Transparent anemone shrimp (Ancylomenes pedersoni) become opaque after exercise and physiological stress in correlation with increased hemolymph perfusion

Cryptic organisms employ various strategies to avoid being detected by predators, including matching the background (Endler, 1978). Background matching is often mediated by pigmentation patterns (e.g. melanism in the peppered moth; reviewed in Ruxton et al., 2004) that allow prey to blend in with their surroundings. However, certain color patterns may match the background of only one specific microhabitat, thus restricting pigmented prey from exploiting larger portions of their environment (Johnsen and Sosik, 2003). Transparency is a cryptic strategy that solves these problems by allowing organisms to match any background. However, unlike colored camouflage, that only involves the body surface, transparency must be maintained throughout the volume of an animal (reviewed in Johnsen, 2001). To maintain transparency, animals may need to limit exercise or other forms of physiological stress, as these may alter the internal ultrastructure and thus increase light scattering.

In any substance, variation in refractive index on size scales greater than a half wavelength of light can lead to observable light scattering and thus increase opacity (Benedek, 1971). For example, snow, while non-absorbing, is opaque because of its complex structure that strongly scatters light. In the case of muscle, any variations in tissue density and composition that can occur during periods of physiological stress may lead to an increase in opacity.

Whole-body transparency is not an unchangeable or static physical state. The siphonophore Hippopodius hippopus becomes opaque when touched because proteins are thought to precipitate in its mesoglea and increase light scattering (Mackie and Mackie, 1967). Opacity also increases in other transparent animals as a result of physiological or environmental stressors. Clear ghost shrimp (Palaemonetes pugio) lose their transparency because of changes in temperature and salinity that occur over the tidal cycle (Bhandiwad and Johnsen, 2011). For example, P. pugio transmits approximately 50% of incident light through its abdomen at salinities of 0 to 20 ppt at 20°C, but nearly 0% at salinities of 30 ppt at 30°C. Surviving shrimp returned to their pre-stress levels of transparency within 24 h of being returned to the control conditions (15 ppt, 20°C) (Bhandiwad and Johnsen, 2011).

Because light scatters when it encounters a medium of a different refractive index, Bhandiwad and Johnsen (2011) hypothesized that the loss in transparency was related to the pooling of hemolymph in the intramuscular space, thus creating regions of low-index fluid between the regions of high-index muscle. An approximate calculation demonstrates how even a seemingly small amount of reflection at one interface can add up to a significant reduction in the amount of light that can be transmitted (Fig. 1) (Bhandiwad and Johnsen, 2011). Using Fresnel's equations, and assuming perpendicular incidence of the light and refractive indices of ∼1.5 for muscle and ∼1.35 for hemolymph, 0.3% of light is reflected at each interface [=(1.5–1.35)²/(1.5+1.35)²]. Over the approximately 2 mm thickness of the shrimp tissue there can be hundreds of such interfaces, substantially reducing the amount of transmitted light (Fig. 1C). This admittedly simplified analysis provides a conservative estimate of light scattering as the interfaces are not always perpendicular to the incident light and thus would scatter even more light. It demonstrates, however, that even small amounts of hemolymph surrounding hundreds of individual fibers can opacify an animal (Fig. 1).

After observing that several species of transparent shrimp (Ancylomenes pedersoni, Periclimenes yucatanicus and Periclimenes rathbunae) become opaque after exercise, we devised a set of four experiments that allowed us to test directly and expand upon the Bhandiwad and Johnsen (2011) hypothesis. In our study, we use the transparent shrimp Ancylomenes pedersoni (Chase 1958) (Crustacea, Decapoda), which inhabits Caribbean reef anemones (Bartholomea annulata, Condylactis gigantea). This shrimp is normally so transparent that when placed on a text background, one can read through its abdomen (Fig. 2A). When threatened, A. pedersoni performs a tail-flipping escape response that propels it backwards multiple times. We noticed, though, that multiple escape responses resulted in both exhaustion and a significant decrease in transparency (Fig. 2B). Although it is clear that maintaining transparency is critically important for camouflage (e.g. Zaret, 1972; Zaret and Kerfoot, 1975; Tsuda et al., 1998; Utne-Palm, 1999), less is known about whether and how transparency can be disrupted. Here, we use several experimental methods to test whether changes in perfusion owing to exercise, as well as three other physiological and environmental stressors, can reversibly disrupt transparency in A. pedersoni.

