INFLUENCE OF FRUIT AVAILABILITY ON MACRONUTRIENT AND ENERGY INTAKE BY FEMALE CHIMPANZEES

Abstract

Daily energy intake of adult female mammals is influenced by environmental conditions and physiological requirements, including reproduction. We examined the effects of fruit availability on macronutrient and metabolisable energy intake by adult female chimpanzees (Pan troglodytes schweinfurthii) of the Kanyawara community in Kibale National Park, Uganda from January 2014 through June 2015. Drupe fruits were abundant for four months, whereas the other fourteen months were dominated by fig fruits. The mean daily intake of food (dry matter) and metabolisable energy, did not differ between drupe-months and fig-months. However, foraging costs were higher during fig-months, as indicated by a 20% increase in feeding time. Furthermore, during drupe-months female chimpanzees ingested more water-soluble carbohydrates and lipids, and less available protein and neutral detergent fibre. Although metabolisable energy intake did not differ consistently between drupe-months and fig-months, they consumed more on days when ripe fruit dominated the diet than when leaves and pithy stems dominated the diet. Our data suggest that differences in diet quality between drupes and figs can have important effects on frugivore foraging, and that they influence net energy gain more by their effects on macronutrient composition or foraging cost than by their direct impact on energy intake.

INTRODUCTION

As frugivorous primates, chimpanzee (Pan troglodytes) foraging success is constrained by fruit availability (Newton-Fisher, 1999; Tutin et al., 1991). Evidence of seasonal fluctuations in fruit availability affecting diet and nutrient composition has been documented for many ape species, including chimpanzees, orangutans (Pongo pygmaeus), western gorillas (Gorilla gorilla), western chimpanzees (Pan troglodytes verus), and mountain gorillas (Gorilla beringei) (Conklin-Brittain, Knott, & Wrangham, 2006; Knott, 1998; Masi et al., 2015; N’guessan, Ortmann, & Boesch, 2009; Rothman, Dierenfeld, Hintz, & Pell, 2008). During months of highest fruit production, orangutans increased their intake of calories by feeding on a diet dominated by fruit (Knott, 1998). Furthermore, during fruit-scarce seasons, frugivorous primates can adjust their behavior by feeding in smaller groups (reducing inter-individual competition), by increasing or reducing their foraging time, travel time or diet breadth, and by incorporating lower-quality foods (Doran, 1997; Felton, Felton, Wood, & Lindenmayer, 2008; Harrison, Morrogh-Bernard, & Chivers, 2010; Irwin, Raharison, Raubenheimer, Chapman, & Rothman, 2014; Lambert & Rothman, 2015; Masi, Cipolletta, & Robbins, 2009; Tutin, Ham, White, & Harrison, 1997; Wright et al., 2015).

Two distinct periods of fruit production, drupe-fruit (or non-fig) and fig-fruit seasons (hereafter referred to as drupe-months and fig-months respectively), have been documented at Kanyawara, Kibale National Park, Uganda (Emery Thompson & Wrangham, 2008; Gilby & Wrangham, 2007; Wrangham et al., 1996). Previous studies show that Kanyawara chimpanzees (Pan troglodytes schweinfurthii) exhibit a range of behavioral and physiological modifications during periods of drupe-fruit scarcity (fig-months) (Conklin-Brittain, et al., 2006; Emery Thompson & Wrangham, 2008; Gilby & Wrangham, 2007; Wrangham et al., 1996). During fig-months, Kanyawara chimpanzees feed in smaller groups and increase their dietary breadth (Wrangham et al., 1996). Another behavioral response associated with drupe-months is increased hunting rates (Gilby & Wrangham, 2007). Furthermore, availability of drupes has been associated with positive energy balance among male chimpanzees in Kanyawara (Emery Thompson, Muller, Wrangham, Lwanga, & Potts, 2009).

Figs (Ficus spp.) have a high fibre content and are lower-quality fruits compared to drupaceous fruits, which tend to be richer in soluble sugars (Conklin-Brittain & Wrangham, 1994; Wrangham et al., 1993). Despite this lower quality, figs are major fallback foods because they are abundant and available year-round, allowing the proportion of figs in the diet to increase when drupes are scarce (Marshall & Wrangham, 2007; Wrangham et al., 1991; 1996). Hence, it has been reported that Kanyawara chimpanzees, like gibbons (Hylobates albibarbis), spider monkeys (Ateles chamek) and orangutans, prefer drupes when they are available (Clink, Dillis, Feilen, Beaudrot, & Marshall, 2017; Felton, Felton, Wood, & Lindenmayer, 2008; Leighton, 1993; Wrangham et al., 1996). Previous studies of Kanyawara chimpanzees revealed that during drupe-months, there was increased intake of non-structural carbohydrates, and reduced intake of fiber ( Conklin-Brittain, Wrangham, & Hunt, 1998; Wrangham, Conklin-Brittain, & Hunt, 1998). By contrast, during fig-months there was an increase in dietary breadth, foraging time and a reduction in group size, suggesting that chimpanzees may experience seasonal nutritional shortfalls when drupe fruits are scarce (Wrangham et al., 1996).

