International Journal of Primatology

, 30:229

Temporal Variation in Insect-eating by Chimpanzees and Gorillas in Southeast Cameroon: Extension of Niche Differentiation

Authors

    • Centre for Research and ConservationRoyal Zoological Society of Antwerp
    • Department of BiologyUniversity of Antwerp
Article

DOI: 10.1007/s10764-009-9337-2

Cite this article as:
Deblauwe, I. Int J Primatol (2009) 30: 229. doi:10.1007/s10764-009-9337-2

Abstract

I studied insect-foraging strategies of great apes and aimed to define niche differentiation in their insect diet. I investigated seasonality in fruit-, foliage-, insect-, and meat-eating by great apes in southeast Cameroon via indirect methods and measured activity and nest densities of insect prey. I used a multinomial logistic regression to analyze the data. Gorilla and chimpanzee insect-, ant-, and termite-eating does not correlate with rainfall. Ant- and nonwinged termite-eating by chimpanzees increased in periods of succulent fruit scarcity and provided protein and energy, which might have compensated for the protein-low foliage eaten then. The apes ate winged termites when succulent fruit was abundant. Ant and winged termite consumption by gorillas correlates positively with that of chimpanzees. Ant-eating by gorillas increased when fruit was scarce, but was also associated with temporal ant activity and nest density. Both ape species also encountered more ant nests and trails in that period, as they predominantly foraged for herbs in vegetation types with high ant availability. In contrast, fruit-eating correlates positively with nonwinged termite-eating by gorillas, but again temporal prey availability is also associated. Termites might have provided 1) supplemental iron when tannin-rich fruits were eaten or 2) antidiarrheal properties when gorillas ate too much laxative fruit. Termite-eating by both ape species is not associated with spatial termite availability. In conclusion, there is niche differentiation in their insect diet. Based on the trade-off between foraging effort and nutritional gain, chimpanzees use a high-energy and gorillas a low-energy strategy when feeding on termites, but both use a low-energy strategy when feeding on ants. However, more information on the consumption of ant larvae is necessary to define niche differentiation in their ant diet.

Keywords

antsforaging strategiesgreat apesseasonalitytermites

Introduction

The closely related and sympatric gorillas and chimpanzees of African lowland rain forests show extensive dietary overlap (Kuroda et al.1996; Yamagiwa et al.1996). However, when fruit is scarce, western gorillas consume large amounts of pith, leaves, bark, and highly fibrous fruit (Rogers et al.2004) and often decrease their daily path lengths and group spread (Doran-Sheehy et al.2004; Goldsmith 1999), whereas chimpanzees travel farther per day and sometimes decrease party size and continue to eat relatively large amounts of ripe fruit (Hashimoto et al.2003; Kuroda et al.1996). In addition to eating more fruit than most eastern gorillas, western gorillas also feed on insects, especially ants and termites, much more than expected from their large body size (Cipolletta et al.2007; Deblauwe et al.2003; Doran et al.2002; Remis 1997; Tutin and Fernandez 1983, 1992). Central African chimpanzees also regularly eat ants and termites (Sanz et al.2004; Suzuki et al.1995; Tutin and Fernandez 1992). Insects contribute only 0.1–1% of the diet by freshweight, but may have considerably more nutritional importance for both ape species than this implies (Cipolletta et al.2007; Deblauwe and Janssens 2008; Doran et al.2002; Hladik 1977; Tutin and Fernandez 1992). Overlap of important gorilla and chimpanzee insect prey species is low at La Belgique (Deblauwe and Janssens 2008) and at Lopé (Tutin and Fernandez 1992). Comparative data on insectivory by sympatric chimpanzees and gorillas at Lopé showed that the monthly variation in the frequency of insect remains in gorilla feces does not correlate significantly with the monthly frequency of remains in chimpanzee feces (Tutin and Fernandez 1992). Tutin and Fernandez (1992) hypothesized that both species would eat more insects when their diets consisted mostly of fruit, which is usually low in protein (Rogers et al.1992), than when fruit was scarce and protein-rich foliage contributed more to their diets. However, data on gorillas did not support this hypothesis, whereas data on chimpanzees were insufficient to test it. Cipolletta et al. (2007) and Remis (1997) also found no significant relationship between fruit intake and insect-eating in western gorillas at Bai Hokou. However, all of these researchers (Cipolletta et al.2007; Doran et al.2002; Goldsmith 1999; Kuroda et al. 1996; Tutin and Fernandez 1992) found that western gorillas ate more insects during rainy seasons, when fruit was abundant. Central African chimpanzees show little temporal variation in insect-eating (Kuroda et al.1996; McGrew et al.1979; Muroyama 1991; Sanz et al.2004; Suzuki et al.1995), and insect-eating does not correlate significantly with rainfall at Okorobiko and Campo (McGrew et al.1979; Muroyama 1991), and with rainfall and termite activity at Ndoki (Suzuki et al.1995).

Explaining the temporal variation in insect-eating by great apes requires data on local availability of prey (Ganas and Robbins 2004; Tutin and Fernandez 1992). I here incorporate data on temporal and spatial insect prey availability in a Central African great ape diet study and investigate possible explanations for seasonality in insectivory. My study focuses on sympatric chimpanzee and gorilla populations at the northern periphery of the Dja Biosphere Reserve, southeast Cameroon. An earlier nutritional study at the site reported that termites provided, in considerable amounts, protein and manganese to chimpanzees and iron to gorillas, while ants provided similar nutrients to both ape species, but in negligible amounts (Deblauwe and Janssens 2008). Based on findings of Tutin and Fernandez (1992), I expect that the seasonal fluctuation of insect abundance in the gorilla diet does not correlate with that in the chimpanzee diet. Because the protein contribution of termites to the chimpanzee diet at our site was sometimes very high (Deblauwe and Janssens 2008), I suggest 2 nutritional explanations for the possible seasonality in termite-eating by chimpanzees. First, as Tutin and Fernandez (1992) hypothesized, termite-eating and frugivory might positively correlate in chimpanzees. Second, termite-eating might increase when fruit is scarce, because the terrestrial herbaceous vegetation (THV) that makes up most of the diet then has less protein and minerals than THV in other periods of the year (Kuroda et al.1996; Rogers et al.1988; White et al.1995). I do not expect similar nutritional correlations between fruit-eating and ant-eating by chimpanzees and between fruit-eating and ant- and termite-eating by gorillas, because these prey items contribute little protein to the apes’ diet (Deblauwe and Janssens 2008). Instead, temporal variation in chimpanzee ant-eating and gorilla insect-eating might result simply from variation in the temporal or spatial availability of these prey species (McGrew et al.1979; Tutin and Fernandez 1992). In summary, I expect chimpanzees and gorillas to use different strategies when foraging for termites, but similar ones when foraging for ants.

