Journal of Comparative Physiology A

, Volume 199, Issue 8, pp 669–680

Polarotaxis and scototaxis in the supratidal amphipod Platorchestia platensis


    • College of Earth, Ocean and Environment, School of Marine Science and PolicyUniversity of Delaware
  • Meagan R. Putts
    • Department of Marine ScienceEckerd College
Original Paper

DOI: 10.1007/s00359-013-0825-7

Cite this article as:
Cohen, J.H. & Putts, M.R. J Comp Physiol A (2013) 199: 669. doi:10.1007/s00359-013-0825-7


Talitrid amphipods use many cues for orientation during forays between temporary burrows and feeding areas, and for locating beaches when submerged, with visual cues being particularly important. Little evidence exists for polarized light among these visual cues despite extensive orientation by celestial and underwater polarized light in other crustaceans and in insects. We used electroretinography to assess spectral sensitivity in the eye of the beach flea Platorchestia platensis, and behavioral studies to test whether linearly polarized light serves as an orientation cue. Two spectral classes were present in the P. platensis eye with maxima at 431 and 520 nm. Non-uniform orientation of amphipods in the laboratory arena required either light/dark or polarized cues. Scototactic movements depended on arena conditions (day/night, wet/dry), while orientation under linearly polarized light was wavelength-dependent and parallel to the e-vector. Subsequent tests presented conflicting and additive scototactic and polarotactic cues to differentiate among these responses. In dry conditions, orientation parallel to the polarization e-vector overcame a dominant negative scototaxis, confirming that polarotaxis and scototaxis are separate orientation responses in this species. These behavioral results demonstrate talitrid amphipods can perceive and orient to linearly polarized light, and may use it to orient toward preferred zones on beaches.


Polarized lightOrientationSpectral sensitivityBehaviorTalitridae


A range of cues are used by talitrid amphipods inhabiting sand beaches to orient between daytime burrows and nocturnal feeding areas, but visual cues have emerged as particularly critical in directing behavior (Scapini 2006). Multiple internal chronometric mechanisms allow talitrids to compensate for movements of the sun and moon over the diel cycle and select bearings relative to the position of these objects in the celestial hemisphere (Ugolini 2003). When the sun and moon are not in clear view, the skylight luminance gradient can be used to guide amphipod orientation (Ugolini et al. 2012). In addition to celestial cues, talitrids may use landscape cues such as relative light/dark areas in the visual surround, local landmarks, and possibly chromatic targets (e.g., green vegetation and blue water) to direct movements toward preferred locations on sand beaches (Nardi et al. 2000; Ugolini et al. 2006; Walsh et al. 2010). These behavioral tasks involve light perception by talitrid compound eyes, which possess at least two photoreceptor spectral classes. A ultraviolet/blue (390–450 nm) sensitive pigment is present in distal R5 cells, while a green sensitive pigment (~520 nm) is present in proximal R1–4 cells (Mezzetti and Scapini 1995; Ugolini et al. 1996, 2010; Cohen et al. 2010). Behavioral studies suggest short-wavelength sensitivity is essential for sun-compass behavior, while longer wavelength sensitivity is required for entrainment of endogenous activity rhythms (Ugolini et al. 1993; Forward et al. 2009a, b). The presence of color vision and the roles played by chromatic cues in other aspects of talitrid vision and orientation remain open questions. Further, it is not known if the compound eyes possess specialized regions for specific visual tasks (e.g., insect dorsal rim area for polarized light detection, Labhart and Meyer 1999), or the functional significance of extraocular photoreceptors such as the intracerebral ocelli present in at least one talitrid species (Frelon-Raimond et al. 2002).

Linearly polarized light is used in orientation by a variety of organisms, most notably insects, for localizing cloud-obscured celestial bodies and identifying landscape cues during both the day and night (reviewed in Wehner and Labhart 2006; Warrant and Dacke 2010; Cronin and Marshall 2011). Polarized light is also utilized for orientation, signaling, and object detection in subtidal arthropods (Goddard and Forward 1991; Shashar et al. 2011). Despite the demonstrated range of light-mediated orientation abilities in talitrid amphipods, little direct evidence exists for either visual polarized light detection or its use in orientation. Pardi and Papi (1953) reported that this ability may exist, but did not clearly differentiate between polarotactic and scototactic behavioral responses that occur when linear polarizing filters are used above circular arenas (Jander and Waterman 1960). Yet, a role for linearly polarized light has been speculated in talitrid orientation, and ultrastructural evidence of orthogonally arranged microvilli in retinular cells of their compound eyes, but not intracerebral ocelli, is consistent with polarization sensitivity (Frelon-Raimond et al. 2002; Ugolini et al. 2010). Clear behavioral and/or electrophysiological evidence for polarization sensitivity in amphipods, talitrids or others, is lacking.