Individual A. pedersoni were collected at 2–20 m depth on coral reefs surrounding Carrie Bow Cay, Belize (16°46′N 88°04′W). Additional specimens were collected in the Florida Keys, USA, by Dynasty Marine Lab (Marathon, FL, USA). Animals were maintained in full-strength, filtered seawater at room temperature (FSW; 35‰ salinity, 21°C) in aerated, recirculating aquaria and were fed fish meal pellets (‘Crab cuisine’, Hikari, Hayward, CA, USA) three times weekly. Animals were not fed in the 24 h prior to any experiments.

Prior to documenting hemolymph flow to tissues, we first determined how much exercise was necessary to turn the shrimp opaque and how long it took them to recover their transparency. Each shrimp (n=23) was placed in an individual aquarium (water maintained at room temperature), and a stimulus (forceps) was placed in front of the rostrum to generate the tail-flip escape response (this response consistently led to the tail-flip escape behavior until the shrimp was too fatigued to move). The number of tail flips was recorded until the animals were fatigued, defined as no longer making escape responses, swimming movements or attempts to evade the forceps. In this preliminary analysis, shrimp were photographed next to a trans-illuminated transparency standard (EIA Halftone Gray Scale, Rochester, NY, USA). For all shrimp, the time to recovery was measured by recording the point at which the shrimp regained their transparency and began to move. At this point, the shrimp were again photographed. We analyzed these photographs before and after treatment, but the all-or-nothing response in opacity (either highly transparent with values greater than 60%, the highest transmission value on the standard, or nearly 100% opaque with light transmission less than or equal to 3%, the lowest transmission value on the standard) led us to simplify the results, which we report as either presence or absence of opacity. This presence or absence of opacity is shown in photographs taken with the shrimp in front of a colored or a black background to best visualize the cloudy-white appearance of the opaque tissue.

In a subsequent experiment using new individuals, the shrimp were divided into either the control group (no exercise, n=5) or the experimental group (exercise/tail flips, n=5). Individual shrimp from both groups were first anesthetized via exposure to 1°C seawater for 10 min. Each shrimp was then injected with 100 µl of Alexa Fluor 594-labeled wheat germ agglutinin (WGA; Molecular Probes) in FSW (1 mg 1⁻¹ ml) into the pericardial cavity using a 32 gauge needle. WGA is a lectin that binds to the sialic acid and N-acetylglucosaminyl residues present on the basement membrane of the muscle fiber sarcolemma and blood vessel endothelium (Wright, 1984). When injected into the pericardial cavity of a shrimp, WGA moves through the circulatory system and tissues, revealing regions that are in contact with hemolymph.

After injection, the shrimp recovered for 15 min at room temperature. The control group was immediately euthanized (without performing a tail-flip response), and the abdominal muscle was immediately fixed in 4% paraformaldehyde in FSW. The experimental exercise group was induced to tail-flip in the same manner as the initial experiment by placing a stimulus (forceps) in front of their rostrum to generate an escape response. After 16 tail flips (all specimens were opaque; duration of 2 min or less), the shrimp was subjected to the same treatment as the control: immediately euthanized, and the abdominal muscle preserved.

Similar to the exercise experimental treatment, this experiment and all subsequent experiments involved anesthetizing each shrimp via chilling to 1°C, injection of 125 µl of Alexa Fluor 594-labeled WGA in FSW into the pericardial cavity followed by recovery for 15 min. Shrimp were then divided into three groups: hyposaline (diluted seawater at 8‰; n=3), control (normal seawater at 35‰; n=6) and hypersaline (concentrated seawater at 55‰; n=3). Exposure was for 30 min before being photographed, euthanized and preserved.

After the WGA injection and recovery period, we perforated the abdomen of the shrimp in this experimental group (n=3) using forceps (holding the shrimp in place with dental wax so that they could not tail flip) in order to trigger localized perfusion as part of the wound healing/inflammation response. We then photographed them and preserved the punctured segment.