Methods

Study Site

The research site La Belgique (40 km2) is located in southeast Cameroon on the northern periphery of the Dja Biosphere Reserve (013°07′–013°11′E and 03°23′–03°27′N; 671 m altitude). The site is officially unprotected and part of one of the logging concessions of FIPCAM (UFA 10 047). The forest is part of the transition zone of the Atlantic coastal rainforests of southern Nigeria and southwest Cameroon and the evergreen forests of Equatorial Guinea and the Congo Basin (Letouzey 1985). Annual rainfall averages 1570 mm, with peaks in May and September (PAP 2001). I collected data from May to August 2003 and September 2004 to April 2005 with an annual rainfall of 1397 mm (pooled data for all years) and distinguished 3 seasons (Fig. 1): early wet season (EW) from March to August (100–160 mm/mo), late wet season (LW) from August to November (>160 mm/mo), and dry season (D) from November to February (<100 mm/mo). The average temperature is 23.3°C (range: 18–27°C) and remains nearly constant throughout the year (PAP 2001).
https://static-content.springer.com/image/art%3A10.1007%2Fs10764-009-9337-2/MediaObjects/10764_2009_9337_Fig1_HTML.gif
Fig. 1

Monthly rainfall (mm) in 2003, 2004, and 2005 with indication of the seasons at the study site (LW=late wet season; EW=early wet season; D=dry season) and the number of gorilla (G) and chimpanzee (C) fecal samples collected per month.

The forest is a mosaic containing 10 types of vegetation (Nguenang and Dupain 2002): near primary forest (NPF, 14%), very old secondary forest (VOSF, 20%), old secondary forest (OSF, 22%), young secondary forest (YSF, 13%), light gap forest (LGF, 11%), gallery forest (GF, 1%), old logging roads (OLR, 1%), riverine forest (RF, 7%), Raphia-swamp (RS, 11%), and swamp clearing (SC, 0.08%). Secondary forest (VOSF, OSF, YSF) covers >50% of the area. Ten parallel transects (NE-SW direction) of ca. 6 km in length cross the study site. They start on a base transect, also 6 km long (N-S direction), and are 600 m apart.

The density of chimpanzees and gorillas is conservatively estimated at 0.9 and 2.1 individuals/km2, respectively (Guislain and Dupain 2005). Chimpanzees and gorillas are intentionally unhabituated to human observers because of the high hunting pressure in the surrounding area. I identified 4 gorilla groups (1–4) and a number of lone gorillas (5). In contrast, I could not identify the number of chimpanzee communities, owing to the difficulties of tracking chimpanzees between nest sites.

Diet Composition

I used indirect methods to investigate diet composition. While tracking apes, I recorded all feeding traces ≤14 d old and distinguished chimpanzee and gorilla feeding traces via other associated traces (foot/knuckle prints, feces, nests). I investigated traces of insect-eating in detail (Deblauwe and Janssens 2008; Deblauwe et al.2006).

I collected fecal samples (<2 d old) from nests or trails. To avoid collecting samples from the same individual twice in 1 d, I collected only 1 fecal sample from each nest or individual trail. I weighed the samples, washed them in running water using a metal sieve (1-mm mesh), and analyzed them according to Tutin and Fernandez (1993). I counted large seeds (>5 mm), scored small seeds on a 4-point abundance scale [absent (0), rare (1), few (2), common (3), and abundant (4)], categorized nonfruit remains as green leaf fragment (GLF) or fiber (F), and scored GLF and F relative to the total mass of the fecal sample on the same 4-point abundance scale. I also assigned percentages to the quantity of found seeds (%S), fruit skin and pulp (%FSP), foliage (%Fo), and rest (%R; insects, meat, unidentifiable parts, etc.) relative to the total mass of the fecal sample and counted the number of fruit species per fecal sample (FS) and per month (FSM). I considered insects to be present in feces when there were ≥1 parts, heads, or mandibles. Subsequently I counted the heads or mandible pairs via the following abundance scores (adjusted per 200 g of fresh feces): 1) rare: 1–14, 2) few: 15–49, 3) common: 50–100, 4) abundant: >100 (Deblauwe et al.2003). I defined the frequency of insect-eating as the percentage of feces that contained insect parts. Except for the monthly frequency of insect-eating in total, all ant, termite, and species-specific monthly frequencies in feces correlate positively to their mean monthly abundance score in feces (Deblauwe, unpubl. data).

Availability of Important Insect Prey Species

Important insect prey species for gorillas are the ants Oecophylla longinoda, Crematogaster spp., and Tetramorium aculeatum and the termites Cubitermes spp. and Thoracotermes macrothorax (Deblauwe and Janssens 2008). For chimpanzees, the important insect prey species are the ants Dorylus kohli, D. opacus, and Oecophylla longinoda and the termites Macrotermes muelleri, M. nobilis, and M. lilljeborgi/renouxi (Deblauwe and Janssens 2008). Winged termites of Macrotermes spp. are also an important prey for both ape species at the study site.

Each month, I measured insect activity by 1) visiting on average 10 colonies or nests of each prey species, chosen randomly along the transects and 2) using pitfall traps at ground level (Deblauwe and Dekoninck 2007). From each of the epigeal (aboveground) nests of the termites Cubitermes spp. (CuW), Thoracotermes macrothorax (ThW), Macrotermes muelleri, and M. nobilis (MW) I cut a small random part of the nest with a machete. I weighed the nest material (earth), together with its occupant termites, and counted and weighed the termites alone. I calculated the biomass (in g) per 100 g nest material and the average for each species per month. I recorded the number of active and nonactive nests in each colony of Oecophylla longinoda and Tetramorium aculeatum. I considered a nest of Oecophylla active when ≥1 individual was present on the surface of the nest. I considered a nest of Tetramorium active if attacking ants exited on moving the leaf with a stick. I calculated the average percentage of active nests of Oecophylla longinoda (O%A) and Tetramorium aculeatum (Te%A) per colony per month. I also estimated the number of ants present on each nest of Oecophylla, and calculated the average per colony and per month (OI/N). Crematogaster spp. have many different nesting sites (Deblauwe and Janssens 2008). I used only large carton nests (of Crematogaster melanogaster? and C. buchneri) high on tree trunks to measure their activity because I could most reliably relocate the nesting sites the following month. The ants forage from and to the nest along the tree trunk or liana. I counted the ants in a 10-cm2 area low and high (using binoculars) on the tree trunk or liana and calculated the average per nest and per month (CrI). I considered a nest or colony that was empty during 2 consecutive visits to be abandoned and selected a new nest. Pitfall trap methods followed Deblauwe and Dekoninck (2007). Some Crematogaster spp. also forage on the ground and Dorylus spp. are nomadic and very active on the ground, so I used pitfall traps to measure their ground activity. I counted individuals of Crematogaster spp. (CrP) and Dorylus species (Dorylus spp.: DP, D. kohli: DKP, D. opacus: DOP) caught in the traps to give a ground activity measure per month and per vegetation type. Macrotermes spp. often forage outside the nest, and also are present in pitfall traps. I gave an abundance score of 1) few: 1–99 individuals, 2) common: 100–499 individuals, or 3) abundant: >500 individuals, to each Macrotermes species (MS) and I counted the number of traps containing each species (MP) per month and per vegetation type. Finally, I recorded the presence of winged termites of Macrotermes, Cubitermes, and Thoracotermes in nests or in pitfall traps each month (Macrotermes: WtM, all termites: Wtall).

