Defense through sensory inactivation: sea hare ink reduces sensory and motor responses of spiny lobsters to food odors

Antipredator defenses assume many forms. Most of them operate through one or more senses of the predators. One possible defensive mechanism is sensory inactivation, in which the predators' senses are impaired by the would-be prey. However, experimental support for sensory inactivation as an antipredatory defense is largely lacking. For example, there is only anecdotal evidence for a ‘flash-bulb effect’ where animals deliver a bright light to temporarily ‘blind’ predators (e.g. Morin, 1983; Morin, 1986; Mackie, 1995; Deheyn and Wilson, 2011) or for inactivation of predators' chemical senses by inking cephalopods (e.g. Eibl-Eibesfeldt and Scheer, 1962; MacGinitie and MacGinitie, 1968; Fox, 1974; Kittredge et al., 1974; Prota et al., 1981; Moynihan and Rodaniche, 1982). One candidate for sensory inactivation as an anti-predatory defense is the ink secretion of sea hares. Sea hares (Aplysia spp.) have many defenses, including crypsis and large size (Johnson and Willows, 1999), but chemical defenses are dominant (Derby, 2007; Derby and Aggio, 2011). Passive chemical defenses are plentiful and effective, and include deterrent molecules in their skin and digestive gland (e.g. Carefoot, 1987; Johnson and Willows, 1999; Wägele and Klussmann-Kolb, 2005; Kamiya et al., 2006). Inking is an active, inducible chemical defense. Sea hares contain two active defensive glands, the ink gland and the opaline gland, whose secretions are released into the mantle cavity and then pumped via a siphon towards the site of predatory attack. Ink is purple due to pigments derived from the sea hare's diet of red algae. Opaline is a whitish liquid that polymerizes and becomes highly viscous upon contact with water. Ink and opaline glands are under separate neural control by different ganglia and can release independently or together (Carew and Kandel, 1977; Tritt and Byrne, 1980; Walters and Erickson, 1986; Nolen and Johnson, 2001). Healthy animals do not release either secretion without provocation, and significant mechanical manipulation such as being bitten or pinched is required.

Although no predator is known to make a regular meal of sea hares, perhaps because they are so well defended, a number of generalist predators from several different phyla, including spiny lobsters, will take sea hares on occasion (reviewed by Johnson and Willows, 1999). Field studies in the marine reserves of Catalina Island, California, demonstrate that spiny lobsters Panulirus interruptus prey on sea hares, probably because their high abundances there have led to depletion of their favorite foods, resulting in hunger-induced acceptance of less-preferred, chemically defended prey including sea hares (W. Wright, personal communication). Laboratory studies show that spiny lobsters will attack and in some cases consume sea hares (Kicklighter et al., 2005).

Sea hare ink has been shown in laboratory behavioral experiments to play a significant protective role against predatory sea anemones, fish and crustaceans including spiny lobsters (Nolen et al., 1995; Kicklighter et al., 2005; Kicklighter and Derby, 2006; Nusnbaum and Derby, 2010b). Sea hare ink acts as a chemical defense through multiple molecules and mechanisms. One mechanism is as a phagomimetic decoy, by virtue of its high concentration of amino acids stimulating chemoreceptors that mediate feeding in predatory crustaceans (Kicklighter et al., 2005). A second mechanism is as a repellent, as demonstrated against predatory fish, spiny lobsters, crabs and sea anemones (Nolen et al., 1995; Johnson and Willows, 1999; Kicklighter and Derby, 2006; Nusnbaum and Derby, 2010a). Several repellent molecules have been identified, including aplysioviolin, phycoerythrobilin and compounds in the ‘escapin’ pathway that include hydrogen peroxide, α-keto-ε-aminocaproic acid, Δ¹-piperidine-2-carboxylic acid and others (Aggio and Derby, 2008; Nusnbaum and Derby, 2010a; Nusnbaum and Derby, 2010b; Kamio et al., 2009; Kamio et al., 2010). A candidate mechanism that has been proposed (Kicklighter et al., 2005; Derby, 2007) but not experimentally tested is sensory inactivation, in which the secretion decreases the activity of the predator's sensors such that the predator cannot detect appetitive stimuli. Anecdotal evidence suggesting that ink may disrupt the sensory systems of predators is that attacks of spiny lobsters on sea hares often result with the lobsters' sensory appendages being covered with the ink secretion and with the lobsters spending considerable time grooming their appendages, especially their antennules (Kicklighter et al., 2005).

