Visual Wavelength Discrimination by the Loggerhead Turtle

Home , Hydrozoa

VISUAL WAVELENGTH DISCRIMINATION BY THE LOGGERHEAD TURTLE,

CARETTA CARETTA

Morgan Young

A Thesis Submitted to the Faculty of

The Charles E. Schmidt College of Science

in Partial Fulfillment of the Requirements for the Degree of

Master of Science

Florida Atlantic University

Boca Raton, Florida

May 2012

ii VISUAL··WAVELENGTH··DISCRIMINATION BY·THE LOGGERHEAD·TURTLE, CARETTA CARETTA

This thesis was prepared under the direction of the candidate's advisor, Dr. Michael Salmon, Department of Biological Sciences, and has been approved by the members of her supervisory committee. It was submitted to the faculty.()f the-Charles E. Schmidt College of Science and was accepted in partial fulfillment ofthe requirements for the degree ofMaster ofScience.

Michal Salmon, Ph.D. Thesis Advisor ~~.tJ~

Gary .PerrPh. ~CharlesE.Sc. .dtCollege Of.• ·Science ~ /~':?7 r.~.~ BarryT. R08s0n, Ph.D. Dean, Graduate College

111 ACKNOWLEDGEMENTS

I would like to thank my graduate advisor, Dr. Michael Salmon, for his support, guidance and patience throughout my graduate studies. I would also like to thank my committee members, Jeanette Wyneken and Tammy Frank. I thank FAU undergraduate students Mrs. Alexandra Kogan and Mr. Justin Vining for their assistance. I greatly appreciate Mark Royer’s help in construction of the training apparatus. I would also like to thank T. Warraich, L. Bachler, A. Lolaver, A. Loson, Z. Anderson, M. Rogers, B.

Resnick, I. Pokotylyuk and M. Glider for their help with the husbandry of turtles used in this study.

I would like to acknowledge and thank the National Save the Sea Turtle

Foundation for funding this research. The hatchlings used in this study came from nests managed by the Gumbo Limbo Nature Center, Boca Raton, Florida, U.S.A; their assistance is greatly appreciated. Permits to complete this study were issued by the

Florida Fish and Wildlife Conservation Commission (TP 173) and the FAU Institutional

Animal Care and Use Committee (A10-12).

iv ABSTRACT

Author: Morgan Young

Title: Visual Wavelength Discrimination by the Loggerhead Turtle, Caretta Caretta

Institution: Florida Atlantic University

Thesis Advisor: Dr. Michael Salmon

Degree: Master of Science

Year: 2012

Little is known about the visual capabilities of marine turtles. The ability to discriminate between colors has not been adequately demonstrated on the basis of behavioral criteria. I used a three-part methodology to determine if color discrimination occurred. First, I exposed naïve, light-adapted hatchlings to either a blue, green or yellow light. I manipulated light intensity to obtain a behavioral phototaxis threshold to each color, which provided a range of intensities we knew turtles could detect. Second, I used food to train older turtles to swim toward one light color, and then to discriminate between the rewarded light and another light color; lights were presented at intensities equally above the phototaxis threshold. Lastly, I varied light intensity so that brightness could not be used as a discrimination cue. Six turtles completed this task and showed a clear ability to select a rewarded over a non-rewarded color, regardless of stimulus intensity. Turtles most rapidly learned to associate shorter wavelengths (blue) with food.

v My results clearly show loggerheads have color vision. Further investigation is required to determine how marine turtles exploit this capability.

vi VISUAL WAVELENGTH DISCRIMINATION BY THE LOGGERHEAD TURTLE, CARETTA CARETTA

LIST OF TABLES ...... ix LIST OF FIGURES ...... x INTRODUCTION ...... 1 METHODS ...... 5 Hatchlings: Study site and turtle acquisition ...... 6 Hatchling experiments ...... 6 Phototaxis testing procedure ...... 7 Training juvenile turtles ...... 9 Apparatus and single light training ...... 10 Paired light training...... 11 Wavelength discrimination ...... 11 RESULTS ...... 13 Stimulus measurements ...... 13 Hatchling phototaxis trials ...... 13 Single light training...... 14 Paired light training...... 15 Wavelength discrimination ...... 15 DISCUSSION ...... 16 Experimental approach ...... 16 Training efficacy and stimulus “bias” ...... 20 APPENDIX ...... 24 Anatomy and physiology of the vertebrate eye ...... 31 Visual ecology ...... 32 Evolution of color vision ...... 34 vii

LITERATURE CITED ...... 37

viii

LIST OF TABLES

Table 1. Hatchlings’ response in phototaxis trials ...... 24 Table 2. Numbers of “assisted”, “unassisted”, and total trials during single light training ...... 25

LIST OF FIGURES

Figure 1. Light training apparatus ...... 26 Figure 2. Spectra for 3 colors used in phototaxis and light training ...... 27 Figure 3. Number of trials required to complete criterion ...... 28 Figure Legends ...... 29

INTRODUCTION

Many animals respond behaviorally to differences in object color. These responses range from phototaxes (orientation towards or away from the light source) to discrimination between objects differing in color. Color vision is defined as the ability to distinguish between light stimuli on this basis, independently of differences in light intensity (Hailman and Jaeger 1971; Cronin 2007). Such a capacity is exploited by animals for many functions such as finding food, avoiding predators, orienting toward favorable habitats, reducing conspicuousness (e.g., camouflage), increasing conspicuousness (distasteful and/or poisonous species, or their mimics), species and/or individual recognition, and mate choice (Bradbury and Vehrencamp 1998; Kelber et al.

2003; Siddiqi et al. 2004; Cronin 2007). Presumably, these capabilities evolve if doing so improves probabilities of survival and/or reproductive success.

If color is to be perceived, two requirements must be met. First, the animal must possess at least two different types of retinal (cone) photoreceptors; second, the animal must have neural circuits in the brain designed to contrast signals received from those two groups of photoreceptors so that they can be perceived and responded to behaviorally

(Jacobs and Rowe 2004). Thus, physiological study alone is unable to provide conclusive evidence for color vision; that can only be done on the basis of appropriate

1 behavioral studies (Bradbury and Vehrencamp 1998; Birgersson et al. 2001; Kelber et al.

2003).

The visual capabilities of marine turtles have been studied primarily from a physiological perspective (Reviewed by Bartol and Musick 2003). Most of the behavioral studies, centering on visual orientation of hatchlings from the nest to the ocean, have been done at night (e.g., Mrosovsky and Carr 1967; Mrosovsky and

Shettleworth 1968; Mrosovsky 1972; Witherington and Bjorndal 1991a, b), when rod vision dominates. Thus, there is no conclusive evidence that these animals perceive color

(see Appendix).

However, it is likely that they do as their closest relatives do. Freshwater turtles have been known to detect color since the 1930’s (Reviewed by Neumeyer 2006) and tortoises since the 1950’s (Quaranta 1952). Marine species, like most turtles, have retinas that contain both rods and several types of cones (Liebman and Granda 1971; Bartol and

Musick 2001). Spectral sensitivities of juvenile green turtles (Chelonia mydas; Granda and O’Shea 1972), as well as hatchling leatherback (Dermochelys coriacea) and loggerhead (Caretta caretta) turtles (Horch et al. 2008) have been described based upon physiology. These results indicate a capacity to respond to broad range of light wavelengths, too large to be explained on the basis of a single cone pigment. Horch et al.

(2008) proposed that at least 3 cone types were likely. In the green turtle, a fourth photopigment, sensitive to the near UV wavelengths, may also be present (Mäthger et al.

2007).

