Comparative locomotor ecology of gibbons and macaques Selection of canopy elements for crossing gaps.код для вставкиСкачать
AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 93:505-524 (1994) Comparative Locomotor Ecology of Gibbons and Macaques: Selection of Canopy Elements for Crossing Gaps CHUCK H. CANNON AND MARK LEIGHTON Department of Anthropology, Haruard Uniuersity, Cambridge, Massachusetts 02138 KEY WORDS Arboreal locomotion, Functional ecology, Canopy structure, Gaps, Selectivity ABSTRACT To examine functional questions of arboreal locomotor ecology, the selection of canopy elements by Bornean agile gibbons (Hylobates agilis) and long-tailed macaques (Macaca fascicularis) was contrasted, and related to locomotor behaviors. The two species, and in some cases, the macaque sexes, varied in their use of most structural elements. Although both species traveled most frequently in the main canopy layer (macaques: 56%, gibbons: 48%),the gibbons strongly preferred the emergent canopy layer and traveled higher than the macaques (31 vs. 23 m above ground) in larger trees (48 vs. 26 cm dbh). Macaques preferred to cross narrower gaps (50%were in the class 0.1-0.5 m wide) than gibbons (42%were 1.6-3.0 m wide), consistent with the maximum gap width each crossed (3.5 m for macaques, 9 m for gibbons). Macaques could cross only 12%of the gaps encountered in the main canopy, and < 5%of the gaps in each of the other four layers. In contrast, all layers appear relatively continuous for gibbons. Specialized locomotor modes were used disproportionately at the beginning and end of travel segments, further indicating that behavior was organized around gap crossings. A model is defined, the Perceived Continuity Index (PCI), which predicts the relative use of canopy strata for each species, based on the percentage of gaps a species can cross, the frequency of gaps, and median length of continuous canopy structure in each canopy layer. The results support the hypothesis that locomotor behaviors, and strategies of selecting canopy strata for travel, are strongly constrained by wide gaps between trees and are ultimately based on selection for efficient direct line travel between distant points. 0 1994 Wiley-Liss, Inc. In the Southeast Asian tropical rainforests, an incredible diversity of mammals share the arboreal environment, using extremely varied locomotor behaviors; sympatric gliding squirrels, jumping and running primates, squirrels, rats and civets, brachiating apes and slow clambering sun bears all exploit canopy plant foods. The niche separations among members of this arboreal guild have traditionally been depicted in terms of the vertical stratification during travel and feeding (e.g., MacKinnon, in Whitmore, 1984). This stratification has often been explained by differences in their locomotor behavior. Although studies relat0 1994 WILEY-LISS, INC ing the locomotor behavior of rainforest mammals to their ecological niches have almost exclusively concerned primates, these have not been designed to test functional hypotheses about locomotor differences among species. Research has focused on the quantitative descriptions of frequencies and Received January 29, 1993; accepted October 17,1993. Address reprint requests to Dr. Mark Leighton,Departmentof Anthropology, Harvard University, Cambridge, MA 02138. 506 C.H. CANNON AND M. LEIGHTON contexts of locomotor behavior by primates and their relative use of canopy strata and structural supports of different sizes. Sympatric species typically differ in their locomotor specializations, and oRen travel through different strata (Curtin, 1977; Morbeck, 1977; Mittermeier, 1978; Fleagle, 1978; Fleagle and Mittermeier, 1980; MacKinnon and MacKinnon, 1980; Gittins, 1983; Crompton, 1984; Yoneda, 1984). These studies do not test hypotheses of niche separation, unless stratification of travel is examined relative to the available structure of the canopy or to the distribution and availability of food resources. Comparative functional analyses of posture and locomotion require examining the benefits or costs of different behaviors in a functional (i.e., selective) context relative t o some correlate of fitness, such as feeding efficiency (Clutton-Brock and Harvey, 1984). Such analyses require different objectives and methods than those of previous studies. Attention must be given to describing the arboreal environment itself in terms of the problems it presents to achieving efficient performance in travel or in feeding (Ripley, 1967; Morbeck, 1977; Temerin and Cant, 1983; Pounds, 1991). The variables and contexts used to describe locomotor behavior must be oriented towards testing hypotheses of solutions to these problems. This paper examines the evolutionary ecology of locomotor behavior through comparative analysis of the selective use of canopy strata during travel, and the constraining factors underlying this selection, by two sympatric Bornean primate species: the agile gibbon (Hylobates agilis, hereafter gibbons) and the long-tailed macaque (Macaca fascicularis, hereafter macaques). An analysis of locomotor behavior that focuses on where and how efficiently mammals travel is relevant to the functional analysis of several problems of their evolutionary ecology, as exemplified by the insights into ranging and feeding ecology from analyses of terrestrial quadrupeds (e.g., Pennycuick, 1979). However, the problem of horizontal travel, of getting from point a to b, is particularly acute for large-bodied, arboreal primates, most of which occupy large home ranges and travel long distances each day. Rainforest canopy structure and the problems it presents for direct, horizontal travel Efficient travel through the rainforest canopy, in terms of minimizing deviations from direct travel between two points, is constrained not only by the ability to use available structure, but also by the ability to cross canopy discontinuities, or gaps (Temerin and Cant, 1983; Cant, 1988). The decision of where to cross gaps not only depends on the direction of travel but also on the availability of sufficiently narrow gaps with appropriate structure at either end for its typical method of crossing these gaps. Therefore, the animal must constantly solve the immediate problem of how to travel through a tree crown in order to cross into the next crown, while minimizing the deviation from directed travel towards its ultimate goal (Temerin and Cant, 1983). Given the complexity of locomotor behavior and the difficulty of evaluating what structure is available ahead, this decision process involves a great number of factors, some less predictable than others. It is not surprising that rainforest arboreal mammals usually stop briefly in each tree crown to choose between the travel paths available to them (see Results). Clearly, the animal’s perspective of the continuity of a given canopy-the relative number and diversity of travel paths available-varies between species, depending on their morphological and behavioral attributes. Past research has focused on describing these latter two aspects of locomotor ecology. Several authors have discussed the importance of demonstrating “selective” use of canopy structure for an effective functional analysis of behavior (Ripley, 1967; Morbeck, 1977; Mittermeier, 1978; Garber, 19841,but few have sampled both the immediate structural context of behavior and the availability of structures, as is necessary to demonstrate preference for a structural feature. Some researchers (Gautier-Hion et al., 1981;Whitten, 1982; Crompton, 1984; Ganzhorn, 1989) have described forest structure by quantifying the density of trees of differ- LOCOMOTOR ECOLOGY OF GIBBONS AND MACAQUES ent sizes and the widths of tree crowns, but have not linked these variables directly to the disproportionate use of available structure. The question why primates with specific locomotor abilities might select particular canopy strata requires this link. Testing hypotheses of selective canopy use requires methods that quantify both the incidence of structures available for travel (Ripley, 1967) and the types of gaps between trees which must be crossed (Napier, 1967). Without these measures, it is impossible to distinguish whether the primate uses a structure merely because of its predominance in the environment or if it is selected because of its structural appropriateness for travel. If the availability of different structural components and the incidence of gaps are measured, however, locomotor decisionmaking can be examined, and we can distinguish casual from causal interactions of behavior and structure. Further, unless the variation in available canopy features is described, field studies completed at different times and in different locations cannot be compared because the structure of the canopy might vary between habitats, forest types, or disturbance histories, confounding their interpretation (e.g., Mittermeier, 1978; Gittins, 1983; Boinski, 1989). So what factors affect stratum choice and how can they be quantified? 