Animal Cognition

, Volume 15, Issue 4, pp 591–596

Cuttlefish rely on both polarized light and landmarks for orientation

Authors

  • Lelia Cartron
    • Groupe Mémoire et Plasticité comportementaleUniversité de Caen Basse-Normandie
  • Anne-Sophie Darmaillacq
    • Groupe Mémoire et Plasticité comportementaleUniversité de Caen Basse-Normandie
  • Christelle Jozet-Alves
    • Groupe Mémoire et Plasticité comportementaleUniversité de Caen Basse-Normandie
  • Nadav Shashar
    • Department of Life SciencesBen Gurion University of the Negev
    • Groupe Mémoire et Plasticité comportementaleUniversité de Caen Basse-Normandie
Original Paper

DOI: 10.1007/s10071-012-0487-9

Cite this article as:
Cartron, L., Darmaillacq, A., Jozet-Alves, C. et al. Anim Cogn (2012) 15: 591. doi:10.1007/s10071-012-0487-9

Abstract

Cuttlefish are sensitive to linear polarization of light, a sensitivity that they use in predation and possibly in intraspecific communication. It has also been shown that cuttlefish are able to solve a maze using visual landmarks. In this study, cuttlefish were trained to solve a Y-maze with the e-vector of a polarized light and landmarks as redundant spatial information. The results showed that cuttlefish can use the e-vector orientation and landmarks in parallel to orient and that they are able to use either type of cue when the other one is missing. When they faced conflicting spatial information in the experimental apparatus, the majority of cuttlefish followed the e-vector rather than landmarks. Differences in response latencies in the different conditions of testing (training with both types of cue, tests with single cue or with conflicting information) were observed and discussed in terms of decision making. The ability to use near field and far field information may enable animals to interpret the partially occluded underwater light field.

Keywords

NavigationPolarized lightLandmarksCephalopods

Introduction

Navigation can be defined as the process which enables a course or path from one place to another to be identified and maintained (Gallistel 1990). Spatial orientation allows animals to reach beneficial places containing food, conspecifics or shelters, and to avoid risky places with predators. Hence, the ability to move from one place to another in an efficient way is an important factor for survival. For this purpose, animals can use a variety of spatial information such as beacons, landmarks, overall spatial geometry, sun or magnetic compasses, celestial polarization and self-generated cues for dead reckoning (reviewed in Rozhok 2008; Shettleworth 2010; see also Kraft et al. 2011 for honeybee navigation using polarization). The way different types of spatial information interact has been extensively studied addressing spatial tasks (Healy 1998 and references therein). Historically, the theory deriving from Pavlovian conditioning was that multiple redundant spatial cues compete for associative learning so one cue can overshadow or block learning about another one (Diez-Chamizo et al. 1985; March et al. 1992; Redhead et al. 1997; Pearce et al. 2001). However, it seems likely that the importance of spatial learning for survival favors the simultaneous acquisition of redundant types of spatial information because it allows secondary cues to be used as backup when primary cues are unavailable (Shettleworth 2010). An increasing number of experiments support this idea in species ranging from insects (Wehner et al. 1996; Steck et al. 2011) to rodents (Lavenex and Schenk 1996; Rossier et al. 2000; Gibson and Shettleworth 2003, 2005).

Spatial orientation in aquatic environments seems to be particularly challenging since visual cues can undergo major changes due to water turbidity and depth, and olfactory cues are highly fugitive due to currents or flow directions. Hence, it would be beneficial to learn and memorize different types of information and to orient according to the cues available under the specific environmental conditions. Many species of crustacean, fish, and sea turtles can rely on a large variety of spatial information to orient such as geomagnetic field, celestial and polarized light cues, ocean currents, olfactory, and other visual cues (Healy 1998; Odling-Smee and Braithwaite 2003; Parkyn et al. 2003; Horvath and Varju 2004; Lohmann et al. 2008; Shettleworth 2010). However, far fewer studies have investigated how different kinds of spatial information interact in a controlled spatial task.