The beach flea Platorchestia platensis (Krøyer 1845) is a talitrid species with a cosmopolitan distribution on tropical and temperate shores throughout the Atlantic, Gulf of Mexico, western Pacific, and Mediterranean (LeCroy 2011). As a primary colonizer of shorelines, P. platensis commonly becomes established prior to more specialized talitrids, and displays considerable flexibility in its use of the littoral/supralittoral zone, from rocky to muddy shores, including beaches, marshes, and mangroves (Bousfield 1973). When present, they are often found in large numbers under beach wrack and decomposing vegetation at or just above the high tide line, where they feed on decaying plant matter, algae, and various invertebrates, and are preyed upon by crabs, spiders, and insects (Behbehani and Croker 1982). As with many talitrids (e.g., Forward et al. 2009b), P. platensis possess a nocturnal activity rhythm that is entrained to the L:D cycle (J. Cohen, unpublished data) and regulates their emergence at sunset to forage between the waterline and upper beach, as well as their subsequent return to refuges beneath beach wrack at dawn. This activity rhythm lacks a tidal component; thus, when P. platensis are under beach wrack during the day, and following their emergence at night during beach foraging, they are vulnerable to being washed into shallow water by waves, particularly on narrow beaches that are inundated at high tides. If submerged during high tides and unable to return to shore, P. platensis is capable of clinging to floating wrack and remaining adrift until it locates a suitable habitat (Wildish 2012).

Platorchestia platensis has need to orient relative to the waterline during periods of nocturnal activity following twilight emergence when celestial polarization patterns are pronounced and stable (e.g., Cronin et al. 2006), as well as to locate shorelines when swept into open water during the day or night where time-dependent polarization patterns are present both underwater and in the sky even if the sun or moon is obscured by clouds (e.g., Cronin and Marshall 2011). Accordingly, this species was chosen as a model in which to investigate polarotaxis, as linearly polarized light could be useful for amphipod orientation relative to the waterline in these scenarios, and is an aspect of talitrid orientation that has not been studied in any detail. This study used electroretinogram recording to assess photoreceptor spectral sensitivity in P. platensis and a series of laboratory behavioral experiments to test whether overhead linearly polarized light serves as an orientation cue. The latter experiments also characterized scototaxis as an orientation cue in P. platensis as the experimental design necessitated differentiating between scototactic and polarotactic responses in the test arena.

Materials and methods

Electrophysiology experiments

Platorchestia platensis was collected from under beach wrack on a southeast-facing sand beach of Boca Ciega Bay, St. Petersburg, Florida (USA) as needed for experiments during June–August 2009. Amphipods were brought to the nearby Galbraith Marine Science Laboratory (Eckerd College), where all experiments were conducted, and placed in a laboratory aquarium filled with 5 cm of seawater-moistened sand. Every other day the sand was aerated by hand and food (cellulose tissues and fish food flakes; Wardley, Secaucus, NJ, USA) was replaced. Animals were maintained under the ambient 14:10 light:dark cycle and a laboratory temperature of 23 °C for no longer than 1 week before being used in experiments.

Electroretinograms (ERGs) were recorded from live P. platensis (mean body length = 6.91 mm ± 1.53 SE, N = 8) restrained on the plastic head of a pin using cyanoacrylate gel adhesive (Fig. 1a). An individual was placed in an acrylic bath filled with 23 °C seawater such that its dorsal surface faced upwards and the anterior–posterior axis of its body paralleled the water. The animal’s pereopods were free to move and touched the seawater surface to keep it hydrated, but its body was not submerged. Differential ERG recordings were made by placing one epoxy-insulated tungsten microelectrode (1 μm diameter exposed tip) sub-corneally, with a similar reference electrode placed in the head away from the eye. The seawater bath was grounded by an AgCl-coated wire. Differential A.C. recordings were amplified and filtered (X-cell-3 amplifier, FHC, Bowdoinham, ME, USA), then digitized and stored in LabView (Version 6.1, National Instruments, Austin, TX, USA) for later analysis of peak-to-peak response heights (see below).
Fig. 1

Platorchestia platensis and the behavioral arena. aP. platensis restrained on a pin for electrophysiology experiments. b Diagram of the laboratory arena used for behavioral experiments. An overhead lamp with wavelength (λ) and neutral density (ND) filters illuminated the arena bowl from above. A polarizing filter layered over wax paper presented either a linearly polarized or unpolarized overhead light field when viewed from inside the bowl depending on which side of the filter faced down toward the bowl. The arena bowl was surrounded by either a solid white collar, or as shown here, a collar with alternating white and black (i.e., light and dark) quadrants. The bowl was positioned on a stand with a mirror, such that an observer could view an amphipod through the bowl’s transparent bottom. For clarity, both the stand and a shield keeping stray light from entering the arena are not shown. c Spectral reflectance of the bowl wall under all experimental collar and polarization conditions with 436-nm filtered light. The left panel shows reflectance with a solid white collar surrounding the arena, while the right panel shows reflectance with an alternating white and black collar (as in panel b) with reflectance measured for both white and black quadrants (solid and dashed lines, respectively). For both collar conditions, reflectance measurements were made under unpolarized (black, red lines) and polarized (blue, green lines) overhead light, parallel (∥) and perpendicular (⊥) to the direction of the polarizing filter e-vector. Unpolarized light perpendicular to the e-vector (left panel, black line) served as a reference spectrum for all measurements. Reflectance data are plotted over the wavelength range of 427–437 nm, which is the full width at half maximum of the light peak present in the arena illuminated by 436-nm filtered light transmitted through the polarizing filter. The internal reflection effect of the polarizer is particularly evident in reflectance from black quadrants