Sensory inactivation could be effected through either of two actions of ink secretion. First, the viscous and sticky secretion might cover the predators' sensors and physically prevent appetitive chemicals from reaching and stimulating its chemosensory neurons (i.e. masking). Second, the highly concentrated amino acids in the secretion might cause high-amplitude, long-lasting stimulation of chemoreceptor neurons, so that they become unresponsive to the food stimulus either through saturation or adaptation.

We tested the sensory inactivation hypothesis and explored its mechanisms using the antennular chemosensory pathway of spiny lobsters (Fig. 1). Antennular chemoreceptors of crustaceans, including spiny lobsters, play important roles in many chemosensory behaviors. First, they are necessary to mediate many responses to intraspecific cues, including alarm cues in the hemolymph that mediate avoidance of injured animals (Shabani et al., 2008) and social cues in the urine that mediate aggression (Shabani et al., 2009). More relevant to their role in predation, they are necessary to initiate searching and orientation towards the source of a distant chemical stimulus (Horner et al., 2004) and complex behaviors such as learning and discrimination (Steullet et al., 2002). They may also play a role in motivation to feed (J.F.A. and C.D.D., unpublished observations). The antennules of spiny lobsters have 10 types of setae comprising two major types and pathways (Cate and Derby, 2001; reviewed by Caprio and Derby, 2008; Schmidt and Mellon, 2011). The aesthetascs are unimodal chemosensilla located in a dense tuft located on the ventral and distal half of the antennular lateral flagella. Environmental chemicals access the dendrites of the chemoreceptor neurons within the aesthetascs by passing through the porous cuticular walls of the setae (Derby et al., 2007). Their chemoreceptor neurons project to the olfactory lobe in the brain and thus are part of the olfactory pathway. All the other types of chemosensilla – of which there are 10 types in spiny lobsters – are bimodal, innervated by both chemosensory neurons and mechanosensory neurons. Collectively, these are called non-aesthetasc, or distributed, chemoreceptor neurons. Chemicals stimulate their chemoreceptor neurons by entering each seta though a distally located pore, and the axons of these neurons project to the lateral antennular neuropil (Cate and Derby, 2001; Schmidt and Derby, 2005).

We performed two types of electrophysiological experiments to determine whether sea hare ink applied to the antennules causes sensory inactivation. We monitored changes in the sensitivity to food-related chemicals for (1) chemoreceptor neurons in the distributed (non-aesthetasc) antennular system and (2) antennular motor neurons controlling food odor-activated movement of the antennules (Maynard, 1966; Daniel and Derby, 1988; Schmidt and Ache, 1993; Schachtner et al., 2005). In both experiments, we used opaline, several other treatments and controls to determine whether effects were due to either of two effects. One possible effect is the physical properties of ink, where the sticky secretion adheres to the surface of the setae, thus impeding chemicals from moving through the setal pores and preventing them from reaching the setal lumen and therein the dendrites of the chemoreceptor neurons. The other possible effect is that chemical in the ink secretion – the millimolar concentrations of amino acids and other chemostimulants – over-stimulate the chemoreceptor neurons such that they become adapted and are unable to respond to subsequent chemical stimuli.

Caribbean spiny lobsters, Panulirus argus (Latreille 1804), were collected in the Florida Keys and shipped to Georgia State University. They were kept individually in 40 l (50×25×30 cm) aquaria containing artificial seawater (ASW) (Instant Ocean, Aquarium Systems, Mentor, OH, USA) until used in experiments. Sea hares Aplysia californica (Cooper 1863) were collected in California by commercial suppliers and shipped to Georgia State University. Upon arrival, they were cooled and anesthetized with an injection of 60 ml of a 0.37 mol l⁻¹ MgCl₂ solution, after which opaline glands were removed by dissection and stored at −80°C until further processing.

