American Journal of Primatology 70:927–938 (2008) RESEARCH ARTICLE Food Preferences of Wild Mountain Gorillas JESSICA GANAS1, SYLVIA ORTMANN2, AND MARTHA M. ROBBINS1 1 Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany 2 Leibniz Institute for Zoo and Wildlife Research, Berlin, Germany Determining the nutritional and phenolic basis of food preference is important for understanding the nutritional requirements of animals. Preference is a measure of which foods would be consumed by an animal if there was no variation in availability among food items. From September 2004 to August 2005, we measured the food preferences of four wild mountain gorilla groups that consume foliage and fruit in Bwindi Impenetrable National Park, Uganda, to determine what nutrients and phenols are preferred and/or avoided. To do so, we asked the following questions: (1) Which plant species do the gorillas prefer? (2) Considering the different plant parts consumed of these preferred species, what nutrients and/or phenols characterize them? (3) Do the nutritional and phenolic characteristics of preferred foods differ among gorilla groups? We found that although some species were preferred and others were not, of those species found in common among the different group home ranges, the same ones were generally preferred by all groups. Second, all groups preferred leaves with relatively high protein content and relatively low fiber content. Third, three out of four groups preferred leaves with relatively high sugar amounts. Fourth, all groups preferred pith with relatively high sugar content. Finally, of the two groups tested, we found that the preferred fruits of one group had relatively high condensed tannin and fiber/sugar contents, whereas the other group’s preferred fruits were not characterized by any particular nutrient/phenol. Overall, there were no differences among gorilla groups in nutritional and phenolic preferences. Our results indicate that protein and sugar are important in the diets of gorillas, and that the gorillas fulfil these nutritional requirements through a combination of different plant parts, shedding new light on how gorillas balance their diets in a variable environment. Am. J. Primatol. 70:927–938, 2008. r 2008 Wiley-Liss, Inc. Key words: nutritional ecology; foraging strategy; protein; sugar; Bwindi Impenetrable National Park INTRODUCTION Animal foraging strategies are based on a complex suite of variables including nutritional requirements, spatial and temporal availability of food, and the amount of energy and time needed to locate and consume food resources [Schoener, 1971; Stephens & Krebs, 1986; Westoby, 1974]. Understanding a species’ foraging strategy includes an examination of food preference, choice, and selectivity. Food preference is a measure of food consumption with the assumption that there is no variation in availability among food items in the animal’s diet [Chesson, 1983; Johnson, 1980]. Many authors claim to measure preference; however, often these calculations do not take into consideration the (equal) availability of dietary items [Calvert, 1985; Hayward et al., 2006; Norscia et al., 2006]. Food preference differs from food choice because although food choice investigates how the attributes of each food species (their differing availabilities and nutrient compositions) may influence the decision of what an animal consumes, preference controls for differences in r 2008 Wiley-Liss, Inc. availability and then calculates which species would be chosen over another. Another measure of foraging behavior, selectivity, measures why certain foods are not consumed by comparing the nutritional contents of foods eaten with those not eaten. Investigating food preference is important because it can lend insight into the nutritional requirements of an animal, which is vital to reproduction, fitness, and survival [Altmann, 1998; Orians & Wittenberger, 1991; Schoener, 1983]. Additionally, information on which nutrients and foods are preferred by an animal can tell us which food species Contract grant sponsors: Max Planck Society; Berrgorilla & Regenwald Direkthilfe; The John Ball Zoo; Leakey Foundation. Correspondence to: Jessica Ganas, Royal Society for the Protection of Birds and the Gola Forest Programme, Kenema, Sierra Leone. E-mail: firstname.lastname@example.org Received 5 August 2007; revised 28 February 2008; revision accepted 19 May 2008 DOI 10.1002/ajp.20584 Published online 19 June 2008 in Wiley InterScience (www. interscience.wiley.com). 928 / Ganas et al. may influence feeding competition and habitat utilization, and which food species and habitats should be considered in management and conservation efforts. Because the availability of a particular food may influence whether it is consumed or not (independent of its nutritional content), determining what nutrients are required by analyses of food choice and/or selectivity may not give a true representation of an animal’s nutritional requirements. For example, during periods of low food availability, animals may eat poor quality, but highly available foods to subsist during lean times. Thus, preference is the most accurate method of investigating the nutritional requirements of animals. Although measures of food preference can be easily conducted in captivity, many studies usually only offer a few food options per experiment, which does not represent what a wild animal experiences, especially for those species with a large dietary repertoire [Benz et al., 1992; Laska et al., 2000; Remis, 2002]. Furthermore, captive studies differ from wild studies in that captive studies do not control for the diet of the animals outside of the trials (which can influence what is preferred), whereas those in the wild can take into consideration the entire diet. Conversely, food preference can be difficult to measure in the wild because there is variation in the availability among most food species, whereas preference controls for availability. However, it is possible to estimate food preferences of wild animals using indices that quantify diet choices based on relative equal availability [Ivlev, 1961; Johnson, 1980]. Although these are not measures of choices made by animals among equally available foods as with experiments, the index is the closest reflection of preference that is possible for wild animals [Chesson, 1983; Johnson, 1980]. Gorillas (Western: Gorilla gorilla and Eastern: Gorilla beringei) are the largest extant primate species; they have an enlarged and highly ciliated hindgut, which facilitates processing of some plant fiber for energy [Milton, 1984; Remis, 2000]. They consume both foliage (nonreproductive plant parts from herbs, shrubs, and trees) and fruit to varying degrees depending on food availability [Ganas et al., 2004; Rogers et al., 2004; Watts, 1984]. Additionally, gorillas generally consume a particular part of a plant (i.e. leaves, pith, or bark) and not the entire plant, suggesting that they selectively consume these parts for specific nutritional reasons. Our knowledge of gorilla food preferences is limited. Food preference experiments on western gorillas conducted in zoos (with fruits and vegetables) found that preferred foods were relatively high in sugar and energy with moderate levels of tannins, with avoided foods having a relatively high protein content [Remis, 2002; Remis & Kerr, 2002]. However these studies did not take into account the regular diet of the gorillas, which can influence which foods Am. J. Primatol. are preferred during the trials. Furthermore, no preference trials were conducted with foliage, a staple of gorilla diets. In a preference study of wild mountain gorillas in Rwanda, Vedder  found no correlation between food preference and levels of protein although research on another gorilla group in the same population found that protein and digestibility positively influenced food choice [Watts, 1983]. To date, we lack information on the food preferences of wild gorilla groups that consume both fruit and foliage and on the nutritional and phenolic attributes that are associated with these preferences. Mountain gorillas (Gorilla beringei beringei) in Bwindi Impenetrable National Park, Uganda, consume both foliage and fruit from a diversity of plant species and experience differences in food availability within and between locations in the park with corresponding diet variability [Ganas et al., 2004]. A study on food choice, which examined how consumption was influenced by both the differing availability of food and food nutritional content, found that Bwindi gorillas chose individual food species based on their relatively high abundance, relatively high sugar contents, and relatively low digestion inhibitor contents [Ganas et al., 2008]. The goal of this study was to measure one facet of Bwindi gorilla foraging strategy, food preference [other aspects of their foraging behavior are treated elsewhere; Ganas et al., 2008]. We asked the following questions considering four gorilla groups at two separate locations: (1) Which plant species do the gorillas prefer? (2) Given the different plant parts consumed of these preferred species, what nutrients and/or phenols characterize them? (3) Were there differences among groups in preference? We predicted that gorillas would prefer leaves and pith with relatively high protein and sugar contents, fruit with relatively high sugar, energy and/or fat contents while avoiding fibers (cellulose, hemicellulose, lignin) and phenols (total phenols, total tannins, condensed tannins) in all food types. As animal nutritional requirements should be the same within a species, we also predicted that preferences would not differ among groups, despite the fact that food availability differed among groups’ home ranges. METHODS Study Groups Data on diet were collected from four habituated gorilla groups from September 2004 to August 2005. Three groups, Mubare, Habinyanja, and Rushegura ranged around Buhoma (1,450–1,800 m). Because groups are used for an ecotourism program, the Uganda Wildlife Authority limits direct contact with these groups and we were not able to conduct direct observations. The fourth group, Kyagurilo, ranges near Ruhija (2,100–2,500 m), and is habituated for research. Although direct observations are possible Gorilla Food Preferences / 929 here, we used indirect methods on all groups for comparative purposes. For details of the study site see McNeilage et al. . All research adhered to the protocols and legal requirements in Uganda. Diet All weaned individuals of a group make nests in close proximity to one another to form a nest site every night. During the day, the gorillas move and feed between each night’s nest location. During this time, the gorillas create obvious trails by trampling vegetation, discarding food, and defecating, which facilitates documentation of the animals’ daily diet. To quantify the frequency of herbs in the gorillas’ diet, we followed each groups’ main trail on a daily basis and recorded observations of each plant species remains left behind and plant part consumed (i.e. leaf, pith, peel [peel is the outer layer of an herb’s stem]). These trails and feeding spots are easy to follow and easy to distinguish from other animals with the assistance of experienced trackers. Although this method does not document the actual amount of food ingested, it is the best approach when indirect means are necessary and is commonly used in dietary studies of gorillas [Calvert, 1985; McNeilage, 1995]. The monthly frequency of each plant species found on these trails was then calculated to represent the relative percentage of foods in a diet [Frequency of species A 5 ] of feeding spots of Species A/total number of feeding spots 100%; following Calvert, 1985; McNeilage, 1995]. We defined ‘‘important’’ herb species as those occurring in Z1% frequency in any month. Although the gorillas eat other plant parts such as flowers and bark, these foods were relatively infrequently eaten. Therefore, these foods are likely not ‘‘important’’ in reference to macronutrient and phenol contents, the focus of this study, and were not included in our analyses. Food species that were consumed infrequently were likely eaten for other reasons such as mineral content or medicinal purposes and for studies that focus on these components, it may be important to consider all foods eaten and use different methodologies [Huffman, 1997; Rothman et al., 2006]. Additionally, for each group there were 0–3 plant species for which peel was important in the gorillas’ diet but owing to the small sample sizes, we were unable to do additional analysis to determine which nutrients they preferred in peel. On average, we analyzed trail signs 19 days per month per group (Mubare monthly range 5 14–26; Habinyanja monthly range 5 15–23; Rushegura range 5 17–25; Kyagurilo range 5 14–19). To determine the frequency and species of fruit consumed, we collected fecal samples from each groups’ night nests, and recorded whether the sample (based on size) was from a silverback, an adult female/blackback (indistinguishable), and a juvenile (defined as sleeps in his/her own nest, sexually immature) nest each day (Schaller, 1963). After collection, fecal samples were washed through a 1 mm sieve and seed species were identified. Important fruit species were defined as those occurring in Z1% of samples per group in any month [modified from Ganas et al., 2004; Remis, 1997]. Because gorillas in Bwindi have not been seen to spit out seeds in over 8 years of observation and also because the vast majority of fruits consumed during this study period were relatively small and consumed in their entirety [mean width of fruit 5 7.7 mm, Ganas, unpublished data; Robbins, personal observation] we assumed that if the gorillas ate fruit, it would be detected in the fecal samples. There were no differences in the frequency of fruit consumption between age and sex classes (using a w2 test) and thus only adult female samples were used in the analysis. Considering all groups, we collected an average of 25 (range 15–26) samples per group per month (Mubare mean 5 23.3, monthly range 5 17–26; Habinyanja mean 5 24, monthly range 17–29; Rushegura mean 5 23.8, monthly range 18–30; Kyagurilo mean 5 26.9, monthly range 12–30). Food Availability Temporal We measured the temporal biomass availability of 20 herb species (Buhoma 5 18, Ruhija 5 11) considered important to the gorillas (see above) in 89 1 m2 permanent plots (Buhoma 5 51, Ruhija 5 38) in the forest [Ganas et al., in press]. Plots were established in areas of high herb density within various areas of the gorillas’ range. An average of 16.8 individual plants per species per month were