Light stimuli were generated from a monochromatic source (100 W voltage-regulated quartz-halogen lamp, HORIBA Jobin–Yvon, Edison, NJ, USA; CM110 monochromator, Spectral Products, Putnam, CT, USA) with spectral purity controlled by blocking filters (7 nm full width at half maximum [fwhm]). A fused-silica neutral-density filter wheel under the control of a stepper motor was used to adjust irradiance. The system was calibrated at 10-nm intervals with a photometer and radiometrically calibrated probe (Models S471 and 260, UDT Instruments, Baltimore, MD, USA). An electromagnetic shutter (VS25, Uniblitz/Vincent Associates, Rochester, NY, USA) under computer control provided a stimulus flash duration of 75 ms, which was delivered to the P. platensis eye via one branch of a bifurcated, randomized fused-silica light guide (EXFO, Richardson, TX, USA). The other branch delivered light from an accessory lamp (DC-950, Dolan-Jenner, Boxborough, MA, USA) onto the eye during specimen preparation and chromatic adaptation experiments.

Spectral sensitivity experiments employed the criterion response method as described by Cohen et al. (2010). An experiment began once the response to a dim test flash remained constant for 1 h, which indicated that the eye was fully dark adapted. Dark-adapted spectral sensitivity curves were generated for each individual by finding the irradiance at test wavelengths required to evoke a criterion response (~40 μV). The presence of multiple photoreceptor spectral classes was analyzed in the P. platensis eye using chromatic adaptation. Five individuals were light adapted with an orange longpass filter (OG590, Schott, Elmsford, NY, USA) for 1 h. Dim test flashes were given periodically over the adapting light until a stable response was observed, and then a spectral sensitivity curve was determined using the criterion response method with the adapting light continuing to illuminate the eye. Test flashes of dim light were given periodically throughout all experiments to ensure the eye remained in its initial state of adaptation.

Predictions of P. platensis photoreceptor spectral classes were modeled from dark- and chromatic-adapted spectral sensitivity curves after normalization to the wavelength of maximum sensitivity. Absorptance of photoreceptors was calculated using a rhodopsin template (Govardovskii et al. 2000). For these calculations, P. platensis rhabdom length (42.5 μm, N = 5) was measured from sections of resin-embedded eyes, and specific absorbance was assumed to be 0.008 μm−1 (Cronin and Forward 1988). The best-fit rhodopsin λmax was determined by minimizing the sum of squared residuals for the model fit to spectral sensitivity data. The effect of chromatic adaptation was assessed using a short:mid wavelength response ratio (Cohen and Frank 2007), constructed as the average log inverse irradiance producing a criterion response at short wavelengths (410, 430, 450 nm) divided by that at middle wavelengths (490, 510, 530 nm) for a given specimen.

Behavioral experiments

Platorchestia platensis was collected from under beach wrack on a southeast-facing beach of Boca Ciega Bay as described above during June–August 2009. Upon returning to the Galbraith Marine Science Laboratory, amphipods were sorted into either a light-proof container of seawater (23 °C; salinity = 34) for experiments in water or in light-proof vials in air for experiments without water, and allowed to dark adapt for a minimum of 1 h before testing. Except where indicated below, testing occurred between 10:00 and 17:00, during the light phase of the ambient light:dark cycle. Times of sunrise and sunset were approximately 6:40 and 20:30, respectively.

The laboratory apparatus and experimental procedures were based upon the classic study by Jander and Waterman (1960). The experimental apparatus (Fig. 1b) was located in a light-tight room and consisted of an arena (glass fingerbowl; 19 cm dia., 7 cm deep), with angles marked on the bottom exterior from 0 to 350°. The arena bowl was placed on an elevated wood/acrylic platform with a mirror underneath such that an observer could view an animal inside the arena from the bottom. Light from a 150-W voltage-regulated quartz-halogen lamp (DC-950, Dolan-Jenner) with fused-silica collimating lens assembly was filtered to 436 nm (10 nm fwhm interference filter, Edmund Optics, Barrington, NJ, USA). This wavelength was used based upon the results from the ERG studies. Light entered the arena bowl from directly above, casting a uniform circle of light downward that extended beyond the bowl walls; a wooden shield painted flat black blocked stray lamp light from reaching the arena. The arena bowl was covered by a polarizing filter consisting of three layers of wax paper and one layer of linear polarizing film (NT-45, Edmund Optics) held between pieces of clear acrylic sheet. With the polarizing film facing down, the arena bowl was illuminated with overhead linearly polarized light; with the wax paper facing down, it was illuminated with diffused, unpolarized light of equal quantal intensity to the polarized treatment (1.5 × 1012 photons cm−2 s−1). A solid white collar or a collar with alternating white and black (i.e., light and dark) quadrants each subtending 90° subsections of the collar blocked light from entering the arena bowl from the sides. Importantly, the white/black collar allowed for the differentiation of polarotaxis (directed movement in response to the plane of polarized light) from a possible scototaxis (directed movement in response to light/dark areas of the visual field). Overhead polarized light interacts with the solid white collar to produce a light and dark reflection pattern such that the areas parallel to the e-vector are relatively dim and the areas perpendicular to the e-vector are relatively bright (Jander and Waterman 1960). This is evident from wall reflectance measurements made within the arena under various collar and polarizer positions (Fig. 1c; Ocean Optics USB2000 spectrometer, 1,000 μm optical fiber, 436 nm-filtered illumination as described above). Thus, an amphipod moving parallel to the e-vector in an arena with a solid white collar could either be orienting to polarized light or showing a positive scototactic response. Accordingly, the procedure for separating polarotaxis from scototaxis is to measure orientation in the presence of polarized light and a collar with alternating light/dark quadrants.

