Making Holes in the Dark

Sexual selection has played a major role in the evolution of the reproductive organs of animals. A full appreciation of this aspect of sexual selection is relatively new, however, and was unknown to Darwin at the time he formulated his theories concerning the pre-copulatory aspects of selection. Here, and in the following two chapters, I consider how sexual selection has affected the evolution of reproductive anatomy and physiology in mammals. Can meaningful correlations be established between measurements of genitalic traits and the types of mating systems displayed by various mammals? If so, how do measurements of these same traits in Homo sapiens fit within the broader, comparative analyses? Can this approach help us to strengthen interpretations of the likely origins of human mating systems based upon the limited fossil evidence (e.g. concerning sexual dimorphism in body size in extinct hominids), outlined in the previous chapter?

Consideration of these questions begins here with an examination of relationships between testes weight, body weight, and mating systems, as this also provides a useful entry point to the history of studies on sexual selection and genitalic evolution in mammals, including human beings.

Primates, like other mammals, exhibit large interspecific differences in their testes sizes, in relation to adult body weight. The first anthropologist to document the extent of these differences was Adolph Schultz who, in 1938, contributed a paper to the Anatomical Record entitled 'The relative weight of the testes in primates'. This presented data on testes weights and body weights for 82 individuals, representing 18 species of monkeys, gibbons, orangutans, chimpanzees, and human males. The bulk of this information had been assembled by Schultz during the 1937 Johns Hopkins Asiatic Primate Expedition

(fifty-five specimens collected in the wild), and from specimens in the anatomical collections of Johns Hopkins University. The only relevant published account available to him at that time was by Hrdlicka (1925), which included measurements of seven adult male howler monkeys. Schultz's data on the human male comprised just three cadavers of African-American men, obtained via the Anatomy Department at Johns Hopkins.

No one who has read Schultz's original account carefully could be under the illusion that human beings (as represented by his tiny sample of humanity) have large testes in relation to body size. He noted, for example, that a bonnet macaque weighing 8 kg had larger testes than a man weighing 69 kg. Some examples selected from Schultz's paper are shown (in ascending order of body weight) in Table 2.1 These differences were not due to any decrease in the bulk of sperm-producing tissue (seminiferous tubules) in those species having larger testes. On the contrary, Schultz was able to demonstrate that 'primates with relatively large testes (chimpanzee, baboon and

Table 2.1 Adolph Schultz's (1938) measurements of testes weights In primates

Testes weight/

Body Testes body Number

Testes weight/

Body Testes body Number


weight (kg)

weight (g)



































macaque) have a considerably greater proportion of sex-cell producing glandular tissue and a smaller proportion of connective tissue than have the primates (langur, gibbon and man) with comparatively small testes.' He calculated, for example, that a macaque has seventeen times more sperm-producing tissue in its testes than a langur of similar size.

Schultz discounted age differences or seasonal effects upon testes size as explanations for the remarkable differences he had discovered. At that time it was thought (incorrectly) that none of the Old World primates were seasonal breeders. Why chimpanzees or macaques should have larger testes and produce more sperm than langurs or men was a mystery. Schultz had provided valuable data, but he remained in the dark as to their functional and evolutionary significance. The darkness was to continue for several decades.

An apocryphal story relates how Robert Louis Stephenson, as a child in Edinburgh, was watching a lamplighter illuminating the gas lamps in the street outside his house. When asked by his mother what he was doing, he replied 'I am watching a man making holes in the dark.' Where reproductive biology and sexual selection are concerned, one of the most important lamplighters has surely been Geoffrey Parker, whose paper on 'Sperm competition and its evolutionary consequences in the insects' (Parker 1970) changed forever the way biologists view this field. When a female mates with two or more males, the capacity exists for competition to occur between the gametes of rival males for access to her ova. Following this line of reasoning, Parker was able to demonstrate that Darwin's concept of sexual selection also operates at the copulatory (and post-copulatory) levels via sperm competition, and that such competition encourages the evolution of mechanisms to optimize sperm numbers in the ejaculate and to enhance fertility.

As knowledge of primate mating systems and sexual behaviour increased, largely as a result of field studies conducted in the 1960s and 1970s, so the conditions were created to apply sperm-competition theory to the questions raised by Schultz's observations on testes sizes and sperm production in monkeys, apes, and human beings.

However, Parker's light did not penetrate all areas of biology immediately. Perhaps, because his work involved insects, and most notably the dung fly

(Scathophaga stercoraria), its influence on vertebrate biologists was slow to take hold. Yet, like all truly great insights, the discovery of sperm competition has come to be applied throughout the animal kingdom, from dung flies to right whales, and from chimpanzees to humans. When Roger Short began to examine relative testes sizes in the great apes and man, working at the Medical Research Council's Reproductive Biology Unit in Edinburgh in the 1970s, he was initially unaware of Parker's work. Independently, Short reasoned that differences in testes sizes between the various apes and man might reflect variations in their mating systems and sperm production in relation to patterns of sexual behaviour. He was especially struck by the modest size of the testes in the gorilla, by comparison with those of its much smaller relative, the chimpanzee. Gorillas usually live in polygy-nous one-male units; females mate primarily with a single partner and copulations are relatively infrequent under natural conditions (Schaller 1963; Har-court et al. 1980). A silverback gorilla, weighing more than 160 kg, had testes weighing about 30 g, by comparison with a male chimpanzee which at 45 kg had much larger testes (weighing approximately 120 g).