monitored in Buhoma (range 2.8–37.9, SD 5 11.5), whereas in Ruhija, an average of 24.2 individuals per species was monitored (range 8.5–60.1, SD 5 15.3). To estimate the biomass of herbs, at approximately the same time every month, for each permanent plot, we first took measurements of particular plant parts. Next, we harvested 40 individuals of each species of varying lengths from outside the plots. For each plant, we measured the length of the plant stem or counted the number of leaves on the individual plant and recorded the wet weight of the part eaten by the gorillas. We then dried the plant parts in sheds with charcoal stoves. After they were dry, we again recorded the weight of each individual. To determine whether there was a significant relationship between the length of the stem or the number of leaves and weight, we plotted length/number of leaves against the weight (one test each for wet and dry weight) and calculated a linear regression that was forced through the origin [Zar, 1999]. We found a significant relationship between these variables, and regression equations were produced that were used to calculate the biomass of Am. J. Primatol. 930 / Ganas et al. each herb species in the plots for each month. This also allowed us to calculate the monthly changes (%) in their temporal availability. For further details, see Ganas et al. [in press]. To calculate the temporal availability of fruit, on a biweekly basis within each location, we monitored 397 trees and herbs of 40 species, 13 of which were found at both sites (211 [mean ] per species 5 7.3, SD 5 4.6] and 186 [mean ] per species 5 7.8, SD 5 4.1] at Ruhija and Buhoma, respectively), which have been known to provide fruits for gorillas. For each species, we recorded the percent abundance of ripe fruit in the crown scoring between zero and four (0 5 0%, 1 5 1–25%, 2 5 26–55%, 3 5 51–75%, and 4 5 76–100% of crown covered) following Sun et al. . The fruits of Trichilia sp. were not previously known to be consumed by gorillas and we did not record phenological data on them and this species was excluded from the analysis. Overall, Trichilia sp. was of low to medium importance for the two groups analyzed (annual % frequency eaten: Mubare: 2.2%, Habinyanja 9.4%). Additionally, the majority of Ficus spp. (excluding F. capensis) was strangler figs and a fruit availability index (FAI) could not be calculated. Spatial availability To determine the spatial distribution of herbs and fruit, we cut and measured 102 and 54 transects of 200 m in length at Buhoma and Ruhija, respectively, placing one transect each within a 500 m2 grid overlaid onto a map of each study location [GreigSmith, 1983]. For each transect, we placed nested quadrats (1 and 10 m2) on alternate sides in intervals of 20 m for 10 quadrats (total transect length 5 200 m) per transect. In these quadrats we documented herb biomass (using the same methods from the permanent plots, length/] leaves) and tree (density and diameter at breast height) availability. To then calculate the total biomass of each herb species in each home range per month, we applied the biomass estimates (regression equations) and the corresponding monthly changes in biomass (%) recorded from the permanent plots to the measurements from the transects. To determine fruit availability at each location for each biweekly period, a score of fruit abundance was calculated using an FAI [following Nkurunungi et al., 2004; calculated as the product of the mean DBH (of phenology trees of each fruit species eaten by the gorillas), density of each species at each location (recorded from the transects), and their mean biweekly abundance score value from the phenology study]. To get a value for total fruit availability for each location, we summed the individual FAI scores for each biweekly period. Am. J. Primatol. Nutritional Sampling We collected 42 important food species (fruit, leaves, pith, peel) consumed by the gorillas at both locations. Fruit crops produced by the herb Rubus sp. were very small and samples could not be obtained for nutritional analysis (this fruit accounted for 1% yearly frequency in diet, and it was not consumed by every group) All food items collected for analyses were processed in a way that mimicked which parts the gorillas consumed. For example, if the gorillas ate the pith from a particular species, we collected only pith. Because indirect observations were made or entire plants eaten, often it was difficult to collect plant parts from the exact tree or herb the part was consumed from. Nonetheless, every attempt was made to sample food items from the specific areas where the gorillas fed. We also made multiple collections of plant species when possible by collecting them from different areas of the gorillas’ home range (where they had been feeding) and collected them during two different wet seasons, mixing samples before analysis. Owing to the difficulty of collecting fruit from tall canopy trees, most fruit samples came from a single tree or location. Because of the short-term availability of fruits in Bwindi, fruit was sampled once when available. We stored samples in cryo tubes, froze them in liquid nitrogen, and then freeze dried them. Dried samples were then stored in a cool, dry place until they were sent to the Institute of Zoo and Wildlife Research for nutritional analysis and the University of Hohenheim for phenolic determination. Phytochemistry Analyses All samples were ground before analysis using a 1 mm screen. Dry matter (DM) content was determined by drying a portion of the sample at 1051C overnight. All data are given as % DM. Samples were analyzed for the following macronutrients using standard techniques: Nitrogen was determined by complete combustion (Dumas combustion) at high temperature (about 9501C) in pure oxygen, using a Rapid N III analyzer (Elementar Analyser Systeme GmbH, Hanau, Germany) and a factor of 6.25 was used for conversion into protein (crude protein (%DM) 5 6.25 N (%DM)). Starch, D-glucose, D-fructose, and sucrose were determined with commercialized enzymatic tests (UV method; R-Biopharm AG, Darmstadt, Germany). Lipids were extracted with ethyl ether using a fully automatic Soxhlett system (Soxtherm; Gerhardt Laboratory Systems, Königswinter, Germany), and gross energy was determined by burning a sample of DM in pure oxygen atmosphere in a bomb calorimeter (C5003 bomb calorimeter; IKA-Werke GmbH & Co. KG, Staufen, Germany). The heat produced is measured in kJ/g DM. Detergent Fiber Analysis was performed following van Soest  with neutral detergent fiber Gorilla Food Preferences / 931 (NDF), acid detergent fiber (ADF), and acid detergent lignin (ADL) being determined sequentially from each sample using an Ankom Fiber Analyser 220 (Ankom Technology, Macedon, NY). Hemicellulose (NDF–ADF) and cellulose (ADF–ADL) were calculated by weight difference. Total phenols were determined according to Makkar et al. . Determination of condensed tannins followed Porter et al. . Statistical Analyses To reduce the large number of macronutrient and phenolic variables tested per herb and fruit species (protein, starch, glucose, fructose, sucrose, water soluble carbohydrates (WSC; sum of glucose, fructose, sucrose), cellulose, hemicellulose, lignin, NDF (sum of hemicellulose, cellulose, lignin), ADF (sum of cellulose and lignin), fat, energy, total