Universidad de los Andes – Carolina Urbina–Tesis de grado para obtener el título de bióloga – Director: Andrés Link–Sustentada 26 de Junio de 2014 Ranging patterns of brown spider monkeys (Ateles hybridus) in a fragmented forest in northern

Key words: Levy pattern, Ateles hybridus, power law distribution, turning angle, step lenght

Abstract

Recent studies on wild spider monkeys have described that their ranging and foraging patterns resemble what is best known in physics as levy-walk patterns. Here we analyzed the ranging patterns of brown spider monkeys (Ateles hybridus) in a small forest fragment using maximum likelihood methods for adjusting power law distributions to step length and resting period times. Daily paths were analyzed using the CPT (change points detection test). Our results suggest that spider monkeys living in a small fragmented forest do not follow Levy walk patterns and that their ranging and foraging strategies seem to be more goal-oriented. As proposed for other , spider monkeys may adjust their ranging behavior to restrictive situations such as habitat fragmentation. Thus, under this scenario spider monkeys may modify their ranging strategies, and this may differ from the results obtained in studies conducted in continuous forests. Thus, we suggest forest fragmentation restricts ´s movement patterns as well as male and female use of their territories.

Introduction

A large number of recent studies have provided consequences of habitat fragmentation on the evidence on the direct and indirect impacts of behavioral ecology of primates. habitat fragmentation on populations and wildlife (Aggimarangsee, 2013). Habitat The brown spider monkey (Ateles hybridus) is fragmentation has four major effects: [I] direct considered to be Critically Endangered (CR) and facing a relatively high risk of extinction (IUCN, decrease on the available habitat, [II] increase in the number of habitat patches, [III] decrease in 2008). It has been included on the 25 most habitat patch size, and [IV] increase in patch of primates and in need of attention for conservation and research isolation (Fahrig, 2003). As the ecosystems where primates currently live are transformed by (Mittermeier et al., 2012). The slow life history anthropogenic activities, species also adapt and variables of brown spider monkeys limit their potential flexibility to adapt to habitat loss and modify the way they use their habitat and exploit the available resources. For example, recent fragmentation. Spider monkeys have extremely studies have found that in smaller fragments, long reproductive cycles; are large-bodied ripe fruit specialists and need large areas to obtain primates travel shorter daily distances, revisit feeding trees more often, and use the forest in a resources for their survival and reproduction (Link more uniform manner (Boyle et al, 2013). et al., 2010). For these primates, an efficient foraging strategy is vital to the survival, especially Although there are several long-term studies on the dynamics of fragmented forests like “The in fragmented forests. The Levy-walk is known as Biological Dynamics of Forest Fragments Project” an efficient foraging strategy during searching for new resources. A Levy-walk is a type of random in Brazil (Boyle et al., 2013), and “Los Tuxtlas” in Mexico (Crtistobal-Azkárate et al., 2013), there is walk in which step lengths are drawn by a power still a gap in our knowledge on the effects and law distribution and turning angles are shaped by a uniform distribution (James et al., 2011). A power law distribution has the general form p(x) use CPT test to detect places where Ateles ~ x -α where α is a constant parameter of the hybridus change the direction within their daily distribution known as the scaling parameter. The paths and to examine whether these points are scaling parameter typically lies in the range related to the visitation of key feeding resources. 2 < α ≤ 3 although there are occasional exceptions These results are then discussed towards better (James et al, 2011). The power law applies only understanding which ecological aspects of habitat for values greater than some minimum xmin. In fragmentation alter spider monkeys movement such cases it is said that the tail of the distribution and foraging patterns. follows a power law (Clauset et al., 2009). The foraging strategies of Ateles geoffroyi have been Methods suggested to follow Levy walk patters, and this is Study site and reflected in the power law distribution of their Data were collected at the Hacienda San Juan de steps which had a scaling parameter of 2.18 and Carare (06 ° 43 'N, 74 ° 09' W), located near the the distributions of rest intervals with exponent of Magdalena river, within the department of 1.7 (Ramos-Fernandez et al., 2004). Santander in Colombia. The study site is within a The Levy hypothesis does not consider the seasonally flooded forest of roughly 65 ha and cognitive ability of individual foragers, and several fragmented about eight years ago due to selective studies have suggested memory can be a logging and the opening of pastures for cattle determinant factor in an efficient foraging ranching. It is comprised by a thin forest fragment strategy of spider monkeys (Di Fiore & Suares, of riverine forest where four species of primates 2007). Even when food patches are common, live: the brown spider monkey (Ateles hybridus), primates might rely on spatial memory to travel to red howler monkeys (Alouatta seniculus), white- highly productive or preferred resources, rejecting fronted capuchins (Cebus albifrons) and night less productive but closer resources (Cunningham monkeys (Aotus griseimembra) (Link et al., 2010). & Janson, 2007). Trying to investigate the spatial Two groups of A. hybridus were habituated to mental representation of primates has multiple researchers since 2007 and studied continuously limitations. The most common observation is that thereafter. Data analyzed for this study comes the paths used by primates when foraging among from two social groups: SJ1 (six females and five spatially fixed resources often appear to be goal- males) and SJ2 (five females and three males). oriented. In nearly all cases, routes taken between known resources are approximate to straight lines Data Collection (Janson & Byrne, 2007). Byrne et al., 2009 Data were collected from January 2010 to designed the (CPT) test, which serves to test December 2012. All monkeys were recognized by hypothesis associated with goal oriented paths. facial markings and unique anatomical This test examines paths composed of short linear characteristics. We conducted behavioral follows segments (vectors) to detect points at which the from dawn to dusk (when possible) and collected direction of these vectors changes significantly, data using focal sampling (Altmann 1974). that is, ‘change-points’ in otherwise more uniform The behavior of the focal subject was recorded at or undirected travel (Asensio et al, 2011). five minute point records. The spatial position was recorded every 15 minutes based on The main goal of this study is to characterize the georeferenced landmarks set at 25m intervals movement patterns of Ateles hybridus in a along the project´s trail system and over 2000 fragmented forest and identify which aspects of feeding trees. All feeding events were registered, the path differ with respect to those reported for taking note of the plant species eaten, the item spider monkeys living in continuous forests. We