507 and patterning of this mosaic is poorly understood. Is most of the forest mature, tall and diversely layered, with small islands of messy, successional forest, or do the patches of less strcturally developed, building phase forest dominate? Do corridors of well developed canopy run through adjacent younger canopy? How does this vary between different geographic locations? Lack of knowledge of spatial patterning of canopy structure and successional stages complicates robust, generalizable descriptions of canopy use. If only the layer of travel is documented, then travel occurring in different forest habitats or types, which are stratified similarly but have different spatial relationships both within and between strata, cannot be distinguished. Two sympatric species could travel through the same piece of forest but select different successional stages, thus selecting different structures within each subhabitat. The actual structural features guiding selection could be obscured, as could differences in the use of the canopy between species. Because even experienced individuals have found it difficult to identify and reliably classify successional stages, several variables should be collected in combination with layer to provide a broad picture of canopy structure and its use (e.g., Whitten, 1982; Ganzhorn, 1989). Height above ground has commonly been recorded by primatologists (MacKinnon and Methodology for examining selective MacKinnon, 1978; Cant, 1987; Boinski, canopy use 19891,but evaluation of its effect requires its Most primatologists have followed the joint analysis with the structural variables. classical profile diagram of primary, well- Variation between species in their height of drained rainforest (Richards, 1952) in de- travel but not in their use of strata would scribing canopy stratification (Napier, 1966; suggest that they are choosing to travel Morbeck, 1977; Fleagle, 1978; Gittins, through differently statured portions of the 1983). Three layers, the understory, the same forest. The size of trees (its “dbh” or main canopy, and the emergent trees, are bole diameter at breast height) is also a distinguished as the organizing structural strong indicator of the stature and maturity components of the canopy. The biological re- of a piece of forest. This data is readily availality of canopy stratification is based on the able and can demonstrate selection for a “inversion surface” caused by light competi- particular forest type. Another variable constraining travel is tion between tree crowns. The height where this competition occurs, and therefore the the degree of horizontal continuity between vertical structure of the canopy, is deter- tree crowns (Napier, 1967; Ripley, 1967; mined by the history of disturbance and re- Cant, 1987). The classical description of the generation a t a particular site, creating a rainforest canopy suggests that horizontal complex mosaic of different successional continuity should vary among canopy laystages (Whitmore, 1984). The spatial scale ers, providing a basis for the selection of can- C.H. CANNON AND M. LEIGHTON 508 opy strata based upon their degree of continuity. By relating the horizontal continuity between tree crowns within each strata to detailed data on how and when the animal crosses gaps of different widths, it would be possible to infer the density of its potential travel paths. If both locomotor behavior and horizontal continuity depend on canopy strata, then the degree of continuity could be a powerful predictor of the strata an animal chooses, or alternatively, to which it is confined. All of the variables discussed above are expected to vary among adjacent rainforest habitats, yet the possible effect of habitat variation has been largely ignored in previous discussions of arboreal locomotion. The rubric of “moist tropical forest,” encompasses a diversity of distinct habitats, distinguished by forest stature, biomass, plant species composition, and rate of natural disturbance, which in turn determines the mosaic of structural successional phases (Whitmore, 1984). In our study, quantitative descriptions of canopy structure in each habitat, when combined with selectivity analysis to delineate which factors underlie choice of travel path, enable us to address why some species are limited to only a few habitats while others appear in almost all habitats available to them. Study species Gibbons and long-tailed macaques are particularly appropriate for comparative analysis (Temerin and Cant, 1983; Grand, 1984; Cant, 1992). They are sympatric primates of similar body mass and have similar ecological relationships with food resources and predators, but vary markedly in locomotor behavior and anatomy and in intriguing aspects of feeding ecology and socioecology. Although the two species are faced with the same distribution of possible food resources from which they choose a diet and feed on the same gross categories of food (MacKmnon and MacKinnon, 1978; Rodman, 1978; Gittins, 1982; von Schaik and van Noordwick, 1986; Leighton unpubl.), they have evolved distinct solutions to the problems of arboreal travel. Gibbons most commonly brachiate or progress bimanually (Fleagle, 1976; Gittins, 1983). Macaques are arboreal quadrupeds (Cant, 1988);their locomotion is cautious and slow in contrast to the gibbons (Grand, 1984). The similarity of their diets may be slightly overstated because of the great difficulty in making careful observations, macaques spending more time foraging for insects and eating more unripe fruit (Cant, pers. comm.). In this study, we contrast the relevant canopy structural variables of several rainforest habitats included within the home ranges of the single groups of macaque and gibbon we studied, and examine how the two species organize their travel in response to these variables. We address the following specific questions: 1. Are these two primate species selective of the canopy layer in which they travel? 2. If so, how are their preferences related t o variables of forest structure? 3. How does each species solve the problem of crossing gaps, and do their abilities in solving this problem constrain their use of canopy strata for travel? 4. Can the use of different locomotor behaviors be related to the problem of crossing gaps? 5. How does the stability of limbs and twigs relate to the diameters of structures used for travel and for crossing gaps, and do the species and for macaque, the sexes, select combinations of diameters and stability for travel? METHODS Study area and subjects The study was conducted at the Cabang Panti Research Station, Gunung Palung National Park, West Kalimantan, Indonesia (1.13”S,110.7”E),20 km from the southwest coast of Borneo (Mitani, 1990). The research site of 15 km2 is comprised of a mosaic of pristine rainforest formations or habitats: peat forest, freshwater swamp, alluvial bench forest, well-drained lowland forest on sandstone-derived soils and submontane forest on granite soils (Whitmore, 1984). 509 LOCOMOTOR ECOLOGY OF GIBBONS AND MACAQUES 0 Riverine Forest Fast-Draining Swamp Slow-Draining Swamp Forest Alluvial Bench Forest Transnion Forest Lowland Sandstone l 0 - - - 250 - ~ 500 Fig. 1. Map of home ranges of the gibbon (solidline) and macaque (dashed line) study groups, and the distributions of six lowland rainforest habitats within their home ranges. These occur over an elevational range from approximately 20 m to 960 m. Rainfall averages about 4,500 mm per annum. Observations were limited to one group of each species, selected because they shared the same home ranges within an area of diverse habitats (Fig. l), and because the macaque group had already been habituated to human observers. Data were collected from January to June 1988 by C. Cannon. A total of 74 segments were collected for the macaques during which they traveled 617 m in 1,883 s, while 37 segments were collected for the gibbons, covering 262.5 m in 208 s. The majority of the data for gibbons were obtained from May to June 1988, following a period spent habituating the gibbons. The gibbon group consisted of one adult pair, a subadult female, and one infant born at the outset of this study. The macaque group consisted of three adult females, one adult male, one subadult female, one subadult male, several adolescents, and one infant carried by its mother. Only adults were used as study subjects to minimize the effects of body size. Travel observations The sampling unit for these observations was a complete segment of locomotion, which began when movement was initiated from a resting posture and ended when movement stopped and the animal returned to a resting posture. This definition of a “segment” of travel here should be distinguished from that used commonly by Fleagle (1976) and other primatologists. In this study, segments of travel, and not specific locomotor patterns, were the focus of sampling. If the individual stopped or diverted movement to forage for insects or to interact with another individual, the segment was not included. This sampling unit was adopted to investigate locomotor strategy as a sequence of locomotor behaviors and to measure the rate and directness of travel (Cannon and Leighton, in prep.). 