Despite the availability of fig fruits throughout the year, it appears that seasonal variation in drupe fruit availability significantly affects reproductive success of female chimpanzees. At Kanyawara, increased consumption of drupe-fruits has been correlated with high conception rates and elevated ovarian hormone production as well as lower cortisol levels in lactating females (Emery Thompson & Wrangham, 2008; Emery Thompson, Kahlenberg, Gilby, & Wrangham, 2007; Emery Thompson, Muller, Kahlenberg, & Wrangham, 2010). Here we examine how the foraging behavior of female chimpanzees varies in relation to drupe fruit abundance, to evaluate the proposition that their reproduction is limited by the availability of drupe fruits.

Optimal foraging theory argues that chimpanzees should maximise the long-term rate of energy intake by choosing foods rich in digestible energy (Lambert & Rothman, 2015; Pyke, Pulliam, & Charnov, 1977). We predict that during the drupe-months, the diet should comprise mostly drupe fruits and female chimpanzees should spend less time feeding since drupe fruits are rich in non-structural carbohydrates. To test these predictions, we: (1) determine the feeding rates, energy intake rates and macronutrient content of the foods eaten by female chimpanzees; and (2) contrast differences in feeding behavior and diet between months when fruit availability and diets were dominated by either drupe fruits or fig-fruits. Comparisons include the daily time spent feeding and total amount of food eaten, as well as intake rates of ripe fruit, terrestrial herbaceous vegetation i.e. pith and leaves, figs, drupes, metabolizable energy and macronutrients.

METHODS

Study site and subjects

Female chimpanzees of the Kanyawara community, in Kibale National Park (KNP) were observed by MU from January 2014 to June 2015. KNP (776 km²) is located in southwestern Uganda (0°13’ to 0°41’N and 30°19’ to 30°32’E) near the foothills of the Rwenzori Mountains (Chapman, Chapman, Kaufman, & Zanne, 1999). The Kanyawara chimpanzee community ranges in the northwestern sector of KNP and has been studied since 1983 (Isabirye-Basuta, 1987; Wrangham et al., 1991). At the time of this study, the Kanyawara chimpanzee community comprised 47 to 51 chimpanzees of which 17 were adult females. Fourteen multiparous female chimpanzees who were fully habituated were selected as study subjects. Estimated ages at the start of the study were between 13.3 and 54.9 years.

Collection of behavioral and feeding data

Behavioral data were collected using continuous focal observations (Altmann, 1974). Ideally, each focal observation (or “follow”) started at dawn when the focal individual left her nest and ended at dusk when she made her sleeping nest. However, sometimes the focal individual had already left the nest before dawn, or we lost the focal individual before she entered her night nest at dusk. Due to the large home ranges and variable sub-group composition of chimpanzees, it was difficult to predetermine the location of a focal individual, hence focal females were located and chosen opportunistically. As far as possible, focal females were selected based on a rotating scheme to ensure that all animals were sampled for a similar amount of time. In each month, focal observations were conducted for a mean of 14±2 days, with each individual sampled as a focal for at least one day per month. On average, each individual was subjected to 15±1 days of focal observations. Additionally, for each focal sample, the foraging group size (the number of individual chimpanzees in the same feeding patch with the focal individual) was recorded.

During each focal sample, we collected the following data for all feeding bouts (a feeding bout is defined as a continuous feeding session on a particular food item that is not interrupted by any non-feeding activity for more than 5 min): the start and end time of feeding on a particular food item to the nearest minute, the plant species, part ingested and its developmental stage (e.g. ripe or unripe fruit, young or mature leaf). When possible, we recorded the number of discrete food units (individual items e.g. one fruit, a strip of leaves or bunch of flowers) consumed per minute (Rothman et al., 2008). The size of discrete food units was determined according to how chimpanzees processed the food item before ingesting it. For example, if the female chimpanzees picked one strip of young leaves at a time, the number of young leaves on each leaflet was estimated and used as the unit. For pith and bark, the length of the pith or bark picked by the individual chimpanzee was estimated. For instance, if on average, a chimpanzee spent 2 min feeding on 50 cm pith of Aframomum angustifolium K.Schum, the discrete food unit was estimated as 25 cm (50/2) per min. To determine the average number of discrete food units ingested per food item, records were noted at five-minute intervals throughout the feeding bout, or when observation conditions allowed. A total of 4314 records of discrete units ingested per minute were taken totaling 648 hrs of observations. A total of 1699 feeding bouts were recorded ranging from 5 feeding bouts on mature leaves to 619 feeding bouts on ripe fruit, which was the common food type. Feeding was defined as reaching for, picking, handling, or chewing a food item as well as short intervals (<5 sec) of searching (defined as searching for the next food item).

