SELECTIVE PRESSURES INFLUENCING COLOR-VISION IN NEOTROPICAL PRIMATES

SELECTIVE PRESSURES INFLUENCING COLOR-VISION IN NEOTROPICAL PRIMATES A thesis submitted to Kent State University in partial fulfillment of the requirements for the degree of Master of Arts by Eric S. Seemiller August, 2011

ii Thesis written by Eric Scott Seemiller B.S., The Pennsylvania State University, 2009 M.A., Kent State University, 2011 Approved by Marilyn Norconk, Advisor Richard Meindl, Chair, Department of Anthropology John R. D. Stalvey, Dean, College of Arts and Sciences ii

iii TABLE OF CONTENTS LIST OF FIGURES... iv LIST OF TABLES......v ACKNOWLEDGEMENTS... vi Chapter I. INTRODUCTION...1 The Front Side: Light Absorption in the Primate Eye...1 The Backside: Interpretation and Comparison of Signals...7 Explanations for the Evolution of Trichromacy in Primates...10 Research Objectives...12 II. METHODS...14 Experiment 1: Foraging Efficiency in Captive Primates...14 Methods and Materials...15 Experiment 2: Within-tree Foraging Preferences in Wild Howlers 19 III. RESULTS...23 Color-based foraging efficiency...23 Feeding preferences in free ranging howlers...28 IV. DISCUSSION...31 Conclusion...36 REFERENCES...37 iii

iv LIST OF FIGURES Figure 1.1: Diagram of the neural physiology of the primate vision subsystems...8 Figure 2.1: Experimental design used in foraging trials...16 Figure 2.2: Schematic of the concentric divisions of trees...22 Figure 3.1: Results of foraging trials for A. nigroviridis and C. wolfi...25 Figure 3.2: Results of foraging trials for S. sciureus...26 Figure 3.3: Results of foraging trials for A. caraya...27 Figure 3.4: Within-tree feeding preferences...29 Figure 3.5: Spatial distribution of feeding trees...30 iv

v LIST OF TABLES Table 1.1: Amino acid substitutions in vertebrate M/L opsin proteins...3 Table 1.2: Census of color-vision among mammals...5 Table 3.1: Results of within-tree feeding preferences and Chi-square tests...30 v

vi ACKNOWLEDGEMENTS Foremost, I must acknowledge my advisor, Dr. Marilyn Norconk, whose invaluable patience and expert guidance brought me to this point. She pulled me through academically and physically trying times and stuck with me when she didn t have to and for that I thank her. Additionally, I must thank the members of my committee, Drs. Chi-hua Chiu and Richard Meindl for their appreciated input in the completion of this thesis. Furthermore, I must thank Drs. Chris Vinyard, Chris Kuhar and Kenneth Glander for letting me use their monkeys and offering their scientific input along the way. Without Dr. Tremie Gregory and Eric Henthorn, who picked me up off the ground on more than one occasion, I would not be in this position and therefore special thanks are directed toward them. Also, I must thank my parents, Donald and Patti Seemiller, and my sisters, Megan and Carrie Seemiller, for not thinking I was crazy when I said that I wanted to study monkeys. vi

1 Chapter One Introduction One of the hallmarks of the primate lineage is an increased reliance on vision, compared with other mammals, often at the expense of the other special senses. Many adaptations have evolved to improve specializations in the primate visual system including binocular vision, increased foveal acuity, and trichromacy in many species (Walls 1942). Trichromacy, or the ability to perceive and interpret light of three different wavelengths (i.e. color-vision), has been the topic of intense interest over the last 30 years because, among mammals, it exists almost exclusively in the primate lineage. Even then, trichromacy in primates exhibits variable phenotypic expression: fixed in catarrhine primates and variable in platyrrhine and strepsirrhine primates (Bowmaker et al. 1991; Jacobs et al. 1996). With the help of genetics, histology and behavioral studies, the full range of variability is emerging (e.g., Jacobs 1993; Cropp et al. 2002; Talebi et al. 2006; Veilleux and Bolnick 2009; Bunce et al. 2011). This thesis seeks to address the variation of trichromacy in the platyrrhine infraorder, while focusing on the selective forces that either drove it to fixation in some species or maintains its variability in others. The Front Side: Light Absorption in the Primate Eye 1

2 Trichromacy, or the ability to see red, green and blue wavelengths discretely, is the phenotype exhibited with the expression of three different opsins in the retina, resulting from a genotype of either three different opsin genes or two different opsins genes with multiple alleles. Opsins are 364-amino-acid septa-transmembrane proteins that exist on the outer layer of cone cells in the retina of the chordate eye. These proteins serve to absorb light of specific wavelengths, activating the neural response of the cone cell. There are two classes of opsin proteins in mammals: the shortwave sensitive opsin (S) and the middle/longwave sensitive opsin (M/L). The former is sensitive to wavelengths nearing 430nm, primarily blue light, while the latter is sensitive to wavelengths between 535 and 562nm, which consists of yellow-green light. While the S opsin is fixed in all mammals with few exceptions (e.g., Aotus, see below), the M/L opsin exhibits considerable genotypic and phenotypic variability, which will be addressed later (Bowmaker et al. 1991; Jacobs 1993) Although opsins are highly conserved exhibiting little variation among taxa in extant mammal lineages, very specific amino-acid substitutions were discovered to yield discrete shifts in spectral sensitivity in primate M/L opsins (Neitz et al. 1991). That is, at three specific amino acid locations in an opsin protein, a substitution will shift the wavelength sensitivity of the molecule in a discrete and predictable way. Initially studied in humans and tamarins, the three amino acids are at sites 180, 277 and 285. More recently, the full range of variation in the primate M/L opsin sensitivity has been mapped (Shyue et al. 1998). Specific amino acid substitutions will yield predictable shifts in color perception from 535 to 562nm (Table 1.1). Despite this system of variation in the M/L 2