510 C.H. CANNON AND M. LEIGHTON The majority of the observations recorded were collected during morning followings typically lasting six hours, from sunup to midday. The same species was not sampled on consecutive days. Individual adults were sampled opportunistically as travel could be observed, but observations were rotated among the adults. Rotation was difficult for gibbons because they would often forage individually; therefore, one individual might be followed for up to a n hour before switching. No samples were taken within fifteen minutes of each other to allow consecutive samples to be reasonably independent. A sample had to meet three criteria to be included: (1)the individual observed must be targeted before movement and therefore before sampling began, (2) the entire segment of travel must be observed, and (3)the end of the segment must be observed. The first criterion avoided focusing on extraordinary eye-catching activities. The second ensured that even brief, possibly critical transitional behaviors were sampled. The third insured that we sampled only “complete” segments of locomotion. These requirements could possibly have biased observation towards shorter segments, but very few (< 5%)were thrown out because of a n inability to follow the individual’s movement. Most segments were rejected because the individual stopped to forage for insects or reacted to the observer. A microcassette recorder and a stopwatch were used to collect observations for the gibbons, but this was unnecessary for the macaques because of their slow rate of travel and limited gymnastic repertoire. All observations were limited to locomotion occurring in the course of travel between feeding locations, with special attention paid to crossings between tree crowns. The individual was identified by sex, and if female, whether or not she carried a n infant. The variables collected for each segment concerned locomotor behavior and the structural context of travel. Variables describing the structural context of travel included habitat type and four measures of vertical stratification: (1) canopy layer, ( 2 ) height above ground (HAG) estimated t o the nearest meter, (3) distance to the most continuous stratum located directly above or below the traveling subject (measured to the nearest meter), and (4)diameter at breast height (dbh) of the tree in which travel ended (estimated to the nearest cm at 137 cm above ground). Travel supports were characterized by their: (1)diameter, estimated to the nearest cm, (2) angle of orientation relative to the horizontal, estimated to the nearest ten degrees, and (3) stability (see Morbeck, 1977). Support stability refers to the flexibility of a structural support when used by a n animal and was not the inherent, mechanical stability of the support. Stability was classified by whether the support: (1)was stable (did not move), (2) moved very slightly, (3) moved > 5 cm and G 0.25 m, (4)moved > 0.25 and s 0.5 m, (5) moved > 0.5 and s 1m and (6) moved > 1 m. The width of crossings between tree crowns was defined as the gaps between available woody structure he., from twig to twig, not leaf to leaf), estimated to the nearest 0.5 m. All distance and support size estimates were made using the average length of a gibbon or macaque torso as a relative guide. Estimates of gap widths, support diameters, or support deformations were made while the subject used them or immediately thereafter. If the focal individual crossed between trees during a segment, the mode of locomotion used was noted. In order to bolster sample sizes, additional crossings were opportunistically sampled, following the three criteria for selecting unbiased samples described above. A total of 83 crossings were observed for the macaques and 49 crossings for the gibbons. Locomotor behavior was described by the mode of locomotion and the distance traveled, to the nearest 0.5 m, while the mode was continuously used. The locomotor modes employed by the two species follow definitions used by previous workers (Cant, 1988; Fleagle, 1976). Leaping is discontinuous progression, the hindlimbs providing all propulsion. Bipedal walking is continuous, using the hindlimbs with the torso erect. Quadrupedalism is continuous, with all four limbs used below the body. Brachiation (arm swinging) is discontinuous progression in which the forelimbs are used in a suspended posture. Scrambling is irregular progression across more than one support, with a variety LOCOMOTOR ECOLOGY OF GIBBONS AND MACAQUES of limbs and postures being used, including pronograde clambering (Cant, 1987). Climbing is continuous progression up or down an approximately vertical support. Bridging is a cautious, discontinuous progression, in which the individual reaches across the distance between supports while remaining anchored on the original support and transferring weight across the gap. Macaques do this forelimbs first, whereas gibbons generally bridge hindlimbs first. Hopping is discontinuous progression using both hindlimbs and forelimbs equally for the propulsive force. A behavior classified as “stepping” was included for gibbons. Often during rapid sequences of brachiation, the individual would step on a support with a hindlimb, bouncing between two supports too wide for normal brachiation. We did not feel this qualified as bipedalism nor should it be included under brachiation. The sequence in which the modes occurred in each sample was recorded; each mode was assigned to one of five classes according to its place in the segment: the first, the second, the middle, the second to last, and the last. Segments of only one mode were placed in the first position. If two modes were used in a segment, they were assigned to the first and the last; if three, to the first, middle, and last; if four, to the first, second, middle, and last. 511 quarter method (Mueller-Dombois and Ellenberg, 1977). At 25 m intervals along the transect (5 points per transect, 60 lines total), a 20 m line was laid perpendicular to the transect on alternating sides of the transect. At 5 m intervals along this line (5 points per line, 300 points total), vertical structure was sampled. The method had to be modified to sample the riverine habitat, which runs along the Air Putih stream. The starting points for each of two 250 m transects, running along each side river, were randomly selected and moved 20 meters away from the river’s edge. The transect then followed the river’s course. The distance to the nearest tree was measured only in the two quadrants away from the river. The perpendicular lines were run every 50 m. The two variables measured using the point-centered quarter method were the distance of the nearest tree from the sample point and the tree’s size (dbh). Variables of vertical structure included the presence or absence of each of five canopy layers and, when present, visually estimating their height above ground and thickness (depth) to the nearest m. Five canopy layers were identified: (1)the understory, which is generally discontinuous horizontally and is discontinuous vertically from the main canopy above; ( 2 ) the low canopy, defined as the bottommost branches of the main canopy Sampling canopy structure trees, which is often discontinuous horizonTo collect measures of the frequency of tally and can be characterized by negatively structural features of forest, twelve 100 m inclined branches and twigs; ( 3 ) the main transects were laid out within the home canopy, which is relatively discontinuous ranges of the two study groups, two in each horizontally and consists of a wide variety of of the following habitat types: (1)riverine branch sizes and inclinations; (4) the high forest, ( 2 ) fast-draining alluvial swamp, ( 3 ) canopy, defined as the topmost branches of slow-draining bench swamp, (4) alluvial the main canopy trees, which are often disbench forest, (5) transition forest between continuous horizontally, and can be characslow-draining swamp and alluvial bench, terized by branches and twigs with high posand (6) lowland sandstone forest. The loca- itive inclinations; and (5) the emergent tion and direction of each transect was de- layers, which are very large individual trees termined in a stratified random fashion, that rise above the rest of the canopy and are with the restriction that the transect had to often discontinuous horizontally and vertiremain entirely within the habitat type and cally. The line-intercept method was used to within the home ranges of both study groups. At 50 m intervals along each measure horizontal continuity a t each of the transect ( 3 points per transect, 36 points to- five points along the perpendiculars to the tal), the distance to the nearest tree > 10 cm main transects. The variable of continuity was measured, using the point-centered was the width of gaps (to the nearest 0.1 m) 512 C.H. CANNON AND M. LEIGHTON between tree crowns for each of the five can- collapsed into a 2 x 2 table for each class, so opy layers. These gaps were measured by that the significant differences, between