phenols, total tannins, condensed tannins), we first inspected correlations between them and in the case of two variables being highly correlated to one another (absolute correlation coefficient40.75) we removed one of them (they included starch, fructose, glucose, lignin, ADF, energy, total tannins). The fruit and herb data sets were treated separately as the nutritional and phenolic components significantly differed between the two food groups [Ganas et al., in preparation]. The individual remaining variables were then subjected to a principal components analysis (PCA). Reducing variables by using correlation tests and PCAs decreases the possible instability in results of subsequent analyses of data sets that consist of large number of correlating variables. It also removes redundant information resulting in a more stable result. Even if some of the variables that are put into the PCA (and group onto the same principal component) are related in some way (i.e. fat content and energy content) putting these variables into a PCA is valid [Field, 2005]. In the case of a principal component where only a single variable had its highest loading, we reran the PCA without that variable and included the variable directly into subsequent analyses. Both the fruit and herb PCAs were justified [Kaiser–Maier–Olkin measure of sampling adequacy: herbs: 0.48; fruit: 0.64; Bartlett’s test of sphericity: herbs: w2 5 117.2, df 5 36, Po0.001; fruit: w2 5 23.3, df 5 15, Po0.08; Field, 2005]. The PCAs revealed three herb and two fruit principal components (with eigenvalues in excess of one), which together explained 71.2% (herb) and 68.2% (fruit) of the total variance. The loadings for the three components for herbs were (a) NDF, hemicellulose, cellulose, condensed tannin, (b) protein, fat, total phenols, and (c) WSC and sucrose (Table I(A)). The loadings for the two components for fruit were (a) NDF and lignin (1)/WSC ( ) and (b) fat and energy (1)/sucrose ( ) (Table I(B)). Condensed tannin is included in the fruit preference analysis as its own variable rather than as a component because it grouped on its own in the PCA. We calculated preference scores for each food species and part consumed (fruit, leaves, and pith) using Ivlev’s electivity index, which is a measure of foraging behavior in relation to food availability [Ivlev, 1961]. Traditionally it is used to determine whether a food species is consumed proportionally to its availability. It can also be used as a relative measure of preference by considering all food species eaten and then ranking them according to both frequency in the diet and their corresponding availability [Johnson, 1980]. By ranking foods rather than using absolute values, it circumvents the problem that arises from accurately measuring rare foods [Johnson, 1980; Lechowicz, 1982]. Although preference measures should control for relative availability, owing to the nature of this index, availability still somewhat influences preference scores. For example, species that have a relatively high abundance may not ever be considered highly preferred regardless of how much is consumed, and very rare foods will often be considered highly preferred [Johnson, 1980; Maitland, 1965]. Lechowicz  reviewed a variety of preference indices, examining the pros and cons of each and determined that the majority provided useful measures of feeding preference, including the one used here. Despite some limitations, we chose Ivlev’s electivity index because it works best with large sample sizes, it has been used often by other authors [Vedder, 1989; Watts, 1984], and it is one of the most appropriate indices available for wild animals [Chesson, 1983; Johnson, 1980; Lechowicz, 1982]. To calculate the index for each biweekly (fruit) or monthly (herb) period, a rank was assigned for both diet frequency and food availability of each species, resulting in 12 (herb) or 24 (fruit) preference scores for each species. Ranks were between 1 and the highest number of fruit or herb species available (considering only those that were consumed) in the time period. The greater the diet frequency/availability, the higher the rank score was. For species that shared the same availability or diet frequency score, tied ranks were assigned. Based on these ranks, we then calculated preference using the formula: Ivlev’s electivity index 5 (rd na)/(rd1na), where rd 5 rank of food item in diet na 5 rank of food item in the home range. Scores between 1 and 0 indicated that a food was not preferred, a score of 0 signified neutrality, and a score between 0 and 1 indicated that it was preferred. Rather than creating categories of ‘‘highly preferred’’ or ‘‘medium preferred’’’ that are not rooted in biological meaning, food preference scores should simply be viewed on a relative continuum where one is regarded as more preferred than Am. J. Primatol. 932 / Ganas et al. TABLE I. Results of the Principal Components Analyses on the (A) herb and (B) fruit nutrient and phenolic values Trait Component Component Component 1 2 3 RESULTS (A) Protein WSC Sucrose NDF Hemicellulose Lignin Fat Total phenols Condensed tannins Eigenvalue % variance explained 0.135 0.324 0.002 0.837 0.549 0.766 0.156 0.172 0.732 2.77 30.8 0.618 0.195 0.067 0.412 0.071 0.269 0.816 0.857 0.139 2.09 23.2 (B) Sucrose WSC Fat Energy Lignin NDF Eigenvalue % variance explained 0.160 0.857 0.441 0.345 0.596 0.862 3.03 50.5 0.755 0.157 0.762 0.636 0.581 0.069 1.06 17.8 0.597 0.773 0.719 0.091 0.184 0.418 0.299 0.152 0.458 1.55 17.2 WSC, water soluble carbohydrates (sum of glucose, fructose, sucrose); NDF, neutral detergent fiber (sum of cellulose, hemicellulose, lignin). Indicated are loadings of the variables on the principal components derived. Bold values indicate the largest absolute loading per variable. another based on its preference score. For most fruit species, there were periods when they were unavailable and thus no preference score was calculated for these time periods. Because we were only able to obtain both nutritional and availability data for 40% of the fruit species eaten by Rushegura and because during the study period Kyagurilo group only ate 1–2 species of fruit per time period, we did not analyze these groups’ fruit preferences. To determine what nutritional and phenolic attributes characterized preferred foods, we correlated each species preference scores for each time period with its nutrient and phenolic composition (the PCA factor scores; each plant part has the same PCA factor score during each biweekly/monthly period as we sampled each plant species once, but the diet frequency values differ per period) using a Pearson’s correlation test. A correlation coefficient between 0 and 1 indicated preference for a particular nutrient/phenol, and a coefficient between 1 and 0 indicated that it was not preferred. We then used a one-sample t-test to determine whether these monthly correlation coefficients on average, equaled zero (a significant result means that they were not and thus a significant correlation). We controlled for multiple testing by using a Fisher’s Omnibus test; Am. J. Primatol. one test for the herb preferences (leaves, pith) and one for fruit. We used two separate tests owing to the two different sets of PCA analyses. Analyses were performed using SPSS 13.0. Plant Species Preferences Herb foods that had