Short, unlike Schultz, had access to information about the sexual behaviour of free-ranging chimpanzees, derived principally from the field work of Jane Goodall and her students, at the Gombe Stream Reserve in Tanzania (Goodall 1965, 1968; McGinnis 1979; Tutin 1979, 1980). Chimpanzees live in multi-male/multi-female communities and females often mate with multiple partners during the follicular phase of the menstrual cycle, when the sexual skin is swollen and ovulation is most likely to occur. In her 1986 review volume on work at Gombe, Goodall records that 'I watched one party as it arrived at a new food source: the attractive female climbed into the tree along with eight bristling males, each of whom copulated with her in quick succession in a period of five minutes.'

These observations of the chimpanzee mating system have since been confirmed by fieldwork at other sites, in Tanzania (Mahale Mountains: Nishida 1990), Ivory Coast (Ta'i Forest: Boesch and Boesch 2000), and Uganda (Budongo Forest: Reynolds 2005). Sometimes male and female chimpanzees pair up to form temporary "consortships" (Tutin 1979) and the most dominant male in a community may monopolize access to particular partners

(Goodall refers to this mate-guarding tactic as 'pos-sessiveness'). We still do not know how successful these alternative tactics are in terms of numbers of offspring sired by males across their lifespan in the community. It seems likely, however, that multiple-partner matings and selection for success in sperm competition have been powerful forces in shaping the evolution of chimpanzee reproductive biology. The increase in sperm-producing tissues in the chimpanzee testis, first measured by Adolph Schultz, makes perfect sense in this context.

Following Short's (1979) work on the great apes, Harcourt, Harvey, Larson, and Short (1981) went on to examine the question of testis weight, body weight, and breeding systems in primates, publishing a paper under this title in Nature in September 1981. Their principal findings are shown in Figure 2.1. This includes data on 33 species of monkeys and apes, as well as human beings. The findings embodied within this graph remain highly influential in discussions about sexual selection, sperm competition, and primate mating systems, so it is worthwhile to examine them in some detail. The double logarithmic plot of testes weights vs. body weights in Figure 2.1 shows that an allometric relationship exists between the two variables. In other words, testes size scales in

Relative testes size and mating system


200 100



A Polygyny

^fe Multi-male / multi-female

Anubis baboon


Pig-tailed macaque »Wra^m tatoop

Bonnet macaqu^^ * «Yellow >; Man baboprt ^^Orangutan

Crab-eating macaque 0

Howler monkey^

Rhesus macaque a amadryas baboon Gorilla Gelada

Cotton top tamarin

Spider monkey' Vervet^ 1 y J

, a A Proboscis monkey

King colobusCommon langur

Molocjhgibbo^A Silver langur

Lar gibbon^ A Dusky langur quirrel monkey

Common marmoset o

Owl monkey

1 10

Body weight (kg)

100 200


Figure 2.1 Relative testes weights and mating systems in anthropoid primates. A double logarithmic plot of combined testes weight versus body weight for anthropoids having monogamous, polygynous, or multi-male/multi-female mating systems. Source: Based on Harcourt et al. (1981); modified from Short (1985).

a predictable way with body size but there is also considerable inter-specific variability in the relationship, as indicated by the distances of various species from the regression line which passes through the 33 data points. Adult males of those species which are situated above the regression line have larger testes than expected in relation to their body weights. As well as the chimpanzee, examples include various baboon and macaque species, howler monkeys and vervets, all of which have multi-male/multi-female mating systems. Sexual selection has favoured the evolution of larger testes in these species, because females commonly mate with multiple partners during the peri-ovulatory phase of the cycle. Sperm from rival males compete for access to ova within the female's reproductive tract. Males that are capable of maintaining high sperm counts in the ejaculate may therefore gain a reproductive advantage via sperm competition. By contrast, species whose data for body weight and testes weight plot below the regression line have smaller testes than expected for this sample of primates. In general these species include polygynous (one-male unit) species such as the gorilla, gelada, proboscis monkey, and black and white colobus as well as pair-living forms, such as the lesser apes (Moloch and lar gibbons), the owl monkey, and the common marmoset of South America. Sperm competition pressures are much less pronounced in these cases than in macaques, baboons, or chimpanzees.

