(Gorilla gorilla gorilla) and Chimpanzees (Pan ... - Springer Link

4 downloads 105 Views 192KB Size Report
Food Preferences Among Captive Western Gorillas. (Gorilla gorilla ..... GO. – gorilla outdoor trials;. GI. – gorilla indoor;. CI. – chimpanzee indoor trials. c. F oods.
P1: Vendor International Journal of Primatology [ijop]

PP324-362463

January 8, 2002

17:0

Style file version Nov. 19th, 1999

C 2002) International Journal of Primatology, Vol. 23, No. 2, April 2002 (°

Food Preferences Among Captive Western Gorillas (Gorilla gorilla gorilla) and Chimpanzees (Pan troglodytes) M. J. Remis1 Received October 18, 2000; revised May 16, 2001; accepted June 1, 2001

I used a zoological park setting to address food preferences among gorillas (Gorilla gorilla gorill) and chimpanzees (Pan troglodytes).Gorillas and chimpanzees are different sizes, and consequently, have been traditionally viewed as ecologically distinct. Sympatric western gorillas and chimpanzees have proved difficult to study in the wild. Limited field data have provided conflicting information about whether gorillas are fundamentally different from chimpanzees in diet and behavior. Fruit eating shapes the behavior of most apes, but it is unclear whether the large-bodied gorillas are an exception to this rule, specifically whether they are less selective and more opportunistic fruit eaters than chimpanzees are. My research provides experimental observational data to complement field data and to better characterize the diets and food preferences of the African apes. During laboratory research at the San Francisco Zoological Gardens, I examined individual and specific differences in food preferences of captive gorillas and chimpanzees via experimental paired-choice food trials with foods that varied in nutritional content. During the study, I offered 2500 paired-food choices to 6 individual gorillas and 2000 additional pairs to them as a group. I also proffered 600 food pairs to 4 individual chimpanzees. Despite expectations of the implications of body size differences for diet, gorillas and chimpanzees exhibited similar food preferences. Both species preferred foods high in non-starch sugars and sugar-to-fiber ratios, and low in total dietary fiber. Neither species avoided foods containing tannins. These data support other suggestions of African apes sharing a frugivorous adaptation. KEY WORDS: frugivory; hominoids; tannins; fiber.

1 To whom correspondence should be addressed at Department of Sociology and Anthropology,

Purdue University, West Lafayette, IN 47907-1365; [email protected]. 231 C 2002 Plenum Publishing Corporation 0164-0291/02/0400-0231/0 °

P1: Vendor International Journal of Primatology [ijop]

PP324-362463

January 8, 2002

232

17:0

Style file version Nov. 19th, 1999

Remis

INTRODUCTION Western gorillas (Gorilla gorilla gorilla) are seasonally frugivorous (Remis, 1997). Dietary overlap with sympatric chimpanzees (Pan troglodytes troglodytes) appears to be high during periods of fleshy fruit abundance throughout central African forests (Tutin et al., 1997). Niche separation becomes apparent during seasons of scarcity, when the larger gorillas exploit fallback foods, including fibrous herbs, leaves and bark (Tutin et al., 1991), while chimpanzees, which have been characterized as more persistent frugivores (Kuroda et al., 1996), maintain a highly frugivorous diet. It is likely that dietary breadth and flexibility among gorillas are related to the effects of large body size (Shea, 1983) and digestive specializations (Chivers and Hladik, 1980; Remis, 2000, Remis et al., 2001). Ecological niche should be related to species-specific food preferences (Ganzhorn, 1989; Laska et al., 2000). Gorilla dietary flexibility in the wild may be shaped by taste perception (Kalmus, 1970; Simmen, 1994) and preferences for a wide array of foods of varying nutritional content (Hladik and Simmen, 1996; Rogers et al., 1990). Conversely, chimpanzees, may be more discriminating in their food preferences, only selecting foods with a highenergy and sugar content, while minimizing fiber and secondary compounds in their diet (Wrangham et al., 1998). Sympatric western gorillas and chimpanzees are difficult to study in the wild (Tutin et al., 1991). Further, although wild weights are poorly documented, body sizes appear to be less divergent among the western gorilla and sympatric chimpanzee subspecies than their eastern counterparts. Wild female Gorilla gorilla gorilla and male Pan troglodytes troglodytes may be of similar size with male gorillas disproportionately larger (Jungers and Susman, 1984). It is therefore difficult to separate species-specific or bodysize related effects on the diets of wild apes from ecological, biomechanical and social influences on food choices (Remis, 1999). No doubt, patterns of food acceptance and aversion are shaped by a combination of taste abilities and preferences, nutritional requirements, past experiences and exposure to conspecifics (Barker et al., 1977; Rozin and Kennel, 1983; Steiner and Glaser, 1984; Watts, 1985). Nevertheless, individuals of a particular species frequently share patterns of food preferences (Simmen and Hladik, 1998). Food preferences have been established based on measures of pleasantness, as well as choices made, order of selection and amounts consumed during experimental trials (Benz et al., 1992; Nunnally, 1978; Simmen, 1994). I used paired-choice food trials to examine the food preferences of gorillas and chimpanzees in a zoo setting, where conditions of food availability and social factors can be controlled. I examined the specific hypothesis

