Expression of squid iridescence depends on environmental luminance and peripheral ganglion control

Eyes evolved as early as 515 million years ago, and thereafter the evolutionary pressure to deceive the vision of predatory animals resulted in the refinement of reflective body elements (Parker, 2005). A myriad of biological reflective elements exists today that produce color through the use of repeated nanostructures, and not pigments. Iridescence is one such type of structural coloration, characterized by a high spectral purity and intensity that is angle dependent (Vukusic et al., 2002). Because iridescence can be produced at any visible wavelength, it confers some advantages over pigment-based coloration for purposes such as signaling or camouflage (Meadows et al., 2009). For example, butterflies and damselflies use blue iridescence for mate recognition and as a badge of fitness (Fitzstephens and Getty, 2000; Kemp, 2007). Given that iridescence is angle-dependent and that most animals cannot alter the nanostructures that produce the color, their placement on the body [e.g. butterflies (Rutowski et al., 2007; Vukusic et al., 2002)] and the animal movements during courtship [e.g. peacock spiders (Girard et al., 2011)] become crucial when producing signals of ecological relevance. Only a few species have evolved ways to quickly alter their iridescent nanostructures. Of those, only a few teleost fish species (Iga et al., 1987; Kasukawa et al., 1986; Muske and Fernald, 1987) and the Atlantic longfin squid, Doryteuthis pealeii (Lesueur 1821) (Wardill et al., 2012), have been proved experimentally to control their iridescence neurally. These two iridescence systems evolved independently, and have different activation mechanisms. In squid, the iridescent cells, called iridocytes, contain platelets made of reflectin (Crookes et al., 2004), a protein that is reversibly condensed upon acetylcholine (ACh) application (Cooper and Hanlon, 1986; DeMartini et al., 2013; Izumi et al., 2010). Reflectin platelets form stacks called iridosomes (Arnold, 1967). As the reflectin refractive index is higher than that of the inter-platelet space, which is made up of cytoplasm and extracellular fluids (Kramer et al., 2007), iridosomes act as biological Bragg stacks (Holt et al., 2011; Wu et al., 2007), interfering with different light wavelengths and producing iridescence. Higher ACh concentrations induce thinner and denser platelets (Cooper and Hanlon, 1986; Cooper et al., 1990; Mäthger et al., 2004), resulting in a greater color shift towards the blue wavelengths (Mäthger et al., 2009; Sutherland et al., 2008; Tao et al., 2010) and an increased refractive index, respectively. In addition, the increase in intracellular calcium elicited by exposure to ACh also causes water to be expelled from the cell, reducing the inter-platelet space further and contributing to the color shift (DeMartini et al., 2013). In contrast, the platelets present in fish iridocytes are made of guanine crystals (Clothier and Lythgoe, 1987). These platelets disperse (Iga et al., 1987) upon norepinephrine exposure (Muske and Fernald, 1987; Oshima and Fujii, 1987), which increases the inter-platelet distance and shifts the reflected coloration from blue towards red wavelengths (Clothier and Lythgoe, 1987; Nagaishi and Oshima, 1989).

The advantages of neurally controlled tunable iridescence are self-evident and our understanding of these dynamic nanostructures has expanded dramatically in recent years (DeMartini et al., 2013; Tao et al., 2010). For instance, the endogenous source of ACh remained controversial until Wardill et al. (Wardill et al., 2012) showed that the site of neurotransmitter released takes place at the iridophore layer. These macroscopic iridescent splotches are created by aggregations of 10s to 100s of iridocytes. However, much remains to be clarified about this system. For example, boutons indicating synaptic contacts between iridocytes and axons are yet to be reported. Thus, the current hypothesis is that within the iridophores, the ACh release may be en passant in nature (Wardill et al., 2012). In addition to this lack of information about the neural–iridocyte interface, every other detail of this neural circuit has remained obscure. This is significant because evolution of iridescence may be linked to phylogenetic relationships in species that are visually guided (Parker, 2005). Moreover, although previous studies have proposed that dynamic iridescence may act as a private intraspecific communication channel (Chiou et al., 2007; Hanlon et al., 1990; Mäthger and Hanlon, 2006) or as an aid in camouflage (Hanlon et al., 1999; Mäthger et al., 2009), direct evidence on the function of cephalopod dynamic iridescence is still lacking. In this study, we used electrophysiological, morphological and behavioral approaches to investigate the neural circuit that controls iridescence in the squid Doryteuthis pealeii. We provide unexpected findings on the neural wiring of the circuit that controls tunable iridescence and the light conditions that elicit changes in reflectivity.