Human males have relatively small testes in relation to body weight; 'Man' falls just below the regression line in Figure 2.1, slightly above the orangutan, but well above the gorilla. It is relevant to note that Harcourt et al. included data on only four men in their studies of relative testes size and mating systems in anthropoid primates. Human testes weight averaged 40.5 g in these individuals, three of whom had been measured by Schultz (1938), whilst the fourth set of measurements was obtained from a paper by Benoit (1922). It may seem unusual that only four men had been sampled. However, the focus of Harcourt et al.'s study was on anthropoids in general and not human beings specifically. For fourteen of the thirty-three species included in Figure 2.1, only one, two, or three sets of adult testes weights were available, less than the information on Homo sapiens.

The huge contribution made by Harcourt et al.'s work as well as some of its limitations were appreciated at the time. Thus, in the same issue of Nature, Martin and May (1981) contributed a 'News and Views' overview in which they pointed out that seasonal changes in testes size, which certainly occur in some monkey species, had not been taken into account. The relatively large size of the testes in the cottontop tamarin (a putatively monogamous species) by comparison with another New World species, the squirrel monkey (which lives in multi-male/multi-female groups), was surprising. Martin and May also drew attention to the relatively small data set employed, which did not include any of the Malagasy lemurs or other prosimian primate species. Other factors besides testes size and sperm production might also be important in determining male reproductive success in competitive situations. Sperm must pass through the epididymis after leaving the testis, and gametes are stored in the distal region (cauda) of the epididymis prior to mating. Thus, the dynamics of sperm storage and transport required study. Males also differ tremendously in their ability to gain access to females and to maximize mating activity during the most fertile (i.e. peri-ovulatory) period of the ovarian cycle. For sperm competition to occur at all, at least two males must mate with the same female during this fertile period. The potential 'window of opportunity' for sperm competition is quite narrow in primates, including human beings. We shall return to this important problem in later chapters.

With regard to the issue of relative testes size in human beings, Martin and May (1981) commented that 'it does not accord with the range of values to be expected for a multimale breeding system.' However, there was insufficient evidence to discriminate between monogamy and polygyny as contributory factors in the evolution of human testes sizes. In practice, both types of mating system occur in present day (or recent) human societies; thus 74 per cent of the 185 societies examined by Ford and Beach (1951) engage in polygynous marriages. For this reason, Homo sapiens has been classified as a polygynous primate species in Figure 2.1.

In the years since the publication of Harcourt et al.'s study of testes sizes and mating systems in the anthropoid primates, comparative studies have been conducted on many vertebrate groups. Among the mammals, sexual selection via sperm competition correlates with the evolution of larger testes sizes among the prosimian primates (Dixson 1987a, 1995a), bats (Hosken 1997; Wilkinson and McCracken 2003), whales and dolphins (Brownell and Ralls 1986; Connor, Read, and Wrangham 2000), eutherian mammals in general (Kenagy and Trombulak 1986), as well as in marsupials and monotremes (Rose, Nevison, and Dixson 1997; Taggart et al. 1998). The larger size of the testes in primates which have multi-male/multi-female mating systems is not significantly influenced by the occurrence of seasonal breeding (Harcourt, Purvis, and Liles 1995). As an example, a restricted mating season and heightened sexual activity in seasonally breeding macaques (e.g. the rhesus monkey)

has not led to the evolution of larger relative testes sizes than in non-seasonal breeders with pronounced sperm competition (e.g. the chimpanzee).

This additional information can help us to reassess the accuracy and comparative significance of measurements of human testes size and its relevance to discussions of human evolution. Any such analysis is only as reliable as the data upon which it is based, however. It is necessary firstly to assemble a data set on testes weights and body weights for a large series of mammals and secondly to gain some clearer perception of individual and ethnic differences in human testes size. Figure 2.2 shows

Log body weight (g)

Figure 2.2 A double logarithmic plot of combined testes weight versus body weight for 339 species, representing a wide spectrum of mammals. Sources: Brownell and Ralls (1986); Kenagy and Trombulak (1986); Harcourt, Purvis, and Liles (1995); Rose, Nevison, and Dixson (1997); Connor, Read, and Wrangham (2000); Wilkinson and McCracken (2003); Anderson, Nyholt, and Dixson (2004); Dixson, Nyholt, and Anderson (2004); Anderson and Dixson, (in press).

a double logarithmic plot of testes weights versus body weights for mammals. There are 339 species represented in this graph, which includes bats, rodents, primates, carnivores, artiodactyls, peris-sodactyls, proboscideans, and cetaceans as well as marsupials and monotremes. This is a respectable sample, but as more than 4,500 extant mammalian species have been named by taxonomists, even the information in Figure 2.2 is still far from complete.