dislaying a measuring tape along the perpen- tributions could be identified, using the dicular and using a clinometer to insure pre- equation of Ynetta (pers. comm.): cise vertical projection; either end of the gap x2ij = [(Eij - O i j ) ' ~ i j ~ / [ ( l - ( C i ~ ) ) ~ ( l - ( R j ~ ) ) l ; was estimated on the tape. These gaps were where E, is the expected and 0, the actual defined to extend between the smallest sup- number of observations, Ci is the total numports used during crossings and not 'foliage' ber of observations in the ith column, % is gaps. These smallest supports used were es- the total number of observations in the jth timated at 0.5 cm in diameter from prior row, and T is the total number of all observaobservations of travel. Occasionally, one end tions. To examine how the three structural feaof a gap lay beyond either end of the lineintercept sample. These partial gaps were tures (layer, dbh, and distance from the included in the data set by doubling their nearest continuous stratum) and species use length, because they will, on average, be half of each structural feature varied with height in and half out of all samples. This was nec- above ground, 2-factor MANOVASwere peressary to ensure that the distribution of gap formed, controlling for appropriate habitat widths as not biased against gaps larger groups which were different in the availabilthan 10 m, which would have been included ity of the feature being examined. Only the within the 20 m samples less than half of the upper three canopy layers were included in time. Measures of variance become inflated the analysis for layer, and only the slowthrough this procedure, but the gap widths draining swamp, transition swamp bench, are only analyzed by their frequency, size alluvial bench, and sandstone habitats were included in the analyses of layer and of disdistributions and median width. tance from the nearest continuous canopy. Statistical methods RESULTS The index used to measure preference for Selection of canopy layer different structural features is Jacob's D value (Jacobs, 1974): Availability by habitat The relative availability of structure in the five canopy layers varied among the six where r is the relative frequency of use and p habitats (x2 = 32.4, 20 d.f., p < .05), but is relative frequency of availability. This only two of the pairwise comparisons bevalue was chosen because it is bounded be- tween habitats were significant; the rivertween -1 and 1, is symmetrical around 0, ine habitat had structure present more frewhich indicates neutrality (neither dispro- quently in the understory and emergent portionate selection nor avoidance) and re- layers than either the slow-draining swamp lies directly on measures of availability (x' = 15.8, 4 d.f., p < ,051 or the alluvial (Cock,1978). Positive values indicate prefer- bench (x2 = 14.7, 4 d.f., p < .05). The riverence while negative values indicate avoid- ine habitat was therefore excluded from the ance. Statistical tests for significance in se- analysis of selectivity by canopy layer belectivity were by contingency table analysis cause of its distinctiveness, and because (x'), Mann-Whitney tests, Kruskal-Wallis only six segments of travel were observed tests, and multivariate analysis of variance there; data from the fast-draining swamp (MANOVA; Conover, 1980). In contingency were excluded because no travel segments analyses, classes were constructed to insure were recorded there. Availability of strucas fine-tuned an analysis as possible with- ture for travel use by canopy layer was out violating rules for applying the test (i.e., therefore compiled from all points sampled all expected values should be a 1.0; in the other four habitats; the main canopy Conover, 1980). When a significant result was the most common layer (present at 82% was obtained in a contingency table, it was of all sample points) and the understory and D = (r-p)/(r+p-2rp), LOCOMOTOR ECOLOGY OF GIBBONS AND MACAQUES I I -1.0 r 1 r I I I Fig. 2. The Jacob’s D preference value for each canopy layer. The blackened symbols indicate significant results ( p < ,051 in a 2 x 2 contingency table for that canopy layer. Positive values indicate preference while negative values indicate avoidance. 513 ground(x2 = 67.4,40d.f.,p < .Ol).Pairwise comparisons among habitats showed that the riverine, slow-draining swamp, and bench habitats were similar to one another; these were combined to form habitat group 1. The transition swamp bench and sandstone habitats had similar HAG distributions and were combined to form habitat group 2. Habitat group 1had more structure available at lower levels (13-20 m HAG) than did group 2, whereas structure was more available at 25-36 m HAG in group 2 (x2 = 26.3,8 d.f.,p < .001). Therefore, group 1 habitats have a higher frequency of structure at lower heights and lower frequency at higher heights than group 2. As was the case for canopy layer, sample points from the fast-draining swamp habitat were excluded from calculations because no travel was observed there. the emergent layers were the least common (both 28%presence). Use by macaques and gibbons Neither species traveled in the canopy layers according to structural availability (macaques, x2 = 14.9, 4 d.f., p < .01; gibbons, x2 = 22.5, 4 d.f., p < .001). The macaques preferred the main canopy (x2 = 10.27, 1 d.f.,p < .001), traveling 55% of the time in that stratum, while neglecting the understory (x2 = 4.95, 1 d.f., p < .05), where only 2% of travel occurred (Fig. 2). The gibbons completely avoided both the understory and the low canopy (x2 = 4.66, 1 d.f.,p < .05; x2 = 10.79, 1 d.f.,p < .001, respectively) and strongly preferred the emergent layer (x2 = 16.93, 1 d.f., p < .0001), where 32% of travel occurred (Figs. 2). The two species differed in relative canopy layer use (Mann-Whitney test, z = 3.4, p < .001), with macaques traveling more frequently in low canopy (x2 = 8.94, 1 d.f., p < .01) and gibbons more in the emergent layer (x2 = 10.54, 1 d . f , p < .001; Fig. 2). Selection of canopy structure by height above ground Availability by habitit The six habitats varied in the availability of structure at different heights above Use by macaques and gibbons Macaques traveled at a median HAG of 23 m in habitat group 1 and 21 m in habitat group 2. Use by HAG was disproportionate to availability in habitat group 1 (x2 = 21.18, 8 d.f., p < .01) but not in group 2 (x2 = 9.2, 5 d.f., p > .lo; classes > 28 m HAG were combined for contingency table analysis). Macaques favored structure 21-24 m HAG (Fig. 3) both in group 1 (x2 = 16.57, d.f., p < .0001) and in group 2 (x2 = 7.64,l d.f.,p < .01). Analysis for gibbons was restricted to habitat group 1 because they traveled too infrequently in group 2 habitats. The median HAG of travel was 31 m. Travel by HAG classes was disproportionate to the availability of structure (x2 = 20.6, 8 d.f., p < .Ol); gibbons preferred to travel in structure 21-24 m high (x2 = 8.07, 1 d.f., p < .01) and 3 7 4 0 m high (x2 = 4.47, 1 d.f., p < .05), and avoided structure 13-16 m high (x2 = 5.19, d.f.,p < .05; Fig. 3). The gibbons traveled higher than the macaques in habitat group 1 (median = 31 vs. 23 m; Mann-Whitney test, z = 3.6, p < .001). The species also varied in their use of the HAG classes in habitat group 1 (x2 = 20.0, 8 d.f., p < .Ol); gibbons traveled C.H. CANNON AND M. LEIGHTON 514 -0- macaques - habitat group 1 - -&- macaques - habitat group 2 .....- m- gibbons - habitat group 1 - I'Ol n Height above ground (m) Fig. 3. The Jacob's D values calculated for each HAG class. Blackened symbols indicate a significant result (p < .05)in a 2 x 2 contingency table for that HAG class. at 3 7 4 0 m HAG (x2 = 4.86, 1 d.f., p < .05) more frequently than macaques. '.O 1 * macaques Selection of different-sized trees Availability by habitat The habitats did not differ in the availability of trees of various diameters (Kruskal-Wallis test, H = 1.45, d.f., p > .95), so the data from all six habitats were pooled into one group. Because neither primate species traveled in trees < 15 cm dbh, they were eliminated from the selectivity analysis. Most trees (76%)were 16-31 cm dbh and few trees were larger than 48 cm dbh. Use by macaques and gibbons Considering only trees 2 15 cm dbh, macaques chose the sizes of trees in which to travel according to their availability (x2= 3.91, 3 d.f.,p > .2; Fig. 4). They traveled mostly in trees 16-31 cm dbh (60%)but showed a slight, nonsignificant preference for trees 64-122 cm dbh (Fig. 4). Gibbons, in contrast, were very selective (x2 = 20.11, 3 d.f., p < .0002). Trees of 16-31 cm dbh (x2 = 4.68, 1 d.f., p < .001) were avoided while trees of 64-122 cm dbh (x2 = 4.10, 1 d.f.,p < .001) were favored (Fig. 4). The gibbons traveled in larger trees than the macaques (median = 48 vs. 26 cm dbh; Mann-Whitney test, z = 3.5, p < .OOl>. The distributions of size classes used are also distinctive (x2 = 11.03, 3 d.f., p < .05) with -1.0 ' , 16-31 I 32-47 I 48-63 I 64-122 Diameter at breast height (cm) Fig. 4. The Jacob's D values calculated for each DBH class. Blackened symbols indicate a significant result 03 < .05) in a 2 x 2 contingency table for that DBH class. the macaques using the smallest trees, 16-31 cm dbh, more frequently than the gibbons (x2 = 9.6, 1 d.f.,p < .001). Use of the most continuous stratum Both species traveled in the most continuous stratum (that with the greatest degree of horizontal continuity relative to their immediate position; see Methods) almost half the time (50%for macaques and 46%for gibbons), and yet their distributions differed (x2 = 13.8, 8 d.f., p < .05 Fig. 5). Gibbons traveled more often > 15 meters above the most continuous stratum than macaques (x2 = 8 . 