors and another occurring later (stage 4, actinula) and consisting of structural and physiological factors.
These data add further support for the actinula (stage 4) as the maximally sensory-equipped larval stage and highlight candidate genes for expression analyses.
We examined the expression of several candidate genes in stage 4 actinula larvae using FISH ( Figure 5). We used the same opsin Together, these data suggest that MSI in E. crocea larval settlement is facilitated by a simple communication circuit between polymodal photo-chemosensory neurons and mechanoreceptive cnidocytes located on the tentacles. We propose that in the absence of chemical and photosensory cues, mechanosensitive cnidocytes are inhibited from discharging and little cnidocyte printing behaviour takes place. However, when light and chemical cues are present, this inhibition is relieved and mechanosensitive cnidocytes are free to fire. The combination of permissive light, chemical and mechanical cues leads to the highest rate of settlement.
MSI is best known in animals with complex brains where specialized brain centers have evolved to facilitate information processing and exchange. Here, we show that MSI is also possible in animals that lack centralized nervous systems and may be facilitated by communication between as few as two cell types: sensory neurons and cnidocytes. Moreover, given the importance of larval settlement dynamics in shaping benthic ecosystems, it is likely that MSI as observed in E. crocea larvae, may be an important determinant of benthic community composition and function.

| CON CLUS ION
Understanding how cnidarians integrate sensory information from the environment is critical to understanding the ecological processes that dictate benthic community composition. At the same time, uncovering the genetic determinants of cnidarian sensory behaviour can illuminate the deep evolutionary histories of the animal senses and provide clues on their early functions. We show that brainless actinula larvae use MSI during the larval settlement decision that incorporates information processing from the light, chemical and mechanical senses. MSI is usually portrayed as a process involving information flow between higher-level brain centres (Currier & Nagel, 2020;Ghosh et al., 2017;Otto et al., 2013;Stein, 1998;Stein et al., 2014); however, our results indicate that MSI may be facilitated by interactions between cells and may have been a prominent feature of the organismal biology of metazoans prior to the evolution of complex brains.

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare no conflict of interest.