Primates

, Volume 48, Issue 1, pp 22–26

Ant fishing by wild chimpanzees is not lateralised

Authors

    • Department of AnthropologyMiami University
  • W. C. McGrew
    • Department of AnthropologyMiami University
    • Department of ZoologyMiami University
    • Leverhulme Centre for Human Evolutionary Studies, Department of Biological AnthropologyUniversity of Cambridge
Original Article

DOI: 10.1007/s10329-006-0020-3

Cite this article as:
Marchant, L.F. & McGrew, W.C. Primates (2007) 48: 22. doi:10.1007/s10329-006-0020-3

Abstract

Right-dominant handedness is unique and universal in Homo sapiens, suggesting that it is a highly derived trait. Our nearest living relations, chimpanzees, show lateralised hand preference when using tools, but not when otherwise manipulating objects. We report the first contrary data, that is, non-lateralised tool-use, for ant fishing as done in the Mahale Mountains of Tanzania. Unlike nut cracking, termite fishing, and fruit pounding, as seen elsewhere, in which most individuals are either significantly or wholly left- or right-biassed, ant fishers are mostly ambilateral. The clue to this exception lies in arboreality; all other patterns of chimpanzee elementary technology are done on the ground. Arboreal tool use usually requires not only that one hand be used to hold the tool, but also that the other hand gives postural support. When the supporting hand is fatigued, then it must be relieved by the other. Terrestrial tool use entails no such trading off. To test the hypothesis, we compared frequency of hand changing with the incidence of major hand support, and found them to be significantly positively correlated. The evolutionary transition from arboreality to terrestriality may have been a key enabler for the origins of human laterality.

Keywords

Pan troglodytesTool useHandednessElementary technologyInsectivory

Introduction

Living humans (Homo s. sapiens) show universal and unique right-hand predominance at the species level, while living apes show only individual-level hand preferences in certain contexts. More precisely, in all human cultures in which handedness has been recorded, about 90% of individuals show overall right handedness. In contrast, at least some wild great apes (Gorilla gorilla, Pan troglodytes), typically show ambilaterality for hand use, except in two contexts of extractive foraging: food processing (= transformation of a natural food item to a more edible state by disabling its mechanical defenses) (Byrne et al. 2001) and tool use (see review in McGrew and Marchant 1997). With one exception (foliage processing by mountain gorillas, Byrne and Byrne 1991), individual lateralisation does not mean collective bias to one side or the other at the population level. Whether cracking nuts, fishing for termites, pounding or peeling hard-shelled fruits, most individuals in a population are significantly biased, but in about equal numbers to left or right (see Table 1).
Table 1

Laterality of hand use in object manipulation of wild chimpanzees, by individual, comparing tool use and non-tool use

No. of individuals

Behavioural pattern

Arboreal/terrestrial

Always LHa

Signific. LHb

Ambi-lateralc

Signific. RHd

Always RHe

Nf

Study site

Non-tool use

 Reach/pick up (Boesch 1991)

T

1

2

13

3

1

20

Tai

 Reach/pick up (Sugiyama et al. 1993)

T

0

1

18

0

0

19

Bossou

 Pluck food (Marchant and McGrew 1996)

A/T

0

0

32

0

0

32

Gombe

 Pluck food (McGrew and Marchant 2001)

A/T

0

1

39

0

0

40

Mahale

 Eat (McGrew and Marchant 2001)

A/T

0

0

40

1

0

41

Mahale

 Peel Saba (Corp and Byrne 2004)

A/T

13

2

5

6

12

38

Mahale

Tool use

 Wadge dip (Boesch 1991)

T

3

1

3

0

9

16

Tai

 Nut crack (Boesch 1991)

T

8

10

5

8

10

41

Tai

 Nut crack (Matsuzawa et al. 2001)

T

6

1

2

1

10

20

Bossou

 Strychnos pound (McGrew et al. 1999)

T

4

1

2

2

5

14

Gombe

 Termite fish (McGrew and Marchant 1996)

T

12

4

9

5

6

36

Gombe

 Termite fish (Lonsdorf and Hopkins 2005)

T

2

10

1

3

1

17

Gombe

 Ant fish (this study)

A

0

1

8

4

2

15

Mahale

On each row, the figure in bold is the mode

aAll bouts done with left hand (LH), n ≥ 6

bStatistically significant left-hand bias, binomial test

cNo statistically significant difference between left and right hand

dStatistically significant right-hand bias

eAll bouts done with right hand (RH)

fNumber of individuals for whom enough data were obtained for binomial testing

Table 1 lists all published studies of wild chimpanzees in which individuals can be statistically tested for classification by hand preference. Table 1 lists examples of object manipulation without tool use: five of the six show clear ambilaterality, and all represent simple food acquisition. The exception is the skilled bimanual peeling of the thick-skinned fruit of Saba florida, as detailed at Mahale by Corp and Byrne (2002). Although they found a strong sex difference in laterality, the data show no overall population bias to one side or the other. They found a similar effect in the same chimpanzees for peeling Citrus lemon (Corp and Byrne 2004).