Further evidence for the possibility of color vision comes from the presence of oil droplets in the cone cells of marine turtles. These reduce overlap in spectral sensitivity between spectrally adjacent cones and in freshwater turtles, increase the potential for color discrimination (Neumeyer 1986; Vorobyev 2003). The green turtle retina contains

3 types of oil droplets: clear (paired with the 440 nm pigment), yellow (paired with the

502 nm pigment), and orange (paired with the 562 nm pigment; Granda and O’Shea

1972; see, also, Mäthger et al. 2007). These may also sharpen wavelength discrimination capabilities but the crucial experiments remain to be done.

There have been many studies describing preferences of turtles for different colors but only a fraction of these have been properly designed. Most yielded inconclusive results because discrimination in these studies (done by Heidt and Burbidge 1966; Ernst and Hamilton 1969) might have occurred on the basis of light intensity rather than light wavelength. Exceptions to poorly designed experiments include studies done by

Quaranta (1952) on two species of giant tortoises (Testudo elephantopus vicina; Testudo gigantea), and by Neumeyer and her collaborators (Neumeyer and Jäger 1985; Arnold and Neumeyer 1987) on red-eared sliders (Pseudemys scripta elegans). Both provided strong evidence for color vision.

Fehring (1972) did the only behavioral study to determine if marine turtles (juvenile loggerheads, Caretta caretta) could discriminate between colors. Unfortunately, while he included what might have been adequate controls for stimulus intensity, he failed to describe precisely how these were incorporated into his experimental design. He also used Wratten filters to generate colored stimuli. These generated irregular and often

3 broadly overlapping spectra, so much so that it was impossible to determine if the turtles were making discriminations on the basis of wavelength differences or intensity differences generated by narrow vs. broadband spectral energy distributions.

In this study, I used innate responses by hatchling loggerheads, coupled with trained responses shown by older (> 1 month old) juveniles, to determine whether the turtles possessed color vision. My results clearly show that loggerheads have this ability.

METHODS

Experiments were done in three steps using turtles at different stages of ontogeny.

Initially, light-adapted hatchlings (which show an innate positive phototaxis) were induced to crawl toward a blue, green or yellow light presented to them from one arm of a Y-maze. Stimulus brightness was decreased during successive trials until this response

(a positive phototaxis) was no longer evident. The lowest intensity of each color that evoked significant attraction was designated as the “phototaxis threshold”. It served as the dimmest stimulus detected by somewhat older (> 1 month) juvenile turtles used in additional light training and discrimination experiments.

Juvenile turtles were conditioned (single light training) to swim in a water-filled Y- maze toward either the same blue, green or yellow light stimulus presented to the hatchlings. A pair of turtles was conditioned to each color using food (small pieces of raw shrimp; loggerheads love raw shrimp!) as a reward. After reaching criterion (see below), that color was paired with a different color that was not rewarded (paired light training). Both colors were presented at a similar perceptual intensity (1 log unit above the hatchling phototaxis threshold).

Once criterion for that task was achieved a final experiment was done to determine whether the turtles could discriminate between two colors when light wavelength was the only reliable cue. In these experiments paired colors were presented at different stimulus

5 intensities relative to one another during each trial. The outcome of these experiments was used to determine if the turtles were using color or intensity to make discriminations.

If, for example, the turtles made light discriminations based upon wavelength they would consistently choose the positively rewarded color, regardless of its intensity or the intensity of the color with which it had been paired. However, if discrimination was based upon light intensity rather than wavelength, then the turtles would consistently choose either the brighter or dimmer light stimulus, regardless of its wavelength.

Hatchlings: Study site and turtle acquisition

Tests with hatchlings were completed during the summer of 2010 using turtles acquired from nests deposited on a 4.5 km long section of beach in Boca Raton (Palm

Beach County), Florida, USA (Lat 26°22’, Long 80° 07). All nests are marked and monitored by staff from the Gumbo Limbo Nature Center. Each morning, they use stakes and signage to mark and date each nest deposited the previous evening. About 52 days after deposition, I checked each nest during the afternoon to determine if hatchlings were present and likely to emerge. If they were, I removed up to 10 turtles from each nest and placed them on a bed of moist sand inside a light-proof cooler. I transported the cooler within 20 min to a non-air conditioned, windowless room in my laboratory where it was stored until evening when the turtles were used in experiments.

Hatchling experiments

The response of hatchlings to three different colored lights was determined by placing them inside a black Plexiglas™ Y-maze (10 cm deep, 48 cm long, 16.5 cm wide; arm

6 length, 29.2 cm). Hatchlings were given a choice between an illuminated and non- illuminated arm. Two Kodak slide projectors (Model 4400) were used to present the light stimuli. A black Plexiglas™ filter chamber (38.5 cm long, 12.7 cm wide; square in cross section) was placed between the projector and frosted panel at the end of each maze arm where a circular light image, ~ 6.0 cm in diameter, was projected. Stimulus intensity was controlled with appropriate combinations of 5 x 5 cm2 square glass neutral density filters (0.3, 0.5, 1.0, 2.0 and 3.0 log unit; Edmund Scientific, Barrington, NJ). Stimulus intensity was measured as irradiance with a UDT (Model 351A) radiometer (Hawthorne,

California), with its sensor ~ 2 cm from the illuminated circle image. Stimulus wavelength was controlled by 5 x 5 cm2 square glass interference filters (blue, 450 nm; green, 500 nm; yellow, 580 nm; Edmund Scientific, Barrington, NJ).

Phototaxis testing procedure

Hatchlings locate the ocean from the nest by crawling away from dark, elevated horizons produced by the dune and vegetation behind the beach, and by crawling toward the lower, flatter and often brighter horizon toward the ocean (Limpus 1971; Mrosovsky

1972, Salmon et al. 1992). I used the brightness component of this response to determine the minimum intensity of the blue, green and yellow light that evoked a positive phototaxis. Experiments were done inside a windowless, dark room that housed the Y- maze. Immediately before testing, I removed up to 10 hatchlings from the cooler where they were stored and placed them in a shallow dry basin illuminated from overhead by a white fluorescent lamp (8 w). The turtles were exposed to this light for a minimum of 15

7 min before I used them in experiments. Light adaptation was accompanied by an increase in the turtles’ crawling activity.

A single light-adapted turtle was placed in the start area of the Y-maze and immediately exposed to a circular, colored light stimulus projected on the end of one maze arm. The turtle was given 2 min to crawl from the start area into either the illuminated or the dark arm. Each turtle was tested only once. Crawls into the illuminated arm were scored as positive (+) responses, crawls into the non-illuminated arm were scored as negative (-) responses, and a failure to crawl into either arm within 2 min was scored as a neutral (0) response.

Experiments with each color began with the light stimulus presented at its brightest intensity. That stimulus induced all of the turtles to crawl into the illuminated arm. In later trials, light intensity were reduced at 1.0 log unit steps until the turtles were equally likely to crawl into either the illuminated or the non-illuminated arm. Stimulus intensity was then increased in 0.5 log unit steps until the positive response returned (the up-down

“staircase” method for measuring thresholds; Dixon and Mood 1948; Cornsweet 1962;

Levitt 1971). For half of the turtles the right arm was illuminated; for the remainder it was the left arm. These procedures were repeated each evening using turtles from a different nest. Once sample sizes for mixed responses (+. – and 0) at a given wavelength and intensity exceeded ~ 20 turtles, I used a binomial test (Zar 1999) to determine whether the ratio of positive to negative and neutral responses resulted in a significant preference for the arm containing the light stimulus. The lowest light intensity at each

8 wavelength that evoked significant attraction was designated as the phototaxis threshold

(pt) for that light wavelength.