relatively high preference scores (40.35) were Aframomum angustifolia pith, Aframomum sanguinum pith, Basella alba leaves, Desmodium repandum leaves, Ipomea wightii leaves, Mormodica calantha leaves, and Palisota mannii pith. Fruit that had relatively high preference scores (40.20) were Ficus capensis and Prunus africana. Species that were highly abundant, which may have resulted in low preference scores, included Cassine aethiopica, Mimulopsis solmsii, and Mimulopsis arborescens. There were no major differences among groups in these scores. For individual food species preference scores, see Tables II and III. Nutritional and Phenolic Characteristics of Preferred Foods A Fisher’s Omnibus test confirmed that the following results are not simply owing to chance (w2 5 148.4, df 5 24, Po0.001). Leaves All groups significantly preferred leaves relatively high in protein, fat, and phenols (Table IV; Mubare one sample t-test t 5 5.6, df 5 10, Po0.001; Habinyanja t 5 5.2, df 5 10, Po0.001; Rushegura t 5 6.1, df 5 10, Po0.001; t 5 11.9, df 5 11, Po0.001; Kyagurilo t 5 5.8, df 5 11, Po0.001) and avoided fiber (Mubare t 5 12.6, df 5 10, Po0.001; Habinyanja t 5 23.2, df 5 10, Po0.001; Rushegura t 5 16.8, df 5 10, Po0.001; Kyagurilo t 5 20.9, df 5 11, Po0.001). Habinyanja, Rushegura, and Kyagurilo preferred leaves relatively high in sugar, whereas Mubare did not (Table IV; Mubare t 5 0.6, df 5 10, P 5 0.57; Habinyanja t 5 3.4, df 5 10, P 5 0.006; Rushegura t 5 2.6, df 5 10, P 5 0.03; Kyagurilo t 5 2.9, df 5 11, P 5 0.02). Pith Mubare preferred pith relatively high in protein, whereas the other groups did not (Table IV; Mubare t 5 3.8, df 5 10, P 5 0.003; Habinyanja t 5 0.3, df 5 10, P 5 0.78; Rushegura t 5 1.2, df 5 10, P 5 0.25). All groups preferred pith high in sugar (Table IV; Mubare t 5 5.0, df 5 10, P 5 0.001; Habinyanja t 5 3.6, df 5 10, P 5 0.005, Rushegura t 5 4.8, df 5 10, P 5 0.001). Gorilla Food Preferences / 933 TABLE II. For each group, the diet frequency, the corresponding yearly average preference rank, food availability, its corresponding average yearly rank, and preference scores of each herb species used in the preference analysis Group Plant species Yearly diet freq. (%) Avg. diet rank (SD) Median avail. g/m2 Avg. avail. rank (SD) ] mo. Pith Leaves Leaves Leaves Pith Leaves Leaves Peel Pith Peel Leaves Pith Leaves Pith Leaves Leaves Leaves Leaves Leaves Pith Pith 5.1 0.3 4.2 13.5 1.7 1.7 0.3 6.6 12.4 7.4 9.9 0.9 3.0 0.9 1.8 0.9 0.7 0.8 0.9 0.7 1.5 13.1(4.7) 3.5(2.6) 12.8(4.0) 17.9(7.0) 7.9(3.2) 8.9(3.7) 4.6(2.2) 15.0(4.5) 17.5(5.2) 15.1(4.6) 16.7(5.1) 6.2(3.0) 11.0(3.5) 4.8(4.1) 8.8(5.1) 5.9(3.3) 4.8(3.2) 5.9(3.2) 4.2(2.9) — — 0.02 0.003 0.05 0.10 0.01 0.05 0.08 0.12 1.9 0.22 5.4 0.04 4.5 0.04 2.4 0.12 0.31 0.22 0.48 — Not detected 4.9(1.9) 1.1(0.4) 5.5(2.4) 8.4(3.7) 3.3(1.2) 5.2(2.6) 2.8(1.4) 11.4(3.5) 16.9(4.9) 14.5(4.3) 17.2(5.1) 7.5(2.2) 18.2(5.5) 7.2(3.8) 15.5(7.2) 11.4(3.5) 9.4(4.5) 14.5(4.3) 11.2(3.8) — — 11 11 11 11 12 11 11 11 11 11 11 12 11 11 11 11 11 11 11 11 12 0.45 0.43 0.41 0.37 0.36 0.26 0.21 0.13 0.02 0.02 0.01 0.13 0.25 0.25 0.30 0.35 0.36 0.45 0.48 — — Pith Leaves Leaves Leaves Leaves Pith Leaves Peel Pith Leaves Peel Leaves Leaves Pith Leaves Leaves Pith Leaves Leaves Peel Pith Pith 0.7 3.7 10.0 2.1 0.4 6.0 0.3 5.4 16.0 1.2 5.4 0.7 5.0 1.1 2.3 0.8 0.3 0.6 0.7 0.7 1.2 1.4 6.8(3.1) 13.3(4.3) 17.5(5.1) 11.0(4.0) 5.3(3.6) 15.0(4.9) 4.8(3.4) 14.6(4.4) 18.3(5.4) 7.2(3.8) 14.3(4.8) 7.8(4.1) 13.8(4.3) 7.5(3.7) 8.1(3.2) 6.8(3.2) 3.5(3.6) 5.4(2.7) 4.8(3.0) 4.1(3.1) — — 0.003 0.10 0.03 0.02 0.003 0.03 0.05 0.10 2.4 0.12 0.20 0.16 4.8 0.05 3.2 0.2 0.02 0.10 1.01 1.01 Not detected — 1.4(0.7) 5.0(2.2) 7.0(2.5) 5.0(2.2) 1.8(1.0) 8.2(2.8) 3.8(1.5) 13.7(4.4) 17.9(5.2) 6.8(3.0) 14.1(4.3) 9.0(4.2) 18.0(5.4) 9.8(2.6) 17.5(4.4) 14.1(4.3) 7.0(3.2) 13.7(4.4) 14.8(6.8) 14.8(6.8) — — 12 11 11 11 11 11 11 11 11 11 11 11 11 12 11 11 11 11 11 11 12 11 0.59 0.46 0.43 0.39 0.38 0.29 0.03 0.03 0.01 0.01 0 0.09 0.13 0.18 0.21 0.37 0.39 0.45 0.53 0.60 — — Pith Leaves Leaves Pith Leaves Pith Peel Leaves Peel Pith Pith Leaves 0.9 2.6 10.5 1.0 4.7 4.4 7.4 0.2 7.2 1.8 22.8 0.7 5.0(4.3) 9.9(4.2) 15.5(5.0) 6.0(2.8) 11.9(4.6) 12.5(4.4) 14.4(5.0) 4.1(2.9) 13.3(4.3) 7.1(4.6) 17.2(5.1) 6.1(3.2) 0.002 0.03 0.13 0.01 0.16 0.07 0.16 0.17 0.06 0.03 4.2 0.16 1(0) 2.9(1.9) 7.3(2.7) 2.9(0.9) 8.5(3.4) 10.3(3.7) 12.4(3.9) 4.0(1.6) 12.3(4.5) 5.7(1.5) 16.9(4.9) 6.8(4.4) 12 11 11 12 11 11 11 11 11 12 11 11 0.54 0.50 0.36 0.29 0.16 0.10 0.07 0.05 0.05 0.04 0.01 0.02 Part eaten Avg. pref. score Mubare Palisota mannii Desmodium repandum Ipomea wightii Basella alba Aframomum angustifolia Mormodica calantha Rubus sp. Laportea aestuans Mimulopsis arborescens Urera sp. Triumfetta sp. Aframomum sp. Mimulopsis solmsii Pennisetum purpureum Ipomea sp. Laportea aestuans Gouania longispicata Urera sp. Mormodica foetida Piper capense Aframomum sanguinum Habinyanja Aframomum angustifolia Ipomea wightii Basella alba Mormodica calantha Desmodium repandum Palisota mannii Rubus sp. Urera sp. Mimulopsis arborescens Mormodica foetida Laportea aestuans Gouania longispicata Triumfetta sp. Aframomum sp. Mimulopsis solmsii Laportea aestuans Pennisetum purpureum Urera sp. Ipomea sp. Ipomea sp. Aframomum sanguinum Piper capense Rushegura Aframomum sanguinum Mormodica calantha Basella alba Aframomum angustifolia Ipomea wightii Palisota mannii Laportea aestuans Rubus sp. Urera sp. Aframomum sp. Mimulopsis arborescens Ipomea sp. Am. J. Primatol. 934 / Ganas et al. TABLE II. Continued Group Plant species Mormodica foetida Triumfetta sp. Gouania longispicata Ipomea sp. Urera sp. Laportea aestuans Desmodium repandum Pennisetum purpureum Piper capense Kyagurilo Mormodica calantha Basella alba Cardus sp. Rubus sp. Ipomea sp. Urera sp. Mimulopsis arborescens Mimulopsis solmsii Piper capense Triumfetta sp. Mimulopsis solmsii Urera sp. Ipomea sp. Mormodica foetida Part eaten Leaves Leaves Leaves Peel Leaves Leaves Leaves Pith Pith Leaves Leaves Stalk/pith Leaves Leaves Peel Pith Bark Pith Leaves Leaves Leaves Peel Leaves Yearly diet freq. (%) Avg. diet rank (SD) Median avail. g/m2 Avg. avail. rank (SD) ] mo. 1.0 5.8 0.4 0.5 0.7 0.5 0.4 0.9 0.9 6.0(3.1) 13.0(4.5) 4.7(3.1) 3.9(3.6) 5.7(2.5) 4.5(2.2) — — — 0.26 6.8 0.22 0.16 0.06 0.16 Not detected Not detected — 7.1(3.7) 15.5(4.6) 7.0(3.6) 6.8(4.4) 12.3(4.5) 12.4(3.9) — — — 11 11 11 11 11 11 — — 11 0.08 0.09 0.22 0.29 0.41 0.48 — — — 2.6 7.5 1.6 5.3 8.9 11.2 14.6 11.1 0.8 7.8 7.1 2.3 1.2 0.1 12.3(1.3) 8.3(1.8) 3.6(1.3) 7.0(1.5) 9.7(2.2) 11.3(2.2) 12.0(2.8) 11.4(1.9) 2.9(0.9) 9.3(1.8) 8.6(2.4) 4.5(1.2) 3.0(3.1) 1.2(0.4) 0.2 0.24 0.08 1.0 0.15 0.2 1.2 6.2 0.91 2.1 6.2 0.2 0.15 0.8 1.8(0.9) 3.1(0.9) 3.0(2.7) 5.2(0.8) 7.4(0.8) 9.3(0.6) 11.3(0.5) 13.5(0) 4.0(1.9) 11.7(0.5) 13.5(0) 9.3(0.6) 7.4(0.8) 4.4(1.4) 12 12 12 12 12 12 12 12 12 12 12 12 12 12 0.75 0.45 0.20 0.14 0.12 0.08 0.01 0.09 0.11 0.12 0.24 0.36 0.44 0.57 Avg. pref. score Food species are ranked from largest to smallest preference score. The average preference score is the average of the monthly preference scores. SD, standard deviation. When two parts are eaten from the same plant species (i.e. Laportea aestuans), the biomass of the two parts is added together. Thus, the median availability for L. aestuans peel includes the biomass from the leaves and peel, and vice versa. ‘‘Not detected’’ means that during the