3 opsin, the S opsin remains largely invariable, exhibiting no noteworthy amino acid polymorphisms, except a loss-of-function mutation in the nocturnal Aotus (Jacobs et al. 1996) (see below). Sensitivity AA 180 AA 277 AA 285 562 nm Serine Tyrosine Threonine 556 nm Alanine Tyrosine Threonine 550 nm Alanine Phenylalanine Threonine 543 nm Alanine Tyrosine Alanine 535 nm Alanine Phenylalanine Alanine Table 1.1- Amino acid substitutions in mammal M/L Opsin proteins. Specific substitutions yield predictable shifts in spectral sensitivity (Neitz et al. 1991; Shyue et al. 1998). In the ancestral dichromatic mammalian condition, one autosomal S opsin gene and one X-linked M/L opsin gene are present, making mammals obligate dichromatic (Jacobs 1993). Obligate trichromats require a third distinct opsin gene. This is achieved in a two-step process. First, there must be allelic variation within the population. That is, the M/L opsin varies in the ways mentioned above, creating a third possible opsin allele within the population. This variation, because it is X-linked in mammals, allows the propensity for females to be trichromatic as they possess two X chromosomes and thus two M/L opsin alleles. Second, in order for males to exhibit trichromacy, gene duplication is necessary. The M/L opsin must duplicate and recombine, resulting in two M/L opsin genes on a single X-chromosome. After this has occurred in evolutionary history, with proper variation, both males and females will express three opsins. All catarrhines are trichromatic; the most recent common ancestor of catarrhines had one autosomal S opsin locus and experienced a duplication of the M/L opsin locus on 3

4 the X chromosome and recombination to create three unlinked opsin genes in both sexes (Nathans et al. 1986). These opsins have spectral tunings at 430, 535 and 562nm (Bowmaker et al. 1991; Hunt et al. 1998). However, there is tremendous variability in the platyrrhine lineage (Table 1.2). With two exceptions, all genera exhibit a sex-based phenotypic difference in color-vision (Jacobs 1984; Jacobs et al. 1996): two-thirds of females are trichromats while one-third of females and all males are dichromats. That is, there is allelic variation within the population, but they lack the genetic recombination necessary for a trichromatic phenotype in males. There are two exceptions to this widespread event in platyrrhines: Aotus (night monkeys), as mentioned above, has a loss-of-function mutation in its S opsin and is monochromatic (Jacobs 1993). Furthermore there is no allelic variation in the M/L opsin, and all night monkeys possess only a 535nm opsin. The other exception in platyrrhines occurs in the genus, Alouatta. All individuals of Alouatta are trichromatic, exhibiting the spectral tuning characteristic of the catarrhine lineage due to a lineage specific duplication of the M/L opsin locus on the X chromosome, leading to three unlinked opsin genes (Hunt et al. 1998; Jacobs et al. 1996). 4

5 Taxon Phenotype Comments and References Non-primate mammals Uniformly dichromatic Excepting marsupials (1) and primates, trichromacy is rare in the mammalian class (2) Catarrhines (including humans) Uniformly trichromatic Platyrrhines Atelinae (Ateles, Brachyteles, Lagothrix) Alouattinae (Alouatta) Dichromatic ; Polymorphic Uniformly trichromatic All possess the same three photopigments, allowing them to distinguish three colors (2) Three M/L opsin alleles exist in the lineage with tunings of 535, 550 and 562nm (3)(4) A gene duplication allows them the same genotype and phenotype as catarrhines (5) Pitheciinae (Pithecia) Dichromatic ; Polymorphic Cebinae (Cebus, Saimiri) Dichromatic ; Polymorphic Callitrichinae (Callithrix, Callimico, Cebuella, Saguinis, Leontopithecus) Dichromatic ; Polymorphic Only Pithecia has been studied, with the same alleles as Atelinae (6) Alleles for all possible opsins have been discovered, though frequencies of each are unknown (6)(7) Some individuals are functionally dichromatic despite being genetically trichromatic (6)(7)(8) Callicebinae (Callicebus) Dichromatic ; Polymorphic Possess only three alleles, similar to Atelinae.(9) Aotinae (Aotus) Monochromatic All individuals possess only one photopigment, making them monochromatic (2) Table 1.2- A census of color-vision among mammals. (1- Arrese et al. 2005; 2- Jacobs 1993; 3- Jacobs and Deegan 2001; 4- Talebi et al. 2006; 5- Jacobs et al. 2006; 6- Jacobs and Deegen 2003; 7- Cropp et al. 2002; 8- Surridge and Mundy 2002; 9- Bunce et al. 2011) 5