Where do human beings lie within this enlarged analysis of mammalian testes sizes and body sizes? Data on testes weights of just four men (Harcourt et al. 1981, 1995) might be sufficient to explore this question if these measurements were representative of humanity as a whole. Unfortunately this is not the case. There are considerable differences in testes sizes between individuals and between human populations around the world. These differences cannot be accounted for solely on the basis of variations in body weight (Short 1984; Diamond 1986). To provide information on possible ethnic differences in human testes size, I have assembled from the published literature measurements of testes weight or volume for more than 7,000 men in 14 countries worldwide (Table 2.2). Even this information is far from exhaustive and must be viewed with appropriate caution. Measurements of testicular volume are useful because they may be converted to testes weight by using a correction factor (volume x 1.05; the specific gravity of tissue: Dahl, Gould, and Nadler 1993). However, when testicular volumes are measured by using orchidometers, the results are subject to considerable inaccuracies. This was demonstrated by Doernberger and Doernberger (1987) in a careful post-mortem comparison of 99 men, using sonog-raphy, water displacement (Archimedes principle), Prader's orchidometer, Schirren's circle, and linear measures (by sliding caliper) to calculate testes volume. Lest the reader find these details tedious (or obsessive!) I should emphasize that exact measurements of human testes size are of great importance if sound judgments are to be made about the possible effects of sexual selection and the origin of human mating systems. Doernberger and Doernberger found that orchidometers may over-estimate testes volume (by an average of 27 per cent using Prad-er's orchidometer and by 52 per cent using Schir-ren's circle) as compared to the volumes obtained

Table 2.2 Average combined human testes weights (grams) or volumes (millilitres) in fourteen countries, worldwide





testes size

Comments Sources

China (Hong


19.01 g

Chang et al.




16.55 g

Age range Short (1984) 13-97 yrs



30.0 ml

Prader Nakamura orchidometer (1961)


35.0 ml

Schirren Fujii et al. orchidometer (1982)

South Korea


38.8 ml

Prader Kim and Lee orchidometer (1982)


31.3 ml

Prader Ku et al. orchidometer (2002)



35.4 g

Weights include Jit and Sanjeev epididymides 1991.



48.2 ml

Sliding calipers; Simmons scrotal skinfold et al. not measured (2004)



36.3 g

Mainly Johnson, Petty, Caucasian and Neaves (1984)


50.2 g

African Schultz American (1938)



46.0 ml

Prader Jorgensen orchidometer et al. (2001)



45.0 ml

Prader Jorgensen orchidometer et al. (2001)



46.0 ml

Prader Jorgensen orchidometer et al. (2001)



47.0 ml

Prader Jorgensen orchidometer et al. (2001)


40.0 ml

Prader Andersen orchidometer et al.




42.0 g

Olesen (1948)


41.5 ml

Linear measures Lambert (1951)



37.2 ml

Prader Zachmann orchidometer et al. (1974)



37.4 ml

Linear measures; Farkas (1971) scrotal skinfold not measured



50.1 ml

Ages 19-23 yrs; Ajmani, linear measures; Jain, and scrotal skinfold Saxena not measured (1985)

Note: g = weight in grams; ml = volume in millilitres.

Note: g = weight in grams; ml = volume in millilitres.

by using Archimedes principle. Caliper measures are more accurate, and a simple formula allows the calculation of testes volume using length and width measurements. However, measurements of living subjects must also take account of scrotal thickness if accurate linear data are to be obtained. Failure to do so can result in overestimation of testes volume, by approximately 25 per cent according to my own calculations. Thus in Table 2.2 I have provided notes on methodology and other problems in relation to the individual reports listed there.

Cross-culturally there is a consistent difference in mean testis size; the right testis is on average 5.5 per cent larger than the left testis in men from eight countries (Table 2.3). Other authors have commented that the right testis tends to be larger but do not provide exact data. Statistically the effect is robust (P< .01 for data in Table 2.3, after conversion of volumetric data to weights). Why the effect occurs is unknown and nor does it necessarily apply to the great apes. Unfortunately relatively few chimpanzees, gorillas, or orangutans have been measured to estimate

Table 2.3 Sizes of the right (R) and left (L) testes In various human populations









China (Hong

9.65 g

9.36 g

0.336 g


Chang et al.



8.57 g

7.98 g

0.59 g


Short (1984)

South Korea

15.9 ml

15.3 ml

0.6 ml


Ku et al. (2002)


18.20 g

17.21 g

0.99 g


Jit and Sanjeev (1991)


24.3 ml

23.9 ml

0.4 ml


Simmons et al. (2004)


19.0 g

17.3 g

1.7 g


Johnson, Petty, and Neaves (1984)


21.6 g

20.4 g

1.2 g


Olesen (1948)


19.59 ml

17.78 ml

1.81 ml


Farkas (1971)


25.66 ml

24.41 ml

1.25 ml


Ajmani et al. (1985)

Notes: ml = mean volume in millilitres; g = mean weight in grams. For information on sample sizes and methods, see Table 2.2.

Notes: ml = mean volume in millilitres; g = mean weight in grams. For information on sample sizes and methods, see Table 2.2.

individual testes sizes, but there are no consistent effects for the available samples.

