Colour vision in fallow deer: a behavioural study

ANIMAL BEHAVIOUR, 2001, 61, 367–371 doi:10.1006/anbe.2000.1603, available online at http://www.idealibrary.com on Colour vision in fallow deer: a behavioural study BJO } RN BIRGERSSON, ULRIKA ALM & BJO } RN FORKMAN Department of Zoology, Stockholm University (Received 28 January 2000; initial acceptance 23 March 2000; final acceptance 30 September 2000; MS. number: 6480) To examine whether fallow deer, Dama dama, have colour vision, we trained four adult females on a two-choice discrimination task, where a positive chromatic stimulus (green) and a negative achromatic stimulus (grey) had similar brightness. The criterion for learning was set at 80% correct responses. To exclude the possibility that the hinds used small differences in brightness to distinguish between the green and the grey stimulus, we conducted a test that differed from the training situation. A light green positive stimulus combined with a dark grey negative stimulus was alternated with a dark green positive stimulus combined with a light grey negative stimulus on every second trial. The positive green stimuli had different reflectance spectra. After training, each of the four hinds showed over 80% correct responses in the test. These results suggest that fallow deer can use colour vision in a discrimination situation by generalizing over slightly different colours, at least in the range of the green spectrum.  for brightness by using either one chromatic and one achromatic stimulus with equal brightness (e.g. Buchenauer & Frisch 1980) or one chromatic stimulus and different achromatic stimuli with more or less brightness (e.g. Zacks & Budde 1983; Smith & Goldman 1999; but see Macuda & Timney 1999 for an alternative method). For the majority of species, however, we do not know the maximum sensitivity of the rods or how skilful animals are at making brightness discriminations. Since animals can, to a large extent generalize over similar perceptual stimuli (Hanson 1959), including wavelengths (Goldsmith 1990; Shettleworth 1998), a more powerful way of controlling for brightness would be to conduct a two-choice discrimination test with different chromatic and achromatic stimuli both varying in brightness. We investigated whether fallow deer are able to see colour, by training hand-reared individuals on a twochoice discrimination task. Two visual stimuli, one negative achromatic (grey) and one positive chromatic (green), with similar brightness were used during the training period. We then did an additional two-choice test, where a light green positive stimulus combined with a dark grey negative stimulus was alternated with a dark green positive stimulus combined with a light grey negative stimulus on every second trial. If the subjects chose the positive green stimuli significantly more often than the negative grey stimuli, this would show that fallow deer can generalize over similar colours and would exclude the possibility that the choice is based on other cues (e.g. brightness). The phenomenon of colour vision in mammals has been investigated by both physiological and behavioural approaches. Studies of the retina by electroretinogram have revealed that the majority of mammals have at least two different photoreceptors sensitive to different wavelengths, which is presumably a fundamental condition for colour vision (Jacobs 1993). One problem in such studies is that cone photoreceptors that are rare can remain undetected, which can lead to different studies of the same species giving conflicting results (see Jacobs 1993; Hemmi 1999). Some features of colour vision cannot be explained with the current di/ trichromatic colour theory (see Coren & Ward 1989; Reitner et al. 1991). Hence, one cannot confirm the existence of colour vision from physiological studies of the retina alone. Independently of the physiological mechanisms for colour vision, an animal must have a nervous system capable of processing and reacting to the information from photoreceptors or similar perceptual structures. Consequently, a behavioural study is the ultimate method to obtain more conclusive evidence for colour vision in mammals. Such a study should be based on a discrimination test, where choices between visual stimuli are based exclusively on colour (Goldsmith 1990). The crucial point in such a study would be to make brightness an impossible cue for making the choice. Earlier investigations have attempted to control Correspondence: B. Birgersson, Department of Zoology, Stockholm University, 106 91 Stockholm, Sweden (email: [email protected]). 