Although circulatory regulation in crustaceans is poorly understood, there is evidence that peptidergic hormones have effects on isolated hearts and cardioarterial valves as well as in vivo (McGaw et al., 1994a). For instance, infusion of the crustacean peptide hormone proctolin resulted in a ‘vasodilation-like’ effect of increased cardiac output and a shunting of blood to the sternal artery of Cancer magister (McGaw et al., 1994a), thereby increasing blood flow to the abdomen.

Fifteen minutes after injecting WGA, we injected 50 µl of proctolin (1 mmol l⁻¹) into the pericardial cavity in the experimental animals and a saline solution in the control animals. In eight out of 10 proctolin injections, the shrimp did not survive the infusion; therefore, we only present the results from the two shrimp that survived. These were photographed, euthanized and preserved.

All tissues were fixed for at least 8 h in 4% paraformaldehyde in FSW, rinsed overnight in 25% sucrose in deionized water, and then mounted with Tissue-Tek® O.C.T. Compound (Sakura® Finetek, USA). Frozen sections were cut at 20 to 30 μm with a Leica Cryocut 1800. Sections were rinsed in PBS, incubated for 30 min in 300 nmol l⁻¹ DAPI, and then rinsed again for 5 min in PBS before being viewed.

Imaging was performed with an Olympus FluoView 1000 laser scanning confocal microscope. Sections from the control specimens were imaged first, and thus served to set the threshold levels required for viewing the red WGA channel, which resulted in some image saturation in the subsequent experimental specimens that were viewed under the same conditions. All confocal micrographs were analyzed using the software package FIJI (Schindelin et al., 2012). For each individual shrimp, at least five sections of muscle tissue were examined and 10 micrographs randomly taken (using systematic uniform random sampling; Howard and Reed, 1998) at the same magnification were analyzed. A stereological point-counting method (Howard and Reed, 1998) was used; a point grid was applied to the image, and all points touching a perfused area (red stain=perfused blood vessel or sarcolemmal area) were tallied and divided by the sum of points to determine the fraction (0 to 1) of perfused tissue, which are reported as mean values per individual. Non-parametric Mann–Whitney U-tests were performed to test for significant (P<0.05) differences between the control and experimental conditions in each of the four treatments: exercise, salinity, perforation and proctolin injection.

To visualize the shrimps' overall muscle morphology and circulatory system, one control (at rest, transparent) and one experimental (tail flip, opaque) A. pedersoni shrimp were scanned at ultra-high resolution using the micro-CT scanner (model: Nikon XTH 225 ST) at the Shared Materials Instrumentation Facility at Duke University. Because previous histological examinations of sectioned tissue from shrimp fixed in 4% paraformaldehyde in phosphate buffer did not differ from that of shrimp fixed in 2.5% glutaraldehyde in phosphate buffer (L.E.B., unpublished data), shrimp from both fixation preparations were utilized. First, the control shrimp was anesthetized via exposure to 1°C seawater for 10 min and then, in its undisturbed, transparent state, fixed in 2.5% glutaraldehyde in phosphate buffer for 24 h before being transferred to a tube containing 10 ml of Lugol's iodine diluted down to 3.4% concentration. The shrimp soaked in this iodine solution for 7 days before being mounted and scanned at a resolution of 6.74 voxels, at 2500 projections with an exposure time of 500 ms, a voltage of 125 Kv, a current of 104 µA and no filter. Frame averaging was set to 4 for this scan. The second shrimp specimen was induced to tail flip 15 times until opaque, was anesthetized via exposure to 1°C seawater for 10 min, and then was fixed in its opaque state in 4% phosphate-buffered paraformaldehyde for 24 h before being transferred to a tube containing 10 ml of Lugol's iodine diluted down to 3.4% concentration. The shrimp soaked in this iodine solution for 7 days before being mounted by being placed in a tube of the original 4% paraformaldehyde fixative and scanned in two parts (and later stitched together) at 7.47 voxels at 2200 projections with an exposure time of 500 ms, a voltage of 120 Kv, a current of 114 µA and no filter. Frame averaging was set to 2 for this scan. Stacks of virtual sections produced by the micro-CT were used for 3-D visualization in the software Avizo version 9.1.1 (FEI Visualization Sciences Group) and the volume rendering function was chosen to maximize contrast between muscular tissue and arterial vessels.