Table 1 lists studies of tool use in food processing. The six published studies show a very similar pattern: only a few individuals are not statistically significantly lateralised to one side or the other. Particularly impressive is Lonsdorf et al.’s (2005) replication of McGrew and Marchant’s (1996) study of termite fishing in the chimpanzees of Gombe, showing an apparent, but not statistically significant left-side tendency.

Ant fishing is a form of extractive foraging in which a flexible probe made of vegetation is inserted into the tiny entry hole of a nest of wood-boring ants (Camponotus spp.) (Nishida 1973). When withdrawn, the fishing tool yields ants to be eaten by the chimpanzee. The task requires manual skill to thread the tool into the ants’ narrow tunnel in the bough of a tree, then careful extraction so that the ants affixed to the tool are not dislodged. As in termite fishing (Goodall 1964), but not in ant dipping (McGrew 1974), sensory feedback from prey to predator is tactile, not visual, hence the term ‘fishing’.

Ant fishing is done arboreally, especially in trees of Brachystegia spp. (Caesalpiniaceae), where the ants have bored into boughs or the trunk. The fishing ape may sit or stand on a nearby bough (if present), suspend itself from an overhead branch, or cling to the vertical trunk of the tree (see Fig. 1). The non-fishing hand may support none of the body weight (idle), a minority of the body weight (minor), or most of the body weight (major).
https://static-content.springer.com/image/art%3A10.1007%2Fs10329-006-0020-3/MediaObjects/10329_2006_20_Fig1_HTML.gif
Fig. 1

Various postures adopted by ant-fishing chimpanzees at Mahale Mountains National Park. Note how the non-tool-using hand is used in support and is not idle. Reprinted from Nishida (1977 )

Two previous studies have yielded laterality data on ant fishing, but both were inconclusive: Nishida (1973; Nishida and Hirawa 1982) recorded too few data to be statistically conclusive; only five individuals presented the six or more bouts needed for applying the binomial test. Norikoshi (1994, 1998) observed 201 bouts of ant fishing by 56 individuals, but pooled his data for analysis, so that statistical testing of individuals was not done. Both observers noted that chimpanzees often used both hands to ant fish. In a chapter in Japanese, Nishida (Nishida 1977, personal communication) noted that chimpanzees are forced to use the left or right hand because of the tree-trunk entry holes of the ants’ nest, relative to the nearest sitting or standing site. Norikoshi (1998) noted that arboreality constrained the expression of ant fishing because the positions of the ants’ holes made them differently accessible. We sought to test Nishida’s and Norikoshi’s hypothesis and its implications.

Methods

The study was done in the Mahale Mountains National Park, Tanzania (6°07′S, 29°44′E). M-group of chimpanzees (Pan troglodytes schweinfurthii) was studied (Nishida 1990). Data were collected on all 44 members of the group, except for seven babes in arms. From September to December 1996, the authors spent 357 h in focal-subject sampling in follows of up to 7 h, and more than 30 h in videotaping of hand use. (For additional details, see McGrew and Marchant 2001.)

Data were taken in two forms: (1) written into notebooks, then transcribed onto standardised tabulation sheets, or (2) recorded onto High-8 videotape, which was later digitised and coded. For 7 of the 15 subjects who ant fished often enough for statistical analyses (see below), there were enough data for the classification of the subjects into five categories (see below) to be compared across notes and video; 6 of the 7 (86%) showed agreement between notes and video.

We recorded only independent bouts, not non-independent events, in an effort to avoid pseudo-replication. We noted only the first performance of an act after there had been an intervening bout of another behavioural pattern, or if the subject changed hands. Thus, if an ape inserted a fishing probe six times in a row (one bout), then paused to scratch three times (one bout), then inserted the probe another four times (one bout), we scored Fish twice and Scratch once only. If an ape fished four times in a row with the left hand, then five in a row with the right hand, we counted these events as one bout of L and one bout of R. Relative frequencies were statistically tested with the binomial test, with chance set at P = q = 0.50, alpha = 0.05, two-tailed.

Each individual ape with six or more bouts of ant fishing was classed into one of five sets: AL = always used the left hand for the task; SL = statistically significantly used the left hand more often than the right; A = ambilateral, that is, frequency of left and right hand use not significantly different; SR = statistically significantly used the right hand more than the left; AR = always used the right hand.

Results

We saw 23 members of M-group fish for Camponotus ants. Fifteen of these had six or more bouts, and so were amenable to statistical analysis individually (see Table 1). No chimpanzee always fished with the left hand, but one fished significantly more often with the left. Two individuals always fished with the right hand, and four others did so significantly more often. Eight individuals used both hands and were not statistically significantly lateralised, that is, the majority were ambilateral.

