Hominin life history: reconstruction and evolution

Abstract

In this review we attempt to reconstruct the evolutionary history of hominin life history from extant and fossil evidence. We utilize demographic life history theory and distinguish life history variables, traits such as weaning, age at sexual maturity, and life span, from life history-related variables such as body mass, brain growth, and dental development. The latter are either linked with, or can be used to make inferences about, life history, thus providing an opportunity for estimating life history parameters in fossil taxa. We compare the life history variables of modern great apes and identify traits that are likely to be shared by the last common ancestor of Pan-Homo and those likely to be derived in hominins. All great apes exhibit slow life histories and we infer this to be true of the last common ancestor of Pan-Homo and the stem hominin. Modern human life histories are even slower, exhibiting distinctively long post-menopausal life spans and later ages at maturity, pointing to a reduction in adult mortality since the Pan-Homo split. We suggest that lower adult mortality, distinctively short interbirth intervals, and early weaning characteristic of modern humans are derived features resulting from cooperative breeding. We evaluate the fidelity of three life history-related variables, body mass, brain growth and dental development, with the life history parameters of living great apes. We found that body mass is the best predictor of great ape life history events. Brain growth trajectories and dental development and eruption are weakly related proxies and inferences from them should be made with caution. We evaluate the evidence of life history-related variables available for extinct species and find that prior to the transitional hominins there is no evidence of any hominin taxon possessing a body size, brain size or aspects of dental development much different from what we assume to be the primitive life history pattern for the Pan-Homo clade. Data for life history-related variables among the transitional hominin grade are consistent and none agrees with a modern human pattern. Aside from mean body mass, adult brain size, crown and root formation times, and the timing and sequence of dental eruption of Homo erectus are inconsistent with that of modern humans. Homo antecessor fossil material suggests a brain size similar to that of Homo erectus s. s., and crown formation times that are not yet modern, though there is some evidence of modern human-like timing of tooth formation and eruption. The body sizes, brain sizes, and dental development of Homo heidelbergensis and Homo neanderthalensis are consistent with a modern human life history but samples are too small to be certain that they have life histories within the modern human range. As more life history-related variable information for hominin species accumulates we are discovering that they can also have distinctive life histories that do not conform to any living model. At least one extinct hominin subclade, Paranthropus, has a pattern of dental life history-related variables that most likely set it apart from the life histories of both modern humans and chimpanzees.

Introduction

Compared to other great apes modern humans have a higher rate of survival, live longer, start reproducing later, and have shorter interbirth intervals (reviewed in Leigh 2001; Robson et al. 2006). To reconstruct the recent evolution of these characteristics of modern human life history we review the life histories of closely related extant and fossil taxa. We also discuss the probable life histories of (1) the hypothetical last common ancestor (LCA) of the chimpanzee/bonobo and modern human (Pan-Homo) clade, (2) the hypothetical stem hominin taxon, (3) the taxa that make up the major grades within the hominin clade, and (4) the evolution of life history within the major subclades within the hominin clade. Comparing the life history of the living primates most closely related to modern humans enables researchers to generate hypotheses about what modern human life history traits are conserved and which are derived.

Direct evidence about non-human great ape life history has been gleaned by meticulous observation both in the field and from captive animals (see Kappeler & Pereira, 2003; van Schaik et al. 2006). These data, combined with molecular and other information about how their phylogenetic histories are related (see Bradley, 2008), contributes to reconstructing the life history of the LCA of the Pan-Homo clade. But in order to investigate the more recent evolutionary context of modern human life history, researchers must examine whatever evidence is available about the life history of closely-related extinct animals. If we make the untested assumption (see below) that the common ancestor of the Pan-Homo clade had a life history that is more like that of modern chimpanzees than that of modern humans, we must look at the fossil evidence of creatures that are more closely related to modern humans than to Pan(that is the hominin clade) to investigate the recent evolution of modern human life history.

Inferences about the life history of extinct hominin taxa must be extracted from fossilized remains of the hard tissues. Even this indirect information about the life history of fossil hominins is useful. If the taxon is directly ancestral to modern humans (but see Wood & Lonergan, 2008; for the reasons why this hypothesis is difficult to test and verify for most early hominin taxa) it provides evidence about an earlier stage in the evolution of modern human life history. If the taxon belongs to an extinct hominin subclade it might help throw light on the factors that determine and constrain how life history is configured more widely within the hominin clade.

In this contribution we have two primary aims: first to reconstruct the recent evolutionary history of hominin life history from extant and fossil evidence, and second to assess when, in what taxon or taxa, and at what pace, the distinctive components of modern human life history appear within the hominin clade. In the first section of our contribution we compare the life histories of the living great apes (orangutans, gorillas, chimpanzees, bonobos and modern humans) to identify traits that are likely to be derived in hominins, and thus suggest the likely life history of the Pan-Homo LCA, and the stem hominin. We distinguish life history variables (LHVs), traits such as age at weaning, age at sexual maturity, and life span that can only be measured in living populations, from life history-related variables (LHRVs). The latter are variables that can be used to make inferences about life history. Given the inability to collect standard life history data from fossil material, we evaluate how well three LHRVs, body mass, brain size and dental development, serve as accurate proxies for the timing of life history events in the extant great apes.

In the second section we address how different taxonomic schemes influence the analysis of hominin life history patterns by using both a relatively speciose (or ‘splitting’) taxonomy, as well as a less speciose (or ‘lumping’) taxonomy (see Wood & Lonergan, 2008). We then summarize what can be deduced about the evolution of the major elements of life history within the hominin clade. This includes an assessment of when, and in which taxa, the distinctive aspects of modern human life history make their appearance.

Finally, we consider the implications of these data for hypotheses about the first appearance of a modern human-like life history and evaluate how well the hominin fossil evidence supports the predictions made using comparative primate data. Specifically, we address three key questions: (1) Did the unique features of modern human life history appear suddenly as one integrated package, or did the components evolve independently and incrementally? (2) Did the onset of modern human life history coincide with the appearance of larger-bodied hominins with a modern human skeletal proportions, or did it appear later in hominin evolution? (3) Are modern human and modern chimpanzee life histories the only ways that life history has been configured within the Pan-Homo clade, or is there evidence within the fossil hominin record of creatures that have a different life history pattern?

Part I. Life history and life history-related variables of extant hominids

All organisms pass through major life stages and life history theory seeks to explain cross-species differences in the timing and covariation of these stages. It has been well established across a broad array of species that the timing of major life events tends to be correlated, even when the effects of body size are removed (Harvey & Read, 1988; Read & Harvey, 1989). A shift in the timing of one event results in a concordant extension or compression in the span between the occurrence of other events (Charnov, 1991). Primates in general, and great apes in particular, have slow life histories, with comparatively long life stages: late ages at maturity, low birth rates with small litter sizes, and long adult life spans (Charnov & Berrigan, 1993). The pace of life history is largely determined by age-specific mortality rates. Generally, species that suffer high rates of adult mortality, that is, a high probability of dying during one's reproductive years, tend to have fast life histories, whereas those with low adult mortality exhibit slower life histories (Harvey et al. 1989). Shifts in adult survival or mortality risk alter the pace of linked life history events, and also the constraints important for optimizing growth and development (Hawkes, 2006a).

Many published lists of life history variables are conflations of two different categories of information (Skinner & Wood, 2006), which we distinguish in Table 1. The first category (A) consists of variables such as gestation length, age at weaning, longevity, interbirth interval, and age of first and last reproduction. These variables reflect population vital rates and the timing of life history events, and we will refer to these as ‘life history variables’ (or LHVs). With the possible exception of weaning (Humphrey et al. 2007), we cannot yet make direct observations about life history variables on extinct taxa and thus we are reduced to making inferences about life history from qualitative or quantitative information about ontogeny gleaned from the hominin fossil record. This second category (B) consists of variables such as body mass and brain size (e.g. Sacher, 1975; Martin, 1981; Martin, 1983; Hofman, 1984; Smith, 1989, 1992; Smith & Tompkins, 1995; Smith et al. 1995; Godfrey et al. 2003) that have been shown empirically within extant primates to be constrained by, or correlated with, LHVs. To distinguish them from first-order life history variables we follow Skinner & Wood (2006) and refer to the second-order category B variables as ‘life history-related variables’ (LHRVs).

Table 1.

Life history and life history-related variables and their present availability for extinct taxa

	Available for extinct taxa*
Life history variables (LHVs)
Gestation length	No
Age at weaning	No?
Age at first reproduction	No
Interbirth interval	No
Mean life span	No
Maximum life span	No
Life history-related variables (LHRVs)
Body mass
Adult	Yes
Neonatal	Yes???
Brain mass†
Adult	Yes
Neonatal	Yes???
Dental crown and root formation times	Yes?
Dental eruption times	Yes?

^*Availability designated as ‘Yes’ means that reasonable sample sizes (but not necessarily reliable estimates) are available for most taxa; ‘Yes?’ means that it is possible to collect data for this variable from the fossil record but sample sizes are currently too small to be meaningful for many taxa; ‘Yes???’ means that it is theoretically possible to get data for this variable in the fossil record, but sample sizes may never be large enough to make meaningful inferences.

^†Estimated from endocranial volume in extinct taxa.

We examine first what LHV data are available for the extant great apes, focusing solely on females for several reasons. Female fertility rates and mortality rates determine population growth and age structure and are typically slower than male potential reproductive rates. Males must compete for paternity opportunities set by female fertilities, a limitation that has important consequences for male life histories, especially with respect to reproductive strategies (Kappeler & Pereira, 2003). In addition, many important life history variables are either restricted to females (such as gestation length, lactation, and interbirth intervals) or are difficult to ascertain for males (such as parity). We then consider in more detail how (and, more importantly, how reliably) LHRVs can be inferred from the evidence provided by the hominin fossil record.

Which apes resemble the first hominins?

Modern humans are part of the wider radiation of great apes as shown in Fig. 1. We follow the standard two species taxonomy for our closest living relatives in the genus Pan: the common chimpanzee (Pan troglodytes) and the bonobo (Pan paniscus). Although differences between the three chimpanzee subspecies are small (Fischer et al. 2006), recent evaluation of genetic differences among chimpanzees supports the traditional taxonomic designation of three geographically distinct lineages (Becquet et al. 2007). The two other non-human great apes, gorillas and orangutans, are currently in a state of taxonomic flux. Gorillas were traditionally classified as a single species with various distinct subspecies, but recently the eastern and western gorilla populations have been accorded species status as Gorilla gorilla and Gorilla beringei, respectively (Groves 2001, 2003; Thalmann et al. 2007). Similarly, two species are recognized within the orangutan genus Pongo, Pongo pygmaeus from Borneo and Pongo abelii from Sumatra (Zhang et al. 2001). While these revisions recognize important species differences within orangutans and gorillas, there are insufficient species-specific long-term life history data to justify us distinguishing them for the purposes of our review, so we pool available life history data on chimpanzees, gorillas and orangutans and deal with these taxa at the generic level.

Fig. 1.

Phylogenetic relationships of the extant great ape species. Estimated time of divergence of the hominid lineage from Glazko & Nei (2003), for chimpanzee/bonobo from Wildman et al. (2003), for the Bornean/Sumatran orangutans from Zhang et al. (2001), and for the eastern/western gorillas from Thalmann et al. (2007).

To use empirical data about the life history of the living great apes to reconstruct the life history of the most recent common ancestor of the Pan-Homo clade, or the life history of the stem hominin, we must make the untested assumption that the life histories of the non-human great apes have undergone relatively little evolution of their own. There is some support for this assumption, for the molecular and morphological similarities among the great apes suggest they have been more conserved than the hominin radiation (Moore, 1996). On the other hand, many assume that some parallel evolution has taken place in the African hominoid lineages, especially with respect to their locomotion. Because chimpanzees and gorillas are terrestrial knuckle-walkers, it has long been considered parsimonious to consider our common ancestor was, too. The wrist morphology of early hominins apparently displays features similar to those seen in our knuckle-walking great ape relatives (Richmond & Strait, 2000), thus supporting this assumption, but a recent examination of the locomotor biomechanics among extant higher primates suggests that hominin bipedalism may have evolved independently from an arboreal ancestor (Schmitt, 2003; Thorpe et al. 2007; Crompton et al. 2008). Given the general correlation between terrestriality and faster life history (van Schaik & Deaner, 2003), and the evidence that the African great apes became more terrestrial over time, it may be argued that the late Miocene ancestors of the Pan-Homo clade probably had slower life histories. If this is the case, the still strictly arboreal orangutan may prove the best extant model for the life history of the earliest hominins. If the African apes did not evolve independently, then the earliest hominins most likely had a life history similar to that of our closest living relatives, the chimpanzee and the bonobo. If they did evolve independently, the best living model would be closer to that of the more arboreal orangutan.

Adult body size is the result of both the duration and rate of growth prior to maturity. Primates on average grow more slowly than other mammals and are therefore smaller compared to non-primate mammals of similar ages at first reproduction. Modern humans, chimpanzees, bonobos, and, orangutans grow even more slowly than the primate average (Blurton Jones, 2006). But this is not true of gorillas; they grow faster than the other great apes, including ourselves. Differences in growth rates across mammals are closely tied to differences in the rate they produce offspring (Charnov, 1991; Charnov & Berrigan, 1993). Gorillas grow more quickly and produce offspring at shorter intervals than do the other non-human great apes (Table 2, see Robson et al. 2006 for discussion).

