American Journal of Primatology 58:91–116 (2002) RESEARCH ARTICLE Western Lowland Gorilla Diet and Resource Availability: New Evidence, Cross-Site Comparisons, and Reflections on Indirect Sampling Methods DIANE M. DORAN1,2, ALASTAIR McNEILAGE2,3, DAVID GREER1,2, CAROLYN BOCIAN1,2, PATRICK MEHLMAN1,2, and NATASHA SHAH4 1 Department of Anthropology, SUNY at Stony Brook, Stony Brook, New York 2 Mondika Research Center, Central African Republic and Republic of Congo 3 Wildlife Conservation Society, Bronx, New York 4 Interdepartmental Doctoral Program in Anthropological Sciences, SUNY at Stony Brook, Stony Brook, New York We describe the resource availability and diet of western lowland gorillas (Gorilla gorilla gorilla) from a new study site in the Central African Republic and Republic of Congo based on 3 years of study. The results, based on 715 fecal samples and 617 days of feeding trails, were similar to those reported from three other sites, in spite of differences in herb and fruit availability. Staple foods (consumed year-round) included highquality herbs (Haumania), swamp herbs (when present), and a minimal diversity of fruit. A variety of fruits (average of 3.5 species per day and 10 per month) were selectively consumed; gorillas ignored some common fruits and incorporated rare fruits to a degree higher than predicted based on availability. During periods of fruit abundance, fruit constituted most of the diet. When succulent fruits were unavailable, gorillas used low-quality herbs (i.e., low-protein), bark, and more fibrous fruits as fallback foods. Fibrous fruit species, such as Duboscia macrocarpa and Klainedoxa gabonensis, were particularly important to gorillas at Mondika and other sites as fallbacks. The densities of these two species are similar across sites for which data are available, in spite of major differences in forest structure, suggesting they may be key species in determining gorilla density. No sex difference in diet was detected. Such little variation in western lowland gorilla diet across sites and between sexes was unexpected and may partly reflect limitations of indirect sampling. Am. J. Primatol. 58:91–116, 2002. r 2002 Wiley-Liss, Inc. Contract grant sponsor: National Science Foundation; Contract grant numbers: SBR-9422438; SBR-9729126; Contract grant sponsor: State University of New York at Stony Brook. n Correspondence to: Diane Doran, Department of Anthropology, SUNY at Stony Brook, Stony Brook, NY 11794. E-mail: ddoran@notes.cc.sunysb.edu Received 23 January 2002; revision accepted 23 September 2002 DOI 10.1002/ajp.10053 Published online in Wiley InterScience (www.interscience.wiley.com). r 2002 Wiley-Liss, Inc. 92 / Doran et al. Key words: Gorilla; foraging strategy; food availability; fecal samples; feeding traces INTRODUCTION At all sites studied to date, western lowland gorillas (Gorilla gorilla gorilla) have greater dietary breadth and eat more fruit than do mountain gorillas (G. g. beringei) [reviewed in Watts, 1996; Doran & McNeilage, 1998, 2001]. Western lowland gorillas eat herbs and fruit throughout the year, with the relative proportion and diversity of food categories shifting with fluctuating fruit availability [Williamson, 1989; Tutin et al., 1991; Nishihara, 1995; Remis, 1997; Goldsmith, 1999]. Gorillas eat herbaceous food throughout the year, but increase their intake of low-quality herbs when succulent fruit is scarce. At most sites, western lowland gorillas eat fruit daily (Z95% of fecal samples contained fruit, an average of Z3 species per day) and include many species (Z70 overall and 10–15 per month) in their diet (reviewed in Doran and McNeilage [2001], but see Calvert [1985]). The incorporation of substantial quantities and varieties of fruit in the diet should have consequences for western lowland gorilla ranging and social behavior, relative to that of mountain gorillas [Doran & McNeilage, 2001]. However, our understanding of western lowland gorilla behavior remains elusive due to the absence of habituated groups. Our current understanding of western lowland gorilla diet is based on indirect sampling, fecal samples, and trail sign data, primarily from four studies at three sites. Western lowland gorilla diets appear similar across sites, but interpretations of foraging strategy differ. Williamson [1989] noted that at Lopé, Gabon, gorillas were selective in their fruit choice, and incorporated fruit in their diet when vegetative plant parts were simultaneously available. The amount of stem-fiber, leaf, and bark in the diet were influenced by the availability of fruit in the environment, rather than by their own availability. She described gorilla dietary strategy as one of selection of succulent fruits. Fruits were complemented nutritionally by the inclusion of other food types when their availability was reduced. Subsequently, Kuroda and colleagues [1996] described gorillas at Ndoki, Republic of Congo, as less selective and persistent in their fruit-eating (compared to chimpanzees), and noted that high-quality herbs (such as Haumania shoots and Hydrocharis roots, which are available and consumed year-round) were the staple components of the diet. They suggested that in areas with access to superabundant, high-quality swamp vegetation (or, presumably, extensive tracts of Haumania), gorillas have few incentives to seek out fruit when its abundance is low, and select fruit only as an alternative food choice [Nishihara, 1995; Kuroda et al., 1996]. Remis’ [1997] and Goldsmith’s [1999] findings support Williamson’s [1989] interpretation. The gorillas at Bai Hokou, Central African Republic, were persistent fruit-eaters; they traveled farther to obtain it when it was available, but were able to subsist entirely on fibrous foods when necessary because of their large body size. Swamp herbs, a key feature of the Ndoki site, were absent from the Bai Hokou and Lopé study sites. Differing opinions concerning western gorilla foraging strategy persist, although no study has explicitly examined how changes in distribution and availability of resources across sites influence diet. In this study, we document the resource availability and diet of western lowland gorillas from a fourth site, the Lowland Gorilla Diet at Mondika / 93 Mondika Research Center, Central African Republic and Republic of Congo, where, as at Ndoki, swamp herbs are available and used regularly. We compare these results to previous studies to examine whether: 1) overall increased herb availability or reliance on superabundant herbs in swamps leads to a decreased reliance on fruit, as proposed by Kuroda et al. [1996]; and 2) higher fruit species diversity is reflected in greater dietary breadth, as predicted by a strategy of dietary opportunism. Since gorillas are extremely sexually dimorphic in body size (male: 170.4 kg, female: 71.5 kg [Smith & Jungers, 1997]), and large body size is a strategy to cope with a more generalized, lower-quality diet [reviewed in Lambert, 1999], we test whether larger male body size is reflected in increased folivory, relative to females. Finally, since we used indirect data to examine diet, as in previous studies, we consider how sampling method may influence results. We report findings from two indirect measures of diet–fecal samples and trail signs– and comment on the strengths and limitations of each. METHODS Study Site and Sampling Methods The Mondika study site (50 km2) straddles the boundaries of the Central African Republic (Dzanga-Ndoki National Park) and Republic of Congo (021 210 85900 N, 0161 160 46500 E). The habitat consists of mixed-species tropical lowland forest, monodominant Gilbertiodendron dewevrei (Caesalpiniaceae) forest, and swamp forest. Mean annual rainfall averaged 1,301 mm during the study (July 1995 to June 1998; SD = 207) and 1,415 mm over the first 5 years (July 1995 to June 2000; SD = 219.1). Rainfall is seasonal, and the study period was characterized by the typical 2–3-month annual