To measure nest densities of the insect prey, I used both belt- and line-transect methods. I randomly placed a single 6 m × 100 m belt-transect (Jones and Eggleton 2000) in each of the 7 most important vegetation types: NPF, VOSF, OSF, YSF, RF, RS, and LGF (caused by tree fall). In August 2003, October and November 2004, and January–April 2005 I ran each belt-transect once, counted all active epigeal nests of Cubitermes spp. and Thoracotermes macrothorax in the 600 m2, and recorded the occupancy (primary and/or secondary occupants). I calculated the number of active nests/ha (CuA and ThA) and the number with primary occupants/ha (CuP and ThP) for each vegetation type. There were no active epigeal nests of Macrotermes muelleri and M. nobilis in the belt-transects. I also counted all intact nests of Oecophylla longinoda and all active nests of Tetramorium aculeatum and converted to nests per ha (ON and TeA) for each vegetation type. Crematogaster spp. have many small nesting sites, so distinguishing their colonies is difficult. Therefore, I recorded the presence of nests of Crematogaster in 20 sections of 6 m × 5 m and calculated relative abundance scores (of 20) of nesting sites (CrA) for each vegetation type. I found no nests of Dorylus spp. in the belt-transects. To obtain the nest density and abundance score per species for the whole study site, I corrected for the percentage of that vegetation type in the site and totaled them per month. In November 2004 (end of late wet season) and February 2005 (end of dry season) I also walked on 5 of the 10 line-transects and recorded all visible and active nests of Cubitermes spp., Thoracotermes macrothorax, Macrotermes spp., and all visible and intact nests of Oecophylla longinoda. Transect distances for each genus were: Cubitermes, 2.60 km; Thoracotermes, 8.63 km; Macrotermes, 9.83 km; and Oecophylla, 6.23 km. I estimated nest densities via the standing crop nest count method (Plumptre and Reynolds 1996) with DISTANCE 4.1 (Thomas et al.2003). I measured the perpendicular distance from the transect line to the middle of the active termite nest or, for Oecophylla, to the center of the colony. Because I recorded only visible nests of Oecophylla, I calculated an average colony size of 7 nests/colony (range with time: 3–11 nests), by averaging the colony sizes from activity visits per month (N = 19 colonies), and afterward over months (N = 17 mo, including 2002 data). I considered nests <20 m apart to belong to the same colony.

Statistical analyses

I used Spearman rank correlations to analyze temporal similarity in insect-eating between gorillas and chimpanzees and whether insectivory varied predictably with rain.

To investigate the correlation of fruit and foliage consumption, insect prey activity, and prey nest density with temporal variation in insectivory, I used a model selection approach. First, I conducted a principal component analysis (PCA) in STATISTICA 5.0 to reduce the number of correlating variables of the plant diet (FS, FSM, %S, %FSP, %Fo, F, GLF and rain). I retained only the first principal component explaining most of the variation. When several variables were associated (eigenvector coefficient: ≥0.60 or ≤−0.60) with the first principal component, I used the factor scores as a new variable. Second, I used the regression procedure GENMOD in SAS 9.1 to perform a multinomial logistic regression. An overview of the explanatory parameters used in the multinomial models is provided in Table I. The categorical response variables (Table II) each had 5 categories (0–4). In all analyses for gorillas, I corrected for repeated measures of gorilla groups. For chimpanzees, I could not correct for repeated measures because the number of communities in the study area is not known. For each response variable, I performed a backward selection of the model, resulting in a model with only the significant variable(s) or no variables at all. I did not check the effect of interactions between variables. I used the Bonferroni method to correct each multinomial model for the total number of explanatory variables in the full model. Finally, I checked the correlation of each response variable with the parameters in its full model and between these parameters using Spearman rank correlation over all fecal samples and over the monthly means, and Bonferroni corrected the correlations for the total number of tests per response variable.
Table I

Explanatory variables in the multinomial model selection for gorillas and chimpanzees of La Belgique

Explanatory variable

Code

Specifier

No. of monthsa

Gorilla

Fruit and foliage score in feces

GFF

PCA scores

12

Percentage of fruit skin and pulp in feces

%FSP

%

12

Monthly average of Oecophylla individuals per nest

OI/N

No. of individuals/nest

12

Percentage of active Oecophylla nests per month

O%A

%

12

Monthly density of intact Oecophylla nests per hectare

ON

No. of nests/ha

7

Monthly average of Crematogaster individuals per 10 cm2

CrI

No. of individuals/10 cm2

12

Monthly number of Crematogaster individuals in pitfalls

CrP

No.

12

Monthly Crematogaster nest abundance score

CrA

No. of sections with nests/20

7

Percentage of active Tetramorium nests per month

Te%A

%

12

Monthly density of active Tetramorium nests per hectare

TeA

No. of nests/ha

7

Monthly Cubitermes biomass per 100 g nest material

CuW

g/100 g

12

Monthly density of active Cubitermes nests per hectare

CuA

No. of nests/ha

7

Monthly density of Cubitermes nests with primary occupants per hectare

CuP

No. of nests/ha

7

Monthly Thoracotermes biomass per 100 g nest material

ThW

g/100 g

12

Monthly density of active Thoracotermes nests per hectare

ThA

No. of nests/ha

7

Monthly density of Thoracotermes nests with primary occupants per hectare

ThP

No. of nests/ha

7

Monthly presence of winged Macrotermes, Cubitermes and Thoracotermes in nests or pitfall traps

Wtall

Present (1), absent (0)

12

Chimpanzee

Fruit and foliage score in feces

CFF

PCA scores

12

Percentage of fruit skin and pulp in feces

%FSP

%

12

Fiber abundance score in feces

F

Absent (0), rare (1), few (2), common (3), abundant (4)

12

Number of fruit species per fecal sample

FS

No. of species/feces

12

Number of fruit species per month in feces

FSM

No. of species/month

12

Presence of evidence of meat-eating in feces

Meat

Present (1), absent (0)

12

Monthly number of individuals of Dorylus in pitfalls

DP

No.

12

Monthly number of individuals of D. kohli in pitfalls

DKP

No.

12

Monthly number of individuals of D. opacus in pitfalls

DOP

No.

12

Monthly average of individuals of Oecophylla per nest

OI/N

No. of individuals/nest

12

Percentage of active nests of Oecophylla per month

O%A

%

12

Monthly density of intact nests of Oecophylla per hectare

ON

No. of nests/ha

7

Monthly Macrotermes biomass per 100 g of nest material

MW

g/100 g

8

Monthly number of pitfall traps containing Macrotermes

MP

No.