Food sample collection and processing

Samples of all the foods eaten during focal follows were collected in air tight bags on the same day or within the same week from the exact plant eaten by chimpanzees or from adjacent plants of the same species. Food samples were processed in a way that mimicked chimpanzee actions. For example, if the chimpanzee peeled the fruit and spat out the seed before eating the fruit, seeds were removed from the collected sample before weighing it. The samples were weighed using a portable balance to yield the “wet unit weight”. At least 30 wet unit weights for each food item were taken at the end of the day (or within 24 hrs) before the same sample was dried to constant weight at ~40°C using a Nesco plant dehydrator (Rothman, Chapman, & Van Soest, 2012). This temperature was chosen to inhibit enzymatic activity and prevent chemical and physical changes, while preserving the sample’s nutritional attributes (Rothman et al., 2012). Dried samples were then weighed to obtain the dry unit weight.

The dried samples were ground using a Wiley mill fitted with a 1 mm screen to obtain uniform particle size. Ground samples were analysed in the nutritional ecology laboratory at Harvard University, Anthropology Department and/or the nutritional ecology laboratory at Hunter College of the City University of New York.

Measuring fruit availability

To describe annual patterns of food availability, we collected phenological data monthly for 18 months (January 2014-June 2015) from 207 trees representing 19 tree species. These trees were located along permanently marked phenology transects (each 200 × 10 m = 0.2 ha) (Wrangham et al., 1998). We estimated each phytophase using visual counts, i.e. young leaves, mature leaves, flowers, ripe fruits, unripe fruits on a 5-point scale (0=0%, 1=1–25% of the crown are, 2=25–50% of the crown area, 3=50–75% of the crown area, 4=>75–100% of the crown area) (Chapman et al., 1992). We recorded phytophases for all trees > 10 cm DBH. Estimates of fruit abundance were based on the number of trees in the phenology sample with non-zero scores (Wrangham et al., 1998). The proportion of trees in fruit for each species per month was determined as the number of trees of that species that had ripe fruit divided by the total number of trees of the species observed with ripe fruit for each month. We calculated the monthly fruit abundance index (FAI_m) as:

$${\text{FAI}}_{\text{m}}=\sum {\text{P}}_{\text{km}}$$

where P_km denotes the proportion of the number of trees in fruit for species k in month m.

Some of the fruit tree species that were included in the phenology sample were not eaten by chimpanzees during the study period, accordingly we calculated FAI using only tree species whose ripe fruits were eaten at least once during the study period (Wrangham et al., 1998).

For each month we identified the dominant fruiting tree species based on which species in the phenology sample had the largest FAI.

Laboratory analyses

Multiple samples of the same foods were pooled to account for intraspecific variation and were analysed in duplicate. We used near-infrared reflectance spectroscopy (NIRS) using a Foss XDS Rapid Content Analyzer to estimate concentrations of total nitrogen; and crude protein (CP) was calculated by multiplying total nitrogen by 6.25 (Rothman, Chapman, Hansen, Cherney, & Pell, 2009). Available protein (AP) (not bound by lignin in the plant cell wall) was estimated by subtracting acid detergent insoluble crude protein (ADCP) from CP. We used ADCP values for Kibale foods estimated by Conklin-Brittain, Dierenfeld, Wrangham, Norconk, & Silver (1999). For the rest of the macronutrients, we used traditional standardized methods. Lipid content was measured using petroleum ether extraction for 4 days at room temperature (Conklin-Brittain & Wrangham, 1994). Water-soluble carbohydrates (WSC) were estimated using a modified phenol/sulfuric acid colorimetric assay with a sucrose standard (Strickland & Parsons, 1972). Neutral detergent fiber (NDF with α-amylase), acid detergent fiber (ADF), and acid detergent lignin (ADL) were measured through sequential analysis using an A200 fiber analyzer (ANKOM, Macedon, NY) (Van Soest, Robertson, & Lewis, 1991). Hemicellulose (HC) was estimated as the difference between NDF and ADF. Cellulose (Cs) was estimated as the difference between ADF and lignin. We estimated total nonstructural carbohydrates (TNC) by subtracting NDF, lipid, AP, and ash from 100%.

We adopted the method used by Conklin-Brittain et al. (2006) to determine dry matter, total ash, and organic matter. Nutrient values were expressed as a percentage of organic matter (OM) instead of dry matter (DM).

A dry-matter correction coefficient (g 105°C DM/g field dried sample) was calculated by drying the field-dried samples in an oven at 105ºC for 8 h and weighing them hot. Total ash (g ash/g field-dried sample) was measured by ashing (burning) the same DM sub-sample at 502°C for 8h and then weighing it at 100°C. Organic matter correction coefficient of the sample was calculated as:

OM correction coefficient (g OM/g field-dried sample) = (1 - total ash) x DM correction coefficient.

These DM and OM correction coefficients were applied to the field-dried DM value to determine the grams of OM in the fresh food:

g OM/ g fresh food = field DM coefficient x final DM correction coefficient x OM correction coefficient.