2 consumed (e.g., fruits, flowers, etc.), the location were obtained by calculating the angles of the event, and the beginning and ending time associated with two consecutive vectors. A of each feeding bout. frequency distribution of the angles of rotation was produced with intervals size of 10 °. Data analysis All statistical analysis was performed using R Travel change-point detection 3.1.1. The probability distributions fit was The change point detection test (CPT) was created performed with a maximum likelihood method by Byrne et al (2009); and was used to reliably and the Kolmogorov Smirnov goodness of fit test determine the points on which there was a (Clauset et al.,2009). Figures were created with significant change in the direction of the path of Arcgis 10.2 and R 3.1.1. monkeys. CPT examines paths composed of short segments (vectors) for detecting points at which Ranging -walk- patterns the direction of these vectors change. This test The trajectories of individual spider monkeys were has a parameter q, which must be fixed for every analyzed according to the methods used by species. The test was performed 2940 times for Ramos-Fernandez et al. (2004) and Turching 490 paths and different values of q oscillating (1998). The path of each focal animal consists of a between 1 and 6 with an alpha value of 0.05. We sequence of Cartesian coordinate pairs, one pair found that the CPT was most sensitive at q = 5. for each 15-min interval. A step was taken as an interval in which one or both of the coordinates in Finally, we performed a Wilcoxon paired test to two consecutive samples differ. The length of make a comparison between the location of each step was calculated as the Euclidean change points and points of nutritional distance between the positions in two importance, (a point of nutritional importance is consecutive samples. In some cases, observers classified as such if the specie of feeding lost sight of the focal animals; steps were not represented 10% or more of the total nutriment calculated for these intervals. Only consecutive consumed in the whole day) these points were routes lasting >6.5 hours of focal animal follows also compared with random locations (a pair of were included in the analysis. The frequency coordinates resulting from a random drawing distribution of step lengths was analyzed using a from all points of the route analyzed ) and with range of 10 meters. Waiting times were calculated less important resources (less than 10% of the from the number of samples in which the focal total consumption). animal did not change its spatial position. The frequency distribution of waiting times was Results analyzed using the size of the sampling interval Overall, we recorded 490 relatively linear (15 min). The best fit for the distribution of the continuous routes throughout this study (Fig.1). length of the steps and the waiting times was The step length fitted a power law distribution, evaluated using the method of Kolmogorov (Fig.2) with a X value of 150.7 and an exponent Smirnov and maximum likelihood (Clauset et al, min of 3.98. 2009). Regarding the turning angle distribution, we In order to calculate the square displacement of found no fit between a uniform distribution and the trajectories, the following procedure was used: First, the length of a line connecting the first the data. This shows that there is not uniformity location of the day with all the recorded locations at different times was calculated. All these distances were then squared. The turning angles