A series of experiments were conducted separately in wet and dry arena conditions. For wet conditions, simulating an animal washed from its beach habitat into shallow water, the arena was filled with 1 cm of ambient seawater (23 °C, salinity = 34). A dark-adapted amphipod was gently pipetted from the holding container into the center of the arena and the filter was quickly placed over the bowl such that unpolarized light entered the arena. Upon release, an animal generally paused for several seconds and then undertook directed swimming from the center towards the perimeter of the arena bowl, where it hit the arena wall and began swimming around the bowl. For each individual, its angular position just before interacting with the arena wall relative to the center of the bowl was recorded at 10° resolution by an observer using the angular grid on the arena bottom. The animal was then removed and not used for further experiments. Some amphipods did not swim immediately, and any individual that did not assume directed swimming after 2 min was considered non-responsive and was removed. After 25 animals were tested, another 25 amphipods were tested with the filter rotated 90° and so forth until 100 individuals were tested. This rotation corrected for any inconsistencies within the arena light field and experimental apparatus. The second experiment repeated the procedure except that the filter was inverted and responses to polarized light were tested. Thus in both experimental series two data sets were collected, in which the filter directions were aligned at 0–180° and 90–270°; the latter data sets were rotated by 90° prior to circular statistical analysis (see below).

During the third experimental series, orientation of 75 amphipods was tested under unpolarized light with the white/black collar. The fourth experimental series also used the white/black collar but the orientation of 75 amphipods was tested with the e-vector aligned through the white quadrants. The fifth set of experiments repeated this procedure but the e-vector was aligned through the black quadrants. The sixth set of experiments repeated experiments 1–5 at night (20:30–22:30) with 50 amphipods tested per condition to test whether the night-active endogenous activity rhythm in P. platensis (J. Cohen, unpublished data) affected polarotactic/scototactic orientation behavior. The seventh set of experiments repeated daytime unpolarized and polarized treatments (experiments 1 and 2) using a solid white collar and either 532 or 589 nm light (10 nm fwhm interference filters, Edmund Optics) adjusted to 1.5 × 1012 photons cm−2 s−1 by neutral density filters to assess whether orientation responses to linearly polarized light were wavelength-dependent. In each condition, 75 amphipods were tested.

Experiments 1–5 were repeated during the day under dry conditions (i.e., no water in the arena bowl), simulating orientation on their home beach. An amphipod was released from a vial into the center of the arena and the filter replaced. The azimuth direction of orientation (jump or walk) toward the perimeter of the arena was recorded at 10° resolution, and the animal was removed. In each experiment, 50–51 animals were tested.

Both e-vector and white/black quadrants were symmetric cues in the laboratory arena. For example, if the e-vector was aligned in the 0–180° axis, an amphipod orienting perpendicular to the e-vector could move toward 90° or 270°. Similarly with the white/black collar (experiment 3) an amphipod could show positive scototaxis by moving in opposite directions toward opposing black quadrants. Accordingly, amphipod orientation in response to these cues was bimodal rather than unimodal. This is a routine condition in polarotaxis and scototaxis experiments (e.g., Goddard and Forward 1991; Freake 1999) and was accounted for by transforming bimodal data into unimodal distributions by doubling the angles (Zar 1999). These unimodal distributions were then tested for uniformity using Rayleigh tests (α = 0.05; Zar 1999) and analyzed for the mean vector angle and length in Oriana 2.02e. Rayleigh P value and mean vector parameters were used as a semi-quantitative test for the effect of conflicting polarotactic and scototactic cues (experiments 3–5). This effect was also assessed by comparing the number of amphipods orienting towards white and black quadrants in unpolarized conditions with those in each polarized condition using Chi square contingency tests with Yates correction in Sigma Plot 12.2.