dry season, with o50 mm of rain per month (Fig. 1a). Mean minimum and maximum daily temperatures were 21.0 7 0.8 and 28.2 7 1.7 (n = 5 years). The area is free of human disturbance, has never been logged, and is rich (10 diurnal species) in primate fauna. Measures of Resource Availability Phenology. Phenological data on fruit, leaf, and flower production were collected beginning October 1996. We sampled important gorilla food trees (see below for definition), rather than overall habitat productivity, to ensure adequate sampling of rare species. We recorded the presence or absence of ripe fruit from 10 individuals (when possible) of 20 tree and three liana species (totaling 197–216 tree and 20–32 liana individuals). We report fruit availability as the number of individuals with ripe fruit per month. Fruit tree identification, density, and diameter at breast height (DBH). Densities of tree species (stems/ha) were determined by counting and identifying stems of all individuals (of DBH Z10 cm) along 28 (200 m  10 m) transects, placed randomly within each 1-km2 grid of the study area (total of 5.6 ha sampled). The size of the gorilla food tree species (large (>80 cm DBH), medium (50–80 cm DBH), small (20–50 cm DBH), and very small (o20 cm)) was based on mean DBH from phenology trees, rather than transects, because phenology trees were more representative of the large fruiting trees gorillas visit, vs. those on transects that included all stages of growth. To assess monthly food tree size, we identified dominant fruit food species each month (those that were present 94 / Doran et al. (a) Rainfall (Jul 95 - Jun 98) 300 millimeters 250 200 150 100 50 1995/96 1996/97 JUN MAY APR MAR FEB JAN DEC NOV OCT SEP AUG JUL 0 1997/98 (b) Phenology (Oct 96 - Sep 98) Oct-96 Nov-96 Dec-96 Jan-97 Feb-97 Mar-97 Apr-97 May-97 Jun-97 Jul-97 Aug-97 Sep-97 Oct-97 Nov-97 Dec-97 Jan-98 Feb-98 Mar-98 Apr-98 May-98 Jun-98 Jul-98 Aug-98 Sep-98 80 70 60 50 40 30 20 10 0 %INDIVIDUALS WITH RIPEFRUIT %SPECIES WITH RIPE FRUIT Fig. 1. Variation in (a) rainfall and (b) phenological patterns at Mondika. Phenological data are based on the presence or absence of ripe gorilla fruit in 197–216 trees and 20–32 lianas (individuals) of 23 species. No phenological data are available for June 1998. in >65% of fecal samples) and calculated the percentage of dominant foods from each tree size category. Herb densities. We determined the densities of all important herb species (occurring in 45% of food trails) by counting and identifying stems rooted within 136 4-m2 circular plots (i.e., of radius 1.13 m), located randomly within each 250  250 grid of the study area. Habitat type (i.e., Gilbertiodendron forest or mixed forest) was noted for each plot. To ensure adequate sample size, we plotted cumulative means of each species against numbers of plots sampled, and continued to add additional plots until changes in mean were negligible [Grieg-Smith, 1983]. Lowland Gorilla Diet at Mondika / 95 Measures of Diet We followed fresh trails of gorilla groups and recorded indirect indicators of diet (trail signs and fecal samples) from July 1995 through June 1998. We collected samples from all areas of the study site, which was used by a minimum of nine genetically distinct gorilla groups with overlapping home ranges (Bradley, personal communication). Groups were not individually recognizable, so samples were most probably collected from all groups, but were not ascribable to any particular group. Fecal analyses. We collected fecal samples from fresh (o24 hr old) gorilla trail and nesting sites and assigned sex on the basis of bolus size (silverback dung can usually be distinguished from smaller individuals (females and blackbacks) because dung diameter is proportional to body size) [reviewed in Tutin & Fernandez, 1993]. Although adult female and blackback males are similar in body size, and thus their fecal samples are indistinguishable, the majority of these samples from group nest sites can be assumed to be from females, given that long-term demographic studies of western gorillas indicate that adult females account for a far larger proportion of adult-sized individuals in mixed-sex groups than do blackbacks (proportion of adult-sized individuals: Maya (n = 588 adult-sized individuals: blackbacks = 4%, females = 96% [Magliocca et al., 1999]; Mbeli (n = 52 individuals in mixed-sex groups: blackbacks = 20%, females = 80% [Parnell, 2002]). We analyzed one fecal sample (of approximately 300 g) from one adult male and one female-sized gorilla per group per day, when possible. We dropped individual months from the analyses when there were less than three of either the male or female samples. As our ability to track gorillas improved, the number of nest sites (and fecal samples) encountered per month increased. On average we analyzed 23 samples per month (SD = 9.5, range = 8–44). Sample size (when >3) had little effect on our measure of the amount of fruit in diet per month: no significant correlation existed between the sample size and 1) the mean number of fruit species per fecal sample, 2) the total number of fruit species consumed per month, or 3) the average fruit or fiber score (Spearman rank correlation: n = 62 months; mean number of fruit species r = –0.2, t = –1.7, P = 0.09: total fruits r = 0.2, t = 1.7, P = 0.09; fruit score r = –0.14, t = –1.1, P = 0.27). Goldsmith [1996] also found no effect of fecal sample size on mean number of fruit species per sample, or on total fruits per month. She did not test for effect on fiber score. Fecal samples were weighed, dissociated in water, and rinsed through a 1mm screen. Wet samples were divided into component parts, including fruit (identified predominantly by seeds and occasionally by pulp), fiber (remnants of leaf and stem), and insects. The fiber score (0–100%) was defined as the proportion (by volume) of fiber, relative to fruit (including seeds, skins, and pulp) in a sample. The fiber score and its inverse, the fruit score (100% - fiber score), were used to track relative amounts of fiber (or fruit) in the diet throughout the year. We assessed fruit content by three measures: 1) quantity of fruit (fruit score), 2) number of different species present per sample, and 3) number of different fruit species occurring in all samples per month (total). We estimated abundance scores to define important fruits based on a scale of 1–4 (few, rare, common, and abundant, respectively). The fruits of most species could be identified to the species level. Exceptions occurred with some genera, such as Aframomum, Landolphia, Dialium, Drypetes, and Ficus, which had more than 96 / Doran et al. one species with indistinguishable seeds. For these, we recorded the genus present in fecal analysis, following Nishihara [1995]. Trail signs. While following the trail of a group, we recorded the presence of each food item the first time we encountered it each day (n = 617 feeding trails of >200 m). Food items were recognizable by the characteristic manner by which each food type had been processed, and by the particular plant parts discarded [Williamson, 1989]. Each month we tallied the percentage of days on which each food item appeared. We limited trail sign data to months (n = 31) where n 4 8 (n = 519 days of trails). We found a significant relationship between the number of herb/ leaf species encountered and trail sample size (r = 0.5, P = 0.004, n = 31 months), but unlike Goldsmith [1996], we found no significant correlation between the number of fruit species encountered each month and trail sample size (r = 0.1, P =0.6, n = 31 months). Important and reliable fruits. We defined fruits as ‘‘important’’ if they were present in fecal samples in at least 50% of months (long duration) in either large (type I) or small (type III) quantities, or if they were used in large quantities over a shorter duration (type II). ‘‘Large quantity’’ is defined as occurring in at least 50% of fecal samples