Table 2.

Primary life history variables of female great apes, mainly for wild populations compared to those of modern humans, mainly foragers

Species	Maximum life span (years)	Age at first birth (years)	Gestation length (days)	Age at weaning (years)	Interbirth interval (years)	Age at last birth (years)	Adult female body mass (kg)
Orangutan (Pongo sp.)	58.7*	15.6§	260§§	7.0¶	8.05§	> 41§	37.81‡‡‡‡
Gorilla (Gorilla sp.)	54*	10.0¶	255§§	4.1***	4.40¶	< 42¶¶¶	95.2‡‡‡‡
Bonobo (P. paniscus)	50.0+†	14.2**	244¶¶		6.25†††		33.35‡‡‡‡
Chimpanzee (P. troglodytes)	53.4*	13.3††	225§§	4.5¶	5.46‡‡‡	42****	35.41‡‡‡‡
Modern human (H. sapiens)	85‡	19.5‡‡	270§§	2.8¶	3.69§§§	45††††	45.5§§§§

^*Judge & Carey (2000).

^†Erwin et al. (2002).

^‡Hill & Hurtado (1996); Howell (1979); Blurton Jones et al. (2002).

^§Wich et al. (2004).

^¶Alvarez (2000); for modern humans, only included data from two foraging populations, the Ache and !Kung.

^**Kuroda (1989).

^††Average age at first birth for five P. troglodytes populations: Bossou (10.9 years) Sugiyama (2004); Gombe (13.3 years) Wallis (1997); Mahale (14.56 years) Nishida et al. (2003); Tai (13.7 years) Boesch and Boesch-Achermann (2000); Kibale (15.4 years) Wrangham in Knott (2001).

^‡‡Average age at first reproduction from four modern human foraging groups: Ache (19.5 years) Hill & Hurtado (1996); !Kung (19.2 years) Howell (1979); Hadza (18.77 years), Blurton Jones (unpublished data); Hiwi (20.5 years) Kaplan et al. (2000).

^§§Harvey et al. (1987).

^¶¶de Waal & Lanting (1997): 190 report median gestation length for bonobos in captivity from Thompson-Handler et al. (1990).

^***Average of median age at last suckle of both species: G. gorilla(4.6 years) Nowell & Fletcher (2007); G. beringei (3.6 years) Fletcher (2001).

^†††Average of two P. paniscus populations: Wamba (4.5 years) Takahata et al. (1996); Lomako (8.0 years) Fruth in Knott (2001).

^‡‡‡Average interbirth interval of five P. troglodytes populations: Bossou (5.3 years) Sugiyama (2004); Gombe (5.2 years) Wallis (1997); Mahale (5.6 years) Nishida et al. (2003); Tai (5.7 years) Boesch & Boesch-Achermann (2000); Kanywara, Kibale (5.4 years) Brewer-Marsden et al. (2006); Budongo (5.6 years) Brewer-Marsden et al. (2006).

^§§§Average modern human interbirth interval averaged from three foraging groups: Ache (3.2 years) Hill & Hurtado (1996); !Kung (4.12 years) Howell (1979); Hiwi (3.76 years) Kaplan et al. (2000).

^¶¶¶Maximum reported age at last birth reported in captivity: Atsalis & Margulis (2006).

^****Average of maximum age at last birth in four P. troglodytes populations: Gombe (44 years) Goodall Institute; Mahale (39 years) Nishida et al. (2003); Tai (44 years) Boesch & Boesch-Achermann (2000); Bossou (41 years) Sugiyama (2004).

^††††Average age at last birth: Hill & Hurtado (1996); Howell (1979); Martin et al. (2003).

^‡‡‡‡Body mass reported for wild populations Plavcan & van Schaik (1997).

^§§§§Average of ethnographic sample reported in Jenike (2001; Table 5).

The rapid growth of gorillas may be related to their diet. Leigh (1994) examined the diet, ecology and growth rates of 42 anthropoid primate species and found that those with more folivorous diets tend to grow faster than those with more frugivorous diets. This association may simply reflect nutritional adaptations, but it is also likely to be influenced by the lowered ecological risks and intraspecific feeding competition associated with a folivorous diet (Janson & van Schaik, 1993). Without these constraints, folivores are able to have faster infant and juvenile growth rates (Leigh, 1994). All great ape species, including gorillas, favor fruit when it is abundant, but chimpanzees and orangutans specialize on fruit and extractive foods (such as insects) and sometimes chimpanzees favor vertebrate meat. In contrast, bonobos to some extent, and gorillas in particular, fall back on vegetative foods that tend to be abundant, but are of lower quality (Conklin-Brittain et al. 2001; Malenky et al. 1994). The diets of archaic hominins are generally reconstructed as being dominated by vegetative items, such as fruits and seeds (e.g. Schoeninger et al. 2001), so if diet influences growth trajectories, then these early hominins would be expected to have growth and reproductive rates closer to those of chimpanzees and orangutans than to gorillas. Also, the available fossil evidence suggests that the body size of archaic hominins is more similar to that of chimpanzees than to gorillas (McHenry, 1994). Average growth rates for modern human females are close to the rates for chimpanzees, bonobos and orangutans (Blurton Jones, 2006). For these reasons, we suggest that chimpanzees and orangutans provide the most appropriate models from which to reconstruct the life history variables of archaic hominins and we refer to data for gorillas only when relevant.

Comparing great ape life history estimates

To develop proper comparisons between modern humans and the other extant great apes we primarily rely on life history parameters estimated from modern human hunter-gatherers, because their diets, mobility, foraging styles, and population densities most likely resemble those of modern humans prior to the introduction of agriculture. While we refer to estimates drawn from a broader range of modern human populations for some of the variables in the text, in Table 2 whenever possible we use estimates derived from detailed studies of extant hunter-gatherers. This reduces concern about possible effects of improvements in diet and medical care on rates of development and senescence. We are aware, however, that it can be argued that the estimates are conservative in that ethnographically known populations of hunter-gatherers mostly occupy environments that are marginal for agriculture, thus these data are likely to sample only a subset of the habitats initially colonized by modern humans.

The non-human great ape data primarily come from long-term field studies and these data are constantly being revised and improved. In all the reports of studies of wild populations, the ages of many adults were estimated and maximum life spans were all based on estimates with unknown errors. The maximum life spans given in Table 2 are therefore taken from captive individuals of known ages. The mortality profiles constructed for wild populations do not suggest either stationary or growing populations, implying that the observed mortalities are higher now than they have been until quite recently.

Comparisons of data in Table 2 show that modern humans differ in the following ways from the other extant great apes.

Maximum potential life span

The maximum potential life span of modern humans exceeds that of the other extant great apes by several decades. Even among modern human foragers with no access to medical support, some individuals live into their 70s and 80s (Blurton Jones et al. 1999, 2002; Hill & Hurtado, 1996; Howell, 1979; Lee, 1968). In contrast, chimpanzees in the wild usually die before they reach 45 (Hill et al. 2001) and orangutans before age 50 (Wich et al. 2004). This difference in life span persists under the best captive conditions; maximum recorded longevity for great apes is around 60 years (Erwin et al. 2002), while the oldest modern human on record died at 122 (Robine & Allard, 1998). These data show that modern humans have an increased maximum life span relative to the inferred ancestral state (i.e. around 45–50 years in non-human great apes) by at least 20–30 years, and maximum life span and average adult life span are correlated (Charnov 1993; Hawkes, 2006a; Sacher, 1959). Chimpanzee (Hill et al. 2001) and orangutan (Wich et al. 2004) females in the wild who survive to age 15 can expect to live only an additional 15–20 years (probably more for orangutans), whereas modern human hunter-gatherers at age 15 can expect to live about twice that long (Howell, 1979; Hill & Hurtado, 1996; Blurton Jones et al. 2002). Among modern human foragers about 30% of those over the age of 15 are past the age of 45, while this is true of less than 3% of wild chimpanzees (Hawkes & Blurton Jones, 2005).

Longer adult life spans reflect lower adult mortality. When extrinsic adult mortality is as low as it is among great apes, adults can live long enough to display signs of declining physiological performance and eventually die from age-specific frailty. Ricklefs (1998) showed that in species with adult life spans similar to chimpanzees, about 69% of adult deaths result from age-related causes. Selection can favor slower rates of aging if the fitness benefits of extending vigorous physical performance exceed the costs of increased somatic maintenance and repair. Slower rates of aging may account for the differences between modern human and non-human great ape maximum life spans (Hawkes, 2003). While there is little systematic evidence documenting age-specific declines in physical performance in the non-human great apes, qualitative descriptions suggest that, as expected from their relatively shorter life spans, chimpanzees do age faster than modern humans. Goodall (1986) classified chimpanzees at Gombe as ‘old’ when they reached the age of 33 years. Finch & Stanford (2004) report that chimpanzee individuals aged 35 years or more ‘show frailty and weight loss’ and the ‘external indications of senescence include sagging skin, slowed movements, and worn teeth’ (ibid, p. 4). Thus, when chimpanzees in the wild reach their mid-30s they appear to age rapidly and die within a decade. In contrast, studies of physical performance among hunters and gatherers show that vigor declines more slowly with age. Measures such as muscle strength in hunter-gatherer women decrease slowly over many decades (Blurton Jones & Marlowe, 2002; Walker & Hill, 2003). Comparable data on the physical performance of the great apes are needed to test whether they do in fact age more quickly than people.

Age at first birth

As expected from an extension in life span, Table 2 shows that age at first reproduction among modern humans is later than in the other great apes, and has increased from what is inferred to be the ancestral state (see below) by 4–6 years. The age at first birth of chimpanzees and bonobos in the wild, while variable, shows a central tendency toward age 13 and 14, respectively. This is the inferred ancestral state for the Pan-Homo and the hominin clades. For gorillas the mean age at first birth is 10 years and orangutans bear their first offspring at around 15.6 years old. Mean age at first birth among modern human foraging populations is 19.5 years.

These central tendencies persist for all great ape species in spite of differences in environment and ecology among populations in the wild. Captivity seems to have only a modest effect on age at first birth (Bentley, 1999). It is often assumed that superabundance of food enhances physical condition, accelerates the timing of first birth and extends longevity. However, there is evidence that the husbandry practices and socioecological conditions of many captive colonies do not always maximize the welfare of great apes and, indeed, often increase the incidence of vascular disease, obesity, and stress (DeRousseau, 1994; Finch & Stanford, 2003). Captive chimpanzees and bonobos bear their first offspring when they are around 11 years old (Bentley, 1999; Knott, 2001; Sugiyama, 2004) and while this mean is earlier than the central tendency of age at first birth among their wild counterparts, it is within the age range of at least one wild population. Age at first birth for gorillas in captivity is virtually identical to those in the wild (9.3 versus 10 years, Harcourt & Stewart, 2007). Captive orangutan females show the largest shift in age at first birth from their wild counterparts. Markham (1995) reports age at first birth for orangutans in captivity as 11.5 years, almost 4 years earlier than orangutans in the wild. However, whether in the wild or captivity, orangutans have the latest age at first birth and are the ‘slowest’ of the non-human great ape species.

There is surprisingly little variation in average age at first birth among modern humans. Even under conditions of ample food supply and medical care, cross-culturally modern human females, on average, bear their first offspring after 18 years of age (Bogin, 1999; Martin et al. 2003). Data from historic records indicate that the average age at first birth occurred even later than at present (LeBourg et al. 1993; Westendorp & Kirkwood, 1998; Korpelainen, 2000, 2003; Low et al. 2002; Smith et al. 2003; Grundy & Tomassini, 2005; Helle et al. 2005; Pettay et al. 2005). These data emphasize the limited plasticity of life history traits even when resources are abundant.

Later age at first birth allows energy to be invested in growth over a longer juvenile period and thus most mammals with slower life histories also have larger body sizes (Purvis & Harvey, 1995). Larger mothers have greater resources for offspring production and great ape mothers translate this energy into larger, more expensive babies than is the case for other primates (Stearns, 1992; Hawkes, 2006b).

Gestation length

Larger primate mothers have larger babies (Robson et al. 2006). The large size of modern human neonates is achieved through a gestation that is between 10 to 30 days longer than for the other great apes (Haig, 1999; Dufour & Sauther, 2002). While this difference appears slight, modern human newborns spend the weeks prior to parturition accumulating large adipose fat stores (Southgate & Hey, 1976) and it is these fat stores that account for the relatively larger size of modern human neonates. Across mammals neonatal fat stores scale allometrically with body size (Widdowson, 1950). Modern human neonates, however, are over three times fatter than expected for a mammal of their size (Kuzawa, 1998). At birth, 12–15% of modern human neonatal body weight is adipose tissue (Fomon et al. 1982). While there are no data documenting the body fat of non-human great ape infants, the qualitative difference in the amount of body fat between modern humans and the other great apes is apparent. Schultz (1969) made the general observation that ‘most human babies are born well padded with a remarkable amount of subcutaneous fat, whereas monkeys and apes have very little, so that they look decidedly ‘skinny’ and horribly wrinkled’ (ibid, p. 152).