6 Although opsins are highly conserved at the amino acid level, especially at spectral tuning sites, there are some significant amino acid differences between catarrhine and platyrrhine opsin proteins (Hunt et al. 1998). This suggests that opsin gene evolution occurred after the Oligocene split of the primate infraorders. Because of the presence of trichromacy in all extant catarrhines, Hunt et al. (1998) propose that allelic variation and recombination occurred very early in the lineage and was quickly driven to fixation. Molecular dating in the platyrrhine lineage, however, shows that allelic variation stalled until 20 mya (Boissinot et al. 1998), and recombination occurred only in one lineage, the Alouattinae, at around 16 mya (Hunt et al. 1998). In addition to inter-specific opsin variation, cone concentrations are also variable among primates. S cones make up roughly 10% of the entire complement of cones with the remaining 90% consisting of one M/L cone type in dichromats or two M/L cone types in trichromats (Calkins 2001). However, individual parts of the eye display higher concentrations of cones in different primates. Cones are densest in the fovea region, where visual acuity is highest. Most primates, including humans, have ca. 200,000 cones/mm 2 (Curcio et al. 1991) in the fovea, but there is a tendency for platyrrhines to have lower foveal cone densities and slightly higher peripheral densities than catarrhines (Franco et al. 2000), although it remains largely unclear how this affects visual acuity. Again the two exceptions among platyrrhines are Aotus with maximum cone densities of only 20,000 cones/mm 2 (Yamada et al. 2001) and Alouatta, which has 376,000 cones/mm 2, nearly twice the density as the average primate (Finlay et al. 2008; Franco et al. 2000). The decreased concentration of cones in Aotus is believed to be beneficial for 6

7 low-light vision (Jacobs et al. 1993), however, it is unknown why Alouatta possesses such a high density of cones in the fovea, and if this impacts visual acuity. The Backside: Interpretation and Comparison of Signals There are two neural visual subsystems in primates (Figure 1.1). The first system is the ancestral mammalian subsystem that is present in all mammals and the second is the derived subsystem that augments the ancestral condition in primates. The ancestral subsystem consists of two classes of bipolar nerve cells that transmit information from S cones and M/L cones, terminating the signal at the Koniocellular layer of the lateral genticular nucleus (LGN) in the thalamus (Dacey and Lee 1994). These cells then compare the ratio of signals, known as blue-yellow opponency. The derived subsystem is made of midget ganglion cells that correspond specifically to M/L cones, with the signal terminating in the parvocellular layer of the LGN, where red-green opponency is calculated (Kremers and Lee 1998). 7

8 Figure 1.1- Diagram of the neural physiology of the primate vision subsystems. In the ancient subsystem, signals are activated through S and M/L cones and, after passing through bipolar cells, activate small bistratified ganglion cells en route to the Koniocellular layer of the LGN where blue-yellow opponency is calculated. In the derived primate subsystem, on M and L cones in the fovea centralis are activated, and the signal goes through midget ganglion cells to the Parvocellular layers of the LGN where red-green opponency is calculated (derived from Regan et al. 2001; Lee 2004). 8

9 The greater bush baby (Otolemur garnetti), an extant representative of the earliest primate lineages, lacks trichromatic vision but employs both subsystems, having bipolar cells and midget ganglion cells (Yamada et al. 1998), suggesting that the advent of the derived primate subsystem occured very early in the primate lineage. Additionally, there is no substantive difference in the backside physiology of vision between catarrhines and platyrrhines (Lee et al. 2000), nor is there one between dichromatic and trichromatic platyrrhines (Solomon 2002). This suggests that the comparison of trichromatic signaling probably co-opted a pre-existing existing primate visual system. It is likely that midget ganglion cells of the derived subsystem evolved as an adaptation to increase visual acuity in the fovea of primates en masse early in their evolutionary history and has since been used in a way to compare and interpret red and green light following the addition of a third opsin (Lee 2004). Furthermore, interpretation of red-green opponency in the ancestral system is dependent upon spectral tuning differentiation: parvocellular cells lack the ability to discriminate colors efficiently when comparing opsin signals that differ by fewer than 7nm, increasing in efficiency when opsin signals differ by more than 20nm (Blessing et al. 2004). For example, a female Saguinus with two M/L opsins tuned to 556 and 562nm (Jacobs and Deegen 2003) may be anatomically and genetically trichromatic, but functionally dichromatic as she is unable to interpret red-green opponency. The spectral difference between opsins tuned at 535 and 562nm is much greater and would allow her the ability interpret red-green opponency. 9

10 Explanations for the Evolution of Trichromacy in Primates The angiosperm radiation hypothesis (Sussman 1991) posits that the primate lineage evolved at the same time as flowering plants, offering many opportunities for coevolutionary relationships. One possible result of this interaction involves the evolution of frugivory and seed dispersal. In this scenario, frugivorous primates are rewarded with carbohydrate and nutrient dense, edible fruits that are often brightly colored in exchange for seed dispersal, as the primate will either carry the seed away from the plant, or most often defecate the seed some time after ingesting the fruit, far from the parent tree (e.g. Janzen 1969, 1970; Clark and Clark 1984). Although the evolution of visual cues to prospective seed dispersers comes at some cost to the plant, colorful pericarps are found in many plant families and the investment in a signal seems worth the benefit of seed removal. According to this hypothesis, color-vision phenotypes will be most advantageous when the possessor is efficient at foraging for colorful fruits, creating a major selective pressure in plant-vertebrate interactions. Mollon (1989) first hypothesized that color vision in primates had primary utility for finding fruit, offering seed dispersal as a reward for plants that invested in making more visible fruit. Through computational modeling, Osorio and Vorobyev (1996) demonstrated that trichromats should be better than dichromats at spotting red fruits contrasted against a green background. Regan et. al (2001) reiterated this finding with a more comprehensive model that included a wide analysis of food and non-food fruits along with foliage in French Guiana. Finally, Caine and Mundy (2000) demonstrated a 10