Young (Young, 1936b) used the terms pallial nerve and mantle connective interchangeably to describe all the axons that leave the brain and travel towards the mantle, some of which enter the stellate ganglion. In this study more precise naming was necessary for clarity. Hence, we have reserved the name ‘pallial nerve’ for the bundle that contains all the axons that descend from the brain. We refer to the branch of the pallial nerve that travels into the stellate ganglion as the ‘stellate connective’ and to the branch that proceeds towards the fin as the ‘fin nerve’ (Fig. 1A,B).

To begin, we re-confirmed that cutting the pallial nerve, just prior to the stellate connective branching point, resulted in immediate skin blanching [due to cessation of chromatophore neural control (Sanders and Young, 1974)], lack of respiratory mantle contractions and a deficiency of undulatory fin movements in the denervated side of the animal (Fig. 1C, pallial nerve). Accordingly, cutting the stellate connective resulted in absence of respiratory movements and a lack of mantle chromatophore activity (Fig. 1C, stellate connective) and cutting the fin nerve caused cessation of undulatory fin movements and chromatophore activity from the base of the fins to the posterior tip of the animal (Fig. 1C, fin nerve). We did not observe an immediate decrease in iridescence in any of the three treatments. This was not surprising, as in our recent experience (Wardill et al., 2012), iridescence decline does not take place immediately after the last neural stimulation. Hence, we returned the animals to their holding tanks. After 24 h, all the animals with a stellate connective cut (N=10) exhibited a complete lack of skin iridescence on the side ipsilateral to the cut (Fig. 2A). However, these animals retained complete control of their fin chromatophores, including expansion of a single color type of chromatophores (yellow; Fig. 2B), full red flash warning display (Fig. 2C; supplementary material Movie 1) and mottled camouflage patterns (Fig. 2D). Moreover, fin undulatory movements remained intact (supplementary material Movie 2). Iridescence was also abolished ipsilaterally in the animals with a pallial nerve cut, but remained bright in the animals with a fin nerve cut. Thus, activation of fin and mantle iridescence requires descending inputs from the brain that enter the stellate ganglion.

The results above were contradictory to the prevailing dogma, which states that all the motor neurons innervating the fin descend directly from the brain through the fin nerve (Hofmann, 1907; Young, 1932). To solve this conundrum, we removed the sheath surrounding the fin nerve and revealed that some of its fibers form a fascicle that originates in the first stellar nerve (Fig. 3A). By electrically stimulating this fascicle, we recovered fin iridescence in the animals whose stellate connective had been severed (Fig. 3B; supplementary material Movie 3). Importantly, this fascicle activated not just the iridophores, but also the scattered iridocytes (supplementary material Fig. S1). Because the fin chromatophores remained relaxed during electrical stimulation of this fascicle, we have named it the ‘fin iridescence nerve’ or ‘FI nerve’. These findings confirm that the first stellar nerve carries the iridescence motor neurons that innervate the fin. We also found that electrical stimulation of the FI nerve resulted in fin contractions that reached tetanus at frequencies >20 Hz (start of stimulation, supplementary material Movie 3).