Food and nutrient intake calculations

First, we computed the wet weight ingested per minute for each food item as the product of its wet unit weight and the corresponding average number of discrete food units eaten per minute.

Feeding rate defined as the dry weight ingested per minute (e.g. Rothman et al., 2008; Schülke, Chalise, & Koenig, 2006) was calculated as:

Feeding rate for each food item= wet weight ingested per minute * (dry unit weight÷wet unit weight *100).

Daily amount of dry food ingested (g/day) was calculated as: dry mass = ∑ (t_i *ʎ_i); where t_i=time spent feeding on food item i per day (mins); ʎ_i= average feeding rate for item i (g/min).

Daily nutrient intake (g/day) was calculated as the product of daily amount of dry food ingested per food item and its corresponding mean nutrient content.

The metabolisable energy (ME) content per food item was computed using the equation below, following Conklin-Brittain, et al. (2006):

$$ME \left(kcal per 100g OM\right)=\left(4*\%TNC\right)+\left(4*\%AP\right)+\left(9*\%lipids\right)+\left(1.6*\%NDF\right)$$

Energy intake rate per food item per minute: average feeding rate (g/ min) * energy content (Kcal/g).

Daily energy intake (Kcal/day) was calculated as: $\sum \left({\text{g}}_{\text{i}}\text{*}\hspace{0.17em}\hspace{0.17em}{\text{e}}_{\text{i}}\right)$; where g_i=amount of dry food of food item i ingested per day (grams); e_i= the energy content of food item i (Kcal/g).

For each day, we calculated the percent grams of fig ripe fruits, drupe ripe fruits or terrestrial herbaceous vegetation (THV: young leaves, pith or flowers) consumed. From this, we determined the dominant food type for each day as the food type with the largest percent either as figs (fig-days), drupes (drupe-days) or THV (THV-days).

Statistical analysis

To examine the differences among food types in feeding and energy intake rates, we used Kruskal-Wallis (H) tests and Dunn’s post-hoc tests for pairwise comparisons (α = 0.05). These tests were chosen because of the small sample sizes for some of the food types and the fact that our dependent variables did not conform to normality and homoscedasticity assumptions. To test the differences in macronutrient content among the different food types, we used one-way ANOVA with Welch’s test because our data violated the assumption for equal variances. Games-Howell’s post-hoc tests were used for pairwise comparisons. Linear mixed effect models were used to analyse if: 1) daily time spent feeding, grams of food ingested, macronutrient and energy intake (response variables) could be predicted by the type of fruit-month (fixed factor), and 2) daily grams of food ingested, macronutrient and energy intake (response variables) could be predicted by the dominant food type (fixed factor). A further predictor variable included in the model was foraging group size, since competition is associated with larger groups and can affect food intake (Janson, 1985; Janson, 1988). All models included the identity of the individuals as a random factor. The models were fitted by REML (Restricted Maximum Likelihood criterion) using the lmer package. The analyses were run in in SPSS version 20.0 software and R version 3.5.1.

RESULTS

For analysis we included only focal follows with a minimum of 10 hrs of observation because the average total time spent observing the female chimpanzees daily was 11 hrs (SEM=0.6 hrs, range 9.6 to 12.5 hrs). Hence from a total of 210 focal follows, only 141 complete focal follows representing 1597 observation hours from January 2014 to June 2015 were used in the analysis. There were 93 complete focal observations during fig-months and 48 during drupe-months. Focal females fed on 90 different plant parts from 61 plant species.

Diet quality based on intake rates and nutrient content

Feeding rate (dry weight ingested per min) and energy intake (Kcal per min) varied across food types (Table 1). Both feeding rate and energy intake rate were fastest for ripe fruits and slowest for woody foods such as pith and wood. Feeding rate on ripe fruits was faster than that on pith (Dunn’s post-hoc test: p=0.03), and energy intake rate of ripe fruits was faster than that of young leaves (p=0.04) and pith (p=0.005).

Table 1:

Feeding rates and energy intake rates by food type

Food type	Feeding rate (dry g/min)	Range for feeding rate	Energy intake rate (Kcal/min)	Range for energy intake rate	N
Ripe fruit
(drupe + fig)	3.4±0.4	0.4–10.8	10.7±1.3	1.5–36.3	36
Drupe fruits	3.0±0.6	4–10.8	9.9±1.8	1.5–36.3	24
Fig fruits	4.2±0.5	1.6–7.6	12.5±1.5	4.1–22.8	12
Young leaves	2.1±0.2	0.7–4.2	6.2±0.6	1.9–11.1	19
Pith	1.8±0.2	1.0–3.1	3.4±2.2	2.3–9.1	13
Unripe fruit	3.0±0.8	1.0–5.9	9.0±3.4	2.5–16.5	7
Flower	2.7±0.3	1.8–3.5	7.7±1.0	4.3–10.1	7
Seed	2.3±1.9	0.3–8.1	6.9±5.7	0.9–23.8	4
Bark	2.9±2.1	0.8–5.1	5.7±4.3	1.5–11.1	2
Wood	0.3±0.1	0.1–0.3	0.6±0.1	0.5–0.9	3

Average intake rates for adult female chimpanzees of the Kanyawara community. Cells show mean ± standard error. Food types and fruit types are arranged from the most frequently consumed through the least consumed plant parts. Differences among the 8 major plant parts analysed were significant in each of the variables measured: Kruskal wallis test (H); feeding rate: H=17.0, df=7, p=0.02; energy intake rate: H=18.2, df=7, p=0.01. N = number of species.