Turning to measurements of combined testes size, there is a strong trend towards the occurrence of smaller testes in men from Asia (China, Japan, Korea, and perhaps in India) as compared to measurements from European and African populations, or their ethnic derivatives (e.g. in the USA and Australia). These differences are quite striking with respect to the information presented in Table 2.2. Smallest by far are the testes of 209 Hong Kong Chinese, weighed in two separate studies. Chang et al. (1960) weighed the testes of 100 men and found the mean combined weight to be only 19.01 g. Short (1984) reported even lower average weights (16.5 g) for 109 Hong Kong males on the basis of an unpublished study (Chan and Short). However, as these subjects varied considerably in age, from 13 to 97 years, it is probable that the inclusion of some adolescent males would have produced a slightly lower average for testes weight. In neither of these studies of Chinese males are their body weights known, but it is highly unlikely that smaller stature alone could account for such small testes sizes by comparison with data for European or African populations. Chan and Short's unpublished study does provide data on the height of the Hong Kong males for whom testes weights were calculated. The average height of their subjects was 157.9 cm (SEM± 1.5 cm). An interesting comparison is provided by Ajmani, Jain, and Saxena's (1985) measurements of Nigerian men. Part of their sample comprised 110 men who were less than 165 cm in height. Combined testes volumes averaged 43.9 ml in this group, which is equivalent to 46.0 g for average weight of the testes. Even if we allow for an overestimate of testicular volume in this study (due to a failure to control for scrotal thickness: see Table 2.2), Nigerian men have testes more than twice as large as those of Hong Kong Chinese subjects of similar height.

Although differences in body size may make some small contribution to the ethnic variations in testes size listed in Table 2.2, it is much more likely that fundamental differences in testes size exist in different human populations. Next smallest are the testes of Japanese men (averaging 30 ml and 35 ml in studies by Nakamura (1961) and Fujii et al. (1982)). The same is true for studies of Korean men. For all the studies in Japan and Korea, measurements were made using orchidometers, and may thus represent overestimates. In India, a study of 325 men yielded an average combined testes weight of 35.4 g but this included the epididymis and hence is also an overestimate. By comparison, the weights or volumes of testes of men in UK, France, Scandinavia and other parts of Europe, as well as in the USA, Australia, and Nigeria are larger on average than for men in those Asian populations for which data exist. Largest of all are the testes of 209 Nigerian medical students (50.1 ml, Ajm-ani et al. 1985) and the three black African Americans measured originally by Schultz (50.2 g).

In Figure 2.3 these data on human testes size in various populations have been log transformed and plotted against body weight in the same manner as for the much larger sample of mammalian species represented in Figure 2.2. The same regression line, calculated for mammals as a whole, has also been plotted on Figure 2.3. The data for the various human populations (represented by circles) are aligned vertically, below the regression line (exact weights of the men in each population are unknown, so an average weight has been used). Smallest of all are the relative testes sizes of Hong Kong Chinese; the circle at the foot of the row, which does not overlap with the others, represents this population. At the opposite end of the row, and closest to the regression line, is the circle representing relative testes size in Nigerian men. Other populations fall between these two extremes. How do these more

Figure 2.3 Relative testes sizes in human populations, as compared to the great apes. The regression line is taken from Figure 2.2. Data are from Tables 2.2 and 2.4. o = Homo; ◊= Gorilla (the western lowland gorilla is plotted separately and falls below the mountain gorilla on the graph); □ = Pongo (two species plotted); A = Pan (two species plotted).

Figure 2.3 Relative testes sizes in human populations, as compared to the great apes. The regression line is taken from Figure 2.2. Data are from Tables 2.2 and 2.4. o = Homo; ◊= Gorilla (the western lowland gorilla is plotted separately and falls below the mountain gorilla on the graph); □ = Pongo (two species plotted); A = Pan (two species plotted).

extensive data on humans compare with those for the apes?

More up-to-date information on relative testes sizes in the apes has also been assembled (Table 2.4) and the great apes have been included in Figure 2.3, for comparison with Homo sapiens. Among the smaller, pair-forming (monogamous) gibbons, adequate data on testes size and body weight are only available for two species (Hylobates lar and H. moloch). They have small testes in relation to body weight, as originally reported by Harcourt et al. (1981). The term 'monogamous' does not necessarily denote a lifelong union and exclusive copulation with a single partner. Extra-pair copulations have been documented in H. lar (Palombit 1994; Reichard 1995). The genetic consequences of such behaviour in terms of offspring sired by males outside the family group are unknown, as no detailed DNA typing study of wild gibbon groups has yet been reported. The prediction is that extra-pair paternity

Table 2.4 Testes weights and body weights of the adult great apes






Number of weights











Dixson et al.




Pongo abelii




Dahl, Gould,


and Nadler






Schultz (1938);



Dixson and



Pan paniscus




Dixson and



(2004) and



Gorilla g.

Three (wild)









G. g. gorilla








rates are likely to be very low and that sperm competition has had little effect upon the evolution of reproductive physiology in gibbons. Males make a considerable investment in maintaining long-term relationships with individual females in these species and their reproductive success depends upon siring offspring and raising them within the family group. Their low relative testes size is commensurate with low sperm-competition pressure.