Chemical stimuli were delivered to the antennule before and after treatments to determine the effect of those treatments on chemosensory responses to appetitive chemical stimuli. Three chemical stimuli were tested: two concentrations of ‘shrimp juice’ and a negative control (ASW). Shrimp juice was made by soaking 1 g of shredded shrimp in 1 l of ASW for 1 h, with occasional stirring. The large shrimp pieces were removed and the solution was filtered using a 0.2 μm membrane (Whatman, Kent, UK). Samples were aliquoted and stored at −20°C. Each test day, an aliquot of shrimp juice was thawed and diluted in ASW to 1 and 10% of the initial concentration.

Five treatments were applied to the medial or lateral flagella of the antennules to test their effects on responses to chemical stimuli. Treatments were applied to the sensory end of the antennules using a small paintbrush. Approximately 0.5 ml was used for each application, which is a sufficient volume to cover the entire stimulation area. This application is meant to simulate the coverage of ink occurring during natural behavioral interactions between spiny lobsters and inking sea hares. An example of ink secretion covering the antennules following an inking bout is shown in supplementary material Movie 1.

The five treatments were as follows:

1. Water-soluble fraction of opaline (opaline_WSF). To obtain opaline_WSF, opaline glands were freeze-dried, crushed with a mortar and pestle, and extracted with 100% methanol. The methanol fraction, which contains amino acids and other major food attractant chemicals, was removed, leaving the pellet, which was stored at −80°C until prepared for testing by rehydrating with ASW to 100%concentration, thus forming opaline_WSF. Opaline_WSF is sticky and mimics the physical nature of opaline.

2. Carboxymethylcellulose (CMC). When mixed at 0.3 g l⁻¹ ASW, CMC yielded a sticky substance that mimics the physical nature of opaline. CMC lacks any of the amino acids or chemical attractant components of opaline.

3. A mixture of the five amino acids in highest concentration in opaline (AAs): taurine (226.1 mmol l⁻¹), l-lysine (105.3 mmol l⁻¹), l-histidine (12.21 μmol l⁻¹), l-aspartic acid (6.867 mmol l⁻¹) and l-glutamic acid (2.793 mmol l⁻¹) (based on Derby et al., 2007). This treatment simulated the appetitive chemical stimuli in opaline for spiny lobsters (see Kicklighter et al., 2005).

4. A combination of CMC and AAs (CMC+AAs), each as described above. This treatment created a mixture that simulates the chemical and physical properties of opaline.

5. A negative control treatment of ASW (i.e. Instant Ocean).

When released naturally, opaline (i.e. ‘raw’ opaline) is viscous but with a fluid consistency. See supplementary material Movie 1 for a view of the consistency of opaline and the ink secretion during natural release by a sea hare. For our laboratory work on opaline, we collected it by dissecting out the glands and freezing them at −80°C for storage. For each experiment, the frozen glands were allowed to reach room temperature and were then gently squeezed. The resulting opaline was very clumpy, less consistent and less fluid than raw opaline. This previously frozen opaline could not be evenly applied to the antennule in a way that is similar to raw opaline in natural encounters between sea hares and lobsters, because it tended to lump together and could not be distributed along the antennule. We found that opaline_WSF was physically much more like raw opaline, and thus we used it rather than the frozen opaline in our studies.

In summary, the five treatments differ in two respects (Table 1): some are sticky and therefore may physically prevent chemical stimuli from reaching the receptors on the antennule (opaline_WSF, CMC); some contain amino acids and by long-term application can lead to sensory adaptation and thus inactivation of chemoreceptor neuronal responses (AAs); some are both (opaline_WSF, CMC+AAs); and some are neither (ASW).