We found that all four experimental treatments – exercise, salinity, perforation and proctolin injection – temporarily disrupted transparency (Figs 2B and 3). Additionally, our stereological analysis of confocal microscope images found that all four experimental groups of shrimp showed substantially more hemolymph perfusion in the abdominal musculature than any of the control groups, though only the exercise and salinity treatments had large enough sample sizes to be statistically significant (Fig. 4).

Three-dimensional reconstructions from CT scans of the shrimp show the abdominal musculature and larger vessels; we use them here as a model to describe the shrimp's general anatomy. Transverse sections taken through the abdomen (Fig. 5) demonstrate the tissue plane appearing in all subsequent confocal images.

First, we observed that some shrimp became opaque after as few as three tail flips. Eliciting additional tail flips to the point of exhaustion (16±1 s.e.m.; n=23) resulted in complete opacity (Fig. 2B) for all shrimp, with original transparency and mobility recovering after 20–60 min; there was no correlation between number of tail flips and time to recovery.

The control group of non-exercised shrimp remained transparent and micrographs of cross-sections revealed that the injected WGA had not perfused to the abdominal muscle tissue, demonstrated by the lack of staining (Fig. 6A) or staining of the main artery only (0.005±0.004, mean±s.d. fraction of tissue perfused; Fig. 4). The experimental (exercised) group showed an obvious difference (0.55±0.06, mean±s.d. fraction of tissue perfused; Figs 4 and 6). After 16 tail flips, the shrimp became opaque and micrographs revealed hemolymph staining of the endothelia and the sarcolemmal boundaries of the abdominal musculature (Fig. 6B). Perfusion was statistically greater in the exercised group than the control, non-exercised group (U=25, P=0.01).

Shrimp placed in hyposaline (8‰) and hypersaline (55‰) water became opaque in fewer than 2 min (Fig. 3A), but recovered transparency within 10 min in FSW.

The control shrimp (in 32‰ seawater) remained transparent, and micrographs revealed little to no hemolymph had perfused through the abdominal muscle tissue; only the main artery showed staining (Fig. 7B). Micrographs of hyposaline and hypersaline treatments showed staining around individual muscle fibers (Fig. 7A,C). The fraction of tissue perfused in both the hyposaline and hypersaline treatments had similar means (0.60±0.08 versus 0.59±0.07 mean±s.d., respectively), thus we lumped the salinity treatments together (n=6) for statistical analysis and found that the hemolymph perfusion after exposure to changed salinity was significantly greater than in the control (U=36, P=0.005; Fig. 4).

Incisions in the abdomens of three shrimp resulted in opacity surrounding the puncture site (Fig. 3B). Micrographs of the WGA staining revealed staining of the tissues surrounding the abdominal vessels at the site of the wound (Fig. 7D). The fraction of tissue perfused at the localized site of the punctures (0.41±0.04 mean±s.d.) was greater than that of the undisturbed control shrimp (0.003±0.004; Fig. 4), though the sample size of only three individual shrimp meant that the ranked, non-parametric Mann–Whitney U-test could not show significance beyond the P=0.1 level (U=9, P=0.1).

Injection (n=2) with proctolin resulted in full-body opacity in one specimen and opacity everywhere but the distal tip of the tail in the second (Fig. 3C). Micrographs from a control (saline-injected) specimen (Fig. 7E) revealed staining confined to the main artery. Micrographs from the proctolin-injected specimens showed staining of additional vessels and around muscle fibers and what appears to be an increase in diameter of the artery (Fig. 7F). Following the same trend as the other treatments (Fig. 4), the fraction of tissue perfused in the proctolin-injected shrimp (0.49±0.04 mean±s.d.) appeared greater than in the saline-injected controls (0.018±0.004), but with a sample size of only two injected shrimp, the Mann–Whitney U-test was not significant (U=4, P=0.3; Fig. 4).

The results of these four experimental treatments show that the transparency of A. pedersoni can be reversibly disrupted after exposure to physiological or environmental stressors, and that the increase in opacity is associated with a corresponding increase in perfusion of hemolymph to the abdominal musculature. While there may be additional mechanisms that disrupt transparency, the correlation of increased perfusion with increased opacity suggests that increased hemolymph flow (and thus presumably widened interfaces) between muscle fibers increases light scattering.