Even smaller, relative to their huge body sizes, are the testes of silverback male gorillas. In Table 2.4 and Figure 2.3 the mountain gorilla (Gorilla g. beringei) and the western lowland gorilla (G.g. gorilla) are shown separately. There are anatomical and ecological differences between these two forms, and some taxonomists accord the western lowland gorilla a specific rank. Mountain gorilla groups may sometimes include more than one sil-verback male, although there is usually a clearly dominant or leader male even in these circumstances (Schaller 1963). DNA typing studies of mountain gorilla groups have demonstrated that when two adult males are present, the dominant silverback sires 85 per cent of the offspring (Bradley et al. 2005). No comparable studies have been carried out on western lowland gorillas. Their groups normally contain a single silverback however, and there is less likelihood that sperm competition might occur. It is interesting that mountain gorillas also have relatively larger testes than western lowland gorillas. However, this finding may represent a sampling bias in the data. Only the testes of captive western lowland gorillas have been measured thus far and some captive males are infertile and exhibit testicular atrophy (Dixson, Moore, and Holt 1980; Enomoto et al. 2004). Individuals displaying extreme testicular pathology have been excluded from Table 2.4. Nonetheless, it is still possible that captive silverback western lowland gorillas are less fertile and have smaller testes than their counterparts in the wild. By contrast, the testes weights listed for mountain gorillas refer to free-ranging silverbacks. Perhaps their testes were larger due to better health and nutrition under natural conditions, or perhaps a true difference in relative testes size exists between the two taxa. More data, especially on free-ranging western lowland gorillas, will be required to resolve this question.

Gorillas are fundamentally polygynous in their mating behaviour and the differences between western lowland and mountain gorillas in relative testes sizes are minor and consistent with a mating system in which sperm competition plays a minimal role. By comparison, differences between gorillas and chimpanzees in relative testes sizes are dramatic, and it is now possible to confirm that the second chimpanzee species (the pygmy chimp or bonobo: Pan paniscus) also has extremely large testes in relation to body size (Table 2.4, Figure 2.3). Unlike chimpanzees, which may form consortships and engage in mate guarding (by alpha males: Goodall 1986) in addition to frequent multi-partner matings, the bonobo mating system is primarily a multi-partner arrangement in which sperm competition between rival males is likely to be intense. The fact that both P. troglodytes and P. paniscus are very similar in relative testis size indicates that sperm competition via multiple partner matings by females is the primary force which has driven selection for their exceptionally large testes. This in turn implies that male reproductive success resulting from consortships and mate guarding in P. troglodytes will likely be smaller, and ancillary to the major investment in sperm competition tactics. This question awaits resolution by detailed DNA typing studies to compare the success of the various sexual strategies employed by male chimpanzees across their lifespan. Where the bonobo is concerned, a DNA typing study of paternity in free-ranging animals by Gerloff et al. (1999) reported that dominant males, which are usually the sons of high-ranking females, have the greatest reproductive success. It must be said, however, that differences in social rank among male bono-bos are not pronounced or linear. On the basis of agonistic interactions and displacement behaviour, Gerloff et al. placed three males in rank category 1, and another three males in the second ranking position. Although these authors state that 'two dominant males together attained the highest paternity success', the observed relationships between male rank and paternity were not pronounced. One of the rank 1 males sired no offspring, whereas all the rank category 2 males had fathered at least one infant. There were at least six potential sires in the group for the eight infants that were DNA typed. These results are more consistent with a multiple-

partner, sperm-competition-driven mating system than with one governed by male dominance in relation to reproductive success. On that note, all eight males studied had high copulatory frequencies (ranging from 0.22-1.29 copulations observed per day) and none were completely excluded from mating with females in the community.

These observations of free-ranging bonobos receive support from recent studies of a captive group, housed at Twycross Zoo in the UK (Marvan et al. 2006). Two infants in the group were sired by two lower-ranking males, whereas the most dominant male in social contexts did not sire any offspring with the group's three adult females.

Turning to the orangutan, some refinements are necessary to the analyses of their relative testes sizes, compared to earlier accounts (Harcourt et al. 1981, 1995). Two species of orangutan are now recognized by some taxonomists (Pongo pygmaeus in Borneo, and P. abelii in Sumatra). Chromosomal and morphological differences between the two forms are significant, but some authorities remain doubtful about separating Bornean and Sumatran orangutans as distinct species (e.g. Courtenay, Groves, and Andrews 1988). Additional data on testes sizes are available for both types of orangutan, and these are included in Table 2.4. The number of specimens measured is still very small, however. I have followed Dahl, Gould, and Nadler (1993) by including only data for adult males weighing more than 85 kg. Males that are smaller than this may not be completely physically mature.