P1: Vendor International Journal of Primatology [ijop]

PP324-362463

January 8, 2002

Food Preferences in Gorillas and Chimpanzees

17:0

Style file version Nov. 19th, 1999

233

that gorillas and chimpanzees may both be selective fruit feeders and exhibit similar food preferences for sugary, low-fiber foods in captivity as in the wild. Alternatively, the larger body size of gorillas may provide a physiological basis for specific differences in food preferences, that relates to differing diets in the wild. Gorillas in captivity may exhibit weak preferences for high-sugar fruits, and like those in the wild, may choose a broad range of low-sugar, fibrous foods or those containing tannins (Simmen and Hladik, 1998). Results have implications for efforts to understand how phylogeny or body size might shape species-specific food preferences and the acceptability of foods of varying taste and nutrient composition to the closely related African apes.

METHODS Staff at the San Francisco Zoological Gardens (SFZ) and I conducted food preference trials on the SFZ housed five adult-sized and one juvenile gorilla, and four adult chimpanzees, July–September, 1999. The subjects had body weights within ranges reported for wild apes, and they are variously wild- or captive-born (Table I). The gorillas born at SFZ were motherreared, but the other gorillas and the chimpanzees were human-reared, and may have had varying exposure to native or domesticated fruits as infants. All subjects were group-housed at the San Francisco Zoo for most of their lives. They receive a similar, diverse diet of seasonal domestic fruits and vegetables and monkey chow and year-round access to browse (Acacia longifolia spp.). The gorilla diet includes approximately 3500–4000 g of mixed vegetables and fruits with 400–800 g Mazuri, Inc. Leaf Eater Primate chow The chimpanzee diet includes approximately 2900–3900 g of mixed fruits and vegetables with 300–400 g Animal Spectrum, Inc., High Fiber primate biscuit. Both diets provide similar amounts of fiber (approximately 18% neutral detergent fiber on a dry matter basis) (Dierenfeld pers. comm.). At the SFZ, the gorillas and chimpanzees eat their main morning and evening meals when they voluntarily separate into different areas of the indoor night-quarters, to prevent conflicts over food. We conducted pairedchoice food trials (Benz et al., 1992; Nunnally, 1978) on individual gorillas and chimpanzees just before mealtimes using foods from the typical diets and several novel foods that varied in nutrient content (Table II). Both gorillas and chimpanzees have daily access to outdoor grottos where foods are scattered mid-day for enrichment. We also offered paired-choice trials to the gorilla group in their grotto and analyzed the results separately from the individual trials.

P1: Vendor International Journal of Primatology [ijop]

PP324-362463

January 8, 2002

17:0

Style file version Nov. 19th, 1999

234

Remis Table I. Study subjects at the San Francisco Zoological Gardens

Species Gorilla gorilla gorilla

Individual Kubwa

Silverback male

23

Shango

Blackback male Lactating female

11

Zura

Nulliparous female

18

Pogo

Nulliparous female

41

Barney Cobby

Juvenile male Male

40

Maggie

Female

30

Minnie

Nulliparous female

30

Tallulah

Nulliparous female

40

Bawang

Pan troglodytes verus

Pan troglodytes troglodytes a Study

Status/sex

Weight (k) Study date agea documented

19

5

History

169.6 (3/96) Captive-born SFZ, parentreared, mother died when he was 3 years old, father of Shango, Barney and Nneka est. 150 Captive-born SFZ, mother reared 87 (4/96) Captive-born Cincinati Z., nursery-reared, arrived at SFZ and group housed at 17 months, mother to Shango, Barney and infant Nneka (10 month old infant in group) 77 (11/93) Captive-born Colombus Z., nursery-reared, arrived at SFZ and group housed at 14 months 109.8 (3/96) Wild-born Cameroon, handreared by missionaries in Cameroon, arrived at SFZ and group housed at 3 years old est. 50 Captive-born SFZ, motherreared 60 (6/95) Wild-born, hand-reared circus, arrived at SFZ and group housed at 10 years old, sired four offspring with Maggie 40.8 (9/95) Wild-born, hand and nursery reared, arrived at SFZ with Minnie at 16 months old, four offspring, none surviving 60.8 (6/95) Wild-born, hand and nursery reared, arrived at SFZ with Maggie at 16 months old 58.9 (6/95) Wild-born, hand-reared pet, arrived at SFZ, and group housed at 8 years old

age: individual’s age when the study took place.