To elucidate whether the motor fibers forming the FI nerve originate in the stellate ganglion, or simply travel through it, we backfilled the FI nerve with Lucifer Yellow. The labeling revealed a large group of cell bodies in the stellate ganglion and a few fibers (six at most) that traveled farther and became part of the stellate connective (Fig. 3C). The labeled cell bodies were large (up to 100 μm), located in the most medial part of the ventral cell wall (adjacent to the giant lobe; Fig. 3C), with a single neurite sporting thin branches that resembled dendritic spines (Fig. 3C). These neurites form part of the ventral root that exits the ganglion through the first stellar nerve. Because of the complexity of the neuropile, it was not possible to resolve whether the fibers that continued to the stellate connective also had cell bodies in the ganglion, or whether they extended dendritic trees or terminals within it. To clarify this point, we carried out double fills of the FI nerve and the stellate connective, with Lucifer Yellow and MicroRuby (Fig. 3D). No fibers were doubly labeled, most likely because of the nerve fills being incomplete (occasionally, axons remained unfilled) and the low number of fibers that travel directly from the FI nerve to the stellate connective. However, the labeling of the stellate connective did fill some blood vessels that surrounded the FI nerve cell bodies (Fig. 3D). In addition, the stellate connective fill also labeled a cell body located between the cell bodies labeled from the FI nerve (Fig. 3D). Thus, in addition to the possibility that these fibers are continuous neurites that descend directly from the brain, or sensory afferents ascending towards the brain, we must add the following two possibilities: they could be (1) monopolar cells extending neurites towards the brain and the periphery, but with cell bodies in the stellate ganglion, or (2) blood vessels supplying the stellate ganglion.

To elucidate whether the iridescence motor neurons belonged to the group of fibers that originated in the stellate ganglion, we used a paradigm based on Young's (Young, 1972) denervation experiments with cuttlefish (Sepia officinalis). Namely, degeneration of the distal stumps of the chromatophore axons is observed 4–7 days after the S. officinalis pallial nerve is severed (animals are kept at 17–23°C) (Young, 1972). This is because the axons of the chromatophore motor neurons descend directly from the brain (Dubas et al., 1986a; Hofmann, 1907; Young, 1932). Messenger et al. (Messenger et al., 1997) further confirmed the validity of this paradigm in squid. Accordingly, we cut the squid stellate connective and waited 7 and 15 days before testing whether the iridescent motor neurons innervating the fin would respond to electrical stimulation [for full protocol, see Wardill et al. (Wardill et al., 2012)]. At maximum stimulation, the color of the iridophores was not significantly different between the animals that had been denervated 7 (N=9 iridophores from 4 animals) and 15 days (N=8 iridophores from 3 animals) previous to the experiment (evidenced by the standard deviation overlap throughout the response; supplementary material Fig. S2). Hence, we pooled these results from here onwards and refer to it as the denervated group. Likewise, the colors obtained in intact (Wardill et al., 2012) and denervated (this study) groups were not significantly different (evidenced by the standard deviation overlap throughout the rising phase of the response; Fig. 3E). The one striking difference between these two groups was the reflectance change (%), which was significantly higher in the denervated group (means ± s.d., denervated, 313.94±164.43%, N=15; intact, 245.17±119.56%, N=13; P>1.428×10⁻³⁰). Moreover, the maximum iridophore reflectance change also belonged to a denervated animal (denervated max, 624.46%, intact max, 440.43%; see example of a 491% change from a denervated animal in supplementary material Fig. S3). This is because the baseline iridescence in denervated animals was extremely low. Taken together, these findings indicate that a stellate connective cut extinguishes iridescence because the descending inputs to the motor neurons are severed, and not the motor neurons themselves.

Elucidating the rise/decay rates for dynamic iridescence may help us understand its behavioral role. In this experiment, we recorded the iridescence decay following the last neural impulse, i.e. the time elapsed until iridophores become transparent following a cut of the stellate connective. Of the 15 animals tested, eight (>50%) exhibited fin iridophores that were transparent within 30 min and 12 (80%) displayed a substantial decline in iridescence within 60 min (Fig. 4A; supplementary material Fig. S4). Three animals (20%) had little or no change in iridescence at 60 min post-denervation. Thus, the passive decline of iridescence is slow, taking >30 min to complete. This experiment does not address the possibility that an active process may exist to reduce iridescence in a shorter time.

Iridophores are efficient light reflectors, hence their reflectivity may increase the risk of predation, both day and night. To test whether squid would reduce their iridescence in response to darkness, we dark-adapted the animals for 1 or 2 h. The side of the animal with a stellate connective cut served as an internal control for no iridescence display. After 1 or 2 h in the dark (0.4 nW), 12 of 18 animals (66%) and 5 of 6 (83%), respectively, showed a substantial decline in iridescence (Fig. 4B).