Cuttlefish (Mollusca: Cephalopoda) are colorblind (Mäthger et al. 2006) but are sensitive to the e-vector orientation of linearly polarized light, a sensitivity which arises from the orthogonal distribution of the microvilli of neighboring photoreceptor cells in their retina (Shashar et al. 1996). In the cuttlefish, polarization sensitivity is involved in predation (Shashar et al. 1998, 2000) and possibly in intraspecific communication (Shashar et al. 1996; Boal et al. 2004). In a range of animals, polarization sensitivity is involved in orientation and may well serve for underwater navigation (Sabbah et al. 2005; Lerner et al. 2011 and references therein). Cuttlefish are known for their remarkable learning abilities (Messenger 1973; Darmaillacq et al. 2004, 2006), notably in spatial learning where they have been shown to solve a maze using landmarks (Alves et al. 2007; Jozet-Alves et al. 2008). However, their ability to use an e-vector orientation for a spatial task has never been investigated.

The first aim of the study was to determine whether cuttlefish could use the e-vector of polarization to find a shelter. The second aim was to test whether cuttlefish were able to learn to find a shelter using e-vector orientation and landmarks in parallel and then to rely on one type of cue when the other one became unavailable. Last, we addressed which type of cue was preferentially followed when cuttlefish faced conflicting information in the experimental apparatus.

Materials and methods

Subjects

Eleven cuttlefish, Sepia officinalis, were tested. They were reared from hatching to 7 months in laboratory conditions with running seawater at 15 ± 1 °C at the Centre de Recherches en Environnement Côtier (CREC, Luc-sur-Mer, France). They were first housed in groups and then, 1 week before behavioral experiments began, housed in individual tanks. The animals were provided with enriched (or “semi natural”) habitats within the tank following Dickel et al. (2000) who showed that an enriched environment has a positive effect on cuttlefish growth rates and the acquisition and the retention of information. They were fed daily with live shrimps (Crangon crangon) and crabs (Carcinus maenas) of suitable size.

Apparatus

Experiments were conducted in a Y-maze (Fig. 1) made of spectrally flat white PVC to reduce potential brightness artifacts, which could arise from the interaction of the polarized light source and the inside walls of the tank (Jander and Waterman 1960). One arm (30 cm long × 10 cm wide × 15 cm high) was used as a start box closed by a sliding glass door. The other two arms (goal arms) each ended with a shelter (20 cm long × 20 cm wide × 15 cm high) which were closed by sliding glass doors. The shelters were made of black PVC and were covered by an opaque top, so as to be attractive to the cuttlefish. Two black-and-white PVC rectangles (4 cm wide × 8 cm high), a striped rectangle (horizontal stripes 2 cm wide), and a spotted rectangle (spots 1 cm in diameter) were used as landmarks. They were placed 5 cm above the water surface at the junction of the Y (Fig. 1; Alves et al. 2007; Jozet-Alves et al. 2008). They could be swapped in position or removed. Black opaque curtains were placed around and above the maze to prevent cuttlefish from seeing external cues and light from the laboratory. The apparatus was illuminated by a 300-W halogen lamp located 1 m above the centre of the maze. The lamp was enclosed in a black baffle (25 cm long × 25 cm wide × 35 cm high), the bottom of which was a layer of commercial wax paper above a linear polarizing filter (American Polarizers, Inc. AP38-030T). The wax paper was used as a light diffuser to remove potential inherent polarization in the optical system. The filter polarized the light along a pre-set e-vector axis. This filter could be rotated to orient the e-vector parallel to one or the other arm (and perpendicular to the other) and could be removed to illuminate the maze with a nonpolarized light. There was no difference in light intensity between the two goal arms regardless of the e-vector orientation (300 lux measured with a luxmeter, Hanna Instruments). Water flow was provided between trials to reduce water heating and remove olfactory cues.
https://static-content.springer.com/image/art%3A10.1007%2Fs10071-012-0487-9/MediaObjects/10071_2012_487_Fig1_HTML.gif
Fig. 1

a Y-maze apparatus. e-vector: illumination of the maze from above with a light polarized in a single e-vector axis (dashed line with arrow); landmarks: black-and-white spotted or striped rectangles; St start box; dashed line start box sliding glass door; Sh: shelter (=reward); dotted line shelter sliding glass door. b Chronology of the experiments. Step 1: training with both the e-vector (light shaded) and the landmarks (striped); it consists in one session (five trials) per day until cuttlefish reach the learning criterion. Step 2: single cue tests (two sessions) alternating with two sessions with the two types of cues. Half of the cuttlefish were presented with the e-vector single cue test first and the other half with the landmarks single cue test first. Step 3: probe trial with conflicting information between the e-vector orientation and the landmarks (dark shaded)