6 , l d.f.,p < .01; Fig. 5). LOCOMOTOR ECOLOGY OF GIBBONS AND MACAQUES Distance from continuous stratum (m) Fig. 5. The relative percentage of travel by macaques (n = 69) and gibbons (n = 29) occurring within different distance classes above (positive values) or below (negative values) the most continuous stratum. Comparison of different canopy use variables 515 Whitney test, z = 2.37, p < .05), regardless of whether a female carried an infant, but the stability and the inclination of these supports did not differ (Table 1).Supports in main canopy and emergent trees were more stable than those in low canopy for both sexes (x2 = 17.5, 8 d.f., p < .05). Also, in the main canopy, more negatively inclined supports (-90" to -20") and fewer positively inclined supports (40"- 60") were used but in the high canopy, more positively inclined supports (40"- 60") were used. No sex, layer, or habitat differences were found in the median diameter (median = 6 cm), stability (class 1 = did not move) and the angle (20")of supports used by gibbons. They used larger diameter supports than the female macaques while traveling (Mann-Whitney test, z = 4 . 2 9 , < ~ .001;Table 11, but not significantly larger diameter supports than the male macaques (Mann-Whitney test, z = 1.43). In the emergent trees, the male macaque used larger diameter supports than gibbons (Mann-Whitney test, z = 2.93, p < .05). The gibbons traveled higher above ground within each layer than the macaques (MANOVA, Fl,2 = 26.9, p < .0001) and Use of supports while crossing these layers were used in relation to HAG between trees (F1,2= 54.3, p < .0001). The interaction of species and layer differences in HAG was Supports were divided between those also significant (F1,2= 3 . 3 , < ~ .04),but this used on "take-off and "landing" (Table 1) result is probably a minor interaction be- This division is necessary because the "landtween factors and due to large differences ing" supports were less stable than the takewithin factors (Snedecor and Cochran, off supports for both species (Wilcoxon 1967). The gibbons also traveled higher in signed-rank, macaques, z = 5 . 0 3 , < ~ .0001; trees matched for the same dbh class gibbons, z = 4 . 4 , < ~ .OOOl). The diameter of (F1,3= 13.1, p < .0005) and the larger size supports did not differ for either species and classes used corresponded to greater heights the angle did not differ for macaques. Gibabove ground for both species (Fl,3 = 29.4, bons used supports which were oriented p < .0001). The interaction between factors more closely to vertical on take-off than was not significant (F1,B= 0.7 p > 5 ) .The landing (Wilcoxon signed-rank, z = 3.7, gibbons traveled at greater heights above p < .001). ground than the macaques while having the The male macaque launched himself off of same relationship to the most continuous larger diameter supports than females (Tastratum (Fl,4 = 1 9 . 1 , < ~ .0001), and gibbon ble 1; Mann-Whitney test, z = 2.5, p < .05), travel which occurred at greater distances and also landed on larger diameter supports above the most continuous stratum occurred than females (Mann-Whitney test, z = 2.8, at greater HAGS (Fl,4= 31.9, p < .0001). p < .05). No differences in the stability or There was a very minor interaction between angle of support were found between sexes, factors (Fl,4 = 2 . 4 , > ~ .05). layers, or habitats in either take-offs or landings. Supports used on take-off for the Use of supports while traveling female gibbon carrying a n infant were less The male macaque traveled on larger di- stable in the high canopy (x2 = 10.6, 4 d.f., ameter supports than the females (Mann- p < .05). No other sex differences were C.H. CANNON AND M. LEJGHTON 516 TABLE 1 . Characteristics of supports' used by macaques and gibbons during travel and while crossinE gavs between trees Travel supports Diameter (n)' Stability3 Angle Male macaaue Female macaque Gibbons 5 (50) 3 (103) 6 (55) 2 10" 2 10" 20" 1 Take-off supports Diameter (n) Stability Angle 3(22) 2(65) 4(51) 2 2 2 0" 0" 10" Landing supports Diameter Stability Angle 3 10" 2 2 2 4 4 10" 30" 'Values are median diameter (cm), stability class and angle of inclination of support. 'Numbers in parentheses are sample sizes used in the calculations; n for landing supports is the same for take-off supports 3Stability classes are: 1 = the support did not move, 2 = support slightly moved, and 4 = support moved >.25 but s.5 m. found in comparisons for the supports used from the understory to the emergent layer. by gibbons on either take-off or landing, and The mean number of gaps steadily devalues were lumped for the sexes (Table 1). creased (Kruskal-Wallis test, H = 81.6, 4 Gibbons took off from larger diameter d.f., p < .0001; Fig. 6a) from the understory supports than either the male macaque to the emergent layer. The median width of (Mann-Whitney test, z = 3.1,p < .01) or fe- gaps initially decreased from the understory male macaques (z = 5.1, p < .0001). On up to the main canopy but gaps were then landing, gibbons used larger diameter sup- much wider in the emergent layer ports than female macaques (z = 4.3, (H = 75.3, 4 d.f., p < .0001; Fig. 6b). The p < .001). The male and female macaques median length of continuous canopy segused larger diameter and more stable sup- ments generally increased, with a small dip ports while traveling than while taking off at the high canopy layer (H = 79.0, 4 d.f., (diameter: males, z = 2.8; females, z = 3.9, p < .0001; Fig. 6c). p < .05; stability: z = 5.7, p < .05) or landing (diameter: males, z = 2.5; females, Use by macaques and gibbons z = 4.5, p < .05; stability: both sexes, Combining data from all six habitats, a z = 9 . 1 ,< ~ .05). Gibbons used larger diammajority of gaps present in all layers were eter and more stable supports while travelwider than 3.5 m, the widest gap crossed by ing than while taking off (z = 3.7, p < .01; (Table 2). The narrowest gap z =5.0,~ < .0001, respectively) and landing macaques width class (0.1-0.5 m) was the one most (z = 4.0,~ < .001; z = 9 . 1 , < ~ .0001). commonly crossed by macaques (50%) but was rarely available (< 5%)in all layers exSelection of gaps of different sizes for cept for the main canopy, where such gaps crossing between tree crowns comprised 12% of all gaps. The macaques' Availability by habitat and by layer use of gap widths strongly favored the narHorizontal continuity in canopy structure rower classes in all layers (Fig. 7) and difamong habitat types was described by three fered from the distribution of gap widths acvariables: (1) the number of gaps encoun- tually available to them. Gibbons crossed tered in each 20 m sample, (2) the longest gaps up to 9 m wide, so a greater number of gap encountered in each sample, and (3)the gaps were available t o them than the longest continuous segment of canopy in macaques. Many gaps were still too wide for each sample. Measures for each of these them t o cross, particularly in the emergent three variables did not differ among habi- layer (65%;Table 2). They crossed the sectats for any layer (Kruskal-Wallis test, ond widest gap class (1.6-3.0 m) most comn = 10 for each of six habitats, total number monly (42%).The use of gaps by gibbons difof comparisons = 5 layers x 3 variables fered from the distribution of gap widths that they could cross (i.e., those G 9 m wide) = 15; all p's > .05). To examine differences among canopy lay- only in the main and the high canopy, and ers, data from all habitats were combined. strongly favored the second class (1.6-3.0 m) Figure 6 illustrates the strong pattern in the in all layers (Table 2, Fig. 8). Finally, gibcentral tendency of each variable, moving up bons typically crossed much wider gaps than LOCOMOTOR ECOLOGY OF GIBBONS AND MACAQUES ". TABLE 2. Comparisons among canopy layers in crossing ofpaps b y macwues and gibbons a) 2.5 1 517 X2 value for selectivity of gaps2 for for for for macaques gibbons macaques gibbons % of gaps too wide to cross1 Layer U L M H E Understory Lowcanopy Main canopy High canopy Emerpents 71.2 58.2 52.4 71.4 87.5 34.9 28.7 16.5 39.6 65.3 51.5*** 49.3*** 11.5 21.7** 25.1*** 7.1 11.8 19.7"" 19.4** 11.0 'Based on maximum width of gap observed to be crossed: = 3.5 m for macaques, 9.0 m for gibbons. 'Selectivity of gaps for crossing (X2 values) are displayed in Figures 7 and 8. **p< .01, - * p < .001. v 0 3 20 layers even though the availability of gap widths varied (Fig. 6). 15- m o, 10- .