After all of the trials each evening were completed, the turtles were weighed (nearest

0.1 g using an electronic scale), measured (straight-line carapace length [SCL], nearest

0.1 mm using calipers), and then returned to the beach and released.

Training juvenile turtles

Loggerheads from nests deposited at Boca Raton were reared at Florida Atlantic

University’s marine laboratory. Each turtle was marked with non-toxic nail polish for individual and nest identification. Because groups of juvenile loggerheads in confined spaces become aggressive, each turtle was isolated individually in a plastic basket (13.4 cm deep, 19.5 cm long, 17.5 cm wide) that floated at the water surface inside shallow tanks furnished with a continuous flow of filtered seawater. Water temperatures varied with natural ocean temperatures, and ranged between 23-30° C. The light cycle was set by timers at 12L:12D. Turtles were fed a formulated diet consisting of turtle chow mixed with ground trout imbedded in gelatin cubes, and supplemented with reptile vitamins and minerals.

I began training 15 turtles, selected based upon a combination of age (minimally, 4 weeks post-emergence), size (> 6.7 cm in SCL) and behavior (consistent feeding vigor, which tended to increase as the turtles grew larger). Training was done once daily.

Ultimately, I used 6 turtles that learned these tasks most rapidly. Variation in learning among individual turtles is common in training studies (Fehring 1972; Bartol et al. 2003).

Three turtles were used for experiments done during the summer of 2010 and three for experiments done during the summer of 2011.

Apparatus and single light training

Daily training took place in a larger Y-maze (10 cm deep, 76 cm long, 13 cm wide;

Fig. 1) filled with seawater to a depth of ~ 7.5 cm. Each turtle was placed in the start area of the maze (Fig. 1) and allowed to acclimate for 2 min before the clear barrier separating the start area from the arms was lifted. During the acclimation period, the turtle could observe a 6 cm diameter circle of blue, green or yellow light (brightness set at 1 log unit above the hatchling pt for that color) at the end of one arm, and another arm that was dark. To obtain a food reward, the turtle had to swim to the end of the illuminated arm.

Because some turtles did not immediately leave the start area after the barrier was removed, I found it necessary to shape this response (in 5 of the 6 subjects) by gently pushing the turtle out of the start area and toward the end of an arm. If the turtles “chose” the illuminated arm, it received a small piece of raw shrimp as a reward before being returned to the start area for the next trial. If it entered the dark arm the turtle was allowed to swim to the end of the arm but then gently removed and placed at the end of the illuminated arm where it was exposed to the light stimulus. It was then given a food reward before being returned to the start area.

On the second day of training, the arm presenting the illuminated target varied (left or right), as dictated by a random number table. Reward followed only when the turtle swam to the end of an illuminated arm. I timed each trial with a stopwatch to determine

10 whether as the turtles learned this response they took less time to swim to the end of the maze arm. Trials were repeated until the turtle reached satiation and became unresponsive (generally, after 15 – 35 trials).

I used a two-tailed Mann-Whitney U test (Zar 1999) to determine if the duration of these responses after day 1 (when the turtles were assisted) shortened as the turtles learned to associate a color with food reward. Criterion was achieved when each turtle selected the illuminated arm 10 times in succession over each of two consecutive days of testing. I used chi-square tests (Zar 1999) to determine whether the number of trials required to reach criterion differed as a function of rewarded light color.

Paired light training

Paired light training followed after single light training. Each turtle was simultaneously presented with its “rewarded” color (end of one arm) and a different color

(end of the other arm). Both colors were presented at a perceptually equal intensity (1 log unit brighter than the hatchling pt for that color). A random numbers table was again used to determine left arm-right arm position for the stimuli. Turtles were rewarded only if they selected the correct color. Trials continued on a daily basis until criterion was achieved (10 correct arm choices in succession over 2 consecutive days of testing).

Wavelength discrimination

In this last series of experiments, each turtle was required to select the rewarded color under two conditions: when the rewarded color was held at a constant intensity (1.0 log unit above the pt) while the unrewarded color varied in intensity, and when the

11 unrewarded color was held at a constant intensity (1.0 log unit above the pt) and the intensity of the rewarded color intensity was similarly varied. Intensity variation encompassed a range of 2.0 log units for blue and yellow, and 2.5 log units for the green stimulus.

One set of trials was run each day until the turtle achieved criterion, followed the next day by the second set of trials. Three of the turtles were first exposed to a constant intensity of the unrewarded color while three of the turtles were first exposed to a constant intensity of the rewarded color.

RESULTS

Stimulus measurements

When projector lighting passed through each interference filter, the result was a narrowband (+/- 5 nm) spectral energy distribution for each color. Peaks were located at

450 nm (blue), 500 nm (green) and 580 nm (yellow; Fig. 2).

Stimulus irradiance was measured at a distance of 2 cm from the projected image of the light stimulus on the vertical wall at the end of the maze arm (phototaxis experiments). In the absence of neutral density filters, irradiances were: blue, 1.63 x 1014 photons/cm2/s, green, 2.87 x 1013 photons/cm2/s and yellow, 5.03 x 1013 photons/cm2/s.

Hatchling phototaxis trials

I used 401 turtles from 32 nests in these trials. Hatchlings averaged 4.32 cm in SCL

(range: 3.81 - 4.79 cm) and 17.61 g in mass (range: 13.48 - 22.94 g).

Hatchlings were attracted to all of the light colors but the turtles were almost equally sensitive to green and yellow, and almost a log unit less sensitive to blue (Table 1). The irradiance pt for green was 2.52 x 1011 photons /cm2 /s while it was 4.96 x 1011 photons

/cm2/s for yellow and 6.75 x 1012 photons/cm2/s for blue.

Single light training

The 6 juvenile turtles that were trained to criterion came from 4 different nests. At the start of training they ranged in SCL between 6.7 - 10.0 cm, and in mass between 64-

173 g.

During single light training there were noticeable changes in behavior as the turtles

“caught on” to the task. With additional days of training, the percentage of correct choices increased and trial duration on average was significantly shorter (Mann-Whitney z = 2.4732, p = 0.01). Some of the turtles swam down the maze arm with their mouths open. All of the turtles soon began searching for food and biting at the maze wall when they reached the end of the arm.

All turtles given single light training learned to orient toward the illuminated arm but there were differences in the number of trials required before the turtles reached criterion

(Table 2). The two turtles trained to a blue light reached criterion most rapidly (after 40 and 73 trials, respectively), whereas turtles trained to green required 91 and 92 trials, and turtles trained to yellow 157 and 186 trials, respectively (Fig. 3). When I compared these trial numbers to an “expected” average, they differed significantly. Turtles 1, 3 and 5 were trained in 2010; their number of trials to reach criterion were 40 (blue), 92 (green) and 186 (yellow). Those values differed from an expected average distribution (106 trials; X2 = 52, p = 0.0001, 2 d.f.). Turtles 2, 4 and 6 were trained in 2011; their trials (73,

91, 157) also differed from an expected average distribution (107 trials; X 2 = 17, p =

0.0001, 2 d.f.). Thus, turtles learned to associate a color with food reward in fewer trials

14 at the shorter wavelengths (blue) than at the longer wavelengths (green, yellow; Table 2,

Fig. 3).

Pairs of turtles trained to green (91 and 92 trials; X2 = .005, n.s.) and to yellow (186 and 157 trials; X2 = 1.07, n.s.; Fig. 3) lights showed no statistical differences in the number of trials required to reach criterion. However, the pair of turtles trained to a blue light differed significantly in performance (40 and 73 trials; X2 = 4.37, p = 0.04).