An object or organism is transparent if it neither scatters nor absorbs light. Most biological tissues do not have pigments and do not absorb light, but crustacean hemocyanin has a weak absorbance band centered at 340 nm when de-oxygenated and at 580 nm when oxygenated (Redfield, 1930; Erker et al., 2008). Despite this, even if the concentration of hemocyanin in the hemolymph of A. pedersoni is as high as in other crustaceans, the optical path length is too short to affect transparency significantly. Additionally, absorbance makes tissue darker or more strongly colored rather than opaque. Thus, the white opacity we see in Figs 2B and 3 must be a result of light scattering.

Modeling light scattering in the abdominal muscle tissue and circulatory system of a crustacean is fundamentally difficult and computationally expensive owing to the extraordinarily complex relationship between tissue ultrastructure and how it scatters light (Bohren and Huffman, 1998; Johnsen and Widder, 1999). In cases of simple or highly regular structures, scattered light can be calculated by solving Maxwell's equations analytically (e.g. Benedek, 1971; Ameen et al., 1998). However, the abdominal tissues in shrimp (unlike that of a cornea) do not possess the required simplicity. Numerical methods such as the finite element method have been used in calculating light scattering based on the actual ultrastructure for individual cells (e.g. Dunn and Richards-Kortum, 1996; Drezek et al., 2000). However, they are computationally expensive for multicellular tissue. In addition, unless many runs are performed including or excluding various components, it is impossible to determine what features are most important for scattering at various angles.

There are several parameters, however, that are known to be important, including the number of interfaces between tissues of different densities (e.g. between muscle fibers and hemolymph; Fig. 1). A shrimp that is approximately 2 to 2.5 mm wide with fiber diameters of approximately 20 μm has roughly 250 interfaces across its body, which would reduce light transmission by at least half (Fig. 1). Again, this is a conservative estimate accounting only for light reflected directly back at the interfaces and not the additional scattering that would occur across all angles of incidence. Vessel diameter may also be an important parameter. For example, for arterioles and capillaries that are at least a few microns in diameter and separated by distances of a similar magnitude or greater, light scattering is linearly correlated with vessel diameter (Bohren and Huffman, 1998). Unfortunately, fixation artifacts and the saturation of the stain in some confocal images make calculating an accurate vessel diameter difficult. For example, it is unclear whether any vessel lumena shrink to less than a wavelength of visible light. The micro-CT scans were not able to resolve details of the smaller arterioles and capillaries, but the data from one control and one experimental shrimp do appear to add further support to our hypothesis that vessels widen and more hemolymph spaces appear in experimental (opaque) shrimp (Fig. 8, Movie 1). Our study does not definitively rule out other mechanisms, such as cellular changes (protein precipitation) or changes to the blood itself that may also increase light scattering.

Taking into account the difficulty of accurately determining light scattering from these complex tissues, our simplified model nevertheless demonstrates that the hemolymph surrounding the individual fibers owing to increased blood flow can create hundreds of interfaces throughout the tissue, resulting in increased opacity.

Our results show that in all experimental treatments that caused an increase in perfusion to the tail, the tissue turned opaque. Just as humans vasodilate (and turn red) during exercise, shrimp apparently perfuse their muscle tissues during tail-flip escape responses and turn white. Similarly, injecting a ‘vasodilator’, or puncturing the shrimp, is expected to increase blood flow, though only locally in the latter case. Additionally, sudden changes to the shrimp's environmental conditions (increasing or decreasing salinity and temperature) also causes an increase in oxygen consumption and metabolic rate, resulting in higher cardiac output and stroke volume (Bhandiwad and Johnsen, 2011; DeWachter and McMahon, 1996).