Given the decline in iridescence after dark adaptation, we tested whether squid would increase their iridescence if exposed to light. We used high-speed videography in case this response was rapid, as it is in chromatophores. Moreover, high speed allows short exposure times, which is crucial in the case that the animal jetted in response to illumination because it was startled. However, our high-speed video showed that although squid did increase their iridescence expression, the change was slow; after 90 s of light exposure, iridescence was detectable and continued to increase during the 4 min that followed (N=4; Fig. 4C).

Dynamic coloration and patterning on the mantle of cephalopods have long been known to be influenced by motor neurons emanating from the posterior chromatophore lobes in the central nervous system (Sereni and Young, 1932). In the loliginid squid Lolliguncula brevis, those neurons travel without synapse from the posterior chromatophore lobes to the chromatophore organs in the skin (Dubas et al., 1986a; Dubas et al., 1986b). Similar investigations focusing on iridescence were lacking, and here we discover in the loliginid D. pealeii that iridescent structural coloration control is routed through the peripheral stellate ganglion. The stellate ganglion contains one of the best-studied synapses known to neuroscience – the giant synapse that mediates jet escape via the third-order giant axons (Otis and Gilly, 1990; Young, 1936a; Young, 1936b; Young, 1938; Young, 1939). It is surprising that the anatomical and neurophysiological features of the iridescence circuit of this large ganglion have not been noted or studied to date.

We have demonstrated, through targeted denervations, that the neural circuits that control iridescence and pigmentary coloration are wired differently. Both circuits originate in the brain, but the iridescence signals are routed through the stellate ganglion while pigmentary signals descend through the fin nerve. Thus, pigmentary and structural skin elements are controlled by different motor neurons. Importantly, our denervation experiments have revealed that iridophores remain visually transparent when neural signals are absent. Furthermore, previous studies showed that not all squid iridescent cells are active; for example, those on the lateral and ventral surfaces of squids such as L. brevis are considered passive as they do not alter their reflectance to bath application of ACh (Cooper et al., 1990; Hanlon et al., 1990). In this study we have shown that the scattered iridocytes of the dorsal fin that do not form aggregations into an iridophore also respond to neural stimulation. Color fluctuations from single iridocytes are too small to be detected by the naked eye. Thus, it is possible that they too are involved in producing background coloration that is tunable. It remains to be clarified whether the iridescent cells of the ventral skin respond to neural stimulation.

Our fluorescent dye fills showed that the cell bodies of the fin iridescence nerve motor neurons arise from the ventral root of the last stellar nerve. This is important because Young (Young, 1972) showed that all the fibers that pass through the stellate ganglion on their way to/from the brain, such as the chromatophore motor neurons, do so as part of the dorsal roots. Taken together, these observations support the premise that FI nerve fibers are unlikely to descend directly from the brain. Moreover, we have shown that the fibers responsible for iridescence activation, which travel via the fin iridescence nerve, are silenced but not ablated when the stellate connective is severed. All such results are supported by those of Young (Young, 1972), who confirmed that cutting the pallial nerve resulted in degeneration of the inputs to the ventral neuropile of the ganglion, but that the postsynaptic cells, whose cell bodies are located in the ventral wall, remained intact. Therefore, although intracellular recording/stimulation is needed for direct proof, thus far all results indicate that the cell bodies of the iridescence motor neurons are located in the ventral wall of the stellate ganglion.