Paired light training

Turtle 1 was trained to distinguish between blue and green while turtle 2 was trained to distinguish between blue and yellow, with blue the rewarded color. Turtles 3 and 4 were trained to distinguish between green and yellow and green and blue, respectively, with green the rewarded color. Turtles 5 and 6 were trained to distinguish between yellow and blue and yellow and green, respectively, with yellow the rewarded color.

All of the turtles reached criterion in 20 - 38 trials (Fig. 3). There were no statistical differences among the turtles in trials required to reach criterion, when compared to an expected average (of 27 trials; X 2 = 4.25, p = 0.5, 5 d.f.).

Wavelength discrimination

All of the turtles achieved criterion for wavelength discrimination in 22 - 50 trials

(Fig. 3). There were no statistical differences among the turtles in the number of trials required to reach criterion when compared to an expected average of the observed numbers (30 trials; X 2 = 8.84, p = 0.12, 5 d.f.).

DISCUSSION

My experiments demonstrate that juvenile loggerheads can be trained to associate visual targets differing in color with food as a reward. The turtles learn this association most rapidly when the colored target consists of shorter rather than longer wavelengths, even when all of the targets are presented at intensities that are equal in perceived brightness (1 log unit above a stimulus level just sufficient to evoke a positive phototactic response; Table 1). However, once that association is learned discrimination between the rewarded color and any unrewarded color is achieved rapidly (Fig. 3). Discrimination persists even when the brightness of each of the light sources is varied, rendering intensity as an unreliable cue. These results show that the turtles can distinguish between visual targets on the basis of light wavelength.

Experimental approach

Phototaxis trials were done to determine the minimum intensity of each light color that evoked orientation toward the light source. Once these thresholds were known they could be used to establish a minimum light intensity at each wavelength that I knew the turtles could detect.

During both single and paired light training, I presented each turtle with a colored target at an intensity that should have been approximately equal in its perceived intensity,

16 as each stimulus was presented at 1 log unit brighter than the hatchling phototaxis threshold for that color. Nevertheless, individual differences between the turtles in spectral sensitivity might have enabled discrimination to occur, even among these

“equivalent brightness” stimulus pairs. To eliminate that possibility, a final set of wavelength discrimination trials was done in which each target varied in brightness from one trial to the next. Even under those conditions, the turtles consistently chose the rewarded color on the basis of the only consistently remaining cue: wavelength.

In previous studies with other species, adequate control over stimulus intensity has not always been a feature of wavelength discrimination studies done with other turtle species. In experiments by Ernst and Hamilton (1969) and by Heidt and Burbidge

(1966), lights differing in color were presented at equal intensities as measured by a light meter. That design assumed that both a turtle eye and a light meter were equally sensitive to each colored target. However, that feature is unique only to some (but not all) light meters and is uncharacteristic of biological systems. The spectral sensitivity characteristics of any animal’s photoreceptors are shaped by its visual ecology, which in turn selects for tuning to those wavelengths important for avoiding predators, detecting prey, locating mates and other functions critical to survival and reproductive success

(Hailman and Jaeger 1971; Bradbury and Vehrencamp 1998). Thus, animals are likely to be more sensitive to some wavelengths than others so that when those wavelengths are paired with others at equal intensity, they will be perceived as brighter. That feature of visual detection makes it imperative to design experiments that distinguish between

17 discrimination based upon intensity differences vs. discrimination based upon wavelength differences.

To demonstrate color vision, the experimental design must eliminate the possibility that discrimination is accomplished by intensity differences. One method is to measure each subject’s spectral sensitivity and then to present lights differing in wavelength at perceptually equal intensities to that subject. Such an approach was used to determine if the red-eared slider was capable of wavelength discrimination and if so, how that ability varied across the entire spectrum of wavelengths that the turtle could detect (Neumeyer and Jäger 1985; Arnold and Neumeyer 1987).

A second approach, and the one I used, is to irregularly vary the intensity of each stimulus presented to the animal so that if discrimination persists, it cannot be explained on the basis of stimulus intensity cues. This approach was used in experiments with tortoises done by Quaranta (1952), and in experiments patterned after Quaranta’s approach done by Fehring (1972) with juvenile loggerhead sea turtles. Quaranta used food reward to train tortoises to discriminate between pairs of light stimuli differing in wavelength. After the response was acquired, he varied stimulus intensity of each of the light targets independently over a broad (3.5 log units) range. In the absence of any spectral sensitivity information, he could not know when his trained tortoises saw one color as brighter or dimmer than the other. But, given the 3 log unit range of intensities he presented, coupled with the separate tests in which either the rewarded or unrewarded color was varied in intensity, it was unlikely that a consistent preference for one color could be explained on the basis of any cue other than wavelength. Fehring’s (1972)

18 design was similar to Quaranta’s in that he also varied light intensity over a broad (2.5 log unit) range. However, he failed to describe how these critical intensities were paired during the trials, nor could he account for why the turtles were unable to distinguish between some of the colored stimuli he presented.

Color vision has been shown to occur in many other animals, beginning with Karl von Frisch’s (1914) classical demonstration using honeybees (Apis mellifera) as subjects.

Von Frisch trained the bees to associate a colored card placed under a glass sugar-water filled bowl with food reward. Once trained, the bees continued to alight on the colored card, even when no bowl and no food was present, and even when the card was placed anywhere within an array of 30 similar cards differing in shade from white through grey to black. The assumption was that at least one of the shaded cards matched the achromatic (intensity) signal of the trained color, and so bees making no errors must have used chromatic (color) cues to make their selection.

The design of recent studies follows similar principles. For example, color vision in the mantis shrimp, Haptosquilla trispinosa, was demonstrated by training the shrimp to discriminate between a rewarded color and various shades of grey, even when the rewarded color was presented at intensities that varied compared to its intensity during training (Marshall et al. 1996). Damsel fish (Pomacentrus amboinensis) were trained using food reward to select a colored target, then to continue to make that selection when the target was paired with other targets differing in intensity (Siebeck et al. 2008). These damselfish live in social groups (“harems”), controlled by a dominant male. Males can discriminate between females in their harem and those who are strangers, and between

19 their species and another that looks identical to a human observer. Discrimination in both instances is based upon individually distinctive patterns of head coloration, expressed as complex UV patterns (Siebeck et al. 2010).

The diurnally active hawkmoth (Macroglossum stellatarum) is a nectar feeder. It can be trained, using food reward, to distinguish among targets either on the basis of differences in wavelength, differences in intensity, or both cues (Kelber 2005). However, hawkmoths normally make flower discriminations based upon color. They require fewer trials to discriminate between visual targets using wavelength than using intensity cues.

Training efficacy and stimulus “bias”

I originally planned to train neonate loggerheads, about 2-3 weeks post emergence, that had been feeding on their own. However, these turtles never learned to associate a color with food, probably because they were satiated before they could complete a sufficient number of trails for learning to occur. However, older juvenile marine turtles had been successfully trained in previous studies (Fehring 1972; Manton et al. 1972;

Bartol 1999). I found that larger neonates were highly motivated to feed and would do so during repeated trials each day. Training was successfully accomplished with these subjects.

There were obvious differences among the turtles in how quickly associations formed between the target colors and food. Turtles were most quickly trained using a blue light stimulus, but training was more difficult to accomplish using a green stimulus, and required persistent training over many days when yellow was the light stimulus (Table 2;

Fig. 3). These differences were not obviously related to light sensitivity, at least as

20 measured by phototaxis, as light intensities at pt for the green and yellow were quite similar (2.52 x 1011 photons /cm2 /s for green; 4.96 x 1011 photons /cm2/s for yellow).