Dark-adapted spectral sensitivity curves for eight individual P. platensis displayed two distinct functional forms: a short-wavelength (i.e., violet) dominant form with a shoulder at middle wavelengths (N = 2) indicative of distal electrode placement nearer the R5 cell, and a middle-wavelength (i.e., blue/green) dominant form with a shoulder at short wavelengths (N = 6) indicative of proximal electrode placement nearer the R1–4 cells (Fig. 2a). The absorptance of the short-wavelength peak (350–450 nm data only) was best fit by a 431-nm rhodopsin λmax, while the middle-wavelength peak (450–650 nm data only) was best fit by a 520 nm rhodopsin λmax (residual sum of squares = 0.0375 and 0.0772, respectively) (Fig. 2a). Chromatic adaptation to >590 nm wavelengths (Schott OG590 longpass filter) resulted in a disproportionate decrease in sensitivity at middle wavelengths relative to short wavelengths (Fig. 2b). The short:mid response ratio for dark-adapted specimens (1.02 ± 0.16 SE) was significantly greater than for those tested upon >590 nm chromatic adaptation (0.97 ± 0.01 SE) (P = 0.004, Mann–Whitney rank sum test). The short-wavelength peak (350–450 nm) resulting from >590 nm chromatic adaptation was best fit by a 429-nm rhodopsin λmax (residual sum of squares = 0.0046).
Fig. 2

Visual spectral sensitivity in P. platensis. a Spectral sensitivity determined from ERGs in violet dominant (closed triangles, N = 2) and blue/green dominant (open circles, N = 6) preparations. Data are normalized to the wavelength of maximum sensitivity and plotted as mean ± SE as a function of wavelength. Best-fit rhodopsin absorptance spectra (solid lines; λmax = 431 and 520 nm) are plotted for each spectral sensitivity curve. b Chromatic-adapted spectral sensitivity is plotted as the log inverse irradiance (Log 1/photons cm−2 s−1) for a single representative blue/green dominant specimen tested first upon dark adaptation (closed circles), and then after light adaptation to an orange OG590 longpass filter (open circles)


For all wavelengths (436, 532, and 589 nm) tested in water during the day using a solid white-collared arena, P. platensis displayed a uniform orientation response under unpolarized light (P = 0.367, 0.093, 0.641, respectively, Rayleigh tests with unimodal doubled data) (Fig. 3a; 532 and 589 nm data not shown). Under 436 nm polarized light, amphipods showed strong orientation along a 2°–182° axis parallel to the polarization e-vector (P < 0.001, Rayleigh test, mean vector length r = 0.492) (Fig. 3b). Under 532 nm polarized light, the orientation response was weaker but remained significant and along a 10°–190° axis parallel to the polarization e-vector (P = 0.004, Rayleigh test, r = 0.274, data not shown). However, when tested under 589 nm light P. platensis orientation was uniform around the arena (P = 0.591, Rayleigh test, data not shown).
Fig. 3

Orientation of P. platensis in a wet arena during the day. Experiments were conducted under 436 nm light. Bearings of individual amphipods (closed circles) under different experimental conditions are plotted as the untransformed bimodal distributions. Sample size of responding individuals (N; total of 100 tested with white collar, 75 tested with white/black collar) and P value for a Rayleigh Uniformity Test on transformed unimodal data are given for each plot. Mean vectors are provided for non-uniform distributions (P < 0.05) as a double-headed arrow symmetric about the origin. Plots are either unpolarized controls (a, c) or polarized treatments (b, d, e). The border of each plot shows the arena collar condition (white or black) for that treatment, while paired triangles outside the border show the alignment of the e-vector with respect to the arena collar for a given treatment

Further experiments with 436 nm light were conducted in water during the day with a white/black collar surrounding the arena to test whether the observed orientation to polarized light was due to polarotaxis or to a scototactic artifact in the arena when polarized light was present. Amphipods displayed a strong positive scototactic response (i.e., orientation toward the black quadrants) when tested under unpolarized light (P < 0.001, Rayleigh test r = 0.606) (Fig. 3c). In trials where polarized light was tested with the e-vector parallel to the white quadrants (i.e., conflicting polarotactic and positive scototactic cues), mean orientation remained toward the black quadrants but the distribution was more scattered (P = 0.003, Rayleigh test r = 0.290) (Fig. 3d). An increased number of amphipods oriented along the e-vector toward the white quadrants under this condition relative to unpolarized controls, but not statistically so (Table 1). When the e-vector was rotated 90° to pass through the black quadrants (i.e., additive polarotactic and positive scototactic cues), positive scototaxis was apparent and mean vector length increased (P < 0.001, Rayleigh test, r = 0.497) (Fig. 3e). Amphipod preference for black quadrants in this condition did not differ from unpolarized controls (Table 1).
Table 1

Comparison of P. platensis orientation to white and black quadrants between unpolarized and polarized treatments for each arena condition

Arena condition




Chi square (χ2)


Wet, day




Wet, day

E-vector white





Wet, day

E-vector black





Wet, night




Wet, night

E-vector white





Wet, night

E-vector black





Dry, day




Dry, day

E-vector white





Dry, day

E-vector black





aThe number of amphipods that oriented towards each quadrant type

To test for a diel component in the observed polarotactic/scototactic behavior in these nocturnal animals, orientation was assessed in water at night using the same arena conditions as for daytime experiments. Orientation of P. platensis under either unpolarized (Fig. 4a) or polarized 436 nm light (Fig. 4b) with a solid white collar resulted in a uniform distribution (P = 0.078, 0.112, respectively, Rayleigh tests). Positive scototaxis was evident under unpolarized light with a white/black collar (P < 0.001, Rayleigh test, r = 0.475) (Fig. 4c), as well as when the e-vector was presented as either a conflicting cue through white quadrants (P < 0.001, Rayleigh test, r = 0.384) (Fig. 4d), or an additive cue through black quadrants (P < 0.001, Rayleigh test, r = 0.520) (Fig. 4e). Mean vector length and Chi square analysis (Table 1) suggest a marginal, non-significant decrease in positive scototaxis with conflicting cues.
Fig. 4