The aim of this experiment was to determine whether responses of chemoreceptor neurons of spiny lobsters to chemical stimulation changed following application of opaline and other treatments to the chemosensory organ, the antennule. We recorded spiking activity from chemosensory neurons in response to a chemical food stimulus, and monitored how responses changed in the various treatments.

Electrophysiological recordings were made from chemoreceptor neurons in the medial or lateral flagellum of the antennule using standard techniques (Derby, 1995). The flagellum was dissected, placed in an olfactometer and supplied with oxygenated saline at ~2 ml min⁻¹ by cannulating its artery. The olfactometer had a U-shaped depression in which the antennular flagellum was situated such that the distal portion of the flagellum (i.e. aesthetasc tuft region) remained submerged at all times with seawater or treatment (see below). It was this distal portion that was chemically stimulated. Extracellular differential recordings were made using a glass suction electrode with silver chloride wire on nerves exposed at the proximal end of the antennule. Electrical activity was amplified and digitized using AxoScope (Molecular Devices, Sunnyvale, CA, USA). Recorded neural activity was in the form of action potentials and, following spike sorting (as described below), was resolvable as responses of single receptor cells. This method is thought to record selectively from non-aesthetasc (distributed) chemoreceptor neurons (e.g. Caprio and Derby, 2008; Schmidt and Mellon, 2011).

The antennule was stimulated with a 2 s pulse of 1% shrimp juice via an automated solution changer (Bio-Logic, Claix, France). The stimulus or seawater flowed from the proximal to the distal end of the antennule at 2 ml min⁻¹ and was removed via suction before reaching the exposed nerves. Responses to the stimulation described above were recorded for the following conditions: before treatment, during treatment and after treatment. All three conditions were repeated for as many stimuli as possible for each antennule. Because a continuous flow of ASW would wash off the treatment, we avoided this using a protocol in which background flow was often interrupted. In this protocol, one stimulation cycle consisted of the following: (1) turn off background flow; (2) apply treatment and wait 1 min; (3) stimulate with either ASW or shrimp juice and record for 20 s; (4) turn on background flow to rinse off the chemical stimulus (40 s); (5–8) repeat steps 1 to 4 using the same treatment in step 2 but the other stimulus in step 3. Steps 1–8 constitute one stimulus cycle, and each chemoreceptor cell was subjected to three cycles. In the first and last cycle, the treatment was ASW (before and after treatment conditions), and in the middle cycle it was the treatment being tested (treatment condition). Steps 1–4 took ~2 min. The persistence of each treatment was guaranteed by interrupting both the background flow and suction from the moment of its application until stimulus onset. In addition, in the case of the sticky stimuli, it could be evaluated visually: they remain on the antennule after application and before chemical stimulation, and even during the 2 s chemical stimulation. As described above, after each chemical stimulation and subsequent rinse, we reapplied the treatment before stimulating with a new chemical.

Data analysis was performed using Spike2 (Cambridge Electronic Devices, Cambridge, England), to sort spikes based on waveform, thus allowing us to quantify responses of single chemoreceptor neurons. For each stimulation, the number of spikes in the 1 s period immediately preceding stimulation was subtracted from the number of spikes in the first 1 s of the response, and the response to ASW was subtracted. Thus, for each treatment, we obtained three such net responses: before treatment, during treatment and after treatment. The overall response was obtained by dividing the during treatment response by the average of the before treatment and after treatment responses. This yielded ΔR, in which ΔR<1 indicates that the treatment reduces the response, ΔR=1 indicates that the treatment has no effect and ΔR>1 indicates that the treatment enhances the response. ΔR values were compared statistically using Kruskal–Wallis tests and post hoc Dunn's tests.

The aim of this experiment was to determine whether a behavioral response to chemical stimulation of a chemosensory organ of spiny lobsters changed following application of opaline or related treatments to that organ. As a measure, we used motor neuron activity associated with movement of the antennules in response to its chemical stimulation with a food chemical stimulus, because antennular movement is a reliable component of chemical-stimulated food-seeking behavior in spiny lobsters (Zimmer-Faust et al., 1984; Zimmer-Faust, 1987; Daniel and Derby, 1988; Derby et al., 2001).