during at least a single month, or a mean abundance fecal score Z2. Additionally, if fruits appeared in Z5% of feeding trails, we classified them as important (type IV). Fruits were considered ‘‘reliable’’ (or ‘‘highly reliable’’) if they occurred in >20% of fecal samples in the same month for 2 or 3 consecutive years. Food list. The food list includes plant items present in Z1% of overall food trails or fecal samples, or Z10% of food trails or fecal samples in any single month. Statistics The results of correlations are reported as ‘‘r’’ and are based on two-tailed Pearson correlation tests, unless otherwise indicated. RESULTS Resource Availability and Seasonality Herb densities. Overall densities of herbs commonly consumed by gorillas were less than one stem per m2 in both Gilbertiodendron and mixed forest (Table I). Marantaceae constituted the majority of stems in both forest types. Haumania danckelmaniana, the herb gorillas ate most frequently, was the most abundant species at the study site. However, its stem densities cannot be directly equated to food availability because gorillas eat only growing shoots, not mature plants (personal observation, Nishihara et al., 1995). Only 5.5% of Haumania stems recorded in the sample plots were growth shoots, effectively reducing the availability of this resource to 0.018 shoots per m2. Herbs followed a clumped distribution, both overall and at the species level. Herb stems were found in only 61% and 45% of plots in mixed and Gilbertiodendron forest, respectively, with individual species found in fewer plots. Overall mean herb density was 0.78 stems per m2 in mixed TABLE I. Stem Densities of Herbaceous Plants in Forest Types Species Mixed forest (n =103 plots) Haumania danckelmaniana Sarcophrynium spp. Megaphrynium macrostachyum/trichogynum Aframomum sp. Palisota ambigua Family Marantaceae Marantaceae Marantaceae Zingiberaceae Commelinaceae Gilbertiodendron forest (n = 33 plots) Haumania danckelmaniana Sarcophrynium spp. Megaphrynium macrostachyum/trichogynum Aframomum spp. Palisota ambigua All Marantaceae All THV Marantaceae Marantaceae Marantaceae Zingiberaceae Commelinaceae 795% confidence interval SD Range (stems/m2) % of plots in which found Variance: mean ratio 0.33 0.07 0.20 0.07 0.11 0.10 0.06 0.19 0.04 0.08 0.51 0.29 0.95 0.20 0.43 0–2.8 0–2.5 0–6 0–1 0–3 46 11 6 16 11 3.14 4.83 18.54 2.13 6.66 0.60 0.78 0.23 0.27 1.20 1.37 0–8.8 0–10 54 61 – – 0.48 0.03 0.24 0.06 0.70 0.17 0–2.2 0–1 42 3 3.99 4.00 0.07 0.08 0.23 0–0.8 9 3.04 0.52 0.58 0.27 0.29 0.77 0.83 0–2.8 0–2.8 42 45 – – Standard deviations (SD), ranges, and percentage occurrence are given as measures of the variation in herb density in different plots. The variance to mean ratio, where greater than one (P o 0.001 in each case), indicates a clumped distribution [Grieg-Smith, 1983] Lowland Gorilla Diet at Mondika / 97 All Marantaceae All herbs Mean (stems/m2) 98 / Doran et al. forest, although stem densities varied considerably (0–10 stems/m2). Standard deviations were correspondingly high. Variance-to-mean ratios (coefficient of dispersion) were greater than one for all species, indicating significantly clumped distributions (significance testing procedure from Grieg-Smith [1983]). Tree densities. Average stem density for all trees (of Z10 cm DBH) was 419.5 trees/ha, with a total of 2,349 trees of 269 species enumerated (5.6 ha). The average DBH of the 2,349 trees was 22.7 7 17.6 cm (range = 13–175, based on transect sampling). Species commonly indicative of recent disturbance or human presence, Musanga cercropoides and Elaeis guineensis, were absent. The forest was diverse, with the most common species, Gilbertiodendron dewevrei (Caesalpiniaceae), accounting for only 6% of trees sampled (Table II). Gorillas selectively consumed fruit species; they ate fruit from only 26% of possible tree species. Trees of important fruit species used in large quantities for long time periods (type I trees) were rare, accounting for r 0.5% of trees. Trees of other important gorilla fruit species were among the 25 most common tree species, including Anonidium mannii (Annonaceae), Polyalthia suaveolens (Annonaceae), Dialium spp. (Caesalpiniaceae), Diospyros ituriensis (Ebenaceae), Angylocalyx pynaertii (Papilionaceae), TABLE II. Relative Availability, Shown in Decreasing Order, of the 25 Most Common Tree Species at Mondika Family Caesalpinaceae Ulmaceae Euphorbiaceae Ebenaceae Olacaceae Annonaceae Unk Annonaceae Caesalpinaceae Sapindaceae Meliaceae Meliaceae Sterculiaceae Rubiaceae Flacourtiaceae Meliaceae Sterculiaceae Euphorbiaceae Ebenaceae Papilionaceae Euphorbiaceae Sapindaceae Ebenaceae Tiliaceae Moraceae Sterculiaceae Species Density % stems Gilbertiodendron dewevrei Celtis mildbraedii/tessmannii/zenkeri* Dichostemma glaucescens Diospyros bipendensis Strombosia nigropunctata/pustulata Anonidium mannii Unk Polyalthia (Greenwayodendron) suaveolens Dialium spp. Pancovia harmsiana/pedicellaris Guarea vel sp aff thompsonnii/laurentii Carapa procera Nesogordonia papaverifera Pausinystalia macroceras Oncoba (Caloncoba) mannii Trichilia prieuriana and other spp. Sterculia (Eribroma) oblonga Macaranga barteri/monandra/spinosa Diospyros ituriensis Angylocalyx pynaertii Grossera macrantha Pancovia laurentii Diospyros canaliculata Desplatsia spp. Myrianthus arboreus Cola lateritia 27 21 19 16 12 11 11 10 10 8 8 8 6 6 6 6 6 5 5 5 5 5 5 4 4 4 6 5 5 4 3 3 3 2 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 1 1 1 A total of 2,349 trees of 269 species were enumerated in 28 (200  10 m) transects. Density is the number of stems per hectare. Percent stems indicates the relative proportion of each species. ‘‘Important’’ gorilla fruit (bold type) and leaf (*) species are indicated. Lowland Gorilla Diet at Mondika / 99 Pancovia laurentii (Sapindaceae), and Myrianthus arboreus (Moraceae). Celtis (Ulmaceae), an important leaf species, was the second most common tree. Seasonality in resource availability. Fruit production varied across months and years (Fig. 1b). An average of 9% of individuals (range = 2–21%) and 38% of species (range = 14–68%) bore ripe fruit per month (n = 23). Generally, fruit availability increased in June or July, peaked in August or September, and was low from November to March. The number of individuals bearing ripe fruit was positively correlated with rainfall, unlike the number of species (individuals r = 0.54, P = 0.009: species r = 0.34, P = 0.12: n = 22 months). Overall Diet Composition Food list. The gorilla plant food list consisted of 127 plant food items (of 100 species), including 70 fruit, 33 leaf, 14 stem, two flower, and eight bark species (Table III). Gorillas also consumed ants, termites, and soil. Composition of the diet based on fecal samples. Fifty-six fruit and one leaf species were identified from fecal samples (n=715; Table III). Gorillas fed on fruit year-round, with an average of 3.5 species per day, 10 species per month, and an average fruit score of 37% (Table IV). Fruit was present in 99.8% of male and 100% of female fecal samples. The nonfruit component of diet (herbs/leaves), present in 99.7% of samples and accounting for 61% of samples by volume (fiber score), dominated the majority of samples. We could not reliably assess herb/leaf species from fecal samples. No sex difference existed in the degree of frugivory or amount of fiber in diet (Table IV). Important fruit species. We classified 24 fruit species as important based on fecal samples, including five type I (present in large quantity for 4 50% of months), three type III (present in over 50% of months but in smaller quantities), and the majority (type II) that were used intensively for shorter periods of time (Table V). We considered an additional species (Irvingia excelsa) important because it appeared on more than 5% of the feeding trails. Important gorilla fruit species were from trees of various sizes, including three large (mean >80 cm DBH), four medium (mean >50– 80 cm), 10 small (mean >20–50 cm), and eight very small (DBH of mean o20 cm, or from herbs) (Table V). Composition of the diet based on trail signs. We identified 81 items of 62 plant and two insect species in Z1% of gorilla trails (n = 619; Table III). Trail signs indicated that gorillas ate an average of 10 fruit species (Table IV) and 15 herb/leaf species per month, including a total of 33 leaf, 14 stem, and eight bark species that were not detected in fecal samples (Table IV). The 24 most frequently encountered food items occurred on Z5% of trails, and could be considered important foods. These included seven species of herb stem, pith, or shoot; nine species of fruit (including six tree and three herb species); six species of leaves; one species of bark (Celtis spp.); and termites (Table VI). The four most frequently encountered items were herbs (Haumania danckelmaniana (Marantaceae), Aframomum limbatum (Zingiberaceae), Family Species Local name F L S F B %TR T 14 10 Acanthaceae Thomandersia hensii Whitfieldia elongata? Ingooka Indolu Anacardiaceae Antrocaryon klaineanum/micraster Trichoscypha acuminata Modiali Indoya/Indolya t,d t,d 1 1 Anonidium mannii Artabotrys sp.? Hexalobus crispiflorus Greenwayodendron (Polyalthia) suaveolens Uvariastrum germainii/ pierreanum Mobei Mbeko Pota Motunga Beti d,t d t,d t,d d 6 Aponcynaceae sp. Funtumia elastica Landolphia sp 1 Landolphia sp 2 Landolphia sp 3 Pleiocarpa pycnantha Tabernaemontana penduliflora Tabernaemontana crassa Pycnobotrya nitida Ivua Landa Ndembo Bosindja Pembe Mosebe Etokoloko Gombo Mongenje Arecaceae Anchomanes difformis Laccosperma secundiflora Ekoule-akoumba Gao Burseraceae Santira trimera Baba d Caesalpiniaceae Copaifera mildbraedii Detarium macrocarpum Dialium sp. Dialium sp. Erythrophleum ivorense Gilbertiodendron dewevrei Mondumba Etebele Mokombe Mbaso Mbanda Bemba d d,t t t t,d Annonaceae Apocynaceae t t 2 t t t T t,d t t t,d t,d t,d %DF %DM 1 1 7 2 3 7 1 8 1 5 6 4 2 5 5 1 1 T 1 2 2 2 3 T T 2 13 t 1 T t t T T T 1 2F,2L 1 1 4 100 / Doran et al. TABLE III. Gorilla Plant Foods Recorded in Either a Minimum of 1% of Fecal (d) or Trail Samples (t) Between June 1995 and July 1998 Chrysobalanaceae Parinari excelsa or Maranthes glabra Mokanja Commelinaceae Palisota ambigua and other spp. Palisota brachythyrsa/tholloni Doto Mangabo t Dioscoreaceae Dioscorea sp.? Mondiki t Ebenaceae Diospyros Diospyros Diospyros Diospyros Lembe Babangu Mulombo Embandja crassiflora ituriensis mannii bipindensis t,d 3 3 43 4L,9S 1 1 2 1 1 1 t Mongamba Kpaya Tembo Embundubundo Mokosa Oncoba (Caloncoba) glauca Oncoba (Caloncoba) welwitschii Unknown sp. 2 Unknown sp. 3 Oncoba (Caloncoba) crepiniana Isoko Ita ti isoku – Batalo Gnetaceae Gnetum africanum Koko Irvingiaceae Irvingia excelsa Klainedoxa gabonensis Payo Bokoko t,d t,d Loganiaceae Strychnos sp. Mobangui d Loranthaceae Helixanthera subalata Tonga Maranthaceae Ataenidia conferta Haumania danckelmaniana Hypselodelphis scadens? Maranthochola congolensis Megaphrynium macrostachyum/trichogynum Sarcophrynium schweinfurthii/brachystachys Trachyphyrnium braunianum Boboko Basele Poposo Mbili Ngungu Kaya Mbonge Flacourtiaceae T T t,d t,d t Dichostemma glaucescens Drypetes diopa & others Drypetes spp. Keayodendron brideloides Manniophytum fulvum Euphorbiaceae 1 t t,d t,d t,d 1 2 4 4 2 4 5 1 4 1 30 4 32 8 10 22 1 19 1 t t 1F,1L t 10 5 11 t t 1 t t,d t,d d t t T T T T T T 1 6F,79S 2 7F,29L 16S 3L,4S 1 Lowland Gorilla Diet at Mondika / 101 D t,d D D D Family Species Local name F L t Menispermaceae Dioscoreophyllum cumminsii Triclisia dictyophylla Tiliacora sp. unknown sp. 6 Mola Ludyasumbu Etokobola – d,t t,d d d Mimosaceae Albizia sp.? Tetrapleura tetraptera Ebamba Ecombolo t,d Ficus natalensis? Ficus spp. Myrianthus arboreus Treculia Africana Ngumu Ngumu/Dobo Ngata Vousa d t,d d,t Olacaceae Strombosia nigropunctata/postulata Embungu t,d Pandaceae Panda oleosa Mokana t Papilionaceae Angylocalyx pynaertii Milletita sp.? Pterocarpus soyauxii Unknown sp. 7 Manjombe Inganda Embema Molinda t,d Passifloraceae Barteria dewevrei/fistulosa Ngumanguma t,d Rubiaceae Colletocoema dewevrei Nauclea diderrichii Mobobo Mosei d t,d Chytranthus gilletii Chytranthus macrobotrys Pancovia laurentii Esekerende Motokodi Ingoyo t,d t t,d Autranella congolensis Chrysophyllum beguei or pruniforme Chrysophyllum (Gambeya) lacourtiana Banga Ingadje Bambu t d t,d Moraceae Sapindaceae Sapotaceae S F B %TR %DF %DM 1 1 2 1 1 1 6 24 27 1 1 4 1 7 7 16 2 20 2 7 5 1 3 3 1 1 1 1 t t,d t t t 1F,6L 1 T T 1 t 1 3 7 1 11 14 16 7 102 / Doran et al. TABLE III. Continued Chrysophyllum (Gambeya) perpulchrum Mannilkara mabokeensis Synsepalum longecuneatum Chrysophyllum bukokense Koloka Mongenja Mokenjenje Mondonge A,B t t 1 Sterculiaceae Cola chlamydantha Bongbo t 1 Tiliaceae Desplatsia spp. Duboscia macrocarpa Grewia oligoneura Liamba Nguluma Boukou t,d t,d d Ulmaceae Celtis adolfi-friderici Celtis mildbraedii/ tessmannii/zenkeri Kakala Ngombe d Verbinaceae Vitex doniana or welwitschii Mogweagwea d Violaceae Rinorea spp. Rinorea spp. Esandja Esandja-mbongo Vitaceae Cissus spp. Mojambi d Zingiberaceae Aframomum limbatum and other spp. Aframomum subsericium Njombo Njokoko t,d t Unknown Unknown sp. 21 Ikumbi Mondamandama t t 1 t 4F,1L 19 1 1 5 74 3 5 75 2 1 t T t t 1 7 31 29 T 9F,58S 1F,5S 2 Parts consumed include fruit (F), leaf (L), stem (S), flower (F), and bark (B). Percentage of trails (%TR) and fecal samples (%DF) = females; %DM = males) in which the item was found are indicated. A question mark indicates that identification remains to be confirmed. Local names are in the BaAka dialect. Lowland Gorilla Diet at Mondika / 103 t t t 10L,7B 104 / Doran et al. TABLE IV. Comparison of Results of Gorilla Diet Based on Female and Male Fecal Samples and Trail Signs Sample size Mean no. of samples/mo No. Fruit species/sample (n = 31 mo) Fruit score (n = 31 mo) Total number of fruits/month (n = 31 mo) Mean no. of herb/leaf species per mo Fecal: female Fecal: male 315 10.274.4 (range=3–21) 3.671.6 (range=1.6–8.3) 37.3714.7% (range=10–82%) 9.573.2 (range = 4–16) NA 400 12.975.6 (range=4–24) 3.571.5 (range=2–8.2) 38.0714.3% (range=15–70) 10.273.0 (range=6–18) NA Sex difference? NS (Mann Whitney U=464, P=0.816, n =31 mo) NS (Mann Whitney U = 469, P=0.871 NS (Mann Whitney U=400, P=0.257 Trail signs: group 617 19.178.8 (range=8–18) NA NA 9.676.8 (range=1–30) 15.076.9 (range=4–31) Fecal data include three measures of frugivory including mean number of fruit species per sample, fruit score, and different number of fruit species consumed per month (7one standard deviation). Trail sign data include the mean number of herb/leaf species and the total number of fruit species consumed per month. TABLE V. Important Fruits in the Diet Species Family Life Form Density (No./ha) DBH Type I, Long duration, large quantity Duboscia macrocarpa Tetrapleura tetraptera Klainedoxa gabonensis Aframomum spp. Megaphrynium macrostachyum Tiliaceae Mimosaceae Irvingiaceae Zingiberaceae Marantaceae Tree, med Tree, med Tree, big Herb Herb Type II. Short duration, large quantity Landolphia spp Anonidium mannii Diospyros ituriensis Dialium spp. (pachyphylum and zenkeri) Hexalobus crispiflorus Barteria dewevrei/fistulosa Haumania danckelmaniana Drypetes spp. (diopa and others) Diospryos mannii Strychnos sp. Oncoba (Caloncoba) welwitschii Pancovia laurentii Tabernaemontana spp. (penduliflora and crassa) Greenwayodendron (Polyalthia) suaveolens Angylocalyx pynaetrii Vitex doniana or welwitschii Apocynaceae Annonaceae Ebenaceae Caesalpiniaceae Annonaceae Passifloraceae Marantaceae Euphorbiaceae Ebenaceae Loganiaceae Flacourtiaceae Sapindaceae Apocynaceae Annonaceae Papilionaceae Verbinaceae Liana Tree, small Shrub Tree, small/very small Tree, med Tree, small Herb Small, tree or shrub Tree, small Liana Tree, small Tree, small Tree, small Tree, small Tree, med Tree, small Type III. Long duration, small quantity Myrianthus arboreus Chrysophyllum (Gambeya) lacourtiana