The stellate ganglion has long been recognized as a center for signal integration (Wilson, 1960) and reflex coordination (Gray, 1960), but only for the purposes of locomotion and respiration (Young, 1972). We have shown that the stellate ganglion does not act as a simple peripheral reflex in the control of iridescence, because descending brain inputs are needed for iridescence expression. However, it is plausible that within the stellate ganglion, the integration of peripheral information modulates the level of iridescence expression. Whether afferent signals from peripheral senses, such as the mechanosensory ‘lateral line’ (Mackie, 2008; Preuss and Budelmann, 1995) and nociception (Crook et al., 2011), relay information in the stellate ganglion remains controversial (e.g. Gray, 1960; Wilson, 1960; Young, 1972). If so, modulation of the iridescence motor neural signal may be inhibitory or excitatory. In this regard, it is noteworthy that Miledi (Miledi, 1972) recorded intracellularly from ventral cells of the stellate ganglion and showed that they receive excitatory and inhibitory inputs from both the stellate connective and stellar nerves. Such findings highlight the integrative role of the ventral neuropile cells. Unfortunately, Miledi (Miledi, 1972) did not suggest a role for the cells that he recorded from.

In addition to activating fin iridescence, electrical stimulation of the FI nerve at 20 Hz produced a tonic contraction of the fin. Such contraction is not produced by giant axons, because we did not see any large axons in our fluorescent fills of the FI nerve and because none of the cell bodies filled were located within the giant axon lobe. Moreover, we also recognize that such contraction is not produced by units that control the undulatory fin motion in a live animal, because our denervations and behavioral results evidenced that such control is wired directly through the fin nerve. Thus, the function of FI nerve fibers that drive fin muscles remains to be elucidated.

At the current time, we are unable to elucidate what percentage of the cell bodies labeled through the FI nerve and located in the stellate ganglion innervate the fin musculature and what proportion target the iridophore layer. Alternatively, it is possible that each of the labeled cells branches at the base of the fin and innervates both iridophores and fin muscles, such that an increase in iridescence would always be coupled to a fin contraction.

Throughout this paper we have used the term motor neuron to refer to neurons that convey impulses to an effector tissue, the iridophore layer in this case. Indeed, within the iridophore layer it remains to be elucidated whether iridocytes are activated via classical synapses, en passant delivery or muscles. It has been suggested that muscles may be involved in the activation of iridescence because single iridocytes can be seen to ‘flicker’ (Wardill et al., 2012) and because muscles have been found closely associated to iridescent cells (Mirow, 1972). Currently, a correlation between muscle activity and iridophore iridescence is still lacking. Moreover, the speed of iridescence activation reported here and in Wardill et al. (Wardill et al., 2012) is not comparable to that of the squid muscular system (see supplementary material Movie 3). Nonetheless, this does not constitute evidence against some (yet unknown) involvement of muscles in the control of iridescence. For an iridocyte to be noticeable, reflectin must first be sufficiently condensed. This intracellular process is likely to be the time-limiting step for the speed of iridescence expression, even if a muscle mechanism with an active role in the iridescence process was activated at the same time.

Wardill et al. (Wardill et al., 2012) concluded that a process different to the reflectin condensation must cause the large and fast color shift seen upon iridophore activation, and suggested that a muscle mechanism may play this role. Shortly after, DeMartini et al. (DeMartini et al., 2013) reported that activated iridocytes expel water, lowering the inter-platelet distance and producing a color shift. Thus, it remains to be clarified whether muscles play a role in iridophore activation, either through the previously observed rapid iridocyte flicker or from fin muscles activated by motor neurons that travel through the FI nerve. Another possibility is that fin musculature activation by FI nerve fibers could modulate the control of iridescence via the release of fin muscle neurotransmitter. This case is unlikely because we did not observe a difference between denervated and wild-type animals upon stimulation of their FI nerve.