In all cases, the control individuals showed little to no evidence of hemolymph perfusion in the abdominal musculature. The main artery was stained with WGA, indicating presence of the labeled hemolymph, but it appeared that hemolymph had not perfused the smaller vessels in the abdomen even 20 to 30 min after injection. This suggests control of perfusion so that blood is shunted away from the abdomen to the thorax. In fact, other studies of decapod crustaceans have found that hemolymph flow in one or more arteries fluctuates independently of flow through the other arteries, suggesting that there are complex underlying control mechanisms of the circulatory system (McGaw et al., 1994a,b). The differential release of peptidergic (e.g. proctolin) or aminergic hormones could function to mediate these kinds of cardiovascular responses either directly or indirectly via their actions with the central nervous system and could divert flow to the anterior vessels and away from the posterior ventral axis (Airriess and McMahon, 1992).

Transparent animals are typically considered to be more fragile and slower than their pigmented congeners (Childress et al., 1990; Johnsen, 2001), but this may be the first documented example of an animal that is diverting blood from its tail muscles presumably to maintain its camouflage, raising questions about the metabolic rate of transparent shrimp and their ability to recover from aerobic activities. The abdominal musculature of these shrimp consists mainly of fast-twitch, anaerobic muscle fibers and few mitochondria (Jimenez et al., 2008). At rest, the metabolic needs of this abdominal tissue are not believed to be great, and it is possible that diffusion or periodically perfusing the abdomen is all that is required to maintain muscle function. The ‘visual-interaction hypothesis’ (Childress et al., 1990) states that, in the absence of visually mediated predator–prey interactions, the distances over which predators and prey interact are reduced and selective pressure for rapid locomotory capacity is relaxed. This may explain why transparent shrimp have limited locomotory abilities. As seen in our study, these benthic anemone shrimp have retained their ability to rapidly tail-flip away from a predator, but are capable of only a limited number of contractions before fatiguing, and performing this escape behavior sacrifices their transparent camouflage and leaves them vulnerable to visual predation during a perhaps extended recovery period.

Additional research on the costs and benefits of transparency as a camouflage strategy is needed. For example, we do not know how these animals may deal with pulsatile oxygen delivery, which in most tissues would lead to a burst of reactive oxygen species and tissue damage. Limiting abdominal blood flow may affect their ability to respond to environmental challenges such as salinity or temperature variation, or leave them vulnerable to predation when their transparency is disrupted under these kinds of changes in environmental conditions. Ancylomenes pedersoni spend much of their lives stationary, waiting for client fish to approach their cleaning station (Huebner and Chadwick, 2012). They do swim toward the client fish to engage in cleaning services, but prolonged tail-flip behaviors that result in opacity have not been reported. Anecdotally, the first author moved an anemone to a new location, which caused the resident shrimp to tail-flip at least 10 times until it turned opaque. At this point, several wrasse chased and one successfully caught and ate the now-visible shrimp (L.E.B., personal observation). As Bhandiwad and Johnsen (2011) demonstrated, the sighting distance for an opaque animal is greatly increased, making it more vulnerable to predation. Thus, one trade-off to transparency may be a prolonged recovery time from exercise where the animal is vulnerable because it is opaque and not able to undergo another escape response. This begs the question of whether pigmented shrimp occupying the same niche are better able to recover more quickly from exercise and respond to changes in environmental conditions more effectively than A. pedersoni, and whether similarly sized pigmented shrimp occupying the same ecological habitat also have little to no perfusion to the abdomen at rest. While more robust testing from a diverse phylogenetic group is currently underway, the results we obtained from one species of pigmented shrimp, Lysmata pederseni, at rest (n=2) lend further support to the hypothesis that hemolymph perfusion disrupts transparency and normally opaque shrimp with pigmented cuticles do not have reduced perfusion (Fig. 9).

Even though we cannot definitively rule out additional mechanisms that may disrupt transparency, the evidence presented here suggests that increased perfusion is one likely mechanism that results in increased light scattering. The fact that transparent shrimp at rest show little to no evidence of perfusion to their abdominal musculature suggests that they may experience significant physiological trade-offs, such as reduced performance and extended recovery from exercise or other stressors.

We thank two anonymous reviewers and William Kier, Kyle Naughton, Chengyi Xu, Robert Fitak, Lorian Schweikert, Eleanor Caves, Julia Notar, Sarah Solie, Kate Thomas and Benjamin Wheeler for comments on earlier versions of this manuscript. We also thank Richard Dillaman, Alison Taylor and Mark Gay (UNC Wilmington) for assistance with confocal microscopy.