- U L M H E Canopy layer Fig. 6. a: The mean number of gaps between tree crowns present per 20 m segment (n = 60 at each layer). Data were combined from all habitats. b: The median gap width for each canopy layer along the 20 m segment. c: The median length of the continuous structure segments in each canopy layer. Incomplete segments were included in all calculations (see Methods). Specialization of locomotor behavior Quadrupedal walking was the specialized mode of locomotion for traveling macaques (60%) whereas bridging (49%) and leaping (36%) were the common modes for crossing between trees (Table 3). Bridging was used for crossing narrow gaps and leaping for crossing wide gaps (Table 4); this distinction is expected because bridging is limited by body length. In contrast, gibbons most often brachiated both during travel along supports (48%) and when crossing gaps (57%), with leaping the next most comrnm mode in both contexts (Table 3). Bridging rarely occurred (one of 21 occurrences). Leaping is used for crossing slightly wider gaps than brachiation (Mann-Whitney test, p < .01, Table 4). Gibbons would often leap from the edge of a tree crown across very wide gaps as the initial mode in the sequence of travel. Sequencing of locomotor modes The sequencing of locomotor modes within each travel segment also varied between the species (Table 3). Quadrupedal walking and running appeared more than expected a t the beginning (x2 = 25.5, 1 d.f., p < .0001) and at the end (x2 = 26.4, 1 d.f., p < .0001) of macaque travel segments and less than expected as the second (x2= 21.4, 1 d.f., the macaques (median = 2.0 vs. 1.0 m; p < .0001) and the second to last modes Mann-Whitney test, z = 7 . 2 , ~ < .0001). For (x2 = 40.4, 1 d.f., p < .0001). Leaping, the both species, the widths of gaps crossed did specialized mode for crossing wide gaps (see not vary by layer; they use the same widths Table 3), occurred more than expected as the of gaps in crossing between trees in different second to last mode (x2 = 12.3, 1 d.f., C.H. CANNON AND M. LEIGHTON 518 1.0- - understory 0.5 0.0 low canopy main canopy - high canopy ernergents - -0.5 1 Fig. 7. Selectivity of gap widths (Jacob's D values) by macaques within each layer. Gaps > 3.5 m were included in the calculations but are not depicted because the D value is always -1.0. Significant results are ( p < .05): in the understory, the low canopy, the high canopy, and the emergent layer, gaps < 1.1 m were preferred (solid symbols). In the understory and the emergent layer, gaps 1.1-1.5 m were preferred. Gaps 2.6-3.0 m were avoided in the low canopy. 1.01 Q) 3 - 0.5 - understory 2 low canopy main canopy 0.0- high canopy v) b 0 0 m 7 emergents - -0.5 -1.0 u! 7 7 0 : : E '4 7 * T x '4 * Gap width 7 rd x x (rn) Fig. 8. Selectivity of gap widths (Jacob's D values) by gibbons within each layer. Gaps > 9.0 m were included in the calculations but are not represented because the D value is always -1.0. Significant results are (p < .05):in the main and the high canopy, gaps 1.5-3.0 m were preferred; in the understory and the emergent layer, gaps 1.1-1.5 m were preferred; gaps &1.5 m were avoided in the main canopy. p < .001) and less than expected as the last mode (x2 = 4 . 8 , l d.f.,p < .05). Bridging, the specialized macaque mode for crossing narrow gaps, appeared more than expected as the second (x2 = 19.1, 1 d.f.,p < .0001) and the second to last modes (x2 = 26.9, 1 d.f., p < .0001) and less than expected a t the be- ginning (x2 = 5 . 9 , l d.f.,p < .05)and the end (x2= 8.4, 1 d.f., p <.01) of travel segments. In the gibbon travel segments, brachiation occurred less than expected at the end (x2 = 9.7, 1 d.f., p < .01; Table 4). Leaping occurred more than expected at the beginning (x2= 12.4, 1 d.f., p < .0001) and less LOCOMOTOR ECOLOGY OF GIBBONS AND MACAQUES 519 TABLE 3. Use of different locomotor modes for travel, for crossing gaps, and during travel sequences Locomotory modes Macaques Quadrupedalism Leaping Bridging Scrambling Climbing Hopping Gibbons Brachiation Leaping Stepping Bipedal walking Climbing Scrambling Bridging Relative % use while traveling’ Relative % use while crossing gaps’ 10.5 59.8** 6.4*** 0.9*** 15.0** 12.0 6.0 48.7 1.3 2.6 1.3 48.0 16.0 14.7 12.0 5.3 4.0 0.0 57.0 33.3 0.0 4.8 0.0 0.0 4.8 35.5 Number of observations for each mode at each step in the travel sequence - First Second Middle 53*** 6 2* 3 9 4*** 7 11*** 4 7 4 35 11 1 15 14*** 0 5 3 0 0 4 1 1 3 0 0 0 11 13 7 4 Second to last 3*** 15*** 15**c 9 4 5 Last 45*** 3* 0** I 3 1 8 2 4 1 3 1 5 18** 1** 1 0 0 0 1 0 1 0 0 3 1 0 ‘For macaques, n = 94 and 68 for traveling and crossing gaps, respectively; for gibbons, n = 75 and 21. *p < .05; * * p < .01; and ***p < .0001: all significance levels are derived from ax2table analysis. TABLE 4. Median width ofgap crossed by macaques and aibbons using each locomotorv mode’ Locomotory Mode Macaques Quadrepalism Leaping Bridging Scrambling Climbing Hopping Gibbons Brachiation Leaping Bipedal walking Bridging Number of observations Median gap width (m) 8 29 41 2 2 0.1 1.5 0.5*** 0.5 0.5 0.5 1 23 23 2.5 3.0** 1.0 1.5 1 2 ’Tests were Man-Whitney U statistic. **p < .01, for leaping vs. brachiation in gibbons. ***p < .0001, for bridging vs. leaping in macaques. than expected a t the end (x2 = 9.0, 1 d.f., p < .01; Table 4). The organization of travel segments themselves was clearly limited by tree crossings. Macaques stopped before leaving the tree in which they began in 26% of all segments, and made only one crossing in 50% of all segments; the remaining fraction involved two or more crossings. Gibbons were short segment specialists; 95% of all segments involved no crossings or only one (47%for each). DISCUSSION The constraint of crossing gaps The results indicate that crossing the gaps between tree crowns is a central problem organizing arboreal locomotion by mammals such as gibbons and macaques, which must cross between hundreds of adjacent tree crowns, most of which are separated spatially by significantly wide gaps, to visit the dispersed patches of fruits and seeds they prefer to eat. For macaques, in all strata, the majority of gaps present between tree crowns are much wider than the gaps usually crossed (Table 2). For gibbons, a third of all gaps are too wide (Table 2); the only circumstance where the widest class of gaps are not the most common gap widths present are for gibbons in the main canopy (Fig. 8). Therefore, although both species are constrained by gaps from traveling directly between two distant points, macaques are much more so than gibbons; this distinction supports a central hypotheses for the evolutionary divergence of apes and monkeys (Temerin and Cant, 1983). The limited choices for both species explains their repeated travel along “arboreal highways” through familiar habitat, commented upon by Ripley (1967) and Whitten (1982), among others. Gap crossings influenced several aspects of macaque locomotor behavior. Their two 520 C.H. CANNON AND M. LEIGHTON specialized modes of locomotion for crossing these gaps, bridging and leaping, seem inadequate to cross most gaps. Bridging, a very slow and relatively safe process, is limited by body length, so it can only occur across very narrow gaps ( S 1.0 m). Leaping, which involves higher energy costs and greater risk, was only used to cross wider gaps. However, the 3.5 m limit observed in this study should not be interpreted as a strict physical limit to macaque abilities; during a confrontation, the male leapt across a gap of more than 6 m while chasing a pair of gibbons. The gaps crossed by leaping reflect a tradeoff between high risk and energy costs and the benefits from greater directness of travel. These specialized crossing modes occur as the second and the second to last modes during travel segments, indicating that locomotor strategy is organized around crossing gaps (Table 3) and few travel segments involve more than one crossing. A segment either begins close to a gap crossing or stops just after one, as the macaque reconsiders the available options. Leaping occurs towards the ends of segments, probably because it is not the preferred method for crossing between crowns and was used only after the animal has traveled into a situation demanding a leap. Thinner and hence more flexible supports are used when macaques cross between trees, both on take-off and landing (Table 1). Even though the male used larger supports than females, because of his greater body mass, his supports were equally unstable. Landing supports, although of diameter equal to take-off supports, were less stable. Therefore, t o infer the forces involved during travel, the stability of that support is essential to consider, as well as the size of support. Because macaques are limited by the width of gaps in their ability to cross between crowns, they are forced to cross from the extreme periphery of crowns in order to minimize the width of crossings. But this also forces them t o use small supports. During powerful leaping or precarious bridging, these are especially flexible, effectively widening the gap and providing an unstable base for propulsion. Our results