The turtles were considerably less sensitive to a blue light (pt was 6.75 x 1012 photons/cm2/s).

Why, then, was training accomplished more easily using a blue stimulus? One possibility is that the turtles are predisposed to associate objects reflecting the shorter wavelengths with food. That hypothesis receives support because at least some prey that small turtles are known to consume are often blue, dark blue or purple in coloration.

Young loggerheads during their oceanic stage of development associate with

Sargassum mats (Carr 1987; Bolten 2003). During this time loggerheads feed primarily on prey within the Sargassum community, as well as other organisms that accumulate nearby in surface waters adjacent to these algal mats (Richardson and McGillivary 1991;

Witherington 1994; Boyle and Limpus 2008). These prey include crustaceans, hydrozoa, insects, gastropods and gelatinous zooplankton identified from the stomach contents of small loggerheads (Brongersma 1972; Hughes 1974; Plotkin 1993; Witherington 1994,

1998; Tomas et al. 2001). A large proportion of these are cnidarians (Witherington 1994,

1998; Richardson and McGillivary 2001; Parker et al. 2005; Boyle and Limpus 2008), including wind-driven jellyfishes that are primarily blue or violet in color such as the

Portuguese man o’ war (Physalia physalis), blue button (Porpita porpita) and by-the wind sailor (Velella velella). Loggerheads also consume pelagic gastropods that emphasize the same colors, such as the common purple snail (Janthina janthina). These prey represent a large component of the loggerhead diet during their oceanic phase of

21 migration (Bolten and Balazs 1995; Bjorndal 1997; Boyle and Limpus 2008). Blue to violet colors are also the most common light wavelengths emitted by bioluminescent marine organisms (Widder 2010) such as jellyfishes and crustaceans, many of which are vertical migrators that might serve as prey for older marine turtles that may forage in oceanic waters at night.

Many other animals show color biases that apparently function to promote their survival and reproductive success. Bumble bees (Bombus terrestris; Raine and Chittka

2007), butterflies (Aglais urticae and Pararge aegeria; Scherer & Kolb 1987) and hawkmoths (Manduca sexta; Goyret et al. 2008) have an innate preference for flowers of certain color (by coincidence, often blues or violets) that have the highest nectar content and therefore provide the maximum reward. Color biases are also common in mate selection, specifically, female preferences for males bearing certain colors associated with characteristics such as reduced parasitism (Godin and McDonough 2003), nest attentiveness (Hill 1991), general health and genetic quality (Folstad and Karter 1992).

These have been studied in a variety of animals including fishes (guppies, Poecilia reticulata; Godin and McDonough 2003), sticklebacks (Gasterosteus aculeatus; Folstad et al. 1994), birds (finches, Carpodacus mexicanus; Hill 1990) and macaques (Macaca mulatta; Waitt et al. 2003). Color biases have not yet been conclusively reported for any species of turtle and virtually nothing is known about how color vision might be advantageous for any chelonian species. One obvious possibility is that color vision may enable land tortoises to discriminate between edible and inedible leaves, or ripened and

22 unripened fruit. A similar function has been proposed to explain the evolution of color vision in the Old World primates (Jacobs and Rowe 2004).

While I presented the turtles with light wavelengths that are visible to humans, previous studies (Mäthger et al. 2007; Horch et. al) have shown that marine turtles also detect near ultraviolet light. This capacity may be advantageous to marine turtles feeding on gelatinous prey near the ocean surface. Many of these species appear transparent to organisms that see only “visible” light (Bjorndal 1997) but may appear to the turtles as either brighter (Lythgoe 1988) or darker (Bowmaker and Kunz 1987; Losey et al. 1999) than the surrounding environment, depending upon whether these prey reflect or absorb

UV light. Overall, how marine turtles may exploit their wavelength discriminating capabilities remains unexplored.

APPENDIX

Table 1. The number of hatchlings that crawled into the illuminated arm (+), the dark arm (-), or that failed to crawl (0) during the phototaxis trials. Data are shown as a function of stimulus intensity beginning with no ND filter present, then as intensity declined at 0.5 – 1.0 log unit steps. The phototaxis threshold (bold) is defined as the minimum intensity evoking significant attraction to the light stimulus (at p < 0.05 by a binomial test).

______

Log Stimulus Blue Green Yellow Intensity + -/0 + -/0 + -/0 ______

No Filter 12 0 14 1 9 2

1.0 10 2 12 3 13 2

1.5 23 2 12 3 18 0

2.0 18* 8 20 6 16*** 4

2.5 16 9 23** 6 15 10

3.0 11 14 16 9 8 7

3.5 5 7 10 10 8 7 ______

* p = 0.04 ** p < 0.001 ***p = 0.006 24

Table 2. Numbers of “assisted”, “unassisted”, and total trials during single light training required to reach criterion (10 correct responses in succession over each of two days).

All turtles (with the exception of turtle 3, the first turtle trained) received “assistance” during training (see text). The rewarded stimulus was blue for turtles 1 and 2, green for turtles 3 and 4, and yellow for turtles 5 and 6.

Turtle Stimulus Assisted Trials Unassisted Trials Total Trials Color to Criterion

1 blue 5 35 40

2 blue 25 48 73

3 green 0 92 92

4 green 54 37 91

5 yellow 92 94 186

6 yellow 136 21 157

______

Figure 1. Light training apparatus

P P

FC 31 cm

CB YMA

76 cm

Start 13 cm x 10 cm area

Figure 2. Spectra for 3 colors used in phototaxis and light training

Relative intensity Relative

Wavelength (nm)

Figure 3. Number of trials required to complete criterion

250 20

Discrimination Paired light 20 25

200 Single light 20 20

38 25

20 50 29 150 2224 50 29 33 31 27

22 30 27 100 33

NumberTrials of 38 50

22 40 73 92 91 186 187 0 11 22 3 43 54 6 57 68

Turtle Number

Figure Legends

Figure 1. Oblique view of the apparatus used for experiments to determine if juvenile loggerheads can be trained to distinguish between colors. Two light sources are presented simultaneously from Kodak projectors (P). The lights pass through an interference and neutral density filter in each filter chamber (FC) before a circular image appears at the frosted end of the two Y-maze arms (YMA; 10 cm deep, 42.3 cm long, 14 cm wide). The turtle is restrained in the start area behind a clear barrier (cb) while the light stimuli are presented. The barrier is lifted to allow the turtle to choose between the rewarded and unrewarded color. Here, the choice is between a yellow and a blue stimulus.

Figure 2. Spectra for the 3 colors used in the trials (based upon data provided by Edmund

Scientific). Colors were produced using Kodak projectors with FHS halogen lamps and

Edmund Optical (5 cm2) interference filters (half band, 5 nm) that produced spectra nearly identical in shape, but different in wavelength.

Figure 3. The number of trials required to complete training during single light (open bars), paired light (grey bars), and wavelength discrimination (black bars) trials. Color pairings for the turtles were: Turtle 1, blue with green; turtle 2, blue with yellow; turtle 3, green with yellow; turtle 4, green with blue; turtle 5, yellow with blue; and turtle 6,

29 yellow with green. On average, fewer trials were required to associate a shorter wavelength color with food than a longer wavelength color. However, once that association was learned the turtles did not differ in the number of trials required to discriminate between pairs of colors. See text for details.