The implications of Nishida’s and Norikoshi’s hypothesis are that the awkward siting of ant holes often causes fishing apes to have to shift positions in order to fish. More precisely, using one upper limb to provide a substantial amount of support of the body weight will lead to fatigue of that limb, so that it has to be rested. Such a switch of supporting hands necessitates a complementary and corresponding change of fishing hands, which leads to the following prediction: The rate of change of fishing hands should be positively correlated with the proportion of bouts in which major support was provided by the non-fishing hand.

To test this (retrospectively), we examined all 14 videotaped ant fishing sessions, only 6 of which had ≥10 bouts (range: 10–55). Of these, we classed each bout as (1) involving major support by the non-fishing hand, versus (2) minor or no support (idle) by the non-fishing hand. We ranked the six sessions by frequency of hand change, then correlated this with the proportion of major support/major + minor + none. The result was a statistically significant positive correlation (Spearman’s rank–correlation coefficient, n = 6, rs = 0.83, P < 0.05, one-tailed; see Table 2).
Table 2

Rate of changes of hand in ant fishing correlated with proportion of bouts with major support provided by non-fishing hand

Subject

Total insertions (A)

Change of hands (B)

B/A

Extent of idle/minor

Support major (C)

C/A

IW

55

0

0

55

0

0

MA

12

0

0

12

0

0

NK

10

3

0.30

3

7

0.70

FU

19

7

0.37

5

14

0.74

RB

13

6

0.46

4

9

0.69

CY

18

11

0.61

2

16

0.89

Discussion

All previous studies of tool use in wild chimpanzees found strong individual lateralisation, whether the extractive task was for animal or plant prey, or whether the skilled movements were deliberate or ballistic. All of the tasks were also done on the ground, where the seated ape had both hands free to perform a one-handed task. Here, the instructive exception is ant fishing, which differs from the other tasks in only one notable way: it is arboreal. Unless an ape can sit on a wide, horizontal bough and fish at its leisure, one of its two hands must always be used to provide support. Thus, posture and positioning can be influential variables, as originally pointed out by MacNeilage et al. (1987).

The picture for object manipulation without tools is less clear because two variables are confounded: simple versus complex task, and one- versus two-handed manipulation. (The latter distinction has been used variously, even when both hands are applied to the same object simultaneously. Two-handed object manipulation can be yoked, as in pushing a lawn mower, or complementary, as in stroking a billiards cue, and sometimes the difference is slight, as in swinging a golf club.) Until we have data on apes doing simple tasks with two hands, as in kneading dough, or complex tasks done with one hand, as in calligraphy, the relative importance of the two variables remains unclear. Artificial tasks induced experimentally in captive apes, such as peanut-butter-filled plastic tubes, gave mixed results, with some showing right-hand task specialisation (Hopkins et al. 2003) and some showing ambilaterality (Kocher and McGrew, unpublished data). What is needed are data on spontaneous, natural tasks that fill all four cells of the 2 × 2 matrix generated by these two variables.

The results of this study yield new implications for Norikoshi and Nishida’s hypothesis: Arboreality seems to be antithetical to species-level laterality of hand function, that is, handedness. Semi-terrestriality yields at least part-time opportunity for lateralisation, as the limbs are freed from keeping the ape from falling to the ground, at least for terrestrial tasks. In this light, Byrne and Byrne’s (1991) exceptional results on population-level lateralisation of foliage processing are suggestive: they came from the most terrestrial of all ape populations, the mountain gorillas of the Virunga Volcanoes. This suggests that the first data on normal laterality from the more frugivorous lowland gorillas could be telling, as they are much more often arboreal.

When we seek to apply this knowledge of apes to modeling the evolution of laterality in the human line, we are again faced with a confounding of variables. In comparison with apes who are facultatively terrestrial and quadrupedal, humans are both obligatorily terrestrial and bipedal. These two traits may be inextricable, both in terms of evolutionary origins and present-day function, but the combination is clearly unique among primates. This suggests that species-level right-handedness is a derived and not a primitive trait in the hominin line nor a stem-trait in the last common ancestor. That is, data from living apes provide a scenario for how an arboreal quadrupedal and unlateralised ancestral ape could have evolved through successive stages to become a terrestrial, bipedal and lateralised human.

Acknowledgments

We thank the Philip and Elaina Hampton Fund (Miami University) for financial support; M. Huffman, K. Kawanaka, M. Nakamura, S. Uehara for assistance in the field; T. Nishida for providing the figure; M. Huffman for providing published materials; A. Kocher and S. Russak for data analysis; K. Harrison, N. Uomini and two anonymous reviewers for critical comments on the manuscript.

Copyright information

© Japan Monkey Centre and Springer 2006