Feeding rate on figs was faster than on drupes (H=5.05, df=1, p=0.03; Table 1). By contrast, the difference between energy intake rate for figs and drupe was not quite significant (H=3.7, df=1, p=0.06) (Table 1).

Macronutrient composition varied considerably among food types (Table 2). Lipid content was greater in seeds than in young leaves, pith, bark and wood. Available protein was greater in young leaves, seeds and flowers than in ripe fruit, pith, unripe fruit and bark. WSC content was greater in ripe fruit than in all foods with the exception of pith and seeds. TNC content was greater in ripe fruit than in young leaves, seed, bark and wood. Fibre content also varied among food types. NDF, ADF, hemicellulose and cellulose contents were greater in pith and woody food types than in ripe fruits.

Table 2:

Macronutrient concentrations (% OM) of foods eaten by female chimpanzees.

Food type	%Lipid	%AP	%NDF	%TNC	%WSC	%ADF	%Ls	%HC	%Cs	N
RF	4.6±1.3	9.3±1.1	41.5±2.4	44.6±2.7	16.5±1.7	28.5±2.3	13.0±1.5	12.9±0.8	15.5±1.2	36
Fig	3.5±0.7	6.8±1.9	49.0±3.8	40.7±3.6	9.2±1.2	38.0±3.7	17.3±2.7	11.0±1.1	20.6±1.4	12
Non-fig	5.2±1.9	10.6±1.3	37.7±2.9	46.6±3.7	20.1±1.2	23.8±2.5	10.9±1.7	13.9±1.1	12.9±1.4	24
YL	0.5±0.2	25.9±1.9	43.1±2.8	30.5±3.1	5.1±0.5	27.7±2.5	10.4±1.5	15.4±1.6	17.2±1.4	19
PT	0.4±0.2	7.5±2.0	58.1±3.1	33.9±2.9	15.5±2.7	32.8±1.8	5.1±0.5	25.3±1.7	27.7±1.6	13
UF	2.9±1.0	7.1±2.3	51.1 ±4.0	38.8±3.0	8.9±1.8	37.1±6.0	17.8±4.6	13.9±2.9	19.3±2.0	7
FL	2.8±1.3	28.4±3.8	55.5±4.1	13.2±4.0	4.5±0.4	37.5±4.0	23.3±3.0	18.0±1.1	14.1±1.1	7
SD	5.3±1.4	30.4±7.5	48.4±4.8	15.8±4.6	11.2±3.1	21.8±2.3	9.3±1.2	26.6±3.5	12.5±1.5	4
BK	0.0±0.0	13.8±5.1	87.0±3.3	1.7±0.7	1.9±0.5	45.8±4.2	7.1±2.5	41.1±0.9	38.7±6.7	2
WD	0.0±0.0	12.0±11.9	71.9±19.5	16.0±7.6	1.9±0.9	40.5±0.0	19.1±0.0	31.4±0.9	21.4±0.1	3

Data source: same as for Table 1. Cells show mean ± standard error.

RF, ripe fruit; YL, young leaves; PT, pith; UF, unripe fruit; FL, flower; SD, seed; BK, bark, WD, wood.

AP, Available protein; WSC, Water soluble carbohydrates; NDF, Neutral detergent fiber; TNC, Total non-structural carbohydrates; Ls, Lignin; ADF, Acid detergent fiber; HC, Hemicellulose; Cs, Cellulose.

Statistical test results for macronutrient composition of different food types: (Games-Howell’s post-hoc tests: lipid: F_6,83=2.1, p=0.06; AP F_6,83=16.3, p=0.001; WSC: F_6,83=7.5, p=0.001; TNC: F_7,83=10.3, p=0.0001; NDF: F_6,83=7.2, p=0.0001; ADF: F_6,83=6.3, p=0.0001; HC: F_6,83=3.8, p=0.002; cellulose: F_6,83=22.3, p=0.001; lignin: F_6,83=4.9, p=0.001). Fruit types, Kruskal-Wallis test: WSC: H=8.5, df=1, p=0.003; NDF: H=5.05, df=1, p=0.03; ADF: H=7.4, df=1, p=0.007; and cellulose content: H=11.0, df=1, p=0.001. Macronutrients where there was no significant difference between the fruit types: lipids: H=0.4, df=1, p=0.5; AP: H=3.1, p=0.08; TNC: H=0.8, df=1, p=0.4, HC: H=2.4, df=1, 0.1; lignin: H=3.7, df=1, p=0.06.