Anatomy and physiology of the vertebrate eye

In all vertebrates the image-forming eye is similar in its basic structure and many properties of its physiology. Light enters the eye by passing through a transparent cornea and into the lens whose refractive index usually differs from the surrounding fluid. The lens focuses the image on the retina either through changing lens shape (mammals) or by changing lens fore-to-aft position (fishes and amphibians). The enormous variation in light intensity entering the eye is controlled by the diameter of the iris opening which decreases in bright light and increases in dim light, much like an f-stop on a camera.

Light then passes through the clear fluid of the vitreous to the retina, which contains several layers of neurons and, at its periphery, the receptors (rods and cones).

These receptors provide a second mechanism for handling the vast range of light intensities present in the environment. Rods are specialized for sensitivity (photon capture) and are used under conditions of dim illumination (scotopic vision at night).

Many of these cells over a relatively large area of the retina converge on a few neurons leading to the ganglion cells (optic nerve axons) so that if any one of them detects light, neural output is affected. That arrangement means that rods provide poor information on spatial detail (a “blurry” image). Cones, in contrast, are relatively insensitive to light but that hardly matters since they operate during bright light conditions (photopic vision in bright light during the day). Cones also are concentrated in the center of the retina, especially in the fovea, where image focus is most precise. Cones converge on relatively few neurons leading to ganglion cells, and thus convey a “finer grained” or sharper image

(Marler and Hamilton 1966).

In both rods and cones, the transduction process (conversion of light stimuli into nerve impulses) is mediated by visual pigments present in the outer (lamellar) segment.

Vertebrates possess one kind of visual pigment (rhodopsin) in the rods, but can have a different iodopsin sequestered in different cones. These iodopsins differ slightly in their amino acid composition in the opsin (protein) portion of pigment, with the result that different visual pigments vary in their spectral tuning. Humans, for example, have three kinds of cones distinguished by their sensitivity to the shorter, longer and medium wavelengths of light that we detect. The colors we perceive are a property of the different combined responses to light stimuli of the three kinds of cones. For example, a light dominated by “red” stimulates the long wavelength cones more than the medium or short wavelength cones whereas a “green” light stimulates the medium cones more than the short or long wavelength cones. In each case, the particular across-cone pattern of stimulation gives rise in the visual brain to a different sensation of color.

Visual ecology

Visual systems among species evolve to best suit the animal’s environmental challenges (finding food; shelter and mates; avoiding unfavorable habitats; thwarting parasites and predators). For those reasons the structure and physiology of the eye reflects an animal’s way of life (Marler and Hamilton 1966). For example, animals can be active during different portions of the light-dark cycle (at night, during the day, or during dawn and dusk [crepuscular] periods). This specialization results in different concentrations of rods and cones in their retinas. Retinas are dominated by rods in night-active vertebrates and cones in day-active animals, so much so that a histological examination of the retina

32 is a reliable indicator of when an animal is active (Marler and Hamilton 1966; Gould

1982).

Terrestrial vertebrates that are nocturnally active are likely to have larger eyes

(with greater light-gathering potential) than those that are closely related, but diurnally- active. The same is true of deep sea fishes compared to their relatives found at shallower depths in the photic zone (Marler and Hamilton 1966).

Among vertebrates, herbivores most often have their eyes placed laterally to maximize their ability to detect objects (food or predators) ahead, to the side, and behind them. That is, they maximize their visual field extent. Predators, however, have eyes placed closer together and in the frontal plane of the head. This adaptation sacrifices visual field extent but provides overlap between the visual field of both eyes and therefore, superior depth perception needed for stalking prey and effectively gauging distance for a strike (Marler and Hamilton 1966).

There are also some interesting differences in the structure of the retina that are correlated with ecology. Animals that roam spatially “open” habitats in bright sunlight

(e.g., terrestrial grazers on open plains, marine grazers on seagrasses or flat reefs;

Pumphrey 1961; Hughes 1977) concentrate cones horizontally across the anterior to posterior, central axis of the retina. This “visual streak” enhances sensitivity to objects moving or approaching on the horizon. Animals that occupy “complex” habitats (forests, coral reefs, mountains) may concentrate cones in circular or oval-shaped areas within the retina (Hughes 1977; Collin and Pettigrew 1988a; 1988b).

Animals also differ in their visual acuity, or resolving power of the eye. This capability is a function, at least in part, of how densely cone cells are packed into the fovea and/or other areas of the retina. Birds, especially birds of prey and vultures, must spot targets from the sky. They do so with ~ 5 times the visual acuity of humans by packing twice as many cones into the same area of fovea. Some hawks also have lenses that act like magnifying glasses to enhance the image. Animals that are acuity champions at night are able to do so presumably by packing more rods into retinal space (e.g., cats and owls).

Finally, the ability to make fine discriminations between colors depends in part upon the number of distinct cone photopigments. Human have three (trichromats) but many reptiles and amphibians have 4 or 5 (tetra- or pentachromats). Honeybees, like humans, are trichromats while some butterflies have up to six photopigments. The most complex visual system is possessed by mantis shrimp (Stomatopods) with 10 or more different spectral receptor types (Cronin and Marshall 1989). Interestingly, those mantis shrimp are found in clear, shallow tropical waters where light levels are bright. Other mantis shrimp, found at greater depth where there is less light, have fewer receptor types and shift their sensitivities to the shorter (blue) wavelengths best transmitted through ocean water.

Evolution of color vision

Color vision is defined as the ability to distinguish between light stimuli on the basis of wavelength, independently of light intensity. This ability, however, involves costs such as those required to synthesize the different classes of visual pigments and to

34 generate the neural machinery required in the brain to analyze the additional visual information transmitted. There must therefore be benefits that exceed the costs, though for some animals, those benefits have yet to be identified.

Most mammals have rather poor color vision capabilities, even though they evolved from reptiles that typically have excellent color vision. Mammals, apparently, lost those capacities when they initially diversified as small, nocturnally-active creatures that co-existed with larger, more dominant and day active reptiles (Kelber et al. 2003).

Most mammals today are dichromats that perceive the world in shades of blue/green.

Primates are the exception among mammals in having well-developed trichromatic color vision, presumably because of their evolved diurnal life-style and diet (Kelber et al.

2003). Primate color vision has been selected for by the necessity of recognizing young leaves and ripe fruit against a background of older leaves produced by plants that depend upon the animals for seed dispersal.

Food recognition may also be a selective pressure favoring the evolution of color vision by nectivorous birds (hummingbirds), as well as by many insects (bees, butterflies) that detect seasonally abundant flowers maturing at different times and places.

Sensitivity to ultraviolet wavelengths also contributes to recognition of, and orientation toward the center of, particular flowers against backgrounds reflecting light of different wavelengths (Bradbury and Vehrencamp 1998).

Color vision is also exploited for purposes other than feeding such as orientation toward favorable habitats, reducing conspicuousness (e.g. camouflage), warning

(distasteful or poisonous species), individual and species recognition, and in choosing

35 mates (squids & octopus, Bradbury and Vehrencamp 1998; brown shrimp, Kelber et al.

2003; strawberry poison frogs, Siddiqi et al. 2004).

Regardless of its present role, an ability to discriminate between wavelengths can be advantageous through small steps that do not immediately result in an ability to discriminate between wavelengths, per se. Suppose a mutation occurs in a monochromat that results in a change in the amino acid composition of an opsin so that the animal now has two photopigments with different (but overlapping) spectral sensitivities. Such an animal may find it advantageous to (i) either more precisely “tune” or even broaden its spectral sensitivity since by doing so it can (ii) enhance visual contrast between objects, each detected best by one of the two populations of cones. These steps can occur even before the neural machinery is in place for distinguishing between the objects on the basis of wavelength. It is for that reason that the possession of multiple pigment systems in cones only indicates a physiological potential for color vision. Proof positive requires experiments that indicate a behavioral ability to discriminate between wavelengths independently of contrasts in light intensity (Bradbury and Vehrencamp 1998; Birgersson et al. 2001; Kelber et al. 2003).