Each spiny lobster was placed in a restraining device dorsal side up, blindfolded by placing aluminum foil caps over its eyes, and then placed into a bath of aerated ASW, ensuring the gills were covered. The lateral flagellum of one of the antennules was secured in an olfactometer, separated from the water bath, allowing for stimulation and application of the treatment as described above. Extracellular differential recordings were made from a pair of silver wire electrodes, insulated with Teflon except for the tip. One electrode of the pair was placed in the antennule joint where the lateral and medial flagellum bifurcate, near the motor neurons controlling movement of the antennular lateral flagellum; the other electrode of the pair was placed in the ASW bath. That we recorded the activity of motor neurons and not mechanoreceptor neurons is supported by several observations. First, the amplitudes of the action potentials were generally very large, as expected for the large-diameter axons of motor neurons. Second, the responses lasted much longer than the duration of the stimulus (see Fig. 3). Third, there were instances of spontaneous bursts of action potentials in the absence of mechanical stimulation (i.e. when the water flow was turned off). Each lobster was given a 30 min rest period after implanting electrodes before presentation of odor stimulus. Motor neuronal activity was recorded using the same amplifiers and software as above.

Stimulation of the antennule was performed using the same equipment exactly as described above, except that two shrimp concentrations (1 and 10%) were tested for each animal. This was done to determine whether the treatment had a different effect on different concentrations of shrimp. A negative control of ASW was still used. Each stimulus was presented twice in random order for each animal and each day. Activity was recorded for both before and during treatments but not for after treatment due to the tolerance of spiny lobsters under testing conditions. Each spiny lobster received only one treatment per day. The protocol used was the same as for the chemoreceptor cells (see above) with the exception that no after treatment condition was required because the preparations were stable.

Data were amplified, digitized and imported into Spike2 as above. The analysis was also similar, but because we were often unable to identify single units from the recordings, we used the following analysis: traces were rectified and smoothed (time constant=0.25 s) and instead of counting the number of spikes, we calculated the area under the resulting curve. For each response, we identified the peak of the rectified and smoothed trace and calculated a 5 s area under it (0.5 s before the peak + 4.5 s after the peak), and divided it by the area under the equivalent length of pre-stimulus curve ending immediately before stimulus onset. Because stimuli were applied twice, we then calculated an average for each one. In this analysis, ΔR for each stimulus was calculated as the response during treatment divided by the response before treatment, but the expected values are the same as above. Data were statistically compared with a repeated-measures ANOVA. The different treatments were compared using planned comparisons and relevant stimulus–treatment combinations with Bonferroni correction for multiple comparisons.

Electrophysiological recordings from non-aesthetasc (distributed) chemoreceptor neurons in the antennules of spiny lobsters were used to evaluate the effect of ink secretions on sensory reception of a food-related chemical stimulus, shrimp juice. Ink secretion of sea hares is composed of two glandular secretions that are co-released: ink, a purple product of the ink gland; and opaline, a white product of the opaline gland (see supplementary material Fig. S1, Movie 1). We focused on opaline because it is the more viscous of the two secretions. Five treatments related to opaline were applied to the antennules to evaluate whether sensory inactivation occurs and to determine its mechanism (Table 1).

Fig. 2A,B shows examples of the effects of two treatments on responses of distributed chemoreceptor neurons to 1% shrimp juice. Fig. 2A shows that treatment with ASW, the negative control, did not affect the response of one chemoreceptor neuron to stimulation with the food odor. Fig. 2B shows that treatment with opaline_WSF reversibly reduces the intensity of the odor-evoked response in another chemoreceptor neuron. Fig. 2C shows composite data for all five treatments. Applying either opaline_WSF, CMC or CMC+AAs to the antennule reversibly decreased the neural response to shrimp juice when compared with ASW. Treatment with AAs caused a decrease that was statistically non-significant. Based on effects of fractions and mimics of opaline, we conclude that the sensory inactivation is principally due to the secretion physically covering the antennular chemosetae and thus blocking chemicals from accessing chemoreceptor neurons.