11 foraging advantage for trichromatic Callithrix geoffryi by simulating fruit-foraging situations. However, a number of researchers have failed to corroborate these findings in other species. For example, researchers have not found color-based fruit or insect foraging advantages in either Cebus capucinus (Melin et al. 2007; Melin et al. 2009), Ateles geoffroyi (Hiramatsu et al. 2008), or in a number of catarrhines including Pan troglodytes, Colobus guereza, Cercopithecus ascanius (Dominy and Lucas 2001) and Macaca fasicularis (Lucas et al. 1998). Other researchers have pointed to long-range visual acuity as the benefit of trichromacy, proposing that color-vision is most effective compared to dichromacy over a long distance (Caine and Mundy 2000). According to this hypothesis, trichromatic females should be more adept at spotting brightly colored fruits from a distance than her dichromatic conspecifics and should provide a social feeding advantage to groups with trichromatic females (Travis et al. 1988). However, troop leaders for groups that have been tested were found to be (obligate dichromatic) males, nullifying any long-range advantage trichromatic females might have (Saguinus imperator and Saguinus fuscicollis- Dominy et al. 2003; Saguinus mystax and Saguinus fuscicollis- Smith et al. 2003). Possible interpretations of these results include a) that there is no feeding benefit to being a trichromat as other cues (e.g., olfactory, tactile, etc.) are equally reliable; b) there is a visual benefit to foraging, but social factors can invalidate these benefits; c) groups accrue a wider variety of benefits (e.g. anti-predation) by having individuals with diverse visual capabilities in the group. 11

12 The finding that Alouatta, with a diet that is more than 50% leaves (Charello 1994; Julliot and Sabatier 1993), is the only uniformly trichromatic platyrrhine has led to a reorganization of evolutionary scenarios. If not red fruit, then perhaps red leaves provide the dietary justification for trichromacy. Dominy, Lucas and colleagues suggest that trichromacy has maximum utility for discerning young, red, nutrient-dense leaves from mature, tough green leaves (Dominy and Lucas 2001; Lucas et al. 1998; Dominy and Lucas 2004). This hypothesis suggests that there is a nutritious advantage in younger (red) leaves that would be beneficial to an animal whose trichromacy enables them to distinguish them from green background foliage. Additionally fewer achromatic cues (i.e. luminance, depth, tactile, scent, etc.) in leaves would make trichromacy vital for the detection of desirable young leaves (Hiramatsu et al. 2008). While trichromacy should benefit animals that can use color cues to detect edible fruits and leaves, there are some costs to giving up dichromacy. For example, dichromats are more efficient foragers in low-light conditions than trichromats (Caine et al. 2010, Yamashita et al. 2005) and actually monochromacy is predicted to be most efficient in low-light settings (Vorobyev et al. 2001). Additionally, dichromats are largely immune to camouflage. Caine and colleagues discovered that they were far more effective foragers when food is camouflaged with its background than trichromatic conspecifics (Caine and Mundy 2000; Caine et al. 2003). These two drawbacks suggest that trichromacy is not without expense and likely has a selective force maintaining it. Research Objectives 12

13 Two short studies were conducted here to assess foraging differences between dichromatic and trichromatic haplorhines. The first experiment addresses the hypothesis that dichromats and trichromats do not utilize color cues to forage for fruit-like objects. In this study conducted at Cleveland Metroparks Zoo, two male platyrrhines were compared and contrasted with two male catarrhines. I predicted that given the uniform nature of the food item, differing only in color, if the individuals rely on color discrimination they will express a preference for the foods that are presented in a colorcontrast situation. I also predict that if there is no difference in food choice, then they likely rely on achromatic cues so that both trichromats and dichromats will forage indiscriminately and without discrepancies in preference between red cereal and green cereal on a green background. The second study took place on free-ranging howling monkeys in Costa Rica. Because trichromacy loses efficiency in low-light settings, as trichromats lose foraging ability as light levels decrease (Caine et al. 2010; Vorobyev et al. 2001), I predict that uniformly trichromatic Alouatta palliata will forage preferentially in areas with higher levels of ambient light, i.e., the periphery of trees. 13

14 Chapter Two Methods Experiment 1: Foraging preference in captive primates at Cleveland Metroparks Zoo The goal of this experiment is to test whether trichromatic and dichromatic primates demonstrate any difference in foraging preference when provided with an identical food in which only color is changed. I hypothesized that color does not play an important role in fruit foraging. I predicted that there would be no difference in foraging preference between the two phenotypes and that both would forage indiscriminate to color. To test this, four species of primates were presented with foraging tests that sought whether color factored in food selection: two platyrrhines (Alouatta caraya and Saimiri sciureus) and two catarrhines (Allenopithecus nigroviridis and Cercopithecus wolfi). A. nigroviridis and C. wolfi are both catarrhines and therefore trichromatic (Bowmaker et al. 1991) as is A. caraya, of the family Alouattinae (Jacobs et al. 1996). Because S. sciureus females are polymorphic, only males were used to ensure that there was limited variability in color perception; these male primates are dichromatic (Jacobs 1984). 14

15 Materials and Methods All subjects are housed in the Primate, Cats and Aquatics building of the Cleveland Metroparks Zoo, Cleveland, Ohio. A total of 5 trials were presented to each individual or group at weekly intervals. Trials for A. nigroviridis, C. wolfi and S. sciureus occurred during March and April 2011, between 9:00 and 10:00 A.M. Because of other time commitments for zoo staff, trials for A. caraya occurred during May and June 2011 between 10:00 and 10:30 A.M. The test apparatus consisted of a plastic tray (42 x 25 x 10cm) lined with green construction paper and bisected with a fastened block of wood. Shredded green construction paper was layered in both sides of the tray to a uniform level to make food choices less obvious. Fruity Cheerios were placed on both sides of the tray. A total of 20 cereal (10 of each color) were used in each trial (Figure 2.1). 15