The social organization and mating system of orangutans is most unusual among the anthropoid primates, for they are comparatively nongregari-ous. Massive adult males, with prominent secondary sexual traits (cheek flanges, throat sac, and long hair) occupy very large, individual home ranges. Their world is centred within the forest canopy, for they only occasionally descend to ground level. Their large arboreal ranges overlap those of a number of adult females and their dependant offspring. Orangutans are the largest arboreal mammals in the world, and although they may congregate occasionally when favoured trees are in fruit, the sexes are normally separate and widely dispersed in the rainforest. Dominant flanged adult males are highly antagonistic to one another, and subordinate males exhibit socially mediated suppression of body growth and secondary sexual traits (Kingsley 1982; Maggioncalda et al. 1999, 2000). This alternative reproductive strategy may be maintained for years in the wild, until some opportunity arises for a non-flanged male to become dominant and transition to full physical development. Non-flanged adults attempt to mate coercively with females they encounter (Mackinnon 1974; Galdikas 1985a, 1985b). DNA typing studies have provided evidence that reproductive success is similar in non-flanged and flanged male Sumatran orangutans (Utami et al 2002; Utami Atmoko and van Hooff 2004), although only eleven infants were included in paternity analyses (Figure 2.4).

Given their widely dispersed social system, and slow rates of travel through the forest canopy, it is most unlikely that female orangutans encounter large numbers of adult males during the few days surrounding ovulation. There is some evidence that females prefer to approach and associate with a fully flanged male at this time (Schurmann 1982). Adult males are aggressive towards each other, and their impressive great calls and snag-crashing displays probably serve to maintain their spatial separation (Galdikas 1983; Mitani 1985). Non-flanged

I Do

males, being more mobile and less conspicuous, can engage in additional copulations, however, so that some potential for sperm competition exists, with respect to the orangutan mating system.

In Figure 2.3 the relative testes sizes of both the Bornean and Sumatran orangutan fall just below the regression line for mammals in general, and are in general no smaller than those recorded for humans. Although the relative testes sizes of some men (e.g. in Nigeria) exceed those of orangutans, others do not, and are smaller in relation to body size (e.g. in Hong Kong, and in Japan). My interpretation of these more detailed comparisons is essentially the same as my view of Harcourt et al.'s (1981, 1995) original findings. Human testes sizes are unexceptional and consistent with an evolutionary history which involved pair formation or polygyny as the principal mating system. Sperm competition pressure would have been low under these circumstances. However, other authors including a number of evolutionary psychologists have interpreted differences between the relative testes sizes of men, the gorilla, and orangutan in a different way. They emphasize that human testes are larger than those of the gorilla and orangutan, and that sperm competition has shaped the evolution of human sexual

Wi nz

Wi nz

^ i i i | i i i i | i i i i | i i i i | i i i i | i i r

1971 1975 1980 1985 1990 1995

Figure 2.4 Paternity in free-ranging (flanged and non-flanged) male orangutans, as revealed by DNA typing studies of offspring and their potential sires. Both flanged and non-flanged males sired similar numbers of offspring. Two adult males, Do (Dobra) and Bo (Boris), developed from non-flanged to full expression of the cheek flanges and other secondary sexual traits during the course of the study.

Source: After Utami Atmoko and Van Hooff (2004).

behaviour (Smith 1984; Baker and Bellis 1995; Miller 2000; Buss 2003; Rolls 2005; Pound, Shackelford, and Goetz 2006). However, I believe this to be a case of wishful thinking; especially in the light of more detailed comparisons of human and great ape relative testes sizes (Figure 2.3). Yet, the issue will probably never be resolved by considering relative testes size in isolation. Relative testes size is, after all, only a crude index of sperm competition pressure. In the next chapter, comparative evidence on a much wider range of anatomical and physiological traits will be sifted in order to achieve a better perspective of the probable evolutionary basis of human mating systems and sexual behaviour.

An extreme view of the importance of human sperm competition is due to Baker and Bellis (1995), who claimed that sexual selection, via sperm competition, has played a crucial role in human evolution. As we shall see in the next two chapters, much of this work has been shown to be seriously flawed. Baker (1997) also reported that men rumoured to be more likely to engage in extrapair copulations had larger testes than those who were not rumoured to behave in this way. Yet in much larger and more carefully controlled studies, Simmons et al. (2004) found that men who reported engaging in extrapair copulations had, on average, smaller testes than men who did not (the difference was not statistically significant). There was a significant correlation between combined testes volume and sperm numbers per ejaculate in 50 men who provided the required testes measurements and semen samples. Simmons et al.'s interpretation of their findings is that human beings have larger testes than expected for a monogamous species. However, I do not think that the data (e.g. as presented in Figures 2.1 and 2.3) support this conclusion.