Electrophysiological recordings from antennular motor neurons were used to evaluate the effect of ink secretions on motor responses to chemical stimulation of a chemosensory organ of spiny lobsters. The activity of antennular motor neurons was used as an assay because antennular movement is a reliable feature of chemically stimulated food-seeking behavior in spiny lobsters (Zimmer-Faust et al., 1984; Zimmer-Faust, 1987; Daniel and Derby, 1988; Derby et al., 2001) and is a behavior that was reliably quantified in our experiments. We used the same five treatments as in the chemoreceptor neuron assay. Fig. 3A,B shows two examples of the effect of treatments on motor neuronal responses to stimulation with 10% shrimp juice. Fig. 3A shows that treatment with ASW did not affect the motor neuronal responses of one animal to chemostimulation, but treatment with opaline_WSF profoundly reduced the intensity of the odor-evoked motor neuronal response in a different animal (Fig. 3B). Composite data for all five treatments are shown in Fig. 3C. Treatment with opaline_WSF, CMC or CMC+AAs significantly decreased the motor neuronal response to shrimp juice compared with treatment with ASW, but effect of treatment with AAs was not different than treatment with ASW. Thus, the effect of the secretion on the chemically evoked motor response is due to it physically covering the antennular chemosetae and blocking access of the chemicals to the chemoreceptor neurons.

Our study examined whether ink secretion protects sea hares from predators by inactivating the predators' sensory systems. We used electrophysiological recordings from chemosensory neurons and motor neurons in a major chemosensory organ, the antennules, of predatory spiny lobsters to evaluate the effect of ink secretions on responses to food odors. We showed that covering the antennules with opaline, which is one of the two glandular secretions comprising the ink secretion of sea hares, significantly reduced the responses of both types of neurons (Figs 2, 3). This result supports the conclusion that opaline acts as a chemical defense because it causes sensory inactivation. Our experimental treatment of the antennules with opaline, by applying 0.5 ml over the cuticle surface using a brush, was meant to simulate the covering, matching what we observed in behavioral interactions between spiny lobsters and inking sea hares. Supplementary material Movie 1 presents one example of the coating of the antennules with ink secretion that occurs during natural encounters between a spiny lobster and a sea hare. The coating of antennules in natural encounters may be more patchy across the antennular surface, but the coverage, where applied on the antennule in our experiments, is, to our best approximation, similar to that occurring in natural encounters. Thus, even if natural encounters lead to more patchy covering of the antennular chemosensors, our techniques should reveal at least qualitatively the inactivating effect of opaline on responsiveness of the chemosensors.

Two properties of opaline might contribute to sensory activation by opaline. First is its physical property: its stickiness. By covering the setae containing the antennular chemoreceptor neurons, opaline might prevent food odors and other soluble appetitive chemicals from moving into the lumen of the setae, binding to the receptor molecules on the neuronal dendrites, and activating these cells. A second property is the chemicals constituting opaline. Opaline and ink contain very high concentrations of small nitrogenous compounds, such as amino acids and ammonium, which are known to be highly stimulatory to chemoreceptor neurons of spiny lobsters and other crustaceans (reviewed by Caprio and Derby, 2008; Schmidt and Mellon, 2011). These stimulatory chemicals, especially embedded into the sticky matrix of opaline, might provide unnaturally long durations of stimulation of these receptor neurons, which would be followed by a sustained response adaptation (e.g. Gomez and Atema, 1996), both of which cause the neurons to be relatively unresponsive to food odors during their exposure to opaline. It is also possible that opaline or ink contains chemicals that suppress the responses of the receptor neurons, as these cells are known to have such inhibitory transduction cascades (Ache and Young, 2005). Our data provide strong support for the first hypothesis – physical blocking. This is especially apparent from the observation that carboxymethylcellulose prepared to a sticky consistency similar to that of opaline had an inactivating effect similar to that of opaline (Figs 2, 3). Because carboxymethylcellulose does not appear to be a chemical stimulant by itself (J.F.A. and C.D.D., personal observations) and it does not contain any other chemicals, its effect appears to be due to its stickiness, probably by physically blocking movement of chemicals through pores in the antennular sensilla (Fig. 1) and into the sensillar lumen where the transduction apparatus of the chemoreceptor neurons is located.