3

in the angles distribution. The turning angles were analyzed between successive steps, it was found that most of the time they follow the same direction, with few changes of direction (Fig.3)

Based on the squared displacement metrics, we also found that the routes of spider monkeys tend to move away from the starting point faster than random motion. The farter point from the start point was on average, about midday, on everyday.

Figure 1. Example of Ateles hybridus daily route, with points taken every 15 minutes. Start point at 7:00 and end of the tracing at 17:00.

Figure 3. Polar plot of the difference between successive turning angles.

Figure 2. Plot of (a) Step length (n= 8334) fitted Figure 4. Plot of the mean squared displacement to a power law distribution with a significant in the daily paths of spider monkeys. level of 0.05 and a critical value of 0.045 for Regarding the distance traveled by solitary spider Smirnov Kolmogorov test. (b) Rest times n= monkeys and groups of 2+ spider monkeys, it was (4245) fitted to a power law distribution with a found that individual’s movement fitted a power level of significance 0.05 and a critical value of law when focal individual was part of a subgroup 0.033

4

Figure 5. Equations for the spider monkeys length of step when the focal individual is alone Figure 7. Lines plot of the distance traveled by and when it has company. The data of lonely female and male spider monkeys. Female data individuals fit the power law distribution with fits with a power law distribution (D= 0.033) α =

(D= 0.03) α = 4.11 and Xmin=158.137. And for 3.99 and Xmin= 143.5274. Males steps are also grouped individuals the power law distribution distributed as a power law with (D= 0.029) α = with (D= 0.022) α = 3.96 and Xmin= 147.21 3.99 and Xmin= 159.6621

Figure 6. Plot of the distance traveled in 15 minutes intervals plotted vs the group size (small=1-3 individuals, medium=4-6 and large=7- Figure 8. Graph of Ateles hybridus route and the 9). Small groups fit the power law distribution resources consumed in the travel. Stars represents the changes points found with the with (D= 0.024) α = 3.73and Xmin=145. For the CPT. Asterisks represent the favorite resources medium groups (D= 0.046) α = 3.86and Xmin=145. For large groups (D= 0.043) α = 4.43and and triangles represent the less favorite resources. Xmin=194.6

5 and when it was alone. Different parameters possibility of finding a favorite resource in the were found in each case (Fig.5). It was noticeable same location of a random point along the route that there is no significant effect of been is almost null (Wilcox paired test x coordinate p accompanied by other group members, but value = 1.549e-13, y coordinate p value= when an individual is part of a subgroup there is 2.656e-15) (Fig.8). a tendency to increase the length of steps as the sizes of the group increases too (Fig. 6). The most important items in the diet of brown spider monkeys were fruits (Table 1),and the plant species that coincide more frequently with change points are large figs (genus spp.).

Table 1. The consumed items and species near change points.

Figure 9. Avoidance of intraspecific competition. Discussion Items consumed by a female and a male of the same group in the same day, when the At first, the results of this study share some respective focal sample starts at the same time. characteristics with those described with levy Important resources are shown as triangles and walks (e.g., the power law distribution that less important resources as circles. describes the steps length (Fig.1-a), and the No difference in the displacement behavior turning angle distribution (Fig 3-a), Beta between males and females was found, the distribution, similar to the typical uniform scaling parameter was 3.99 for both sexes (Fig. 7). distribution of Levy processes). But after analyzing We also did not found any difference between the the power law exponents and figuring out which duration of rest times between genders. models fit better with our study case; we can reject the hypothesis that spider monkeys follow a Of 490 routes analyzed we found that 274 of levy walk ranging and foraging pattern in the those have significant “turning” points. We also fragmented forest where this study took place. found that the location of favorite resources corresponds to the same location of changes Levy walks are characterized by a distribution -α points (Wilcox paired test x coordinate p value = function f(x) = x with 1 <α ≤ 3 where x is the 0.2296, y coordinate p value= 0.4377) walk length, and α is the scaling parameter (Viswanathan et al., 1999). In our study, the Almost all the changes points locations coincide power law distribution found for steps magnitude with favorite resources location (Wilcox paired has a scaling factor of α = 3.98, and thus the test x coordinate p value = 1.549e-13, y function decreases more rapidly than a Levy coordinate p value= 2.656e-15). The function, and has a thinner tail. Fragmentation