16 Figure 2.1- Experimental design used in foraging trials. Above is a grayscale representation, which demonstrates no difference in achromatic contrast between red and green cereal, with variable difference against the background. Below is the full color display. 16

17 Zoo employees isolated subjects from their enclosures briefly so that they could place the testing apparatus in a position that could be viewed by the researchers from outside the enclosure. Since this isolation is part of their daily routine for enclosure cleaning and maintenance, they are trained to transfer in such a way and were not handled in the course of the experiment. Though the catarrhines involved were tested alone, due to cage configurations, S. sciureus and A. caraya individuals were unable to be isolated for the experiment allowing other individuals in the group to interfere (see below). Once the experiment was completed, the subject was again transferred out of the enclosure so that the test apparatus could be removed. All remaining paper materials were discarded between trials. Generally, three species were tested on a single morning. Feeding was not restricted prior to the test, although that did not appear to affect the subjects interest in the experiment. The red and green sides of the tray were alternated with each trial, so that handedness or simple convenience did not have an effect on the results. Trials were filmed using a Panasonic video recorder on a tripod to prevent errors in data collection during the trials. Subjects were allowed to forage for a maximum of five minutes or until all food items were eaten. Adobe Photoshop (7.0) was used to ascertain luminance data of the two colors of cereal (Bergman and Beehner 2008). A color photograph was taken with a Nikon D3100 camera and a grayscale image was rendered. Using this image, achromatic percentages (i.e. the percentage of light and dark, with absolute black being 100%) were taken of both colors of cereal and the background. All cereal ranged between 48-52%, while the 17

18 background ranged between 54-60%. This ensures that the cereal does not differ in luminance, but only in color. Also, both colors are different than the background. Essentially, this means that both colors of cereal are equally bright, precluding the subjects from using this as a way to discriminate between cereal colors, but still allowing it to be used to discriminate from the background. Trials were scored in one of two ways. If there was not interference from other monkeys (in the case of A. nigroviridis and C. wolfi) rank-scores were calculated. To do this, the order in which each individual cereal item was eaten was documented. Then, each cereal was given a rank based on the order it was eaten, between 1 and 20. The first cereal eaten was ranked with a 1, the second was ranked with a 2 and so on. Two rank sum scores were summed, based on the two colors. Therefore, the rank-sum score for green cereal in a particular trial is the sum of the ranks for all green cereal eaten. A complementary red rank-sum score was calculated in the same way. Scores ranged from 55, if the subject ate all of that color before moving to the other color, (i.e., 1+2+3 +10 = 55) to 155, if the subject ate all of that color last (i.e., 11+12+13 +20 = 155). Lower scores for a color indicate that color was eaten first. Because these values are complementary, only the green rank-sums were used for analysis. Graphs were made for both subjects that showed the green-rank sum of each trial. T-tests were used to determine if the trials differed significantly from an expected mean of 105. The second option for quantification was similar. If interference from another individual occurred causing the subject to not eat all of the cereal in the test (as was the case for S. sciureus and A. caraya), the mean-rank sum was ascertained. To do this, the 18

19 rank-sums were calculated in the same way as the first option. However, this value was then divided by the total number of that color eaten during the trial. For example, if the subject only ate 5 red cereal, with a rank-sum of 35, this value was divided by 5, to arrive at a mean rank-sum of 7 for that color in that trial. These red and green values for each trial are not complementary, however they can be compared. The values were compiled into a chart for analysis. Again, a lower mean-rank sum indicates that the subject ate more of that color first. There are two scenarios in which the mean-rank sum could be a deceptive figure, although only one was encountered. If the subject did not eat any of one color, it would have a rank-sum of zero for that trial, as there would be no values to sum. If this happened, it was specifically noted. Additionally, if, for example, the subject ate only one or two of one color, this could generate a misleadingly low score and would need to be noted. However, this was not observed for any trial in this study. Experiment 2: Within-tree foraging preferences for wild howlers in Costa Rica Variable ambient light conditions are known to affect the ability of wild trichromats to perceive color variation in food items (Yamashita 2005). I hypothesized that trichromatic primates would attempt to behaviorally overcome the drawbacks of trichromacy. I predicted that monkeys would selectively feed in areas with higher light levels, such as the periphery of tree, to offset some of the poor low-light vision that comes with trichromacy. To test this, a habituated group of semi-free ranging mantled howling monkeys (Alouatta paliatta) was followed for 264 hours between July and 19

20 August 2010 in La Pacifica, Costa Rica (10 28 N, 85 07 W). The site is situated in the lowland tropical dry forest zone (Holdridge 1967) meaning that the forest changes dramatically seasonally, with a number of deciduous trees that lose their leaves. The study group resided on a 1330 ha cattle ranch and tilapia fish farm called Hacienda La Pacifica (Glander 1979), in which 42 distinct social groups have been recognized (Teaford and Glander 1996). Further, A. paliatta is the only nonhuman primate on the ranch (Teaford and Glander 1996). Though much of the forest on the ranch is connected, this study focused on Group 1, which inhabited a partially isolated 3 ha forest segment established as wind-break forest strip (Glander 1992). The forest fragment is surrounded by two pastures and a major road, which restricted the movement of the monkeys. Although there is a connection to larger forest patches, Group 1 remained cohesive and stayed inside this forest patch for the entire study period. Group 1 consisted of eight individuals: 1 adult male, 4 adult females, 2 juveniles and 1 infant. One adult female died during the study period due to a fall. The study occurred during the long wet season (June through December), predominately during the peak weeks of the wet season (mid to late July) (Glander et al. 1991). During the period daily rainfall ranged from 0 mm to 44 mm, with a daily average of 9.3 mm. The neardaily rainfall allowed all vegetation in the range to remain lush, effectively precluding food availability as a confounding factor in this study. Data were collected during a series of all-day follows, in which animals were followed from 6:00 to 18:00, approximately dawn to dusk. A random sequence of focal animals was determined each day and included all individuals, as the infant was weaned 20