Age at weaning and interbirth intervals

Species with slow life histories generally have relatively later ages at weaning and longer interbirth intervals. Great apes, especially the frugivorous chimpanzees and orangutans, wean their offspring relatively late (around ages 4–5 and 6–8, respectively) and have long interbirth intervals (around 5–6 and 7–9 years, respectively). However, while modern humans have the slowest life history in many respects, we wean our infants comparatively early. Modern human foragers typically wean their infants by 3 years of age and have mean interbirth intervals of around 3.7 years. Like age at first birth, modern human weaning ages are consistent across a broad range of ecologies, so that weaning in modern humans occurs ‘between 2–3 years and generally occurs about midway in that range’ (Kennedy 2005: p. 7).

Many different ways have been proposed to estimate expected (‘natural’) weaning age from other modern human life history variables and most predict later weaning ages than have been observed (Sellen, 2001). Harvey & Clutton-Brock (1985) predict an average weaning age of 3.36 years based on a correlation between maternal and infant body size, but Charnov & Berrigan (1993) note that mammalian infants are generally weaned when they achieve one-third of maternal body weight (Lee et al. 1991), which for modern humans occurs around 6.4 years. Smith (1992), following Schultz (1956), found that across a sample of primates weaning age correlated with the eruption of the first permanent molar, an event that occurs around 6 years in modern humans (see Table 4). The observed modern human weaning age of 2–3 years is substantially earlier than these predictions, and this is all the more remarkable because other aspects of our life history have slowed down relative to the ancestral state (Smith & Tompkins, 1995).

Table 4.

Eruption and crown formation schedules for permanent teeth of extant great ape species

	I1	I1	I2	I2	M1	M1	M3	M3
(A) Chronological age at crown completion (years)
Orangutan (Pongo sp.)					2.9–3.1‡	2.81§
Gorilla (Gorilla sp.)				2.7§	2.9§
Chimpanzee (P. troglodytes)	4.0*	4.5–5.4*	4.5*	5.0–5.2*	2.1–2.3¶	1.69–3.05¶		6.9–8.0¶
Modern human (H. sapiens)†	4.2–5.0	3.4–3.8	4.8–5.1	3.8–4.2	3.0	3.1–3.3	9.3–9.4	11.2–11.3

	M1	Age at weaning	M3	Age at first birth
(B) Chronological age at molar eruption and corresponding life history event (years)
Orangutan (Pongo sp.)	∼3.5–4.9**	7.0	∼10**	15.6
Gorilla (Gorilla sp.)	3.0–4.0**	4.1	8.7–13.1**	10
Chimpanzee (P. troglodytes)	2.66–4.08††	4.5	8–14††	13.3
Modern human (H. sapiens)	5.84 (4.74–7.0)‡‡	2.8	19.8–20.4**	19.5

^*Reid et al. (1998).

^†Reid & Dean (2006). Initiation ages: UI1 = 128 days, UI2 = 383 days, LI1 = 90 days, LI2 = 146 days, M1 = birth, M3 = 8 years old.

^‡Macho (2001); Kelley & Schwartz (2005).

^§Macho (2001).

^¶Smith et al. (2007c): Ranges reported from radiographic and histological studies of wild-born, captive-born, and unknown provanence samples. M1 initiation age = 1–2 months prior to birth.

^**Smith et al. (1994); Kelley & Schwartz (2005).

^††Smith et al. (2007b; consensus range from Table 11).

^‡‡Liversidge (2003): mean (range) of 56 world-wide modern human populations.

Age at last birth and menopause

In mammals, oocytes are produced in the fetal ovaries until the third trimester of gestation when mitosis of germ cells ends. This fixed store of oocytes is subject to a process of continual depletion, or atresia, over the individual's life time (vom Saal et al. 1994; O’Connor et al. 2001; Cohen, 2004). In all higher primates, including modern humans, the cycle of ovulation and menstruation is generated by an endocrinological feedback loop that requires a sufficient store of oocytes (Wood, 1994). When insufficient oocytes remain to stimulate ovulation (estimated at around 1000 follicles, Richerson et al. 1987) cycling ceases. All menstruating primates can potentially experience the senescent cessation of menses, or menopause, if they live long enough. However, in non-human species reproductive senescence usually corresponds with somatic senescence and few species live beyond the depletion of their oocyte store.

Menopause has been well documented in captive populations of macaques (e.g. Macaca fuscata, Nozaki et al. 1995; Macaca mulatta, Walker 1995; Macaca nemestrina, Short et al. 1989) where individuals with senescent impairments live longer than they can in the wild. Data on reproductive senescence in great apes are scant, but histological examination of captive chimpanzee ovaries suggests that the process of oocyte reduction is similar to that in modern humans (Gould et al. 1981). The few captive chimpanzee females that have survived to menopause exhibited the same pattern of declining fecundity and variable cycling experienced by women (Tutin & McGinnis, 1981) and they did so around the same age (Gould et al. 1981). Counts of primordial oocytes for a sample of chimpanzees from 3 months to 47 years show the same exponential rate of decline as the rate documented in modern humans (Jones et al. 2007).

Several years prior to menopause in modern human women, the hypothalamic-pituitary-ovarian (HPO) axis begins to break down due to the number of oocytes falling below the level necessary for ovarian steroid production. During this period of ‘perimenopause’, cycle lengths become long and irregular, and many are anovulatory. The age at menopause, the permanent cessation of menstruation, is assessed retrospectively, after 1 year of no menstrual bleeding. Inconsistent functioning of the HPO axis and the increase in pregnancy failure during perimenopause results in a steep decline in the fertility of modern human females (Holman & Wood, 2001). Though age at menopause varies, Treloar (1981) found in his classic prospective study an average age of 50–51 for the complete cessation of menses. In non-contracepting modern human populations the average age at last birth precedes the average age at menopause by about 10 years (Gosden, 1985) and this pattern is similar globally. ‘With few exceptions the means [of age at last birth] fall in the 39–41-year range even when subpopulations with different ages at marriage, occupations of husbands, and numbers of infant deaths are considered’ (Bongaarts & Potter, 1983: p. 43).

There are few data documenting the pattern of age-specific fertility decline in non-human great apes, but the data available for chimpanzees suggest that fertility is close to zero at 45 years of age (Nishida et al. 1990; Boesch & Boesch-Achermann, 2000; Sugiyama, 2004; Emery Thompson et al. 2007), much as it is in modern humans (Howell, 1979; Hill & Hurtado, 1996; Muller et al. 2002; Martin et al. 2003). It appears that the age at which fertility declines in the other great apes is similar to that in modern humans (see Wich et al. 2004 on orangutans). This similarity suggests that all higher primates share the ancestral pattern of ovarian ontogeny. What is derived in modern humans is not an unusual rate, and thus an unusual timing, of reproductive decline, but a slowed rate of somatic aging, distinctively low adult mortality, and, in females, a vigorous post-menopausal life.

Life history-related variables

Many characteristics of growth and development that depend on life history are not life history traits themselves. The first-order life history variables (LHVs) described above – maximum potential life span (or average adult life span), age at first birth, gestation length, interbirth intervals and age at weaning, and age at last birth – directly summarize rates of survival and reproduction across the life span. In this section we discuss three attributes that are strongly linked with life history – body mass, brain size, and the timing and sequence of tooth formation and eruption – and evaluate how well these variables correspond with the timing of major life history events in the extant higher primates. These life history-related variables (LHRVs) are particularly relevant to palaeoanthropology because, unlike first-order life history variables, they are attributes whose values can potentially be derived from hominin fossil evidence.

Body mass

Body size plays an important role in mammalian life histories (Charnov, 1993, see Hawkes, 2006a for discussion of Charnov's model) and is positively correlated with many life history variables across a range of mammalian taxa (Harvey & Read, 1988). Specifically, there is a strong correlation across subfamilies of primates between body size and LHVs such as gestation length, weaning age, age at first reproduction, interbirth interval and maximum life span (Harvey & Clutton-Brock, 1985).

Great apes are the longest-lived and latest maturing as well as the largest of all primates. Chimpanzees, bonobos, orangutans, and modern humans all have late ages at first birth, and this allows energy to be invested in growth over a longer juvenile period and thus most mammals with slower life histories are also large (Purvis & Harvey, 1995). As previously discussed, gorillas are unusual in that they grow faster than the other great apes, including modern humans, and thus they achieve a larger adult size. The remaining great ape species share a similar growth rate (Table 2) and, as expected, achieve body sizes that generally vary with the duration of growth before maturity (Blurton Jones, 2006). Chimpanzees, bonobos, and orangutans bear their first offspring between 13 and 16 years of age, and they have similar body weights around 35 kg. Modern human females have a later average age at first birth (19.5), and grow 4–6 years longer than either Pan or Pongo. As a result, modern human females in extant foraging societies are about 10–15 kg larger than chimpanzee, bonobo, or orangutan females. Modern human foragers are generally smaller than body sizes estimated for pre-Mesolithic people (Jenike 2001; Ruff et al. 1997). Ethnographic hunter-gatherer means may therefore underestimate the average maternal size differences between humans and the hypothetical common ancestor of the Pan-Homo and hominin clades.

Brain growth trajectories and adult brain size

Encephalization is often linked to the slow pace of modern human life history because adult brain size has been shown to be correlated with many life history variables (Sacher, 1975; Harvey & Clutton Brock, 1985; Deaner et al. 2003). Having a larger than expected adult brain size for a given body size can be achieved either by extending the period of brain growth, increasing the rate of brain growth, or both (see Vinicius 2005 for review). Because most relatively large-brained mammals also have slow life histories, and because large brain size is strongly correlated with many life history events, most researchers assume that brain size and the pace of life history are physiologically linked and that encephalization causes a slowdown in life history. The idea that large brain size slows life history implies that subadulthood is extended because it takes a longer time to grow a larger brain (Kaplan et al. 2000). However, few studies have systematically examined the rate and timing of brain growth between modern humans and the other great apes to test this assumption.

There are few published datasets of brain sizes for modern human individuals of known ages. Most authors summarize their original data in figures and report parameters instead of original values, making intraspecies comparisons difficult (Jolicoeur et al. 1988; Cabana et al. 1993). Of the complete datasets published, most are derived from autopsy and necropsy records. Because these samples are made up of individuals with various pathologies it is more than likely that they do not represent the ‘normal’ population. These are cross-sectional data, not longitudinal, repeated measurements on the same individual, but these data currently provide the only opportunity to quantify brain growth and development in modern humans.

We used Marchand's (1902) dataset that reports brain weight (wet, including meninges, in grams), stature (in centimeters), sex, and known or estimated chronological age, assembled from German autopsy records documented between 1885 and 1900. The original data include a total of 716 modern human males and 452 females from birth to over 80 years old and the variation in brain size with age and sex compares favorably with other reports (Dekaban & Sadowsky, 1978; Kretschmann et al. 1979), indicating that Marchand's series can serve as a representative sample. Brain weights for chimpanzees (Pan troglodytes) of known ages were drawn from necropsy data reported by Herndon et al. (1999). Brain weights were obtained fresh, from 76 captive individuals (33 females and 43 males) at Yerkes Regional Primate Center who died from natural causes or were euthanized when natural death was imminent.

Using these two datasets, shown in Fig. 2 and summarized in Table 3, we examined how well the timing of brain growth and development corresponds proportionately with life history events. More specifically, we investigated whether a longer period of postnatal brain growth is associated with a longer subadulthood, whether a longer period of postnatal brain growth is associated with a smaller portion of adult brain size at birth, and whether a longer subadult period is commensurate with a slower rate of brain growth. We find that none of these predictions are supported. Firstly, although modern human subadulthood is over 6 years longer than that of chimpanzees (19.5 vs. 13.3 years), only one additional year is spent growing a larger brain. The outlined portion of the shaded bands in Fig. 2a highlights the length of brain growth during subadulthood and shows that, compared to modern humans, chimpanzees devote a relatively longer period of their subadulthood to brain growth. Modern humans reach adult brain size much earlier than widely claimed, some as early as 3 years of age. Kretschmann et al. (1979) used the Marchand (1902) data to show that on average modern human males achieve 95% of total brain size by 3.82 years old and females by 3.44 years old. On average, modern humans in this dataset achieve 90% of adult brain size by 5 years old, only 1 year later than the chimpanzee average (around 4 years) and much earlier than widely assumed for our long subadulthoods and slower life history.

Fig. 2.

Comparison of modern human and chimpanzee absolute (panel A) and relative (panel B) brain growth trajectories. Black triangles are chimpanzees (Herndon et al. 1999; n = 26; males = 16, females = 10); open circles are modern humans (Marchand, 1902; n = 160; males = 111, females = 49). Shaded bands in panel A represent the period of subadulthood with the duration of brain growth outlined and darkened.

Table 3.

Comparison modern human and chimpanzee absolute and relative brain size

	Average neonatal brain size (g)*	Average adult brain size(g)†	% adult brain size at birth	Age 90% of adult brain size attained	Age at sexual maturity (years)	Years from adult brain size to maturity (years)	% subadult pd left after reaching adult brain size
Modern human‡	364	1352	27%	5	19.5	14.5	74%
Chimpanzee§	137	384	36%	4	13.3	9.3	70%

^*Neonate defined as individuals from birth to 10 days old.

^†Average adult brain size was calculated as the mean of all individuals between 20–40 years old for modern humans and the mean of all individuals between 7 and 30 years old in chimpanzees because this range safely precedes a known trend toward declining brain weight with age (Dekaban & Sadowsky (1978); Herndon et al. (1999).

^‡Modern human brain data from Marchand (1902).