Foods Selected for Use in the Study In order to prevent over-feeding, we adjusted the fruit portion of the zoo diets to include foods to be used in the trials. During a 7-day pretrial period, we gave individuals an opportunity to develop taste preferences

N N Y N N N N Y N Y N N N Y N N Y Y N N N N

Banana GI,CI Apple GI,CI Mango, flesh GI,GO Orange GI Melon, f flesh GO Cucumber GO Onion GO Fig, black GI,CI Pepper GO Kumquat GI Tomato GO Beet GO Carrot GO Parsnip GI,CI Cabbage GO Kale GO Tamarind GI,CI Lemon GI,CI Broccoli GO Celery GO Corn GO Potato baked GO

(Nutrient units g/100 grams dry wt.) 0 75.1 83.94 12.45 93.17 0 84.5 67.74 11.61 76.13 20.7e 82.4 78.41 16.48 80.11 0 86.10 61.15 12.95 61.15 0 92.1 53.16 11.39 69.62 Na 96.0 46.15 10.26 71.80 Na 89.7 50.39 12.6 74.61 0 84.6 61.69 22.89 61.69 Na 92.2 63.12 23.05 63.12 0 83.4 56.02 22.89 56.02 Na 93.8 44.87 24.04 57.05 Na 87.5 48 32 53.04 Na 87.7 44.23 23.75 66.26 0 82.6 50.57 22.99 67.24 Na 92.2 43.83 34.39 43.83 Na 84.5 20.00 12.87 61.39 e 29.2–88.8 35.8 7.94 8.10 88.01 0 86.3 23.36 34.31 23.36 Na 90.7 25.56 38.67 25.56 Na 94.6 22.39 33.58 49.81 Na 76.92 7.36 13.86 70.45 Na 71.2 2.08 8.68 81.95 381.53 303.23 323.86 266.19 303.80 371.74 357.36 279.22 319.95 259.04 330.29 278.80 312.78 308.01 279.64 371.02 370.72 138.69 264.12 278.73 366.20 367.01

4.82 2.58 3.98 7.9 7.59 15.38 11.24 8.44 11.4 5.42 13.62 13.6 8.44 9.77 18.34 21.24 3.58 7.3 32.01 13.99 11.35 7.99

Vit C

Fe

K

Mg

(Mineral units g/100 mg dry wt.) 1.2 13.9 0.4 535.7 43.7 0.65 32.3 0.8 709.7 25.8 1.14 210.2 3.98 1022.7 73.9 0.72 388.5 0.72 1079.1 71.9 1.56 329.1 1.3 2658.2 86.6 2.56 132.86 6.52 3609.0 275.69 1.55 62.0 2.13 1521.3 96.9 1.95 13.0 1.9 1298.7 97.4 2.4 1143.4 5.9 2266.3 128.0 3.01 234.9 3.6 1084.3 78.3 5.29 306.1 7.2 3557.7 176.3 1.36 39.2 6.4 2600.0 184.0 1.56 76.17 4.10 2645.4 122.8 Tr 86.21 82.9 1954.0 126.4 3.44 410.2 7.5 3133.8 191.1 4.5 772.2 10.9 2876.5 218.8 0.47 4.7 2.8 934.6 143.3 2.19 423.4 3.7 1094.9 87.6 3.76 1001.1 9.5 3490.9 268.5 2.61 130.6 7.46 5354.5 205.2 4.33 51.94 4.76 1399.5 199.13 0.35 44.8 4.7 – 93.8

Fat

PP324-362463

235

January 8, 2002

given in trials administered to gorillas (g) and chimpanzees (c), results are expressed as percent dry weight per 100 g of those food parts consumed by apes, except where noted (Data are taken from Holland et al., 1991, 1992). % H2 O: percent water. Total fiber: today dietary fiber; TNC: total nonstructural carbohydrates and includes simple sugars + starch expressed as monosaccharides. b Foods used in trials: GO – gorilla outdoor trials; GI – gorilla indoor; CI – chimpanzee indoor trials. c Foods not in typical diet scored as novel. d Proportion of non-starch sugars to total dietary fiber. e RD Tannins are measured by the RD analysis and expressed as mg/gram dry weight. RD tannins: mango without peel = 0; with peel = 20.7; tamarinds RD = 29.2, 86.3, 88.8, (n = 3). f Melon: cantaloupe.