0003–3472/01/020367+05 $35.00/0 2001 The Association for the Study of Animal Behaviour 367  2001 The Association for the Study of Animal Behaviour 368 ANIMAL BEHAVIOUR, 61, 2 (a) Wooden containers Start box 4m (b) the distance from the start box to each of the containers was 9 m. During training the choice stimuli consisted of six medium dark green plates and six medium dark grey plates, which could be fastened on to the flap doors of the containers. During the test the choice stimuli consisted of six light green, six dark green, six light grey and six dark grey plates. We used six plates of each stimulus to minimize the risk that the hinds could use any feature of a specific painted plate to discriminate between the stimuli. The colours were ‘acrylatex-colour for outdoor use’. The reflection spectrum (Fig. 2) for each stimulus was measured with a spectrometer (Ocean Optics S1000, used with a deuterium/halogen light source which approximates daylight). The reward for the positive stimulus consisted of 6 g of raisins and 20 cm3 of pellets. To control for any odour effect, both of the containers always contained a reward, but the flap door was locked for the negative stimulus (i.e. nonrewarded grey colour). During training and testing, both the rewarding and the nonrewarding box were treated identically with respect to the opening and closing of the lid, and the changing of the coloured plates. Training Figure 1. Schematic illustrations of (a) the experimental enclosure and (b) one of the wooden containers on to which the choice stimuli were fastened. METHODS Study Animals and Material We conducted the study at Tovetorp zoological field station in south-central Sweden from June to August 1999. Six 3-year-old tame fallow deer hinds, Dama dama, originally reared by hand in 1996 for other purposes (see Birgersson et al. 1998), were used. During the study the hinds were kept in a 4-ha enclosure with woodland, pasture and access to water and a salt stone. Throughout the study, except during the experimental sessions, the animals could forage ad libitum in the enclosure. The experimental enclosure (100 m2) was placed within the larger enclosure (Fig. 1a). The walls of the experimental enclosure that faced the larger enclosure were solid, so that there was no risk of observational learning by hinds not participating in the experiment. A hind could be let in to the test area from the start box by means of a double door. Two identical wooden containers were placed in the test area, each measuring 3531 cm and 50 cm high. Each container had a flap door (3630 cm) which could be locked (Fig. 1b). The lock was placed inside the box (reached through the top of the box) and so could not be seen from the outside. The containers were 5 m apart and The hinds were habituated to the experimental enclosure for 14 days, then trained to open the containers. During training, we kept the flap door of one of the containers partially open, with the reward clearly visible. The rewarding container had a medium green plate fastened on the flap door (the positive stimulus); the other flap door, which was locked, had a medium grey plate (the negative stimulus). The flap door of the rewarding container was gradually lowered and the procedure repeated until the hinds could walk into the enclosure and open the flap door. After 2 days of training four of the hinds fulfilled the criteria, while two did not. These two were excluded from the rest of the experiments. The next stage was discrimination training, when the hinds were rewarded for choosing a container with a medium green over one with a medium grey plate. During the discrimination training we swapped the positions of the containers randomly from one day to another. The six medium green and six medium grey plates were changed for each trial according to a semirandom protocol (i.e. the plates were changed randomly, but with each colour on one side no more than three times in a row). Each session continued until the hind refused to go back to the starting pen within 10 min (this resulted in each session consisting of between 1 and 17 trials). We used the odour of pellets to call in the hind to the starting pen. A new trial was initiated when the hind had been in the starting pen for at least 1 min. The mean time for a trial was 2.5 min. A hind was said to have made a choice when she had touched one of the plates with her nose. Since