concur with Cant’s (1988), although he observed a higher use of lianas in crossing gaps, probably because of habitat differences in structure. Brachiating gibbons, on the other hand, did not seem t o be as limited in their choice of travel path nor does their locomotor strategy seem to be as constrained by the problem of crossing between gaps. They did not exhibit any strong specialization of locomotor mode or sequencing of behavior for crossing gaps (Table 31, but they certainly modified their behavior while crossing gaps. First, the median gap width (2.5 m) crossed while brachiating is probably longer than the “stride length commonly used within a tree crown (Prueschoft and Demes, 1984). Second, leaping occurred twice as often during crossings than during travel and was used to cross wider gaps than was brachiation, although small samples probably prevented statistical significance (Table 3). Travel supports, both on landing and takeoff, were also smaller and less stable during crossings than during travel. A model predicting use of canopy strata As suggested for macaques, the choice of a stratum is influenced by three factors: the availability of gaps which can be easily crossed, the frequency with which gaps are encountered per unit distance, and the length of continuous segments of structure (Fig. 6 ) .The first variable is determined by a species’ locomotor morphology and behavior while the other two depend on canopy structure. To examine whether these three variables can predict strata use by each primate species, we derived and calculated a Perceived Continuity Index (PCI) for each strata. The PCI for each species for each stratrum, i, is defined as: PCIi = (Ci* Gi/2 * Fi) in which Ci equals the median length of continuous canopy in the ith layer, Gi equals the percentage of gaps narrow enough to be crossed by each species, and Fi equals the frequency of gaps per 20 meter sample. The factor 2 in the denominator merely scales the two species for easy comparison and does not affect the relative suitabilities of the strata. A relatively high value of PCI for a strata indicates it is more continuous from 521 LOCOMOTOR ECOLOGY OF GIBBONS AND MACAQUES X 4- zc .- 3- 3 ._ c c 0 2- - macaques --*- gibbons 1- Fig. 9. The continuity index for each canopy layer for macaques and gibbons, illustrating how each species can use the different layers as pathways for continuous travel. The index is the product of the median length of continuous structure by the percentage of gaps between tree crowns which can be crossed for each layer, divided by the product of two times the mean number of gaps per 20 m sample. the animal’s perspective and thus is a more favorable layer for direct, uninterrupted travel. Figure 9 shows the PCI values for each of the five measured strata for each primate species; these relationships are identical to those for the frequency of use of each layer (see Fig. 2). This congruence supports the hypothesis that the nature of the horizontal continuity within a stratum can predict which strata will be used by a species, when modified by its limits to crossing wide gaps. Further, this match between the use of canopy strata and PCI can be interpreted as support for the hypothesis (Cant, 1987) that these two primate species travel in layers which maximize travel efficiency rather than because food might be located there. This PCI model assumes that the immediate goal of locomotion is to move directly between two points separated by several tree crowns. In comparing the five canopy strata in their general suitability for direct travel, it is clear why the main canopy is preferred for travel by both species. The combination of narrow gaps, few gaps per unit distance and a long median length of continuous structure, provide the greatest number of options for direct travel in the main canopy (Fig. 6). The two primates diverge in their PCIs especially in the upper strata (Fig. 9). The high canopy and emergent layers are perceived by gibbons as more continuous. The wider gaps in these layers (Fig. 6b) are less of an impediment because gibbons cross wide gaps, so the longer segments of continuous canopy (Fig. 6c) can be exploited for travel. In contrast, the understory and low canopy combine the unfavorable features of short canopy segments between gaps and wide gaps (Fig. 6). The application of this model may be limited to arboreal mammals with medium body sizes feeding on widely dispersed patches of food such as fruit. If food patches are sparse, then the cost of moving between patches is more significant than moving up or down through the strata after arriving at a food patch. For small insectivorous primates, travel may typically be combined with more continuous prey searching, so that travel layer would likely reflect where food occurs. For large-bodied males or females carrying infants, and especially for the very large-bodied orangutan (Cant, 1987, 1992), the actual strength of available supports may be an important variable in addition to the nature of gaps. Also, orangutans frequently travel through small trees by using their large body mass to sway a small tree back and forth until the tree leans over enough for them to bridge t o the next tree (Cant, 1992; pers. obs.). Therefore, although this model predicts the location of travel by macaques and gibbons quite well, over a wider range of body sizes and locomo- 522 C.H. CANNON AND M. LEIGHTON tor morphologies and behaviors there may be other confounding factors that must be added to the model. Other factors, such as predation pressure, may also strongly affect the suitability of a canopy layer for travel, with the relative danger of travel in each strata depending on the type of predator. The congruence between the PCI model and the use of canopy layers suggests, however, that predators have only a slight, if any, effect on the canopy layer selected for travel. would be required to evaluate other plausible factors. Because these factors are likely to covary, it remains unclear what governs macaque decision-making about travel path. Do macaques select the main canopy, the most continuous stratum, or merely use the nearest tree which will support them? Testing ecological and evolutionary hypotheses of arboreal locomotion In choosing how and where to travel in the arboreal environment, primates presumably integrate an enormously complex set of The explanatory power of canopy costs and benefits, involving tradeoffs structural variables among time efficiency, energy efficiency and No single variable suffices to describe and risk avoidance (see Cant, 1992). Our samexplain canopy travel by these two primates, ples are too small to allow us to assess how arguing for detailed investigations. For in- choices of travel path and modes are influstance, within each class of the two struc- enced by the spatial pattern of forest succestural variables, canopy layer and dbh, gib- sional phases and of forest habitats. The six bons traveled higher than macaques. habitats were delineated by topography and Further, gibbons traveled higher above the drainage, but differed also in the availabilground than macaques when their relation- ity of canopy layers and structure at various ship to the most continuous stratum was heights above ground (Fig. 3). The effects of similar (Fig. 5), indicating that the most these aspects of canopy structure on locomocontinuous stratum was higher above tion have been underemphasized. More deground in tree crowns through which gib- tailed analyses, such as those of Whitten bons traveled. Because each canopy layer is (1982) and Ganzhorn (19891, if they were to often several meters thick and the depth of a incorporate the structural variables demonsingle tree crown can be 10 m or more, it is strated to be important in this study, could probable that gibbons simply travel through link locomotor modes to use of different habthe upper regions of the same type of forest itats, possibly successional stages, and even as macaques. In sum, a combination of plant species assemblages. Most primate height descriptors is therefore necessary to home ranges contain several distinct habiexpose subtle differences in travel not cap- tats. If detailed analysis of selectivity, of structure use versus availability, were made tured by a single structural variable. Lack of selectivity among tree sizes for within each habitat, the degree of flexibility travel (Fig. 4) supports the inference that of locomotor behavior could be evaluated macaques adopt the simple decision rule of and linked to habitat selection. choosing the nearest tree crown with the This study demonstrates that sampling structural strength and stability to support methodologies can be readily applied to imthem. This makes sense if the height at prove the description of vegetative structure which macaques travel is driven by the need as it relates to the functional ecology of arboto efficiently cross gaps between tree real locomotion. More detailed measures of crowns. An animal unable to cross wide habitat structure are an essential next step gaps, which has to cross numerous