Unlike pigmentary chromatophore changes, the passive decline of iridescence is slow; the decline was substantial within 30 min, but some iridescence was still visible in nearly half of the tested animals after 1 h. Because squid in these experiments reduced their iridescence in response to dark adaptation and increased reflectance upon light exposure, we interpret this to mean that the slow iridescence changes are related to different light levels experienced in the wild. It is probable that squid tailor their iridescence expression to improve their camouflage depending on the amount of light present during the day. Moreover, a substantial iridescence reduction in low lighting conditions or at nighttime may also reduce the risk of predation. Squid are likely to detect such overall environmental luminance through their eyes, although neurons in the photosensitive vesicles (Young, 1978) or rhodopsin located in the skin (Mäthger et al., 2010) could also be employed. Squid such as D. pealeii are found nearshore at depths from 3 to 30 m when inshore during the summer (Shashar and Hanlon, 2013; Summers, 1983) and often forage near surface waters of 1–10 m depth at nighttime (Serchuk and Rathjen, 1974). In winter, D. pealeii lives offshore in canyons on the continental slope at depths up to 170 m (Summers, 1967) where light is dim even during the day. Even so, we recognize that in our testing setup the lighting was extremely low (0.4 nW) and such luminance may not be encountered by this squid species in the wild, especially because calculated luminance at 100 m depth of clear ocean water is 18 mW m⁻¹ m⁻¹ [calculated using water type values from Jerlov (Jerlov, 1968) and light penetration values from Gordon and McCluney (Gordon and McCluney, 1975)]. To clarify this point, it would be necessary to test iridescence output in a variety of low light intensities, as well as the outcome in the presence of conspecifics.

The only extant cephalopod species known to lack a stellate ganglion is Nautilus pompilius (Young, 1972; Sasaki et al., 2010), which is not a coleoid and does not possess neural control of its skin (Crook and Basil, 2008). The extinct ancestors of current coleoid cephalopods, named belemnoids, had elongated bodies and dorsal fins (Lewy, 2009), but it is not known whether they had evolved a stellate ganglion. Moreover, a comparative demonstration in sepioids and octopods of neural stimulation activating iridescence is currently lacking. Hence a comprehensive comparative approach among extant coleoids will be the most efficient route for understanding the evolution of iridescence control among coleoid groups.

This is the first study to describe part of the neural circuit controlling iridescence in a cephalopod. We have shown that (1) iridescence is controlled independently from pigmentary elements (chromatophores), (2) fin iridescence neural signals are routed through the stellate ganglion (instead of descending directly through the fin nerve) and the iridescent motorneurons likely originate within it, (3) iridescence expression requires the input of neurons descending from the brain, (4) passive decline of iridescence is slow and (5) squid turn their iridescence off and on in response to ambient darkness and light, respectively. At present, the brain area where the iridescence signals are computed remains to be located. In addition, single iridescence motor neurons will need to be stimulated to fully understand the role of each iridescence motor unit. Further behavioral studies are necessary to test whether squid can tune their iridescence output depending on its role for either camouflage or signaling. Our current research directions focus on these questions.

Squid were collected by trawling outside of Woods Hole, MA, USA [for more details, see Wardill et al. (Wardill et al., 2012)]. To minimize stress, squid were held at low density (1–5 squid) in each holding pen, made using divisions within large tanks. The tanks had a continuous high flow supply of seawater at 22°C. The squid were fed twice daily with live fish (Fundulus sp.). Healthy animals were put into a separate tank. Only the animals that had fed well for at least 1 day, a sign of strength and drop in stress levels, were used for surgical experiments. For animal transfer, squid were caught with a net and carefully moved by hand between enclosures. Any signs of stress displayed by the squid did not last more than a few seconds. For the experiments involving nerve cuts, the animals were held for brief periods of time (15 s), from which they recovered almost immediately. However, if an animal was restless when first held, it was lightly anesthetized by submersion for <30 s in 1.5% EtOH in seawater before denervation. For the experiments that required dissection of the fin tissues and nerves, the animals were first deeply anesthetized with 3% EtOH, and their unresponsiveness was tested before proceeding with decapitation and decerebration.

All procedures carried out in this study comply with institutional recommendations for cephalopods.

For denervation, the squid was held out of the water by gently wrapping the left hand around the squid mantle, between the fins and the head. Squid usually maintained their head firmly in a straight line with the body, which allowed the experimenter to look into the mantle cavity and visually identify the location of the stellate ganglion. Next, micro-scissors held with the right hand were used to cut the descending fibers. The complete process takes less than 20 s. All squid were denervated on the same side; the other half of the animal was left intact and served as an internal control.