the hinds showed a strong tendency to alternate between containers from one trial to another, irrespective of whether they had received a reward, we fixed the BIRGERSSON ET AL.: COLOUR VISION IN FALLOW DEER Dark green 60 50 40 30 20 10 0 300 400 500 600 Dark grey 700 800 60 50 40 30 20 10 0 300 800 60 50 40 30 20 10 0 300 800 60 50 40 30 20 10 0 300 400 Reflectance (%) Medium green 60 50 40 30 20 10 0 300 400 500 600 400 500 600 600 700 800 700 800 700 800 Medium grey 700 400 Light green 60 50 40 30 20 10 0 300 500 500 600 Light grey 700 400 500 600 Wavelength (nm) Figure 2. The reflectance spectra for each of the six colours. Values given are the mean of five measurements on one plate of the relevant colour. The mean standard deviation as a percentage of the total mean value for each colour was: dark green 24%; medium green 29%; light green 19%; dark grey 72%; medium grey 25%; light grey 10%. location of the plates until the hind had chosen the green side for two consecutive trials. We started doing this at the same day for all hinds (after 50–70 trials for each hind). After ca. 165 trials/individual (4–7 days), we reverted back to the semirandom protocol. The criteria for starting the test sessions, after we reverted to the original training procedure, was that the hinds had completed 10 sessions with 10 trials per session and 80% correct choices or more in each session. Trials were conducted between 0700 and 1700 hours GMT when daylight conditions were relatively constant. The discrimination training until the test sessions lasted for ca. 15 days. Depending on the specific spectral sensitivity of the species, there is always a risk with this experimental design that the animal perceives the brightness of the reflected light from the stimuli differently from humans (see Endler 1990). This seems unlikely for fallow deer, however, as they have rods with a maximum sensitivity of about 497 nm (498 nm for humans) and two classes of cone pigments with maximum sensitivity between 450–460 nm (420 nm for humans) and 530–550 nm (534+565 nm for humans), respectively (Jacobs et al. 1994). Test All the hinds chose the green plate in 80%, or more, of the trials, during the first test series. Over all the series the hinds chose the green plates on average more than 80% of the time (Fig 3; difference from random: hind A:
21 =67; B:
21 =64; C:
21 =55; D:
21 =97; P for all <0.001). Errors were more often made when the hinds had a choice between a light green and a dark grey plate than when the choice was between dark green and light grey plates (Table 1; hind A:
21 =10, P<0.001; B:
21 =9, P<0.003; C:
21 =7, P<0.009; D:
21 =4.5, P<0.03). Nevertheless, they chose the green plate in at least 80%, or more, of the trials for both two-choice alternatives. The total number of errors made in the light green/dark grey combination during the first test series was six out of 20 trials (
21 =3.2, To exclude the possibility that the hinds used small differences in reflectance, or perceived brightness, to distinguish between the medium grey and the medium green plate, we tested their ability to see colour by using six light grey, six dark grey, six light green and six dark green plates. These were combined so that a light grey plate was paired with a dark green, and a dark grey with a light green. As before, the selection of the green plate was rewarded. The two combinations were alternated from one trial to the next. The location of the green plate (the positive stimulus) was changed according to the semirandom protocol. RESULTS 369 ANIMAL BEHAVIOUR, 61, 2 brighter or darker than the corresponding negative stimulus, and consisted of two different green colours, which differed from the one used during the training period. This suggests that the fallow deer hinds made their choice by generalizing over slightly different colours. Perceptual generalization is a widespread phenomenon among animals and is probably also important for the ability to see colour (Shettleworth 1998). All four hinds performed significantly better when they had to discriminate between dark green and light grey than between light green and dark grey, although they showed more than 80% correct responses for both of the two-choice alternatives. During the training period, the subjects may have experienced the positive stimulus as somewhat darker than the negative stimulus, either because of a minor difference in reflectance (see Fig. 2) or because of a slightly different spectral luminosity function (see Jacobs 1981; Endler 1990). If so, the subjects could have continued to choose the darker stimulus in the light grey/dark green combination and then learned to avoid the even darker stimulus in the dark grey/light green combination. This seems unlikely, however, since the errors made in the first test series were almost significantly fewer than expected by chance and thereafter significantly fewer. It is unlikely that after five trials the subjects could have learned to choose the darker stimulus when the difference was small and avoid it when the difference was large. Nor is it likely, although not impossible, that this result depends on a slighly different perceived brightness for fallow deer, since the difference in reflectance between light green and dark grey was much larger than between dark green and light grey (Fig. 2). As long as our knowledge of the visual mechanisms in fallow deer is limited, the explanation for this asymmetry remains open. Although the majority of physiological studies have shown that the mammalian retina contains at least two classes of cones sensitive to different wavelengths (a prerequisite for colour vision), there are very few behavioural studies confirming that these two classes of cones are important parts of a larger colour visual system that makes an individual able to discriminate visual stimuli solely on wavelength. The advantage of having colour vision is probably at least partly founded on the ability to discriminate between different plant species, or parts of plants, that presumably vary in nutrients and toxins. Diet selection in browsing mammals involves visual, olfactory, gustatory and tactile mechanisms (Bryant et al. 1991; Augner et al. 1997); by using colour as an additional cue to make a correct choice, fallow deer, categorized as a 100 Correct choices (%) 370 75 A B C D 50 25 0 1 2 3 4 5 6 Series 7 8 9 10 Figure 3. Percentage of correct choices over the 10 test series for hinds A–D. One series consisted of 10 trials. P=0.07); for test series 2–10 the total numbers of errors were 4, 3, 5, 5, 2, 4, 2, 1 and 3, respectively. DISCUSSION After the training period, the fallow deer hinds were able to choose correctly between a positive green and a negative grey stimulus with similar brightness. In the following two-choice test, the hinds chose the green stimulus independently of whether it was lighter or darker than the corresponding grey stimulus. Our conclusion is that fallow deer can use colour to discriminate between two visual stimuli. This is the first behavioural study of fallow deer that has investigated their ability to respond to colour. All four individuals were able to discriminate green from grey independently of brightness. This finding is in line with the results of a physiological study of the retina by retinogram, where it was reported that fallow deer have two types of cones: one concentration of sensitive cones in the range of blue wavelengths (450–460 nm) and a corresponding peak in the range of green wavelengths (530–550 nm, Jacobs et al. 1994). Among other ungulates, the majority of reported indications for colour vision are indirect evidence from studies of the retina. In whitetailed deer, Odocoileus virginianus (Jacobs et al. 1994), pigs, Sus domestica (Neitz & Jacobs 1989), horses, Equus caballus (Sandmann et al. 1996), sheep, Ovis aries, goats, Capra hircus, and cows, Bos taurus (Jacobs et al. 1998), the retina contains two classes of cone pigments. Hence, the majority of authors seem to agree that ungulates are most likely dichromats. All hinds showed 80%, or more, correct responses during the first 10 trials, immediately after the training period. In this test the positive stimulus was either Table 1. Number of choices of each stimulus during the test sessions Hind A B C D Light green Dark grey χ21 P Dark green Light grey χ21 P 40 41 43 41 10 9 7 9 18 20 26 20 <0.001 <0.001 <0.001 <0.001 50 50 50 48 0 0 0 2 50 50 50 42 <0.001 <0.001 <0.001 <0.001 BIRGERSSON ET AL.: COLOUR VISION IN FALLOW DEER selective grazer (Hofmann 1989), can probably forage more effectively. For a typical prey species such as fallow deer, colour vision may also enhance predator detection. The most conclusive evidence for dichromatic colour vision in ungulates comes from three comprehensive studies of the horse. These studies differ from ours in that the most troublesome discrimination appeared to be in the green wavelength area (Pick et al. 1994; Macuda & Timney 1999; Smith & Goldman 1999), perhaps