6 leads to a reduction of the habitat and increases movement, another distribution of the turning edge areas, usually decreasing the available angles may be due to different factors. In the Levy resources and habitat quality for forest-dwelling forager hypothesis, the feeding trees must be species, including primates (Rodriguez et al., distributed randomly, which is in general far from 2006). A direct consequence of habitat the reality within a small fragment in which the fragmentation is the increasing restrictions on resources are clustered in patches (Arroyo- movement possibilities within their home ranges. Rodríguez & Mandujano, 2006). At San Juan, the At San Juan, we found that brown spider monkeys limited area of the spider monkey group´s home have a very low frequency of longer steps (the range (< 40 ha) and its elongated and thin shape ones that define the tail of a power law produces a bias towards the geographical axis of distribution) in relation to the ones that could be the fragment. These two factors, amongst others, generated by a Levy walk. Moreover, we found might influence the non-typical turning angle extremely frequent short steps. Steps of spider distribution observed for spider monkeys at the monkeys that inhabits continuous forests are small forest fragment at San Juan. typically distributed as a power law distribution, the results shown here fit with a extreme case of According to theoretical studies on Levy walk a power law distribution, the one almost tailless. patterns, the mean-squared displacement in this movement model increases faster than in other We propose that the reported pattern can be due to the space limitations caused by fragmentation. random walks such as a Brownian random walk Because the distribution of resources promotes (Shlesinger et al., 1993). Mean squared displacement can be viewed as the amount of the movement and the territory´s shape limits the movement, in San Juan these two aspects are system "explored" by the random walker. In the present study the mean squared displacement determined by the fragmentation. took more time to increase than a Levy walk. Primates can adjust their activity budget to Instead, it increased faster than a linear relation, stressful conditions like fragmentation as and had a faster return to the start point. This described by Boyle & Smith (2010), where saki result supports the notion that spider monkeys monkeys employ a greater proportion of time make use of spatial memory in their resting in fragmented forests than in continuous displacement. If it is true that spider monkeys forests. Ramos Fernandez et al. (2004) found that have a spatial memory about their territory,, they spider monkeys (Ateles geoffroyi) inhabiting a would not need to explore it as much as a Levy continuous forest in the Yucatan Peninsula have walker. resting times that fit a power law distribution with α =1.7. We found that our data fit a power law In a comparative field study between spider with α =3.3 with most of the resting events monkeys and chimpanzees, Chapman et al., (1995) suggested that group size is a function of characterized short resting intervals, different from those expected from a Levy walk that would travel costs that can be indexed by the size, have a larger proportion of records towards the density, and distribution of food resources. Food competition limits group size, when monkeys tail of the distribution (i.e., longer rest times). travel in larger groups there is a need to travel Several studies have considered turning angle greater distances in order to get resources for all distributions to be by-products of complex group members (Janson & Goldsmith, 1995). behavioral processes related to orientation Ramos-Fernandez et al., (2004) found that solitary mechanisms (Bartumeus et al., 2008). The spider monkeys in continuous forests travel uniformity of the turning angle in a Levy process is greater distances than when they are part of a then product of the random character of this larger subgroup. According to the data reported in