All values represent percentage of organic matter.

As expected, the concentration of WSC was greater in drupes than in figs, whereas some of the fibre indices such as NDF, ADF and cellulose content were greater in figs than in drupes. For the rest of the macronutrients, there was no significant difference in their composition among the fruit types (Table 2).

Contrasts in behavior and intakes between drupe-months and fig-months

Based on the fruit abundance indices, we identified 4 months as drupe-months and 14 months as fig-months (Table 3). Although the monthly proportion of trees in fruit was consistently lower than the proportion containing young leaves or flowers (Figure 1a), ripe fruit dominated the diet of chimpanzees (62% ripe fruit, 21% young leaves, and 1 % flowers: based on grams ingested) (Figure 1b). Mean foraging group size was larger during drupe-months (13±2 chimpanzees) than in fig-months (10±1 chimpanzees), but the difference was not significant (F_1,139=3.2, p=0.07).

Table 3:

Summary of fruit abundant tree species, feeding observations and group size used in the analysis

Month	Fruit-month	Dominant fruit available	Total number of feeding bouts	No. of food types eaten	Average group size	FAI
January 2014	Drupe	Mimusops bagshawei S.Moore	70	18	19	12.4
February 2014	Fig	Figs	81	20	27	7.3
March 2014	Fig	Figs	87	25	13	9.0
April 2014	Fig	Ficus dawei Hutchinson	138	30	11	7.9
May 2014	Fig	Ficus dawei Hutchinson	55	23	17	7.9
June 2014	Drupe	Aningeria altissima A. Chev and Uvariopsis congensis Robyns & Ghesq	94	30	15	7.3
July 2014	Fig	Ficus natalensis Hochst	137	39	14	8.4
August 2014	Fig	Ficus capensis Thunb	121	28	8	5.1
September 2014	Fig	Ficus natalensis Hochst	126	36	4	6.7
October 2014	Drupe	Celtis gomphopylla Baker	100	29	5	15.7
November 2014	Fig	Figs	86	22	6	12.4
December 2014	Fig	Figs	103	25	6	13.4
January 2015	Fig	Ficus natalensis Hochst	79	24	9	8.9
February 2015	Fig	Ficus brachylepis Welw. ex Hiern	127	25	11	9.0
March 2015	Fig	Figs	77	26	10	14.0
April 2015	Fig	Figs	70	24	4	10.1
May 2015	Fig	Figs	68	24	14	18.0
June 2015	Drupe	Uvariopsis congensis Robyns & Ghesq	80	15	20	24.7

Dominant fruit available was based on the fruit abundance index (FAI) for each month. In some of the months, more than one fruit species was abundant. If they were all Ficus species, then the dominant fruit available was scored as “figs” (FF). “Total number of feeding bouts” represents the number of times a particular food item was continuously fed on for atleast 5 minutes or more before the focal individual changed to feed on a different food item.

Figure 1.

Figure 1: (a) Fluctuation in availability of flowers, ripe fruits and young leaves and (b) variation in the intake of flowers, ripe fruits and young leaves (based on weight) by female chimpanzees between January 2014 and June 2015. Number of trees monitored for phenology=209 trees representing 19 species.

Across all the months, the mean daily time spent feeding, dry mass ingested and energy intake were 308.7±85 mins, 872.6±289 g, and 2479.4±858.1 Kcal respectively. Throughout the study period, figs accounted for the largest proportion of ripe fruit intake (69.2%). Drupe fruits accounted for a larger proportion in drupe-months than in fig-months (F_1,17=18.9, p=0.001), whereas figs accounted for the largest proportion during fig-months (F_1,17=12.5, p=0.003) (Figure 2). The total dry weight of food ingested per day did not vary between drupe-months and fig-months (F_1,139=0.2, p=0.6) (Figure 2). However, compared to fig-months, during drupe-months, a greater percentage of dry-weight intake came from ripe fruits (F_1,139=5.2, p=0.02), a lower percentage came from THV (F_1,139=5.7, p=0.02; Figure 2), and the time spent feeding per day was 20% less (F_1,139=10.0, p=0.002) (Figure 2).

Figure 2.

Variation in the proportion of drupes and figs ingested, time spent feeding and daily food intake and the proportion of ripe fruit and THV consumed between the fruit months. N for drupe-months and fig-months= 4 and 14 months respectively. The diamond symbol indicates the mean; boxes show the lower and upper quartiles, horizontal middle line indicates the median; whiskers denote the minimum and maximum values; black circles denote the outliers

As with food intake, metabolisable energy intake did not differ significantly between drupe-months and fig-months (F_1,139=0.01, p=0.9; Figure 3). Nutrient composition did vary, however, in drupe-months female chimpanzees consumed more WSC and lipid and less NDF (Kcal/day) than in fig-months (WSC: F_1,139=13.2, p=0.0004, lipid: F_1,139=4.5, p=0.04; NDF: F_1,139=4.0, p=0.05). In drupe-months intake of AP (Kcal/day) was also slightly less than in fig-months (F_1,139=3.6, p=0.06). TNC intake did not vary by type of fruit-month (TNC: F_1,139 = 0.9, p=0.3) (Figure 3).