^§Chimpanzee brain weight data from Herndon et al. (1999).

Second, chimpanzee and modern human infants are more similar in the percentage of adult brain size achieved at birth than previously assumed. It has conventionally been reported (e.g. Dienske, 1986) that modern human neonatal brain weight is only 25% of adult size at birth, whereas chimpanzee neonates have achieved 50% of their adult brain weight at birth. But this estimate of relative chimpanzee neonatal brain size is based on the estimated cranial capacity of a single specimen (Schultz, 1941). A recent re-examination of that specimen has revealed that it was not a neonate, but was 74 days old at death (Vinicius, 2005). When plotted against the Herndon et al. (1999) values, this specimen falls in the scatter where it should be given an age of 2.5 months (Robson et al. 2006). Thus, the interspecific difference in relative brain size at birth is reduced from 25% to only 10% (see Table 3). Additional data may shrink the difference even further, weakening any remaining association between relative neonatal brain size and the length of subadulthood.

Third, chimpanzees and modern humans share a similar pattern of relative brain growth trajectories (Fig. 2b). The large brain size of modern human adults is primarily achieved by a faster rate, and not by a longer relative duration, of post-natal brain growth. Leigh (2004) conducted similar analyses using the same data and concluded that ‘after the first 18 months of life, Pan and Homo are not substantially different in terms of growth rates’ (p. 152).

These similarities between chimpanzees and modern humans do not support the view that our juvenility is longer because of the growth requirements of our large brains. Whereas adult brain size is strongly correlated with the length of subadulthood (Leigh, 2004), age at brain growth cessation is not. These data show that encephalization in primates is achieved through an increased velocity, not longer relative duration, of brain growth and challenge the widely held assumption that the length of brain growth is linked to, and sets the pace of, life history. Rather, external adult mortality and demographic profiles probably determine the pace of mammalian life history schedules and patterns of growth and development adjust to these life history constraints (Dean, 2006). From this perspective, slower life history provides an opportunity for shifts in the rate and timing of brain growth.

This analysis is important because recent studies have drawn conclusions about the developmental patterns and cognitive abilities of fossil hominins based on comparison of modern human and chimpanzee brain growth trajectories (Coqueugniot et al. 2004; Alemseged et al. 2006). We and others (Leigh, 2004; Vinicius, 2005) show that there is substantial overlap in brain growth trajectories between modern humans and chimpanzees, thus undercutting the usual basis for inferences about cognition and development.

Dental development

Any consistent relationships between dental growth and development and life history would provide a means for making direct interpretations of maturation schedules within the hominin clade. Teeth are less sensitive to developmental insults and short-term ecological fluctuations than other tissues (Nissen & Riessen, 1964; Garn et al. 1973; Liversidge, 2003), thus making them relatively reliable maturation indicators. We evaluate two forms of dental data. Firstly, we examine the potential of dental microstructure, the rate and pattern of crown and root formation, as a means of comparing life histories. Second, we evaluate the information available about the timing and sequence of tooth eruption into the jaws in the same light. Because the timing and pattern of overall dental development are considered proxies for somatic growth, and this is constrained by life history, it should in theory be possible to make inferences about shared or distinct life history patterns from these data.

Crown and root formation times

Enamel and dentin formation are especially promising lines of evidence for linking dental development with absolute calendar time (Moorrees et al. 1963; Bromage & Dean, 1985; Beynon & Dean, 1987). Because the rhythms of the incremental growth of the dental hard tissues are regular, it is possible to use those cycles of cellular activity as clocks to time the onset, duration and offset of the cellular activity responsible for the deposition of dental hard tissues (Dean, 1987; Macho & Wood, 1995b; Schwartz & Dean, 2000; Wood, 2000; and Dean, 2006 all provide reviews of the cellular basis of dental ontogeny). Specifically, the crystalline matrix secreted by enamel-forming cells (ameloblasts) and dentin-forming cells (odontoblasts) shows two discrete periodicities, a ‘short period’ (c. 24 h) and a ‘long period’ (c. 6–9 days). In enamel these physical manifestations are called ‘cross-striations’ and the ‘brown striae of Retzius’, respectively (Schwartz & Dean, 2000.) Their equivalents in dentin are ‘von Ebner's’ and ‘Andresen's lines’, respectively (Dean, 1995b, 1998; Fitzgerald, 1998; Dean, 2000).

Macho (2001) found that crown formation is broadly correlated with life history across the anthropoid primates. However, several studies have found similarities between the molar formation times of modern humans and chimpanzees (Reid et al. 1998; Smith et al. 2007a), and preliminary data suggest that this is also true for bonobos (Ramirez-Rozzi & Lacruz, 2007). We show, below, that the broader correlation of crown formation variables with life history does not operate within the narrower confines of the extant great apes.

Comparison of crown formation rates in the extant higher primates (Table 4a) shows a poor correspondence between dental microstructure and life history variables, such as age at weaning and age at first birth. Whereas the timing of life history events among the great apes fall along a continuum, crown formation times for these species are quite similar, and thus fail to track weaning ages or age at maturity. There is ‘considerable overlap among great apes and humans’ in the formation rates of both incisors and molars (Macho & Wood, 1995b: p. 23). These data show that researchers must temper expectations that individual aspects of dental development (such as anterior crown formation times) are tightly tied to age at weaning (Macho, 2001), or to age at first birth (Ramirez-Rozzi & Bermudez de Castro, 2004).

Timing of tooth formation and eruption

Schultz's much reproduced graph depicting differences in the timing of life stages across primates (e.g. Schultz, 1969) used the emergence of the first permanent teeth to mark the end of infancy, and the emergence of the last permanent teeth to mark the beginning of adulthood. Schultz (1949) also observed differences in the sequence of tooth eruption across primates. In species that are weaned relatively early, molars erupt before the deciduous teeth are lost and prior to the emergence of the anterior permanent dentition. Schultz suggested that permanent molars erupted first so that infants would be prepared to masticate food when weaned, a generalization that Smith (2000) has called ‘Schultz's rule’. Slower developing modern humans show a distinctive eruption sequence, with the permanent anterior dentition emerging before the molars. In the non-human great apes the first molar is the first permanent tooth to erupt, followed by the incisors and premolars, the second molar, and then the canine. In modern humans the first molar and first incisor erupt close together, followed by the second incisor, with the canine, premolars and second molar subsequently erupting close together (Mann et al. 1990; Conroy & Vannier, 1991a).

Dean & Wood (1981) published a provisional chart comparing modern human, chimpanzee and gorilla tooth crown and root development, and with subsequent important modifications by Anemone, Conroy and Kuykendall (summarized in Kuykendall, 2002) the chart is still used today. However, the proximate cause of these differences in eruption sequence has more to do with the roots than with the crowns. For example, one of the main differences between the dental development of modern humans and chimpanzees and gorillas, the late eruption of the first molar in the former, is caused by a temporal retardation in the final stages of root formation so that first molar eruption in modern humans occurs long after the crown and most of the root are formed (Dean, 1995a; Macho & Wood, 1995b).

Schultz speculated that the shift in eruption sequence seen in modern humans is directly connected to our slower life history and in particular to our much longer period of juvenility. Building on Schultz's recognition of a connection between dental development and life history, Smith (1989) showed that across the primates there is a correlation between the eruption of the first permanent molar (M1) and weaning age, and between the eruption of the third molar (M3) and age at first birth. However, a narrower (sensuSmith (1989) re. allometry) examination of just the great apes (Table 4b) shows that the patterns of dental maturation and eruption do not always correspond with one another, nor with the pace of life histories among these species. A comparison of age at weaning with M1 eruption and age at first birth with M3 eruption in Table 4b illustrates this lack of correspondence. The eruption of M1 precedes weaning age in gorillas and chimpanzees for a period that varies from several months to more than 1 year and in modern humans by more than 3 years. In orang-utans, M1 eruption lags behind weaning by 3 years. The age of M3 eruption is later in modern humans, but M3s do not erupt later in the later-breeding chimps and orangutans compared to gorillas. The eruption of M3 inaccurately estimates age at first birth in all the non-human great ape species by one to 5.5 years. For example, the M3 erupts at around 11 years in gorillas and chimpanzees and 10 years in orangutans, while age at first birth occurs around 10, 13.3, and 15.6 years, respectively, in these animals. These data show that among the living great apes differences in life history are not necessarily reflected in their molar eruption schedules.

The timing of tooth eruption, crown maturation, and other aspects of dental development (Godfrey et al. 2003) varies among great ape species. While the range of this variation is not independent of life history, the evidence reveals that the link is not a tight one. The robust associations among life history traits themselves reflect the necessary interdependence of population vital rates (Hawkes, 2006a), but the demographic constraints on growth and development are indirect. Life histories may change without concomitant shifts in all aspects of growth and development, and conversely selection might favor ontogenetic adjustments that are adaptations to particular problems faced by infants and juveniles in each species (Godfrey et al. 2003).

Summary

There is a distinction between first order life history variables such as age at weaning, age at sexual maturity, and life span, and second order life history variables such as body mass, brain size, and dental development. The latter, which we refer to as life history-related variables (LHRVs), are not life history variables as such, but are either linked with, or can be used to make inferences about life history variables. Life history variables can only be recorded from observations of individual living animals, which can then be pooled to generate species parameters. To the extent that LHRVs correspond with LHVs, they offer an opportunity to estimate life history parameters for fossil taxa.

A general feature of living great apes is a slow life history, so we infer this was also true of both the hypothetical Pan-Homo LCA and the stem hominin. Within the great apes, there is a distinct species order in the pace of life history. Modern humans have the slowest life history, followed by orangutans, chimpanzees and bonobos, and gorillas. Compared to chimpanzees (see Table 2), modern humans live at least 25 years longer and become sexually mature more than 6 years later. Late age at maturity results in larger mothers who then bear absolutely and relatively larger, fatter babies. These characteristics point to a lowering in adult mortality rates in the Homo lineage since the Pan-Homo split. The age at which female fertility declines to menopause appears to be the same in women as in the other extant apes, indicating that this trait has been conserved. However, modern humans have the shortest interbirth intervals and experience an earlier age at weaning than expected for an ape of our age at maturity. The distinctively fast rate of modern human reproduction results in ‘stacking’ weaned but nutritionally dependent offspring. This unique pattern is likely a derived feature of our genus and could only have evolved if mothers had a reliable source of help with food acquisition for provisioning dependent youngsters. Vigorous, postmenopausal grandmothers and adolescents, without infant dependants of their own, are unique age stages of modern human life history, and likely provided that help (Robson et al. 2006).

Constructing life histories for extinct hominin species is problematic because it depends on the extent to which LHRVs are correlated with life history. We evaluated three LHRVs, body mass, brain growth and dental development, and found that many aspects of these variables correspond imperfectly with a species life history. Previous research has shown that aspects of life history strongly correlate with these LHRVs across broad primate taxonomic groups. Our evaluation shows that these correlations do not hold within the narrow range of taxa we examine here. Many aspects in the timing of growth and development do not accurately correspond with the timing of life history in the higher primate clade.

Of the three LHRVs we examined, body mass is the best predictor of great ape life history events. While adult brain size has been found to strongly predict aspects of life history (Deaner et al. 2003), we show that the timing of brain growth is a less effective measure because it does not match up with the length of subadulthood between modern humans or chimpanzees. Both species complete brain growth between 4–5 years old and, despite their significantly larger adult brain sizes, modern humans spend relatively less time during subadulthood growing a large brain. Similarly, dental development and eruption is also a weakly related proxy for the timing of life history events and inferences about the latter from tooth formation and eruption times should be made with caution.

Part II. Inferring the life history of extinct hominin taxa

Organizing the hominin fossil record

The classification of the hominin fossil evidence is controversial, nonetheless a sound taxonomy is a prerequisite for any paleobiological investigation, including one that addresses the evolution of modern human life history. This is because the allocation of individual fossils to each hominin taxon determines the inferences drawn about the life history of that taxon. There is lively debate about how to define living species (for a discussion see Wood & Lonergan 2008), so we should not be surprised that there is a spectrum of opinion about how the species category should be applied to fossil evidence.

One of the many factors that paleoanthropologists must take into account is that the fossil record they have to work with is confined to the remains of hard tissues (bones and teeth). We know from living animals that many uncontested species (for example, Cercopithecus species) are difficult to distinguish using bones and teeth, thus there are logical reasons to suspect that a hard tissue-bound fossil record is always likely to underestimate the number of species. This has recently been referred to as ‘Tattersall's Rule’ (Antón, 2003). When discontinuities are stressed (as in so-called ‘taxic’ interpretations), and if a punctuated equilibrium model of evolution is adopted along with a branching, or cladogenetic, interpretation of the fossil record, then researchers will tend to split the hominin fossil record into a larger rather than a smaller number of species. This should be the preferred approach for life history studies for the results will be less prone to producing ‘chimeric’ life histories (Smith et al. 1994). Conversely, other researchers emphasize morphological continuity instead of morphological discontinuity, and see species as longer-lived and more prone to substantial changes in morphology through time. When this philosophy is combined with a more gradualistic or anagenetic interpretation of evolution, researchers tend to resolve the hominin fossil record into fewer, more inclusive, species. This will also be the case if researchers think in terms of allotaxa (e.g. Jolly, 2001; Antón, 2003) and allow a single species to manifest substantial regional and temporal variation.