12

Monthly number of pitfall traps containing M. muelleri

MmP

No.

12

Monthly number of pitfall traps containing M. nobilis

MnP

No.

12

Monthly number of pitfall traps containing M. lilljeborgi/renouxi

MlrP

No.

12

Monthly abundance score of Macrotermes in the pitfall traps

MS

Absent (0), few (1), common (2), abundant (3)

12

Monthly abundance score of M. muelleri in the pitfall traps

MmS

Absent (0), few (1), common (2), abundant (3)

12

Monthly abundance score of M. nobilis in the pitfall traps

MnS

Absent (0), few (1), common (2), abundant (3)

12

Monthly abundance score of M. lilljeborgi/renouxi in the pitfall traps

MlrS

Absent (0), few (1), common (2), abundant (3)

12

Monthly presence of winged Macrotermes in nests or pitfall traps

WtM

Present (1), absent (0)

12

aNumber of months in which I collected data.

Table II

The final model for each response variable (RV) of gorillas and chimpanzees with the degrees of freedom (df), the χ2, and the p-value after Bonferroni correction (p) for each significant explanatory variable (EV) and the Spearman rank correlations (rs) over all fecal samples and over the monthly meansa

Gorilla

Chimpanzee

RV

EV

Multinomial model

N=feces

N=month

RV

EV

Multinomial model

N=feces

N=month

 

 

df

χ2

p

N

rs

N

rs

 

 

df

χ2

p

N

rs

N

rs

IAS

%FSP

1

9.63

0.0038

178

-0.17*

12

-0.63*

IAS

CFF

1

17.95

<0.0006

135

-0.35***

12

-0.85***

AAS

GFF

1

26.59

<0.0002

178

0.37****

12

0.77**

AAS

CFF

1

20.02

<0.0006

135

-0.38***

12

-0.80**

OAS1

GFF

1

11.91

0.0024

178

0.27***

12

0.70

DAS

Meat

1

14.57

0.0007

135

0.36***

12

0.40

%FSP

1

9.19

0.0096

178

-0.14

12

-0.56

DkAS

x

       

O%A

1

7.71

0.0220

178

-0.18

12

-0.30

DoAS

CFF

1

7.52

0.0305

135

-0.27**

12

-0.48

OAS2

%FSP

1

24.30

<0.0005

108

-0.12

7

-0.30

 

Meat

1

8.87

0.0145

135

0.74***

12

0.74

OI/N

1

19.30

<0.0005

108

-0.03

7

-0.19

OAS1

x

O%A

1

14.93

0.0005

108

-0.14

7

-0.41

OAS2

x

ON

1

18.25

<0.0005

108

0.18

7

0.59

TAS

CFF

1

9.54

0.0120

135

-0.26**

12

-0.69

CrAS1

GFF

1

27.47

<0.0004

178

0.17

12

0.53

MAS1

CFF

1

10.39

0.0104

135

-0.27**

12

-0.68

CrP

1

8.69

0.0128

178

-0.06

12

-0.23

MAS2

CFF

1

10.39

0.0117

135

-0.27**

12

-0.68

CrAS2

CrA

1

212.64

<0.0005

108

0.31**

7

0.74

MmAS

CFF

1

15.55

<0.0008

135

-0.32***

12

-0.81*

TeAS1

GFF

1

28.12

<0.0003

178

0.30****

12

0.39

MnAS

x

TeAS2

GFF

1

28.12

<0.0004

178

0.30****

12

0.39

MlrAS

x

TAS

GFF

1

8.26

0.0080

178

-0.14

12

-0.63*

WtAS

CFF

1

11.58

0.0042

135

0.28**

12

0.52

%FSP

1

37.63

<0.0002

178

-0.11

12

0.25

         

CuAS1

GFF

1

9.64

0.0057

178

-0.01

12

-0.03

         

%FSP

1

9.39

0.0066

178

-0.10

12

-0.00

         

CuW

1

13.71

0.0006

178

0.07

12

0.14

         

CuAS2

%FSP

1

9.80

0.0085

178

-0.10

12

-0.00

         

ThAS1

GFF

1

27.23

<0.0003

178

-0.11

12

-0.65

         

%FSP

1

6.45

0.0333

178

-0.18

12

0.50

         

ThAS2

ThW

1

60.57

<0.0005

108

0.16

7

0.71

         

ThA

1

10.24

0.0070

108

-0.06

7

-0.33

         

ThP

1

11.98

0.0025

108

-0.11

7

-0.17

         

WtAS

GFF

1

48.95

<0.0003

178

-0.22**

12

-0.28

         

%FSP

1

13.82

0.0006

178

-0.26***

12

-0,14

         

Wtall

1

17.70

<0.0003

178

0.43****

12

0.83***

         

aResponse variables: IAS=insect abundance score, AAS=ant AS, TAS=termite AS, WtAS=winged termite AS, OAS=Oecophylla AS, CrAS=Crematogaster AS, TeAS=Tetramorium AS, DAS=Dorylus AS, DkAS=Dorylus kohli AS, DoAS=D. opacus AS, CuAS=Cubitermes AS, ThAS=Thoracotermes AS, MAS=Macrotermes AS, MmAS=Macrotermes muelleri AS, MnAS=M. nobilis AS, MlrAS=M. lilljeborgi/renouxi AS. Explanatory variables (for abbreviations see Table I) in all models for gorillas were GFF and %FSP, while for chimpanzees these were CFF, FS, FSM, %FSP, F and Meat. Depending on the termite or ant species of the response variable I included monthly activity and nest density parameters in the models (models 1 without and models 2 with nest density or biomass parameter(s), which I measured only during 7 or 8 mo). Spearman rank correlations: *p < 0.1 (trend), **p < 0.05, ***p < 0.01, *** * p < 0.001 (Bonferroni corrected; n=sample number).

I used t-tests to compare the estimated nest densities from line transects between November 2004 and February 2005 (Plumptre 2000).

Finally, to check if the preferred vegetation types of gorillas and chimpanzees for the consumption of insects corresponded to the vegetation types with the highest insect availability, I used χ2 tests. Because I found only traces of feeding on termites (Deblauwe and Janssens 2008), I compared the distribution of termite-feeding traces with the distribution of termite nests (Cubitermes and Thoracotermes, line-transect data) or of pitfall occurrences (Macrotermes). I first corrected the distributions for the percentage of each vegetation type in the study area (traces) or on the line-transects (nests). I calculated a preference index (PI) as follows: PI=percentage of termite feeding traces in a vegetation type - percentage of nests or occurrences in that vegetation type. For the ant Oecophylla, I also corrected the distribution of nests across vegetation types for the percentage of each vegetation type on the line-transects.