Ficus spp. Moraceae Sapotaceae Moraceae Tree, small Tree, big Strangler 4.1 2.3 36.3 (31.3) 82.7 (23.8) Tree, big 0.7 123.7 (52.9) 69.3 (36.3) 52.4 (39.2) 111.1 (38.1) – 10.9 5.2 1.25 and 0.53 1.1 2.1 49.3 (26.9) (13.8) (11.6 and 19.6) 67.6 (24.7) (17.2) 0.4 and 3.0 1.2 20.6 2.5 4.8 0.9 and 1.25 10.2 5.2 2.1 21.6 36.3 (12.0 38.1 51.2 37.8 (22.5) (23.2) and 12.1) (23.9) Definitions are modified after Nishihara [1995]. Long duration fruits were present in fecal samples during 50% of months. Large quantity occurred in 4than 50% of samples in at least a single month or had a mean abundance fecal score of 2. Sizes of trees include: big (480 cm), med (50–80 cm), small (20–50 cm), and very small (o20 cm). DBH in parentheses is from transects. Lowland Gorilla Diet at Mondika / 105 Type IV. Additional fruits that are present in 45% of feeding trails Irvingia excelsa Sapotaceae 2.1 1.6 0.7 106 / Doran et al. TABLE VI. Important (Appearing on 5% Trails) Gorilla Foods Based on Trail Sign Data Family Species Haumania danckelmaniana Aframomum limbatum and other spp. Palistota ambigua and other spp. Megaphrynium macrostachyum Duboscia macrocarpa Cubitermes sp. Marantaceae Megaphrynium macrostachyum Acanthaceae Thomandersia hensii Arecaceae Laccosperma secundiflora Sapotaceae Chrysophyllum (Gambeya) lacourtiana Irvingiaceae Klainedoxa gabonensis Ulmaceae Celtis mildbraedii/tessmannii/zenkeri Gnetacae Gnetum africanum Acanthaceae Whitfieldia elongata? Zingiberaceae Aframomum limbatum and others Commelinaceae Palisota brachythyrsa/tholloni Ulmaceae Celtis mildbraedii/tessmannii/zenkeri Marantaceae Megaphrynium macrostachyum/ trichogynum Marantaceae Haumania danckelmaniana Papilionaceae Angylocalyx pynaertii Mimosaceae Tetrapleura tetraptera Annonaceae Anonidium mannii Irvingiaceae Irvingia excelsa Zingiberaceae Aframomum subsericium Marantaceae Zingiberaceae Commelinaceae Marantaceae Tiliaceae Local name Life form Part %TR Basele Njombo Doto Ngungu Nguluma Kusu Ngungu Ingoka Gao Bambu Bokoko Ngombe Koko Indolu Njombo Mangabo Ngombe Ngungu Herb Herb Herb Herb Tree, med Insect Herb Shrub Herb Tree, big Tree, big Tree, big Vine Shrub Herb Herb Tree, big Herb sh st st sh fr st lf st fr fr lf lf lf fr st bk fr 79 58 43 29 19 19 16 14 13 11 11 10 10 10 9 9 7 7 Basele Manjombe Ecombolo Mobei Payo Njokoko Herb Tree, Tree, Tree, Tree, Herb fr lf fr fr fr st 6 6 6 6 5 5 med med small big Percent of days on which each food was encountered (%TR), life form, and plant parts eaten including stem (st), shoot (sh), leaf (lf), fruit (fr), and bark (bk) are indicated. Palisota ambigua (Commelinaceae), and Megaphrynium macrostachyum (Marantaceae). Two species of herbs, Haumania danckelmaniana and Aframomum limbatum, were eaten on more than 50% of days. Four species of leaf, including Thomandersia hensii (Acanthaceae), Celtis mildbraedii (Ulmaceae), Gnetum africanum (Gnetaceae), and Whitfieldia elongata (Acanthaceae), were consumed on more than 10% of days. None of these herb or leaves from woody plants were identified as important on the basis of fecal sampling. Gorillas ate termites on 19% of days. Three species of fruit were present on more than 10% of days. These include Duboscia macrocarpa (19%), Klainedoxa gabonensis (11%), and Chrysophyllum (Gambeya) lacourtiana (11%). Of the nine species of fruit classified as important on the basis of feeding trails, all but one, Irvingia excelsa, was identified as important on the basis of fecal samples. Sixteen additional species of important fruits (based on fecal samples) were not observed on 4 5% of feeding trails, but appeared in 4 10% of trails during at least 1 month. Seasonal Influence on Diet Fecal samples. Fruit and herbs were present in the gorilla diet throughout the year, with their proportions fluctuating relative to changes in fruit availability (Fig. 2). As fruit availability increased, the mean number of fruit species per sample Lowland Gorilla Diet at Mondika / 107 (a) Number of fruit species per sample 10 8 6 4 2 NOV DEC DEC 1997 OCT SEP AUG JUL 1996 NOV 1995 JUN MAY APR MAR FEB JAN 0 1998 (b) Total fruit species per month 25 20 15 10 5 95/96 96/97 OCT SEP AUG JUL JUN MAY APR MAR FEB JAN 0 97/98 (c) Fiber score per sample 60 40 1995 1996 1997 DEC NOV OCT SEPT AUG JUL JUN MAY APR MAR FEB 20 0 JAN % fiber 100 80 1998 Fig. 2. Monthly and annual variation in male gorilla diet (fecal samples), including (a) mean number of fruit species per fecal sample per month, (b) total number of different fruit species consumed each month, and (c) relative fiber score. Although only the male pattern is shown, the female pattern is not significantly different. increased and fiber score decreased for both males and females [r (fruit): males = 0.67, P = 0.001; females =0.60, P = 0.004: r (fiber score): males = 0.69, P o 0.001; females = 0.69, P o 0.001: n = 21 months). Fiber score and the number of fruit species per fecal sample were inversely correlated (r: male = –0.68, P o 0.001, female = –0.46, P = 0.04; n = 21 months). The total number of different fruit species appearing in fecal samples each month was neither significantly positively correlated with fruit availability (male r = 0.41, P = 0.06; female r = 0.29, P = 0.2) nor negatively correlated with fiber score (male r = –0.32, P = 0.14; female r = –0.15, P = 0.5; n = 22 months), suggesting that a minimum variety of 108 / Doran et al. fruit is maintained in the diet throughout the year, even when fruit availability is minimal. Fruit availability and the number of reliable fruits per month were not significantly correlated (r = 0.54, P = 0.09, n = 11 months). On average, five species of reliable fruit, including three from trees (Duboscia macrocarpa, Tetrapleura tetraptera, and Desplatsia dewevrei) and two herbs (Aframomum spp. and Megaphrynium macrostachyum), were particularly important during the period of lowest fruit availability (October–February). Seasonal variation existed in the size of food trees that dominate gorilla diet each month (Fig. 3). Gorillas commonly used medium-sized trees throughout the year (with a slight drop in September–October). Small trees were used most frequently during the major fruiting season (June–September), with seven of 10 important small species (Table V) among those most commonly available (Table II). Large trees, used immediately prior to (March–June), and after (September– October), the major fruiting season, are rare in the environment and require longer distances to travel between them. Trail signs. Seasonal differences in diet determined from trail signs broadly resembled those from fecal sampling; as fruit availability increased, herb/leaf diversity in diet decreased (r = –0.57, P = 0.007; n = 21 months; Fig. 4a). However, this was not uniformly true across all herb species. Protein-rich Haumania danckelmaniana shoots [Nishihara, 1995] were a dietary staple; they were eaten on average 79% of days each month (Fig. 4b; SD = 13, range = 55–94), and their use was not correlated with fruit availability (Fig. 4b; r = 0.11, P = 0.63, n = 21 months). Other food items were used (to varying extents) as fallback foods, as indicated by an inverse relationship with fruit availability (Fig. 4c; n = 21 months in all cases). These included the pith of the second most frequently encountered herb, Aframomum limbatum (r = –0.72, P 4 0.001), the four most important leaf species, significantly in three of the four cases and approaching significance in the fourth (T. hensii r = –0.61, P = 0.003; C. mildbraedii/tessmannii/zenkeri r = –0.49, P = 0.02; G. africanum r = –0.46, P = 0.03; W. elongata r = –0.40, P = 0.07) and bark (r = –0.77, P o 0.0001). Termite use increased with rainfall (r = 0.4, P = 0.01, n = 36 months) but was not significantly correlated with fruit availability (r = 0.13, P = 0.58, n = 21 months). DISCUSSION Site Differences in Resource Availability Herb availability. Overall herb densities (of important gorilla foods) at Mondika were comparable