For the reasons given above the taxonomic hypothesis we favor is the relatively speciose taxonomy in Table 5A, but in Table 5B we also provide an example of how inferences about life history would map onto the less speciose taxonomy (both taxonomies are set out in Wood & Lonergan 2008). While some researchers might contest the specific details of each of these taxonomies, we offer them as a pragmatic way to address whether and how differences in taxonomic hypotheses affect the way we interpret the evolution of modern human life history. Further details about most of the taxa and a more extensive bibliography can be found in Wood & Richmond (2000), and more recent reviews of many of these taxa can be found in Hartwig (2002), Wood & Constantino (2004) and Henry & Wood (2007).

Table 5.

(A) Splitting and (B) lumping hominin taxonomies and skeletal representation^* within the taxa in the more speciose taxonomic scheme

(A) Splitting taxonomy
Informal group	Taxa	Age (Ma)	Type specimen	Crania	Dentition	Axial	Upper limb	Lower Limb
Possible and probable primitive hominins	S. tchadensis	7.0–6.0	TM 266-01-060-1	X	X
	Orronin tugenensis	6.0	BAR 1000’00		X		X	X
	Ar. ramidus s. s.†	5.7–4.5	ARA-VP-6/1	X	X		X	ff
Archaic hominins	Australopithecus anamensis	4.2–3.9	KNM-KP 29281	ff	X		X	X
	Australopithecus afarensis s. s.	4.0–3.0	LH4	X	X	X	X	X
	Kenyanthropus platyops	3.5–3.3	KNM-WT 40000	X	X
	Australopithecus bahrelghazali	3.5–3.0	KT 12/H1		X
	Au. africanus	3.0–2.4	Taung 1	X	X	ff	X	X
Megadont archaic hominins	Au. garhi	2.5	BOU-VP-12/130	X	X		?	?
	P. aethiopicus	2.5–2.3	Omo 18.18	X	X
	P. boisei s. s.	2.3–1.3	OH 5	X	X		?	?
	P. robustus	2.0–1.5	TM1517	X	X		X	X
Transitional hominins	H. habilis s. s.	2.4–1.6	OH 7	X	X	X	X	X
	H. rudolfensis	2.4–1.6	KNM-ER 1470	X	X			?
Pre-modern Homo	H. ergaster	1.9–1.5	KNM-ER 992	X	X	X	X	X
	H. erectus s. s.	1.8–0.2	Trinil 2	X	X		X	X
	H. floresiensis‡	0.074–0.012	LB1	X	X	ff	X	X
	H. antecessor	0.7–0.5	ATD6-5	X	X
	H. heidelbergensis	0.6–0.1	Mauer 1	X	X		ff	X
	H. neanderthalensis	0.2–0.03	Neanderthal 1	X	X	X	X	X
Anatomically modern humans	H. sapiens s. s.	0.19-present	None designated	X	X	X	X	X

(B) Lumping taxonomy
Informal group	Taxa	Age (Ma)	Taxa subsumed from the splitting taxonomy
Possible and probable primitive hominins	Ar. ramidus s. l.	7.0–4.5	S. tchadensis, O. tugenensis, Ar. ramidus s. s.†
Archaic hominins	Au. afarensis s. l.	4.2–3.0	Au. anamensis, Au. afarensis s. s., Au. bahrelghazali, K. platyops
	Au. africanus	3.0–2.4	Au. africanus
Megadont archaic hominins	P. boisei s. l.	2.5–1.3	Au. garhi, P. aethiopicus, P. boisei s. s.
	P. robustus	2.0–1.5	P. robustus
Transitional hominins	H. habilis s. l.	2.4–1.6	H. habilis s. s., H. rudolfensis
Pre-modern Homo	H. erectus s. l.	1.9–0.018	H. erectus s. s., H. ergaster, H. floresiensis
Anatomically-modern humans	H. sapiens s. l.	0.7-present	H. antecessor, H. heidelbergensis, H. neanderthalensis, H. sapiens s. s.

^*Skeletal representation key: X = present, ff = fragmentary specimens, ? = taxanomic affiliation of fossil specimen(s) uncertain.

^†Recently, some specimens included in the Ar. ramidus s. s. have been raised to a separate species, Ar. kadabba (Haile-Selassie et al. (2004); however, this taxonomic distinction has not been incorporated into our analyses.

^‡Given the recent and limited publication of this taxon and its current interpretation as an isolated endemic dwarf descendent of H. erectus s. s., H. floresiensis is not included in our comparisons or analyses of life history patterns in fossil hominins.

We use the same six informal grade-based groupings (Table 5; Fig. 3) of hominin taxa that are used by Wood & Lonergan (2008). The first group, possible and probable primitive hominins, comprises Late Miocene/Early Pliocene taxa that are temporally relatively close to the estimated 5–8 Ma split between hominins and panins (taxa more closely related to modern chimpanzees than to modern humans). In the early stages of hominin evolution it may be either the lack of panin synapomorphies, or relatively subtle derived differences in the size and shape of the canines, the detailed morphology of the limbs or some unique combination of such traits, which mark out the creatures that are more closely related to modern humans than they are to chimpanzees and bonobos. This group contains a mix of taxa, some of which may belong in the hominin clade, and others of which may belong to clades that have no living representatives. The second grade grouping, archaic hominins, includes Pliocene taxa from East and southern Africa that exhibit morphology consistent with facultative bipedalism, but cranially these taxa are broadly similar to chimpanzees with regards to brain and body size. The third group, megadont archaic hominins, includes Plio-Pleistocene taxa from southern and East Africa. This group includes taxa many researchers include in the genus Paranthropus, and the distinctive cranial and dental morphology of Paranthropus includes large and robust mandibular bodies and extremely large postcanine teeth. Few, if any, postcranial fossils are unambiguously linked with any of the three taxa concerned. The fourth group, transitional hominins, includes Late Pliocene/Early Pleistocene taxa from East and southern Africa, which exhibit morphology consistent with facultative bipedalism, and some individuals in this grade have a slightly larger brain and postcanine teeth that are absolutely smaller than those of archaic hominin taxa. We place the two taxa concerned, Homo habilis sensu stricto and Homo rudolfensis in their own grade to recognize the ongoing debate about whether they should be included in the genus Homo (see Wood & Collard, 1999b). The fifth grade grouping, pre-modern Homo, includes Pleistocene taxa present in Africa and Asia, which possess morphology that is consistent with obligate bipedalism, brains that range from medium to large, and small postcanine teeth. This is the grade to which we allocate the recently reported taxon Homo floresiensis from the island of Flores, Indonesia (Brown et al. 2004; Morwood et al. 2004). This species appears to represent a late surviving hominin descendant; however, given its unique morphology and probable life history within the hominin clade it is not included in comparisons among hominin taxa. The final, sixth grade grouping, referred to as anatomically modern Homo, includes specimens located across the globe which exhibit morphology that is similar to, if not identical with, that of modern Homo sapiens (the only extant hominin taxon).

Fig. 3.

The more speciose (splitting) taxonomy. Informal groupings are based on brain size, body mass, postcanine tooth-size estimates, and locomotor mode. No ancestor-descendant relationships are implied among taxa.

Readers should be aware of two caveats with respect to the speciose taxonomy illustrated in Fig. 3. First, the age of the first and last appearances of any taxon in the fossil record (called the ‘first appearance datum’, or FAD, and ‘last appearance datum’, or LAD, respectively) almost certainly underestimates the temporal range of each taxon. It is very unlikely that we have a complete record of hominin taxonomic diversity, particularly in the pre-4 Ma phase of hominin evolution. This is because intensive explorations of sediments of this age have only been conducted for less than a decade, and because these investigations have been restricted in their geographical scope. Thus, the dataset we are working with in the early phase of hominin evolution is almost certainly incomplete. We should bear this in mind when formulating and testing hypotheses about any aspect of hominin evolution, including the evolution of modern human life history. Nonetheless, FADs and LADs provide an approximate temporal sequence for the hominin taxa. Second, we made a deliberate decision not to use lines to connect the taxa in Fig. 3. This reflects our view that within the constraints of existing knowledge there are only two relatively well-supported subclades within the hominin clade, one for Paranthropus taxa and the other for post-Homo ergaster pre-modern Homo taxa. Without well-supported subclades in the early part of the hominin fossil record it is probably unwise to begin to try to identify specific taxa as ancestors or descendants of other taxa.

Body mass

How reliably can we estimate body mass using skeletal fragments sampled from extinct taxa? Did increases in hominin body mass occur gradually within the history of species, or did it increase relatively quickly with the appearance of new species? When in hominin evolution did body mass reach the levels we see in contemporary and subrecent modern humans?

The most reliable estimates of body mass are made when the skeletal fragment is known to belong to a group for which regressions can be determined using actual body masses and skeletal measurements. This is clearly not the case for fossil hominins, for the regressions used have to be generated using data from extant, more or less, closely related groups such as the hominids, hominoids, anthropoids or simians (e.g. Aiello & Wood, 1994). In addition to this potential source of error, Richard Smith (1996) has cautioned that because paleontologists have to rely on proxies for body mass in fossil-only taxa this inevitably introduces additional error into attempts to estimate the body mass of fossil hominin taxa.

Traditionally, the most reliable body mass estimates for living taxa have come from the postcranial skeleton. But, reliably associated postcranial remains are rare in most of the hominin fossil record, and some early hominin taxon hypodigms (e.g. Paranthropus boisei) include little, or no, postcranial evidence. This has led to attempts to use cranial variables as proxies for body mass (e.g. Aiello & Wood, 1994; Kappelman, 1996; Spocter & Manger, 2007). We have compiled body mass estimates from the literature using both postcranial and cranial methods for taxa in the splitting and lumping hominin taxonomies (Table 6B,C). The published body mass estimates for H. rudolfensis used in Table 6 are more speculative than most because they are based on postcranial fossils whose links to H. rudolfensis are tentative and questionable. However, when Aiello & Wood (1994) used orbit dimensions to predict body mass directly from the KNM-ER 1470 cranium (the lectotype of H. rudolfensis), the 95% CIs (confidence intervals) they derived for its body mass (c. 43–67 kg) (Aiello & Wood, 1994, Table 8: p. 421) are very similar to the species 95% CIs given in Table 6.

Table 6.

Body mass estimates for extant great ape species, modern humans and the hominin taxa as defined in the splitting and lumping hominin taxonomies

	Species adult average		Male mean (kg)	Female mean (kg)	Sexual dimorphism‡	Method§
	Mean (kg)	95% CI†
A) Extant apes
Orangutans (Pongo sp.)	64		80	38	2.12	A
Gorillas (Gorilla sp.)	128		160	95	1.68	A
Bonobos (P. paniscus)	39		45	33	1.35	A
Chimpanzees (P. troglodytes)	41		46	35	1.31	A
Modern humans (world-wide)	49		53	46	1.16	A
(B) Splitting taxonomy
S. tchadensis	?	?	?	?
O. tugenensis	?	?	?	?
Ar. ramidus s. s.†	40	?	?	?		E
Au. anamensis	42	–72–156	51	33	1.54	B
Au. afarensis s. s.	38	31–45	45	29	1.55	B
K. platyops	?	?	?	?
Au. bahrelghazali	?	?	?	?
Au. africanus	34	30–38	41	30	1.36	B
Au. garhi	?	?	?	?
P. aethiopicus	38	?	38	?		C
P. boisei s. s.	41	–52–134	49	34	1.44	B
P. robustus	36	27–45	40	32	1.25	B
H. habilis s. s.	33	25–41	37	32	1.16	B
H. rudolfensis	55	46–64	60	51	1.18	B
H. ergaster	64	53–76	68	54	1.26	F
H. erectus s. s.	58	50–65	59	57	1.04	D,C,F
H. antecessor	?	?	?	?
H. heidelbergensis	71	62–80	84	78	1.08	F
H. neanderthalensis	72	69–76	76	65	1.17	F
H. sapiens s. s.	64	63–66	68	57	1.19	F
(C) Lumping taxonomy
Ar. ramidus s. l.	40	?	?	?		G
Au. afarensis s. l.	39	32–45	46	30	1.53	G
Au. africanus	34	30–38	41	30	1.36	G
P. boisei s. l.	40	21–59	43	34	1.26	G
P. robustus	36	27–45	40	32	1.25	G
H. habilis s. l.	46	34–57	52	41	1.27	G
H. erectus s. l.	61	55–66	65	57	1.14	G
H. sapiens s. l.	66	6–67	70	59	1.19	G

*See Appendix I for the fossil specimens used to estimate body mass for each taxon.

^†The 95% confidence intervals are calculated using a quantile from Student's t distribution, instead of a quantile of 1.96 from the normal distribution. This gives a more realistic estimate of the confidence interval for a mean derived from a small sample size (e.g. P. boisei s. s.).

^‡Body-mass sexual dimorphism calculated as the ratio of the estimated male mean and the estimated female mean body mass.

^§Method key: A = sex-specific body mass reported for wild (Plavcan & van Schaik, 1997) or ethnographic (Jenike, 2001) populations, B = based on a modern human regression of hindlimb joint size, C = based on a hominiod-derived regression of orbital area, D = based on a hominoid-derived regression of orbital height, E = a comparative estimate of upper limb joint size of Ar. ramidus and AL 288-1 (Au. afarensis), F = based on regressions of femoral head diameter and/or stature and bi-iliac breadth (see Ruff et al. 1997), and G = body mass estimates for the more inclusive taxa, calculated as the mean value of all specimens from appropriate individual taxa listed in the splitting hominin taxonomy.