Carboxymethylcellulose spiked with a mixture of the major amino acids in opaline at their natural concentrations was equally effective as carboxymethylcellulose alone in reducing responses to food odors (Figs 2, 3), suggesting either that the inactivating effect of opaline is due either solely to its physical properties, or that the effect of its physical properties is so dominant that any additional effect of chemicals is inconsequential. Treatment with the amino acid mixture by itself was not at all effective in reducing responses to food odors. This provides additional support for a physical-only effect. However, the absence of an effect of treatment with the amino acid mixture might be partially due to technical issues. During treatment, we exposed the antennule to treatments for 1 min without allowing any seawater flow or rinsing over the antennule. In the ensuing test of the effect of treatment, we restarted the seawater flow and presented the food odor. Two factors in this protocol might have made it more difficult to observe an effect of the amino acid mixture. First, because the amino acid mixture lacks the sticky consistency of opaline or carboxymethylcellulose, it does not adhere to the antennular cuticle. Thus, although the apparatus kept the antennule bathed in the amino acid mixture, the amino acids were quickly washed off when the flow was restarted. Taken together, our experiments provide strong support that the sensory inactivation is principally due to the secretion physically covering the antennule and thus blocking chemicals from accessing chemosensory neurons. Our experiments do not provide evidence for the chemical properties of opaline, either excitatory or suppressive, contributing to the inactivating effect, but experimental design issues allow that there might be some chemical effect that we could not resolve.

Our demonstration of sensory inactivation focused on antennular chemoreception. The antennular chemoreceptors play a specific role in feeding behavior and its chemical activation. The fact that the antennules of spiny lobsters are many centimeters long and can be actively moved allows the animal to use them to sample the chemical space over a large three-dimensional area in front of the animal. This helps the antennular chemoreceptors in their major function, which is to initiate searching upon identifying appetitive chemicals and to orient within an odor plume during tracking towards its source (Horner et al., 2004; Weissburg, 2011). Our demonstration that sensory inactivation of the antennules impairs the ability of motor neurons to respond to food odors is consistent with an effect of sensory inactivation on initiating searching and tracking during searches.

Typically, a sea hare is in the grasp of a spiny lobster before the sea hare inks. Our observations are that the ink sticks to all of the sensory appendages in the anterior end, including the antennules, mouthparts and anterior legs. We would expect an effect on these other chemoreceptors similar to that we have demonstrated for antennular chemoreceptors. While all of the chemoreceptors in the anterior end of the animal will be affected, one might expect the mouthpart and oesophageal chemoreceptors to be major functional targets of the defensive ink. Chemosensilla in the legs, mouthparts and esophagus control other aspects of feeding behavior. Leg chemoreceptors control local searching and grasping responses, and delivery to the mouth. Once the material is in the lobster's mouth, chemoreceptors on the mouthparts (mandibles, maxillae and maxillipeds) mediate manipulation of the food and biting (Derby et al., 2001; Garm et al., 2003; Garm et al., 2005). The decision to ingest food is controlled by appetitive and deterrent receptors in the mouthparts and esophagus (Garm et al., 2003; Garm et al., 2005; Aggio et al., 2012). Thus, if our demonstration of ink's sensory inactivation of antennules generalizes to chemosensors on other sensory appendages, which we expect that it does, then sea hare ink will have effects on other aspects of feeding behavior.