Results

Plant Diet Composition and Temporal Variation

I collected 178 gorilla (G) and 135 chimpanzee (C) fecal samples, an average of 15 (N = 12; range 6–20) and 11 (N = 12; range 3–21), respectively, on an average of 5 d (N = 12; G: range 2–9 and C: range 2–10) per month (Fig. 1). I recorded 2106 gorilla and 781 chimpanzee feeding traces. Gorillas consumed 96 plant species and chimpanzees, 80. However, for gorillas and chimpanzees, respectively 36% and 31% of the plant species consumed are not yet identified by genus. Dietary overlap is high: both species ate 63% of chimpanzee and 48% of gorilla plant species. Both great apes also ate mushrooms and many insects, while chimpanzees sometimes ate meat. Chimpanzees and gorillas ate a similar number of fruit species (C: 68; G: 70) and shared >50% (G: 53%, C: 54%) of these, including the important species (sensu Doran et al.2002) for both apes. Ninety-seven percent of gorilla feces and 100% of chimpanzee feces contained fruit. The mean number of fruit species per fecal sample was 3.1 (range 0–9) for gorillas and 3.3 (range 0–7) for chimpanzees. Gorilla feces were dominated by foliage (mean foliage score: 5.5 and mean %Fo: 57.7 vs. mean %Fr: 40.1), while chimpanzee feces were fruit dominated (mean foliage score: 2.3 and mean %Fo: 15.3 vs. mean %Fr: 80).

For gorillas, the first axis of the PCA accounted for 50% of the total variation in the 8 plant diet variables. The number of fruit species per fecal sample (FS) and per month (FSM), and the percentage of seeds (%S) were negatively associated with the first axis, while the percentage of foliage (%Fo), the fiber (F), and green leaf fragment (GLF) abundance scores were positively associated. I used the scores of the PCA axis as a new variable for gorillas, called fruit and foliage score (GFF). In September and October 2004 gorillas consumed many succulent fruit species, e.g., Landolphia spp., Cissus dinklagei, Celtis spp., while in March, April 2005, and May 2003 gorillas fed almost exclusively on foliage and much less fruit was eaten (Fig. 2a). Between these periods, the diet remained foliage-dominated. The percentage of fruit skin and pulp (%FSP) seems to represent, in particular, the amount of fibrous fruits, e.g., Desplatia spp. and Duboscia macrocarpa, and Gambeya lacourtiana, consumed. Some fruits were eaten during >7 mo: Duboscia spp. (11 mo), Desplatia spp. (11 mo), Aframomum spp. (11 mo), Nauclea diderrichii (10 mo), and Uapaca spp. (8 mo).
https://static-content.springer.com/image/art%3A10.1007%2Fs10764-009-9337-2/MediaObjects/10764_2009_9337_Fig2_HTML.gif
Fig. 2

(a) Mean gorilla fruit and foliage PCA score (GFF) per month and (b) mean chimpanzee fruit and foliage PCA score per month (CFF). The signs of the CFF scores are inverted to obtain a graph similar to that for GFF; the direction of the arrows indicates an increase of fruit or foliage in the feces; bars represent the standard deviations.

For chimpanzees, the first axis of the PCA accounted for 34% of the total variation in the 8 plant diet variables. The percentage of foliage in the feces (%Fo) and the green leaf fragment abundance score (GLF) were negatively associated with the first PCA axis, and the percentage of seeds (%S) positively. I used the scores of the PCA axis as the new variable fruit and foliage score (CFF) for chimpanzees. Chimpanzees ate large amounts of succulent fruits during a longer period than gorillas (September–November 2004), but also had a foliage-dominated diet in May 2003 (Fig. 2b). In all months except May, the chimpanzee diet was fruit-dominated. Only 1 fruit genus was eaten during >7 mo, Ficus spp. (9 mo).

Temporal Variation in Insect Diet and Explanatory Variables

The mean insect (IAS) (Fig. 3a), Oecophylla (OAS), and termite (TAS) (Fig. 3b) abundance scores of gorilla feces/mo do not significantly correlate with those of chimpanzee feces (IAS: N = 12, rs = 0.40, p = 0.193; OAS: N = 12, rs = –0.03, p = 0.921, TAS: N = 12, rs = 0.00, p = 0.991). In contrast, the mean ant (AAS) (Fig. 3c) and winged termite (WtAS) abundance scores of gorilla feces/mo correlate positively and significantly with those of chimpanzee feces (AAS: N = 12, rs = 0.85, p < 0.001; WtAS: N = 12, rs = 0.58, p = 0.048). None of the abundance scores correlates with monthly rainfall (p > 0.05). Termite and Macrotermes abundance scores (TAS and MAS) of chimpanzee feces negatively, but nonsignificantly, correlate with rain (TAS: N = 12, rs = −0.55, p = 0.067; MAS: N = 12, rs = −0.55, p = 0.062).
https://static-content.springer.com/image/art%3A10.1007%2Fs10764-009-9337-2/MediaObjects/10764_2009_9337_Fig3_HTML.gif
Fig. 3

The mean monthly insect abundance score (IAS, a), termite abundance score (TAS, b), and ant abundance score (AAS, c) of chimpanzees and gorillas at La Belgique.

Both the plant diet variables of gorillas (GFF and %FSP) and some insect activity and nest density parameters are significantly associated with the amount of insects eaten by gorillas (Table II). However, the correlations are not always significant. Gorillas ate the largest quantities of insects when few fibrous and fruits of Gambeya lacourtiana (%FSP) were included in the diet, especially in April 2005 and May 2003 (Fig. 3a). Although the fruit and foliage score (GFF) correlates negatively with %FSP (N = 178, rs = −0.46, p < 0.001) and positively with IAS (N = 178, rs = 0.20, p = 0.023), the correlations are not significant in the model. In contrast, gorillas included more ants in their diet when they ate more foliage and less succulent fruit. When considering only important ant prey, foliage consumption was positively associated with consumption of Oecophylla, Crematogaster, and Tetramorium. Oecophylla fecal abundance score (OAS) is negatively associated with %FSP, but correlations are not significant. In addition, monthly nest density of Oecophylla (ON) and Crematogaster (CrA), which correlate positively with GFF (ON: N = 108, rs = 0.62, p < 0.002; CrA: N = 108, rs = 0.80, p < 0.002), and their activities (OI/N, O%A, and CrP) are associated with the abundance scores of the ants in feces (OAS and CrAS). However, only CrA correlates significantly with CrAS. The percentage of active nests of Oecophylla per month (O%A) correlates positively with the monthly average number of individuals per nest of Oecophylla (OI/N) (N = 178, rs = 0.60, p < 0.001). The monthly nest density and the percentage of active nests of Tetramorium (TeA and Te%A), which correlate with GFF (TeA: N = 108, rs = 0.56, p<0.001; Te%A: N = 178, rs = −0.52, p < 0.001), are not associated with the abundance score of Tetramorium (TeAS). In contrast to ant-eating, termite-eating by gorillas increased when less foliage and fibrous fruits, but more succulent fruits, were included in the diet. Although the models included GFF and %FSP, there is no correlation with the abundance scores for termites (TAS), Cubitermes (CuAS), and Thoracotermes (ThAS) in feces. The monthly biomass/100 g of nest material (CuW and ThW) is positively associated with CuAS (in the model without CuA and CuP) and with ThAS (in the model with ThA and ThP), but there is no significant correlation. However, CuW correlates negatively with GFF (N = 178, rs = −0.23, p < 0.010) and with monthly nest density of Cubitermes (CuP; N = 108, rs = −0.67, p < 0.002). The model included the monthly nest densities of Thoracotermes (ThA and ThP), but they do not correlate with ThAS. Finally, winged termites were eaten when they were available, which coincided with the presence of succulent fruit and a low amount of fibrous fruit in the diet.