Table 8.

The presence of modern human-like LHRVs within the taxa recognized in a splitting hominin taxonomy (Y = present; N = not present; ? = not known)

Informal Group	Splitting taxonomy	Body Size	Brain Mass	Dental crown and root formation	Timing of tooth formation and eruption
Basal hominins	S. tchadensis	?	N	?	?
	O. tugenensis	?	?	?	?
	Ar. ramidus s. s.	N	?	?	?
Archaic hominins	Au. anamensis	N	?	?	?
	Au. afarensis s. s.	N	N	N	?
	K. platyops	?	?	?	?
	Au. bahrelghazali	?	?	?	?
	Au. africanus	N	N	N	N
Megadont archaic hominins	Au. garhi	?	N	?	?
	P. aethiopicus	N	N	?	?
	P. boisei s. s.	N	N	N	N*
	P. robustus	N	N	N	N*
Transitional hominins	H. habilus s. s.	N	N	N	N
	H. rudolfensis	Y	N	?	?
Pre-modern Homo	H. ergaster	Y	N	N	N
	H. erectus s. s.	Y	N	N	N
	H. antecessor	?	N	N	Y
	H. heidelbergensis	Y	Y	N	Y
	H. neanderthalensis	Y	Y	Y	Y
Anatomically-modern humans	H. sapiens s. s.	Y	Y	Y	Y

^*Sequence but not timing.

The 95% CIs around the means show that the body mass estimates vary greatly in their reliability, and there are (as one would expect) differences in the parameters of those taxa that have more inclusive and less inclusive interpretations (for example, H. habilis sensu lato and H. habilis sensu stricto). However, whether one uses the lumping or the splitting taxonomy, there is apparently a substantial increase in the mean body mass of some hominin taxa with FADs around 2 Ma (Figs 4, 5). Prior to 2 Ma the estimated body mass of each hominin taxon did not appear to differ markedly from each other or from the average body mass of extant chimpanzees (c. 35–45 kg; Table 6A). Exceptions to this pattern are the estimated body masses of H. rudolfensis and H. habilis s. l. (Taxon F in Fig. 4 and D in Fig. 5), which at 2.4 Ma have estimates of mean body mass of 55 kg and 46 kg, respectively. It is important to note that in both cases the specimens from which body mass is actually being estimated and which give a reasonably large body mass estimate for H. rudolfensis and H. habilis s. l., respectively, date to ∼1.8 Ma and not to 2.4 Ma, the first appearance datum for this taxon. This apparent difference in the pattern and timing of body size evolution within hominins is an example of the influence of differing taxonomic hypotheses on the interpretation of life history evolution.

Fig. 4.

Estimated body mass plotted against first appearance date for the fossil hominin taxa recognized in the splitting taxonomy. Box and whisker plots show the median, upper, and lower quartiles (box) and the maximum and minimum values (whiskers). The number of individual estimates (n) used for each variable in this comparison is listed in the legend. Taxa represented by a single horizontal line have only a single estimate for this variable. Taxa with no data for this variable appear between question marks; their position along the vertical axis is determined by their informal group membership.

Fig. 5.

Estimated body mass plotted against first appearance date for the fossil hominin taxa recognized in the lumping taxonomy. Box and whisker plots show the median, upper, and lower quartiles (box) and the maximum and minimum values (whiskers). The number of individual estimates (n) used for each variable in this comparison is listed in the legend. Taxa represented by a single horizontal line have only a single estimate for this variable. Taxa with no data for this variable appear between question marks; their position along the vertical axis is determined by their informal group membership.

The body mass of a species can increase during hominin evolution because both males and females within a taxon are larger, or because there is a selective increase in female body mass and thus a reduction in body mass sexual dimorphism. Female body mass has long been considered a critical life history-related variable as we noted above, so it is of particular interest to see when in hominin evolution there is evidence of any significant reduction in the relatively high levels of sexual dimorphism seen in Miocene higher primates and in at least some archaic hominin taxa, such as Australopithecus afarensis and P. boisei (Lockwood et al. 1996; Silverman et al. 2001– but see Reno et al. 2003 for a different interpretation of the extent of sexual dimorphism in the former).

We calculated sexual dimorphism as the ratio of male to female estimated body mass. In the speciose hominin taxonomy (Table 6B) body mass sexual dimorphism appears to be greater than or equal to that of chimpanzees (∼ 1.3, Table 6A) until the appearance of early Homo. The less speciose hominin taxonomy (Table 6C) presents a similar pattern, with early archaic hominin taxa (e.g. Au. afarensis s. l. and Australopithecus africanus) exhibiting higher levels of body mass sexual dimorphism than chimpanzees. Paranthropus taxa and Homo habilis s. l. exhibit levels of sexual dimorphism that are similar to those of chimpanzees and sexual dimorphism decreases to modern human levels with the appearance of Homo erectus s. l.(but see the arguments in Spoor et al. 2007 and Lockwood et al. 2007 for more substantial sexual dimorphism in H. erectus s. l. and Paranthropus robustus, respectively). Thus, working back from extant H. sapiens the pattern of moderate levels of body mass sexual dimorphism seems to be consistent back to, and including, H. heidelbergensis, with greater body mass differences between presumed males and presumed females now thought to be more likely in Homo erectus s. l. and almost certainly the case in archaic hominins. The larger mean body mass of H. ergaster, which is temporally the earliest taxon included in H. erectus s. l., may be because there are no small individuals in the sample that was used to generate that estimate.

Brain mass/endocranial volume

Though measures of brain growth and development do not correspond with the timing of life history events in the extant great apes, adult brain size has been shown to be strongly correlated with many life history variables (Sacher, 1975; Harvey & Clutton-Brock, 1985; Deaner et al. 2003; and see above). While it is not possible to make direct measurements of brain size using fossil evidence, it is possible, with varying degrees of precision, to measure the volume of the cranial cavity, otherwise known as endocranial volume. Brain mass can be derived from brain volume, and brain volume can be derived from endocranial volume if allowance is made for the space occupied by endocranial vasculature and the intracranial extracerebral cerebrospinal fluid. Few fossil hominin crania are well enough preserved to be able to measure endocranial volume with the precision and accuracy one can achieve using museum specimens of extant taxa. Holloway (1983a) attempted to classify endocranial volumes recorded from fossil hominin crania according to what he considered was the likelihood that the estimated volumes were an accurate reflection of the actual volume, but most published endocranial volumes of fossil hominins lack any assessment of the precision or accuracy of the estimated volumes.

Parameters for the cranial capacity (i.e. endocranial volume) of hominin taxa in the splitting and lumping taxonomies are listed in Table 7 and illustrated in Figs 6 and 7. The confidence intervals (CIs) provided in Table 7 reflect interindividual variation within each taxon, but they take no account of the precision and accuracy of each individual endocranial volume measurement. All archaic hominins have brain sizes that do not differ significantly from P. troglodytes(∼400 cm³). The brain sizes of H. habilis s. s., H. rudolfensis, H. habilis s. l., H. ergaster and H. erectus s. s. are intermediate between the values for P. troglodytes and H. sapiens(Table 7B). The value for H. erectus s. s. is the only one in this group that is closer to that for H. sapiens than it is to that of P. troglodytes. Only Homo neanderthalensis and Homo heidelbergensis have brain sizes that are indistinguishable from those of H. sapiens(Table 7B). Thus, there appears to be a discontinuity between two LHRVs (body mass and brain size) in the timing of the appearance of the modern human expression of those variables.

Table 7.

Cranial capacity estimates for the hominin taxa recognized in the splitting and lumping hominin taxonomies^*

	Mean cranial capacity (cm3)	95% CI†	Sample size
(A) Splitting
S. tchadensis	365	?	1
O. tugenensis	?	?
Ar. ramidus s. s.†	?	?
Au. anamensis	?	?
Au. afarensis s. s.	458	335–580	4
K. platyops	?	?
Au. bahrelghazali	?	?
Au. africanus	464	426–502	8
Au. garhi	450	?	1
P. aethiopicus	410	?	1
P. boisei s. s.	481	454–507	10
P. robustus	563	–542–1668	2
H. habilis s. s.	609	544–674	6
H. rudolfensis	726	501–950	3
H. ergaster	764	640–888	6
H. erectus s. s.	1003	956–1051	36
H. antecessor	1000	?	1
H. heidelbergensis	1204	1130–1278	17
H. neanderthalensis	1426	1351–1501	23
H. sapiens s. s.	1478	1444–1512	66
(B) Lumping‡
Ar. ramidus s. l.	365	?	1
Au. afarensis s. l.	458	335–580	6
Au. africanus	464	426–502	8
P. boisei s. l.	472	447–498	12
P. robustus	563	–542–1668	2
H. habilis s. l.	648	579–716	9
H. erectus s. l.	969	919–1019	42
H. sapiens s. l.	1418	1384–1452	108

^*See Appendix I for fossil specimens included in the estimation of cranial capacity for each taxon.

^†The 95% confidence intervals are calculated using a quantile from Student's t distribution, instead of a quantile of 1.96 from the normal distribution. This gives a more realistic estimate of the confidence interval for a mean derived from very small sample sizes (for example, P. robustus).

^‡Cranial capacity estimates for these more inclusive taxa are calculated as the mean value of all specimens from appropriate individual taxa listed in the splitting hominin taxonomy above.

Fig. 6.

Estimated endocranial volume plotted against first appearance date for the fossil hominin taxa recognized in the splitting taxonomy. Box and whisker plots show the median, upper, and lower quartiles (box) and the maximum and minimum values (whiskers). The number of individual estimates (n) used for each variable in this comparison is listed in the legend. Taxa represented by a single horizontal line have only a single estimate for this variable. Taxa with no data for this variable appear between question marks; their position along the vertical axis is determined by their informal group membership.

Fig. 7.

Estimated endocranial volume plotted against first appearance date for the fossil hominin taxa recognized in the lumping taxonomy. Box and whisker plots show the median, upper, and lower quartiles (box) and the maximum and minimum values (whiskers). The number of individual estimates (n) used for each variable in this comparison is listed in the legend. Taxa represented by a single horizontal line have only a single estimate for this variable.

Dental LHRVs

Crown and root formation times: extinct species

For fossil teeth (which are not naturally fractured or from which thin-sections cannot be made) determining crown formation time involves summing the estimated duration of appositional enamel growth (that is, enamel covering the cusp of a tooth whose long period lines do not reach the surface of the crown) and the duration of imbricational enamel growth (that is, the product of the number of perikymata, defined as striae of Retzius that reach the surface of the enamel in the form of steps that resemble those of a tiled roof, and an estimated long period duration of 6–9 days).

In a recent analysis of enamel formation times in the incisors and canines of early hominins, Dean et al. (2001) counted long-period cross-striations, then used an empirically derived modal periodicity of 9 days to calculate enamel formation times, and plotted these against enamel thickness. These analyses show that archaic hominins take on average 100 days less than modern humans to reach an enamel thickness of 1000 µm. The authors conclude that ‘none of the trajectories of enamel growth in apes, australopiths or fossils attributed to H. habilis, H. rudolfensis or H. erectus falls within that of the sample from modern humans’ (Dean et al. 2001, p. 629). Similarly, in his analysis of root formation time in OH 16 (a specimen assigned to H. habilis s. s.) Dean (1995b) identified a pattern unlike modern humans.

Generally, crown formation times of anterior teeth are related to crown height (the taller the tooth, the longer it takes to form) and those of postcanine teeth are related to overall crown size (Macho & Wood, 1995b). Within fossil hominin taxa the major exception to these generalizations is that the premolar and molar crowns of P. boisei take the same time, or less, to form than in modern humans and chimpanzees, despite having crowns that are approximately twice the overall size of those of modern humans. This is due to a combination of more enamel secretion per day by ameloblasts, and a faster rate of ameloblast activation (Beynon & Wood, 1987). But we need more information before we can determine whether these differences are due to selection operating on life history, or diet, or on a combination of the two. In her analysis of crown formation times and life history evolution Macho (2001) suggests that the rapid crown formation times of P. boisei are due to a disjunction between body mass and brain mass. However, she uses an estimated body mass for P. boisei that is little different from modern humans. In fact the available evidence suggests that neither P. boisei s. s. nor P. boisei s. l. is likely to have been significantly heavier than other archaic hominin taxa (Table 6). Thus, in this respect at least, there is no evidence for a unique life history pattern for P. boisei.

Dean et al. (2001) concluded, albeit based on analysis of a single specimen, that H. neanderthalensis shared similar enamel growth rates with modern humans. Using perikymata packing patterns on the anterior dentition as a proxy for crown formation times (that is, closely spaced perikymata reflect decreased rates of maturation of enamel-forming ameloblasts and thus longer crown formation times) Ramirez-Rozzi & Bermúdez de Castro (2004) countered Dean et al., concluding that Homo antecessor and H. heidelbergensis had shorter periods of dental growth than H. sapiens (both modern and Upper Paleolithic-Mesolithic) and that H. neanderthalensis had decreased crown formation times that were derived with respect to H. antecessor and H. heidelbergensis, suggesting a shorter period of somatic growth in this taxon.