Figure 3.

Variation of average daily metabolizable energy and macronutrient intake across the fruit months. N for drupe and fig-months= 4 months and 14 months respectively. The boxes show the lower and upper quartiles, horizontal middle line indicates the median; whiskers denote the minimum and maximum values; black circles denote the outliers.

In addition to comparing feeding success between months, we also considered variation between days. We found that on drupe-days female chimpanzees ingested more food (dry weight) and metabolizable energy than on THV-days (Table 4). Intake of WSC was greater on drupe-days than on either fig-days or THV-days, whereas available protein intake was greater on THV-days than on drupe-days or fig-days. TNC intake on drupe-days and fig-days was greater than that on THV-days. Intake of NDF was greater on fig-days than on THV-days. Lipid intake did not vary among the days dominated by different food types (Table 4).

Table 4:

Variation in daily food, macronutrient and metabolisable energy intake among days dominated by different food types

Daily intake	Dominant food type			Range	F and p-value
	Non-fig days (22)	Fig-days (76)	THV-days (43)
Food ingested (grams)	930.7±70.9	924.7±32.2	750.8±37.8	409.0–1716.7	F(2,138)=5.9, p=0.004
Lipid (Kcal)	202.8±49.5	200.6±14.4	149.5±12.7	8.1–726.2	F(2,138)=2.1, p=0.13
Available protein (Kcal)	317.7±38.5	340.9±14.5	410.3±31.0	92.2–1420.4	F(2,138)=3.3, p=0.04
WSC (Kcal)	792.3±89.2	400.2±19.3	344.4±31.4	83.2–2033.3	F(2,138)=29.6, p=0.0001
TNC (Kcal)	1547.8±138.3	1325.2±60.9	1073.9±74.3	453.1–3373.8	F(2,138)=6.1, p=0.003
NDF (Kcal)	638.0±46.7	722.9±25.5	534.9±27.8	243.9–1390.2	F(2,138)=11.0, p=0.0001
Metabolisable energy (Kcal)	2706.2±221.4	2589.5±94.5	2168.6±113.7	1240.3–4931.0	F(2,138)=4.4, p=0.01

Cells show mean ± standard error. The number in parenthesis is the number of days when that food type dominated the diet.

DISCUSSION

The diet of adult female chimpanzees in Kanyawara was dominated by ripe fruits whether calculated in terms of contributing species, time spent feeding or amount of food ingested. Nevertheless as in previous studies, female chimpanzees also fed on a diverse array of food types from different plant species including young leaves, pith and unripe fruit (Basabose, 2002; Newton-Fisher, 1999; Watts, Potts, Lwanga, & Mitani, 2012).

Ripe fruits are high-quality foods with high soluble sugar content and low fibre content compared to other diet items that were eaten by female chimpanzees (Conklin-Brittain et al., 1998; Wrangham et al., 1998). While ripe fruits are an important source of calories in the form of digestible carbohydrates, they have lower levels of protein and minerals. Therefore, despite being ripe fruit specialists, chimpanzees may have to complement and balance their fruit-based diet with other foods such as young leaves which are rich in protein. Furthermore, we noted that feeding rates of fibrous foods such as pith were generally slow compared to other food types. This is because high-fibre foods tend to require longer mastication, as has been demonstrated in mammalian herbivores such as deer and some primates (Oftedal, 1991; Spalinger, Hanley, & Robbins, 1988).

Our findings are commensurate with previous studies of Kanyawara chimpanzees, in that energy intake rates of figs and drupes were comparable, and figs contained less soluble carbohydrates and more fibre than drupes (Conklin-Brittain & Wrangham, 1994; Wrangham et al., 1993). Additionally, our findings show that female chimpanzees were likely to ingest more dry grams per minute when feeding on figs than when feeding on drupes, apparently due to the low handling cost for figs given that figs are normally consumed whole, without any prior processing (Leighton, 1993). However, the high water and fibre content of figs leads to lower energy contributions than drupe fruit (Leighton, 1993; Wrangham et al., 1993). Compared to gorillas who have a relatively longer hind-gut, chimpanzees may digest fibre less efficiently (National Research Council, 2003). High-fibre parts of the food were often either discarded by chimpanzees as a wadge or were passed out in faeces.

Compared to drupe-months, in fig-months female chimpanzees increased the proportion of fibrous foods in their diet (i.e. figs and THV), and spent more time feeding. However despite the increase in feeding time, there was no commensurate increase in food intake, presumably because the fibre-rich foods that were eaten in greater amounts were eaten more slowly (Oftedal, 1991). The foraging strategy employed by female chimpanzees during fig-months was thus typical of frugivorous primates during fruit-scarce periods, involving changes both in diet composition and in time spent feeding (Clink et al., 2017; Conklin-Brittain et al., 1998; Doran, 1997; Felton et al., 2008; Irwin et al., 2014; Tutin et al., 1997).

