Polarization sensitivity is the ability to distinguish between different directions of oscillation (e-vectors) of light. Many species of fish were determined to possess polarization sensitivity and allied mechanisms were proposed (Bradbury & Vehrencamp, 2011; Kamermans & Hawryshyn, 2011). A behavioural study of the Lake Malawi cichlid Pseudotropheus macrophthalmus demonstrated its ability to discriminate between alternating e-vectors  (Davitz & McKaye, 1978). Even though the authors admitted imperfections, their conclusions suggest an option to solve the presented problem.
Polarization sensitivity could play a prominent role in target detection by fish (Marshall & Cronin, 2011; Pignatelli et al., 2011). Especially in murky water where colour vision becomes less reliable, polarization sensitivity may be incorporated in contrast enhancement,  camouflage breaking, object recognition and signal detection and discrimination (Cronin et al., 2003). Although transparent zooplankton and insect larvae – important recourses to many cichlids – are strikingly birefringent (i.e. they can partially polarize light), their increased visibility to a predator with polarization sensitivity primarily depends on other factors, like the polarization of the horizontal background radiance and the partially depolarized light scattering from the prey (Johnson et al., 2011). These environmental conditions could have been decisive to Lake Victoria cichlids with polarization sensitivity in order to recover from the initial heavy predation by the Nile Perch and to survive the changed ecosystem.
The innate optokinetic or optomotor response (OMR) finds expression in fish by its swimming along with a horizontally moving vertical contrastive grating. The OMR behavioural paradigm has been used to determine varies optical thresholds in fish (Smith et al., 2011; Maan et al., 2006; Dobberfuhl et al., 2005; Kröger et al., 2003; Van der Meer, 1994). The use of OMR to establish the existence of polarization sensitivity in aquatic animals has been applied to some species of cuttlefish  (Talbot & Marshall, 2010; Darmaillacq & Shashar, 2008), crayfish (Glantz & Schroeter, 2006), salmon (Hasegawa, 2007) and zebrafish (see elsewhere on this site).

Studied species
One South American and four African cichlid species were tested, using OMR with the apparatus accurately described elsewhere on this site. Two specimens of the Lake Malawi cichlid Cynotilapia afra (ca. 4 months old; body-lengths of ca. 4 cm) were collected at a local hatchery where the species had been cultivated for several generations. The species is known to be UV-sensitive (Dalton et al., 2010). Hybridization with the also UV-sensitive Metriaclima zebra cannot be excluded (Streelman et al., 2004). Two juveniles (ca. 4 months old; body-lengths of 3,5 and 4,0 cm) of the Central American Convict cichlid Amatitlania (Cichlasoma) nigrofasciata were also collected at a local hatchery where the species had been cultivated for several generations. Three species of Lake Victoria cichlids were obtained from the life stock of the Biological Institute of Leiden University (IBL). They included adult females of Haplochromis (Yssichromis) piceatus (ca. 18 months old; body-lengths ca. 4-5 cm), adult males and females of Haplochromis (Yssichromis) pyrrhocephalus (ca. 14-18 months old; body lengths ca. 4-5,5 cm) and five adults of Platytaeniodus degeni (ca. 18 months old; body lengths ca. 5,5-6,5 cm).
H. piceatus had been raised for many generations since the 1980s in the laboratory (they practically disappeared in their original habitat and are considered to be extinct; Frans Witte, pers. comm.). One of the females was ‘pregnant’, i.e. she carried juveniles inside her mouth. Two females of H. pyrrhocephalus, an endemic insectivore, descended from the ‘old’ life stock that was raised for at least six generations since the 1980s and two males were recently (March 2011) caught as hatchlings and belonged to the ‘new’ population of recovered haplochromines after the Nile Perch boom. An additional set of three females and two males of H. pyrrhocephalus belonged to the F1 of the hatchlings caught in 2011. The five specimens of P. degeni, an endemicoral shelling/crushing molluscivore, were derived as the fry of a ‘pregnant’ female caught in March 2011 and since then raised under laboratory conditions.

Testing procedure
The presence or absence of polarization sensitivity was tested, using optomotor response with the modified apparatus and techniques described in detail elsewhere on this site. The method was tested with the aid of the polarisation sensitive goldfish. The absolute and relative measurements of swimming activity (clockwise or counter clockwise) were carried out with respect to normal optomotor response (Control 1), polarisation filters with alternating e-vectors (Test-carousel) and neutral density filters (Control 2). In all specimens  the positive and negative responses were repeatedly measured during sessions of ten minutes. Significance levels were calculated with the use of Fisher’s exact test.