LITERATURE CITED

Arnold, K. & Neumeyer, C. 1987. Wavelength discrimination in the turtle Pseudemys scripta elegans. Vision Research, 27, 1501-1511.

Balazs, G. 1995. Growth rates of immature green turtles in the Hawaiian Archipelago. In: Biology and Conservation of Sea Turtles (Ed. by K. A. Bjorndal), pp. 117-125. Washington D.C.: Smithsonian Institution Press.

Bartol, S. M. 1999. Morphological, electrophysiological and behavioral investigation of the visual acuity of juvenile loggerhead turtles (Caretta caretta). Ph.D. Thesis. Virginia: Faculty of the School of Marine Science, College of William and Mary.

Bartol, S. M., Mellgren, R. L. & Musick, J. A. 2003. Visual acuity of juvenile loggerhead sea turtles (Caretta caretta): a behavioral approach. International Journal of Comparative Physiology, 16, 143-155.

Bartol, S. M. & Musick, J. A. 2001. Morphology and topographical organization of the retina of juvenile Loggerhead Sea Turtles (Caretta caretta). Copeia, 2001, 718-725.

Bartol, S. M. & Musick, J. A. 2003. Sensory biology of sea turtles. In: The biology of sea turtles, Vol. II. (Ed. by P. Lutz, J. A. Musick & J. Wyneken). Boca Raton, Florida: CRC Press LLC.

Birgersson, B., Alm, U. & Forkman, B. 2001. Colour vision in fallow deer: a behavioural study. Animal Behaviour, 61, 367-371.

Bjorndal, K. A. 1997. Foraging ecology and nutrition of sea turtles. In: The biology of sea turtles, Vol. I (Ed. by P. L. Lutz & J. A. Musick), pp. 199-203. Boca Raton, Florida: CRC Press.

Bolten, A. B. 2003. Variation in sea turtle life history patterns: neritic vs oceanic developmental stages. In: The biology of sea turtles, Vol. II (Ed. by P. L. Lutz, J. A. Musick & J. Wyneken), pp. 243-257. Boca Raton, Florida: CRC Press.

Bowmaker, J. & Kunz, Y. W. 1987. Ultraviolet receptors, tetrachromatic color vision and retinal mosaics in the Brown trout (Salmo trutta): age dependent changes. Vision Research, 27, 2101-2108. 37

Boyle, M. C. & Limpus, C. J. 2008. The stomach contents of post-hatchling green and loggerhead sea turtles in the southwest Pacific: an insight into habitat association. Marine Biology, 155, 233-241.

Bradbury, J. W. & Vehrencamp, S. L. 1998. Principles of Animal Communication. Publishers Sunderland, MA: Sinauer Associates, Inc.

Brongersma, L. D. 1972. European Atlantic turtles. Zoologische Verhandelingen (Leiden), 121, 1-318.

Carr, A. 1987. New perspectives on the pelagic stage of sea turtle development. Conservative Biology, 1, 103-121.

Collin, S.P. & Pettigrew, J.D. 1988a. Retinal topography in reef teleosts I, Some species with well-developed areae but poorly-developed streaks. Brain, Behavior and Evolution, 31, 269-282.

Collin, S.P. & Pettigrew, J.D. 1988b. Retinal topography in reef teleosts II, Some species with prominent horizontal streaks and high density areae. Brain, Behavior and Evolution, 31, 283-295.

Cornsweet, T. N. 1962. The staircase method in psychophysics. American Journal of Psychology, 75, 485-491.

Cronin, T.W. & Marshall, N.J. 1989. Multiple spectral classes of photoreceptors in the retinas of gonodactylidstomatopod shrimps. Journal of Comparative Physiology A, 166, 261-275.

Cronin, T. 2007. Evolution of Color Vision and Visual Pigments in Invertebrates. In: Evolution of nervous systems, pp. 361-365. Amsterdam: Elsevier Academic Press.

Dixon, J. W. & Mood, A. M. 1948. A method for obtaining and analyzing sensitivity data. Journal of the American Statistical Association, 43, 109-126.

Ernst, C. & Hamilton, H. 1969. Color preferences of some North American turtles. Journal of Herpetology, 3, 176-180.

Fehring, W. K. 1972. Hue discrimination in hatchling loggerhead turtles (Caretta caretta caretta). Animal Behaviour, 20, 632-637.

Folstad, I., Hope, A. M., Karter, A. & Skorping, A. 1994. Sexually selected color in male sticklebacks – a signal of both parasite exposure and parasite resistance. Oikos, 69, 511-515.

Folstad, I. & Karter, A. J. 1992. Parasites, bright males, and the immunocompetence handicap. The American Naturalist, 139, 603-622.

Frisch, von K. 1914. Der Farbensinn und Formensinn der Biene. Zoologische Jahrbucher. Abteilung fur allgemeine Zoologie und Physiologie der Tiere, 35, 1-188.

Godin, J. G. J. & McDonough, H. E. 2003. Predator preference for brightly colored males in the guppy: a viability cost for a sexually selected trait. Behavioral Ecology, 14, 194-200.

Gould, J.L. 1982. Ethology: The mechanisms and evolution of behavior. W.W. Norton & Company, New York & London.

Goyret, J., Pfaff, M., Raguso, R. A. & Kelber, A. 2008. Why do Manduca sexta feed from white flowers? Innate and learnt colour preferences in a hawkmoth. Naturwissenschaften, 95, 569-576.

Granda, A. M. & O’Shea, P. J. 1972. Spectral sensitivity of the green turtle (Chelonia mydas mydas) determined by electrical responses to heterochromatic light. Brain, Behavior and Evolution, 5, 143-154.

Hailman, J. P. & Jaeger, R. G. 1971. On criteria for color preferences in turtles. Journal of herpetology, 5, 83-85.

Heidt, G. A. & Burbidge, R. G. 1966. Some aspects of color preference, substrate preference, and learning in hatchling Chrysemys picta. Herpetologica, 22, 288-292.

Hill, G. E. 1990. Female house finches prefer colorful males: sexual selection for a condition dependent trait. Animal Behaviour, 40, 563-572.

Hill, G. E. 1991. Plumage color is a sexually selected indicator of male quality. Nature, 50, 337-339.

Horch, K. W., Gocke, J. P., Salmon, M. & Forward, R. B. 2008. Visual spectral sensitivity of hatchling loggerhead (Caretta caretta L.) and leatherback (Dermochelys coriacea L.) sea turtles, as determined by single-flash electroretinography. Marine and Freshwater Behavior and Physiology, 41, 79-91.

Hughes, G. R. 1974. The sea turtles of south-east Africa II. The biology of the Tongaland loggerhead turtle Caretta caretta L. with comments on the leatherback turtle Dermochelys coriacea L. and the green turtle Chelonia mydas L. in the study region. Oceanographic Research Institute Invest Rep, 36, 1-48.

Hughes, A. 1977. The topography of vision in mammals of contrasting life style: comparative optics and retinal organization. In: Handbook of Sensory Physiology (Ed. By F. Cresticelli), pp. 613–756. Berlin, Germany: Springer-Verlag.

Jacobs, G. H. & Rowe, M. P. 2004. Evolution of vertebrate colour vision. Clinical and Experimental Optometry, 87, 206-216.