measurements of plant spatial availability, we did not encounter these plants in this particular home range, indicating that these species were at an extremely low density. For the majority of species in Buhoma, monitoring of their biomass began 1 month after the beginning of the study, thus availability is 11 months rather than 12, and was thus considered available every month of the study. Fruit A Fisher’s Omnibus test confirmed that the following results were not owing to chance (w2 5 51.7 df 5 16, Pr0.001). For the two groups tested, the Mubare group preferred fruits with a relatively large NDF and lignin/WSC ratio, whereas the Habinyanja group did not (Table IV; t 5 2.8, df 5 19, P 5 0.011; Habinyanja t 5 1.6, df 5 19, P 5 0.13). Neither group significantly preferred fat and energy/sucrose in fruit over the year (Table IV). Furthermore, periods in which there was a positive correlation between fruit preference and fat and energy/sucrose were mostly during the times that NDF and lignin/WSC was not preferred. This could indicate that this group preferred some type of energy source every month. The Mubare group preferred fruits with condensed tannin, whereas the Habinyanja group did not (Table IV; Mubare t 5 2.8, df 5 19, P 5 0.01; Habinyanja t 5 1.7, df 5 19, P 5 0.11). DISCUSSION Our research, which gives an insight into gorilla nutritional requirements, found that although the gorillas preferred some plant species over others, Am. J. Primatol. these preferences were related to particular nutrients and phenols. Groups generally preferred leaves with relatively high levels of protein, fat, phenols and sugar and low amounts of fiber, and pith with relatively high amounts of sugar. Fruit preferences were less clear, and the result that one group of gorillas preferred digestion inhibitors (condensed tannins and fiber) was possibly owing to simultaneously ingesting relatively high amounts of sugar. Despite differences in spatial and temporal variability in food availability among gorilla home ranges, there were no large differences in either preference for a particular plant species or preference for particular nutrients and phenols among gorilla groups. All groups preferred leaves with relatively high amounts of protein, fat, and phenols and relatively low amounts of fiber and condensed tannins, concurring with other studies of primate herbivores, which found that the protein/fiber ratio is an important component of their foraging strategy [Chapman et al., 2004; Ganzhorn, 1992; Milton, 1979; Oates et al., 1990]. Curiously, these results differ from similar work conducted on mountain gorilla preference in Rwanda (using the same methodology as this study), which found no relationship between preference and protein [Vedder, 1989]. Gorilla Food Preferences / 935 TABLE III. For each group, the diet frequency, the corresponding yearly average preference rank, food availability, its corresponding average yearly rank, and preference scores of each fruit species used in the preference analysis Group Plant species Yearly diet freq. (%) Avg. diet rank (SD) Avg. avail. Avg. avail. rank ] mo. Avg. pref. score Mubare Prunus africana Ficus capensis Myrianthus holstii Syzygium guineense Cassine aethiopica Habinyanja Ficus capensis Prunus africana Syzygium guineense Maesa lanceolata Cassine aethiopica Myrianthus holstii 1.8 5.1 29.5 19.6 34.6 3.0(1.0) 2.0(0.9) 2.6(1.1) 1.8(1.0) 2.7 (1.0) 2.4 20.5 20.8 1.0 35.4 33.3 2.3(0.9) 2.5(1.0) 3.4(1.9) 1.7 (0.6) 3.6(1.2) 3.2(1.3) 0.22 10.5 400.2 52.0 996.1 8.5 1.0 46.9 23.3 1156.8 392.7 1.0(0.3) 1.2(0.4) 2.9(0.9) 2.0(1.1) 3.4(0.9) 1.5 12 12 5 9.5 0.47 0.20 0.06 0.08 0.13 1.1(0.4) 2.0(0.7) 3.1(1.6) 2.0(0.6) 4.3(1.1) 3.8(0.8) 12 1.5 5 10.5 9.5 12 0.29 0.10 0.02 0.03 0.08 0.09 Food species are ranked from largest to smallest preference score. Availability was calculated using the FAI (fruit availability index), which is the product of the mean DBH (of phenology trees of fruits eaten by the gorillas), density of those species in the gorillas’ range, and their mean monthly abundance score value from the phenology for each species). The average preference score is the average of the biweekly periods that fruit was available. ] mo 5 number of months fruit species was available during the study period. TABLE IV. Nutritional and phenolic attributes of preferred foods/plant parts (individual species preference scores [per plant part] compared with their nutrient compositions) Plant part Leaf Leaf Leaf Leaf Leaf Leaf Leaf Leaf Leaf Leaf Leaf Leaf Pith Pith Pith Pith Pith Pith Pith Pith Pith Fruit Fruit Fruit Fruit Fruit Fruit Group Mubare Habinyanja Rushegura Kyagurilo Mubare Habinyanja Rushegura Kyagurilo Mubare Habinyanja Rushegura Kyagurilo Mubare Habinyanja Rushegura Mubare Habinyanja Rushegura Mubare Habinyanja Rushegura Mubare Habinyanja Mubare Habinyanja Mubare Habinyanja PCA Component Avg Rho Protein, fat, total phenols Protein, fat, total phenols Protein, fat, total phenols Protein, fat, total phenols WSC and sucrose WSC and sucrose WSC and sucrose WSC and sucrose NDF, hemicellulose, cellulose, NDF, hemicellulose, cellulose, NDF, hemicellulose, cellulose, NDF, hemicellulose, cellulose, Protein, fat, total phenols Protein, fat, total phenols Protein, fat, total phenols WSC and sucrose WSC and sucrose WSC and sucrose NDF, hemicellulose, cellulose, CT NDF, hemicellulose, cellulose, CT NDF, hemicellulose, cellulose, CT NDF and lignin/WSC NDF and lignin/WSC Fat and energy/sucrose Fat and energy/sucrose Condensed tannin Condensed tannin CT CT CT CT 0.30 0.34 0.45 0.36 0.03 0.17 0.18 0.09 0.35 0.68 0.58 0.72 0.07 0.02 0.07 0.14 0.20 0.36 0.14 0.10 0.05 0.28 0.17 0.05 0.15 0.28 0.18 SD 0.20 0.23 0.29 0.20 0.12 0.14 0.21 0.09 0.19 0.36 0.30 0.38 0.06 0.14 0.14 0.01 0.16 0.25 0.16 0.13 0.28 0.44 0.46 0.47 0.51 0.44 0.46 Range 0.04–0.60 0.07–0.68 0.06–0.82 0.14–0.57 0.19–0.32 0.02–0.35 0.22–0.77 0.07–0.21 0.47–0.16 0.82–0.54 0.7–0.43 0.89–0.50 0.03–0.17 0.24–0.27 0.04–0.33 0.04–0.30 0.06–0.43 0.24–0.59 0.1–0.5 0.2–0.5 0.89–0.5 0.32–0.74 0.63–0.46 0.8–0.74 0.5–1 0.32–0.74 0.87–0.62 Nutritional composition of plant parts was represented by principal component analysis scores. The results are displayed as the yearly average, SD and range of the biweekly/monthly correlation coefficients for each plant part (Avg Rho) A significant correlation for the year (tested via a one sample t-test with each biweekly or monthly period correlation coefficient as a data point) is indicated in bold. WSC, water soluble carbohydrates; NDF, neutral detergent fiber. Because the same methods were used in both studies, it is unknown why these differences occurred. One possibility could be that because mountain gorillas in Rwanda consume only a small number of highly abundant species, the limitations of Ivlev’s electivity index could have contributed to this result. Am. J. Primatol. 