During the experimental procedures C. afra demonstrated agitated behaviour with rather abrupt swimming movements, body shaking and biting at the cylinder wall. Both individualsseemed to have a preference to swim counter-clockwise. In one session the response to the rotating polarisation filters was significantly higher than to the neutral density filters. Since the mean response to the Test-carousel was significantly different from Control1 (p < 0,0001) and not significantly different from Control 2 (p = 0,8876) the species was considered to be ‘polarization blind’.
The juveniles of A. nigrofasciata showed a more quiet behaviour although the two specimens had quite different personalities. Their individual preferences for a swimming direction were in balance. Only one session showed a significant positive response to the Test-carousel. Since the mean response to the Test-carousel was significantly different from Control 1 (p < 0,0001) and not significantly different from Control 2 (p > 0,9999) the species was considered to be ‘polarization blind’.
The haplochromine species showed a very strong optomotor response by almost constantly swimming along with the rotating stimulus. In contrast to C. afra and A. nigrofaciata,  H. piceatus avoided contact with the cylinder wall during its swimming movements. The ‘pregnant’ specimen needed a little more time to adapt to the experimental tank, initially showing only optokinetic eyeball reflexes and scarcely swimming movements. None of the specimens showed an obvious preferred swimming direction. Sporadically some of the tested individuals demonstrated a similar swimming behaviour in response to the rotating Test-carousel as to Control 1. One session showed a significant positive response to the Test-carousel and due to very little swimming activity some positive responses to the neutral density filters were relatively high. Since the mean response to the Test-carousel was significantly different from Control 1 (p < 0,0001) and not significantly different from Control 2 (p > 0,4752) this species was considered to be ‘polarization blind’ too.
It was more difficult to induce any response to the presented stimuli in the first set of H. pyrrhocephalus than it was in the foregoing species. The optomotor response of the dominant male (largest specimen) was performed poorly and further measurements were therefore neglected in this specimen. The other specimens only responded after a notable period of accommodation by lying still with quivering pectoral fins. After some time their measured responses (swimming behaviour during the sessions) remained acceptable closed up. The small female predominantly responded just below the water surface with every once in a while fast criss-cross swimming movements (these were not registered as optomotor responses). Sometimes specimens showed a preferred swimming direction although this seemed not to be consistent. No specific preferred swimming direction was established. The exposure to the test carousel evoked much more activity and apparent optomotor response than the neutral stimulus did. Accordingly, the species was considered to be able to discriminate between e-vectors and therefore seemed to have some polarization vision. As this was rather unexpected, another set of five specimens of this species was investigated. They appeared to show a similar positive response to the Test-carousel. Since the mean response to the Test-carousel was significantly different from both Control 1 (p < 0,0001) and Control 2 (p = 0,0013) this species was considered to have some polarization sensitivity.
The slightly larger specimens of P. degeni behaved much the same way as H. piceatus by showing a strong optomotor response when exposed to Control 1. Neither the Test-carousel nor Control 2 evoked much movement in some of the sessions, and it took considerable time to collect useful data. Possibly due to the low swimming activity there was a significant difference between the Test-carousel and Control 2 in six sessions. However, since the mean response to the Test-carousel was significantly different from Control 1 (p < 0,0001) and not significantly different from Control 2 (p > 0,5702) this species too was considered to be ‘polarization blind’.

It has been suggested that polarization vision in fish is related to the orientation of the tangent plane between the closely attached ellipsoids of the double cones in the retinal cone mosaic (Hawryshyn, 2000; Novales Flamarique et al., 1998; Rowe et al., 1994) or indeed to the cone mosaic with respect to the mutual position of the outer segments of the cones (Hawryshyn, 2010). The theory suggests a dual explanation: on one hand it presents the involved photoreceptor mechanism to detect polarized light and, on the other hand, it clears up the significance of the double cone mosaic. The latter is quite common among vertebrates (with the exception of placental animals) without anyone understanding its function (Rowe, 2000). The second part of the explanation would also contribute to the understanding why mammals, including Homo sapiens, are incapable of polarization detection. The theory, however, does not answer the question why a double cone mosaic exists in fish that have no polarization sensitivity (see above; Novales Flamarique & Hawryshyn, 1998a; 1998b; 1997).
One photoreceptor mechanism to detect polarized light is based on unequal double cones with different shaped ellipsoids together with the presence of UV-sensitive additional single cones. These features are not characteristic of the retina of percomorph fish, including cichlidae (Van der Meer, 1992). Cichlids lack additional single cones, and UV-light is restricted to clear water. Whereas under the reduced light conditions where polarization sensitivity could be helpful to spot prey, the orthogonal structure of the cone pattern in cichlids is affected at the cost of short wave photon catching ability (Van der Meer et al., 1995), also called quantum catch efficiency (Kröger et al., 1999). The additional single or small corner cones form no part of the percomorph cone mosaic and the UV-sensitivity of C. afra is based on the absorption maxima of the photopigment in the central single cones (l_max = 358 nm; Dalton et al., 2010).
Another mechanism involves axial dichroism based on tilted membranes in the outer segments of some cones (Roberts et al., 2004).With the correct opponent processing of these cellular outputs from orthogonal channels, unique information can be obtained about the surrounding polarization field (Roberts & Needham, 2007). The oligomaric shift from retinol to dehydroretinol may substantially increase the anistropic quality of the photopigments in the membranes of the outer segments, which could explain the observed polarization vision in H. pyrrhocephalus. Although no direct molecular analyses of the photopigments are available, the species is unique for its maximum sensitivity at longer wavelengths in comparison with other cichlids from the East African Lakes (Carleton, 2009; Van der Meer & Bowmaker, 1995).