6.7 5.8 4.8 4.7 4.7 4.5 4 2.9 2.9 2.4 1.9 1.9 1.9 1.8 1.4 1 0.9 0.7 0.7 0.7 0.5 0.2

Sugar fiber RD tannin % Non-starch Total Energy ratiod mg/g H2 O sugars fiber TNC (Kcal) Protein

International Journal of Primatology [ijop]

a Foods

Novelc

Food used in trialb

Table II. Nutrient content of foods used in ape paired choice trials, ranked by proportion of sugar to fiber a

P1: Vendor 17:0 Style file version Nov. 19th, 1999

P1: Vendor International Journal of Primatology [ijop]

PP324-362463

January 8, 2002

17:0

236

Style file version Nov. 19th, 1999

Remis

to novel foods. Among foods seasonally available in San Francisco produce markets, we selected foods for the trials based on variation in sugars, fiber and secondary compound content, as well as their acceptability to the animals (Table II). For example, gorillas refused to eat passion fruit or tamarillos. Figs were selected because of the importance of wild varieties in ape diet and because they have been reported to be high in dietary fiber and polyphenols (Vinson, 1999). We included mangos (with peels) and tamarinds based on their tannin content, rare in most domestic fruits, but common in foods consumed by primates in the wild. In addition, tamarinds and other hard-to-process legumes are important foods for many primate species. We used nutritional information on sample foods from the literature, and report them on a percent dry weight basis (Holland et al., 1991, 1992). In addition, Dr. C. Mowry assayed the foods used in the indoor trials for total tannins at his laboratory at Berry College, Georgia via radial diffusion analysis. The radial diffusion (RD) method produces a measure of the protein binding effect of tannins in plants, including both condensed and hydrolyzable tannins (Hagerman, 1987). The assay is correlated with other measures of condensed and hydrolyzable tannins (Remis et al., 2001; Wrangham et al., 1998). We selected 14 foods for use in the outdoor gorilla trials (n = 2000 trials). We presented 9 of them to 6 gorillas in indoor trials (n = 2500 trials). We tested the 4 chimpanzees via a narrower range of 6 foods (n = 600 trials), primarily because of logistical constraints with the chimpanzee housing and zoo schedules. We selected individual foods to be given to the chimpanzees from the broader range used in the gorilla trials, in order to preserve the variation in nutrient content. For example, we did not use mangos during the chimpanzee trials, but the mango flesh was nutritionally similar to bananas (Table II).

Outdoor Food Preference Experiments During the food choice trials, we divided fruits and other foods into similarly-sized small pieces, and the number of paired choices offered per day was limited. Each day before gorillas were allowed outdoor access, we strategically scattered between 80 and 120 pairs of vegetables (n = 12) and fruits (n = 2) in front of observation stations around the grotto. We placed food piles ≥2 m apart throughout the 22,000-ft2 gorilla grotto to minimize feeding-related agonism. We offered each possible combination ≥7 times over the course of the study.

P1: Vendor International Journal of Primatology [ijop]

PP324-362463

January 8, 2002

Food Preferences in Gorillas and Chimpanzees

17:0

Style file version Nov. 19th, 1999

237

Each day, between 4 and 8 observers collected all-occurrence data on individual approaches to food piles, from 1000 h when gorillas received outdoor access until 1100 h, by which time >98% of piles had been visited. We considered each approach by a gorilla to ≤0.5 m of a pile as a trial. We recorded gorillas’ order of selection, order and the amount of food consumed and social interactions at outdoor food piles using ad libitum sampling methods. When ≥1 food was selected from a pile, we scored the trial as completed. We disregarded subsequent approaches to single remaining items. We based food preference assessments on order of selection.

Indoor Food Preference Trials In order to provide greater controls on food choices and to eliminate the effects of social facilitation, we offered a series of indoor paired-choice trials to individual gorillas and chimpanzees in their night housing. During the course of the indoor trials, we offered approximately 15 paired-food trials sequentially to individuals just before they received the remainder of each of their morning and evening meals. We dropped paired foods simultaneously and in random combination and position (left or right) into the metal food trays affixed to the side of the cage. We considered the first food picked up by the subject as selected and recorded the order and amount consumed. The subjects generally consumed both foods ≤15 sec. If no selection was made within 2 min, the trial was discarded. If a single food item was not consumed within the trial, we scored it as having been rejected. Before each subsequent trial, the keeper (or the subject) removed any previously rejected foods from the tray and surrounds. We presented each possible pair of foods to each subject ≥10 times to ensure that food preferences measured were not simply artifacts of the test. If subjects preferred two of them equally, we expected them to select both of them in equal proportions in the paired choice trials.