Procedure

The procedure consisted of three steps (Fig. 1). First, cuttlefish were trained to find a shelter (the reward; see Alves et al. 2007). A combination of a landmark and an e-vector orientation was randomly chosen for each cuttlefish to indicate the goal arm (e.g., the black-and-white spotted rectangle and the e-vector perpendicular to the longitudinal axis of the goal arm). The sliding glass door at the entrance of the shelter was opened only if the cuttlefish made the correct choice. The goal arm (i.e., correct choice) for each trial was randomly chosen, and the cues were changed accordingly between trials (rotation of the linear polarizer and permutation of the right/left position of the landmarks). Cuttlefish were given five trials per session, with one training session per day, until they reached the learning criterion (i.e., eight correct choices out of 10, significant choice at 10 % level of significance with a binomial test; Jozet-Alves et al. 2008).

In step 2, cuttlefish were given two single cue tests, each composed of two consecutive sessions (Fig. 1). In the e-vector single cue test, cuttlefish had to find the shelter using only the e-vector orientation, whereas in the landmarks single cue test, cuttlefish had to find the shelter using only the landmarks. The single cue tests were conducted to determine whether the cuttlefish had learned to use the e-vector orientation, the landmarks, or both, to find the shelter. The order of single cue tests was randomly chosen for each cuttlefish. Each single cue test was followed by two training sessions with both types of cue.

In step 3, cuttlefish were given one probe trial with conflicting information: the e-vector orientation indicating one arm as rewarded while the landmarks indicated the other arm as rewarded.

Statistical analyses

Data were analyzed using StatXact 7 (Cytel Studio software). All analyses used a significance threshold of α = 0.05 and the tests were two-tailed. In step 1, the percentage success between the first and the last session was compared with a permutation exact test for paired samples. In steps 2 and 3, the mean latencies to choose the goal arm were compared using an exact permutation test for paired samples. In step 3, the choice for the e-vector orientation or the landmarks was compared using the exact Chi-square test.

Results

In step 1, all cuttlefish (n = 11) reached the learning criterion in 6–12 sessions (i.e., 30–60 trials), with a mean ± SEM of 8 ± 1 sessions. Percentage success (mean ± SEM) was significantly higher during the last session for each cuttlefish (91 ± 3 %) than during the first session (35 ± 7 %; exact permutation test for paired samples: n = 11, P = 0.002; Fig. 2). It made no difference which e-vector orientation and which landmark signaled the goal arm.
https://static-content.springer.com/image/art%3A10.1007%2Fs10071-012-0487-9/MediaObjects/10071_2012_487_Fig2_HTML.gif
Fig. 2

Mean percentages of correct choices per session (±SEM) during the first and last sessions of training and during the single cue tests. Asterisks indicate significant differences **P < 0.01; exact permutation tests for paired samples

In step 2, one cuttlefish died after the e-vector single cue test due to unrelated reasons so that the final sample size for the landmarks single cue test was 10. All cuttlefish succeeded in choosing the goal arm with only the e-vector orientation (n = 11, mean percentage success ± SEM = 87 ± 2 %) and with only the landmarks (n = 10; 88 ± 2 %; Fig. 2). For the cuttlefish that were given the e-vector single cue test first (n = 5), and those that were given the landmarks single cue test first (n = 5), the latency to enter the goal arm increased, though not significantly, in the first trial of the single cue test compared with the first trial of the last training session (respectively: mean latency ± SEM = 67 ± 9 s vs. 83 ± 7 s and 55 ± 15 s vs. 82 ± 11 s; exact permutation test for paired samples: P = 0.062 and P = 0.062; Fig. 3).
https://static-content.springer.com/image/art%3A10.1007%2Fs10071-012-0487-9/MediaObjects/10071_2012_487_Fig3_HTML.gif
Fig. 3