Kelber, A. 2005. Alternative use of chromatic and achromatic cues in a hawkmoth. Proceedings of the Royal Society: Biological Sciences, 272, 2143-2147.

Kelber, A., Vorobyev, M. & Osorio, D. 2003. Animal colour vision- behavioural tests and physiological concepts. Biological reviews, 78, 81-118.

Levitt, H. 1971. Transformed up-down methods in psychoacoustics. Journal of the acoustical society of america, 49, 467-476.

Liebman, P. A. & Granda, A. M. 1971. Microspectrophotometric measurements of visual Pseudemys scripta and Chelonia mydas. Vision research, 11, 105-114.

Limpus, C. 1971. Sea turtle ocean finding behaviour. Search, 2, 385-387.

Losey, G. S., Cronin, T. W., Goldsmith, T. H., Hyde, D., Marshall, N. J. & McFarland, W. N. 1999. The UV visual world of fishes: a review. Journal of Fish Biology, 54, 921-943.

Lythgoe, J. N. 1988. Light and vision in the aquatic environment. In: Sensory biology of aquatic animals (Ed. by J. Atema, R. Fay, A. N. Popper & W. L. Tavolga). New York: Springer-Verlag.

Manton, M. L., Karr, A. & Ehrenfeld, D. W. 1972. An operant method for the study of chemoreception in the green turtle, Chelonia mydas. Brain, Behavior, and Evolution, 5, 188-201.

Marler, P. and W. Hamilton III. 1966. Mechanisms of Animal Behavior. John Wiley & Sons, Inc. New York, London, Sydney.

Marshall, N. J., Jones, J. P. & Cronin, T. W. 1996. Behavioural evidence for color vision in stomatopod crustaceans. Journal of Comparative Physiology, 179, 473-481.

Mäthger, L. M., Litherland, L. & Fritsches, K. A. 2007. An anatomical study of the visual Chelonia mydas. Copeia, 2007, 169-179.

Mrosovsky, N. 1972. The water-finding ability of sea turtles. Behavioral studies and physiological speculations. Brain, Behavior and Evolution, 5, 202-225.

Mrosovsky, N. & Carr, A. 1967. Preference for light of short wave lengths in hatchling green sea turtles, Chelonia mydas, tested on their natural nesting beaches. Behaviour, 28, 217-231.

Mrosovsky, N. & Shettleworth, S. J. 1968. Wavelength preferences and brightness cues in the water finding behavior of sea turtles. Behaviour, 32, 211-257.

Neumeyer, C. 1986. Wavelength discrimination in the goldfish. Journal of Comparative Physiology A: Neuroethology, Sensory, Neural, and Behavioral Physiology, 158, 203- 213.

Neumeyer, C. 1998. Color vision in lower vertebrates. In: Color vision: perspectives from different disciplines (Ed. by W. Backhaus, R. Kliegl & J. S. Werner), pp. 149-162. Berlin; New York: Walter de Gruyter.

Neumeyer, C. & Jäger, J. 1985. Spectral sensitivity of the freshwater turtle Pseudomys scripta elegans : evidence for the filter effect of colored oil droplets Vision Research, 25, 833-838.

Parker, D. M., Cooke, W. J. & Balazs, G. H. 2005. Diet of oceanic loggerhead sea turtles (Caretta caretta) in the central North Pacific. Fishery Bulletin, 103, 142-152.

Plotkin, P. T., Wicksten, M. K. & Amos, A. F. 1993. Feeding ecology of the loggerhead sea turtle Caretta caretta in the Northwestern Gulf of Mexico. Marine Biology, 115, 1-5.

Pumphrey, R.J. 1961. Concerning vision. In: The cell and the organism (Ed. By J.A. Ramsay and V.B. Wigglesworth), pp. 193-208. London: Cambridge University Press.

Quaranta, J. V. 1952. An experimental study of the colour vision of the giant tortoise. Zoologica, 37, 295-311.

Raine, N. E. & Chittka, L. 2007. The adaptive significance of sensory bias in a foraging context: floral colour preferences in the bumblebee Bombus terrestris. PLoS One, 2, 556- 563.

Richardson, J. I. & McGillivary, P. 2001. Post-hatchling loggerhead turtles eat insects in Sargassum community. Marine Turtle Newsletter, 55, 2-5.

Salmon, M., Wyneken, J., Fritz, E. & Lucas, M. 1992. Seafinding by hatchling sea turtles: role of brightness, silhouette and beach slope as orientation cues. Behaviour, 122, 56-77.

Scherer, C. & Kolb, G. 1987. The influence of color stimuli on visually controlled behavior in Aglais urticae L. and Pararge aegeria L. (Lepidoptera). Journal of Comparative Physiology A, 161, 891-898.

Siddiqi, A., Cronin, T. W., Loew, E. R., Vorobyev, M. & Summers, K. 2004. Interspecific and intraspecific views of color signals in the strawberry poison frog Dendrobates pumilio. Journal of Experimental Biology, 207, 2471-2485.

Siebeck, U. E., Parker, A. N., Sprenger, D., Mäthger, L. M. & Wallis, G. 2010. UV patterns for covert face recognition. Current Biology, 20, 407-410.

Siebeck, U. E., Wallis, G. M. & Litherland, L. 2008. Colour vision in coral reef fish. Journal of Experimental Biology, 211, 354-360.

Tomas, J., Aznar, F. J. & Raga, J. A. 2001. Feeding ecology of the loggerhead turtle Caretta caretta in the western Mediterranean. Journal of Zoology, 255, 525-532.

Vorobyev, M. 2003. Coloured oil droplets enhance colour discrimination. Proceedings of the Royal Society B: Biological Sciences, 270, 1255-1261.

Waitt, C., Little, A. C., Wolfensohn, S., Honess, P., Brown, A. P., Buchanan-Smith, H. M. & Perrett, D. I. 2003. Evidence from rhesus macaques suggests that male coloration plays a role in female primate mate choice. Proceedings of the Royal Society: Biological Sciences, 270, 144-146.

Widder, E. A. 2010. Bioluminescence in the Ocean: Origins of biological, chemical and ecological diversity. Science, 328, 704-708.

Witherington, B. E. 1994. Flotsam, jetsam, post-hatchling loggerheads, and the advecting surface smorgasbord. In: Proceedings of the 14th international symposium on sea turtle biology and conservation (Ed. by K. A. Bjorndal, A. B. Bolten, D. A. Johnson & P. J. Eliazar), pp. 166-168. Hilton Head, South Carolina: U.S. Dept. Commerce NOAA Tech Memo NMFS-SEFSC-351.

Witherington, B. E. 1998. Habitats and bad habits of young loggerhead turtles in the open ocean. In: Proceedings of the 18th international symposium on sea turtle biology and conservation (Ed. by F. A. Abreu-Grobois, R. Briseño-Dueñas, R. Márquez-Millán & L. Sarti-Martínez), pp. 34-35. Mazatlán, Sinaloa México: U.S. Dept. Commerce NOAA Tech Memo MFS-SEFSC-436.

Witherington, B. E. & Bjorndal, K. A. 1991a. Influences of artificial lighting on the seaward orientation of hatchling loggerhead turtles "Caretta caretta". Biological Conservation, 55, 139-149.

Witherington, B. E. & Bjorndal, K. A. 1991b. Influences of wavelength and intensity on hatchling sea turtle phototaxis: implications for sea-finding behavior. Copeia, 4, 1060- 1069.

Zar, J. H. 1999. Biostatistical Analysis. New Jersey: Prentice Hall.