To fill inputs and outputs of the stellate ganglion with fluorescent dyes, the stellate ganglia were removed and bathed in a Ca²⁺-free solution (Strathmann, 1987) for 10 min at room temperature. Then, they were transferred to a Petri dish with a bed of Sylgard 184 (Dow Corning, Midland, MI, USA). While the dish was dry, small cubes of Sylgard were cut out, forming a well that was lined with Vaseline applied through a blunted syringe tip. The nerve of interest was pinned with an insect pin into the inside of the well. The rest of the ganglion was pinned down to avoid movement. Next, the Vaseline syringe was used to make the walls of the well taller, trapping the nerve between the Vaseline layers. Subsequently, a few microliters of dye were delivered into the well. After confirming that the tip of the nerve was well submerged, the Vaseline syringe was used to close up the well in the shape of an igloo, and the preparation was covered with Tris-buffered seawater and left in the fridge (at 4°C) overnight. We followed this protocol for all the nerves of interest. Most ganglia were double filled (following Chrachri, 1995) through application of Lucifer Yellow to the output of the stellate ganglion and MicroRuby to the descending inputs. Tissue was prepared as shown in Gonzalez-Bellido and Wardill (Gonzalez-Bellido and Wardill, 2012).

In squid, the name iridophore refers to a group of iridescent cells (named iridocytes) that form an iridescent ‘splotch’ (Wardill et al., 2012). We recorded the changes of reflectivity from single iridophores in response to neural stimulation. The animals were tested either 7 (N=9) or 15 days (N=8) post-denervation. The same protocol as that of Wardill et al. (Wardill et al., 2012) was used, so that results between the two studies could be compared. Briefly, nerve fascicles named dermal nerves, which radiate from the base of the large fin nerve and innervate the skin, were exposed and stimulated via suction electrode. Electrical pulses of 7 V and 0.6 ms in duration were delivered at a frequency of 10 Hz for 15 s while iridescence was monitored with a spectrometer. The spectrometer was calibrated with a white standard, and the spectral reflectance was recorded every 0.1 s over 3–8 min. For analysis, the background reflectance was subtracted and the data were normalized and then smoothed with a Savitzky–Golay filter using MATLAB (The MathWorks, Natick, MA, USA). Normalized reflected intensity could be read directly after smoothing and the change in color over time was determined by finding the peak reflecting wavelength at each time point from the smoothed data.

Because the time taken for motor neurons to go through apoptosis depends on the temperature of the water, we monitored the denervated squid as they went though the described sequence, which consisted of relaxed chromatophores for approximately 2 days, expanded chromatophores on days 3–4, and the development of ‘passing clouds’ on day 5, which form perfect waves by day 7. Thus, according to Young (Young, 1972), at day 7 post denervation the cut neurons should have died.

After the stellate connective was cut, each animal (N=15) was placed back into a small round arena (25 cm diameter) with black felt-lined walls, a smooth gray substrate and a continuous seawater supply. The animal was videoed for 60 min. To assess iridescence change over time, we extracted images from the video at 5, 10, 20, 30, 40, 50 and 60 min. At each time point, the iridescence was scored as one of the three options: no change, substantial decline or completely absent.

Squid were placed in an arena with black lined walls and black pebbled substrate 24 h after denervation. The whole ensemble was placed inside a blackened tent surrounded with galvanized sheet metal (0.4 nW) and the animal was left to dark adapt for either 1 (N=17) or 2 h (N=6). After that time, a picture was taken with a remote trigger using the Canon EOS 5D MarkII camera with flash (Melville, NY, USA) to capture the body patterning of the animal.

To assess iridescence changes due to light exposure, animals (N=4) were dark adapted as explained above. The arena was covered with a glass panel to prevent ripples on the water surface reflecting light into the camera. Animals were imaged at 300 frames s⁻¹ with a Casio Exilim Pro EX-F1 camera (Dover, NJ, USA) as they were exposed to continuous light.

We thank fellow lab members for their support and discussion of this study. We thank MBL Equipment Resources, MBL Apparatus Department and Zeiss Microscopes for assistance with equipment. We also thank Nathan Boor (Aimed Research, Burghill, OH, USA) for his generosity in lending us the high-speed video camera. We thank the MBL Central Microscopy facility for providing imaging resources and the Aquatic Resources Division of MBL for supplying squid.