This is the first experimental demonstration of sensory inactivation as a chemical defense. While it has been proposed previously as a mechanism of defense by cephalopods, in which ink over-stimulates chemoreceptors of predators so that those predators can no longer sense chemicals released by prey (Eibl-Eibesfeldt and Scheer, 1962; MacGinitie and MacGinitie, 1968; Fox, 1974; Kittredge et al., 1974; Prota et al., 1981; Moynihan and Rodaniche, 1982), it has not been experimentally demonstrated. Sensory inactivation has been suggested as functioning in the visual modality, but the evidence there too is anecdotal. For example, a ‘flash-bulb effect’ has been hypothesized, whereby animals deliver a bright light to temporarily ‘blind’ predators (Morin, 1983; Morin, 1986; Mackie, 1995; Deheyn and Wilson, 2011). In the acoustic channel, some moth species produce ultrasound, which causes jamming of echolocating bats (Tougaard et al., 1998; Corcoran et al., 2009). The mechanism underlying this jamming is not inactivation of the bats' periphery auditory system, as bats can detect their ultrasonic echoes bouncing off moths. Rather, the moths' ultrasound interferes with the ability of the bats' central nervous system to calculate correctly the distance of the moth (Corcoran et al., 2011). Other moth species use their ultrasound as an aposematic defense, or as Batesian or Müllerian mimics, rather than for jamming (Hristov and Conner, 2005a; Hristov and Conner, 2005b; Conner and Corcoran, 2012). Our demonstration of the effect of sea hare ink on spiny lobsters is the first experimental demonstration of sensory inactivation as a chemical defense.

Ink of sea hares acts through multiple mechanisms of chemical defense against predators. These mechanisms include phagomimicry, in which the chemical acts as a decoy (Kicklighter et al., 2005); chemical deterrency, in which the chemical is a repellent (Kicklighter et al., 2005; Aggio and Derby, 2008; Kamio et al., 2009; Kamio et al., 2010; Nusnbaum and Derby, 2010a; Nusnbaum and Derby, 2010b; Nusnbaum et al., 2012); and, as shown in this paper, sensory inactivation. These various forms of chemical defense in sea hares prove to be effective anti-predatory chemical defenses against diverse species under various conditions, such as hunger state of the predator and environmental availability of certain species of algae and thus diet-derived acquisition of deterrent compounds by the herbivorous sea hares (Derby, 2007; Derby and Aggio, 2011).

Taxa throughout the animal kingdom have been shown to use similar antipredator defensive strategies. Decoys, in which the defense is produced to distract the predator from the would-be prey, can also be seen in other sensory channels. For example, limb autonomy, in which an appendage is detached and distracts the predator, is found in octopuses and lizards (Arnold, 1994; Bateman and Fleming, 2009). Decoys can also be visual, such as the bioluminescent ink clouds that are released in many marine species, including squid and ostracods, and act as misdirectional cues (Morin, 1983; Morin, 1986; Grober, 1990; Herring, 1990; Bush and Robison, 2007; Zoerner and Fischer, 2007; Haddock et al., 2010). Repellent or deterrent chemicals are commonly used to create startle or escape responses in predators; for example, the spray of skunks (Wood et al., 2002) and bombardier beetles (Eisner et al., 2006; Eisner et al., 2007). Startle responses can also be produced through other sensory channels to repel or startle predators. Visual cues mediating startle include the sudden flashing of eyespots on the wings of butterflies (Caro, 2005; Stevens, 2005; Langridge, 2009; Janzen et al., 2010) and bioluminescent flashes by planktonic animals (Morin, 1983; Grober, 1990; Mackie, 1995). Auditory cues used in defense include stridulation noises by wasps and beetles against wolf spiders (Masters, 1979) and by spiny lobsters against octopuses or other predators (Bouwma and Herrnkind, 2009), which can be used as a startle defense, an aposematic signal or other defensive functions (Hoy, 1989; Sargent, 1990; Ruxton et al., 2004; Staaterman et al., 2010). In summary, the ubiquity of decoy and repellent defenses indicates that sensory inactivation may be prevalent in antipredator defenses in diverse taxa, but at present experimental tests of this hypothesis are lacking. Further study of sensory inactivation could provide crucial information on predator–prey interactions in diverse groups of animals.

We thank the staff of the Keys Marine Laboratory for providing spiny lobsters. We also thank the entire Derby laboratory for their help and support.