21 and could eat on its own. This sequence was used to determine the order of observation once group feeding began in order to ensure that there was no bias toward individuals that were more easily observed. When feeding in the group commenced, the first individual to start eating was observed for 20 minutes or until it stopped feeding. There was no individual in the group who demonstrated any bias toward eating first, and thus this generally proved to be a random individual each time. Data were recorded at five-minute intervals including: within-tree location (defined below), tree ID number and food type. After the focal animal stopped feeding, or at 20 minutes, the next individual on the list was observed for 20 minutes in the same way. If the next individual was not feeding or could not be seen, the next individual in the observation order was chosen. This process continued until all feeding stopped. If one animal was observed to feed alone, either before or after a general group feeding time, data were recorded in the same way. To determine in which part of the tree the animal was feeding, each tree was divided into 4 concentric spheres: inner, middle, outer and edge (Figure 2.2). The edge was defined as the area that is within arms length of the outer-most part of the tree. The remainder of the tree was trisected into equal concentric layers. Also, food type was determined as fruit, leaf or other. Only trees that were visited more than once were included in the analysis of within-tree foraging preferences. This served to eliminate passive feeding as a confounding factor and to focus on deliberate feeding. Chi-square tests were used for each tree to determine if there was a significant difference between sections of the tree. Though there are volumetric differences between each section of the 21

22 tree, they were variable for each tree analyzed. Therefore, I assumed equal likelihood for each part of the tree as a null hypothesis (α =.05). Figure 2.2- Schematic of the concentric divisions of trees. This was used to determine the part of the tree in which a focal animal was feeding. It is important to note that in reality, the divisions would be hemispheric, and this is only a two-dimensional rendering. 22

23 Chapter Three Results Color-based Foraging Preference The results of the foraging preference trials satisfied the predictions of the hypothesis that primates are indifferent to color when foraging for foods with many properties similar to fruit. In five trials, neither trichromatic catarrhine showed a significant deviation from the expected green rank-sum value of 105 (A. nigroviridis: t= 0.729, p=.506; C. wolfi: t= 1.233, p=.285), meaning that they did not use color to discriminate the two halves of the test apparatus. A. nigroviridis showed a tendency toward extremes, generally eating all of one color first, which could be interpreted as a preference. However, if it was a preference, it changed between colors over the course of the five trials. Conversely, C. wolfi s results were distributed normally (Figure 3.1). Because females interfered with the dichromatic test subject male in S. sciureus the total green rank-sum could not be ascertained and compared against an expected value. Therefore the mean rank-sum of both colors in five trials is shown in Figure 3.2. As predicted, the subject showed no preference for either color. Like S. sciureus, the A. caraya subject could not be isolated and the mean rank-sum of both colors was used (Figure 3.3). Though the subject showed no interest in the test during trial one, he was receptive for the remaining four trials. The results mimicked those of the other primates in the study; the subject showed no preference for either color, and foraged 23

24 indiscriminately. Because there was no expected value or comparable mean, hypothesis testing was not possible for S. sciureus and A. caraya. 24

25 Figure 3.1- Results of foraging trials for A. nigroviridis and C. wolfi. The dotted line represents the expected results of the trial. A. nigroviridis tended toward extremes, while C. wolfi distributed normally around the expected value. Neither differed significantly from expectations. 25

26 Figure 3.2- Results of foraging trials for S. sciureus. Because females interfered with the test subject, the mean rank-sum for each color was used. With the exception of trial 1, in which the subject ate only green cereal, there is no difference between red and green cereal in any trial. Because the total number of cereal eaten in each trial was different, there was no expected value, and thus hypothesis testing was not possible. 26

27 Figure 3.3- Results of foraging trials for A. caraya. Despite no participation in the first trial, the remaining four trials went as expected, with no clear preference for either color. 27

28 Feeding Preferences in Free Ranging Howlers The spatial distribution of the trees is shown in Figure 3.5. The feeding samples are concentrated heavily in the northwest corner of the range, where 3 of the 4 most frequented trees were clustered (Trees #411, 67, 412). The remaining trees are distributed evenly throughout the range. Members of Group 1 were observed feeding on more than one occasion in seven different trees: 3 Enterolobrium cyclocarpum, Muntingia calabura, Mastichodendron tempisque, Maclura tinctoria, and one unknown species. The remaining observed feedings were deemed to be only passive feeding and not included in the analyses. Individuals were observed feeding from these trees a total of 360 times. Of these observations, 213 were from Tree 411 (Maclura tinctoria), 69 were from Tree 67 (Enterolobrium cyclocarpum) and the remaining 5 trees were each sampled fewer than 30 times (Figure 3.4). Howlers fed on the edge greater than 50% of the time in all trees except one, and percentages declined toward the inner part of the tree. Trees #411 (χ 2 =142.9, p<.0001), #67 (χ 2 = 79.1, p<.0001), #412 (χ 2 = 13.3, p =.001), #447 (χ 2 = 19.1, p<.0001), #627 (χ 2 = 6.4, p =.011), and #417 (χ 2 = 6.2, p =.045) were all significant for a feeding bias on the periphery of the tree compared with other portions of the tree. Conversely, only Tree #416 (χ 2 = 1.4, p =.497) showed no significant feeding bias (Table 3.1). 28