Guatelli-Steinberg et al. (2005, 2007) also analyzed Neanderthal enamel formation times by counting perikymata-packing rates, but their sample was not the same as that used by Ramirez-Rozzi & Bermúdez de Castro (2004), and contra the latter authors, Guatelli-Steinberg and colleagues report growth rates within the range of modern human variation. Guatelli-Steinberg et al. (2007) suggest that ‘the most important question of all is the degree to which variation in lateral-enamel formation time of anterior teeth reflects life-history differences among and within species’ (ibid, p. 117). The high variation in anterior tooth growth rates within modern humans (Reid & Dean, 2006) and between modern humans and other great ape populations (Dean & Reid, 2001) suggests these rates are not reliable predictors of life history (Smith et al. 2007d).

Tooth eruption is considered the best predictor of life history (Smith, 1991; Smith et al. 1994) and two recent studies (Macchiarelli et al. 2006; Smith et al. 2007d) have conducted the first preliminary attempts to associate enamel formation rates with age of tooth eruption in Neanderthals based on internal molar microstructure visible using high-resolution microcomputed tomography, but with conflicting conclusions. Macchiarelli et al. (2006), based on a single permanent Neanderthal M1 (from La Chaise-de-Vouthon, Charente, France), show enamel formation times and root completion times comparable to modern humans. They conclude that these data ‘firmly place Neanderthal life history variables within those known for modern humans’ (p. 748). Smith et al. (2007d) conducted similar analyses on the entire dentition of juvenile Neanderthal (from Scladina, Belgium) and determined the opposite, that formation times were shorter and eruption times earlier than in modern humans. Smith et al. conclude that ‘a prolonged childhood and slow life history are unique to Homo sapiens’ (p. 20220). While the majority of available evidence suggests Neanderthals, and perhaps earlier Homo species, share a similar pattern of dental growth and development with extant modern humans (Dean 2007), the scant evidence is equivocal and conclusions about similarities or differences in the pattern of dental growth and development modern humans and Neanderthals must wait until further data are accumulated.

Timing of tooth formation and eruption

The extent of root development in the teeth of living taxa can be assessed relatively crudely by radiography, and more precisely if the teeth are available for sectioning and histological analysis (Anemone, 2002). Unfortunately, all these methods are more difficult to apply to fossil hominin jaws. The mineralized bone of most fossils is resistant to conventional radiographic techniques, but images can be obtained by using computerized tomography (e.g. Conroy & Vannier, 1987). Developments in hardware and software are leading to expanded datasets for those fossil hominin taxa with large hypodigms, but even so the data for most extinct hominin taxa are still not sufficient to form definite conclusions. As noted more than a decade ago by Conroy & Vannier (1991b), just because the eruption sequence differs between modern humans and living chimpanzees it does not follow that fossil hominin taxa, whose eruption sequence is the same as that as modern humans, will have modern human rates of dental development.

Figure 8 emphasizes the complexity of the interactions between several aspects of the development of lower incisors and molars in modern humans, living chimpanzees and Paranthropus taxa. Despite similarities in gross dental ontogeny between Pan and Paranthropus (that is, eruption of M1 at ∼3 years of age), different incisor crown formation times in Pan and Paranthropus result in different eruption sequences. However, despite modern humans and Paranthropus having similar eruption sequences there are marked differences in their rates of crown and root formation. Even though similar eruption sequences can mask differences in other aspects of dental development, it is nonetheless a truism that eruption sequences are bound to differ among hominin taxa unless all aspects of dental ontogeny change their rates proportionally (Macho & Wood, 1995a).

Fig. 8.

The relationship between crown formation and eruption sequence in modern humans, Pan, and P. boisei. The vertical dashed line represents the time from the onset of crown formation to eruption. The height of the crown represents the approximate time taken for crown formation; the balance of the period to eruption represents the time taken for the root to form. The tooth crowns are approximately to scale. Infancy is taken to cease at the time of M1 eruption (*). The vertical gray bars indicate rates and patterns in common among the taxa. All three genera share similar molar crown formation times, but Pan differs from the other two in eruption schedules and Homo in root formation times. Adapted from Macho & Wood (1995b).

Bermúdez de Castro et al. (2003) compared the relative timing of tooth formation in a variety of hominin specimens, and in samples of modern humans and the non-human great apes, and found similarities between the non-human great apes and archaic hominins on the one hand, and H. antecessor, H. erectus s. s., H. heidelbergensis, and modern humans on the other, with H. ergaster (or early H. erectus s. l., depending on your taxonomic hypothesis) specimens appearing to be intermediate between these two groups.

In appropriate juvenile fossil hominin specimens it is possible to use aspects of dental microstructure, assessments of dental attrition and the sequence of eruption of the dentition to determine the age-at-death and thus compare dental development among extant great apes, modern humans and fossil hominin specimens of the same chronological age. Bromage & Dean (1985) pioneered this approach by using counts of perikymata on the central incisor crown, together with assumptions about the time it takes to begin calcification of the tooth and the time it takes to begin root formation, to more accurately age fossil specimens and thus enable comparisons with modern human dental specimens at a comparable stage of development. They did this for several fossil hominin mandibles, LH 2 (Au. afarensis), Sts 24 (Au. africanus), SK 63 (P. robustus), and KNM-ER 820 (H. ergaster) and concluded that the timing and duration of the dental development of these specimens was much closer to that of extant chimpanzees than to modern humans. However, although perikymata counts made up c. 90% of the age estimates for LH 2 and Sts 24, for KNM-ER 820 the majority of the elapsed time was based on assumptions, not observations, about ontogeny.

Subsequent studies have achieved greater accuracy and precision by sectioning whole teeth to recover information about the cellular events involved in tooth development (e.g. Dean et al. 1993; Moggi-Cecchi et al. 1998). Age at death estimates for other early Homo specimens (e.g. KNM-ER 1590 and KNM-WT 15000) assigned to H. rudolfensis and H. ergaster(or H. erectus s. l.), respectively, also suggest that the timing of dental development of these taxa was not modern human-like (Smith, 1991). However, any inferences drawn from these results must be tentative until we better understand the extent of variation of dental development within regional samples of H. sapiens (Liversidge, 2003; Reid & Dean, 2006).

Within the context of dental LHRVs such as crown and root formation time and the relative timing of tooth formation and eruption, H. neanderthalensis and Upper Paleolithic H. sapiens exhibit a modern-human like pattern, whereas the available evidence suggests that archaic and transitional hominins were more chimpanzee-like. Dental development in later H. erectus s. s., H. antecessor, and H. heidelbergensis is more derived in the modern direction than that of the archaic and transitional hominins, but the pattern is still not like that of anatomically modern humans. This would suggest that these pre-modern Homo taxa have life histories that are unlike those of either archaic hominins or modern humans.

Phylogenetic trends in fossil hominin life history-related variables

If the application of cladistic methods to the hominin fossil record was known to generate robust hypotheses about the structure of the hominin clade, then in theory we should be able to predict the primitive condition of LHRVs for each of the hominin subclades, look for any evidence of homoplasy in life history, and determine at what stage in hominin evolution the distinctive aspects of modern human life history make their appearance. However, there is disagreement about the reliability of the results of cladistic analyses of the hominin fossil record that are based on traditional metrical or non-metrical data. Some researchers (e.g. Strait & Grine, 2001, 2004) are more willing than we are to accept as reliable the results of hominin cladistic analyses. Other researchers (e.g. Corruccini, 1994), especially those who have tried to test the validity of these methods by applying them to living higher taxa for which we have independent molecular evidence about taxonomic relationships (e.g. Collard & Wood, 2000), are more skeptical about the reliability of cladistic analyses of early hominins that are based on conventional (i.e. non 3D) metric and non-metrical data.

Just as a well-supported hypothesis about evolutionary relationships among the living higher primates (see above) is essential for predicting the primitive condition for life history in the Pan-Homo and hominin clades, a robust hypothesis about evolutionary relationships among extinct hominin taxa is required to enable us to explore the evolution of life history within the hominin clade. There have been many attempts to determine phylogenetic relationships within the hominin clade. Most differ in their detailed conclusions, but nearly all (e.g. Chamberlain & Wood, 1987; Skelton & McHenry, 1992; Strait et al. 1997, 2007) share the conclusion that around 2.5 Ma the hominin clade split into two major subclades. One is the subclade that contains the megadont archaic hominins assigned to the genus Paranthropus; the other subclade includes taxa assigned to Homo(i.e. H. erectus s. l., H. heidelbergensis, and H. neanderthalensis, and the only living hominin, H. sapiens).

If we accept that a genus should be both a clade and a grade (see Wood & Collard, 1999a, 2001 and Wood & Lonergan, this volume for a discussion) then it would be natural to want to know whether all the taxa included in Paranthropus, on the one hand, and Homo, on the other, have the same life history. With respect to the Paranthropus clade, the data gathered for this review suggest that there is little evidence for any significant increase in body mass (but see Lockwood et al. 2007). There is, however, evidence for a slight increase in endocranial volume compared to modern chimpanzees (Elton et al. 2001). However, enamel and dentin formation are faster in Paranthropus taxa (see above) than they are for any other member of the Pan-Homo clade for which data are available. This suggests that if the pattern of Paranthropus life history mirrors its dental growth and development, then it was most likely distinct from that of modern humans, on the one hand, and from chimpanzees and bonobos, on the other (Kuykendall, 2003).

With regard to the Homo clade, there is disagreement about the criteria used to determine whether a taxon should be included within Homo, and thus where we should place the boundary between Homo and non-Homo hominin taxa (Wood & Collard, 1999a,b). As seen below in a summary of LHRVs present in Homo taxa (which is just one of the several categories of evidence that could be used to determine the boundaries of a genus), there is little evidence to support a grade distinction that applies to all the taxa presently included in the genus Homo.

Implications of fossil hominin life history-related data for hypotheses about the evolution of modern human life history

There are a substantial number of differences between the life history of modern humans and the life history of our closest living relatives within the genus Pan. A summary table outlining the presence of modern human-like LHRVs within a speciose hominin taxonomy is presented in Table 8. Prior to the transitional hominins, there is no evidence of any hominin taxon possessing a body size, brain size or aspects of dental development that differed significantly from what we assume (but remember that this is an untested assumption) to be the primitive life history pattern for the Pan-Homo clade.

Within the transitional hominin grade, that is H. habilis s. s. and H. rudolfensis (or H. habilis s. l. for those unconvinced that this hypodigm subsumes more than one taxon), what can be inferred about LHRVs is consistent. No LHRVs (with the possible exception of H. rudolfensis body mass) are consistent with the type of prolonged ontogeny seen in modern humans. The situation is only slightly different for H. ergaster, the mean body mass estimates for which are similar to those of modern humans. Neither its adult brain size, nor its crown and root formation times, nor the timing and sequence of its dental eruption are consistent with a modern human pattern. Middle Pleistocene H. erectus s. s. may be more modern human-like in its dental development, although the evidence is conflicting (for example, Sangiran 4 being more modern human-like and Sangiran 7 less so). Non-craniodental evidence for fossil hominin growth and development in H. ergaster/H. erectus s. s. is sparse and conflicting. Whereas some workers interpret the pattern of growth and development of the postcranial skeleton in these taxa as compatible with that of modern humans (Clegg & Aiello, 1999; Smith, 2004), others point to subtle but significant differences (Tardieu, 1998) from the ontogeny of modern humans.

The fossil material attributed to H. antecessor does not provide a good estimate of body mass, and it indicates an adult brain size similar to that of Homo erectus s. s. The crown formation times of H. antecessor are not yet modern, but there is some evidence for modern human-like timing of tooth formation and eruption. The body and brain sizes of H. heidelbergensis and H. neanderthalensis are consistent with a modern human life history. However, although both of these taxa appear to possess a modern human-like pattern of dental development, the crown formation times of the former are similar to H. antecessor and those of the latter appear to be autapomorphically rapid. Thus, depending upon the weight one wants to give to these LHRVs, and it is possible that a modern human pattern of life history was present in H. heidelbergensis and H. neanderthalensis.

Summary

Using the hominin fossil record, the second part of this contribution attempted to answer the following questions:

1) Did the unique features of modern human life history (or LHRVs) appear piecemeal, or did they appear suddenly as an integrated package?

2) If they did appear as an integrated package, did that package appear when large-bodied hominins with modern human skeletal proportions emerged?

3) Were modern human and modern chimpanzee life histories the only ways that life history has been configured within the Pan-Homo clade?

The clear contrasts between the life history of modern humans and the life history of our closest living relatives, the chimpanzees, perhaps promoted the expectation that there would be a point in evolutionary history when all these variables switched simultaneously from their primitive non-modern human condition to the modern human condition. The reality seems to be more complicated. Some LHRVs (for example, body mass) shift to the modern human condition earlier while others, for example, some aspects of dental development, do not appear until the Middle/Late Pleistocene with H. neanderthalensis and Homo sapiens.

Initial attempts to describe the dental ontogeny of fossil hominins were mostly confined to statements about whether it was ‘modern human-like’ or ‘ape-like’. Additional data and more sophisticated ways of displaying those data resulted in the realization that the dental ontogeny of many early hominins was distinctive, and was not an amalgam of some modern human-like characteristics and some ape-like ones (Bromage, 1987; Kuykendall, 2003). As we come to know more of the life histories of early, and most likely also later, hominins we are also discovering that they can have distinctive life histories that do not conform to any living model (see Kelley, 2002, 2004 for insightful reviews of life history evolution within living and extinct higher primates). At least one extinct hominin subclade, Paranthropus, has a pattern of dental LHRVs that most likely set it apart from the life histories of both modern humans and chimpanzees.