Insect-eating by chimpanzees is significantly associated with the fruit and foliage score (CFF) and meat, but not with insect activity and nest density (Table II). Chimpanzees ate more insects, ants, and termites, especially Macrotermes, when more foliage, especially green leaves, and less succulent fruits were eaten. Ant abundance in the feces (AAS) peaked in May 2003 (Fig. 3c) when the chimpanzee diet was foliage-dominated. Although %FSP and fiber abundance score (F) correlate negatively with CFF (%FSP: N = 135, rs = −0.39, p < 0.002; F: N = 135, rs = −0.55, p < 0.002) and %FSP positively with AAS (N = 135, rs = 0.30, p = 0.008), they are not significant in the models. A peak in meat consumption in May 2003 is positively associated with the consumption of Dorylus army ants (DAS), especially of D. opacus (DoAS). DoAS is also associated with CFF and had the highest peak in May 2003. Although the monthly number of Dorylus opacus in pitfalls (DOP) correlates positively with the abundance score in feces (DoAS; N = 135, rs = 0.28, p = 0.025), the model did not include it. The abundance scores of Dorylus kohli (DkAS) and Oecophylla (OAS) in feces, which also peaked in May 2003, are not associated with any of the variables. The abundance score of Macrotermes muelleri (MmAS) increased with foliage consumption and decreased with succulent fruit consumption. The amounts of Macrotermes nobilis and M. lilljeborgi/renouxi eaten by chimpanzees (MnAS and MlrAS) are not associated with any of the variables. In contrast to the soldiers of Macrotermes, their winged termites were eaten only in October and November 2004, and the fruit and foliage score (CFF) has a significant positive correlation with their abundance score in feces (WtAS).

Ant and Termite Nest Densities Based on Line-Transects

The results of the DISTANCE analyses are provided and compared with insect nest density estimates of belt-transects and with insect abundance scores in gorilla and chimpanzee feces in Table III. I tried to detect a minimum of 60–80 nests or colonies (Buckland et al., 1993), but achieved this only for active nests of Cubitermes. I estimated the density of active nests of Thoracotermes and colonies of Oecophylla with a coefficient of variance of <34%, but detected too few active, epigeal nests of Macrotermes (M. nobilis and M. muelleri) to do more than estimate encounter rates per kilometer (ERK). In general, belt-transect estimates were much higher than line-transect estimates, but relative nest densities stayed the same across species (Cubitermes > Oecophylla > Thoracotermes > Macrotermes). In addition, the differences between November and February are similar for both methods. The densities of active nests of Cubitermes and Thoracotermes estimated by the line- and belt-transects (CuA and ThA) did not differ between the 2 mo. Still, the abundance score in gorilla feces decreased in February for Thoracotermes (ThAS), but not for Cubitermes (CuAS). This is in contrast with the multinomial models of CuAS and ThAS, which included ThA, but excluded CuA. Although nest density of Oecophylla did not differ significantly between November and February, both methods showed a clear increase in February. This increase was reflected by the Oecophylla abundance score (OAS) in gorilla feces and is consistent with the inclusion of monthly nest density (ON) in the multinomial model of OAS. In contrast, chimpanzees did not eat Oecophylla in both months. Finally, ERK of active nests of Macrotermes did not differ between November and February, but abundance scores for Macrotermes muelleri and M. nobilis in chimpanzee feces increased in February.
Table III

Estimated nest densities of Cubitermes, Thoracotermes, and Oecophylla from line-transects (DLT) and from belt-transects (DBT); the encounter rate per km (ERK) for epigeal Macrotermes nests and the abundance scores of the insects in the gorilla (ASG) and chimpanzee (ASC) feces in November 2004 (N) and February 2005 (F)a

Species

Mo.

N

DLT (/ha)

95%CI

CV (%)

t

ERK

DBT (/ha)

ASG

ASC

Cubitermes

N

79

23.5

18.7–27.4

9.3

0.59

 

50.5

0.7

 

F

77

21.7

18.6–25.8

8.7

  

50.0

0.8

 

Thoracotermes

N

34

3.7

1.8–7.9

32.0

0.02

 

10.7

0.8

 

F

32

3.8

2.8–6.8

27.3

  

9.5

0.4

 

Oecophylla

N

26

13.7 (2.0)

6.5–29.0 (0.9–4.1)

33.5

0.99

 

14.2

1.0

0.0

F

43

20.2 (2.9)

15.2–30.3 (2.2–4.3)

19.5

  

34.7

1.5

0.0

Macrotermes

N

14

    

1.5 ± 1.2

  

0.4

F

12

    

1.0 ± 0.8

  

0.8

aDISTANCE: N=number of detections, 95%CI = 95% confidence interval, CV=percent coefficient of variation, t=t-statistic of T-test (all are not significant). The colony density and its 95% CI of Oecophylla are in parentheses. The ERK of Macrotermes is the average ERK over the 5 transects. The abundance score (AS) of Macrotermes in chimpanzee feces is the average of the AS of M. muelleri and M. nobilis (epigeal nest builders).

Spatial Variation in Termite and Ant Availability

The distribution of gorilla termite-feeding traces across vegetation types differed from the distribution of termite nests (Cubitermes: χ2 = 116.3, df = 7, p < 0.001; Thoracotermes: χ2 = 59.0, df = 7, p < 0.001). Preference indices show that gorillas broke more nests of Cubitermes in NPF, YSF, RS, and LGF, and more nests of Thoracotermes in NPF, RS, and LGF than expected from the proportion of nests in these vegetation types (Fig. 4a). In addition, the distribution of chimpanzee feeding traces of Macrotermes across vegetation types differed from that based on pitfall occurrences (χ2 = 98.1, df = 7, p < 0.001). Chimpanzees fished more for soldiers of Macrotermes in OSF, YSF, and RF than expected (Fig. 4a). However, on line-transects epigeal nests of Macrotermes were present only in OSF and RS. Most nests of Oecophylla were present in YSF and especially LGF, most nests of Tetramorium in YSF and RS, while those of Crematogaster were present in all vegetations with most nests in RF and LGF (Fig. 4b). Dorylus were active in all vegetation types, but had their highest activity in NPF and LGF (Fig. 4b). Dorylus kohli activity was highest in OSF, YSF, and especially RS, while D. opacus activity in YSF and LGF (Fig. 4b).
https://static-content.springer.com/image/art%3A10.1007%2Fs10764-009-9337-2/MediaObjects/10764_2009_9337_Fig4_HTML.gif
Fig. 4