Orientation of P. platensis in a wet arena during the night. Experiments were conducted under 436 nm light. Bearings of individual amphipods (closed circles) under different experimental conditions are plotted as the untransformed bimodal distributions. Sample size of responding individuals (N, total of 50 tested) and P value for a Rayleigh Uniformity Test on transformed unimodal data are given for each plot. Mean vectors are provided for non-uniform distributions (P < 0.05) as a double-headed arrow symmetric about the origin. Plots are either unpolarized controls (a, c) or polarized treatments (b, d, e). The border of each plot shows the arena collar condition (white or black) for that treatment, while paired triangles outside the border show the alignment of the e-vector with respect to the arena collar for a given treatment

Orientation of P. platensis in a dry arena during the day with a solid white collar under unpolarized 436 nm light resulted in a uniform distribution (P = 0.676, Rayleigh test) (Fig. 5a). When tested in these same conditions under polarized light, significant orientation along a 1°–181° axis parallel to the polarization e-vector was observed (P < 0.001, Rayleigh test, r = 0.463) (Fig. 5b). With the arena surrounded by a white/black collar under unpolarized light, P. platensis displayed significant negative scototaxis (i.e., orientation towards the white quadrants) (P = 0.013, Rayleigh test, r = 0.293) (Fig. 5c). Under polarized light with the e-vector aligned to the white quadrants, in this case presenting additive polarotactic and scototactic cues, orientation toward these quadrants paralleled the e-vector (P < 0.001, Rayleigh test, r = 0.381) (Fig. 5d). Orientation in this condition was stronger than with scototaxis alone as suggested by mean vector length, but white quadrants were equally favored in this and unpolarized conditions (Table 1). However, when the e-vector was aligned with the black quadrants, presenting conflicting polarotactic and scototactic cues, orientation was uniform around the arena (P = 0.713, Rayleigh test) (Fig. 5e) and orientation toward black quadrants was significantly greater than in unpolarized controls (Table 1).
Fig. 5

Orientation of P. platensis in a dry arena during the day. Experiments were conducted under 436 nm light. Bearings of individual amphipods (closed circles) under different experimental conditions are plotted as the untransformed bimodal distributions. Sample size of responding individuals (N; total of 50–51 tested) and P value for a Rayleigh Uniformity Test on transformed unimodal data are given for each plot. Mean vectors are provided for non-uniform distributions (P < 0.05) as a double-headed arrow symmetric about the origin. Plots are either unpolarized controls (a, c) or polarized treatments (b, d, e). The border of each plot shows the arena collar condition (white or black) for that treatment, while paired triangles outside the border show the alignment of the e-vector with respect to the arena collar for a given treatment


The use of celestial and landscape visual cues in orientation of talitrid amphipods is well established (reviewed by Ugolini et al. 2002; Scapini 2006; Warrant and Dacke 2010). It is increasingly recognized that these behaviors are facilitated by light intensity gradients (Ugolini et al. 2009, 2012), in addition to the spectral quality of light (Ugolini et al. 1996; Forward et al. 2009a). Support for the latter has also come from the observation that several talitrid species possess a dichromatic visual system with sensitivity maxima around 430–450 nm (R5 cells) and 520–525 nm (R1–4 cells) (Mezzetti and Scapini 1995; Ugolini et al. 1996, 2006, 2010; Galanti et al. 2007; Cohen et al. 2010). Spectral sensitivity in Talitrus saltator determined by ERG recording found maxima at ~450 and 525 nm (Ugolini et al. 2006) with subsequent experiments reporting peak sensitivity at 380 and 510 nm (Ugolini et al. 2010), while maxima in Talorchestia longicornis were at 430 and 522 nm depending on electrode placement relative to the R5 cell (Cohen et al. 2010). Visual spectral sensitivity in P. platensis shows maxima at 431 and 520 nm, with chromatic adaptation to >590 nm light selectively reducing sensitivity at longer wavelengths. These findings provide evidence for a dichromatic visual system in P. platensis as observed for other talitrids, with the short-wavelength R5 receptor class more similar in spectral maximum to T. longicornis than to Talitrus saltator. It is unclear if this difference in short-wavelength spectral sensitivity among species is an artifact of differences in electrophysiological techniques employed, if it reflects the closer evolutionary relatedness between Platorchestia and Talorchestia (Conceição et al. 1998), and/or if it is functionally significant.