How frequent are extrapair copulations among human beings, and how often do they result in pregnancies? These are exceedingly difficult questions to answer, especially in view of the covert nature of such behaviour, and the cultural differences in sexual attitudes which pertain in different parts of the world. Events in modern-day New York or Paris probably bear little or no relation to patterns of extrapair copulation and paternity in remote ancestral populations of Homo sapiens or their African precursors. In their review of five studies conducted in UK, France, Australia, and the USA, Simmons et al. (2004) record that between 6.9-44 per cent of men and 5-51.7 per cent of women report having engaged in extrapair copulations. For people under 30 years of age the range is approximately 5-27 per cent. Of course, very few of these copulations may result in pregnancies and the birth of offspring. Rates of extrapair paternity in various human populations may range from as low as 0.03 to 11.8 per cent (see Table 2.5). The median value is 1.82 per cent, but very few studies have involved non-random samples of subjects or precise techniques to pinpoint paternity of offspring. The study by Sasse et al. (1994), conducted in Switzerland, involved 1,607 subjects and employed DNA fingerprinting techniques. Extrapair paternity was very low (0.7%) in this case. By contrast, Chagnon (1979), in a most interesting

Table 2.5 Rates of extrapair paternity in human populations


% Extrapair paternity



Michigan, USA



Schacht and



Detroit, USA



Potthoff and






Peritz and Rust







Ashton (1980)




Le Roux et al.





Sasse et al.


West Middlesex



Edwards (1957)

Sykes family, UK



Sykes and Irven





Brock and



Nuevo León,





et al. (1999)

South America,



Chagnon (1979)



Source: From Simmons et al. (2004).

Source: From Simmons et al. (2004).

study of the Yanomamo Indians of South America, records extrapair paternity frequencies of 10 per cent for 132 cases, based upon blood group comparisons (Table 2.5).

Thus, rates of extrapair paternity are, in general, very low for the few human populations that have been sampled. The data are not sufficient to justify generalizations about the occurrence, or importance, of sperm competition in human reproduction. I shall return to this subject in Chapters 5 and 6, when the evolution of human copulatory behaviour is discussed in greater detail.

Ethnic differences in human testes sizes, discussed previously and summarized in Table 2.2, may result from a number of causes. Not all morphological differences between human populations are necessarily the result of natural, or sexual selection. As human beings dispersed across the globe, relatively small founder populations may have given rise to much larger numbers of modern day descendants, at least in some areas of the world. A limited founder gene pool may give rise to significant differences in morphological and physiological traits due to genetic drift. Genes may also influence a variety of pleiotropic effects. Mayr (1963) pointed out that 'a gene elaborates a gene product, which may be utilized in the differentiation of several organs (pleiotropy), and, conversely, any one character may be affected by many genes (polygeny).' In this regard it is interesting to consider the genetic link between gonadal size and fertility in males and females of the same species. Research on mice and sheep has demonstrated a connection between the occurrence of higher ovulation rates in females and larger testes sizes among males belonging to the same genetic strain. Short (1984) extended these findings on animals to propose that ethnic differences in human testes size might correlate with frequencies of dizygotic twinning among females belonging to the same populations. Smaller testes sizes of Hong Kong Chinese might thus be linked to lower frequencies of non-identical twin births among Chinese women, as compared to women in Africa or Europe. In Table 2.6, I have assembled data on human twinning rates (from Short 1984 and Diamond 1986) for comparison with the much larger body of data which is now available on testes sizes

Table 2.6 Human testes sizes and frequencies of dizygotic twinning: Data for nine countries

Dizygotic twinning rates

Combined testes

(per 1000


weight (g)*




















43.6) 46.6 49.6)










32.9) 36.8












Q 1


31.5) 34.1


8.1 2.3



China (Hong

19.0) 17.7






*Testes weights from Table 2.2; volumes converted to weights by using a correction factor (x 1.05).

Source: Dizygotic twinning rates from Short (1984) and Diamond (1986).

*Testes weights from Table 2.2; volumes converted to weights by using a correction factor (x 1.05).

Source: Dizygotic twinning rates from Short (1984) and Diamond (1986).

for human populations in nine countries. There is a positive correlation between dizygotic twinning rates and testes sizes (Spearman correlation coefficient = 0.8, P< . 02). A word of caution, however, relates to the earlier discussion of the limits of error for some of these data, and especially those obtained by using orchidometers to assess testes volumes (Table 2.2). Unfortunately, if one attempts to correct for the various possible errors of measurement, using for example Doernberger and Doernberger's (1987) comparative studies of volumetric methods, then the correlation between dizygotic twinning rates and testes sizes shown in Table 2.2 loses statistical significance. Until the task of obtaining truly accurate data on testes weights for a large enough sample of human populations has been accomplished, some doubt remains as to whether the correlation with twinning rates is a genuine effect. The available data are suggestive of such a relationship. Might it be possible that, as human populations spread from Africa to Asia, and human physiques changed, twinning rates underwent negative selection due to the smaller sizes of women and higher mortality rates caused by multiple births? This possibility is worthy of further study, and provides a feasible alternative to notions of decreased sperm competition and reductions in relative testes sizes in Asia.

Twinning occurs in some other primate species besides Homo sapiens. It is especially prevalent among the marmosets and tamarins (Callitrichi-dae) of South America. The cotton-topped tamarin, which Harcourt et al. (1981) reported as having surprisingly large testes (see Figure 2.1), normally gives birth to dizygotic twins. Larger than expected testes sizes in some tamarins and marmosets may be linked to the genetic predisposition to produce larger gonads and higher levels of reproductive hormones in both sexes of these monkeys.

0 0

Post a comment