7 this study, when monkeys are part of a subgroup evidence of a spatial memory representation in they travel a higher proportion of long steps Ateles belzebuth (Di fiore & Suares, 2007).This compared to short ones. We found differences in study also evidenced a repetitive route pattern the power law exponents for solitary individuals used by brown spider monkeys at San Juan. The and larger subgroups. The tail of the power law reduced area of the forest fragment makes easily distribution of a group of monkeys is heavier than a complete cognizance of the location of the distribution for a lonely individual. We also important areas. The thin and elongated shape of found a positive relation between group size and the spider monkeys home range requires the the distance travelled by it, as proposed by establishment of certain routes over time due to (Janson & Goldsmith 1995). These relation could memory effects or simply the reduced be the result of high intraspecific competition probabilities of moving in different directions to generated by a large number of animals living in a visit novel areas. small forest fragment, where resources might be predictably located by spider monkeys. Under this In our study most of the travel change points match with important dietary resources. Thus, we scenario as subgroup size increases travel distances should increase if all animals are to suggest that resource distribution is an important criterion on the daily travel decisions of spider fulfill their nutritional requirements. monkeys. Similarly, travel routes of gibbons and Previous studies on wild spider monkeys have sakis appear to be determined by a few selected evidenced sexual differences in the ranging feeding sites (Asensio et al.,2011 Cunningham and patterns and the relative use of home range Janson 2007). As Valero and Byrne (2007) we have (Shimooka 2005). In general, males use a bigger found that spider monkeys do not generally proportion of the home range and walk longer continue in the same direction after exploiting a distances than females (Ateles geoffroyi Ramos resource; they change their direction to reach Fernandez et al .,2004; Ateles belzebuth belzebuth with high linearity another resource. Our results Shimooka, 2005). Here we found no difference in rather indicate that spider monkeys might have a the travel characteristics between males and good spatial representation of resources females. Our results show the limitation that locations. (Shimooka 2005). It has also been experiences Ateles hybridus due the proposed that monkeys may construct a spatial fragmentation and loss of habitat. cognitive map of their home range that enables them to reach favorite feeding trees that are out Linear travel paths to out-of-sight food resources of their visual range in an efficient way. Similar have traditionally been used as evidence that results have been found for gibbons and sakis animals had a travel route or goal in mind at the (Asensio et al.,2011; Cunningham & Janson, beginning of a travelling event and therefore 2007). probably knew where they were heading to (Asensio et al., 2011). The difference between Figs are usually regarded as a key resource due to consecutive angles shows that there is a its importance for frugivores in times of fruit predominant direction in spider monkeys daily scarcity (Terborgh, 1986)- In a study on spider paths. The results from CTP analysis reveal that monkeys in the Bolivian Amazon, Felton et almost half of the analyzed routes were linear. al.(2007) found that the diet of Ateles chamek This supports the notion that Ateles hybridus are strongly depends on figs year round leading them able to travel efficiently through their home range to consider figs, not only as fallback resources but and could also be seen as evidence of the use of as staple resources which are available cognitive maps throughout their habitat. The throughout the year. In our study, we also found repeated use of the same paths through time is an that the main food source route enhancers were

8 fig tree species (Genus Ficus), and it is the most Arroyo-Rodríguez, V., & Mandujano, S. (2006). important genus in the diet of wild spider Forest fragmentation modifies habitat quality for monkeys at San Juan (Link et al. unpublished Alouatta palliata. International Journal of data). Primatology, 27, 1079–1096.

Based on our results we conclude that brown Asensio, N., Brockelman, W. Y., Malaivijitnond, S., spider monkeys (Ateles hybridus) at San Juan de & Reichard, U. H. (2011). Gibbon travel paths are Carare do not follow Levy walk patterns. In goal oriented. Animal cognition, 14(3), 395-405. contrast, they employ a foraging strategy Bartumeus, F. (2007). Lévy processes in animal determined by the frequent visit of known resources. Thus, our results suggest that spider movement: an evolutionary hypothesis. Fractals, monkeys at San Juan follow goal oriented daily 15(02), 151-162. trajectories which might serve as an effective Bartumeus, F., Catalan, J., Viswanathan, G. M., strategy to reduce intra and intergroup Raposo, E. P., & da Luz, M. G. E. (2008). The competition. influence of turning angles on the success of non- oriented animal searches. Journal of Theoretical Aknowledgments Biology, 252(1), 43-55. Boyle, S. A., Lenz, B. B., Gilbert, K. A., Sprionello, W. R., Gómez, M. S., Setz, E. Z., ... & Pinto, F. I thank the members of Proyecto primates (2013). Primates of the Biological Dynamics of Colombia who collected field data at San Juan de Forest Fragments Project: A History. In Primates in Carare. I am also thankful with Lalinde and De Fragments (pp. 57-74). Springer New York. Greiff families who have allowed us to work at their land for the past seven years. National Boyle, S. A., & Smith, A. T. (2010). Behavioral Science Foundation, National Geographic Society, modifications in northern bearded saki monkeys Idea Wild and Margo Marsh Biodiversity Fund (Chiropotes satanas chiropotes) in forest provided the financial support for this study. fragments of central Amazonia. Primates, 51(1), Special thanks to Gaston Lyons for all the 43-51. assistance and Pablo Montes for all advice. I also thank Andres Link for allowing me to be part of Byrne, R. W., et al. "How did they get here from this research and conservation project. there? Detecting changes of direction in terrestrial ranging." Animal Behaviour 77.3 (2009): 619-631.