The fluctuations in drupe fruit availability resulted in monthly variations in macronutrient intake. Increased consumption of fig-fruits resulted in reduced intake of WSC and lipids. There was also an increased intake of available protein and NDF, with the increase in available protein coming from the greater intake of THV, whereas the increase in NDF intake came mainly from figs, and not from THV as is sometimes assumed (Conklin-Brittain et al., 1998; Wrangham et al., 1998). These adjustments in behavior and diet to reductions in diet-quality during fig-months allow chimpanzees to compensate for energy shortages in the diet (Chapman, Wrangham, & Chapman, 1995; Conklin-Brittain & Wrangham, 1994). The larger foraging group size during drupe-months may be an indication that larger groups have better quality diets and habitats (Wright et al., 2015).

The average daily calorie intake of female chimpanzees estimated in this study (2500 Kcal) was slightly greater than but comparable to the estimate from a previous study of Kanyawara chimpanzees (2340 Kcal) (Conklin-Brittain et al., 2006). The difference might be due to the fact that the previous study combined feeding rates of both male and female chimpanzees and did not estimate feeding rates for piths, which are some of the important fallback foods for chimpanzees. Furthermore, in the previous study, researchers did not continuously follow one individual per day to estimate their food intake, but instead changed among focal animals every 10 minutes. Finally, the subjects in the present study were nursing mothers that have high caloric needs to meet the energy costs of milk production (Gittleman & Thompson, 1988; Oftedal & Hudson, 1985).

Critically, our data indicate that despite the variation in energy density among available foods, there was no monthly variation in metabolisable energy intake. This finding is commensurate with a previous study of Kanyawara chimpanzees (Conklin-Brittain, et al., 2006), and with results from mountain gorillas, western gorillas and western chimpanzees (Masi et al., 2015; N’guessan et al., 2009; Rothman et al., 2008; Wright et al., 2015).

The lack of significant monthly variation in metabolisable energy intake at Kanyawara may be due partly to the lack of mast fruiting periods, unlike those reported to cause large fluctuations in net caloric gain rates among Ngogo chimpanzees (10 km to the south of Kanyawara) and orangutans (Knott, 1998; Potts, Baken, Ortmann, Watts, & Wrangham, 2015). Additionally, unlike orangutans which feed on a 100% fruit based diet during the months of high fruit production (Knott, 1998), female chimpanzees included lower-quality fall back foods in their diet even when preferred foods were abundant, contrary to our prediction. The fact that female chimpanzees ingested more calories on days when the diets were dominated by ripe fruits (whether figs or drupes) than on days when THV dominated the diets is an indication that if female chimpanzees fed exclusively on ripe fruit when drupe or fig fruits are abundant, there would be significant peaks in energy intake across the fruit-months. However, it seems that female chimpanzees are not interested in maximizing energy intake at the expense of other macronutrients.

Although we found no difference in energy intake between drupe-months and fig-months, we found two other changes that could be important. Firstly, macronutrient composition changed: there was reduced intake of WSC and lipids during fig-months. An indication that chimpanzees just like other primates such as orangutans, red colobus monkeys (Piliocolobus trephosceles), black howler monkeys (Alouatta pigra), and yellow baboons (Papio cynocephalus) are nutritionally stressed when preferred foods are scarce (Behie, Pavelka, & Chapman, 2010; Chapman, Saj, & Snaith, 2007; Gesquiere et al., 2008).

Secondly, time spent feeding was reduced during drupe-months, suggesting that there was a reduction in foraging costs. This would likely result in increased positive energy balance during drupe-months. And since increased intake of digestible carbohydrates seems to fluctuate in parallel with environmental food abundance, increased intake of WSC and lipids could in theory serve as a cue to trigger increased reproductive effort (Emery Thompson, 2013). Our findings thus suggest that variation in energy intake may not be the only reason why reproductive effort is associated with nutritional quality, as indicated in chimpanzees and other primates such as Hanuman langurs (Presbytis entellus), Assamese macaques (Macaca assamensis) (Emery Thompson & Wrangham, 2008; Heesen, Rogahn, Ostner, & Schülke, 2013; Koenig, Borries, Chalise, & Winkler, 1997; Wallis, 1995). Further research is needed to test the idea that macronutrient composition, and foraging costs imposed by increasing feeding time, can affect reproductive effort independently of energy intake.

Conclusion

We found that on days when more ripe fruit were eaten the amount of metabolisable energy ingested by female chimpanzees was greater. In contrast to the comparison among days, however, when we compared among months (drupe-months versus fig-months), females maintained a consistent mean energy intake of around 2500 Kcal per day. The fact that mean energy intake did not vary in relation to monthly changes in the availability of preferred foods requires further investigations of the commonly found positive correlations between reproductive performance and food quality. Variations in macronutrient composition and time spent foraging provide candidate explanations.