In the talitrid T. longicornis, sensitivity to short-wavelength blue light (~430 nm) is required for celestial sun-compass orientation (Forward et al. 2009a), whereas green wavelengths (~520 nm) are needed to entrain the endogenous activity rhythm (Forward et al. 2009b). Similarly, ultraviolet/short-wavelength blue light is needed for celestial orientation in Talitrus saltator (Ugolini et al. 1993), and it responds phototactically to light in the 350–436 nm range (Galanti et al. 2007; Ugolini et al. 2010). In the present study, movement of P. platensis in the laboratory arena was more directed under overhead polarized light at shorter (436 nm) rather than longer (532 or 589 nm) wavelengths of equal quantal intensity, as evidenced by the greatest mean vector length for the amphipod distribution under 436 nm light. This is consistent with the involvement of the shorter photoreceptor spectral class in orientation to celestial polarized light. Within ommatidia of talitrid amphipod eyes, the four blue/green sensitive retinular cells (R1–R4) are arranged in pairs with orthogonally oriented microvilli, while the violet-sensitive R5 cell is situated distally and projects microvilli at 45° to both sets of receptors in the main rhabdom (Hallberg et al. 1980; Ugolini et al. 2010). Detection of linearly polarized light in most crustaceans and many other taxa typically involves differential e-vector discrimination by rhodopsin molecules of the same spectral class aligned along orthogonal microvilli in two pairs of retinular cells of the main rhabdom (reviewed in Wehner and Labhart 2006). Polarization sensitivity observed behaviorally in P. platensis may similarly be conferred by such a process in the blue/green-sensitive R1–4 retinular cells, but the involvement of this receptor class conflicts with the present data suggesting behavioral orientation to e-vector stimuli was stronger at 436 nm than at 532 nm. Given the behavioral data, it is possible that the violet-sensitive R5 cells are involved in polarized light detection; indeed, short-wavelength (ultra-violet) visual sensitivity has previously been speculated to be part of a skylight polarization detection system in talitrids (Ugolini et al. 2002), as it functions in some insects (Labhart and Meyer 1999). However, given that R5 cells are not paired with orthogonal microvilli in individual ommatidia, it is unclear at present what mechanism would be used to deduce e-vector direction. Intracellular recordings from R1–5 cells are needed to differentiate which cells are responsible for polarization sensitivity in talitrids and to determine whether this sensitivity is evenly distributed across the eye or localized to a specific eye region (e.g., Labhart 1986; Stowasser and Buschbeck 2012).

Visual cues have been shown to aid talitrid orientation toward shore when in water (terHorst 2012), and the orientation responses of P. platensis to brightness cues in their visual surround demonstrated in the present study would facilitate return of individuals to shore once submerged. When P. platensis was tested in a water-filled arena during either the day or night, simulating animals having been washed from the beach into the water, they exhibited movement toward dark black quadrants (positive scototaxis). In shallow estuarine/marine habitats, this response would orient them towards the shore, which appears as a relatively dark region in the horizontal visual field due to submerged vegetation landward as opposed to open water seaward. Likewise, when looking up within Snell’s window the open ocean would be light, whereas shoreward vegetation would be dark. Hence, moving shoreward in response to the light pattern in Snell’s window would be positive scototaxis and would facilitate movement onto the home beach when washed into the water. Such positive scototactic responses are common among shore-dwelling crustaceans (e.g., Langdon and Herrnkind 1985) but may reverse in direction within an individual depending on habitat, time of day, ontogenetic stage, and size/shape of the dark stimulus (Goddard and Forward 1991; Díaz et al. 1995; Scapini 1997). No such diel change in scototaxis occurred in P. platensis during tests in water, supporting a constant role for this response to brightness cues throughout the diel cycle, such as in shore-seeking behavior that would direct these animals to their beach habitat.

When tested in a dry bowl during the day simulating an animal on the beach, however, P. platensis did show a change in its scototactic response relative to the daytime response in water with amphipods moving instead toward the light white quadrants (negative scototaxis). Experiments with P. platensis were not repeated at night in a dry bowl, so it is unknown if a diel change in scototaxis occurs. The magnitude and direction of scototactic responses are variable among talitrids, but if present, the dominant sign during the day in dry situations tends to be positive scototaxis with movement toward dark areas. This has been interpreted as part of a burrow-seeking behavior during the inactive phase in these nocturnal organisms (Edwards and Naylor 1987). When negative scototaxis is present in talitrids, it typically is displayed at night and is thought to help in burrow emergence and/or movement away from dunes and towards brighter sandy foraging areas (Ugolini et al. 1986; Mezzetti et al. 1994, 1997; Nardi et al. 2000). Negative scototaxis on the sand during the day in P. platensis contradicts the general trends for talitrids, but a possible explanation is that this species may be displaying a predator-avoidance response. If disturbed during the day, P. platensis may actively avoid dark areas as they are interpreted as predators. Alternatively, negative scototaxis away from dark shoreward vegetation and toward a lighter beach or water area would move the amphipod seaward away from terrestrial predators. Similar involvement of scototaxis in predator-avoidance was suggested for the talitrid Atlantorchestoidea brasiliensis (Fanini et al. 2009), as well as the mangrove crab Aratus pisonii (Díaz et al. 1995).