936 / Ganas et al. TABLE V. The average and standard deviations (represented in parenthesis after the average) of the nutritional and phenolic contents (% dry matter) of important plant parts used in the preference tests Buhoma Leaves n 5 11 PT ST FC GC SC NDF ADF LN CL HC FT EN TP TT CT 26.6 2.6 1.1 0.9 0.3 35.3 17.3 4.4 12.9 18.0 1.9 19.6 4.4 3.3 1.0 (3.8) (2.1) (0.9) (0.7) (0.2) (8.1) (5.6) (2.6) (3.3) (5.7) (0.7) (1.3) (3.3) (3.0) (1.9) Ruhija Pith n 5 7 Peel n 5 2 Fruit n 5 5 10.4 1.8 9.5 9.2 0.9 38.3 23.1 1.5 21.6 15.2 1.3 15.1 0.7 0.3 0.4 13.0 0.7 0.4 0.3 0.1 61.6 52.3 8.1 44.1 9.3 0.6 16.3 1.0 0.7 0.7 8.1(2.8) 6.3(12.3) 7.5(8.6) 7.1(7.9) 1.5(2.5) 32.7(13.0) 20.1(8.3) 7.5(4.8) 12.7(4.4) 12.6(8.9) 6.1(7.2) 18.8(2.1) 4.2(3.6) 3.6(3.7) 4.4(4.5) (2.2) (2.6) (5.2) (3.8) (0.3) (6.5) (3.4) (1.4) (2.3) (4.9) (0.7) (0.8) (0.4) (0.4) (0.5) (0.6) (0.5) (0.1) (0.1) (0.1) (0.8) (0.8) (7.0) (7.8) (2.6) (0.4) (0.5) (0.1) (0.01) (0.2) Leaves n 5 8 24.7 3.6 1.3 1.2 0.2 41.4 16.7 4.4 12.3 24.6 0.9 18.7 3.5 2.7 0.3 (2.9) (2.0) (1.3) (1.4) (0.4) (10.0) (4.7) (2.6) (3.1) (7.9) (0.6) (1.2) (4.1) (3.8) (0.7) Pith n 5 3 7.8 2.4 1.2 1.8 0.2 41.4 31.2 4.8 26.3 10.3 2.7 13.6 1.2 0.6 0.1 (2.3) (3.8) (1.0) (1.4) (0.1) (15.7) (12.9) (5.9) (7.0) (2.9) (3.3) (1.3) (0.4) (0.1) (0.001) Peel n 5 2 Fruit n 5 4 12.5 0.5 1.0 0.5 0.3 55.1 37.0 8.1 28.9 18.1 2.1 18.1 1.0 0.5 0.03 9.4(3.4) 3.0(5.6) 10.2(5.7) 6.3(2.7) 0.05(0.1) 25.1(9.2) 16.8(7.8) 7.0(3.4) 9.8(5.1) 8.3(2.5) 7.5(11.0) 20.2(2.1) 4.0(4.7) 2.7(2.9) 5.1(7.2) (5.6) (0.2) (0.5) (0.3) (0.5) (8.6) (2.1) (5.4) (7.5) (6.5) (2.0) (1.7) (0.5) (0.3) (0.01) PT, protein; ST, starch; FC, fructose; GC, glucose; SC, sucrose; NDF, neutral detergent fiber; ADF, acid detergent fiber; LN, lignin; CL, cellulose; HC, hemicellulose; FT, fat; EN, energy; TP, total phenols; TT, total tannins; CT, condensed tannins. One of the most interesting and novel results of this study was that the gorillas preferred leaves and pith of herbs that contained relatively high amounts of sugar. Similarly, sugar also influenced the choice of herbs [Ganas et al., 2008]. Although the average sugar content of leaves is relatively low compared with other plant parts (0.2–1.3% DM; Table V), as leaves constitute the greatest amount of wet mass intake to the diet, at least for the Kyagurilo group [Rothman et al., 2007]; sugar intake from leaves could be substantial. For herbivores that live in habitats where fruit availability is low or nonexistent, sugar in nonfruit plant parts may provide a required nutrient previously not associated with this type of vegetation [Danish et al., 2006]. Sugar in different foliage parts, as well as fruit, may enable the gorillas to be flexible and to exploit a variety of foods and habitat types when trying to fulfil nutritional requirements. Surprisingly, the Mubare group preferred fruit with relatively high condensed tannin amounts. However, this appears to be largely driven by the fruit of Myrianthus holstii, which contains much greater amounts of condensed tannin that other fruits [Ganas et al., in preparation]. Perhaps consuming fruit that contains a relatively high amount of condensed tannin is sometimes necessary to ingest substantial amounts of sugar. Other studies have also found that some animals tolerate relatively high tannin amounts in food if that food is consumed infrequently or the food’s consumption allowed a concurrent ingestion of a relatively high nutrient amount [Oates et al., 1980; Remis and Kerr, 2002]. To better understand the importance of sugar in fruit to Bwindi gorillas, we compared the sugar Am. J. Primatol. contents of the top five fruits consumed with five highly available, but avoided fruits (selectivity) and found that consumed fruits contained significantly more sugar than those not consumed [Ganas et al., in preparation]. This example highlights the importance of examining different measures of an animal’s foraging behavior to fully understand their foraging strategy. Owing to the nature of Ivlev’s electivity index (foods relatively high in availability can rarely score high preference scores or rare foods usually score high preference scores), the preference scores for some highly abundant foods may be biased compared with experimental studies of food preference. For example, widely abundant foods such as C. aethiopica (fruit) and M. solmsii (foliage) scored as avoided or neutral, despite their high frequency in the diet (Tables II and III). Therefore, it is important to individually examine preference scores of species that are of high availability (or rare) to determine whether their preference scores could be simply owing to this inherent limitation of this index. For example, C. aethiopica fruits were not preferred; yet their nutrient profile typifies what a high-quality fruit contains (relatively high amounts of sugar), suggesting that these fruits may be consumed for their sugar contents. Our preference results differed in some aspects from our concurrent research on food choice, which measured how the fluctuating availability of foods as well as the nutrient and phenolic composition influenced what foods were eaten. In that study, the yearround availability of herbs high in protein led to the result that protein did not influence the choice of herb foods [unlike this study, which showed that protein is Gorilla Food Preferences / 937 preferred, and thus is nutritionally important; Ganas et al., 2008, in press]. Together, these results show that protein is important to Bwindi gorillas, but that they do not need to specifically seek it out owing to its high availability. These studies demonstrate the importance of calculating both food preference (to determine nutritional requirements) and food choice (to determine which factors influence the consumption of particular foods in a variable environment) to better understand an animals’ foraging strategy. From a broader perspective, although preferred food species are found in various habitat types [i.e. open, mixed, regenerating, swamp forests; Nkurunungi et al., 2004], which likely plays a role in gorilla habitat use, open forest at both study locations in Bwindi contributed the greatest proportion of herb biomass [considering gorilla food; Ganas et al., in press], in both locations and contained many preferred foods [Basella alba, Ipomea spp. Mormodica calantha, etc.; Ganas et al., in press]. Therefore, in terms of mountain gorilla conservation, open forests can be considered a habitat that should be prioritized for conservation efforts. Overall, our results on food preference are not atypical of what we would expect for gorillas. However, the ways in which the gorillas fulfil these requirements, through a combination of different plant parts, shed new light on how gorillas can balance their diet in a variable environment. These results, together with our concurrent study of food choice, tell us the following information about the Bwindi gorilla foraging strategy: First, Bwindi gorillas need protein in their diets, but owing to the yearround high availability of herbs high in protein, it is easy to fulfil this requirement. Second, sugar is also important to the gorillas’ diet, and gorillas can eat fruit, pith, and/or leaves to obtain this nutrient. These studies underscore the importance of investigating the different facets of feeding behavior, nutrition, and food availability to get a comprehensive understanding of an animals’ foraging strategy. ACKNOWLEDGMENTS We thank the Uganda Wildlife Authority and the Uganda National Council of Science and Technology for permission to conduct this research. All research complied with animal care regulations and national laws. We appreciate the work of our field assistants who are too numerous to name. Further thanks to Bosco Nkurunungi, Alastair McNeilage, Robert Barigira, the Institute of Tropical Forest