Conclusion

Life history is an important component of the shared adaptive mix that justifies grouping taxa into genera. The tantalizing glimpses existing data and methods have provided into the life history of taxa included in Homo, suggest that this genus, as traditionally defined, subsumes at least two different patterns of life history. If LHRVs are used to reconstruct life history, then the life histories of transitional hominins and pre-modern Homo appear to differ from each other as well as from the life history of anatomically modern Homo. How these differences relate to hominin ecology and patterns of social and cultural evolution within the hominin clade are pressing research problems. The evolution of modern human life history, as well as the evolution of life history in other parts of the hominin clade, are clearly complex though there have been some attempts to reconstruct evolutionary scenarios (Hawkes et al. 1998; O’Connell et al. 1999; Kaplan et al. 2000). The task of using the hominin fossil record to document and help understand these complexities has barely begun.

Appendix I

Notes for body mass and brain size data as used in Tables 3–4.

Table 6.Body mass estimates

Splitting taxonomy

Ar. ramidus s. s.Wood & Richmond (2000). Estimate based on the observation that shoulder joint size of Ar. ramidus is 30% larger than that of AL 288-1 (30 kg).

Au. anamensis Male estimate from Leakey et al. (1995), and female estimate from McHenry & Coffing (2000) (calculated from the ratio of male to female in Au. afarensis).

Au. afarensis s. s. Adapted from McHenry (1992). Based on A.L. 333-3, 333-4, 333-7, 333w-56 and 333x-26 for male, and 129-1a, 129-1b, 288-1, 333-6 for female.

Au. africanus Adapted from McHenry (1992). Based on Sts 34, Stw 99, 311, 389 for male, and Sts 14, Stw 25, 102, 347, 358, 392, and TM 1512 for female.

Paranthropus aethiopicus Taken from Kappelman (1996). Based on KNM-WT 17000 for male.

P. boisei s. s. Adapted from McHenry (1992). Based on KNM-ER 1464 for male, and KNM-ER 1500 for female [but see Wood & Constantino (2004) for a discussion of whether KNM-ER 1500 can be confidently assigned to P. boisei].

P. robustus Adapted from McHenry (1992). Based on SK 82 and 97 for male, and SK 3155 and TM 1517 for female.

H. habilis s. s. Adapted from McHenry (1992). Based on KNM-ER 3735 (1503) for male, and OH 8 and 35 for female.

H. rudolfensis Adapted from McHenry (1992). Based on KNM-ER 1481 and 3228 for male, and KNM-ER 813 and 1472 for female. It is possible that some or all of these specimens belong to H. ergaster and not H. rudolfensis.

H. ergaster Adapted from Ruff et al. (1997). Based on KNM-ER 736, 1808 and KNM-WT 15000 for male and KNM-ER 737 for female.

H. erectus s. s. Taken from Aiello & Wood (1994) (Sangiran 17 = male), Kappelman (1996) (Zhoukoudian XI = female) and adapted from Ruff et al. (1997) (OH 28 and OH 34 = female; Zhoukoudian FeIV = no sex determination).

H. heidelbergensis Adapted from Ruff et al. (1997), Rosenberg et al. (1999) (Jinniushan) and Arsuaga et al. (1999) (Atapuerca-SH-1). Species estimate based on Atapuerca (SH) Pelvis 1 (m), Broken Hill 689, Broken Hill 690, Broken Hill 691, Broken Hill 719 (m), Broken Hill 907, Boxgrove 1 (m), Jinniushan (f), Arago 44 (m), Gesher-Benot-Ya’acov, KNM-BK 66, Ain Maarouf 1. Male (n = 4) and female (n = 1) estimates based on sex determinations taken from references and denoted by (m) and (f), respectively.

H. neanderthalensis Adapted from Ruff et al. (1997). Species estimate based on Amud 1 (m), La Chapelle-aux-Saints (m), La Ferrassie 1 (m), La Ferrassie 2 (f), Kebara 2 (m), Neanderthal 1 (m), La Quina 5, Regourdou 1, Saint-Cesaire 1 (m), Spy 1 (f), Spy 2 (m), Shanidar 1 (m), Shanidar 3 (m), Shanidar 5 (m), Krapina 207 (m), Krapina 208 (f), Krapina 209 (f), Krapina 213 (m), Krapina 214 (f), Shanidar 2 (m), Shanidar 4 (m), Shanidar 6 (f), Tabun C1 (f). Male (n = 14) and female (n = 7) estimates based on sex determinations taken from Ruff et al. (1997) and denoted by (m) and (f), respectively.

H. sapiens s. s. Adapted from Ruff et al. (1997). Species estimate based on Qafzeh 3 (f), Qafzeh 7 (m), Qafzeh 8 (m), Qafzeh 9 (f), Skhul 4 (m), Skhul 5 (m), Skhul 6 (m), Skhul 7 (m), Skhul 7a (f), Skhul 9 (m) and 104 specimens (49 male, 31 female and 24 unsexed) dated to between 10 ky and 35 ky BP. Male (n = 56) and female (n = 36) estimates based on sex determinations from Ruff et al. (1997) and denoted by (m) and (f), respectively.

Lumping taxonomy

Au. afarensis s. l. Includes specimens attributed to Au. afarensis s. s. and Au. anamensis. Sample sizes: species estimate (n = 11); male (n = 6), female (n = 5).

P. boisei s. l. Includes specimens attributed to P. boisei s. s. and P. aethiopicus. Sample sizes: species estimate (n = 3); male (n = 2), female (n = 1).

H. habilis s. l. Includes specimens attributed to H. habilis s. s. and H. rudolfensis. Sample sizes: species estimate (n = 7); male (n = 3), female (n = 4). It is possible that some or all of the specimens attributed to H. rudolfensis in this calculation actually belong to H. ergaster.

H. erectus s. l. Includes specimens attributed to H. erectus s. s. and H. ergaster. Sample sizes: species estimate (n = 9); male (n = 4), female (n = 4).

H. sapiens s. l. Includes specimens attributed to H. sapiens s. s., H. neanderthalensis, and H. heidelbergensis. Sample sizes: species estimate (n = 148); male (n = 74), female (n = 42).

Table 7.Cranial capacity estimates

Splitting taxonomy

Sahelanthropus tchadensis Based on TM 266-01-060-1 (Zollikofer et al. 2005)

Au. afarensis s. s. Based on AL 162-28, 333-45 (Delson et al. 2000), 333-105 – adult est. (Holloway, 1983b), 444-2 (Kimbel et al. 2004).

Au. africanus Based on MLD 1, 37/38; Sts 19/58, 5, 60, 71; Taung – adult est. (Delson et al. 2000), Stw 505 (550 cc. Holloway pers. comm. 2003).

Australopithecus garhi Based on BOU-VP-12/130 (Asfaw et al. 1999).

P. aethiopicus Based on KNM-WT 17000 (Walker et al. 1986)

P. boisei s. s. Based on KGA 10-125 (Suwa et al. 1997), KNM-ER 406, 13750; Omo L338y-6 (Delson et al. 2000), KNM-ER 407, 732; OH 5 (Falk et al. 2000), KNM-ER 23000; Omo 323-1976-896, KNM-WT 17400 (Brown et al. 1993); and (Holloway, 1988).

P. robustus Based on SK 1585 (Falk et al. 2000), TM 1517 (Broom & Robinson, 1948).

H. habilis s. s. Based on KNM-ER 1805, 1813; OH 7, 13, 24 (Delson et al. 2000), OH 16 – adult est. (Tobias, 1971).

H. rudolfensis Based on KNM-ER 1470, 1590, 3732 (Delson et al. 2000).

H. ergaster Based on D2280, 2282 (Gabunia et al. 2000), 2700 (Vekua et al. 2002); KNM-ER 3733, 3883 (Delson et al. 2000); KNM-WT 15000 (Begun & Walker, 1993). Note that the inclusion of the recently discovered KNM-ER 42700 does not change the average cranial capacity of H. ergaster by more than 10 cm³ if the actual capacity is close to 720 as tentatively reported (Leakey et al. 2003).

H. erectus s. s. Based on BOU-VP-2/66 (Asfaw et al. 2002); Ceprano (Ascenzi et al. 2000); Gongwangling 1/Lantian (Woo, 1966); Hexian/PA 830 (Wu & Dong, 1982); Narmada/Hathnora [mean of 1155 and 1421cc in Wolpoff (1999)]; Ngandong 1, 5, 6, 10, 11; Sambungmacan 1; Sangiran 4/Pith IV (Delson et al. 2000), Ngandong 7, 12; Sangiran 2/Pith II, 10/Pith VI, 17/Pith VIII; Trinil 2/Pith I; Zhoukoudian II/D, X/L1, XI/L2, XII/L3 (Grimaud-Herve, 1997); Ngandong 9; Zhoukoudian III/E1, Zhoukoudian VI (Weidenreich, 1943); Ngawi [Wolpoff cited in Antón (2002)]; OH 9, 12 (Holloway, 1983b); Perning 1/Mojokerto – adult est. (Antón, 1997); PL-1/Poyolo (Mowbray et al. 2000); Sambungmacan 3 (Márquez et al. 2001), 4 (Baba et al. 2003); Nanjing 1 (Liu, Zhang and Wu, in press); Sangiran 12/Pith VII (Holloway, 1981a), Sangiran IX (Anton and Swisher III pers. comm. 2003); Zhoukoudian V/H3 (Chiu et al. 1973).

H. antecessor Based on ATD-15 (Bermúdez de Castro et al. 1997).

H. heidelbergensis Based on Arago 21; Broken Hill-1/Kabwe; Petralona 1; Reilingen; Swanscombe 1; Vertesszollos II (Delson et al. 2000); Atapuerca 4, 5 (Arsuaga et al. 1997), 6 (Ruff et al. 1997); Bodo (Conroy et al. 2000); Dali 1 (Wu, 1981); Florisbad [Beaumont et al. cited in Aiello & Dean (1990)]; Jinniushan (Wolpoff, 1999); Ndutu (Brauer, 1984); Saldanha/Hopefield/Elandsfontein [Drennan cited in Brauer (1984)]; Sale (Holloway, 1981b); Steinheim (Ruff et al. 1997).

H. neanderthalensis Based on Amud 1; Biache-Saint Vaast; Ganovce 1; Krapina 2/B, 3/C, 4/D; La Quina 5; Monte Circeo I/Guattari 1; Neanderthal; Saccopastore I, II; Tabun C1 (Delson et al. 2000); Ehringsdorf 9; Gibraltar 1; La Chapelle-aux-Saints; La Ferrassie 1; Le Moustier 1; Teshik-Tash 1 (Grimaud-Herve, 1997); La Quina 18-adult est.; Shanidar 5 (Ruff, et al. 1997), Shanidar 1 [Stewart cited in Day (1986)]; Spy 1, 2 (Holloway, 1983b).

H. sapiens s. s. Arene Candide 1, 1-IP, 2, 4, 5; Barma Grande 2; Bruniquel 2; Cap Blanc 1; Dolni Vestonice III; Grotte des Enfants 4, 5, 6; Minatogawa 1, 2, 4; Mladec 1; Nazlet Khater 1; Oberkassel 2; Paderbourne; Pataud 1; Qafzeh 11; San Teodoro 1, 2, 3, 5; St. Germain-la-Riviere 1; Veryier 1; Zhoukoudian Up. Cave 1, 2, 3 (Ruff et al. 1997); Asselar (Tobias, 1971); Border Cave 1 (de Villiers, 1973); BOU-VP-16/1 (White et al. 2003); Brno II, III; Dolni Vestonice XIII, XIV, XV, XVI; Pavlov 1 [Vlcek cited in Schwartz & Tattersall (2002)]; Combe-Cappelle; Predmosti 3, 9 (Grimaud-Herve, 1997); Cro-Magnon 3; Mladec 2, 5 (Wolpoff, 1999); Eyasi 1 [Protsch cited in Brauer (1984)]; Jebel Irhoud 1 (Holloway, 1981b) Jebel Irhoud 2 [Ennouchi cited in Brauer (1984)]; Kanjera 1 [Coon cited in Brauer (1984)]; LH 18/Ngaloba (Brauer, 1984); Omo-Kibish 1 Omo-Kibish 2 [Day cited in Brauer (1984)]; Qafzeh 6 (Vallois & Vandermeersch, 1972) Qafzeh 9 [Genet-Varcin cited in Brauer (1984)]; Singa 1 [Wells cited in Stringer (1979)]; Skhul 4 [McCown and Keith cited in Brauer (1984)]; Brno I; Chancelade 1; Cro-Magnon 1; Oberkassel 1; Predmosti 4, 10, Skhul 5, 9; Yinkou (Delson et al. 2000).

Lumping taxonomy

P. boisei s. l. Includes specimens attributed to P. boisei s. s. and P. aethiopicus.

H. habilis s. l. Includes specimens attributed to H. habilis s. s. and H. rudolfensis.

H. erectus s. l. Includes specimens attributed to H. erectus s. s. and H. ergaster.

H. sapiens s. l. Includes specimens attributed to H. sapiens s. s., H. neanderthalensis, H. heidelbergensis, H. antecessor, and Fontechevade (Delson et al. 2000).