Celestial skylight polarization has been suggested as a reliable cue for orientation in talitrids since early work by Pardi and Papi (1953), but studies have yet to convincingly demonstrate this ability. The present behavioral experiments with P. platensis in all conditions, particularly those in a dry bowl, demonstrate this species is capable of polarotaxis. In the presence of a white collar, polarized light interacts with the walls of the arena such that reflectance is low parallel to the e-vector and high perpendicular to the e-vector, which itself can present a scototactic cue (Fig. 1c) (Jander and Waterman 1960). In dry conditions, P. platensis showed strong negative scototaxis moving to the light quadrants under unpolarized light with a white/black collar. Yet with a solid-white collar it orients parallel to the polarization e-vector, which extends between the relatively dark quadrants of the bowl created by brighter internal reflection perpendicular to the plane of polarization. Thus, orienting parallel to the e-vector with the solid white collar is a polarotactic response, not positive scototaxis. Polarized light experiments in the presence of the white/black collar provide further evidence that polarotaxis and scototaxis are separate responses. When the e-vector was aligned with the white quadrants P. platensis oriented strongly in this direction. However, orientation was statistically uniform under polarized light when the e-vector direction opposed the preferred negative scototactic direction and was aligned through the dark quadrants. In the latter case, orientation along the e-vector towards dark quadrants occurred in some individuals while movement towards light quadrants occurred in others. The polarotactic response was weaker when amphipods were tested underwater, but the same general trend can be observed when mean vector lengths are compared among experiments with additive and conflicting polarotactic and scototactic cues.

This laboratory assay presented a celestial e-vector cue to amphipods in the test arena irrespective of whether amphipods were tested in water or air. This would be equivalent to their viewing a patch of sky while on a beach or viewing the same patch of sky within Snell’s window when submerged in very shallow water (e.g., Horváth and Varjú 1995; Cronin and Shashar 2001). The latter case was confirmed by Goddard and Forward (1991) for grass shrimp Palaemonetes vulgaris when tested in a comparable laboratory apparatus. These subtidal shrimp oriented uniformly under unpolarized light, but parallel to the e-vector of polarized light in a white visual surround. When these authors tested P. vulgaris in a white/black visual surround as in the present study, an open water shrimp population showed significant negative scototaxis under unpolarized conditions, but were uniformly distributed when the e-vector was oriented through the dark quadrants thus conflicting scototactic and polarotactic cues. Subsequent field experiments confirmed that P. vulgaris was capable of using celestial polarized light cues to orient relative to the sun’s position for sun-compass orientation. A similar use of celestial polarized light in locating an obscured sun and/or moon for celestial compass orientation is likely in P. platensis, whether viewing the sky directly when on land or through the refractive cone of Snell’s window when submerged in shallow water. It is well established that talitrids can use both the sun and moon to guide their orientation on beaches (Ugolini 2003). Orientation by celestial polarized light would provide these animals with a cue in the day time or night time sky when celestial bodies themselves may be obscured by clouds, enabling them to direct their movements toward the wrack line located in a band of moist sand between the waterline and desiccating conditions on the upper beach, or perhaps directing swimming movements toward the beach when submerged in shallow water. Subsequent work characterizing P. platensis locomotory responses relative to ambient polarized skylight (e.g., Goddard and Forward 1991) or to a rotating e-vector stimulus (e.g., Brunner and Labhart 1987) would provide further information on how these animals are using polarized light in celestial orientation throughout the day and night.

Given that this species enters the active phase of its endogenous activity rhythm at twilight, it is possible that orientation to celestial polarized light at dusk is particularly useful in directing amphipod movement to their preferred zone of moist sand. At twilight, the sky is highly polarized in a wide band extending along the North–South axis, with the e-vector also directed along this axis; this pattern is stable throughout twilight and occurs at both sunset and sunrise (Cronin et al. 2006). As the collection beach for P. platensis in this study faces the southeast, this celestial band of polarized light would generally be perpendicular to the waterline. Thus, orienting parallel to the celestial e-vector as occurred in the behavioral assay would direct amphipod movement between the waterline and the upper-beach, consistent with how they orient in response to the sun/moon (e.g., Ugolini 2003). Talitrid movement in response to celestial cues depends on the land–sea axis of their home beach (e.g., Forward et al. 2009a), so it would be expected that populations of P. platensis on beaches facing other directions would orient differently with respect to the e-vector, but their resulting movement would be perpendicular to the waterline. Given that the eye design and visual capabilities of talitrid species appear similar, and many also have nocturnal activity rhythms and forage between the waterline and the upper-beach, it is possible that use of celestial polarized light, particularly at twilight, is a common orientation mechanism in these organisms.


Rebecca Barkdoll, Caleb Bartles, Joseph Conrad, and Kelly Vasbinder provided valuable assistance with experiments. We thank Dr. Richard Forward, Lillian McCormick and Corie Charpentier for comments on the manuscript. This work was funded in part by the Eckerd College Natural Sciences Summer Research Program. All experiments complied with regulations for use of invertebrates in research at Eckerd College where experiments were conducted.

Copyright information

© Springer-Verlag Berlin Heidelberg 2013