Conservation, Paul Kakende and the Biochemistry department at Makerere University, Heidrun Barleben, and the lab of Dr. Klaus Becker. This article benefited from statistical assistance from Roger Mundry as well as comments to previous versions by Oliver Schülke, Joanna Lambert, Shelly Masi, Colin Chapman, and four anonymous reviewers. This research was funded by the Max Planck Society, Berrgorilla & Regenwald Direkthilfe, The John Ball Zoo, and the Leakey Foundation. REFERENCES Altmann SA. 1998. Foraging for survival: Yearling baboons in Africa, Chicago: University of Chicago Press. Benz JJ, Leger DW, French JA. 1992. Relation between food preference and food-elicited vocalizations in golden lion tamarins (Leontopithicus rosalia). J Comp Psychol 106:142–149. Calvert JJ. 1985. Food selection by western gorillas (G. g. gorilla) in relation to food chemistry. Oecologia 65:236–246. Chapman CA, Chapman LJ, Naughton-Treves L, Lawes ML, McDowell LR. 2004. Predicting folivore primate abundance: validation of a nutritional model. Am J Primatol 62:55–69. Chesson J. 1983. The estimation and analysis of preference and its relationship to foraging models. Ecology 64:1297–1304. Danish L, Chapman CA, Hall MB, Rode KD, Worman CO. 2006. The role of sugar in diet selection in red tail and red colobus monkeys. In: Hohmann G, Robbins MM, Boesch C, editors. Feeding ecology in apes and other primates. Cambridge: Cambridge University Press. Field A. 2005. Discovering statistics using SPSS. London: Sage. Ganas J, Nkurunungi JB, Robbins MM, Kaplin BA, McNeilage A. 2004. Dietary variability of mountain gorillas in Bwindi Impenetrable National Park, Uganda. Int J Primatol 25:1043–1072. Ganas J, Nkurunungi JB, Robbins MM. A preliminary study of the temporal and spatial availability of herbaceous vegetation consumed by mountain gorillas in an afromontane rainforest. Biotropica, in press. Ganas J, Ortmann S, Robbins MM. 2008b. Food choice decisions of mountain gorillas in Bwindi Impenetrable National Park, Uganda: the influence nutrients, phenols, and availability, in review. Ganzhorn JU. 1992. Leaf chemistry and the biomass of folivorous primates in tropical forests. Oecologia 91:540–547. Grieg-Smith P. 1983. Quantitative plant ecology. Berkeley: University of California Press. Hayward MW, Hofmeyr M, O’Brien J, Kerley GIH. 2006. Prey preferences of the cheetah (Acinonyx jubatus) (Felidae:Carnivora) morphological limitations or the need to capture rapidly consumable prey before kleptoparasites arrive? J Zool 270:615–627. Huffman M. 1997. Current evidence for self-medication in primates: a multidisciplinary perspective. Yearbook Phys Anthropol 40:171–200. Ivlev VS. 1961. Experimental ecology of the feeding of fishes. New Haven: Yale University Press. Johnson DH. 1980. The comparison of usage and availability measurements for evaluating resource preference. Ecology 61:65–71. Laska M, Salazar LTH, Luna ER. 2000. Food preferences and nutrient composition in captive spider monkeys, Ateles geoffroyi. Int J Primatol 21:671–683. Lechowicz MJ. 1982. The sampling characteristics of electivity indices. Oecologia 52:22–30. Makkar HPS, Bluemmel M, Borowy NK, Becker K. 1993. Gravimetric determination of tannins and their correlations with chemical and protein precipitation methods. J Sci Food Agr 61:161–164. McNeilage A. 1995. Mountain gorillas in the Virunga Volcanoes: ecology and carrying capacity. Bristol University. Unpublished dissertation. Am. J. Primatol. 938 / Ganas et al. McNeilage A, Plumptre AJ, Brock-Doyle A, Vedder A. 2001. Bwindi Impenetrable National Park, Uganda: gorilla census 1997. Oryx 35:39–47. Milton K. 1979. Factors influencing leaf choice by howler monkeys: a test of some hypotheses of food selection by generalist herbivores. Am Nat 114:362–378. Milton K. 1984. The role of food processing factors in primate food choice. In: Rodman PS, Cant JGH, editors. Adaptations for foraging in non-human primates: contributions to an organismal biology of prosimians, monkeys, and apes. New York: Columbia University Press. p 249–279. Nkurunungi JB, Ganas J, Robbins MM, Stanford CB. 2004. A comparison of two mountain gorilla habitats in Bwindi Impenetrable National Park, Uganda. Afr J Ecol 42: 289–297. Norscia I, Carrai V, Borgognini-Tarli SM. 2006. Influence of dry season and food quality and quantity on behavior and feeding strategy of Propithecus verreauxi in Kirindy, Madagascar. Int J Primatol 27:1001–1022. Oates JF, Waterman PG, Choo G. 1980. Food selection by the south Indian leaf monkey, Presbytis johnii, in relation to leaf chemistry. Oecologia 45:45–56. Oates JF, Whitesides GH, Davies AG, Waterman PG, Green SM, DaSilva GL, Mole S. 1990. Determinants of variation in tropical forest primate biomass: new evidence from West Africa. Ecology 71:328–343. Orians GH, Wittenberger JF. 1991. Spatial and temporal scales in habital selection. American Naturalist 137:29–49. Porter LJ, Hrstich LN, Chan BG. 1986. The conversion of procyanadins and prodelphinidins to cyanidin and delphinidin. Phytochemistry 25:223–230. Remis MJ. 1997. Western lowland gorillas (Gorilla gorilla gorilla) as seasonal frugivores: use of variable resources. Am J Primatol 43:87–109. Remis MJ. 2000. Initial studies on the contributions of body size and gastrointestinal passage rates to dietary flexibility among gorillas. Am J Phys Anthropol 112:171–180. Remis MJ. 2002. Food preferences of captive western gorillas (Gorilla gorilla gorilla) and chimpanzees (Pan troglodytes). Int J Primatol 23:231–249. Am. J. Primatol. Remis MJ, Kerr ME. 2002. Taste responses to fructose and tannic acid among gorillas (Gorilla gorilla gorilla). Int J Primatol 23:251–261. Rogers ME, Abernathy K, Bermejo M, Cipolletta C, Doran DM, McFarland K, Nishihara T, Remis M, Tutin CEG. 2004. Western gorilla diet: a synthesis from six sites. Am J Primatol 64:173–192. Rothman JM, van Soest PJ, Pell AN. 2006. Decaying wood is a sodium source for mountain gorillas. Biol Lett 2:321–324. Rothman JM, Plumptre AJ, Dierenfeld ES, Pell AN. 2007. Foraging ecology and nutritional composition of the diets of gorillas (Gorilla beringei): a comparison between two montane habitats. J Trop Ecol 23:673–682. Schaller GB. 1963. The mountain gorilla: ecology and behavior. Chicago: University of Chicago Press. Schoener TW. 1971. Theory of feeding strategies. Ann Rev Ecol Syst 2:370–404. Schoener TW. 1983. Simple models of optimal feeding territory size: A reconciliation. American Naturalist 121: 608–629. Stephens DW, Krebs JR. 1986. Foraging theory. Princeton: Princeton University Press. Sun C, Kaplin BA, Kristensen KA, Munyaligoga V, Mvukiyumwami J, Kajonda K, Moermond TC. 1996. Tree phenology in a tropical montane forest in Rwanda. Biotropica 28:668–681. van Soest PJ. 1991. Nutritional ecology of the ruminant. Ithaca: Cornell University Press. Vedder AL. 1989. Feeding ecology and conservation of the mountain gorilla. Madison: University of Wisconsin. 262p. Watts DP. 1983. Foraging strategy and socioecology of mountain gorillas (Pan gorilla beringei). PhD dissertation. University of Chicago. Watts DP. 1984. Composition and variability of mountain gorilla diets in the central Virungas. Am J Primatol 7:323–356. Westoby M. 1974. An analysis of diet selection by large generalist herbivores. Am Nat 108:290–304. Zar JH. 1999. Biostatistical analysis. Upper Saddle River, NJ: Prentice-Hall.