crowns in the field study of primate arboreal locoduring its daily ranging, might minimize its motion. Additional variables to be measured deviation from direct line travel between include the densities of available support points by choosing the nearest trees which types and the frequency of lianas that cross offer structural support. Studies of how tree gaps between trees. The variation in the size (dbh) relates to vertical stratification availability of structure within a forest haband to the interconnectedness of tree crowns itat, among habitats, among maturity LOCOMOTOR ECOLOGY OF GIBBONS AND MACAQUES phases, and across an animal's geographic range, and its behavioral responses, and that of sympatric species to this variation, should be essential elements in studies that explore the fundamental problems of primate ecology, such as habitat selection, niche relationships and limits to the distributions of species. Much more detailed, quantitative analysis of the structural environment in which arboreal mammals live is necessary to separate hypotheses of the function of postcranial morphology that are central to the study of primate evolutionary ecology. We argue that the approach adopted in this study, in which selectivity analyses is used to demonstrate preferences for canopy structures and identify constraints to canopy use, is essential to testing these ecological and evolutionary hypotheses. ACKNOWLEDGMENTS We thank our sponsors, the Center for Research and Development in Biology, of the Indonesian Institute of Sciences (LIPI), in particular Dr. Soetikno Wijoatmodjo, Ms. Moertini Atmowidjojo and Mr. H. Napitupulu, and the Subdirectorate of Nature Conservation (PHPA) of the Ministry of Forestry, in particular Dr. E. Sumardja and Mr. H. Prayitno, for permission to conduct research in the Gunung Palung National Park and for logistic support. We thank L. Curran, T. Laman, and P. Palmiotto for assistance at Cabang Panti, and especially D. Lawrence for support during every phase of this research. C. Chapman, T. Goldberg, S. Zens and especially K. Hunt and three anonomous reviewers kindly helped clarify and refine drafts of this paper. M. Leighton appreciates grants from the National Science Foundation (BNS-8409299), the Conservation Food & Health Foundation and the John Merck Fund in support of this research. The Goelet Fund and the Ford Foundation Fund for Undergraduate Research, both of Harvard University, the Explorers' Club Youth Activity Fund, and the Research Experience for Undergraduates (REU) program of the NSF supported C. Cannon. 523 LITERATURE CITED Boinski S (1989) The positional behavior and substrate use of squirrel monkeys: Ecological implications. J. Hum. Evol. 18:659. Cant JGH (1986) Positional behavior of spider and howling monkeys: Field study and evolutionary interpretation. Folia Primatol. 46:l-14. Cant JGH (1987) Positional behavior of female Bornean orangutans (Pongo pygmaeus). Am. J . Primatol. 12: 71-90. Cant JGH (1988) Positional behavior of long-tailed macaques (Macacu fuscicularis) in northern Sumatra. Am. J . Phys. Anthropol. 76:29-37. Cant JGH (1992) Positional behavior and body size of arboreal primates: A theoretical framework for field studies and a n illustration of its application. Am. J. Phys. Anthropol. 88:273-283. Clutton-Brock TH, and Harvey PH (1984) Comparative approaches to investigating adaptation. In J R Krebs and NB Davies (eds.): Behavioural Ecology:An Evolutionary Approach. Oxford: Blackwell Scientific Publications, pp. 7-29. Cock MJW (1978) The assessment of preference. J. h i m . Ecol. 47:805-816. Conover WJ (1980) Practical Nonparametric Statistics. 2nd. ed. New York John Wiley & Sons. Crompton RH (1984) Foraging, habitat structure, and locomotion in two species of Galago. In PS Rodman and JGH Cant (eds.): Adaptations for Foraging in Nonhuman Primates. New York: Columbia Univ. Press, pp. 73-111. Curtin SH (1977)Niche separation in sympatric Malaysian leaf-monkeys (Presbytis obscura and Presbytis melulophus). Ybk. Phys. Anthropol. 20:421--150. Fleagle J G (1976) Locomotion and posture of the Malayan siamang and implications for hominoid evolution. Folia Primatol. 26:245-269. Fleagle J G (1978) Locomotion, posture and habitat utilization in two sympatric, Malaysian leaf-monkeys (Presbytis obscura and Presbytis melalophus). In GG Montgomery (ed.): The Ecology of Arboreal Folivores. Washington, D.C.: Smithsonian Institution Press, pp. 243-251. Fleagle J G (1985) Size and adaptation in primates. In WL Jungers (ed.): Size and Scaling in Primate Biology. New York: Plenum Press. Fleagle JG, and Mittermeier RA (1980) Locomotor behavior, body size, and comparative ecology of seven Surinam monkeys. Am. J. Phys. Anthropol. 52:301314. Ganzhorn J U (1989) Niche separation of seven lemur species in the eastern rainforest of Madagascar. Oecologia 79:279-286. Garber PA (1984) Use of habitat and positional behavior in a neotropical primate, Suguznus oedipus. In PS Rodman and JGH Cant (eds.): Adaptations for Foraging in Nonhuman Primates. New York: Columbia Univ. Press, pp. 113-133. Gautier-Hion A, Gautier JP, and Quris R (1981) Forest structure and fruit availability as complementary factors influencing habitat use by a troop of monkeys 524 C.H. CANNON AND M. LEIGHTON (Cercopithecus cephusi. Rev. Ecol. (Terre et Vie) 35: 511-536. Gittins SP (1982) Feeding and ranging in the agile gibbon. Folia Primatol. 38:39-71. Gittins SP (1983) Use of the forest canopy by the agile gibbon. Folia Primatol. 40:134-144. Grand TI (1984) Motion economy within the canopy: Four strategies for mobility. In PS Rodman and JGH Cant (eds.): Adaptations for Foraging in Nonhuman Primates. New York: Columbia Univ. Press, pp. 5 P 72. Hunt KD (1992) Positional behavior of Pun troglodytes in the Mahale Mountains and Gombe Stream National Parks, Tanzania. Am. J . Phys. Anthropol. 87: 8S-105. Jacobs J (1974) Quantitative measurement of food selection. Oecologia 14:413-417. Leighton D (1987) Gibbons: Territoriality and monogamy. In BB Smits, DL Cheney, RM Seyfarth, RW Wrangham, and ‘IT Struhsaker (eds.): Primate Societies. Chicago: Univ. of Chicago Press, pp. 135-145. MacKinnon JR, and MacKinnon KS (1978) Comparative feeding ecology of six sympatric primates in West Malaysia. In D J Chivers and J H Chivers (eds.): Recent Advances in Primatology. New York: Academic Press, pp. 304-321. MacKinnon JR, and MacKmnon KS (1980) Niche differentiation in a primate community. In DJ Chivers (ed.): Malayan Forest Primates. New York: Plenum Press, pp. 167-190. Mitani J (1990) Demography of agile gibbons (Hylobutes agilis). Inter. J. Primatol. 11:411-424. Mittermeier RA (1978) Locomotion and posture in Ateles geoffroyi and Ateles paniscus. Folia Primatol. 30:161193. Morbeck ME (1977) Positional behavior, selective use of habitat substrate and associated non-positional behavior in free-ranging Colobus guereza. Primates 18: 35-58. Mueller-Dombois D, and Ellenberg H (1977) Aims and Methods of Vegetation Ecology. New York John Wiley. Napier J R (1966) Stratification and primate ecology. J. Anim. Ecol. 35:411. Napier JR (1967) Evolutionary aspects of primate locomotion. Am. J . Phys. Anthropol. 27~333-342. Pennycuick CJ (1979) Energy costs of locomotion and the concept of “foraging radius”. In ARE Sinclair and MN Norton-Griffiths (eds.): Serengeti, Dynamics of an Ecosystem. Chicago: Univ. of Chicago Press, pp. 164-184. Pound JA (1991) Habitat structure and morphological patterns in arboreal vertebrates. In SS Bell, ED McCoy, and HR Mushinsky (eds.): Habitat Structure: Physical Arrangements of Objects in Space. New York: Chapman and Hall, pp. 109-119. Prueschoft H, and Demes B (1984) Biomechanics of brachiation. In H Prueschoft, DJ Chivers, WY Brockelman, and N Creel (eds.): The Lesser Apes. Edinburgh: Edinburgh Univ. Press. Richards PW (1952) The Tropical Rain Forest. Cambridge, U.K.: Cambridge Univ. Press. Ripley S (1967) The leaping of langurs: A problem in the study of locomotor adaptation. Am J . Phys. Anthropol. 26: 149-1 70. Rodman PS (1978) Diets, densities and distributions of Bornean primates. In GG Montgomery (ed.): The EcoIogy ofArboreal Folivores. Washington, D.C.: Smithsonian Institution Press, pp. 465430. Snedecor GW, and Cochran WG (1967) Statistical Methods. Ames, Iowa: Iowa State University. Temerin LA, and Cant JGH (1983) The evolutionary divergence of Old World monkeys and apes. Amer. Nat. 122:335-351. van Schaik CP, and van Noordwijk M (1986) The hidden costs of sociality: Intra-group variation in feeding strategies in Sumatran long-tailed macaques (Macuca fascicularis). Behavior 99t296-319. Whitmore TC (1984) Tropical Rain Forests of the Far East. Oxford: Clarendon. Whitten AJ (1982) A numerical analysis of tropical rain forest, using floristic and structural data, and its application to an analysis of gibbon ranging behaviour. J . Ecol. 70.949-271. Yoneda M (1984) Comparative studies on vertical separation, foraging behavior and traveling mode of saddle-backed tamarins (Scguinus fuscicollisi and redchested moustached tamarins (S. labiatus) in Northern Bolivia. Primates 25:414-422.