to those at Bai Hokou, but lower than those at Lopé and Ndoki (Table VII). In particular, the density of Haumania spp., arguably the most important herb across sites, was two to three times lower at Mondika (0.33 stems/ m2) and Bai Hokou (0.44 stems/m2 [Goldsmith, 1996]) relative to Lopé (0.90 stems/m2 [White et al., 1995]). Reduced herb densities should result in smaller group size or longer daily path lengths, if other factors are equal [Doran & McNeilage, 2001]. In addition to herbs in terra firma forest, swamp herbs were present (but not measured) at Mondika and Ndoki, but absent from Bai Hokou and Lopé. The four sites, ranked in descending order of herb density, are Ndoki 95/96 96/97 97/98 May Apr Mar Feb Jan Dec Nov Oct Sep Jul Aug (a) Large Trees (> 80 cm DBH) 120 100 80 60 40 20 0 Jun % of fecal samples Lowland Gorilla Diet at Mondika / 109 98 97/98 May Apr Mar Feb Jan Dec 96/97 98 95/96 96/97 97/98 May Apr Mar Feb Jan Dec Nov Oct Sep Jul Aug (c) Small Trees (20 - 50 cm DBH) 120 100 80 60 40 20 0 Jun % of fecal samples 95/96 Nov Oct Sep Aug Jul Jun % of fecal samples (b) Medium trees (50 - 80 cm DBH) 120 100 80 60 40 20 0 98 Fig. 3. Monthly variation in the relative size of fruit trees used. Proportion of dominant food species that were from (a) large (>80 cm), (b) medium (50–80 cm), and (c) small (20–50 cm) trees each month are indicated. (with high herb density plus presence of swamp vegetation), followed by Lopé, Mondika, and Bai Hokou (low density and absence of swamp herbs). Fruit availability. Comparable data on fruit availability are limited to those from Mondika and Lopé. The forest at Mondika was more diverse in its species composition than Lopé, with higher tree density (stems/ha: Mondika = 419.5; Lopé = 384.5 (a) Fruit and herb consumption (trails) 40 30 20 10 Fruit May-98 Feb-98 Nov-97 Aug-97 May-97 Feb-97 Nov-96 Aug-96 May-96 Feb-96 Nov-95 0 Aug-95 Number of species 110 / Doran et al. Leaf/Herb (b) Monthly herb consumption (trails) % of trails 100 80 60 40 20 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Haumania Aframomum % of trails (c) Monthly leaf consumption (trails) 45 40 35 30 25 20 15 10 5 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Fig. 4. Monthly variability in diet (trail signs), including (a) total number of fruit and herb/leaf species encountered each month, (b) frequency of Haumania danckelmaniana and Aframomum sp., and (c) frequency of the four most important leaf species (Thomandersia hensii, Celtis mildbraedii/ tessmannii/zenkeri, Gnetum africanum, and Whitfieldia elongata). [Williamson, 1989]), a greater number of species contributing to make up 50% of the population (Mondika = >20: Lopé = 6 [Williamson, 1989]), and a lower proportion of forest formed by the 10 most common trees (Mondika = 32%: Lopé = 61% [Williamson, 1989]). At Lopé, one species (Cola lizae) accounted for a quarter of all trees, and was an important food (found in 16.7% of fecal samples) [Williamson, 1989]; the most common important fruit tree species at Mondika accounted for only 3% of trees (Table III). In spite of considerable differences in forest structure, habitat (Lopé includes seven habitat types compared to three at Mondika), and species composition, densities of two key fallback species of fruit Lowland Gorilla Diet at Mondika / 111 Table VII. Resource Availability and Western Lowland Gorilla Diet at 4 Sites Sites Ndoki Lopé Mondika Bai Hokoua Bai Hokoub Length of study Average rainfall Number of dry season months Fecal sample size Fecal samples per month Resource availability Herb density Gorilla diet: staple foods Herbs, Haumania sp. Swamp Herbs Fruit No. species overall Average no. species/mo Species/d Average fruit score Fallback foods Duboscia Low quality herbs Bark Leaves Seasonally important Klainedoxa Tetrapleura Herb fruits 1 yr 1,430 3 522 42 (9–53) 3 yr 1,536 3 180* 5 (5) 3 yr 1,415 3 715 21 (6–45)** 27 mo 1,401 3 859 35 (3–102) 19 mo 1,952 3 528 31 (6–56) 2.25 1.87 0.78 0.82 0.82 + + + – + + + – + – 115 91 70 77 17(8–36) 14.2 (7–20) 9–10 (4–16) 13 (5–28) 3 3.0 3.5 3.2 48% 39% 15.7 3.5 + + + + + + + + + + + + + + + + + – + + + + + + – + + + + + + + + + + Sites are listed in decreasing rank order of herb availability. Staple foods are used throughout the year. Fallback foods are used more frequently when succulent fruits are not available, although generally available throughout the year. Fruit is included as a staple food, because a minimum diversity is maintained in diet throughout year (although no single species is used throughout the year) in spite of major changes in availability. Diet data are from Nishihara [1995] (Ndoki); Tutin et al. [1991]; Rogers & Williamson [1988]; (Lopé); this study (Mondika); Remis [1997] (Bai Hokoua); Goldsmith, 1996 (Bai Hokoub). Resource availability data are from Malenky et al. [1994]; White et al. [1995]; this study, and Goldsmith [1996]. *Sample size for Lopé is from five random samples per month. **Sample size for Mondika is from combined male and female samples. trees were similar and relatively rare at both sites (Mondika vs. Lopé (stems/ha): Duboscia macrocarpa (2.1 vs. 2.2), and Klainedoxa gabonensis (0.7 vs.0.7)). Gorilla Diet at Mondika Rainfall, length of dry season, and fruit availability were seasonal at Mondika, and similar to other western lowland gorillas sites (Table VII). Previous researchers reported gorilla diet on a monthly basis, but distinguished between two different periods of fruit availability: rainy vs. dry season [Remis, 1997; Goldsmith, 1999], or good fruit and bad fruit months [Tutin, 1996]. We distinguish three annual time periods at Mondika, which differ in the size, density, and distribution of available gorilla foods in a manner we expect is associated with predictable differences in ranging and/or group cohesion. During the first period (November–February), which included but was not restricted to the dry season, fruit availability was low (Fig. 1). Nonfruit foods, including low-quality herbs (e.g., Aframomum spp. and Palisota), a wide variety of leaves, and Celtis bark, dominated the diet. These resources were commonly available, and those in trees (particularly Celtis sp.) occurred in patches large 112 / Doran et al. enough to permit many gorillas to feed. In spite of reduced fruit availability, gorillas continued to feed on five species of reliable fruit. These were also commonly available, either in herb patches (Aframomum spp. and Megaphrynium macrostachyum) or in trees that were relatively common (intermediate in density = 4 o x o 1 stem/ha) and 450 cm DBH in size (Tetrapleura tetraptera and Duboscia macrocarpa), or small and commonly distributed (Desplatsia dewevrei, four stems/ha). During this period, gorillas ate two species of succulent fruit (Pancovia laurentii and Angylocalyx pynaertii) that were among the most common species in the study site. However, their fruiting patterns were synchronous (within species), brief and unpredictable, and less likely to influence ranging. The first annual period was thus characterized by low fruit availability combined with relatively evenly distributed and commonly available resources. Important fallback tree species (Tetrapleura tetraptera and Duboscia macrocarpa) were large enough to permit all individuals in a group to feed (average nest group size = 7.4, SD = 3.7, n = 514 nest sites [Mehlman & Doran, 2002]). During this period, daily path lengths should be at their lowest and groups should be relatively cohesive. The second period (April–June) was characterized by increased rainfall and fruit availability. Gorillas responded by eating a greater quantity and variety of fruit, and reduced variety of leaf species. Both protein-rich (Haumania danckelmaniana) and lower-quality herb consumption was high. Termite consumption increased as a function of rainfall. Gorillas used three reliable fibrous (vs. succulent) fruit species, including Duboscia macrocarpa, the large and rare Klainedoxa gabonesis, and