29 Figure 3.4- Within-tree feeding preferences. 29

30 Edge (%) 53.5 Outside (%) 34.3 Middle (%) 11.7 Inner (%) 0.5 Food n χ2 p Fruit 213 142.9 <.0001 Enterolobrium cyclocarpum Mastichodendron tempisque Unknown 69.6 21.7 7.2 1.4 Leaf 69 79.1 <.0001 71.4 21.4 7.1 0 Leaf 28 19.1 <.0001 70.0 25.0 5.0 0 Leaf 20 13.3.001 Muntingia calabura Enterolobrium cyclocarpum Enterolobrium cyclocarpum 90.0 0 0 1.0 Fruit 10 6.4.011 70.0 20.0 10.0 0 Leaf 10 6.2.045 30.0 50.0 20.0 0 Leaf 10 1.4.497 Tree ID 411 Species 67 447 412 627 417 416 Maclura tinctoria Table 3.1- Results of within-tree feeding preferences and Chi-square tests. Percentages of each tree sample are shown for each part of the tree. Of the seven feeding-trees visited more than once during the study, six showed a significant deviation from expected within-tree feeding frequencies. Figure 3.5- Spatial distribution of feeding trees. The tree locations are graded in size based on frequency of feeding. 30

31 Chapter Four Discussion The findings of this research corroborate the ideas presented by Lucas et. al (1998, 2003) and Dominy and Lucas (2001), in that color-based foraging preference is not readily demonstrated by primates when dealing with a three-dimensional object such as fruit. Though the experiments used cereal, it does mimic the texture and dimensionality that a primate might face when foraging for fruit in the wild. When foraging, there are numerous cues that a primate can consider: visual cues like color or depth, olfactory cues, tactile cues, etc. This experiment controlled many of these, because the cereal only differed in color. When everything else is the same, both trichromatic and dichromatic primates in this study seemed to ignore color when foraging, and ate the cereal indiscriminately with no preference for either color. However, when a primate is feeding on a flat object, such as a leaf, there are fewer non-chromatic cues available, as the forager often forgoes depth and possibly olfaction and texture depending on the species of leaf. Furthermore, young red tropical leaves are protein-rich and often less hard and easier to eat (Dominy and Lucas 2004) and unlike fruit that can ripen to a multitude of colors, they operate primarily on a red-green axis (Dominy 2004) almost requiring trichromacy for maximum foraging efficiency. But the majority of trichromatic primates do not subsist entirely, or even in any large capacity 31

32 on leaves. How then does folivory provide an adequately powerful selective force to favor trichromacy? For many primates, leaves are a fallback food. A consensus definition of fallback foods is that they are low-quality foods that which, despite augmenting the diet year round, are vastly important during seasonal periods when higher-quality foods are unavailable (Lambert 2007; Marshall et al. 2009; Wrangham et al. 1998). Though periods of fallback food subsistence can be rare (e.g. El Niño years), they are potent and are widely considered to have an intense influence on morphological adaptations: dental (Lambert et al. 2004; Lucas 2000; Rosenberger and Kinzey 1976), digestive (Davies et al. 1988; Porter et al. 2009; Remis et al. 2001), and now even visual. The experiment conducted here as well as others have demonstrated that color-vision is largely irrelevant when foraging for fruit. Therefore, the most parsimonious explanation for the evolution of trichromacy is that it is most advantageous during periods of fallback subsistence, when the primary food source is leaves. It is during this period that the capacity for distinguishing young, nutritious, red leaves from more mature green leaves is most essential, and also when trichromacy is most beneficial due to the increased reliance on chromatic cues when foraging. So what is so special about Alouatta that led to its trichromacy while other platyrrhines have maintained polymorphic color vision? The South and Central American rainforests are equally susceptible to seasonal variation and periods of fallback subsistence as the Old World rainforests, and often, due to coincidental flushing and fruiting, experience exaggerated fallback periods (van Schaik et al. 1993) forcing a 32

33 number of morphological adaptations for fallback subsistence (Rosenberger 1992). Though other Neotropical primates exploit an array of fallback foods (e.g. seeds, insects) howlers are much quicker to revert to fallback folivory subsistence. Greater than half of their diet consists of leaves (Di Fiore and Campbell 2007), causing them to undergo a suite of adaptations for folivory (Milton 1977; Teaford and Glander 1996). Then the question becomes which came first: trichromacy or folivory? There are two distinct possibilities: 1) Alouatta is the only taxon that managed the mutation necessary for uniform trichromacy, opening the door to such intense reliance on folivory or 2) the selective force for trichromacy is just simply higher in howlers than in other platyrrhine species because of a reliance on folivory. The fossil record of Neotropical primates contains a Miocene folivore similar to extant Alouatta from 12.6 mya called Stirtonia (Hershkovitz 1970). Molecular evidence, however, suggests that Alouatta split from the other Atelids at 16.0 mya (Cortes-Ortiz 2003). Because the same recombination is present in all extant Alouatta species, it almost certainly preceded the radiation of howlers, thus making it older than the oldest known fossil folivore in the lineage. This evidence suggests that trichromacy preceded other folivorous adaptations in Alouatta, and likely set the stage for a dependence on fallback folivory; trichromacy came first in Alouatta. If that is the case, folivory might not be the selective force favoring trichromacy, but instead a convenient by-product. My second hypothesis was also supported: in the wild howler study, subjects fed in the periphery of the tree more often than expected by chance. This could demonstrate a behavioral response to maximize utility in a morphological adaptation. Because 33