because these studies used monochromatic light. For a dichromat, the monochromatic green of a particular hue may fall close to the neutral point making the discrimination between grey and green very difficult. Considering the great radiation in the evolution of colour vision in mammals (Goldsmith 1990) it seems likely that these two ungulates from different genera evolved cones sensitive to different wavelengths. If so, we need to find the different selective advantages of responding to these different wavelengths for fallow deer and horses. It is first necessary, however, to examine the nature of colour vision in more detail in both species. The question of whether mammals can see colour has a long history. The answers have shifted dramatically since Walls (1942) concluded that ‘within the mammals, colour vision is by no means widespread’. Jacobs (1993) stated that almost all species studied in detail have shown an ability to see colour. Our study supports this view. Acknowledgments We are grateful to Staffan Andersson and Birgitta Tullberg for helping us with the spectrometer, to Olle Brick, Olof Leimar and an anonymous referee for constructive criticisms on the manuscript and to Anders Bylin for practical help. References Augner, M., Provenza, F. D. & Villalba, J. J. 1997. A rule of thumb in mammalian herbivores? Animal Behaviour, 56, 337–345. Birgersson, B., Tillbom, M. & Ekvall, K. 1998. Male-biased investment in fallow deer: an experimental study. Animal Behaviour, 56, 301–307. Bryant, J. P., Provenza, F. D., Pastor, J., Reichardt, P. B., Clausen, T. P. & du Toit, J. T. 1991. Interactions between woody plants and browsing mammals mediated by secondary metabolites. Annual Review of Ecology, 22, 431–446. Buchenauer, D. & Fritsch, B. 1980. Color perception in domestic goats. Zeitschrift für Tierpsychologie, 53, 225–230. Coren, S. & Ward, L. M. 1989. Sensation and Perception. 3rd edn. San Diego: Harcourt Brace Jovanovich. Endler, J. 1990. On the measurement and classification of colour in studies of animal colour patterns. Biological Journal of the Linnean Society, 41, 315–352. Goldsmith, T. H. 1990. Optimization, constraint, and history in the evolution of eyes. Quarterly Review of Biology, 65, 281–322. Hanson, H. M. 1959. Effects of discrimination training on stimulus generalization. Journal of Experimental Psychology, 58, 321–334. Hemmi, J. M. 1999. Dichromatic colour vision in an Australian marsupial, the tammar wallaby. Journal of Comparative Physiology A., 185, 509–515. Hofmann, R. R. 1989. Evolutionary steps of ecophysiological adaption and diversification of ruminants: a comparative view of their digestive system. Oecologica, 78, 443–457. Jacobs, G. H. 1981. Comparative Color Vision. New York: Academic Press. Jacobs, G. H. 1993. The distribution of colour vision among the mammals. Biological Reviews, 68, 413–471. Jacobs, G. H., Deegan, J. F., Neitz, J., Murohy, B. P., Miller, K. V. & Marchinton, R. L. 1994. Electrophysical measurements of spectral mechanisms in the retinas of two cervids: white-tailed deer (Odocoileus virginianus) and fallow deer (Dama dama). Journal of Comparative Physiology A, 174, 551–557. Jacobs, G. H., Deegan, J. F. & Neitz, J. 1998. Photopigment basis for dichromatic colour vision in cows, goats and sheep. Visual Neuroscience, 15, 581–584. Macuda, T. & Timney, B. 1999. Luminance and chromatic discrimination in the horse (Equus caballus). Behavioural Processes, 44, 301–307. Neitz, J. & Jacobs, G. H. 1989. Spectral sensitivity of cones in an ungulate. Visual Neuroscience, 2, 97–100. Pick, D. F., Lovell, G., Brown, S. & Dail, D. 1994. Equine color perception revisited. Applied Animal Behaviour Science, 42, 61–65. Reitner, A., Sharpe, L. T. & Zrenner, E. 1991. Is colour vision possible with only rods and blue-sensitive cones? Nature, 29, 798–800. Sandmann, D., Boycott, B. B. & Peichl, L. 1996. Blue-cone horizontal cells in the retinae of horses and other equidae. Journal of Neuroscience, 16, 3381–3396. Shettleworth, S. 1998. Cognition, Evolution and Behavior. New York: Oxford University Press. Smith, S. & Goldman, L. 1999. Colour discrimination in horses. Applied Animal Behaviour Science, 62, 13–25. Walls, G. L. 1942. The Vertebrate Eye and its Adaptive Radiation. Bloomfield Hills, Missouri: Cranbrook Institute of Science. Zacks, J. L. & Budde, W. 1983. Behavioural investigations of colour vision in the white-tailed deer (Odocoileus virginianus). Investigative Ophthalmology and Visual Science, 24, 183. 371