References Clauset, A., Shalizi, C. R., & Newman, M. E. (2009). Power-law distributions in empirical data. SIAM review, 51(4), 661-703. Aggimarangsee, N. (2013). Status Monitoring of Isolated Populations of Macaques and Other Chapman, C. A., Chapman, L. J., & Wrangham, R. Nonhuman Primates in Thailand. In Primates in W. (1995). Ecological constraints on group size: an Fragments (pp. 147-158). Springer New York. analysis of spider monkey and chimpanzee subgroups. Behavioral Ecology and Sociobiology, Altman, P. L. (Ed.). (1974). Biology data book. 36(1), 59-70. Bethesda: Federation of American Societies for Experimental Biology. Cristóbal-Azkarate, J., & Dunn, J. C. (2013). Lessons from Los Tuxtlas: 30 Years of Research

9 into Primates in Fragments. In Primates in Shimooka, Y. (2005). Sexual differences in ranging Fragments (pp. 75-88). Springer New York. of Ateles belzebuth belzebuth at La Macarena, Colombia. International Journal of Primatology, Cunningham, E., & Janson, C. (2007). Integrating 26(2), 385-406. information about location and value of resources by white-faced saki monkeys (Pithecia pithecia). Ramos-Fernández, G., Mateos, J. L., Miramontes, Animal cognition, 10(3), 293-304. O., Cocho, G., Larralde, H., & Ayala-Orozco, B. (2004). Lévy walk patterns in the foraging Di Fiore, A., & Suarez, S. A. (2007). Route-based movements of spider monkeys (Ateles geoffroyi). travel and shared routes in sympatric spider and Behavioral Ecology and Sociobiology, 55(3), 223- woolly monkeys: cognitive and evolutionary 230. implications.Animal Cognition, 10(3), 317-329. Shlesinger, M. F., Zaslavsky, G. M., & Klafter, J. Fahrig, L. (2003). Effects of habitat fragmentation (1993). Strange kinetics. Nature,363(6424), 31-37. on biodiversity. Annual review of ecology, evolution, and systematics, 487-515. Terborgh, J. (1986). Community aspects of frugivory in tropical forests. In Frugivores and IUCN.2008.The IUCN red list of threatened seed dispersal (pp. 371-384). Springer species. Available online: Netherlands. http://www.iucnredlist.org/details/39961/0 (accessed 8 August 2014). Wallace, R. B. (2001). Diurnal activity budgets of black spider monkeys, Ateles chamek, in a James, A., Plank, M. J., & Edwards, A. M. (2011). southern Amazonian tropical forest. Neotropical Assessing Lévy walks as models of animal Primates, 9(3), 101-107. foraging. Journal of The Royal Society Interface, rsif20110200.

Janson, C. H., & Byrne, R. (2007). What wild primates know about resources: opening up the black box. Animal cognition, 10(3), 357-367.

Janson, C. H., & Goldsmith, M. L. (1995). Predicting group size in primates: foraging costs and predation risks. Behavioral Ecology, 6(3), 326- 336.

Link, A., De Luna, A. G., Alfonso, F., Giraldo- Beltran, P., & Ramirez, F. (2010). Initial effects of fragmentation on the density of three neotropical primate species in two lowland forests of Colombia. Endangered Species Research, 13(1), 41-50.

Mittermeier¹, R. A., Wallis, J., & Rylands¹, A. B. (2012). The world’s 25 most endangered primates 2010–2012.

10