34 trichromatic primates forfeit foraging efficiency as light levels decrease (Caine et al. 2010; Yamashita et al. 2005), a trichromatic primate should feed in areas with maximum ambient light, like the outside of a tree. Though Marshall and Wrangham (2007) argue that behavioral adaptations are generally responses to primary food sources, this appears to be an exception. As hypothesis one reiterates, color-vision is not necessary for fruit foraging, thus there would be no need to maximize trichromatic efficiency when foraging for fruit. Therefore, this behavior is likely an adaptation to maximize visual efficiency for leaf foraging. One might argue that there are other benefits to feeding of the periphery of tree crowns, namely nutritional ones. For instance, the terminal buds in the periphery might offer more nutritious value. However, this experiment controlled for that; terminal buds are equally likely to exist in all parts of the tree, or at the very least are present in all parts of the tree. Further, there was no difference in texture or color between parts of the tree, two characteristics that can often be correlated with nutritious attributes (Lucas 2000). Though it is possible that there may be outstanding, uncontrollable advantages to peripheral feeding, this experiment has made sufficient attempts to control for them. Variation among food types is equally likely in all parts of the tree. This hypothesis requires further investigation in other primates. In order to verify the utility of peripheral feeding, dichromatic arborealists and polymorphic arborealists must be studied in the same context. If a dichromatic primate is also foraging on the periphery, it could nullify the results of hypothesis 2 of this thesis, at least in the context 34

35 of color-vision, and peripheral feeding would likely be a response to some extraneous cue, one for which I was unable to control. Though they help highlight the relationship of folivory and trichromacy in Alouatta, the results of these two studies do not satisfactorily explain the maintenance of polymorphisms in other platyrrhines. Though the lineage exhibits an array of fallback food-types, trichromacy is still maintained in females when available, and it has been that way for 20 million years, the age of the allelic variation (Boissinot et al. 1998). By exploiting alternative fallback niches, according to my findings, trichromacy should not be excessively beneficial to most Neotropical primates. This leads to two possibilities: 1) the trait is either drifting, with no significant purifying selection or 2) the trait is extrinsically beneficial to all possessors and there is merely a mutational barrier which Alouatta and Old World monkeys have overcome. The results of the second study, in which howlers behaviorally counteracted a disadvantage to trichromacy, suggests that downfalls of trichromacy are real and have the potential to be powerful purifying selective agents. If these disadvantages are strong enough, this would preclude the first possibility and lead to more questions: if the heterozygote advantage in polymorphic platyrrhines is not for eating fruit, or insects (Melin 2007), is it for another food type, or something altogether different? The catarrhine lineage seems to have exploited trichromacy for a diverse set of diets, suggesting that it might supersede feeding altogether. Though much of the research in trichromacy is relevant to feeding, it is possible that predator defense or sexual selection offer selective pressures (Bradley and Mundy 2008), and it is in this direction that some future research should be directed. 35

36 Conclusion Though the findings of this thesis are not in conflict with any recent findings, they do lead to more questions. These results, and others, have demonstrated that frugivory is not substantially impacted by a trichromatic phenotype. One benefit of trichromacy is that it allows the exploitation of a leaf-eating niche, which happened in Alouatta after recombination led to its fixation. However, its maintenance in platyrrhines that do not heavily rely on leaves is baffling. Further, trichromacy has several apparent disadvantages that must be overcome. Based on that, is unlikely that genetic drift has led to color-vision polymorphisms, and the phenotype has been actively maintained by some selective force. Additional research is required to paint a complete picture on the selective benefits of trichromacy. Future research in this field has a macro and micro approach. Additional ecological research is required to analyze the patterns of color-based foraging, which can be a murky endeavor as food color has many confounding attributes that must be considered (e.g. blue-yellow opponency, luminance, ripening, etc.). Other aspects of color-vision ecology must also be analyzed. Are there benefits outside of feeding, like predator defense, sexual selection or conspecific recognition, that are aided by trichromacy? Likewise, advances in retinal mapping and topography have opened the door to a micro approach. Retinal mapping on a broad sample of primates will provide a more complete picture on the physiological nature of color-vision. Both of these approaches must be considered as the study of primate vision moves forward. 36

37 References Arrese CA, Oddy AY, Runham PB, Hart NS, Shand J, Hunt DM, and Beazley LD. 2005. Cone topography and spectral sensitivity in two potentially trichromatic marsupials, the quokka (Setonix brachyurus) and quenda (Isoodon obesulus). Proceedings. Biological sciences / The Royal Society 272:791-6. Bergman TJ, and Beehner JC. 2008. A simple method for measuring colour in wild animals: validation and use on chest patch colour in geladas (Theropithecus gelada). Biological Journal of the Linnean Society 94:231-240. Blessing EM, Solomon SG, Hashemi-Nezhad M, Morris BJ, and Martin PR. 2004. Chromatic and spatial properties of parvocellular cells in the lateral geniculate nucleus of the marmoset (Callithrix jacchus). The Journal of Physiology 557:229-45. Boissinot S, Tan Y, Shyue SK, Schneider H, Sampaio I, Neiswanger K, Hewett-Emmett D, and Li WH. 1998. Origins and antiquity of X-linked triallelic color vision systems in New World monkeys. Proceedings of the National Academy of Sciences 95:13749-54. Bowmaker JK, Astell S, Hunt DM, and Mollon JD. 1991. Photosensitive and photostable pigments in the retinae of Old World monkeys. Journal of Experimental Biology 156:1-19. 37