(a) Preference indices of gorilla and chimpanzee termite-feeding in the 7 important vegetation types and (b) proportion (%) of nests of Oecophylla (ON), Crematogaster (CrN), and Tetramorium (TeN) and of individuals of Dorylus (DA), D. kohli (DkA) and D. opacus (DoA) in pitfall traps (activity) in the 7 vegetation types. Cu=Cubitermes, Th=Thoracotermes, and M=Macrotermes; NPF=near primary forest, VOSF=very old secondary forest, OSF=old secondary forest, YSF=young secondary forest, RS=Raphia swamp, RF=riverine forest, and LGF=light gap forest.

Discussion

In general, chimpanzees and gorillas at La Belgique have plant foraging strategies similar to those of other Central African great ape populations. Chimpanzees maintained a high overall intake of ripe fruit, except in May, when even nonsucculent fruit species were rare and chimpanzees ate more foliage, especially leaves. Party size decreased when fewer succulent fruits were available (Guislain and Dupain 2007; Hashimoto et al.2003). As at Ndoki (Kuroda et al.1996), fruits of Ficus were an important fallback food for chimpanzees, but less so for gorillas. Gorillas mostly fed on succulent fruits when these were available and increased their daily path length in this period (Guislain and Dupain 2007). During periods of fruit scarcity, especially in April and May, gorillas relied heavily on THV and on less palatable fruits such as Duboscia, Desplatia, and Nauclea diderrichii.

Temporal Variation in Insect-Eating

As expected, chimpanzees and gorillas showed different patterns of temporal variation in insect-eating, which is consistent with findings from Lopé (Tutin and Fernandez 1992). I also found different explanations for their insect-eating seasonality. Succulent fruit scarcity coincided with increased chimpanzee insectivory, as it did for eastern chimpanzees in the montane forest of Kahuzi-Biega (DRC; Yamagiwa and Basabose 2006). For gorillas, consumption of fibrous fruits and Gambeya lacourtiana coincided with decreased insect-eating. Ants and termites represent staple foods (sensu Rogers et al.2004) for chimpanzees and gorillas at La Belgique: they are important food items (Deblauwe and Janssens 2008) and are eaten on a daily or weekly basis.

When fruit was extremely scarce in April and May, both gorillas and chimpanzees consumed large amounts of ants. Temporal and spatial variation in ant nest availability was also associated with ant consumption by gorillas, as expected. Gorillas foraged for THV predominantly in young secondary forest (YSF), light gap forest (LGF), and Raphia swamp (RS; Guislain and Dupain 2007), which have the highest ant prey nest abundances during these months (Fig. 4b). Increased ant consumption seemed to be partly a consequence of increased foraging in these vegetation types (cf. Tutin and Fernandez 1992). Both ape species travel through all vegetation types at our site, but distribute their foraging time differently across vegetation types (Guislain and Dupain 2007). Extreme fruit scarcity and spatial variation in ant availability were associated with ant-eating in chimpanzees in the same way as in gorillas. Chimpanzees fed on fruits mostly in older vegetation types and riverine forest (RF), and fed on THV mostly in YSF, LGF, and RS (Guislain and Dupain 2007). They ate more THV when fruit was scarce and also encountered more ant trails and nests (Fig. 4b). In contrast to that in gorillas, ant-eating in chimpanzees was not associated with temporal ant availability, but with meat consumption. The percentage of chimpanzee feces containing meat remains at our site was highest in the period of extreme fruit scarcity (early wet season).

The nutritional importance of ants might have been high relative to the amount ingested, especially because no remains of the protein-rich larvae and pupae are recovered in fecal samples (Deblauwe and Janssens 2008). The extra protein supply to the apes’ diet could complement the low protein content of low-quality THV, e.g., Aframomum spp. (Rogers et al.1988) when high-quality THV, e.g., Haumania danckelmanniana, is scarce.

Finally, the single important ant prey eaten by both ape species, Oecophylla longinoda, showed no similar seasonality in their diet, which is consistent with findings from Lopé, where this ant made up the bulk of the insect part of the apes’ diet (Tutin and Fernandez 1992). However, none of the explanatory factors, except spatial availability, could account for the temporal variation in Oecophylla-eating by chimpanzees, while for gorillas it was associated with several factors. A possible explanation that needs further investigation is the medicinal use of Oecophylla by chimpanzees. Native Australians use this ant species against gastrointestinal disorders, fever, pain, inflammations, etc. (Devanesen 2000).

In contrast to ant-eating, chimpanzees and gorillas displayed different fluctuations in nonwinged termite-eating. Whereas savanna chimpanzees at Fongoli increased termite-eating when fruit availability was high (Bogart and Pruetz 2008), chimpanzees at La Belgique increased termite-eating when succulent fruit consumption was low. This is consistent with their ant-eating habits and supports the second explanation that termites complement the protein from low-quality THV during periods of fruit scarcity. The consumption of protein-rich green leaves also increased during those periods. However, secondary compounds and high amounts of fiber make them less digestible than insects (Milton 1999), and insects might compensate for the lack of essential amino acids in leaves (Hladik 1977). Further, the high manganese content of termites (Deblauwe and Janssens 2008) might compensate for the low content in low-quality THV and leaves. Temporal variation in availability of Macrotermes and variation in the frequency with which the apes encountered termites while foraging for other foods are not associated with the frequency of consumption. However, only epigeal nests are visible, and estimates of Macrotermes abundance are based on activity rather than on nest density. Chimpanzees dig and fish for termites in subterranean or half-subterranean nests (Deblauwe et al.2006), and I did not measure the density of such inconspicuous nests.

Seasonality in termite-eating by gorillas was more complex and attributable to different factors. Termite-eating by gorillas correlates positively with the amount of succulent fruit in the diet, which was unexpected, but also with temporal variation in availability of Cubitermes and Thoracotermes, especially biomass, which was expected. Our results support the finding of other studies that gorillas feed more on termites during the rainy season when fruit is abundant (Doran et al.2002; Goldsmith 1999). Gorillas might eat soil-feeding termites as a high-quality alternative for geophagy (Deblauwe and Janssens 2008). The termites are rich in iron. Although bioavailability of micronutrients in ripe fruits is predicted to be higher than in leaves (Milton 2003), at Bai Hokou (CAR), gorilla fruits with high iron levels also had high condensed tannin (CT) levels (Remis et al., 2001). Therefore, fruit iron (Fe3+), which is less easily absorbed by gorillas, might be used in detoxification, while termite iron (Fe2+) might complement iron requirements in periods of fruit abundance (Deblauwe and Janssens 2008). Second, soil-feeding termites might also be consumed for antidiarrheal purposes (Deblauwe and Janssens 2008) when a large amount of succulent fruit is eaten. A dietary change to a high intake of succulent fruits can cause gastrointestinal disorders in animals (Putz 1993), including gorillas (Savini et al. 2000; Tutin and Fernandez 1992). For example, an important fruit eaten by gorillas at the study site during the rainy season is that of Aframomum spp., which is used as a laxative in Congo (Tane et al.2006). In addition, increased consumption of this fruit by gorillas can lead to diarrhea (Schaller 1963).