Mean latencies (±SEM) to enter the goal arm. ae-vector single cue test first (n = 5): comparison between the first trial of the last training session and the first trial of each single cue test. b Landmarks single cue test first (n = 5): comparison between the first trial of the last training session and the first trial of each single cue test. c Comparison between the first trial of the last training session and the probe trial (n = 10). Asterisks indicate significant differences ***P < 0.001; exact permutation tests for paired samples

In step 3, eight out of 10 cuttlefish followed the e-vector orientation and two followed the landmarks. They tended to follow the e-vector rather than landmarks (exact Chi-square test, χ² = 3.6, P = 0.058). The latency to enter the goal arm (mean ± SEM) increased significantly between the first trial of the training session (with both types of cue, performed just before the probe trial; 52.1 ± 7.8 s) and the probe trial (126.4 ± 15.3 s; exact permutation test for paired samples P = 0.007; Fig. 3). Furthermore, this latency (mean ± SEM) increase (196.7 ± 57.1 %) was significantly greater than the increase between the training and the first single cue test (55.6 ± 19.3 %; exact permutation test for paired samples P = 0.047).

Discussion

The aim of this study was threefold: (1) to determine whether cuttlefish could use the e-vector of polarized light to find a shelter; (2) to test whether cuttlefish were able to learn to find a shelter using e-vector orientation and landmarks in parallel and then to rely on one type of cue when the other one became unavailable; (3) to test which type of cue was preferentially followed when cuttlefish faced conflicting information in the experimental apparatus. Here, we showed that cuttlefish learned to solve a maze with two distinct visual cues (step 1) but were able to solve it when only one type of cue was available (step 2). This demonstrates that they are able to learn to use landmarks, as already shown by Alves et al. (2007); this also shows for the first time that cuttlefish can use an e-vector orientation to find a shelter. To date, it was only known that polarization sensitivity is used by cuttlefish for predation (Shashar et al. 2000) and possibly in communication (Shashar et al. 1996; Boal et al. 2004). Squid (Jander et al. 1963) exhibit spontaneous preferential swimming direction relative to the e-vector orientation, like some species of crustaceans (Horvath and Varju 2004; Sabbah et al. 2005; Lerner et al. 2011). Cuttlefish learn to use landmarks and the e-vector orientation in parallel. In the water, the ability to memorize redundant spatial information at the same time when finding a goal location should be advantageous since landmarks may not be visible from large distances due to water turbidity or bottom topography. They should therefore be considered as near field visual cues. On the other hand, polarization information is based on the sun’s position in the sky (far field). Thus, it may be altered, for example, by waves or clouds (Horvath and Varju 2004; Sabbah et al. 2005; Shashar et al. 2011). Luschi et al. (1997) demonstrated that the crab Dotilla wichmanni mainly relies on skylight polarization to find food patches. However, when this polarization information is not available (e.g., under a cloudy sky), the crab is able to switch to visual landmarks which can be more advantageous under those environmental conditions.

During the probe trial, we observed that the latency to choose an arm was significantly greater under conflicting information than when the two types of cue were congruent. Additionally, the absence of one of the two types of cue in the maze induced a slower response which confirms that cuttlefish used both landmarks and the e-vector orientation in step 1. The response was even slower when visual cues were in conflict (step 3), suggesting that cuttlefish responded to the incongruence between the cues. The change of the response latency is what would be predicted if more complex cognitive processes were required before the cuttlefish made a decision, though there may be other explanations. Nippak and Milgram (2005) suggested that the response latency in dogs was longer for more complex visual discrimination tasks compared with simple tasks and reflected the cognitive strategy used by the animal. We also found that when cuttlefish faced conflicting information in the experimental apparatus (step 3), they tended to follow the e-vector orientation rather than the landmarks. The degree to which one cue rather than another controls behavior depends on their salience and their validity, that is, on both the strength and the information content of the signals. Field measurements are needed to examine the relative stability of different cues available to cuttlefish in the wild. Although little is known about cuttlefish behavior in their natural habitat, we demonstrate here for the first time that cuttlefish can use both polarized light and landmarks to find and return to specific sites, such as patches of food or safe places.

Acknowledgments

We thank the staff of the CREC for their technical assistance. This research was supported by a grant from the Ministère de l’Enseignement Supérieur et de la Recherche to L.C.

Copyright information

© Springer-Verlag 2012