the small and common Myrianthus arboreus. Additionally, Chrysophyllum (Gambeya) lacourtiana, a succulent fruit found in large and rare trees, was eaten frequently when available, although its availability was less predictable between years than the other species. Thus during the second period, reliance on commonly available tree leaves decreased, and as a result fewer preferred foods were available in a given area. Additionally, key fallback fruit trees were large and rare, and as a result daily path length increased, but groups could remain cohesive. During the third period (July–September), fruit availability peaked. Fruit consumption was at its highest, and herb (with one exception), leaf, and bark consumption was at its lowest. Haumania danckelmaniana consumption remained high, consistent with Kuroda et al.’s [1996] finding that protein-rich herbs were ingested throughout the year, unlike lower-quality herb use, which diminished with increasing fruit availability. During this period, succulent fruit was readily available, although patches were too small to permit all group individuals to feed, and too widely spaced to allow individuals to feed in separate patches and retain short-range vocal contact. We expect the number of trees visited and the daily path length to increase, or groups to be less cohesive– a pattern not predicted during the second period. Thus, throughout the year gorilla diet consisted of fluctuating proportions of herbs, leaves, and fruit. Fruit remained a consistent component in diet yearround at Mondika, even in periods of low fruit availability. Gorillas ate an average of 3.5 fruit species per day and were selective in their fruit choice, ignoring some common fruits and incorporating rare fruits to a degree higher than predicted on availability. Although ranging data are not available from the same time period of this study, more recent data, including those from all other sites, indicate that gorillas travel farther to add fruit to their diet when it is available, rather than subsist on lower-quality forage [Doran & Greer, 2002; Goldsmith, 1999; Remis, 1997; Tutin, 1996]. This was not the case for herb consumption [Doran & Greer, Lowland Gorilla Diet at Mondika / 113 2002; Goldsmith, 1999], except on occasions when gorillas traveled long distances to reach swamps to feed on aquatic herbs (unpublished results). Our results did not indicate a sex difference in diet. Large male body size was not associated with greater folivory; males and females did not differ in daily or monthly fruit consumption or fiber content. Similarly, Remis [1997] did not detect a sex difference in the mean monthly number of fruits consumed, although she reported that females ate more species of fruit per day than did males during one poor fruit season. Given that sex differences have been reported in other ape species, with less body size dimorphism [e.g., Galdikas & Teleki, 1981; McGrew, 1992], perhaps indirect methods are simply not fine-grained enough to detect subtle differences in diet. We discuss this topic further below. Gorilla Diet Across Sites In spite of differences in resource availability across sites, gorilla diet, as currently measured, was remarkably similar (Table VII). Reduced herb density at Mondika (relative to Ndoki and Lopé) was not associated with increased reliance on fruit, as proposed by Kuroda et al. [1996], nor was a greater number of available fruit species (at Mondika relative to Lopé) coincident with a greater number of fruits recorded in the diet (Table VI). At all sites, fruit and herbs were important components of diet throughout the year, fruit production was highly seasonal. The quantity and quality of fruit and fiber, and the number of fruit species in the diet fluctuated in response to changes in fruit availability. Gorillas ate an average of three and maximum of 12 species of fruit per day, and an average of 10–17 and maximum of 36 species per month. Greatest monthly fruit diversity in the diet occurred at Ndoki, the site of greatest herb availability. Average fiber scores were also equivalent across sites, although less directly comparable across sites due to differences in methods. Gorillas shared key components of diet across sites. Staple foods, eaten yearround, included high-quality herbs (Haumania), swamp herbs when present, and (we would argue) a minimal diversity of fruit. Fallback foods (those present yearround but incorporated into the diet when more preferred foods are absent) included leaves, bark, low-quality herbs (but see Goldsmith [1996]), and two species of fruit (Duboscia macrocarpa and Klainedoxa gabonensis). Densities of these two species were similar across sites for which data are available, in spite of major differences in forest structure, suggesting that these may influence gorilla density. Our findings are consistent with Williamson’s [1989] interpretation that western lowland gorillas pursue succulent fruits at some cost, while incorporating high-quality herbs as a staple and complementary dietary component. Costs to frugivory entail greater day ranges and possible constraints on group size (which can be buffered somewhat in areas with higher herb densities). However, such little variation in diet across sites is surprising, given differences in resource availability, such as a twofold difference in herb density. This may partly reflect the fact that researchers have thus far sampled primarily a somewhat restricted range of habitat types (but see Sabater Pi [1977] and Calvert [1985]). Further sampling of gorilla diet from areas of much higher (as at Lossi) or lower (as at Petit Loango) herb density may provide more evidence of greater cross-site variation in gorilla diet. 114 / Doran et al. Fecal Samples and Trail Signs: Pros and Cons The findings of no sex difference and little cross-site variation in diet may reflect the limitations of indirect sampling methods. Fecal samples and trail sign data can greatly inform our understanding of diet composition and the relative importance of different components during the year, as indicated in this and previous studies [Williamson, 1989; Tutin et al., 1991; Nishihara, 1995; Remis, 1997; Goldsmith, 1999]. However, both methods can potentially bias the results. Fecal samples provided the most effective means of assessing diet during the early stage of our study, because: 1) they are easy to collect (relative to following a trail for an entire day), 2) they represent a diet ‘‘snapshot’’ for one individual (vs. the entire group), and 3) results are not dependent upon length of trail followed and are less dependent upon monthly sample size (compared to trail signs). Fecal samples provided a good assessment of the diversity and frequency of fruit consumption, because seeds of most fruits were swallowed whole, passed through the gut intact, and were identifiable to species, as reviewed previously [Tutin & Fernandez, 1993]. However, quantitative measures of fruit consumption were, at best, poor approximations, as noted previously [Tutin & Fernandez, 1985]. Although relative changes in fiber content could be monitored, we could not identify species of leaf, pith, or bark macroscopically from fecal samples, and therefore failed to detect the variety of leaf, stem, and bark species in the diet. This apparently was less of a problem in previous studies [e.g., Tutin et al., 1991; Remis, 1997; Goldsmith, 1996]. Feeding-trail data added important additional information on the diversity and frequency of herb and leaf consumption to fecal samples: we identified 33 leaf, 14 stem, and eight bark species that were undetected from fecal samples (Table III). Fewer studies have conducted extensive trail sampling (but see Goldsmith [1996]), because it is difficult to track gorillas (the average daily path length of western gorillas exceeds 2 km/day at all sites). Additionally, sample size appears to influence results to a greater degree than in fecal sampling. Trail