Estimated termite biomass was lowest in the dry season. Some termites move deeper into the soil during the dry season (Wood et al.1982). Cubitermes and Thoracotermes might do the same in pursuit of moisture, resulting in reduced availability to gorillas. The seasonality of active nest density is related both to the foundation of new colonies after swarming of reproductive termites and to the abandoning of old nests. Active nest density differed between vegetation types and preference indices show that gorillas did not always break mounds of Thoracotermes in vegetation types with the highest density. Thus, the temporal association of active nest density of Thoracotermes with its abundance score in gorilla feces is complex. There were many termite-feeding traces in vegetation types where gorillas fed on THV, e.g., Aframomum pith, which supports the suggested relationship between feeding on termites and feeding on the succulent fruits of Aframomum.

Finally, consumption of winged termites by chimpanzees and gorillas shows similar seasonality. Winged termites have a high fat content (53% dry weight; Hladik 1977) and might be a source of energy. Gorillas ate alates of Macrotermes, Cubitermes, Thoracotermes, and other genera when they were available, while chimpanzees ate alates of Macrotermes only in the late wet season when succulent fruits were eaten. Availability, based on presence or absence, was not associated with consumption of winged termites by chimpanzees. However, in the late wet season I encountered more alates than in other seasons. Thus, winged termite availability, instead of fruit consumption, positively influenced consumption of alates by chimpanzees, as for gorillas.

Extension of Niche Differentiation

Chimpanzees and gorillas feed selectively on insects. The foraging strategies used for feeding on insects, especially termites, are similar to those used for feeding on plant food, which also causes niche differentiation in insect diet. Chimpanzees are fruit pursuers even in periods of fruit scarcity, which is a high-energy strategy, while gorillas are fruit pursuers only when fruit is abundant and switch to a low-energy strategy in periods of fruit scarcity (Rogers et al.2004). The fact that termite availability did not seem to influence chimpanzee termite-feeding behavior suggests that, when feeding on termites, chimpanzees choose a similar high-energy strategy. They search for nests of Macrotermes, which are less visible (subterranean) or available (epigeal) than those of gorilla termite prey, and use a tool set to obtain mainly the major soldiers. These methods are time- and energy-consuming, but the high nutritional benefits —energy, protein, and manganese— of the prey may account for this behavior (Deblauwe and Janssens 2008; McGrew 2001; McGrew et al.1979). Although gorillas could also feed on Macrotermes (Deblauwe and Janssens 2008), searching for nests of Cubitermes and Thoracotermes and catching the termites by hand is less time- and energy-consuming. However, they also provide a lower nutritional return. Although spatial termite availability did not influence gorilla termite-feeding behavior, temporal availability did. This suggests that gorillas adopt a low-energy strategy when feeding on termites, but factors other than energy or protein, e.g., iron, may explain the importance of the activity.

Both ape species seemed to use a rather low-energy strategy for foraging on ants. Spatial availability influenced ant-eating and chimpanzees apparently did not use tools to obtain ants. Ants provide less nutritional benefit than termites (Deblauwe and Janssens 2008). However, chimpanzee ant-eating did not reflect temporal variation in ant availability. Further, energy and protein derived from ants could be underestimated due to the undetectability of larvae in fecal samples. If chimpanzees focus on larvae of Dorylus by opening the nests by hand, as in Tai (Boesch and Boesch 1990), biomass intake may be similar to or higher than that of gorillas (Deblauwe and Janssens 2008) and a high-energy strategy may be used. In addition, temporal and spatial nest densities of Dorylus prey need to be measured to check the effect of nest availability on army ant-eating. The ants are nomadic and nests subterranean (Deblauwe and Janssens 2008) and their nests are difficult to detect. Researchers have not yet documented the nests of the important chimpanzee prey species Dorylus kohli and D. opacus (Schöning et al.2008). I encountered visible nests of Dorylus sjöstedti and D. wilverthi occasionally, but never in belt-transects or along line-transects, and hence they are probably less available than nests of Oecophylla, Crematogaster, and Tetramorium.

Conclusion

Niche differentiation is apparent not only in the plant diet of great apes, but also in their insect diet. Although fecal analysis has limits, I identified clear seasonal patterns in insect-eating by both ape species. However, especially for gorillas, a complex combination of ecological and nutritional factors explained the fluctuation. Further analyses of the nutritional value and availability of fruit and foliage, together with density measures of inconspicuous nests of Macrotermes and Dorylus at the study site, are needed to confirm my hypotheses. Ideally, researchers should conduct direct observation of habituated individuals to investigate time allocation patterns, manipulation and amount of insect prey ingested, sex-age differences, inter- and intraspecific competition for insects, etc. However, at La Belgique habituation is currently not an option because hunting pressure remains high in the surrounding area. Therefore, similar insectivory studies are required at sites where sympatric chimpanzees and gorillas are protected, incorporating nutritional analyses and accessibility and availability studies of the insect prey: prey characteristics, activity, and nest density. This will allow a comparison of geographic variation in great ape insect-eating behavior against an ecological and nutritional background, helping us to distinguish cultural from ecological differences.

Acknowledgments

I thank the RZSA, Durell Wildlife Conservation Fund, the logging company FIPCAM, MINRESI (Ministère de la Recherche Scientifique et de l’Innovation, Cameroon), MINFoF (Ministère de Forêt et de la Faune, Cameroon), and the Service de Conservation de la Reserve de Faune du Dja for supporting this project. I thank the Badjoué trackers who guided me and Patrick Guislain and the many volunteers who helped with data collection. The King Leopold III fund for Nature Exploration and Conservation (RBINS), the Flemish Inter-University Council (VLIR), Father Louis Bruyns Foundation (UA), and the Fund to Promote Scientific Research in Africa (NBG, Meisse) funded the fieldwork. I was supported by a grant from the UA and the RZSA. I thank the Flemish government for structural support to the CRC of the RZSA. Thanks are also due to Stefan Van Dongen for statistical advice; to Eric Arnhem for helping me with DISTANCE 4.1.; and to Nikki Tagg, Linda Van Elsacker, Jeroen Stevens, Patrick Guislain, David Watts, and an anonymous reviewer for helpful comments on the manuscript.

Copyright information

© Springer Science+Business Media, LLC 2009