Analysis of Food Preferences The paired choice method allows the construction of an interval-level measurement scale on which the resulting food preference scores can be tested for significant differences. The dependent variable is the proportion of time that each food type was chosen for each pair presented. I determined individual and species-specific food preference scores from proportions of

P1: Vendor International Journal of Primatology [ijop]

PP324-362463

January 8, 2002

17:0

Style file version Nov. 19th, 1999

238

Remis

Table III. Sample individual choice matrix (Kubwa, the silverback gorilla) to show the percentages of first choice for each food when paired with every other food Chosen food

Paired food

Fig Mango Banana Tamarind Apple Orange Kumquat Parsnip Lemon

Fig Mango Banana Tamarind Apple Orange Kumquat Parsnip Lemon

50 72 79 100 70 92 92 100 92

27 50 35 100 42 70 100 100 90

21 65 50 100 53 100 100 91 100

0 0 0 50 0 9 18 40 40

30 58 47 100 50 40 100 100 92

0 30 0 91 60 50 90 92 90

8 0 0 82 0 10 50 80 60

0 0 9 60 0 8 20 50 70

8 10 0 60 8 10 40 30 50

Note. The numbers in the cells in each column (read down) represent the percentages of times foods were chosen first (wins) for each item in a pair. The numbers read in rows represent percentages of losses for each item in a pair. See text for explanation of method of determining food preference scores from contingency table matrices.

first selections or wins during trials via standard methods (Benz et al., 1992; Nunnally, 1978). I constructed individual and group contingency tables (similar to sociometric matrices in Altmann, 1974) wherein foods selected are arrayed across the top, and those not selected are arrayed down the side of a 9 × 9 matrix. Table III examplifies an individual contingency table (matrix). Values entered in each cell represent the proportion of times each food in a pair won (columns) or lost (rows). Following standard procedures, I entered values of 0.5 on the diagonal of the matrix in order to drop these non-trials from the analysis when they were converted into zscores. I then converted the percentages of wins and losses into z-scores, i.e. 50% had a z-score of 0, 0–49% had increasingly negative z-scores, 51–100% had increasingly positive z-scores. I summed and averaged the z-scores for each food type. This procedure yielded individual and overall specific positive and negative food preference scores. In order to produce food preference hierarchies that had arbitrary zero points and meaningful interval scale distances between the foods, in each case, I added the absolute value of the lowest preference score for the individual or species to the subjects’ scores. I evaluated differences in individual and mean specific preferences via SPSS Pearson’s correlations, MANOVA and ANOVA procedures (Benz et al., 1992). I used principle components analysis (PCA) to arrange foods along axes based on their nutrient composition (ter Braak, 1995). Subsequently I used the PCA factor scores as the fruit values in Pearson’s correlations to test the relationship between ape preference and nutrient composition.

P1: Vendor International Journal of Primatology [ijop]

PP324-362463

January 8, 2002

Food Preferences in Gorillas and Chimpanzees

17:0

Style file version Nov. 19th, 1999

239

RESULTS Outdoor Food Trials The gorillas generally scrambled to choose all available fruit from choice piles within 5 min of release into the outdoor exhibit. Individuals would then generally retreat to a safe location to consume their loot before returning to the remaining intact piles to choose vegetable foods. Each pile was isolated, consisted of a choice of two and was scored as an individual trial. Therefore, although gorillas were choosing from multiple individual piles throughout the enclosure, preferences were evaluated only between pairs of food at each approach. The results are consistent with those obtained during the individual paired-choice trials. Because piles were widely distributed, very little food-related agonism occurred. Ninety-two percent of outdoor trials were completed (n = 2000). All individuals participated in the trials. Adult males and females are fairly equally represented in the sample. Nevertheless, each of the adult-sized males made more selections than smaller females or the juvenile did, representation may reflect differing energy requirements or dominance effects or both (silverback: 22%, blackback: 24%, lactating female: 17%, nulliparous female: 15%, nulliparous old female: 12%, juvenile: 10%; χ2 test p < 0.001). The overall preference scores of individual foods in the outdoor trials varied (ANOVA F13,140 = 10.196, p < 0.001). Preferred foods had preference ratings >2, less preferred foods had ratings 2, whereas foods that were not preferred had mean ratings 2.5, and least preferred foods have mean ratings