sample size, unlike fecal sample size, was correlated with the number of species detected in this study as well as previously [Goldsmith, 1996]. When shorter trails are followed, foods occurring in highest density, such as herbs, tend to be oversampled. Because trail signs cannot be ascribed to particular individuals, results tend to overestimate individual daily dietary breadth. This was true in our study, as we detected more fruit species (5.4 vs. 3.8) on a daily basis from trail signs vs. fecal samples. However, this difference averaged out, since we found no difference between the monthly number of fruit species detected by either method. The degree to which oversampling individuals or undersampling group diet occurs should vary throughout the year, and lead to seasonal biases in data. For example, when group cohesion is high, trails of a greater number of individuals are encountered and thus more food items are detected, relative to when interindividual spacing is greater. All trail sign feeding remains can be quantified to estimate quantity of leaf and herb (but not fruit) in the diet [Goldsmith, 1996], but only if habituation is not a priority, since it is timeconsuming. However, these results should be interpreted with caution because it is nearly impossible to know how many individuals the trail represents. For these reasons, and to minimize sampling effort, we chose to sample only the presence or absence of each food on trails each day, and to collect fecal samples from only two individuals (male and female) per day vs. the entire group (as was done in previous studies). In spite of these differences in sampling Lowland Gorilla Diet at Mondika / 115 protocol, the results are nearly identical across sites. This leads us to conclude that a basic knowledge of diet can be obtained with smaller sampling effort on a daily basis than is currently used, and that more extensive sampling does not provide results that are more accurate or finer in detail enough to warrant the additional effort. We conclude that indirect sampling of diet provides a broad estimate of diet, but is not sufficiently informative to capture the range of interindividual and intersite variation in diet predicted. ACKNOWLEDGMENTS We gratefully acknowledge the Ministries of Eaux et Forets and Recherches Scientifiques in Central African Republic, and the Ministry of l’Enseignement Primaire, Secondaire et Superieur Charge de la Recherche Scientifique in Republic of Congo for permission to carry out research at the Mondika Research Center. We thank the late M. Urbain Ngatoua, former National Director of the Dzanga-Ndoki National Park; Allard Blom and Lisa Steel of WWF, and Bryan Curran and Fiona Maisels of WCS for continued support and logistical assistance; David Harris for assistance in identifying botanical specimens; Terry Brncic for collecting much of the tree transect data; and Richard Malenky for many ‘‘fruitful’’ discussions on ape foraging behavior. The manuscript was greatly improved as a result of comments from Richard Malenky, Sharon Pochron, and four anonymous reviewers. Finally, the work at Mondika would not have been possible without the efforts of many people in the field, including Cleve Hicks, Shannon Crowley, and Patrice Mongo, and the skilled tracking and botanical lessons of many BaAka field colleagues, especially Mangombe, Mokonjo, Ndima, Mamandele, Bakombo, and Mbokola. REFERENCES Calvert JJ. 1985. Food selection by western gorillas (G.g. gorilla) in relation to food chemistry. Oecologia 65:236–246. Doran DM, McNeilage A. 1998. Gorilla ecology and behavior. Evol Anthropol 6:120–131. Doran DM, McNeilage A. 2001. Subspecific variation in gorilla behavior: the influence of ecological and social factors. In: Robbins MM, Sicotte P, Stewart KJ, editors. Mountain gorillas: three decades of research at Karisoke. Cambridge: Cambridge University Press. p 123–149. Doran DM, Greer D. 2002. The influence of swamp use and fruit consumption on western lowland gorilla (Gorilla gorilla gorilla) ranging behavior at Mondika Research Center. Am J Phys Anthropol 34: 64–65. Galdikas B, Teleki G. 1981. Variations in subsistence activities of female and male pongids: new perspectives on the origins of hominid labor division. Curr Anthropol 22:241–256. Goldsmith ML. 1996. Ecological influences on the ranging and grouping behavior of western lowland gorillas at Bai Hohou, Central African Republic. Ph.D. thesis, SUNY at Stony Brook, Stony Brook, New York. Goldsmith ML. 1999. Ecological constraints on the foraging effort of western gorillas (Gorilla gorilla gorilla) at Bai Hokou, Central African Republic. Int J Primatol 20:1–23. Grieg-Smith P. 1983. Quantitative plant ecology. Berkeley: University of California Press. 399 p. Kuroda S, Nishihara T, Suzuki S, Oko RA. 1996. Sympatric chimpanzees and gorillas in the Ndoki Forest, Congo. In: McGrew WC, Marchant LF, Nishida T, editors. Great ape societies. Cambridge: Cambridge University Press. p 71–81. Lambert JE. 1999. Primate digestion: interactions among anatomy, physiology, and feeding ecology. Evol Anthropol 7:8–20. Magliocca F, Querouil S, Gautier-Hion A. 1999. Population structure and group composition of western lowland gorillas in north-western Republic of Congo. Am J Primatol 48:1–14. Malenky RK, Kuroda S, Vineberg E, Wrangham RW. 1994. The significance of terrestrial herbaceous foods for bonobos, 116 / Doran et al. chimpanzees and gorillas. In: Wrangham RW, McGrew WC, deWaal FBM, Heltne P, editors. Chimpanzee cultures. Cambridge, MA: Harvard University Press. p 93–108. McGrew WC. 1992. Chimpanzee material culture. Cambridge: Cambridge University Press. Mehlman PT, Doran DM. 2002. When do western lowland gorillas (Gorilla gorilla gorilla) build nests? Factors influencing gorilla nest construction at Mondika Research Center. Int J Primatol 23:1257–1285. Nishihara T. 1995. Feeding ecology of western lowland gorillas in the Nouabale-Ndoki National Park, Congo. Primates 36:151– 168. Parnell RJ. 2002. Group size and structure in western lowland gorillas (Gorilla gorilla gorilla) at Mbeli Bai, Republic of Congo. Am J Primatol 56:193–206 Remis MJ. 1997. Western lowland gorillas (Gorilla gorilla gorilla) as seasonal frugivores: use of variable resources. Am J Primatol 43:87–109. Rogers ME, Williamson EA, Tutin CEG, Fernandez M. 1988. Effects of the dry season on gorilla diet in Gabon. Primate Rep 22:25–33. Sabater Pi J. 1977. Contribution to the study of alimentation of lowland gorillas in the natural state, in Rio Muni, Republic of Equatorial Guinea (West Africa). Primates 18:183–204. Smith R, Jungers WL. 1997. Body mass in comparative primatology. J Hum Evol 32:523–559. Tutin CEG, Fernandez M. 1985. Foods consumed by sympatric populations of Gorilla g. gorilla and Pan t. troglodytes in Gabon: some preliminary data. Int J Primatol 6:27–43. Tutin CEG, Fernandez M, Rogers ME, Williamson EA, McGrew WC. 1991. Foraging profiles of sympatric lowland gorillas and chimpanzees in the Lope Reserve, Gabon. Phil Trans R Soc Lond B 334:179–186. Tutin CEG, Fernandez M. 1993. Composition of the diet of chimpanzees and comparisons with that of sympatric lowland gorillas in the Lope Reserve, Gabon. Am J Primatol 30:195–211. Tutin CEG. 1996. Ranging and social structure of lowland gorillas in the Lope Reserve, Gabon. In: McGrew WC, LF Marchant LF, Nishida T, editors. Great ape societies. Cambridge: Cambridge University Press. p 58–70. Watts DP. 1996. Comparative socio-ecology of gorillas. In: McGrew WC, LF Marchant LF, Nishida T, editors. Great ape societies. Cambridge: Cambridge University Press. p 16–28. White LJT, Rogers ME, Tutin CEG, Williamson E, Fernandez M. 1995. Herbaceous vegetation in different forest types in the Lope Forest Reserve, Gabon: implications for keystone food availability. Afr J Ecol 33:124–141. Williamson EA. 1989. Behavioural ecology of western lowland gorillas in Gabon. Ph.D. dissertation, University of Stirling, Stirling, UK.